[Telemetry]: Add PeaceKeeper Benchmark suite(Render, DOM, String and Array tests).

tools/perf/benchmarks/peacekeeper.py

+# Copyright (c) 2013 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.
+
+"""PeaceKeeper benchmark suite.
+
+Peacekeeper measures browser's performance by testing its JavaScript
+functionality. JavaScript is a widely used programming language used in the
+creation of modern websites to provide features such as animation, navigation,
+forms and other common requirements. By measuring a browser's ability to handle
+commonly used JavaScript functions Peacekeeper can evaluate its performance.
+Peacekeeper scores are measured in operations per second or rendered frames per
+second depending on the test. Final Score is computed by calculating geometric
+mean of individual tests scores.
+"""
+
+import os
+
+from metrics import statistics
+from telemetry import test
+from telemetry.core import util
+from telemetry.page import page_measurement
+from telemetry.page import page_set
+
+
+class PeaceKeeperMeasurement(page_measurement.PageMeasurement):
+
+  def WillNavigateToPage(self, page, tab):
+    page.script_to_evaluate_on_commit = """
+        var __results = {};
+        var _done = false;
+        var __real_log = window.console.log;
+        var test_frame = null;
+        var benchmark = null;
+        window.console.log = function(msg) {
+          if (typeof(msg) == "string" && (msg.indexOf("benchmark")) == 0) {
+            test_frame = document.getElementById("testFrame");
+            benchmark =  test_frame.contentWindow.benchmark;
+            test_frame.contentWindow.onbeforeunload = {};
+            if ((msg.indexOf("Submit ok.")) != -1) {
+              _done = true;
+              var __data = {};
+              __results["test"] = benchmark.testObjectName;
+              __results["score"] = benchmark.test.result;
+              if (typeof(benchmark.test.unit) != "undefined") {
+                __results["unit"] = benchmark.test.unit;
+              } else {
+                __results["unit"] = benchmark.test.isFps ? "fps" : "ops";
+              }
+            }
+          }
+          __real_log.apply(this, [msg]);
+        }
+        """
+
+  def MeasurePage(self, _, tab, results):
+    def _IsDone():
+      res = tab.EvaluateJavaScript('_done')
+      return res
+    util.WaitFor(_IsDone, 600)

[tonyg, 2013/11/06 21:51:55]

We have a new primitive now, tab.WaitForJavaScriptExpression(). Please use that
instead.

[prasadv, 2013/11/06 22:54:45]

Done.

+    result = tab.EvaluateJavaScript('__results')
+
+    results.Add('Score', 'score', int(result['score']), result['test'],
+                'unimportant')
+
+  def DidRunTest(self, browser, results):
+    # Calculate geometric mean as the total for the combined tests.
+    scores = []
+    for result in results.page_results:
+      scores.append(result['Score'].output_value)
+    total = statistics.GeometricMean(scores)
+    results.AddSummary('Score', 'score', total, 'Total')
+
+
+class PeaceKeeperBenchmark(test.Test):
+  """A base class for Peackeeper benchmarks."""
+  test = PeaceKeeperMeasurement
+
+  def CreatePageSet(self, options):
+    """Makes a PageSet for PeaceKeeper benchmarks."""
+    # Subclasses are expected to define a class member called query_param.
+    if not hasattr(self, 'test_param'):
+      raise NotImplementedError('test_param not in PeaceKeeper benchmark.')
+
+    # The docstring of benchmark classes may also be used as a description
+    # when 'run_benchmarks list' is run.
+    description = self.__doc__ or 'PeaceKeeper Benchmark'
+    test_urls = []
+    for test_name in self.test_param:
+      test_urls.append(
+        {"url": ("http://peacekeeper.futuremark.com/run.action?debug=true&"
+                 "repeat=false&forceSuiteName=%s&forceTestName=%s") %

[tonyg, 2013/11/06 21:51:55]

Nice find! I'm glad you figured out a way to split these up and automate them :)

Annie and I both looking into this and never figured it out.

[prasadv, 2013/11/06 22:54:45]

Done.

