A few OEM devices have introduced a memory corruption bug
that is only triggered when an application's AndroidManifest.xml
uses UTF8 string pools. Workaround this by only encoding
AndroidManifest.xml with UTF16 strings.
Bug: 64434571
Test: manual
Change-Id: I2ee50d1b2b5942d971ef2a544c878de63f67c652
This is the desired behavior if resource IDs may overlap, and also happens to
improve the behavior of:
aapt dump -I base.apk badging feature.apk
when both APKs use the same package name with different package IDs
(base=0x7F, feature=0x80). Previously, the final call to
DynamicRefTable::addMapping was for base.apk, and all references to 0x80 were
rewritten in DynamicRefTable::lookupResourceId to 0x7F. After this
change resources defined in both APKs are resolved correctly.
Note that this is a bit of kludge, and APKs should use different names
to avoid conflating the package IDs.
Fixes: 67070085
Bug: 64058531
Test: manual with sample in b/67070085#comment2
Change-Id: I285adb6f44a297440b08fc7a1f9ad73c700eb9bc
ASM 6 is required to read version 53 (OpenJDK 9) class files.
Test: make EXPERIMENTAL_USE_OPENJDK9=true services.core
(in the presence of other unsubmitted CLs that fix
other issues)
Test: Copied all required dependencies into a subdirectory lib/,
then ran the steps from TestMain.java, both for OpenJDK 8
and OpenJDK 9 toolchains.
Bug: 67676752
Change-Id: If324d8ea28450d9a8426a2279190b2f4d9213ed5
Bug: 66729158
Test: Manually build and flash the system.img and test "activity --proto
broadcasts"
Change-Id: I3cb7474c4a3ab092f38ed6a121b4d034cadfc402
It is very similiar to protoc-gen-javastream, which generates field Ids used
by ProtoOutputStream.cpp to dump protobuf data.
Bug: 65641021
Test: compile the streaming_proto:
$ mmm -j frameworks/base/tools/streaming_proto/
and run:
$ PATH=$PATH:out/host/linux-x86/bin/protoc-gen-cppstream aprotoc
--cppstream_out=tmp/
frameworks/base/core/proto/android/service/procstats.proto
frameworks/base/core/proto/android/util/common.proto
Change-Id: I68becc80b5166455455c5df28cd698601b4a1c1d
Also add check for when enum value can't be found and fixing output
formatting for some int types.
BUG: 65633831
Test: check incident_report output of repeated enums.
Change-Id: I329eef9b3bb0026259ea465ab67e71952df1a4e4
Bug: 34740546
Test: build with WITH_TIDY=1 and
WITH_TIDY_CHECKS="*,-readability-*,-google-readability-*,-google-runtime-references"
Change-Id: If99c75cab6a2bec0c6b38aa17189668b4943478d
If an artifact in a multi-apk set is for a scpecific set of screen
densities then update the manifest to set the compatible-screens element
to match.
Test: ran unit tests
Test: manually ran optimize command
Change-Id: I07e90850ab2f4647a8623acbeaccbebbe81c3a6a
Set versionCode in the AndroidManifest.xml so each artifact has a unique
version code. This will allow the play store to serve up the correct
APKs.
The versionCode is determined by taking the versionCode from the input
APK and adding the configured version to that. If an artifact does not
have a version, the next version is used based on the order in the
configuration file. If the firs artifact does not have a version, it
starts at 1, so that all artifacts have a different version code to the
base APK.
Bug: 64972562
Test: unit tests
Test: manually optimized an APK into multiple APKs and confirmed versions:
aapt dump badging ${apk} | grep versionCode
Change-Id: Ie12b3e42bbd8ce9a2819712c2c56b24c0a901e6b
When generating multiple APKs from a configuration file, check to see if
we have filtered resource by minSdkVersion and update the manifest to
reflect this. We only want to inflate and modify the manifest file if
there is an update to be applied.
Bug: 37944703
Bug: 67005138
Test: Ran unit tests
Test: Manually split an APK and verified the manifest by dumping with
AAPT (both xmltree and badging).
Change-Id: I64a0e4889d7d9e57373369b044a091287b06cc35
Make sure that Styleables are directly followed by their indices.
If not, Robolectric breaks. This is not strictly incorrect to have
an arbitrary ordering in R.java, but its easier to just support
Robolectric in this case.
Bug: 65837293
Merged-In: Ia59ba58427ade386d075ca9fc9eb5b53e35beca0
Test: make aapt2_tests
(cherry picked from commit af85c4deb667843a227d62275fe6992005f4c38d)
Change-Id: Ia59ba58427ade386d075ca9fc9eb5b53e35beca0
Make sure that Styleables are directly followed by their indices.
