Reject branded/generic share images; hero prefers a clean illustrated story

- og:image enrichment now skips branded/generic share images (BBC 'branded_news'
  with its burned-in logo, NPR 'facebook-default', etc.) and keeps the first
  real article image — so no competitor logo lands on our hero. Cleared the few
  already-stored branded URLs so they re-enrich.
- Hero selection now prefers a gentle + readable story that also HAS a (clean)
  image, falling back to gentle-readable, then gentle. The lead is visual when
  possible, typographic otherwise — never branded.

(The '7 cards' report was a stale browser cache: the brief stores 7 and the
built JS requests 7; a hard refresh shows all seven.)

Tests: branded/generic image rejection (87 total).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

goodnews/api.py

@@ -58,13 +58,18 @@ def get_conn():
 
 
 def _pick_lead(items: list[dict]) -> list[dict]:
-    """Lead with a gentle, readable story when possible.
+    """Lead with a gentle, readable, ideally illustrated story.
 
-    Tries gentle-and-readable first, then gentle, then leaves the order alone.
+    Preference order: gentle + readable + has an image, then gentle + readable,
-    Charged or paywalled stories still appear in the set — they just don't lead.
+    then gentle, then leave the order alone. Charged/paywalled/imageless stories
+    still appear in the set — they just don't lead.
     """
+    def gentle(a: dict) -> bool:
+        return safe_to_lead(a) and not is_paywalled(a.get("canonical_url"))
+
     for ok in (
-        lambda a: safe_to_lead(a) and not is_paywalled(a.get("canonical_url")),
+        lambda a: gentle(a) and bool(a.get("image_url")),
+        gentle,
         safe_to_lead,
     ):
         for i, a in enumerate(items):

goodnews/enrich.py

@@ -34,6 +34,28 @@ MAX_REDIRECTS = 3
 _META_RE = re.compile(rb"<meta\b[^>]*>", re.IGNORECASE)
 _HEAD_END_RE = re.compile(rb"</head>", re.IGNORECASE)
 
+# Substrings that mark a branded/generic share image rather than the article's
+# own picture (e.g. BBC burns "BBC NEWS" into branded_news; NPR uses a default).
+# We'd rather show no image (typographic hero) than a competitor logo.
+_GENERIC_IMAGE_MARKERS = (
+    "branded_news",
+    "facebook-default",
+    "default-wide",
+    "default-fb",
+    "og-default",
+    "default-og",
+    "twitter-default",
+    "default-image",
+    "/placeholder",
+    "share-default",
+    "social-default",
+)
+
+
+def _is_generic_image(url: str) -> bool:
+    lowered = url.lower()
+    return any(marker in lowered for marker in _GENERIC_IMAGE_MARKERS)
+
 
 def _attr(tag: bytes, name: bytes) -> bytes | None:
     m = re.search(name + rb"""\s*=\s*["']([^"']*)["']""", tag, re.IGNORECASE)
@@ -47,8 +69,12 @@ def og_image_from_html(html: bytes) -> str | None:
         key = _attr(tag, b"property") or _attr(tag, b"name")
         if key and key.lower() in (b"og:image", b"og:image:url", b"twitter:image"):
             content = _attr(tag, b"content")
-            if content:
+            if not content:
-                return canonicalize_url(content.decode("utf-8", "replace"))
+                continue
+            image = canonicalize_url(content.decode("utf-8", "replace"))
+            # Skip branded/generic share images; keep scanning for a real one.
+            if image and not _is_generic_image(image):
+                return image
     return None
 
 

tests/test_enrich.py

@@ -69,3 +69,14 @@ def test_enrich_upgrades_existing_feed_image(tmp_path):
     assert found == 1
     assert c.execute("SELECT image_url FROM articles WHERE id=1").fetchone()["image_url"] == "https://bbc.com/big-og.jpg"
     c.close()
+
+
+def test_rejects_branded_and_generic_share_images():
+    assert og_image_from_html(b'<meta property="og:image" content="https://ichef.bbci.co.uk/news/1024/branded_news/x.jpg">') is None
+    assert og_image_from_html(b'<meta name="twitter:image" content="https://media.npr.org/include/images/facebook-default-wide-s.jpg">') is None
+    # a real article image still comes through
+    assert og_image_from_html(b'<meta property="og:image" content="https://x.com/real-photo.jpg">') == "https://x.com/real-photo.jpg"
+    # if the first og is branded but a later one is real, take the real one
+    html = (b'<meta property="og:image" content="https://x.com/branded_news/logo.jpg">'
+            b'<meta property="og:image" content="https://x.com/article.jpg">')
+    assert og_image_from_html(html) == "https://x.com/article.jpg"