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