logo

searx

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

test_yahoo.py (7264B)

    

  1. # -*- coding: utf-8 -*-

  2. from collections import defaultdict

  3. import mock

  4. from searx.engines import yahoo

  5. from searx.testing import SearxTestCase

  6. 

  7. 

  8. class TestYahooEngine(SearxTestCase):

  9. 

  10.     def test_parse_url(self):

  11.         test_url = 'http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\

  12.                    '2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=' +\

  13.                    'dtcJsfP4mEeBOjnVfUQ-'

  14.         url = yahoo.parse_url(test_url)

  15.         self.assertEqual('https://this.is.the.url/', url)

  16. 

  17.         test_url = 'http://r.search.yahoo.com/_ylt=A0LElb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\

  18.                    '2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RS=' +\

  19.                    'dtcJsfP4mEeBOjnVfUQ-'

  20.         url = yahoo.parse_url(test_url)

  21.         self.assertEqual('https://this.is.the.url/', url)

  22. 

  23.         test_url = 'https://this.is.the.url/'

  24.         url = yahoo.parse_url(test_url)

  25.         self.assertEqual('https://this.is.the.url/', url)

  26. 

  27.     def test_request(self):

  28.         yahoo.supported_languages = ['en', 'fr', 'zh-CHT', 'zh-CHS']

  29.         query = 'test_query'

  30.         dicto = defaultdict(dict)

  31.         dicto['pageno'] = 1

  32.         dicto['time_range'] = ''

  33.         dicto['language'] = 'fr-FR'

  34.         params = yahoo.request(query, dicto)

  35.         self.assertIn('url', params)

  36.         self.assertIn(query, params['url'])

  37.         self.assertIn('search.yahoo.com', params['url'])

  38.         self.assertIn('fr', params['url'])

  39.         self.assertIn('cookies', params)

  40.         self.assertIn('sB', params['cookies'])

  41.         self.assertIn('fr', params['cookies']['sB'])

  42. 

  43.         dicto['language'] = 'zh'

  44.         params = yahoo.request(query, dicto)

  45.         self.assertIn('zh_chs', params['url'])

  46.         self.assertIn('zh_chs', params['cookies']['sB'])

  47. 

  48.         dicto['language'] = 'zh-TW'

  49.         params = yahoo.request(query, dicto)

  50.         self.assertIn('zh_cht', params['url'])

  51.         self.assertIn('zh_cht', params['cookies']['sB'])

  52. 

  53.     def test_no_url_in_request_year_time_range(self):

  54.         dicto = defaultdict(dict)

  55.         query = 'test_query'

  56.         dicto['time_range'] = 'year'

  57.         params = yahoo.request(query, dicto)

  58.         self.assertEqual({}, params['url'])

  59. 

  60.     def test_response(self):

  61.         self.assertRaises(AttributeError, yahoo.response, None)

  62.         self.assertRaises(AttributeError, yahoo.response, [])

  63.         self.assertRaises(AttributeError, yahoo.response, '')

  64.         self.assertRaises(AttributeError, yahoo.response, '[]')

  65. 

  66.         response = mock.Mock(text='<html></html>')

  67.         self.assertEqual(yahoo.response(response), [])

  68. 

  69.         html = """

  70. <ol class="reg mb-15 searchCenterMiddle">

  71.     <li class="first">

  72.         <div class="dd algo fst Sr">

  73.             <div class="compTitle">

  74.                 <h3 class="title"><a class=" td-u" href="http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;

  75.                      _ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10

  76.                      /RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"

