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
  59     # Chromium code in third_party.
  60     os.path.join('third_party','fuzzymatch'),
  61     os.path.join('tools', 'swarming_client'),
  62
  63     # Stuff pulled in from chrome-internal for official builds/tools.
  64     os.path.join('third_party', 'clear_cache'),
  65     os.path.join('third_party', 'gnu'),
  66     os.path.join('third_party', 'googlemac'),
  67     os.path.join('third_party', 'pcre'),
  68     os.path.join('third_party', 'psutils'),
  69     os.path.join('third_party', 'sawbuck'),
  70
  71     # Redistribution does not require attribution in documentation.
  72     os.path.join('third_party','directxsdk'),
  73     os.path.join('third_party','platformsdk_win2008_6_1'),
  74     os.path.join('third_party','platformsdk_win7'),
  75 ])
  76
  77 # Directories we don't scan through.
  78 VCS_METADATA_DIRS = ('.svn', '.git')
  79 PRUNE_DIRS = (VCS_METADATA_DIRS +
  80               ('out', 'Debug', 'Release',  # build files
  81                'layout_tests'))            # lots of subdirs
  82
  83 ADDITIONAL_PATHS = (
  84     os.path.join('breakpad'),
  85     os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
  86     os.path.join('chrome', 'test', 'chromeos', 'autotest'),
  87     os.path.join('chrome', 'test', 'data'),
  88     os.path.join('native_client'),
  89     os.path.join('net', 'tools', 'spdyshark'),
  90     os.path.join('sdch', 'open-vcdiff'),
  91     os.path.join('testing', 'gmock'),
  92     os.path.join('testing', 'gtest'),
  93     # The directory with the word list for Chinese and Japanese segmentation
  94     # with different license terms than ICU.
  95     os.path.join('third_party','icu','source','data','brkitr'),
  96     os.path.join('tools', 'grit'),
  97     os.path.join('tools', 'gyp'),
  98     os.path.join('tools', 'page_cycler', 'acid3'),
  99     os.path.join('url', 'third_party', 'mozilla'),
 100     os.path.join('v8'),
 101     # Fake directory so we can include the strongtalk license.
 102     os.path.join('v8', 'strongtalk'),
 103 )
 104
 105
 106 # Directories where we check out directly from upstream, and therefore
 107 # can't provide a README.chromium.  Please prefer a README.chromium
 108 # wherever possible.
 109 SPECIAL_CASES = {
 110     os.path.join('native_client'): {
 111         "Name": "native client",
 112         "URL": "http://code.google.com/p/nativeclient",
 113         "License": "BSD",
 114     },
 115     os.path.join('sdch', 'open-vcdiff'): {
 116         "Name": "open-vcdiff",
 117         "URL": "http://code.google.com/p/open-vcdiff",
 118         "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
 119         "License Android Compatible": "yes",
 120     },
 121     os.path.join('testing', 'gmock'): {
 122         "Name": "gmock",
 123         "URL": "http://code.google.com/p/googlemock",
 124         "License": "BSD",
 125         "License File": "NOT_SHIPPED",
 126     },
 127     os.path.join('testing', 'gtest'): {
 128         "Name": "gtest",
 129         "URL": "http://code.google.com/p/googletest",
 130         "License": "BSD",
 131         "License File": "NOT_SHIPPED",
 132     },
 133     os.path.join('third_party', 'angle'): {
 134         "Name": "Almost Native Graphics Layer Engine",
 135         "URL": "http://code.google.com/p/angleproject/",
 136         "License": "BSD",
 137     },
 138     os.path.join('third_party', 'cros_system_api'): {
 139         "Name": "Chromium OS system API",
 140         "URL": "http://www.chromium.org/chromium-os",
 141         "License": "BSD",
 142         # Absolute path here is resolved as relative to the source root.
