Merge pull request 'v26.05.26.2 — hotfix: alembic 0022 Post-collision pre-merge + ci.yml cache continue-on-error' (#24) from dev into main

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

.forgejo/workflows/ci.yml

@@ -25,6 +25,15 @@ jobs:
       - uses: actions/checkout@v4
 
       - name: Cache pip wheels
+        # continue-on-error: act_runner's cache backend has been broken on
+        # this homelab setup since 2026-05-15 — bolt.db + cache/ dir exist
+        # but the action's JS bundle now fails to load at all
+        # ("Cannot find module .../dist/restore/index.js"), hard-failing
+        # the whole job. The install step that follows handles cold caches
+        # natively (uv pip / pip install both work without it), so net
+        # cost of disabling the cache here is ~30s of wheel downloads per
+        # job. Re-enable once the runner-side cache backend is fixed.
+        continue-on-error: true
         uses: actions/cache@v4
         with:
           path: ~/.cache/pip
@@ -125,6 +134,15 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - name: Cache pip wheels
+        # continue-on-error: act_runner's cache backend has been broken on
+        # this homelab setup since 2026-05-15 — bolt.db + cache/ dir exist
+        # but the action's JS bundle now fails to load at all
+        # ("Cannot find module .../dist/restore/index.js"), hard-failing
+        # the whole job. The install step that follows handles cold caches
+        # natively (uv pip / pip install both work without it), so net
+        # cost of disabling the cache here is ~30s of wheel downloads per
+        # job. Re-enable once the runner-side cache backend is fixed.
+        continue-on-error: true
         uses: actions/cache@v4
         with:
           path: ~/.cache/pip
@@ -190,6 +208,15 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - name: Cache pip wheels
+        # continue-on-error: act_runner's cache backend has been broken on
+        # this homelab setup since 2026-05-15 — bolt.db + cache/ dir exist
+        # but the action's JS bundle now fails to load at all
+        # ("Cannot find module .../dist/restore/index.js"), hard-failing
+        # the whole job. The install step that follows handles cold caches
+        # natively (uv pip / pip install both work without it), so net
+        # cost of disabling the cache here is ~30s of wheel downloads per
+        # job. Re-enable once the runner-side cache backend is fixed.
+        continue-on-error: true
         uses: actions/cache@v4
         with:
           path: ~/.cache/pip
@@ -255,6 +282,15 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - name: Cache pip wheels
+        # continue-on-error: act_runner's cache backend has been broken on
+        # this homelab setup since 2026-05-15 — bolt.db + cache/ dir exist
+        # but the action's JS bundle now fails to load at all
+        # ("Cannot find module .../dist/restore/index.js"), hard-failing
+        # the whole job. The install step that follows handles cold caches
+        # natively (uv pip / pip install both work without it), so net
+        # cost of disabling the cache here is ~30s of wheel downloads per
+        # job. Re-enable once the runner-side cache backend is fixed.
+        continue-on-error: true
         uses: actions/cache@v4
         with:
           path: ~/.cache/pip

alembic/versions/0021_image_provenance_unique.py

@@ -0,0 +1,54 @@
+"""provenance-race: dedupe + UNIQUE(image_record_id, post_id) on image_provenance
+
+Revision ID: 0021
+Revises: 0020
+Create Date: 2026-05-26
+
+Closes the race in Importer._apply_sidecar's existence-check + INSERT pattern.
+Two workers writing for the same (image, post) pair both saw no existing row
+and both inserted, leaving duplicates that then broke .scalar_one_or_none()
+on every subsequent deep-scan rederive against those images
+(MultipleResultsFound). Most plausibly seeded when the 5-min recovery sweep
+re-enqueued a still-running long-import task and the second worker collided
+with the first inside _apply_sidecar.
+
+Migration steps:
+  1. DELETE all but min(id) per (image_record_id, post_id) pair. Operator's
+     DB had 2 affected pairs at write-time; harmless no-op if zero.
+  2. Add UNIQUE constraint so the importer's new savepoint+IntegrityError
+     recovery path can trip on collision and re-select, mirroring
+     uq_source_artist_platform_url and uq_post_source_external_id.
+"""
+from typing import Sequence, Union
+
+from alembic import op
+
+revision: str = "0021"
+down_revision: Union[str, None] = "0020"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    op.execute(
+        """
+        DELETE FROM image_provenance ip1
+        USING image_provenance ip2
+        WHERE ip1.image_record_id = ip2.image_record_id
+          AND ip1.post_id = ip2.post_id
+          AND ip1.id > ip2.id
+        """
+    )
+    op.create_unique_constraint(
+        "uq_image_provenance_image_post",
+        "image_provenance",
+        ["image_record_id", "post_id"],
+    )
+
+
+def downgrade() -> None:
+    op.drop_constraint(
+        "uq_image_provenance_image_post",
+        "image_provenance",
+        type_="unique",
+    )

alembic/versions/0022_source_per_artist_platform.py

@@ -0,0 +1,198 @@
+"""source-collapse: one Source per (artist, platform) — consolidate junk per-post Sources
+
+Revision ID: 0022
+Revises: 0021
+Create Date: 2026-05-26
+
+Closes the operator-flagged 2026-05-26 issue where the filesystem importer
+called _find_or_create_source(url=sd.post_url), creating one Source row per
+imported post URL. Operator's Atole artist had 406 Source rows where there
+should have been 1 (the /cw/Atole subscription Source).
+
+Source represents a subscription feed (one per artist+platform — the
+gallery-dl URL polled by the FC-3 downloader). Posts hang off it. The
+filesystem importer was misusing Source as a per-post key.
+
+Migration steps per (artist_id, platform) group with >1 Source:
+  1. Pick canonical — prefer a URL NOT matching '/posts/<id>$' (real
+     campaign URL like /cw/Atole); else min(id).
+  2. PRE-merge any Posts under non-canonical sources whose
+     external_post_id ALREADY exists under the canonical source. (Same
+     gallery-dl post imported via two different sidecar paths can plant
+     two Post rows with identical external_post_id under different
+     Sources for the same artist.) Repoint ImageProvenance +
+     ImageRecord.primary_post_id to the canonical-side Post, dedupe
+     ImageProvenance against alembic 0021's uq, then delete the
+     non-canonical-side Post. This MUST happen before step 3 — Postgres
+     fires uq_post_source_external_id row-by-row during the bulk UPDATE
+     and the merge-after-reparent ordering 500s on first collision
+     (operator-hit during v26.05.26.1 deploy, 2026-05-26).
+  3. Reparent remaining Posts onto canonical (no collisions possible now).
+  4. Reparent ImageProvenance.source_id off the non-canonical sources.
+  5. Delete the orphan Source rows.
+  6. If the canonical Source's URL still looks like a per-post URL (no
+     campaign URL existed among candidates), rewrite it to
+     'sidecar:<platform>:<artist_slug>' so the artist detail page shows
+     something readable.
+"""
+from typing import Sequence, Union
+
+from alembic import op
+from sqlalchemy import text
+
+revision: str = "0022"
+down_revision: Union[str, None] = "0021"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+_POST_URL_RE = r"/posts/[^/]+$"
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+
+    # Find (artist_id, platform) groups with > 1 Source row.
+    groups = conn.execute(text("""
+        SELECT artist_id, platform
+        FROM source
+        GROUP BY artist_id, platform
+        HAVING COUNT(*) > 1
+    """)).fetchall()
+
+    for artist_id, platform in groups:
+        rows = conn.execute(
+            text("""
+                SELECT id, url FROM source
+                WHERE artist_id = :a AND platform = :p
+                ORDER BY id ASC
+            """),
+            {"a": artist_id, "p": platform},
+        ).fetchall()
+
+        # Canonical: first row whose URL doesn't look like a per-post URL;
+        # else min(id).
+        canonical_id = None
+        for sid, url in rows:
+            if not _matches_post_url(url):
+                canonical_id = sid
+                break
+        if canonical_id is None:
+            canonical_id = rows[0][0]
+
+        other_ids = [sid for sid, _ in rows if sid != canonical_id]
+        if not other_ids:
+            continue
+
+        # STEP 2: PRE-merge colliding Posts BEFORE the bulk reparent.
+        # Find pairs (keep, drop) where:
+        #   keep  = a Post under the canonical source
+        #   drop  = a Post under one of the non-canonical sources whose
+        #           external_post_id equals keep.external_post_id
+        # If we let the bulk UPDATE try to repoint drop onto canonical,
+        # uq_post_source_external_id fires row-by-row and aborts the
+        # whole migration.
+        colliding = conn.execute(
+            text("""
+                SELECT keep.id AS keep_id, drop_.id AS drop_id
+                FROM post AS keep
+                JOIN post AS drop_
+                  ON drop_.external_post_id = keep.external_post_id
+                 AND drop_.id != keep.id
+                WHERE keep.source_id = :canonical
+                  AND drop_.source_id = ANY(:others)
+            """),
+            {"canonical": canonical_id, "others": other_ids},
+        ).fetchall()
+        for keep_id, drop_id in colliding:
+            conn.execute(
+                text("""
+                    UPDATE image_provenance SET post_id = :keep
+                    WHERE post_id = :drop_
+                """),
+                {"keep": keep_id, "drop_": drop_id},
+            )
+            conn.execute(
+                text("""
+                    UPDATE image_record SET primary_post_id = :keep
+                    WHERE primary_post_id = :drop_
+                """),
+                {"keep": keep_id, "drop_": drop_id},
+            )
+            # Repointed provenance may now collide on
+            # uq_image_provenance_image_post (alembic 0021). Dedupe:
+            # keep min(id) per (image_record_id, post_id).
+            conn.execute(text("""
+                DELETE FROM image_provenance ip1
+                USING image_provenance ip2
+                WHERE ip1.image_record_id = ip2.image_record_id
+                  AND ip1.post_id = ip2.post_id
+                  AND ip1.id > ip2.id
+            """))
+            conn.execute(
+                text("DELETE FROM post WHERE id = :drop_"),
+                {"drop_": drop_id},
+            )
+
+        # STEP 3: Bulk reparent the remaining Posts off the other
+        # Sources. After step 2, no collisions on
+        # (canonical, external_post_id) are possible.
+        conn.execute(
+            text("""
+                UPDATE post SET source_id = :canonical
+                WHERE source_id = ANY(:others)
+            """),
+            {"canonical": canonical_id, "others": other_ids},
+        )
+
+        # STEP 4: Reparent ImageProvenance.source_id (denormalized FK).
+        # No UNIQUE on source_id; safe bulk update.
+        conn.execute(
+            text("""
+                UPDATE image_provenance SET source_id = :canonical
+                WHERE source_id = ANY(:others)
+            """),
+            {"canonical": canonical_id, "others": other_ids},
+        )
+
+        # STEP 5: Drop the orphan Sources.
+        conn.execute(
+            text("DELETE FROM source WHERE id = ANY(:others)"),
+            {"others": other_ids},
+        )
+
+        # If the canonical's URL still looks per-post (no campaign URL
+        # existed among the candidates), rewrite to a synthetic anchor so
+        # the artist detail page renders something readable.
+        canonical_url = conn.execute(
+            text("SELECT url FROM source WHERE id = :id"),
+            {"id": canonical_id},
+        ).scalar_one()
+        if _matches_post_url(canonical_url):
+            slug = conn.execute(
+                text("SELECT slug FROM artist WHERE id = :id"),
+                {"id": artist_id},
+            ).scalar_one()
+            conn.execute(
+                text("""
+                    UPDATE source
+                    SET url = :new_url, enabled = false
+                    WHERE id = :id
+                """),
+                {
+                    "id": canonical_id,
+                    "new_url": f"sidecar:{platform}:{slug}",
+                },
+            )
+
+
+def downgrade() -> None:
+    # Lossy migration — orphan Sources deleted, Posts reparented, Posts
+    # merged. No safe downgrade. If you need to roll back the schema
+    # invariant, fork from 0021 and re-run filesystem imports.
+    pass
+
+
+def _matches_post_url(url: str) -> bool:
+    """True if url ends with /posts/<token> (gallery-dl-style per-post URL)."""
+    import re
+    return bool(re.search(_POST_URL_RE, url or ""))

