Added a FilterChain that can apply multiple filter steps to an APK file
as it is being written to disk. The first filter applied is by ABI. If
a library in the APK does not match the filter it is skipped.
Added an AbiFilter that keeps files that are either not native libs or
are for the set of wanted ABIs
Test: ran unit tests locally
Test: ran against an APK with ARM and x68 libs and diffed the results
Change-Id: I3fb901d3de3513e85f2a2763a8e4487a28ed4881
Generating splits should be possible to do from the optimize command.
This means that a lot of infrastructure around split APKs can be
shared by both the optimize and link phase.
Bug: 35925830
Change-Id: Ia88b9e4bff300a56353b2f7a4a2547c8eb43a299
Test: manual
This also enables an AAPT behavior that CTS tests have
come to depend on.
Small files that compress negatively (get larger) are stored
uncompressed. Some CTS tests assume this and try to open these
files by mmapping them, which is only possible if they are
uncompressed.
Bug: 35461578
Test: make aapt2_tests
Change-Id: Id622a6150fe72477ad65d67d1bad897a8ee2ffb9
Benchmarks on bullhead-userdebug show that there is a negligent
performance impact when using sparse entries on a 30% loaded
sparse type of 1000 resources.
Benchmark Time CPU Iterations
-----------------------------------------------------------------------------------
BM_SparseEntryGetResourceSparseLarge 255 ns 254 ns 2751408
BM_SparseEntryGetResourceNotSparseLarge 254 ns 254 ns 2756534
Bug: 27381711
Test: make libandroidfw_tests aapt2_tests
Change-Id: I051ea22f2f6b2bc3696e446adc9e2a34be18009f
