com.android.builder.dexing

Class DexArchiveMerger

java.lang.Object

com.android.builder.dexing.DexArchiveMerger

public class DexArchiveMerger
extends java.lang.Object

Merges DEX files found in DexArchives, and produces the final DEX file(s). Inputs can come from one or more dex archives. In order to process the dex archives, one should invoke merge(Collection, Path, Set) method.

In order to merge individual DEX files, we are using DexMergingStrategy to determine how many input DEX files can fit into a single output DEX.

Merging is actually executed by invoking the DexMerger, which is part of dx.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	DexArchiveMerger.DexArchiveMergerException Exception thrown when dex archive merging goes wrong.

Constructor Summary

Constructors

Constructor and Description
DexArchiveMerger(DexMergerConfig config, DexMergingStrategy mergingStrategy, java.util.concurrent.ForkJoinPool forkJoinPool)
DexArchiveMerger(DexMergerConfig config, java.util.concurrent.ForkJoinPool forkJoinPool) Creates an instance of merger.

Method Summary

Modifier and Type	Method and Description
void	merge(java.util.Collection<java.nio.file.Path> inputs, java.nio.file.Path outputDir, java.util.Set<java.lang.String> mainDexClasses) Triggers the actual merging by processing all DEX files from dex archives, and outputting the DEX file(s) in the specified directory.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DexArchiveMerger

public DexArchiveMerger(@NonNull
                        DexMergerConfig config,
                        @NonNull
                        java.util.concurrent.ForkJoinPool forkJoinPool)

Creates an instance of merger. The executor that is specified in parameters will be used to schedule tasks. Important to notice is that merging, triggered by invoking merge(Collection, Path, Set) method, might return before final DEX file(s) are merged and written out. Therefore, the invoker will have to block on the executor, in order to be sure the merging is finished.

Parameters:

config - configuration for this merging

forkJoinPool - executor used to schedule tasks in the merging process

DexArchiveMerger

public DexArchiveMerger(@NonNull
                        DexMergerConfig config,
                        @NonNull
                        DexMergingStrategy mergingStrategy,
                        @NonNull
                        java.util.concurrent.ForkJoinPool forkJoinPool)

Method Detail

merge

public void merge(@NonNull
                  java.util.Collection<java.nio.file.Path> inputs,
                  @NonNull
                  java.nio.file.Path outputDir,
                  @Nullable
                  java.util.Set<java.lang.String> mainDexClasses)
           throws java.io.IOException

Triggers the actual merging by processing all DEX files from dex archives, and outputting the DEX file(s) in the specified directory. DEX files will have names classes.dex, classes2.dex etc. It is the responsibility of the invoker to make sure the output directory is empty and writable. Merging might return before the DEX file(s) are merged and written out.

In case of mono-dex, it is expected that all DEX files will fit into a single output DEX file. If this is not possible, DexIndexOverflowException exception will be thrown.

When merging native multidex, multiple DEX files might be created.

In case of legacy multidex, list of classes in the main dex should be specified as well. When using legacy multidex mode, only classes explicitly specified in this list will be in the main dex file. This is equivalent of invoking dx with --minimal-main-dex option.

Throws:

java.io.IOException