Fix rewrite: dedup sentences, complete sentence trim, merge images from direct+Jina scrape

#1

Fixes:

  1. _clean_ai_output: complete sentence trimming, word-overlap dedup (not just substring)
  2. Prompt: explicit "KHÔNG sao chép nguyên văn" rule + "Chỉ viết phần tóm tắt" suffix
  3. Image extraction: 12 strategies including fig-parent, content-block-img, article-scope
  4. scrape_any_url: MERGE images from direct scrape + Jina instead of overwriting
  5. Frontend: voice selector always shown, gallery with swapGalleryHero, wall-video-badge CSS
bep40 changed pull request status to open
bep40 changed pull request status to merged

Sign up or log in to comment