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
  56     # Chromium code in third_party.
  57     os.path.join('third_party','fuzzymatch'),
  58
  59     # Stuff pulled in from chrome-internal for official builds/tools.
  60     os.path.join('third_party', 'clear_cache'),
  61     os.path.join('third_party', 'gnu'),
  62     os.path.join('third_party', 'googlemac'),
  63     os.path.join('third_party', 'pcre'),
  64     os.path.join('third_party', 'psutils'),
  65     os.path.join('third_party', 'sawbuck'),
  66
  67     # Redistribution does not require attribution in documentation.
  68     os.path.join('third_party','directxsdk'),
  69     os.path.join('third_party','platformsdk_win2008_6_1'),
  70     os.path.join('third_party','platformsdk_win7'),
  71 ])
  72
  73 # Directories we don't scan through.
  74 PRUNE_DIRS = ('.svn', '.git',             # VCS metadata
  75               'out', 'Debug', 'Release',  # build files
  76               'layout_tests')             # lots of subdirs
  77
  78 ADDITIONAL_PATHS = (
  79     os.path.join('breakpad'),
  80     os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
  81     os.path.join('chrome', 'test', 'chromeos', 'autotest'),
  82     os.path.join('chrome', 'test', 'data'),
  83     os.path.join('googleurl'),
  84     os.path.join('native_client'),
  85     os.path.join('native_client_sdk'),
  86     os.path.join('net', 'tools', 'spdyshark'),
  87     os.path.join('ppapi'),
  88     os.path.join('sandbox', 'linux', 'seccomp-legacy'),
  89     os.path.join('sdch', 'open-vcdiff'),
  90     os.path.join('testing', 'gmock'),
  91     os.path.join('testing', 'gtest'),
  92     # The directory with the word list for Chinese and Japanese segmentation
  93     # with different license terms than ICU.
  94     os.path.join('third_party','icu','source','data','brkitr'),
  95     os.path.join('tools', 'grit'),
  96     os.path.join('tools', 'gyp'),
  97     os.path.join('tools', 'page_cycler', 'acid3'),
  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('googleurl'): {
 109         "Name": "google-url",
 110         "URL": "http://code.google.com/p/google-url/",
 111         "License": "BSD and MPL 1.1/GPL 2.0/LGPL 2.1",
 112         "License File": "LICENSE.txt",
 113     },
 114     os.path.join('native_client'): {
 115         "Name": "native client",
 116         "URL": "http://code.google.com/p/nativeclient",
 117         "License": "BSD",
 118     },
 119     os.path.join('sandbox', 'linux', 'seccomp-legacy'): {
 120         "Name": "seccompsandbox",
 121         "URL": "http://code.google.com/p/seccompsandbox",
 122         "License": "BSD",
 123     },
 124     os.path.join('sdch', 'open-vcdiff'): {
 125         "Name": "open-vcdiff",
 126         "URL": "http://code.google.com/p/open-vcdiff",
 127         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 128         "License Android Compatible": "yes",
 129     },
 130     os.path.join('testing', 'gmock'): {
 131         "Name": "gmock",
 132         "URL": "http://code.google.com/p/googlemock",
 133         "License": "BSD",
 134         "License File": "NOT_SHIPPED",
 135     },
 136     os.path.join('testing', 'gtest'): {
 137         "Name": "gtest",
 138         "URL": "http://code.google.com/p/googletest",
 139         "License": "BSD",
 140         "License File": "NOT_SHIPPED",
 141     },
 142     os.path.join('third_party', 'angle'): {
 143         "Name": "Almost Native Graphics Layer Engine",
 144         "URL": "http://code.google.com/p/angleproject/",
 145         "License": "BSD",
 146     },
 147     os.path.join('third_party', 'cros_system_api'): {
 148         "Name": "Chromium OS system API",
 149         "URL": "http://www.chromium.org/chromium-os",
 150         "License": "BSD",
 151         # Absolute path here is resolved as relative to the source root.
