diagnose_apk_bloat.py: add native diffs.

tools/binary_size/diagnose_apk_bloat.py

 #!/usr/bin/env python
 # Copyright 2017 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """Tool for finding the cause of APK bloat.
 
 Run diagnose_apk_bloat.py -h for detailed usage help.
 """
 
 import argparse
 import collections
 from contextlib import contextmanager
 import distutils.spawn
 import json
 import multiprocessing
 import os
+import re
 import shutil
 import subprocess
 import sys
 import tempfile
 import zipfile
 
 _COMMIT_COUNT_WARN_THRESHOLD = 15
 _ALLOWED_CONSECUTIVE_FAILURES = 2
+_DIFF_DETAILS_LINES_THRESHOLD = 100
 _BUILDER_URL = \
     'https://build.chromium.org/p/chromium.perf/builders/Android%20Builder'
 _CLOUD_OUT_DIR = os.path.join('out', 'Release')
 _SRC_ROOT = os.path.abspath(
     os.path.join(os.path.dirname(__file__), os.pardir, os.pardir))
 _DEFAULT_ARCHIVE_DIR = os.path.join(_SRC_ROOT, 'binary-size-bloat')
 _DEFAULT_OUT_DIR = os.path.join(_SRC_ROOT, 'out', 'diagnose-apk-bloat')
 _DEFAULT_TARGET = 'monochrome_public_apk'
 
 
@@ skipping 11 matching lines @@
   def __init__(self, name):
     self.name = name
     self.banner = '\n' + '*' * 30 + name + '*' * 30
 
   def AppendResults(self, logfile):
     """Print and write diff results to an open |logfile|."""
     _PrintAndWriteToFile(logfile, self.banner)
     _PrintAndWriteToFile(logfile, 'Summary:')
     _PrintAndWriteToFile(logfile, self.Summary())
     _PrintAndWriteToFile(logfile, '\nDetails:')
-    for l in self.DetailedResults():
-      _PrintAndWriteToFile(logfile, l)
+    _PrintAndWriteToFile(logfile, self.DetailedResults())
 
   def Summary(self):
     """A short description that summarizes the source of binary size bloat."""
     raise NotImplementedError()
 
   def DetailedResults(self):
     """An iterable description of the cause of binary size bloat."""
     raise NotImplementedError()
 
   def ProduceDiff(self, archive_dirs):
     """Prepare a binary size diff with ready to print results."""
     raise NotImplementedError()
 
   def RunDiff(self, logfile, archive_dirs):
-    _Print('Creating {}', self.name)
     self.ProduceDiff(archive_dirs)
     self.AppendResults(logfile)
 
 
+class NativeDiff(BaseDiff):
+  _RE_SUMMARY = re.compile(
+      r'.*(Section Sizes .*? object files added, \d+ removed).*',
+      flags=re.DOTALL)
+
+  def __init__(self, size_name, supersize_path):
+    self._size_name = size_name
+    self._supersize_path = supersize_path
+    self._diff = []
+    super(NativeDiff, self).__init__('Native Diff')
+
+  def DetailedResults(self):
+    return self._diff.splitlines()
+
+  def Summary(self):
+    return NativeDiff._RE_SUMMARY.match(self._diff).group(1)
+
+  def ProduceDiff(self, archive_dirs):
+    size_files = [os.path.join(a, self._size_name)
+                  for a in reversed(archive_dirs)]
+    q = 'Print(Diff(size_info1, size_info2), use_pager=False);'

[agrieve, 2017/04/24 20:43:21]

nit: shouldn't need to specify use_pager=False. It should default to false if it
detects the output is being piped.

[estevenson, 2017/04/24 22:39:40]

Switched to diff instead of console. Done.

+    cmd = [self._supersize_path, 'console'] + size_files + ['--query', q]
+    self._diff = _RunCmd(cmd)[0].replace('{', '{{').replace('}', '}}')
+
+
 _ResourceSizesDiffResult = collections.namedtuple(
     'ResourceSizesDiffResult', ['section', 'value', 'units'])
 
 
 class ResourceSizesDiff(BaseDiff):
   _RESOURCE_SIZES_PATH = os.path.join(
       _SRC_ROOT, 'build', 'android', 'resource_sizes.py')
 
