test_wikidata.py 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502
  1. # -*- coding: utf-8 -*-
  2. from json import loads
  3. from lxml.html import fromstring
  4. from collections import defaultdict
  5. import mock
  6. from searx.engines import wikidata
  7. from searx.testing import SearxTestCase
  8. class TestWikidataEngine(SearxTestCase):
  9. def test_request(self):
  10. query = 'test_query'
  11. dicto = defaultdict(dict)
  12. dicto['language'] = 'all'
  13. params = wikidata.request(query, dicto)
  14. self.assertIn('url', params)
  15. self.assertIn(query, params['url'])
  16. self.assertIn('wikidata.org', params['url'])
  17. self.assertIn('en', params['url'])
  18. dicto['language'] = 'es_ES'
  19. params = wikidata.request(query, dicto)
  20. self.assertIn(query, params['url'])
  21. self.assertIn('es', params['url'])
  22. # successful cases are not tested here to avoid sending additional requests
  23. def test_response(self):
  24. self.assertRaises(AttributeError, wikidata.response, None)
  25. self.assertRaises(AttributeError, wikidata.response, [])
  26. self.assertRaises(AttributeError, wikidata.response, '')
  27. self.assertRaises(AttributeError, wikidata.response, '[]')
  28. response = mock.Mock(content='<html></html>', search_params={"language": "all"})
  29. self.assertEqual(wikidata.response(response), [])
  30. def test_getDetail(self):
  31. response = {}
  32. results = wikidata.getDetail(response, "Q123", "en", "en-US")
  33. self.assertEqual(results, [])
  34. title_html = '<div><div class="wikibase-title-label">Test</div></div>'
  35. html = """
  36. <div>
  37. <div class="wikibase-entitytermsview-heading-description">
  38. </div>
  39. <div>
  40. <ul class="wikibase-sitelinklistview-listview">
  41. <li data-wb-siteid="enwiki"><a href="http://en.wikipedia.org/wiki/Test">Test</a></li>
  42. </ul>
  43. </div>
  44. </div>
  45. """
  46. response = {"parse": {"displaytitle": title_html, "text": html}}
  47. results = wikidata.getDetail(response, "Q123", "en", "en-US")
  48. self.assertEqual(len(results), 1)
  49. self.assertEqual(results[0]['url'], 'https://en.wikipedia.org/wiki/Test')
  50. title_html = """
  51. <div>
  52. <div class="wikibase-title-label">
  53. <span lang="en">Test</span>
  54. <sup class="wb-language-fallback-indicator">English</sup>
  55. </div>
  56. </div>
  57. """
  58. html = """
  59. <div>
  60. <div class="wikibase-entitytermsview-heading-description">
  61. <span lang="en">Description</span>
  62. <sup class="wb-language-fallback-indicator">English</sup>
  63. </div>
  64. <div id="P856">
  65. <div class="wikibase-statementgroupview-property-label">
  66. <a href="/wiki/Property:P856">
  67. <span lang="en">official website</span>
  68. <sup class="wb-language-fallback-indicator">English</sup>
  69. </a>
  70. </div>
  71. <div class="wikibase-statementview-mainsnak">
  72. <a class="external free" href="https://officialsite.com">
  73. https://officialsite.com
  74. </a>
  75. </div>
  76. </div>
  77. <div>
  78. <ul class="wikibase-sitelinklistview-listview">
  79. <li data-wb-siteid="enwiki"><a href="http://en.wikipedia.org/wiki/Test">Test</a></li>
  80. </ul>
  81. </div>
  82. </div>
  83. """
  84. response = {"parse": {"displaytitle": title_html, "text": html}}
  85. results = wikidata.getDetail(response, "Q123", "yua", "yua_MX")
  86. self.assertEqual(len(results), 2)
  87. self.assertEqual(results[0]['title'], 'official website')
  88. self.assertEqual(results[0]['url'], 'https://officialsite.com')
  89. self.assertEqual(results[1]['infobox'], 'Test')
  90. self.assertEqual(results[1]['id'], None)
  91. self.assertEqual(results[1]['content'], 'Description')
  92. self.assertEqual(results[1]['attributes'], [])
  93. self.assertEqual(results[1]['urls'][0]['title'], 'official website')
  94. self.assertEqual(results[1]['urls'][0]['url'], 'https://officialsite.com')
  95. self.assertEqual(results[1]['urls'][1]['title'], 'Wikipedia (en)')
  96. self.assertEqual(results[1]['urls'][1]['url'], 'https://en.wikipedia.org/wiki/Test')
