Docserver: Only fetch content versions in the crons, not their contents.

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

Index: chrome/common/extensions/docs/server2/api_models.py
diff --git a/chrome/common/extensions/docs/server2/api_models.py b/chrome/common/extensions/docs/server2/api_models.py
index ac164fb4e6d331288c797e0449ee424df649d9b3..ab6e0bd5db343204f6f3c1dad4f0ce5a92c7e64f 100644
--- a/chrome/common/extensions/docs/server2/api_models.py
+++ b/chrome/common/extensions/docs/server2/api_models.py
@@ -8,8 +8,9 @@ from compiled_file_system import SingleFile, Unicode
 from extensions_paths import API_PATHS
 from features_bundle import HasParent, GetParentName
 from file_system import FileNotFoundError
-from future import All, Future
+from future import All, Future, Race
 from operator import itemgetter
+from path_util import Join
 from platform_util import PlatformToExtensionType
 from schema_util import ProcessSchema
 from third_party.json_schema_compiler.json_schema import DeleteNodes
@@ -58,6 +59,7 @@ class APIModels(object):
                object_store_creator,
                platform):
     self._features_bundle = features_bundle
+    self._file_system = file_system
     self._platform = PlatformToExtensionType(platform)
     self._model_cache = compiled_fs_factory.Create(
         file_system, self._CreateAPIModel, APIModels, category=self._platform)
@@ -86,7 +88,10 @@ class APIModels(object):
     return [name for name, feature in api_features.iteritems()
             if not HasParent(name, feature, api_features)]
 
-  def GetModel(self, api_name):
+  def _GetPotentialPathsForModel(self, api_name):
+    '''Returns the list of file system paths that the model for |api_name|
+    might be located at.
+    '''
     # By default |api_name| is assumed to be given without a path or extension,
     # so combinations of known paths and extension types will be searched.
     api_extensions = ('.json', '.idl')
@@ -116,19 +121,13 @@ class APIModels(object):
           'devtools', file_name.replace(basename,
                                         basename.replace('devtools_' , '')))
 
-    futures = [self._model_cache.GetFromFile(
-                   posixpath.join(path, '%s%s' % (file_name, ext)))
-               for ext in api_extensions
-               for path in api_paths]
-    def resolve():
-      for future in futures:
-        try:
-          return future.Get()
-        # Either the file wasn't found or there was no schema for the file
-        except (FileNotFoundError, ValueError): pass
-      # Propagate the first error if neither were found.
-      futures[0].Get()
-    return Future(callback=resolve)
+    return [Join(path, file_name + ext) for ext in api_extensions
+                                        for path in api_paths]
+
+  def GetModel(self, api_name):
+    futures = [self._model_cache.GetFromFile(path)
+               for path in self._GetPotentialPathsForModel(api_name)]
+    return Race(futures, except_pass=(FileNotFoundError, ValueError))
 
   def GetContentScriptAPIs(self):
     '''Creates a dict of APIs and nodes supported by content scripts in
@@ -169,10 +168,6 @@ class APIModels(object):
       return content_script_apis
     return Future(callback=resolve)
 
-  def Cron(self):
-    futures = [self.GetModel(name) for name in self.GetNames()]
-    return All(futures, except_pass=(FileNotFoundError, ValueError))
-
   def IterModels(self):
     future_models = [(name, self.GetModel(name)) for name in self.GetNames()]
     for name, future_model in future_models:
@@ -182,3 +177,23 @@ class APIModels(object):
         continue
       if model:
         yield name, model
+
+  def Cron(self):
+    def map_cron_paths(op):
+      results = []
+      for name in self.GetNames():
+        results += [op(path) for path in self._GetPotentialPathsForModel(name)]
+      return results
+
+    # Immediately stat everything so that files are guaranteed to be eventually
+    # up to date. See http://crbug.com/398042 for background.
+    # XXX see comment below, return a future for this, hopefully crons will
+    # complete and we don't need to immediately resolve?
+    All(map_cron_paths(self._file_system.StatAsync),
+        except_pass=(FileNotFoundError, ValueError)).Get()
+
+    # Update content in the future.
+    # XXX don't do this for now, see how the crons and instances behave with
+    # just getting the versions.
+    return All(map_cron_paths(self.GetModel),
+               except_pass=(FileNotFoundError, ValueError))