Generate tests/Android.mk from gyp

Generate tests/Android.mk from gyp

gyp/apptype_console.gypi:
Don't include console app dependencies when building for Android framework.

gyp/tests.gyp:
Add/remove libraries for framework build.

platform_tools/android/bin/android_framework_gyp.py:
Moved to gyp_gen.
clean_up_gypd_files moved to this script.

platform_tools/android/bin/gyp_to_android.py:
Call new function for generating tool makefile.
Set LOCAL_MODULE.

platform_tools/android/gyp_gen/gypd_parser.py:
Support relative paths.

platform_tools/android/gyp_gen/makefile_writer.py:
Factor out helper functions to be used by tool writer.
LOCAL_MODULE is set elsewhere.

platform_tools/android/gyp_gen/tool_makefile_writer.py:
Multipurpose file for writing makefiles for tools. Should be able to use
it mostly unchanged for bench, gm, etc.

platform_tools/android/gyp_gen/vars_dict_lib.py:
Make the comments follow the style guide.
Add set().
Rename __li to __ordered_set

More/update tests:
platform_tools/android/tests/android_framework_gyp_tests.py
platform_tools/android/tests/expectations/Android.mk
platform_tools/android/tests/expectations/tool/Android.mk
platform_tools/android/tests/expectations/write_local_vars_append_arm
platform_tools/android/tests/expectations/write_local_vars_append_foo
platform_tools/android/tests/expectations/write_local_vars_append_no_name
platform_tools/android/tests/expectations/write_local_vars_no_append_arm
platform_tools/android/tests/expectations/write_local_vars_no_append_foo
platform_tools/android/tests/expectations/write_local_vars_no_append_no_name
platform_tools/android/tests/gyp_to_android_tests.py
platform_tools/android/tests/makefile_writer_tests.py
platform_tools/android/tests/ordered_set_tests.py
platform_tools/android/tests/test_variables.py

BUG=skia:2447
May require an update to the bot to remove pyc files.

Committed: http://code.google.com/p/skia/source/detail?r=14408

[scroggo, 2014-04-25 22:10:55]

https://codereview.chromium.org/235883015/diff/100001/Android.mk
File Android.mk (left):

https://codereview.chromium.org/235883015/diff/100001/Android.mk#oldcode661
Android.mk:661: # TODO include those sources!
These are included in tests now.

https://codereview.chromium.org/235883015/diff/100001/Android.mk
File Android.mk (right):

https://codereview.chromium.org/235883015/diff/100001/Android.mk#newcode564
Android.mk:564: LOCAL_MODULE := \
As usual, the Android.mk files are not checked into Skia, but I have included
them here for reference. The diff will show the actual diff when checking them
in.

https://codereview.chromium.org/235883015/diff/100001/platform_tools/android/...
File platform_tools/android/bin/android_framework_gyp.py (left):

https://codereview.chromium.org/235883015/diff/100001/platform_tools/android/...
platform_tools/android/bin/android_framework_gyp.py:9: Modified version of
gyp_skia, used by gyp_to_android.py to generate Android.mk
This file has actually been moved into platform_tools/android/gyp_gen.

https://codereview.chromium.org/235883015/diff/100001/platform_tools/android/...
File platform_tools/android/bin/gyp_to_android.py (left):

https://codereview.chromium.org/235883015/diff/100001/platform_tools/android/...
platform_tools/android/bin/gyp_to_android.py:38: def clean_gypd_files(folder):
This moved to android_framework_gyp.py, unchanged besides comments.

[djsollen, 2014-04-28 13:52:22]

just one comment and I'll let someone else review the python code.

If putting a *.a file in gyp translates to something else can we at a minimum
document that behavior?

https://codereview.chromium.org/235883015/diff/100001/gyp/tests.gyp
File gyp/tests.gyp (right):

https://codereview.chromium.org/235883015/diff/100001/gyp/tests.gyp#newcode22
gyp/tests.gyp:22: 'stlport_static.a',
are we sure that is the right way to include stlport?  Do we still need it for
something in tests?

[scroggo, 2014-04-28 14:37:24]

On 2014/04/28 13:52:22, djsollen wrote:
> If putting a *.a file in gyp translates to something else can we at a minimum
> document that behavior?

Patch set 2 adds a comment in gypd_parser. I think that's not incorrect, but I'm
not sure it's the only right place for it. gyp_to_android.py probably isn't the
right place for it either, since that is just the entry point. It probably
belongs
in a doc/on a webpage somewhere. I have filed skbug.com/2467.) FWIW, this is not
new behavior, and IIRC, it mimics behavior in the other undocumented Android.mk
generator.

> just one comment and I'll let someone else review the python code.

Hal, will you please review the python code?

https://codereview.chromium.org/235883015/diff/100001/gyp/tests.gyp
File gyp/tests.gyp (right):

https://codereview.chromium.org/235883015/diff/100001/gyp/tests.gyp#newcode22
gyp/tests.gyp:22: 'stlport_static.a',
On 2014/04/28 13:52:23, djsollen wrote:
> are we sure that is the right way to include stlport? 

This matches what the makefile already stated.

> Do we still need it for something in tests?

It turns out we do not. Removed.

