tweak paths and find links

This commit is contained in:
maximator 2026-04-03 15:50:40 +02:00
parent 36c8bb2354
commit e4aaa33137
7 changed files with 309 additions and 10 deletions

View file

@ -3,8 +3,8 @@ import shutil
import re
import json
INPUT_DIR = Path("unique_pages")
OUTPUT_DIR = Path("classified_pages")
INPUT_DIR = Path("../unique_pages")
OUTPUT_DIR = Path("../classified_pages")
CATEGORY_RE = re.compile(r'"wgCategories":\[(.*?)\]')