  97. def test_add_image(self):
  98. image_src = wikidata.add_image(fromstring("<div></div>"))
  99. self.assertEqual(image_src, None)
  100. html = u"""
  101. <div>
  102. <div id="P18">
  103. <div class="wikibase-statementgroupview-property-label">
  104. <a href="/wiki/Property:P18">
  105. image
  106. </a>
  107. </div>
  108. <div class="wikibase-statementlistview">
  109. <div class="wikibase-statementview listview-item">
  110. <div class="wikibase-statementview-rankselector">
  111. <span class="wikibase-rankselector-normal"></span>
  112. </div>
  113. <div class="wikibase-statementview-mainsnak">
  114. <div>
  115. <div class="wikibase-snakview-value">
  116. <a href="https://commons.wikimedia.org/wiki/File:image.png">
  117. image.png
  118. </a>
  119. </div>
  120. </div>
  121. </div>
  122. </div>
  123. </div>
  124. </div>
  125. </div>
  126. """
  127. html_etree = fromstring(html)
  128. image_src = wikidata.add_image(html_etree)
  129. self.assertEqual(image_src, "https://commons.wikimedia.org/wiki/Special:FilePath/image.png?width=500")
  130. html = u"""
  131. <div>
  132. <div id="P2910">
  133. <div class="wikibase-statementgroupview-property-label">
  134. <a href="/wiki/Property:P2910">
  135. icon
  136. </a>
  137. </div>
  138. <div class="wikibase-statementlistview">
  139. <div class="wikibase-statementview listview-item">
  140. <div class="wikibase-statementview-rankselector">
  141. <span class="wikibase-rankselector-normal"></span>
  142. </div>
  143. <div class="wikibase-statementview-mainsnak">
  144. <div>
  145. <div class="wikibase-snakview-value">
  146. <a href="https://commons.wikimedia.org/wiki/File:icon.png">
  147. icon.png
  148. </a>
  149. </div>
  150. </div>
  151. </div>
  152. </div>
  153. </div>
  154. </div>
  155. <div id="P154">
  156. <div class="wikibase-statementgroupview-property-label">
  157. <a href="/wiki/Property:P154">
  158. logo
  159. </a>
  160. </div>
  161. <div class="wikibase-statementlistview">
  162. <div class="wikibase-statementview listview-item">
  163. <div class="wikibase-statementview-rankselector">
  164. <span class="wikibase-rankselector-normal"></span>
  165. </div>
  166. <div class="wikibase-statementview-mainsnak">
  167. <div>
  168. <div class="wikibase-snakview-value">
  169. <a href="https://commons.wikimedia.org/wiki/File:logo.png">
  170. logo.png
  171. </a>
  172. </div>
  173. </div>
  174. </div>
  175. </div>
  176. </div>
  177. </div>
  178. </div>
  179. """
  180. html_etree = fromstring(html)
  181. image_src = wikidata.add_image(html_etree)
  182. self.assertEqual(image_src, "https://commons.wikimedia.org/wiki/Special:FilePath/logo.png?width=500")
  183. def test_add_attribute(self):
  184. html = u"""
  185. <div>
  186. <div id="P27">
  187. <div class="wikibase-statementgroupview-property-label">
  188. <a href="/wiki/Property:P27">
  189. country of citizenship
  190. </a>
  191. </div>
  192. <div class="wikibase-statementlistview">
  193. <div class="wikibase-statementview listview-item">
  194. <div class="wikibase-statementview-rankselector">
  195. <span class="wikibase-rankselector-normal"></span>
  196. </div>
  197. <div class="wikibase-statementview-mainsnak">
  198. <div>
  199. <div class="wikibase-snakview-value">
  200. <a href="/wiki/Q145">
  201. United Kingdom
  202. </a>
  203. </div>
  204. </div>
  205. </div>
  206. </div>
  207. </div>
  208. </div>
  209. </div>
  210. """
  211. attributes = []
  212. html_etree = fromstring(html)
  213. wikidata.add_attribute(attributes, html_etree, "Fail")
  214. self.assertEqual(attributes, [])
  215. wikidata.add_attribute(attributes, html_etree, "P27")
  216. self.assertEqual(len(attributes), 1)
  217. self.assertEqual(attributes[0]["label"], "country of citizenship")
  218. self.assertEqual(attributes[0]["value"], "United Kingdom")
  219. html = u"""
  220. <div>
  221. <div id="P569">
  222. <div class="wikibase-statementgroupview-property-label">
  223. <a href="/wiki/Property:P569">
  224. date of birth
  225. </a>
  226. </div>
  227. <div class="wikibase-statementlistview">