   def __init__(self, apk_name, slow_options=False):
     self._apk_name = apk_name
     self._slow_options = slow_options
     self._diff = None  # Set by |ProduceDiff()|
     super(ResourceSizesDiff, self).__init__('Resource Sizes Diff')
 
   def DetailedResults(self):
-    for section, value, units in self._diff:
-      yield '{:>+10,} {} {}'.format(value, units, section)
+    return ['{:>+10,} {} {}'.format(value, units, section)
+            for section, value, units in self._diff]
 
   def Summary(self):
     for s in self._diff:
       if 'normalized' in s.section:
         return 'Normalized APK size: {:+,} {}'.format(s.value, s.units)
     return ''
 
   def ProduceDiff(self, archive_dirs):
     chartjsons = self._RunResourceSizes(archive_dirs)
     diff = []
@@ skipping 54 matching lines @@
   def apk_name(self):
     # Only works on apk targets that follow: my_great_apk naming convention.
     apk_name = ''.join(s.title() for s in self.target.split('_')[:-1]) + '.apk'
     return apk_name.replace('Webview', 'WebView')
 
   @property
   def apk_path(self):
     return os.path.join('apks', self.apk_name)
 
   @property
-  def main_lib_name(self):
+  def main_lib_path(self):
     # TODO(estevenson): Get this from GN instead of hardcoding.
     if self.IsLinux():
       return 'chrome'
     elif 'monochrome' in self.target:
       return 'lib.unstripped/libmonochrome.so'
     else:
       return 'lib.unstripped/libchrome.so'
 
   @property
-  def main_lib_path(self):
-    return os.path.join(self.output_directory, self.main_lib_name)
+  def abs_main_lib_path(self):
+    return os.path.join(self.output_directory, self.main_lib_path)
 
   @property
-  def map_file_name(self):
-    return self.main_lib_name + '.map.gz'
+  def map_file_path(self):
+    return self.main_lib_path + '.map.gz'
+
+  @property
+  def size_name(self):
+    return os.path.splitext(os.path.basename(self.main_lib_path))[0] + '.size'
 
   def _SetDefaults(self):
     has_goma_dir = os.path.exists(os.path.join(os.path.expanduser('~'), 'goma'))
     self.use_goma = self.use_goma or has_goma_dir
     self.max_load_average = (self.max_load_average or
                              str(multiprocessing.cpu_count()))
     if not self.max_jobs:
       self.max_jobs = '10000' if self.use_goma else '500'
 
     if os.path.exists(os.path.join(os.path.dirname(_SRC_ROOT), 'src-internal')):
@@ skipping 40 matching lines @@
 
 class _BuildArchive(object):
   """Class for managing a directory with build results and build metadata."""
   def __init__(self, rev, base_archive_dir, build, subrepo):
     self.build = build
     self.dir = os.path.join(base_archive_dir, rev)
     metadata_path = os.path.join(self.dir, 'metadata.txt')
     self.rev = rev
     self.metadata = _GenerateMetadata([self], build, metadata_path, subrepo)
 
-  def ArchiveBuildResults(self, bs_dir):
+  def ArchiveBuildResults(self, supersize_path):
     """Save build artifacts necessary for diffing."""
     _Print('Saving build results to: {}', self.dir)
     _EnsureDirsExist(self.dir)
     build = self.build
-    self._ArchiveFile(build.main_lib_path)
-    lib_name_noext = os.path.splitext(os.path.basename(build.main_lib_path))[0]
-    size_path = os.path.join(self.dir, lib_name_noext + '.size')
-    supersize_path = os.path.join(bs_dir, 'supersize')
+    self._ArchiveFile(build.abs_main_lib_path)
     tool_prefix = _FindToolPrefix(build.output_directory)
+    size_path = os.path.join(self.dir, build.size_name)
     supersize_cmd = [supersize_path, 'archive', size_path, '--elf-file',
-                     build.main_lib_path, '--tool-prefix', tool_prefix,
+                     build.abs_main_lib_path, '--tool-prefix', tool_prefix,
                      '--output-directory', build.output_directory,
                      '--no-source-paths']
     if build.IsAndroid():
       supersize_cmd += ['--apk-file', build.abs_apk_path]
       self._ArchiveFile(build.abs_apk_path)
 
