logo

searx

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

gigablast.py (2947B)


  1. """
  2. Gigablast (Web)
  3. @website https://gigablast.com
  4. @provide-api yes (https://gigablast.com/api.html)
  5. @using-api yes
  6. @results XML
  7. @stable yes
  8. @parse url, title, content
  9. """
  10. import random
  11. from json import loads
  12. from time import time
  13. from lxml.html import fromstring
  14. from searx.url_utils import urlencode
  15. # engine dependent config
  16. categories = ['general']
  17. paging = True
  18. number_of_results = 10
  19. language_support = True
  20. safesearch = True
  21. # search-url
  22. base_url = 'https://gigablast.com/'
  23. search_string = 'search?{query}'\
  24. '&n={number_of_results}'\
  25. '&c=main'\
  26. '&s={offset}'\
  27. '&format=json'\
  28. '&qh=0'\
  29. '&qlang={lang}'\
  30. '&ff={safesearch}'\
  31. '&rxiec={rxieu}'\
  32. '&ulse={ulse}'\
  33. '&rand={rxikd}' # current unix timestamp
  34. # specific xpath variables
  35. results_xpath = '//response//result'
  36. url_xpath = './/url'
  37. title_xpath = './/title'
  38. content_xpath = './/sum'
  39. supported_languages_url = 'https://gigablast.com/search?&rxikd=1'
  40. # do search-request
  41. def request(query, params):
  42. offset = (params['pageno'] - 1) * number_of_results
  43. language = params['language'].replace('-', '_').lower()
  44. if language.split('-')[0] != 'zh':
  45. language = language.split('-')[0]
  46. if params['safesearch'] >= 1:
  47. safesearch = 1
  48. else:
  49. safesearch = 0
  50. # rxieu is some kind of hash from the search query, but accepts random atm
  51. search_path = search_string.format(query=urlencode({'q': query}),
  52. offset=offset,
  53. number_of_results=number_of_results,
  54. rxikd=int(time() * 1000),
  55. rxieu=random.randint(1000000000, 9999999999),
  56. ulse=random.randint(100000000, 999999999),
  57. lang=language,
  58. safesearch=safesearch)
  59. params['url'] = base_url + search_path
  60. return params
  61. # get response from search-request
  62. def response(resp):
  63. results = []
  64. # parse results
  65. response_json = loads(resp.text)
  66. for result in response_json['results']:
  67. # append result
  68. results.append({'url': result['url'],
  69. 'title': result['title'],
  70. 'content': result['sum']})
  71. # return results
  72. return results
  73. # get supported languages from their site
  74. def _fetch_supported_languages(resp):
  75. supported_languages = []
  76. dom = fromstring(resp.text)
  77. links = dom.xpath('//span[@id="menu2"]/a')
  78. for link in links:
  79. href = link.xpath('./@href')[0].split('lang%3A')
  80. if len(href) == 2:
  81. code = href[1].split('_')
  82. if len(code) == 2:
  83. code = code[0] + '-' + code[1].upper()
  84. else:
  85. code = code[0]
  86. supported_languages.append(code)
  87. return supported_languages