test_wikipedia.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266
  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. <<<<<<< HEAD
  24. dicto['language'] = 'nb'
  25. params = wikipedia.request(query, dicto)
  26. self.assertIn('no.wikipedia.org', params['url'])
  27. =======
  28. dicto['language'] = 'all'
  29. params = wikipedia.request(query, dicto)
  30. self.assertIn('en', params['url'])
  31. >>>>>>> Revert "remove 'all' option from search languages"
  32. dicto['language'] = 'xx'
  33. params = wikipedia.request(query, dicto)
  34. self.assertIn('en.wikipedia.org', params['url'])
  35. def test_response(self):
  36. dicto = defaultdict(dict)
  37. dicto['language'] = 'fr'
  38. self.assertRaises(AttributeError, wikipedia.response, None)
  39. self.assertRaises(AttributeError, wikipedia.response, [])
  40. self.assertRaises(AttributeError, wikipedia.response, '')
  41. self.assertRaises(AttributeError, wikipedia.response, '[]')
  42. # page not found
  43. json = """
  44. {
  45. "batchcomplete": "",
  46. "query": {
  47. "normalized": [],
  48. "pages": {
  49. "-1": {
  50. "ns": 0,
  51. "title": "",
  52. "missing": ""
  53. }
  54. }
  55. }
  56. }"""
  57. response = mock.Mock(text=json, search_params=dicto)
  58. self.assertEqual(wikipedia.response(response), [])
  59. # normal case
  60. json = """
  61. {
  62. "batchcomplete": "",
  63. "query": {
  64. "normalized": [],
  65. "pages": {
  66. "12345": {
  67. "pageid": 12345,
  68. "ns": 0,
  69. "title": "The Title",
  70. "extract": "The Title is...",
  71. "thumbnail": {
  72. "source": "img_src.jpg"
  73. },
  74. "pageimage": "img_name.jpg"
  75. }
  76. }
  77. }
  78. }"""
  79. response = mock.Mock(text=json, search_params=dicto)
  80. results = wikipedia.response(response)
  81. self.assertEqual(type(results), list)
  82. self.assertEqual(len(results), 2)
  83. self.assertEqual(results[0]['title'], u'The Title')
  84. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[0]['url'])
  85. self.assertEqual(results[1]['infobox'], u'The Title')
  86. self.assertIn('fr.wikipedia.org/wiki/The_Title', results[1]['id'])
  87. self.assertIn('The Title is...', results[1]['content'])
  88. self.assertEqual(results[1]['img_src'], 'img_src.jpg')
  89. # disambiguation page
  90. json = """
  91. {
  92. "batchcomplete": "",
  93. "query": {
  94. "normalized": [],
  95. "pages": {
  96. "12345": {
  97. "pageid": 12345,
  98. "ns": 0,
  99. "title": "The Title",
  100. "extract": "The Title can be:\\nThe Title 1\\nThe Title 2\\nThe Title 3\\nThe Title 4......................................................................................................................................." """ # noqa
  101. json += """
  102. }
  103. }
  104. }
  105. }"""
  106. response = mock.Mock(text=json, search_params=dicto)
  107. results = wikipedia.response(response)
  108. self.assertEqual(type(results), list)
  109. self.assertEqual(len(results), 2)
  110. # no image
  111. json = """
  112. {
  113. "batchcomplete": "",
  114. "query": {
  115. "normalized": [],
  116. "pages": {
  117. "12345": {
  118. "pageid": 12345,
  119. "ns": 0,
  120. "title": "The Title",
  121. "extract": "The Title is......................................................................................................................................................................................." """ # noqa
  122. json += """
  123. }
  124. }
  125. }
  126. }"""
  127. response = mock.Mock(text=json, search_params=dicto)
  128. results = wikipedia.response(response)
  129. self.assertEqual(type(results), list)
  130. self.assertEqual(len(results), 2)
  131. self.assertIn('The Title is...', results[1]['content'])
  132. self.assertEqual(results[1]['img_src'], None)