 143         "License File": "/LICENSE.chromium_os",
 144     },
 145     os.path.join('third_party', 'lss'): {
 146         "Name": "linux-syscall-support",
 147         "URL": "http://code.google.com/p/linux-syscall-support/",
 148         "License": "BSD",
 149         "License File": "/LICENSE",
 150     },
 151     os.path.join('third_party', 'ots'): {
 152         "Name": "OTS (OpenType Sanitizer)",
 153         "URL": "http://code.google.com/p/ots/",
 154         "License": "BSD",
 155     },
 156     os.path.join('third_party', 'pdfium'): {
 157         "Name": "PDFium",
 158         "URL": "http://code.google.com/p/pdfium/",
 159         "License": "BSD",
 160     },
 161     os.path.join('third_party', 'pdfsqueeze'): {
 162         "Name": "pdfsqueeze",
 163         "URL": "http://code.google.com/p/pdfsqueeze/",
 164         "License": "Apache 2.0",
 165         "License File": "COPYING",
 166     },
 167     os.path.join('third_party', 'ppapi'): {
 168         "Name": "ppapi",
 169         "URL": "http://code.google.com/p/ppapi/",
 170     },
 171     os.path.join('third_party', 'scons-2.0.1'): {
 172         "Name": "scons-2.0.1",
 173         "URL": "http://www.scons.org",
 174         "License": "MIT",
 175         "License File": "NOT_SHIPPED",
 176     },
 177     os.path.join('third_party', 'trace-viewer'): {
 178         "Name": "trace-viewer",
 179         "URL": "http://code.google.com/p/trace-viewer",
 180         "License": "BSD",
 181         "License File": "NOT_SHIPPED",
 182     },
 183     os.path.join('third_party', 'v8-i18n'): {
 184         "Name": "Internationalization Library for v8",
 185         "URL": "http://code.google.com/p/v8-i18n/",
 186         "License": "Apache 2.0",
 187     },
 188     os.path.join('third_party', 'WebKit'): {
 189         "Name": "WebKit",
 190         "URL": "http://webkit.org/",
 191         "License": "BSD and GPL v2",
 192         # Absolute path here is resolved as relative to the source root.
 193         "License File": "/webkit/LICENSE",
 194     },
 195     os.path.join('third_party', 'webpagereplay'): {
 196         "Name": "webpagereplay",
 197         "URL": "http://code.google.com/p/web-page-replay",
 198         "License": "Apache 2.0",
 199         "License File": "NOT_SHIPPED",
 200     },
 201     os.path.join('tools', 'grit'): {
 202         "Name": "grit",
 203         "URL": "http://code.google.com/p/grit-i18n",
 204         "License": "BSD",
 205         "License File": "NOT_SHIPPED",
 206     },
 207     os.path.join('tools', 'gyp'): {
 208         "Name": "gyp",
 209         "URL": "http://code.google.com/p/gyp",
 210         "License": "BSD",
 211         "License File": "NOT_SHIPPED",
 212     },
 213     os.path.join('v8'): {
 214         "Name": "V8 JavaScript Engine",
 215         "URL": "http://code.google.com/p/v8",
 216         "License": "BSD",
 217     },
 218     os.path.join('v8', 'strongtalk'): {
 219         "Name": "Strongtalk",
 220         "URL": "http://www.strongtalk.org/",
 221         "License": "BSD",
 222         # Absolute path here is resolved as relative to the source root.
 223         "License File": "/v8/LICENSE.strongtalk",
 224     },
 225 }
 226
 227 # Special value for 'License File' field used to indicate that the license file
 228 # should not be used in about:credits.
 229 NOT_SHIPPED = "NOT_SHIPPED"
 230
 231
 232 class LicenseError(Exception):
 233     """We raise this exception when a directory's licensing info isn't
 234     fully filled out."""
 235     pass
 236
 237 def AbsolutePath(path, filename, root):
 238     """Convert a path in README.chromium to be absolute based on the source
 239     root."""
 240     if filename.startswith('/'):
 241         # Absolute-looking paths are relative to the source root
 242         # (which is the directory we're run from).
