pboProperty Ontology
python19 lines562 B
RawDownload
1#!/usr/bin/env python3
2"""Reformat completed scraped documents without shortening their content."""
3
4from pathlib import Path
5import re
6
7from scrape_bookmarks import clean_markdown
8
9
10ROOT = Path(__file__).resolve().parent
11OUTPUT = ROOT / "property-pages"
12
13
14for path in sorted(OUTPUT.glob("*.md")):
15 original = path.read_text(encoding="utf-8")
16 heading = re.search(r"^#\s+(.+)$", original, flags=re.M)
17 title = heading.group(1).strip() if heading else path.stem
18 path.write_text(clean_markdown(original, title), encoding="utf-8")
19 print(path.name)