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','pywebsocket'),
  53     os.path.join('third_party','syzygy'),
  54     os.path.join('tools','gn'),
  55
  56     # Chromium code in third_party.
  57     os.path.join('third_party','fuzzymatch'),
  58     os.path.join('tools', 'swarming_client'),
  59
  60     # Stuff pulled in from chrome-internal for official builds/tools.
  61     os.path.join('third_party', 'clear_cache'),
  62     os.path.join('third_party', 'gnu'),
  63     os.path.join('third_party', 'googlemac'),
  64     os.path.join('third_party', 'pcre'),
  65     os.path.join('third_party', 'psutils'),
  66     os.path.join('third_party', 'sawbuck'),
  67
  68     # Redistribution does not require attribution in documentation.
  69     os.path.join('third_party','directxsdk'),
  70     os.path.join('third_party','platformsdk_win2008_6_1'),
  71     os.path.join('third_party','platformsdk_win7'),
  72 ])
  73
  74 # Directories we don't scan through.
  75 VCS_METADATA_DIRS = ('.svn', '.git')
  76 PRUNE_DIRS = (VCS_METADATA_DIRS +
  77               ('out', 'Debug', 'Release',  # build files
  78                'layout_tests'))            # lots of subdirs
  79
  80 ADDITIONAL_PATHS = (
  81     os.path.join('breakpad'),
  82     os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
  83     os.path.join('chrome', 'test', 'chromeos', 'autotest'),
  84     os.path.join('chrome', 'test', 'data'),
  85     os.path.join('native_client'),
  86     os.path.join('net', 'tools', 'spdyshark'),
  87     os.path.join('sdch', 'open-vcdiff'),
  88     os.path.join('testing', 'gmock'),
  89     os.path.join('testing', 'gtest'),
  90     # The directory with the word list for Chinese and Japanese segmentation
  91     # with different license terms than ICU.
  92     os.path.join('third_party','icu','source','data','brkitr'),
  93     os.path.join('tools', 'grit'),
  94     os.path.join('tools', 'gyp'),
  95     os.path.join('tools', 'page_cycler', 'acid3'),
  96     os.path.join('url', 'third_party', 'mozilla'),
  97     os.path.join('v8'),
  98     # Fake directory so we can include the strongtalk license.
  99     os.path.join('v8', 'strongtalk'),
 100 )
 101
 102
 103 # Directories where we check out directly from upstream, and therefore
 104 # can't provide a README.chromium.  Please prefer a README.chromium
 105 # wherever possible.
 106 SPECIAL_CASES = {
 107     os.path.join('native_client'): {
 108         "Name": "native client",
 109         "URL": "http://code.google.com/p/nativeclient",
 110         "License": "BSD",
 111     },
 112     os.path.join('sdch', 'open-vcdiff'): {
 113         "Name": "open-vcdiff",
 114         "URL": "http://code.google.com/p/open-vcdiff",
 115         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 116         "License Android Compatible": "yes",
 117     },
 118     os.path.join('testing', 'gmock'): {
 119         "Name": "gmock",
 120         "URL": "http://code.google.com/p/googlemock",
 121         "License": "BSD",
 122         "License File": "NOT_SHIPPED",
 123     },
 124     os.path.join('testing', 'gtest'): {
 125         "Name": "gtest",
 126         "URL": "http://code.google.com/p/googletest",
 127         "License": "BSD",
 128         "License File": "NOT_SHIPPED",
 129     },
 130     os.path.join('third_party', 'angle'): {
 131         "Name": "Almost Native Graphics Layer Engine",
 132         "URL": "http://code.google.com/p/angleproject/",
 133         "License": "BSD",
 134     },
 135     os.path.join('third_party', 'cros_system_api'): {
 136         "Name": "Chromium OS system API",
 137         "URL": "http://www.chromium.org/chromium-os",
 138         "License": "BSD",
 139         # Absolute path here is resolved as relative to the source root.
