fix(recapture): link on-disk images to their post (#1288)

Recapture disk-skips already-downloaded media, and upsert_post_record only
writes Post fields — so a pre-existing image (e.g. one pulled under the old
gallery-dl path, imported bare with no post) stays orphaned even after its post
record is (re)written. Confirmed on the operator's instance: 329 pixiv images
with primary_post_id NULL, 694 pixiv posts with content but no linked images, 0
duplicate posts.

Fix: the recapture relink channel now carries the media's post_id (2- → 3-tuple
path/url/post_id), and phase 3 calls importer.link_existing_image_to_post — match
the on-disk image by path, find its Post by (source, external_post_id), upsert
image_provenance + primary_post_id. Factored the provenance-linking out of
_apply_sidecar into a shared _attach_provenance so the fresh-import and
recapture-backlink paths can't diverge. Idempotent; generic across native
platforms (no-op for already-linked Patreon/SubscribeStar). Re-running recapture
now repairs orphaned images; future walks never orphan.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM

tests/test_patreon_ingester.py

@@ -667,10 +667,12 @@ async def test_backfill_skips_already_captured_post_but_recapture_forces_it(
     assert downloader2.post_records == 1
     assert res_recap.files_downloaded == 0          # no media re-download
     assert downloader2.download_calls == 0
-    # The on-disk media is surfaced as a (path, CDN url) relink pair.
+    # The on-disk media is surfaced as a (path, CDN url, post_id) relink triple
+    # (post_id drives the #1288 image→post back-link in phase 3).
     assert len(res_recap.relink_source_paths) == 1
-    rel_path, rel_url = res_recap.relink_source_paths[0]
+    rel_path, rel_url, rel_post_id = res_recap.relink_source_paths[0]
     assert rel_url == m1.url
+    assert rel_post_id == m1.post_id
     # Diagnostic summary surfaces the post-record + relink counts (operator reads
     # this off the event stdout to see what a recapture actually did).
     assert "1 post-record(s), 1 relinked" in res_recap.stdout

tests/test_sidecar_import.py

@@ -158,6 +158,60 @@ def test_relink_source_filehash_no_match_is_noop(importer, import_layout):
     ) is False
 
 
+def test_link_existing_image_to_post_backfills_provenance(importer, import_layout, db_sync):
+    """#1288: an on-disk image imported BARE (no sidecar → no post link, as the
+    pre-cutover gallery-dl pixiv images were) gets linked to its post from the
+    recapture walk's (path, external_post_id) pairing — image_provenance +
+    primary_post_id — without re-importing. Idempotent."""
+    import_root, _ = import_layout
+    m = import_root / "Alice" / "img.jpg"
+    _split(m, "v")
+    rec = importer.session.get(ImageRecord, importer.import_one(m).image_id)
+    assert rec.primary_post_id is None          # bare: no post link yet
+    artist = db_sync.get(Artist, rec.artist_id)
+    src = Source(
+        artist_id=artist.id, platform="pixiv",
+        url="https://www.pixiv.net/users/1", enabled=True,
+    )
+    db_sync.add(src)
+    db_sync.flush()
+    # The post already exists (upsert_post_record wrote it earlier in recapture).
+    post = Post(
+        source_id=src.id, artist_id=artist.id,
+        external_post_id="146132304", post_title="t",
+    )
+    db_sync.add(post)
+    db_sync.commit()
+
+    assert importer.link_existing_image_to_post(
+        Path(rec.path), "146132304", source=src, artist=artist,
+    ) is True
+    importer.session.refresh(rec)
+    assert rec.primary_post_id == post.id
+    prov = db_sync.execute(select(ImageProvenance).where(
+        ImageProvenance.image_record_id == rec.id,
+        ImageProvenance.post_id == post.id,
+    )).scalar_one()
+    assert prov.source_id == src.id
+
+    # Idempotent: a second call re-affirms without duplicating the row.
+    assert importer.link_existing_image_to_post(
+        Path(rec.path), "146132304", source=src, artist=artist,
+    ) is True
+    n = db_sync.execute(select(func.count(ImageProvenance.id)).where(
+        ImageProvenance.image_record_id == rec.id,
+        ImageProvenance.post_id == post.id,
+    )).scalar_one()
+    assert n == 1
+
+
+def test_link_existing_image_to_post_no_match_is_noop(importer):
+    """A path with no ImageRecord, or an empty external id, is a clean no-op."""
+    assert importer.link_existing_image_to_post(
+        Path("/nonexistent/x.jpg"), "999",
+    ) is False
+
+
 def test_reimport_same_post_idempotent(importer, import_layout):
     import_root, _ = import_layout
     # Threshold 0: only an exact phash match collapses. Orthogonal splits