1 # Copyright 2014 The Chromium Authors. All rights reserved.
2 # Use of this source code is governed by a BSD-style license that can be
3 # found in the LICENSE file.
5 from telemetry
.page
import page
6 from telemetry
.page
import page_set
10 'http://www.facebook.com/barackobama',
11 'https://www.google.com/search?q=barack%20obama',
14 'http://www.baidu.com/s?wd=barack+obama',
15 'http://en.wikipedia.org/wiki/Wikipedia',
17 'http://www.amazon.com/Kindle-Fire-Amazon-Tablet/dp/B0051VVOB2',
18 'http://googleblog.blogspot.com/',
20 'http://www.linkedin.com/in/linustorvalds',
24 'http://yandex.ru/yandsearch?text=barack+obama',
25 'http://translation.babylon.com/',
26 'http://www.bing.com/search?q=barack+obama',
27 'http://wordpress.org/news/',
28 'http://www.ebay.com/sch/i.html?_nkw=antiques',
30 'http://www.soso.com/q?w=barack+obama',
31 'http://www.microsoft.com/en-us/default.aspx',
32 'http://go.mail.ru/search?mailru=1&mg=1&q=barack+obama',
33 'http://vk.com/id118712387',
34 'http://staff.tumblr.com/',
36 'http://sfbay.craigslist.org/mis/',
37 'http://www.ask.com/web?q=barack+obama&search=&qsrc=0&o=0&l=dir',
38 'http://www.apple.com/ipodtouch/',
39 'http://blog.pinterest.com/',
40 'http://pinterest.com/backdrophome/',
43 'http://www.avg.com/us-en/avg-premium-security',
44 'http://googlesystem.blogspot.com/',
46 'http://blog.fc2.com/en/',
49 'http://www.flickr.com/photos/thomashawk/',
50 'http://www.flickr.com/photos/thomashawk/sets/72157600284219965/detail/',
51 # pylint: disable=C0301
52 'http://search.yahoo.com/search?ei=UTF-8&trackingType=go_search_home&p=barack+obama&fr=hsusgo1&sa.x=0&sa.y=0',
53 'http://www.conduit.com/',
59 # pylint: disable=C0301
60 'http://search.mywebsearch.com/mywebsearch/GGmain.jhtml?searchfor=barack+obama',
62 'http://portal.ebay.de/deutschland-schraubt-angebote',
63 'http://www.adobe.com/products/photoshopfamily.html?promoid=JOLIW',
64 'http://global.rakuten.com/us/',
65 # pylint: disable=C0301
66 'http://laundry.about.com/od/kidsandlaundry/f/How-Do-I-Wash-A-Backpack.htm',
67 'http://thepiratebay.se/search/barack%20obama/0/99/0',
69 'http://huffingtonpost.com',
72 'http://www.sogou.com/web?query=barack+obama',
73 # pylint: disable=C0301
74 ('http://www.amazon.de/gp/product/B0051QVF7A/ref=amb_link_170625867_1/'
75 '275-4711375-4099801?ie=UTF8&nav_sdd=aps&pf_rd_m=A3JWKAKR8XB7XF&'
76 'pf_rd_s=center-1&pf_rd_r=1C0XDBPB12WHDM63V11R&pf_rd_t=101&pf_rd_p'
77 '=320475427&pf_rd_i=301128'),
79 'http://mediafire.com',
82 'http://www.godaddy.com/products/secure-hosting.aspx?ci=72738',
83 'http://imgur.com/gallery/b90ZE',
84 'http://home.alipay.com/bank/paymentKJ.htm',
85 'http://amazon.co.jp',
86 # pylint: disable=C0301
87 'http://stackoverflow.com/questions/11227809/why-is-processing-a-sorted-array-faster-than-an-unsorted-array',
88 'http://www.google.com/doubleclick/',
89 'http://search.4shared.com/q/CCAD/1/barack%20obama',
90 'http://dailymotion.com',
92 'http://instagram.com/developer/',
93 'http://livedoor.com',
94 'http://wordpress.org/showcase/',
95 'http://bp.blogspot.com',
96 'http://wigetmedia.com/advertisers',
97 'http://www.search-results.com/web?&q=barack%20obama',
100 'http://torrentz.eu/search?f=barack+obama',
101 'http://livejournal.com',
103 'http://www.weather.com/weather/right-now/Mountain+View+CA+94043',
104 'http://dailymail.co.uk',
105 'http://www.tianya.cn/bbs/index.shtml',
107 'http://theproject.badoo.com/final.phtml',
108 # pylint: disable=C0301
109 'http://www.bankofamerica.com/deposits/checksave/index.cfm?template=check_eBanking',
112 'http://indiatimes.com',
113 'http://deviantart.com',
116 'http://warriorforum.com',
118 'http://pconline.com.cn',
119 'http://mozilla.org',
120 'http://booking.com',
122 'https://www.chase.com/online/Home-Lending/mortgages.htm',
123 'http://addthis.com',
125 'http://news.blogfa.com/',
126 'http://www.stumbleupon.com/jobs',
127 'https://www.dropbox.com/about',
128 'http://www.clicksor.com/publishers/adformat',
129 'http://answers.com',
130 'http://en.softonic.com/',
131 'http://walmart.com',
132 'http://pengyou.com',
133 'http://outbrain.com',
134 'http://comcast.net',
135 'http://foxnews.com',
136 'http://photobucket.com/findstuff/photography%20styles/',
