test_wikipedia.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260
  1. # -*- coding: utf-8 -*-
  2. from collections import defaultdict
  3. import mock
  4. from searx.engines import wikipedia
  5. from searx.testing import SearxTestCase
  6. class TestWikipediaEngine(SearxTestCase):
  7. def test_request(self):
  8. wikipedia.supported_languages = ['fr', 'en']
  9. query = 'test_query'
  10. dicto = defaultdict(dict)
  11. dicto['language'] = 'fr-FR'
  12. params = wikipedia.request(query.encode('utf-8'), dicto)
  13. self.assertIn('url', params)
  14. self.assertIn(query, params['url'])
  15. self.assertIn('test_query', params['url'])
  16. self.assertIn('Test_Query', params['url'])
  17. self.assertIn('fr.wikipedia.org', params['url'])
  18. query = u'Test_Query'
  19. params = wikipedia.request(query.encode('utf-8'), dicto)
  20. self.assertIn('Test_Query', params['url'])
  21. self.assertNotIn('test_query', params['url'])
  22. dicto['language'] = 'all'
  23. params = wikipedia.request(query, dicto)
  24. self.assertIn('en', params['url'])
  25. dicto['language'] = 'xx'
  26. params = wikipedia.request(query, dicto)
  27. self.assertIn('en', params['url'])
  28. def test_response(self):
  29. dicto = defaultdict(dict)
  30. dicto['language'] = 'fr'
  31. self.assertRaises(AttributeError, wikipedia.response, None)
  32. self.assertRaises(AttributeError, wikipedia.response, [])
  33. self.assertRaises(AttributeError, wikipedia.response, '')
  34. self.assertRaises(AttributeError, wikipedia.response, '[]')
  35. # page not found
  36. json = """
  37. {
  38. "batchcomplete": "",
  39. "query": {
  40. "normalized": [],
  41. "pages": {
  42. "-1": {
  43. "ns": 0,
  44. "title": "",
  45. "missing": ""
  46. }
  47. }
  48. }
  49. }"""
  50. response = mock.Mock(text=json, search_params=dicto)
  51. self.assertEqual(wikipedia.response(response), [])
  52. # normal case
  53. json = """
  54. {
  55. "batchcomplete": "",
  56. "query": {
  57. "normalized": [],
  58. "pages": {
  59. "12345": {
  60. "pageid": 12345,
  61. "ns": 0,
  62. "title": "The Title",
  63. "extract": "The Title is...",
  64. "thumbnail": {
  65. "source": "img_src.jpg"
  66. },
  67. "pageimage": "img_name.jpg"
  68. }
  69. }
  70. }
  71. }"""
  72. response = mock.Mock(text=json, search_params=dicto)
  73. results = wikipedia.response(response)
  74. self.assertEqual(type(results), list)
  75. self.assertEqual(len(results), 2)
  76. self.assertEqual(results[0]['title'], u'The Title')
  77. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[0]['url'])
  78. self.assertEqual(results[1]['infobox'], u'The Title')
  79. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[1]['id'])
  80. self.assertIn('The Title is...', results[1]['content'])
  81. self.assertEqual(results[1]['img_src'], 'img_src.jpg')
  82. # disambiguation page
  83. json = """
  84. {
  85. "batchcomplete": "",
  86. "query": {
  87. "normalized": [],
  88. "pages": {
  89. "12345": {
  90. "pageid": 12345,
  91. "ns": 0,
  92. "title": "The Title",
  93. "extract": "The Title can be:\\nThe Title 1\\nThe Title 2\\nThe Title 3\\nThe Title 4......................................................................................................................................." """ # noqa
  94. json += """
  95. }
  96. }
  97. }
  98. }"""
  99. response = mock.Mock(text=json, search_params=dicto)
  100. results = wikipedia.response(response)
  101. self.assertEqual(type(results), list)
  102. self.assertEqual(len(results), 2)
  103. # no image
  104. json = """
  105. {
  106. "batchcomplete": "",
  107. "query": {
  108. "normalized": [],
  109. "pages": {
  110. "12345": {
  111. "pageid": 12345,
  112. "ns": 0,
  113. "title": "The Title",
  114. "extract": "The Title is......................................................................................................................................................................................." """ # noqa
  115. json += """
  116. }
  117. }
  118. }
  119. }"""
