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