fix(posts): link duplicate items to every post + prune bare shells

The native Patreon backfill flooded the feed with bare 'Post <id>' shells
(1589 for Anduo). Root cause: PostAttachment.sha256 was GLOBALLY unique, so a
non-art file reused across posts only ever linked to the first one, and
_capture_attachment created the Post before that dedup check — leaving later
posts with no image and no attachment. Duplicate IMAGES had the mirror gap:
attach_in_place returned duplicate_hash/duplicate_phash before _apply_sidecar,
so the second post got no provenance row, and the feed only rendered via
primary_post_id (one post per image).

Operator requirement: a duplicate item must show on EVERY post it appears in.
Unify the fix as link-not-suppress:

- importer: on duplicate_hash / duplicate_phash(larger_exists), append an
  image_provenance row for the new post (keep primary on the first). Both the
  download path (attach_in_place) and the filesystem path (_import_media).
- post_feed_service: render thumbnails by image_provenance UNION primary_post_id,
  so a cross-posted image shows on every post (and legacy primary-only images
  still show).
- PostAttachment: per-post uniqueness — drop UNIQUE(sha256), add partial
  UNIQUE(post_id, sha256) + partial UNIQUE(sha256) WHERE post_id IS NULL
  (migration 0043); _capture_attachment dedups per-(post,sha) over the shared
  sha-addressed blob, so no post is left bare.
- cleanup: new prune-bare-posts maintenance action (cleanup_service
  _bare_post_conditions shared by preview/count/delete per preview/apply parity;
  admin endpoint; PostMaintenanceCard). Deletes posts with zero image links
  (primary or provenance) AND zero attachments. Run after the feed fix so a
  hidden provenance link spares the post instead of deleting it.

Tests: dup image shows on both posts; dup attachment shows on both posts; feed
renders provenance-linked duplicates; prune-bare delete-path == preview.

Operator redeploys (migration 0043) then runs the prune to clear the shells.

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

tests/test_cleanup_service.py

@@ -8,7 +8,15 @@ re-read ORM attributes after a service mutates and re-fetches.
 import pytest
 from sqlalchemy import func, select
 
-from backend.app.models import Artist, ImageRecord, Tag, TagKind
+from backend.app.models import (
+    Artist,
+    ImageProvenance,
+    ImageRecord,
+    Post,
+    PostAttachment,
+    Tag,
+    TagKind,
+)
 from backend.app.models.series_chapter import SeriesChapter
 from backend.app.models.series_page import SeriesPage
 from backend.app.models.tag import image_tag
@@ -399,6 +407,39 @@ def test_prune_unused_tags_commit_spares_fandom_and_chaptered_series(
     assert "GenuinelyUnused" not in surviving
 
 
+def test_prune_bare_posts_spares_linked_and_deletes_bare(db_sync, tmp_path):
+    # The LIVE delete (not just the preview) shares _bare_post_conditions, so a
+    # post is spared if it has ANY link — a primary image, a provenance (cross-
+    # posted/duplicate) image, or an attachment — and only the truly-bare shell
+    # is deleted (the empty-post flood, operator-flagged 2026-06-08).
+    a = _make_artist(db_sync, slug="pb")
+    img = _make_image(db_sync, artist=a, path=str(tmp_path / "x.jpg"), sha256="a" * 64)
+
+    bare = Post(artist_id=a.id, external_post_id="bare1")
+    has_primary = Post(artist_id=a.id, external_post_id="prim")
+    has_prov = Post(artist_id=a.id, external_post_id="prov")
+    has_att = Post(artist_id=a.id, external_post_id="att")
+    db_sync.add_all([bare, has_primary, has_prov, has_att])
+    db_sync.flush()
+
+    img.primary_post_id = has_primary.id
+    db_sync.add(ImageProvenance(image_record_id=img.id, post_id=has_prov.id))
+    db_sync.add(PostAttachment(
+        post_id=has_att.id, artist_id=a.id, sha256="b" * 64,
+        path="/store/b", original_filename="f.pdf", ext=".pdf", size_bytes=1,
+    ))
+    db_sync.commit()
+
+    # dry-run count agrees with the delete: only the 1 truly-bare post.
+    assert cleanup_service.prune_bare_posts(db_sync, dry_run=True)["count"] == 1
+    result = cleanup_service.prune_bare_posts(db_sync, dry_run=False)
+    assert result["deleted"] == 1
+
+    surviving = db_sync.execute(select(Post.external_post_id)).scalars().all()
+    assert "bare1" not in surviving
+    assert set(surviving) == {"prim", "prov", "att"}
+
+
 # --- reset_content_tagging ------------------------------------------
 
 

tests/test_importer_attach_in_place.py

@@ -150,6 +150,87 @@ def test_attach_in_place_phash_smaller_skips_without_modifying_disk(
     assert smaller.exists()
 
