apkzlib 7.1.0-alpha10 (#19)

apkzlib/.gitignore

@@ -0,0 +1 @@
+/build

apkzlib/build.gradle

@@ -0,0 +1,18 @@
+plugins {
+    id 'java-library'
+}
+
+java {
+    sourceCompatibility = JavaVersion.VERSION_11
+    targetCompatibility = JavaVersion.VERSION_11
+}
+
+dependencies {
+    implementation 'com.google.code.findbugs:jsr305:3.0.2'
+    implementation 'org.bouncycastle:bcpkix-jdk15on:1.69'
+    implementation 'org.bouncycastle:bcprov-jdk15on:1.69'
+    api 'com.google.guava:guava:30.1.1-jre'
+    api 'com.android.tools.build:apksig:7.0.1'
+    compileOnlyApi 'com.google.auto.value:auto-value-annotations:1.8.2'
+    annotationProcessor 'com.google.auto.value:auto-value:1.8.2'
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/AbstractCloseableByteSourceFromOutputStreamBuilder.java

@@ -0,0 +1,86 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.base.Preconditions;
+import java.io.IOException;
+
+/**
+ * Abstract implementation of a {@link CloseableByteSourceFromOutputStreamBuilder} that simplifies
+ * the implementation of concrete instances. It implements the state machine implied by the
+ * interface contract and requires subclasses to implement two methods:
+ * {@link #doWrite(byte[], int, int)} -- that actually does writing and {@link #doBuild()} that
+ * builds the {@link CloseableByteSource].
+ */
+abstract class AbstractCloseableByteSourceFromOutputStreamBuilder
+    extends CloseableByteSourceFromOutputStreamBuilder {
+
+  /**
+   * Array that allows {@link #write(int)} to delegate to {@link #write(byte[], int, int)} without
+   * having to create an array for each invocation.
+   */
+  private final byte[] tempByte;
+
+  /**
+   * Has the builder been closed? If it has, then {@link #build()} may be called, but none of the
+   * writing methods can.
+   */
+  private boolean closed;
+
+  /**
+   * Has the builder been built? If this is {@code true} then {@link #closed} is also {@code true}.
+   */
+  private boolean built;
+
+  /** Creates a new builder. */
+  AbstractCloseableByteSourceFromOutputStreamBuilder() {
+    tempByte = new byte[1];
+    closed = false;
+    built = false;
+  }
+
+  @Override
+  public void write(byte[] b, int off, int len) throws IOException {
+    Preconditions.checkState(!closed);
+    doWrite(b, off, len);
+  }
+
+  @Override
+  public void write(int b) throws IOException {
+    tempByte[0] = (byte) b;
+    write(tempByte, 0, 1);
+  }
+
+  @Override
+  public void close() throws IOException {
+    closed = true;
+  }
+
+  @Override
+  public CloseableByteSource build() throws IOException {
+    Preconditions.checkState(!built);
+    closed = true;
+    built = true;
+
+    return doBuild();
+  }
+
+  /**
+   * Same as {@link #write(byte[], int, int)}, but with the guarantee that the source has not been
+   * built and the builder is still open.
+   *
+   * @param b see {@link #write(byte[], int, int)}
+   * @param off see {@link #write(byte[], int, int)}
+   * @param len see {@link #write(byte[], int, int)}
+   * @throws IOException see {@link #write(byte[], int, int)}
+   */
+  protected abstract void doWrite(byte[] b, int off, int len) throws IOException;
+
+  /**
+   * Builds the {@link CloseableByteSource} from the written data. This method is at most invoked
+   * once.
+   *
+   * @return the new source that will contain all data written to the builder so far
+   * @throws IOException failed to create the byte source
+   */
+  protected abstract CloseableByteSource doBuild() throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/ByteStorage.java

@@ -0,0 +1,72 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.io.ByteSource;
+import java.io.Closeable;
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Interface for a storage that will temporarily save bytes. There are several factory methods to
+ * create byte sources from several inputs, all of which may be discarded after the byte source has
+ * been created. The data is saved in the storage and will be kept until the byte source is closed.
+ */
+public interface ByteStorage extends Closeable {
+  /**
+   * Creates a new byte source by fully reading an input stream.
+   *
+   * @param stream the input stream
+   * @return a byte source containing the cached data from the given stream
+   * @throws IOException failed to read the stream
+   */
+  CloseableByteSource fromStream(InputStream stream) throws IOException;
+
+  /**
+   * Creates a builder that is an output stream and can create a byte source.
+   *
+   * @return a builder where data can be written to and a {@link CloseableByteSource} can eventually
+   *     be obtained from
+   * @throws IOException failed to create the builder; this may happen if the builder require some
+   *     preparation such as temporary storage allocation that may fail
+   */
+  CloseableByteSourceFromOutputStreamBuilder makeBuilder() throws IOException;
+
+  /**
+   * Creates a new byte source from another byte source.
+   *
+   * @param source the byte source to copy data from
+   * @return the tracked byte source
+   * @throws IOException failed to read data from the byte source
+   */
+  CloseableByteSource fromSource(ByteSource source) throws IOException;
+
+  /**
+   * Obtains the number of bytes currently used.
+   *
+   * @return the number of bytes
+   */
+  long getBytesUsed();
+
+  /**
+   * Obtains the maximum number of bytes ever used by this tracker.
+   *
+   * @return the number of bytes
+   */
+  long getMaxBytesUsed();
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/ByteStorageFactory.java

@@ -0,0 +1,14 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.IOException;
+
+/** Factory that creates {@link ByteStorage}. */
+public interface ByteStorageFactory {
+
+  /**
+   * Creates a new storage.
+   *
+   * @return a storage that should be closed when no longer used.
+   */
+  ByteStorage create() throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/ChunkBasedByteStorage.java

@@ -0,0 +1,143 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.io.ByteSource;
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.List;
+import javax.annotation.Nullable;
+
+/**
+ * Byte storage that breaks byte sources into smaller byte sources. This storage uses another
+ * storage as a delegate and, when a source is requested, it will allocate one or more sources from
+ * the delegate to build the requested source.
+ */
+public class ChunkBasedByteStorage implements ByteStorage {
+
+  /** Size of the default chunk size. */
+  private static final long DEFAULT_CHUNK_SIZE_BYTES = 10 * 1024 * 1024;
+
+  /** Maximum size of each chunk. */
+  private final long maxChunkSize;
+
+  /** Byte storage where the data is actually stored. */
+  private final ByteStorage delegate;
+
+  /**
+   * Creates a new storage breaking sources in chunks with the default maximum size and allocating
+   * each chunk from {@code delegate}.
+   */
+  ChunkBasedByteStorage(ByteStorage delegate) {
+    this(DEFAULT_CHUNK_SIZE_BYTES, delegate);
+  }
+
+  /**
+   * Creates a new storage breaking sources in chunks with the maximum of {@code maxChunkSize} and
+   * allocating each chunk from {@code delegate}.
+   */
+  ChunkBasedByteStorage(long maxChunkSize, ByteStorage delegate) {
+    this.maxChunkSize = maxChunkSize;
+    this.delegate = delegate;
+  }
+
+  /** Obtains the byte storage chunks are allocated from. */
+  @VisibleForTesting // private otherwise.
+  public ByteStorage getDelegate() {
+    return delegate;
+  }
+
+  @Override
+  public CloseableByteSource fromStream(InputStream stream) throws IOException {
+    List<CloseableByteSource> sources = new ArrayList<>();
+    while (true) {
+      LimitedInputStream limitedInput = new LimitedInputStream(stream, maxChunkSize);
+      sources.add(delegate.fromStream(limitedInput));
+      if (limitedInput.isInputFinished()) {
+        break;
+      }
+    }
+
+    return new ChunkBasedCloseableByteSource(sources);
+  }
+
+  @Override
+  public CloseableByteSourceFromOutputStreamBuilder makeBuilder() throws IOException {
+    return new AbstractCloseableByteSourceFromOutputStreamBuilder() {
+      private final List<CloseableByteSource> sources = new ArrayList<>();
+      @Nullable private CloseableByteSourceFromOutputStreamBuilder currentBuilder = null;
+      private long written = 0;
+
+      @Override
+      protected void doWrite(byte[] b, int off, int len) throws IOException {
+        int actualOffset = off;
+        int remaining = len;
+
+        while (remaining > 0) {
+          // Since we're writing data, make sure we have a builder to create the new source.
+          if (currentBuilder == null) {
+            currentBuilder = delegate.makeBuilder();
+            written = 0;
+          }
+
+          // See how much we can write without exceeding maxChunkSize in the current builder.
+          int maxWrite = (int) Math.min(maxChunkSize - written, remaining);
+          currentBuilder.write(b, actualOffset, maxWrite);
+          written += maxWrite;
+
+          remaining -= maxWrite;
+          actualOffset += maxWrite;
+
+          // If we've reached the end of the chunk, create the source for the part we have and reset
+          // to builder so we start a new one if there is more data.
+          if (written == maxChunkSize) {
+            sources.add(currentBuilder.build());
+            currentBuilder = null;
+          }
+        }
+      }
+
+      @Override
+      protected CloseableByteSource doBuild() throws IOException {
+        // If we were writing a chunk, close it.
+        if (currentBuilder != null) {
+          sources.add(currentBuilder.build());
+          currentBuilder = null;
+        }
+
+        return new ChunkBasedCloseableByteSource(sources);
+      }
+    };
+  }
+
+  @Override
+  public CloseableByteSource fromSource(ByteSource source) throws IOException {
+    List<CloseableByteSource> sources = new ArrayList<>();
+
+    long end = source.size();
+    long start = 0;
+    while (start < end) {
+      long chunkSize = Math.min(end - start, maxChunkSize);
+      sources.add(delegate.fromSource(source.slice(start, chunkSize)));
+      start += chunkSize;
+    }
+
+    return new ChunkBasedCloseableByteSource(sources);
+  }
+
+  @Override
+  public long getBytesUsed() {
+    return delegate.getBytesUsed();
+  }
+
+  @Override
+  public long getMaxBytesUsed() {
+    return delegate.getMaxBytesUsed();
+  }
+
+  @Override
+  public void close() throws IOException {
+    delegate.close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/ChunkBasedByteStorageFactory.java

@@ -0,0 +1,40 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/**
+ * {@link ByteStorageFactory} that creates {@link ByteStorage} instances that keep all data in
+ * memory.
+ */
+public class ChunkBasedByteStorageFactory implements ByteStorageFactory {
+
+  /** Factory to create the delegate storages. */
+  private final ByteStorageFactory delegate;
+
+  /** Maximum size for chunks, if any. */
+  @Nullable private final Long maxChunkSize;
+
+  /** Creates a new factory whose storages are created using delegates from the given factory. */
+  public ChunkBasedByteStorageFactory(ByteStorageFactory delegate) {
+    this(delegate, /*maxChunkSize=*/ null);
+  }
+
+  /**
+   * Creates a new factory whose storages use the given maximum chunk size and are created using
+   * delegates from the given factory.
+   */
+  public ChunkBasedByteStorageFactory(ByteStorageFactory delegate, @Nullable Long maxChunkSize) {
+    this.delegate = delegate;
+    this.maxChunkSize = maxChunkSize;
+  }
+
+  @Override
+  public ByteStorage create() throws IOException {
+    if (maxChunkSize == null) {
+      return new ChunkBasedByteStorage(delegate.create());
+    } else {
+      return new ChunkBasedByteStorage(maxChunkSize, delegate.create());
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/ChunkBasedCloseableByteSource.java

@@ -0,0 +1,44 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.android.tools.build.apkzlib.zip.utils.CloseableDelegateByteSource;
+import com.google.common.collect.ImmutableList;
+import com.google.common.io.ByteSource;
+import com.google.common.io.Closer;
+import java.io.IOException;
+import java.util.List;
+
+/**
+ * Byte source that has its data spread over several chunks, each with its own {@link
+ * CloseableByteSource}.
+ */
+class ChunkBasedCloseableByteSource extends CloseableDelegateByteSource {
+
+  /** The sources for data of all the chunks, in order. */
+  private final ImmutableList<CloseableByteSource> sources;
+
+  /** Creates a new source from the given sources. */
+  ChunkBasedCloseableByteSource(List<CloseableByteSource> sources) throws IOException {
+    super(ByteSource.concat(sources), sumSizes(sources));
+    this.sources = ImmutableList.copyOf(sources);
+  }
+
+  /** Computes the size of this source by summing the sizes of all sources. */
+  private static long sumSizes(List<CloseableByteSource> sources) throws IOException {
+    long sum = 0;
+    for (CloseableByteSource source : sources) {
+      sum += source.size();
+    }
+
+    return sum;
+  }
+
+  @Override
+  protected synchronized void innerClose() throws IOException {
+    try (Closer closer = Closer.create()) {
+      for (CloseableByteSource source : sources) {
+        closer.register(source);
+      }
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/CloseableByteSourceFromOutputStreamBuilder.java

@@ -0,0 +1,22 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import java.io.IOException;
+import java.io.OutputStream;
+
+/**
+ * Output stream that creates a {@link CloseableByteSource} from the data that was written to it.
+ * Calling {@link #close} is optional as {@link #build()} will also close the output stream.
+ */
+public abstract class CloseableByteSourceFromOutputStreamBuilder extends OutputStream {
+
+  /**
+   * Creates the source from the data that has been written to the stream. No more data can be
+   * written to the output stream after this method has been called.
+   *
+   * @return a source that will provide the data that was written to the stream before this method
+   *     is invoked; where this data is stored is not specified by this interface
+   * @throws IOException failed to build the byte source
+   */
+  public abstract CloseableByteSource build() throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/InMemoryByteStorage.java

@@ -0,0 +1,104 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.android.tools.build.apkzlib.zip.utils.CloseableDelegateByteSource;
+import com.google.common.io.ByteSource;
+import com.google.common.io.ByteStreams;
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+
+/** Keeps track of used bytes allowing gauging memory usage. */
+public class InMemoryByteStorage implements ByteStorage {
+
+  /** Number of bytes currently in use. */
+  private long bytesUsed;
+
+  /** Maximum number of bytes used. */
+  private long maxBytesUsed;
+
+  @Override
+  public CloseableByteSource fromStream(InputStream stream) throws IOException {
+    byte[] data = ByteStreams.toByteArray(stream);
+    updateUsage(data.length);
+    return new CloseableDelegateByteSource(ByteSource.wrap(data), data.length) {
+      @Override
+      public synchronized void innerClose() throws IOException {
+        super.innerClose();
+        updateUsage(-sizeNoException());
+      }
+    };
+  }
+
+  @Override
+  public CloseableByteSourceFromOutputStreamBuilder makeBuilder() throws IOException {
+    ByteArrayOutputStream output = new ByteArrayOutputStream();
+    return new AbstractCloseableByteSourceFromOutputStreamBuilder() {
+      @Override
+      protected void doWrite(byte[] b, int off, int len) throws IOException {
+        output.write(b, off, len);
+        updateUsage(len);
+      }
+
+      @Override
+      protected CloseableByteSource doBuild() throws IOException {
+        byte[] data = output.toByteArray();
+        return new CloseableDelegateByteSource(ByteSource.wrap(data), data.length) {
+          @Override
+          protected synchronized void innerClose() throws IOException {
+            super.innerClose();
+            updateUsage(-data.length);
+          }
+        };
+      }
+    };
+  }
+
+  @Override
+  public CloseableByteSource fromSource(ByteSource source) throws IOException {
+    return fromStream(source.openStream());
+  }
+
+  /**
+   * Updates the memory used by this tracker.
+   *
+   * @param delta the number of bytes to add or remove, if negative
+   */
+  private synchronized void updateUsage(long delta) {
+    bytesUsed += delta;
+    if (maxBytesUsed < bytesUsed) {
+      maxBytesUsed = bytesUsed;
+    }
+  }
+
+  @Override
+  public synchronized long getBytesUsed() {
+    return bytesUsed;
+  }
+
+  @Override
+  public synchronized long getMaxBytesUsed() {
+    return maxBytesUsed;
+  }
+
+  @Override
+  public void close() throws IOException {
+    // Nothing to do on close.
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/InMemoryByteStorageFactory.java

@@ -0,0 +1,15 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.IOException;
+
+/**
+ * {@link ByteStorageFactory} that creates {@link ByteStorage} instances that keep all data in
+ * memory.
+ */
+public class InMemoryByteStorageFactory implements ByteStorageFactory {
+
+  @Override
+  public ByteStorage create() throws IOException {
+    return new InMemoryByteStorage();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/LimitedInputStream.java

@@ -0,0 +1,88 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Input stream that reads only a limited number of bytes from another input stream before reporting
+ * EOF. When closed, this stream will not close the underlying stream.
+ *
+ * <p>If the underlying stream does not have enough data, this stream will read all available data
+ * from the underlying stream.
+ */
+class LimitedInputStream extends InputStream {
+  /** Where the data comes from. */
+  private final InputStream input;
+
+  /** How many bytes remain in this stream. */
+  private long remaining;
+
+  /** Has EOF been detected in {@link #input}? */
+  private boolean eofDetected;
+
+  /**
+   * Creates a new input stream.
+   *
+   * @param input where to read data from
+   * @param maximum the maximum number of bytes to read from {@code input}
+   */
+  LimitedInputStream(InputStream input, long maximum) {
+    this.input = input;
+    this.remaining = maximum;
+    this.eofDetected = false;
+  }
+
+  @Override
+  public int read() throws IOException {
+    if (remaining == 0) {
+      return -1;
+    }
+
+    int r = input.read();
+    if (r >= 0) {
+      remaining--;
+    } else {
+      eofDetected = true;
+    }
+
+    return r;
+  }
+
+  @Override
+  public int read(byte[] whereTo, int offset, int length) throws IOException {
+    if (remaining == 0) {
+      return -1;
+    }
+
+    int toRead = (int) Math.min(remaining, length);
+    int r = input.read(whereTo, offset, toRead);
+    if (r >= 0) {
+      remaining -= r;
+    } else {
+      eofDetected = true;
+    }
+
+    return r;
+  }
+
+  /** Returns {@code true} if EOF has been detected in the {@code input} stream. */
+  boolean isInputFinished() {
+    return eofDetected;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/LruTrackedCloseableByteSource.java

@@ -0,0 +1,80 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.base.Preconditions;
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Byte source that, until switched, will keep itself in the LRU queue. The byte source will
+ * automatically remove itself from the queue once closed or moved to disk (see {@link
+ * #moveToDisk(ByteStorage)}. This source should not be switched explicitly or tracking will not
+ * work.
+ *
+ * <p>The source will consider an access to be opening a stream. Every time a stream is open the
+ * source will move itself to the top of the LRU list.
+ */
+class LruTrackedCloseableByteSource extends SwitchableDelegateCloseableByteSource {
+  /** The tracker being used. */
+  private final LruTracker<LruTrackedCloseableByteSource> tracker;
+
+  /** Are we still tracking usage? */
+  private boolean tracking;
+
+  /** Has the byte source been closed? */
+  private boolean closed;
+
+  /** Creates a new byte source based on the given source and using the provided tracker. */
+  LruTrackedCloseableByteSource(
+      CloseableByteSource delegate, LruTracker<LruTrackedCloseableByteSource> tracker)
+      throws IOException {
+    super(delegate);
+    this.tracker = tracker;
+    tracker.track(this);
+    tracking = true;
+    closed = false;
+  }
+
+  @Override
+  public synchronized InputStream openStream() throws IOException {
+    Preconditions.checkState(!closed);
+    if (tracking) {
+      tracker.access(this);
+    }
+
+    return super.openStream();
+  }
+
+  @Override
+  protected synchronized void innerClose() throws IOException {
+    closed = true;
+
+    untrack();
+    super.innerClose();
+  }
+
+  /**
+   * Marks this source as not being tracked any more. May be called multiple times (only the first
+   * one will do anything).
+   */
+  private synchronized void untrack() {
+    if (tracking) {
+      tracking = false;
+      tracker.untrack(this);
+    }
+  }
+
+  /**
+   * Moves the contents of this source to a storage. This will untrack the source and switch its
+   * contents to a new delegate provided by {@code diskStorage}.
+   */
+  synchronized void move(ByteStorage diskStorage) throws IOException {
+    if (closed) {
+      return;
+    }
+
+    CloseableByteSource diskSource = diskStorage.fromSource(this);
+    untrack();
+    switchSource(diskSource);
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/LruTracker.java

@@ -0,0 +1,85 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.google.common.base.Preconditions;
+import com.google.common.collect.BiMap;
+import com.google.common.collect.HashBiMap;
+import java.util.TreeSet;
+import javax.annotation.Nullable;
+
+/**
+ * A tracker that keeps a list of the last-recently-used objects of type {@code T}. The tracker
+ * doesn't define what LRU means, it has a method, {@link #access(Object)} that marks the object as
+ * being accessed and moves it to the top of the queue.
+ *
+ * <p>This implementation is O(log(N)) on all operations.
+ *
+ * <p>Implementation note: we don't keep track of time. Instead we use a counter that is incremented
+ * every time a new access is done or a new object is tracked. Because of this, each access time is
+ * unique for each object (although it will change after each access).
+ */
+class LruTracker<T> {
+
+  /** Maps each object to its unique access time and vice-versa. */
+  private final BiMap<T, Integer> objectToAccessTime;
+
+  /**
+   * Ordered set of all object's access times. This set has the same contents as {@code
+   * objectToAccessTime.value()}. It is sorted from the highest access time (newest) to the lowest
+   * access time (oldest).
+   */
+  private final TreeSet<Integer> accessTimes;
+
+  /** Next access time to use for tracking or accessing. */
+  private int currentTime;
+
+  /** Creates a new tracker without any objects. */
+  LruTracker() {
+    currentTime = 1;
+    objectToAccessTime = HashBiMap.create();
+    accessTimes = new TreeSet<>((i0, i1) -> i1 - i0);
+  }
+
+  /** Starts tracking an object. This object's will be the most recently used. */
+  synchronized void track(T object) {
+    Preconditions.checkState(!objectToAccessTime.containsKey(object));
+    objectToAccessTime.put(object, currentTime);
+    accessTimes.add(currentTime);
+    currentTime++;
+  }
+
+  /** Stops tracking an object. */
+  synchronized void untrack(T object) {
+    Preconditions.checkState(objectToAccessTime.containsKey(object));
+    accessTimes.remove(objectToAccessTime.get(object));
+    objectToAccessTime.remove(object);
+  }
+
+  /** Marks the given object as having been accessed promoting it as the most recently used. */
+  synchronized void access(T object) {
+    untrack(object);
+    track(object);
+  }
+
+  /**
+   * Obtains the position of an object in the queue. It will be {@code 0} for the most recently used
+   * object.
+   */
+  synchronized int positionOf(T object) {
+    Preconditions.checkState(objectToAccessTime.containsKey(object));
+    int lastAccess = objectToAccessTime.get(object);
+    return accessTimes.headSet(lastAccess).size();
+  }
+
+  /**
+   * Obtains the last element, the one last accessed earliest. Will return empty if there are no
+   * objects being tracked.
+   */
+  @Nullable
+  synchronized T last() {
+    if (accessTimes.isEmpty()) {
+      return null;
+    }
+
+    return objectToAccessTime.inverse().get(accessTimes.last());
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/OverflowToDiskByteStorage.java

@@ -0,0 +1,167 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.io.ByteSource;
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Byte storage that keeps data in memory up to a certain size. After that, older sources are moved
+ * to disk and the newer ones served from memory.
+ *
+ * <p>Once unloaded to disk, sources are not reloaded into memory as that would be in direct
+ * conflict with the filesystem's caching and the costs would probably outweight the benefits.
+ *
+ * <p>The maximum memory used by storage is actually larger than the maximum provided. It may exceed
+ * the limit by the size of one source. That is because sources are always loaded into memory before
+ * the storage decides to flush them to disk.
+ */
+public class OverflowToDiskByteStorage implements ByteStorage {
+
+  /** Size of the default memory cache. */
+  private static final long DEFAULT_MEMORY_CACHE_BYTES = 50 * 1024 * 1024;
+
+  /** In-memory storage. */
+  private final InMemoryByteStorage memoryStorage;
+
+  /** Disk-based storage. */
+  @VisibleForTesting // private otherwise.
+  final TemporaryDirectoryStorage diskStorage;
+
+  /** Tracker that keeps all memory sources. */
+  private final LruTracker<LruTrackedCloseableByteSource> memorySourcesTracker;
+
+  /** Maximum amount of data to keep in memory. */
+  private final long memoryCacheSize;
+
+  /** Maximum amount of data used. */
+  private long maxBytesUsed;
+
+  /**
+   * Creates a new byte storage with the default memory cache using the provided temporary directory
+   * to write data that overflows the memory size.
+   *
+   * @param temporaryDirectoryFactory the factory used to create a temporary directory where to
+   *     overflow to; the created directory will be closed when the {@link
+   *     OverflowToDiskByteStorage} object is closed
+   * @throws IOException failed to create the temporary directory
+   */
+  public OverflowToDiskByteStorage(TemporaryDirectoryFactory temporaryDirectoryFactory)
+      throws IOException {
+    this(DEFAULT_MEMORY_CACHE_BYTES, temporaryDirectoryFactory);
+  }
+
+  /**
+   * Creates a new byte storage with the given memory cache size using the provided temporary
+   * directory to write data that overflows the memory size.
+   *
+   * @param memoryCacheSize the in-memory cache; a value of {@link 0} will effectively disable
+   *     in-memory caching
+   * @param temporaryDirectoryFactory the factory used to create a temporary directory where to
+   *     overflow to; the created directory will be closed when the {@link
+   *     OverflowToDiskByteStorage} object is closed
+   * @throws IOException failed to create the temporary directory
+   */
+  public OverflowToDiskByteStorage(
+      long memoryCacheSize, TemporaryDirectoryFactory temporaryDirectoryFactory)
+      throws IOException {
+    memoryStorage = new InMemoryByteStorage();
+    diskStorage = new TemporaryDirectoryStorage(temporaryDirectoryFactory);
+    this.memoryCacheSize = memoryCacheSize;
+    this.memorySourcesTracker = new LruTracker<>();
+  }
+
+  @Override
+  public CloseableByteSource fromStream(InputStream stream) throws IOException {
+    CloseableByteSource memSource =
+        new LruTrackedCloseableByteSource(memoryStorage.fromStream(stream), memorySourcesTracker);
+    checkMaxUsage();
+    reviewSources();
+    return memSource;
+  }
+
+  @Override
+  public CloseableByteSourceFromOutputStreamBuilder makeBuilder() throws IOException {
+    CloseableByteSourceFromOutputStreamBuilder memBuilder = memoryStorage.makeBuilder();
+    return new AbstractCloseableByteSourceFromOutputStreamBuilder() {
+      @Override
+      protected void doWrite(byte[] b, int off, int len) throws IOException {
+        memBuilder.write(b, off, len);
+      }
+
+      @Override
+      protected CloseableByteSource doBuild() throws IOException {
+        CloseableByteSource memSource =
+            new LruTrackedCloseableByteSource(memBuilder.build(), memorySourcesTracker);
+        checkMaxUsage();
+        reviewSources();
+        return memSource;
+      }
+    };
+  }
+
+  @Override
+  public CloseableByteSource fromSource(ByteSource source) throws IOException {
+    CloseableByteSource memSource =
+        new LruTrackedCloseableByteSource(memoryStorage.fromSource(source), memorySourcesTracker);
+    checkMaxUsage();
+    reviewSources();
+    return memSource;
+  }
+
+  @Override
+  public synchronized long getBytesUsed() {
+    return memoryStorage.getBytesUsed() + diskStorage.getBytesUsed();
+  }
+
+  @Override
+  public synchronized long getMaxBytesUsed() {
+    return maxBytesUsed;
+  }
+
+  /** Checks if we have reached a new high of data usage and set it. */
+  private synchronized void checkMaxUsage() {
+    if (getBytesUsed() > maxBytesUsed) {
+      maxBytesUsed = getBytesUsed();
+    }
+  }
+
+  /** Checks if any of the sources needs to be written to disk or loaded into memory. */
+  private synchronized void reviewSources() throws IOException {
+    // Move data from memory to disk until we have at most memoryCacheSize bytes in memory.
+    while (memoryStorage.getBytesUsed() > memoryCacheSize) {
+      LruTrackedCloseableByteSource last = memorySourcesTracker.last();
+      if (last != null) {
+        LruTrackedCloseableByteSource lastSource = last;
+        lastSource.move(diskStorage);
+      }
+    }
+  }
+
+  /** Obtains the number of bytes stored in memory. */
+  public long getMemoryBytesUsed() {
+    return memoryStorage.getBytesUsed();
+  }
+
+  /** Obtains the maximum number of bytes ever stored in memory. */
+  public long getMaxMemoryBytesUsed() {
+    return memoryStorage.getMaxBytesUsed();
+  }
+
+  /** Obtains the number of bytes stored in disk. */
+  public long getDiskBytesUsed() {
+    return diskStorage.getBytesUsed();
+  }
+
+  /** Obtains the maximum number of bytes ever stored in disk. */
+  public long getMaxDiskBytesUsed() {
+    return diskStorage.getMaxBytesUsed();
+  }
+
+  @Override
+  public void close() throws IOException {
+    memoryStorage.close();
+    diskStorage.close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/OverflowToDiskByteStorageFactory.java

@@ -0,0 +1,50 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/**
+ * {@link ByteStorageFactory} that creates instances of {@link ByteStorage} that will keep some data
+ * in memory and will overflow to disk when necessary.
+ */
+public class OverflowToDiskByteStorageFactory implements ByteStorageFactory {
+
+  /** How much data we want to keep in cache? If {@code null} then we want the default value. */
+  @Nullable private final Long memoryCacheSizeInBytes;
+
+  /** Factory that creates temporary directories. */
+  private final TemporaryDirectoryFactory temporaryDirectoryFactory;
+
+  /**
+   * Creates a new factory with an optional in-memory size and a temporary directory for overflow.
+   *
+   * @param temporaryDirectoryFactory a factory that creates temporary directories that will be used
+   *     for overflow of the {@link ByteStorage} instances created by this factory
+   */
+  public OverflowToDiskByteStorageFactory(TemporaryDirectoryFactory temporaryDirectoryFactory) {
+    this(null, temporaryDirectoryFactory);
+  }
+
+  /**
+   * Creates a new factory with an optional in-memory size and a temporary directory for overflow.
+   *
+   * @param memoryCacheSizeInBytes how many bytes to keep in memory? If {@code null} then a default
+   *     value will be used
+   * @param temporaryDirectoryFactory a factory that creates temporary directories that will be used
+   *     for overflow of the {@link ByteStorage} instances created by this factory
+   */
+  public OverflowToDiskByteStorageFactory(
+      Long memoryCacheSizeInBytes, TemporaryDirectoryFactory temporaryDirectoryFactory) {
+    this.memoryCacheSizeInBytes = memoryCacheSizeInBytes;
+    this.temporaryDirectoryFactory = temporaryDirectoryFactory;
+  }
+
+  @Override
+  public ByteStorage create() throws IOException {
+    if (memoryCacheSizeInBytes == null) {
+      return new OverflowToDiskByteStorage(temporaryDirectoryFactory);
+    } else {
+      return new OverflowToDiskByteStorage(memoryCacheSizeInBytes, temporaryDirectoryFactory);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/SwitchableDelegateCloseableByteSource.java

@@ -0,0 +1,122 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.io.Closer;
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * Byte source that delegates to another byte source that can be switched dynamically.
+ *
+ * <p>This byte source encloses another byte source (the delegate) and allows switching the
+ * delegate. Switching is done transparently for the user (as long as the new byte source represents
+ * the same data) maintaining all open streams working, but now streaming from the new source.
+ */
+class SwitchableDelegateCloseableByteSource extends CloseableByteSource {
+
+  /** The current delegate. */
+  private CloseableByteSource delegate;
+
+  /** Has the byte source been closed? */
+  private boolean closed;
+
+  /**
+   * Streams that have been opened, but not yet closed. These are all the streams that have to be
+   * switched when we switch delegates.
+   */
+  private final List<SwitchableDelegateInputStream> nonClosedStreams;
+
+  /** Creates a new source using {@code source} as delegate. */
+  SwitchableDelegateCloseableByteSource(CloseableByteSource source) {
+    this.delegate = source;
+    nonClosedStreams = new ArrayList<>();
+  }
+
+  @Override
+  protected synchronized void innerClose() throws IOException {
+    closed = true;
+
+    try (Closer closer = Closer.create()) {
+      for (SwitchableDelegateInputStream stream : nonClosedStreams) {
+        closer.register(stream);
+      }
+
+      nonClosedStreams.clear();
+    }
+
+    delegate.close();
+  }
+
+  @Override
+  public synchronized InputStream openStream() throws IOException {
+    SwitchableDelegateInputStream stream =
+        new SwitchableDelegateInputStream(delegate.openStream()) {
+          // Can't have a lock on the stream while we synchronize the removal of nonClosedStreams
+          // because it can deadlock when called in parallel with switchSource as the lock order is
+          // reversed. The lack of synchronization is OK because we don't access any data on the
+          // stream anyway until super.close() is called.
+          @SuppressWarnings("UnsynchronizedOverridesSynchronized")
+          @Override
+          public void close() throws IOException {
+            // Remove the stream on close.
+            synchronized (SwitchableDelegateCloseableByteSource.this) {
+              nonClosedStreams.remove(this);
+            }
+
+            super.close();
+          }
+        };
+
+    nonClosedStreams.add(stream);
+    return stream;
+  }
+
+  /**
+   * Switches the current source for {@code source}. All streams are kept valid. The current source
+   * is closed.
+   *
+   * <p>If the current source has already been closed, {@code source} will also be closed and
+   * nothing else is done.
+   *
+   * <p>Otherwise, as long as it is possible to open enough input streams from {@code source} to
+   * replace all current input streams, the source if changed. Any errors while closing input
+   * streams (which happens during switching -- see {@link
+   * SwitchableDelegateInputStream#switchStream(InputStream)}) or closing the old source are
+   * reported as thrown {@code IOException}
+   */
+  synchronized void switchSource(CloseableByteSource source) throws IOException {
+    if (source == delegate) {
+      return;
+    }
+
+    if (closed) {
+      source.close();
+      return;
+    }
+
+    List<InputStream> switchStreams = new ArrayList<>();
+    for (int i = 0; i < nonClosedStreams.size(); i++) {
+      switchStreams.add(source.openStream());
+    }
+
+    CloseableByteSource oldDelegate = delegate;
+    delegate = source;
+
+    // A bit of trickery. We want to call switchStream for all streams. switchStream will
+    // successfully switch the stream even if it throws an exception (if it does, it means it
+    // failed to close the old stream). So we want to continue switching and recording all
+    // exceptions. Closer() has that logic already so we register each stream switch as a close
+    // operation.
+    try (Closer closer = Closer.create()) {
+      for (int i = 0; i < nonClosedStreams.size(); i++) {
+        SwitchableDelegateInputStream nonClosedStream = nonClosedStreams.get(i);
+        InputStream switchStream = switchStreams.get(i);
+        closer.register(() -> nonClosedStream.switchStream(switchStream));
+      }
+
+      closer.register(oldDelegate);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/SwitchableDelegateInputStream.java

@@ -0,0 +1,181 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.google.common.annotations.VisibleForTesting;
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Input stream that delegates to another input stream, but can switch transparently the source
+ * input stream.
+ *
+ * <p>Given a set of input streams that return the same data, this input stream will read from one
+ * and allow switching to read from other streams continuing from the offset that was initially
+ * read. The result is only meaningful if all streams read the same data.
+ *
+ * <p>This class allows transparently to switch between different implementations of the underlying
+ * streams (memory, disk, etc.) while transparently providing data to users. It does not support
+ * marking and it is multi-thread safe.
+ */
+class SwitchableDelegateInputStream extends InputStream {
+
+  /** The input stream that is currently providing data. */
+  private InputStream delegate;
+
+  /**
+   * Current offset in the input stream. We keep track of this to allow skipping data when switching
+   * input streams.
+   */
+  private long currentOffset;
+
+  /** Have we reached the end of stream? */
+  @VisibleForTesting // private otherwise.
+  boolean endOfStreamReached;
+
+  /**
+   * If a switch has occurred, how many bytes still need to be skipped in the input stream to
+   * continue reading from the same position?
+   */
+  private long needsSkipping;
+
+  SwitchableDelegateInputStream(InputStream delegate) {
+    this.delegate = delegate;
+    currentOffset = 0;
+    endOfStreamReached = false;
+    needsSkipping = 0;
+  }
+
+  /**
+   * Skips data in the input stream if it has been switched and there is data to skip. Will fail if
+   * we can't skip all the data.
+   */
+  private void skipDataIfNeeded() throws IOException {
+    while (needsSkipping > 0) {
+      long skipped = delegate.skip(needsSkipping);
+      if (skipped == 0) {
+        throw new IOException("Skipping InputStream after switching failed");
+      }
+
+      needsSkipping -= skipped;
+    }
+  }
+
+  /** Same as {@link #increaseOffset(long)}. */
+  private int increaseOffset(int amount) {
+    return (int) increaseOffset((long) amount);
+  }
+
+  /**
+   * Increases the current offset after reading. {@code amount} will indicate how many bytes we have
+   * read. It {@code -1} then we know we've reached the end of the stream and {@link
+   * #endOfStreamReached} is set to {@code true}.
+   */
+  private long increaseOffset(long amount) {
+    if (amount > 0) {
+      currentOffset += amount;
+    }
+
+    if (amount == -1) {
+      endOfStreamReached = true;
+    }
+
+    return amount;
+  }
+
+  @Override
+  public synchronized int read(byte[] b) throws IOException {
+    if (endOfStreamReached) {
+      return -1;
+    }
+
+    skipDataIfNeeded();
+    return increaseOffset(delegate.read(b));
+  }
+
+  @Override
+  public synchronized int read(byte[] b, int off, int len) throws IOException {
+    if (endOfStreamReached) {
+      return -1;
+    }
+
+    skipDataIfNeeded();
+    return increaseOffset(delegate.read(b, off, len));
+  }
+
+  @Override
+  public synchronized int read() throws IOException {
+    if (endOfStreamReached) {
+      return -1;
+    }
+
+    skipDataIfNeeded();
+    int r = delegate.read();
+    if (r == -1) {
+      endOfStreamReached = true;
+    } else {
+      increaseOffset(1);
+    }
+
+    return r;
+  }
+
+  @Override
+  public synchronized long skip(long n) throws IOException {
+    if (endOfStreamReached) {
+      return 0;
+    }
+
+    skipDataIfNeeded();
+    return increaseOffset(delegate.skip(n));
+  }
+
+  @Override
+  public synchronized int available() throws IOException {
+    if (endOfStreamReached) {
+      return 0;
+    }
+
+    skipDataIfNeeded();
+    return delegate.available();
+  }
+
+  @Override
+  public synchronized void close() throws IOException {
+    endOfStreamReached = true;
+    delegate.close();
+  }
+
+  @Override
+  public void mark(int readlimit) {
+    // We don't support marking.
+  }
+
+  @Override
+  public void reset() throws IOException {
+    throw new IOException("Mark not supported");
+  }
+
+  @Override
+  public boolean markSupported() {
+    return false;
+  }
+
+  /**
+   * Switches the stream used.
+   *
+   * <p>The stream that is currently in use and the new stream will be used in further operations.
+   * If this stream has already reached the end, {@code newStream} will be closed immediately and no
+   * other action is taken. If the stream has not reached the end, any exception reported is due to
+   * closing the stream currently in use, the new stream is not affected and this stream can still
+   * be used to read from {@code newStream}.
+   */
+  synchronized void switchStream(InputStream newStream) throws IOException {
+    if (newStream == delegate) {
+      return;
+    }
+
+    try (InputStream oldDelegate = delegate) {
+      delegate = newStream;
+      needsSkipping = currentOffset;
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/TemporaryDirectory.java

@@ -0,0 +1,77 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.google.common.annotations.VisibleForTesting;
+import java.io.Closeable;
+import java.io.File;
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Path;
+
+/**
+ * A temporary directory is a directory that creates temporary files. Upon close, all temporary
+ * files are removed. Whether the directory itself is removed is dependent on the actual
+ * implementation.
+ */
+public interface TemporaryDirectory extends Closeable {
+
+  /**
+   * Creates a new file in the directory. This method returns a new file that deleted, recreated,
+   * read and written freely by the caller. No assumptions are made on the contents of this file
+   * except that it will be deleted it if it still exists when the temporary directory is closed.
+   */
+  File newFile() throws IOException;
+
+  /** Obtains the directory, only useful for tests. */
+  @VisibleForTesting // private otherwise.
+  File getDirectory();
+
+  /**
+   * Creates a new temporary directory in the system's temporary directory. All files created will
+   * be created in this directory. The directory will be deleted (as long as all the files in it)
+   * when closed.
+   */
+  static TemporaryDirectory newSystemTemporaryDirectory() throws IOException {
+    Path tempDir = Files.createTempDirectory("tempdir_");
+    TemporaryFile tempDirFile = new TemporaryFile(tempDir.toFile());
+    return new TemporaryDirectory() {
+      @Override
+      public File newFile() throws IOException {
+        return Files.createTempFile(tempDir, "temp_", ".data").toFile();
+      }
+
+      @Override
+      public File getDirectory() {
+        return tempDir.toFile();
+      }
+
+      @Override
+      public void close() throws IOException {
+        tempDirFile.close();
+      }
+    };
+  }
+
+  /**
+   * Creates a new temporary directory that uses a fixed directory.
+   *
+   * @param directory the directory that will be returned; this directory won't be deleted when the
+   *     {@link TemporaryDirectory} objects are closed
+   * @return a {@link TemporaryDirectory} that will create files in {@code directory}
+   */
+  static TemporaryDirectory fixed(File directory) {
+    return new TemporaryDirectory() {
+      @Override
+      public File newFile() throws IOException {
+        return Files.createTempFile(directory.toPath(), "temp_", ".data").toFile();
+      }
+
+      @Override
+      public File getDirectory() {
+        return directory;
+      }
+
+      @Override
+      public void close() throws IOException {}
+    };
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/TemporaryDirectoryFactory.java

@@ -0,0 +1,31 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import java.io.File;
+import java.io.IOException;
+
+/**
+ * Factory that creates temporary directories. {@link
+ * TemporaryDirectory#newSystemTemporaryDirectory()} conforms to this interface.
+ */
+public interface TemporaryDirectoryFactory {
+
+  /**
+   * Creates a new temporary directory.
+   *
+   * @return the new temporary directory that should be closed when finished
+   * @throws IOException failed to create the temporary directory
+   */
+  TemporaryDirectory make() throws IOException;
+
+  /**
+   * Obtains a factory that creates temporary directories using {@link
+   * TemporaryDirectory#fixed(File)}.
+   *
+   * @param directory the directory where all temporary files will be created
+   * @return a factory that creates instances of {@link TemporaryDirectory} that creates all files
+   *     inside {@code directory}
+   */
+  static TemporaryDirectoryFactory fixed(File directory) {
+    return () -> TemporaryDirectory.fixed(directory);
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/TemporaryDirectoryStorage.java