  133. # title not in first paragraph
  134. json = u"""
  135. {
  136. "batchcomplete": "",
  137. "query": {
  138. "normalized": [],
  139. "pages": {
  140. "12345": {
  141. "pageid": 12345,
  142. "ns": 0,
  143. "title": "披頭四樂隊",
  144. "extract": "披头士乐队....................................................................................................................................................................................................\\n披頭四樂隊...", """ # noqa
  145. json += """
  146. "thumbnail": {
  147. "source": "img_src.jpg"
  148. },
  149. "pageimage": "img_name.jpg"
  150. }
  151. }
  152. }
  153. }"""
  154. response = mock.Mock(text=json, search_params=dicto)
  155. results = wikipedia.response(response)
  156. self.assertEqual(type(results), list)
  157. self.assertEqual(len(results), 2)
  158. self.assertEqual(results[1]['infobox'], u'披頭四樂隊')
  159. self.assertIn(u'披头士乐队...', results[1]['content'])
  160. def test_fetch_supported_languages(self):
  161. html = u"""<html></html>"""
  162. response = mock.Mock(text=html)
  163. languages = wikipedia._fetch_supported_languages(response)
  164. self.assertEqual(type(languages), dict)
  165. self.assertEqual(len(languages), 0)
  166. html = u"""
  167. <html>
  168. <body>
  169. <div>
  170. <div>
  171. <h3>Table header</h3>
  172. <table class="sortable jquery-tablesorter">
  173. <thead>
  174. <tr>
  175. <th>N</th>
  176. <th>Language</th>
  177. <th>Language (local)</th>
  178. <th>Wiki</th>
  179. <th>Articles</th>
  180. </tr>
  181. </thead>
  182. <tbody>
  183. <tr>
  184. <td>2</td>
  185. <td><a>Swedish</a></td>
  186. <td><a>Svenska</a></td>
  187. <td><a>sv</a></td>
  188. <td><a><b>3000000</b></a></td>
  189. </tr>
  190. <tr>
  191. <td>3</td>
  192. <td><a>Cebuano</a></td>
  193. <td><a>Sinugboanong Binisaya</a></td>
  194. <td><a>ceb</a></td>
  195. <td><a><b>3000000</b></a></td>
  196. </tr>
  197. </tbody>
  198. </table>
  199. <h3>Table header</h3>
  200. <table class="sortable jquery-tablesorter">
  201. <thead>
  202. <tr>
  203. <th>N</th>
  204. <th>Language</th>
  205. <th>Language (local)</th>
  206. <th>Wiki</th>
  207. <th>Articles</th>
  208. </tr>
  209. </thead>
  210. <tbody>
  211. <tr>
  212. <td>2</td>
  213. <td><a>Norwegian (Bokmål)</a></td>
  214. <td><a>Norsk (Bokmål)</a></td>
  215. <td><a>no</a></td>
  216. <td><a><b>100000</b></a></td>
  217. </tr>
  218. </tbody>
  219. </table>
  220. </div>
  221. </div>
  222. </body>
  223. </html>
  224. """
  225. response = mock.Mock(text=html)
  226. languages = wikipedia._fetch_supported_languages(response)
  227. self.assertEqual(type(languages), dict)
  228. self.assertEqual(len(languages), 3)
  229. self.assertIn('sv', languages)
  230. self.assertIn('ceb', languages)
  231. self.assertIn('no', languages)
  232. self.assertEqual(type(languages['sv']), dict)
  233. self.assertEqual(type(languages['ceb']), dict)
  234. self.assertEqual(type(languages['no']), dict)
  235. self.assertIn('name', languages['sv'])
  236. self.assertIn('english_name', languages['sv'])
  237. self.assertIn('articles', languages['sv'])
  238. self.assertEqual(languages['sv']['name'], 'Svenska')
  239. self.assertEqual(languages['sv']['english_name'], 'Swedish')
  240. self.assertEqual(languages['sv']['articles'], 3000000)
  241. self.assertEqual(languages['ceb']['name'], 'Sinugboanong Binisaya')
  242. self.assertEqual(languages['ceb']['english_name'], 'Cebuano')
  243. self.assertEqual(languages['ceb']['articles'], 3000000)
  244. self.assertEqual(languages['no']['name'], u'Norsk (Bokmål)')
  245. self.assertEqual(languages['no']['english_name'], u'Norwegian (Bokmål)')
  246. self.assertEqual(languages['no']['articles'], 100000)