test_wikipedia.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263
  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'] = 'all'
  27. params = wikipedia.request(query, dicto)
  28. self.assertIn('en', params['url'])
  29. dicto['language'] = 'xx'
  30. params = wikipedia.request(query, dicto)
  31. self.assertIn('en.wikipedia.org', params['url'])
  32. def test_response(self):
  33. dicto = defaultdict(dict)
  34. dicto['language'] = 'fr'
  35. self.assertRaises(AttributeError, wikipedia.response, None)
  36. self.assertRaises(AttributeError, wikipedia.response, [])
  37. self.assertRaises(AttributeError, wikipedia.response, '')
  38. self.assertRaises(AttributeError, wikipedia.response, '[]')
  39. # page not found
  40. json = """
  41. {
  42. "batchcomplete": "",
  43. "query": {
  44. "normalized": [],
  45. "pages": {
  46. "-1": {
  47. "ns": 0,
  48. "title": "",
  49. "missing": ""
  50. }
  51. }
  52. }
  53. }"""
  54. response = mock.Mock(text=json, search_params=dicto)
  55. self.assertEqual(wikipedia.response(response), [])
  56. # normal case
  57. json = """
  58. {
  59. "batchcomplete": "",
  60. "query": {
  61. "normalized": [],
  62. "pages": {
  63. "12345": {
  64. "pageid": 12345,
  65. "ns": 0,
  66. "title": "The Title",
  67. "extract": "The Title is...",
  68. "thumbnail": {
  69. "source": "img_src.jpg"
  70. },
  71. "pageimage": "img_name.jpg"
  72. }
  73. }
  74. }
  75. }"""
  76. response = mock.Mock(text=json, search_params=dicto)
  77. results = wikipedia.response(response)
  78. self.assertEqual(type(results), list)
  79. self.assertEqual(len(results), 2)
  80. self.assertEqual(results[0]['title'], u'The Title')
  81. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[0]['url'])
  82. self.assertEqual(results[1]['infobox'], u'The Title')
  83. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[1]['id'])
  84. self.assertIn('The Title is...', results[1]['content'])
  85. self.assertEqual(results[1]['img_src'], 'img_src.jpg')
  86. # disambiguation page
  87. json = """
  88. {
  89. "batchcomplete": "",
  90. "query": {
  91. "normalized": [],
  92. "pages": {
  93. "12345": {
  94. "pageid": 12345,
  95. "ns": 0,
  96. "title": "The Title",
  97. "extract": "The Title can be:\\nThe Title 1\\nThe Title 2\\nThe Title 3\\nThe Title 4......................................................................................................................................." """ # noqa
  98. json += """
  99. }
  100. }
  101. }
  102. }"""
  103. response = mock.Mock(text=json, search_params=dicto)
  104. results = wikipedia.response(response)
  105. self.assertEqual(type(results), list)
  106. self.assertEqual(len(results), 2)
  107. # no image
  108. json = """
  109. {
  110. "batchcomplete": "",
  111. "query": {
  112. "normalized": [],
  113. "pages": {
  114. "12345": {
  115. "pageid": 12345,
  116. "ns": 0,
  117. "title": "The Title",
  118. "extract": "The Title is......................................................................................................................................................................................." """ # noqa
  119. json += """
  120. }
  121. }
  122. }
  123. }"""
  124. response = mock.Mock(text=json, search_params=dicto)
  125. results = wikipedia.response(response)
  126. self.assertEqual(type(results), list)
  127. self.assertEqual(len(results), 2)
  128. self.assertIn('The Title is...', results[1]['content'])
  129. self.assertEqual(results[1]['img_src'], None)
  130. # title not in first paragraph
  131. json = u"""
  132. {
  133. "batchcomplete": "",
  134. "query": {