  228. <div class="wikibase-statementview listview-item">
  229. <div class="wikibase-statementview-rankselector">
  230. <span class="wikibase-rankselector-normal"></span>
  231. </div>
  232. <div class="wikibase-statementview-mainsnak">
  233. <div>
  234. <div class="wikibase-snakview-value">
  235. 27 January 1832
  236. <sup class="wb-calendar-name">
  237. Gregorian
  238. </sup>
  239. </div>
  240. </div>
  241. </div>
  242. </div>
  243. </div>
  244. </div>
  245. </div>
  246. """
  247. attributes = []
  248. html_etree = fromstring(html)
  249. wikidata.add_attribute(attributes, html_etree, "P569", date=True)
  250. self.assertEqual(len(attributes), 1)
  251. self.assertEqual(attributes[0]["label"], "date of birth")
  252. self.assertEqual(attributes[0]["value"], "27 January 1832")
  253. html = u"""
  254. <div>
  255. <div id="P6">
  256. <div class="wikibase-statementgroupview-property-label">
  257. <a href="/wiki/Property:P27">
  258. head of government
  259. </a>
  260. </div>
  261. <div class="wikibase-statementlistview">
  262. <div class="wikibase-statementview listview-item">
  263. <div class="wikibase-statementview-rankselector">
  264. <span class="wikibase-rankselector-normal"></span>
  265. </div>
  266. <div class="wikibase-statementview-mainsnak">
  267. <div>
  268. <div class="wikibase-snakview-value">
  269. <a href="/wiki/Q206">
  270. Old Prime Minister
  271. </a>
  272. </div>
  273. </div>
  274. </div>
  275. </div>
  276. <div class="wikibase-statementview listview-item">
  277. <div class="wikibase-statementview-rankselector">
  278. <span class="wikibase-rankselector-preferred"></span>
  279. </div>
  280. <div class="wikibase-statementview-mainsnak">
  281. <div>
  282. <div class="wikibase-snakview-value">
  283. <a href="/wiki/Q3099714">
  284. Actual Prime Minister
  285. </a>
  286. </div>
  287. </div>
  288. </div>
  289. </div>
  290. </div>
  291. </div>
  292. </div>
  293. """
  294. attributes = []
  295. html_etree = fromstring(html)
  296. wikidata.add_attribute(attributes, html_etree, "P6")
  297. self.assertEqual(len(attributes), 1)
  298. self.assertEqual(attributes[0]["label"], "head of government")
  299. self.assertEqual(attributes[0]["value"], "Old Prime Minister, Actual Prime Minister")
  300. attributes = []
  301. html_etree = fromstring(html)
  302. wikidata.add_attribute(attributes, html_etree, "P6", trim=True)
  303. self.assertEqual(len(attributes), 1)
  304. self.assertEqual(attributes[0]["value"], "Actual Prime Minister")
  305. def test_add_url(self):
  306. html = u"""
  307. <div>
  308. <div id="P856">
  309. <div class="wikibase-statementgroupview-property-label">
  310. <a href="/wiki/Property:P856">
  311. official website
  312. </a>
  313. </div>
  314. <div class="wikibase-statementlistview">
  315. <div class="wikibase-statementview listview-item">
  316. <div class="wikibase-statementview-mainsnak">
  317. <div>
  318. <div class="wikibase-snakview-value">
  319. <a class="external free" href="https://searx.me">
  320. https://searx.me/
  321. </a>
  322. </div>
  323. </div>
  324. </div>
  325. </div>
  326. </div>
  327. </div>
  328. </div>
  329. """
  330. urls = []
  331. html_etree = fromstring(html)
  332. wikidata.add_url(urls, html_etree, 'P856')
  333. self.assertEquals(len(urls), 1)
  334. self.assertIn({'title': 'official website', 'url': 'https://searx.me/'}, urls)
  335. urls = []
  336. results = []
  337. wikidata.add_url(urls, html_etree, 'P856', 'custom label', results=results)
  338. self.assertEquals(len(urls), 1)
  339. self.assertEquals(len(results), 1)
  340. self.assertIn({'title': 'custom label', 'url': 'https://searx.me/'}, urls)
  341. self.assertIn({'title': 'custom label', 'url': 'https://searx.me/'}, results)
  342. html = u"""
  343. <div>
  344. <div id="P856">
  345. <div class="wikibase-statementgroupview-property-label">
  346. <a href="/wiki/Property:P856">
  347. official website
  348. </a>
  349. </div>
  350. <div class="wikibase-statementlistview">
  351. <div class="wikibase-statementview listview-item">