 140         "License File": "/LICENSE.chromium_os",
 141     },
 142     os.path.join('third_party', 'lss'): {
 143         "Name": "linux-syscall-support",
 144         "URL": "http://code.google.com/p/linux-syscall-support/",
 145         "License": "BSD",
 146         "License File": "/LICENSE",
 147     },
 148     os.path.join('third_party', 'ots'): {
 149         "Name": "OTS (OpenType Sanitizer)",
 150         "URL": "http://code.google.com/p/ots/",
 151         "License": "BSD",
 152     },
 153     os.path.join('third_party', 'pdfium'): {
 154         "Name": "PDFium",
 155         "URL": "http://code.google.com/p/pdfium/",
 156         "License": "BSD",
 157     },
 158     os.path.join('third_party', 'pdfsqueeze'): {
 159         "Name": "pdfsqueeze",
 160         "URL": "http://code.google.com/p/pdfsqueeze/",
 161         "License": "Apache 2.0",
 162         "License File": "COPYING",
 163     },
 164     os.path.join('third_party', 'ppapi'): {
 165         "Name": "ppapi",
 166         "URL": "http://code.google.com/p/ppapi/",
 167     },
 168     os.path.join('third_party', 'scons-2.0.1'): {
 169         "Name": "scons-2.0.1",
 170         "URL": "http://www.scons.org",
 171         "License": "MIT",
 172         "License File": "NOT_SHIPPED",
 173     },
 174     os.path.join('third_party', 'trace-viewer'): {
 175         "Name": "trace-viewer",
 176         "URL": "http://code.google.com/p/trace-viewer",
 177         "License": "BSD",
 178         "License File": "NOT_SHIPPED",
 179     },
 180     os.path.join('third_party', 'v8-i18n'): {
 181         "Name": "Internationalization Library for v8",
 182         "URL": "http://code.google.com/p/v8-i18n/",
 183         "License": "Apache 2.0",
 184     },
 185     os.path.join('third_party', 'WebKit'): {
 186         "Name": "WebKit",
 187         "URL": "http://webkit.org/",
 188         "License": "BSD and GPL v2",
 189         # Absolute path here is resolved as relative to the source root.
 190         "License File": "/webkit/LICENSE",
 191     },
 192     os.path.join('third_party', 'webpagereplay'): {
 193         "Name": "webpagereplay",
 194         "URL": "http://code.google.com/p/web-page-replay",
 195         "License": "Apache 2.0",
 196         "License File": "NOT_SHIPPED",
 197     },
 198     os.path.join('tools', 'grit'): {
 199         "Name": "grit",
 200         "URL": "http://code.google.com/p/grit-i18n",
 201         "License": "BSD",
 202         "License File": "NOT_SHIPPED",
 203     },
 204     os.path.join('tools', 'gyp'): {
 205         "Name": "gyp",
 206         "URL": "http://code.google.com/p/gyp",
 207         "License": "BSD",
 208         "License File": "NOT_SHIPPED",
 209     },
 210     os.path.join('v8'): {
 211         "Name": "V8 JavaScript Engine",
 212         "URL": "http://code.google.com/p/v8",
 213         "License": "BSD",
 214     },
 215     os.path.join('v8', 'strongtalk'): {
 216         "Name": "Strongtalk",
 217         "URL": "http://www.strongtalk.org/",
 218         "License": "BSD",
 219         # Absolute path here is resolved as relative to the source root.
 220         "License File": "/v8/LICENSE.strongtalk",
 221     },
 222 }
 223
 224 # Special value for 'License File' field used to indicate that the license file
 225 # should not be used in about:credits.
 226 NOT_SHIPPED = "NOT_SHIPPED"
 227
 228
 229 class LicenseError(Exception):
 230     """We raise this exception when a directory's licensing info isn't
 231     fully filled out."""
 232     pass
 233
 234 def AbsolutePath(path, filename, root):
 235     """Convert a path in README.chromium to be absolute based on the source
 236     root."""