137 'http://bleach.wikia.com/?redirect=no',
138 'http://sourceforge.net/projects/xoops/?source=frontpage&position=1',
140 'http://guardian.co.uk',
141 # pylint: disable=C0301
142 'https://www.wellsfargo.com/jump/enterprise/doublediscount?msc=5589&mplx=10918-70119-3408-64',
143 'http://wikimediafoundation.org/wiki/Home',
145 'http://as.58.com/shuma/',
149 # pylint: disable=C0301
150 'http://search.naver.com/search.naver?where=nexearch&query=barack+obama&sm=top_hty&fbm=0&ie=utf8',
151 'http://statcounter.com/features/?PHPSESSID=bbjcvjr681bcul4vqvgq2qgmo7',
154 'http://www.myspace.com/browse/people',
155 'http://allegro.pl/antyki-i-sztuka',
157 'http://justbeenpaid.com',
158 'http://adultfriendfinder.com',
160 'http://www.leboncoin.fr/annonces/offres/centre/',
161 'http://dictionary.reference.com/',
162 'http://realtime.rediff.com/instasearch#!barack%20obama',
165 'http://www.filestube.com/search.html?q=barack+obama&select=All',
166 'http://xinhuanet.com',
167 'http://www.salesforce.com/sales-cloud/overview/',
168 # pylint: disable=C0301
169 'http://www.squidoo.com/make-cards-and-gift-bags-with-antique-photos',
170 'http://www.domaintools.com/research/',
171 'http://download.cnet.com/windows/?tag=hdr;brandnav',
172 'https://rapidshare.com/#!shop',
173 'http://people.com.cn',
176 'http://nicovideo.jp',
177 # pylint: disable=C0301
178 'http://www.yelp.com/search?find_desc=food&find_loc=San+Jose%2C+CA&ns=1',
179 'http://slideshare.net',
180 'http://archive.org/web/web.php',
181 'http://www.cntv.cn/index.shtml',
182 'http://english.cntv.cn/01/index.shtml',
183 'http://abonnez-vous.orange.fr/residentiel/accueil/accueil.aspx',
184 'http://v.it168.com/',
185 'http://nbcolympics.com',
186 'http://hootsuite.com',
187 # pylint: disable=C0301
188 'http://www.scribd.com/doc/52210329/The-Masters-Augusta-National-s-Amen-Corner-up-close',
189 'http://themeforest.net',
191 'http://www.soku.com/v?keyword=barack%20obama',
193 'http://funmoods.com',
195 'http://telegraph.co.uk',
196 'http://taringa.net',
197 # pylint: disable=C0301
198 'http://www.tripadvisor.com/Tourism-g32701-Mendocino_California-Vacations.html',
201 'http://soundcloud.com/flosstradamus/tracks',
202 'http://w3schools.com/html/default.asp',
203 'http://ameblo.jp/staff/',
206 'http://sweetim.com',
209 'http://www.indeed.com/jobs?q=software&l=Mountain+View%2C+CA',
211 'http://www.xing.com/search/people?search%5Bq%5D=lufthansa',
212 'http://reuters.com',
213 'http://hostgator.com',
214 'http://www.ikea.com/us/en/catalog/categories/departments/living_room/',
215 'http://www.kaixin001.com/award2012/wenming/index.php',
218 'http://samsung.com',
220 'http://espncricinfo.com',
222 # pylint: disable=C0301
223 'http://www.ups.com/content/us/en/bussol/browse/smallbiz/new-to-ups.html?WT.svl=SolExp',
225 'http://ero-advertising.com',
226 'http://mashable.com',
227 'http://iminent.com',
228 'http://rutracker.org',
229 # pylint: disable=C0301
230 'http://www.shopping.hp.com/en_US/home-office/-/products/Laptops/Laptops',
231 # pylint: disable=C0301
232 'http://www.clickbank.com/buy_products.htm?dores=true&mainCategoryId=1340&sortField=POPULARITY&b1=1340',
233 'http://b.hatena.ne.jp/',
234 # pylint: disable=C0301
235 'http://www.youdao.com/search?q=barack+obama&ue=utf8&keyfrom=web.index',
237 'http://nbcnews.com',
238 'http://bitauto.com',
240 'http://www.target.com/c/women/-/N-5xtd3#?lnk=nav_t_spc_1_0',
242 'http://www.aizhan.com/siteall/www.youboy.com/',
243 'http://veiculos-home.mercadolivre.com.br/',
245 'http://flipkart.com',
250 class Top2012Q3Page(page
.Page
):
252 def __init__(self
, url
, ps
):
253 super(Top2012Q3Page
, self
).__init
__(
254 url
=url
, page_set
=ps
, credentials_path
= 'data/credentials.json')
255 self
.archive_data_file
= 'data/2012Q3.json'
257 def RunPageInteractions(self
, action_runner
):
258 with action_runner
.CreateGestureInteraction('ScrollAction'):
259 action_runner
.ScrollPage()
262 class Top2012Q3PageSet(page_set
.PageSet
):
263 """ Pages hand-picked from top-lists in Q32012. """
266 super(Top2012Q3PageSet
, self
).__init
__(
267 archive_data_file
='data/2012Q3.json',
268 bucket
=page_set
.PARTNER_BUCKET
)
271 for url
in TOP_2013_URLS
:
272 self
.AddUserStory(Top2012Q3Page(url
, self
))