query.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. # pylint: disable=invalid-name, missing-module-docstring, missing-class-docstring
  3. from abc import abstractmethod, ABC
  4. import re
  5. from searx import settings
  6. from searx.sxng_locales import sxng_locales
  7. from searx.engines import categories, engines, engine_shortcuts
  8. from searx.external_bang import get_bang_definition_and_autocomplete
  9. from searx.search import EngineRef
  10. from searx.webutils import VALID_LANGUAGE_CODE
  11. class QueryPartParser(ABC):
  12. __slots__ = "raw_text_query", "enable_autocomplete"
  13. @staticmethod
  14. @abstractmethod
  15. def check(raw_value):
  16. """Check if raw_value can be parsed"""
  17. def __init__(self, raw_text_query, enable_autocomplete):
  18. self.raw_text_query = raw_text_query
  19. self.enable_autocomplete = enable_autocomplete
  20. @abstractmethod
  21. def __call__(self, raw_value):
  22. """Try to parse raw_value: set the self.raw_text_query properties
  23. return True if raw_value has been parsed
  24. self.raw_text_query.autocomplete_list is also modified
  25. if self.enable_autocomplete is True
  26. """
  27. def _add_autocomplete(self, value):
  28. if value not in self.raw_text_query.autocomplete_list:
  29. self.raw_text_query.autocomplete_list.append(value)
  30. class TimeoutParser(QueryPartParser):
  31. @staticmethod
  32. def check(raw_value):
  33. return raw_value[0] == '<'
  34. def __call__(self, raw_value):
  35. value = raw_value[1:]
  36. found = self._parse(value) if len(value) > 0 else False
  37. if self.enable_autocomplete and not value:
  38. self._autocomplete()
  39. return found
  40. def _parse(self, value):
  41. if not value.isdigit():
  42. return False
  43. raw_timeout_limit = int(value)
  44. if raw_timeout_limit < 100:
  45. # below 100, the unit is the second ( <3 = 3 seconds timeout )
  46. self.raw_text_query.timeout_limit = float(raw_timeout_limit)
  47. else:
  48. # 100 or above, the unit is the millisecond ( <850 = 850 milliseconds timeout )
  49. self.raw_text_query.timeout_limit = raw_timeout_limit / 1000.0
  50. return True
  51. def _autocomplete(self):
  52. for suggestion in ['<3', '<850']:
  53. self._add_autocomplete(suggestion)
  54. class LanguageParser(QueryPartParser):
  55. @staticmethod
  56. def check(raw_value):
  57. return raw_value[0] == ':'
  58. def __call__(self, raw_value):
  59. value = raw_value[1:].lower().replace('_', '-')
  60. found = self._parse(value) if len(value) > 0 else False
  61. if self.enable_autocomplete and not found:
  62. self._autocomplete(value)
  63. return found
  64. def _parse(self, value):
  65. found = False
  66. # check if any language-code is equal with
  67. # declared language-codes
  68. for lc in sxng_locales:
  69. lang_id, lang_name, country, english_name, _flag = map(str.lower, lc)
  70. # if correct language-code is found
  71. # set it as new search-language
  72. if (
  73. value == lang_id or value == lang_name or value == english_name or value.replace('-', ' ') == country
  74. ) and value not in self.raw_text_query.languages:
  75. found = True
  76. lang_parts = lang_id.split('-')
  77. if len(lang_parts) == 2:
  78. self.raw_text_query.languages.append(lang_parts[0] + '-' + lang_parts[1].upper())
  79. else:
  80. self.raw_text_query.languages.append(lang_id)
  81. # to ensure best match (first match is not necessarily the best one)
  82. if value == lang_id:
  83. break
  84. # user may set a valid, yet not selectable language
  85. if VALID_LANGUAGE_CODE.match(value) or value == 'auto':
  86. lang_parts = value.split('-')
  87. if len(lang_parts) > 1:
  88. value = lang_parts[0].lower() + '-' + lang_parts[1].upper()
  89. if value not in self.raw_text_query.languages:
  90. self.raw_text_query.languages.append(value)
  91. found = True
  92. return found
  93. def _autocomplete(self, value):
  94. if not value:
  95. # show some example queries
  96. if len(settings['search']['languages']) < 10:
  97. for lang in settings['search']['languages']:
  98. self.raw_text_query.autocomplete_list.append(':' + lang)
  99. else:
  100. for lang in [":en", ":en_us", ":english", ":united_kingdom"]:
  101. self.raw_text_query.autocomplete_list.append(lang)
  102. return
  103. for lc in sxng_locales:
  104. if lc[0] not in settings['search']['languages']:
  105. continue
  106. lang_id, lang_name, country, english_name, _flag = map(str.lower, lc)
  107. # check if query starts with language-id
  108. if lang_id.startswith(value):
  109. if len(value) <= 2:
  110. self._add_autocomplete(':' + lang_id.split('-')[0])
  111. else:
  112. self._add_autocomplete(':' + lang_id)
  113. # check if query starts with language name
  114. if lang_name.startswith(value) or english_name.startswith(value):
  115. self._add_autocomplete(':' + lang_name)
  116. # check if query starts with country
  117. # here "new_zealand" is "new-zealand" (see __call__)
  118. if country.startswith(value.replace('-', ' ')):
  119. self._add_autocomplete(':' + country.replace(' ', '_'))
  120. class ExternalBangParser(QueryPartParser):
  121. @staticmethod
  122. def check(raw_value):
  123. return raw_value.startswith('!!') and len(raw_value) > 2
  124. def __call__(self, raw_value):
  125. value = raw_value[2:]
  126. found, bang_ac_list = self._parse(value) if len(value) > 0 else (False, [])
  127. if self.enable_autocomplete:
  128. self._autocomplete(bang_ac_list)
  129. return found
  130. def _parse(self, value):
  131. found = False
  132. bang_definition, bang_ac_list = get_bang_definition_and_autocomplete(value)
  133. if bang_definition is not None:
  134. self.raw_text_query.external_bang = value
  135. found = True
  136. return found, bang_ac_list
  137. def _autocomplete(self, bang_ac_list):
  138. if not bang_ac_list:
  139. bang_ac_list = ['g', 'ddg', 'bing']
  140. for external_bang in bang_ac_list:
  141. self._add_autocomplete('!!' + external_bang)
  142. class BangParser(QueryPartParser):
  143. @staticmethod
  144. def check(raw_value):
  145. # make sure it's not any bang with double '!!'
  146. return raw_value[0] == '!' and (len(raw_value) < 2 or raw_value[1] != '!')
  147. def __call__(self, raw_value):
  148. value = raw_value[1:].replace('-', ' ').replace('_', ' ')
  149. found = self._parse(value) if len(value) > 0 else False
  150. if found and raw_value[0] == '!':
  151. self.raw_text_query.specific = True
  152. if self.enable_autocomplete:
  153. self._autocomplete(raw_value[0], value)
  154. return found
  155. def _parse(self, value):
  156. # check if prefix is equal with engine shortcut
  157. if value in engine_shortcuts: # pylint: disable=consider-using-get
  158. value = engine_shortcuts[value]
  159. # check if prefix is equal with engine name
  160. if value in engines:
  161. self.raw_text_query.enginerefs.append(EngineRef(value, 'none'))
  162. return True
  163. # check if prefix is equal with category name
  164. if value in categories:
  165. # using all engines for that search, which
  166. # are declared under that category name
  167. self.raw_text_query.enginerefs.extend(
  168. EngineRef(engine.name, value)
  169. for engine in categories[value]
  170. if (engine.name, value) not in self.raw_text_query.disabled_engines
  171. )
  172. return True
  173. return False
  174. def _autocomplete(self, first_char, value):
  175. if not value:
  176. # show some example queries
  177. for suggestion in ['images', 'wikipedia', 'osm']:
  178. if suggestion not in self.raw_text_query.disabled_engines or suggestion in categories:
  179. self._add_autocomplete(first_char + suggestion)
  180. return
  181. # check if query starts with category name
  182. for category in categories:
  183. if category.startswith(value):
  184. self._add_autocomplete(first_char + category.replace(' ', '_'))
  185. # check if query starts with engine name
  186. for engine in engines:
  187. if engine.startswith(value):
  188. self._add_autocomplete(first_char + engine.replace(' ', '_'))
  189. # check if query starts with engine shortcut
  190. for engine_shortcut in engine_shortcuts:
  191. if engine_shortcut.startswith(value):
  192. self._add_autocomplete(first_char + engine_shortcut)
  193. class FeelingLuckyParser(QueryPartParser):
  194. @staticmethod
  195. def check(raw_value):
  196. return raw_value == '!!'
