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', 'test', 'chromeos', 'autotest'),
  88     os.path.join('chrome', 'test', 'data'),
  89     os.path.join('native_client'),
  90     os.path.join('net', 'tools', 'spdyshark'),
  91     os.path.join('sdch', 'open-vcdiff'),
  92     os.path.join('testing', 'gmock'),
  93     os.path.join('testing', 'gtest'),
  94     os.path.join('tools', 'grit'),
  95     os.path.join('tools', 'gyp'),
  96     os.path.join('tools', 'page_cycler', 'acid3'),
  97     os.path.join('url', 'third_party', 'mozilla'),
  98     os.path.join('v8'),
  99     # Fake directories to include the strongtalk and fdlibm licenses.
 100     os.path.join('v8', 'strongtalk'),
 101     os.path.join('v8', 'fdlibm'),
 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('native_client'): {
 110         "Name": "native client",
 111         "URL": "http://code.google.com/p/nativeclient",
 112         "License": "BSD",
 113     },
 114     os.path.join('sdch', 'open-vcdiff'): {
 115         "Name": "open-vcdiff",
 116         "URL": "http://code.google.com/p/open-vcdiff",
 117         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 118         "License Android Compatible": "yes",
 119     },
 120     os.path.join('testing', 'gmock'): {
 121         "Name": "gmock",
 122         "URL": "http://code.google.com/p/googlemock",
 123         "License": "BSD",
 124         "License File": "NOT_SHIPPED",
 125     },
 126     os.path.join('testing', 'gtest'): {
 127         "Name": "gtest",
 128         "URL": "http://code.google.com/p/googletest",
 129         "License": "BSD",
 130         "License File": "NOT_SHIPPED",
 131     },
 132     os.path.join('third_party', 'angle'): {
 133         "Name": "Almost Native Graphics Layer Engine",
 134         "URL": "http://code.google.com/p/angleproject/",
 135         "License": "BSD",
 136     },
 137     os.path.join('third_party', 'cros_system_api'): {
 138         "Name": "Chromium OS system API",
 139         "URL": "http://www.chromium.org/chromium-os",
 140         "License": "BSD",
 141         # Absolute path here is resolved as relative to the source root.
 142         "License File": "/LICENSE.chromium_os",
 143     },
 144     os.path.join('third_party', 'lss'): {
 145         "Name": "linux-syscall-support",
 146         "URL": "http://code.google.com/p/linux-syscall-support/",
 147         "License": "BSD",
 148         "License File": "/LICENSE",
 149     },
 150     os.path.join('third_party', 'ots'): {
 151         "Name": "OTS (OpenType Sanitizer)",
 152         "URL": "http://code.google.com/p/ots/",
 153         "License": "BSD",
 154     },
 155     os.path.join('third_party', 'pdfium'): {
 156         "Name": "PDFium",
 157         "URL": "http://code.google.com/p/pdfium/",
 158         "License": "BSD",
 159     },
 160     os.path.join('third_party', 'pdfsqueeze'): {
 161         "Name": "pdfsqueeze",
 162         "URL": "http://code.google.com/p/pdfsqueeze/",
 163         "License": "Apache 2.0",
 164         "License File": "COPYING",
 165     },
 166     os.path.join('third_party', 'ppapi'): {
 167         "Name": "ppapi",
 168         "URL": "http://code.google.com/p/ppapi/",
 169     },
 170     os.path.join('third_party', 'scons-2.0.1'): {
 171         "Name": "scons-2.0.1",
 172         "URL": "http://www.scons.org",
 173         "License": "MIT",
 174         "License File": "NOT_SHIPPED",
 175     },
 176     os.path.join('third_party', 'trace-viewer'): {
 177         "Name": "trace-viewer",
 178         "URL": "http://code.google.com/p/trace-viewer",
 179         "License": "BSD",
 180         "License File": "NOT_SHIPPED",
 181     },
 182     os.path.join('third_party', 'v8-i18n'): {
 183         "Name": "Internationalization Library for v8",
 184         "URL": "http://code.google.com/p/v8-i18n/",
 185         "License": "Apache 2.0",
 186     },
 187     os.path.join('third_party', 'WebKit'): {
 188         "Name": "WebKit",
 189         "URL": "http://webkit.org/",
 190         "License": "BSD and GPL v2",
 191         # Absolute path here is resolved as relative to the source root.