backend/app/api/__init__.py

@@ -38,6 +38,7 @@ def all_blueprints() -> list[Blueprint]:
     from .system_activity import system_activity_bp
     from .system_backup import system_backup_bp
     from .tags import tags_bp
+    from .thumbnails import thumbnails_bp
     return [
         api_bp,
         attachments_bp,
@@ -58,6 +59,7 @@ def all_blueprints() -> list[Blueprint]:
         allowlist_bp,
         aliases_bp,
         ml_admin_bp,
+        thumbnails_bp,
         sources_bp,
         platforms_bp,
         posts_bp,

backend/app/api/import_admin.py

@@ -103,17 +103,22 @@ async def list_tasks():
 
 @import_admin_bp.route("/retry-failed", methods=["POST"])
 async def retry_failed():
+    # Fold SELECT into UPDATE…WHERE…RETURNING — the prior SELECT-then-
+    # UPDATE-WHERE-id-IN pattern blew past psycopg's 65535-parameter
+    # ceiling once failed_ids exceeded ~65k rows.
     async with get_session() as session:
-        failed_ids = (
+        result = await session.execute(
-            await session.execute(select(ImportTask.id).where(ImportTask.status == "failed"))
+            update(ImportTask)
-        ).scalars().all()
+            .where(ImportTask.status == "failed")
+            .values(
+                status="queued", error=None,
+                started_at=None, finished_at=None,
+            )
+            .returning(ImportTask.id)
+        )
+        failed_ids = [row[0] for row in result.all()]
         if not failed_ids:
             return jsonify({"retried": 0})
-        await session.execute(
-            update(ImportTask)
-            .where(ImportTask.id.in_(failed_ids))
-            .values(status="queued", error=None, started_at=None, finished_at=None)
-        )
         await session.commit()
 
     from ..tasks.import_file import import_media_file
@@ -138,28 +143,26 @@ async def clear_stuck():
     autoretry-looped for 2 days after a corrupt-data PIL OSError.
     """
     async with get_session() as session:
-        stuck_ids = (
+        # Fold SELECT into UPDATE…WHERE — see /retry-failed for the
-            await session.execute(
+        # 65535-parameter ceiling rationale. rowcount is enough here
-                select(ImportTask.id).where(
+        # because we don't need the ids afterward (no .delay()).
-                    ImportTask.status.in_(["pending", "queued", "processing"])
+        clear_result = await session.execute(
-                )
+            update(ImportTask)
+            .where(
+                ImportTask.status.in_(["pending", "queued", "processing"])
             )
-        ).scalars().all()
+            .values(
-        if stuck_ids:
+                status="failed",
-            await session.execute(
+                finished_at=datetime.now(UTC),
-                update(ImportTask)
+                error=(
-                .where(ImportTask.id.in_(stuck_ids))
+                    "manually cleared via /api/import/clear-stuck "
-                .values(
+                    "— stuck in non-terminal state; retry once "
-                    status="failed",
+                    "underlying cause (corrupt file, missing model, "
-                    finished_at=datetime.now(UTC),
+                    "etc.) is resolved"
-                    error=(
+                ),
-                        "manually cleared via /api/import/clear-stuck "
-                        "— stuck in non-terminal state; retry once "
-                        "underlying cause (corrupt file, missing model, "
-                        "etc.) is resolved"
-                    ),
-                )
             )
+        )
+        tasks_failed = clear_result.rowcount or 0
 
         # Finalize any 'running' ImportBatch that no longer has any
         # active children. The "Scanning..." banner is driven by
@@ -195,7 +198,7 @@ async def clear_stuck():
         await session.commit()
 
     return jsonify({
-        "tasks_failed": len(stuck_ids),
+        "tasks_failed": tasks_failed,
         "batches_finalized": finalized_batches,
     })
 

backend/app/api/thumbnails.py

@@ -0,0 +1,13 @@
+"""Thumbnail admin API: backfill trigger."""
+
+from quart import Blueprint, jsonify
+
+thumbnails_bp = Blueprint("thumbnails", __name__, url_prefix="/api/thumbnails")
+
+
+@thumbnails_bp.route("/backfill", methods=["POST"])
+async def trigger_backfill():
+    from ..tasks.thumbnail import backfill_thumbnails
+
+    r = backfill_thumbnails.delay()
+    return jsonify({"celery_task_id": r.id}), 202

backend/app/models/image_provenance.py

@@ -1,14 +1,18 @@
 """ImageProvenance — links an ImageRecord to a Post.
 
-Many-to-one (one image, many provenance rows) enables the enrich-on-duplicate
+One image can have many provenance rows — different posts each contribute
-rule (spec §3): when a downloaded image is a pHash dupe of an existing
+metadata (enrich-on-duplicate rule, spec §3: a downloaded image that is a
-record, we append a new provenance row to the existing record rather than
+pHash dupe of an existing record gets a NEW provenance row for the new post
-dropping the metadata.
+appended, rather than the metadata being dropped). But the (image, post)
+pair is unique — alembic 0021 enforces uq_image_provenance_image_post
+after operator-flagged 2026-05-26 saw _apply_sidecar's existence-check +
+INSERT race plant duplicates that then broke .scalar_one_or_none() on
+every later deep-scan rederive (MultipleResultsFound).
 """
 
 from datetime import datetime
 
-from sqlalchemy import JSON, DateTime, ForeignKey, Integer, func
+from sqlalchemy import JSON, DateTime, ForeignKey, Integer, UniqueConstraint, func
 from sqlalchemy.orm import Mapped, mapped_column
 
 from .base import Base
@@ -16,6 +20,12 @@ from .base import Base
 
 class ImageProvenance(Base):
     __tablename__ = "image_provenance"
+    __table_args__ = (
+        UniqueConstraint(
+            "image_record_id", "post_id",
+            name="uq_image_provenance_image_post",
+        ),
+    )
 
     id: Mapped[int] = mapped_column(Integer, primary_key=True)
     image_record_id: Mapped[int] = mapped_column(

backend/app/services/artist_service.py

@@ -111,12 +111,22 @@ class ArtistService:
             )
         ).all()
 
+        post_count = (
+            await self.session.execute(
+                select(func.count(func.distinct(Post.id)))
+                .select_from(Post)
+                .join(Source, Source.id == Post.source_id)
+                .where(Source.artist_id == aid)
+            )
+        ).scalar_one()
+
         return {
             "id": artist.id,
             "name": artist.name,
             "slug": artist.slug,
             "is_subscription": bool(artist.is_subscription),
             "image_count": int(image_count),
+            "post_count": int(post_count),
             "date_range": {
                 "min": dmin.isoformat() if dmin else None,
                 "max": dmax.isoformat() if dmax else None,

backend/app/services/extension_service.py

@@ -11,6 +11,7 @@ from __future__ import annotations
 import re
 
 from sqlalchemy import select
+from sqlalchemy.exc import IntegrityError
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from ..models import Artist, Source
@@ -97,20 +98,40 @@ class ExtensionService:
         raise UnknownPlatformError(f"no platform pattern matched {url!r}")
 
     async def _find_or_create_artist(self, raw_name: str) -> tuple[Artist, bool]:
+        """Race-safe find-or-create on Artist by slug. Mirrors the
+        savepoint + IntegrityError recovery pattern used in
+        Importer._find_or_create_source/post (see
+        reference_scalar_one_or_none_duplicates memory). Without this,
+        two concurrent quick-add-source calls hitting the same artist
+        would both miss the existence check and the second INSERT would
+        500 against uq_artist_slug.
+        """
         slug = slugify(raw_name)
         existing = (await self.session.execute(
             select(Artist).where(Artist.slug == slug)
         )).scalar_one_or_none()
         if existing is not None:
             return existing, False
-        artist = Artist(name=raw_name, slug=slug, is_subscription=True)
+        sp = await self.session.begin_nested()
-        self.session.add(artist)
+        try:
-        await self.session.flush()
+            artist = Artist(name=raw_name, slug=slug, is_subscription=True)
-        return artist, True
+            self.session.add(artist)
+            await self.session.flush()
+            await sp.commit()
+            return artist, True
+        except IntegrityError:
+            await sp.rollback()
+            recovered = (await self.session.execute(
+                select(Artist).where(Artist.slug == slug)
+            )).scalar_one()
+            return recovered, False
 
     async def _find_or_create_source(
         self, *, artist_id: int, platform: str, url: str,
     ) -> tuple[Source, bool]:
+        """Race-safe — same pattern as _find_or_create_artist above. The
+        uq_source_artist_platform_url constraint catches the duplicate
+        insert; we roll the savepoint back and re-select."""
         existing = (await self.session.execute(
             select(Source).where(
                 Source.artist_id == artist_id,
@@ -120,8 +141,24 @@ class ExtensionService:
         )).scalar_one_or_none()
         if existing is not None:
             return existing, False
-        src = Source(artist_id=artist_id, platform=platform, url=url, enabled=True)
+        sp = await self.session.begin_nested()
-        self.session.add(src)
+        try:
-        await self.session.flush()
+            src = Source(
+                artist_id=artist_id, platform=platform,
+                url=url, enabled=True,
+            )
+            self.session.add(src)
+            await self.session.flush()
+            await sp.commit()
+        except IntegrityError:
+            await sp.rollback()
+            recovered = (await self.session.execute(
+                select(Source).where(
+                    Source.artist_id == artist_id,
+                    Source.platform == platform,
+                    Source.url == url,
+                )
+            )).scalar_one()
+            return recovered, False
         await self.session.commit()
         return src, True

backend/app/services/importer.py

@@ -247,6 +247,62 @@ class Importer:
                 )
             ).scalar_one()
 
