Update license tool to check several additional directories

tools/licenses.py

 #!/usr/bin/env python
 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """Utility for checking and processing licensing information in third_party
 directories.
 
 Usage: licenses.py <command>
 
 Commands:
   scan     scan third_party directories, verifying that we have licensing info
   credits  generate about:credits on stdout
 
 (You can also import this as a module.)
 """
 
 import cgi
 import os
 import sys
 
 # Paths from the root of the tree to directories to skip.
 PRUNE_PATHS = set([
-    # Same module occurs in both the top-level third_party and others.
-    os.path.join('base','third_party','icu'),
-
-    # Assume for now that breakpad has their licensing in order.
-    os.path.join('breakpad'),
-
-    # Assume for now that native client has their licensing in order.
-    os.path.join('native_client'),
-
-    # Same module occurs in chrome/ and in net/, so skip one of them.
-    os.path.join('net','third_party','mozilla_security_manager'),
-
-    # Same module occurs in base/, net/, and src/ so skip all but one of them.
+    # Same module occurs in crypto/third_party/nss and net/third_party/nss, so
+    # skip this one.
     os.path.join('third_party','nss'),
-    os.path.join('net','third_party','nss'),
 
     # Placeholder directory only, not third-party code.
     os.path.join('third_party','adobe'),
 
-    # Same license as Chromium.
-    os.path.join('third_party','lss'),
-
     # Only binaries, used during development.
     os.path.join('third_party','valgrind'),
 
-    # Directories that are the same as those in base/third_party.
-    os.path.join('v8','src','third_party','valgrind'),
-
     # Used for development and test, not in the shipping product.
     os.path.join('third_party','bidichecker'),
     os.path.join('third_party','cygwin'),
     os.path.join('third_party','gold'),
     os.path.join('third_party','lighttpd'),
     os.path.join('third_party','mingw-w64'),
     os.path.join('third_party','pefile'),
     os.path.join('third_party','python_26'),
 
     # Stuff pulled in from chrome-internal for official builds/tools.
     os.path.join('third_party', 'clear_cache'),
     os.path.join('third_party', 'gnu'),
     os.path.join('third_party', 'googlemac'),
     os.path.join('third_party', 'pcre'),
     os.path.join('third_party', 'psutils'),
     os.path.join('third_party', 'sawbuck'),
 
     # Redistribution does not require attribution in documentation.
     os.path.join('third_party','directxsdk'),
     os.path.join('third_party','platformsdk_win2008_6_1'),
     os.path.join('third_party','platformsdk_win7'),
-
-    # Harfbuzz-ng is not currently shipping in any product:
-    os.path.join('third_party','harfbuzz-ng'),
 ])
 
 # Directories we don't scan through.
 PRUNE_DIRS = ('.svn', '.git',             # VCS metadata
               'out', 'Debug', 'Release',  # build files
               'layout_tests')             # lots of subdirs
 
 ADDITIONAL_PATHS = (
+    os.path.join('breakpad'),
+    os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
+    os.path.join('chrome', 'test', 'chromeos', 'autotest'),
+    os.path.join('chrome', 'test', 'data'),
     os.path.join('googleurl'),
+    os.path.join('native_client'),
     os.path.join('native_client_sdk'),
+    os.path.join('net', 'tools', 'spdyshark'),
     os.path.join('ppapi'),
+    os.path.join('sandbox', 'linux', 'seccomp-legacy'),
+    os.path.join('sdch', 'open-vcdiff'),
+    os.path.join('testing', 'gmock'),
+    os.path.join('testing', 'gtest'),
     # The directory with the word list for Chinese and Japanese segmentation
     # with different license terms than ICU.
     os.path.join('third_party','icu','source','data','brkitr'),
+    os.path.join('tools', 'grit'),
+    os.path.join('tools', 'gyp'),
+    os.path.join('tools', 'page_cycler', 'acid3'),
+    os.path.join('v8'),
     # Fake directory so we can include the strongtalk license.
     os.path.join('v8', 'strongtalk'),
 )
 
