Durable image quality: stop trusting feed thumbnails; cycle enriches Latest

Make "no blurry images" sustainable, not a one-off cleanup. RSS feed thumbnails
(~44% were ~90px) were stored at ingest and upscaled to mush, so new articles
would reintroduce them. Now image_url is filled ONLY by the quality-gated
og:image enrichment:

* insert_article no longer stores the feed image (was canonicalize_url(item...)).
* enrich_recent_images(): the cycle fetches a quality og:image for the newest
  accepted, imageless articles each run (bounded), keeping Latest photo-rich.
* Brief + on-open enrichment unchanged.

Net: every stored image is a validated, ≥450px og:image; the rest are clean
placeholders.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
jay
2026-06-07 15:55:57 -04:00
parent b134c2dab6
commit 50dc2167cd
4 changed files with 56 additions and 2 deletions
+16
View File
@@ -78,3 +78,19 @@ def test_image_dimensions_parses_headers():
gif = b"GIF89a" + struct.pack("<HH", 90, 90)
assert enrich._image_dimensions(gif) == (90, 90)
assert enrich._image_dimensions(b"not an image at all") is None
def test_enrich_recent_targets_newest_imageless_accepted(tmp_path):
c = _setup(tmp_path)
# newest first by published_at; summary not required (unlike the backfill)
c.execute("INSERT INTO articles (id,source_id,canonical_url,title,url_hash,image_url,published_at) "
"VALUES (1,1,'u1','T1','h1',NULL,'2026-01-01T00:00:00')")
c.execute("INSERT INTO articles (id,source_id,canonical_url,title,url_hash,image_url,published_at) "
"VALUES (2,1,'u2','T2','h2',NULL,'2026-06-01T00:00:00')") # newest
c.execute("INSERT INTO article_scores (article_id,accepted) VALUES (1,1),(2,1)")
c.commit()
seen = []
enrich.enrich_recent_images(c, fetch=lambda url: seen.append(url) or "http://og.jpg", limit=10)
# both enriched; newest processed first
assert seen[0].endswith("u2")
assert c.execute("SELECT image_url FROM articles WHERE id=2").fetchone()[0] == "http://og.jpg"