test_bing.py 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178
  1. # -*- coding: utf-8 -*-
  2. from collections import defaultdict
  3. import mock
  4. from searx.engines import bing
  5. from searx.testing import SearxTestCase
  6. class TestBingEngine(SearxTestCase):
  7. def test_request(self):
  8. bing.supported_languages = ['en', 'fr', 'zh-CHS', 'zh-CHT', 'pt-PT', 'pt-BR']
  9. query = u'test_query'
  10. dicto = defaultdict(dict)
  11. dicto['pageno'] = 1
  12. dicto['language'] = 'fr-FR'
  13. params = bing.request(query.encode('utf-8'), dicto)
  14. self.assertTrue('url' in params)
  15. self.assertTrue(query in params['url'])
  16. self.assertTrue('language%3AFR' in params['url'])
  17. self.assertTrue('bing.com' in params['url'])
  18. dicto['language'] = 'all'
  19. params = bing.request(query.encode('utf-8'), dicto)
  20. self.assertTrue('language' in params['url'])
  21. def test_response(self):
  22. dicto = defaultdict(dict)
  23. dicto['pageno'] = 1
  24. dicto['language'] = 'fr-FR'
  25. self.assertRaises(AttributeError, bing.response, None)
  26. self.assertRaises(AttributeError, bing.response, [])
  27. self.assertRaises(AttributeError, bing.response, '')
  28. self.assertRaises(AttributeError, bing.response, '[]')
  29. response = mock.Mock(text='<html></html>')
  30. response.search_params = dicto
  31. self.assertEqual(bing.response(response), [])
  32. response = mock.Mock(text='<html></html>')
  33. response.search_params = dicto
  34. self.assertEqual(bing.response(response), [])
  35. html = """
  36. <div>
  37. <div id="b_tween">
  38. <span class="sb_count" data-bm="4">23 900 000 résultats</span>
  39. </div>
  40. <ol id="b_results" role="main">
  41. <div class="sa_cc" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
  42. <div Class="sa_mc">
  43. <div class="sb_tlst">
  44. <h3>
  45. <a href="http://this.should.be.the.link/" h="ID=SERP,5124.1">
  46. <strong>This</strong> should be the title</a>
  47. </h3>
  48. </div>
  49. <div class="sb_meta"><cite><strong>this</strong>.meta.com</cite>
  50. <span class="c_tlbxTrg">
  51. <span class="c_tlbxH" H="BASE:CACHEDPAGEDEFAULT" K="SERP,5125.1">
  52. </span>
  53. </span>
  54. </div>
  55. <p><strong>This</strong> should be the content.</p>
  56. </div>
  57. </div>
  58. </ol>
  59. </div>
  60. """
  61. response = mock.Mock(text=html)
  62. response.search_params = dicto
  63. results = bing.response(response)
  64. self.assertEqual(type(results), list)
  65. self.assertEqual(len(results), 2)
  66. self.assertEqual(results[0]['title'], 'This should be the title')
  67. self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
  68. self.assertEqual(results[0]['content'], 'This should be the content.')
  69. self.assertEqual(results[-1]['number_of_results'], 23900000)
  70. html = """
  71. <div>
  72. <div id="b_tween">
  73. <span class="sb_count" data-bm="4">9-18 résultats sur 23 900 000</span>
  74. </div>
  75. <ol id="b_results" role="main">
  76. <li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
  77. <div Class="sa_mc">
  78. <div class="sb_tlst">
  79. <h2>
  80. <a href="http://this.should.be.the.link/" h="ID=SERP,5124.1">
  81. <strong>This</strong> should be the title</a>
  82. </h2>
  83. </div>
  84. <div class="sb_meta"><cite><strong>this</strong>.meta.com</cite>
  85. <span class="c_tlbxTrg">
  86. <span class="c_tlbxH" H="BASE:CACHEDPAGEDEFAULT" K="SERP,5125.1">
  87. </span>
  88. </span>
  89. </div>
  90. <p><strong>This</strong> should be the content.</p>
  91. </div>
  92. </li>
  93. </ol>
  94. </div>
  95. """
  96. dicto['pageno'] = 2
  97. response = mock.Mock(text=html)
  98. response.search_params = dicto
  99. results = bing.response(response)
  100. self.assertEqual(type(results), list)
  101. self.assertEqual(len(results), 2)
  102. self.assertEqual(results[0]['title'], 'This should be the title')
  103. self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
  104. self.assertEqual(results[0]['content'], 'This should be the content.')
  105. self.assertEqual(results[-1]['number_of_results'], 23900000)
  106. html = """
  107. <div>
  108. <div id="b_tween">
  109. <span class="sb_count" data-bm="4">23 900 000 résultats</span>
  110. </div>
  111. <ol id="b_results" role="main">
  112. <li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
  113. <div Class="sa_mc">
  114. <div class="sb_tlst">
  115. <h2>
  116. <a href="http://this.should.be.the.link/" h="ID=SERP,5124.1">
  117. <strong>This</strong> should be the title</a>
  118. </h2>
  119. </div>
  120. <div class="sb_meta"><cite><strong>this</strong>.meta.com</cite>
  121. <span class="c_tlbxTrg">
  122. <span class="c_tlbxH" H="BASE:CACHEDPAGEDEFAULT" K="SERP,5125.1">
  123. </span>
  124. </span>
  125. </div>
  126. <p><strong>This</strong> should be the content.</p>
  127. </div>
  128. </li>
  129. </ol>
  130. </div>
  131. """
  132. dicto['pageno'] = 33900000
  133. response = mock.Mock(text=html)
  134. response.search_params = dicto
  135. results = bing.response(response)
  136. self.assertEqual(bing.response(response), [])
  137. def test_fetch_supported_languages(self):
  138. html = """<html></html>"""
  139. response = mock.Mock(text=html)
  140. results = bing._fetch_supported_languages(response)
  141. self.assertEqual(type(results), list)
  142. self.assertEqual(len(results), 0)
  143. html = """
  144. <html>
  145. <body>
  146. <form>
  147. <div id="limit-languages">
  148. <div>
  149. <div><input id="es" value="es"></input></div>
  150. </div>
  151. <div>
  152. <div><input id="pt_BR" value="pt_BR"></input></div>
  153. <div><input id="pt_PT" value="pt_PT"></input></div>
  154. </div>
  155. </div>
  156. </form>
  157. </body>
  158. </html>
  159. """
  160. response = mock.Mock(text=html)
  161. languages = bing._fetch_supported_languages(response)
  162. self.assertEqual(type(languages), list)
  163. self.assertEqual(len(languages), 3)
  164. self.assertIn('es', languages)
  165. self.assertIn('pt-BR', languages)
  166. self.assertIn('pt-PT', languages)