 
+def test_attach_in_place_duplicate_image_links_both_posts(importer, db_sync):
+    """A byte-identical image appearing in a SECOND post must link to BOTH posts
+    (enrich-on-duplicate) so it shows on every post — not be silently dropped,
+    which left the second post a bare shell. Operator-flagged 2026-06-08."""
+    from backend.app.models import ImageProvenance, Post
+
+    images_root = importer.images_root
+    artist = Artist(name="Hana", slug="hana")
+    db_sync.add(artist)
+    db_sync.flush()
+
+    first = images_root / "hana" / "patreon" / "post_a" / "img.jpg"
+    bytes_ = _make_jpg(first, color=(120, 40, 90))
+    first.with_suffix(first.suffix + ".json").write_text(
+        '{"category": "patreon", "id": "111", "title": "Post 111"}'
+    )
+    r1 = importer.attach_in_place(first, artist=artist)
+    assert r1.status == "imported"
+
+    second = images_root / "hana" / "patreon" / "post_b" / "img.jpg"
+    second.parent.mkdir(parents=True, exist_ok=True)
+    second.write_bytes(bytes_)  # identical bytes → sha256 duplicate
+    second.with_suffix(second.suffix + ".json").write_text(
+        '{"category": "patreon", "id": "222", "title": "Post 222"}'
+    )
+    r2 = importer.attach_in_place(second, artist=artist)
+    assert r2.status == "skipped"
+    assert r2.skip_reason.value == "duplicate_hash"
+    assert r2.image_id == r1.image_id
+
+    linked = db_sync.execute(
+        select(Post.external_post_id)
+        .join(ImageProvenance, ImageProvenance.post_id == Post.id)
+        .where(ImageProvenance.image_record_id == r1.image_id)
+    ).scalars().all()
+    assert set(linked) == {"111", "222"}  # shows on BOTH posts
+
+    primary = db_sync.execute(
+        select(Post.external_post_id)
+        .join(ImageRecord, ImageRecord.primary_post_id == Post.id)
+        .where(ImageRecord.id == r1.image_id)
+    ).scalar_one()
+    assert primary == "111"  # primary stays on the first post
+
+
+def test_attach_in_place_duplicate_attachment_links_both_posts(importer, db_sync):
+    """The same non-art file attached to two posts gets a row on EACH post over a
+    single sha-addressed blob — neither post is left bare. Operator-flagged
+    2026-06-08 (1589 empty Anduo shells from globally-unique attachment sha)."""
+    from backend.app.models import Post, PostAttachment
+
+    images_root = importer.images_root
+    artist = Artist(name="Iris", slug="iris")
+    db_sync.add(artist)
+    db_sync.flush()
+
+    payload = b"the same readme creator attaches to every post"
+    a = images_root / "iris" / "patreon" / "post_a" / "readme.txt"
+    a.parent.mkdir(parents=True, exist_ok=True)
+    a.write_bytes(payload)
+    a.with_suffix(a.suffix + ".json").write_text(
+        '{"category": "patreon", "id": "301", "title": "Post 301"}'
+    )
+    assert importer.attach_in_place(a, artist=artist).status == "attached"
+
+    b = images_root / "iris" / "patreon" / "post_b" / "readme.txt"
+    b.parent.mkdir(parents=True, exist_ok=True)
+    b.write_bytes(payload)
+    b.with_suffix(b.suffix + ".json").write_text(
+        '{"category": "patreon", "id": "302", "title": "Post 302"}'
+    )
+    assert importer.attach_in_place(b, artist=artist).status == "attached"
+
+    linked = db_sync.execute(
+        select(Post.external_post_id)
+        .join(PostAttachment, PostAttachment.post_id == Post.id)
+        .where(PostAttachment.original_filename == "readme.txt")
+    ).scalars().all()
+    assert set(linked) == {"301", "302"}  # one row per post, both present
+
+
 def test_attach_in_place_invalid_image_returns_skipped(importer):
     images_root = importer.images_root
     bad = images_root / "fred" / "bad.jpg"

tests/test_post_feed_service.py

@@ -11,6 +11,7 @@ from sqlalchemy import select
 
 from backend.app.models import (
     Artist,
+    ImageProvenance,
     ImageRecord,
     Post,
     PostAttachment,
@@ -283,6 +284,42 @@ async def test_scroll_thumbnails_show_first_six_with_overflow_count(db):
         assert t["mime"] == "image/jpeg"
 
 
+@pytest.mark.asyncio
+async def test_scroll_thumbnails_include_provenance_linked_duplicates(db):
+    """A cross-posted/duplicate image is linked to a second post via
+    image_provenance (its primary_post_id stays on the first post). The feed
+    renders by provenance UNION primary, so the image shows on BOTH posts —
+    not just its primary. Operator requirement 2026-06-08."""
+    artist = await _seed_artist(db, "alice-prov")
+    src = await _seed_source(db, artist.id, "patreon", "https://p/alice-prov")
+    primary_post = await _seed_post(
+        db, src.id, external_id="PRIMARY", post_date=datetime.now(UTC),
+    )
+    other_post = await _seed_post(
+        db, src.id, external_id="OTHER",
+        post_date=datetime.now(UTC) + timedelta(minutes=1),
+    )
+    img = ImageRecord(
+        path="/images/dup.jpg", sha256="d" * 64, size_bytes=10,
+        mime="image/jpeg", width=10, height=10, origin="downloaded",
+        primary_post_id=primary_post.id, artist_id=artist.id,
+    )
+    db.add(img)
+    await db.flush()
+    # The duplicate links to the OTHER post via provenance only.
+    db.add(ImageProvenance(image_record_id=img.id, post_id=other_post.id))
+    await db.commit()
+
+    page = await PostFeedService(db).scroll(cursor=None, limit=10)
+    by_ext = {it["external_post_id"]: it for it in page["items"]}
+    assert len(by_ext["PRIMARY"]["thumbnails"]) == 1
+    assert len(by_ext["OTHER"]["thumbnails"]) == 1  # shown via provenance
+    assert (
+        by_ext["PRIMARY"]["thumbnails"][0]["image_id"]
+        == by_ext["OTHER"]["thumbnails"][0]["image_id"]
+    )
+
+
 # --- Attachments composition ---------------------------------------------