     _RunCmd(supersize_cmd)
     _WriteMetadata(self.metadata)
 
   def Exists(self):
@@ skipping 27 matching lines @@
       _Print('Skipping diff for {} due to missing build archives.', diff_path)
       return
 
     metadata_path = self._DiffMetadataPath(archives)
     metadata = _GenerateMetadata(
         archives, self.build, metadata_path, self.subrepo)
     if _MetadataExists(metadata):
       _Print('Skipping diff for {} and {}. Matching diff already exists: {}',
              archives[0].rev, archives[1].rev, diff_path)
     else:
+      if os.path.exists(diff_path):
+        os.remove(diff_path)
       archive_dirs = [archives[0].dir, archives[1].dir]
       with open(diff_path, 'a') as diff_file:
         for d in self.diffs:
           d.RunDiff(diff_file, archive_dirs)
+        _Print('See detailed diff results here: {}.', diff_path)
       _WriteMetadata(metadata)
 
   def _CanDiff(self, archives):
     return all(a.Exists() for a in archives)
 
   def _DiffFilePath(self, archives):
     return os.path.join(self._DiffDir(archives), 'diff_results.txt')
 
   def _DiffMetadataPath(self, archives):
     return os.path.join(self._DiffDir(archives), 'metadata.txt')
@@ skipping 154 matching lines @@
     _Print('Failure: please ensure working directory is clean.')
     sys.exit()
 
 
 def _Die(s, *args, **kwargs):
   _Print('Failure: ' + s, *args, **kwargs)
   _global_restore_checkout_func()
   sys.exit(1)
 
 
-def _DownloadBuildArtifacts(archive, build, bs_dir, depot_tools_path):
+def _DownloadBuildArtifacts(archive, build, supersize_path, depot_tools_path):
   """Download artifacts from arm32 chromium perf builder."""
   if depot_tools_path:
     gsutil_path = os.path.join(depot_tools_path, 'gsutil.py')
   else:
     gsutil_path = distutils.spawn.find_executable('gsutil.py')
 
   if not gsutil_path:
     _Die('gsutil.py not found, please provide path to depot_tools via '
          '--depot-tools-path or add it to your PATH')
 
   download_dir = tempfile.mkdtemp(dir=_SRC_ROOT)
   try:
-    _DownloadAndArchive(gsutil_path, archive, download_dir, build, bs_dir)
+    _DownloadAndArchive(
+        gsutil_path, archive, download_dir, build, supersize_path)
   finally:
     shutil.rmtree(download_dir)
 
 
-def _DownloadAndArchive(gsutil_path, archive, dl_dir, build, bs_dir):
+def _DownloadAndArchive(gsutil_path, archive, dl_dir, build, supersize_path):
   dl_file = 'full-build-linux_%s.zip' % archive.rev
   dl_url = 'gs://chrome-perf/Android Builder/%s' % dl_file
   dl_dst = os.path.join(dl_dir, dl_file)
   _Print('Downloading build artifacts for {}', archive.rev)
   # gsutil writes stdout and stderr to stderr, so pipe stdout and stderr to
   # sys.stdout.
   retcode = subprocess.call([gsutil_path, 'cp', dl_url, dl_dir],
                              stdout=sys.stdout, stderr=subprocess.STDOUT)
   if retcode:
       _Die('unexpected error while downloading {}. It may no longer exist on '
            'the server or it may not have been uploaded yet (check {}). '
            'Otherwise, you may not have the correct access permissions.',
            dl_url, _BUILDER_URL)
 
   # Files needed for supersize and resource_sizes. Paths relative to out dir.
-  to_extract = [build.main_lib_name, build.map_file_name, 'args.gn',
+  to_extract = [build.main_lib_path, build.map_file_path, 'args.gn',
                 'build_vars.txt', build.apk_path]
   extract_dir = os.path.join(os.path.splitext(dl_dst)[0], 'unzipped')
   # Storage bucket stores entire output directory including out/Release prefix.
   _Print('Extracting build artifacts')
   with zipfile.ZipFile(dl_dst, 'r') as z:
     _ExtractFiles(to_extract, _CLOUD_OUT_DIR, extract_dir, z)
     dl_out = os.path.join(extract_dir, _CLOUD_OUT_DIR)
     build.output_directory, output_directory = dl_out, build.output_directory
-    archive.ArchiveBuildResults(bs_dir)
+    archive.ArchiveBuildResults(supersize_path)
     build.output_directory = output_directory
 
