scripts/disambiguate-location-identifiers.py

   1 # Encoding: UTF-8
   2 """Automatically disambiguate location identifiers
   3
   4 This is an unmaintained one-shot script, only included in the repo for reference.
   5
   6
   7 Disambiguates identifiers that aren't unique, Routes and Sea Routes, and
   8 generic names like 'villa' or 'game corner' that could appear in future
   9 generations again.
  10
  11 Does this by prepending the region name, and if that isn't enough, appends
  12 numbers.
  13 """
  14
  15 import sys
  16 import re
  17 from collections import defaultdict
  18
  19 from pokedex.db import connect, tables
  20
  21 ambiguous_re = re.compile(r'^(sea-)?route-\d+$')
  22
  23 ambiguous_set = set('foreign-building game-corner global-terminal lighthouse '
  24     'restaurant flower-shop cycle-shop cafe shopping-mall villa'.split())
  25
  26 def main(*argv):
  27     session = connect()
  28
  29     location_dict = defaultdict(list)
  30     for location in session.query(tables.Location).order_by(tables.Location.id):
  31         location_dict[location.identifier].append(location)
  32
  33     changes = False
  34     for identifier, locations in sorted(location_dict.items()):
  35         disambiguate = any((
  36                 len(locations) > 1,
  37                 ambiguous_re.match(identifier),
  38                 identifier in ambiguous_set,
  39             ))
  40         print len(locations), ' *'[disambiguate], identifier,
  41         if disambiguate:
  42             changes = True
  43             print u'→'.encode('utf-8'),
  44             by_region = defaultdict(list)
  45             for location in locations:
  46                 if location.region:
  47                     by_region[location.region.identifier].append(location)
  48                 else:
  49                     by_region[None].append(location)
  50             for region_identifier, region_locations in by_region.items():
  51                 if region_identifier:
  52                     new_identifier = '%s-%s' % (region_identifier, identifier)
  53                 else:
  54                     # No change
  55                     new_identifier = identifier
  56                 if len(region_locations) == 1:
  57                     location = region_locations[0]
  58                     # The region was enough
  59                     print new_identifier,
  60                     location.identifier = new_identifier
  61                 else:
  62                     # Need to number the locations :(
  63                     for i, location in enumerate(region_locations, start=1):
  64                         numbered_identifier = '%s-%s' % (new_identifier, i)
  65                         print numbered_identifier,
  66                         location.identifier = numbered_identifier
  67         print
  68
  69     if changes:
  70         if argv and argv[0] == '--commit':
  71             session.commit()
  72             print 'Committed'
  73         else:
  74             print 'Run with --commit to commit changes'
  75     else:
  76         print 'No changes needed'
  77
  78
  79 if __name__ == '__main__':
  80     main(*sys.argv[1:])