tools/licenses.py

   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 """Utility for checking and processing licensing information in third_party
   7 directories.
   8
   9 Usage: licenses.py <command>
  10
  11 Commands:
  12   scan     scan third_party directories, verifying that we have licensing info
  13   credits  generate about:credits on stdout
  14
  15 (You can also import this as a module.)
  16 """
  17
  18 import cgi
  19 import os
  20 import sys
  21
  22 # Paths from the root of the tree to directories to skip.
  23 PRUNE_PATHS = set([
  24     # Same module occurs in crypto/third_party/nss and net/third_party/nss, so
  25     # skip this one.
  26     os.path.join('third_party','nss'),
  27
  28     # Placeholder directory only, not third-party code.
  29     os.path.join('third_party','adobe'),
  30
  31     # Build files only, not third-party code.
  32     os.path.join('third_party','widevine'),
  33
  34     # Only binaries, used during development.
  35     os.path.join('third_party','valgrind'),
  36
  37     # Used for development and test, not in the shipping product.
  38     os.path.join('third_party','bison'),
  39     os.path.join('third_party','cygwin'),
  40     os.path.join('third_party','gnu_binutils'),
  41     os.path.join('third_party','gold'),
  42     os.path.join('third_party','gperf'),
  43     os.path.join('third_party','lighttpd'),
  44     os.path.join('third_party','llvm'),
  45     os.path.join('third_party','llvm-build'),
  46     os.path.join('third_party','mingw-w64'),
  47     os.path.join('third_party','nacl_sdk_binaries'),
  48     os.path.join('third_party','pefile'),
  49     os.path.join('third_party','perl'),
  50     os.path.join('third_party','psyco_win32'),
  51     os.path.join('third_party','pylib'),
  52     os.path.join('third_party','python_26'),
  53     os.path.join('third_party','pywebsocket'),
  54     os.path.join('third_party','syzygy'),
  55
  56     # Chromium code in third_party.
  57     os.path.join('third_party','fuzzymatch'),
  58
  59     # Stuff pulled in from chrome-internal for official builds/tools.
  60     os.path.join('third_party', 'clear_cache'),
  61     os.path.join('third_party', 'gnu'),
  62     os.path.join('third_party', 'googlemac'),
  63     os.path.join('third_party', 'pcre'),
  64     os.path.join('third_party', 'psutils'),
  65     os.path.join('third_party', 'sawbuck'),
  66
  67     # Redistribution does not require attribution in documentation.
  68     os.path.join('third_party','directxsdk'),
  69     os.path.join('third_party','platformsdk_win2008_6_1'),
  70     os.path.join('third_party','platformsdk_win7'),
  71 ])
  72
  73 # Directories we don't scan through.
  74 VCS_METADATA_DIRS = ('.svn', '.git')
  75 PRUNE_DIRS = (VCS_METADATA_DIRS +
  76               ('out', 'Debug', 'Release',  # build files
  77                'layout_tests'))            # lots of subdirs
  78
  79 ADDITIONAL_PATHS = (
  80     os.path.join('breakpad'),
  81     os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
  82     os.path.join('chrome', 'test', 'chromeos', 'autotest'),
  83     os.path.join('chrome', 'test', 'data'),
  84     os.path.join('googleurl'),
  85     os.path.join('native_client'),
  86     os.path.join('native_client_sdk'),
  87     os.path.join('net', 'tools', 'spdyshark'),
  88     os.path.join('ppapi'),
  89     os.path.join('sandbox', 'linux', 'seccomp-legacy'),
  90     os.path.join('sdch', 'open-vcdiff'),
  91     os.path.join('testing', 'gmock'),
  92     os.path.join('testing', 'gtest'),
  93     # The directory with the word list for Chinese and Japanese segmentation
  94     # with different license terms than ICU.
  95     os.path.join('third_party','icu','source','data','brkitr'),
  96     os.path.join('tools', 'grit'),
  97     os.path.join('tools', 'gyp'),
  98     os.path.join('tools', 'page_cycler', 'acid3'),
  99     os.path.join('v8'),
 100     # Fake directory so we can include the strongtalk license.
 101     os.path.join('v8', 'strongtalk'),
 102 )
 103
 104
 105 # Directories where we check out directly from upstream, and therefore
 106 # can't provide a README.chromium.  Please prefer a README.chromium
 107 # wherever possible.