 152         "License File": "/LICENSE.chromium_os",
 153     },
 154     os.path.join('third_party', 'GTM'): {
 155         "Name": "Google Toolbox for Mac",
 156         "URL": "http://code.google.com/p/google-toolbox-for-mac/",
 157         "License": "Apache 2.0",
 158         "License File": "COPYING",
 159     },
 160     os.path.join('third_party', 'lss'): {
 161         "Name": "linux-syscall-support",
 162         "URL": "http://code.google.com/p/linux-syscall-support/",
 163         "License": "BSD",
 164         "License File": "/LICENSE",
 165     },
 166     os.path.join('third_party', 'ots'): {
 167         "Name": "OTS (OpenType Sanitizer)",
 168         "URL": "http://code.google.com/p/ots/",
 169         "License": "BSD",
 170     },
 171     os.path.join('third_party', 'pdfsqueeze'): {
 172         "Name": "pdfsqueeze",
 173         "URL": "http://code.google.com/p/pdfsqueeze/",
 174         "License": "Apache 2.0",
 175         "License File": "COPYING",
 176     },
 177     os.path.join('third_party', 'ppapi'): {
 178         "Name": "ppapi",
 179         "URL": "http://code.google.com/p/ppapi/",
 180     },
 181     os.path.join('third_party', 'scons-2.0.1'): {
 182         "Name": "scons-2.0.1",
 183         "URL": "http://www.scons.org",
 184         "License": "MIT",
 185         "License File": "NOT_SHIPPED",
 186     },
 187     os.path.join('third_party', 'trace-viewer'): {
 188         "Name": "trace-viewer",
 189         "URL": "http://code.google.com/p/trace-viewer",
 190         "License": "BSD",
 191         "License File": "NOT_SHIPPED",
 192     },
 193     os.path.join('third_party', 'v8-i18n'): {
 194         "Name": "Internationalization Library for v8",
 195         "URL": "http://code.google.com/p/v8-i18n/",
 196         "License": "Apache 2.0",
 197     },
 198     os.path.join('third_party', 'WebKit'): {
 199         "Name": "WebKit",
 200         "URL": "http://webkit.org/",
 201         "License": "BSD and GPL v2",
 202         # Absolute path here is resolved as relative to the source root.
 203         "License File": "/webkit/LICENSE",
 204     },
 205     os.path.join('third_party', 'webpagereplay'): {
 206         "Name": "webpagereplay",
 207         "URL": "http://code.google.com/p/web-page-replay",
 208         "License": "Apache 2.0",
 209         "License File": "NOT_SHIPPED",
 210     },
 211     os.path.join('tools', 'grit'): {
 212         "Name": "grit",
 213         "URL": "http://code.google.com/p/grit-i18n",
 214         "License": "BSD",
 215         "License File": "NOT_SHIPPED",
 216     },
 217     os.path.join('tools', 'gyp'): {
 218         "Name": "gyp",
 219         "URL": "http://code.google.com/p/gyp",
 220         "License": "BSD",
 221         "License File": "NOT_SHIPPED",
 222     },
 223     os.path.join('v8'): {
 224         "Name": "V8 JavaScript Engine",
 225         "URL": "http://code.google.com/p/v8",
 226         "License": "BSD",
 227     },
 228     os.path.join('v8', 'strongtalk'): {
 229         "Name": "Strongtalk",
 230         "URL": "http://www.strongtalk.org/",
 231         "License": "BSD",
 232         # Absolute path here is resolved as relative to the source root.
 233         "License File": "/v8/LICENSE.strongtalk",
 234     },
 235 }
 236
 237 # Special value for 'License File' field used to indicate that the license file
 238 # should not be used in about:credits.
 239 NOT_SHIPPED = "NOT_SHIPPED"
 240
 241
 242 class LicenseError(Exception):
 243     """We raise this exception when a directory's licensing info isn't
 244     fully filled out."""
