nominatim/tokenizer/sanitizers/tag_japanese.py

   1 # SPDX-License-Identifier: GPL-3.0-or-later
   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 This sanitizer maps OSM data to Japanese block addresses.
   9 It replaces blocknumber and housenumber with housenumber,
  10 and quarter and neighbourhood with place.
  11 """
  12
  13
  14 from typing import Callable
  15 from typing import List
  16
  17 from nominatim.tokenizer.sanitizers.base import ProcessInfo
  18 from nominatim.tokenizer.sanitizers.config import SanitizerConfig
  19 from nominatim.data.place_name import PlaceName
  20
  21 def create(_: SanitizerConfig) -> Callable[[ProcessInfo], None]:
  22 #def create(config: SanitizerConfig) -> Callable[[ProcessInfo],None]:
  23     """Set up the sanitizer
  24     """
  25     return tag_japanese
  26     #return tag_japanese(config)
  27
  28 def convert_kanji_sequence_to_number(sequence: str) -> str:
  29     """Converts Kanji numbers to Arabic numbers
  30     """
  31     kanji_map = {
  32       '零': '0',
  33       '一': '1',
  34       '二': '2',
  35       '三': '3',
  36       '四': '4',
  37       '五': '5',
  38       '六': '6',
  39       '七': '7',
  40       '八': '8',
  41       '九': '9'
  42     }
  43     converted = ''
  44     current_number = ''
  45     for char in sequence:
  46         if char in kanji_map:
  47             current_number += kanji_map[char]
  48         else:
  49             converted += current_number
  50             current_number = ''
  51             converted += char
  52     converted += current_number
  53     return converted
  54
  55 def reconbine_housenumber(
  56     new_address: List[PlaceName],
  57     tmp_housenumber: str | None,
  58     tmp_blocknumber: str | None
  59 ) -> List[PlaceName]:
  60     """ Recombine the tag of housenumber by using housenumber and blocknumber
  61     """
  62     if tmp_blocknumber and tmp_housenumber:
  63         new_address.append(
  64             PlaceName(
  65                 kind='housenumber',
  66                 name=f'{tmp_blocknumber}-{tmp_housenumber}',
  67                 suffix=''
  68             )
  69         )
  70     elif tmp_blocknumber:
  71         new_address.append(
  72             PlaceName(
  73                 kind='housenumber',
  74                 name=f'{tmp_blocknumber}',
  75                 suffix=''
  76             )
  77         )
  78     elif tmp_housenumber:
  79         new_address.append(
  80             PlaceName(
  81                 kind='housenumber',
  82                 name=f'{tmp_housenumber}',
  83                 suffix=''
  84             )
  85         )
  86     return new_address
  87
  88 def reconbine_place(
  89     new_address: List[PlaceName],
  90     tmp_neighbourhood: str | None,
  91     tmp_quarter: str | None
  92 ) -> List[PlaceName]:
  93     """ Recombine the tag of place by using neighbourhood and quarter
  94     """
  95     if tmp_neighbourhood and tmp_quarter:
  96         new_address.append(
  97             PlaceName(
  98                 kind='place',
  99                 name=f'{tmp_quarter}{tmp_neighbourhood}',
 100                 suffix=''
 101             )
 102         )
 103     elif tmp_neighbourhood:
 104         new_address.append(
 105             PlaceName(
 106                 kind='place',
 107                 name=f'{tmp_neighbourhood}',
 108                 suffix=''
 109             )
 110         )
 111     elif tmp_quarter:
 112         new_address.append(
 113             PlaceName(
 114                 kind='place',
 115                 name=f'{tmp_quarter}',
 116                 suffix=''
 117             )
 118         )
 119     return new_address
 120 def tag_japanese(obj: ProcessInfo) -> None:
 121     """Recombine kind of address
 122     """
 123     if obj.place.country_code != 'jp':
 124         return
 125     tmp_housenumber = None
 126     tmp_blocknumber = None
 127     tmp_neighbourhood = None
 128     tmp_quarter = None
 129
 130     new_address = []
 131     for item in obj.names:
 132         item.name = convert_kanji_sequence_to_number(item.name)
 133
 134     for item in obj.address:
 135         item.name = convert_kanji_sequence_to_number(item.name)
 136         if item.kind == 'housenumber':
 137             tmp_housenumber = item.name
 138         elif item.kind == 'block_number':
 139             tmp_blocknumber = item.name
 140         elif item.kind == 'neighbourhood':
 141             tmp_neighbourhood = item.name
 142         elif item.kind == 'quarter':
 143             tmp_quarter = item.name
 144         else:
 145             new_address.append(item)
 146
 147     new_address = reconbine_housenumber(new_address,tmp_housenumber,tmp_blocknumber)
 148     new_address = reconbine_place(new_address,tmp_neighbourhood,tmp_quarter)
 149
 150     obj.address = [item for item in new_address if item.name is not None]