1+ from pathlib import Path
2+ import re
3+
4+ def load_existing_list (path ):
5+ with open (path , encoding = 'utf-8' , newline = '' ) as f :
6+ return f .read ()
7+
8+ japanese_list = '../msgothic_0_charset_Japanese.txt'
9+ multilang_list = '../msgothic_2_charset_OtherLang.txt'
10+ out_char_list = '../msgothic_2_charset_JP_and_OtherLang.txt'
11+
12+ jp = load_existing_list (japanese_list )
13+ multi = load_existing_list (multilang_list )
14+
15+ chars_to_add = set (jp )
16+ existing_chars = set (multi )
17+
18+
19+ with open (out_char_list , 'w' , encoding = 'utf-8' , newline = '' ) as f :
20+ for i , c in enumerate (multi ):
21+ f .write (c )
22+
23+ # This is very bad for performance if there are lots of new chars found, but it works for now to maintain ordering
24+ remove_list = []
25+ for new_character in chars_to_add :
26+ if new_character < c :
27+ f .write (new_character )
28+ remove_list .append (new_character )
29+ print (f"Inserting new character { new_character } { i } { c } )
30+
31+ for item in remove_list :
32+ chars_to_add .remove (item )
33+
34+ remove_list = []
35+ for char in chars_to_add :
36+ if char not in existing_chars :
37+ f .write (char )
38+ else :
39+ print (f"WARNING: character { char } )
40+ remove_list .append (char )
41+
42+ for item in remove_list :
43+ chars_to_add .remove (item )
44+
45+
46+ if chars_to_add :
47+ raise Exception (f"One or more characters were not added { chars_to_add } )
0 commit comments