def build_matcher_from_lists(*collections: List[str]) -> StringMatcher:
matcher = StringMatcher(MatchStrategy.TrieTree, NumberWithUnitTokenizer())
matcher_list = []
for collection in collections:
list(map(lambda x: matcher_list.append(x.strip().lower()), collection))
matcher_list = TimeZoneUtility.distinct(matcher_list)
matcher.init(matcher_list)
return matcher
def build_matcher_from_lists(self, collections: []):
matcher: StringMatcher = StringMatcher(MatchStrategy.TrieTree, NumberWithUnitTokenizer())
matcher_list = []
for collection in collections:
list(map(lambda x: matcher_list.append(x.strip().lower()), collection))
matcher_list = self.distinct(matcher_list)
matcher.init(matcher_list)
return matcher
def _build_matcher_from_set(self, definitions) -> StringMatcher:
matcher = StringMatcher(match_strategy=MatchStrategy.TrieTree, tokenizer=NumberWithUnitTokenizer())
match_term_list = list(map(lambda words:
list(filter(lambda word: not str.isspace(word) and word is not None,
str(words).strip().split('|'))),
definitions))
match_terms = self.distinct(match_term_list)
flatten = [item for sublist in match_terms for item in sublist]
matcher.init(flatten)
return matcher