 192         "License File": "/third_party/WebKit/LICENSE_FOR_ABOUT_CREDITS",
 193     },
 194     os.path.join('third_party', 'webpagereplay'): {
 195         "Name": "webpagereplay",
 196         "URL": "http://code.google.com/p/web-page-replay",
 197         "License": "Apache 2.0",
 198         "License File": "NOT_SHIPPED",
 199     },
 200     os.path.join('tools', 'grit'): {
 201         "Name": "grit",
 202         "URL": "http://code.google.com/p/grit-i18n",
 203         "License": "BSD",
 204         "License File": "NOT_SHIPPED",
 205     },
 206     os.path.join('tools', 'gyp'): {
 207         "Name": "gyp",
 208         "URL": "http://code.google.com/p/gyp",
 209         "License": "BSD",
 210         "License File": "NOT_SHIPPED",
 211     },
 212     os.path.join('v8'): {
 213         "Name": "V8 JavaScript Engine",
 214         "URL": "http://code.google.com/p/v8",
 215         "License": "BSD",
 216     },
 217     os.path.join('v8', 'strongtalk'): {
 218         "Name": "Strongtalk",
 219         "URL": "http://www.strongtalk.org/",
 220         "License": "BSD",
 221         # Absolute path here is resolved as relative to the source root.
 222         "License File": "/v8/LICENSE.strongtalk",
 223     },
 224     os.path.join('v8', 'fdlibm'): {
 225         "Name": "fdlibm",
 226         "URL": "http://www.netlib.org/fdlibm/",
 227         "License": "Freely Distributable",
 228         # Absolute path here is resolved as relative to the source root.
 229         "License File" : "/v8/src/third_party/fdlibm/LICENSE",
 230         "License Android Compatible" : "yes",
 231     },
 232     os.path.join('third_party', 'khronos_glcts'): {
 233         # These sources are not shipped, are not public, and it isn't
 234         # clear why they're tripping the license check.
 235         "Name": "khronos_glcts",
 236         "URL": "http://no-public-url",
 237         "License": "Khronos",
 238         "License File": "NOT_SHIPPED",
 239     },
 240     os.path.join('tools', 'telemetry', 'third_party', 'gsutil'): {
 241         "Name": "gsutil",
 242         "URL": "https://cloud.google.com/storage/docs/gsutil",
 243         "License": "Apache 2.0",
 244         "License File": "NOT_SHIPPED",
 245     },
 246 }
 247
 248 # Special value for 'License File' field used to indicate that the license file
 249 # should not be used in about:credits.
 250 NOT_SHIPPED = "NOT_SHIPPED"
 251
 252
 253 class LicenseError(Exception):
 254     """We raise this exception when a directory's licensing info isn't
 255     fully filled out."""
 256     pass
 257
 258 def AbsolutePath(path, filename, root):
 259     """Convert a path in README.chromium to be absolute based on the source
 260     root."""
 261     if filename.startswith('/'):
 262         # Absolute-looking paths are relative to the source root
 263         # (which is the directory we're run from).
 264         absolute_path = os.path.join(root, filename[1:])
 265     else:
 266         absolute_path = os.path.join(root, path, filename)
 267     if os.path.exists(absolute_path):
 268         return absolute_path
 269     return None
 270
 271 def ParseDir(path, root, require_license_file=True, optional_keys=None):
 272     """Examine a third_party/foo component and extract its metadata."""
 273
 274     # Parse metadata fields out of README.chromium.
 275     # We examine "LICENSE" for the license file by default.
 276     metadata = {
 277         "License File": "LICENSE",  # Relative path to license text.
 278         "Name": None,               # Short name (for header on about:credits).
 279         "URL": None,                # Project home page.
 280         "License": None,            # Software license.
 281         }
 282
 283     if optional_keys is None:
 284         optional_keys = []
 285
 286     if path in SPECIAL_CASES:
 287         metadata.update(SPECIAL_CASES[path])
 288     else:
 289         # Try to find README.chromium.
 290         readme_path = os.path.join(root, path, 'README.chromium')
 291         if not os.path.exists(readme_path):
 292             raise LicenseError("missing README.chromium or licenses.py "
 293                                "SPECIAL_CASES entry")
 294
 295         for line in open(readme_path):
 296             line = line.strip()
 297             if not line:
 298                 break
 299             for key in metadata.keys() + optional_keys:
 300                 field = key + ": "
 301                 if line.startswith(field):
 302                     metadata[key] = line[len(field):]
 303
 304     # Check that all expected metadata is present.
 305     for key, value in metadata.iteritems():
 306         if not value:
 307             raise LicenseError("couldn't find '" + key + "' line "
 308                                "in README.chromium or licences.py "
 309                                "SPECIAL_CASES")
 310
 311     # Special-case modules that aren't in the shipping product, so don't need
 312     # their license in about:credits.
 313     if metadata["License File"] != NOT_SHIPPED:
 314         # Check that the license file exists.
 315         for filename in (metadata["License File"], "COPYING"):
 316             license_path = AbsolutePath(path, filename, root)
 317             if license_path is not None:
 318                 break
 319
 320         if require_license_file and not license_path:
 321             raise LicenseError("License file not found. "
 322                                "Either add a file named LICENSE, "
 323                                "import upstream's COPYING if available, "
 324                                "or add a 'License File:' line to "
 325                                "README.chromium with the appropriate path.")
