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('build','secondary'),
  39     os.path.join('third_party','bison'),
  40     os.path.join('third_party','blanketjs'),
  41     os.path.join('third_party','cygwin'),
  42     os.path.join('third_party','gnu_binutils'),
  43     os.path.join('third_party','gold'),
  44     os.path.join('third_party','gperf'),
  45     os.path.join('third_party','lighttpd'),
  46     os.path.join('third_party','llvm'),
  47     os.path.join('third_party','llvm-build'),
  48     os.path.join('third_party','mingw-w64'),
  49     os.path.join('third_party','nacl_sdk_binaries'),
  50     os.path.join('third_party','pefile'),
  51     os.path.join('third_party','perl'),
  52     os.path.join('third_party','psyco_win32'),
  53     os.path.join('third_party','pylib'),
  54     os.path.join('third_party','pywebsocket'),
  55     os.path.join('third_party','qunit'),
  56     os.path.join('third_party','sinonjs'),
  57     os.path.join('third_party','syzygy'),
  58     os.path.join('tools', 'profile_chrome', 'third_party'),
  59
  60     # Chromium code in third_party.
  61     os.path.join('third_party','fuzzymatch'),
  62     os.path.join('tools', 'swarming_client'),
  63
  64     # Stuff pulled in from chrome-internal for official builds/tools.
  65     os.path.join('third_party', 'clear_cache'),
  66     os.path.join('third_party', 'gnu'),
  67     os.path.join('third_party', 'googlemac'),
  68     os.path.join('third_party', 'pcre'),
  69     os.path.join('third_party', 'psutils'),
  70     os.path.join('third_party', 'sawbuck'),
  71
  72     # Redistribution does not require attribution in documentation.
  73     os.path.join('third_party','directxsdk'),
  74     os.path.join('third_party','platformsdk_win2008_6_1'),
  75     os.path.join('third_party','platformsdk_win7'),
  76 ])
  77
  78 # Directories we don't scan through.
  79 VCS_METADATA_DIRS = ('.svn', '.git')
  80 PRUNE_DIRS = (VCS_METADATA_DIRS +
  81               ('out', 'Debug', 'Release',  # build files
  82                'layout_tests'))            # lots of subdirs
  83
  84 ADDITIONAL_PATHS = (
  85     os.path.join('breakpad'),
  86     os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
  87     os.path.join('chrome', 'browser', 'resources', 'settings', 'routing',
  88                  'more_routing'),
  89     os.path.join('chrome', 'test', 'chromeos', 'autotest'),
  90     os.path.join('chrome', 'test', 'data'),
  91     os.path.join('native_client'),
  92     os.path.join('net', 'tools', 'spdyshark'),
  93     os.path.join('sdch', 'open-vcdiff'),
  94     os.path.join('testing', 'gmock'),
  95     os.path.join('testing', 'gtest'),
  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     os.path.join('v8', 'third_party', 'fdlibm'),
 104 )
 105
 106
 107 # Directories where we check out directly from upstream, and therefore
 108 # can't provide a README.chromium.  Please prefer a README.chromium
 109 # wherever possible.
 110 SPECIAL_CASES = {
 111     os.path.join('native_client'): {
 112         "Name": "native client",
 113         "URL": "http://code.google.com/p/nativeclient",
 114         "License": "BSD",
 115     },
 116     os.path.join('sdch', 'open-vcdiff'): {
 117         "Name": "open-vcdiff",
 118         "URL": "http://code.google.com/p/open-vcdiff",
 119         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 120         "License Android Compatible": "yes",
 121     },
 122     os.path.join('testing', 'gmock'): {
 123         "Name": "gmock",
 124         "URL": "http://code.google.com/p/googlemock",
 125         "License": "BSD",
 126         "License File": "NOT_SHIPPED",
 127     },
 128     os.path.join('testing', 'gtest'): {
 129         "Name": "gtest",
 130         "URL": "http://code.google.com/p/googletest",
 131         "License": "BSD",
 132         "License File": "NOT_SHIPPED",
 133     },
 134     os.path.join('third_party', 'angle'): {
 135         "Name": "Almost Native Graphics Layer Engine",
 136         "URL": "http://code.google.com/p/angleproject/",
 137         "License": "BSD",
 138     },
 139     os.path.join('third_party', 'cros_system_api'): {
 140         "Name": "Chromium OS system API",
 141         "URL": "http://www.chromium.org/chromium-os",
 142         "License": "BSD",
 143         # Absolute path here is resolved as relative to the source root.
