Made synonym list all lower()

2024-10-28 10:56:10 -07:00
parent 558c53c34c
commit 8a3de562eb
22 changed files with 22398 additions and 22395 deletions
--- a/tools/ucsxls2json.py
+++ b/tools/ucsxls2json.py
@@ -77,8 +77,11 @@ for lang in langs:
            if key_name == 'Synonyms':
                # synonyms are stored in the spreadsheet as CSV, we should 
                # normalize this. 
-                category['Synonyms'] = re.split(r'\W+', category['Synonyms'])
-                
+                syns_raw = category['Synonyms']
+                assert type(syns_raw) == str, \
+                        f"Synonym list (lang: {lang}, {category['CatID']}) was not readable"
+                syn_list = re.split(r'\W+', syns_raw)
+                category['Synonyms'] = [s.lower() for s in syn_list]
        
        schedule.append(category)