+    def _source_for_sidecar(
+        self, *, artist_id: int, platform: str, artist_slug: str,
+    ) -> Source:
+        """Filesystem-import sidecar Source resolver.
+
+        Source represents a subscription feed (one per artist+platform — the
+        gallery-dl URL polled by the FC-3 downloader). The filesystem importer
+        used to call _find_or_create_source(url=sd.post_url), which created
+        one Source row per post URL — 100s of junk Sources per artist, all
+        with enabled=True, polluting the artist detail page and tricking the
+        subscription checker into trying to poll patreon post URLs as feeds.
+        Operator-flagged 2026-05-26.
+
+        New behaviour: if any Source row exists for (artist_id, platform),
+        reuse it regardless of its URL — the artist's real subscription Source
+        (created by the downloader / extension / UI) is the canonical
+        attachment point for filesystem-imported posts. If none exists, create
+        ONE synthetic anchor with url='sidecar:<platform>:<artist_slug>' and
+        enabled=False (so the subscription checker doesn't poll it).
+        """
+        existing = self.session.execute(
+            select(Source)
+            .where(
+                Source.artist_id == artist_id,
+                Source.platform == platform,
+            )
+            .order_by(Source.id.asc())
+            .limit(1)
+        ).scalar_one_or_none()
+        if existing is not None:
+            return existing
+        synthetic_url = f"sidecar:{platform}:{artist_slug}"
+        sp = self.session.begin_nested()
+        try:
+            row = Source(
+                artist_id=artist_id,
+                platform=platform,
+                url=synthetic_url,
+                enabled=False,
+            )
+            self.session.add(row)
+            self.session.flush()
+            sp.commit()
+            return row
+        except IntegrityError:
+            sp.rollback()
+            return self.session.execute(
+                select(Source)
+                .where(
+                    Source.artist_id == artist_id,
+                    Source.platform == platform,
+                )
+                .order_by(Source.id.asc())
+                .limit(1)
+            ).scalar_one()
+
     def _find_or_create_post(
         self, *, source_id: int, external_post_id: str,
     ) -> Post:
@@ -315,9 +371,8 @@ class Importer:
             return None
         sd = parse_sidecar(data)
         platform = sd.platform or "unknown"
-        url = sd.post_url or f"sidecar:{platform}"
+        src = self._source_for_sidecar(
-        src = self._find_or_create_source(
+            artist_id=artist.id, platform=platform, artist_slug=artist.slug,
-            artist_id=artist.id, platform=platform, url=url,
         )
         epid = sd.external_post_id or sc.stem
         return self._find_or_create_post(
@@ -763,9 +818,9 @@ class Importer:
             src = explicit_source
         else:
             platform = sd.platform or "unknown"
-            url = sd.post_url or f"sidecar:{platform}"
+            src = self._source_for_sidecar(
-            src = self._find_or_create_source(
+                artist_id=artist.id, platform=platform,
-                artist_id=artist.id, platform=platform, url=url,
+                artist_slug=artist.slug,
             )
 
         epid = sd.external_post_id or sc.stem
@@ -784,6 +839,15 @@ class Importer:
             post.attachment_count = sd.attachment_count
         post.raw_metadata = sd.raw
 
+        # Race-safe (image_record_id, post_id) upsert — mirrors the
+        # _find_or_create_source/post savepoint pattern. The plain
+        # SELECT-then-INSERT pattern lost a race when two workers ran
+        # _apply_sidecar on the same (image, post) pair (e.g. the 5-min
+        # recovery sweep re-enqueued a still-running long import), planting
+        # duplicates that then broke .scalar_one_or_none() on every later
+        # deep-scan rederive (MultipleResultsFound). Alembic 0021 adds the
+        # uq_image_provenance_image_post UNIQUE so this savepoint actually
+        # trips on collision.
         exists = self.session.execute(
             select(ImageProvenance.id).where(
                 ImageProvenance.image_record_id == record.id,
@@ -791,14 +855,20 @@ class Importer:
             )
         ).scalar_one_or_none()
         if exists is None:
-            self.session.add(
+            sp = self.session.begin_nested()
-                ImageProvenance(
+            try:
-                    image_record_id=record.id,
+                self.session.add(
-                    post_id=post.id,
+                    ImageProvenance(
-                    source_id=src.id,
+                        image_record_id=record.id,
-                    captured_metadata=sd.raw,
+                        post_id=post.id,
+                        source_id=src.id,
+                        captured_metadata=sd.raw,
+                    )
                 )
-            )
+                self.session.flush()
+                sp.commit()
+            except IntegrityError:
+                sp.rollback()
         if record.primary_post_id is None:
             record.primary_post_id = post.id
         self.session.flush()

backend/app/tasks/maintenance.py

@@ -54,41 +54,40 @@ def recover_interrupted_tasks() -> int:
     processing_cutoff = now - timedelta(minutes=STUCK_THRESHOLD_MINUTES)
     orphan_cutoff = now - timedelta(minutes=ORPHAN_PENDING_THRESHOLD_MINUTES)
     with SessionLocal() as session:
-        stuck_ids = session.execute(
+        # Both sweeps used to be SELECT ids → UPDATE WHERE id IN (...) which
-            select(ImportTask.id)
+        # blew past psycopg's 65535-parameter ceiling once a sweep covered
+        # tens of thousands of rows (operator hit it 2026-05-26 after the
+        # /import deep scan piled up orphans). Folding the SELECT into the
+        # UPDATE eliminates the IN-list entirely. RETURNING gives us back
+        # exactly the ids that flipped so the stuck sweep can still
+        # .delay() each one.
+        stuck_result = session.execute(
+            update(ImportTask)
             .where(ImportTask.status == "processing")
             .where(ImportTask.started_at < processing_cutoff)
-        ).scalars().all()
+            .values(
+                status="queued",
+                started_at=None,
+                error="recovered from stuck state",
+            )
+            .returning(ImportTask.id)
+        )
+        stuck_ids = [row[0] for row in stuck_result.all()]
 
-        orphan_ids = session.execute(
+        orphan_result = session.execute(
-            select(ImportTask.id)
+            update(ImportTask)
             .where(ImportTask.status.in_(["pending", "queued"]))
             .where(ImportTask.created_at < orphan_cutoff)
-        ).scalars().all()
+            .values(
-
+                status="failed",
-        if not stuck_ids and not orphan_ids:
+                error=(
-            return 0
+                    "orphan pending/queued swept by recover_interrupted_tasks "
-
+                    "(scanner likely crashed mid-enqueue); retry via "
-        if stuck_ids:
+                    "/api/import/retry-failed"
-            session.execute(
+                ),
-                update(ImportTask)
-                .where(ImportTask.id.in_(stuck_ids))
-                .values(status="queued", started_at=None, error="recovered from stuck state")
-            )
-
-        if orphan_ids:
-            session.execute(
-                update(ImportTask)
-                .where(ImportTask.id.in_(orphan_ids))
-                .values(
-                    status="failed",
-                    error=(
-                        "orphan pending/queued swept by recover_interrupted_tasks "
-                        "(scanner likely crashed mid-enqueue); retry via "
-                        "/api/import/retry-failed"
-                    ),
-                )
             )
+        )
+        orphan_count = orphan_result.rowcount or 0
 
         session.commit()
 
@@ -97,7 +96,7 @@ def recover_interrupted_tasks() -> int:
             for tid in stuck_ids:
                 import_media_file.delay(tid)
 
-        return len(stuck_ids) + len(orphan_ids)
+        return len(stuck_ids) + orphan_count
 
 
 @celery.task(name="backend.app.tasks.maintenance.cleanup_old_tasks")

backend/app/tasks/thumbnail.py

@@ -17,6 +17,30 @@ from ._sync_engine import sync_session_factory as _sync_session_factory
 
 IMAGES_ROOT = Path("/images")
 
+THUMB_MAGIC_JPEG = b"\xff\xd8\xff"
+THUMB_MAGIC_PNG = b"\x89PNG\r\n\x1a\n"
+
+
+def _thumb_is_valid(path: Path) -> bool:
+    """Return True iff `path` exists and starts with a JPEG or PNG magic header.
+
+    The on-disk thumbnail format is set by services/thumbnailer.py — JPEG for
+    opaque sources, PNG for alpha sources. Anything else (missing file, OSError,
+    truncated, wrong magic) is invalid.
+    """
+    try:
+        with path.open("rb") as f:
+            head = f.read(12)
+    except OSError:
+        return False
+    if len(head) < 8:
+        return False
+    if head[:3] == THUMB_MAGIC_JPEG:
+        return True
+    if head[:8] == THUMB_MAGIC_PNG:
+        return True
+    return False
+
 
 @celery.task(
     name="backend.app.tasks.thumbnail.generate_thumbnail",
@@ -50,3 +74,58 @@ def generate_thumbnail(self, image_id: int) -> dict:
         session.add(record)
         session.commit()
         return {"status": "ok", "image_id": image_id, "path": str(result.path)}
+
+
+@celery.task(
+    name="backend.app.tasks.thumbnail.backfill_thumbnails",
+    bind=True,
+)
+def backfill_thumbnails(self) -> dict:
+    """Scan ImageRecord and enqueue generate_thumbnail for rows whose
+    thumbnail is missing, gone from disk, or has wrong magic bytes.
+
+    Keyset paginates by id ASC, page size 500. NULLs out thumbnail_path for
+    rows that point at a missing or corrupt file before enqueueing — keeps
+    the DB self-consistent on partial runs and makes re-runs safe.
+
+    Returns {"enqueued": N, "ok": M, "regenerated": K} where:
+      - enqueued = total generate_thumbnail.delay() calls
+      - ok       = rows whose existing thumbnail file is valid (skipped)
+      - regenerated = subset of enqueued that had a non-NULL thumbnail_path
+                      cleared (i.e. missing + corrupt)
+    """
+    from sqlalchemy import select, update
+
+    SessionLocal = _sync_session_factory()
+    enqueued = 0
+    ok = 0
+    regenerated = 0
+    last_id = 0
+    with SessionLocal() as session:
+        while True:
+            rows = session.execute(
+                select(ImageRecord.id, ImageRecord.thumbnail_path)
+                .where(ImageRecord.id > last_id)
+                .order_by(ImageRecord.id.asc())
+                .limit(500)
+            ).all()
+            if not rows:
+                break
+            for image_id, thumb_path in rows:
+                if thumb_path is None:
+                    generate_thumbnail.delay(image_id)
+                    enqueued += 1
+                elif _thumb_is_valid(Path(thumb_path)):
+                    ok += 1
+                else:
+                    session.execute(
+                        update(ImageRecord)
+                        .where(ImageRecord.id == image_id)
+                        .values(thumbnail_path=None)
+                    )
+                    generate_thumbnail.delay(image_id)
+                    enqueued += 1
+                    regenerated += 1
+            session.commit()
+            last_id = rows[-1][0]
+    return {"enqueued": enqueued, "ok": ok, "regenerated": regenerated}