 108 SPECIAL_CASES = {
 109     os.path.join('googleurl'): {
 110         "Name": "google-url",
 111         "URL": "http://code.google.com/p/google-url/",
 112         "License": "BSD and MPL 1.1/GPL 2.0/LGPL 2.1",
 113         "License File": "LICENSE.txt",
 114     },
 115     os.path.join('native_client'): {
 116         "Name": "native client",
 117         "URL": "http://code.google.com/p/nativeclient",
 118         "License": "BSD",
 119     },
 120     os.path.join('sandbox', 'linux', 'seccomp-legacy'): {
 121         "Name": "seccompsandbox",
 122         "URL": "http://code.google.com/p/seccompsandbox",
 123         "License": "BSD",
 124     },
 125     os.path.join('sdch', 'open-vcdiff'): {
 126         "Name": "open-vcdiff",
 127         "URL": "http://code.google.com/p/open-vcdiff",
 128         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 129         "License Android Compatible": "yes",
 130     },
 131     os.path.join('testing', 'gmock'): {
 132         "Name": "gmock",
 133         "URL": "http://code.google.com/p/googlemock",
 134         "License": "BSD",
 135         "License File": "NOT_SHIPPED",
 136     },
 137     os.path.join('testing', 'gtest'): {
 138         "Name": "gtest",
 139         "URL": "http://code.google.com/p/googletest",
 140         "License": "BSD",
 141         "License File": "NOT_SHIPPED",
 142     },
 143     os.path.join('third_party', 'angle'): {
 144         "Name": "Almost Native Graphics Layer Engine",
 145         "URL": "http://code.google.com/p/angleproject/",
 146         "License": "BSD",
 147     },
 148     os.path.join('third_party', 'cros_system_api'): {
 149         "Name": "Chromium OS system API",
 150         "URL": "http://www.chromium.org/chromium-os",
 151         "License": "BSD",
 152         # Absolute path here is resolved as relative to the source root.
 153         "License File": "/LICENSE.chromium_os",
 154     },
 155     os.path.join('third_party', 'GTM'): {
 156         "Name": "Google Toolbox for Mac",
 157         "URL": "http://code.google.com/p/google-toolbox-for-mac/",
 158         "License": "Apache 2.0",
 159         "License File": "COPYING",
 160     },
 161     os.path.join('third_party', 'lss'): {
 162         "Name": "linux-syscall-support",
 163         "URL": "http://code.google.com/p/linux-syscall-support/",
 164         "License": "BSD",
 165         "License File": "/LICENSE",
 166     },
 167     os.path.join('third_party', 'ots'): {
 168         "Name": "OTS (OpenType Sanitizer)",
 169         "URL": "http://code.google.com/p/ots/",
 170         "License": "BSD",
 171     },
 172     os.path.join('third_party', 'pdfsqueeze'): {
 173         "Name": "pdfsqueeze",
 174         "URL": "http://code.google.com/p/pdfsqueeze/",
 175         "License": "Apache 2.0",
 176         "License File": "COPYING",
 177     },
 178     os.path.join('third_party', 'ppapi'): {
 179         "Name": "ppapi",
 180         "URL": "http://code.google.com/p/ppapi/",
 181     },
 182     os.path.join('third_party', 'scons-2.0.1'): {
 183         "Name": "scons-2.0.1",
 184         "URL": "http://www.scons.org",
 185         "License": "MIT",
 186         "License File": "NOT_SHIPPED",
 187     },
 188     os.path.join('third_party', 'trace-viewer'): {
 189         "Name": "trace-viewer",
 190         "URL": "http://code.google.com/p/trace-viewer",
 191         "License": "BSD",
 192         "License File": "NOT_SHIPPED",
 193     },
 194     os.path.join('third_party', 'v8-i18n'): {
 195         "Name": "Internationalization Library for v8",
 196         "URL": "http://code.google.com/p/v8-i18n/",
 197         "License": "Apache 2.0",
 198     },
 199     os.path.join('third_party', 'WebKit'): {
 200         "Name": "WebKit",
 201         "URL": "http://webkit.org/",
 202         "License": "BSD and GPL v2",
 203         # Absolute path here is resolved as relative to the source root.