 144         "License File": "/LICENSE.chromium_os",
 145     },
 146     os.path.join('third_party', 'lss'): {
 147         "Name": "linux-syscall-support",
 148         "URL": "http://code.google.com/p/linux-syscall-support/",
 149         "License": "BSD",
 150         "License File": "/LICENSE",
 151     },
 152     os.path.join('third_party', 'ots'): {
 153         "Name": "OTS (OpenType Sanitizer)",
 154         "URL": "http://code.google.com/p/ots/",
 155         "License": "BSD",
 156     },
 157     os.path.join('third_party', 'pdfium'): {
 158         "Name": "PDFium",
 159         "URL": "http://code.google.com/p/pdfium/",
 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     os.path.join('v8', 'third_party', 'fdlibm'): {
 227         "Name": "fdlibm",
 228         "URL": "http://www.netlib.org/fdlibm/",
 229         "License": "Freely Distributable",
 230         # Absolute path here is resolved as relative to the source root.
 231         "License File" : "/v8/third_party/fdlibm/LICENSE",
 232         "License Android Compatible" : "yes",
 233     },
 234     os.path.join('third_party', 'khronos_glcts'): {
 235         # These sources are not shipped, are not public, and it isn't
 236         # clear why they're tripping the license check.
 237         "Name": "khronos_glcts",
 238         "URL": "http://no-public-url",
 239         "License": "Khronos",
 240         "License File": "NOT_SHIPPED",
 241     },
 242     os.path.join('tools', 'telemetry', 'third_party', 'gsutil'): {
 243         "Name": "gsutil",
 244         "URL": "https://cloud.google.com/storage/docs/gsutil",
 245         "License": "Apache 2.0",
 246         "License File": "NOT_SHIPPED",
 247     },
 248 }
 249
 250 # Special value for 'License File' field used to indicate that the license file
 251 # should not be used in about:credits.
 252 NOT_SHIPPED = "NOT_SHIPPED"
 253
 254
 255 class LicenseError(Exception):
 256     """We raise this exception when a directory's licensing info isn't
 257     fully filled out."""
 258     pass
 259
 260 def AbsolutePath(path, filename, root):
 261     """Convert a path in README.chromium to be absolute based on the source
 262     root."""
 263     if filename.startswith('/'):
 264         # Absolute-looking paths are relative to the source root
 265         # (which is the directory we're run from).
 266         absolute_path = os.path.join(root, filename[1:])
 267     else:
 268         absolute_path = os.path.join(root, path, filename)
 269     if os.path.exists(absolute_path):
 270         return absolute_path
 271     return None
 272
 273 def ParseDir(path, root, require_license_file=True, optional_keys=None):
 274     """Examine a third_party/foo component and extract its metadata."""
 275
 276     # Parse metadata fields out of README.chromium.
 277     # We examine "LICENSE" for the license file by default.
 278     metadata = {
 279         "License File": "LICENSE",  # Relative path to license text.
 280         "Name": None,               # Short name (for header on about:credits).
 281         "URL": None,                # Project home page.
 282         "License": None,            # Software license.
 283         }
 284
 285     if optional_keys is None:
 286         optional_keys = []
 287
 288     if path in SPECIAL_CASES:
 289         metadata.update(SPECIAL_CASES[path])
 290     else:
 291         # Try to find README.chromium.
 292         readme_path = os.path.join(root, path, 'README.chromium')
 293         if not os.path.exists(readme_path):
 294             raise LicenseError("missing README.chromium or licenses.py "
 295                                "SPECIAL_CASES entry")
 296
 297         for line in open(readme_path):
 298             line = line.strip()
 299             if not line:
 300                 break
 301             for key in metadata.keys() + optional_keys:
 302                 field = key + ": "
 303                 if line.startswith(field):
 304                     metadata[key] = line[len(field):]
 305
 306     # Check that all expected metadata is present.
 307     for key, value in metadata.iteritems():
 308         if not value:
 309             raise LicenseError("couldn't find '" + key + "' line "
 310                                "in README.chromium or licences.py "
 311                                "SPECIAL_CASES")
 312
 313     # Special-case modules that aren't in the shipping product, so don't need
 314     # their license in about:credits.
