Method migrate

scripts/migrate_posts.py:255–321

Main migration entry point.

(self)

            print(f"  {subcat:<30} {count:3d} posts")

    def migrate(self) -> None:
        """Main migration entry point."""
        mode = "[DRY RUN] " if self.dry_run else ""
        print(f"\n{mode}Migrating WordPress posts to insights/")
        print(f"Output directory: {self.output_dir}\n")

        posts = self.get_all_posts()
        print(f"\nTotal posts to process: {len(posts)}")

        path_entries: Dict[str, List[dict]] = {}

        for i, post in enumerate(posts, 1):
            self.stats["total"] += 1
            title = self.clean_title(post.get("title", {}).get("rendered", ""))
            date = post.get("date", "")[:10]
            print(f"[{i:3d}/{len(posts)}] [{date}] {title[:60]}")

            # Resolve category slugs (with delay to avoid rate limiting)
            cat_ids = post.get("categories", [])
            for cat_id in cat_ids:
                self.get_category_slug(cat_id)
            if cat_ids:
                time.sleep(self.delay)

            try:
                zh_file, fm = self.save_post(post)
                if fm is None:
                    self.stats["skipped"] += 1
                    print(f"        SKIP (empty content)")
                    continue

                output_path = fm["category"]
                if output_path not in path_entries:
                    path_entries[output_path] = []
                path_entries[output_path].append({
                    "id": fm["id"],
                    "title": fm["title"],
                    "published_date": fm["published_date"],
                    "subcategory": fm["subcategory"],
                    "source_url": fm["source_url"],
                    "excerpt": fm.get("excerpt", {}),
                })

                self.stats["success"] += 1
                if self.dry_run:
                    print(f"        -> [DRY] {output_path}/{post.get('slug', '')}.zh.md")
                else:
                    print(f"        -> {zh_file.relative_to(self.output_dir)}")
            except Exception as e:
                self.stats["error"] += 1
                print(f"        ERROR: {e}")

            time.sleep(self.delay)

        if not self.dry_run:
            print("\nUpdating _index.json files...")
            for path, entries in path_entries.items():
                self.update_index(path, entries)