  135. "normalized": [],
  136. "pages": {
  137. "12345": {
  138. "pageid": 12345,
  139. "ns": 0,
  140. "title": "披頭四樂隊",
  141. "extract": "披头士乐队....................................................................................................................................................................................................\\n披頭四樂隊...", """ # noqa
  142. json += """
  143. "thumbnail": {
  144. "source": "img_src.jpg"
  145. },
  146. "pageimage": "img_name.jpg"
  147. }
  148. }
  149. }
  150. }"""
  151. response = mock.Mock(text=json, search_params=dicto)
  152. results = wikipedia.response(response)
  153. self.assertEqual(type(results), list)
  154. self.assertEqual(len(results), 2)
  155. self.assertEqual(results[1]['infobox'], u'披頭四樂隊')
  156. self.assertIn(u'披头士乐队...', results[1]['content'])
  157. def test_fetch_supported_languages(self):
  158. html = u"""<html></html>"""
  159. response = mock.Mock(text=html)
  160. languages = wikipedia._fetch_supported_languages(response)
  161. self.assertEqual(type(languages), dict)
  162. self.assertEqual(len(languages), 0)
  163. html = u"""
  164. <html>
  165. <body>
  166. <div>
  167. <div>
  168. <h3>Table header</h3>
  169. <table class="sortable jquery-tablesorter">
  170. <thead>
  171. <tr>
  172. <th>N</th>
  173. <th>Language</th>
  174. <th>Language (local)</th>
  175. <th>Wiki</th>
  176. <th>Articles</th>
  177. </tr>
  178. </thead>
  179. <tbody>
  180. <tr>
  181. <td>2</td>
  182. <td><a>Swedish</a></td>
  183. <td><a>Svenska</a></td>
  184. <td><a>sv</a></td>
  185. <td><a><b>3000000</b></a></td>
  186. </tr>
  187. <tr>
  188. <td>3</td>
  189. <td><a>Cebuano</a></td>
  190. <td><a>Sinugboanong Binisaya</a></td>
  191. <td><a>ceb</a></td>
  192. <td><a><b>3000000</b></a></td>
  193. </tr>
  194. </tbody>
  195. </table>
  196. <h3>Table header</h3>
  197. <table class="sortable jquery-tablesorter">
  198. <thead>
  199. <tr>
  200. <th>N</th>
  201. <th>Language</th>
  202. <th>Language (local)</th>
  203. <th>Wiki</th>
  204. <th>Articles</th>
  205. </tr>
  206. </thead>
  207. <tbody>
  208. <tr>
  209. <td>2</td>
  210. <td><a>Norwegian (Bokmål)</a></td>
  211. <td><a>Norsk (Bokmål)</a></td>
  212. <td><a>no</a></td>
  213. <td><a><b>100000</b></a></td>
  214. </tr>
  215. </tbody>
  216. </table>
  217. </div>
  218. </div>
  219. </body>
  220. </html>
  221. """
  222. response = mock.Mock(text=html)
  223. languages = wikipedia._fetch_supported_languages(response)
  224. self.assertEqual(type(languages), dict)
  225. self.assertEqual(len(languages), 3)
  226. self.assertIn('sv', languages)
  227. self.assertIn('ceb', languages)
  228. self.assertIn('no', languages)
  229. self.assertEqual(type(languages['sv']), dict)
  230. self.assertEqual(type(languages['ceb']), dict)
  231. self.assertEqual(type(languages['no']), dict)
  232. self.assertIn('name', languages['sv'])
  233. self.assertIn('english_name', languages['sv'])
  234. self.assertIn('articles', languages['sv'])
  235. self.assertEqual(languages['sv']['name'], 'Svenska')
  236. self.assertEqual(languages['sv']['english_name'], 'Swedish')
  237. self.assertEqual(languages['sv']['articles'], 3000000)
  238. self.assertEqual(languages['ceb']['name'], 'Sinugboanong Binisaya')
  239. self.assertEqual(languages['ceb']['english_name'], 'Cebuano')
  240. self.assertEqual(languages['ceb']['articles'], 3000000)
  241. self.assertEqual(languages['no']['name'], u'Norsk (Bokmål)')
  242. self.assertEqual(languages['no']['english_name'], u'Norwegian (Bokmål)')
  243. self.assertEqual(languages['no']['articles'], 100000)