 326         metadata["License File"] = license_path
 327
 328     return metadata
 329
 330
 331 def ContainsFiles(path, root):
 332     """Determines whether any files exist in a directory or in any of its
 333     subdirectories."""
 334     for _, dirs, files in os.walk(os.path.join(root, path)):
 335         if files:
 336             return True
 337         for vcs_metadata in VCS_METADATA_DIRS:
 338             if vcs_metadata in dirs:
 339                 dirs.remove(vcs_metadata)
 340     return False
 341
 342
 343 def FilterDirsWithFiles(dirs_list, root):
 344     # If a directory contains no files, assume it's a DEPS directory for a
 345     # project not used by our current configuration and skip it.
 346     return [x for x in dirs_list if ContainsFiles(x, root)]
 347
 348
 349 def FindThirdPartyDirs(prune_paths, root):
 350     """Find all third_party directories underneath the source root."""
 351     third_party_dirs = set()
 352     for path, dirs, files in os.walk(root):
 353         path = path[len(root)+1:]  # Pretty up the path.
 354
 355         if path in prune_paths:
 356             dirs[:] = []
 357             continue
 358
 359         # Prune out directories we want to skip.
 360         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 361         # list that we're simultaneously mutating.)
 362         for skip in PRUNE_DIRS:
 363             if skip in dirs:
 364                 dirs.remove(skip)
 365
 366         if os.path.basename(path) == 'third_party':
 367             # Add all subdirectories that are not marked for skipping.
 368             for dir in dirs:
 369                 dirpath = os.path.join(path, dir)
 370                 if dirpath not in prune_paths:
 371                     third_party_dirs.add(dirpath)
 372
 373             # Don't recurse into any subdirs from here.
 374             dirs[:] = []
 375             continue
 376
 377         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 378         # third_party/foo paths.
 379         if path in ADDITIONAL_PATHS:
 380             dirs[:] = []
 381
 382     for dir in ADDITIONAL_PATHS:
 383         if dir not in prune_paths:
 384             third_party_dirs.add(dir)
 385
 386     return third_party_dirs
 387
 388
 389 def ScanThirdPartyDirs(root=None):
 390     """Scan a list of directories and report on any problems we find."""
 391     if root is None:
 392       root = os.getcwd()
 393     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 394     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 395
 396     errors = []
 397     for path in sorted(third_party_dirs):
 398         try:
 399             metadata = ParseDir(path, root)
 400         except LicenseError, e:
 401             errors.append((path, e.args[0]))
 402             continue
 403
 404     for path, error in sorted(errors):
 405         print path + ": " + error
 406
 407     return len(errors) == 0
 408
 409
 410 def GenerateCredits():
 411     """Generate about:credits."""
 412
 413     if len(sys.argv) not in (2, 3):
 414       print 'usage: licenses.py credits [output_file]'
 415       return False
 416
 417     def EvaluateTemplate(template, env, escape=True):
 418         """Expand a template with variables like {{foo}} using a
 419         dictionary of expansions."""
 420         for key, val in env.items():
 421             if escape:
 422                 val = cgi.escape(val)
 423             template = template.replace('{{%s}}' % key, val)
 424         return template
 425
 426     root = os.path.join(os.path.dirname(__file__), '..')
 427     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 428
 429     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 430                                        'about_credits_entry.tmpl'), 'rb').read()
 431     entries = []
 432     for path in sorted(third_party_dirs):
 433         try:
 434             metadata = ParseDir(path, root)
 435         except LicenseError:
 436             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 437             continue
 438         if metadata['License File'] == NOT_SHIPPED:
 439             continue
 440         env = {
 441             'name': metadata['Name'],
 442             'url': metadata['URL'],
 443             'license': open(metadata['License File'], 'rb').read(),
 444         }
 445         entries.append(EvaluateTemplate(entry_template, env))
 446
 447     file_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 448                                       'about_credits.tmpl'), 'rb').read()
 449     template_contents = "<!-- Generated by licenses.py; do not edit. -->"
 450     template_contents += EvaluateTemplate(file_template,
 451                                           {'entries': '\n'.join(entries)},
 452                                           escape=False)
 453
 454     if len(sys.argv) == 3:
 455       with open(sys.argv[2], 'w') as output_file:
 456         output_file.write(template_contents)
 457     elif len(sys.argv) == 2:
 458       print template_contents
 459
 460     return True
 461
 462
 463 def main():
 464     command = 'help'
 465     if len(sys.argv) > 1:
 466         command = sys.argv[1]
 467
 468     if command == 'scan':
 469         if not ScanThirdPartyDirs():
 470             return 1
 471     elif command == 'credits':
 472         if not GenerateCredits():
 473             return 1
 474     else:
 475         print __doc__
 476         return 1
 477
 478
 479 if __name__ == '__main__':
 480   sys.exit(main())