@@ -0,0 +1,102 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.io.ByteSource;
+import com.google.common.io.ByteStreams;
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+
+/**
+ * Byte storage that keeps all byte sources as files in a temporary directory. Each data stored is
+ * stored as a new file. The file is deleted as soon as the byte source is closed.
+ */
+public class TemporaryDirectoryStorage implements ByteStorage {
+
+  /** Temporary directory to use. */
+  @VisibleForTesting // private otherwise.
+  final TemporaryDirectory temporaryDirectory;
+
+  /** Number of bytes currently used. */
+  private long bytesUsed;
+
+  /** Maximum number of bytes used. */
+  private long maxBytesUsed;
+
+  /**
+   * Creates a new storage using the provided temporary directory.
+   *
+   * @param temporaryDirectoryFactory a factory used to create the directory to use for temporary
+   *     files; this directory will be closed when the {@link TemporaryDirectoryStorage} is closed.
+   * @throws IOException failed to create the temporary directory
+   */
+  public TemporaryDirectoryStorage(TemporaryDirectoryFactory temporaryDirectoryFactory)
+      throws IOException {
+    this.temporaryDirectory = temporaryDirectoryFactory.make();
+  }
+
+  @Override
+  public CloseableByteSource fromStream(InputStream stream) throws IOException {
+    File temporaryFile = temporaryDirectory.newFile();
+    try (FileOutputStream output = new FileOutputStream(temporaryFile)) {
+      ByteStreams.copy(stream, output);
+    }
+
+    long size = temporaryFile.length();
+    incrementBytesUsed(size);
+    return new TemporaryFileCloseableByteSource(temporaryFile, () -> incrementBytesUsed(-size));
+  }
+
+  @Override
+  public CloseableByteSourceFromOutputStreamBuilder makeBuilder() throws IOException {
+    File temporaryFile = temporaryDirectory.newFile();
+    return new AbstractCloseableByteSourceFromOutputStreamBuilder() {
+      private final FileOutputStream output = new FileOutputStream(temporaryFile);
+
+      @Override
+      protected void doWrite(byte[] b, int off, int len) throws IOException {
+        output.write(b, off, len);
+        incrementBytesUsed(len);
+      }
+
+      @Override
+      protected CloseableByteSource doBuild() throws IOException {
+        output.close();
+        long size = temporaryFile.length();
+        return new TemporaryFileCloseableByteSource(temporaryFile, () -> incrementBytesUsed(-size));
+      }
+    };
+  }
+
+  @Override
+  public CloseableByteSource fromSource(ByteSource source) throws IOException {
+    try (InputStream stream = source.openStream()) {
+      return fromStream(stream);
+    }
+  }
+
+  @Override
+  public synchronized long getBytesUsed() {
+    return bytesUsed;
+  }
+
+  @Override
+  public synchronized long getMaxBytesUsed() {
+    return maxBytesUsed;
+  }
+
+  /** Increments the byte counter by the given amount (decrements if {@code amount} is negative). */
+  private synchronized void incrementBytesUsed(long amount) {
+    bytesUsed += amount;
+    if (bytesUsed > maxBytesUsed) {
+      maxBytesUsed = bytesUsed;
+    }
+  }
+
+  @Override
+  public void close() throws IOException {
+    temporaryDirectory.close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/TemporaryFile.java

@@ -0,0 +1,64 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.google.common.base.Preconditions;
+import java.io.Closeable;
+import java.io.File;
+import java.io.IOException;
+
+/**
+ * A temporary file or directory. Wraps a file or directory and deletes it (recursively, if it is a
+ * directory) when closed.
+ */
+public class TemporaryFile implements Closeable {
+
+  /** Has the file or directory represented by {@link #file} been deleted? */
+  private boolean deleted;
+
+  /**
+   * The file or directory that will be deleted on close. May no longer exist if {@link #deleted} is
+   * {@code true}.
+   */
+  private final File file;
+
+  /**
+   * Creates a new wrapper around the given file. The file or directory {@code file} will be deleted
+   * (recursively, if it is a directory) on close.
+   */
+  public TemporaryFile(File file) {
+    deleted = false;
+    this.file = file;
+  }
+
+  /** Obtains the file or directory this temporary file refers to. */
+  public File getFile() {
+    Preconditions.checkState(!deleted, "File already deleted");
+    return file;
+  }
+
+  @Override
+  public void close() throws IOException {
+    if (deleted) {
+      return;
+    }
+
+    deleted = true;
+
+    deleteFile(file);
+  }
+
+  /** Deletes a file or directory if it exists. */
+  private void deleteFile(File file) throws IOException {
+    if (file.isDirectory()) {
+      File[] contents = file.listFiles();
+      if (contents != null) {
+        for (File subFile : contents) {
+          deleteFile(subFile);
+        }
+      }
+    }
+
+    if (file.exists() && !file.delete()) {
+      throw new IOException("Failed to delete '" + file.getAbsolutePath() + "'");
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/bytestorage/TemporaryFileCloseableByteSource.java

@@ -0,0 +1,37 @@
+package com.android.tools.build.apkzlib.bytestorage;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableDelegateByteSource;
+import com.google.common.io.Files;
+import java.io.File;
+import java.io.IOException;
+
+/**
+ * Closeable byte source that uses a temporary file to store its contents. The file is deleted when
+ * the byte source is closed.
+ */
+class TemporaryFileCloseableByteSource extends CloseableDelegateByteSource {
+
+  /** Temporary file backing the byte source. */
+  private final TemporaryFile temporaryFile;
+
+  /** Callback to notify when the byte source is closed. */
+  private final Runnable closeCallback;
+
+  /**
+   * Creates a new byte source based on the given file. The provided callback is executed when the
+   * source is deleted. There is no guarantee about which thread invokes the callback (it is the
+   * thread that closes the source).
+   */
+  TemporaryFileCloseableByteSource(File file, Runnable closeCallback) {
+    super(Files.asByteSource(file), file.length());
+    temporaryFile = new TemporaryFile(file);
+    this.closeCallback = closeCallback;
+  }
+
+  @Override
+  protected synchronized void innerClose() throws IOException {
+    super.innerClose();
+    temporaryFile.close();
+    closeCallback.run();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/DigestAlgorithm.java

@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.sign;
+
+
+/** Message digest algorithms. */
+public enum DigestAlgorithm {
+  /**
+   * SHA-1 digest.
+   *
+   * <p>Android 2.3 (API Level 9) to 4.2 (API Level 17) (inclusive) do not support SHA-2 JAR
+   * signatures.
+   *
+   * <p>Moreover, platforms prior to API Level 18, without the additional Digest-Algorithms
+   * attribute, only support SHA or SHA1 algorithm names in .SF and MANIFEST.MF attributes.
+   */
+  SHA1("SHA1", "SHA-1"),
+
+  /** SHA-256 digest. */
+  SHA256("SHA-256", "SHA-256");
+
+  /**
+   * API level which supports {@link #SHA256} with {@link SignatureAlgorithm#RSA} and {@link
+   * SignatureAlgorithm#ECDSA}.
+   */
+  public static final int API_SHA_256_RSA_AND_ECDSA = 18;
+
+  /**
+   * API level which supports {@link #SHA256} for all {@link SignatureAlgorithm}s.
+   *
+   * <p>Before that, SHA256 can only be used with RSA and ECDSA.
+   */
+  public static final int API_SHA_256_ALL_ALGORITHMS = 21;
+
+  /** Name of algorithm for message digest. */
+  public final String messageDigestName;
+
+  /** Name of attribute in signature file with the manifest digest. */
+  public final String manifestAttributeName;
+
+  /** Name of attribute in entry (both manifest and signature file) with the entry's digest. */
+  public final String entryAttributeName;
+
+  /**
+   * Creates a digest algorithm.
+   *
+   * @param attributeName attribute name in the signature file
+   * @param messageDigestName name of algorithm for message digest
+   */
+  DigestAlgorithm(String attributeName, String messageDigestName) {
+    this.messageDigestName = messageDigestName;
+    this.entryAttributeName = attributeName + "-Digest";
+    this.manifestAttributeName = attributeName + "-Digest-Manifest";
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/ManifestGenerationExtension.java

@@ -0,0 +1,222 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.sign;
+
+import com.android.tools.build.apkzlib.utils.CachedSupplier;
+import com.android.tools.build.apkzlib.utils.IOExceptionRunnable;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.android.tools.build.apkzlib.zfile.ManifestAttributes;
+import com.android.tools.build.apkzlib.zip.StoredEntry;
+import com.android.tools.build.apkzlib.zip.ZFile;
+import com.android.tools.build.apkzlib.zip.ZFileExtension;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.util.jar.Attributes;
+import java.util.jar.Manifest;
+import javax.annotation.Nullable;
+
+/**
+ * Extension to {@link ZFile} that will generate a manifest. The extension will register
+ * automatically with the {@link ZFile}.
+ *
+ * <p>Creating this extension will ensure a manifest for the zip exists. This extension will
+ * generate a manifest if one does not exist and will update an existing manifest, if one does
+ * exist. The extension will also provide access to the manifest so that others may update the
+ * manifest.
+ *
+ * <p>Apart from standard manifest elements, this extension does not handle any particular manifest
+ * features such as signing or adding custom attributes. It simply generates a plain manifest and
+ * provides infrastructure so that other extensions can add data in the manifest.
+ *
+ * <p>The manifest itself will only be written when the {@link ZFileExtension#beforeUpdate()}
+ * notification is received, meaning all manifest manipulation is done in-memory.
+ */
+public class ManifestGenerationExtension {
+
+  /** Name of META-INF directory. */
+  private static final String META_INF_DIR = "META-INF";
+
+  /** Name of the manifest file. */
+  static final String MANIFEST_NAME = META_INF_DIR + "/MANIFEST.MF";
+
+  /** Who should be reported as the manifest builder. */
+  private final String builtBy;
+
+  /** Who should be reported as the manifest creator. */
+  private final String createdBy;
+
+  /** The file this extension is attached to. {@code null} if not yet registered. */
+  @Nullable private ZFile zFile;
+
+  /** The zip file's manifest. */
+  private final Manifest manifest;
+
+  /**
+   * Byte representation of the manifest. There is no guarantee that two writes of the java's {@code
+   * Manifest} object will yield the same byte array (there is no guaranteed order of entries in the
+   * manifest).
+   *
+   * <p>Because we need the byte representation of the manifest to be stable if there are no changes
+   * to the manifest, we cannot rely on {@code Manifest} to generate the byte representation every
+   * time we need the byte representation.
+   *
+   * <p>This cache will ensure that we will request one byte generation from the {@code Manifest}
+   * and will cache it. All further requests of the manifest's byte representation will receive the
+   * same byte array.
+   */
+  private final CachedSupplier<byte[]> manifestBytes;
+
+  /**
+   * Has the current manifest been changed and not yet flushed? If {@link #dirty} is {@code true},
+   * then {@link #manifestBytes} should not be valid. This means that marking the manifest as dirty
+   * should also invalidate {@link #manifestBytes}. To avoid breaking the invariant, instead of
+   * setting {@link #dirty}, {@link #markDirty()} should be called.
+   */
+  private boolean dirty;
+
+  /** The extension to register with the {@link ZFile}. {@code null} if not registered. */
+  @Nullable private ZFileExtension extension;
+
+  /**
+   * Creates a new extension. This will not register the extension with the provided {@link ZFile}.
+   * Until {@link #register(ZFile)} is invoked, this extension is not used.
+   *
+   * @param builtBy who built the manifest?
+   * @param createdBy who created the manifest?
+   */
+  public ManifestGenerationExtension(String builtBy, String createdBy) {
+    this.builtBy = builtBy;
+    this.createdBy = createdBy;
+    manifest = new Manifest();
+    dirty = false;
+    manifestBytes =
+        new CachedSupplier<>(
+            () -> {
+              ByteArrayOutputStream outBytes = new ByteArrayOutputStream();
+              try {
+                manifest.write(outBytes);
+              } catch (IOException e) {
+                throw new IOExceptionWrapper(e);
+              }
+
+              return outBytes.toByteArray();
+            });
+  }
+
+  /**
+   * Marks the manifest as being dirty, <i>i.e.</i>, its data has changed since it was last read
+   * and/or written.
+   */
+  private void markDirty() {
+    dirty = true;
+    manifestBytes.reset();
+  }
+
+  /**
+   * Registers the extension with the {@link ZFile} provided in the constructor.
+   *
+   * @param zFile the zip file to add the extension to
+   * @throws IOException failed to analyze the zip
+   */
+  public void register(ZFile zFile) throws IOException {
+    Preconditions.checkState(extension == null, "register() has already been invoked.");
+    this.zFile = zFile;
+
+    rebuildManifest();
+
+    extension =
+        new ZFileExtension() {
+          @Nullable
+          @Override
+          public IOExceptionRunnable beforeUpdate() {
+            return ManifestGenerationExtension.this::updateManifest;
+          }
+        };
+
+    this.zFile.addZFileExtension(extension);
+  }
+
+  /** Rebuilds the zip file's manifest, if it needs changes. */
+  private void rebuildManifest() throws IOException {
+    Verify.verifyNotNull(zFile, "zFile == null");
+
+    StoredEntry manifestEntry = zFile.get(MANIFEST_NAME);
+
+    if (manifestEntry != null) {
+      /*
+       * Read the manifest entry in the zip file. Make sure we store these byte sequence
+       * because writing the manifest may not generate the same byte sequence, which may
+       * trigger an unnecessary re-sign of the jar.
+       */
+      manifest.clear();
+      byte[] manifestBytes = manifestEntry.read();
+      manifest.read(new ByteArrayInputStream(manifestBytes));
+      this.manifestBytes.precomputed(manifestBytes);
+    }
+
+    Attributes mainAttributes = manifest.getMainAttributes();
+    String currentVersion = mainAttributes.getValue(ManifestAttributes.MANIFEST_VERSION);
+    if (currentVersion == null) {
+      setMainAttribute(
+          ManifestAttributes.MANIFEST_VERSION, ManifestAttributes.CURRENT_MANIFEST_VERSION);
+    } else {
+      if (!currentVersion.equals(ManifestAttributes.CURRENT_MANIFEST_VERSION)) {
+        throw new IOException("Unsupported manifest version: " + currentVersion + ".");
+      }
+    }
+
+    /*
+     * We "blindly" override all other main attributes.
+     */
+    setMainAttribute(ManifestAttributes.BUILT_BY, builtBy);
+    setMainAttribute(ManifestAttributes.CREATED_BY, createdBy);
+  }
+
+  /**
+   * Sets the value of a main attribute.
+   *
+   * @param attribute the attribute
+   * @param value the value
+   */
+  private void setMainAttribute(String attribute, String value) {
+    Attributes mainAttributes = manifest.getMainAttributes();
+    String current = mainAttributes.getValue(attribute);
+    if (!value.equals(current)) {
+      mainAttributes.putValue(attribute, value);
+      markDirty();
+    }
+  }
+
+  /**
+   * Updates the manifest in the zip file, if it has been changed.
+   *
+   * @throws IOException failed to update the manifest
+   */
+  private void updateManifest() throws IOException {
+    Verify.verifyNotNull(zFile, "zFile == null");
+
+    if (!dirty) {
+      return;
+    }
+
+    zFile.add(MANIFEST_NAME, new ByteArrayInputStream(manifestBytes.get()));
+    dirty = false;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/SignatureAlgorithm.java

@@ -0,0 +1,94 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.sign;
+
+import java.security.NoSuchAlgorithmException;
+
+/** Signature algorithm. */
+public enum SignatureAlgorithm {
+  /** RSA algorithm. */
+  RSA("RSA", 1, "withRSA"),
+
+  /** ECDSA algorithm. */
+  ECDSA("EC", 18, "withECDSA"),
+
+  /** DSA algorithm. */
+  DSA("DSA", 1, "withDSA");
+
+  /** Name of the private key as reported by {@code PrivateKey}. */
+  public final String keyAlgorithm;
+
+  /** Minimum SDK version that allows this signature. */
+  public final int minSdkVersion;
+
+  /** Suffix appended to digest algorithm to obtain signature algorithm. */
+  public final String signatureAlgorithmSuffix;
+
+  /**
+   * Creates a new signature algorithm.
+   *
+   * @param keyAlgorithm the name as reported by {@code PrivateKey}
+   * @param minSdkVersion minimum SDK version that allows this signature
+   * @param signatureAlgorithmSuffix suffix for signature name with used with a digest
+   */
+  SignatureAlgorithm(String keyAlgorithm, int minSdkVersion, String signatureAlgorithmSuffix) {
+    this.keyAlgorithm = keyAlgorithm;
+    this.minSdkVersion = minSdkVersion;
+    this.signatureAlgorithmSuffix = signatureAlgorithmSuffix;
+  }
+
+  /**
+   * Obtains the signature algorithm that corresponds to a private key name applicable to a SDK
+   * version.
+   *
+   * @param keyAlgorithm the named referred in the {@code PrivateKey}
+   * @param minSdkVersion minimum SDK version to run
+   * @return the algorithm that has {@link #keyAlgorithm} equal to {@code keyAlgorithm}
+   * @throws NoSuchAlgorithmException if no algorithm was found for the given private key; an
+   *     algorithm was found but is not applicable to the given SDK version
+   */
+  public static SignatureAlgorithm fromKeyAlgorithm(String keyAlgorithm, int minSdkVersion)
+      throws NoSuchAlgorithmException {
+    for (SignatureAlgorithm alg : values()) {
+      if (alg.keyAlgorithm.equalsIgnoreCase(keyAlgorithm)) {
+        if (alg.minSdkVersion > minSdkVersion) {
+          throw new NoSuchAlgorithmException(
+              "Signatures with "
+                  + keyAlgorithm
+                  + " keys are not supported on minSdkVersion "
+                  + minSdkVersion
+                  + ". They are supported only for minSdkVersion >= "
+                  + alg.minSdkVersion);
+        }
+
+        return alg;
+      }
+    }
+
+    throw new NoSuchAlgorithmException("Signing with " + keyAlgorithm + " keys is not supported");
+  }
+
+  /**
+   * Obtains the name of the signature algorithm when used with a digest algorithm.
+   *
+   * @param digestAlgorithm the digest algorithm to use
+   * @return the name of the signature algorithm
+   */
+  public String signatureAlgorithmName(DigestAlgorithm digestAlgorithm) {
+    return digestAlgorithm.messageDigestName.replace("-", "") + signatureAlgorithmSuffix;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/SigningExtension.java

@@ -0,0 +1,437 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package com.android.tools.build.apkzlib.sign;
+
+import com.android.apksig.ApkSignerEngine;
+import com.android.apksig.ApkVerifier;
+import com.android.apksig.DefaultApkSignerEngine;
+import com.android.apksig.apk.ApkFormatException;
+import com.android.apksig.internal.apk.ApkSigningBlockUtils;
+import com.android.apksig.util.DataSink;
+import com.android.apksig.util.DataSource;
+import com.android.apksig.util.DataSources;
+import com.android.tools.build.apkzlib.utils.IOExceptionRunnable;
+import com.android.tools.build.apkzlib.utils.SigningBlockUtils;
+import com.android.tools.build.apkzlib.zip.StoredEntry;
+import com.android.tools.build.apkzlib.zip.ZFile;
+import com.android.tools.build.apkzlib.zip.ZFileExtension;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
+import com.google.common.collect.ImmutableList;
+import com.google.common.collect.ImmutableSet;
+import com.google.common.collect.Iterables;
+import com.google.common.primitives.Bytes;
+import java.io.BufferedInputStream;
+import java.io.ByteArrayInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.ByteBuffer;
+import java.security.InvalidKeyException;
+import java.security.NoSuchAlgorithmException;
+import java.security.SignatureException;
+import java.security.cert.CertificateEncodingException;
+import java.security.cert.X509Certificate;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+import javax.annotation.Nullable;
+
+/**
+ * {@link ZFile} extension which signs the APK.
+ *
+ * <p>This extension is capable of signing the APK using JAR signing (aka v1 scheme) and APK
+ * Signature Scheme v2 (aka v2 scheme). Which schemes are actually used is specified by parameters
+ * to this extension's constructor.
+ */
+public class SigningExtension {
+  private static final int MAX_READ_CHUNK_SIZE = 65536;
+
+  // IMPLEMENTATION NOTE: Most of the heavy lifting is performed by the ApkSignerEngine primitive
+  // from apksig library. This class is an adapter between ZFile extension and ApkSignerEngine.
+  // This class takes care of invoking the right methods on ApkSignerEngine in response to ZFile
+  // extension events/callbacks.
+  //
+  // The main issue leading to additional complexity in this class is that the current build
+  // pipeline does not reuse ApkSignerEngine instances (or ZFile extension instances for that
+  // matter) for incremental builds. Thus:
+  // * ZFile extension receives no events for JAR entries already in the APK whereas
+  //   ApkSignerEngine needs to know about all JAR entries to be covered by signature. Thus, this
+  //   class, during "beforeUpdate" ZFile event, notifies ApkSignerEngine about JAR entries
+  //   already in the APK which ApkSignerEngine hasn't yet been told about -- these are the JAR
+  //   entries which the incremental build session did not touch.
+  // * The build pipeline expects the APK not to change if no JAR entry was added to it or removed
+  //   from it whereas ApkSignerEngine produces no output only if it has already produced a signed
+  //   APK and no changes have since been made to it. This class addresses this issue by checking
+  //   in its "register" method whether the APK is correctly signed and, only if that's the case,
+  //   doesn't modify the APK unless a JAR entry is added to it or removed from it after
+  //   "register".
+
+  /** APK signer which performs most of the heavy lifting. */
+  private final ApkSignerEngine signer;
+
+  /** Names of APK entries which have been processed by {@link #signer}. */
+  private final Set<String> signerProcessedOutputEntryNames = new HashSet<>();
+
+  /** Signing block Id for SDK dependency block. */
+  static final int DEPENDENCY_INFO_BLOCK_ID = 0x504b4453;
+
+  /** SDK dependencies of the APK */
+  @Nullable private byte[] sdkDependencyData;
+
+  /**
+   * Cached contents of the most recently output APK Signing Block or {@code null} if the block
+   * hasn't yet been output.
+   */
+  @Nullable private byte[] cachedApkSigningBlock;
+
+  /**
+   * {@code true} if signatures may need to be output, {@code false} if there's no need to output
+   * signatures. This is used in an optimization where we don't modify the APK if it's already
+   * signed and if no JAR entries have been added to or removed from the file.
+   */
+  private boolean dirty;
+
+  /** The extension registered with the {@link ZFile}. {@code null} if not registered. */
+  @Nullable private ZFileExtension extension;
+
+  /** The file this extension is attached to. {@code null} if not yet registered. */
+  @Nullable private ZFile zFile;
+
+  /** A buffer used to read data from entries to feed to digests */
+  private final Supplier<byte[]> digestBuffer =
+      Suppliers.memoize(() -> new byte[MAX_READ_CHUNK_SIZE]);
+
+  /** An object that has all necessary information to sign the zip file and verify its signature */
+  private final SigningOptions options;
+
+  public SigningExtension(SigningOptions opts) throws InvalidKeyException {
+    DefaultApkSignerEngine.SignerConfig signerConfig =
+        new DefaultApkSignerEngine.SignerConfig.Builder(
+                "CERT", opts.getKey(), opts.getCertificates())
+            .build();
+    signer =
+        new DefaultApkSignerEngine.Builder(ImmutableList.of(signerConfig), opts.getMinSdkVersion())
+            .setOtherSignersSignaturesPreserved(false)
+            .setV1SigningEnabled(opts.isV1SigningEnabled())
+            .setV2SigningEnabled(opts.isV2SigningEnabled())
+            .setV3SigningEnabled(false)
+            .setCreatedBy("1.0 (Android)")
+            .build();
+    if (opts.getSdkDependencyData() != null) {
+      sdkDependencyData = opts.getSdkDependencyData();
+    }
+    if (opts.getExecutor() != null) {
+      signer.setExecutor(opts.getExecutor());
+    }
+    this.options = opts;
+  }
+
+  public void register(ZFile zFile) throws NoSuchAlgorithmException, IOException {
+    Preconditions.checkState(extension == null, "register() already invoked");
+    this.zFile = zFile;
+    switch (options.getValidation()) {
+      case ALWAYS_VALIDATE:
+        dirty = !isCurrentSignatureAsRequested();
+        break;
+      case ASSUME_VALID:
+        if (options.isV1SigningEnabled()) {
+          Set<String> entryNames =
+              ImmutableSet.copyOf(
+                  Iterables.transform(
+                      zFile.entries(), e -> e.getCentralDirectoryHeader().getName()));
+          StoredEntry manifestEntry = zFile.get(ManifestGenerationExtension.MANIFEST_NAME);
+
+          Preconditions.checkNotNull(
+              manifestEntry,
+              "No manifest found in apk for incremental build with enabled v1 signature");
+          signerProcessedOutputEntryNames.addAll(
+              this.signer.initWith(manifestEntry.read(), entryNames));
+        }
+
+        dirty = false;
+        break;
+      case ASSUME_INVALID:
+        dirty = true;
+        break;
+    }
+    extension =
+        new ZFileExtension() {
+          @Override
+          public IOExceptionRunnable added(StoredEntry entry, @Nullable StoredEntry replaced) {
+            return () -> onZipEntryOutput(entry);
+          }
+
+          @Override
+          public IOExceptionRunnable removed(StoredEntry entry) {
+            String entryName = entry.getCentralDirectoryHeader().getName();
+            return () -> onZipEntryRemovedFromOutput(entryName);
+          }
+
+          @Override
+          public IOExceptionRunnable beforeUpdate() throws IOException {
+            return () -> onOutputZipReadyForUpdate();
+          }
+
+          @Override
+          public void entriesWritten() throws IOException {
+            onOutputZipEntriesWritten();
+          }
+
+          @Override
+          public void closed() {
+            onOutputClosed();
+          }
+        };
+    this.zFile.addZFileExtension(extension);
+  }
+
+  /**
+   * Returns {@code true} if the APK's signatures are as requested by parameters to this signing
+   * extension.
+   */
+  private boolean isCurrentSignatureAsRequested() throws IOException, NoSuchAlgorithmException {
+    ApkVerifier.Result result;
+    try {
+      result =
+          new ApkVerifier.Builder(zFile.asDataSource())
+              .setMinCheckedPlatformVersion(options.getMinSdkVersion())
+              .build()
+              .verify();
+    } catch (ApkFormatException e) {
+      // Malformed APK
+      return false;
+    }
+
+    if (!result.isVerified()) {
+      // Signature(s) did not verify
+      return false;
+    }
+
+    if ((result.isVerifiedUsingV1Scheme() != options.isV1SigningEnabled())
+        || (result.isVerifiedUsingV2Scheme() != options.isV2SigningEnabled())) {
+      // APK isn't signed with exactly the schemes we want it to be signed
+      return false;
+    }
+
+    List<X509Certificate> verifiedSignerCerts = result.getSignerCertificates();
+    if (verifiedSignerCerts.size() != 1) {
+      // APK is not signed by exactly one signer
+      return false;
+    }
+
+    byte[] expectedEncodedCert;
+    byte[] actualEncodedCert;
+    try {
+      expectedEncodedCert = options.getCertificates().get(0).getEncoded();
+      actualEncodedCert = verifiedSignerCerts.get(0).getEncoded();
+    } catch (CertificateEncodingException e) {
+      // Failed to encode signing certificates
+      return false;
+    }
+
+    if (!Arrays.equals(expectedEncodedCert, actualEncodedCert)) {
+      // APK is signed by a wrong signer
+      return false;
+    }
+
+    // APK is signed the way we want it to be signed
+    return true;
+  }
+
+  private void onZipEntryOutput(StoredEntry entry) throws IOException {
+    setDirty();
+    String entryName = entry.getCentralDirectoryHeader().getName();
+    // This event may arrive after the entry has already been deleted. In that case, we don't
+    // report the addition of the entry to ApkSignerEngine.
+    if (entry.isDeleted()) {
+      return;
+    }
+    ApkSignerEngine.InspectJarEntryRequest inspectEntryRequest = signer.outputJarEntry(entryName);
+    signerProcessedOutputEntryNames.add(entryName);
+    if (inspectEntryRequest != null) {
+      try (InputStream inputStream = new BufferedInputStream(entry.open())) {
+        copyStreamToDataSink(inputStream, inspectEntryRequest.getDataSink());
+      }
+      inspectEntryRequest.done();
+    }
+  }
+
+  private void copyStreamToDataSink(InputStream inputStream, DataSink dataSink) throws IOException {
+    int bytesRead;
+    byte[] buffer = digestBuffer.get();
+    while ((bytesRead = inputStream.read(buffer)) > 0) {
+      dataSink.consume(buffer, 0, bytesRead);
+    }
+  }
+
+  private void onZipEntryRemovedFromOutput(String entryName) {
+    setDirty();
+    signer.outputJarEntryRemoved(entryName);
+    signerProcessedOutputEntryNames.remove(entryName);
+  }
+
+  private void onOutputZipReadyForUpdate() throws IOException {
+    if (!dirty) {
+      return;
+    }
+
+    // Notify signer engine about ZIP entries that have appeared in the output without the
+    // engine knowing. Also identify ZIP entries which disappeared from the output without the
+    // engine knowing.
+    Set<String> unprocessedRemovedEntryNames = new HashSet<>(signerProcessedOutputEntryNames);
+    for (StoredEntry entry : zFile.entries()) {
+      String entryName = entry.getCentralDirectoryHeader().getName();
+      unprocessedRemovedEntryNames.remove(entryName);
+      if (!signerProcessedOutputEntryNames.contains(entryName)) {
+        // Signer engine is not yet aware that this entry is in the output
+        onZipEntryOutput(entry);
+      }
+    }
+
+    // Notify signer engine about entries which disappeared from the output without the engine
+    // knowing
+    for (String entryName : unprocessedRemovedEntryNames) {
+      onZipEntryRemovedFromOutput(entryName);
+    }
+
+    // Check whether we need to output additional JAR entries which comprise the v1 signature
+    ApkSignerEngine.OutputJarSignatureRequest addV1SignatureRequest;
+    try {
+      addV1SignatureRequest = signer.outputJarEntries();
+    } catch (Exception e) {
+      throw new IOException("Failed to generate v1 signature", e);
+    }
+    if (addV1SignatureRequest == null) {
+      return;
+    }
+
+    // We need to output additional JAR entries which comprise the v1 signature
+    List<ApkSignerEngine.OutputJarSignatureRequest.JarEntry> v1SignatureEntries =
+        new ArrayList<>(addV1SignatureRequest.getAdditionalJarEntries());
+
+    // Reorder the JAR entries comprising the v1 signature so that MANIFEST.MF is the first
+    // entry. This ensures that it cleanly overwrites the existing MANIFEST.MF output by
+    // ManifestGenerationExtension.
+    for (int i = 0; i < v1SignatureEntries.size(); i++) {
+      ApkSignerEngine.OutputJarSignatureRequest.JarEntry entry = v1SignatureEntries.get(i);
+      String name = entry.getName();
+      if (!ManifestGenerationExtension.MANIFEST_NAME.equals(name)) {
+        continue;
+      }
+      if (i != 0) {
+        v1SignatureEntries.remove(i);
+        v1SignatureEntries.add(0, entry);
+      }
+      break;
+    }
+
+    // Output the JAR entries comprising the v1 signature
+    for (ApkSignerEngine.OutputJarSignatureRequest.JarEntry entry : v1SignatureEntries) {
+      String name = entry.getName();
+      byte[] data = entry.getData();
+      zFile.add(name, new ByteArrayInputStream(data));
+    }
+
+    addV1SignatureRequest.done();
+  }
+
+  private void onOutputZipEntriesWritten() throws IOException {
+    if (!dirty) {
+      return;
+    }
+
+    // Check whether we should output an APK Signing Block which contains v2 signatures
+    byte[] apkSigningBlock;
+    byte[] centralDirBytes = zFile.getCentralDirectoryBytes();
+    byte[] eocdBytes = zFile.getEocdBytes();
+    ApkSignerEngine.OutputApkSigningBlockRequest2 addV2SignatureRequest;
+    // This event may arrive a second time -- after we write out the APK Signing Block. Thus, we
+    // cache the block to speed things up. The cached block is invalidated by any changes to the
+    // file (as reported to this extension).
+    if (cachedApkSigningBlock != null) {
+      apkSigningBlock = cachedApkSigningBlock;
+      addV2SignatureRequest = null;
+    } else {
+      DataSource centralDir = DataSources.asDataSource(ByteBuffer.wrap(centralDirBytes));
+      DataSource eocd = DataSources.asDataSource(ByteBuffer.wrap(eocdBytes));
+      long zipEntriesSizeBytes =
+          zFile.getCentralDirectoryOffset() - zFile.getExtraDirectoryOffset();
+      DataSource zipEntries = zFile.asDataSource(0, zipEntriesSizeBytes);
+      try {
+        addV2SignatureRequest = signer.outputZipSections2(zipEntries, centralDir, eocd);
+      } catch (NoSuchAlgorithmException
+          | InvalidKeyException
+          | SignatureException
+          | ApkFormatException
+          | IOException e) {
+        throw new IOException("Failed to generate v2 signature", e);
+      }
+
+      if (addV2SignatureRequest != null) {
+        apkSigningBlock = addV2SignatureRequest.getApkSigningBlock();
+        if (sdkDependencyData != null) {
+          apkSigningBlock =
+              SigningBlockUtils.addToSigningBlock(
+                  apkSigningBlock, sdkDependencyData, DEPENDENCY_INFO_BLOCK_ID);
+        }
+        apkSigningBlock =
+            Bytes.concat(
+                new byte[addV2SignatureRequest.getPaddingSizeBeforeApkSigningBlock()],
+                apkSigningBlock);
+      } else {
+        apkSigningBlock = new byte[0];
+        if (sdkDependencyData != null) {
+          apkSigningBlock =
+              SigningBlockUtils.addToSigningBlock(
+                  apkSigningBlock, sdkDependencyData, DEPENDENCY_INFO_BLOCK_ID);
+          int paddingSize =
+              ApkSigningBlockUtils.generateApkSigningBlockPadding(
+                      zipEntries, /* apkSigningBlockPaddingSupported */ true)
+                  .getSecond();
+          apkSigningBlock = Bytes.concat(new byte[paddingSize], apkSigningBlock);
+        }
+      }
+      cachedApkSigningBlock = apkSigningBlock;
+    }
+
+    // Insert the APK Signing Block into the output right before the ZIP Central Directory and
+    // accordingly update the start offset of ZIP Central Directory in ZIP End of Central
+    // Directory.
+    zFile.directWrite(
+        zFile.getCentralDirectoryOffset() - zFile.getExtraDirectoryOffset(), apkSigningBlock);
+    zFile.setExtraDirectoryOffset(apkSigningBlock.length);
+
+    if (addV2SignatureRequest != null) {
+      addV2SignatureRequest.done();
+    }
+  }
+
+  private void onOutputClosed() {
+    if (!dirty) {
+      return;
+    }
+    signer.outputDone();
+    dirty = false;
+  }
+
+  private void setDirty() {
+    dirty = true;
+    cachedApkSigningBlock = null;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/SigningOptions.java

@@ -0,0 +1,102 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.sign;
+
+import com.android.apksig.util.RunnablesExecutor;
+import com.google.auto.value.AutoValue;
+import com.google.common.base.Preconditions;
+import com.google.common.collect.ImmutableList;
+import java.security.PrivateKey;
+import java.security.cert.X509Certificate;
+import javax.annotation.Nonnull;
+import javax.annotation.Nullable;
+
+/** A class that contains data to initialize SigningExtension. */
+@AutoValue
+public abstract class SigningOptions {
+
+    /** An implementation of builder pattern to create a {@link SigningOptions} object. */
+    @AutoValue.Builder
+    public abstract static class Builder {
+        public abstract Builder setKey(@Nonnull PrivateKey key);
+        public abstract Builder setCertificates(@Nonnull ImmutableList<X509Certificate> certs);
+        public abstract Builder setCertificates(X509Certificate... certs);
+        public abstract Builder setV1SigningEnabled(boolean enabled);
+        public abstract Builder setV2SigningEnabled(boolean enabled);
+        public abstract Builder setMinSdkVersion(int version);
+        public abstract Builder setValidation(@Nonnull Validation validation);
+        public abstract Builder setExecutor(@Nullable RunnablesExecutor executor);
+        public abstract Builder setSdkDependencyData(@Nullable byte[] sdkDependencyData);
+
+        abstract SigningOptions autoBuild();
+
+        public SigningOptions build() {
+            SigningOptions options = autoBuild();
+            Preconditions.checkArgument(options.getMinSdkVersion() >= 0, "minSdkVersion < 0");
+            Preconditions.checkArgument(
+                    !options.getCertificates().isEmpty(),
+                    "There should be at least one certificate in SigningOptions");
+            return options;
+        }
+    }
+
+    public static Builder builder() {
+        return new AutoValue_SigningOptions.Builder()
+                .setV1SigningEnabled(false)
+                .setV2SigningEnabled(false)
+                .setValidation(Validation.ALWAYS_VALIDATE);
+    }
+
+    /** {@link PrivateKey} used to sign the archive. */
+    public abstract PrivateKey getKey();
+
+    /**
+     * A list of the {@link X509Certificate}s to embed in the signed APKs. The first
+     * element of the list must be the certificate associated with the private key.
+     */
+    public abstract ImmutableList<X509Certificate> getCertificates();
+
+    /** Shows whether signing with JAR Signature Scheme (aka v1 signing) is enabled. */
+    public abstract boolean isV1SigningEnabled();
+
+    /** Shows whether signing with APK Signature Scheme v2 (aka v2 signing) is enabled. */
+    public abstract boolean isV2SigningEnabled();
+
+    /** Minimum SDK version supported. */
+    public abstract int getMinSdkVersion();
+
+    /** Strategy of package signature validation */
+    public abstract Validation getValidation();
+
+    @Nullable
+    public abstract RunnablesExecutor getExecutor();
+
+  /** SDK dependencies of the APK */
+  @Nullable
+  public abstract byte[] getSdkDependencyData();
+
+    public enum Validation {
+        /** Always perform signature validation */
+        ALWAYS_VALIDATE,
+        /**
+         * Assume the signature is valid without validation i.e. don't resign if no files changed
+         */
+        ASSUME_VALID,
+        /** Assume the signature is invalid without validation i.e. unconditionally resign */
+        ASSUME_INVALID,
+    }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/sign/package-info.java

@@ -0,0 +1,157 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * The {@code sign} package provides extensions for the {@code zip} package that allow:
+ *
+ * <ul>
+ *   <li>Adding a {@code MANIFEST.MF} file to a zip making a jar.
+ *   <li>Signing a jar.
+ *   <li>Fully signing a jar using v2 apk signature.
+ * </ul>
+ *
+ * <p>Because the {@code zip} package is completely independent of the {@code sign} package, the
+ * actual coordination between the two is complex. The {@code sign} package works by registering
+ * extensions with the {@code zip} package. These extensions are notified in changes made in the zip
+ * and will change the zip file itself.
+ *
+ * <p>The {@link com.android.apkzlib.sign.ManifestGenerationExtension} extension will ensure the zip
+ * has a manifest file and is, therefore, a valid jar. The {@link
+ * com.android.apkzlib.sign.SigningExtension} extension will ensure the jar is signed.
+ *
+ * <p>The extension mechanism used is the one provided in the {@code zip} package (see {@link
+ * com.android.apkzlib.zip.ZFile} and {@link com.android.apkzlib.zip.ZFileExtension}. Building the
+ * zip and then operating the extensions is not done sequentially, as we don't want to build a zip
+ * and then sign it. We want to build a zip that is automatically signed. Extension are basically
+ * observers that register on the zip and are notified when things happen in the zip. They will then
+ * modify the zip accordingly.
+ *
+ * <p>The zip file notifies extensions in 4 critical moments: when a file is added or removed from
+ * the zip, when the zip is about to be flushed to disk and when the zip's entries have been flushed
+ * but the central directory not. At these moments, the extensions can act to update the zip in any
+ * way they need.
+ *
+ * <p>To see how this works, consider the manifest generation extension: when the extension is
+ * created, it checks the zip file to see if there is a manifest. If a manifest exists and does not
+ * need updating, it does not change anything, otherwise it generates a new manifest for the zip
+ * file. At this point, the extension could write the manifest to the zip, but we opted not to. It
+ * would be irrelevant anyway as the zip will only be written when flushed.
+ *
+ * <p>Now, when the {@code ZFile} notifies the extension that it is about to start writing the zip
+ * file, the manifest extension, if it has noted that the manifest needs to be rewritten, will --
+ * before the {@code ZFile} actually writes anything -- modify the zip and add or replace the
+ * existing manifest file. So, process-wise, the zip is written only once with the correct manifest.
+ * The flow is as follows (if only the manifest generation extension was added to the {@code
+ * ZFile}):
+ *
+ * <ol>
+ *   <li>{@code ZFile.update()} is called.
+ *   <li>{@code ZFile} calls {@code beforeUpdate()} for all {@code ZFileExtensions} registered, in
+ *       this case, only the instance of the anonymous inner class generated in the {@code
+ *       ManifestGenerationExtension} constructor is invoked.
+ *   <li>{@code ManifestGenerationExtension.updateManifest()} is called.
+ *   <li>If the manifest does not need to be updated, {@code updateManifest()} returns immediately.
+ *   <li>If the manifest needs updating, {@code ZFile.add()} is invoked to add or replace the
+ *       manifest.
+ *   <li>{@code ManifestGenerationExtension.updateManifest()} returns.
+ *   <li>{@code ZFile.update()} continues and writes the zip file, containing the manifest.
+ *   <li>The zip is finally written with an updated manifest.
+ * </ol>
+ *
+ * <p>To generate a signed apk, we need to add a second extension, the {@code SigningExtension}.
+ * This extension will also register listeners with the {@code ZFile}.
+ *
+ * <p>In this case the flow would be (starting a bit earlier for clarity and assuming a package task
+ * in the build process):
+ *
+ * <ol>
+ *   <li>Package task creates a {@code ZFile} on the target apk (or non-existing file, if there is
+ *       no target apk in the output directory).
+ *   <li>Package task configures the {@code ZFile} with alignment rules.
+ *   <li>Package task creates a {@code ManifestGenerationExtension}.
+ *   <li>Package task registers the {@code ManifestGenerationExtension} with the {@code ZFile}.
+ *   <li>The {@code ManifestGenerationExtension} looks at the {@code ZFile} to see if there is valid
+ *       manifest. No changes are done to the {@code ZFile}.
+ *   <li>Package task creates a {@code SigningExtension}.
+ *   <li>Package task registers the {@code SigningExtension} with the {@code ZFile}.
+ *   <li>The {@code SigningExtension} registers a {@code ZFileExtension} with the {@code ZFile} and
+ *       look at the {@code ZFile} to see if there is a valid signature file.
+ *   <li>If there are changes to the digital signature file needed, these are marked internally in
+ *       the extension. If there are changes needed to the digests, the manifest is updated (by
+ *       calling {@code ManifestGenerationExtension}.<br>
+ *       <em>(note that this point, the apk file, if any existed, has not been touched, the manifest
+ *       is only updated in memory and the digests of all files in the apk, if any, have been
+ *       computed and stored in memory only; the digital signature of the {@code SF} file has not
+ *       been computed.) </em>
+ *   <li>The Package task now adds all files to the {@code ZFile}.
+ *   <li>For each file that is added (*), {@code ZFile} calls the added {@code ZFileExtension.added}
+ *       method of all registered extensions.
+ *   <li>The {@code ManifestGenerationExtension} ignores added invocations.
+ *   <li>The {@code SigningExtension} computes the digest for the added file and stores them in the
+ *       manifest.<br>
+ *       <em>(when all files are added to the apk, all digests are computed and the manifest is
+ *       updated but only in memory; the apk file has not been touched; also note that {@code ZFile}
+ *       has not actually written anything to disk at this point, all files added are kept in
+ *       memory).</em>
+ *   <li>Package task calls {@code ZFile.update()} to update the apk.
+ *   <li>{@code ZFile} calls {@code before()} for all {@code ZFileExtensions} registered. This is
+ *       done before anything is written. In this case both the {@code ManifestGenerationExtension}
+ *       and {@code SigningExtension} are invoked.
+ *   <li>The {@code ManifestGenerationExtension} will update the {@code ZFile} with the new
+ *       manifest, unless nothing has changed, in which case it does nothing.
+ *   <li>The {@code SigningExtension} will add the SF file (unless nothing has changed), will
+ *       compute the digital signature of the SF file and write it to the {@code ZFile}.<br>
+ *       <em>(note that the order by which the {@code ManifestGenerationExtension} and {@code
+ *       SigningExtension} are called is non-deterministic; however, this is not a problem because
+ *       the manifest is already computed by the {@code ManifestGenerationExtension} at this time
+ *       and the {@code SigningExtension} will obtain the manifest data from the {@code
+ *       ManifestGenerationExtension} and not from the {@code ZFile}; this means that the {@code SF}
+ *       file may be added to the {@code ZFile} before the {@code MF} file, but that is
+ *       irrelevant.)</em>
+ *   <li>Once both extensions have finished doing the {@code beforeUpdate()} method, the {@code
+ *       ZFile.update()} method continues.
+ *   <li>{@code ZFile.update()} writes all changes and new entries to the zip file.
+ *   <li>{@code ZFile.update()} calls {@code ZFileExtension.entriesWritten()} for all registered
+ *       extensions. {@code SigningExtension} will kick in at this point, if v2 signature has
+ *       changed.
+ *   <li>{@code ZFile} writes the central directory and EOCD.
+ *   <li>{@code ZFile.update()} returns control to the package task.
+ *   <li>The package task finishes.
+ * </ol>
+ *
+ * <em>(*) There is a number of optimizations if we're adding files from another {@code ZFile},
+ * which is the case when we add the output of aapt to the apk. In particular, files from the aapt
+ * are ignored if they are already in the apk (same name, same CRC32) and also files copied from the
+ * aapt's output are not recompressed (the binary compressed data is directly copied to the
+ * zip).</em>
+ *
+ * <p>If there are no changes to the {@code ZFile} made by the package task and the file's manifest
+ * and v1 signatures are correct, neither the {@code ManifestGenerationExtension} nor the {@code
+ * SigningExtension} will not do anything on the {@code beforeUpdate()} and the {@code ZFile} won't
+ * even be open for writing.
+ *
+ * <p>This implementation provides perfect incremental updates.
+ *
+ * <p>Additionally, by adding/removing extensions we can configure what type of apk we want:
+ *
+ * <ul>
+ *   <li>No SigningExtension ⇒ Aligned, unsigned apk.
+ *   <li>SigningExtension ⇒ Aligned, signed apk.
+ * </ul>
+ *
+ * So, by configuring which extensions to add, the package task can decide what type of apk we want.
+ */
+package com.android.apkzlib.sign;