 245     pass
 246
 247 def AbsolutePath(path, filename, root):
 248     """Convert a path in README.chromium to be absolute based on the source
 249     root."""
 250     if filename.startswith('/'):
 251         # Absolute-looking paths are relative to the source root
 252         # (which is the directory we're run from).
 253         absolute_path = os.path.join(root, filename[1:])
 254     else:
 255         absolute_path = os.path.join(root, path, filename)
 256     if os.path.exists(absolute_path):
 257         return absolute_path
 258     return None
 259
 260 def ParseDir(path, root, require_license_file=True):
 261     """Examine a third_party/foo component and extract its metadata."""
 262
 263     # Parse metadata fields out of README.chromium.
 264     # We examine "LICENSE" for the license file by default.
 265     metadata = {
 266         "License File": "LICENSE",  # Relative path to license text.
 267         "Name": None,               # Short name (for header on about:credits).
 268         "URL": None,                # Project home page.
 269         "License": None,            # Software license.
 270         }
 271
 272     # Relative path to a file containing some html we're required to place in
 273     # about:credits.
 274     optional_keys = ["Required Text", "License Android Compatible"]
 275
 276     if path in SPECIAL_CASES:
 277         metadata.update(SPECIAL_CASES[path])
 278     else:
 279         # Try to find README.chromium.
 280         readme_path = os.path.join(root, path, 'README.chromium')
 281         if not os.path.exists(readme_path):
 282             raise LicenseError("missing README.chromium or licenses.py "
 283                                "SPECIAL_CASES entry")
 284
 285         for line in open(readme_path):
 286             line = line.strip()
 287             if not line:
 288                 break
 289             for key in metadata.keys() + optional_keys:
 290                 field = key + ": "
 291                 if line.startswith(field):
 292                     metadata[key] = line[len(field):]
 293
 294     # Check that all expected metadata is present.
 295     for key, value in metadata.iteritems():
 296         if not value:
 297             raise LicenseError("couldn't find '" + key + "' line "
 298                                "in README.chromium or licences.py "
 299                                "SPECIAL_CASES")
 300
 301     # Special-case modules that aren't in the shipping product, so don't need
 302     # their license in about:credits.
 303     if metadata["License File"] != NOT_SHIPPED:
 304         # Check that the license file exists.
 305         for filename in (metadata["License File"], "COPYING"):
 306             license_path = AbsolutePath(path, filename, root)
 307             if license_path is not None:
 308                 break
 309
 310         if require_license_file and not license_path:
 311             raise LicenseError("License file not found. "
 312                                "Either add a file named LICENSE, "
 313                                "import upstream's COPYING if available, "
 314                                "or add a 'License File:' line to "
 315                                "README.chromium with the appropriate path.")
 316         metadata["License File"] = license_path
 317
 318     if "Required Text" in metadata:
 319         required_path = AbsolutePath(path, metadata["Required Text"], root)
 320         if required_path is not None:
 321             metadata["Required Text"] = required_path
 322         else:
 323             raise LicenseError("Required text file listed but not found.")
 324
 325     return metadata
 326
 327
 328 def ContainsFiles(path, root):
 329     """Determines whether any files exist in a directory or in any of its
 330     subdirectories."""
 331     for _, _, files in os.walk(os.path.join(root, path)):
 332         if files:
 333             return True
 334     return False
 335
 336
 337 def FilterDirsWithFiles(dirs_list, root):
 338     # If a directory contains no files, assume it's a DEPS directory for a
 339     # project not used by our current configuration and skip it.
 340     return [x for x in dirs_list if ContainsFiles(x, root)]
 341
 342
 343 def FindThirdPartyDirs(prune_paths, root):
 344     """Find all third_party directories underneath the source root."""
 345     third_party_dirs = []
 346     for path, dirs, files in os.walk(root):
 347         path = path[len(root)+1:]  # Pretty up the path.
 348
 349         if path in prune_paths:
 350             dirs[:] = []
 351             continue
 352
 353         # Prune out directories we want to skip.
