Run Chrome Endure tests with network simulation via Web Page Replay.

chrome/test/functional/record_endure.py

Index: chrome/test/functional/record_endure.py
diff --git a/chrome/test/functional/record_endure.py b/chrome/test/functional/record_endure.py
new file mode 100755
index 0000000000000000000000000000000000000000..6edb4df3a0aa565ed5f709ca0dcd1dbc1a810c23
--- /dev/null
+++ b/chrome/test/functional/record_endure.py
@@ -0,0 +1,106 @@
+#!/usr/bin/env python
+# Copyright (c) 2012 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.
+
+"""Record webapp interactions using Web Page Replay for Endure test.
+
+Usage:
+  ./record_endure.py [options] test_or_suite_1 [test_or_suite_2] ...

[Nirnimesh, 2012/08/06 21:36:45]

remove ./
prefix python (so that it works on win)

+  Please make sure that perf_endure.py is in the current working directory.
+Examples:
+  1) ./record_endure.py perf.ChromeEndureGmailTest
+
+    The above command runs each test in perf.ChromeEndureGmailTest. By default,
+    it runs each test for 30 secs and saves the archive file to 'archive.wpr' in
+    the current working directory.
+
+  2) ./record_endure.py --test-length=40 \\

[dennis_jeffrey, 2012/08/06 21:33:46]

just put a single \ at the end of this line, not 2 of them

+     --archive-file="/home/user/ChromeEndureGmailTest.wpr" \\

[dennis_jeffrey, 2012/08/06 21:33:46]

just put a single \ at the end of this line, not 2 of them

+     perf.ChromeEndureGmailTest
+
+    The above command runs each test for 40 seconds and saves the archive to
+    /home/user/ChromeEndureGmailTest.wpr
+"""
+
+import optparse
+import os
+import subprocess
+import sys
+
+import perf
+import webpagereplay
+
+
+class PlainHelpFormatter(optparse.IndentedHelpFormatter):
+  """Format the help message of this script."""
+
+  def format_description(self, description):
+    if description:
+      return description + '\n'
+    else:
+      return ''
+
+
+class Recorder(object):
+  """Record webapp interactions of Endure tests via Web Page Replay."""
+
+  @classmethod
+  def _ParseArgs(cls):
+    parser = optparse.OptionParser(
+        usage='%prog [options] test_or_suite_1 [test_or_suite_2] ...',
+        formatter=PlainHelpFormatter(),
+        description=__doc__)
+    message = 'Length of each test in seconds. Default value is 30 secs.'
+    parser.add_option('--test-length', type='int', default=30,
+                      help=message)
+    message = 'Path and name of the archive file. Default value is ' \
+              '<CURRENT_WORKING_DIR>/archive.wpr'
+    parser.add_option('--archive-file', type='string', default='archive.wpr',
+                      help=message)
+    options, args = parser.parse_args()
+    if len(args) < 1:
+      parser.error('Invalid arguments. Use --help to see usage.')
+    return options, args
+
+  @classmethod
+  def Main(cls):
+    """Set up Web Page Replay and run Endure tests."""
+    options, args = cls._ParseArgs()
+    test_length = options.test_length
+    # Use a greater value than test length to disable stats.
+    perf_stats_interval = test_length + 60
+    archive_path = os.path.abspath(options.archive_file)
+    tests = args[0:]
+    script_path = 'src/chrome/test/data/chrome_endure/' \
+                  'webpagereplay/wpr_deterministic.js'

[slamm_google, 2012/08/06 20:58:44]

It is redundant to have this path here.
Is there a way to have perf_endure.py run WPR instead of doing it here?

[dennis_jeffrey, 2012/08/06 21:33:46]

We need to be able to start WPR in record mode, run one or more pyauto tests,
then stop WPR.  This is so that we can use multiple pyauto tests to record a
single archive file (e.g, some tests that are very similar probably shouldn't
have different archive files, since many things would be duplicated in the
archives).

This script is really just a wrapper around pyauto to (1) start WPR; (2) run one
or more pyauto tests; (3) stop WPR.  We need this because perf_endure.py would
otherwise start and stop WPR for each individual test.

If WPR had an "append" mode for recording archives, we could use that instead. 
Does it have such a feature, or would that be easy to add?  If that was
available, we could start and stop WPR for each test but still have the same
archive file get updated.

+    replay_options = [
+        '--record',
+        '--inject_scripts', perf.FormatChromePath(script_path),
+        ]
+    with webpagereplay.ReplayServer(archive_path, replay_options):
+      env = os.environ.copy()
+      env.update({
+          'TEST_LENGTH': str(test_length),
+          'PERF_STATS_INTERVAL': str(perf_stats_interval),
+          'WPR_RECORD': '1',
+          'DO_NOT_RESTART_PYTHON_FOR_PYAUTO': '1',
+      })
+
+      cmd = ['python', 'perf_endure.py']
+      cmd.extend(tests)
+      print cmd
+      code = subprocess.call(cmd, env=env)
+      print 'Shutting down server.'
+    if (code == 0 and os.path.exists(archive_path) and
+        os.path.isfile(archive_path)):
+      print 'Archive is saved to %s' % archive_path
+      return 0
+    else:
+      print ('Error in recording. The output archive might be invalid. ' +
+             'See log: %s' % webpagereplay.LOG_PATH)
+      return 1
+
+
+if __name__ == '__main__':
+  sys.exit(Recorder().Main())