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/ApkZLibPair.java

@@ -0,0 +1,38 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+/** Pair implementation to use with the {@code apkzlib} library. */
+public class ApkZLibPair<T1, T2> {
+
+  /** First value. */
+  public T1 v1;
+
+  /** Second value. */
+  public T2 v2;
+
+  /**
+   * Creates a new pair.
+   *
+   * @param v1 the first value
+   * @param v2 the second value
+   */
+  public ApkZLibPair(T1 v1, T2 v2) {
+    this.v1 = v1;
+    this.v2 = v2;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/CachedFileContents.java

@@ -0,0 +1,160 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import com.google.common.base.Objects;
+import com.google.common.hash.HashCode;
+import com.google.common.hash.Hashing;
+import com.google.common.io.Files;
+import java.io.File;
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/**
+ * A cache for file contents. The cache allows closing a file and saving in memory its contents (or
+ * some related information). It can then be used to check if the contents are still valid at some
+ * later time. Typical usage flow is:
+ *
+ * <p>
+ *
+ * <pre>{@code
+ * Object fileRepresentation = // ...
+ * File toWrite = // ...
+ * // Write file contents and update in memory representation
+ * CachedFileContents<Object> contents = new CachedFileContents<Object>(toWrite);
+ * contents.closed(fileRepresentation);
+ *
+ * // Later, when data is needed:
+ * if (contents.isValid()) {
+ *     fileRepresentation = contents.getCache();
+ * } else {
+ *     // Re-read the file and recreate the file representation
+ * }
+ * }</pre>
+ *
+ * @param <T> the type of cached contents
+ */
+public class CachedFileContents<T> {
+
+  /** The file. */
+  private final File file;
+
+  /** Time when last closed (time when {@link #closed(Object)} was invoked). */
+  private long lastClosed;
+
+  /** Size of the file when last closed. */
+  private long size;
+
+  /** Hash of the file when closed. {@code null} if hashing failed for some reason. */
+  @Nullable private HashCode hash;
+
+  /** Cached data associated with the file. */
+  @Nullable private T cache;
+
+  /**
+   * Creates a new contents. When the file is written, {@link #closed(Object)} should be invoked to
+   * set the cache.
+   *
+   * @param file the file
+   */
+  public CachedFileContents(File file) {
+    this.file = file;
+  }
+
+  /**
+   * Should be called when the file's contents are set and the file closed. This will save the cache
+   * and register the file's timestamp to later detect if it has been modified.
+   *
+   * <p>This method can be called as many times as the file has been written.
+   *
+   * @param cache an optional cache to save
+   */
+  public void closed(@Nullable T cache) {
+    this.cache = cache;
+    lastClosed = file.lastModified();
+    size = file.length();
+    hash = hashFile();
+  }
+
+  /**
+   * Are the cached contents still valid? If this method determines that the file has been modified
+   * since the last time {@link #closed(Object)} was invoked.
+   *
+   * @return are the cached contents still valid? If this method returns {@code false}, the cache is
+   *     cleared
+   */
+  public boolean isValid() {
+    boolean valid = true;
+
+    if (!file.exists()) {
+      valid = false;
+    }
+
+    if (valid && file.lastModified() != lastClosed) {
+      valid = false;
+    }
+
+    if (valid && file.length() != size) {
+      valid = false;
+    }
+
+    if (valid && !Objects.equal(hash, hashFile())) {
+      valid = false;
+    }
+
+    if (!valid) {
+      cache = null;
+    }
+
+    return valid;
+  }
+
+  /**
+   * Obtains the cached data set with {@link #closed(Object)} if the file has not been modified
+   * since {@link #closed(Object)} was invoked.
+   *
+   * @return the last cached data or {@code null} if the file has been modified since {@link
+   *     #closed(Object)} has been invoked
+   */
+  @Nullable
+  public T getCache() {
+    return cache;
+  }
+
+  /**
+   * Computes the hashcode of the cached file.
+   *
+   * @return the hash code
+   */
+  @Nullable
+  private HashCode hashFile() {
+    try {
+      return Files.asByteSource(file).hash(Hashing.crc32());
+    } catch (IOException e) {
+      return null;
+    }
+  }
+
+  /**
+   * Obtains the file used for caching.
+   *
+   * @return the file; this file always exists and contains the old (cached) contents of the file
+   */
+  public File getFile() {
+    return file;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/CachedSupplier.java

@@ -0,0 +1,109 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import com.google.common.base.Supplier;
+
+/**
+ * Supplier that will cache a computed value and always supply the same value. It can be used to
+ * lazily compute data. For example:
+ *
+ * <pre>{@code
+ * CachedSupplier<Integer> value = new CachedSupplier<>(() -> {
+ *     Integer result;
+ *     // Do some expensive computation.
+ *     return result;
+ * });
+ *
+ * if (a) {
+ *     // We need the result of the expensive computation.
+ *     Integer r = value.get();
+ * }
+ *
+ * if (b) {
+ *     // We also need the result of the expensive computation.
+ *     Integer r = value.get();
+ * }
+ *
+ * // If neither a nor b are true, we avoid doing the computation at all.
+ * }</pre>
+ */
+public class CachedSupplier<T> {
+
+  /**
+   * The cached data, {@code null} if computation resulted in {@code null}. It is also {@code null}
+   * if the cached data has not yet been computed.
+   */
+  private T cached;
+
+  /** Is the current data in {@link #cached} valid? */
+  private boolean valid;
+
+  /** Actual supplier of data, if computation is needed. */
+  private final Supplier<T> supplier;
+
+  /** Creates a new supplier. */
+  public CachedSupplier(Supplier<T> supplier) {
+    valid = false;
+    this.supplier = supplier;
+  }
+
+  /**
+   * Obtains the value.
+   *
+   * @return the value, either cached (if one exists) or computed
+   */
+  public synchronized T get() {
+    if (!valid) {
+      cached = supplier.get();
+      valid = true;
+    }
+
+    return cached;
+  }
+
+  /**
+   * Resets the cache forcing a {@code get()} on the supplier next time {@link #get()} is invoked.
+   */
+  public synchronized void reset() {
+    cached = null;
+    valid = false;
+  }
+
+  /**
+   * In some cases, we may be able to precompute the cache value (or load it from somewhere we had
+   * previously stored it). This method allows the cache value to be loaded.
+   *
+   * <p>If this method is invoked, then an invocation of {@link #get()} will not trigger an
+   * invocation of the supplier provided in the constructor.
+   *
+   * @param t the new cache contents; will replace any currently cache content, if one exists
+   */
+  public synchronized void precomputed(T t) {
+    cached = t;
+    valid = true;
+  }
+
+  /**
+   * Checks if the contents of the cache are valid.
+   *
+   * @return are there valid contents in the cache?
+   */
+  public synchronized boolean isValid() {
+    return valid;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/IOExceptionConsumer.java

@@ -0,0 +1,31 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/** Consumer that can throw an {@link IOException}. */
+public interface IOExceptionConsumer<T> {
+
+  /**
+   * Performs an operation on the given input.
+   *
+   * @param input the input
+   */
+  void accept(@Nullable T input) throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/IOExceptionFunction.java

@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import com.google.common.base.Function;
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/** Function that can throw an I/O Exception */
+public interface IOExceptionFunction<F, T> {
+
+  /**
+   * Applies the function to the given input.
+   *
+   * @param input the input
+   * @return the function result
+   */
+  @Nullable
+  T apply(@Nullable F input) throws IOException;
+
+  /**
+   * Wraps a function that may throw an IO Exception throwing an {@link IOExceptionWrapper}.
+   *
+   * @param f the function
+   */
+  static <F, T> Function<F, T> asFunction(IOExceptionFunction<F, T> f) {
+    return i -> {
+      try {
+        return f.apply(i);
+      } catch (IOException e) {
+        throw new IOExceptionWrapper(e);
+      }
+    };
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/IOExceptionRunnable.java

@@ -0,0 +1,45 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import java.io.IOException;
+
+/** Runnable that can throw I/O exceptions. */
+public interface IOExceptionRunnable {
+
+  /**
+   * Runs the runnable.
+   *
+   * @throws IOException failed to run
+   */
+  void run() throws IOException;
+
+  /**
+   * Wraps a runnable that may throw an IO Exception throwing an {@code UncheckedIOException}.
+   *
+   * @param r the runnable
+   */
+  static Runnable asRunnable(IOExceptionRunnable r) {
+    return () -> {
+      try {
+        r.run();
+      } catch (IOException e) {
+        throw new IOExceptionWrapper(e);
+      }
+    };
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/IOExceptionWrapper.java

@@ -0,0 +1,40 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import java.io.IOException;
+
+/**
+ * Runtime exception used to encapsulate an IO Exception. This is used to allow throwing I/O
+ * exceptions in functional interfaces that do not allow it and catching the exception afterwards.
+ */
+public class IOExceptionWrapper extends RuntimeException {
+
+  /**
+   * Creates a new exception.
+   *
+   * @param e the I/O exception to encapsulate
+   */
+  public IOExceptionWrapper(IOException e) {
+    super(e);
+  }
+
+  @Override
+  public IOException getCause() {
+    return (IOException) super.getCause();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/SigningBlockUtils.java

@@ -0,0 +1,185 @@
+/*
+ * Copyright (C) 2019 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.utils;
+
+import static java.nio.ByteOrder.LITTLE_ENDIAN;
+
+import com.android.apksig.apk.ApkSigningBlockNotFoundException;
+import com.android.apksig.apk.ApkUtils;
+import com.android.apksig.apk.ApkUtils.ApkSigningBlock;
+import com.android.apksig.internal.apk.ApkSigningBlockUtils;
+import com.android.apksig.internal.util.Pair;
+import com.android.apksig.util.DataSource;
+import com.android.apksig.util.DataSources;
+import com.android.apksig.zip.ZipFormatException;
+import com.google.common.collect.ImmutableList;
+import com.google.common.primitives.Ints;
+import java.io.File;
+import java.io.IOException;
+import java.io.RandomAccessFile;
+import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
+import javax.annotation.Nullable;
+
+/** Generates and appends a new block to APK v2 Signature block. */
+public final class SigningBlockUtils {
+
+  private static final int MAGIC_NUM_BYTES = 16;
+  private static final int BLOCK_LENGTH_NUM_BYTES = 8;
+  static final int SIZE_OF_BLOCK_NUM_BYTES = 8;
+  static final int BLOCK_ID_NUM_BYTES = 4;
+
+  static final int ANDROID_COMMON_PAGE_ALIGNMENT_NUM_BYTES = 4096;
+  static final int VERITY_PADDING_BLOCK_ID = 0x42726577;
+
+  /**
+   * Generates a new block with the given block value and block id, and appends it to the signing
+   * block.
+   *
+   * @param signingBlock Block containing v2 signature and (optionally) padding block or null.
+   * @param blockValue byte array containing block value of the new block or null.
+   * @param blockId block id of the new block.
+   * @return APK v2 block with signatures and the new block. If {@code blockValue} is null the
+   *     {@code signingBlock} is returned without any modification. If {@code signingBlock} is null,
+   *     a new signature block is created containing the new block and, optionally, padding block.
+   */
+  public static byte[] addToSigningBlock(byte[] signingBlock, byte[] blockValue, int blockId)
+      throws IOException {
+    if (blockValue == null || blockValue.length == 0) {
+      return signingBlock;
+    }
+    if (signingBlock == null || signingBlock.length == 0) {
+      return createSigningBlock(blockValue, blockId);
+    }
+    return appendToSigningBlock(signingBlock, blockValue, blockId);
+  }
+
+  /**
+   * Adds a new block to the signature block and a padding block, if required.
+   *
+   * @param signingBlock APK v2 signing block containing : length prefix, signers (can include
+   *     padding block), length postfix and APK sig v2 block magic.
+   * @param blockValue byte array containing block value of the new block.
+   * @param blockId block id of the new block.
+   * @return APK v2 signing block containing : length prefix, signers including the new block (may
+   *     include padding block as well), length postfix and APK sig v2 block magic.
+   */
+  private static byte[] appendToSigningBlock(byte[] signingBlock, byte[] blockValue, int blockId)
+      throws IOException {
+    ImmutableList<Pair<byte[], Integer>> entries =
+        ImmutableList.<Pair<byte[], Integer>>builder()
+            .addAll(extractAllSigners(DataSources.asDataSource(ByteBuffer.wrap(signingBlock))))
+            .add(Pair.of(blockValue, blockId))
+            .build();
+    return ApkSigningBlockUtils.generateApkSigningBlock(entries);
+  }
+
+  /**
+   * Generate APK sig v2 block containing a block composed of the provided block value and id, and
+   * (optionally) padding block.
+   */
+  private static byte[] createSigningBlock(byte[] blockValue, int blockId) {
+    return ApkSigningBlockUtils.generateApkSigningBlock(
+        ImmutableList.of(Pair.of(blockValue, blockId)));
+  }
+
+  /**
+   * Extracts all signing block entries except padding block.
+   *
+   * @param signingBlock APK v2 signing block containing: length prefix, signers (can include
+   *     padding block), length postfix and APK sig v2 block magic.
+   * @return list of block entry value and block entry id pairs.
+   */
+  private static ImmutableList<Pair<byte[], Integer>> extractAllSigners(DataSource signingBlock)
+      throws IOException {
+    long wholeBlockSize = signingBlock.size();
+    // Take the segment of the existing signing block without the length prefix (8 bytes)
+    // at the beginning and the length and magic (24 bytes) at the end, so it is just the sequence
+    // of length prefix id value pairs.
+    DataSource lengthPrefixedIdValuePairsSource =
+        signingBlock.slice(
+            SIZE_OF_BLOCK_NUM_BYTES,
+            wholeBlockSize - 2 * SIZE_OF_BLOCK_NUM_BYTES - MAGIC_NUM_BYTES);
+    final int lengthAndIdByteCount = BLOCK_LENGTH_NUM_BYTES + BLOCK_ID_NUM_BYTES;
+    ByteBuffer lengthAndId = ByteBuffer.allocate(lengthAndIdByteCount).order(LITTLE_ENDIAN);
+    ImmutableList.Builder<Pair<byte[], Integer>> idValuePairs = ImmutableList.builder();
+
+    for (int index = 0; index <= lengthPrefixedIdValuePairsSource.size() - lengthAndIdByteCount; ) {
+      lengthPrefixedIdValuePairsSource.copyTo(index, lengthAndIdByteCount, lengthAndId);
+      lengthAndId.flip();
+      int blockLength = Ints.checkedCast(lengthAndId.getLong());
+      int id = lengthAndId.getInt();
+      lengthAndId.clear();
+
+      if (id != VERITY_PADDING_BLOCK_ID) {
+        int blockValueSize = blockLength - BLOCK_ID_NUM_BYTES;
+        ByteBuffer blockValue = ByteBuffer.allocate(blockValueSize);
+        lengthPrefixedIdValuePairsSource.copyTo(
+            index + BLOCK_LENGTH_NUM_BYTES + BLOCK_ID_NUM_BYTES, blockValueSize, blockValue);
+        idValuePairs.add(Pair.of(blockValue.array(), id));
+      }
+
+      index += blockLength + BLOCK_LENGTH_NUM_BYTES;
+    }
+    return idValuePairs.build();
+  }
+
+  /**
+   * Extract a block with the given id from the APK. If there is more than one block with the same
+   * ID, the first block will be returned. If there are no block with the give id, {@code null} will
+   * be returned.
+   *
+   * @param apk APK file
+   * @param blockId id of the block to be extracted.
+   */
+  @Nullable
+  public static ByteBuffer extractBlock(File apk, int blockId)
+      throws IOException, ZipFormatException, ApkSigningBlockNotFoundException {
+    try (RandomAccessFile file = new RandomAccessFile(apk, "r")) {
+      DataSource apkDataSource = DataSources.asDataSource(file);
+      ApkSigningBlock signingBlockInfo =
+          ApkUtils.findApkSigningBlock(apkDataSource, ApkUtils.findZipSections(apkDataSource));
+
+      DataSource wholeV2Block = signingBlockInfo.getContents();
+      final int lengthAndIdByteCount = BLOCK_LENGTH_NUM_BYTES + BLOCK_ID_NUM_BYTES;
+      DataSource signingBlock =
+          wholeV2Block.slice(
+              SIZE_OF_BLOCK_NUM_BYTES,
+              wholeV2Block.size() - SIZE_OF_BLOCK_NUM_BYTES - MAGIC_NUM_BYTES);
+      ByteBuffer lengthAndId =
+          ByteBuffer.allocate(lengthAndIdByteCount).order(ByteOrder.LITTLE_ENDIAN);
+      for (int index = 0; index <= signingBlock.size() - lengthAndIdByteCount; ) {
+        signingBlock.copyTo(index, lengthAndIdByteCount, lengthAndId);
+        lengthAndId.flip();
+        int blockLength = (int) lengthAndId.getLong();
+        int id = lengthAndId.getInt();
+        lengthAndId.flip();
+        if (id == blockId) {
+          ByteBuffer block = ByteBuffer.allocate(blockLength - BLOCK_ID_NUM_BYTES);
+          signingBlock.copyTo(
+              index + lengthAndIdByteCount, blockLength - BLOCK_ID_NUM_BYTES, block);
+          block.flip();
+          return block;
+        }
+        index += blockLength + BLOCK_LENGTH_NUM_BYTES;
+      }
+      return null;
+    }
+  }
+
+  private SigningBlockUtils() {}
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/utils/package-info.java

@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/** Utilities to work with {@code apkzlib}. */
+package com.android.tools.build.apkzlib.utils;

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ApkCreator.java

@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+import com.google.common.base.Function;
+import com.google.common.base.Predicate;
+import java.io.Closeable;
+import java.io.File;
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/** Creates or updates APKs based on provided entries. */
+public interface ApkCreator extends Closeable {
+
+  /**
+   * Copies the content of a Jar/Zip archive into the receiver archive.
+   *
+   * <p>An optional predicate allows to selectively choose which files to copy over and an option
+   * function allows renaming the files as they are copied.
+   *
+   * @param zip the zip to copy data from
+   * @param transform an optional transform to apply to file names before copying them
+   * @param isIgnored an optional filter or {@code null} to mark which out files should not be
+   *     added, even through they are on the zip; if {@code transform} is specified, then this
+   *     predicate applies after transformation
+   * @throws IOException I/O error
+   */
+  void writeZip(
+      File zip, @Nullable Function<String, String> transform, @Nullable Predicate<String> isIgnored)
+      throws IOException;
+
+  /**
+   * Writes a new {@link File} into the archive. If a file already existed with the given path, it
+   * should be replaced.
+   *
+   * @param inputFile the {@link File} to write.
+   * @param apkPath the filepath inside the archive.
+   * @throws IOException I/O error
+   */
+  void writeFile(File inputFile, String apkPath) throws IOException;
+
+  /**
+   * Deletes a file in a given path.
+   *
+   * @param apkPath the path to remove
+   * @throws IOException failed to remove the entry
+   */
+  void deleteFile(String apkPath) throws IOException;
+
+  /** Returns true if the APK will be rewritten on close. */
+  boolean hasPendingChangesWithWait() throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ApkCreatorFactory.java

@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+import com.android.tools.build.apkzlib.sign.SigningOptions;
+import com.google.auto.value.AutoValue;
+import com.google.common.base.Optional;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Predicate;
+import java.io.File;
+import javax.annotation.Nonnull;
+import javax.annotation.Nullable;
+
+/** Factory that creates instances of {@link ApkCreator}. */
+public interface ApkCreatorFactory {
+
+  /**
+   * Creates an {@link ApkCreator} with a given output location, and signing information.
+   *
+   * @param creationData the information to create the APK
+   */
+  ApkCreator make(CreationData creationData);
+
+  /**
+   * Data structure with the required information to initiate the creation of an APK. See {@link
+   * ApkCreatorFactory#make(CreationData)}.
+   */
+  @AutoValue
+  abstract class CreationData {
+
+    /** An implementation of builder pattern to create a {@link CreationData} object. */
+    @AutoValue.Builder
+    public abstract static class Builder {
+      public abstract Builder setApkPath(@Nonnull File apkPath);
+
+      public abstract Builder setSigningOptions(@Nonnull SigningOptions signingOptions);
+
+      public abstract Builder setBuiltBy(@Nullable String buildBy);
+
+      public abstract Builder setCreatedBy(@Nullable String createdBy);
+
+      public abstract Builder setNativeLibrariesPackagingMode(
+          NativeLibrariesPackagingMode packagingMode);
+
+      public abstract Builder setNoCompressPredicate(Predicate<String> predicate);
+
+      public abstract Builder setIncremental(boolean incremental);
+
+      abstract CreationData autoBuild();
+
+      public CreationData build() {
+        CreationData data = autoBuild();
+        Preconditions.checkArgument(data.getApkPath() != null, "Output apk path is not set");
+        return data;
+      }
+    }
+
+    public static Builder builder() {
+      return new AutoValue_ApkCreatorFactory_CreationData.Builder()
+          .setBuiltBy(null)
+          .setCreatedBy(null)
+          .setNoCompressPredicate(s -> false)
+          .setIncremental(false);
+    }
+
+    /**
+     * Obtains the path where the APK should be located. If the path already exists, then the APK
+     * may be updated instead of re-created.
+     *
+     * @return the path that may already exist or not
+     */
+    public abstract File getApkPath();
+
+    /**
+     * Obtains the data used to sign the APK.
+     *
+     * @return the SigningOptions
+     */
+    @Nonnull
+    public abstract Optional<SigningOptions> getSigningOptions();
+
+    /**
+     * Obtains the "built-by" text for the APK.
+     *
+     * @return the text or {@code null} if the default should be used
+     */
+    @Nullable
+    public abstract String getBuiltBy();
+
+    /**
+     * Obtains the "created-by" text for the APK.
+     *
+     * @return the text or {@code null} if the default should be used
+     */
+    @Nullable
+    public abstract String getCreatedBy();
+
+    /** Returns the packaging policy that the {@link ApkCreator} should use for native libraries. */
+    public abstract NativeLibrariesPackagingMode getNativeLibrariesPackagingMode();
+
+    /** Returns the predicate to decide which file paths should be uncompressed. */
+    public abstract Predicate<String> getNoCompressPredicate();
+
+    /**
+     * Returns if this apk build is incremental.
+     *
+     * As mentioned in {@link getApkPath} description, we may already have an existing apk in place.
+     * This is the case when e.g. building APK via build system and this is not the first build.
+     * In that case the build is called incremental and internal APK data might be reused speeding
+     * the build up.
+     */
+    public abstract boolean isIncremental();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ApkZFileCreator.java

@@ -0,0 +1,176 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+import com.android.tools.build.apkzlib.zip.AlignmentRule;
+import com.android.tools.build.apkzlib.zip.AlignmentRules;
+import com.android.tools.build.apkzlib.zip.StoredEntry;
+import com.android.tools.build.apkzlib.zip.ZFile;
+import com.android.tools.build.apkzlib.zip.ZFileOptions;
+import com.google.common.base.Function;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Predicate;
+import com.google.common.io.Closer;
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import javax.annotation.Nullable;
+
+/** {@link ApkCreator} that uses {@link ZFileOptions} to generate the APK. */
+class ApkZFileCreator implements ApkCreator {
+
+  /** Suffix for native libraries. */
+  private static final String NATIVE_LIBRARIES_SUFFIX = ".so";
+
+  /** Shared libraries are alignment at 4096 boundaries. */
+  private static final AlignmentRule SO_RULE =
+      AlignmentRules.constantForSuffix(NATIVE_LIBRARIES_SUFFIX, 4096);
+
+  /** The zip file. */
+  private final ZFile zip;
+
+  /** Has the zip file been closed? */
+  private boolean closed;
+
+  /** Predicate defining which files should not be compressed. */
+  private final Predicate<String> noCompressPredicate;
+
+  /**
+   * Creates a new creator.
+   *
+   * @param creationData the data needed to create the APK
+   * @param options zip file options
+   * @throws IOException failed to create the zip
+   */
+  ApkZFileCreator(ApkCreatorFactory.CreationData creationData, ZFileOptions options)
+      throws IOException {
+
+    switch (creationData.getNativeLibrariesPackagingMode()) {
+      case COMPRESSED:
+        noCompressPredicate = creationData.getNoCompressPredicate();
+        break;
+      case UNCOMPRESSED_AND_ALIGNED:
+        Predicate<String> baseNoCompressPredicate = creationData.getNoCompressPredicate();
+        noCompressPredicate =
+            name -> baseNoCompressPredicate.apply(name) || name.endsWith(NATIVE_LIBRARIES_SUFFIX);
+        options.setAlignmentRule(AlignmentRules.compose(SO_RULE, options.getAlignmentRule()));
+        break;
+      default:
+        throw new AssertionError();
+    }
+    // In case of incremental build we can skip validation since we generated the previous apk and
+    // we trust ourselves
+    options.setSkipValidation(creationData.isIncremental());
+
+    zip =
+        ZFiles.apk(
+            creationData.getApkPath(),
+            options,
+            creationData.getSigningOptions(),
+            creationData.getBuiltBy(),
+            creationData.getCreatedBy());
+    closed = false;
+  }
+
+  @Override
+  public void writeZip(
+      File zip, @Nullable Function<String, String> transform, @Nullable Predicate<String> isIgnored)
+      throws IOException {
+    Preconditions.checkState(!closed, "closed == true");
+    Preconditions.checkArgument(zip.isFile(), "!zip.isFile()");
+
+    Closer closer = Closer.create();
+    try {
+      ZFile toMerge = closer.register(ZFile.openReadWrite(zip));
+
+      Predicate<String> ignorePredicate;
+      if (isIgnored == null) {
+        ignorePredicate = s -> false;
+      } else {
+        ignorePredicate = isIgnored;
+      }
+
+      // Files that *must* be uncompressed in the result should not be merged and should be
+      // added after. This is just very slightly less efficient than ignoring just the ones
+      // that were compressed and must be uncompressed, but it is a lot simpler :)
+      Predicate<String> noMergePredicate =
+          v -> ignorePredicate.apply(v) || noCompressPredicate.apply(v);
+
+      this.zip.mergeFrom(toMerge, noMergePredicate);
+
+      for (StoredEntry toMergeEntry : toMerge.entries()) {
+        String path = toMergeEntry.getCentralDirectoryHeader().getName();
+        if (noCompressPredicate.apply(path) && !ignorePredicate.apply(path)) {
+          // This entry *must* be uncompressed so it was ignored in the merge and should
+          // now be added to the apk.
+          try (InputStream ignoredData = toMergeEntry.open()) {
+            this.zip.add(path, ignoredData, false);
+          }
+        }
+      }
+    } catch (Throwable t) {
+      throw closer.rethrow(t);
+    } finally {
+      closer.close();
+    }
+  }
+
+  @Override
+  public void writeFile(File inputFile, String apkPath) throws IOException {
+    Preconditions.checkState(!closed, "closed == true");
+
+    boolean mayCompress = !noCompressPredicate.apply(apkPath);
+
+    Closer closer = Closer.create();
+    try {
+      FileInputStream inputFileStream = closer.register(new FileInputStream(inputFile));
+      zip.add(apkPath, inputFileStream, mayCompress);
+    } catch (IOException e) {
+      throw closer.rethrow(e, IOException.class);
+    } catch (Throwable t) {
+      throw closer.rethrow(t);
+    } finally {
+      closer.close();
+    }
+  }
+
+  @Override
+  public void deleteFile(String apkPath) throws IOException {
+    Preconditions.checkState(!closed, "closed == true");
+
+    StoredEntry entry = zip.get(apkPath);
+    if (entry != null) {
+      entry.delete();
+    }
+  }
+
+  @Override
+  public boolean hasPendingChangesWithWait() throws IOException {
+    return zip.hasPendingChangesWithWait();
+  }
+
+  @Override
+  public void close() throws IOException {
+    if (closed) {
+      return;
+    }
+
+    zip.close();
+    closed = true;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ApkZFileCreatorFactory.java

@@ -0,0 +1,46 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.android.tools.build.apkzlib.zip.ZFileOptions;
+import java.io.IOException;
+
+/** Creates instances of {@link ApkZFileCreator}. */
+public class ApkZFileCreatorFactory implements ApkCreatorFactory {
+
+  /** Options for the {@link ZFileOptions} to use in all APKs. */
+  private final ZFileOptions options;
+
+  /**
+   * Creates a new factory.
+   *
+   * @param options the options to use for all instances created
+   */
+  public ApkZFileCreatorFactory(ZFileOptions options) {
+    this.options = options;
+  }
+
+  @Override
+  public ApkCreator make(CreationData creationData) {
+    try {
+      return new ApkZFileCreator(creationData, options);
+    } catch (IOException e) {
+      throw new IOExceptionWrapper(e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ManifestAttributes.java

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+/** Java manifest attributes and some default values. */
+public interface ManifestAttributes {
+  /** Manifest attribute with the built by information. */
+  String BUILT_BY = "Built-By";
+
+  /** Manifest attribute with the created by information. */
+  String CREATED_BY = "Created-By";
+
+  /** Manifest attribute with the manifest version. */
+  String MANIFEST_VERSION = "Manifest-Version";
+
+  /** Manifest attribute value with the manifest version. */
+  String CURRENT_MANIFEST_VERSION = "1.0";
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/NativeLibrariesPackagingMode.java

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+/** Describes how native libs should be packaged. */
+public enum NativeLibrariesPackagingMode {
+  /** Native libs are packaged as any other file. */
+  COMPRESSED,
+
+  /**
+   * Native libs are packaged uncompressed and page-aligned, so they can be mapped into memory at
+   * runtime.
+   *
+   * <p>Support for this mode was added in Android 23, it only works if the {@code
+   * extractNativeLibs} attribute is set in the manifest.
+   */
+  UNCOMPRESSED_AND_ALIGNED;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/ZFiles.java

@@ -0,0 +1,139 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zfile;
+
+import com.android.tools.build.apkzlib.sign.ManifestGenerationExtension;
+import com.android.tools.build.apkzlib.sign.SigningExtension;
+import com.android.tools.build.apkzlib.sign.SigningOptions;
+import com.android.tools.build.apkzlib.zip.AlignmentRule;
+import com.android.tools.build.apkzlib.zip.AlignmentRules;
+import com.android.tools.build.apkzlib.zip.ZFile;
+import com.android.tools.build.apkzlib.zip.ZFileOptions;
+import com.google.common.base.Optional;
+import java.io.File;
+import java.io.IOException;
+import java.security.InvalidKeyException;
+import java.security.NoSuchAlgorithmException;
+import javax.annotation.Nullable;
+
+/** Factory for {@link ZFile}s that are specifically configured to be APKs, AARs, ... */
+public class ZFiles {
+
+  /** By default all non-compressed files are alignment at 4 byte boundaries.. */
+  private static final AlignmentRule APK_DEFAULT_RULE = AlignmentRules.constant(4);
+
+  /** Default build by string. */
+  private static final String DEFAULT_BUILD_BY = "Generated-by-ADT";
+
+  /** Default created by string. */
+  private static final String DEFAULT_CREATED_BY = "Generated-by-ADT";
+
+  /**
+   * Creates a new zip file configured as an apk, based on a given file.
+   *
+   * @param f the file, if this path does not represent an existing path, will create a {@link
+   *     ZFile} based on an non-existing path (a zip will be created when {@link ZFile#close()} is
+   *     invoked)
+   * @param options the options to create the {@link ZFile}
+   * @return the zip file
+   * @throws IOException failed to create the zip file
+   */
+  public static ZFile apk(File f, ZFileOptions options) throws IOException {
+    options.setAlignmentRule(AlignmentRules.compose(options.getAlignmentRule(), APK_DEFAULT_RULE));
+    return ZFile.openReadWrite(f, options);
+  }
+
+  /**
+   * Creates a new zip file configured as an apk, based on a given file.
+   *
+   * @param f the file, if this path does not represent an existing path, will create a {@link
+   *     ZFile} based on an non-existing path (a zip will be created when {@link ZFile#close()} is
+   *     invoked)
+   * @param options the options to create the {@link ZFile}
+   * @param signingOptions the options to sign the apk
+   * @param builtBy who to mark as builder in the manifest
+   * @param createdBy who to mark as creator in the manifest
+   * @return the zip file
+   * @throws IOException failed to create the zip file
+   */
+  public static ZFile apk(
+      File f,
+      ZFileOptions options,
+      Optional<SigningOptions> signingOptions,
+      @Nullable String builtBy,
+      @Nullable String createdBy)
+      throws IOException {
+    return apk(
+        f, options, signingOptions, builtBy, createdBy, options.getAlwaysGenerateJarManifest());
+  }
+
+  /**
+   * Creates a new zip file configured as an apk, based on a given file.
+   *
+   * @param f the file, if this path does not represent an existing path, will create a {@link
+   *     ZFile} based on an non-existing path (a zip will be created when {@link ZFile#close()} is
+   *     invoked)
+   * @param options the options to create the {@link ZFile}
+   * @param signingOptions the options to sign the apk
+   * @param builtBy who to mark as builder in the manifest
+   * @param createdBy who to mark as creator in the manifest
+   * @param writeManifest a migration parameter that forces keeping (useless) manifest.mf file in
+   *     apk file in order to prevent breaking changes. Clients of the previous interface will still
+   *     get apk with manifest.mf because the flag is true by default
+   * @return the zip file
+   * @throws IOException failed to create the zip file
+   * @deprecated Use ZFileOptions.setAlwaysGenerateJarManifest() instead.
+   */
+  @Deprecated
+  // This method can be removed once ZFileOptions.getAlwaysGenerateJarManifest() is on Maven.
+  public static ZFile apk(
+      File f,
+      ZFileOptions options,
+      Optional<SigningOptions> signingOptions,
+      @Nullable String builtBy,
+      @Nullable String createdBy,
+      boolean writeManifest)
+      throws IOException {
+    ZFile zfile = apk(f, options);
+
+    if ((signingOptions.isPresent() && signingOptions.get().isV1SigningEnabled())
+        || writeManifest) {
+      if (builtBy == null) {
+        builtBy = DEFAULT_BUILD_BY;
+      }
+
+      if (createdBy == null) {
+        createdBy = DEFAULT_CREATED_BY;
+      }
+      ManifestGenerationExtension manifestExt = new ManifestGenerationExtension(builtBy, createdBy);
+      manifestExt.register(zfile);
+    }
+
+    if (signingOptions.isPresent()) {
+      SigningOptions signOptions = signingOptions.get();
+      try {
+        new SigningExtension(signOptions).register(zfile);
+      } catch (NoSuchAlgorithmException | InvalidKeyException e) {
+        throw new IOException("Failed to create signature extensions", e);
+      }
+    }
+
+    return zfile;
+  }
+
+  private ZFiles() {}
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zfile/package-info.java

@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/** The {@code zfile} package contains */
+package com.android.tools.build.apkzlib.zfile;

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/AlignmentRule.java

@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+
+/** An alignment rule defines how to a file should be aligned in a zip, based on its name. */
+public interface AlignmentRule {
+
+  /** Alignment value of files that do not require alignment. */
+  int NO_ALIGNMENT = 1;
+
+  /**
+   * Obtains the alignment this rule computes for a given path.
+   *
+   * @param path the path in the zip file
+   * @return the alignment value, always greater than {@code 0}; if this rule places no restrictions
+   *     on the provided path, then {@link AlignmentRule#NO_ALIGNMENT} is returned
+   */
+  int alignment(String path);
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/AlignmentRules.java

@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.google.common.base.Preconditions;
+
+/** Factory for instances of {@link AlignmentRule}. */
+public final class AlignmentRules {
+
+  private AlignmentRules() {}
+
+  /**
+   * A rule that defines a constant alignment for all files.
+   *
+   * @param alignment the alignment
+   * @return the rule
+   */
+  public static AlignmentRule constant(int alignment) {
+    Preconditions.checkArgument(alignment > 0, "alignment <= 0");
+
+    return (String path) -> alignment;
+  }
+
+  /**
+   * A rule that defines constant alignment for all files with a certain suffix, placing no
+   * restrictions on other files.
+   *
+   * @param suffix the suffix
+   * @param alignment the alignment for paths that match the provided suffix
+   * @return the rule
+   */
+  public static AlignmentRule constantForSuffix(String suffix, int alignment) {
+    Preconditions.checkArgument(!suffix.isEmpty(), "suffix.isEmpty()");
+    Preconditions.checkArgument(alignment > 0, "alignment <= 0");
+
+    return (String path) -> path.endsWith(suffix) ? alignment : AlignmentRule.NO_ALIGNMENT;
+  }
+
+  /**
+   * A rule that applies other rules in order.
+   *
+   * @param rules all rules to be tried; the first rule that does not return {@link
+   *     AlignmentRule#NO_ALIGNMENT} will define the alignment for a path; if there are no rules
+   *     that return a value different from {@link AlignmentRule#NO_ALIGNMENT}, then {@link
+   *     AlignmentRule#NO_ALIGNMENT} is returned
+   * @return the composition rule
+   */
+  public static AlignmentRule compose(AlignmentRule... rules) {
+    return (String path) -> {
+      for (AlignmentRule r : rules) {
+        int align = r.alignment(path);
+        if (align != AlignmentRule.NO_ALIGNMENT) {
+          return align;
+        }
+      }
+
+      return AlignmentRule.NO_ALIGNMENT;
+    };
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/CentralDirectory.java

