logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git

mediawiki.py (2352B)


  1. """
  2. general mediawiki-engine (Web)
  3. @website websites built on mediawiki (https://www.mediawiki.org)
  4. @provide-api yes (http://www.mediawiki.org/wiki/API:Search)
  5. @using-api yes
  6. @results JSON
  7. @stable yes
  8. @parse url, title
  9. @todo content
  10. """
  11. from json import loads
  12. from string import Formatter
  13. from searx.url_utils import urlencode, quote
  14. # engine dependent config
  15. categories = ['general']
  16. language_support = True
  17. paging = True
  18. number_of_results = 1
  19. search_type = 'nearmatch' # possible values: title, text, nearmatch
  20. # search-url
  21. base_url = 'https://{language}.wikipedia.org/'
  22. search_postfix = 'w/api.php?action=query'\
  23. '&list=search'\
  24. '&{query}'\
  25. '&format=json'\
  26. '&sroffset={offset}'\
  27. '&srlimit={limit}'\
  28. '&srwhat={searchtype}'
  29. # do search-request
  30. def request(query, params):
  31. offset = (params['pageno'] - 1) * number_of_results
  32. string_args = dict(query=urlencode({'srsearch': query}),
  33. offset=offset,
  34. limit=number_of_results,
  35. searchtype=search_type)
  36. format_strings = list(Formatter().parse(base_url))
  37. language = params['language'].split('-')[0]
  38. # format_string [('https://', 'language', '', None), ('.wikipedia.org/', None, None, None)]
  39. if any(x[1] == 'language' for x in format_strings):
  40. string_args['language'] = language
  41. # write search-language back to params, required in response
  42. params['language'] = language
  43. search_url = base_url + search_postfix
  44. params['url'] = search_url.format(**string_args)
  45. return params
  46. # get response from search-request
  47. def response(resp):
  48. results = []
  49. search_results = loads(resp.text)
  50. # return empty array if there are no results
  51. if not search_results.get('query', {}).get('search'):
  52. return []
  53. # parse results
  54. for result in search_results['query']['search']:
  55. if result.get('snippet', '').startswith('#REDIRECT'):
  56. continue
  57. url = base_url.format(language=resp.search_params['language']) +\
  58. 'wiki/' + quote(result['title'].replace(' ', '_').encode('utf-8'))
  59. # append result
  60. results.append({'url': url,
  61. 'title': result['title'],
  62. 'content': ''})
  63. # return results
  64. return results