[system-health] Add full Google Search story to the System Health story sets

tools/perf/page_sets/system_health/single_page_stories.py

 # 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.
 
 import logging
 import sys
 
 from page_sets.login_helpers import dropbox_login
 from page_sets.login_helpers import google_login
 
@@ skipping 68 matching lines @@
     yield cls
 
 
 ################################################################################
 # Search and e-commerce.
 ################################################################################
 
 
 class LoadGoogleStory(_SinglePageStory):
   NAME = 'load:search:google'
-  URL = 'https://www.google.com/#hl=en&q=science'
+  URL = 'https://www.google.co.uk/'
+
+
+class SearchGoogleStory(_SinglePageStory):

[petrcermak, 2016/07/13 15:47:15]

I'll move this to a separate file (search_stories.py) after Ulan's patch lands.

+  NAME = 'search:search:google'

[petrcermak, 2016/07/13 15:47:15]

I'm still not completely sure about the names:

1. I'm thinking about dropping the second component of the name (search, social,
games, ...) as it is always unambiguously implied by the third component.

2. As explained in
https://docs.google.com/spreadsheets/d/1t15Ya5ssYBeXAZhHm3RJqfwBRpgWsxoib8_kw...,
I would like to have gradual stories as well:

  * load (currently 'load:*')
  * load -> type
  * load -> type -> return
  * load -> type -> return -> scroll
  * load -> type -> return -> scroll -> click (currently 'search:*')

However, I'm not completely sure how this maps onto story names and if it's even
feasible (in terms of benchmark running time).

[nednguyen, 2016/07/13 16:03:24]

This seems like it a type of browsing stories so should go into
browsing_stories.py? 

[petrcermak, 2016/07/14 13:20:15]

Done.

[nednguyen, 2016/07/14 13:25:17]

Now that I know search is a major activity that people spend time doing online,
I think it's ok to put seach stories to their own group instead of merging to
browsing.
When we talk about search in general, I think there are quite a few major
scenarios that I can think about:
+ search for news
+ search for video (google video search, youtube)
+ search for images
+ search for location 
+ search for a product (e.g: google shopping/bing/amazon/ebay)
+ search for a person in social network

Based on these different themes, we can name things the following way:
"search:news:google", "search:video:bing", "search:product:amazon",...

What do you think?

+  URL = 'https://www.google.co.uk/'
+
+  _SEARCH_BOX_SELECTOR = 'input[aria-label="Search"]'
+  _RESULT_SELECTOR = '.r > a[href*="wikipedia"]'
+
+  def _DidLoadDocument(self, action_runner):
+    # Click on the search box.
+    action_runner.Wait(1)
+    action_runner.WaitForElement(selector=self._SEARCH_BOX_SELECTOR)
+    action_runner.TapElement(selector=self._SEARCH_BOX_SELECTOR)
+
+    # Submit search query.
+    action_runner.Wait(1)
+    action_runner.EnterText('what is science')
+    action_runner.Wait(0.5)
+    action_runner.PressKey('Return')
+
+    # Scroll to the Wikipedia result.
+    action_runner.Wait(1)
+    action_runner.tab.WaitForDocumentReadyStateToBeComplete()
+    action_runner.WaitForElement(selector=self._RESULT_SELECTOR)
+    # TODO(petrcermak): Turn this into a proper Telemetry API (ScrollToElement).
+    result_visible_expression = ('''
+        (function() {
+          var resultElem = document.querySelector(\'%s\');
+          var boundingRect = resultElem.getBoundingClientRect();
+          return boundingRect.bottom >= window.innerHeight
+        })()''' % self._RESULT_SELECTOR)
+    while action_runner.EvaluateJavaScript(result_visible_expression):

[petrcermak, 2016/07/13 15:47:15]

Note that we can't a single RepeatableBrowserDrivenScroll (with repeat_count >
1) since it would scroll too far (due to fling).

+      action_runner.RepeatableBrowserDrivenScroll(y_scroll_distance_ratio=0.75,
+                                                  prevent_fling=False)
+      action_runner.Wait(0.2)
+
+    # Click on the Wikipedia result.
+    action_runner.Wait(1)
+    action_runner.TapElement(selector=self._RESULT_SELECTOR)
+    action_runner.tab.WaitForDocumentReadyStateToBeComplete()
 
 
 class LoadBaiduStory(_SinglePageStory):
   NAME = 'load:search:baidu'
   URL = 'https://www.baidu.com/s?word=google'
 
 
 class LoadYahooStory(_SinglePageStory):
   NAME = 'load:search:yahoo'
   URL = 'https://search.yahoo.com/search;_ylt=?p=google'
@@ skipping 298 matching lines @@
   URL = 'http://www.miniclip.com/games/en/'
   # Desktop only (requires Flash).
   SUPPORTED_PLATFORMS = _DESKTOP_ONLY
 
 
 class LoadAlphabettyStory(_SinglePageStory):
   NAME = 'load:games:alphabetty'
   URL = 'https://king.com/play/alphabetty'
   # Desktop only (requires Flash).
   SUPPORTED_PLATFORMS = _DESKTOP_ONLY