+                 (self.tag, test_name)
+        })
+
+    page_set_dict = {
+        'description': description,
+        'archive_data_file': '../page_sets/data/peacekeeper_%s.json' % self.tag,
+        'make_javascript_deterministic': False,
+        'pages': test_urls,
+    }
+    return page_set.PageSet.FromDict(page_set_dict, os.path.abspath(__file__))
+
+
+class PeaceKeeperRender(PeaceKeeperBenchmark):
+  """PeaceKeeper rendering benchmark suite.
+
+  These tests measure your browser's ability to render and modify specific
+  elements used in typical web pages. Rendering tests manipulate the DOM tree in
+  real-time. The tests measure display updating speed (frames per seconds).
+  """
+  tag = 'render'
+  test_param = ['renderGrid01', 'renderGrid02', 'renderGrid03',
+                'renderPhysics']
+
+
+class PeaceKeeperData(PeaceKeeperBenchmark):
+  """PeaceKeeper Data operations benchmark suite.
+
+  These tests measure your browser's ability to add, remove and modify data
+  stored in an array. The Data suite consists of two tests:
+  1. arrayCombined: This test uses all features of the JavaScript Array object.
+  This is a technical test that is not based on profiled data.
+  The source data are different sized arrays of numbers.
+  2. arrayWeighted: This test is similar to 'arrayCombined', but the load is
+  balanced based on profiled data. The source data is a list of all the
+  countries in the world.
+  """
+
+  tag = 'array'
+  test_param = ['arrayCombined01', 'arrayWeighted']
+
+
+class PeaceKeeperDom(PeaceKeeperBenchmark):
+  """PeaceKeeper DOM operations benchmark suite.
+
+  These tests emulate the methods used to create typical dynamic webpages.
+  The DOM tests are based on development experience and the capabilities of the
+  jQuery framework.
+  1. domGetElements: This test uses native DOM methods getElementById and
+    getElementsByName. The elements are not modified.
+  2. domDynamicCreationCreateElement: A common use of DOM is to dynamically
+    create content with JavaScript, this test measures creating objects
+    individually and then appending them to DOM.
+  3. domDynamicCreationInnerHTML: This test is similarl to the previous one,
+    but uses the innerHTML-method.
+  4. domJQueryAttributeFilters: This test does a DOM query with jQuery.
+    It searches elements with specific attributes.
+  5. domJQueryBasicFilters: This test uses filters to query elements from DOM.
+  6. domJQueryBasics: This test queries elements from DOM with basic methods.
+    It is similar to domGetElements, but uses jQuery rather than native methods.
+  7. domJQueryContentFilters: Query elements based on content. This does string
+    searching and these methods are assumed to be time consuming.
+  8. domJQueryHierarchy: Query elements based on hierarchy, such as getting
+    sibling, parent or child nodes from a DOM tree.
+  9. domQueryselector: QuerySelector, which allows JavaScript to search elements
+    from the DOM tree directly without the need to iterate the whole tree
+    through domGetElements.
+  """
+
+  tag = 'dom'
+  test_param = ['domGetElements', 'domDynamicCreationCreateElement',

[tonyg, 2013/11/06 21:51:55]

Recommend one string per line just for readability/maintainability.

[prasadv, 2013/11/06 22:54:45]

Done.

+                'domDynamicCreationInnerHTML', 'domJQueryAttributeFilters',
+                'domJQueryBasicFilters', 'domJQueryBasics',
+                'domJQueryContentFilters', 'domJQueryHierarchy',
+                'domQueryselector']
+
+
+class PeaceKeeperTextParsing(PeaceKeeperBenchmark):
+  """PeaceKeeper Text Parsing benchmark suite.
+
+  These tests measure your browser's performance in typical text manipulations
+  such as using a profanity filter for chats, browser detection and form
+  validation.
+  1. stringChat: This test removes swearing from artificial chat messages.
+    Test measures looping and string replace-method.
+  2. stringDetectBrowser: This test uses string indexOf-method to detect browser
+    and operating system.
+  3. stringFilter: This test filters a list of movies with a given keyword.
+    The behaviour is known as filtering select or continuous filter. It's used
+    to give real time suggestions while a user is filling input fields.
+    The test uses simple regular expressions.
+  4. stringValidateForm: This test uses complex regular expressions to validate
+    user input.
+  5. stringWeighted: This is an artificial test. Methods used and their
+    intensities are chosen based on profiled data.
+  """
+
+  tag = 'string'
+  test_param = ['stringChat', 'stringDetectBrowser', 'stringFilter',
+                'stringWeighted', 'stringValidateForm']
+