 
 # Directories where we check out directly from upstream, and therefore
 # can't provide a README.chromium.  Please prefer a README.chromium
 # wherever possible.
 SPECIAL_CASES = {
-    'googleurl': {
+    os.path.join('googleurl'): {
         "Name": "google-url",
         "URL": "http://code.google.com/p/google-url/",
         "License": "BSD and MPL 1.1/GPL 2.0/LGPL 2.1",
         "License File": "LICENSE.txt",
     },
+    os.path.join('native_client'): {
+        "Name": "native client",
+        "URL": "http://code.google.com/p/nativeclient",
+        "License": "BSD",
+    },
+    os.path.join('sandbox', 'linux', 'seccomp-legacy'): {
+        "Name": "seccompsandbox",
+        "URL": "http://code.google.com/p/seccompsandbox",
+        "License": "BSD",
+    },
+    os.path.join('sdch', 'open-vcdiff'): {
+        "Name": "open-vcdiff",
+        "URL": "http://code.google.com/p/open-vcdiff",
+        "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
+    },
+    os.path.join('testing', 'gmock'): {
+        "Name": "gmock",
+        "URL": "http://code.google.com/p/googlemock",
+        "License": "BSD",
+    },
+    os.path.join('testing', 'gtest'): {
+        "Name": "gtest",
+        "URL": "http://code.google.com/p/googletest",
+        "License": "BSD",
+    },
     os.path.join('third_party', 'angle'): {
         "Name": "Almost Native Graphics Layer Engine",
         "URL": "http://code.google.com/p/angleproject/",
         "License": "BSD",
     },
     os.path.join('third_party', 'cros_system_api'): {
         "Name": "Chromium OS system API",
         "URL": "http://www.chromium.org/chromium-os",
         "License": "BSD",
         # Absolute path here is resolved as relative to the source root.
         "License File": "/LICENSE.chromium_os",
     },
     os.path.join('third_party', 'GTM'): {
         "Name": "Google Toolbox for Mac",
         "URL": "http://code.google.com/p/google-toolbox-for-mac/",
         "License": "Apache 2.0",
         "License File": "COPYING",
     },
     os.path.join('third_party', 'lss'): {
         "Name": "linux-syscall-support",
         "URL": "http://code.google.com/p/lss/",
+        "License": "BSD",
+        "License File": "/LICENSE",
     },
     os.path.join('third_party', 'ots'): {
         "Name": "OTS (OpenType Sanitizer)",
         "URL": "http://code.google.com/p/ots/",
         "License": "BSD",
     },
     os.path.join('third_party', 'pdfsqueeze'): {
         "Name": "pdfsqueeze",
         "URL": "http://code.google.com/p/pdfsqueeze/",
         "License": "Apache 2.0",
@@ skipping 23 matching lines @@
         "URL": "http://webkit.org/",
         "License": "BSD and GPL v2",
         # Absolute path here is resolved as relative to the source root.
         "License File": "/webkit/LICENSE",
     },
     os.path.join('third_party', 'webpagereplay'): {
         "Name": "webpagereplay",
         "URL": "http://code.google.com/p/web-page-replay",
         "License": "Apache 2.0",
     },
+    os.path.join('tools', 'grit'): {
+        "Name": "grit",
+        "URL": "http://code.google.com/p/grit-i18n",
+        "License": "BSD",
+    },
+    os.path.join('tools', 'gyp'): {
+        "Name": "gyp",
+        "URL": "http://code.google.com/p/gyp",
+        "License": "BSD",
+    },
+    os.path.join('v8'): {
+        "Name": "gyp",
+        "URL": "http://code.google.com/p/v8",
+        "License": "BSD",
+    },
     os.path.join('v8', 'strongtalk'): {
         "Name": "Strongtalk",
         "URL": "http://www.strongtalk.org/",
         "License": "BSD",
         # Absolute path here is resolved as relative to the source root.
         "License File": "/v8/LICENSE.strongtalk",
     },
 }
 
+# Special value for 'License File' field used to indicate that the license file
+# should not be used in about:credits.
+NOT_SHIPPED = "NOT_SHIPPED"
+
+
 class LicenseError(Exception):
     """We raise this exception when a directory's licensing info isn't
     fully filled out."""
     pass
 
 def AbsolutePath(path, filename):
     """Convert a path in README.chromium to be absolute based on the source
     root."""
     if filename.startswith('/'):
         # Absolute-looking paths are relative to the source root
@@ skipping 39 matching lines @@
                 if line.startswith(field):
                     metadata[key] = line[len(field):]
 
     # Check that all expected metadata is present.
     for key, value in metadata.iteritems():
         if not value:
             raise LicenseError("couldn't find '" + key + "' line "
                                "in README.chromium or licences.py "
                                "SPECIAL_CASES")
 
-    # Check that the license file exists.
-    for filename in (metadata["License File"], "COPYING"):
-        license_path = AbsolutePath(path, filename)
-        if license_path is not None:
-            metadata["License File"] = license_path
-            break
+    # Special-case modules that aren't in the shipping product, so don't need
+    # their license in about:credits.
+    if metadata["License File"] != NOT_SHIPPED:
+        # Check that the license file exists.
+        for filename in (metadata["License File"], "COPYING"):
+            license_path = AbsolutePath(path, filename)
+            if license_path is not None:
+                metadata["License File"] = license_path
+                break
 
-    if not license_path:
-        raise LicenseError("License file not found. "
-                           "Either add a file named LICENSE, "
-                           "import upstream's COPYING if available, "
-                           "or add a 'License File:' line to README.chromium "
-                           "with the appropriate path.")
+        if not license_path:
+            raise LicenseError("License file not found. "
+                               "Either add a file named LICENSE, "
+                               "import upstream's COPYING if available, "
+                               "or add a 'License File:' line to "
+                               "README.chromium with the appropriate path.")
 
     if "Required Text" in metadata:
         required_path = AbsolutePath(path, metadata["Required Text"])
         if required_path is not None:
             metadata["Required Text"] = required_path
         else:
             raise LicenseError("Required text file listed but not found.")
 
     return metadata
 
@@ skipping 83 matching lines @@
     entry_template = open('chrome/browser/resources/about_credits_entry.tmpl',
                           'rb').read()
     entries = []
     for path in sorted(third_party_dirs):
         try:
             metadata = ParseDir(path)
         except LicenseError:
             print >>sys.stderr, ("WARNING: licensing info for " + path +
                                  " is incomplete, skipping.")
             continue
+        if metadata['License File'] == NOT_SHIPPED:
+            print >>sys.stderr, ("Path " + path + " marked as " + NOT_SHIPPED +
+                                 ", skipping.")
+            continue
         env = {
             'name': metadata['Name'],
             'url': metadata['URL'],
             'license': open(metadata['License File'], 'rb').read(),
             'license_unescaped': '',
         }
         if 'Required Text' in metadata:
             required_text = open(metadata['Required Text'], 'rb').read()
             env["license_unescaped"] = required_text
         entries.append(EvaluateTemplate(entry_template, env))
@@ skipping 16 matching lines @@
     elif command == 'credits':
         if not GenerateCredits():
             return 1
     else:
         print __doc__
         return 1
 
 
 if __name__ == '__main__':
   sys.exit(main())