  77.                      target="_blank" data-bid="54e712e13671c">

  78.                      <b><b>This is the title</b></b></a>

  79.                 </h3>

  80.             </div>

  81.             <div class="compText aAbs">

  82.                 <p class="lh-18"><b><b>This is the </b>content</b>

  83.                 </p>

  84.             </div>

  85.         </div>

  86.     </li>

  87.     <li>

  88.         <div class="dd algo lst Sr">

  89.             <div class="compTitle">

  90.             </div>

  91.             <div class="compText aAbs">

  92.                 <p class="lh-18">This is the second content</p>

  93.             </div>

  94.         </div>

  95.     </li>

  96. </ol>

  97. <div class="dd assist fst lst AlsoTry" data-bid="54e712e138d04">

  98.     <div class="compTitle mb-4 h-17">

  99.         <h3 class="title">Also Try</h3> </div>

  100.     <table class="compTable m-0 ac-1st td-u fz-ms">

  101.         <tbody>

  102.             <tr>

  103.                 <td class="w-50p pr-28"><a href="https://search.yahoo.com/"><B>This is the </B>suggestion<B></B></a>

  104.                 </td>

  105.             </tr>

  106.     </table>

  107. </div>

  108.         """

  109.         response = mock.Mock(text=html)

  110.         results = yahoo.response(response)

  111.         self.assertEqual(type(results), list)

  112.         self.assertEqual(len(results), 2)

  113.         self.assertEqual(results[0]['title'], 'This is the title')

  114.         self.assertEqual(results[0]['url'], 'https://this.is.the.url/')

  115.         self.assertEqual(results[0]['content'], 'This is the content')

  116.         self.assertEqual(results[1]['suggestion'], 'This is the suggestion')

  117. 

  118.         html = """

  119. <ol class="reg mb-15 searchCenterMiddle">

  120.     <li class="first">

  121.         <div class="dd algo fst Sr">

  122.             <div class="compTitle">

  123.                 <h3 class="title"><a class=" td-u" href="http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;

  124.                      _ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10

  125.                      /RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"

  126.                      target="_blank" data-bid="54e712e13671c">

  127.                   <b><b>This is the title</b></b></a>

  128.                 </h3>

  129.             </div>

  130.             <div class="compText aAbs">

  131.                 <p class="lh-18"><b><b>This is the </b>content</b>

  132.                 </p>

  133.             </div>

  134.         </div>

  135.     </li>

  136. </ol>

  137.         """

  138.         response = mock.Mock(text=html)

  139.         results = yahoo.response(response)

  140.         self.assertEqual(type(results), list)

  141.         self.assertEqual(len(results), 1)

  142.         self.assertEqual(results[0]['title'], 'This is the title')

  143.         self.assertEqual(results[0]['url'], 'https://this.is.the.url/')

  144.         self.assertEqual(results[0]['content'], 'This is the content')

  145. 

  146.         html = """

  147.         <li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">

  148.         </li>

  149.         """

  150.         response = mock.Mock(text=html)

  151.         results = yahoo.response(response)

  152.         self.assertEqual(type(results), list)

  153.         self.assertEqual(len(results), 0)

  154. 

  155.     def test_fetch_supported_languages(self):

  156.         html = """<html></html>"""

  157.         response = mock.Mock(text=html)

  158.         results = yahoo._fetch_supported_languages(response)

  159.         self.assertEqual(type(results), list)

  160.         self.assertEqual(len(results), 0)

  161. 

  162.         html = """

  163.         <html>

  164.             <div>

  165.                 <div id="yschlang">

  166.                     <span>

  167.                         <label><input value="lang_ar"></input></label>

  168.                     </span>

  169.                     <span>

  170.                         <label><input value="lang_zh_chs"></input></label>

  171.                         <label><input value="lang_zh_cht"></input></label>

  172.                     </span>

  173.                 </div>

  174.             </div>

  175.         </html>

  176.         """

  177.         response = mock.Mock(text=html)

  178.         languages = yahoo._fetch_supported_languages(response)

  179.         self.assertEqual(type(languages), list)

  180.         self.assertEqual(len(languages), 3)

  181.         self.assertIn('ar', languages)

  182.         self.assertIn('zh-CHS', languages)

  183.         self.assertIn('zh-CHT', languages)