 237     if filename.startswith('/'):
 238         # Absolute-looking paths are relative to the source root
 239         # (which is the directory we're run from).
 240         absolute_path = os.path.join(root, filename[1:])
 241     else:
 242         absolute_path = os.path.join(root, path, filename)
 243     if os.path.exists(absolute_path):
 244         return absolute_path
 245     return None
 246
 247 def ParseDir(path, root, require_license_file=True):
 248     """Examine a third_party/foo component and extract its metadata."""
 249
 250     # Parse metadata fields out of README.chromium.
 251     # We examine "LICENSE" for the license file by default.
 252     metadata = {
 253         "License File": "LICENSE",  # Relative path to license text.
 254         "Name": None,               # Short name (for header on about:credits).
 255         "URL": None,                # Project home page.
 256         "License": None,            # Software license.
 257         }
 258
 259     # Relative path to a file containing some html we're required to place in
 260     # about:credits.
 261     optional_keys = ["Required Text", "License Android Compatible"]
 262
 263     if path in SPECIAL_CASES:
 264         metadata.update(SPECIAL_CASES[path])
 265     else:
 266         # Try to find README.chromium.
 267         readme_path = os.path.join(root, path, 'README.chromium')
 268         if not os.path.exists(readme_path):
 269             raise LicenseError("missing README.chromium or licenses.py "
 270                                "SPECIAL_CASES entry")
 271
 272         for line in open(readme_path):
 273             line = line.strip()
 274             if not line:
 275                 break
 276             for key in metadata.keys() + optional_keys:
 277                 field = key + ": "
 278                 if line.startswith(field):
 279                     metadata[key] = line[len(field):]
 280
 281     # Check that all expected metadata is present.
 282     for key, value in metadata.iteritems():
 283         if not value:
 284             raise LicenseError("couldn't find '" + key + "' line "
 285                                "in README.chromium or licences.py "
 286                                "SPECIAL_CASES")
 287
 288     # Special-case modules that aren't in the shipping product, so don't need
 289     # their license in about:credits.
 290     if metadata["License File"] != NOT_SHIPPED:
 291         # Check that the license file exists.
 292         for filename in (metadata["License File"], "COPYING"):
 293             license_path = AbsolutePath(path, filename, root)
 294             if license_path is not None:
 295                 break
 296
 297         if require_license_file and not license_path:
 298             raise LicenseError("License file not found. "
 299                                "Either add a file named LICENSE, "
 300                                "import upstream's COPYING if available, "
 301                                "or add a 'License File:' line to "
 302                                "README.chromium with the appropriate path.")
 303         metadata["License File"] = license_path
 304
 305     if "Required Text" in metadata:
 306         required_path = AbsolutePath(path, metadata["Required Text"], root)
 307         if required_path is not None:
 308             metadata["Required Text"] = required_path
 309         else:
 310             raise LicenseError("Required text file listed but not found.")
 311
 312     return metadata
 313
 314
 315 def ContainsFiles(path, root):
 316     """Determines whether any files exist in a directory or in any of its
 317     subdirectories."""
 318     for _, dirs, files in os.walk(os.path.join(root, path)):
 319         if files:
 320             return True
 321         for vcs_metadata in VCS_METADATA_DIRS:
 322             if vcs_metadata in dirs:
 323                 dirs.remove(vcs_metadata)
 324     return False
 325
 326
 327 def FilterDirsWithFiles(dirs_list, root):
 328     # If a directory contains no files, assume it's a DEPS directory for a
 329     # project not used by our current configuration and skip it.
 330     return [x for x in dirs_list if ContainsFiles(x, root)]
 331
 332
 333 def FindThirdPartyDirs(prune_paths, root):
 334     """Find all third_party directories underneath the source root."""
 335     third_party_dirs = set()
 336     for path, dirs, files in os.walk(root):
 337         path = path[len(root)+1:]  # Pretty up the path.
 338
 339         if path in prune_paths:
 340             dirs[:] = []
 341             continue
 342
 343         # Prune out directories we want to skip.
