1#!/usr/bin/env python
2# Copyright (c) 2012 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"""Verifies that GRD resource files define all the strings used by a given
7set of source files. For file formats where it is not possible to infer which
8strings represent message identifiers, localized strings should be explicitly
9annotated with the string "i18n-content", for example:
10
11  LocalizeString(/*i18n-content*/"PRODUCT_NAME");
12
13This script also recognises localized strings in HTML and manifest.json files:
14
15  HTML:          i18n-content="PRODUCT_NAME"
16              or i18n-value-name-1="BUTTON_NAME"
17              or i18n-title="TOOLTIP_NAME"
18  manifest.json: __MSG_PRODUCT_NAME__
19
20Note that these forms must be exact; extra spaces are not permitted, though
21either single or double quotes are recognized.
22
23In addition, the script checks that all the messages are still in use; if
24this is not the case then a warning is issued, but the script still succeeds.
25"""
26
27import json
28import os
29import optparse
30import re
31import sys
32import xml.dom.minidom as minidom
33
34WARNING_MESSAGE = """
35To remove this warning, either remove the unused tags from
36resource files, add the files that use the tags listed above to
37remoting.gyp, or annotate existing uses of those tags with the
38prefix /*i18n-content*/
39"""
40
41def LoadTagsFromGrd(filename):
42  xml = minidom.parse(filename)
43  android_tags = []
44  other_tags = []
45  msgs_and_structs = xml.getElementsByTagName("message")
46  msgs_and_structs.extend(xml.getElementsByTagName("structure"))
47  for res in msgs_and_structs:
48    name = res.getAttribute("name")
49    if not name or not name.startswith("IDS_"):
50      raise Exception("Tag name doesn't start with IDS_: %s" % name)
51    name = name[4:]
52    if 'android_java' in res.getAttribute('formatter_data'):
53      android_tags.append(name)
54    else:
55      other_tags.append(name)
56  return android_tags, other_tags
57
58
59def ExtractTagFromLine(file_type, line):
60  """Extract a tag from a line of HTML, C++, JS or JSON."""
61  if file_type == "html":
62    # HTML-style (tags)
63    m = re.search('i18n-content=[\'"]([^\'"]*)[\'"]', line)
64    if m: return m.group(1)
65    # HTML-style (titles)
66    m = re.search('i18n-title=[\'"]([^\'"]*)[\'"]', line)
67    if m: return m.group(1)
68    # HTML-style (substitutions)
69    m = re.search('i18n-value-name-[1-9]=[\'"]([^\'"]*)[\'"]', line)
70    if m: return m.group(1)
71  elif file_type == 'js':
72    # Javascript style
73    m = re.search('/\*i18n-content\*/[\'"]([^\`"]*)[\'"]', line)
74    if m: return m.group(1)
75  elif file_type == 'cc' or file_type == 'mm':
76    # C++ style
77    m = re.search('IDS_([A-Z0-9_]*)', line)
78    if m: return m.group(1)
79    m = re.search('/\*i18n-content\*/["]([^\`"]*)["]', line)
80    if m: return m.group(1)
81  elif file_type == 'json.jinja2':
82    # Manifest style
83    m = re.search('__MSG_(.*)__', line)
84    if m: return m.group(1)
85  elif file_type == 'jinja2':
86    # Jinja2 template file
87    m = re.search('\{\%\s+trans\s+\%\}([A-Z0-9_]+)\{\%\s+endtrans\s+\%\}', line)
88    if m: return m.group(1)
89  return None
90
91
92def VerifyFile(filename, messages, used_tags):
93  """
94  Parse |filename|, looking for tags and report any that are not included in
95  |messages|. Return True if all tags are present and correct, or False if
96  any are missing.
97  """
98
99  base_name, file_type = os.path.splitext(filename)
100  file_type = file_type[1:]
101  if file_type == 'jinja2' and base_name.endswith('.json'):
102    file_type = 'json.jinja2'
103  if file_type not in ['js', 'cc', 'html', 'json.jinja2', 'jinja2', 'mm']:
104    raise Exception("Unknown file type: %s" % file_type)
105
106  result = True
107  matches = False
108  f = open(filename, 'r')
109  lines = f.readlines()
110  for i in xrange(0, len(lines)):
111    tag = ExtractTagFromLine(file_type, lines[i])
112    if tag:
113      tag = tag.upper()
114      used_tags.add(tag)
115      matches = True
116      if not tag in messages:
117        result = False
118        print '%s/%s:%d: error: Undefined tag: %s' % \
119            (os.getcwd(), filename, i + 1, tag)
120  f.close()
121  return result
122
123
124def main():
125  parser = optparse.OptionParser(
126      usage='Usage: %prog [options...] [source_file...]')
127  parser.add_option('-t', '--touch', dest='touch',
128                    help='File to touch when finished.')
129  parser.add_option('-r', '--grd', dest='grd', action='append',
130                    help='grd file')
131  parser.add_option('--strict', dest='strict', action='store_true',
132                    help='Use strict verification checks.')
133
134  options, args = parser.parse_args()
135  if not options.touch:
136    print '-t is not specified.'
137    return 1
138  if len(options.grd) == 0 or len(args) == 0:
139    print 'At least one GRD file needs to be specified.'
140    return 1
141
142  all_resources = []
143  non_android_resources = []
144  for f in options.grd:
145    android_tags, other_tags = LoadTagsFromGrd(f)
146    all_resources.extend(android_tags + other_tags)
147    non_android_resources.extend(other_tags)
148
149  used_tags = set([])
150  exit_code = 0
151  for f in args:
152    if not VerifyFile(f, all_resources, used_tags):
153      exit_code = 1
154
155  if options.strict:
156    warnings = False
157    # Determining if a resource is being used in the Android app is tricky
158    # because it requires annotating and parsing Android XML layout files.
159    # For now, exclude Android strings from this check.
160    for tag in non_android_resources:
161      if tag not in used_tags:
162        print ('%s/%s:0: warning: %s is defined but not used') % \
163            (os.getcwd(), sys.argv[2], tag)
164        warnings = True
165    if warnings:
166      print WARNING_MESSAGE
167
168  if exit_code == 0:
169    f = open(options.touch, 'a')
170    f.close()
171    os.utime(options.touch, None)
172
173  return exit_code
174
175
176if __name__ == '__main__':
177  sys.exit(main())
178