Reland of land #2 of Enable whitelist generation for official builds.

tools/resources/generate_resource_whitelist.py

Index: tools/resources/generate_resource_whitelist.py
diff --git a/tools/resources/generate_resource_whitelist.py b/tools/resources/generate_resource_whitelist.py
new file mode 100755
index 0000000000000000000000000000000000000000..527ceb484158c9fcf32cb5d9dc34e7561d7471d9
--- /dev/null
+++ b/tools/resources/generate_resource_whitelist.py
@@ -0,0 +1,82 @@
+#!/usr/bin/env python
+# Copyright 2016 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.
+
+import argparse
+import os
+import re
+import sys
+
+USAGE = """generate_resource_whitelist.py [-h] [-i INPUT] [-o OUTPUT]
+
+INPUT specifies a file containing existing resource IDs that should be
+whitelisted, where each line of INPUT contains a single resource ID.
+
+Creates a resource whitelist by collecting existing resource IDs that are part
+of unknown pragma warnings and adds additional arch specfic resource IDs.
+
+This script is used to find the resources that are actually compiled in Chrome
+in order to only include the needed strings/images in Chrome PAK files.
+These resource IDs show up in the build output after building Chrome with
+gn variable enable_resource_whitelist_generation set to true.
+This causes the compiler to print out an UnknownPragma message every time a
+resource ID is used.
+
+E.g. foo.cc:22:0: warning: ignoring #pragma whitelisted_resource_12345
+[-Wunknown-pragmas]
+
+On Windows, the message is simply a message via __pragma(message(...)).
+
+"""
+
+COMPONENTS_STRINGS_HEADER = 'gen/components/strings/grit/components_strings.h'
+
+# We don't want the resources are different between 32-bit and 64-bit build,
+# added arch related resources even they are not used.
+ARCH_SPECIFIC_RESOURCES = [
+  'IDS_VERSION_UI_64BIT',
+  'IDS_VERSION_UI_32BIT',
+]
+
+
+def _FindResourceIds(header, resource_names):
+  """Returns the numerical resource IDs that correspond to the given resource
+     names, as #defined in the given header file."
+  """
+  pattern = re.compile(
+      r'^#define (%s) _Pragma\S+ (\d+)$' % '|'.join(resource_names))
+  with open(header, 'r') as f:
+    res_ids = [ int(pattern.match(line).group(2))
+                 for line in f if pattern.match(line) ]
+  if len(res_ids) != len(resource_names):
+    raise Exception('Find resource id failed: the result is ' +
+                    ', '.join(str(i) for i in res_ids))
+  return set(res_ids)
+
+
+def main():
+  parser = argparse.ArgumentParser(usage=USAGE)
+  parser.add_argument(
+      '-i', '--input', type=argparse.FileType('r'), default=sys.stdin,
+      help='A resource whitelist where each line contains one resource ID')
+  parser.add_argument(
+      '-o', '--output', type=argparse.FileType('w'), default=sys.stdout,
+      help='The resource list path to write (default stdout)')
+  parser.add_argument(
+      '--out-dir', required=True,
+      help='The out target directory, for example out/Release')
+
+  args = parser.parse_args()
+
+  used_resources = set()
+  used_resources.update([int(resource_id) for resource_id in args.input])
+  used_resources |= _FindResourceIds(
+      os.path.join(args.out_dir, COMPONENTS_STRINGS_HEADER),
+      ARCH_SPECIFIC_RESOURCES)
+
+  for resource_id in sorted(used_resources):
+    args.output.write('%d\n' % resource_id)
+
+if __name__ == '__main__':
+  main()