Source servers return 503 (Cloudflare) for cross-origin image requests
from browsers. Solution: download and cache images server-side during
scraping, serve from /static/uploads/zopk/.
- Scraper now downloads og:image and stores locally during article
scraping (max 2MB, supports jpg/png/webp)
- Backfill script downloads images for all existing articles server-side
- Template fallback shows domain initial letter when image unavailable
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix broken news thumbnails by adding og:image extraction during content
scraping (replaces Brave proxy URLs that block hotlinking)
- Add image onerror fallback in templates showing domain favicon when
original image fails to load
- Decode Brave proxy image URLs to original source URLs before saving
- Enforce English-only entity types in AI extraction prompt to prevent
mixed Polish/English type names
- Add migration 083 to normalize 14 existing Polish entity types and
clean up 5 stale fetch jobs stuck in 'running' status
- Add backfill script for existing articles with broken image URLs
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