frontend/src/components/artist/ArtistGalleryTab.vue

@@ -0,0 +1,48 @@
+<template>
+  <div class="fc-artist-gallery">
+    <MasonryGrid
+      :items="store.images"
+      :loading="store.imagesLoading"
+      :has-more="store.hasMoreImages"
+      @load-more="store.loadMoreImages(props.slug)"
+      @open="openImage"
+    />
+  </div>
+</template>
+
+<script setup>
+import { onMounted, watch } from 'vue'
+import { useRoute, useRouter } from 'vue-router'
+
+import { useArtistStore } from '../../stores/artist.js'
+import { useModalStore } from '../../stores/modal.js'
+import MasonryGrid from '../discovery/MasonryGrid.vue'
+
+const props = defineProps({
+  slug: { type: String, required: true },
+})
+
+const store = useArtistStore()
+const modal = useModalStore()
+const route = useRoute()
+const router = useRouter()
+
+onMounted(() => {
+  const initial = parseInt(route.query.image, 10)
+  if (!isNaN(initial)) modal.open(initial)
+})
+
+watch(() => route.query.image, (q) => {
+  const id = parseInt(q, 10)
+  if (!isNaN(id) && id !== modal.currentImageId) modal.open(id)
+  else if (isNaN(id) && modal.currentImageId !== null) modal.close()
+})
+
+function openImage (id) {
+  router.push({ query: { ...route.query, image: id } })
+}
+</script>
+
+<style scoped>
+.fc-artist-gallery { min-width: 0; }
+</style>

frontend/src/components/artist/ArtistHeader.vue

@@ -0,0 +1,116 @@
+<template>
+  <header class="fc-artist-header">
+    <div class="fc-artist-header__left">
+      <h1 class="fc-artist-header__name">{{ name }}</h1>
+      <span v-if="stats" class="fc-artist-header__stats">{{ stats }}</span>
+    </div>
+    <v-tabs
+      :model-value="modelValue"
+      color="accent"
+      density="compact"
+      class="fc-artist-header__tabs"
+      @update:model-value="$emit('update:modelValue', $event)"
+    >
+      <v-tab value="posts">
+        Posts
+        <span v-if="postCount != null" class="fc-artist-header__tab-count">
+          ({{ postCount }})
+        </span>
+      </v-tab>
+      <v-tab value="gallery">
+        Gallery
+        <span v-if="imageCount != null" class="fc-artist-header__tab-count">
+          ({{ imageCount }})
+        </span>
+      </v-tab>
+      <v-tab value="management">Management</v-tab>
+    </v-tabs>
+  </header>
+</template>
+
+<script setup>
+import { computed } from 'vue'
+
+const props = defineProps({
+  name:        { type: String, required: true },
+  imageCount:  { type: Number, default: null },
+  postCount:   { type: Number, default: null },
+  lastAdded:   { type: String, default: null },
+  modelValue:  { type: String, required: true },
+})
+
+defineEmits(['update:modelValue'])
+
+const stats = computed(() => {
+  const parts = []
+  if (props.imageCount != null) {
+    parts.push(`${props.imageCount} image${props.imageCount === 1 ? '' : 's'}`)
+  }
+  if (props.lastAdded) {
+    parts.push(`last added ${props.lastAdded.slice(0, 10)}`)
+  }
+  return parts.join(' · ')
+})
+</script>
+
+<style scoped>
+/* Matches TopNav.vue's frosted recipe exactly — top:64px parks it under
+   the 64px-tall TopNav with no visible seam. */
+.fc-artist-header {
+  position: sticky;
+  top: 64px;
+  z-index: 4;
+  display: flex;
+  align-items: center;
+  gap: 1rem;
+  padding: 0.5rem 1rem;
+  background: linear-gradient(
+    to bottom,
+    rgba(20, 23, 26, 0.92) 0%,
+    rgba(20, 23, 26, 0.65) 60%,
+    rgba(20, 23, 26, 0) 100%
+  );
+  backdrop-filter: blur(2px);
+  -webkit-backdrop-filter: blur(2px);
+}
+
+.fc-artist-header__left {
+  flex: 1 1 0;
+  min-width: 0;
+  display: flex;
+  align-items: baseline;
+  gap: 12px;
+  overflow: hidden;
+}
+
+.fc-artist-header__name {
+  font-family: 'Fraunces', Georgia, serif;
+  font-size: 24px;
+  font-weight: 500;
+  margin: 0;
+  color: rgb(var(--v-theme-on-surface));
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
+}
+
+.fc-artist-header__stats {
+  font-size: 13px;
+  color: rgb(var(--v-theme-on-surface-variant));
+  font-variant-numeric: tabular-nums;
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
+}
+
+.fc-artist-header__tabs {
+  flex: 0 0 auto;
+}
+
+.fc-artist-header__tab-count {
+  margin-left: 4px;
+  font-size: 12px;
+  color: rgb(var(--v-theme-on-surface-variant));
+  font-variant-numeric: tabular-nums;
+}
+</style>

frontend/src/components/artist/ArtistManagementTab.vue

@@ -0,0 +1,133 @@
+<template>
+  <div class="fc-artist-mgmt">
+    <section class="fc-artist-mgmt__sec">
+      <h2 class="fc-h2">Overview</h2>
+      <div class="fc-artist-mgmt__chips">
+        <v-chip
+          size="small"
+          :variant="overview.is_subscription ? 'flat' : 'outlined'"
+          :color="overview.is_subscription ? 'accent' : undefined"
+          prepend-icon="mdi-rss"
+        >{{ overview.is_subscription ? 'Subscription' : 'One-off' }}</v-chip>
+        <v-chip
+          size="small" variant="outlined" prepend-icon="mdi-link-variant"
+          :to="`/subscriptions?artist_id=${overview.id}`"
+        >{{ overview.sources.length }} subscription{{ overview.sources.length === 1 ? '' : 's' }}</v-chip>
+      </div>
+    </section>
+
+    <section v-if="overview.cooccurring_tags.length" class="fc-artist-mgmt__sec">
+      <h2 class="fc-h2">Frequent tags</h2>
+      <div class="fc-artist-mgmt__tags">
+        <v-chip
+          v-for="t in overview.cooccurring_tags" :key="t.id"
+          size="small" @click="openTag(t.id)"
+        >{{ t.name }} <span class="fc-artist-mgmt__tagc">{{ t.count }}</span></v-chip>
+      </div>
+    </section>
+
+    <section v-if="overview.activity.length" class="fc-artist-mgmt__sec">
+      <h2 class="fc-h2">Activity</h2>
+      <svg
+        class="fc-artist-mgmt__spark" :viewBox="`0 0 ${sparkW} ${sparkH}`"
+        preserveAspectRatio="none" role="img" aria-label="posts over time"
+      >
+        <polyline :points="sparkPoints" fill="none"
+                  stroke="rgb(var(--v-theme-accent))" stroke-width="2" />
+      </svg>
+    </section>
+
+    <section v-if="overview.sources.length" class="fc-artist-mgmt__sec">
+      <div class="fc-artist-mgmt__sec-head">
+        <h2 class="fc-h2">Subscriptions</h2>
+        <RouterLink
+          :to="`/subscriptions?artist_id=${overview.id}`"
+          class="fc-artist-mgmt__manage"
+        >Manage subscriptions →</RouterLink>
+      </div>
+      <v-table density="compact">
+        <thead>
+          <tr><th>Platform</th><th>URL</th><th class="text-right">Images</th></tr>
+        </thead>
+        <tbody>
+          <tr v-for="s in overview.sources" :key="s.id">
+            <td>{{ s.platform }}</td>
+            <td class="fc-artist-mgmt__url">{{ s.url }}</td>
+            <td class="text-right">{{ s.image_count }}</td>
+          </tr>
+        </tbody>
+      </v-table>
+    </section>
+
+    <section class="fc-artist-mgmt__sec">
+      <h2 class="fc-h2">Danger zone</h2>
+      <ArtistDangerZone
+        :slug="overview.slug"
+        :artist-id="overview.id"
+        :artist-name="overview.name"
+      />
+    </section>
+  </div>
+</template>
+
+<script setup>
+import { computed } from 'vue'
+import { useRouter, RouterLink } from 'vue-router'
+
+import ArtistDangerZone from './ArtistDangerZone.vue'
+
+const props = defineProps({
+  overview: { type: Object, required: true },
+})
+
+const router = useRouter()
+
+const sparkW = 600
+const sparkH = 80
+const sparkPoints = computed(() => {
+  const a = props.overview.activity ?? []
+  if (a.length === 0) return ''
+  const max = Math.max(...a.map(p => p.count), 1)
+  const stepX = a.length > 1 ? sparkW / (a.length - 1) : 0
+  return a.map((p, i) => {
+    const x = i * stepX
+    const y = sparkH - (p.count / max) * (sparkH - 4) - 2
+    return `${x.toFixed(1)},${y.toFixed(1)}`
+  }).join(' ')
+})
+
+function openTag (tagId) {
+  router.push({ name: 'gallery', query: { tag_id: tagId } })
+}
+</script>
+
+<style scoped>
+.fc-artist-mgmt { padding-top: 1rem; }
+.fc-h2 {
+  font-family: 'Fraunces', Georgia, serif;
+  font-size: 20px; font-weight: 500; margin-bottom: 8px;
+}
+.fc-artist-mgmt__sec { margin-bottom: 28px; }
+.fc-artist-mgmt__chips { display: flex; gap: 8px; flex-wrap: wrap; }
+.fc-artist-mgmt__tags { display: flex; flex-wrap: wrap; gap: 6px; }
+.fc-artist-mgmt__tagc {
+  color: rgb(var(--v-theme-on-surface-variant));
+  margin-left: 4px;
+  font-variant-numeric: tabular-nums;
+}
+.fc-artist-mgmt__spark { width: 100%; height: 80px; }
+.fc-artist-mgmt__url {
+  max-width: 380px; overflow: hidden; text-overflow: ellipsis;
+  white-space: nowrap;
+}
+.fc-artist-mgmt__sec-head {
+  display: flex; align-items: baseline; justify-content: space-between;
+  margin-bottom: 0.25rem;
+}
+.fc-artist-mgmt__manage {
+  font-size: 0.85rem;
+  color: rgb(var(--v-theme-accent));
+  text-decoration: none;
+}
+.fc-artist-mgmt__manage:hover { text-decoration: underline; }
+</style>