@@ -0,0 +1,482 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.utils.CachedSupplier;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.android.tools.build.apkzlib.zip.utils.MsDosDateTimeUtils;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Preconditions;
+import com.google.common.collect.ImmutableMap;
+import com.google.common.collect.Lists;
+import com.google.common.collect.Maps;
+import com.google.common.primitives.Ints;
+import com.google.common.util.concurrent.Futures;
+import com.google.common.util.concurrent.ListenableFuture;
+import java.io.IOException;
+import java.io.UncheckedIOException;
+import java.nio.ByteBuffer;
+import java.util.Collections;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+
+/** Representation of the central directory of a zip archive. */
+class CentralDirectory {
+
+  /** Field in the central directory with the central directory signature. */
+  private static final ZipField.F4 F_SIGNATURE = new ZipField.F4(0, 0x02014b50, "Signature");
+
+  /** Field in the central directory with the "made by" code. */
+  private static final ZipField.F2 F_MADE_BY =
+      new ZipField.F2(F_SIGNATURE.endOffset(), "Made by", new ZipFieldInvariantNonNegative());
+
+  /** Field in the central directory with the minimum version required to extract the entry. */
+  @VisibleForTesting
+  static final ZipField.F2 F_VERSION_EXTRACT =
+      new ZipField.F2(
+          F_MADE_BY.endOffset(), "Version to extract", new ZipFieldInvariantNonNegative());
+
+  /** Field in the central directory with the GP bit flag. */
+  private static final ZipField.F2 F_GP_BIT =
+      new ZipField.F2(F_VERSION_EXTRACT.endOffset(), "GP bit");
+
+  /**
+   * Field in the central directory with the code of the compression method. See {@link
+   * CompressionMethod#fromCode(long)}.
+   */
+  private static final ZipField.F2 F_METHOD = new ZipField.F2(F_GP_BIT.endOffset(), "Method");
+
+  /**
+   * Field in the central directory with the last modification time in MS-DOS format (see {@link
+   * MsDosDateTimeUtils#packTime(long)}).
+   */
+  private static final ZipField.F2 F_LAST_MOD_TIME =
+      new ZipField.F2(F_METHOD.endOffset(), "Last modification time");
+
+  /**
+   * Field in the central directory with the last modification date in MS-DOS format. See {@link
+   * MsDosDateTimeUtils#packDate(long)}.
+   */
+  private static final ZipField.F2 F_LAST_MOD_DATE =
+      new ZipField.F2(F_LAST_MOD_TIME.endOffset(), "Last modification date");
+
+  /**
+   * Field in the central directory with the CRC32 checksum of the entry. This will be zero for
+   * directories and files with no content.
+   */
+  private static final ZipField.F4 F_CRC32 = new ZipField.F4(F_LAST_MOD_DATE.endOffset(), "CRC32");
+
+  /**
+   * Field in the central directory with the entry's compressed size, <em>i.e.</em>, the file on the
+   * archive. This will be the same as the uncompressed size if the method is {@link
+   * CompressionMethod#STORE}.
+   */
+  private static final ZipField.F4 F_COMPRESSED_SIZE =
+      new ZipField.F4(F_CRC32.endOffset(), "Compressed size", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the central directory with the entry's uncompressed size, <em>i.e.</em>, the size the
+   * file will have when extracted from the zip. This will be zero for directories and empty files
+   * and will be the same as the compressed size if the method is {@link CompressionMethod#STORE}.
+   */
+  private static final ZipField.F4 F_UNCOMPRESSED_SIZE =
+      new ZipField.F4(
+          F_COMPRESSED_SIZE.endOffset(), "Uncompressed size", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the central directory with the length of the file name. The file name is stored after
+   * the offset field ({@link #F_OFFSET}). The number of characters in the file name are stored in
+   * this field.
+   */
+  private static final ZipField.F2 F_FILE_NAME_LENGTH =
+      new ZipField.F2(
+          F_UNCOMPRESSED_SIZE.endOffset(), "File name length", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the central directory with the length of the extra field. The extra field is stored
+   * after the file name ({@link #F_FILE_NAME_LENGTH}). The contents of this field are partially
+   * defined in the zip specification but we do not parse it.
+   */
+  private static final ZipField.F2 F_EXTRA_FIELD_LENGTH =
+      new ZipField.F2(
+          F_FILE_NAME_LENGTH.endOffset(), "Extra field length", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the central directory with the length of the comment. The comment is stored after the
+   * extra field ({@link #F_EXTRA_FIELD_LENGTH}). We do not parse the comment.
+   */
+  private static final ZipField.F2 F_COMMENT_LENGTH =
+      new ZipField.F2(
+          F_EXTRA_FIELD_LENGTH.endOffset(), "Comment length", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Number of the disk where the central directory starts. Because we do not support multi-file
+   * archives, this field has to have value {@code 0}.
+   */
+  private static final ZipField.F2 F_DISK_NUMBER_START =
+      new ZipField.F2(F_COMMENT_LENGTH.endOffset(), 0, "Disk start");
+
+  /** Internal attributes. This field can only contain one bit set, the {@link #ASCII_BIT}. */
+  private static final ZipField.F2 F_INTERNAL_ATTRIBUTES =
+      new ZipField.F2(F_DISK_NUMBER_START.endOffset(), "Int attributes");
+
+  /** External attributes. This field is ignored. */
+  private static final ZipField.F4 F_EXTERNAL_ATTRIBUTES =
+      new ZipField.F4(F_INTERNAL_ATTRIBUTES.endOffset(), "Ext attributes");
+
+  /**
+   * Offset into the archive where the entry starts. This is the offset to the local header (see
+   * {@link StoredEntry} for information on the local header), not to the file data itself. The file
+   * data, if there is any, will be stored after the local header.
+   */
+  private static final ZipField.F4 F_OFFSET =
+      new ZipField.F4(
+          F_EXTERNAL_ATTRIBUTES.endOffset(), "Offset", new ZipFieldInvariantNonNegative());
+
+  /** Maximum supported version to extract. */
+  private static final int MAX_VERSION_TO_EXTRACT = 20;
+
+  /**
+   * Bit that can be set on the internal attributes stating that the file is an ASCII file. We don't
+   * do anything with this information, but we check that nothing unexpected appears in the internal
+   * attributes.
+   */
+  private static final int ASCII_BIT = 1;
+
+  /** Contains all entries in the directory mapped from their names. */
+  private final Map<String, StoredEntry> entries;
+
+  /** The file where this directory belongs to. */
+  private final ZFile file;
+
+  /** Supplier that provides a byte representation of the central directory. */
+  private final CachedSupplier<byte[]> bytesSupplier;
+
+  /** Verify log for the central directory. */
+  private final VerifyLog verifyLog;
+
+  /**
+   * Creates a new, empty, central directory, for a given zip file.
+   *
+   * @param file the file
+   */
+  CentralDirectory(ZFile file) {
+    entries = Maps.newHashMap();
+    this.file = file;
+    bytesSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+    verifyLog = file.getVerifyLog();
+  }
+
+  /**
+   * Reads the central directory data from a zip file, parses it, and creates the in-memory
+   * structure representing the directory.
+   *
+   * @param bytes the data of the central directory; the directory is read from the buffer's current
+   *     position; when this method terminates, the buffer's position is the first byte after the
+   *     directory
+   * @param count the number of entries expected in the central directory (usually read from the
+   *     {@link Eocd}).
+   * @param file the zip file this central directory belongs to
+   * @param storage the storage used to generate sources with entry data
+   * @return the central directory
+   * @throws IOException failed to read data from the zip, or the central directory is corrupted or
+   *     has unsupported features
+   */
+  static CentralDirectory makeFromData(ByteBuffer bytes, long count, ZFile file, ByteStorage storage)
+      throws IOException {
+    Preconditions.checkNotNull(bytes, "bytes == null");
+    Preconditions.checkArgument(count >= 0, "count < 0");
+
+    CentralDirectory directory = new CentralDirectory(file);
+
+    for (long i = 0; i < count; i++) {
+      try {
+        directory.readEntry(bytes, storage);
+      } catch (IOException e) {
+        throw new IOException(
+            "Failed to read directory entry index "
+                + i
+                + " (total "
+                + "directory bytes read: "
+                + bytes.position()
+                + ").",
+            e);
+      }
+    }
+
+    return directory;
+  }
+
+  /**
+   * Creates a new central directory from the entries. This is used to build a new central directory
+   * from entries in the zip file.
+   *
+   * @param entries the entries in the zip file
+   * @param file the zip file itself
+   * @return the created central directory
+   */
+  static CentralDirectory makeFromEntries(Set<StoredEntry> entries, ZFile file) {
+    CentralDirectory directory = new CentralDirectory(file);
+    for (StoredEntry entry : entries) {
+      CentralDirectoryHeader cdr = entry.getCentralDirectoryHeader();
+      Preconditions.checkArgument(
+          !directory.entries.containsKey(cdr.getName()), "Duplicate filename");
+      directory.entries.put(cdr.getName(), entry);
+    }
+
+    return directory;
+  }
+
+  /**
+   * Reads the next entry from the central directory and adds it to {@link #entries}.
+   *
+   * @param bytes the central directory's data, positioned starting at the beginning of the next
+   *     entry to read; when finished, the buffer's position will be at the first byte after the
+   *     entry
+   * @param storage the storage used to generate sources to store entry data
+   * @throws IOException failed to read the directory entry, either because of an I/O error, because
+   *     it is corrupt or contains unsupported features
+   */
+  private void readEntry(ByteBuffer bytes, ByteStorage storage) throws IOException {
+    F_SIGNATURE.verify(bytes);
+    long madeBy = F_MADE_BY.read(bytes);
+
+    long versionNeededToExtract = F_VERSION_EXTRACT.read(bytes);
+    verifyLog.verify(
+        versionNeededToExtract <= MAX_VERSION_TO_EXTRACT,
+        "Ignored unknown version needed to extract in zip directory entry: %s.",
+        versionNeededToExtract);
+
+    long gpBit = F_GP_BIT.read(bytes);
+    GPFlags flags = GPFlags.from(gpBit);
+
+    long methodCode = F_METHOD.read(bytes);
+    CompressionMethod method = CompressionMethod.fromCode(methodCode);
+    verifyLog.verify(method != null, "Unknown method in zip directory entry: %s.", methodCode);
+
+    long lastModTime;
+    long lastModDate;
+    if (file.areTimestampsIgnored()) {
+      lastModTime = 0;
+      lastModDate = 0;
+      F_LAST_MOD_TIME.skip(bytes);
+      F_LAST_MOD_DATE.skip(bytes);
+    } else {
+      lastModTime = F_LAST_MOD_TIME.read(bytes);
+      lastModDate = F_LAST_MOD_DATE.read(bytes);
+    }
+
+    long crc32 = F_CRC32.read(bytes);
+    long compressedSize = F_COMPRESSED_SIZE.read(bytes);
+    long uncompressedSize = F_UNCOMPRESSED_SIZE.read(bytes);
+    int fileNameLength = Ints.checkedCast(F_FILE_NAME_LENGTH.read(bytes));
+    int extraFieldLength = Ints.checkedCast(F_EXTRA_FIELD_LENGTH.read(bytes));
+    int fileCommentLength = Ints.checkedCast(F_COMMENT_LENGTH.read(bytes));
+
+    F_DISK_NUMBER_START.verify(bytes, verifyLog);
+    long internalAttributes = F_INTERNAL_ATTRIBUTES.read(bytes);
+    verifyLog.verify(
+        (internalAttributes & ~ASCII_BIT) == 0,
+        "Ignored invalid internal attributes: %s.",
+        internalAttributes);
+
+    long externalAttributes = F_EXTERNAL_ATTRIBUTES.read(bytes);
+    long entryOffset = F_OFFSET.read(bytes);
+
+    long remainingSize = (long) fileNameLength + extraFieldLength + fileCommentLength;
+
+    if (bytes.remaining() < fileNameLength + extraFieldLength + fileCommentLength) {
+      throw new IOException(
+          "Directory entry should have "
+              + remainingSize
+              + " bytes remaining (name = "
+              + fileNameLength
+              + ", extra = "
+              + extraFieldLength
+              + ", comment = "
+              + fileCommentLength
+              + "), but it has "
+              + bytes.remaining()
+              + ".");
+    }
+
+    byte[] encodedFileName = new byte[fileNameLength];
+    bytes.get(encodedFileName);
+    String fileName = EncodeUtils.decode(encodedFileName, flags);
+
+    byte[] extraField = new byte[extraFieldLength];
+    bytes.get(extraField);
+
+    byte[] fileCommentField = new byte[fileCommentLength];
+    bytes.get(fileCommentField);
+
+    /*
+     * Tricky: to create a CentralDirectoryHeader we need the future that will hold the result
+     * of the compress information. But, to actually create the result of the compress
+     * information we need the CentralDirectoryHeader
+     */
+    ListenableFuture<CentralDirectoryHeaderCompressInfo> compressInfo =
+        Futures.immediateFuture(
+            new CentralDirectoryHeaderCompressInfo(method, compressedSize, versionNeededToExtract));
+    CentralDirectoryHeader centralDirectoryHeader =
+        new CentralDirectoryHeader(
+            fileName,
+            encodedFileName,
+            uncompressedSize,
+            compressInfo,
+            flags,
+            file,
+            lastModTime,
+            lastModDate);
+    centralDirectoryHeader.setMadeBy(madeBy);
+    centralDirectoryHeader.setLastModTime(lastModTime);
+    centralDirectoryHeader.setLastModDate(lastModDate);
+    centralDirectoryHeader.setCrc32(crc32);
+    centralDirectoryHeader.setInternalAttributes(internalAttributes);
+    centralDirectoryHeader.setExternalAttributes(externalAttributes);
+    centralDirectoryHeader.setOffset(entryOffset);
+    centralDirectoryHeader.setExtraFieldNoNotify(new ExtraField(extraField));
+    centralDirectoryHeader.setComment(fileCommentField);
+
+    StoredEntry entry;
+
+    try {
+      entry = new StoredEntry(centralDirectoryHeader, file, null, storage);
+    } catch (IOException e) {
+      throw new IOException("Failed to read stored entry '" + fileName + "'.", e);
+    }
+
+    if (entries.containsKey(fileName)) {
+      verifyLog.log("File file contains duplicate file '" + fileName + "'.");
+    }
+
+    entries.put(fileName, entry);
+  }
+
+  /**
+   * Obtains all the entries in the central directory.
+   *
+   * @return all entries on a non-modifiable map
+   */
+  Map<String, StoredEntry> getEntries() {
+    return ImmutableMap.copyOf(entries);
+  }
+
+  /**
+   * Obtains whether the Central Directory contains any files with Zip64 file extensions.
+   *
+   * <p>At the present time, files in the Zip64 format are not supported, so this method returns
+   * false.
+   *
+   * @return false, as Zip64 formatted files are not supported
+   */
+  boolean containsZip64Files() {
+    return false;
+  }
+
+  /**
+   * Obtains the byte representation of the central directory.
+   *
+   * @return a byte array containing the whole central directory
+   * @throws IOException failed to write the byte array
+   */
+  byte[] toBytes() throws IOException {
+    return bytesSupplier.get();
+  }
+
+  /**
+   * Computes the byte representation of the central directory.
+   *
+   * @return a byte array containing the whole central directory
+   * @throws UncheckedIOException failed to write the byte array
+   */
+  private byte[] computeByteRepresentation() {
+
+    List<StoredEntry> sorted = Lists.newArrayList(entries.values());
+    Collections.sort(sorted, StoredEntry.COMPARE_BY_NAME);
+
+    CentralDirectoryHeader[] cdhs = new CentralDirectoryHeader[entries.size()];
+    CentralDirectoryHeaderCompressInfo[] compressInfos =
+        new CentralDirectoryHeaderCompressInfo[entries.size()];
+    byte[][] encodedFileNames = new byte[entries.size()][];
+    byte[][] extraFields = new byte[entries.size()][];
+    byte[][] comments = new byte[entries.size()][];
+
+    try {
+      /*
+       * First collect all the data and compute the total size of the central directory.
+       */
+      int idx = 0;
+      int total = 0;
+      for (StoredEntry entry : sorted) {
+        cdhs[idx] = entry.getCentralDirectoryHeader();
+        compressInfos[idx] = cdhs[idx].getCompressionInfoWithWait();
+        encodedFileNames[idx] = cdhs[idx].getEncodedFileName();
+        extraFields[idx] = new byte[cdhs[idx].getExtraField().size()];
+        cdhs[idx].getExtraField().write(ByteBuffer.wrap(extraFields[idx]));
+        comments[idx] = cdhs[idx].getComment();
+
+        total +=
+            F_OFFSET.endOffset()
+                + encodedFileNames[idx].length
+                + extraFields[idx].length
+                + comments[idx].length;
+        idx++;
+      }
+
+      ByteBuffer out = ByteBuffer.allocate(total);
+
+      for (idx = 0; idx < entries.size(); idx++) {
+        F_SIGNATURE.write(out);
+        F_MADE_BY.write(out, cdhs[idx].getMadeBy());
+        F_VERSION_EXTRACT.write(out, compressInfos[idx].getVersionExtract());
+        F_GP_BIT.write(out, cdhs[idx].getGpBit().getValue());
+        F_METHOD.write(out, compressInfos[idx].getMethod().methodCode);
+
+        if (file.areTimestampsIgnored()) {
+          F_LAST_MOD_TIME.write(out, 0);
+          F_LAST_MOD_DATE.write(out, 0);
+        } else {
+          F_LAST_MOD_TIME.write(out, cdhs[idx].getLastModTime());
+          F_LAST_MOD_DATE.write(out, cdhs[idx].getLastModDate());
+        }
+
+        F_CRC32.write(out, cdhs[idx].getCrc32());
+        F_COMPRESSED_SIZE.write(out, compressInfos[idx].getCompressedSize());
+        F_UNCOMPRESSED_SIZE.write(out, cdhs[idx].getUncompressedSize());
+
+        F_FILE_NAME_LENGTH.write(out, cdhs[idx].getEncodedFileName().length);
+        F_EXTRA_FIELD_LENGTH.write(out, cdhs[idx].getExtraField().size());
+        F_COMMENT_LENGTH.write(out, cdhs[idx].getComment().length);
+        F_DISK_NUMBER_START.write(out);
+        F_INTERNAL_ATTRIBUTES.write(out, cdhs[idx].getInternalAttributes());
+        F_EXTERNAL_ATTRIBUTES.write(out, cdhs[idx].getExternalAttributes());
+        F_OFFSET.write(out, cdhs[idx].getOffset());
+
+        out.put(encodedFileNames[idx]);
+        out.put(extraFields[idx]);
+        out.put(comments[idx]);
+      }
+
+      return out.array();
+    } catch (IOException e) {
+      throw new IOExceptionWrapper(e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/CentralDirectoryHeader.java

@@ -0,0 +1,414 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.MsDosDateTimeUtils;
+import com.google.common.base.Verify;
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.Future;
+
+/**
+ * The Central Directory Header contains information about files stored in the zip. Instances of
+ * this class contain information for files that already are in the zip and, for which the data was
+ * read from the Central Directory. But some instances of this class are used for new files. Because
+ * instances of this class can refer to files not yet on the zip, some of the fields may not be
+ * filled in, or may be filled in with default values.
+ *
+ * <p>Because compression decision is done lazily, some data is stored with futures.
+ */
+public class CentralDirectoryHeader implements Cloneable {
+
+  /**
+   * Default "version made by" field: upper byte needs to be 0 to set to MS-DOS compatibility. Lower
+   * byte can be anything, really. We use 18 because aapt uses 17 :)
+   */
+  private static final int DEFAULT_VERSION_MADE_BY = 0x0018;
+
+  private static final byte[] EMPTY_COMMENT = new byte[0];
+
+  /** Name of the file. */
+  private final String name;
+
+  /** CRC32 of the data. 0 if not yet computed. */
+  private long crc32;
+
+  /** Size of the file uncompressed. 0 if the file has no data. */
+  private long uncompressedSize;
+
+  /** Code of the program that made the zip. We actually don't care about this. */
+  private long madeBy;
+
+  /** General-purpose bit flag. */
+  private GPFlags gpBit;
+
+  /** Last modification time in MS-DOS format (see {@link MsDosDateTimeUtils#packTime(long)}). */
+  private long lastModTime;
+
+  /** Last modification time in MS-DOS format (see {@link MsDosDateTimeUtils#packDate(long)}). */
+  private long lastModDate;
+
+  /**
+   * Extra data field contents. This field follows a specific structure according to the
+   * specification.
+   */
+  private ExtraField extraField;
+
+  /** File comment. */
+  private byte[] comment;
+
+  /** File internal attributes. */
+  private long internalAttributes;
+
+  /** File external attributes. */
+  private long externalAttributes;
+
+  /**
+   * Offset in the file where the data is located. This will be -1 if the header corresponds to a
+   * new file that is not yet written in the zip and, therefore, has no written data.
+   */
+  private long offset;
+
+  /** Encoded file name. */
+  private byte[] encodedFileName;
+
+  /** Compress information that may not have been computed yet due to lazy compression. */
+  private final Future<CentralDirectoryHeaderCompressInfo> compressInfo;
+
+  /** The file this header belongs to. */
+  private final ZFile file;
+
+  /**
+   * Creates data for a file.
+   *
+   * @param name the file name
+   * @param encodedFileName the encoded file name, this array will be owned by the header
+   * @param uncompressedSize the uncompressed file size
+   * @param compressInfo computation that defines the compression information
+   * @param flags flags used in the entry
+   * @param zFile the file this header belongs to
+   */
+  CentralDirectoryHeader(
+      String name,
+      byte[] encodedFileName,
+      long uncompressedSize,
+      Future<CentralDirectoryHeaderCompressInfo> compressInfo,
+      GPFlags flags,
+      ZFile zFile) {
+    this(
+        name,
+        encodedFileName,
+        uncompressedSize,
+        compressInfo,
+        flags,
+        zFile,
+        MsDosDateTimeUtils.packCurrentTime(),
+        MsDosDateTimeUtils.packCurrentDate());
+  }
+
+  CentralDirectoryHeader(
+      String name,
+      byte[] encodedFileName,
+      long uncompressedSize,
+      Future<CentralDirectoryHeaderCompressInfo> compressInfo,
+      GPFlags flags,
+      ZFile zFile,
+      long currentTime,
+      long currentDate) {
+    this.name = name;
+    this.uncompressedSize = uncompressedSize;
+    crc32 = 0;
+
+    /*
+     * Set sensible defaults for the rest.
+     */
+    madeBy = DEFAULT_VERSION_MADE_BY;
+
+    gpBit = flags;
+    lastModTime = currentTime;
+    lastModDate = currentDate;
+    extraField = ExtraField.EMPTY;
+    comment = EMPTY_COMMENT;
+    internalAttributes = 0;
+    externalAttributes = 0;
+    offset = -1;
+    this.encodedFileName = encodedFileName;
+    this.compressInfo = compressInfo;
+    file = zFile;
+  }
+
+  /**
+   * Obtains the name of the file.
+   *
+   * @return the name
+   */
+  public String getName() {
+    return name;
+  }
+
+  /**
+   * Obtains the size of the uncompressed file.
+   *
+   * @return the size of the file
+   */
+  public long getUncompressedSize() {
+    return uncompressedSize;
+  }
+
+  /**
+   * Obtains the CRC32 of the data.
+   *
+   * @return the CRC32, 0 if not yet computed
+   */
+  public long getCrc32() {
+    return crc32;
+  }
+
+  /**
+   * Sets the CRC32 of the data.
+   *
+   * @param crc32 the CRC 32
+   */
+  void setCrc32(long crc32) {
+    this.crc32 = crc32;
+  }
+
+  /**
+   * Obtains the code of the program that made the zip.
+   *
+   * @return the code
+   */
+  public long getMadeBy() {
+    return madeBy;
+  }
+
+  /**
+   * Sets the code of the progtram that made the zip.
+   *
+   * @param madeBy the code
+   */
+  void setMadeBy(long madeBy) {
+    this.madeBy = madeBy;
+  }
+
+  /**
+   * Obtains the general-purpose bit flag.
+   *
+   * @return the bit flag
+   */
+  public GPFlags getGpBit() {
+    return gpBit;
+  }
+
+  /**
+   * Obtains the last modification time of the entry.
+   *
+   * @return the last modification time in MS-DOS format (see {@link
+   *     MsDosDateTimeUtils#packTime(long)})
+   */
+  public long getLastModTime() {
+    return lastModTime;
+  }
+
+  /**
+   * Sets the last modification time of the entry.
+   *
+   * @param lastModTime the last modification time in MS-DOS format (see {@link
+   *     MsDosDateTimeUtils#packTime(long)})
+   */
+  void setLastModTime(long lastModTime) {
+    this.lastModTime = lastModTime;
+  }
+
+  /**
+   * Obtains the last modification date of the entry.
+   *
+   * @return the last modification date in MS-DOS format (see {@link
+   *     MsDosDateTimeUtils#packDate(long)})
+   */
+  public long getLastModDate() {
+    return lastModDate;
+  }
+
+  /**
+   * Sets the last modification date of the entry.
+   *
+   * @param lastModDate the last modification date in MS-DOS format (see {@link
+   *     MsDosDateTimeUtils#packDate(long)})
+   */
+  void setLastModDate(long lastModDate) {
+    this.lastModDate = lastModDate;
+  }
+
+  /**
+   * Obtains the data in the extra field.
+   *
+   * @return the data (returns an empty array if there is none)
+   */
+  public ExtraField getExtraField() {
+    return extraField;
+  }
+
+  /**
+   * Sets the data in the extra field.
+   *
+   * @param extraField the data to set
+   */
+  public void setExtraField(ExtraField extraField) {
+    setExtraFieldNoNotify(extraField);
+    file.centralDirectoryChanged();
+  }
+
+  /**
+   * Sets the data in the extra field, but does not notify {@link ZFile}. This method is invoked
+   * when the {@link ZFile} knows the extra field is being set.
+   *
+   * @param extraField the data to set
+   */
+  void setExtraFieldNoNotify(ExtraField extraField) {
+    this.extraField = extraField;
+  }
+
+  /**
+   * Obtains the entry's comment.
+   *
+   * @return the comment (returns an empty array if there is no comment)
+   */
+  public byte[] getComment() {
+    return comment;
+  }
+
+  /**
+   * Sets the entry's comment.
+   *
+   * @param comment the comment
+   */
+  void setComment(byte[] comment) {
+    this.comment = comment;
+  }
+
+  /**
+   * Obtains the entry's internal attributes.
+   *
+   * @return the entry's internal attributes
+   */
+  public long getInternalAttributes() {
+    return internalAttributes;
+  }
+
+  /**
+   * Sets the entry's internal attributes.
+   *
+   * @param internalAttributes the entry's internal attributes
+   */
+  void setInternalAttributes(long internalAttributes) {
+    this.internalAttributes = internalAttributes;
+  }
+
+  /**
+   * Obtains the entry's external attributes.
+   *
+   * @return the entry's external attributes
+   */
+  public long getExternalAttributes() {
+    return externalAttributes;
+  }
+
+  /**
+   * Sets the entry's external attributes.
+   *
+   * @param externalAttributes the entry's external attributes
+   */
+  void setExternalAttributes(long externalAttributes) {
+    this.externalAttributes = externalAttributes;
+  }
+
+  /**
+   * Obtains the offset in the zip file where this entry's data is.
+   *
+   * @return the offset or {@code -1} if the file has no data in the zip and, therefore, data is
+   *     stored in memory
+   */
+  public long getOffset() {
+    return offset;
+  }
+
+  /**
+   * Sets the offset in the zip file where this entry's data is.
+   *
+   * @param offset the offset or {@code -1} if the file is new and has no data in the zip yet
+   */
+  void setOffset(long offset) {
+    this.offset = offset;
+  }
+
+  /**
+   * Obtains the encoded file name.
+   *
+   * @return the encoded file name
+   */
+  public byte[] getEncodedFileName() {
+    return encodedFileName;
+  }
+
+  /** Resets the deferred CRC flag in the GP flags. */
+  void resetDeferredCrc() {
+    /*
+     * We actually create a new set of flags. Since the only information we care about is the
+     * UTF-8 encoding, we'll just create a brand new object.
+     */
+    gpBit = GPFlags.make(gpBit.isUtf8FileName());
+  }
+
+  @Override
+  protected CentralDirectoryHeader clone() throws CloneNotSupportedException {
+    CentralDirectoryHeader cdr = (CentralDirectoryHeader) super.clone();
+    cdr.extraField = extraField;
+    cdr.comment = Arrays.copyOf(comment, comment.length);
+    cdr.encodedFileName = Arrays.copyOf(encodedFileName, encodedFileName.length);
+    return cdr;
+  }
+
+  /**
+   * Obtains the future with the compression information.
+   *
+   * @return the information
+   */
+  public Future<CentralDirectoryHeaderCompressInfo> getCompressionInfo() {
+    return compressInfo;
+  }
+
+  /**
+   * Equivalent to {@code getCompressionInfo().get()} but masking the possible exceptions and
+   * guaranteeing non-{@code null} return.
+   *
+   * @return the result of the future
+   * @throws IOException failed to get the information
+   */
+  public CentralDirectoryHeaderCompressInfo getCompressionInfoWithWait() throws IOException {
+    try {
+      CentralDirectoryHeaderCompressInfo info = getCompressionInfo().get();
+      Verify.verifyNotNull(info, "info == null");
+      return info;
+    } catch (InterruptedException e) {
+      throw new IOException("Interrupted while waiting for compression information.", e);
+    } catch (ExecutionException e) {
+      throw new IOException("Execution of compression failed.", e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/CentralDirectoryHeaderCompressInfo.java

@@ -0,0 +1,110 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+
+/**
+ * Information stored in the {@link CentralDirectoryHeader} that is related to compression and may
+ * need to be computed lazily.
+ */
+public class CentralDirectoryHeaderCompressInfo {
+
+  /** Version of zip file that only supports stored files. */
+  public static final long VERSION_WITH_STORE_FILES_ONLY = 10L;
+
+  /** Version of zip file that only supports directories and deflated files. */
+  public static final long VERSION_WITH_DIRECTORIES_AND_DEFLATE = 20L;
+
+  /** Version of zip file that only supports ZIP64 format extensions */
+  public static final long VERSION_WITH_ZIP64_EXTENSIONS = 45L;
+
+  /** Version of zip file that uses central file encryption and version 2 of the Zip64 EOCD */
+  public static final long VERSION_WITH_CENTRAL_FILE_ENCRYPTION = 62L;
+
+  /** The compression method. */
+  private final CompressionMethod method;
+
+  /** Size of the file compressed. 0 if the file has no data. */
+  private final long compressedSize;
+
+  /** Version needed to extract the zip. */
+  private final long versionExtract;
+
+  /**
+   * Creates new compression information for the central directory header.
+   *
+   * @param method the compression method
+   * @param compressedSize the compressed size
+   * @param versionToExtract minimum version to extract (typically {@link
+   *     #VERSION_WITH_STORE_FILES_ONLY} or {@link #VERSION_WITH_DIRECTORIES_AND_DEFLATE})
+   */
+  public CentralDirectoryHeaderCompressInfo(
+      CompressionMethod method, long compressedSize, long versionToExtract) {
+    this.method = method;
+    this.compressedSize = compressedSize;
+    versionExtract = versionToExtract;
+  }
+
+  /**
+   * Creates new compression information for the central directory header.
+   *
+   * @param header the header this information relates to
+   * @param method the compression method
+   * @param compressedSize the compressed size
+   */
+  public CentralDirectoryHeaderCompressInfo(
+      CentralDirectoryHeader header, CompressionMethod method, long compressedSize) {
+    this.method = method;
+    this.compressedSize = compressedSize;
+
+   if (header.getName().endsWith("/") || method == CompressionMethod.DEFLATE) {
+      /*
+       * Directories and compressed files only in version 2.0.
+       */
+      versionExtract = VERSION_WITH_DIRECTORIES_AND_DEFLATE;
+    } else {
+      versionExtract = VERSION_WITH_STORE_FILES_ONLY;
+    }
+  }
+
+  /**
+   * Obtains the compression data size.
+   *
+   * @return the compressed data size
+   */
+  public long getCompressedSize() {
+    return compressedSize;
+  }
+
+  /**
+   * Obtains the compression method.
+   *
+   * @return the compression method
+   */
+  public CompressionMethod getMethod() {
+    return method;
+  }
+
+  /**
+   * Obtains the minimum version for extract.
+   *
+   * @return the minimum version
+   */
+  long getVersionExtract() {
+    return versionExtract;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/CompressionMethod.java

@@ -0,0 +1,57 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import javax.annotation.Nullable;
+
+/** Enumeration with all known compression methods. */
+public enum CompressionMethod {
+  /** STORE method: data is stored without any compression. */
+  STORE(0),
+
+  /** DEFLATE method: data is stored compressed using the DEFLATE algorithm. */
+  DEFLATE(8);
+
+  /** Code, within the zip file, that identifies this compression method. */
+  int methodCode;
+
+  /**
+   * Creates a new compression method.
+   *
+   * @param methodCode the code used in the zip file that identifies the compression method
+   */
+  CompressionMethod(int methodCode) {
+    this.methodCode = methodCode;
+  }
+
+  /**
+   * Obtains the compression method that corresponds to the provided code.
+   *
+   * @param code the code
+   * @return the method or {@code null} if no method has the provided code
+   */
+  @Nullable
+  static CompressionMethod fromCode(long code) {
+    for (CompressionMethod method : values()) {
+      if (method.methodCode == code) {
+        return method;
+      }
+    }
+
+    return null;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/CompressionResult.java

@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+
+/** Result of compressing data. */
+public class CompressionResult {
+
+  /** The compression method used. */
+  private final CompressionMethod compressionMethod;
+
+  /** The resulting data. */
+  private final CloseableByteSource source;
+
+  /**
+   * Size of the compressed source. Kept because {@code source.size()} can throw {@code
+   * IOException}.
+   */
+  private final long mSize;
+
+  /**
+   * Creates a new compression result.
+   *
+   * @param source the data source
+   * @param method the compression method
+   */
+  public CompressionResult(CloseableByteSource source, CompressionMethod method, long size) {
+    compressionMethod = method;
+    this.source = source;
+    mSize = size;
+  }
+
+  /**
+   * Obtains the compression method.
+   *
+   * @return the compression method
+   */
+  public CompressionMethod getCompressionMethod() {
+    return compressionMethod;
+  }
+
+  /**
+   * Obtains the compressed data.
+   *
+   * @return the data, the resulting array should not be modified
+   */
+  public CloseableByteSource getSource() {
+    return source;
+  }
+
+  /**
+   * Obtains the size of the compression result.
+   *
+   * @return the size
+   */
+  public long getSize() {
+    return mSize;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/Compressor.java

@@ -0,0 +1,38 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.util.concurrent.ListenableFuture;
+
+/**
+ * A compressor is capable of, well, compressing data. Data is read from an {@code ByteSource}.
+ * Compressors are asynchronous: compressing results in a {@code ListenableFuture} that will contain
+ * the compression result.
+ */
+public interface Compressor {
+
+  /**
+   * Compresses an entry source.
+   *
+   * @param source the source to compress
+   * @param storage a byte storage from where the compressor can obtain byte sources to work
+   * @return a future that will eventually contain the compression result
+   */
+  ListenableFuture<CompressionResult> compress(CloseableByteSource source, ByteStorage storage);
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/DataDescriptorType.java

@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/**
+ * Type of data descriptor that an entry has. Data descriptors are used if the CRC and sizing data
+ * is not known when the data is being written and cannot be placed in the file's local header. In
+ * those cases, after the file data itself, a data descriptor is placed after the entry's contents.
+ *
+ * <p>While the zip specification says the data descriptor should be used but it is optional. We
+ * record also whether the data descriptor contained the 4-byte signature at the start of the block
+ * or not.
+ */
+public enum DataDescriptorType {
+  /** The entry has no data descriptor. */
+  NO_DATA_DESCRIPTOR(0),
+
+  /** The entry has a data descriptor that does not contain a signature. */
+  DATA_DESCRIPTOR_WITHOUT_SIGNATURE(12),
+
+  /** The entry has a data descriptor that contains a signature. */
+  DATA_DESCRIPTOR_WITH_SIGNATURE(16);
+
+  /** The number of bytes the data descriptor spans. */
+  public int size;
+
+  /**
+   * Creates a new data descriptor.
+   *
+   * @param size the number of bytes the data descriptor spans
+   */
+  DataDescriptorType(int size) {
+    this.size = size;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/EncodeUtils.java

@@ -0,0 +1,136 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import static java.nio.charset.StandardCharsets.US_ASCII;
+import static java.nio.charset.StandardCharsets.UTF_8;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.charset.CharacterCodingException;
+import java.nio.charset.Charset;
+import java.nio.charset.CodingErrorAction;
+
+/** Utilities to encode and decode file names in zips. */
+public class EncodeUtils {
+
+  /** Utility class: no constructor. */
+  private EncodeUtils() {
+    /*
+     * Nothing to do.
+     */
+  }
+
+  /**
+   * Decodes a file name.
+   *
+   * @param bytes the raw data buffer to read from
+   * @param length the number of bytes in the raw data buffer containing the string to decode
+   * @param flags the zip entry flags
+   * @return the decode file name
+   */
+  public static String decode(ByteBuffer bytes, int length, GPFlags flags) throws IOException {
+    if (bytes.remaining() < length) {
+      throw new IOException(
+          "Only "
+              + bytes.remaining()
+              + " bytes exist in the buffer, but "
+              + "length is "
+              + length
+              + ".");
+    }
+
+    byte[] stringBytes = new byte[length];
+    bytes.get(stringBytes);
+    return decode(stringBytes, flags);
+  }
+
+  /**
+   * Decodes a file name.
+   *
+   * @param data the raw data
+   * @param flags the zip entry flags
+   * @return the decode file name
+   */
+  public static String decode(byte[] data, GPFlags flags) {
+    return decode(data, flagsCharset(flags));
+  }
+
+  /**
+   * Decodes a file name.
+   *
+   * @param data the raw data
+   * @param charset the charset to use
+   * @return the decode file name
+   */
+  private static String decode(byte[] data, Charset charset) {
+    try {
+      return charset
+          .newDecoder()
+          .onMalformedInput(CodingErrorAction.REPORT)
+          .decode(ByteBuffer.wrap(data))
+          .toString();
+    } catch (CharacterCodingException e) {
+      // If we're trying to decode ASCII, try UTF-8. Otherwise, revert to the default
+      // behavior (usually replacing invalid characters).
+      if (charset.equals(US_ASCII)) {
+        return decode(data, UTF_8);
+      } else {
+        return charset.decode(ByteBuffer.wrap(data)).toString();
+      }
+    }
+  }
+
+  /**
+   * Encodes a file name.
+   *
+   * @param name the name to encode
+   * @param flags the zip entry flags
+   * @return the encoded file name
+   */
+  public static byte[] encode(String name, GPFlags flags) {
+    Charset charset = flagsCharset(flags);
+    ByteBuffer bytes = charset.encode(name);
+    byte[] result = new byte[bytes.remaining()];
+    bytes.get(result);
+    return result;
+  }
+
+  /**
+   * Obtains the charset to encode and decode zip entries, given a set of flags.
+   *
+   * @param flags the flags
+   * @return the charset to use
+   */
+  private static Charset flagsCharset(GPFlags flags) {
+    if (flags.isUtf8FileName()) {
+      return UTF_8;
+    } else {
+      return US_ASCII;
+    }
+  }
+
+  /**
+   * Checks if some text may be encoded using ASCII.
+   *
+   * @param text the text to check
+   * @return can it be encoded using ASCII?
+   */
+  public static boolean canAsciiEncode(String text) {
+    return US_ASCII.newEncoder().canEncode(text);
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/Eocd.java

@@ -0,0 +1,277 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.utils.CachedSupplier;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import com.google.common.primitives.Ints;
+import java.io.IOException;
+import java.io.UncheckedIOException;
+import java.nio.ByteBuffer;
+
+/** End Of Central Directory record in a zip file. */
+class Eocd {
+
+  /** Max total records that can be specified by the standard EOCD. */
+  static final long MAX_TOTAL_RECORDS = 0xFFFFL;
+
+  /** Max size of the Central Directory that can be specified by the standard EOCD. */
+  static final long MAX_CD_SIZE = 0xFFFFFFFFL;
+
+  /** Max offset of the Central Directory that can be specified by the standard EOCD. */
+  static final long MAX_CD_OFFSET = 0xFFFFFFFFL;
+
+  /** Field in the record: the record signature, fixed at this value by the specification. */
+  private static final ZipField.F4 F_SIGNATURE = new ZipField.F4(0, 0x06054b50, "EOCD signature");
+
+  /**
+   * Field in the record: the number of the disk where the EOCD is located. It has to be zero
+   * because we do not support multi-file archives.
+   */
+  private static final ZipField.F2 F_NUMBER_OF_DISK =
+      new ZipField.F2(F_SIGNATURE.endOffset(), 0, "Number of this disk");
+
+  /**
+   * Field in the record: the number of the disk where the Central Directory starts. Has to be zero
+   * because we do not support multi-file archives.
+   */
+  private static final ZipField.F2 F_DISK_CD_START =
+      new ZipField.F2(F_NUMBER_OF_DISK.endOffset(), 0, "Disk where CD starts");
+
+  /**
+   * Field in the record: the number of entries in the Central Directory on this disk. Because we do
+   * not support multi-file archives, this is the same as {@link #F_RECORDS_TOTAL}.
+   */
+  private static final ZipField.F2 F_RECORDS_DISK =
+      new ZipField.F2(
+          F_DISK_CD_START.endOffset(), "Record on disk count", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the record: the total number of entries in the Central Directory. This value will be
+   * {@link #MAX_TOTAL_RECORDS} if the file is in the Zip64 format, and the Central Directory holds
+   * at least {@link #MAX_TOTAL_RECORDS} entries.
+   */
+  private static final ZipField.F2 F_RECORDS_TOTAL =
+      new ZipField.F2(
+          F_RECORDS_DISK.endOffset(),
+          "Total records",
+          new ZipFieldInvariantNonNegative(),
+          new ZipFieldInvariantMaxValue(Integer.MAX_VALUE));
+
+  /**
+   * Field in the record: number of bytes of the Central Directory. This is not private because it
+   * is required in unit tests. This value will be {@link #MAX_CD_SIZE} if the file is in the Zip64
+   * format, and the Central Directory is at least {@link #MAX_CD_SIZE} bytes.
+   */
+  @VisibleForTesting
+  static final ZipField.F4 F_CD_SIZE =
+      new ZipField.F4(
+          F_RECORDS_TOTAL.endOffset(), "Directory size", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the record: offset, from the archive start, where the Central Directory starts. This
+   * is not private because it is required in unit tests. This value will be {@link #MAX_CD_OFFSET}
+   * if the file is in the Zip64 format, and the Central Directory is at least
+   * {@link #MAX_CD_OFFSET} bytes.
+   */
+  @VisibleForTesting
+  static final ZipField.F4 F_CD_OFFSET =
+      new ZipField.F4(
+          F_CD_SIZE.endOffset(), "Directory offset", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the record: number of bytes of the file comment (located at the end of the EOCD
+   * record).
+   */
+  private static final ZipField.F2 F_COMMENT_SIZE =
+      new ZipField.F2(
+          F_CD_OFFSET.endOffset(), "File comment size", new ZipFieldInvariantNonNegative());
+
+  /** Number of entries in the central directory. */
+  private final long totalRecords;
+
+  /** Offset from the beginning of the archive where the Central Directory is located. */
+  private final long directoryOffset;
+
+  /** Number of bytes of the Central Directory. */
+  private final long directorySize;
+
+  /** Contents of the EOCD comment. */
+  private final byte[] comment;
+
+  /** Supplier of the byte representation of the EOCD. */
+  private final CachedSupplier<byte[]> byteSupplier;
+
+  /**
+   * Creates a new EOCD, reading it from a byte source. This method will parse the byte source and
+   * obtain the EOCD. It will check that the byte source starts with the EOCD signature.
+   *
+   * @param bytes the byte buffer with the EOCD data; when this method finishes, the byte buffer's
+   *     position will have moved to the end of the EOCD
+   * @throws IOException failed to read information or the EOCD data is corrupt or invalid
+   */
+  Eocd(ByteBuffer bytes) throws IOException {
+
+    /*
+     * Read the EOCD record.
+     */
+    F_SIGNATURE.verify(bytes);
+    F_NUMBER_OF_DISK.verify(bytes);
+    F_DISK_CD_START.verify(bytes);
+    long totalRecords1 = F_RECORDS_DISK.read(bytes);
+    long totalRecords2 = F_RECORDS_TOTAL.read(bytes);
+    long directorySize = F_CD_SIZE.read(bytes);
+    long directoryOffset = F_CD_OFFSET.read(bytes);
+    int commentSize = Ints.checkedCast(F_COMMENT_SIZE.read(bytes));
+
+    /*
+     * Some sanity checks.
+     */
+    if (totalRecords1 != totalRecords2) {
+      throw new IOException(
+          "Zip states records split in multiple disks, which is not " + "supported.");
+    }
+
+    Verify.verify(totalRecords1 <= Integer.MAX_VALUE);
+
+    totalRecords = Ints.checkedCast(totalRecords1);
+    this.directorySize = directorySize;
+    this.directoryOffset = directoryOffset;
+
+    if (bytes.remaining() < commentSize) {
+      throw new IOException(
+          "Corrupt EOCD record: not enough data for comment (comment "
+              + "size is "
+              + commentSize
+              + ").");
+    }
+
+    comment = new byte[commentSize];
+    bytes.get(comment);
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * Creates a new EOCD. This is used when generating an EOCD for an Central Directory that has just
+   * been generated. The EOCD will be generated without any comment.
+   *
+   * @param totalRecords total number of records in the directory
+   * @param directoryOffset offset, since beginning of archive, where the Central Directory is
+   *     located
+   * @param directorySize number of bytes of the Central Directory
+   * @param comment the EOCD comment
+   */
+  Eocd(long totalRecords, long directoryOffset, long directorySize, byte[] comment) {
+    Preconditions.checkArgument(totalRecords >= 0, "totalRecords < 0");
+    Preconditions.checkArgument(directoryOffset >= 0, "directoryOffset < 0");
+    Preconditions.checkArgument(directorySize >= 0, "directorySize < 0");
+
+    this.totalRecords = totalRecords;
+    this.directoryOffset = directoryOffset;
+    this.directorySize = directorySize;
+    this.comment = comment;
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * Obtains the number of records in the Central Directory.
+   *
+   * @return the number of records
+   */
+  long getTotalRecords() {
+    return totalRecords;
+  }
+
+  /**
+   * Obtains the offset since the beginning of the zip archive where the Central Directory is
+   * located.
+   *
+   * @return the offset where the Central Directory is located
+   */
+  long getDirectoryOffset() {
+    return directoryOffset;
+  }
+
+  /**
+   * Obtains the size of the Central Directory.
+   *
+   * @return the number of bytes that make up the Central Directory
+   */
+  long getDirectorySize() {
+    return directorySize;
+  }
+
+  /**
+   * Obtains the size of the EOCD.
+   *
+   * @return the size, in bytes, of the EOCD
+   */
+  long getEocdSize() {
+    return (long) F_COMMENT_SIZE.endOffset() + comment.length;
+  }
+
+  /**
+   * Generates the EOCD data.
+   *
+   * @return a byte representation of the EOCD that has exactly {@link #getEocdSize()} bytes
+   * @throws IOException failed to generate the EOCD data
+   */
+  byte[] toBytes() throws IOException {
+    return byteSupplier.get();
+  }
+
+  /**
+   * Obtains the comment in the EOCD.
+   *
+   * @return the comment exactly as it is represented in the file (no encoding conversion is
+   * done)
+   */
+  byte[] getComment() {
+    byte[] commentCopy = new byte[comment.length];
+    System.arraycopy(comment, 0, commentCopy, 0, comment.length);
+    return commentCopy;
+  }
+
+  /**
+   * Computes the byte representation of the EOCD.
+   *
+   * @return a byte representation of the EOCD that has exactly {@link #getEocdSize()} bytes
+   * @throws UncheckedIOException failed to generate the EOCD data
+   */
+  private byte[] computeByteRepresentation() {
+    ByteBuffer out = ByteBuffer.allocate(F_COMMENT_SIZE.endOffset() + comment.length);
+
+    try {
+      F_SIGNATURE.write(out);
+      F_NUMBER_OF_DISK.write(out);
+      F_DISK_CD_START.write(out);
+      F_RECORDS_DISK.write(out, totalRecords);
+      F_RECORDS_TOTAL.write(out, totalRecords);
+      F_CD_SIZE.write(out, directorySize);
+      F_CD_OFFSET.write(out, directoryOffset);
+      F_COMMENT_SIZE.write(out, comment.length);
+      out.put(comment);
+
+      return out.array();
+    } catch (IOException e) {
+      throw new IOExceptionWrapper(e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/EocdGroup.java

