Report all flakes reported to issue tracker also to FindIt

appengine/chromium_try_flakes/endpoints/endpoints.py

Index: appengine/chromium_try_flakes/endpoints/endpoints.py
diff --git a/appengine/chromium_try_flakes/endpoints/endpoints.py b/appengine/chromium_try_flakes/endpoints/endpoints.py
new file mode 100644
index 0000000000000000000000000000000000000000..44ee42652f875ae706ec4996c991bd7d43c5b12b
--- /dev/null
+++ b/appengine/chromium_try_flakes/endpoints/endpoints.py
@@ -0,0 +1,82 @@
+# Copyright 2016 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.
+
+"""Provides functions to work with AppEngine endpoints"""
+
+import httplib2
+import logging
+import time
+
+import apiclient.discovery
+import apiclient.errors
+import oauth2client.appengine
+
+
+AUTH_SCOPE = 'https://www.googleapis.com/auth/userinfo.email'
+
+
+# TODO(akuegel): Do we want to use a different timeout? Do we want to use a
+# cache? See documentation here:
+# https://github.com/jcgregorio/httplib2/blob/master/python2/httplib2/__init__.py#L1142
+def _authenticated_http(http, scope):
+  credentials = oauth2client.appengine.AppAssertionCredentials(scope=scope)
+  return credentials.authorize(http or httplib2.Http())
+
+
+def build_client(api_name, api_version, discovery_url, http=None,
+                 num_tries=5):
+  """Creates HTTP endpoints client, retries connection errors.
+
+  All requests to the endpoints will be authenticated with AppEngine app
+  crendetials.
+
+  Args:
+    api_name: Name of the endpoints API.
+    api_version: Version of the endpoints API.
+    discovery_url: URL of the discovery endpoint. Should contain {api} and
+        {apiVersion} placeholders, e.g. https://your-app.appspot.com/_ah/api/
+        discovery/v1/apis/{api}/{apiVersion}/rest.
+    http: Optional HTTP object. If not specified httplib2.Http() will be used.
+    num_retries: Maximum number of retries to create client.
+
+  Returns:
+    Constructed client.
+  """
+  tries = 0
+  while True:
+    tries += 1
+    try:
+      return apiclient.discovery.build(
+          api_name, api_version,
+          discoveryServiceUrl=discovery_url,
+          http=_authenticated_http(http, AUTH_SCOPE))
+    except apiclient.errors.HttpError as e:
+      if tries == num_tries:
+        logging.exception(
+            'apiclient.discovery.build() failed for %s too many times.',
+            api_name)
+        raise e
+
+      delay = 2 ** (tries - 1)
+      logging.warn(
+          'apiclient.discovery.build() failed for %s: %s', api_name, e)
+      logging.warn(
+          'Retrying apiclient.discovery.build() in %s seconds.', delay)
+      time.sleep(delay)
+
+def retry_request(request, num_tries=5):
+  """Retries provided endpoint request up to num_retries times."""
+  tries = 0
+  while True:
+    tries += 1
+    try:
+      return request.execute()
+    except apiclient.errors.HttpError as e:
+      # This retries internal server (500, 503) and quota (403) errors.
+      # TODO(sergiyb): Figure out if we still need to retry 403 errors. They
+      # were used by codesite to fail on quota errors, but it is unclear whether
+      # Monorail uses same logic or not.
+      if tries == num_tries or e.resp.status not in [403, 500, 503]:
+        raise
+      time.sleep(2 ** (tries - 1))