scan.py

import re
import conf

findVoiceRegex = re.compile(r'"voice\\(\d\d)\\(.*?)\.ogg"')

umitweak_script = r'c:\temp\umitweak_original.utf'
current_script = r'C:\drojf\large_projects\umineko\git_repository\0.utf'

git_text = None
with open(current_script, encoding=conf.encoding) as script_in:
    git_text = script_in.read()


git_id_set = set()

for match in re.finditer(findVoiceRegex, git_text):
    character_no = match[1]
    ogg_number = match[2]
    git_id_set.add(ogg_number)


with open(umitweak_script, encoding=conf.encoding) as script_in:
    for i, line in enumerate(script_in):
        for match in re.finditer(findVoiceRegex, line):
            full_match = match[0]
            ogg_number = match[2]

            if ogg_number not in git_id_set:

                print('ut_line:{:6d} path:[{:26}] full_line:[{}]'.format(i, full_match, line.strip()))