(doc_id)
| 17 | ] # EXCLUDING 2025-6 which are just table checkmarks |
| 18 | |
| 19 | def get_url(doc_id): |
| 20 | md_file = MDCG_DATA_DIR / f"{doc_id}.zh.md" |
| 21 | if not md_file.exists(): return None |
| 22 | content = md_file.read_text(encoding="utf-8") |
| 23 | m = re.search(r'source_url:\s*(.+)$', content, re.M) |
| 24 | return m.group(1).strip() if m else None |
| 25 | |
| 26 | pipeline_options = PdfPipelineOptions() |
| 27 | pipeline_options.generate_picture_images = True |