webadapter.py 6.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160
  1. from searx.exceptions import SearxParameterException
  2. from searx.query import RawTextQuery, VALID_LANGUAGE_CODE
  3. from searx.engines import categories, engines
  4. from searx.search import SearchQuery, EngineRef
  5. # remove duplicate queries.
  6. # FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
  7. def deduplicate_engineref_list(engineref_list):
  8. engineref_dict = {q.category + '|' + q.name: q for q in engineref_list}
  9. return engineref_dict.values()
  10. def get_search_query_from_webapp(preferences, form):
  11. # no text for the query ?
  12. if not form.get('q'):
  13. raise SearxParameterException('q', '')
  14. # set blocked engines
  15. disabled_engines = preferences.engines.get_disabled()
  16. # parse query, if tags are set, which change
  17. # the serch engine or search-language
  18. raw_text_query = RawTextQuery(form['q'], disabled_engines)
  19. # set query
  20. query = raw_text_query.getQuery()
  21. # get and check page number
  22. pageno_param = form.get('pageno', '1')
  23. if not pageno_param.isdigit() or int(pageno_param) < 1:
  24. raise SearxParameterException('pageno', pageno_param)
  25. query_pageno = int(pageno_param)
  26. # get language
  27. # set specific language if set on request, query or preferences
  28. # TODO support search with multible languages
  29. if len(raw_text_query.languages):
  30. query_lang = raw_text_query.languages[-1]
  31. elif 'language' in form:
  32. query_lang = form.get('language')
  33. else:
  34. query_lang = preferences.get_value('language')
  35. # check language
  36. if not VALID_LANGUAGE_CODE.match(query_lang):
  37. raise SearxParameterException('language', query_lang)
  38. # get safesearch
  39. if 'safesearch' in form:
  40. query_safesearch = form.get('safesearch')
  41. # first check safesearch
  42. if not query_safesearch.isdigit():
  43. raise SearxParameterException('safesearch', query_safesearch)
  44. query_safesearch = int(query_safesearch)
  45. else:
  46. query_safesearch = preferences.get_value('safesearch')
  47. # safesearch : second check
  48. if query_safesearch < 0 or query_safesearch > 2:
  49. raise SearxParameterException('safesearch', query_safesearch)
  50. # get time_range
  51. query_time_range = form.get('time_range')
  52. # check time_range
  53. if query_time_range not in ('None', None, '', 'day', 'week', 'month', 'year'):
  54. raise SearxParameterException('time_range', query_time_range)
  55. # query_engines
  56. query_engineref_list = raw_text_query.enginerefs
  57. # timeout_limit
  58. query_timeout = raw_text_query.timeout_limit
  59. if query_timeout is None and 'timeout_limit' in form:
  60. raw_time_limit = form.get('timeout_limit')
  61. if raw_time_limit in ['None', '']:
  62. raw_time_limit = None
  63. else:
  64. try:
  65. query_timeout = float(raw_time_limit)
  66. except ValueError:
  67. raise SearxParameterException('timeout_limit', raw_time_limit)
  68. # query_categories
  69. query_categories = []
  70. # if engines are calculated from query,
  71. # set categories by using that informations
  72. if query_engineref_list and raw_text_query.specific:
  73. additional_categories = set()
  74. for engineref in query_engineref_list:
  75. if engineref.from_bang:
  76. additional_categories.add('none')
  77. else:
  78. additional_categories.add(engineref.category)
  79. query_categories = list(additional_categories)
  80. # otherwise, using defined categories to
  81. # calculate which engines should be used
  82. else:
  83. # set categories/engines
  84. load_default_categories = True
  85. for pd_name, pd in form.items():
  86. if pd_name == 'categories':
  87. query_categories.extend(categ for categ in map(str.strip, pd.split(',')) if categ in categories)
  88. elif pd_name == 'engines':
  89. pd_engines = [EngineRef(engineref, engines[engineref].categories[0])
  90. for engine in map(str.strip, pd.split(',')) if engine in engines]
  91. if pd_engines:
  92. query_engineref_list.extend(pd_engines)
  93. load_default_categories = False
  94. elif pd_name.startswith('category_'):
  95. category = pd_name[9:]
  96. # if category is not found in list, skip
  97. if category not in categories:
  98. continue
  99. if pd != 'off':
  100. # add category to list
  101. query_categories.append(category)
  102. elif category in query_categories:
  103. # remove category from list if property is set to 'off'
  104. query_categories.remove(category)
  105. if not load_default_categories:
  106. if not query_categories:
  107. query_categories = list(set(engine['category']
  108. for engine in query_engineref_list))
  109. else:
  110. # if no category is specified for this search,
  111. # using user-defined default-configuration which
  112. # (is stored in cookie)
  113. if not query_categories:
  114. cookie_categories = preferences.get_value('categories')
  115. for ccateg in cookie_categories:
  116. if ccateg in categories:
  117. query_categories.append(ccateg)
  118. # if still no category is specified, using general
  119. # as default-category
  120. if not query_categories:
  121. query_categories = ['general']
  122. # using all engines for that search, which are
  123. # declared under the specific categories
  124. for categ in query_categories:
  125. query_engineref_list.extend(EngineRef(engine.name, categ)
  126. for engine in categories[categ]
  127. if (engine.name, categ) not in disabled_engines)
  128. query_engineref_list = deduplicate_engineref_list(query_engineref_list)
  129. external_bang = raw_text_query.external_bang
  130. return (SearchQuery(query, query_engineref_list, query_categories,
  131. query_lang, query_safesearch, query_pageno,
  132. query_time_range, query_timeout, preferences,
  133. external_bang=external_bang),
  134. raw_text_query)