@@ -0,0 +1,696 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.android.tools.build.apkzlib.zip.utils.LittleEndianUtils;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import com.google.common.primitives.Ints;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import javax.annotation.Nullable;
+
+/**
+ * The collection of all data stored in all End of Central Directory records in the zip file. The
+ * {@code EOCDGroup} is meant to collect and manage all the information about the {@link Eocd},
+ * {@link Zip64EocdLocator}, and the {@link Zip64Eocd} in one place.
+ */
+public class EocdGroup {
+
+  /** Minimum size the EOCD can have. */
+  private static final int MIN_EOCD_SIZE = 22;
+
+  /** Maximum size for the EOCD. */
+  private static final int MAX_EOCD_COMMENT_SIZE = 65535;
+
+  /** How many bytes to look back from the end of the file to look for the EOCD signature. */
+  private static final int LAST_BYTES_TO_READ = MIN_EOCD_SIZE + MAX_EOCD_COMMENT_SIZE;
+
+  /** Signature of the Zip64 EOCD locator record. */
+  private static final int ZIP64_EOCD_LOCATOR_SIGNATURE = 0x07064b50;
+
+  /** Signature of the EOCD record. */
+  private static final long EOCD_SIGNATURE = 0x06054b50;
+
+  /**
+   * The EOCD entry. Will be {@code null} if there is no EOCD (because the zip is new) or the one
+   * that exists on disk is no longer valid (because the zip has been changed).
+   *
+   * <p>If the EOCD is deleted because the zip has been changed and the old EOCD was no longer
+   * valid, then {@link #eocdComment} will contain the comment saved from the EOCD.
+   */
+  @Nullable
+  private FileUseMapEntry<Eocd> eocdEntry;
+
+  /**
+   * The EOCD locator entry. Will be {@code null} if there is no EOCD (because the zip is new),
+   * the EOCD on disk is no longer valid (because the zip has been changed), or the zip file is not
+   * in Zip64 format (There are no values in the EOCD that overflow or any files with Zip64
+   * extended information.)
+   *
+   * <p> If this value is {@code nonnull} then the EOCD exists and is in Zip64 format (<i>i.e.</i>
+   * both {@link #eocdEntry} and {@link #eocd64Entry} will be {@code nonnull}).
+   */
+  @Nullable
+  private FileUseMapEntry<Zip64EocdLocator> eocd64Locator;
+
+  /**
+   * The Zip64 EOCD entry. Will be {@code null} if there is no EOCD (because the zip is new),
+   * the EOCD on disk is no longer valid (because the zip has been changed), or the zip file is not
+   * in Zip64 format (There are no values in the EOCD that overflow or any files with Zip64
+   * extended information.)
+   *
+   * <p> If this value is {@code nonnull} then the EOCD exists and is in Zip64 format (<i>i.e.</i>
+   * both {@link #eocdEntry} and {@link #eocd64Locator} will be {@code nonnull}).
+   */
+  @Nullable
+  private FileUseMapEntry<Zip64Eocd> eocd64Entry;
+
+  /**
+   * This field contains the comment in the zip's EOCD if there is no in-memory EOCD structure. This
+   * may happen, for example, if the zip has been changed and the Central Directory and EOCD have
+   * been deleted (in-memory). In that case, this field will save the comment to place on the EOCD
+   * once it is created.
+   *
+   * <p>This field will only be non-{@code null} if there is no in-memory EOCD structure
+   * (<i>i.e.</i>, {@link #eocdEntry} is {@code null}, If there is an {@link #eocdEntry}, then the
+   * comment will be there instead of being in this field.
+   */
+  @Nullable
+  private byte[] eocdComment;
+
+  /**
+   * This field contains the extensible data sector in the zip's Zip64 EOCD if there is no EOCD
+   * in-memory. This may happen if the zip has been modified and the Central Directory and EOCD have
+   * been deleted (in-memory). In that case, this field will save the data sector to place in the
+   * Zip64 EOCD once it is created.
+   *
+   * <p>This field will only be non-{@code null} if there is no in-memory EOCD structure
+   * (<i>i.e.</i>, {@link #eocdEntry} is {@code null}, If there is an {@link #eocdEntry}, then the
+   * data sector will be in the {@link #eocd64Entry} instead of being in this field.
+   */
+  @Nullable
+  private Zip64ExtensibleDataSector eocdDataSector;
+
+  /**
+   * Specifies whether the Zip64 Eocd will be in Version 2 or Version 1 format when it is
+   * constructed.
+   */
+  private boolean useVersion2Header;
+
+  /** The zip file to which this EOCD record belongs. */
+  private final ZFile file;
+
+  /** The in-memory map of the pieces of the zip-file. */
+  private final FileUseMap map;
+
+  /** The zip file's log. */
+  private final VerifyLog verifyLog;
+
+  /**
+   * Constructs an empty EOCD group, which will have no in-memory EOCD structure.
+   *
+   * @param file The zip file to which this EOCD record belongs.
+   * @param map he in-memory map of the zip file.
+   */
+  EocdGroup(ZFile file, FileUseMap map) {
+
+    eocd64Entry = null;
+    eocd64Locator = null;
+    eocdEntry = null;
+    eocdComment = new byte[0];
+    eocdDataSector = new Zip64ExtensibleDataSector();
+    this.file = file;
+    this.map = map;
+    this.verifyLog = file.getVerifyLog();
+    useVersion2Header = false;
+  }
+
+  /**
+   * Attempts to read the EOCD record into the {@link EocdGroup} from disk specified by
+   * {@link #file}. It will populate the in-memory EOCD structure (<i>i.e.</i> {@link #eocdEntry}),
+   * including the Zip64 EOCD record and locator if applicable.
+   *
+   * @param fileLength The length of the file on disk, used to help find the EOCD record.
+   * @throws IOException Failed to read the EOCD.
+   */
+  void readRecord(long fileLength) throws IOException {
+    /*
+     * Read the last part of the zip into memory. If we don't find the EOCD signature by then,
+     * the file is corrupt.
+     */
+    int lastToRead = LAST_BYTES_TO_READ;
+    if (lastToRead > fileLength) {
+      lastToRead = Ints.checkedCast(fileLength);
+    }
+
+    byte[] last = new byte[lastToRead];
+    file.directFullyRead(fileLength - lastToRead, last);
+
+    /*
+     * Start endIdx at the first possible location where the signature can be located and then
+     * move backwards. Because the EOCD must have at least MIN_EOCD size, the first byte of the
+     * signature (and first byte of the EOCD) must be located at last.length - MIN_EOCD_SIZE.
+     *
+     * Because the EOCD signature may exist in the file comment, when we find a signature we
+     * will try to read the Eocd. If we fail, we continue searching for the signature. However,
+     * we will keep the last exception in case we don't find any signature.
+     */
+    Eocd eocd = null;
+    int foundEocdSignatureIdx = -1;
+    IOException errorFindingSignature = null;
+    long eocdStart = -1;
+
+    for (int endIdx = last.length - MIN_EOCD_SIZE;
+        endIdx >= 0 && foundEocdSignatureIdx == -1;
+        endIdx--) {
+
+      ByteBuffer potentialLocator = ByteBuffer.wrap(last, endIdx, 4);
+      if (LittleEndianUtils.readUnsigned4Le(potentialLocator) == EOCD_SIGNATURE) {
+
+        /*
+         * We found a signature. Try to read the EOCD record.
+         */
+
+        foundEocdSignatureIdx = endIdx;
+        ByteBuffer eocdBytes =
+            ByteBuffer.wrap(last, foundEocdSignatureIdx, last.length - foundEocdSignatureIdx);
+
+        try {
+          eocd = new Eocd(eocdBytes);
+
+          eocdStart = fileLength - lastToRead + foundEocdSignatureIdx;
+
+          /*
+           * Make sure the EOCD takes the whole file up to the end. Log an error if it
+           * doesn't.
+           */
+          if (eocdStart + eocd.getEocdSize() != fileLength) {
+            verifyLog.log(
+                "EOCD starts at "
+                    + eocdStart
+                    + " and has "
+                    + eocd.getEocdSize()
+                    + " bytes, but file ends at "
+                    + fileLength
+                    + ".");
+          }
+        } catch (IOException e) {
+          if (errorFindingSignature != null) {
+            e.addSuppressed(errorFindingSignature);
+          }
+
+          errorFindingSignature = e;
+          foundEocdSignatureIdx = -1;
+          eocd = null;
+        }
+      }
+    }
+
+    if (foundEocdSignatureIdx == -1) {
+      throw new IOException(
+          "EOCD signature not found in the last " + lastToRead + " bytes of the file.",
+          errorFindingSignature);
+    }
+
+    Verify.verify(eocdStart >= 0);
+    eocdEntry = map.add(eocdStart, eocdStart + eocd.getEocdSize(), eocd);
+
+    /*
+     * Look for the Zip64 central directory locator. If we find it, then this file is a Zip64
+     * file and we need to read both the Zip64 EOCD locator and Zip64 EOCD
+     */
+    long zip64LocatorStart = eocdStart - Zip64EocdLocator.LOCATOR_SIZE;
+    if (zip64LocatorStart >= 0) {
+      byte[] possibleZip64Locator = new byte[Zip64EocdLocator.LOCATOR_SIZE];
+      file.directFullyRead(zip64LocatorStart, possibleZip64Locator);
+      if (LittleEndianUtils.readUnsigned4Le(ByteBuffer.wrap(possibleZip64Locator))
+          == ZIP64_EOCD_LOCATOR_SIGNATURE) {
+
+        /* found the locator. Read it into memory. */
+
+        Zip64EocdLocator locator = new Zip64EocdLocator(ByteBuffer.wrap(possibleZip64Locator));
+        eocd64Locator = map.add(
+            zip64LocatorStart, zip64LocatorStart + locator.getSize(), locator);
+
+        /* Find the size of the Zip64 EOCD by reading its size field */
+        byte[] zip64EocdSizeHolder = new byte[8];
+        file.directFullyRead(
+            locator.getZ64EocdOffset() + Zip64Eocd.SIZE_OFFSET, zip64EocdSizeHolder);
+        long zip64EocdSize =
+            LittleEndianUtils.readUnsigned8Le(ByteBuffer.wrap(zip64EocdSizeHolder))
+                + Zip64Eocd.TRUE_SIZE_DIFFERENCE;
+
+        /* read the Zip64 EOCD into memory */
+
+        byte[] zip64EocdBytes = new byte[Ints.checkedCast(zip64EocdSize)];
+        file.directFullyRead(locator.getZ64EocdOffset(), zip64EocdBytes);
+        Zip64Eocd zip64Eocd = new Zip64Eocd(ByteBuffer.wrap(zip64EocdBytes));
+        useVersion2Header =
+            zip64Eocd.getVersionToExtract()
+                >= CentralDirectoryHeaderCompressInfo.VERSION_WITH_CENTRAL_FILE_ENCRYPTION;
+
+        long zip64EocdEnd = locator.getZ64EocdOffset() + zip64EocdSize;
+        if (zip64EocdEnd != zip64LocatorStart) {
+          String msg =
+              "Zip64 EOCD record is stored in ["
+                  + locator.getZ64EocdOffset()
+                  + " - "
+                  + zip64EocdEnd
+                  + "] and EOCD starts at "
+                  + zip64LocatorStart
+                  + ".";
+
+          /*
+           * If there is an empty space between the Zip64 EOCD and the EOCD locator, we proceed
+           * logging an error. If the Zip64 EOCD ends after the start of the EOCD locator (and
+           * therefore, they overlap), throw an exception.
+           */
+          if (zip64EocdEnd > zip64LocatorStart) {
+            throw new IOException(msg);
+          } else {
+            verifyLog.log(msg);
+          }
+        }
+
+        eocd64Entry = map.add(
+            locator.getZ64EocdOffset(), zip64EocdEnd, zip64Eocd);
+      }
+    }
+
+  }
+
+  /**
+   * Computes the EOCD record from the given Central Directory entry in memory. This will populate
+   * the EOCD in-memory and possibly the Zip64 EOCD and Locator if applicable.
+   *
+   * @param directoryEntry The entry to create the EOCD record from.
+   * @param extraDirectoryOffset The offset between the last local entry and the Central Directory.
+   * This will be preserved by the EOCD if the Central Directory is empty.
+   * @throws IOException Failed to create the EOCD record.
+   */
+  void computeRecord(
+      @Nullable FileUseMapEntry<CentralDirectory> directoryEntry,
+      long extraDirectoryOffset) throws IOException {
+
+    long dirStart;
+    long dirSize;
+    long dirNumEntries;
+
+    if (directoryEntry != null) {
+      dirStart = directoryEntry.getStart();
+      dirSize = directoryEntry.getSize();
+      dirNumEntries = directoryEntry.getStore().getEntries().size();
+    } else {
+      // if we do not have a directory, then we must leave any required offset.
+      dirStart = extraDirectoryOffset;
+      dirSize = 0;
+      dirNumEntries = 0;
+    }
+
+    /*
+     * We need a Zip64 EOCD if any value overflows or if Zip64 file extensions are used as stated
+     * in the Zip Specification.
+     */
+
+    boolean useZip64Eocd =
+        dirStart > Eocd.MAX_CD_OFFSET ||
+            dirSize > Eocd.MAX_CD_SIZE ||
+            dirNumEntries > Eocd.MAX_TOTAL_RECORDS ||
+            (directoryEntry != null && directoryEntry.getStore().containsZip64Files());
+
+    /* construct the Zip64 EOCD and locator first, as they come before the standard EOCD */
+    if (useZip64Eocd) {
+      Verify.verify(eocdDataSector != null);
+      Zip64Eocd zip64Eocd =
+          new Zip64Eocd(dirNumEntries, dirStart, dirSize, useVersion2Header, eocdDataSector);
+      eocdDataSector = null;
+      byte[] zip64EocdBytes = zip64Eocd.toBytes();
+      long zip64Offset = map.size();
+      map.extend(zip64Offset + zip64EocdBytes.length);
+      eocd64Entry = map.add(zip64Offset, zip64Offset + zip64EocdBytes.length, zip64Eocd);
+
+      Zip64EocdLocator locator = new Zip64EocdLocator(eocd64Entry.getStart());
+      byte[] locatorBytes = locator.toBytes();
+      long locatorOffset = map.size();
+      map.extend(locatorOffset + locatorBytes.length);
+      eocd64Locator = map.add(locatorOffset, locatorOffset + locatorBytes.length, locator);
+    }
+
+    /* add the EOCD to the end of the file */
+
+    Verify.verify(eocdComment != null);
+    Eocd eocd = new Eocd(
+        Math.min(dirNumEntries, Eocd.MAX_TOTAL_RECORDS),
+        Math.min(dirStart, Eocd.MAX_CD_OFFSET),
+        Math.min(dirSize, Eocd.MAX_CD_SIZE),
+        eocdComment);
+    eocdComment = null;
+    byte[] eocdBytes = eocd.toBytes();
+    long eocdOffset = map.size();
+    map.extend(eocdOffset + eocdBytes.length);
+    eocdEntry = map.add(eocdOffset, eocdOffset + eocdBytes.length, eocd);
+  }
+
+  /**
+   * Writes the entire EOCD record to the end of the file. The EOCDGroup must <i>not</i> be empty
+   * ({@link #isEmpty()}) by being populated by a call to
+   * {@link #computeRecord(FileUseMapEntry, long)}, and the Central Directory must already be
+   * written to the file. If the CentralDirectory has not written, then {@link #file} should have
+   * no entries.
+   *
+   * @throws IOException Failed to write the EOCD record.
+   */
+  void appendToFile() throws IOException {
+    Preconditions.checkNotNull(eocdEntry, "eocdEntry == null");
+
+    if (eocd64Entry != null) {
+      Zip64Eocd zip64Eocd = eocd64Entry.getStore();
+      Preconditions.checkNotNull(zip64Eocd);
+      Zip64EocdLocator locator = eocd64Locator.getStore();
+      Preconditions.checkNotNull(locator);
+
+      file.directWrite(eocd64Entry.getStart(), zip64Eocd.toBytes());
+      file.directWrite(eocd64Locator.getStart(), locator.toBytes());
+    }
+
+    Eocd eocd = eocdEntry.getStore();
+    Preconditions.checkNotNull(eocd, "eocd == null");
+
+    byte[] eocdBytes = eocd.toBytes();
+    long eocdOffset = eocdEntry.getStart();
+
+    file.directWrite(eocdOffset, eocdBytes);
+  }
+
+  /**
+   * Obtains the byte array representation of the EOCD. The EOCD must have already been computed for
+   * this method to be invoked.
+   *
+   * @return The byte representation of the EOCD.
+   * @throws IOException Failed to obtain the byte representation of the EOCD.
+   */
+  byte[] getEocdBytes() throws IOException {
+    Preconditions.checkNotNull(eocdEntry, "eocdEntry == null");
+
+    Eocd eocd = eocdEntry.getStore();
+    Preconditions.checkNotNull(eocd, "eocd == null");
+
+    return eocd.toBytes();
+  }
+
+  /**
+   * Obtains the byte array representation of the Zip64 EOCD Locator. The EOCD record must already
+   * have been computed for this method to be invoked.
+   *
+   * @return The byte representation of the Zip64 EOCD Locator, or null if the EOCD record is not
+   * in Zip64 format.
+   * @throws IOException Failed to obtain the byte representation of the EOCD Locator.
+   */
+  @VisibleForTesting
+  @Nullable
+  byte[] getEocdLocatorBytes() throws IOException {
+    Preconditions.checkNotNull(eocdEntry);
+
+    if (eocd64Locator == null) {
+      return null;
+    }
+
+    return eocd64Locator.getStore().toBytes();
+  }
+
+  /**
+   * Obtains the byte array representation of the Zip64 EOCD. The EOCD record must already
+   * have been computed for this method to be invoked.
+   *
+   * @return The byte representation of the Zip64 EOCD, or null if the EOCD record is not
+   * in Zip64 format.
+   * @throws IOException Failed to obtain the byte representation of the Zip64 EOCD.
+   */
+  @VisibleForTesting
+  @Nullable
+  byte[] getZ64EocdBytes() throws IOException {
+    Preconditions.checkNotNull(eocdEntry);
+
+    if (eocd64Entry == null) {
+      return null;
+    }
+
+    return eocd64Entry.getStore().toBytes();
+  }
+
+  /**
+   * Checks whether the EOCD record is presently in-memory. (<i>i.e.</i> the EOCD was either read
+   * from disk and is still valid, or has been computed from the Central Directory).
+   *
+   * @return True iff the EOCD record is in-memory.
+   */
+  boolean isEmpty() {
+    return eocdEntry == null;
+  }
+
+  /**
+   * Sets whether or not the EOCD record should use the Version 1 or Version 2 of the Zip64 EOCD
+   * (iff the file needs a Zip64 record). The EOCD record should not be in-memory when trying to set
+   * this value, and the EOCD will need to be recomputed to have any affect.
+   *
+   * @param useVersion2Header True if the Version 2 header is to be used, and false for the Version
+   * 1 header.
+   */
+  void setUseVersion2Header(boolean useVersion2Header) {
+    verifyLog.verify(eocdEntry == null, "eocdEntry != null");
+
+    this.useVersion2Header = useVersion2Header;
+  }
+
+  /**
+   * Specifies if the EOCD Group will be using a Version 2 Zip64 EOCD record or a Version 1 record
+   * if the file needs to be in Zip64 format.
+   *
+   * @return True if the Version 2 record will be used, and false if the Version 1 record will be
+   * used.
+   */
+  boolean usingVersion2Header() {
+    return useVersion2Header;
+  }
+
+  /**
+   * Removes the EOCD record from memory.
+   */
+  void deleteRecord() {
+    if (eocdEntry != null) {
+      map.remove(eocdEntry);
+
+      Eocd eocd = eocdEntry.getStore();
+      Verify.verify(eocd != null);
+      eocdComment = eocd.getComment();
+      eocdEntry = null;
+    }
+
+    if (eocd64Locator != null) {
+      Verify.verify(eocd64Entry != null);
+      eocdDataSector = eocd64Entry.getStore().getExtraFields();
+      map.remove(eocd64Locator);
+      map.remove(eocd64Entry);
+      eocd64Locator = null;
+      eocd64Entry = null;
+    } else {
+      eocdDataSector = new Zip64ExtensibleDataSector();
+    }
+  }
+
+  /**
+   * Sets the EOCD comment.
+   *
+   * @param comment The new comment; no conversion is done, these exact bytes will be placed in the
+   *     EOCD comment.
+   * @throws IllegalArgumentException If the comment corrupts the ZipFile by having a valid EOCD
+   *     record in it.
+   */
+  void setEocdComment(byte[] comment) {
+    if (comment.length > MAX_EOCD_COMMENT_SIZE) {
+      throw new IllegalArgumentException(
+          "EOCD comment size ("
+              + comment.length
+              + ") is larger than the maximum allowed ("
+              + MAX_EOCD_COMMENT_SIZE
+              + ")");
+    }
+
+    // Check if the EOCD signature appears anywhere in the comment we need to check if it
+    // is valid.
+    for (int i = 0; i < comment.length - MIN_EOCD_SIZE; i++) {
+      // Remember: little endian...
+      ByteBuffer potentialSignature = ByteBuffer.wrap(comment, i, 4);
+      try {
+        if (LittleEndianUtils.readUnsigned4Le(potentialSignature) == EOCD_SIGNATURE) {
+          // We found a possible EOCD signature at position i. Try to read it.
+          ByteBuffer bytes = ByteBuffer.wrap(comment, i, comment.length - i);
+          try {
+            new Eocd(bytes);
+            // If a valid record is found in the comment then this corrupts the Zip file record
+            // as we look for the EOCD at the back of the file (where the comment is) first.
+            throw new IllegalArgumentException(
+                "Position " + i + " of the comment contains a valid EOCD record.");
+          } catch (IOException e) {
+            // Fine, this is an invalid record. Move along...
+          }
+        }
+      } catch (IOException e) {
+        throw new IOExceptionWrapper(e);
+      }
+    }
+
+    deleteRecord();
+    eocdComment = new byte[comment.length];
+    System.arraycopy(comment, 0, eocdComment, 0, comment.length);
+  }
+
+  /**
+   * Returns the start of the EOCD record location in the file or -1 if the EOCD is not in memory.
+   *
+   * @return The start of the record.
+   */
+  long getOffset() {
+    if (eocdEntry == null) {
+      return -1;
+    }
+    return getRecordStart();
+  }
+
+  /**
+   * Gets the comment in the EOCD.
+   *
+   * @return The comment exactly as it was encoded in the EOCD, no encoding is done.
+   */
+  byte[] getEocdComment() {
+    if (eocdEntry == null) {
+      Verify.verify(eocdComment != null);
+      byte[] eocdCommentCopy = eocdComment.clone();
+      return eocdCommentCopy;
+    }
+
+    Eocd eocd = eocdEntry.getStore();
+    Verify.verify(eocd != null);
+    return eocd.getComment();
+  }
+
+  /**
+   * Gets the size of the central directory as specified from the EOCD record. The EOCD must be in
+   * memory before this method is invoked.
+   *
+   * @return The directory's size.
+   */
+  long getDirectorySize() {
+    Preconditions.checkNotNull(eocdEntry, "eocdEntry == null");
+
+    Eocd eocd = eocdEntry.getStore();
+
+    if (eocd64Entry != null && eocd.getDirectorySize() == Eocd.MAX_CD_SIZE) {
+      return eocd64Entry.getStore().getDirectorySize();
+    } else {
+      return eocd.getDirectorySize();
+    }
+  }
+
+  /**
+   * Gets the offset of the Central Directory from the start of the archive as specified from the
+   * EOCD record. The EOCD must be in memory before this method is invoked.
+   *
+   * @return The offset of the start of the Central Directory.
+   */
+  long getDirectoryOffset() {
+    Preconditions.checkNotNull(eocdEntry, "eocdEntry == null");
+
+    Eocd eocd = eocdEntry.getStore();
+
+    if (eocd64Entry != null && eocd.getDirectoryOffset() == Eocd.MAX_CD_OFFSET) {
+      return eocd64Entry.getStore().getDirectoryOffset();
+    } else {
+      return eocd.getDirectoryOffset();
+    }
+  }
+
+  /**
+   * Gets the total number of entries in the Central Directory as specified from the EOCD record.
+   * The EOCD must be in memory before this method is invoked.
+   *
+   * @return The total number of records in the Central Directory.
+   */
+  long getTotalDirectoryRecords() {
+    Preconditions.checkNotNull(eocdEntry, "eocdEntry == null");
+
+    Eocd eocd = eocdEntry.getStore();
+    if (eocd64Entry != null && eocd.getTotalRecords() == Eocd.MAX_TOTAL_RECORDS) {
+      return eocd64Entry.getStore().getTotalRecords();
+    }
+
+    return eocd.getTotalRecords();
+  }
+
+  /**
+   * Returns the start of the EOCD record from the start of the archive. This will be the same as
+   * the start of the standard EOCD in a Zip32 file or in a Zip64 file will be the start of the
+   * Zip64 Eocd record. The EOCD must be in memory for this method to be invoked.
+   *
+   * @return The start of the entire EOCD record.
+   */
+  long getRecordStart() {
+    Verify.verify(eocdEntry != null, "eocdEntry == null");
+    if (eocd64Entry != null) {
+      return eocd64Entry.getStart();
+    }
+    return eocdEntry.getStart();
+  }
+
+  /**
+   * Returns the total size of the EOCD record. This will be the same as the standard EOCD size for
+   * a Zip32 file or in a Zip64 file will be the start of the Zip64 record to the end of the
+   * standard EOCD. the EOCD must be in memory for this method to be invoked.
+   *
+   * @return The total size of the EOCD record.
+   */
+  public long getRecordSize() {
+    if (eocd64Entry != null) {
+      Verify.verify(eocdEntry != null);
+      return eocdEntry.getEnd() - eocd64Entry.getStart();
+    }
+    if (eocdEntry == null) {
+      return -1;
+    }
+
+    return eocdEntry.getSize();
+  }
+
+  /**
+   * Returns the Zip64 Extensible Data Sector, or {@code null} if the EOCD record is not in the
+   * Zip64 format. The EOCD must be in memory for this method to be invoked.
+   *
+   * @return The Extensible data sector, or {@code null} if none exists.
+   */
+  @Nullable
+  public Zip64ExtensibleDataSector getExtensibleData() {
+    Verify.verify(eocdEntry != null);
+    if (eocd64Entry != null) {
+      return eocd64Entry.getStore().getExtraFields();
+    }
+
+    return null;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ExtraField.java

@@ -0,0 +1,385 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.LittleEndianUtils;
+import com.google.common.base.Preconditions;
+import com.google.common.collect.ImmutableList;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.util.ArrayList;
+import java.util.List;
+import javax.annotation.Nullable;
+
+/**
+ * Contains an extra field.
+ *
+ * <p>According to the zip specification, the extra field is composed of a sequence of fields. This
+ * class provides a way to access, parse and modify that information.
+ *
+ * <p>The zip specification calls fields to the fields inside the extra field. Because this
+ * terminology is confusing, we use <i>segment</i> to refer to a part of the extra field. Each
+ * segment is represented by an instance of {@link Segment} and contains a header ID and data.
+ *
+ * <p>Each instance of {@link ExtraField} is immutable. The extra field of a particular entry can be
+ * changed by creating a new instanceof {@link ExtraField} and pass it to {@link
+ * StoredEntry#setLocalExtra(ExtraField)}.
+ *
+ * <p>Instances of {@link ExtraField} can be created directly from the list of segments in it or
+ * from the raw byte data. If created from the raw byte data, the data will only be parsed on
+ * demand. So, if neither {@link #getSegments()} nor {@link #getSingleSegment(int)} is invoked, the
+ * extra field will not be parsed. This guarantees low performance impact of the using the extra
+ * field unless its contents are needed.
+ */
+public class ExtraField {
+  public static final ExtraField EMPTY = new ExtraField();
+
+  /** Header ID for field with zip alignment. */
+  static final int ALIGNMENT_ZIP_EXTRA_DATA_FIELD_HEADER_ID = 0xd935;
+
+  /**
+   * The field's raw data, if it is known. Either this variable or {@link #segments} must be
+   * non-{@code null}.
+   */
+  @Nullable private final byte[] rawData;
+
+  /**
+   * The list of field's segments. Will be populated if the extra field is created based on a list
+   * of segments; will also be populated after parsing if the extra field is created based on the
+   * raw bytes.
+   */
+  @Nullable private ImmutableList<Segment> segments;
+
+  /**
+   * Creates an extra field based on existing raw data.
+   *
+   * @param rawData the raw data; will not be parsed unless needed
+   */
+  public ExtraField(byte[] rawData) {
+    this.rawData = rawData;
+    segments = null;
+  }
+
+  /** Creates a new extra field with no segments. */
+  public ExtraField() {
+    rawData = null;
+    segments = ImmutableList.of();
+  }
+
+  /**
+   * Creates a new extra field with the given segments.
+   *
+   * @param segments the segments
+   */
+  public ExtraField(ImmutableList<Segment> segments) {
+    rawData = null;
+    this.segments = segments;
+  }
+
+  /**
+   * Obtains all segments in the extra field.
+   *
+   * @return all segments
+   * @throws IOException failed to parse the extra field
+   */
+  public ImmutableList<Segment> getSegments() throws IOException {
+    if (segments == null) {
+      parseSegments();
+    }
+
+    Preconditions.checkNotNull(segments);
+    return segments;
+  }
+
+  /**
+   * Obtains the only segment with the provided header ID.
+   *
+   * @param headerId the header ID
+   * @return the segment found or {@code null} if no segment contains the provided header ID
+   * @throws IOException there is more than one header with the provided header ID
+   */
+  @Nullable
+  public Segment getSingleSegment(int headerId) throws IOException {
+    List<Segment> found = new ArrayList<>();
+    for (Segment s : getSegments()) {
+      if (s.getHeaderId() == headerId) {
+        found.add(s);
+      }
+    }
+
+    if (found.isEmpty()) {
+      return null;
+    } else if (found.size() == 1) {
+      return found.get(0);
+    } else {
+      throw new IOException(found.size() + " segments with header ID " + headerId + "found");
+    }
+  }
+
+  /**
+   * Parses the raw data and generates all segments in {@link #segments}.
+   *
+   * @throws IOException failed to parse the data
+   */
+  private void parseSegments() throws IOException {
+    Preconditions.checkNotNull(rawData);
+    Preconditions.checkState(segments == null);
+
+    List<Segment> segments = new ArrayList<>();
+    ByteBuffer buffer = ByteBuffer.wrap(rawData);
+
+    while (buffer.remaining() > 0) {
+      int headerId = LittleEndianUtils.readUnsigned2Le(buffer);
+      int dataSize = LittleEndianUtils.readUnsigned2Le(buffer);
+      if (dataSize < 0) {
+        throw new IOException(
+            "Invalid data size for extra field segment with header ID "
+                + headerId
+                + ": "
+                + dataSize);
+      }
+
+      byte[] data = new byte[dataSize];
+      if (buffer.remaining() < dataSize) {
+        throw new IOException(
+            "Invalid data size for extra field segment with header ID "
+                + headerId
+                + ": "
+                + dataSize
+                + " (only "
+                + buffer.remaining()
+                + " bytes are available)");
+      }
+      buffer.get(data);
+
+      SegmentFactory factory = identifySegmentFactory(headerId);
+      Segment seg = factory.make(headerId, data);
+      segments.add(seg);
+    }
+
+    this.segments = ImmutableList.copyOf(segments);
+  }
+
+  /**
+   * Obtains the size of the extra field.
+   *
+   * @return the size
+   */
+  public int size() {
+    if (rawData != null) {
+      return rawData.length;
+    } else {
+      Preconditions.checkNotNull(segments);
+      int sz = 0;
+      for (Segment s : segments) {
+        sz += s.size();
+      }
+
+      return sz;
+    }
+  }
+
+  /**
+   * Writes the extra field to the given output buffer.
+   *
+   * @param out the output buffer to write the field; exactly {@link #size()} bytes will be written
+   * @throws IOException failed to write the extra fields
+   */
+  public void write(ByteBuffer out) throws IOException {
+    if (rawData != null) {
+      out.put(rawData);
+    } else {
+      Preconditions.checkNotNull(segments);
+      for (Segment s : segments) {
+        s.write(out);
+      }
+    }
+  }
+
+  /**
+   * Identifies the factory to create the segment with the provided header ID.
+   *
+   * @param headerId the header ID
+   * @return the segmnet factory that creates segments with the given header
+   */
+  private static SegmentFactory identifySegmentFactory(int headerId) {
+    if (headerId == ALIGNMENT_ZIP_EXTRA_DATA_FIELD_HEADER_ID) {
+      return AlignmentSegment::new;
+    }
+
+    return RawDataSegment::new;
+  }
+
+  /**
+   * Field inside the extra field. A segment contains a header ID and data. Specific types of
+   * segments implement this interface.
+   */
+  public interface Segment {
+
+    /**
+     * Obtains the segment's header ID.
+     *
+     * @return the segment's header ID
+     */
+    int getHeaderId();
+
+    /**
+     * Obtains the size of the segment including the header ID.
+     *
+     * @return the number of bytes needed to write the segment
+     */
+    int size();
+
+    /**
+     * Writes the segment to a buffer.
+     *
+     * @param out the buffer where to write the segment to; exactly {@link #size()} bytes will be
+     *     written
+     * @throws IOException failed to write segment data
+     */
+    void write(ByteBuffer out) throws IOException;
+  }
+
+  /** Factory that creates a segment. */
+  interface SegmentFactory {
+
+    /**
+     * Creates a new segment.
+     *
+     * @param headerId the header ID
+     * @param data the segment's data
+     * @return the created segment
+     * @throws IOException failed to create the segment from the data
+     */
+    Segment make(int headerId, byte[] data) throws IOException;
+  }
+
+  /**
+   * Segment of raw data: this class represents a general segment containing an array of bytes as
+   * data.
+   */
+  public static class RawDataSegment implements Segment {
+
+    /** Header ID. */
+    private final int headerId;
+
+    /** Data in the segment. */
+    private final byte[] data;
+
+    /**
+     * Creates a new raw data segment.
+     *
+     * @param headerId the header ID
+     * @param data the segment data
+     */
+    RawDataSegment(int headerId, byte[] data) {
+      this.headerId = headerId;
+      this.data = data;
+    }
+
+    @Override
+    public int getHeaderId() {
+      return headerId;
+    }
+
+    @Override
+    public void write(ByteBuffer out) throws IOException {
+      LittleEndianUtils.writeUnsigned2Le(out, headerId);
+      LittleEndianUtils.writeUnsigned2Le(out, data.length);
+      out.put(data);
+    }
+
+    @Override
+    public int size() {
+      return 4 + data.length;
+    }
+  }
+
+  /**
+   * Segment with information on an alignment: this segment contains information on how an entry
+   * should be aligned and contains zero-filled data to force alignment.
+   *
+   * <p>An alignment segment contains the header ID, the size of the data, the alignment value and
+   * zero bytes to pad
+   */
+  public static class AlignmentSegment implements Segment {
+
+    /** Minimum size for an alignment segment. */
+    public static final int MINIMUM_SIZE = 6;
+
+    /** The alignment value. */
+    private int alignment;
+
+    /** How many bytes of padding are in this segment? */
+    private int padding;
+
+    /**
+     * Creates a new alignment segment.
+     *
+     * @param alignment the alignment value
+     * @param totalSize how many bytes should this segment take?
+     */
+    public AlignmentSegment(int alignment, int totalSize) {
+      Preconditions.checkArgument(alignment > 0, "alignment <= 0");
+      Preconditions.checkArgument(totalSize >= MINIMUM_SIZE, "totalSize < MINIMUM_SIZE");
+
+      /*
+       * We have 6 bytes of fixed data: header ID (2 bytes), data size (2 bytes), alignment
+       * value (2 bytes).
+       */
+      this.alignment = alignment;
+      padding = totalSize - MINIMUM_SIZE;
+    }
+
+    /**
+     * Creates a new alignment segment from extra data.
+     *
+     * @param headerId the header ID
+     * @param data the segment data
+     * @throws IOException failed to create the segment from the data
+     */
+    public AlignmentSegment(int headerId, byte[] data) throws IOException {
+      Preconditions.checkArgument(headerId == ALIGNMENT_ZIP_EXTRA_DATA_FIELD_HEADER_ID);
+
+      ByteBuffer dataBuffer = ByteBuffer.wrap(data);
+      alignment = LittleEndianUtils.readUnsigned2Le(dataBuffer);
+      if (alignment <= 0) {
+        throw new IOException("Invalid alignment in alignment field: " + alignment);
+      }
+
+      padding = data.length - 2;
+    }
+
+    @Override
+    public void write(ByteBuffer out) throws IOException {
+      LittleEndianUtils.writeUnsigned2Le(out, ALIGNMENT_ZIP_EXTRA_DATA_FIELD_HEADER_ID);
+      LittleEndianUtils.writeUnsigned2Le(out, padding + 2);
+      LittleEndianUtils.writeUnsigned2Le(out, alignment);
+      out.put(new byte[padding]);
+    }
+
+    @Override
+    public int size() {
+      return padding + 6;
+    }
+
+    @Override
+    public int getHeaderId() {
+      return ALIGNMENT_ZIP_EXTRA_DATA_FIELD_HEADER_ID;
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/FileUseMap.java

