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