webadapter.py 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162
  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
  5. # remove duplicate queries.
  6. # FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
  7. def deduplicate_query_engines(query_engines):
  8. uniq_query_engines = {q["category"] + '|' + q["name"]: q for q in query_engines}
  9. return uniq_query_engines.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_engines = raw_text_query.engines
  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_engines and raw_text_query.specific:
  73. additional_categories = set()
  74. for engine in query_engines:
  75. if 'from_bang' in engine and engine['from_bang']:
  76. additional_categories.add('none')
  77. else:
  78. additional_categories.add(engine['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 = [{'category': engines[engine].categories[0],
  90. 'name': engine}
  91. for engine in map(str.strip, pd.split(',')) if engine in engines]
  92. if pd_engines:
  93. query_engines.extend(pd_engines)
  94. load_default_categories = False
  95. elif pd_name.startswith('category_'):
  96. category = pd_name[9:]
  97. # if category is not found in list, skip
  98. if category not in categories:
  99. continue
  100. if pd != 'off':
  101. # add category to list
  102. query_categories.append(category)
  103. elif category in query_categories:
  104. # remove category from list if property is set to 'off'
  105. query_categories.remove(category)
  106. if not load_default_categories:
  107. if not query_categories:
  108. query_categories = list(set(engine['category']
  109. for engine in query_engines))
  110. else:
  111. # if no category is specified for this search,
  112. # using user-defined default-configuration which
  113. # (is stored in cookie)
  114. if not query_categories:
  115. cookie_categories = preferences.get_value('categories')
  116. for ccateg in cookie_categories:
  117. if ccateg in categories:
  118. query_categories.append(ccateg)
  119. # if still no category is specified, using general
  120. # as default-category
  121. if not query_categories:
  122. query_categories = ['general']
  123. # using all engines for that search, which are
  124. # declared under the specific categories
  125. for categ in query_categories:
  126. query_engines.extend({'category': categ,
  127. 'name': engine.name}
  128. for engine in categories[categ]
  129. if (engine.name, categ) not in disabled_engines)
  130. query_engines = deduplicate_query_engines(query_engines)
  131. external_bang = raw_text_query.external_bang
  132. return (SearchQuery(query, query_engines, query_categories,
  133. query_lang, query_safesearch, query_pageno,
  134. query_time_range, query_timeout, preferences,
  135. external_bang=external_bang),
  136. raw_text_query)