@@ -0,0 +1,598 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import com.google.common.collect.Lists;
+import com.google.common.collect.Sets;
+import com.google.common.primitives.Ints;
+import java.util.List;
+import java.util.Set;
+import java.util.SortedSet;
+import java.util.TreeSet;
+import javax.annotation.Nullable;
+
+/**
+ * The file use map keeps track of which parts of the zip file are used which parts are not. It
+ * essentially maintains an ordered set of entries ({@link FileUseMapEntry}). Each entry either has
+ * some data (an entry, the Central Directory, the EOCD) or is a free entry.
+ *
+ * <p>For example: [0-95, "foo/"][95-260, "xpto"][260-310, free][310-360, Central Directory]
+ * [360-390,EOCD]
+ *
+ * <p>There are a few invariants in this structure:
+ *
+ * <ul>
+ *   <li>there are no gaps between map entries;
+ *   <li>the map is fully covered up to its size;
+ *   <li>there are no two free entries next to each other; this is guaranteed by coalescing the
+ *       entries upon removal (see {@link #coalesce(FileUseMapEntry)});
+ *   <li>all free entries have a minimum size defined in the constructor, with the possible
+ *       exception of the last one
+ * </ul>
+ */
+class FileUseMap {
+  /**
+   * Size of the file according to the map. This should always match the last entry in {@code #map}.
+   */
+  private long size;
+
+  /**
+   * Tree with all intervals ordered by position. Contains coverage from 0 up to {@link #size}. If
+   * {@link #size} is zero then this set is empty. This is the only situation in which the map will
+   * be empty.
+   */
+  private final TreeSet<FileUseMapEntry<?>> map;
+
+  /**
+   * Tree with all free blocks ordered by size. This is essentially a view over {@link #map}
+   * containing only the free blocks, but in a different order.
+   */
+  private final TreeSet<FileUseMapEntry<?>> freeBySize;
+  private final TreeSet<FileUseMapEntry<?>> freeByStart;
+
+  /** If defined, defines the minimum size for a free entry. */
+  private int mMinFreeSize;
+
+  /**
+   * Creates a new, empty file map.
+   *
+   * @param size the size of the file
+   * @param minFreeSize minimum size of a free entry
+   */
+  FileUseMap(long size, int minFreeSize) {
+    Preconditions.checkArgument(size >= 0, "size < 0");
+    Preconditions.checkArgument(minFreeSize >= 0, "minFreeSize < 0");
+
+    this.size = size;
+    map = new TreeSet<>(FileUseMapEntry.COMPARE_BY_START);
+    freeBySize = new TreeSet<>(FileUseMapEntry.COMPARE_BY_SIZE);
+    freeByStart = new TreeSet<>(FileUseMapEntry.COMPARE_BY_START);
+    mMinFreeSize = minFreeSize;
+
+    if (size > 0) {
+      internalAdd(FileUseMapEntry.makeFree(0, size));
+    }
+  }
+
+  /**
+   * Adds an entry to the internal structures.
+   *
+   * @param entry the entry to add
+   */
+  private void internalAdd(FileUseMapEntry<?> entry) {
+    map.add(entry);
+
+    if (entry.isFree()) {
+      freeBySize.add(entry);
+      freeByStart.add(entry);
+    }
+  }
+
+  /**
+   * Removes an entry from the internal structures.
+   *
+   * @param entry the entry to remove
+   */
+  private void internalRemove(FileUseMapEntry<?> entry) {
+    boolean wasRemoved = map.remove(entry);
+    Preconditions.checkState(wasRemoved, "entry not in map");
+
+    if (entry.isFree()) {
+      freeBySize.remove(entry);
+      freeByStart.remove(entry);
+    }
+  }
+
+  /**
+   * Adds a new file to the map. The interval specified by {@code entry} must fit inside an empty
+   * entry in the map. That entry will be replaced by entry and additional free entries will be
+   * added before and after if needed to make sure no spaces exist on the map.
+   *
+   * @param entry the entry to add
+   */
+  private void add(FileUseMapEntry<?> entry) {
+    Preconditions.checkArgument(entry.getStart() < size, "entry.getStart() >= size");
+    Preconditions.checkArgument(entry.getEnd() <= size, "entry.getEnd() > size");
+    Preconditions.checkArgument(!entry.isFree(), "entry.isFree()");
+
+    FileUseMapEntry<?> container = findContainer(entry);
+    Verify.verify(container.isFree(), "!container.isFree()");
+
+    Set<FileUseMapEntry<?>> replacements = split(container, entry);
+    internalRemove(container);
+    for (FileUseMapEntry<?> r : replacements) {
+      internalAdd(r);
+    }
+  }
+
+  /**
+   * Adds a new file to the map. The interval specified by ({@code start}, {@code end}) must fit
+   * inside an empty entry in the map. That entry will be replaced by entry and additional free
+   * entries will be added before and after if needed to make sure no spaces exist on the map.
+   *
+   * <p>The entry cannot extend beyong the end of the map. If necessary, extend the map using {@link
+   * #extend(long)}.
+   *
+   * @param start the start of this entry
+   * @param end the end of the entry
+   * @param store extra data to store with the entry
+   * @param <T> the type of data to store in the entry
+   * @return the new entry
+   */
+  <T> FileUseMapEntry<T> add(long start, long end, T store) {
+    Preconditions.checkArgument(start >= 0, "start < 0");
+    Preconditions.checkArgument(end > start, "end < start");
+
+    FileUseMapEntry<T> entry = FileUseMapEntry.makeUsed(start, end, store);
+    add(entry);
+    return entry;
+  }
+
+  /**
+   * Removes a file from the map, replacing it with an empty one that is then coalesced with
+   * neighbors (if the neighbors are free).
+   *
+   * @param entry the entry
+   */
+  void remove(FileUseMapEntry<?> entry) {
+    Preconditions.checkState(map.contains(entry), "!map.contains(entry)");
+    Preconditions.checkArgument(!entry.isFree(), "entry.isFree()");
+
+    internalRemove(entry);
+
+    FileUseMapEntry<?> replacement = FileUseMapEntry.makeFree(entry.getStart(), entry.getEnd());
+    internalAdd(replacement);
+    coalesce(replacement);
+  }
+
+  /**
+   * Finds the entry that fully contains the given one. It is assumed that one exists.
+   *
+   * @param entry the entry whose container we're looking for
+   * @return the container
+   */
+  private FileUseMapEntry<?> findContainer(FileUseMapEntry<?> entry) {
+    FileUseMapEntry container = map.floor(entry);
+    Verify.verifyNotNull(container);
+    Verify.verify(container.getStart() <= entry.getStart());
+    Verify.verify(container.getEnd() >= entry.getEnd());
+
+    return container;
+  }
+
+  /**
+   * Splits a container to add an entry, adding new free entries before and after the provided entry
+   * if needed.
+   *
+   * @param container the container entry, a free entry that is in {@link #map} that that encloses
+   *     {@code entry}
+   * @param entry the entry that will be used to split {@code container}
+   * @return a set of non-overlapping entries that completely covers {@code container} and that
+   *     includes {@code entry}
+   */
+  private static Set<FileUseMapEntry<?>> split(
+      FileUseMapEntry<?> container, FileUseMapEntry<?> entry) {
+    Preconditions.checkArgument(container.isFree(), "!container.isFree()");
+
+    long farStart = container.getStart();
+    long start = entry.getStart();
+    long end = entry.getEnd();
+    long farEnd = container.getEnd();
+
+    Verify.verify(farStart <= start, "farStart > start");
+    Verify.verify(start < end, "start >= end");
+    Verify.verify(farEnd >= end, "farEnd < end");
+
+    Set<FileUseMapEntry<?>> result = Sets.newHashSet();
+    if (farStart < start) {
+      result.add(FileUseMapEntry.makeFree(farStart, start));
+    }
+
+    result.add(entry);
+
+    if (end < farEnd) {
+      result.add(FileUseMapEntry.makeFree(end, farEnd));
+    }
+
+    return result;
+  }
+
+  /**
+   * Coalesces a free entry replacing it and neighboring free entries with a single, larger entry.
+   * This method does nothing if {@code entry} does not have free neighbors.
+   *
+   * @param entry the free entry to coalesce with neighbors
+   */
+  private void coalesce(FileUseMapEntry<?> entry) {
+    Preconditions.checkArgument(entry.isFree(), "!entry.isFree()");
+
+    FileUseMapEntry<?> prevToMerge = null;
+    long start = entry.getStart();
+    if (start > 0) {
+      /*
+       * See if we have a previous entry to merge with this one.
+       */
+      prevToMerge = map.floor(FileUseMapEntry.makeFree(start - 1, start));
+      Verify.verifyNotNull(prevToMerge);
+      if (!prevToMerge.isFree()) {
+        prevToMerge = null;
+      }
+    }
+
+    FileUseMapEntry<?> nextToMerge = null;
+    long end = entry.getEnd();
+    if (end < size) {
+      /*
+       * See if we have a next entry to merge with this one.
+       */
+      nextToMerge = map.ceiling(FileUseMapEntry.makeFree(end, end + 1));
+      Verify.verifyNotNull(nextToMerge);
+      if (!nextToMerge.isFree()) {
+        nextToMerge = null;
+      }
+    }
+
+    if (prevToMerge == null && nextToMerge == null) {
+      return;
+    }
+
+    long newStart = start;
+    if (prevToMerge != null) {
+      newStart = prevToMerge.getStart();
+      internalRemove(prevToMerge);
+    }
+
+    long newEnd = end;
+    if (nextToMerge != null) {
+      newEnd = nextToMerge.getEnd();
+      internalRemove(nextToMerge);
+    }
+
+    internalRemove(entry);
+    internalAdd(FileUseMapEntry.makeFree(newStart, newEnd));
+  }
+
+  /** Truncates map removing the top entry if it is free and reducing the map's size. */
+  void truncate() {
+    if (size == 0) {
+      return;
+    }
+
+    /*
+     * Find the last entry.
+     */
+    FileUseMapEntry<?> last = map.last();
+    Verify.verifyNotNull(last, "last == null");
+    if (last.isFree()) {
+      internalRemove(last);
+      size = last.getStart();
+    }
+  }
+
+  /**
+   * Obtains the size of the map.
+   *
+   * @return the size
+   */
+  long size() {
+    return size;
+  }
+
+  /**
+   * Obtains the largest used offset in the map. This will be size of the map after truncation.
+   *
+   * @return the size of the file discounting the last block if it is empty
+   */
+  long usedSize() {
+    if (size == 0) {
+      return 0;
+    }
+
+    /*
+     * Find the last entry to see if it is an empty entry. If it is, we need to remove its size
+     * from the returned value.
+     */
+    FileUseMapEntry<?> last = map.last();
+    Verify.verifyNotNull(last, "last == null");
+    if (last.isFree()) {
+      return last.getStart();
+    } else {
+      Verify.verify(last.getEnd() == size);
+      return size;
+    }
+  }
+
+  /**
+   * Extends the map to guarantee it has at least {@code size} bytes. If the current size is as
+   * large as {@code size}, this method does nothing.
+   *
+   * @param size the new size of the map that cannot be smaller that the current size
+   */
+  void extend(long size) {
+    Preconditions.checkArgument(size >= this.size, "size < size");
+
+    if (this.size == size) {
+      return;
+    }
+
+    FileUseMapEntry<?> newBlock = FileUseMapEntry.makeFree(this.size, size);
+    internalAdd(newBlock);
+
+    this.size = size;
+
+    coalesce(newBlock);
+  }
+
+  /**
+   * Locates a free area in the map with at least {@code size} bytes such that {@code ((start +
+   * alignOffset) % align == 0} and such that the free space before {@code start} is not smaller
+   * than the minimum free entry size. This method will follow the algorithm specified by {@code
+   * alg}.
+   *
+   * <p>If no free contiguous block exists in the map that can hold the provided size then the first
+   * free index at the end of the map is provided. This means that the map may need to be extended
+   * before data can be added.
+   *
+   * @param size the size of the contiguous area requested
+   * @param alignOffset an offset to which alignment needs to be computed (see method description)
+   * @param align alignment at the offset (see method description)
+   * @param alg which algorithm to use
+   * @return the location of the contiguous area; this may be located at the end of the map
+   */
+  long locateFree(long size, long alignOffset, long align, PositionAlgorithm alg) {
+    Preconditions.checkArgument(size > 0, "size <= 0");
+
+    FileUseMapEntry<?> minimumSizedEntry = FileUseMapEntry.makeFree(0, size);
+    SortedSet<FileUseMapEntry<?>> matches;
+
+    switch (alg) {
+      case BEST_FIT:
+        matches = freeBySize.tailSet(minimumSizedEntry);
+        break;
+      case FIRST_FIT:
+        matches = freeByStart;
+        break;
+      default:
+        throw new AssertionError();
+    }
+
+    FileUseMapEntry<?> best = null;
+    long bestExtraSize = 0;
+    for (FileUseMapEntry<?> curr : matches) {
+      /*
+       * We don't care about blocks that aren't free.
+       */
+      if (!curr.isFree()) {
+        continue;
+      }
+
+      /*
+       * Compute any extra size we need in this block to make sure we verify the alignment.
+       * There must be a better to do this...
+       */
+      long extraSize;
+      if (align == 0) {
+        extraSize = 0;
+      } else {
+        extraSize = (align - ((curr.getStart() + alignOffset) % align)) % align;
+      }
+
+      /*
+       * We can't leave than mMinFreeSize before. So if the extraSize is less than
+       * mMinFreeSize, we have to increase it by 'align' as many times as needed. For
+       * example, if mMinFreeSize is 20, align 4 and extraSize is 5. We need to increase it
+       * to 21 (5 + 4 * 4)
+       */
+      if (extraSize > 0 && extraSize < mMinFreeSize) {
+        int addAlignBlocks = Ints.checkedCast((mMinFreeSize - extraSize + align - 1) / align);
+        extraSize += addAlignBlocks * align;
+      }
+
+      /*
+       * We don't care about blocks where we don't fit in.
+       */
+      if (curr.getSize() < (size + extraSize)) {
+        continue;
+      }
+
+      /*
+       * We don't care about blocks that leave less than the minimum size after. There are
+       * two exceptions: (1) this is the last block and (2) the next block is free in which
+       * case, after coalescing, the free block with have at least the minimum size.
+       */
+      long emptySpaceLeft = curr.getSize() - (size + extraSize);
+      if (emptySpaceLeft > 0 && emptySpaceLeft < mMinFreeSize) {
+        FileUseMapEntry<?> next = map.higher(curr);
+        if (next != null && !next.isFree()) {
+          continue;
+        }
+      }
+
+      /*
+       * We don't care about blocks that are bigger than the best so far (otherwise this
+       * wouldn't be a best-fit algorithm).
+       */
+      if (best != null && best.getSize() < curr.getSize()) {
+        continue;
+      }
+
+      best = curr;
+      bestExtraSize = extraSize;
+
+      /*
+       * If we're doing first fit, we don't want to search for a better one :)
+       */
+      if (alg == PositionAlgorithm.FIRST_FIT) {
+        break;
+      }
+    }
+
+    /*
+     * If no entry that could hold size is found, get the first free byte.
+     */
+    long firstFree = this.size;
+    if (best == null && !map.isEmpty()) {
+      FileUseMapEntry<?> last = map.last();
+      if (last.isFree()) {
+        firstFree = last.getStart();
+      }
+    }
+
+    /*
+     * We're done: either we found something or we didn't, in which the new entry needs to
+     * be added to the end of the map.
+     */
+    if (best == null) {
+      long extra = (align - ((firstFree + alignOffset) % align)) % align;
+
+      /*
+       * If adding this entry at the end would create a space smaller than the minimum,
+       * push it for 'align' bytes forward.
+       */
+      if (extra > 0) {
+        if (extra < mMinFreeSize) {
+          extra += align * (((mMinFreeSize - extra) + (align - 1)) / align);
+        }
+      }
+
+      return firstFree + extra;
+    } else {
+      return best.getStart() + bestExtraSize;
+    }
+  }
+
+  /**
+   * Obtains all free areas of the map, excluding any trailing free area.
+   *
+   * @return all free areas, an empty set if there are no free areas; the areas are returned in file
+   *     order, that is, if area {@code x} starts before area {@code y}, then area {@code x} will be
+   *     stored before area {@code y} in the list
+   */
+  List<FileUseMapEntry<?>> getFreeAreas() {
+    List<FileUseMapEntry<?>> freeAreas = Lists.newArrayList();
+
+    for (FileUseMapEntry<?> area : map) {
+      if (area.isFree() && area.getEnd() != size) {
+        freeAreas.add(area);
+      }
+    }
+
+    return freeAreas;
+  }
+
+  /**
+   * Obtains the entry that is located before the one provided.
+   *
+   * @param entry the map entry to get the previous one for; must belong to the map
+   * @return the entry before the provided one, {@code null} if {@code entry} is the first entry in
+   *     the map
+   */
+  @Nullable
+  FileUseMapEntry<?> before(FileUseMapEntry<?> entry) {
+    Preconditions.checkNotNull(entry, "entry == null");
+
+    return map.lower(entry);
+  }
+
+  /**
+   * Obtains the entry that is located after the one provided.
+   *
+   * @param entry the map entry to get the next one for; must belong to the map
+   * @return the entry after the provided one, {@code null} if {@code entry} is the last entry in
+   *     the map
+   */
+  @Nullable
+  FileUseMapEntry<?> after(FileUseMapEntry<?> entry) {
+    Preconditions.checkNotNull(entry, "entry == null");
+
+    return map.higher(entry);
+  }
+
+  /**
+   * Obtains the entry at the given offset.
+   *
+   * @param offset the offset to look for
+   * @return the entry found or {@code null} if there is no entry (not even a free one) at the given
+   *     offset
+   */
+  @Nullable
+  FileUseMapEntry<?> at(long offset) {
+    Preconditions.checkArgument(offset >= 0, "offset < 0");
+    Preconditions.checkArgument(offset < size, "offset >= size");
+
+    FileUseMapEntry<?> entry = map.floor(FileUseMapEntry.makeFree(offset, offset + 1));
+    if (entry == null) {
+      return null;
+    }
+
+    Verify.verify(entry.getStart() <= offset);
+    Verify.verify(entry.getEnd() > offset);
+
+    return entry;
+  }
+
+  @Override
+  public String toString() {
+    StringBuilder builder = new StringBuilder();
+    boolean first = true;
+    for (FileUseMapEntry<?> entry : map) {
+      if (first) {
+        first = false;
+      } else {
+        builder.append(", ");
+      }
+
+      builder.append(entry.getStart());
+      builder.append(" - ");
+      builder.append(entry.getEnd());
+      builder.append(": ");
+      builder.append(entry.getStore());
+    }
+    return builder.toString();
+  }
+
+  /** Algorithms used to position entries in blocks. */
+  public enum PositionAlgorithm {
+    /** Best fit: finds the smallest free block that can receive the entry. */
+    BEST_FIT,
+
+    /** First fit: finds the first free block that can receive the entry. */
+    FIRST_FIT
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/FileUseMapEntry.java

@@ -0,0 +1,151 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.google.common.base.MoreObjects;
+import com.google.common.base.Preconditions;
+import com.google.common.primitives.Ints;
+import java.util.Comparator;
+import javax.annotation.Nullable;
+
+/**
+ * Represents an entry in the {@link FileUseMap}. Each entry contains an interval of bytes. The end
+ * of the interval is exclusive.
+ *
+ * <p>Entries can either be free or used. Used entries <em>must</em> store an object. Free entries
+ * do not store anything.
+ *
+ * <p>File map entries are used to keep track of which parts of a file map are used and not.
+ *
+ * @param <T> the type of data stored
+ */
+class FileUseMapEntry<T> {
+
+  /** Comparator that compares entries by their start date. */
+  public static final Comparator<FileUseMapEntry<?>> COMPARE_BY_START =
+      (o1, o2) -> Ints.saturatedCast(o1.getStart() - o2.getStart());
+
+  /** Comparator that compares entries by their size. */
+  public static final Comparator<FileUseMapEntry<?>> COMPARE_BY_SIZE =
+      (o1, o2) -> Ints.saturatedCast(o1.getSize() - o2.getSize());
+
+  /** The first byte in the entry. */
+  private final long start;
+
+  /** The first byte no longer in the entry. */
+  private final long end;
+
+  /** The stored data. If {@code null} then this entry represents a free entry. */
+  @Nullable private final T store;
+
+  /**
+   * Creates a new map entry.
+   *
+   * @param start the start of the entry
+   * @param end the end of the entry (first byte no longer in the entry)
+   * @param store the data to store in the entry or {@code null} if this is a free entry
+   */
+  private FileUseMapEntry(long start, long end, @Nullable T store) {
+    Preconditions.checkArgument(start >= 0, "start < 0");
+    Preconditions.checkArgument(end > start, "end <= start");
+
+    this.start = start;
+    this.end = end;
+    this.store = store;
+  }
+
+  /**
+   * Creates a new free entry.
+   *
+   * @param start the start of the entry
+   * @param end the end of the entry (first byte no longer in the entry)
+   * @return the entry
+   */
+  public static FileUseMapEntry<Object> makeFree(long start, long end) {
+    return new FileUseMapEntry<>(start, end, null);
+  }
+
+  /**
+   * Creates a new used entry.
+   *
+   * @param start the start of the entry
+   * @param end the end of the entry (first byte no longer in the entry)
+   * @param store the data to store in the entry
+   * @param <T> the type of data to store in the entry
+   * @return the entry
+   */
+  public static <T> FileUseMapEntry<T> makeUsed(long start, long end, T store) {
+    Preconditions.checkNotNull(store, "store == null");
+    return new FileUseMapEntry<>(start, end, store);
+  }
+
+  /**
+   * Obtains the first byte in the entry.
+   *
+   * @return the first byte in the entry (if the same value as {@link #getEnd()} then the entry is
+   *     empty and contains no data)
+   */
+  long getStart() {
+    return start;
+  }
+
+  /**
+   * Obtains the first byte no longer in the entry.
+   *
+   * @return the first byte no longer in the entry
+   */
+  long getEnd() {
+    return end;
+  }
+
+  /**
+   * Obtains the size of the entry.
+   *
+   * @return the number of bytes contained in the entry
+   */
+  long getSize() {
+    return end - start;
+  }
+
+  /**
+   * Determines if this is a free entry.
+   *
+   * @return is this entry free?
+   */
+  boolean isFree() {
+    return store == null;
+  }
+
+  /**
+   * Obtains the data stored in the entry.
+   *
+   * @return the data stored or {@code null} if this entry is a free entry
+   */
+  @Nullable
+  T getStore() {
+    return store;
+  }
+
+  @Override
+  public String toString() {
+    return MoreObjects.toStringHelper(this)
+        .add("start", start)
+        .add("end", end)
+        .add("store", store)
+        .toString();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/GPFlags.java

@@ -0,0 +1,159 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import java.io.IOException;
+
+/**
+ * General purpose bit flags. Contains the encoding of the zip's general purpose bits.
+ *
+ * <p>We don't really care about the method bit(s). These are bits 1 and 2. Here are the values:
+ *
+ * <ul>
+ *   <li>0 (00): Normal (-en) compression option was used.
+ *   <li>1 (01): Maximum (-exx/-ex) compression option was used.
+ *   <li>2 (10): Fast (-ef) compression option was used.
+ *   <li>3 (11): Super Fast (-es) compression option was used.
+ * </ul>
+ */
+class GPFlags {
+
+  /** Is the entry encrypted? */
+  private static final int BIT_ENCRYPTION = 1;
+
+  /** Has CRC computation been deferred and, therefore, does a data description block exist? */
+  private static final int BIT_DEFERRED_CRC = (1 << 3);
+
+  /** Is enhanced deflating used? */
+  private static final int BIT_ENHANCED_DEFLATING = (1 << 4);
+
+  /** Does the entry contain patched data? */
+  private static final int BIT_PATCHED_DATA = (1 << 5);
+
+  /** Is strong encryption used? */
+  private static final int BIT_STRONG_ENCRYPTION = (1 << 6) | (1 << 13);
+
+  /**
+   * If this bit is set the filename and comment fields for this file must be encoded using UTF-8.
+   */
+  private static final int BIT_EFS = (1 << 11);
+
+  /** Unused bits. */
+  private static final int BIT_UNUSED =
+      (1 << 7) | (1 << 8) | (1 << 9) | (1 << 10) | (1 << 14) | (1 << 15);
+
+  /** Bit flag value. */
+  private final long value;
+
+  /** Has the CRC computation beeen deferred? */
+  private boolean deferredCrc;
+
+  /** Is the file name encoded in UTF-8? */
+  private boolean utf8FileName;
+
+  /**
+   * Creates a new flags object.
+   *
+   * @param value the value of the bit mask
+   */
+  private GPFlags(long value) {
+    this.value = value;
+
+    deferredCrc = ((value & BIT_DEFERRED_CRC) != 0);
+    utf8FileName = ((value & BIT_EFS) != 0);
+  }
+
+  /**
+   * Obtains the flags value.
+   *
+   * @return the value of the bit mask
+   */
+  public long getValue() {
+    return value;
+  }
+
+  /**
+   * Is the CRC computation deferred?
+   *
+   * @return is the CRC computation deferred?
+   */
+  public boolean isDeferredCrc() {
+    return deferredCrc;
+  }
+
+  /**
+   * Is the file name encoded in UTF-8?
+   *
+   * @return is the file name encoded in UTF-8?
+   */
+  public boolean isUtf8FileName() {
+    return utf8FileName;
+  }
+
+  /**
+   * Creates a new bit mask.
+   *
+   * @param utf8Encoding should UTF-8 encoding be used?
+   * @return the new bit mask
+   */
+  static GPFlags make(boolean utf8Encoding) {
+    long flags = 0;
+
+    if (utf8Encoding) {
+      flags |= BIT_EFS;
+    }
+
+    return new GPFlags(flags);
+  }
+
+  /**
+   * Creates the flag information from a byte. This method will also validate that only supported
+   * options are defined in the flag.
+   *
+   * @param bits the bit mask
+   * @return the created flag information
+   * @throws IOException unsupported options are used in the bit mask
+   */
+  static GPFlags from(long bits) throws IOException {
+    if ((bits & BIT_ENCRYPTION) != 0) {
+      throw new IOException("Zip files with encrypted of entries not supported.");
+    }
+
+    if ((bits & BIT_ENHANCED_DEFLATING) != 0) {
+      throw new IOException("Enhanced deflating not supported.");
+    }
+
+    if ((bits & BIT_PATCHED_DATA) != 0) {
+      throw new IOException("Compressed patched data not supported.");
+    }
+
+    if ((bits & BIT_STRONG_ENCRYPTION) != 0) {
+      throw new IOException("Strong encryption not supported.");
+    }
+
+    if ((bits & BIT_UNUSED) != 0) {
+      throw new IOException(
+          "Unused bits set in directory entry. Weird. I don't know what's " + "going on.");
+    }
+
+    if ((bits & 0xffffffff00000000L) != 0) {
+      throw new IOException("Unsupported bits after 32.");
+    }
+
+    return new GPFlags(bits);
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/InflaterByteSource.java

@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import java.io.ByteArrayInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.SequenceInputStream;
+import java.util.zip.Inflater;
+import java.util.zip.InflaterInputStream;
+
+/**
+ * Byte source that inflates another byte source. It assumed the inner byte source has deflated
+ * data.
+ */
+public class InflaterByteSource extends CloseableByteSource {
+
+  /** The stream factory for the deflated data. */
+  private final CloseableByteSource deflatedSource;
+
+  /**
+   * Creates a new source.
+   *
+   * @param byteSource the factory for deflated data
+   */
+  public InflaterByteSource(CloseableByteSource byteSource) {
+    deflatedSource = byteSource;
+  }
+
+  @Override
+  public InputStream openStream() throws IOException {
+    /*
+     * The extra byte is a dummy byte required by the inflater. Weirdo.
+     * (see the java.util.Inflater documentation). Looks like a hack...
+     * "Oh, I need an extra dummy byte to allow for some... err... optimizations..."
+     */
+    ByteArrayInputStream hackByte = new ByteArrayInputStream(new byte[] {0});
+    return new InflaterInputStream(
+        new SequenceInputStream(deflatedSource.openStream(), hackByte), new Inflater(true));
+  }
+
+  @Override
+  public void innerClose() throws IOException {
+    deflatedSource.close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/LazyDelegateByteSource.java

@@ -0,0 +1,153 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.hash.HashCode;
+import com.google.common.hash.HashFunction;
+import com.google.common.io.ByteProcessor;
+import com.google.common.io.ByteSink;
+import com.google.common.io.ByteSource;
+import com.google.common.io.CharSource;
+import com.google.common.util.concurrent.ListenableFuture;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.OutputStream;
+import java.nio.charset.Charset;
+import java.util.concurrent.ExecutionException;
+
+/**
+ * {@code ByteSource} that delegates all operations to another {@code ByteSource}. The other byte
+ * source, the <em>delegate</em>, may be computed lazily.
+ */
+public class LazyDelegateByteSource extends CloseableByteSource {
+
+  /** Byte source where we delegate operations to. */
+  private final ListenableFuture<CloseableByteSource> delegate;
+
+  /**
+   * Creates a new byte source that delegates operations to the provided source.
+   *
+   * @param delegate the source that will receive all operations
+   */
+  public LazyDelegateByteSource(ListenableFuture<CloseableByteSource> delegate) {
+    this.delegate = delegate;
+  }
+
+  /**
+   * Obtains the delegate future.
+   *
+   * @return the delegate future, that may be computed or not
+   */
+  public ListenableFuture<CloseableByteSource> getDelegate() {
+    return delegate;
+  }
+
+  /**
+   * Obtains the byte source, waiting for the future to be computed.
+   *
+   * @return the byte source
+   * @throws IOException failed to compute the future :)
+   */
+  private CloseableByteSource get() throws IOException {
+    try {
+      CloseableByteSource r = delegate.get();
+      if (r == null) {
+        throw new IOException("Delegate byte source computation resulted in null.");
+      }
+
+      return r;
+    } catch (InterruptedException e) {
+      throw new IOException("Interrupted while waiting for byte source computation.", e);
+    } catch (ExecutionException e) {
+      throw new IOException("Failed to compute byte source.", e);
+    }
+  }
+
+  @Override
+  public CharSource asCharSource(Charset charset) {
+    try {
+      return get().asCharSource(charset);
+    } catch (IOException e) {
+      throw new RuntimeException(e);
+    }
+  }
+
+  @Override
+  public InputStream openBufferedStream() throws IOException {
+    return get().openBufferedStream();
+  }
+
+  @Override
+  public ByteSource slice(long offset, long length) {
+    try {
+      return get().slice(offset, length);
+    } catch (IOException e) {
+      throw new RuntimeException(e);
+    }
+  }
+
+  @Override
+  public boolean isEmpty() throws IOException {
+    return get().isEmpty();
+  }
+
+  @Override
+  public long size() throws IOException {
+    return get().size();
+  }
+
+  @Override
+  public long copyTo(OutputStream output) throws IOException {
+    return get().copyTo(output);
+  }
+
+  @Override
+  public long copyTo(ByteSink sink) throws IOException {
+    return get().copyTo(sink);
+  }
+
+  @Override
+  public byte[] read() throws IOException {
+    return get().read();
+  }
+
+  @Override
+  public <T> T read(ByteProcessor<T> processor) throws IOException {
+    return get().read(processor);
+  }
+
+  @Override
+  public HashCode hash(HashFunction hashFunction) throws IOException {
+    return get().hash(hashFunction);
+  }
+
+  @Override
+  public boolean contentEquals(ByteSource other) throws IOException {
+    return get().contentEquals(other);
+  }
+
+  @Override
+  public InputStream openStream() throws IOException {
+    return get().openStream();
+  }
+
+  @Override
+  public void innerClose() throws IOException {
+    get().close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ProcessedAndRawByteSources.java

@@ -0,0 +1,77 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.io.Closer;
+import java.io.Closeable;
+import java.io.IOException;
+
+/**
+ * Container that has two bytes sources: one representing raw data and another processed data. In
+ * case of compression, the raw data is the compressed data and the processed data is the
+ * uncompressed data. It is valid for a RaP ("Raw-and-Processed") to contain the same byte sources
+ * for both processed and raw data.
+ */
+public class ProcessedAndRawByteSources implements Closeable {
+
+  /** The processed byte source. */
+  private final CloseableByteSource processedSource;
+
+  /** The processed raw source. */
+  private final CloseableByteSource rawSource;
+
+  /**
+   * Creates a new container.
+   *
+   * @param processedSource the processed source
+   * @param rawSource the raw source
+   */
+  public ProcessedAndRawByteSources(
+      CloseableByteSource processedSource, CloseableByteSource rawSource) {
+    this.processedSource = processedSource;
+    this.rawSource = rawSource;
+  }
+
+  /**
+   * Obtains a byte source that read the processed contents of the entry.
+   *
+   * @return a byte source
+   */
+  public CloseableByteSource getProcessedByteSource() {
+    return processedSource;
+  }
+
+  /**
+   * Obtains a byte source that reads the raw contents of an entry. This is the data that is
+   * ultimately stored in the file and, in the case of compressed files, is the same data in the
+   * source returned by {@link #getProcessedByteSource()}.
+   *
+   * @return a byte source
+   */
+  public CloseableByteSource getRawByteSource() {
+    return rawSource;
+  }
+
+  @Override
+  public void close() throws IOException {
+    Closer closer = Closer.create();
+    closer.register(processedSource);
+    closer.register(rawSource);
+    closer.close();
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/StoredEntry.java

@@ -0,0 +1,775 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.bytestorage.CloseableByteSourceFromOutputStreamBuilder;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.annotations.VisibleForTesting;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Supplier;
+import com.google.common.base.Suppliers;
+import com.google.common.base.Verify;
+import com.google.common.io.ByteStreams;
+import com.google.common.primitives.Ints;
+import java.io.BufferedInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.ByteBuffer;
+import java.util.Comparator;
+import javax.annotation.Nullable;
+
+/**
+ * A stored entry represents a file in the zip. The entry may or may not be written to the zip file.
+ *
+ * <p>Stored entries provide the operations that are related to the files themselves, not to the
+ * zip. It is through the {@code StoredEntry} class that entries can be deleted ({@link #delete()},
+ * open ({@link #open()}) or realigned ({@link #realign()}).
+ *
+ * <p>Entries are not created directly. They are created using {@link ZFile#add(String, InputStream,
+ * boolean)} and obtained from the zip file using {@link ZFile#get(String)} or {@link
+ * ZFile#entries()}.
+ *
+ * <p>Most of the data in the an entry is in the Central Directory Header. This includes the name,
+ * compression method, file compressed and uncompressed sizes, CRC32 checksum, etc. The CDH can be
+ * obtained using the {@link #getCentralDirectoryHeader()} method.
+ */
+public class StoredEntry {
+
+  /** Comparator that compares instances of {@link StoredEntry} by their names. */
+  static final Comparator<StoredEntry> COMPARE_BY_NAME =
+      (o1, o2) -> {
+        if (o1 == null && o2 == null) {
+          return 0;
+        }
+
+        if (o1 == null) {
+          return -1;
+        }
+
+        if (o2 == null) {
+          return 1;
+        }
+
+        String name1 = o1.getCentralDirectoryHeader().getName();
+        String name2 = o2.getCentralDirectoryHeader().getName();
+        return name1.compareTo(name2);
+      };
+
+  /** Signature of the data descriptor. */
+  private static final int DATA_DESC_SIGNATURE = 0x08074b50;
+
+  /** Local header field: signature. */
+  private static final ZipField.F4 F_LOCAL_SIGNATURE = new ZipField.F4(0, 0x04034b50, "Signature");
+
+  /** Local header field: version to extract, should match the CDH's. */
+  @VisibleForTesting
+  static final ZipField.F2 F_VERSION_EXTRACT =
+      new ZipField.F2(
+          F_LOCAL_SIGNATURE.endOffset(), "Version to extract", new ZipFieldInvariantNonNegative());
+
+  /** Local header field: GP bit flag, should match the CDH's. */
+  private static final ZipField.F2 F_GP_BIT =
+      new ZipField.F2(F_VERSION_EXTRACT.endOffset(), "GP bit flag");
+
+  /** Local header field: compression method, should match the CDH's. */
+  private static final ZipField.F2 F_METHOD =
+      new ZipField.F2(
+          F_GP_BIT.endOffset(), "Compression method", new ZipFieldInvariantNonNegative());
+
+  /** Local header field: last modification time, should match the CDH's. */
+  private static final ZipField.F2 F_LAST_MOD_TIME =
+      new ZipField.F2(F_METHOD.endOffset(), "Last modification time");
+
+  /** Local header field: last modification time, should match the CDH's. */
+  private static final ZipField.F2 F_LAST_MOD_DATE =
+      new ZipField.F2(F_LAST_MOD_TIME.endOffset(), "Last modification date");
+
+  /** Local header field: CRC32 checksum, should match the CDH's. 0 if there is no data. */
+  private static final ZipField.F4 F_CRC32 = new ZipField.F4(F_LAST_MOD_DATE.endOffset(), "CRC32");
+
+  /** Local header field: compressed size, size the data takes in the zip file. */
+  private static final ZipField.F4 F_COMPRESSED_SIZE =
+      new ZipField.F4(F_CRC32.endOffset(), "Compressed size", new ZipFieldInvariantNonNegative());
+
+  /** Local header field: uncompressed size, size the data takes after extraction. */
+  private static final ZipField.F4 F_UNCOMPRESSED_SIZE =
+      new ZipField.F4(
+          F_COMPRESSED_SIZE.endOffset(), "Uncompressed size", new ZipFieldInvariantNonNegative());
+
+  /** Local header field: length of the file name. */
+  private static final ZipField.F2 F_FILE_NAME_LENGTH =
+      new ZipField.F2(
+          F_UNCOMPRESSED_SIZE.endOffset(), "@File name length", new ZipFieldInvariantNonNegative());
+
+  /** Local header filed: length of the extra field. */
+  private static final ZipField.F2 F_EXTRA_LENGTH =
+      new ZipField.F2(
+          F_FILE_NAME_LENGTH.endOffset(), "Extra length", new ZipFieldInvariantNonNegative());
+
+  /** Local header size (fixed part, not counting file name or extra field). */
+  static final int FIXED_LOCAL_FILE_HEADER_SIZE = F_EXTRA_LENGTH.endOffset();
+
+  /** Type of entry. */
+  private final StoredEntryType type;
+
+  /** The central directory header with information about the file. */
+  private final CentralDirectoryHeader cdh;
+
+  /** The file this entry is associated with */
+  private final ZFile file;
+
+  /** Has this entry been deleted? */
+  private boolean deleted;
+
+  /** Extra field specified in the local directory. */
+  private ExtraField localExtra;
+
+  /** Type of data descriptor associated with the entry. */
+  private Supplier<DataDescriptorType> dataDescriptorType;
+
+  /**
+   * Source for this entry's data. If this entry is a directory, this source has to have zero size.
+   */
+  private ProcessedAndRawByteSources source;
+
+  /** Verify log for the entry. */
+  private final VerifyLog verifyLog;
+
+  /** Storage used to create buffers when loading storage into memory. */
+  private final ByteStorage storage;
+
+  /**
+   * Creates a new stored entry.
+   *
+   * @param header the header with the entry information; if the header does not contain an offset
+   *     it means that this entry is not yet written in the zip file
+   * @param file the zip file containing the entry
+   * @param source the entry's data source; it can be {@code null} only if the source can be read
+   *     from the zip file, that is, if {@code header.getOffset()} is non-negative
+   * @throws IOException failed to create the entry
+   */
+  StoredEntry(
+      CentralDirectoryHeader header,
+      ZFile file,
+      @Nullable ProcessedAndRawByteSources source,
+      ByteStorage storage)
+      throws IOException {
+    cdh = header;
+    this.file = file;
+    deleted = false;
+    verifyLog = file.makeVerifyLog();
+    this.storage = storage;
+
+    if (header.getOffset() >= 0) {
+      readLocalHeader();
+
+      Preconditions.checkArgument(
+          source == null, "Source was defined but contents already exist on file.");
+
+      /*
+       * Since the file is already in the zip, dynamically create a source that will read
+       * the file from the zip when needed. The assignment is not really needed, but we
+       * would get a warning because of the @NotNull otherwise.
+       */
+      this.source = createSourceFromZip(cdh.getOffset());
+    } else {
+      /*
+       * There is no local extra data for new files.
+       */
+      localExtra = new ExtraField();
+
+      Preconditions.checkNotNull(source, "Source was not defined, but contents are not on file.");
+      this.source = source;
+    }
+
+    /*
+     * It seems that zip utilities store directories as names ending with "/".
+     * This seems to be respected by all zip utilities although I could not find there anywhere
+     * in the specification.
+     */
+    if (cdh.getName().endsWith(Character.toString(ZFile.SEPARATOR))) {
+      type = StoredEntryType.DIRECTORY;
+      verifyLog.verify(
+          this.source.getProcessedByteSource().isEmpty(), "Directory source is not empty.");
+      verifyLog.verify(cdh.getCrc32() == 0, "Directory has CRC32 = %s.", cdh.getCrc32());
+      verifyLog.verify(
+          cdh.getUncompressedSize() == 0,
+          "Directory has uncompressed size = %s.",
+          cdh.getUncompressedSize());
+
+      /*
+       * Some clever (OMG!) tools, like jar will actually try to compress the directory
+       * contents and generate a 2 byte compressed data. Of course, the uncompressed size is
+       * zero and we're just wasting space.
+       */
+      long compressedSize = cdh.getCompressionInfoWithWait().getCompressedSize();
+      verifyLog.verify(
+          compressedSize == 0 || compressedSize == 2,
+          "Directory has compressed size = %s.",
+          compressedSize);
+    } else {
+      type = StoredEntryType.FILE;
+    }
+
+    /*
+     * By default we assume there is no data descriptor unless the CRC is marked as deferred
+     * in the header's GP Bit.
+     */
+    dataDescriptorType = Suppliers.ofInstance(DataDescriptorType.NO_DATA_DESCRIPTOR);
+    if (header.getGpBit().isDeferredCrc()) {
+      /*
+       * If the deferred CRC bit exists, then we have an extra descriptor field. This extra
+       * field may have a signature.
+       */
+      Verify.verify(
+          header.getOffset() >= 0,
+          "Files that are not on disk cannot have the " + "deferred CRC bit set.");
+
+      dataDescriptorType =
+          Suppliers.memoize(
+              () -> {
+                try {
+                  return readDataDescriptorRecord();
+                } catch (IOException e) {
+                  throw new IOExceptionWrapper(
+                      new IOException("Failed to read data descriptor record.", e));
+                }
+              });
+    }
+  }
+
+  /**
+   * Obtains the size of the local header of this entry.
+   *
+   * @return the local header size in bytes
+   */
+  public int getLocalHeaderSize() {
+    Preconditions.checkState(!deleted, "deleted");
+    return FIXED_LOCAL_FILE_HEADER_SIZE + cdh.getEncodedFileName().length + localExtra.size();
+  }
+
+  /**
+   * Obtains the size of the whole entry on disk, including local header and data descriptor. This
+   * method will wait until compression information is complete, if needed.
+   *
+   * @return the number of bytes
+   * @throws IOException failed to get compression information
+   */
+  long getInFileSize() throws IOException {
+    Preconditions.checkState(!deleted, "deleted");
+    return cdh.getCompressionInfoWithWait().getCompressedSize()
+        + getLocalHeaderSize()
+        + dataDescriptorType.get().size;
+  }
+
+  /**
+   * Obtains a stream that allows reading from the entry.
+   *
+   * @return a stream that will return as many bytes as the uncompressed entry size
+   * @throws IOException failed to open the stream
+   */
+  public InputStream open() throws IOException {
+    return source.getProcessedByteSource().openStream();
+  }
+
+  /**
+   * Obtains the contents of the file.
+   *
+   * @return a byte array with the contents of the file (uncompressed if the file was compressed)
+   * @throws IOException failed to read the file
+   */
+  public byte[] read() throws IOException {
+    try (InputStream is = new BufferedInputStream(open())) {
+      return ByteStreams.toByteArray(is);
+    }
+  }
+
+  /**
+   * Obtains the contents of the file in an existing buffer.
+   *
+   * @param bytes buffer to read the file contents in.
+   * @return the number of bytes read
+   * @throws IOException failed to read the file.
+   */
+  public int read(byte[] bytes) throws IOException {
+    if (bytes.length < getCentralDirectoryHeader().getUncompressedSize()) {
+      throw new RuntimeException(
+          "Buffer to small while reading {}" + getCentralDirectoryHeader().getName());
+    }
+    try (InputStream is = new BufferedInputStream(open())) {
+      return ByteStreams.read(is, bytes, 0, bytes.length);
+    }
+  }
+
+  /**
+   * Obtains the type of entry.
+   *
+   * @return the type of entry
+   */
+  public StoredEntryType getType() {
+    Preconditions.checkState(!deleted, "deleted");
+    return type;
+  }
+
+  /**
+   * Deletes this entry from the zip file. Invoking this method doesn't update the zip itself. To
+   * eventually write updates to disk, {@link ZFile#update()} must be called.
+   *
+   * @throws IOException failed to delete the entry
+   * @throws IllegalStateException if the zip file was open in read-only mode
+   */
+  public void delete() throws IOException {
+    delete(true);
+  }
+
+  /**
+   * Deletes this entry from the zip file. Invoking this method doesn't update the zip itself. To
+   * eventually write updates to disk, {@link ZFile#update()} must be called.
+   *
+   * @param notify should listeners be notified of the deletion? This will only be {@code false} if
+   *     the entry is being removed as part of a replacement
+   * @throws IOException failed to delete the entry
+   * @throws IllegalStateException if the zip file was open in read-only mode
+   */
+  void delete(boolean notify) throws IOException {
+    Preconditions.checkState(!deleted, "deleted");
+    file.delete(this, notify);
+    deleted = true;
+    source.close();
+  }
+
+  /** Returns {@code true} if this entry has been deleted/replaced. */
+  public boolean isDeleted() {
+    return deleted;
+  }
+
+  /**
+   * Obtains the CDH associated with this entry.
+   *
+   * @return the CDH
+   */
+  public CentralDirectoryHeader getCentralDirectoryHeader() {
+    return cdh;
+  }
+
+  /**
+   * Reads the file's local header and verifies that it matches the Central Directory Header
+   * provided in the constructor. This method should only be called if the entry already exists on
+   * disk; new entries do not have local headers.
+   *
+   * <p>This method will define the {@link #localExtra} field that is only defined in the local
+   * descriptor.
+   *
+   * @throws IOException failed to read the local header
+   */
+  private void readLocalHeader() throws IOException {
+    byte[] localHeader = new byte[FIXED_LOCAL_FILE_HEADER_SIZE];
+    file.directFullyRead(cdh.getOffset(), localHeader);
+
+    CentralDirectoryHeaderCompressInfo compressInfo = cdh.getCompressionInfoWithWait();
+
+    ByteBuffer bytes = ByteBuffer.wrap(localHeader);
+    F_LOCAL_SIGNATURE.verify(bytes);
+    F_VERSION_EXTRACT.verify(bytes, compressInfo.getVersionExtract(), verifyLog);
+    F_GP_BIT.verify(bytes, cdh.getGpBit().getValue(), verifyLog);
+    F_METHOD.verify(bytes, compressInfo.getMethod().methodCode, verifyLog);
+
+    if (file.areTimestampsIgnored()) {
+      F_LAST_MOD_TIME.skip(bytes);
+      F_LAST_MOD_DATE.skip(bytes);
+    } else {
+      F_LAST_MOD_TIME.verify(bytes, cdh.getLastModTime(), verifyLog);
+      F_LAST_MOD_DATE.verify(bytes, cdh.getLastModDate(), verifyLog);
+    }
+
+    /*
+     * If CRC-32, compressed size and uncompressed size are deferred, their values in Local
+     * File Header must be ignored and their actual values must be read from the Data
+     * Descriptor following the contents of this entry. See readDataDescriptorRecord().
+     */
+    if (cdh.getGpBit().isDeferredCrc()) {
+      F_CRC32.skip(bytes);
+      F_COMPRESSED_SIZE.skip(bytes);
+      F_UNCOMPRESSED_SIZE.skip(bytes);
+    } else {
+      F_CRC32.verify(bytes, cdh.getCrc32(), verifyLog);
+      F_COMPRESSED_SIZE.verify(bytes, compressInfo.getCompressedSize(), verifyLog);
+      F_UNCOMPRESSED_SIZE.verify(bytes, cdh.getUncompressedSize(), verifyLog);
+    }
+
+    F_FILE_NAME_LENGTH.verify(bytes, cdh.getEncodedFileName().length);
+    long extraLength = F_EXTRA_LENGTH.read(bytes);
+    long fileNameStart = cdh.getOffset() + F_EXTRA_LENGTH.endOffset();
+    byte[] fileNameData = new byte[cdh.getEncodedFileName().length];
+    file.directFullyRead(fileNameStart, fileNameData);
+
+    String fileName = EncodeUtils.decode(fileNameData, cdh.getGpBit());
+    if (!fileName.equals(cdh.getName())) {
+      verifyLog.log(
+          String.format(
+              "Central directory reports file as being named '%s' but local header"
+                  + "reports file being named '%s'.",
+              cdh.getName(), fileName));
+    }
+
+    long localExtraStart = fileNameStart + cdh.getEncodedFileName().length;
+    byte[] localExtraRaw = new byte[Ints.checkedCast(extraLength)];
+    file.directFullyRead(localExtraStart, localExtraRaw);
+    localExtra = new ExtraField(localExtraRaw);
+  }
+
+  /**
+   * Reads the data descriptor record. This method can only be invoked once it is established that a
+   * data descriptor does exist. It will read the data descriptor and check that the data described
+   * there matches the data provided in the Central Directory.
+   *
+   * <p>This method will set the {@link #dataDescriptorType} field to the appropriate type of data
+   * descriptor record.
+   *
+   * @throws IOException failed to read the data descriptor record
+   */
+  private DataDescriptorType readDataDescriptorRecord() throws IOException {
+    CentralDirectoryHeaderCompressInfo compressInfo = cdh.getCompressionInfoWithWait();
+
+    long ddStart =
+        cdh.getOffset()
+            + FIXED_LOCAL_FILE_HEADER_SIZE
+            + cdh.getName().length()
+            + localExtra.size()
+            + compressInfo.getCompressedSize();
+    byte[] ddData = new byte[DataDescriptorType.DATA_DESCRIPTOR_WITH_SIGNATURE.size];
+    file.directFullyRead(ddStart, ddData);
+
+    ByteBuffer ddBytes = ByteBuffer.wrap(ddData);
+
+    ZipField.F4 signatureField = new ZipField.F4(0, "Data descriptor signature");
+    int cpos = ddBytes.position();
+    long sig = signatureField.read(ddBytes);
+    DataDescriptorType result;
+    if (sig == DATA_DESC_SIGNATURE) {
+      result = DataDescriptorType.DATA_DESCRIPTOR_WITH_SIGNATURE;
+    } else {
+      result = DataDescriptorType.DATA_DESCRIPTOR_WITHOUT_SIGNATURE;
+      ddBytes.position(cpos);
+    }
+
+    ZipField.F4 crc32Field = new ZipField.F4(0, "CRC32");
+    ZipField.F4 compressedField = new ZipField.F4(crc32Field.endOffset(), "Compressed size");
+    ZipField.F4 uncompressedField =
+        new ZipField.F4(compressedField.endOffset(), "Uncompressed size");
+
+    crc32Field.verify(ddBytes, cdh.getCrc32(), verifyLog);
+    compressedField.verify(ddBytes, compressInfo.getCompressedSize(), verifyLog);
+    uncompressedField.verify(ddBytes, cdh.getUncompressedSize(), verifyLog);
+    return result;
+  }
+
+  /**
+   * Creates a new source that reads data from the zip.
+   *
+   * @param zipOffset the offset into the zip file where the data is, must be non-negative
+   * @throws IOException failed to close the old source
+   * @return the created source
+   */
+  private ProcessedAndRawByteSources createSourceFromZip(final long zipOffset) throws IOException {
+    Preconditions.checkArgument(zipOffset >= 0, "zipOffset < 0");
+
+    final CentralDirectoryHeaderCompressInfo compressInfo;
+    try {
+      compressInfo = cdh.getCompressionInfoWithWait();
+    } catch (IOException e) {
+      throw new RuntimeException(
+          "IOException should never occur here because compression "
+              + "information should be immediately available if reading from zip.",
+          e);
+    }
+
+    /*
+     * Create a source that will return whatever is on the zip file.
+     */
+    CloseableByteSource rawContents =
+        new CloseableByteSource() {
+          @Override
+          public long size() throws IOException {
+            return compressInfo.getCompressedSize();
+          }
+
+          @Override
+          public InputStream openStream() throws IOException {
+            Preconditions.checkState(!deleted, "deleted");
+
+            long dataStart = zipOffset + getLocalHeaderSize();
+            long dataEnd = dataStart + compressInfo.getCompressedSize();
+
+            file.openReadOnlyIfClosed();
+            return file.directOpen(dataStart, dataEnd);
+          }
+
+          @Override
+          protected void innerClose() throws IOException {
+            /*
+             * Nothing to do here.
+             */
+          }
+        };
+
+    return createSourcesFromRawContents(rawContents);
+  }
+
+  /**
+   * Creates a {@link ProcessedAndRawByteSources} from the raw data source . The processed source
+   * will either inflate or do nothing depending on the compression information that, at this point,
+   * should already be available
+   *
+   * @param rawContents the raw data to create the source from
+   * @return the sources for this entry
+   */
+  private ProcessedAndRawByteSources createSourcesFromRawContents(CloseableByteSource rawContents) {
+    CentralDirectoryHeaderCompressInfo compressInfo;
+    try {
+      compressInfo = cdh.getCompressionInfoWithWait();
+    } catch (IOException e) {
+      throw new RuntimeException(
+          "IOException should never occur here because compression "
+              + "information should be immediately available if creating from raw "
+              + "contents.",
+          e);
+    }
+
+    CloseableByteSource contents;
+
+    /*
+     * If the contents are deflated, wrap that source in an inflater source so we get the
+     * uncompressed data.
+     */
+    if (compressInfo.getMethod() == CompressionMethod.DEFLATE) {
+      contents = new InflaterByteSource(rawContents);
+    } else {
+      contents = rawContents;
+    }
+
+    return new ProcessedAndRawByteSources(contents, rawContents);
+  }
+
+  /**
+   * Replaces {@link #source} with one that reads file data from the zip file.
+   *
+   * @param zipFileOffset the offset in the zip file where data is written; must be non-negative
+   * @throws IOException failed to replace the source
+   */
+  void replaceSourceFromZip(long zipFileOffset) throws IOException {
+    Preconditions.checkArgument(zipFileOffset >= 0, "zipFileOffset < 0");
+
+    ProcessedAndRawByteSources oldSource = source;
+    source = createSourceFromZip(zipFileOffset);
+    cdh.setOffset(zipFileOffset);
+    oldSource.close();
+  }
+
+  /**
+   * Loads all data in memory and replaces {@link #source} with one that contains all the data in
+   * memory.
+   *
+   * <p>If the entry's contents are already in memory, this call does nothing.
+   *
+   * @throws IOException failed to replace the source
+   */
+  void loadSourceIntoMemory() throws IOException {
+    if (cdh.getOffset() == -1) {
+      /*
+       * No offset in the CDR means data has not been written to disk which, in turn,
+       * means data is already loaded into memory.
+       */
+      return;
+    }
+
+    CloseableByteSourceFromOutputStreamBuilder rawBuilder = storage.makeBuilder();
+    try (InputStream input = source.getRawByteSource().openStream()) {
+      ByteStreams.copy(input, rawBuilder);
+    }
+
+    CloseableByteSource newRaw = rawBuilder.build();
+    ProcessedAndRawByteSources newSources = createSourcesFromRawContents(newRaw);
+
+    try (ProcessedAndRawByteSources oldSource = source) {
+      source = newSources;
+      cdh.setOffset(-1);
+    }
+  }
+
+  /**
+   * Obtains the source data for this entry. This method can only be called for files, it cannot be
+   * called for directories.
+   *
+   * @return the entry source
+   */
+  ProcessedAndRawByteSources getSource() {
+    return source;
+  }
+
+  /**
+   * Obtains the type of data descriptor used in the entry.
+   *
+   * @return the type of data descriptor
+   */
+  public DataDescriptorType getDataDescriptorType() {
+    return dataDescriptorType.get();
+  }
+
+  /**
+   * Removes the data descriptor, if it has one and resets the data descriptor bit in the central
+   * directory header.
+   *
+   * @return was the data descriptor remove?
+   */
+  boolean removeDataDescriptor() {
+    if (dataDescriptorType.get() == DataDescriptorType.NO_DATA_DESCRIPTOR) {
+      return false;
+    }
+
+    dataDescriptorType = Suppliers.ofInstance(DataDescriptorType.NO_DATA_DESCRIPTOR);
+    cdh.resetDeferredCrc();
+    return true;
+  }
+
+  /**
+   * Obtains the local header data.
+   *
+   * @param buffer a buffer to write header data to
+   * @return the header data size
+   * @throws IOException failed to get header byte data
+   */
+  int toHeaderData(byte[] buffer) throws IOException {
+    Preconditions.checkArgument(
+        buffer.length
+            >= F_EXTRA_LENGTH.endOffset() + cdh.getEncodedFileName().length + localExtra.size(),
+        "Buffer should be at least the header size");
+
+    ByteBuffer out = ByteBuffer.wrap(buffer);
+    writeData(out);
+    return out.position();
+  }
+
+  private void writeData(ByteBuffer out) throws IOException {
+    CentralDirectoryHeaderCompressInfo compressInfo = cdh.getCompressionInfoWithWait();
+
+    F_LOCAL_SIGNATURE.write(out);
+    F_VERSION_EXTRACT.write(out, compressInfo.getVersionExtract());
+    F_GP_BIT.write(out, cdh.getGpBit().getValue());
+    F_METHOD.write(out, compressInfo.getMethod().methodCode);
+
+    if (file.areTimestampsIgnored()) {
+      F_LAST_MOD_TIME.write(out, 0);
+      F_LAST_MOD_DATE.write(out, 0);
+    } else {
+      F_LAST_MOD_TIME.write(out, cdh.getLastModTime());
+      F_LAST_MOD_DATE.write(out, cdh.getLastModDate());
+    }
+
+    F_CRC32.write(out, cdh.getCrc32());
+    F_COMPRESSED_SIZE.write(out, compressInfo.getCompressedSize());
+    F_UNCOMPRESSED_SIZE.write(out, cdh.getUncompressedSize());
+    F_FILE_NAME_LENGTH.write(out, cdh.getEncodedFileName().length);
+    F_EXTRA_LENGTH.write(out, localExtra.size());
+
+    out.put(cdh.getEncodedFileName());
+    localExtra.write(out);
+  }
+
+  /**
+   * Requests that this entry be realigned. If this entry is already aligned according to the rules
+   * in {@link ZFile} then this method does nothing. Otherwise it will move the file's data into
+   * memory and place it in a different area of the zip.
+   *
+   * @return has this file been changed? Note that if the entry has not yet been written on the
+   *     file, realignment does not count as a change as nothing needs to be updated in the file;
+   *     also, if the entry has been changed, this object may have been marked as deleted and a new
+   *     stored entry may need to be fetched from the file
+   * @throws IOException failed to realign the entry; the entry may no longer exist in the zip file
+   */
+  public boolean realign() throws IOException {
+    Preconditions.checkState(!deleted, "Entry has been deleted.");
+
+    return file.realign(this);
+  }
+
+  /**
+   * Obtains the contents of the local extra field.
+   *
+   * @return the contents of the local extra field
+   */
+  public ExtraField getLocalExtra() {
+    return localExtra;
+  }
+
+  /**
+   * Sets the contents of the local extra field.
+   *
+   * @param localExtra the contents of the local extra field
+   * @throws IOException failed to update the zip file
+   */
+  public void setLocalExtra(ExtraField localExtra) throws IOException {
+    boolean resized = setLocalExtraNoNotify(localExtra);
+    file.localHeaderChanged(this, resized);
+  }
+
+  /**
+   * Sets the contents of the local extra field, does not notify the {@link ZFile} of the change.
+   * This is used internally when the {@link ZFile} itself wants to change the local extra and
+   * doesn't need the callback.
+   *
+   * @param localExtra the contents of the local extra field
+   * @return has the local header size changed?
+   * @throws IOException failed to load the file
+   */
+  boolean setLocalExtraNoNotify(ExtraField localExtra) throws IOException {
+    boolean sizeChanged;
+
+    /*
+     * Make sure we load into memory.
+     *
+     * If we change the size of the local header, the actual start of the file changes
+     * according to our in-memory structures so, if we don't read the file now, we won't be
+     * able to load it later :)
+     *
+     * But, even if the size doesn't change, we need to read it force the entry to be
+     * rewritten otherwise the changes in the local header aren't written. Of course this case
+     * may be optimized with some extra complexity added :)
+     */
+    loadSourceIntoMemory();
+
+    if (this.localExtra.size() != localExtra.size()) {
+      sizeChanged = true;
+    } else {
+      sizeChanged = false;
+    }
+
+    this.localExtra = localExtra;
+    return sizeChanged;
+  }
+
+  /**
+   * Obtains the verify log for the entry.
+   *
+   * @return the verify log
+   */
+  public VerifyLog getVerifyLog() {
+    return verifyLog;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/StoredEntryType.java

