nominatim/tokenizer/sanitizers/clean_postcodes.py

   1 # SPDX-License-Identifier: GPL-2.0-only
   2 #
   3 # This file is part of Nominatim. (https://nominatim.org)
   4 #
   5 # Copyright (C) 2022 by the Nominatim developer community.
   6 # For a full list of authors see the git log.
   7 """
   8 Sanitizer that filters postcodes by their officially allowed pattern.
   9
  10 Arguments:
  11     convert-to-address: If set to 'yes' (the default), then postcodes that do
  12                         not conform with their country-specific pattern are
  13                         converted to an address component. That means that
  14                         the postcode does not take part when computing the
  15                         postcode centroids of a country but is still searchable.
  16                         When set to 'no', non-conforming postcodes are not
  17                         searchable either.
  18 """
  19 import re
  20
  21 from nominatim.errors import UsageError
  22 from nominatim.tools import country_info
  23
  24 class _PostcodeMatcher:
  25     """ Matches and formats a postcode according to the format definition.
  26     """
  27     def __init__(self, country_code, config):
  28         if 'pattern' not in config:
  29             raise UsageError("Field 'pattern' required for 'postcode' "
  30                              f"for country '{country_code}'")
  31
  32         self.pattern = re.compile(config['pattern'].replace('d', '[0-9]')
  33                                                    .replace('l', '[A-Z]'))
  34
  35
  36     def normalize(self, postcode):
  37         """ Return the normalized version of the postcode. If the given postcode
  38             does not correspond to the usage-pattern, return null.
  39         """
  40         normalized = postcode.strip().upper()
  41
  42         return normalized if self.pattern.fullmatch(normalized) else None
  43
  44
  45 class _PostcodeSanitizer:
  46
  47     def __init__(self, config):
  48         self.convert_to_address = config.get_bool('convert-to-address', True)
  49         # Objects without a country code can't have a postcode per definition.
  50         self.country_without_postcode = {None}
  51         self.country_matcher = {}
  52
  53         for ccode, prop in country_info.iterate('postcode'):
  54             if prop is False:
  55                 self.country_without_postcode.add(ccode)
  56             elif isinstance(prop, dict):
  57                 self.country_matcher[ccode] = _PostcodeMatcher(ccode, prop)
  58             else:
  59                 raise UsageError(f"Invalid entry 'postcode' for country '{ccode}'")
  60
  61
  62     def __call__(self, obj):
  63         if not obj.address:
  64             return
  65
  66         postcodes = ((i, o) for i, o in enumerate(obj.address) if o.kind == 'postcode')
  67
  68         for pos, postcode in postcodes:
  69             formatted = self.scan(postcode.name, obj.place.country_code)
  70
  71             if formatted is None:
  72                 if self.convert_to_address:
  73                     postcode.kind = 'unofficial_postcode'
  74                 else:
  75                     obj.address.pop(pos)
  76             else:
  77                 postcode.name = formatted
  78
  79
  80     def scan(self, postcode, country):
  81         """ Check the postcode for correct formatting and return the
  82             normalized version. Returns None if the postcode does not
  83             correspond to the oficial format of the given country.
  84         """
  85         if country in self.country_without_postcode:
  86             return None
  87
  88         if country in self.country_matcher:
  89             return self.country_matcher[country].normalize(postcode)
  90
  91         return postcode.upper()
  92
  93
  94
  95 def create(config):
  96     """ Create a housenumber processing function.
  97     """
  98
  99     return _PostcodeSanitizer(config)