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