@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/** Type of stored entry. */
+public enum StoredEntryType {
+  /** Entry is a file. */
+  FILE,
+
+  /** Entry is a directory. */
+  DIRECTORY
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/VerifyLog.java

@@ -0,0 +1,54 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.google.common.collect.ImmutableList;
+
+/**
+ * The verify log contains verification messages. It is used to capture validation issues with a zip
+ * file or with parts of a zip file.
+ */
+public interface VerifyLog {
+
+  /**
+   * Logs a message.
+   *
+   * @param message the message to verify
+   */
+  void log(String message);
+
+  /**
+   * Obtains all save logged messages.
+   *
+   * @return the logged messages
+   */
+  ImmutableList<String> getLogs();
+
+  /**
+   * Performs verification of a non-critical condition, logging a message if the condition is not
+   * verified.
+   *
+   * @param condition the condition
+   * @param message the message to write if {@code condition} is {@code false}.
+   * @param args arguments for formatting {@code message} using {@code String.format}
+   */
+  default void verify(boolean condition, String message, Object... args) {
+    if (!condition) {
+      log(String.format(message, args));
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/VerifyLogs.java

@@ -0,0 +1,67 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.google.common.collect.ImmutableList;
+import java.util.ArrayList;
+import java.util.List;
+
+/** Factory for verification logs. */
+final class VerifyLogs {
+
+  private VerifyLogs() {}
+
+  /**
+   * Creates a {@link VerifyLog} that ignores all messages logged.
+   *
+   * @return the log
+   */
+  static VerifyLog devNull() {
+    return new VerifyLog() {
+      @Override
+      public void log(String message) {}
+
+      @Override
+      public ImmutableList<String> getLogs() {
+        return ImmutableList.of();
+      }
+    };
+  }
+
+  /**
+   * Creates a {@link VerifyLog} that stores all log messages.
+   *
+   * @return the log
+   */
+  static VerifyLog unlimited() {
+    return new VerifyLog() {
+
+      /** All saved messages. */
+      private final List<String> messages = new ArrayList<>();
+
+      @Override
+      public void log(String message) {
+        messages.add(message);
+      }
+
+      @Override
+      public ImmutableList<String> getLogs() {
+        return ImmutableList.copyOf(messages);
+      }
+    };
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZFileExtension.java

@@ -0,0 +1,140 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.utils.IOExceptionRunnable;
+import java.io.IOException;
+import javax.annotation.Nullable;
+
+/**
+ * An extension of a {@link ZFile}. Extensions are notified when files are open, updated, closed and
+ * when files are added or removed from the zip. These notifications are received after the zip has
+ * been updated in memory for open, when files are added or removed and when the zip has been
+ * updated on disk or closed.
+ *
+ * <p>An extension is also notified before the file is updated, allowing it to modify the file
+ * before the update happens. If it does, then all extensions are notified of the changes on the zip
+ * file. Because the order of the notifications is preserved, all extensions are notified in the
+ * same order. For example, if two extensions E1 and E2 are registered and they both add a file at
+ * update time, this would be the flow:
+ *
+ * <ul>
+ *   <li>E1 receives {@code beforeUpdate} notification.
+ *   <li>E1 adds file F1 to the zip (notifying the addition is suspended because another
+ *       notification is in progress).
+ *   <li>E2 receives {@code beforeUpdate} notification.
+ *   <li>E2 adds file F2 to the zip (notifying the addition is suspended because another
+ *       notification is in progress).
+ *   <li>E1 is notified F1 was added.
+ *   <li>E2 is notified F1 was added.
+ *   <li>E1 is notified F2 was added.
+ *   <li>E2 is notified F2 was added.
+ *   <li>(zip file is updated on disk)
+ *   <li>E1 is notified the zip was updated.
+ *   <li>E2 is notified the zip was updated.
+ * </ul>
+ *
+ * <p>An extension should not modify the zip file when notified of changes. If allowed, this would
+ * break event notification order in case multiple extensions are registered with the zip file. To
+ * allow performing changes to the zip file, all notification method return a {@code
+ * IOExceptionRunnable} that is invoked when {@link ZFile} has finished notifying all extensions.
+ */
+public abstract class ZFileExtension {
+
+  /**
+   * The zip file has been open and the zip's contents have been read. The default implementation
+   * does nothing and returns {@code null}.
+   *
+   * @return an optional runnable to run when notification of all listeners has ended
+   * @throws IOException failed to process the event
+   */
+  @Nullable
+  public IOExceptionRunnable open() throws IOException {
+    return null;
+  }
+
+  /**
+   * The zip will be updated. This method allows the extension to register changes to the zip file
+   * before the file is written. The default implementation does nothing and returns {@code null}.
+   *
+   * <p>After this notification is received, the extension will receive further {@link
+   * #added(StoredEntry, StoredEntry)} and {@link #removed(StoredEntry)} notifications if it or
+   * other extensions add or remove files before update.
+   *
+   * <p>When no more files are updated, the {@link #entriesWritten()} notification is sent.
+   *
+   * @return an optional runnable to run when notification of all listeners has ended
+   * @throws IOException failed to process the event
+   */
+  @Nullable
+  public IOExceptionRunnable beforeUpdate() throws IOException {
+    return null;
+  }
+
+  /**
+   * This notification is sent when all entries have been written in the file but the central
+   * directory and the EOCD have not yet been written. No entries should be added, removed or
+   * updated during this notification. If this method forces an update of either the central
+   * directory or EOCD, then this method will be invoked again for all extensions with the new
+   * central directory and EOCD.
+   *
+   * <p>After this notification, {@link #updated()} is sent.
+   *
+   * @throws IOException failed to process the event
+   */
+  public void entriesWritten() throws IOException {}
+
+  /**
+   * The zip file has been updated on disk. The default implementation does nothing.
+   *
+   * @throws IOException failed to perform update tasks
+   */
+  public void updated() throws IOException {}
+
+  /**
+   * The zip file has been closed. Note that if {@link ZFile#close()} requires that the zip file be
+   * updated (because it had in-memory changes), {@link #updated()} will be called before this
+   * method. The default implementation does nothing.
+   */
+  public void closed() {}
+
+  /**
+   * A new entry has been added to the zip, possibly replacing an entry in there. The default
+   * implementation does nothing and returns {@code null}.
+   *
+   * @param entry the entry that was added
+   * @param replaced the entry that was replaced, if any
+   * @return an optional runnable to run when notification of all listeners has ended
+   */
+  @Nullable
+  public IOExceptionRunnable added(StoredEntry entry, @Nullable StoredEntry replaced) {
+    return null;
+  }
+
+  /**
+   * An entry has been removed from the zip. This method is not invoked for entries that have been
+   * replaced. Those entries are notified using <em>replaced</em> in {@link #added(StoredEntry,
+   * StoredEntry)}. The default implementation does nothing and returns {@code null}.
+   *
+   * @param entry the entry that was deleted
+   * @return an optional runnable to run when notification of all listeners has ended
+   */
+  @Nullable
+  public IOExceptionRunnable removed(StoredEntry entry) {
+    return null;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZFileOptions.java

@@ -0,0 +1,260 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorageFactory;
+import com.android.tools.build.apkzlib.bytestorage.ChunkBasedByteStorageFactory;
+import com.android.tools.build.apkzlib.bytestorage.OverflowToDiskByteStorageFactory;
+import com.android.tools.build.apkzlib.bytestorage.TemporaryDirectory;
+import com.android.tools.build.apkzlib.zip.compress.DeflateExecutionCompressor;
+import com.android.tools.build.apkzlib.zip.utils.ByteTracker;
+import com.google.common.base.Supplier;
+import java.util.zip.Deflater;
+
+/** Options to create a {@link ZFile}. */
+public class ZFileOptions {
+
+  /** The storage to use. */
+  private ByteStorageFactory storageFactory;
+
+  /** The compressor to use. */
+  private Compressor compressor;
+
+  /** Should timestamps be zeroed? */
+  private boolean noTimestamps;
+
+  /** The alignment rule to use. */
+  private AlignmentRule alignmentRule;
+
+  /** Should the extra field be used to cover empty space? */
+  private boolean coverEmptySpaceUsingExtraField;
+
+  /** Should files be automatically sorted before update? */
+  private boolean autoSortFiles;
+
+  /**
+   * Skip expensive validation during {@link ZFile} creation?
+   *
+   * <p>During incremental build we are absolutely sure that the zip file is valid, so we do not
+   * have to spend time verifying different fields (some of these checks are relatively expensive
+   * and should be skipped if possible for performance)
+   */
+  private boolean skipValidation;
+
+  /** Factory creating verification logs to use. */
+  private Supplier<VerifyLog> verifyLogFactory;
+
+  /**
+   * Whether to always generate the MANIFEST.MF file regardless whether the APK will be signed with
+   * v1 signing scheme (i.e. jar signing).
+   */
+  private boolean alwaysGenerateJarManifest;
+
+  /** Creates a new options object. All options are set to their defaults. */
+  public ZFileOptions() {
+    storageFactory =
+        new ChunkBasedByteStorageFactory(
+            new OverflowToDiskByteStorageFactory(TemporaryDirectory::newSystemTemporaryDirectory));
+    compressor = new DeflateExecutionCompressor(Runnable::run, Deflater.DEFAULT_COMPRESSION);
+    alignmentRule = AlignmentRules.compose();
+    verifyLogFactory = VerifyLogs::devNull;
+
+    // We set this to true because many utilities stream the zip and expect no space between entries
+    // in the zip file.
+    coverEmptySpaceUsingExtraField = true;
+    skipValidation = false;
+    // True by default for backwards compatibility.
+    alwaysGenerateJarManifest = true;
+  }
+
+  /**
+   * Obtains the ZFile's byte storage factory.
+   *
+   * @return the factory used to create byte storages used to store data
+   */
+  public ByteStorageFactory getStorageFactory() {
+    return storageFactory;
+  }
+
+  @Deprecated
+  public ByteTracker getTracker() {
+    return new ByteTracker();
+  }
+
+  /**
+   * Sets the byte storage factory to use.
+   *
+   * @param storage the factory to use to create storage for new instances of {@link ZFile} created
+   *     for these options.
+   */
+  public ZFileOptions setStorageFactory(ByteStorageFactory storage) {
+    this.storageFactory = storage;
+    return this;
+  }
+
+  /**
+   * Obtains the compressor to use.
+   *
+   * @return the compressor
+   */
+  public Compressor getCompressor() {
+    return compressor;
+  }
+
+  /**
+   * Sets the compressor to use.
+   *
+   * @param compressor the compressor
+   */
+  public ZFileOptions setCompressor(Compressor compressor) {
+    this.compressor = compressor;
+    return this;
+  }
+
+  /**
+   * Obtains whether timestamps should be zeroed.
+   *
+   * @return should timestamps be zeroed?
+   */
+  public boolean getNoTimestamps() {
+    return noTimestamps;
+  }
+
+  /**
+   * Sets whether timestamps should be zeroed.
+   *
+   * @param noTimestamps should timestamps be zeroed?
+   */
+  public ZFileOptions setNoTimestamps(boolean noTimestamps) {
+    this.noTimestamps = noTimestamps;
+    return this;
+  }
+
+  /**
+   * Obtains the alignment rule.
+   *
+   * @return the alignment rule
+   */
+  public AlignmentRule getAlignmentRule() {
+    return alignmentRule;
+  }
+
+  /**
+   * Sets the alignment rule.
+   *
+   * @param alignmentRule the alignment rule
+   */
+  public ZFileOptions setAlignmentRule(AlignmentRule alignmentRule) {
+    this.alignmentRule = alignmentRule;
+    return this;
+  }
+
+  /**
+   * Obtains whether the extra field should be used to cover empty spaces. See {@link ZFile} for an
+   * explanation on using the extra field for covering empty spaces.
+   *
+   * @return should the extra field be used to cover empty spaces?
+   */
+  public boolean getCoverEmptySpaceUsingExtraField() {
+    return coverEmptySpaceUsingExtraField;
+  }
+
+  /**
+   * Sets whether the extra field should be used to cover empty spaces. See {@link ZFile} for an
+   * explanation on using the extra field for covering empty spaces.
+   *
+   * @param coverEmptySpaceUsingExtraField should the extra field be used to cover empty spaces?
+   */
+  public ZFileOptions setCoverEmptySpaceUsingExtraField(boolean coverEmptySpaceUsingExtraField) {
+    this.coverEmptySpaceUsingExtraField = coverEmptySpaceUsingExtraField;
+    return this;
+  }
+
+  /**
+   * Obtains whether files should be automatically sorted before updating the zip file. See {@link
+   * ZFile} for an explanation on automatic sorting.
+   *
+   * @return should the file be automatically sorted?
+   */
+  public boolean getAutoSortFiles() {
+    return autoSortFiles;
+  }
+
+  /**
+   * Sets whether files should be automatically sorted before updating the zip file. See {@link
+   * ZFile} for an explanation on automatic sorting.
+   *
+   * @param autoSortFiles should the file be automatically sorted?
+   */
+  public ZFileOptions setAutoSortFiles(boolean autoSortFiles) {
+    this.autoSortFiles = autoSortFiles;
+    return this;
+  }
+
+  /**
+   * Sets the verification log factory.
+   *
+   * @param verifyLogFactory verification log factory
+   */
+  public ZFileOptions setVerifyLogFactory(Supplier<VerifyLog> verifyLogFactory) {
+    this.verifyLogFactory = verifyLogFactory;
+    return this;
+  }
+
+  /**
+   * Obtains the verification log factory. By default, the verification log doesn't store anything
+   * and will always return an empty log.
+   *
+   * @return the verification log factory
+   */
+  public Supplier<VerifyLog> getVerifyLogFactory() {
+    return verifyLogFactory;
+  }
+
+  /**
+   * Sets whether expensive validation should be skipped during {@link ZFile} creation
+   *
+   * @param skipValidation during creation?
+   */
+  public ZFileOptions setSkipValidation(boolean skipValidation) {
+    this.skipValidation = skipValidation;
+    return this;
+  }
+
+  /**
+   * Gets whether expensive validation should be performed during {@link ZFile} creation
+   *
+   * @return skip verification during creation?
+   */
+  public boolean getSkipValidation() {
+    return skipValidation;
+  }
+
+  /**
+   * Sets whether to always generate the MANIFEST.MF file, regardless whether the APK is signed with
+   * v1 signing scheme.
+   */
+  public ZFileOptions setAlwaysGenerateJarManifest(boolean alwaysGenerateJarManifest) {
+    this.alwaysGenerateJarManifest = alwaysGenerateJarManifest;
+    return this;
+  }
+
+  /** Returns whether the MANIFEST.MF file should always be generated. */
+  public boolean getAlwaysGenerateJarManifest() {
+    return alwaysGenerateJarManifest;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/Zip64Eocd.java

@@ -0,0 +1,408 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.utils.CachedSupplier;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.google.common.primitives.Ints;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+
+/**
+ * Zip64 End of Central Directory record in a zip file.
+ */
+public class Zip64Eocd {
+
+  /**
+   * Default "version made by" field: upper byte needs to be 0 to set to MS-DOS compatibility. Lower
+   * byte can be anything, really. We use 0x18 because aapt uses 0x17 :)
+   */
+  private static final int DEFAULT_VERSION_MADE_BY = 0x0018;
+
+  /**
+   * Minimum size that can be stored in the {@link #F_EOCD_SIZE} field of the record.
+   */
+  private static final int MIN_EOCD_SIZE = 44;
+
+  /** Field in the record: the record signature, fixed at this value by the specification */
+  private static final ZipField.F4 F_SIGNATURE =
+      new ZipField.F4(0, 0x06064b50, "Zip64 EOCD signature");
+
+  /**
+   * Field in the record: the size of the central directory record, not including the first 12
+   * bytes of data (the signature and this size information). Therefore this variable should be:
+   *
+   * <code>size = sizeOfFixedFields + sizeOfVariableData - 12</code>
+   *
+   * as specified by the zip specification.
+   */
+  private static final ZipField.F8 F_EOCD_SIZE =
+      new ZipField.F8(
+          F_SIGNATURE.endOffset(),
+          "Zip64 EOCD size",
+          new ZipFieldInvariantMinValue(MIN_EOCD_SIZE));
+
+  /** Field in the record: ID program that made the zip (we don't actually use this). */
+  private static final ZipField.F2 F_MADE_BY =
+      new ZipField.F2(F_EOCD_SIZE.endOffset(), "Made by", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the record: Version needed to extract the Zip. We expect this value to be at least
+   * {@link CentralDirectoryHeaderCompressInfo#VERSION_WITH_ZIP64_EXTENSIONS}. This value also
+   * determines whether we are using Version 1 or Version 2 of the Zip64 EOCD record.
+   */
+  private static final ZipField.F2 F_VERSION_EXTRACT =
+      new ZipField.F2(
+          F_MADE_BY.endOffset(),
+          "Version to extract",
+          new ZipFieldInvariantMinValue(
+              CentralDirectoryHeaderCompressInfo.VERSION_WITH_ZIP64_EXTENSIONS));
+
+  /**
+   * Field in the record: the number of disk where the Zip64 EOCD is located. It must be zero
+   * as multi-file archives are not supported.
+   */
+  private static final ZipField.F4 F_NUMBER_OF_DISK =
+      new ZipField.F4(F_VERSION_EXTRACT.endOffset(), 0, "Number of this disk");
+
+  /**
+   * Field in the record: the number of the disk where the central directory resides. This must be
+   * zero as multi-file archives are not supported.
+   */
+  private static final ZipField.F4 F_DISK_CD_START =
+      new ZipField.F4(F_NUMBER_OF_DISK.endOffset(), 0, "Disk where CD starts");
+
+  /**
+   * Field in the record: the number of entries in the Central Directory on this disk. Because we do
+   * not support multi-file archives, this is the same as {@link #F_RECORDS_TOTAL}
+   */
+  private static final ZipField.F8 F_RECORDS_DISK =
+      new ZipField.F8(
+          F_DISK_CD_START.endOffset(),
+          "Record on disk count",
+          new ZipFieldInvariantNonNegative());
+
+  /** Field in the record: the total number of entries in the Central Directory. */
+  private static final ZipField.F8 F_RECORDS_TOTAL =
+      new ZipField.F8(
+          F_RECORDS_DISK.endOffset(),
+          "Total records",
+          new ZipFieldInvariantNonNegative());
+
+  /** Field in the record: number of bytes of the Central Directory. */
+  private static final ZipField.F8 F_CD_SIZE =
+      new ZipField.F8(
+          F_RECORDS_TOTAL.endOffset(), "Directory size", new ZipFieldInvariantNonNegative());
+
+  /** Field in the record: offset, from the archive start, where the Central Directory starts. */
+  private static final ZipField.F8 F_CD_OFFSET =
+      new ZipField.F8(
+          F_CD_SIZE.endOffset(), "Directory offset", new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in Version 2 of the record: The compression method used for the Central Directory in the
+   * given Zip file. Although we do support version 2 of the Zip64 EOCD, we presently do not support
+   * any compression method, and thus this value must be zero.
+   */
+  private static final ZipField.F2 F_V2_CD_COMPRESSION_METHOD =
+      new ZipField.F2(
+          F_CD_OFFSET.endOffset(), 0, "Version 2: Directory Compression method");
+
+  /**
+   * Field in Version 2 of the record: The compressed size of the Central Directory. As Compression
+   * is not supported for the CD, this value should always be the same as
+   * {@link #F_V2_CD_UNCOMPRESSED_SIZE}.
+   */
+  private static final ZipField.F8 F_V2_CD_COMPRESSED_SIZE =
+      new ZipField.F8(
+          F_V2_CD_COMPRESSION_METHOD.endOffset(),
+          "Version 2: Directory Compressed Size",
+          new ZipFieldInvariantNonNegative());
+
+  /** Field in Version 2 of the record: The uncompressed size of the Central Directory. */
+  private static final ZipField.F8 F_V2_CD_UNCOMPRESSED_SIZE =
+      new ZipField.F8(
+          F_V2_CD_COMPRESSED_SIZE.endOffset(),
+          "Version 2: Directory Uncompressed Size",
+          new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in Version 2 of the record: The ID for the type of encryption used to encrypt the Central
+   * directory. Since Central Directory encryption is not supported, this value has to be zero.
+   */
+  private static final ZipField.F2 F_V2_CD_ENCRYPTION_ID =
+      new ZipField.F2(
+          F_V2_CD_UNCOMPRESSED_SIZE.endOffset(),
+          0,
+          "Version 2: Directory Encryption");
+
+  /**
+   * Field in Version 2 of the record: The length of the encryption key for the encryption of the
+   * Central Directory given by {@link #F_V2_CD_ENCRYPTION_ID}. Since encryption of the Central
+   * Directory is not supported, this value has to be zero.
+   */
+  private static final ZipField.F2 F_V2_CD_ENCRYPTION_KEY_LENGTH =
+      new ZipField.F2(
+          F_V2_CD_ENCRYPTION_ID.endOffset(),
+          0,
+          "Version 2: Directory Encryption key length");
+
+  /**
+   * Field in Version 2 of the record: The flags for the encryption method used on the Central
+   * Directory. As encryption of the Central Directory is not supported, this value has to be zero.
+   */
+  private static final ZipField.F2 F_V2_CD_ENCRYPTION_FLAGS =
+      new ZipField.F2(
+          F_V2_CD_ENCRYPTION_KEY_LENGTH.endOffset(),
+          0,
+          "Version 2: Directory Encryption Flags");
+
+  /**
+   * Field in Version 2 of the record: ID of the algorithm used to hash the Central Directory data.
+   * Hashing of the Central Directory is not supported, so this value has to be zero.
+   */
+  private static final ZipField.F2 F_V2_HASH_ID =
+      new ZipField.F2(
+          F_V2_CD_ENCRYPTION_FLAGS.endOffset(),
+          0,
+          "Version 2: Hash algorithm ID");
+
+  /**
+   * Field in Version 2 of the record: Length of the data for the hash of the Central Directory.
+   * Hashing of the Central Directory is not supported, so this value has to be zero.
+   */
+  private static final ZipField.F2 F_V2_HASH_LENGTH =
+      new ZipField.F2(
+          F_V2_HASH_ID.endOffset(),
+          0,
+          "Version 2: Hash length");
+
+  /** The location of the Zip64 size field relative to the start of the Zip64 EOCD. */
+  public static final int SIZE_OFFSET = F_EOCD_SIZE.offset();
+
+  /**
+   * The difference between the size in the size field and the true size of the Zip64 EOCD. The size
+   * field in the EOCD does not consider the size field and the identifier field when calculating
+   * the size of the Zip64 EOCD record.
+   */
+  public static final int TRUE_SIZE_DIFFERENCE = F_EOCD_SIZE.endOffset();
+
+  /** Code of the program that made the zip. We actually don't care about this. */
+  private final long madeBy;
+
+  /** Version needed to extract the zip. */
+  private final long versionToExtract;
+
+  /** Number of entries in the Central Directory. */
+  private final long totalRecords;
+
+  /** Offset from the beginning of the archive where the Central Directory is located. */
+  private final long directoryOffset;
+
+  /** Number of bytes of the Central Directory. */
+  private final long directorySize;
+
+  /** The variable extra fields at the end of the Zip64 EOCD (in both Version 1 and 2). */
+  private final Zip64ExtensibleDataSector extraFields;
+
+  /** Supplier of the byte representation of the Zip64 EOCD. */
+  private final CachedSupplier<byte[]> byteSupplier;
+
+  /**
+   * Creates a Zip64Eocd record from the given information from the central directory record.
+   *
+   * @param totalRecords the number of entries in the central directory.
+   * @param directoryOffset the offset of the central directory from the start of the archive.
+   * @param directorySize the size (in bytes) of the central directory record.
+   * @param useVersion2 whether we want to use Version 2 of the Zip64 EOCD.
+   * @param dataSector the extensible data sector.
+   */
+  Zip64Eocd(
+      long totalRecords,
+      long directoryOffset,
+      long directorySize,
+      boolean useVersion2,
+      Zip64ExtensibleDataSector dataSector) {
+    this.madeBy = DEFAULT_VERSION_MADE_BY;
+    this.totalRecords = totalRecords;
+    this.directorySize = directorySize;
+    this.directoryOffset = directoryOffset;
+    this.versionToExtract =
+        useVersion2
+            ? CentralDirectoryHeaderCompressInfo.VERSION_WITH_CENTRAL_FILE_ENCRYPTION
+            : CentralDirectoryHeaderCompressInfo.VERSION_WITH_ZIP64_EXTENSIONS;
+    extraFields = dataSector;
+
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * Creates a Zip64 EOCD from the given byte information. It does verify that the record starts
+   * with the correct header information.
+   *
+   * @param bytes the bytes to be read as a Zip64 EOCD
+   * @throws IOException the bytes could not be read as a Zip64 EOCD
+   */
+  Zip64Eocd(ByteBuffer bytes) throws IOException {
+
+    F_SIGNATURE.verify(bytes);
+    long eocdSize = F_EOCD_SIZE.read(bytes);
+    long madeBy = F_MADE_BY.read(bytes);
+    long versionToExtract = F_VERSION_EXTRACT.read(bytes);
+    F_NUMBER_OF_DISK.verify(bytes);
+    F_DISK_CD_START.verify(bytes);
+    long totalRecords1 = F_RECORDS_DISK.read(bytes);
+    long totalRecords2 = F_RECORDS_TOTAL.read(bytes);
+    long directorySize = F_CD_SIZE.read(bytes);
+    long directoryOffset = F_CD_OFFSET.read(bytes);
+    long sizeOfFixedFields = F_CD_OFFSET.endOffset();
+
+    // sanity checks for Version 1 fields.
+    if (totalRecords1 != totalRecords2) {
+      throw new IOException(
+          "Zip states records split in multiple disks, which is not supported");
+    }
+
+    // read Version 2 fields if necessary
+    if (versionToExtract
+        >= CentralDirectoryHeaderCompressInfo.VERSION_WITH_CENTRAL_FILE_ENCRYPTION) {
+      if (eocdSize < F_V2_HASH_LENGTH.endOffset() - F_EOCD_SIZE.endOffset()) {
+        throw new IOException(
+            "Zip states the size of Zip64 EOCD is too small for version 2 format.");
+      }
+
+      F_V2_CD_COMPRESSION_METHOD.verify(bytes);
+      long compressedSize = F_V2_CD_COMPRESSED_SIZE.read(bytes);
+      long uncompressedSize = F_V2_CD_UNCOMPRESSED_SIZE.read(bytes);
+      F_V2_CD_ENCRYPTION_ID.verify(bytes);
+      F_V2_CD_ENCRYPTION_KEY_LENGTH.verify(bytes);
+      F_V2_CD_ENCRYPTION_FLAGS.verify(bytes);
+      F_V2_HASH_ID.verify(bytes);
+      F_V2_HASH_LENGTH.verify(bytes);
+      sizeOfFixedFields = F_V2_HASH_LENGTH.endOffset();
+
+      // sanity checks for version 2 fields.
+      if (compressedSize != uncompressedSize) {
+        throw new IOException(
+            "Zip states Central Directory Compression is used, which is not supported");
+      }
+      directorySize = uncompressedSize;
+    }
+
+    this.madeBy = madeBy;
+    this.versionToExtract = versionToExtract;
+    this.totalRecords = totalRecords1;
+    this.directorySize = directorySize;
+    this.directoryOffset = directoryOffset;
+
+    long extensibleDataSize = eocdSize - (sizeOfFixedFields - F_EOCD_SIZE.endOffset());
+
+    if (extensibleDataSize > Integer.MAX_VALUE) {
+      throw new IOException("Extensible data of size: " + extensibleDataSize + "not supported");
+    }
+    byte[] rawData = new byte[Ints.checkedCast(extensibleDataSize)];
+    bytes.get(rawData);
+    extraFields = new Zip64ExtensibleDataSector(rawData);
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * The size of the fixed field in the Zip64 EOCD. This vaue may be different if we are using a
+   * version 1 or version 2 record.
+   *
+   * @return the size of the fixed fields.
+   */
+  private int sizeOfFixedFields() {
+    return versionToExtract
+        >= CentralDirectoryHeaderCompressInfo.VERSION_WITH_CENTRAL_FILE_ENCRYPTION
+          ? F_V2_HASH_LENGTH.endOffset()
+          : F_CD_OFFSET.endOffset();
+  }
+
+  /**
+   * Gets the size (in bytes) of the Zip64 EOCD record.
+   *
+   * @return the size of the record.
+   */
+  public int size() {
+    return sizeOfFixedFields() + extraFields.size();
+  }
+
+  public long getTotalRecords() {
+    return totalRecords;
+  }
+
+  public long getDirectorySize() {
+    return directorySize;
+  }
+
+  public long getDirectoryOffset() {
+    return directoryOffset;
+  }
+
+  public Zip64ExtensibleDataSector getExtraFields() {
+    return extraFields;
+  }
+
+  public long getVersionToExtract() { return versionToExtract; }
+
+  /**
+   * Gets the byte representation of The Zip64 EOCD record.
+   *
+   * @return the bytes of the EOCD.
+   */
+  public byte[] toBytes() {
+    return byteSupplier.get();
+  }
+
+  private byte[] computeByteRepresentation() {
+    int size = size();
+    ByteBuffer out = ByteBuffer.allocate(size);
+
+    try {
+      F_SIGNATURE.write(out);
+      F_EOCD_SIZE.write(out, size - F_EOCD_SIZE.endOffset());
+      F_MADE_BY.write(out, madeBy);
+      F_VERSION_EXTRACT.write(out, versionToExtract);
+      F_NUMBER_OF_DISK.write(out);
+      F_DISK_CD_START.write(out);
+      F_RECORDS_DISK.write(out, totalRecords);
+      F_RECORDS_TOTAL.write(out, totalRecords);
+      F_CD_SIZE.write(out, directorySize);
+      F_CD_OFFSET.write(out, directoryOffset);
+
+      // write version 2 fields if necessary.
+      if (versionToExtract
+          >= CentralDirectoryHeaderCompressInfo.VERSION_WITH_CENTRAL_FILE_ENCRYPTION) {
+        F_V2_CD_COMPRESSION_METHOD.write(out);
+        F_V2_CD_COMPRESSED_SIZE.write(out, directorySize);
+        F_V2_CD_UNCOMPRESSED_SIZE.write(out, directorySize);
+        F_V2_CD_ENCRYPTION_ID.write(out);
+        F_V2_CD_ENCRYPTION_KEY_LENGTH.write(out);
+        F_V2_CD_ENCRYPTION_FLAGS.write(out);
+        F_V2_HASH_ID.write(out);
+        F_V2_HASH_LENGTH.write(out);
+      }
+
+      extraFields.write(out);
+
+      return out.array();
+    } catch (IOException e) {
+      throw new IOExceptionWrapper(e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/Zip64EocdLocator.java

@@ -0,0 +1,155 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.utils.CachedSupplier;
+import com.android.tools.build.apkzlib.utils.IOExceptionWrapper;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import java.io.IOException;
+import java.io.UncheckedIOException;
+import java.nio.ByteBuffer;
+
+/**
+ * Zip64 End of Central Directory Locator. Used to locate the Zip64 EOCD record in
+ * the Zip64 format. This will be located right above the standard EOCD record, if it exists.
+ */
+class Zip64EocdLocator {
+  /** Field in the record: the record signature, fixed at this value by the specification. */
+  private static final ZipField.F4 F_SIGNATURE =
+      new ZipField.F4(0, 0x07064b50, "Zip64 EOCD Locator signature");
+
+  /**
+   * Field in the record: the number of the disk where the Zip64 EOCD is located. This has to be
+   * zero because multi-file archives are not supported.
+   */
+  private static final ZipField.F4 F_NUMBER_OF_DISK =
+      new ZipField.F4(F_SIGNATURE.endOffset(), 0, "Number of disk with Zip64 EOCD");
+
+  /**
+   * Field in the record: the location of the zip64 EOCD record on the disk specified by
+   * {@link #F_NUMBER_OF_DISK}.
+   */
+  private static final ZipField.F8 F_Z64_EOCD_OFFSET =
+      new ZipField.F8(
+          F_NUMBER_OF_DISK.endOffset(),
+          "Offset of Zip64 EOCD",
+          new ZipFieldInvariantNonNegative());
+
+  /**
+   * Field in the record: the total number of disks in the archive. This has to be zero because
+   * multi-file archives are not supported.
+   */
+  private static final ZipField.F4 F_TOTAL_NUMBER_OF_DISKS =
+      new ZipField.F4(
+          F_Z64_EOCD_OFFSET.endOffset(), 0,"Total number of disks");
+
+
+  public static final int LOCATOR_SIZE = F_TOTAL_NUMBER_OF_DISKS.endOffset();
+
+  /**
+   * Offset from the beginning of the archive to where the Zip64 End of Central Directory record
+   * is located.
+   */
+  private final long z64EocdOffset;
+
+  /** Supplier of the byte representation of the zip64 Eocd Locator. */
+  private final CachedSupplier<byte[]> byteSupplier;
+
+  /**
+   * Creates a new Zip64 EOCD Locator, reading it from a byte source. This method will parse the
+   * byte source and obtain the EOCD Locator. It will check that the byte source starts with the
+   * EOCD Locator signature.
+   *
+   * @param bytes the byte buffer with the Locator data; when this method finishes, the byte buffer
+   *     will have its position moved to the end of the Locator (the beginning of the standard EOCD)
+   * @throws IOException failed to read information or the EOCD data is corrupt or invalid.
+   */
+  Zip64EocdLocator(ByteBuffer bytes) throws IOException {
+    F_SIGNATURE.verify(bytes);
+    F_NUMBER_OF_DISK.verify(bytes);
+    long z64EocdOffset = F_Z64_EOCD_OFFSET.read(bytes);
+    F_TOTAL_NUMBER_OF_DISKS.verify(bytes);
+
+    Verify.verify(z64EocdOffset >= 0);
+    this.z64EocdOffset = z64EocdOffset;
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * Creates a new Zip64 EOCD Locator. This is used when generating an EOCD Locator for a
+   * Zip64 EOCD that has been generated.
+   *
+   * @param z64EocdOffset offset position of the Zip64 EOCD from the beginning of the archive
+   */
+  Zip64EocdLocator(long z64EocdOffset) {
+    Preconditions.checkArgument(z64EocdOffset >= 0, "z64EocdOffset < 0");
+
+    this.z64EocdOffset = z64EocdOffset;
+    byteSupplier = new CachedSupplier<>(this::computeByteRepresentation);
+  }
+
+  /**
+   * Obtains the offset from the beginning of the archive to where the Zip64 EOCD is located.
+   *
+   * @return the Zip64 EOCD offset.
+   */
+  long getZ64EocdOffset() {
+    return z64EocdOffset;
+  }
+
+  /**
+   * Obtains the size of the Zip64 EOCD Locator
+   *
+   * @return the size, in bytes, of the EOCD Locator.<em> i.e. </em> 20.
+   */
+  long getSize() {
+    return F_TOTAL_NUMBER_OF_DISKS.endOffset();
+  }
+
+  /**
+   * Generates the EOCD Locator data.
+   *
+   * @return a byte representation of the EOCD Locator that has exactly {@link #getSize()} bytes
+   * @throws IOException failed to generate the EOCD data.
+   */
+  byte[] toBytes() throws IOException {
+    return byteSupplier.get();
+  }
+
+  /**
+   * Computes the byte representation of the EOCD Locator.
+   *
+   * @return a byte representation of the Zip64 EOCD Locator that has exactly {@link #getSize()}
+   *     bytes
+   * @throws UncheckedIOException failed to generate the EOCD Locator data
+   */
+  private byte[] computeByteRepresentation() {
+    ByteBuffer out = ByteBuffer.allocate(F_TOTAL_NUMBER_OF_DISKS.endOffset());
+
+    try {
+      F_SIGNATURE.write(out);
+      F_NUMBER_OF_DISK.write(out);
+      F_Z64_EOCD_OFFSET.write(out, z64EocdOffset);
+      F_TOTAL_NUMBER_OF_DISKS.write(out);
+
+      return out.array();
+    } catch (IOException e) {
+      throw new IOExceptionWrapper(e);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/Zip64ExtensibleDataSector.java