  120. response = mock.Mock(text=json, search_params=dicto)
  121. results = wikipedia.response(response)
  122. self.assertEqual(type(results), list)
  123. self.assertEqual(len(results), 2)
  124. self.assertIn('The Title is...', results[1]['content'])
  125. self.assertEqual(results[1]['img_src'], None)
  126. # title not in first paragraph
  127. json = u"""
  128. {
  129. "batchcomplete": "",
  130. "query": {
  131. "normalized": [],
  132. "pages": {
  133. "12345": {
  134. "pageid": 12345,
  135. "ns": 0,
  136. "title": "披頭四樂隊",
  137. "extract": "披头士乐队....................................................................................................................................................................................................\\n披頭四樂隊...", """ # noqa
  138. json += """
  139. "thumbnail": {
  140. "source": "img_src.jpg"
  141. },
  142. "pageimage": "img_name.jpg"
  143. }
  144. }
  145. }
  146. }"""
  147. response = mock.Mock(text=json, search_params=dicto)
  148. results = wikipedia.response(response)
  149. self.assertEqual(type(results), list)
  150. self.assertEqual(len(results), 2)
  151. self.assertEqual(results[1]['infobox'], u'披頭四樂隊')
  152. self.assertIn(u'披头士乐队...', results[1]['content'])
  153. def test_fetch_supported_languages(self):
  154. html = u"""<html></html>"""
  155. response = mock.Mock(text=html)
  156. languages = wikipedia._fetch_supported_languages(response)
  157. self.assertEqual(type(languages), dict)
  158. self.assertEqual(len(languages), 0)
  159. html = u"""
  160. <html>
  161. <body>
  162. <div>
  163. <div>
  164. <h3>Table header</h3>
  165. <table class="sortable jquery-tablesorter">
  166. <thead>
  167. <tr>
  168. <th>N</th>
  169. <th>Language</th>
  170. <th>Language (local)</th>
  171. <th>Wiki</th>
  172. <th>Articles</th>
  173. </tr>
  174. </thead>
  175. <tbody>
  176. <tr>
  177. <td>2</td>
  178. <td><a>Swedish</a></td>
  179. <td><a>Svenska</a></td>
  180. <td><a>sv</a></td>
  181. <td><a><b>3000000</b></a></td>
  182. </tr>
  183. <tr>
  184. <td>3</td>
  185. <td><a>Cebuano</a></td>
  186. <td><a>Sinugboanong Binisaya</a></td>
  187. <td><a>ceb</a></td>
  188. <td><a><b>3000000</b></a></td>
  189. </tr>
  190. </tbody>
  191. </table>
  192. <h3>Table header</h3>
  193. <table class="sortable jquery-tablesorter">
  194. <thead>
  195. <tr>
  196. <th>N</th>
  197. <th>Language</th>
  198. <th>Language (local)</th>
  199. <th>Wiki</th>
  200. <th>Articles</th>
  201. </tr>
  202. </thead>
  203. <tbody>
  204. <tr>
  205. <td>2</td>
  206. <td><a>Norwegian (Bokmål)</a></td>
  207. <td><a>Norsk (Bokmål)</a></td>
  208. <td><a>no</a></td>
  209. <td><a><b>100000</b></a></td>
  210. </tr>
  211. </tbody>
  212. </table>
  213. </div>
  214. </div>
  215. </body>
  216. </html>
  217. """
  218. response = mock.Mock(text=html)
  219. languages = wikipedia._fetch_supported_languages(response)
  220. self.assertEqual(type(languages), dict)
  221. self.assertEqual(len(languages), 3)
  222. self.assertIn('sv', languages)
  223. self.assertIn('ceb', languages)
  224. self.assertIn('no', languages)
  225. self.assertEqual(type(languages['sv']), dict)
  226. self.assertEqual(type(languages['ceb']), dict)
  227. self.assertEqual(type(languages['no']), dict)
  228. self.assertIn('name', languages['sv'])
  229. self.assertIn('english_name', languages['sv'])
  230. self.assertIn('articles', languages['sv'])
  231. self.assertEqual(languages['sv']['name'], 'Svenska')
  232. self.assertEqual(languages['sv']['english_name'], 'Swedish')
  233. self.assertEqual(languages['sv']['articles'], 3000000)
  234. self.assertEqual(languages['ceb']['name'], 'Sinugboanong Binisaya')
  235. self.assertEqual(languages['ceb']['english_name'], 'Cebuano')
  236. self.assertEqual(languages['ceb']['articles'], 3000000)
  237. self.assertEqual(languages['no']['name'], u'Norsk (Bokmål)')
  238. self.assertEqual(languages['no']['english_name'], u'Norwegian (Bokmål)')
  239. self.assertEqual(languages['no']['articles'], 100000)