def make_de_definitions_parser():
split_params = LineSplitParams()
split_params.line_breaks = {'\n', '.', ';', '!', '?'}
split_params.abbreviations = DeLanguageTokens.abbreviations
split_params.abbr_ignore_case = True
functions = [CommonDefinitionPatterns.match_es_def_by_semicolon,
CommonDefinitionPatterns.match_acronyms,
DeutscheParsingMethods.match_ist_jeder,
DeutscheParsingMethods.match_im_sinne]
p = UniversalDefinitionsParser(functions, split_params)
p.prohibited_words = {w for w in DeLanguageTokens.articles + DeLanguageTokens.conjunctions}
return p
def make_es_definitions_parser():
split_params = LineSplitParams()
split_params.line_breaks = {'\n', '.', ';', '!', '?'}
split_params.abbreviations = EsLanguageTokens.abbreviations
split_params.abbr_ignore_case = True
functions = [CommonDefinitionPatterns.match_es_def_by_semicolon,
CommonDefinitionPatterns.match_acronyms,
SpanishParsingMethods.match_es_def_by_hereafter,
SpanishParsingMethods.match_es_def_by_reffered,
SpanishParsingMethods.match_first_word_is]
return UniversalDefinitionsParser(functions, split_params)