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