 204         "License File": "/webkit/LICENSE",
 205     },
 206     os.path.join('third_party', 'webpagereplay'): {
 207         "Name": "webpagereplay",
 208         "URL": "http://code.google.com/p/web-page-replay",
 209         "License": "Apache 2.0",
 210         "License File": "NOT_SHIPPED",
 211     },
 212     os.path.join('tools', 'grit'): {
 213         "Name": "grit",
 214         "URL": "http://code.google.com/p/grit-i18n",
 215         "License": "BSD",
 216         "License File": "NOT_SHIPPED",
 217     },
 218     os.path.join('tools', 'gyp'): {
 219         "Name": "gyp",
 220         "URL": "http://code.google.com/p/gyp",
 221         "License": "BSD",
 222         "License File": "NOT_SHIPPED",
 223     },
 224     os.path.join('v8'): {
 225         "Name": "V8 JavaScript Engine",
 226         "URL": "http://code.google.com/p/v8",
 227         "License": "BSD",
 228     },
 229     os.path.join('v8', 'strongtalk'): {
 230         "Name": "Strongtalk",
 231         "URL": "http://www.strongtalk.org/",
 232         "License": "BSD",
 233         # Absolute path here is resolved as relative to the source root.
 234         "License File": "/v8/LICENSE.strongtalk",
 235     },
 236 }
 237
 238 # Special value for 'License File' field used to indicate that the license file
 239 # should not be used in about:credits.
 240 NOT_SHIPPED = "NOT_SHIPPED"
 241
 242
 243 class LicenseError(Exception):
 244     """We raise this exception when a directory's licensing info isn't
 245     fully filled out."""
 246     pass
 247
 248 def AbsolutePath(path, filename, root):
 249     """Convert a path in README.chromium to be absolute based on the source
 250     root."""
 251     if filename.startswith('/'):
 252         # Absolute-looking paths are relative to the source root
 253         # (which is the directory we're run from).
 254         absolute_path = os.path.join(root, filename[1:])
 255     else:
 256         absolute_path = os.path.join(root, path, filename)
 257     if os.path.exists(absolute_path):
 258         return absolute_path
 259     return None
 260
 261 def ParseDir(path, root, require_license_file=True):
 262     """Examine a third_party/foo component and extract its metadata."""
 263
 264     # Parse metadata fields out of README.chromium.
 265     # We examine "LICENSE" for the license file by default.
 266     metadata = {
 267         "License File": "LICENSE",  # Relative path to license text.
 268         "Name": None,               # Short name (for header on about:credits).
 269         "URL": None,                # Project home page.
 270         "License": None,            # Software license.
 271         }
 272
 273     # Relative path to a file containing some html we're required to place in
 274     # about:credits.
 275     optional_keys = ["Required Text", "License Android Compatible"]
 276
 277     if path in SPECIAL_CASES:
 278         metadata.update(SPECIAL_CASES[path])
 279     else:
 280         # Try to find README.chromium.
 281         readme_path = os.path.join(root, path, 'README.chromium')
 282         if not os.path.exists(readme_path):
 283             raise LicenseError("missing README.chromium or licenses.py "
 284                                "SPECIAL_CASES entry")
 285
 286         for line in open(readme_path):
 287             line = line.strip()
 288             if not line:
 289                 break
 290             for key in metadata.keys() + optional_keys:
 291                 field = key + ": "
 292                 if line.startswith(field):
 293                     metadata[key] = line[len(field):]
 294
 295     # Check that all expected metadata is present.
 296     for key, value in metadata.iteritems():
 297         if not value:
 298             raise LicenseError("couldn't find '" + key + "' line "
 299                                "in README.chromium or licences.py "
 300                                "SPECIAL_CASES")
 301
 302     # Special-case modules that aren't in the shipping product, so don't need
 303     # their license in about:credits.
 304     if metadata["License File"] != NOT_SHIPPED:
 305         # Check that the license file exists.
 306         for filename in (metadata["License File"], "COPYING"):
 307             license_path = AbsolutePath(path, filename, root)
 308             if license_path is not None:
 309                 break
 310
 311         if require_license_file and not license_path:
 312             raise LicenseError("License file not found. "
 313                                "Either add a file named LICENSE, "
 314                                "import upstream's COPYING if available, "
 315                                "or add a 'License File:' line to "
 316                                "README.chromium with the appropriate path.")
 317         metadata["License File"] = license_path
 318
 319     if "Required Text" in metadata:
 320         required_path = AbsolutePath(path, metadata["Required Text"], root)
 321         if required_path is not None:
 322             metadata["Required Text"] = required_path
 323         else:
 324             raise LicenseError("Required text file listed but not found.")
 325
 326     return metadata
 327
 328
 329 def ContainsFiles(path, root):
 330     """Determines whether any files exist in a directory or in any of its
 331     subdirectories."""
 332     for _, dirs, files in os.walk(os.path.join(root, path)):
 333         if files:
 334             return True
 335         for vcs_metadata in VCS_METADATA_DIRS:
 336             if vcs_metadata in dirs:
 337                 dirs.remove(vcs_metadata)
 338     return False
 339
 340
 341 def FilterDirsWithFiles(dirs_list, root):
 342     # If a directory contains no files, assume it's a DEPS directory for a
 343     # project not used by our current configuration and skip it.