frontend/src/components/artist/ArtistPostsTab.vue

@@ -0,0 +1,90 @@
+<template>
+  <div class="fc-artist-posts">
+    <v-alert v-if="store.error" type="error" variant="tonal" closable class="mb-3">
+      {{ String(store.error) }}
+    </v-alert>
+
+    <div v-if="store.loading && store.items.length === 0" class="fc-artist-posts__loading">
+      <v-progress-circular indeterminate color="accent" size="36" />
+    </div>
+
+    <div v-else-if="store.items.length === 0 && store.done" class="fc-artist-posts__empty">
+      <p>No posts for this artist yet. Switch to
+        <a href="#" @click.prevent="$emit('switch-tab', 'gallery')">Gallery</a>
+        to see imported images, or visit
+        <RouterLink to="/subscriptions">Subscriptions</RouterLink>
+        to start capturing posts.
+      </p>
+    </div>
+
+    <div v-else>
+      <PostCard v-for="p in store.items" :key="p.id" :post="p" />
+      <div ref="sentinel" class="fc-artist-posts__sentinel">
+        <v-progress-circular v-if="store.loading" indeterminate color="accent" size="24" />
+        <span v-else-if="store.done" class="fc-artist-posts__end">End of stream</span>
+      </div>
+    </div>
+  </div>
+</template>
+
+<script setup>
+import { onMounted, onUnmounted, ref, watch } from 'vue'
+import { RouterLink } from 'vue-router'
+
+import { usePostsStore } from '../../stores/posts.js'
+import PostCard from '../posts/PostCard.vue'
+
+const props = defineProps({
+  artistId: { type: Number, required: true },
+})
+
+defineEmits(['switch-tab'])
+
+const store = usePostsStore()
+const sentinel = ref(null)
+let observer = null
+
+async function reload () {
+  await store.loadInitial({ artist_id: props.artistId, platform: null })
+}
+
+watch(() => props.artistId, reload)
+
+onMounted(async () => {
+  await reload()
+  observer = new IntersectionObserver((entries) => {
+    if (entries.some(e => e.isIntersecting)) {
+      store.loadMore()
+    }
+  }, { rootMargin: '400px 0px' })
+  if (sentinel.value) observer.observe(sentinel.value)
+})
+
+onUnmounted(() => {
+  if (observer) observer.disconnect()
+})
+</script>
+
+<style scoped>
+.fc-artist-posts {
+  max-width: 900px;
+  margin: 0 auto;
+}
+.fc-artist-posts__loading,
+.fc-artist-posts__empty {
+  display: flex;
+  justify-content: center;
+  padding: 2rem;
+  color: rgb(var(--v-theme-on-surface-variant));
+}
+.fc-artist-posts__sentinel {
+  display: flex;
+  justify-content: center;
+  padding: 1.5rem 0;
+  min-height: 2rem;
+}
+.fc-artist-posts__end {
+  color: rgb(var(--v-theme-on-surface-variant));
+  font-size: 0.85rem;
+}
+</style>

frontend/src/components/common/ErrorDetailModal.vue

@@ -4,14 +4,20 @@
     <v-card>
       <v-card-title class="d-flex align-center" style="gap: 12px;">
         <v-icon icon="mdi-alert-circle-outline" color="error" />
-        <span>{{ title }}</span>
+        <span>{{ displayTitle }}</span>
         <v-spacer />
         <v-btn icon variant="text" size="small" @click="close">
           <v-icon>mdi-close</v-icon>
         </v-btn>
       </v-card-title>
       <v-card-text>
-        <pre class="fc-err-pre">{{ message || '(no error message)' }}</pre>
+        <dl v-if="contextRows.length" class="fc-err-context">
+          <template v-for="(row, idx) in contextRows" :key="idx">
+            <dt>{{ row[0] }}</dt>
+            <dd>{{ row[1] }}</dd>
+          </template>
+        </dl>
+        <pre class="fc-err-pre">{{ displayMessage || '(no error message)' }}</pre>
       </v-card-text>
       <v-card-actions>
         <v-btn
@@ -27,12 +33,21 @@
 </template>
 
 <script setup>
-import { ref, watch } from 'vue'
+import { computed, ref, watch } from 'vue'
+
+import { copyText } from '../../utils/clipboard.js'
 
 const props = defineProps({
   modelValue: { type: Boolean, default: false },
+  // Legacy mode: pass title + message strings. Used by callers whose row
+  // shape lacks structured context (e.g. ImportTaskList where `error` is
+  // a plain string on the import_task row, not a TaskRun).
   title: { type: String, default: 'Error details' },
   message: { type: String, default: '' },
+  // Row mode: pass the full TaskRun-shaped dict from /api/system_activity.
+  // When set, displayTitle/displayMessage derive from the row and a context
+  // panel of task_name/queue/target/duration/etc. renders above the error.
+  row: { type: Object, default: null },
 })
 
 const emit = defineEmits(['update:modelValue'])
@@ -40,6 +55,47 @@ const emit = defineEmits(['update:modelValue'])
 const copied = ref(false)
 let copiedTimer = null
 
+const displayTitle = computed(() => {
+  if (props.row) return props.row.error_type || 'Error details'
+  return props.title
+})
+
+const displayMessage = computed(() => {
+  if (props.row) return props.row.error_message || ''
+  return props.message
+})
+
+function _shortTaskName (name) {
+  if (!name) return ''
+  const parts = String(name).split('.')
+  return parts[parts.length - 1]
+}
+
+function _formatDuration (ms) {
+  if (ms == null) return null
+  if (ms < 1000) return `${ms} ms`
+  if (ms < 60_000) return `${(ms / 1000).toFixed(1)} s`
+  return `${(ms / 60_000).toFixed(1)} min`
+}
+
+const contextRows = computed(() => {
+  const r = props.row
+  if (!r) return []
+  const rows = []
+  if (r.task_name) rows.push(['Task', _shortTaskName(r.task_name)])
+  if (r.queue) rows.push(['Queue', r.queue])
+  if (r.target_id != null) rows.push(['Target', r.target_id])
+  const dur = _formatDuration(r.duration_ms)
+  if (dur != null) rows.push(['Duration', dur])
+  if (r.started_at) rows.push(['Started', r.started_at])
+  if (r.finished_at) rows.push(['Finished', r.finished_at])
+  if (r.retry_count) rows.push(['Retries', r.retry_count])
+  if (r.worker_hostname) rows.push(['Worker', r.worker_hostname])
+  if (r.celery_task_id) rows.push(['Celery ID', r.celery_task_id])
+  if (r.args_summary) rows.push(['Args', r.args_summary])
+  return rows
+})
+
 watch(() => props.modelValue, (open) => {
   if (!open) {
     copied.value = false
@@ -47,13 +103,18 @@ watch(() => props.modelValue, (open) => {
   }
 })
 
-function close() {
+function close () {
   emit('update:modelValue', false)
 }
 
-async function onCopy() {
+async function onCopy () {
+  let text = displayMessage.value || ''
+  if (contextRows.value.length) {
+    const header = contextRows.value.map(([k, v]) => `${k}: ${v}`).join('\n')
+    text = `${header}\n\nError: ${displayTitle.value}\n${text}`
+  }
   try {
-    await navigator.clipboard.writeText(props.message || '')
+    await copyText(text)
     copied.value = true
     if (copiedTimer) clearTimeout(copiedTimer)
     copiedTimer = setTimeout(() => { copied.value = false }, 1500)
@@ -64,19 +125,43 @@ async function onCopy() {
 </script>
 
 <style scoped>
-/* The full error often contains a SQLAlchemy statement + parameters
+/* Context panel: muted labels in vellum, crisp values in parchment. The
-   block + multi-line traceback. Pre-wrap keeps long lines readable;
+   2-column key/value grid keeps rows visually scannable when there are
-   monospace + tabular layout keeps the structure scannable. The
+   many fields (Task, Queue, Target, Duration, Started, Finished, Retries,
-   max-height + overflow-auto prevents a 50-line traceback from
+   Worker, Celery ID, Args). */
-   pushing the Close button off-screen. Operator-flagged 2026-05-26:
+.fc-err-context {
-   the prior :title="..." tooltip was unusable for content this long. */
+  display: grid;
+  grid-template-columns: max-content 1fr;
+  column-gap: 14px;
+  row-gap: 4px;
+  margin: 0 0 14px 0;
+  font-size: 13px;
+}
+.fc-err-context dt {
+  color: rgb(var(--v-theme-on-surface-variant));
+  font-weight: 500;
+  white-space: nowrap;
+}
+.fc-err-context dd {
+  color: rgb(var(--v-theme-on-surface));
+  margin: 0;
+  font-variant-numeric: tabular-nums;
+  word-break: break-all;
+}
+
+/* Error pre block: high-contrast pairing. The page's `background` token
+   (obsidian #14171A) is darker than the modal card's `surface` (iron
+   #1E2228), so parchment text reads crisply against it. The prior pairing
+   used `surface-variant` which Vuetify auto-derives to a near-parchment
+   light value in this theme — pale-on-pale and unreadable. Operator-
+   flagged 2026-05-26 ("ui contrast is poor"). */
 .fc-err-pre {
-  font-family: ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
+  font-family: 'JetBrains Mono', ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
   font-size: 12px;
   line-height: 1.5;
   white-space: pre-wrap;
   word-break: break-word;
-  background: rgb(var(--v-theme-surface-variant, 38 36 41));
+  background: rgb(var(--v-theme-background));
   color: rgb(var(--v-theme-on-surface));
   padding: 12px 14px;
   border-radius: 6px;

frontend/src/components/credentials/ExtensionKeyBar.vue

@@ -28,6 +28,7 @@
 <script setup>
 import { onMounted, ref } from 'vue'
 import { useCredentialsStore } from '../../stores/credentials.js'
+import { copyText } from '../../utils/clipboard.js'
 
 const store = useCredentialsStore()
 const showRotateConfirm = ref(false)
@@ -37,7 +38,7 @@ onMounted(() => store.loadKey())
 async function copyKey() {
   if (!store.extensionKey) return
   try {
-    await navigator.clipboard.writeText(store.extensionKey)
+    await copyText(store.extensionKey)
     globalThis.window?.__fcToast?.({ text: 'Copied', type: 'success' })
   } catch {
     globalThis.window?.__fcToast?.({ text: 'Copy failed', type: 'error' })

frontend/src/components/settings/BrowserExtensionCard.vue

@@ -110,6 +110,7 @@
 <script setup>
 import { computed, onMounted, ref } from 'vue'
 import { useApi } from '../../composables/useApi.js'
+import { copyText } from '../../utils/clipboard.js'
 
 const api = useApi()
 