  352. <div class="wikibase-statementview-mainsnak">
  353. <div>
  354. <div class="wikibase-snakview-value">
  355. <a class="external free" href="http://www.worldofwarcraft.com">
  356. http://www.worldofwarcraft.com
  357. </a>
  358. </div>
  359. </div>
  360. </div>
  361. </div>
  362. <div class="wikibase-statementview listview-item">
  363. <div class="wikibase-statementview-mainsnak">
  364. <div>
  365. <div class="wikibase-snakview-value">
  366. <a class="external free" href="http://eu.battle.net/wow/en/">
  367. http://eu.battle.net/wow/en/
  368. </a>
  369. </div>
  370. </div>
  371. </div>
  372. </div>
  373. </div>
  374. </div>
  375. </div>
  376. """
  377. urls = []
  378. html_etree = fromstring(html)
  379. wikidata.add_url(urls, html_etree, 'P856')
  380. self.assertEquals(len(urls), 2)
  381. self.assertIn({'title': 'official website', 'url': 'http://www.worldofwarcraft.com'}, urls)
  382. self.assertIn({'title': 'official website', 'url': 'http://eu.battle.net/wow/en/'}, urls)
  383. def test_get_imdblink(self):
  384. html = u"""
  385. <div>
  386. <div class="wikibase-statementview-mainsnak">
  387. <div>
  388. <div class="wikibase-snakview-value">
  389. <a class="wb-external-id" href="http://www.imdb.com/tt0433664">
  390. tt0433664
  391. </a>
  392. </div>
  393. </div>
  394. </div>
  395. </div>
  396. """
  397. html_etree = fromstring(html)
  398. imdblink = wikidata.get_imdblink(html_etree, 'https://www.imdb.com/')
  399. html = u"""
  400. <div>
  401. <div class="wikibase-statementview-mainsnak">
  402. <div>
  403. <div class="wikibase-snakview-value">
  404. <a class="wb-external-id"
  405. href="href="http://tools.wmflabs.org/...http://www.imdb.com/&id=nm4915994"">
  406. nm4915994
  407. </a>
  408. </div>
  409. </div>
  410. </div>
  411. </div>
  412. """
  413. html_etree = fromstring(html)
  414. imdblink = wikidata.get_imdblink(html_etree, 'https://www.imdb.com/')
  415. self.assertIn('https://www.imdb.com/name/nm4915994', imdblink)
  416. def test_get_geolink(self):
  417. html = u"""
  418. <div>
  419. <div class="wikibase-statementview-mainsnak">
  420. <div>
  421. <div class="wikibase-snakview-value">
  422. 60°N, 40°E
  423. </div>
  424. </div>
  425. </div>
  426. </div>
  427. """
  428. html_etree = fromstring(html)
  429. geolink = wikidata.get_geolink(html_etree)
  430. self.assertIn('https://www.openstreetmap.org/', geolink)
  431. self.assertIn('lat=60&lon=40', geolink)
  432. html = u"""
  433. <div>
  434. <div class="wikibase-statementview-mainsnak">
  435. <div>
  436. <div class="wikibase-snakview-value">
  437. 34°35'59"S, 58°22'55"W
  438. </div>
  439. </div>
  440. </div>
  441. </div>
  442. """
  443. html_etree = fromstring(html)
  444. geolink = wikidata.get_geolink(html_etree)
  445. self.assertIn('https://www.openstreetmap.org/', geolink)
  446. self.assertIn('lat=-34.59', geolink)
  447. self.assertIn('lon=-58.38', geolink)
  448. def test_get_wikilink(self):
  449. html = """
  450. <div>
  451. <div>
  452. <ul class="wikibase-sitelinklistview-listview">
  453. <li data-wb-siteid="arwiki"><a href="http://ar.wikipedia.org/wiki/Test">Test</a></li>
  454. <li data-wb-siteid="enwiki"><a href="http://en.wikipedia.org/wiki/Test">Test</a></li>
  455. </ul>
  456. </div>
  457. <div>
  458. <ul class="wikibase-sitelinklistview-listview">
  459. <li data-wb-siteid="enwikiquote"><a href="https://en.wikiquote.org/wiki/Test">Test</a></li>
  460. </ul>
  461. </div>
  462. </div>
  463. """
  464. html_etree = fromstring(html)
  465. wikilink = wikidata.get_wikilink(html_etree, 'nowiki')
  466. self.assertEqual(wikilink, None)
  467. wikilink = wikidata.get_wikilink(html_etree, 'enwiki')
  468. self.assertEqual(wikilink, 'https://en.wikipedia.org/wiki/Test')
  469. wikilink = wikidata.get_wikilink(html_etree, 'arwiki')
  470. self.assertEqual(wikilink, 'https://ar.wikipedia.org/wiki/Test')
  471. wikilink = wikidata.get_wikilink(html_etree, 'enwikiquote')
  472. self.assertEqual(wikilink, 'https://en.wikiquote.org/wiki/Test')