tools/resources/find_unused_resources.py

   1 #!/usr/bin/env python
   2 # Copyright 2013 The Chromium Authors. All rights reserved.
   3 # Use of this source code is governed by a BSD-style license that can be
   4 # found in the LICENSE file.
   5
   6 """This script searches for unused art assets listed in a .grd file.
   7
   8 It uses git grep to look for references to the IDR resource id or the base
   9 filename. If neither is found, the file is reported unused.
  10
  11 Requires a git checkout. Must be run from your checkout's "src" root.
  12
  13 Example:
  14   cd /work/chrome/src
  15   tools/resources/find_unused_resouces.py ash/resources/ash_resources.grd
  16 """
  17
  18 __author__ = 'jamescook@chromium.org (James Cook)'
  19
  20
  21 import os
  22 import re
  23 import subprocess
  24 import sys
  25
  26
  27 def GetBaseResourceId(resource_id):
  28   """Removes common suffixes from a resource ID.
  29
  30   Removes suffixies that may be added by macros like IMAGE_GRID or IMAGE_BORDER.
  31   For example, converts IDR_FOO_LEFT and IDR_FOO_RIGHT to just IDR_FOO.
  32
  33   Args:
  34     resource_id: String resource ID.
  35
  36   Returns:
  37     A string with the base part of the resource ID.
  38   """
  39   suffixes = [
  40       '_TOP_LEFT', '_TOP', '_TOP_RIGHT',
  41       '_LEFT', '_CENTER', '_RIGHT',
  42       '_BOTTOM_LEFT', '_BOTTOM', '_BOTTOM_RIGHT',
  43       '_TL', '_T', '_TR',
  44       '_L', '_M', '_R',
  45       '_BL', '_B', '_BR']
  46   # Note: This does not check _HOVER, _PRESSED, _HOT, etc. as those are never
  47   # used in macros.
  48   for suffix in suffixes:
  49     if resource_id.endswith(suffix):
  50       resource_id = resource_id[:-len(suffix)]
  51   return resource_id
  52
  53
  54 def FindFilesWithContents(string_a, string_b):
  55   """Returns list of paths of files that contain |string_a| or |string_b|.
  56
  57   Uses --name-only to print the file paths. The default behavior of git grep
  58   is to OR together multiple patterns.
  59
  60   Args:
  61     string_a: A string to search for (not a regular expression).
  62     string_b: As above.
  63
  64   Returns:
  65     A list of file paths as strings.
  66   """
  67   matching_files = subprocess.check_output([
  68       'git', 'grep', '--name-only', '--fixed-strings', '-e', string_a,
  69       '-e', string_b])
  70   files_list = matching_files.split('\n')
  71   # The output ends in a newline, so slice that off.
  72   files_list = files_list[:-1]
  73   return files_list
  74
  75
  76 def GetUnusedResources(grd_filepath):
  77   """Returns a list of resources that are unused in the code.
  78
  79   Prints status lines to the console because this function is quite slow.
  80
  81   Args:
  82     grd_filepath: Path to a .grd file listing resources.
  83
  84   Returns:
  85     A list of pairs of [resource_id, filepath] for the unused resources.
  86   """
  87   unused_resources = []
  88   grd_file = open(grd_filepath, 'r')
  89   grd_data = grd_file.read()
  90   print 'Checking:'
  91   # Match the resource id and file path out of substrings like:
  92   # ...name="IDR_FOO_123" file="common/foo.png"...
  93   # by matching between the quotation marks.
  94   pattern = re.compile(
  95       r"""name="([^"]*)"  # Match resource ID between quotes.
  96       \s*                 # Run of whitespace, including newlines.
  97       file="([^"]*)"      # Match file path between quotes.""",
  98       re.VERBOSE)
  99   # Use finditer over the file contents because there may be newlines between
 100   # the name and file attributes.
 101   searched = set()
 102   for result in pattern.finditer(grd_data):
 103     # Extract the IDR resource id and file path.
 104     resource_id = result.group(1)
 105     filepath = result.group(2)
 106     filename = os.path.basename(filepath)
 107     base_resource_id = GetBaseResourceId(resource_id)
 108
 109     # Do not bother repeating searches.
 110     key = (base_resource_id, filename)
 111     if key in searched:
 112       continue
 113     searched.add(key)
 114
 115     # Print progress as we go along.
 116     print resource_id
 117
 118     # Ensure the resource isn't used anywhere by checking both for the resource
 119     # id (which should appear in C++ code) and the raw filename (in case the
 120     # file is referenced in a script, test HTML file, etc.).
 121     matching_files = FindFilesWithContents(base_resource_id, filename)
 122
 123     # Each file is matched once in the resource file itself. If there are no
 124     # other matching files, it is unused.
 125     if len(matching_files) == 1:
 126       # Give the user some happy news.
 127       print 'Unused!'
 128       unused_resources.append([resource_id, filepath])
 129
 130   return unused_resources
 131
 132
 133 def GetScaleDirectories(resources_path):
 134   """Returns a list of paths to per-scale-factor resource directories.
 135
 136   Assumes the directory names end in '_percent', for example,
 137   ash/resources/default_200_percent or
 138   chrome/app/theme/resources/touch_140_percent
 139
 140   Args:
 141     resources_path: The base path of interest.
 142
 143   Returns:
 144     A list of paths relative to the 'src' directory.
 145   """
 146   file_list = os.listdir(resources_path)
 147   scale_directories = []
 148   for file_entry in file_list:
 149     file_path = os.path.join(resources_path, file_entry)
 150     if os.path.isdir(file_path) and file_path.endswith('_percent'):
 151       scale_directories.append(file_path)
 152
 153   scale_directories.sort()
 154   return scale_directories
 155
 156
 157 def main():
 158   # The script requires exactly one parameter, the .grd file path.
 159   if len(sys.argv) != 2:
 160     print 'Usage: tools/resources/find_unused_resources.py <path/to/grd>'
 161     sys.exit(1)
 162   grd_filepath = sys.argv[1]
 163
 164   # Try to ensure we are in a source checkout.
 165   current_dir = os.getcwd()
 166   if os.path.basename(current_dir) != 'src':
 167     print 'Script must be run in your "src" directory.'
 168     sys.exit(1)
 169
 170   # We require a git checkout to use git grep.
 171   if not os.path.exists(current_dir + '/.git'):
 172     print 'You must use a git checkout for this script to run.'
 173     print current_dir + '/.git', 'not found.'
 174     sys.exit(1)
 175
 176   # Look up the scale-factor directories.
 177   resources_path = os.path.dirname(grd_filepath)
 178   scale_directories = GetScaleDirectories(resources_path)
 179   if not scale_directories:
 180     print 'No scale directories (like "default_100_percent") found.'
 181     sys.exit(1)
 182
 183   # |unused_resources| stores pairs of [resource_id, filepath] for resource ids
 184   # that are not referenced in the code.
 185   unused_resources = GetUnusedResources(grd_filepath)
 186   if not unused_resources:
 187     print 'All resources are used.'
 188     sys.exit(0)
 189
 190   # Dump our output for the user.
 191   print
 192   print 'Unused resource ids:'
 193   for resource_id, filepath in unused_resources:
 194     print resource_id
 195   # Print a list of 'git rm' command lines to remove unused assets.
 196   print
 197   print 'Unused files:'
 198   for resource_id, filepath in unused_resources:
 199     for directory in scale_directories:
 200       print 'git rm ' + os.path.join(directory, filepath)
 201
 202
 203 if __name__ == '__main__':
 204   main()