@@ -172,7 +173,7 @@ async function rotateKey() {
 
 async function copy(text, label) {
   try {
-    await navigator.clipboard.writeText(text)
+    await copyText(text)
     window.__fcToast?.({ text: `${label} copied.`, type: 'success' })
   } catch (e) {
     window.__fcToast?.({ text: `Copy failed: ${e.message}`, type: 'error' })

frontend/src/components/settings/MaintenancePanel.vue

@@ -1,13 +1,15 @@
 <template>
   <div class="fc-maint">
     <p class="text-body-2 mb-4">
-      Machine-assisted tagging controls. Backfill and centroid recompute run
+      Operational backfills and tagging controls. The ML backfill and centroid
-      nightly automatically; the allowlist auto-applies accepted tags to new
+      recompute run nightly automatically; the allowlist auto-applies accepted
-      and existing images.
+      tags to new and existing images. Use the cards below to trigger a
+      one-off pass.
     </p>
     <div class="fc-maint__grid">
       <MLBackfillCard />
       <CentroidRecomputeCard />
+      <ThumbnailBackfillCard />
     </div>
     <MLThresholdSliders class="mt-4" />
     <AllowlistTable class="mt-4" />
@@ -23,6 +25,7 @@
 <script setup>
 import MLBackfillCard from './MLBackfillCard.vue'
 import CentroidRecomputeCard from './CentroidRecomputeCard.vue'
+import ThumbnailBackfillCard from './ThumbnailBackfillCard.vue'
 import MLThresholdSliders from './MLThresholdSliders.vue'
 import AllowlistTable from './AllowlistTable.vue'
 import AliasTable from './AliasTable.vue'

frontend/src/components/settings/SystemActivityTab.vue

@@ -62,7 +62,7 @@
               <td>
                 <button
                   type="button" class="fc-err-link"
-                  @click="openError(r.error_type, r.error_message)"
+                  @click="openError(r)"
                   :title="'Click for full error'"
                 >{{ r.error_type }}</button>
               </td>
@@ -145,8 +145,7 @@
 
     <ErrorDetailModal
       v-model="showErrorModal"
-      :title="errorModalTitle"
+      :row="errorModalRow"
-      :message="errorModalMessage"
     />
   </div>
 </template>
@@ -163,12 +162,10 @@ import QueuesTable from './QueuesTable.vue'
 // rollback + traceback content rendered as a cramped browser tooltip
 // you couldn't copy from or scroll within).
 const showErrorModal = ref(false)
-const errorModalTitle = ref('')
+const errorModalRow = ref(null)
-const errorModalMessage = ref('')
 
-function openError(title, message) {
+function openError(row) {
-  errorModalTitle.value = title || 'Error details'
+  errorModalRow.value = row
-  errorModalMessage.value = message || ''
   showErrorModal.value = true
 }
 

frontend/src/components/settings/ThumbnailBackfillCard.vue

@@ -0,0 +1,30 @@
+<template>
+  <v-card>
+    <v-card-title>Thumbnail backfill</v-card-title>
+    <v-card-text>
+      <p class="text-body-2 mb-3">
+        Scan the library for images with no thumbnail, or whose thumbnail file
+        is missing or corrupt on disk. Repair candidates are re-enqueued for
+        thumbnail generation. Safe to re-run.
+      </p>
+      <v-btn color="primary" rounded="pill" :loading="busy" @click="run">
+        <v-icon start>mdi-image-refresh</v-icon> Run backfill now
+      </v-btn>
+      <span v-if="done" class="ml-3 text-caption">Enqueued.</span>
+    </v-card-text>
+  </v-card>
+</template>
+
+<script setup>
+import { ref } from 'vue'
+import { useThumbnailsStore } from '../../stores/thumbnails.js'
+const store = useThumbnailsStore()
+const busy = ref(false)
+const done = ref(false)
+async function run () {
+  busy.value = true
+  try { await store.triggerBackfill(); done.value = true }
+  catch (e) { window.__fcToast?.({ text: e.message, type: 'error' }) }
+  finally { busy.value = false }
+}
+</script>

frontend/src/stores/artist.js

@@ -1,6 +1,7 @@
 import { defineStore } from 'pinia'
 import { ref, computed } from 'vue'
 import { useApi } from '../composables/useApi.js'
+import { usePostsStore } from './posts.js'
 
 const PAGE = 60
 
@@ -15,7 +16,11 @@ export const useArtistStore = defineStore('artist', () => {
   const notFound = ref(false)
   let started = false
 
-  async function load(slug) {
+  async function load (slug) {
+    // Cross-artist reset: clear this store AND the posts store so the new
+    // artist doesn't briefly render with the previous artist's content
+    // when the user is on the Posts tab. (Gallery tab uses this artist
+    // store's own images list — cleared above.)
     overview.value = null
     images.value = []
     nextCursor.value = null
@@ -23,6 +28,7 @@ export const useArtistStore = defineStore('artist', () => {
     started = false
     error.value = null
     loading.value = true
+    usePostsStore().$reset?.()
     try {
       overview.value = await api.get(`/api/artist/${encodeURIComponent(slug)}`)
       await loadMoreImages(slug)
@@ -34,7 +40,7 @@ export const useArtistStore = defineStore('artist', () => {
     }
   }
 
-  async function loadMoreImages(slug) {
+  async function loadMoreImages (slug) {
     if (imagesLoading.value) return
     if (started && nextCursor.value === null) return
     imagesLoading.value = true
@@ -55,9 +61,13 @@ export const useArtistStore = defineStore('artist', () => {
   }
 
   const hasMoreImages = computed(() => !started || nextCursor.value !== null)
+  const postCount = computed(() => overview.value?.post_count ?? null)
+  const imageCount = computed(() => overview.value?.image_count ?? null)
+  const lastAdded = computed(() => overview.value?.date_range?.max ?? null)
 
   return {
     overview, images, loading, imagesLoading, error, notFound,
-    hasMoreImages, load, loadMoreImages
+    hasMoreImages, postCount, imageCount, lastAdded,
+    load, loadMoreImages,
   }
 })

frontend/src/stores/thumbnails.js

@@ -0,0 +1,12 @@
+import { defineStore } from 'pinia'
+import { useApi } from '../composables/useApi.js'
+
+export const useThumbnailsStore = defineStore('thumbnails', () => {
+  const api = useApi()
+
+  async function triggerBackfill () {
+    await api.post('/api/thumbnails/backfill')
+  }
+
+  return { triggerBackfill }
+})

frontend/src/utils/clipboard.js

@@ -0,0 +1,35 @@
+// Clipboard write that works on plain-HTTP self-hosted deployments.
+//
+// navigator.clipboard is gated by the browser's Secure Context restriction
+// (HTTPS or localhost only). FabledCurator runs over plain HTTP per the
+// homelab posture, so the modern API is undefined in production. We fall
+// back to the legacy execCommand('copy') path via a temporary off-screen
+// textarea — wide browser support, no HTTPS requirement.
+
+export async function copyText (text) {
+  const str = text == null ? '' : String(text)
+  if (typeof navigator !== 'undefined' && navigator.clipboard?.writeText) {
+    try {
+      await navigator.clipboard.writeText(str)
+      return
+    } catch {
+      // Fall through to the legacy path. Some browsers throw even when
+      // the API exists (permission denied, focus loss, etc.).
+    }
+  }
+  const ta = document.createElement('textarea')
+  ta.value = str
+  ta.setAttribute('readonly', '')
+  ta.style.position = 'fixed'
+  ta.style.top = '0'
+  ta.style.left = '0'
+  ta.style.opacity = '0'
+  ta.style.pointerEvents = 'none'
+  document.body.appendChild(ta)
+  ta.select()
+  ta.setSelectionRange(0, str.length)
+  let ok = false
+  try { ok = document.execCommand('copy') } catch { ok = false }
+  document.body.removeChild(ta)
+  if (!ok) throw new Error('clipboard copy not supported')
+}

frontend/src/views/ArtistView.vue

@@ -1,220 +1,96 @@
 <template>
-  <v-container fluid class="py-6">
+  <div v-if="store.loading && !store.overview" class="fc-artist__loading">
-    <div v-if="store.loading && !store.overview" class="fc-artist__loading">
+    <v-progress-circular indeterminate color="accent" size="36" />
-      <v-progress-circular indeterminate color="accent" size="36" />
+  </div>
-    </div>
 
-    <v-alert v-else-if="store.notFound" type="warning" variant="tonal">
+  <v-container v-else-if="store.notFound" class="py-6">
-      Artist not found.
+    <v-alert type="warning" variant="tonal">Artist not found.</v-alert>
-    </v-alert>
+  </v-container>
 
-    <v-alert v-else-if="store.error" type="error" variant="tonal" closable>
+  <v-container v-else-if="store.error && !store.overview" class="py-6">
-      {{ store.error }}
+    <v-alert type="error" variant="tonal" closable>{{ store.error }}</v-alert>
-    </v-alert>
+  </v-container>
-
-    <template v-else-if="store.overview">
-      <header class="fc-artist__head">
-        <h1 class="fc-h1">{{ store.overview.name }}</h1>
-        <div class="fc-artist__stats">
-          <span>{{ store.overview.image_count }} images</span>
-          <span v-if="dateRange">· {{ dateRange }}</span>
-        </div>
-      </header>
-
-      <div class="fc-artist__fc4">
-        <v-chip
-          size="small"
-          :variant="store.overview.is_subscription ? 'flat' : 'outlined'"
-          :color="store.overview.is_subscription ? 'accent' : undefined"
-          prepend-icon="mdi-rss"
-        >{{ store.overview.is_subscription ? 'Subscription' : 'One-off' }}</v-chip>
-
-        <v-chip
-          size="small" variant="outlined" prepend-icon="mdi-link-variant"
-          :to="`/subscriptions?artist_id=${store.overview.id}`"
-        >{{ store.overview.sources.length }} source{{ store.overview.sources.length === 1 ? '' : 's' }}</v-chip>
-
-        <v-chip
-          size="small" variant="outlined" prepend-icon="mdi-rss"
-          :to="`/posts?artist_id=${store.overview.id}`"
-        >View posts</v-chip>
-
-        <v-chip
-          size="small" variant="outlined" disabled
-          prepend-icon="mdi-clock-outline"
-        >Credential health · FC-3b</v-chip>
-      </div>
-
-      <!-- Tabs split (2026-05-25): Settings was previously slotted at the
-           bottom of the page after the infinite-scroll image grid, which
-           made it effectively unreachable for any artist with more than
-           a couple of pages of content. The Settings tab now hosts
-           destructive admin actions (artist+content cascade-delete) and
-           any future per-artist management UI. v-tabs is `position:
-           sticky; top: 64px` (under the 64px AppShell TopNav) so it
-           stays parked while the gallery scrolls. -->
-      <v-tabs
-        v-model="tab" color="accent" class="mb-4"
-        style="position: sticky; top: 64px; z-index: 4;
-               background: rgb(var(--v-theme-surface));"
-      >
-        <v-tab value="overview">Overview</v-tab>
-        <v-tab value="settings">Settings</v-tab>
-      </v-tabs>
 