 243         absolute_path = os.path.join(root, filename[1:])
 244     else:
 245         absolute_path = os.path.join(root, path, filename)
 246     if os.path.exists(absolute_path):
 247         return absolute_path
 248     return None
 249
 250 def ParseDir(path, root, require_license_file=True):
 251     """Examine a third_party/foo component and extract its metadata."""
 252
 253     # Parse metadata fields out of README.chromium.
 254     # We examine "LICENSE" for the license file by default.
 255     metadata = {
 256         "License File": "LICENSE",  # Relative path to license text.
 257         "Name": None,               # Short name (for header on about:credits).
 258         "URL": None,                # Project home page.
 259         "License": None,            # Software license.
 260         }
 261
 262     # Relative path to a file containing some html we're required to place in
 263     # about:credits.
 264     optional_keys = ["Required Text", "License Android Compatible"]
 265
 266     if path in SPECIAL_CASES:
 267         metadata.update(SPECIAL_CASES[path])
 268     else:
 269         # Try to find README.chromium.
 270         readme_path = os.path.join(root, path, 'README.chromium')
 271         if not os.path.exists(readme_path):
 272             raise LicenseError("missing README.chromium or licenses.py "
 273                                "SPECIAL_CASES entry")
 274
 275         for line in open(readme_path):
 276             line = line.strip()
 277             if not line:
 278                 break
 279             for key in metadata.keys() + optional_keys:
 280                 field = key + ": "
 281                 if line.startswith(field):
 282                     metadata[key] = line[len(field):]
 283
 284     # Check that all expected metadata is present.
 285     for key, value in metadata.iteritems():
 286         if not value:
 287             raise LicenseError("couldn't find '" + key + "' line "
 288                                "in README.chromium or licences.py "
 289                                "SPECIAL_CASES")
 290
 291     # Special-case modules that aren't in the shipping product, so don't need
 292     # their license in about:credits.
 293     if metadata["License File"] != NOT_SHIPPED:
 294         # Check that the license file exists.
 295         for filename in (metadata["License File"], "COPYING"):
 296             license_path = AbsolutePath(path, filename, root)
 297             if license_path is not None:
 298                 break
 299
 300         if require_license_file and not license_path:
 301             raise LicenseError("License file not found. "
 302                                "Either add a file named LICENSE, "
 303                                "import upstream's COPYING if available, "
 304                                "or add a 'License File:' line to "
 305                                "README.chromium with the appropriate path.")
 306         metadata["License File"] = license_path
 307
 308     if "Required Text" in metadata:
 309         required_path = AbsolutePath(path, metadata["Required Text"], root)
 310         if required_path is not None:
 311             metadata["Required Text"] = required_path
 312         else:
 313             raise LicenseError("Required text file listed but not found.")
 314
 315     return metadata
 316
 317
 318 def ContainsFiles(path, root):
 319     """Determines whether any files exist in a directory or in any of its
 320     subdirectories."""
 321     for _, dirs, files in os.walk(os.path.join(root, path)):
 322         if files:
 323             return True
 324         for vcs_metadata in VCS_METADATA_DIRS:
 325             if vcs_metadata in dirs:
 326                 dirs.remove(vcs_metadata)
 327     return False
 328
 329
 330 def FilterDirsWithFiles(dirs_list, root):
 331     # If a directory contains no files, assume it's a DEPS directory for a
 332     # project not used by our current configuration and skip it.
 333     return [x for x in dirs_list if ContainsFiles(x, root)]
 334
 335
 336 def FindThirdPartyDirs(prune_paths, root):
 337     """Find all third_party directories underneath the source root."""
 338     third_party_dirs = set()
 339     for path, dirs, files in os.walk(root):
 340         path = path[len(root)+1:]  # Pretty up the path.
 341
 342         if path in prune_paths:
 343             dirs[:] = []
 344             continue
 345
 346         # Prune out directories we want to skip.
 347         # (Note that we loop over PRUNE_DIRS so we're not iterating over a
 348         # list that we're simultaneously mutating.)