 344         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 345         # list that we're simultaneously mutating.)
 346         for skip in PRUNE_DIRS:
 347             if skip in dirs:
 348                 dirs.remove(skip)
 349
 350         if os.path.basename(path) == 'third_party':
 351             # Add all subdirectories that are not marked for skipping.
 352             for dir in dirs:
 353                 dirpath = os.path.join(path, dir)
 354                 if dirpath not in prune_paths:
 355                     third_party_dirs.add(dirpath)
 356
 357             # Don't recurse into any subdirs from here.
 358             dirs[:] = []
 359             continue
 360
 361         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 362         # third_party/foo paths.
 363         if path in ADDITIONAL_PATHS:
 364             dirs[:] = []
 365
 366     for dir in ADDITIONAL_PATHS:
 367         if dir not in prune_paths:
 368             third_party_dirs.add(dir)
 369
 370     return third_party_dirs
 371
 372
 373 def ScanThirdPartyDirs(root=None):
 374     """Scan a list of directories and report on any problems we find."""
 375     if root is None:
 376       root = os.getcwd()
 377     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 378     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 379
 380     errors = []
 381     for path in sorted(third_party_dirs):
 382         try:
 383             metadata = ParseDir(path, root)
 384         except LicenseError, e:
 385             errors.append((path, e.args[0]))
 386             continue
 387
 388     for path, error in sorted(errors):
 389         print path + ": " + error
 390
 391     return len(errors) == 0
 392
 393
 394 def GenerateCredits():
 395     """Generate about:credits."""
 396
 397     if len(sys.argv) not in (2, 3):
 398       print 'usage: licenses.py credits [output_file]'
 399       return False
 400
 401     def EvaluateTemplate(template, env, escape=True):
 402         """Expand a template with variables like {{foo}} using a
 403         dictionary of expansions."""
 404         for key, val in env.items():
 405             if escape and not key.endswith("_unescaped"):
 406                 val = cgi.escape(val)
 407             template = template.replace('{{%s}}' % key, val)
 408         return template
 409
 410     root = os.path.join(os.path.dirname(__file__), '..')
 411     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 412
 413     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 414                                        'about_credits_entry.tmpl'), 'rb').read()
 415     entries = []
 416     for path in sorted(third_party_dirs):
 417         try:
 418             metadata = ParseDir(path, root)
 419         except LicenseError:
 420             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 421             continue
 422         if metadata['License File'] == NOT_SHIPPED:
 423             continue
 424         env = {
 425             'name': metadata['Name'],
 426             'url': metadata['URL'],
 427             'license': open(metadata['License File'], 'rb').read(),
 428             'license_unescaped': '',
 429         }
 430         if 'Required Text' in metadata:
 431             required_text = open(metadata['Required Text'], 'rb').read()
 432             env["license_unescaped"] = required_text
 433         entries.append(EvaluateTemplate(entry_template, env))
 434
 435     file_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 436                                       'about_credits.tmpl'), 'rb').read()
 437     template_contents = "<!-- Generated by licenses.py; do not edit. -->"
 438     template_contents += EvaluateTemplate(file_template,
 439                                           {'entries': '\n'.join(entries)},
 440                                           escape=False)
 441
 442     if len(sys.argv) == 3:
 443       with open(sys.argv[2], 'w') as output_file:
 444         output_file.write(template_contents)
 445     elif len(sys.argv) == 2:
 446       print template_contents
 447
 448     return True
 449
 450
 451 def main():
 452     command = 'help'
 453     if len(sys.argv) > 1:
 454         command = sys.argv[1]
 455
 456     if command == 'scan':
 457         if not ScanThirdPartyDirs():
 458             return 1
 459     elif command == 'credits':
 460         if not GenerateCredits():
 461             return 1
 462     else:
 463         print __doc__
 464         return 1
 465
 466
 467 if __name__ == '__main__':
 468   sys.exit(main())