+  <template v-else-if="store.overview">
+    <ArtistHeader
+      v-model="tab"
+      :name="store.overview.name"
+      :image-count="store.imageCount"
+      :post-count="store.postCount"
+      :last-added="store.lastAdded"
+    />
+    <v-container fluid class="py-4">
       <v-window v-model="tab">
-        <v-window-item value="overview">
+        <v-window-item value="posts">
-          <section v-if="store.overview.cooccurring_tags.length" class="fc-artist__sec">
+          <ArtistPostsTab
-            <h2 class="fc-h2">Frequent tags</h2>
-            <div class="fc-artist__tags">
-              <v-chip
-                v-for="t in store.overview.cooccurring_tags" :key="t.id"
-                size="small" @click="openTag(t.id)"
-              >{{ t.name }} <span class="fc-artist__tagc">{{ t.count }}</span></v-chip>
-            </div>
-          </section>
-
-          <section v-if="store.overview.activity.length" class="fc-artist__sec">
-            <h2 class="fc-h2">Activity</h2>
-            <svg class="fc-artist__spark" :viewBox="`0 0 ${sparkW} ${sparkH}`"
-                 preserveAspectRatio="none" role="img" aria-label="posts over time">
-              <polyline :points="sparkPoints" fill="none"
-                        stroke="rgb(var(--v-theme-accent))" stroke-width="2" />
-            </svg>
-          </section>
-
-          <section v-if="store.overview.sources.length" class="fc-artist__sec">
-            <div class="fc-artist__sec-head">
-              <h2 class="fc-h2">Sources</h2>
-              <RouterLink
-                :to="`/subscriptions?artist_id=${store.overview.id}`"
-                class="fc-artist__manage"
-              >Manage subscriptions →</RouterLink>
-            </div>
-            <v-table density="compact">
-              <thead>
-                <tr><th>Platform</th><th>URL</th><th class="text-right">Images</th></tr>
-              </thead>
-              <tbody>
-                <tr v-for="s in store.overview.sources" :key="s.id">
-                  <td>{{ s.platform }}</td>
-                  <td class="fc-artist__url">{{ s.url }}</td>
-                  <td class="text-right">{{ s.image_count }}</td>
-                </tr>
-              </tbody>
-            </v-table>
-          </section>
-
-          <section class="fc-artist__sec">
-            <h2 class="fc-h2">Images</h2>
-            <MasonryGrid
-              :items="store.images"
-              :loading="store.imagesLoading"
-              :has-more="store.hasMoreImages"
-              @load-more="store.loadMoreImages(slug)"
-              @open="openImage"
-            />
-          </section>
-        </v-window-item>
-
-        <v-window-item value="settings">
-          <ArtistDangerZone
-            :slug="slug"
             :artist-id="store.overview.id"
-            :artist-name="store.overview.name"
+            @switch-tab="(t) => tab = t"
           />
         </v-window-item>
+        <v-window-item value="gallery">
+          <ArtistGalleryTab :slug="slug" />
+        </v-window-item>
+        <v-window-item value="management">
+          <ArtistManagementTab :overview="store.overview" />
+        </v-window-item>
       </v-window>
-    </template>
+    </v-container>
-  </v-container>
+  </template>
 </template>
 
 <script setup>
 import { computed, ref, watch } from 'vue'
-import { useRoute, useRouter, RouterLink } from 'vue-router'
+import { useRoute, useRouter } from 'vue-router'
+
 import { useArtistStore } from '../stores/artist.js'
-import { useModalStore } from '../stores/modal.js'
+import ArtistHeader from '../components/artist/ArtistHeader.vue'
-import MasonryGrid from '../components/discovery/MasonryGrid.vue'
+import ArtistPostsTab from '../components/artist/ArtistPostsTab.vue'
-import ArtistDangerZone from '../components/artist/ArtistDangerZone.vue'
+import ArtistGalleryTab from '../components/artist/ArtistGalleryTab.vue'
+import ArtistManagementTab from '../components/artist/ArtistManagementTab.vue'
+
+const VALID_TABS = ['posts', 'gallery', 'management']
 
 const route = useRoute()
 const router = useRouter()
 const store = useArtistStore()
-const modal = useModalStore()
 
 const slug = computed(() => route.params.slug)
-// Per-artist tab — defaults to Overview. Settings tab hosts destructive
+const tab = ref('posts')
-// admin actions (DangerZone). Switching artists resets to Overview so the
-// destructive surface isn't re-shown by accident when navigating between
-// artists.
-const tab = ref('overview')
 
-watch(slug, (s) => {
+function resolveDefaultTab () {
-  if (s) {
+  const fromUrl = route.query.tab
-    store.load(s)
+  if (VALID_TABS.includes(fromUrl)) return fromUrl
-    tab.value = 'overview'
+  if ((store.postCount ?? 0) > 0) return 'posts'
-  }
+  return 'gallery'
+}
+
+watch(slug, async (s) => {
+  if (!s) return
+  await store.load(s)
+  document.title = store.overview
+    ? `${store.overview.name} — FabledCurator`
+    : 'FabledCurator'
+  tab.value = resolveDefaultTab()
 }, { immediate: true })
 
-const dateRange = computed(() => {
+// Reflect tab changes back into the URL so refresh/back/forward work.
-  const r = store.overview?.date_range
+watch(tab, (newTab) => {
-  if (!r || !r.min) return null
+  if (route.query.tab === newTab) return
-  const fmt = (iso) => iso.slice(0, 10)
+  router.replace({
-  return r.min === r.max ? fmt(r.min) : `${fmt(r.min)} → ${fmt(r.max)}`
+    query: { ...route.query, tab: newTab },
+  })
 })
 
-const sparkW = 600
+// React to URL-tab changes (e.g., back/forward).
-const sparkH = 80
+watch(() => route.query.tab, (q) => {
-const sparkPoints = computed(() => {
+  if (q && VALID_TABS.includes(q) && tab.value !== q) {
-  const a = store.overview?.activity ?? []
+    tab.value = q
-  if (a.length === 0) return ''
+  }
-  const max = Math.max(...a.map(p => p.count), 1)
-  const stepX = a.length > 1 ? sparkW / (a.length - 1) : 0
-  return a.map((p, i) => {
-    const x = i * stepX
-    const y = sparkH - (p.count / max) * (sparkH - 4) - 2
-    return `${x.toFixed(1)},${y.toFixed(1)}`
-  }).join(' ')
 })
-
-function openImage(id) {
-  modal.open(id)
-}
-function openTag(tagId) {
-  router.push({ name: 'gallery', query: { tag_id: tagId } })
-}
 </script>
 
 <style scoped>
-.fc-h1 {
+.fc-artist__loading {
-  font-family: 'Fraunces', Georgia, serif;
+  display: flex; justify-content: center; padding: 64px 0;
-  font-size: 32px; font-weight: 500;
-  color: rgb(var(--v-theme-on-surface));
 }
-.fc-h2 {
-  font-family: 'Fraunces', Georgia, serif;
-  font-size: 20px; font-weight: 500; margin-bottom: 8px;
-}
-.fc-artist__loading { display: flex; justify-content: center; padding: 64px 0; }
-.fc-artist__head { margin-bottom: 12px; }
-.fc-artist__stats { opacity: 0.75; margin-top: 4px; }
-.fc-artist__fc4 { display: flex; gap: 8px; flex-wrap: wrap; margin-bottom: 24px; }
-.fc-artist__sec { margin-bottom: 28px; }
-.fc-artist__tags { display: flex; flex-wrap: wrap; gap: 6px; }
-.fc-artist__tagc { opacity: 0.6; margin-left: 4px; font-variant-numeric: tabular-nums; }
-.fc-artist__spark { width: 100%; height: 80px; }
-.fc-artist__url {
-  max-width: 380px; overflow: hidden; text-overflow: ellipsis;
-  white-space: nowrap;
-}
-.fc-artist__sec-head {
-  display: flex; align-items: baseline; justify-content: space-between;
-  margin-bottom: 0.25rem;
-}
-.fc-artist__manage {
-  font-size: 0.85rem;
-  color: rgb(var(--v-theme-accent));
-  text-decoration: none;
-}
-.fc-artist__manage:hover { text-decoration: underline; }
 </style>

tests/test_api_artist.py

@@ -29,6 +29,30 @@ async def test_artist_overview_ok(client, db):
     body = await resp.get_json()
     assert body["name"] == "Mira"
     assert body["image_count"] == 0
+    assert body["post_count"] == 0
+
+
+@pytest.mark.asyncio
+async def test_artist_overview_post_count(client, db):
+    from backend.app.models import Post, Source
+
+    a = Artist(name="Lyra", slug="lyra")
+    db.add(a)
+    await db.flush()
+    s = Source(
+        artist_id=a.id, platform="patreon",
+        url="https://patreon.com/cw/lyra", enabled=True,
+    )
+    db.add(s)
+    await db.flush()
+    db.add(Post(source_id=s.id, external_post_id="p1"))
+    db.add(Post(source_id=s.id, external_post_id="p2"))
+    await db.flush()
+    await db.commit()
+    resp = await client.get("/api/artist/lyra")
+    assert resp.status_code == 200
+    body = await resp.get_json()
+    assert body["post_count"] == 2
 
 
 @pytest.mark.asyncio

tests/test_api_thumbnails.py

@@ -0,0 +1,32 @@
+import pytest
+
+from backend.app import create_app
+from backend.app.celery_app import celery
+
+pytestmark = pytest.mark.integration
+
+
+@pytest.fixture(autouse=True)
+def eager():
+    celery.conf.task_always_eager = True
+    yield
+    celery.conf.task_always_eager = False
+
+
+@pytest.fixture
+async def app():
+    return create_app()
+
+
+@pytest.fixture
+async def client(app):
+    async with app.test_client() as c:
+        yield c
+
+
+@pytest.mark.asyncio
+async def test_trigger_thumbnail_backfill(client):
+    r = await client.post("/api/thumbnails/backfill")
+    assert r.status_code == 202
+    body = await r.get_json()
+    assert "celery_task_id" in body

