Loading in an APk changed the ordering of strings in the string pool.
When loading an apk, assign the strings to the same index as they
are in the ResStringPool.
Bug: 118831219
Test: "aapt2 dump strings left.apk" prints in the correct order,
"aapt2 convert left.apk --output-format binary -o left_binary.apk" has
entries in the correct order, and aapt2_tests
Change-Id: I00014c02195f39c1152a110e90083d9b14e9216e
Codepoints that are encoded to 4 bytes in UTF-8 are not allowed in
Modified UTF-8. They instead should be encoded as surrogate pairs in the
same way that CESU-8 allows for surrogate pairs. This will also cause 4
byte UTF-8 codes to be represented in 6 bytes.
Bug: 37140916
Test: aapt2_tests
Change-Id: I155dc24f166139d1d36a16bac088dcfcd59eb321
b/77862560 detected that when converting an apk to binary using aapt2,
all resource ids of attributes that have been replaced with resource
identifiers become set to the identifier of the first attribute. This is
because the attribute names are all empty because the names are not necessary
since the resource ids are present. The empty attribute names all map to
the same string pool reference and cause all the ids to be the first
empty string into the string pool.
The ag/3897499 approach to fix the specified bug was extremely inefficient
and caused hour long builds. This change takes advantage of the multimap
data structure to do lookups efficiently.
Bug: 77862560
Test: Converted apk in listed bug from proto to binary and observed
correct resource ids and correct badging. Also built the Android tree to
check for regressions in build time.
Change-Id: I27a9ee4ffbed8b9ff6f238ad315cdf87b588947c
This reverts commit 35ecb89a8aa68f24d2e991df5bb9964ad15075dc.
Reason for revert: <Long make duration (infinite loop)>
Change-Id: I10d1cf53ca3054d40e23b06368ebaff6af69beab
b/77862560 detected that when converting an apk to binary using aapt2,
all resource ids of attributes that have been replaced with resource
identifiers become set to the identifier of the first attribute. This is
because the attribute names are all empty because the names are not necessary
since the resource ids are present. The empty attribute names all map to
the same string pool reference and cause all the ids to be the first
empty string into the string pool.
Bug: 77862560
Test: Converted apk in listed bug from proto to binary and observed
correct resource ids and correct badging.
Change-Id: I635c13cd1ad7a395fe40a57198cfe5ec91602d01
Test: Tested for normal functionality when string does not exceed
maximum length and tests for detection of string that is too lonhg for
UTF8i
Bug: b/74176037
Change-Id: Ic71d3671a069e7012e8ca107e79e071499eebbf6
(cherry picked from commit a15c2a8957b9883cb293fdacaeabd7f2e037a0a5)
Keep strings sorted by configuration so that strings likely
to be selected (all match the same locale, for instance) are
close together.
Bug: 67958501
Test: make aapt2_tests
Change-Id: Id17d93bf2e03ce408a6f619d3ea6dc313e393b76
Move the styled strings to a separate section of the StringPool so
that sorting can never mess up the order of Styles.
Bug: 63570514
Test: make aapt2_tests
Change-Id: Id2ce1355b92be1bb31ce0daa7e54ae9b5b6c2ffe
The XLIFF 'g' tag specifies content that should NOT be translated.
AAPT2's pseudolocalization process should respect it.
Bug:34064599
Test: make libandroidfw_tests
Change-Id: Ice437d7f0ff246730ee04896fd035e2d846148fb
libandroidfw needs to make use of StringPiece, so
move it to libandroidfw and update all code referencing
StringPiece in aapt2.
Test: make libandroidfw_tests libaapt2_tests
Change-Id: I68d7f0fc7c651b048d9d1f5e7971f10ef5349fa1
Use Google3 naming style to match new
projects' and open source google projects' style.
Preferred to do this in a massive CL so as to avoid
style inconsistencies that plague legacy code bases.
This is a relatively NEW code base, may as well keep
it up to date.
Test: name/style refactor - existing tests pass
Change-Id: Ie80ecb78d46ec53efdfca2336bb57d96cbb7fb87
For legacy reasons, we kept around the use of UTF-16 internally
in AAPT2. We don't need this and this CL removes all instances of
std::u16string and StringPiece16. The only places still needed
are when interacting with the ResTable APIs that only operate in
UTF16.
Change-Id: I492475b84bb9014fa13bf992cff447ee7a5fe588
- Add option to rename package in AndroidManifest.xml
- Support default versionName and versionCode
- Accept True and False as valid booleans
Change-Id: I400e350b9dcd0fd1c197d1929144299c7823617d
An early refactor. Some ideas became clearer as
development continued. Now the various phases are much
clearer and more easily reusable.
Also added a ton of tests!
Change-Id: Ic8f0a70c8222370352e63533b329c40457c0903e
Previously, you could only reference namespace prefixes in attribute names:
<View xmlns:appcompat="http://schemas.android.com/apk/res/android.support.v7.appcompat"
appcompat:name="hey"
...
Now you can also reference them in resource names within an attribute value:
...
android:text="@appcompat:string/confirm"
...
Which will be treated as "@android.support.v7.appcompat:string/confirm".
Change-Id: Ib076e867a990c80cf877a704eb77cd1ef0b23b52
First checking of AAPT2. The individual phases of AAPT2 work, but there
are some missing pieces.
For early testing we are missing:
- Need to properly mark file references and include them in package
- Need to package into zip
Final AAPT for apps we are missing:
- Need to crush PNGs
- Need to parse 9-patches
- Need to validate all of AndroidManifest.xml
- Need to write align method to align resource tables for splits.
Final AAPT for apps + system we are missing:
- Need to handle overlays
- Need to store comments for R file
- Need to handle --shared-lib (dynamic references too).
New AAPT features coming:
- Need to import compiled libraries
- Name mangling
- R file generation for library code
Change-Id: I95f8a63581b81a1f424ae6fb2c373c883b72c18d