If not, Robolectric breaks. This is not strictly incorrect to have
an arbitrary ordering in R.java, but its easier to just support
Robolectric in this case.
Bug: 65837293
Test: make aapt2_tests
(cherry picked from commit af85c4deb667843a227d62275fe6992005f4c38d)
Change-Id: Ia59ba58427ade386d075ca9fc9eb5b53e35beca0
This fixes the warning
Call to function 'strcpy' is insecure as it does not provide bounding of
the memory buffer. As a side effect, it sliences the warning
frameworks/base/tools/aapt/Images.cpp:1270:50: warning: Potential leak
of memory pointed to by field 'data' [clang-analyzer-unix.Malloc]
frameworks/base/tools/aapt2/compile/Png.cpp:562:42: warning: Potential
leak of memory pointed to by field 'data' [clang-analyzer-unix.Malloc].
Bug: None
Test: The warning is gone.
Change-Id: I25f68ff85bea7069c21549c7deb7920d1877069e
We're going to be using the Any proto field in ConfigDescription,
so migrate to proto3.
Test: make aapt2_tests
Change-Id: I43b91f4dfc448251f4352e48586b3b15502c033d
Resource deduping relies on the definitions of
ConfigDescription.ConflictsWith, ConfigDescription.IsCompatibleWith,
and ConfigDescription.Dominates.
ConflictsWith is supposed to ignore range-based qualifiers, like
version, density, smallest width, screen size, etc.
This was not the case for screen size, and was assumed that the
choice of screen size is mutually exclusive.
This CL fixes the assumption and includes screen size (small, normal, large, xlarge)
as a qualifier that does not conflict.
Bug: 64397629
Bug: 65645766
Test: make aapt2_tests
Change-Id: I573a6735fedd7721a10ba32902bc5d27ef99b88e
Merged-In: I573a6735fedd7721a10ba32902bc5d27ef99b88e
A bug in the dump badging command caused strings to be looked up in the
first loaded package only.
Bug: 64948230
Bug: 65645766
Test: none
Change-Id: Ia804777fe3f963004a5c053129ef6e0c94b7f6a3
Merged-In: Ia804777fe3f963004a5c053129ef6e0c94b7f6a3
Accept a set of arguments separated by newlines.
This avoids path separator conflicts with the argument
format for passing splits.
Test: manual
Bug: 65645766
Change-Id: Ia68122cb77b7dde2292a0fd953e79f02996ac01c
Merged-In: Ia68122cb77b7dde2292a0fd953e79f02996ac01c
Permissions defined with the same leaf name emit the same
string symbol, which causes collisions. AAPT would override
the symbol with the last one seen.
Do the same thing as AAPT, but emit a warning.
Bug: 64472942
Bug: 65645766
Test: make aapt2_tests
Change-Id: I17b9dc7e8d8bd80db98869394c93695cb453bebd
Merged-In: I17b9dc7e8d8bd80db98869394c93695cb453bebd
This proto format is meant to encapsulate more information
that is specific to Android and allows for easier validation
and manipulation across tools.
Test: make aapt2_tests
Bug: 65645766
Change-Id: I13bc34a460671fc0a36246be0d287a3d37d244d6
Merged-In: I13bc34a460671fc0a36246be0d287a3d37d244d6
Gradle and other build tools that need to frequently interact with aapt2
can use the daemon mode to issue regular CLI commands without paying
the process start cost for each invocation.
Test: manual
Bug: 65645766
Change-Id: I543858ed46496278bad643f0dae688c924ec6c3f
Merged-In: I543858ed46496278bad643f0dae688c924ec6c3f
Once switched to using std::string, the mkdirs implementation was
trying to create an empty string when an absolute file path on
linux or macOS was used.
Bug: 62336414
Bug: 65645766
Test: manual
Change-Id: I52f3050b410a923ca48f353b0983667c16d00ee8
Merged-In: I52f3050b410a923ca48f353b0983667c16d00ee8
In preparation for exporting an XML proto format for UAM to consume,
this change brings the XML DOM API more in line with other APIs that
do not make the Namespace a separate node.
Treating Namespace declarations as just properties of an Element
node makes the implementation of algorithms much simpler, as
the constraints that Namespace nodes have only one child
are now built in and traversing to find Element nodes
is much simpler.
Also made a bunch of quality of life improvements, like formatting and
comment style.
Test: make aapt2_tests
Bug: 65645766
Change-Id: Ib97ff1c4252b7907e2cc1f13a448dc4ca3b809a4
Merged-In: Ib97ff1c4252b7907e2cc1f13a448dc4ca3b809a4