 349         for skip in PRUNE_DIRS:
 350             if skip in dirs:
 351                 dirs.remove(skip)
 352
 353         if os.path.basename(path) == 'third_party':
 354             # Add all subdirectories that are not marked for skipping.
 355             for dir in dirs:
 356                 dirpath = os.path.join(path, dir)
 357                 if dirpath not in prune_paths:
 358                     third_party_dirs.add(dirpath)
 359
 360             # Don't recurse into any subdirs from here.
 361             dirs[:] = []
 362             continue
 363
 364         # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
 365         # third_party/foo paths.
 366         if path in ADDITIONAL_PATHS:
 367             dirs[:] = []
 368
 369     for dir in ADDITIONAL_PATHS:
 370         if dir not in prune_paths:
 371             third_party_dirs.add(dir)
 372
 373     return third_party_dirs
 374
 375
 376 def ScanThirdPartyDirs(root=None):
 377     """Scan a list of directories and report on any problems we find."""
 378     if root is None:
 379       root = os.getcwd()
 380     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 381     third_party_dirs = FilterDirsWithFiles(third_party_dirs, root)
 382
 383     errors = []
 384     for path in sorted(third_party_dirs):
 385         try:
 386             metadata = ParseDir(path, root)
 387         except LicenseError, e:
 388             errors.append((path, e.args[0]))
 389             continue
 390
 391     for path, error in sorted(errors):
 392         print path + ": " + error
 393
 394     return len(errors) == 0
 395
 396
 397 def GenerateCredits():
 398     """Generate about:credits."""
 399
 400     if len(sys.argv) not in (2, 3):
 401       print 'usage: licenses.py credits [output_file]'
 402       return False
 403
 404     def EvaluateTemplate(template, env, escape=True):
 405         """Expand a template with variables like {{foo}} using a
 406         dictionary of expansions."""
 407         for key, val in env.items():
 408             if escape and not key.endswith("_unescaped"):
 409                 val = cgi.escape(val)
 410             template = template.replace('{{%s}}' % key, val)
 411         return template
 412
 413     root = os.path.join(os.path.dirname(__file__), '..')
 414     third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
 415
 416     entry_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 417                                        'about_credits_entry.tmpl'), 'rb').read()
 418     entries = []
 419     for path in sorted(third_party_dirs):
 420         try:
 421             metadata = ParseDir(path, root)
 422         except LicenseError:
 423             # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
 424             continue
 425         if metadata['License File'] == NOT_SHIPPED:
 426             continue
 427         env = {
 428             'name': metadata['Name'],
 429             'url': metadata['URL'],
 430             'license': open(metadata['License File'], 'rb').read(),
 431             'license_unescaped': '',
 432         }
 433         if 'Required Text' in metadata:
 434             required_text = open(metadata['Required Text'], 'rb').read()
 435             env["license_unescaped"] = required_text
 436         entries.append(EvaluateTemplate(entry_template, env))
 437
 438     file_template = open(os.path.join(root, 'chrome', 'browser', 'resources',
 439                                       'about_credits.tmpl'), 'rb').read()
 440     template_contents = "<!-- Generated by licenses.py; do not edit. -->"
 441     template_contents += EvaluateTemplate(file_template,
 442                                           {'entries': '\n'.join(entries)},
 443                                           escape=False)
 444
 445     if len(sys.argv) == 3:
 446       with open(sys.argv[2], 'w') as output_file:
 447         output_file.write(template_contents)
 448     elif len(sys.argv) == 2:
 449       print template_contents
 450
 451     return True
 452
 453
 454 def main():
 455     command = 'help'
 456     if len(sys.argv) > 1:
 457         command = sys.argv[1]
 458
 459     if command == 'scan':
 460         if not ScanThirdPartyDirs():
 461             return 1
 462     elif command == 'credits':
 463         if not GenerateCredits():
 464             return 1
 465     else:
 466         print __doc__
 467         return 1
 468
 469
 470 if __name__ == '__main__':
 471   sys.exit(main())