Docserver: refactor the Handler/ServerInstance relationship into a servlet

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

Index: chrome/common/extensions/docs/server2/handler.py
diff --git a/chrome/common/extensions/docs/server2/handler.py b/chrome/common/extensions/docs/server2/handler.py
index af2682356785cfc3cab939cbf1f47a3c6e1c7bf2..8fd58124cf868e099135171a653940287f412430 100644
--- a/chrome/common/extensions/docs/server2/handler.py
+++ b/chrome/common/extensions/docs/server2/handler.py
@@ -2,164 +2,24 @@
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
-import logging
-import os
-from StringIO import StringIO
-import traceback
+from appengine_wrappers import webapp
+from cron_servlet import CronServlet
+from render_servlet import RenderServlet
 
-from appengine_wrappers import (
-    DeadlineExceededError, IsDevServer, logservice, memcache, urlfetch, webapp)
-from branch_utility import BranchUtility
-from server_instance import ServerInstance
-import svn_constants
-import time
-
-# The default channel to serve docs for if no channel is specified.
-_DEFAULT_CHANNEL = 'stable'
+_SERVLETS = {
+  'cron': CronServlet,
+}
 
 class Handler(webapp.RequestHandler):
-  # AppEngine instances should never need to call out to SVN. That should only
-  # ever be done by the cronjobs, which then write the result into DataStore,
-  # which is as far as instances look.
-  #
-  # Why? SVN is slow and a bit flaky. Cronjobs failing is annoying but
-  # temporary. Instances failing affects users, and is really bad.
-  #
-  # Anyway - to enforce this, we actually don't give instances access to SVN.
-  # If anything is missing from datastore, it'll be a 404. If the cronjobs
-  # don't manage to catch everything - uhoh. On the other hand, we'll figure it
-  # out pretty soon, and it also means that legitimate 404s are caught before a
-  # round trip to SVN.
-  #
-  # However, we can't expect users of preview.py nor the dev server to run a
-  # cronjob first, so, this is a hack allow that to be online all of the time.
-  # TODO(kalman): achieve this via proper dependency injection.
-  ALWAYS_ONLINE = IsDevServer()
-
   def __init__(self, request, response):
     super(Handler, self).__init__(request, response)
 
