logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://anongit.hacktivis.me/git/searx.git/

test_yahoo_news.py (5420B)

    

  1. # -*- coding: utf-8 -*-
  2. from collections import defaultdict
  3. from datetime import datetime
  4. import mock
  5. from searx.engines import yahoo_news
  6. from searx.testing import SearxTestCase
  9. class TestYahooNewsEngine(SearxTestCase):
  11.     def test_request(self):
  12.         yahoo_news.supported_languages = ['en', 'fr']
  13.         query = 'test_query'
  14.         dicto = defaultdict(dict)
  15.         dicto['pageno'] = 1
  16.         dicto['language'] = 'fr-FR'
  17.         params = yahoo_news.request(query, dicto)
  18.         self.assertIn('url', params)
  19.         self.assertIn(query, params['url'])
  20.         self.assertIn('news.search.yahoo.com', params['url'])
  21.         self.assertIn('fr', params['url'])
  22.         self.assertIn('cookies', params)
  23.         self.assertIn('sB', params['cookies'])
  24.         self.assertIn('fr', params['cookies']['sB'])
  26.     def test_sanitize_url(self):
  27.         url = "test.url"
  28.         self.assertEqual(url, yahoo_news.sanitize_url(url))
  30.         url = "www.yahoo.com/;_ylt=test"
  31.         self.assertEqual("www.yahoo.com/", yahoo_news.sanitize_url(url))
  33.     def test_response(self):
  34.         self.assertRaises(AttributeError, yahoo_news.response, None)
  35.         self.assertRaises(AttributeError, yahoo_news.response, [])
  36.         self.assertRaises(AttributeError, yahoo_news.response, '')
  37.         self.assertRaises(AttributeError, yahoo_news.response, '[]')
  39.         response = mock.Mock(text='<html></html>')
  40.         self.assertEqual(yahoo_news.response(response), [])
  42.         html = """
  43.         <ol class=" reg searchCenterMiddle">
  44.             <li class="first">
  45.                 <div class="compTitle">
  46.                     <h3>
  47.                         <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
  48.                            This is
  49.                            the <b>title</b>...
  50.                         </a>
  51.                     </h3>
  52.                 </div>
  53.                 <div>
  54.                     <span class="cite">Business via Yahoo!</span>
  55.                     <span class="tri fc-2nd ml-10">May 01 10:00 AM</span>
  56.                 </div>
  57.                 <div class="compText">
  58.                    This is the content
  59.                </div>
  60.             </li>
  61.             <li class="first">
  62.                 <div class="compTitle">
  63.                     <h3>
  64.                         <a class="yschttl spt" target="_blank">
  65.                         </a>
  66.                     </h3>
  67.                 </div>
  68.                 <div class="compText">
  69.                </div>
  70.             </li>
  71.         </ol>
  72.         """
  73.         response = mock.Mock(text=html)
  74.         results = yahoo_news.response(response)
  75.         self.assertEqual(type(results), list)
  76.         self.assertEqual(len(results), 1)
  77.         self.assertEqual(results[0]['title'], 'This is the title...')
  78.         self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
  79.         self.assertEqual(results[0]['content'], 'This is the content')
  81.         html = """
  82.         <ol class=" reg searchCenterMiddle">
  83.             <li class="first">
  84.                 <div class="compTitle">
  85.                     <h3>
  86.                         <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
  87.                             This is
  88.                             the <b>title</b>...
  89.                         </a>
  90.                     </h3>
  91.                 </div>
  92.                 <div>
  93.                     <span class="cite">Business via Yahoo!</span>
  94.                     <span class="tri fc-2nd ml-10">2 hours, 22 minutes ago</span>
  95.                 </div>
  96.                 <div class="compText">
  97.                     This is the content
  98.                 </div>
  99.             </li>
  100.             <li>
  101.                 <div class="compTitle">
  102.                     <h3>
  103.                         <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
  104.                             This is
  105.                             the <b>title</b>...
  106.                         </a>
  107.                     </h3>
  108.                 </div>
  109.                 <div>
  110.                     <span class="cite">Business via Yahoo!</span>
  111.                     <span class="tri fc-2nd ml-10">22 minutes ago</span>
  112.                 </div>
  113.                 <div class="compText">
  114.                     This is the content
  115.                 </div>
  116.             </li>
  117.             <li>
  118.                 <div class="compTitle">
  119.                     <h3>
  120.                         <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
  121.                             This is
  122.                             the <b>title</b>...
  123.                         </a>
  124.                     </h3>
  125.                 </div>
  126.                 <div>
  127.                     <span class="cite">Business via Yahoo!</span>
  128.                     <span class="tri fc-2nd ml-10">Feb 03 09:45AM 1900</span>
  129.                 </div>
  130.                 <div class="compText">
  131.                     This is the content
  132.                 </div>
  133.             </li>
  134.         </ol>
  135.         """
  136.         response = mock.Mock(text=html)
  137.         results = yahoo_news.response(response)
  138.         self.assertEqual(type(results), list)
  139.         self.assertEqual(len(results), 3)
  140.         self.assertEqual(results[0]['title'], 'This is the title...')
  141.         self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
  142.         self.assertEqual(results[0]['content'], 'This is the content')
  143.         self.assertEqual(results[2]['publishedDate'].year, datetime.now().year)