1 # SPDX-License-Identifier: GPL-2.0-only
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2022 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 Specialized processor for housenumbers. Analyses common housenumber patterns
9 and creates variants for them.
13 from nominatim.tokenizer.token_analysis.generic_mutation import MutationVariantGenerator
15 RE_NON_DIGIT = re.compile('[^0-9]')
16 RE_DIGIT_ALPHA = re.compile(r'(\d)\s*([^\d\s␣])')
17 RE_ALPHA_DIGIT = re.compile(r'([^\s\d␣])\s*(\d)')
18 RE_NAMED_PART = re.compile(r'[a-z]{4}')
20 ### Configuration section
22 def configure(rules, normalization_rules): # pylint: disable=W0613
23 """ All behaviour is currently hard-coded.
29 def create(normalizer, transliterator, config): # pylint: disable=W0613
30 """ Create a new token analysis instance for this module.
32 return HousenumberTokenAnalysis(normalizer, transliterator)
35 class HousenumberTokenAnalysis:
36 """ Detects common housenumber patterns and normalizes them.
38 def __init__(self, norm, trans):
42 self.mutator = MutationVariantGenerator('␣', (' ', ''))
44 def normalize(self, name):
45 """ Return the normalized form of the housenumber.
47 # shortcut for number-only numbers, which make up 90% of the data.
48 if RE_NON_DIGIT.search(name) is None:
51 norm = self.trans.transliterate(self.norm.transliterate(name))
52 # If there is a significant non-numeric part, use as is.
53 if RE_NAMED_PART.search(norm) is None:
54 # Otherwise add optional spaces between digits and letters.
55 (norm_opt, cnt1) = RE_DIGIT_ALPHA.subn(r'\1␣\2', norm)
56 (norm_opt, cnt2) = RE_ALPHA_DIGIT.subn(r'\1␣\2', norm_opt)
57 # Avoid creating too many variants per number.
63 def get_variants_ascii(self, norm_name):
64 """ Compute the spelling variants for the given normalized housenumber.
66 Generates variants for optional spaces (marked with '␣').
68 return list(self.mutator.generate([norm_name]))