 315     if metadata["License File"] != NOT_SHIPPED:
 316         # Check that the license file exists.
 317         for filename in (metadata["License File"], "COPYING"):
 318             license_path = AbsolutePath(path, filename, root)
 319             if license_path is not None:
 320                 break
 321
 322         if require_license_file and not license_path:
 323             raise LicenseError("License file not found. "
 324                                "Either add a file named LICENSE, "
 325                                "import upstream's COPYING if available, "
 326                                "or add a 'License File:' line to "
 327                                "README.chromium with the appropriate path.")
 328         metadata["License File"] = license_path
 329
 330     return metadata
 331
 332
 333 def ContainsFiles(path, root):
 334     """Determines whether any files exist in a directory or in any of its
 335     subdirectories."""
 336     for _, dirs, files in os.walk(os.path.join(root, path)):
 337         if files:
 338             return True
 339         for vcs_metadata in VCS_METADATA_DIRS:
 340             if vcs_metadata in dirs:
 341                 dirs.remove(vcs_metadata)
 342     return False
 343
 344
 345 def FilterDirsWithFiles(dirs_list, root):
 346     # If a directory contains no files, assume it's a DEPS directory for a
 347     # project not used by our current configuration and skip it.
 348     return [x for x in dirs_list if ContainsFiles(x, root)]
 349
 350
 351 def FindThirdPartyDirs(prune_paths, root):
 352     """Find all third_party directories underneath the source root."""
 353     third_party_dirs = set()
 354     for path, dirs, files in os.walk(root):
 355         path = path[len(root)+1:]  # Pretty up the path.
 356
 357         if path in prune_paths:
 358             dirs[:] = []
 359             continue
 360
 361         # Prune out directories we want to skip.
 362         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 363         # list that we're simultaneously mutating.)
 364         for skip in PRUNE_DIRS:
 365             if skip in dirs:
 366                 dirs.remove(skip)
 367
 368         if os.path.basename(path) == 'third_party':
 369             # Add all subdirectories that are not marked for skipping.
 370             for dir in dirs:
 371                 dirpath = os.path.join(path, dir)
 372                 if dirpath not in prune_paths:
 373                     third_party_dirs.add(dirpath)
 374
 375             # Don't recurse into any subdirs from here.
 376             dirs[:] = []
 377             continue
 378
 379         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 380         # third_party/foo paths.
 381         if path in ADDITIONAL_PATHS:
 382             dirs[:] = []
 383
 384     for dir in ADDITIONAL_PATHS:
 385         if dir not in prune_paths:
 386             third_party_dirs.add(dir)
 387
 388     return third_party_dirs
 389
 390
 391 def ScanThirdPartyDirs(root=None):
 392     """Scan a list of directories and report on any problems we find."""
 393     if root is None:
 394       root = os.getcwd()
 395     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 396     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 397
 398     errors = []
 399     for path in sorted(third_party_dirs):
 400         try:
 401             metadata = ParseDir(path, root)
 402         except LicenseError, e:
 403             errors.append((path, e.args[0]))
 404             continue
 405
 406     for path, error in sorted(errors):
 407         print path + ": " + error
 408
 409     return len(errors) == 0
 410
 411
 412 def GenerateCredits():
 413     """Generate about:credits."""
 414
 415     if len(sys.argv) not in (2, 3):
 416       print 'usage: licenses.py credits [output_file]'
 417       return False
 418
 419     def EvaluateTemplate(template, env, escape=True):
 420         """Expand a template with variables like {{foo}} using a
 421         dictionary of expansions."""
 422         for key, val in env.items():
 423             if escape:
 424                 val = cgi.escape(val)
 425             template = template.replace('{{%s}}' % key, val)
 426         return template
 427
 428     root = os.path.join(os.path.dirname(__file__), '..')
 429     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 430
 431     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 432                                        'about_credits_entry.tmpl'), 'rb').read()
 433     entries = []
 434     for path in sorted(third_party_dirs):
 435         try:
 436             metadata = ParseDir(path, root)
 437         except LicenseError:
 438             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 439             continue
 440         if metadata['License File'] == NOT_SHIPPED:
 441             continue
 442         env = {
 443             'name': metadata['Name'],
 444             'url': metadata['URL'],
 445             'license': open(metadata['License File'], 'rb').read(),
 446         }
 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())