 
 def _ExtractFiles(to_extract, prefix, dst, z):
   zip_infos = z.infolist()
   assert all(info.filename.startswith(prefix) for info in zip_infos), (
       'Storage bucket folder structure doesn\'t start with %s' % prefix)
   to_extract = [os.path.join(prefix, f) for f in to_extract]
   for f in to_extract:
     z.extract(f, path=dst)
 
 
 def _Print(s, *args, **kwargs):
   print s.format(*args, **kwargs)
 
 
 def _PrintAndWriteToFile(logfile, s):
-  """Print |s| to |logfile| and stdout."""
-  _Print(s)
-  logfile.write('%s\n' % s)
+  """Write and print |s| thottling output if |s| is a large list."""
+  if isinstance(s, basestring):
+    _Print(s)
+  else:
+    for l in s[:_DIFF_DETAILS_LINES_THRESHOLD]:
+      _Print(l)
+    if len(s) > _DIFF_DETAILS_LINES_THRESHOLD:
+      _Print('\nOutput truncated, see {} for more.', logfile.name)
+  logfile.write('\n'.join(s))
 
 
 @contextmanager
-def _TmpBinarySizeDir():
-  """Recursively copy files to a temp dir and yield the tmp binary_size dir."""
+def _TmpCopyBinarySizeDir():
+  """Recursively copy files to a temp dir and yield supersize path."""
   # Needs to be at same level of nesting as the real //tools/binary_size
   # since supersize uses this to find d3 in //third_party.
   tmp_dir = tempfile.mkdtemp(dir=_SRC_ROOT)
   try:
     bs_dir = os.path.join(tmp_dir, 'binary_size')
     shutil.copytree(os.path.join(_SRC_ROOT, 'tools', 'binary_size'), bs_dir)
-    yield bs_dir
+    yield os.path.join(bs_dir, 'supersize')
   finally:
     shutil.rmtree(tmp_dir)
 
 
 def main():
   parser = argparse.ArgumentParser(
       description='Find the cause of APK size bloat.')
   parser.add_argument('--archive-directory',
                       default=_DEFAULT_ARCHIVE_DIR,
                       help='Where results are stored.')
@@ skipping 63 matching lines @@
     if args.subrepo:
       parser.error('--subrepo doesn\'t work with --cloud')
 
   subrepo = args.subrepo or _SRC_ROOT
   _EnsureDirectoryClean(subrepo)
   _SetRestoreFunc(subrepo)
   revs = _GenerateRevList(args.rev,
                           args.reference_rev or args.rev + '^',
                           args.all,
                           subrepo)
-  diffs = []
-  if build.IsAndroid():
-    diffs +=  [
-        ResourceSizesDiff(
-            build.apk_name, slow_options=args.include_slow_options)
-    ]
-  diff_mngr = _DiffArchiveManager(
-      revs, args.archive_directory, diffs, build, subrepo)
-  consecutive_failures = 0
-  with _TmpBinarySizeDir() as bs_dir:
+  with _TmpCopyBinarySizeDir() as supersize_path:
+    diffs = [NativeDiff(build.size_name, supersize_path)]
+    if build.IsAndroid():
+      diffs +=  [
+          ResourceSizesDiff(
+              build.apk_name, slow_options=args.include_slow_options)
+      ]
+    diff_mngr = _DiffArchiveManager(
+        revs, args.archive_directory, diffs, build, subrepo)
+    consecutive_failures = 0
     for i, archive in enumerate(diff_mngr.IterArchives()):
       if archive.Exists():
         _Print('Found matching metadata for {}, skipping build step.',
                archive.rev)
       else:
         if build.IsCloud():
-          _DownloadBuildArtifacts(archive, build, bs_dir, args.depot_tools_path)
+          _DownloadBuildArtifacts(
+              archive, build, supersize_path, args.depot_tools_path)
         else:
           build_success = _SyncAndBuild(archive, build, subrepo)
           if not build_success:
             consecutive_failures += 1
             if consecutive_failures > _ALLOWED_CONSECUTIVE_FAILURES:
               _Die('{} builds failed in a row, last failure was {}.',
                    consecutive_failures, archive.rev)
           else:
-            archive.ArchiveBuildResults(bs_dir)
+            archive.ArchiveBuildResults(supersize_path)
             consecutive_failures = 0
 
       if i != 0:
         diff_mngr.MaybeDiff(i - 1, i)
 
   _global_restore_checkout_func()
 
 if __name__ == '__main__':
   sys.exit(main())