 344     return [x for x in dirs_list if ContainsFiles(x, root)]
 345
 346
 347 def FindThirdPartyDirs(prune_paths, root):
 348     """Find all third_party directories underneath the source root."""
 349     third_party_dirs = []
 350     for path, dirs, files in os.walk(root):
 351         path = path[len(root)+1:]  # Pretty up the path.
 352
 353         if path in prune_paths:
 354             dirs[:] = []
 355             continue
 356
 357         # Prune out directories we want to skip.
 358         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 359         # list that we're simultaneously mutating.)
 360         for skip in PRUNE_DIRS:
 361             if skip in dirs:
 362                 dirs.remove(skip)
 363
 364         if os.path.basename(path) == 'third_party':
 365             # Add all subdirectories that are not marked for skipping.
 366             for dir in dirs:
 367                 dirpath = os.path.join(path, dir)
 368                 if dirpath not in prune_paths:
 369                     third_party_dirs.append(dirpath)
 370
 371             # Don't recurse into any subdirs from here.
 372             dirs[:] = []
 373             continue
 374
 375         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 376         # third_party/foo paths.
 377         if path in ADDITIONAL_PATHS:
 378             dirs[:] = []
 379
 380     for dir in ADDITIONAL_PATHS:
 381         if dir not in prune_paths:
 382             third_party_dirs.append(dir)
 383
 384     return third_party_dirs
 385
 386
 387 def ScanThirdPartyDirs(root=None):
 388     """Scan a list of directories and report on any problems we find."""
 389     if root is None:
 390       root = os.getcwd()
 391     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 392     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 393
 394     errors = []
 395     for path in sorted(third_party_dirs):
 396         try:
 397             metadata = ParseDir(path, root)
 398         except LicenseError, e:
 399             errors.append((path, e.args[0]))
 400             continue
 401
 402     for path, error in sorted(errors):
 403         print path + ": " + error
 404
 405     return len(errors) == 0
 406
 407
 408 def GenerateCredits():
 409     """Generate about:credits."""
 410
 411     if len(sys.argv) not in (2, 3):
 412       print 'usage: licenses.py credits [output_file]'
 413       return False
 414
 415     def EvaluateTemplate(template, env, escape=True):
 416         """Expand a template with variables like {{foo}} using a
 417         dictionary of expansions."""
 418         for key, val in env.items():
 419             if escape and not key.endswith("_unescaped"):
 420                 val = cgi.escape(val)
 421             template = template.replace('{{%s}}' % key, val)
 422         return template
 423
 424     root = os.path.join(os.path.dirname(__file__), '..')
 425     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 426
 427     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 428                                        'about_credits_entry.tmpl'), 'rb').read()
 429     entries = []
 430     for path in sorted(third_party_dirs):
 431         try:
 432             metadata = ParseDir(path, root)
 433         except LicenseError:
 434             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 435             continue
 436         if metadata['License File'] == NOT_SHIPPED:
 437             continue
 438         env = {
 439             'name': metadata['Name'],
 440             'url': metadata['URL'],
 441             'license': open(metadata['License File'], 'rb').read(),
 442             'license_unescaped': '',
 443         }
 444         if 'Required Text' in metadata:
 445             required_text = open(metadata['Required Text'], 'rb').read()
 446             env["license_unescaped"] = required_text
 447         entries.append(EvaluateTemplate(entry_template, env))
 448
 449     file_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 450                                       'about_credits.tmpl'), 'rb').read()
 451     template_contents = "<!-- Generated by licenses.py; do not edit. -->"
 452     template_contents += EvaluateTemplate(file_template,
 453                                           {'entries': '\n'.join(entries)},
 454                                           escape=False)
 455
 456     if len(sys.argv) == 3:
 457       with open(sys.argv[2], 'w') as output_file:
 458         output_file.write(template_contents)
 459     elif len(sys.argv) == 2:
 460       print template_contents
 461
 462     return True
 463
 464
 465 def main():
 466     command = 'help'
 467     if len(sys.argv) > 1:
 468         command = sys.argv[1]
 469
 470     if command == 'scan':
 471         if not ScanThirdPartyDirs():
 472             return 1
 473     elif command == 'credits':
 474         if not GenerateCredits():
 475             return 1
 476     else:
 477         print __doc__
 478         return 1
 479
 480
 481 if __name__ == '__main__':
 482   sys.exit(main())