  197. def __call__(self, raw_value):
  198. self.raw_text_query.redirect_to_first_result = True
  199. return True
  200. class RawTextQuery:
  201. """parse raw text query (the value from the html input)"""
  202. PARSER_CLASSES = [
  203. TimeoutParser, # force the timeout
  204. LanguageParser, # force a language
  205. ExternalBangParser, # external bang (must be before BangParser)
  206. BangParser, # force an engine or category
  207. FeelingLuckyParser, # redirect to the first link in the results list
  208. ]
  209. def __init__(self, query, disabled_engines):
  210. assert isinstance(query, str)
  211. # input parameters
  212. self.query = query
  213. self.disabled_engines = disabled_engines if disabled_engines else []
  214. # parsed values
  215. self.enginerefs = []
  216. self.languages = []
  217. self.timeout_limit = None
  218. self.external_bang = None
  219. self.specific = False
  220. self.autocomplete_list = []
  221. # internal properties
  222. self.query_parts = [] # use self.getFullQuery()
  223. self.user_query_parts = [] # use self.getQuery()
  224. self.autocomplete_location = None
  225. self.redirect_to_first_result = False
  226. self._parse_query()
  227. def _parse_query(self):
  228. """
  229. parse self.query, if tags are set, which
  230. change the search engine or search-language
  231. """
  232. # split query, including whitespaces
  233. raw_query_parts = re.split(r'(\s+)', self.query)
  234. last_index_location = None
  235. autocomplete_index = len(raw_query_parts) - 1
  236. for i, query_part in enumerate(raw_query_parts):
  237. # part does only contain spaces, skip
  238. if query_part.isspace() or query_part == '':
  239. continue
  240. # parse special commands
  241. special_part = False
  242. for parser_class in RawTextQuery.PARSER_CLASSES:
  243. if parser_class.check(query_part):
  244. special_part = parser_class(self, i == autocomplete_index)(query_part)
  245. break
  246. # append query part to query_part list
  247. qlist = self.query_parts if special_part else self.user_query_parts
  248. qlist.append(query_part)
  249. last_index_location = (qlist, len(qlist) - 1)
  250. self.autocomplete_location = last_index_location
  251. def get_autocomplete_full_query(self, text):
  252. qlist, position = self.autocomplete_location
  253. qlist[position] = text
  254. return self.getFullQuery()
  255. def changeQuery(self, query):
  256. self.user_query_parts = query.strip().split()
  257. self.query = self.getFullQuery()
  258. self.autocomplete_location = (self.user_query_parts, len(self.user_query_parts) - 1)
  259. self.autocomplete_list = []
  260. return self
  261. def getQuery(self):
  262. return ' '.join(self.user_query_parts)
  263. def getFullQuery(self):
  264. """
  265. get full query including whitespaces
  266. """
  267. return '{0} {1}'.format(' '.join(self.query_parts), self.getQuery()).strip()
  268. def __str__(self):
  269. return self.getFullQuery()
  270. def __repr__(self):
  271. return (
  272. f"<{self.__class__.__name__} "
  273. + f"query={self.query!r} "
  274. + f"disabled_engines={self.disabled_engines!r}\n "
  275. + f"languages={self.languages!r} "
  276. + f"timeout_limit={self.timeout_limit!r} "
  277. + f"external_bang={self.external_bang!r} "
  278. + f"specific={self.specific!r} "
  279. + f"enginerefs={self.enginerefs!r}\n "
  280. + f"autocomplete_list={self.autocomplete_list!r}\n "
  281. + f"query_parts={self.query_parts!r}\n "
  282. + f"user_query_parts={self.user_query_parts!r} >\n"
  283. + f"redirect_to_first_result={self.redirect_to_first_result!r}"
  284. )