-  def _HandleGet(self, path):
-    channel_name, real_path = BranchUtility.SplitChannelNameFromPath(path)
-
-    if channel_name == _DEFAULT_CHANNEL:
-      self.redirect('/%s' % real_path)
-      return
-
-    if channel_name is None:
-      channel_name = _DEFAULT_CHANNEL
-
-    # TODO(kalman): Check if |path| is a directory and serve path/index.html
-    # rather than special-casing apps/extensions.
-    if real_path.strip('/') == 'apps':
-      real_path = 'apps/index.html'
-    if real_path.strip('/') == 'extensions':
-      real_path = 'extensions/index.html'
-
-    constructor = (
-        ServerInstance.CreateOnline if Handler.ALWAYS_ONLINE else
-        ServerInstance.GetOrCreateOffline)
-    server_instance = constructor(channel_name)
-
-    canonical_path = server_instance.path_canonicalizer.Canonicalize(real_path)
-    if real_path != canonical_path:
-      self.redirect(canonical_path)
-      return
-
-    server_instance.Get(real_path, self.request, self.response)
-
-  def _HandleCron(self, path):
-    # Cron strategy:
-    #
-    # Find all public template files and static files, and render them. Most of
-    # the time these won't have changed since the last cron run, so it's a
-    # little wasteful, but hopefully rendering is really fast (if it isn't we
-    # have a problem).
-    class MockResponse(object):
-      def __init__(self):
-        self.status = 200
-        self.out = StringIO()
-        self.headers = {}
-      def set_status(self, status):
-        self.status = status
-      def clear(self, *args):
-        pass
-
-    class MockRequest(object):
-      def __init__(self, path):
-        self.headers = {}
-        self.path = path
-        self.url = '//localhost/%s' % path
-
-    channel = path.split('/')[-1]
-    logging.info('cron/%s: starting' % channel)
-
-    server_instance = ServerInstance.CreateOnline(channel)
-
-    def run_cron_for_dir(d, path_prefix=''):
-      success = True
-      start_time = time.time()
-      files = [f for f in server_instance.content_cache.GetFromFileListing(d)
-               if not f.endswith('/')]
-      logging.info('cron/%s: rendering %s files from %s...' % (
-          channel, len(files), d))
-      for i, f in enumerate(files):
-        error = None
-        path = '%s%s' % (path_prefix, f)
-        try:
-          response = MockResponse()
-          server_instance.Get(path, MockRequest(path), response)
-          if response.status != 200:
-            error = 'Got %s response' % response.status
-        except DeadlineExceededError:
-          logging.error(
-              'cron/%s: deadline exceeded rendering %s (%s of %s): %s' % (
-                  channel, path, i + 1, len(files), traceback.format_exc()))
-          raise
-        except error:
-          pass
-        if error:
-          logging.error('cron/%s: error rendering %s: %s' % (
-              channel, path, error))
-          success = False
-      logging.info('cron/%s: rendering %s files from %s took %s seconds' % (
-          channel, len(files), d, time.time() - start_time))
-      return success
-
-    success = True
-    for path, path_prefix in (
-        # Note: rendering the public templates will pull in all of the private
-        # templates.
-        (svn_constants.PUBLIC_TEMPLATE_PATH, ''),
-        # Note: rendering the public templates will have pulled in the .js and
-        # manifest.json files (for listing examples on the API reference pages),
-        # but there are still images, CSS, etc.
-        (svn_constants.STATIC_PATH, 'static/'),
-        (svn_constants.EXAMPLES_PATH, 'extensions/examples/')):
-      try:
-        # Note: don't try to short circuit any of this stuff. We want to run
-        # the cron for all the directories regardless of intermediate failures.
-        success = run_cron_for_dir(path, path_prefix=path_prefix) and success
-      except DeadlineExceededError:
-        success = False
-        break
-
-    if success:
-      self.response.status = 200
-      self.response.out.write('Success')
-    else:
-      self.response.status = 500
-      self.response.out.write('Failure')
-
-    logging.info('cron/%s: finished' % channel)
-
   def _RedirectSpecialCases(self, path):
-    google_dev_url = 'http://developer.google.com/chrome'
-    if path == '/' or path == '/index.html':
-      self.redirect(google_dev_url)
+    if not path or path == 'index.html':
+      self.redirect('http://developer.google.com/chrome')
       return True
 
-    if path == '/apps.html':
+    if path == 'apps.html':
       self.redirect('/apps/about_apps.html')
       return True
 
@@ -189,7 +49,9 @@ class Handler(webapp.RequestHandler):
     return True
 
   def get(self):
-    path = self.request.path
+    path, request, response = (self.request.path.lstrip('/'),
+                               self.request,
+                               self.response)
 
     if path in ['favicon.ico', 'robots.txt']:
       response.set_status(404)
@@ -197,25 +59,25 @@ class Handler(webapp.RequestHandler):
 
     if self._RedirectSpecialCases(path):
       return
-
-    if path.startswith('/cron'):
-      # Crons often time out, and when they do *and* then eventually try to
-      # flush logs they die. Turn off autoflush and manually do so at the end.
-      logservice.AUTOFLUSH_ENABLED = False
-      try:
-        self._HandleCron(path)
-      finally:
-        logservice.flush()
+    if self._RedirectFromCodeDotGoogleDotCom(path):
       return
 
-    # Redirect paths like "directory" to "directory/". This is so relative
-    # file paths will know to treat this as a directory.
-    if os.path.splitext(path)[1] == '' and path[-1] != '/':
-      self.redirect(path + '/')
-      return
+    if path.startswith('_'):
+      servlet_path = path[1:]
+      if servlet_path.find('/') == -1:
+        servlet_path += '/'
+      servlet_name, servlet_path = servlet_path.split('/', 1)
+      servlet = _SERVLETS.get(servlet_name)
+      if servlet is None:
+        response.out.write('"%s" servlet not found' %  servlet_path)
+        response.set_status(404)
+        return
+    else:
+      servlet_path = path
+      servlet = RenderServlet
 
-    path = path.strip('/')
-    if self._RedirectFromCodeDotGoogleDotCom(path):
-      return
+    servlet_response = servlet(servlet_path, request.headers).Get()
 
-    self._HandleGet(path)
+    response.out.write(servlet_response.content.ToString())
+    response.headers.update(servlet_response.headers)
+    response.status = servlet_response.status