nominatim/tokenizer/sanitizers/strip_brace_terms.py

   1 """
   2 This sanitizer creates additional name variants for names that have
   3 addendums in brackets (e.g. "Halle (Saale)"). The additional variant contains
   4 only the main name part with the bracket part removed.
   5 """
   6
   7 def create(_):
   8     """ Create a name processing function that creates additional name variants
   9         for bracket addendums.
  10     """
  11     def _process(obj):
  12         """ Add variants for names that have a bracket extension.
  13         """
  14         if obj.names:
  15             new_names = []
  16             for name in (n for n in obj.names if '(' in n.name):
  17                 new_name = name.name.split('(')[0].strip()
  18                 if new_name:
  19                     new_names.append(name.clone(name=new_name))
  20
  21             obj.names.extend(new_names)
  22
  23     return _process