reland "rebaseline_server: download actual-results.json files from GCS instead of SVN"

gm/rebaseline_server/download_actuals.py

Index: gm/rebaseline_server/download_actuals.py
diff --git a/gm/rebaseline_server/download_actuals.py b/gm/rebaseline_server/download_actuals.py
index 636958be6b545c2251628a3b0f65f2d2db0c25d1..2f92898fd4b0451feb3dd1895157f21b51283919 100755
--- a/gm/rebaseline_server/download_actuals.py
+++ b/gm/rebaseline_server/download_actuals.py
@@ -10,44 +10,19 @@ Download actual GM results for a particular builder.
 """
 
 # System-level imports
-import contextlib
 import optparse
 import os
 import posixpath
 import re
-import shutil
-import sys
-import urllib
 import urllib2
-import urlparse
 
 # Imports from within Skia
-#
-# We need to add the 'gm' and 'tools' directories, so that we can import
-# gm_json.py and buildbot_globals.py.
-#
-# Make sure that these dirs are in the PYTHONPATH, but add them at the *end*
-# so any dirs that are already in the PYTHONPATH will be preferred.
-#
-# TODO(epoger): Is it OK for this to depend on the 'tools' dir, given that
-# the tools dir is dependent on the 'gm' dir (to import gm_json.py)?
-TRUNK_DIRECTORY = os.path.dirname(os.path.dirname(os.path.dirname(__file__)))
-GM_DIRECTORY = os.path.join(TRUNK_DIRECTORY, 'gm')
-TOOLS_DIRECTORY = os.path.join(TRUNK_DIRECTORY, 'tools')
-if GM_DIRECTORY not in sys.path:
-  sys.path.append(GM_DIRECTORY)
-if TOOLS_DIRECTORY not in sys.path:
-  sys.path.append(TOOLS_DIRECTORY)
+import fix_pythonpath  # must do this first
+from pyutils import gs_utils
+from pyutils import url_utils
 import buildbot_globals
 import gm_json
 
-# Imports from third-party code
-APICLIENT_DIRECTORY = os.path.join(
-    TRUNK_DIRECTORY, 'third_party', 'externals', 'google-api-python-client')
-if APICLIENT_DIRECTORY not in sys.path:
-  sys.path.append(APICLIENT_DIRECTORY)
-from googleapiclient.discovery import build as build_service
-
 
 GM_SUMMARIES_BUCKET = buildbot_globals.Get('gm_summaries_bucket')
 DEFAULT_ACTUALS_BASE_URL = (
@@ -105,98 +80,19 @@ class Download(object):
             test_name=test, hash_type=hash_type, hash_digest=hash_digest,
             gm_actuals_root_url=self._gm_actuals_root_url)
         dest_path = os.path.join(dest_dir, config, test + '.png')
-        # TODO(epoger): To speed this up, we should only download files that
-        # we don't already have on local disk.
-        copy_contents(source_url=source_url, dest_path=dest_path,
-                      create_subdirs_if_needed=True)
-
-
-def create_filepath_url(filepath):
-  """ Returns a file:/// URL pointing at the given filepath on local disk.
+        url_utils.copy_contents(source_url=source_url, dest_path=dest_path,
+                                create_subdirs_if_needed=True)
 
-  For now, this is only used by unittests, but I anticipate it being useful
-  in production, as a way for developers to run rebaseline_server over locally
-  generated images.
 
-  TODO(epoger): Move this function, and copy_contents(), into a shared
-  utility module.  They are generally useful.
+def get_builders_list(summaries_bucket=GM_SUMMARIES_BUCKET):
+  """ Returns the list of builders we have actual results for.
 
   Args:
-    filepath: string; path to a file on local disk (may be absolute or relative,
-        and the file does not need to exist)
-
-  Returns:
-    A file:/// URL pointing at the file.  Regardless of whether filepath was
-        specified as a relative or absolute path, the URL will contain an
-        absolute path to the file.
-
-  Raises:
-    An Exception, if filepath is already a URL.
+    summaries_bucket: Google Cloud Storage bucket containing the summary
+        JSON files
   """
-  if urlparse.urlparse(filepath).scheme:
-    raise Exception('"%s" is already a URL' % filepath)
-  return urlparse.urljoin(
-      'file:', urllib.pathname2url(os.path.abspath(filepath)))
-
-
-def copy_contents(source_url, dest_path, create_subdirs_if_needed=False):
-  """ Copies the full contents of the URL 'source_url' into
-  filepath 'dest_path'.
-
-  Args:
-    source_url: string; complete URL to read from
-    dest_path: string; complete filepath to write to (may be absolute or
-        relative)
-    create_subdirs_if_needed: boolean; whether to create subdirectories as
-        needed to create dest_path
-
-  Raises:
-    Some subclass of Exception if unable to read source_url or write dest_path.
-  """
-  if create_subdirs_if_needed:
-    dest_dir = os.path.dirname(dest_path)
-    if not os.path.exists(dest_dir):
-      os.makedirs(dest_dir)
-  with contextlib.closing(urllib.urlopen(source_url)) as source_handle:
-    with open(dest_path, 'wb') as dest_handle:
-      shutil.copyfileobj(fsrc=source_handle, fdst=dest_handle)
-
-
-def gcs_list_bucket_contents(bucket, subdir=None):
-  """ Returns files in the Google Cloud Storage bucket as a (dirs, files) tuple.
-
-  Uses the API documented at
-  https://developers.google.com/storage/docs/json_api/v1/objects/list
-
-  Args:
-    bucket: name of the Google Storage bucket
-    subdir: directory within the bucket to list, or None for root directory
-  """
-  # The GCS command relies on the subdir name (if any) ending with a slash.
-  if subdir and not subdir.endswith('/'):
-    subdir += '/'
-  subdir_length = len(subdir) if subdir else 0
-
-  storage = build_service('storage', 'v1')
-  command = storage.objects().list(
-      bucket=bucket, delimiter='/', fields='items(name),prefixes',
-      prefix=subdir)
-  results = command.execute()
-
-  # The GCS command returned two subdicts:
-  # prefixes: the full path of every directory within subdir, with trailing '/'
-  # items: property dict for each file object within subdir
-  #        (including 'name', which is full path of the object)
-  dirs = []
-  for dir_fullpath in results.get('prefixes', []):
-    dir_basename = dir_fullpath[subdir_length:]
-    dirs.append(dir_basename[:-1])  # strip trailing slash
-  files = []
-  for file_properties in results.get('items', []):
-    file_fullpath = file_properties['name']
-    file_basename = file_fullpath[subdir_length:]
-    files.append(file_basename)
-  return (dirs, files)
+  dirs, _ = gs_utils.list_bucket_contents(bucket=GM_SUMMARIES_BUCKET)
+  return dirs
 
 
 def main():
@@ -234,8 +130,7 @@ def main():
   (params, remaining_args) = parser.parse_args()
 
   if params.list_builders:
-    dirs, _ = gcs_list_bucket_contents(bucket=GM_SUMMARIES_BUCKET)
-    print '\n'.join(dirs)
+    print '\n'.join(get_builders_list())
     return
 
   # Make sure all required options were set,