Docserver: Fixes for GitilesFileSystem

chrome/common/extensions/docs/server2/gitiles_file_system.py

 # Copyright 2014 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.
 
 from base64 import b64decode
 from itertools import izip
 import json
 import posixpath
 import traceback
 
 from appengine_url_fetcher import AppEngineUrlFetcher
 from appengine_wrappers import IsDownloadError
 from docs_server_utils import StringIdentity
 from file_system import (FileNotFoundError,
                          FileSystem,
                          FileSystemError,
                          StatInfo)
 from future import All, Future
-from path_util import AssertIsValid, IsDirectory, SplitParent, ToDirectory
+from path_util import AssertIsValid, IsDirectory, ToDirectory
 from third_party.json_schema_compiler.memoize import memoize
 from url_constants import GITILES_BASE, GITILES_BRANCH_BASE
 
 
 _JSON_FORMAT = '?format=JSON'
 _TEXT_FORMAT = '?format=TEXT'
 
 
 def _ParseGitilesJson(json_data):
   '''json.loads with fix-up for non-executable JSON. Use this to parse any JSON
@@ skipping 67 matching lines @@
     #   {
     #     "id": "12a5464de48d2c46bc0b2dc78fafed75aab554fa", # The tree ID.
     #     "entries": [
     #       {
     #         "mode": 33188,
     #         "type": "blob",
     #           "id": "ab971ca447bc4bce415ed4498369e00164d91cb6", # File ID.
     #         "name": ".gitignore"
     #       },
     #       ...
+    #     ]
     #   }
     def list_dir(json_data):
-      return [e['name'] for e in _ParseGitilesJson(json_data)['entries']]
+      entries = _ParseGitilesJson(json_data).get('entries', [])
+      return [e['name'] + ('/' if e['type'] == 'tree' else '') for e in entries]
 
     def fixup_url_format(path):
       # By default, Gitiles URLs display resources in HTML. To get resources
       # suitable for our consumption, a '?format=' string must be appended to
       # the URL. The format may be one of 'JSON' or 'TEXT' for directory or
       # text resources, respectively.
       return path + (_JSON_FORMAT if IsDirectory(path) else _TEXT_FORMAT)
 
     # A list of tuples of the form (path, Future).
     fetches = ((path, self._FetchAsync(fixup_url_format(path)))
                for path in paths)
 
     def parse_contents(results):
       value = {}
       for path, content in izip(paths, results):
         if content is None:
           continue
         # Gitiles encodes text content in base64 (see
         # http://tools.ietf.org/html/rfc4648 for info about base64).
         value[path] = (list_dir if IsDirectory(path) else b64decode)(content)
       return value
     return All(self._ResolveFetchContent(path, future, skip_not_found)
                for path, future in fetches).Then(parse_contents)
 
   def Refresh(self):
     return Future(value=())
 
   @memoize
-  def GetCommitID(self):
-    '''Returns a future that resolves to the commit ID for this branch.
+  def _GetCommitInfo(self, key):
+    '''Gets the commit information specified by |key|.
+
+    The JSON view for commit info looks like:
+      {
+        "commit": "8fd578e1a7b142cd10a4387861f05fb9459b69e2", # Commit ID.
+        "tree": "3ade65d8a91eadd009a6c9feea8f87db2c528a53",   # Tree ID.
+        "parents": [
+          "a477c787fe847ae0482329f69b39ce0fde047359" # Previous commit ID.
+        ],
+        "author": {
+          "name": "...",
+          "email": "...",
+          "time": "Tue Aug 12 17:17:21 2014"
+        },
+        "committer": {
+          "name": "...",
+          "email": "...",
+          "time": "Tue Aug 12 17:18:28 2014"
+        },
+        "message": "...",
+        "tree_diff": [...]
+      }
     '''
     # Commit information for a branch is obtained by appending '?format=JSON'
     # to the branch URL. Note that '<gitiles_url>/<branch>?format=JSON' is
     # different from '<gitiles_url>/<branch>/?format=JSON': the latter serves
     # the root directory JSON content, whereas the former serves the branch
     # commit info JSON content.
     fetch_future = self._fetcher.FetchAsync(self._base_url + _JSON_FORMAT)
     content_future = self._ResolveFetchContent(self._base_url, fetch_future)
-    # The commit info JSON looks like:
-    #
-    #   {
-    #     "commit": "8fd578e1a7b142cd10a4387861f05fb9459b69e2", # Commit ID.
-    #     "tree": "3ade65d8a91eadd009a6c9feea8f87db2c528a53",   # Tree ID.
-    #     "author": {...},
-    #     "committer": {...},
-    #     "message": <codereview message>,
-    #     ...
-    #   }
-    return content_future.Then(lambda json: _ParseGitilesJson(json)['commit'])
+    return content_future.Then(lambda json: _ParseGitilesJson(json)[key])
+
+  def GetCommitID(self):
+    '''Returns a future that resolves to the commit ID for this branch.
+    '''
+    return self._GetCommitInfo('commit')
 
   def Stat(self, path):
     return self.StatAsync(path).Get()
 
   def StatAsync(self, path):
     dir_, filename = posixpath.split(path)
     def stat(content):
       stat_info = _CreateStatInfo(content)
       if stat_info.version is None:
         raise FileSystemError('Failed to find version of dir %s' % dir_)
       if IsDirectory(path):
         return stat_info
       if filename not in stat_info.child_versions:
         raise FileNotFoundError(
             '%s from %s was not in child versions for Stat' % (filename, path))
       return StatInfo(stat_info.child_versions[filename])
     fetch_future = self._FetchAsync(ToDirectory(dir_) + _JSON_FORMAT)
     return self._ResolveFetchContent(path, fetch_future).Then(stat)
 
   def GetIdentity(self):
-    return '@'.join((self.__class__.__name__,
-                     StringIdentity(self._commit or self._branch)))
+    # NOTE: Do not use commit information to create the string identity.
+    # Doing so will mess up caching.
+    if self._commit is None and self._branch != 'master':
+      str_id = GITILES_BRANCH_BASE
+    else:
+      str_id = GITILES_BASE
+    return '@'.join((self.__class__.__name__, StringIdentity(str_id)))