@@ -0,0 +1,219 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.LittleEndianUtils;
+import com.google.common.base.Preconditions;
+import com.google.common.collect.ImmutableList;
+import com.google.common.primitives.Ints;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.util.ArrayList;
+import java.util.List;
+import javax.annotation.Nullable;
+
+/**
+ * Contains the special purpose data for the Zip64 EOCD record.
+ *
+ * <p>According to the zip specification, the Zip64 EOCD is composed of a sequence of zero or more
+ * Special Purpose Data fields. This class provides a way to access, parse, and modify that
+ * information.
+ *
+ * <p>Each Special Purpose Data is represented by an instance of {@link Z64SpecialPurposeData} and
+ * contains a header ID and data.
+ */
+public class Zip64ExtensibleDataSector {
+
+  /**
+   * The extensible data sector's raw data, if it is known. Either this variable or {@link #fields}
+   * must be non-{@code null}.
+   */
+  @Nullable private final byte[] rawData;
+
+  /**
+   * The list of fields in the data sector. Will be populated if the data sector is created based on
+   * a list of special purpose data; will also be populated after parsing if the Data Sector is
+   * created based on the raw bytes.
+   */
+  @Nullable private ImmutableList<Z64SpecialPurposeData> fields;
+
+  /**
+   * Creates a Zip64 Extensible Data Sector based on existing raw data.
+   *
+   * @param rawData the raw data; will only be parsed if needed.
+   */
+  public Zip64ExtensibleDataSector(byte[] rawData) {
+    this.rawData = rawData;
+    fields = null;
+  }
+
+  /**
+   * Creates an Extensible Data Sector with no special purpose data.
+   */
+  public Zip64ExtensibleDataSector() {
+    rawData = null;
+    fields = ImmutableList.of();
+  }
+
+  /**
+   * Creates a Zip64 Extensible Data with the given Special purpose data.
+   *
+   * @param fields all special purpose data.
+   */
+  public Zip64ExtensibleDataSector(ImmutableList<Z64SpecialPurposeData> fields) {
+    rawData = null;
+    this.fields = fields;
+  }
+
+  int size() {
+    if (rawData != null) {
+      return rawData.length;
+    } else {
+      Preconditions.checkNotNull(fields);
+      int sumSizes = 0;
+      for (Z64SpecialPurposeData data : fields){
+        sumSizes += data.size();
+      }
+      return sumSizes;
+    }
+  }
+
+  void write(ByteBuffer out) throws IOException {
+    if (rawData != null) {
+      out.put(rawData);
+    } else {
+      Preconditions.checkNotNull(fields);
+      for (Z64SpecialPurposeData data : fields) {
+        data.write(out);
+      }
+    }
+  }
+
+  public ImmutableList<Z64SpecialPurposeData> getFields() throws IOException {
+    if (fields == null) {
+      parseData();
+    }
+
+    Preconditions.checkNotNull(fields);
+    return fields;
+  }
+
+  private void parseData() throws IOException {
+    Preconditions.checkNotNull(rawData);
+    Preconditions.checkState(fields == null);
+
+    List<Z64SpecialPurposeData> fields = new ArrayList<>();
+    ByteBuffer buffer = ByteBuffer.wrap(rawData);
+
+    while (buffer.remaining() > 0) {
+      int headerId = LittleEndianUtils.readUnsigned2Le(buffer);
+      long dataSize = LittleEndianUtils.readUnsigned4Le(buffer);
+
+      byte[] data = new byte[Ints.checkedCast(dataSize)];
+      if (dataSize < 0) {
+        throw new IOException(
+            "Invalid data size for special purpose data with header ID "
+                + headerId
+                + ": "
+                + dataSize);
+      }
+      buffer.get(data);
+
+      SpecialPurposeDataFactory factory = RawSpecialPurposeData::new;
+      Z64SpecialPurposeData spd = factory.make(headerId, data);
+      fields.add(spd);
+    }
+    this.fields = ImmutableList.copyOf(fields);
+  }
+
+  public interface Z64SpecialPurposeData {
+
+    /** Length of header id and the size length fields that comes before the data */
+    int PREFIX_LENGTH = 6;
+
+    /**
+     * Obtains the Special purpose data's header id.
+     *
+     * @return the data's header id.
+     */
+    int getHeaderId();
+
+    /**
+     * Obtains the size of the data in this special purpose data.
+     *
+     * @return the number of bytes needed to write the data.
+     */
+    int size();
+
+    /**
+     * Writes the special purpose data to the buffer.
+     *
+     * @param out the buffer where to write the data to; exactly {@link #size()} bytes will be
+     *     written.
+     * @throws IOException failed to write special purpose data.
+     */
+    void write(ByteBuffer out) throws IOException;
+  }
+
+  public  interface SpecialPurposeDataFactory {
+
+    /**
+     * Creates a new special purpose data.
+     *
+     * @param headerId the header ID
+     * @param data the data in the special purpose data
+     * @return the created special purpose data.
+     * @throws IOException failed to create the special purpose data from the data given.
+     */
+    Z64SpecialPurposeData make(int headerId, byte[] data) throws IOException;
+  }
+
+  /**
+   * Special Purpose Data containing raw data: this class represents a general "special purpose
+   * data" containing an array of bytes as data.
+   */
+  public static class RawSpecialPurposeData implements Z64SpecialPurposeData {
+
+    /** Header ID. */
+    private final int headerId;
+
+    /** Data in the segment */
+    private final byte[] data;
+
+    RawSpecialPurposeData(int headerId, byte[] data) {
+      this.headerId = headerId;
+      this.data = data;
+    }
+
+    @Override
+    public int getHeaderId() {
+      return headerId;
+    }
+
+    @Override
+    public int size() {
+      return PREFIX_LENGTH + data.length;
+    }
+
+    @Override
+    public void write(ByteBuffer out) throws IOException {
+      LittleEndianUtils.writeUnsigned2Le(out, headerId);
+      LittleEndianUtils.writeUnsigned4Le(out, data.length);
+      out.put(data);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipField.java

@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+import com.android.tools.build.apkzlib.zip.utils.LittleEndianUtils;
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import com.google.common.collect.Sets;
+import com.google.common.primitives.Ints;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
+import java.util.Set;
+import javax.annotation.Nullable;
+
+/**
+ * The ZipField class represents a field in a record in a zip file. Zip files are made with records
+ * that have fields. This class makes it easy to read, write and verify field values.
+ *
+ * <p>There are three main types of fields: 2-byte, 4-byte, and 8-byte fields. We represent each
+ * one as a subclass of {@code ZipField}, {@code F2} for the 2-byte field, {@code F4} for the 4-byte
+ * field and {@code F8} for the 8-byte field. Because Java's {@code int} data type is guaranteed to
+ * be 4-byte, all methods use Java's native {@link int} as data type.
+ *
+ * <p>The {@code F8} subclass is to support the 8-byte fields in the Zip64 specification. Because
+ * Java's 8-byte {@code long} does not support unsigned types, which reduces the support to 8-byte
+ * numbers of the form 2^63-1 or less. As {@code F8} fields refer to file sizes, this should be
+ * sufficient.
+ *
+ * <p>For each field we can either read, write or verify. Verification is used for fields whose
+ * value we know. Some fields, <em>e.g.</em> signature fields, have fixed value. Other fields have
+ * variable values, but in some situations we know which value they have. For example, the last
+ * modification time of a file's local header will have to match the value of the file's
+ * modification time as stored in the central directory.
+ *
+ * <p>Because records are compact, <em>i.e.</em> fields are stored sequentially with no empty
+ * spaces, fields are generally created in the sequence they exist and the end offset of a field is
+ * used as the offset of the next one. The end of a field can be obtained by invoking {@link
+ * #endOffset()}. This allows creating fields in sequence without doing offset computation:
+ *
+ * <pre>
+ * ZipField.F2 firstField = new ZipField.F2(0, "First field");
+ * ZipField.F4 secondField = new ZipField(firstField.endOffset(), "Second field");
+ * ZipField.F8 thirdField = new ZipField(secondField.endOffset(), "Third field");
+ * </pre>
+ */
+abstract class ZipField {
+
+  /** Field name. Used for providing (more) useful error messages. */
+  private final String name;
+
+  /** Offset of the file in the record. */
+  protected final int offset;
+
+  /** Size of the field. Only 2, 4, or 8 allowed. */
+  private final int size;
+
+  /** If a fixed value exists for the field, then this attribute will contain that value. */
+  @Nullable private final Long expected;
+
+  /** All invariants that this field must verify. */
+  private final Set<ZipFieldInvariant> invariants;
+
+  /**
+   * Creates a new field that does not contain a fixed value.
+   *
+   * @param offset the field's offset in the record
+   * @param size the field size
+   * @param name the field's name
+   * @param invariants the invariants that must be verified by the field
+   */
+  ZipField(int offset, int size, String name, ZipFieldInvariant... invariants) {
+    Preconditions.checkArgument(offset >= 0, "offset >= 0");
+    Preconditions.checkArgument(
+        size == 2 || size == 4 || size == 8,
+        "size != 2 && size != 4 && size != 8");
+
+    this.name = name;
+    this.offset = offset;
+    this.size = size;
+    expected = null;
+    this.invariants = Sets.newHashSet(invariants);
+  }
+
+  /**
+   * Creates a new field that contains a fixed value.
+   *
+   * @param offset the field's offset in the record
+   * @param size the field size
+   * @param expected the expected field value
+   * @param name the field's name
+   */
+  ZipField(int offset, int size, long expected, String name) {
+    Preconditions.checkArgument(offset >= 0, "offset >= 0");
+    Preconditions.checkArgument(
+        size == 2 || size == 4 || size == 8,
+        "size != 2 && size != 4 && size != 8");
+
+    this.name = name;
+    this.offset = offset;
+    this.size = size;
+    this.expected = expected;
+    invariants = Sets.newHashSet();
+  }
+
+  /**
+   * Checks whether a value verifies the field's invariants. Nothing happens if the value verifies
+   * the invariants.
+   *
+   * @param value the value
+   * @throws IOException the invariants are not verified
+   */
+  private void checkVerifiesInvariants(long value) throws IOException {
+    for (ZipFieldInvariant invariant : invariants) {
+      if (!invariant.isValid(value)) {
+        throw new IOException(
+            "Value "
+                + value
+                + " of field "
+                + name
+                + " is invalid "
+                + "(fails '"
+                + invariant.getName()
+                + "').");
+      }
+    }
+  }
+
+  /**
+   * Advances the position in the provided byte buffer by the size of this field.
+   *
+   * @param bytes the byte buffer; at the end of the method its position will be greater by the size
+   *     of this field
+   * @throws IOException failed to advance the buffer
+   */
+  void skip(ByteBuffer bytes) throws IOException {
+    if (bytes.remaining() < size) {
+      throw new IOException(
+          "Cannot skip field "
+              + name
+              + " because only "
+              + bytes.remaining()
+              + " remain in the buffer.");
+    }
+
+    bytes.position(bytes.position() + size);
+  }
+
+  /**
+   * Reads a field value.
+   *
+   * @param bytes the byte buffer with the record data; after this method finishes, the buffer will
+   *     be positioned at the first byte after the field
+   * @return the value of the field
+   * @throws IOException failed to read the field
+   */
+  long read(ByteBuffer bytes) throws IOException {
+    if (bytes.remaining() < size) {
+      throw new IOException(
+          "Cannot skip field "
+              + name
+              + " because only "
+              + bytes.remaining()
+              + " remain in the buffer.");
+    }
+
+    bytes.order(ByteOrder.LITTLE_ENDIAN);
+
+    long r;
+    if (size == 2) {
+      r = LittleEndianUtils.readUnsigned2Le(bytes);
+    } else if (size == 4) {
+      r = LittleEndianUtils.readUnsigned4Le(bytes);
+    } else {
+      r = LittleEndianUtils.readUnsigned8Le(bytes);
+    }
+
+    checkVerifiesInvariants(r);
+    return r;
+  }
+
+  /**
+   * Verifies that the field at the current buffer position has the expected value. The field must
+   * have been created with the constructor that defines the expected value.
+   *
+   * @param bytes the byte buffer with the record data; after this method finishes, the buffer will
+   *     be positioned at the first byte after the field
+   * @throws IOException failed to read the field or the field does not have the expected value
+   */
+  void verify(ByteBuffer bytes) throws IOException {
+    verify(bytes, null);
+  }
+
+  /**
+   * Verifies that the field at the current buffer position has the expected value. The field must
+   * have been created with the constructor that defines the expected value.
+   *
+   * @param bytes the byte buffer with the record data; after this method finishes, the buffer will
+   *     be positioned at the first byte after the field
+   * @param verifyLog if non-{@code null}, will log the verification error
+   * @throws IOException failed to read the data or the field does not have the expected value; only
+   *     thrown if {@code verifyLog} is {@code null}
+   */
+  void verify(ByteBuffer bytes, @Nullable VerifyLog verifyLog) throws IOException {
+    Preconditions.checkState(expected != null, "expected == null");
+    verify(bytes, expected, verifyLog);
+  }
+
+  /**
+   * Verifies that the field has an expected value.
+   *
+   * @param bytes the byte buffer with the record data; after this method finishes, the buffer will
+   *     be positioned at the first byte after the field
+   * @param expected the value we expect the field to have; if this field has invariants, the value
+   *     must verify them
+   * @throws IOException failed to read the data or the field does not have the expected value
+   */
+  void verify(ByteBuffer bytes, long expected) throws IOException {
+    verify(bytes, expected, null);
+  }
+
+  /**
+   * Verifies that the field has an expected value.
+   *
+   * @param bytes the byte buffer with the record data; after this method finishes, the buffer will
+   *     be positioned at the first byte after the field
+   * @param expected the value we expect the field to have; if this field has invariants, the value
+   *     must verify them
+   * @param verifyLog if non-{@code null}, will log the verification error
+   * @throws IOException failed to read the data or the field does not have the expected value; only
+   *     thrown if {@code verifyLog} is {@code null}
+   */
+  void verify(ByteBuffer bytes, long expected, @Nullable VerifyLog verifyLog) throws IOException {
+    checkVerifiesInvariants(expected);
+    long r = read(bytes);
+    if (r != expected) {
+      String error =
+          String.format(
+              "Incorrect value for field '%s': value is %s but %s expected.", name, r, expected);
+
+      if (verifyLog == null) {
+        throw new IOException(error);
+      } else {
+        verifyLog.log(error);
+      }
+    }
+  }
+
+  /**
+   * Writes the value of the field.
+   *
+   * @param output where to write the field; the field will be written at the current position of
+   *     the buffer
+   * @param value the value to write
+   * @throws IOException failed to write the value in the stream
+   */
+  void write(ByteBuffer output, long value) throws IOException {
+    checkVerifiesInvariants(value);
+
+    Preconditions.checkArgument(value >= 0, "value (%s) < 0", value);
+
+    if (size == 2) {
+      Preconditions.checkArgument(value <= 0x0000ffff, "value (%s) > 0x0000ffff", value);
+      LittleEndianUtils.writeUnsigned2Le(output, Ints.checkedCast(value));
+    } else if (size == 4) {
+      Preconditions.checkArgument(
+          value <= 0x00000000ffffffffL, "value (%s) > 0x00000000ffffffffL", value);
+      LittleEndianUtils.writeUnsigned4Le(output, value);
+    } else {
+      Verify.verify(size == 8);
+      LittleEndianUtils.writeUnsigned8Le(output, value);
+    }
+  }
+
+  /**
+   * Writes the value of the field. The field must have an expected value set in the constructor.
+   *
+   * @param output where to write the field; the field will be written at the current position of
+   *     the buffer
+   * @throws IOException failed to write the value in the stream
+   */
+  void write(ByteBuffer output) throws IOException {
+    Preconditions.checkState(expected != null, "expected == null");
+    write(output, expected);
+  }
+
+  /**
+   * Obtains the offset at which the field starts.
+   *
+   * @return the start offset
+   */
+  int offset() {
+    return offset;
+  }
+
+  /**
+   * Obtains the offset at which the field ends. This is the exact offset at which the next field
+   * starts.
+   *
+   * @return the end offset
+   */
+  int endOffset() {
+    return offset + size;
+  }
+
+  /** Concrete implementation of {@link ZipField} that represents a 2-byte field. */
+  static class F2 extends ZipField {
+
+    /**
+     * Creates a new field.
+     *
+     * @param offset the field's offset in the record
+     * @param name the field's name
+     * @param invariants the invariants that must be verified by the field
+     */
+    F2(int offset, String name, ZipFieldInvariant... invariants) {
+      super(offset, 2, name, invariants);
+    }
+
+    /**
+     * Creates a new field that contains a fixed value.
+     *
+     * @param offset the field's offset in the record
+     * @param expected the expected field value
+     * @param name the field's name
+     */
+    F2(int offset, long expected, String name) {
+      super(offset, 2, expected, name);
+    }
+  }
+
+  /** Concrete implementation of {@link ZipField} that represents a 4-byte field. */
+  static class F4 extends ZipField {
+    /**
+     * Creates a new field.
+     *
+     * @param offset the field's offset in the record
+     * @param name the field's name
+     * @param invariants the invariants that must be verified by the field
+     */
+    F4(int offset, String name, ZipFieldInvariant... invariants) {
+      super(offset, 4, name, invariants);
+    }
+
+    /**
+     * Creates a new field that contains a fixed value.
+     *
+     * @param offset the field's offset in the record
+     * @param expected the expected field value
+     * @param name the field's name
+     */
+    F4(int offset, long expected, String name) {
+      super(offset, 4, expected, name);
+    }
+  }
+
+  /** Concrete implementation of {@link ZipField} that represents a 8-byte field. */
+  static class F8 extends ZipField {
+
+    /**
+     * Creates a new field
+     *
+     * @param offset offset the field's offset in the record
+     * @param name the field's name
+     * @param invariants the invariants that must be verified by the field
+     */
+    F8(int offset, String name, ZipFieldInvariant... invariants) {
+      super(offset, 8, name, invariants);
+    }
+
+    /**
+     * Creates a new field that contains a fixed value.
+     *
+     * @param offset the field's offset in the record
+     * @param expected the expected field value
+     * @param name the field's name
+     */
+    F8(int offset, long expected, String name) {
+      super(offset, 8, expected, name);
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipFieldInvariant.java

@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/**
+ * A field rule defines an invariant (<em>i.e.</em>, a constraint) that has to be verified by a
+ * field value.
+ */
+interface ZipFieldInvariant {
+
+  /**
+   * Evalutes the invariant against a value.
+   *
+   * @param value the value to check the invariant
+   * @return is the invariant valid?
+   */
+  boolean isValid(long value);
+
+  /**
+   * Obtains the name of the invariant. Used for information purposes.
+   *
+   * @return the name of the invariant
+   */
+  String getName();
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipFieldInvariantMaxValue.java

@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/** Invariant checking a zip field does not exceed a threshold. */
+class ZipFieldInvariantMaxValue implements ZipFieldInvariant {
+
+  /** The maximum value allowed. */
+  private final long max;
+
+  /**
+   * Creates a new invariant.
+   *
+   * @param max the maximum value allowed for the field
+   */
+  ZipFieldInvariantMaxValue(long max) {
+    this.max = max;
+  }
+
+  @Override
+  public boolean isValid(long value) {
+    return value <= max;
+  }
+
+  @Override
+  public String getName() {
+    return "Maximum value " + max;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipFieldInvariantMinValue.java

@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2018 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/** Invariant checking a zip field doesn't go below a given value.*/
+class ZipFieldInvariantMinValue implements ZipFieldInvariant {
+
+  /** The minimum value allowed. */
+  private final long min;
+
+  /**
+   * Creates a new invariant.
+   *
+   * @param min the minimum value allowed for the field
+   */
+  ZipFieldInvariantMinValue(long min) {
+    this.min = min;
+  }
+
+  @Override
+  public boolean isValid(long value) {
+    return value >= min;
+  }
+
+  @Override
+  public String getName() {
+    return "Min value " + min;
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipFieldInvariantNonNegative.java

@@ -0,0 +1,31 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/** Invariant that verifies a field's value is not negative. */
+class ZipFieldInvariantNonNegative implements ZipFieldInvariant {
+
+  @Override
+  public boolean isValid(long value) {
+    return value >= 0;
+  }
+
+  @Override
+  public String getName() {
+    return "Is positive";
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/ZipFileState.java

@@ -0,0 +1,29 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip;
+
+/** The {@code ZipFileState} enumeration holds the state of a {@link ZFile}. */
+enum ZipFileState {
+  /** Zip file is closed. */
+  CLOSED,
+
+  /** File file is open in read-only mode. */
+  OPEN_RO,
+
+  /** File file is open in read-write mode. */
+  OPEN_RW
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/compress/BestAndDefaultDeflateExecutorCompressor.java

@@ -0,0 +1,84 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.compress;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.zip.CompressionResult;
+import com.android.tools.build.apkzlib.zip.utils.ByteTracker;
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.base.Preconditions;
+import java.util.concurrent.Executor;
+import java.util.zip.Deflater;
+
+/**
+ * Compressor that tries both the best and default compression algorithms and picks the default
+ * unless the best is at least a given percentage smaller.
+ */
+public class BestAndDefaultDeflateExecutorCompressor extends ExecutorCompressor {
+
+  /** Deflater using the default compression level. */
+  private final DeflateExecutionCompressor defaultDeflater;
+
+  /** Deflater using the best compression level. */
+  private final DeflateExecutionCompressor bestDeflater;
+
+  /**
+   * Minimum best compression size / default compression size ratio needed to pick the default
+   * compression size.
+   */
+  private final double minRatio;
+
+  /**
+   * Creates a new compressor.
+   *
+   * @param executor the executor used to perform compression activities.
+   * @param minRatio the minimum best compression size / default compression size needed to pick the
+   *     default compression size; if {@code 0.0} then the default compression is always picked, if
+   *     {@code 1.0} then the best compression is always picked unless it produces the exact same
+   *     size as the default compression.
+   */
+  public BestAndDefaultDeflateExecutorCompressor(Executor executor, double minRatio) {
+    super(executor);
+
+    Preconditions.checkArgument(minRatio >= 0.0, "minRatio < 0.0");
+    Preconditions.checkArgument(minRatio <= 1.0, "minRatio > 1.0");
+
+    defaultDeflater = new DeflateExecutionCompressor(executor, Deflater.DEFAULT_COMPRESSION);
+    bestDeflater = new DeflateExecutionCompressor(executor, Deflater.BEST_COMPRESSION);
+    this.minRatio = minRatio;
+  }
+
+  @Deprecated
+  public BestAndDefaultDeflateExecutorCompressor(
+      Executor executor, ByteTracker tracker, double minRatio) {
+    this(executor, minRatio);
+  }
+
+  @Override
+  protected CompressionResult immediateCompress(CloseableByteSource source, ByteStorage storage)
+      throws Exception {
+    CompressionResult defaultResult = defaultDeflater.immediateCompress(source, storage);
+    CompressionResult bestResult = bestDeflater.immediateCompress(source, storage);
+
+    double sizeRatio = bestResult.getSize() / (double) defaultResult.getSize();
+    if (sizeRatio >= minRatio) {
+      return defaultResult;
+    } else {
+      return bestResult;
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/compress/DeflateExecutionCompressor.java

@@ -0,0 +1,73 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.compress;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.bytestorage.CloseableByteSourceFromOutputStreamBuilder;
+import com.android.tools.build.apkzlib.zip.CompressionMethod;
+import com.android.tools.build.apkzlib.zip.CompressionResult;
+import com.android.tools.build.apkzlib.zip.utils.ByteTracker;
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.io.ByteStreams;
+import java.io.InputStream;
+import java.util.concurrent.Executor;
+import java.util.zip.Deflater;
+import java.util.zip.DeflaterOutputStream;
+
+/** Compressor that uses deflate with an executor. */
+public class DeflateExecutionCompressor extends ExecutorCompressor {
+
+  /** Deflate compression level. */
+  private final int level;
+
+  /**
+   * Creates a new compressor.
+   *
+   * @param executor the executor to run deflation tasks
+   * @param level the compression level
+   */
+  public DeflateExecutionCompressor(Executor executor, int level) {
+    super(executor);
+
+    this.level = level;
+  }
+
+  @Deprecated
+  public DeflateExecutionCompressor(Executor executor, ByteTracker tracker, int level) {
+    this(executor, level);
+  }
+
+  @Override
+  protected CompressionResult immediateCompress(CloseableByteSource source, ByteStorage storage)
+      throws Exception {
+    Deflater deflater = new Deflater(level, true);
+    CloseableByteSourceFromOutputStreamBuilder resultBuilder = storage.makeBuilder();
+
+    try (InputStream inputStream = source.openBufferedStream();
+            DeflaterOutputStream dos = new DeflaterOutputStream(resultBuilder, deflater)) {
+      ByteStreams.copy(inputStream, dos);
+    }
+
+    CloseableByteSource result = resultBuilder.build();
+    if (result.size() >= source.size()) {
+      result.close();
+      return new CompressionResult(source, CompressionMethod.STORE, source.size());
+    } else {
+      return new CompressionResult(result, CompressionMethod.DEFLATE, result.size());
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/compress/ExecutorCompressor.java

@@ -0,0 +1,71 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.compress;
+
+import com.android.tools.build.apkzlib.bytestorage.ByteStorage;
+import com.android.tools.build.apkzlib.zip.CompressionResult;
+import com.android.tools.build.apkzlib.zip.Compressor;
+import com.android.tools.build.apkzlib.zip.utils.CloseableByteSource;
+import com.google.common.util.concurrent.ListenableFuture;
+import com.google.common.util.concurrent.SettableFuture;
+import java.util.concurrent.Executor;
+
+/**
+ * A synchronous compressor is a compressor that computes the result of compression immediately and
+ * never returns an uncomputed future object.
+ */
+public abstract class ExecutorCompressor implements Compressor {
+
+  /** The executor that does the work. */
+  private final Executor executor;
+
+  /**
+   * Compressor that delegates execution into the given executor.
+   *
+   * @param executor the executor that will do the compress
+   */
+  public ExecutorCompressor(Executor executor) {
+    this.executor = executor;
+  }
+
+  @Override
+  public ListenableFuture<CompressionResult> compress(
+      CloseableByteSource source, ByteStorage storage) {
+    final SettableFuture<CompressionResult> future = SettableFuture.create();
+    executor.execute(
+        () -> {
+          try {
+            future.set(immediateCompress(source, storage));
+          } catch (Throwable e) {
+            future.setException(e);
+          }
+        });
+
+    return future;
+  }
+
+  /**
+   * Immediately compresses a source.
+   *
+   * @param source the source to compress
+   * @param storage a byte storage where the compressor can obtain data sources from
+   * @return the result of compression
+   * @throws Exception failed to compress
+   */
+  protected abstract CompressionResult immediateCompress(
+      CloseableByteSource source, ByteStorage storage) throws Exception;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/compress/Zip64NotSupportedException.java

@@ -0,0 +1,27 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.compress;
+
+import java.io.IOException;
+
+/** Exception raised by ZFile when encountering unsupported Zip64 format jar files. */
+public class Zip64NotSupportedException extends IOException {
+
+  public Zip64NotSupportedException(String message) {
+    super(message);
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/compress/package-info.java

@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2017 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/** Compressors to use with the {@code zip} package. */
+package com.android.tools.build.apkzlib.zip.compress;

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/ByteTracker.java

@@ -0,0 +1,27 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+
+/**
+ * Keeps track of used bytes allowing gauging memory usage.
+ *
+ * @deprecated will be removed shortly.
+ */
+@Deprecated
+public class ByteTracker {
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/CloseableByteSource.java

@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+import com.google.common.io.ByteSource;
+import java.io.Closeable;
+import java.io.IOException;
+
+/**
+ * Byte source that can be closed. Closing a byte source allows releasing any resources associated
+ * with it. This should not be confused with closing streams. For example, {@link ByteTracker} uses
+ * {@code CloseableByteSources} to know when the data associated with the byte source can be
+ * released.
+ */
+public abstract class CloseableByteSource extends ByteSource implements Closeable {
+
+  /** Has the source been closed? */
+  private boolean closed;
+
+  /** Creates a new byte source. */
+  public CloseableByteSource() {
+    closed = false;
+  }
+
+  @Override
+  public final synchronized void close() throws IOException {
+    if (closed) {
+      return;
+    }
+
+    try {
+      innerClose();
+    } finally {
+      closed = true;
+    }
+  }
+
+  /**
+   * Closes the by source. This method is only invoked once, even if {@link #close()} is called
+   * multiple times.
+   *
+   * @throws IOException failed to close
+   */
+  protected abstract void innerClose() throws IOException;
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/CloseableDelegateByteSource.java

@@ -0,0 +1,159 @@
+/*
+ * Copyright (C) 2016 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+import com.google.common.hash.HashCode;
+import com.google.common.hash.HashFunction;
+import com.google.common.io.ByteProcessor;
+import com.google.common.io.ByteSink;
+import com.google.common.io.ByteSource;
+import com.google.common.io.CharSource;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.OutputStream;
+import java.nio.charset.Charset;
+import javax.annotation.Nullable;
+
+/** Closeable byte source that delegates to another byte source. */
+public class CloseableDelegateByteSource extends CloseableByteSource {
+
+  /** The byte source we delegate all operations to. {@code null} if disposed. */
+  @Nullable private ByteSource inner;
+
+  /**
+   * Size of the byte source. This is the same as {@code inner.size()} (when {@code inner} is not
+   * {@code null}), but we keep it separate to avoid calling {@code inner.size()} because it might
+   * throw {@code IOException}.
+   */
+  private final long mSize;
+
+  /**
+   * Creates a new byte source.
+   *
+   * @param inner the inner byte source
+   * @param size the size of the source
+   */
+  public CloseableDelegateByteSource(ByteSource inner, long size) {
+    this.inner = inner;
+    mSize = size;
+  }
+
+  /**
+   * Obtains the inner byte source. Will throw an exception if the inner by byte source has been
+   * disposed of.
+   *
+   * @return the inner byte source
+   */
+  private synchronized ByteSource get() {
+    if (inner == null) {
+      throw new ByteSourceDisposedException();
+    }
+
+    return inner;
+  }
+
+  /** Mark the byte source as disposed. */
+  @Override
+  protected synchronized void innerClose() throws IOException {
+    if (inner == null) {
+      return;
+    }
+
+    inner = null;
+  }
+
+  /**
+   * Obtains the size of this byte source. Equivalent to {@link #size()} but not throwing {@code
+   * IOException}.
+   *
+   * @return the size of the byte source
+   */
+  public long sizeNoException() {
+    return mSize;
+  }
+
+  @Override
+  public CharSource asCharSource(Charset charset) {
+    return get().asCharSource(charset);
+  }
+
+  @Override
+  public InputStream openBufferedStream() throws IOException {
+    return get().openBufferedStream();
+  }
+
+  @Override
+  public ByteSource slice(long offset, long length) {
+    return get().slice(offset, length);
+  }
+
+  @Override
+  public boolean isEmpty() throws IOException {
+    return get().isEmpty();
+  }
+
+  @Override
+  public long size() throws IOException {
+    return get().size();
+  }
+
+  @Override
+  public long copyTo(OutputStream output) throws IOException {
+    return get().copyTo(output);
+  }
+
+  @Override
+  public long copyTo(ByteSink sink) throws IOException {
+    return get().copyTo(sink);
+  }
+
+  @Override
+  public byte[] read() throws IOException {
+    return get().read();
+  }
+
+  @Override
+  public <T> T read(ByteProcessor<T> processor) throws IOException {
+    return get().read(processor);
+  }
+
+  @Override
+  public HashCode hash(HashFunction hashFunction) throws IOException {
+    return get().hash(hashFunction);
+  }
+
+  @Override
+  public boolean contentEquals(ByteSource other) throws IOException {
+    return get().contentEquals(other);
+  }
+
+  @Override
+  public InputStream openStream() throws IOException {
+    return get().openStream();
+  }
+
+  /** Exception thrown when trying to use a byte source that has been disposed. */
+  private static class ByteSourceDisposedException extends RuntimeException {
+
+    /** Creates a new exception. */
+    private ByteSourceDisposedException() {
+      super(
+          "Byte source was created by a ByteTracker and is now disposed. If you see "
+              + "this message, then there is a bug.");
+    }
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/LittleEndianUtils.java

@@ -0,0 +1,157 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+import com.google.common.base.Preconditions;
+import com.google.common.base.Verify;
+import java.io.EOFException;
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
+
+/**
+ * Utilities to read and write 16, 32, and 64 bit integers with support for little-endian encoding,
+ * as used in zip files. Zip files actually use unsigned data types. We use Java's native (signed)
+ * data types but will use long (64 bit) to ensure we can fit the whole range for the 16 and 32
+ * bit fields.
+ */
+public class LittleEndianUtils {
+  /** Utility class, no constructor. */
+  private LittleEndianUtils() {}
+
+  /**
+   * Reads 8 bytes in little-endian format and converts them into a 64-bit value.
+   *
+   * @param bytes from where should the bytes be read; the first 8 bytes of the source will be read.
+   * @return the 64-bit value
+   * @throws IOException failed to read the value.
+   */
+  public static long readUnsigned8Le(ByteBuffer bytes) throws IOException {
+    Preconditions.checkNotNull(bytes, "bytes == null");
+
+    if (bytes.remaining() < 8) {
+      throw new EOFException(
+              "Not enough data: 8 bytes expected, " + bytes.remaining() + " available.");
+    }
+
+    ByteOrder order = bytes.order();
+    bytes.order(ByteOrder.LITTLE_ENDIAN);
+    long r = bytes.getLong();
+    bytes.order(order);
+    return r;
+  }
+
+  /**
+   * Reads 4 bytes in little-endian format and converts them into a 32-bit value.
+   *
+   * @param bytes from where should the bytes be read; the first 4 bytes of the source will be read
+   * @return the 32-bit value
+   * @throws IOException failed to read the value
+   */
+  public static long readUnsigned4Le(ByteBuffer bytes) throws IOException {
+    Preconditions.checkNotNull(bytes, "bytes == null");
+
+    if (bytes.remaining() < 4) {
+      throw new EOFException(
+          "Not enough data: 4 bytes expected, " + bytes.remaining() + " available.");
+    }
+
+    byte b0 = bytes.get();
+    byte b1 = bytes.get();
+    byte b2 = bytes.get();
+    byte b3 = bytes.get();
+    long r = (b0 & 0xff) | ((b1 & 0xff) << 8) | ((b2 & 0xff) << 16) | ((b3 & 0xffL) << 24);
+    Verify.verify(r >= 0);
+    Verify.verify(r <= 0x00000000ffffffffL);
+    return r;
+  }
+
+  /**
+   * Reads 2 bytes in little-endian format and converts them into a 16-bit value.
+   *
+   * @param bytes from where should the bytes be read; the first 2 bytes of the source will be read
+   * @return the 16-bit value
+   * @throws IOException failed to read the value
+   */
+  public static int readUnsigned2Le(ByteBuffer bytes) throws IOException {
+    Preconditions.checkNotNull(bytes, "bytes == null");
+
+    if (bytes.remaining() < 2) {
+      throw new EOFException(
+          "Not enough data: 2 bytes expected, " + bytes.remaining() + " available.");
+    }
+
+    byte b0 = bytes.get();
+    byte b1 = bytes.get();
+    int r = (b0 & 0xff) | ((b1 & 0xff) << 8);
+
+    Verify.verify(r >= 0);
+    Verify.verify(r <= 0x0000ffff);
+    return r;
+  }
+
+  /**
+   * Writes 8 bytes in little-endian format, converting them from a <em> signed </em> 64-bit value.
+   *
+   * @param output the output stream where the bytes will be written.
+   * @param value the 64-bit value to convert.
+   * @throws IOException failed to write the value data.
+   */
+  public static void writeUnsigned8Le(ByteBuffer output, long value) throws IOException {
+    Preconditions.checkNotNull(output, "output == null");
+
+    ByteOrder order = output.order();
+    output.order(ByteOrder.LITTLE_ENDIAN);
+    output.putLong(value);
+    output.order(order);
+  }
+
+  /**
+   * Writes 4 bytes in little-endian format, converting them from a 32-bit value.
+   *
+   * @param output the output stream where the bytes will be written
+   * @param value the 32-bit value to convert
+   * @throws IOException failed to write the value data
+   */
+  public static void writeUnsigned4Le(ByteBuffer output, long value) throws IOException {
+    Preconditions.checkNotNull(output, "output == null");
+    Preconditions.checkArgument(value >= 0, "value (%s) < 0", value);
+    Preconditions.checkArgument(
+        value <= 0x00000000ffffffffL, "value (%s) > 0x00000000ffffffffL", value);
+
+    output.put((byte) (value & 0xff));
+    output.put((byte) ((value >> 8) & 0xff));
+    output.put((byte) ((value >> 16) & 0xff));
+    output.put((byte) ((value >> 24) & 0xff));
+  }
+
+  /**
+   * Writes 2 bytes in little-endian format, converting them from a 16-bit value.
+   *
+   * @param output the output stream where the bytes will be written
+   * @param value the 16-bit value to convert
+   * @throws IOException failed to write the value data
+   */
+  public static void writeUnsigned2Le(ByteBuffer output, int value) throws IOException {
+    Preconditions.checkNotNull(output, "output == null");
+    Preconditions.checkArgument(value >= 0, "value (%s) < 0", value);
+    Preconditions.checkArgument(value <= 0x0000ffff, "value (%s) > 0x0000ffff", value);
+
+    output.put((byte) (value & 0xff));
+    output.put((byte) ((value >> 8) & 0xff));
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/MsDosDateTimeUtils.java

@@ -0,0 +1,105 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+import com.google.common.base.Verify;
+import java.util.Calendar;
+import java.util.Date;
+
+/** Yes. This actually refers to MS-DOS in 2015. That's all I have to say about legacy stuff. */
+public class MsDosDateTimeUtils {
+  /** Utility class: no constructor. */
+  private MsDosDateTimeUtils() {}
+
+  /**
+   * Packs java time value into an MS-DOS time value.
+   *
+   * @param time the time value
+   * @return the MS-DOS packed time
+   */
+  public static int packTime(long time) {
+    Calendar c = Calendar.getInstance();
+    c.setTime(new Date(time));
+
+    int seconds = c.get(Calendar.SECOND);
+    int minutes = c.get(Calendar.MINUTE);
+    int hours = c.get(Calendar.HOUR_OF_DAY);
+
+    /*
+     * Here is how MS-DOS packs a time value:
+     * 0-4: seconds (divided by 2 because we only have 5 bits = 32 different numbers)
+     * 5-10: minutes (6 bits = 64 possible values)
+     * 11-15: hours (5 bits = 32 possible values)
+     *
+     * source: https://msdn.microsoft.com/en-us/library/windows/desktop/ms724247(v=vs.85).aspx
+     */
+    return (hours << 11) | (minutes << 5) | (seconds / 2);
+  }
+
+  /**
+   * Packs the current time value into an MS-DOS time value.
+   *
+   * @return the MS-DOS packed time
+   */
+  public static int packCurrentTime() {
+    return packTime(new Date().getTime());
+  }
+
+  /**
+   * Packs java time value into an MS-DOS date value.
+   *
+   * @param time the time value
+   * @return the MS-DOS packed date
+   */
+  public static int packDate(long time) {
+    Calendar c = Calendar.getInstance();
+    c.setTime(new Date(time));
+
+    /*
+     * Even MS-DOS used 1 for January. Someone wasn't really thinking when they decided on Java
+     * it would start at 0...
+     */
+    int day = c.get(Calendar.DAY_OF_MONTH);
+    int month = c.get(Calendar.MONTH) + 1;
+
+    /*
+     * MS-DOS counts years starting from 1980. Since its launch date was in 81, it was obviously
+     * not necessary to talk about dates earlier than that.
+     */
+    int year = c.get(Calendar.YEAR) - 1980;
+    Verify.verify(year >= 0 && year < 128);
+
+    /*
+     * Here is how MS-DOS packs a date value:
+     * 0-4: day (5 bits = 32 values)
+     * 5-8: month (4 bits = 16 values)
+     * 9-15: year (7 bits = 128 values)
+     *
+     * source: https://msdn.microsoft.com/en-us/library/windows/desktop/ms724247(v=vs.85).aspx
+     */
+    return (year << 9) | (month << 5) | day;
+  }
+
+  /**
+   * Packs the current time value into an MS-DOS date value.
+   *
+   * @return the MS-DOS packed date
+   */
+  public static int packCurrentDate() {
+    return packDate(new Date().getTime());
+  }
+}

apkzlib/src/main/java/com/android/tools/build/apkzlib/zip/utils/RandomAccessFileUtils.java

@@ -0,0 +1,55 @@
+/*
+ * Copyright (C) 2015 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package com.android.tools.build.apkzlib.zip.utils;
+
+import java.io.IOException;
+import java.io.RandomAccessFile;
+
+/** Utility class with utility methods for random access files. */
+public final class RandomAccessFileUtils {
+
+  private RandomAccessFileUtils() {}
+
+  /**
+   * Reads from an random access file until the provided array is filled. Data is read from the
+   * current position in the file.
+   *
+   * @param raf the file to read data from
+   * @param data the array that will receive the data
+   * @throws IOException failed to read the data
+   */
+  public static void fullyRead(RandomAccessFile raf, byte[] data) throws IOException {
+    int r;
+    int p = 0;
+
+    while ((r = raf.read(data, p, data.length - p)) > 0) {
+      p += r;
+      if (p == data.length) {
+        break;
+      }
+    }
+
+    if (p < data.length) {
+      throw new IOException(
+          "Failed to read "
+              + data.length
+              + " bytes from file. Only "
+              + p
+              + " bytes could be read.");
+    }
+  }
+}

build.gradle

@@ -4,12 +4,9 @@ buildscript {
     repositories {
         google()
         mavenCentral()
-        maven { url "https://jcenter.bintray.com" }
-        maven { url "https://jitpack.io" }
     }
     dependencies {
         classpath 'com.android.tools.build:gradle:7.1.0-alpha10'
-        classpath "org.jetbrains.kotlin:kotlin-gradle-plugin:1.5.21"
     }
 }
 
@@ -32,8 +29,6 @@ allprojects {
     repositories {
         google()
         mavenCentral()
-        maven { url "https://jcenter.bintray.com" }
-        maven { url "https://jitpack.io" }
     }
 }
 

patch/build.gradle

@@ -13,8 +13,8 @@ dependencies {
     implementation fileTree(dir: 'libs', include: ['*.jar'])
     implementation project(':axmlprinter')
     implementation project(':share')
-    implementation 'commons-io:commons-io:2.10.0'
-    implementation 'com.android.tools.build:apkzlib:4.2.2'
+    implementation project(':apkzlib')
+    implementation 'commons-io:commons-io:2.11.0'
     implementation 'com.beust:jcommander:1.81'
 }
 
@@ -40,7 +40,7 @@ jar {
         from fileTree(dir: "$rootProject.projectDir/out/so")
     }
 
-    exclude 'META-INF/*.SF', 'META-INF/*.DSA', 'META-INF/*.RSA', 'META-INF/*.MF', 'META-INF/*.txt'
+    exclude 'META-INF/*.SF', 'META-INF/*.DSA', 'META-INF/*.RSA', 'META-INF/*.MF', 'META-INF/*.txt', "META-INF/versions/**"
 }
 
 tasks.register("buildDebug") {

settings.gradle

@@ -14,3 +14,4 @@ include ':patch'
 include ':axmlprinter'
 include ':share'
 include ':appstub'
+include ':apkzlib'