Import škol: Základní zjednodušování názvů

bin/init-schools

@@ -11,6 +11,7 @@ import argparse
 from collections import defaultdict
 import csv
 from dataclasses import dataclass
+import re
 from sqlalchemy.orm import joinedload
 import sys
 from typing import List, Dict, DefaultDict, Tuple, Optional
@@ -244,6 +245,15 @@ def get_old_schools() -> DefaultDict[str, List[ProtoSchool]]:
     return old_schools
 
 
+def simplify_name(name: str, town: str) -> str:
+    name = re.sub('základní škola', 'ZŠ', name, flags=re.IGNORECASE)
+    name = re.sub('mateřská škola', 'MŠ', name, flags=re.IGNORECASE)
+    name = re.sub('střední škola', 'SŠ', name, flags=re.IGNORECASE)
+    name = re.sub('gymnázium', 'G', name, flags=re.IGNORECASE)
+    name = re.sub(r',?\s*s\.r\.o\.', "", name)
+    return name + ', ' + town
+
+
 def apply_single_change(old: Optional[ProtoSchool], new: Optional[ProtoSchool]) -> None:
     if new is None:
         assert old is not None
@@ -253,10 +263,11 @@ def apply_single_change(old: Optional[ProtoSchool], new: Optional[ProtoSchool])
             print(f'\t{field}: {getattr(old, field)}')
     elif old is None:
         assert new is not None
+        simple_name = simplify_name(new.official_name, new.town_name)
         place = db.Place(
             level=4,
             parent=new.town.place_id,
-            name=new.official_name,
+            name=simple_name,
             type=db.PlaceType.school)
         school = db.School(
             place=place,
@@ -271,6 +282,7 @@ def apply_single_change(old: Optional[ProtoSchool], new: Optional[ProtoSchool])
 
         if args.update:
             print(f'Škola #{school.place_id} (RED IZO {new.red_izo}): NOVÁ')
+            print(f'\tname: {simple_name}')
             for field in fields:
                 print(f'\t{field}: {getattr(new, field)}')