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, Optional
  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 KANJI_MAP = {
  22       ord('零'): '0',
  23       ord('一'): '1',
  24       ord('二'): '2',
  25       ord('三'): '3',
  26       ord('四'): '4',
  27       ord('五'): '5',
  28       ord('六'): '6',
  29       ord('七'): '7',
  30       ord('八'): '8',
  31       ord('九'): '9'
  32     }
  33
  34 def convert_kanji_sequence_to_number(sequence: str) -> str:
  35     """Converts Kanji numbers to Arabic numbers
  36     """
  37     converted = sequence.translate(KANJI_MAP)
  38     return converted
  39
  40 def create(_: SanitizerConfig) -> Callable[[ProcessInfo], None]:
  41     """Set up the sanitizer
  42     """
  43     return tag_japanese
  44
  45 def reconbine_housenumber(
  46     new_address: List[PlaceName],
  47     tmp_housenumber: Optional[str],
  48     tmp_blocknumber: Optional[str]
  49 ) -> List[PlaceName]:
  50     """ Recombine the tag of housenumber by using housenumber and blocknumber
  51     """
  52     if tmp_blocknumber:
  53         tmp_blocknumber = convert_kanji_sequence_to_number(tmp_blocknumber)
  54     if tmp_housenumber:
  55         tmp_housenumber = convert_kanji_sequence_to_number(tmp_housenumber)
  56
  57     if tmp_blocknumber and tmp_housenumber:
  58         new_address.append(
  59             PlaceName(
  60                 kind='housenumber',
  61                 name=f'{tmp_blocknumber}-{tmp_housenumber}',
  62                 suffix=''
  63             )
  64         )
  65     elif tmp_blocknumber:
  66         new_address.append(
  67             PlaceName(
  68                 kind='housenumber',
  69                 name=tmp_blocknumber,
  70                 suffix=''
  71             )
  72         )
  73     elif tmp_housenumber:
  74         new_address.append(
  75             PlaceName(
  76                 kind='housenumber',
  77                 name=tmp_housenumber,
  78                 suffix=''
  79             )
  80         )
  81     return new_address
  82
  83 def reconbine_place(
  84     new_address: List[PlaceName],
  85     tmp_neighbourhood: Optional[str],
  86     tmp_quarter: Optional[str]
  87 ) -> List[PlaceName]:
  88     """ Recombine the tag of place by using neighbourhood and quarter
  89     """
  90     if tmp_neighbourhood:
  91         tmp_neighbourhood = convert_kanji_sequence_to_number(tmp_neighbourhood)
  92     if tmp_quarter:
  93         tmp_quarter = convert_kanji_sequence_to_number(tmp_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=tmp_neighbourhood,
 108                 suffix=''
 109             )
 110         )
 111     elif tmp_quarter:
 112         new_address.append(
 113             PlaceName(
 114                 kind='place',
 115                 name=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.address:
 132         if item.kind == 'housenumber':
 133             tmp_housenumber = item.name
 134         elif item.kind == 'block_number':
 135             tmp_blocknumber = item.name
 136         elif item.kind == 'neighbourhood':
 137             tmp_neighbourhood = item.name
 138         elif item.kind == 'quarter':
 139             tmp_quarter = item.name
 140         else:
 141             new_address.append(item)
 142
 143     new_address = reconbine_housenumber(new_address, tmp_housenumber, tmp_blocknumber)
 144     new_address = reconbine_place(new_address, tmp_neighbourhood, tmp_quarter)
 145
 146     obj.address = [item for item in new_address if item.name is not None]