[hal.canary, 2014-04-28 15:35:18]

lgtm

[scroggo, 2014-04-28 15:37:19]

The CQ bit was checked by scroggo@google.com

[commit-bot: I haz the power, 2014-04-28 15:37:39]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/scroggo@google.com/235883015/120001

[commit-bot: I haz the power, 2014-04-28 15:37:45]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-28 15:37:45]

Failed to apply patch for Android.mk:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  can't find file to patch at input line 6
  Perhaps you used the wrong -p or --strip option?
  The text leading up to this was:
  --------------------------
  |Index: Android.mk
  |diff --git a/Android.mk b/Android.mk
  |index
6c967df24b5a9f0c14442b3f6dd66b17bc4747ce..2e9a31e31819a50f975a455e83fd2f69d30c0aaf
100644
  |--- a/Android.mk
  |+++ b/Android.mk
  --------------------------
  No file to patch.  Skipping patch.
  3 out of 3 hunks ignored

Patch:       Android.mk
Index: Android.mk
diff --git a/Android.mk b/Android.mk
index
6c967df24b5a9f0c14442b3f6dd66b17bc4747ce..2e9a31e31819a50f975a455e83fd2f69d30c0aaf
100644
--- a/Android.mk
+++ b/Android.mk
@@ -561,6 +561,9 @@ LOCAL_EXPORT_C_INCLUDE_DIRS := \
 	$(LOCAL_PATH)/include/utils \
 	$(LOCAL_PATH)/src/utils
 
+LOCAL_MODULE := \
+	libskia
+
 LOCAL_SRC_FILES_arm += \
 	src/core/SkUtilsArm.cpp \
 	src/opts/memset.arm.S \
@@ -641,7 +644,6 @@ LOCAL_SRC_FILES_arm64 += \
 	src/opts/SkXfermode_opts_arm_neon.cpp
 
 include external/stlport/libstlport.mk
-LOCAL_MODULE:= libskia
 include $(BUILD_SHARED_LIBRARY)
 
 #############################################################
@@ -655,7 +657,4 @@ include $(BUILD_SHARED_LIBRARY)
 #include $(BASE_PATH)/gm/Android.mk
 
 # unit-tests
-#include $(BASE_PATH)/tests/Android.mk
-
-# pathOps unit-tests
-# TODO include those sources!
+include $(BASE_PATH)/tests/Android.mk

[scroggo, 2014-04-28 15:50:43]

The CQ bit was checked by scroggo@google.com

[commit-bot: I haz the power, 2014-04-28 15:51:00]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/scroggo@google.com/235883015/140001

[commit-bot: I haz the power, 2014-04-28 16:00:59]

Change committed as 14408

[epoger, 2014-04-28 18:44:51]

Looks like this broke runandroidplatformselftests ; see
http://108.170.220.102:10117/builders/Housekeeper-PerCommit/builds/1042

[scroggo, 2014-04-28 18:50:53]

On 2014/04/28 18:44:51, epoger wrote:
> Looks like this broke runandroidplatformselftests ; see
> http://108.170.220.102:10117/builders/Housekeeper-PerCommit/builds/1042

Ah yes. I got a similar error on the merge bot. Since the bot had previously run
these scripts, it had .pyc files laying around with the old code. It calls them
instead of the new versions. On the merge bot I logged in and deleted those
files. Is there a better way?

(Also, I seem to recall there is a way to log in to the buildbots, but I'm not
sure how to log in to this one.)

[epoger, 2014-04-28 18:53:47]

On 2014/04/28 18:50:53, scroggo wrote:
> On 2014/04/28 18:44:51, epoger wrote:
> > Looks like this broke runandroidplatformselftests ; see
> > http://108.170.220.102:10117/builders/Housekeeper-PerCommit/builds/1042
> 
> Ah yes. I got a similar error on the merge bot. Since the bot had previously
run
> these scripts, it had .pyc files laying around with the old code. It calls
them
> instead of the new versions. On the merge bot I logged in and deleted those
> files. Is there a better way?
> 
> (Also, I seem to recall there is a way to log in to the buildbots, but I'm not
> sure how to log in to this one.)

Very strange.  When .py files are updated, Python should regenerate the .pyc
files automatically.

We should not need to log into any machines and clean out the .pyc files.

I will open a bug.

[epoger, 2014-04-28 19:00:27]

On 2014/04/28 18:53:47, epoger wrote:
> On 2014/04/28 18:50:53, scroggo wrote:
> > On 2014/04/28 18:44:51, epoger wrote:
> > > Looks like this broke runandroidplatformselftests ; see
> > > http://108.170.220.102:10117/builders/Housekeeper-PerCommit/builds/1042
> > 
> > Ah yes. I got a similar error on the merge bot. Since the bot had previously
> run
> > these scripts, it had .pyc files laying around with the old code. It calls
> them
> > instead of the new versions. On the merge bot I logged in and deleted those
> > files. Is there a better way?
> > 
> > (Also, I seem to recall there is a way to log in to the buildbots, but I'm
not
> > sure how to log in to this one.)
> 
> Very strange.  When .py files are updated, Python should regenerate the .pyc
> files automatically.
> 
> We should not need to log into any machines and clean out the .pyc files.
> 
> I will open a bug.

Filed http://skbug.com/2468 ('pyc file cleanup required after r14408 ('Generate
tests/Android.mk from gyp')')