 354         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 355         # list that we're simultaneously mutating.)
 356         for skip in PRUNE_DIRS:
 357             if skip in dirs:
 358                 dirs.remove(skip)
 359
 360         if os.path.basename(path) == 'third_party':
 361             # Add all subdirectories that are not marked for skipping.
 362             for dir in dirs:
 363                 dirpath = os.path.join(path, dir)
 364                 if dirpath not in prune_paths:
 365                     third_party_dirs.append(dirpath)
 366
 367             # Don't recurse into any subdirs from here.
 368             dirs[:] = []
 369             continue
 370
 371         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 372         # third_party/foo paths.
 373         if path in ADDITIONAL_PATHS:
 374             dirs[:] = []
 375
 376     for dir in ADDITIONAL_PATHS:
 377         if dir not in prune_paths:
 378             third_party_dirs.append(dir)
 379
 380     return third_party_dirs
 381
 382
 383 def ScanThirdPartyDirs(root=None):
 384     """Scan a list of directories and report on any problems we find."""
 385     if root is None:
 386       root = os.getcwd()
 387     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 388     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 389
 390     errors = []
 391     for path in sorted(third_party_dirs):
 392         try:
 393             metadata = ParseDir(path, root)
 394         except LicenseError, e:
 395             errors.append((path, e.args[0]))
 396             continue
 397
 398     for path, error in sorted(errors):
 399         print path + ": " + error
 400
 401     return len(errors) == 0
 402
 403
 404 def GenerateCredits():
 405     """Generate about:credits."""
 406
 407     if len(sys.argv) not in (2, 3):
 408       print 'usage: licenses.py credits [output_file]'
 409       return False
 410
 411     def EvaluateTemplate(template, env, escape=True):
 412         """Expand a template with variables like {{foo}} using a
 413         dictionary of expansions."""
 414         for key, val in env.items():
 415             if escape and not key.endswith("_unescaped"):
 416                 val = cgi.escape(val)
 417             template = template.replace('{{%s}}' % key, val)
 418         return template
 419
 420     root = os.path.join(os.path.dirname(__file__), '..')
 421     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 422
 423     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 424                                        'about_credits_entry.tmpl'), 'rb').read()
 425     entries = []
 426     for path in sorted(third_party_dirs):
 427         try:
 428             metadata = ParseDir(path, root)
 429         except LicenseError:
 430             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 431             continue
 432         if metadata['License File'] == NOT_SHIPPED:
 433             continue
 434         env = {
 435             'name': metadata['Name'],
 436             'url': metadata['URL'],
 437             'license': open(metadata['License File'], 'rb').read(),
 438             'license_unescaped': '',
 439         }
 440         if 'Required Text' in metadata:
 441             required_text = open(metadata['Required Text'], 'rb').read()
 442             env["license_unescaped"] = required_text
 443         entries.append(EvaluateTemplate(entry_template, env))
 444
 445     file_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 446                                       'about_credits.tmpl'), 'rb').read()
 447     template_contents = "<!-- Generated by licenses.py; do not edit. -->"
 448     template_contents += EvaluateTemplate(file_template,
 449                                           {'entries': '\n'.join(entries)},
 450                                           escape=False)
 451
 452     if len(sys.argv) == 3:
 453       with open(sys.argv[2], 'w') as output_file:
 454         output_file.write(template_contents)
 455     elif len(sys.argv) == 2:
 456       print template_contents
 457
 458     return True
 459
 460
 461 def main():
 462     command = 'help'
 463     if len(sys.argv) > 1:
 464         command = sys.argv[1]
 465
 466     if command == 'scan':
 467         if not ScanThirdPartyDirs():
 468             return 1
 469     elif command == 'credits':
 470         if not GenerateCredits():
 471             return 1
 472     else:
 473         print __doc__
 474         return 1
 475
 476
 477 if __name__ == '__main__':
 478   sys.exit(main())