tests/test_backfill_thumbnails.py

@@ -0,0 +1,242 @@
+"""Thumbnail backfill: _thumb_is_valid helper + backfill_thumbnails planner."""
+
+from pathlib import Path
+
+import pytest
+
+from backend.app.models import ImageRecord
+from backend.app.tasks.thumbnail import _thumb_is_valid
+
+pytestmark = pytest.mark.integration
+
+
+def test_thumb_is_valid_jpeg(tmp_path):
+    p = tmp_path / "good.jpg"
+    p.write_bytes(b"\xff\xd8\xff\xe0" + b"\x00" * 100)
+    assert _thumb_is_valid(p) is True
+
+
+def test_thumb_is_valid_png(tmp_path):
+    p = tmp_path / "good.png"
+    p.write_bytes(b"\x89PNG\r\n\x1a\n" + b"\x00" * 100)
+    assert _thumb_is_valid(p) is True
+
+
+def test_thumb_is_valid_too_short(tmp_path):
+    p = tmp_path / "tiny"
+    p.write_bytes(b"\xff\xd8")
+    assert _thumb_is_valid(p) is False
+
+
+def test_thumb_is_valid_wrong_magic(tmp_path):
+    p = tmp_path / "garbage"
+    p.write_bytes(b"\x00" * 12)
+    assert _thumb_is_valid(p) is False
+
+
+def test_thumb_is_valid_missing_file(tmp_path):
+    assert _thumb_is_valid(tmp_path / "nope") is False
+
+
+# --- backfill_thumbnails planner tests ------------------------------------
+
+
+class _Ctx:
+    def __init__(self, s):
+        self.s = s
+
+    def __enter__(self):
+        return self.s
+
+    def __exit__(self, *a):
+        return False
+
+
+def _sf(db_sync):
+    """sessionmaker-like returning the test's bound session, matching the
+    pattern in tests/test_backfill_phash.py."""
+    class _SM:
+        def __call__(self):
+            return _Ctx(db_sync)
+
+    return _SM()
+
+
+def _sha(prefix: str) -> str:
+    return f"{prefix}".ljust(64, "0")[:64]
+
+
+def _rec(db_sync, path, *, sha, thumb_path=None, mime="image/jpeg"):
+    rec = ImageRecord(
+        path=str(path), sha256=sha, size_bytes=1, mime=mime,
+        width=64, height=64, origin="imported_filesystem",
+        integrity_status="unknown",
+        thumbnail_path=str(thumb_path) if thumb_path is not None else None,
+    )
+    db_sync.add(rec)
+    db_sync.flush()
+    return rec
+
+
+def _write_jpeg(p: Path) -> Path:
+    p.parent.mkdir(parents=True, exist_ok=True)
+    p.write_bytes(b"\xff\xd8\xff\xe0" + b"\x00" * 100)
+    return p
+
+
+def _write_png(p: Path) -> Path:
+    p.parent.mkdir(parents=True, exist_ok=True)
+    p.write_bytes(b"\x89PNG\r\n\x1a\n" + b"\x00" * 100)
+    return p
+
+
+def _write_garbage(p: Path) -> Path:
+    p.parent.mkdir(parents=True, exist_ok=True)
+    p.write_bytes(b"\x00" * 12)
+    return p
+
+
+def test_backfill_null_path_enqueued(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src = tmp_path / "a.bin"
+    src.write_bytes(b"x")
+    rec = _rec(db_sync, src, sha=_sha("a"), thumb_path=None)
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    assert result == {"enqueued": 1, "ok": 0, "regenerated": 0}
+    assert delayed == [rec.id]
+
+
+def test_backfill_missing_file_clears_and_enqueues(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src = tmp_path / "b.bin"
+    src.write_bytes(b"x")
+    rec = _rec(
+        db_sync, src, sha=_sha("b"),
+        thumb_path=tmp_path / "thumbs" / "missing.jpg",
+    )
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    db_sync.expire_all()
+    assert result == {"enqueued": 1, "ok": 0, "regenerated": 1}
+    assert delayed == [rec.id]
+    assert db_sync.get(ImageRecord, rec.id).thumbnail_path is None
+
+
+def test_backfill_valid_jpeg_skipped(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src = tmp_path / "c.bin"
+    src.write_bytes(b"x")
+    thumb = _write_jpeg(tmp_path / "thumbs" / "c.jpg")
+    rec = _rec(db_sync, src, sha=_sha("c"), thumb_path=thumb)
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    db_sync.expire_all()
+    assert result == {"enqueued": 0, "ok": 1, "regenerated": 0}
+    assert delayed == []
+    assert db_sync.get(ImageRecord, rec.id).thumbnail_path == str(thumb)
+
+
+def test_backfill_valid_png_skipped(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src = tmp_path / "d.bin"
+    src.write_bytes(b"x")
+    thumb = _write_png(tmp_path / "thumbs" / "d.png")
+    _rec(db_sync, src, sha=_sha("d"), thumb_path=thumb)
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    assert result == {"enqueued": 0, "ok": 1, "regenerated": 0}
+    assert delayed == []
+
+
+def test_backfill_corrupt_magic_clears_and_enqueues(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src = tmp_path / "e.bin"
+    src.write_bytes(b"x")
+    thumb = _write_garbage(tmp_path / "thumbs" / "e.jpg")
+    rec = _rec(db_sync, src, sha=_sha("e"), thumb_path=thumb)
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    db_sync.expire_all()
+    assert result == {"enqueued": 1, "ok": 0, "regenerated": 1}
+    assert delayed == [rec.id]
+    assert db_sync.get(ImageRecord, rec.id).thumbnail_path is None
+
+
+def test_backfill_mixed_aggregate(db_sync, tmp_path, monkeypatch):
+    from backend.app.tasks import thumbnail as m
+
+    src_null = tmp_path / "src_null.bin"
+    src_null.write_bytes(b"x")
+    src_jpeg = tmp_path / "src_jpeg.bin"
+    src_jpeg.write_bytes(b"x")
+    src_png = tmp_path / "src_png.bin"
+    src_png.write_bytes(b"x")
+    src_missing = tmp_path / "src_missing.bin"
+    src_missing.write_bytes(b"x")
+    src_bad = tmp_path / "src_bad.bin"
+    src_bad.write_bytes(b"x")
+
+    jpeg = _write_jpeg(tmp_path / "thumbs" / "ok.jpg")
+    png = _write_png(tmp_path / "thumbs" / "ok.png")
+    bad = _write_garbage(tmp_path / "thumbs" / "bad.jpg")
+
+    r_null = _rec(db_sync, src_null, sha=_sha("aa"), thumb_path=None)
+    _rec(db_sync, src_jpeg, sha=_sha("bb"), thumb_path=jpeg)
+    _rec(db_sync, src_png, sha=_sha("cc"), thumb_path=png)
+    r_missing = _rec(
+        db_sync, src_missing, sha=_sha("dd"),
+        thumb_path=tmp_path / "thumbs" / "missing.jpg",
+    )
+    r_bad = _rec(db_sync, src_bad, sha=_sha("ee"), thumb_path=bad)
+    db_sync.commit()
+
+    monkeypatch.setattr(m, "_sync_session_factory", lambda: _sf(db_sync))
+    delayed: list[int] = []
+    monkeypatch.setattr(
+        m.generate_thumbnail, "delay", lambda image_id: delayed.append(image_id)
+    )
+
+    result = m.backfill_thumbnails()
+    assert result == {"enqueued": 3, "ok": 2, "regenerated": 2}
+    assert sorted(delayed) == sorted([r_null.id, r_missing.id, r_bad.id])

tests/test_gallery_provenance_filter.py

@@ -69,20 +69,13 @@ async def test_scroll_post_id_filter(db):
     assert {x.id for x in page.images} == {i1.id, i2.id}
 
 
-@pytest.mark.asyncio
+# test_scroll_post_id_dedups_multi_rows removed 2026-05-26: it deliberately
-async def test_scroll_post_id_dedups_multi_rows(db):
+# inserted two ImageProvenance rows with the same (image_record_id, post_id),
-    i1 = await _img(db, 1)
+# now prevented at the DB layer by uq_image_provenance_image_post (alembic
-    _, s, p = await _post(db, "A", "a", "10")
+# 0021). The EXISTS-based dedup in _provenance_clause is still useful for the
-    # two provenance rows, same image+post (enrich-on-duplicate shape)
+# artist-id filter (one image legitimately joins many provenance rows via
-    db.add(ImageProvenance(image_record_id=i1.id, post_id=p.id,
+# different posts under the same artist), so the gallery_service logic is
-                           source_id=s.id))
+# unchanged.
-    await db.flush()
-    db.add(ImageProvenance(image_record_id=i1.id, post_id=p.id,
-                           source_id=s.id))
-    await db.flush()
-    svc = GalleryService(db)
-    page = await svc.scroll(cursor=None, limit=10, post_id=p.id)
-    assert [x.id for x in page.images] == [i1.id]  # appears once
 
 
 @pytest.mark.asyncio

tests/test_importer_provenance_race.py

@@ -0,0 +1,184 @@
+"""Race-safe ImageProvenance insert in Importer._apply_sidecar.
+
+Operator-flagged 2026-05-26: the prior SELECT-then-INSERT pattern lost a
+race when two workers ran _apply_sidecar on the same (image, post) pair
+(plausibly seeded when the 5-min recovery sweep re-enqueued a still-running
+long import). Duplicates then broke .scalar_one_or_none() on every later
+deep-scan rederive (MultipleResultsFound). Alembic 0021 added
+uq_image_provenance_image_post; the importer's new savepoint+IntegrityError
+recovery path now trips on collision and gracefully recovers.
+
+Tests cover:
+  - idempotent: re-running _apply_sidecar via _deep_rederive produces
+    exactly one ImageProvenance row.
+  - IntegrityError recovery: pre-seed a provenance row, force the first
+    SELECT to return None (simulating the race window where two workers
+    both observed no row), call _apply_sidecar — the savepoint INSERT
+    trips uq_image_provenance_image_post, gets rolled back, no exception
+    escapes, still exactly one row.
+"""
+
+import json
+from pathlib import Path
+
+import pytest
+from PIL import Image
+from sqlalchemy import func, select
+
+from backend.app.models import (
+    ImageProvenance,
+    ImageRecord,
+    ImportSettings,
+    Source,
+)
+from backend.app.services.importer import Importer
+from backend.app.services.thumbnailer import Thumbnailer
+
+pytestmark = pytest.mark.integration
+
+
+@pytest.fixture
+def import_layout(tmp_path):
+    import_root = tmp_path / "import"
+    images_root = tmp_path / "images"
+    import_root.mkdir()
+    images_root.mkdir()
+    return import_root, images_root
+
+
+@pytest.fixture
+def importer(db_sync, import_layout):
+    import_root, images_root = import_layout
+    settings = db_sync.execute(
+        select(ImportSettings).where(ImportSettings.id == 1)
+    ).scalar_one()
+    return Importer(
+        session=db_sync,
+        images_root=images_root,
+        import_root=import_root,
+        thumbnailer=Thumbnailer(images_root=images_root),
+        settings=settings,
+    )
+
+
+@pytest.fixture
+def deep_importer(db_sync, import_layout):
+    import_root, images_root = import_layout
+    settings = db_sync.execute(
+        select(ImportSettings).where(ImportSettings.id == 1)
+    ).scalar_one()
+    return Importer(
+        session=db_sync,
+        images_root=images_root,
+        import_root=import_root,
+        thumbnailer=Thumbnailer(images_root=images_root),
+        settings=settings,
+        deep=True,
+    )
+
+
+def _split(path: Path, orient, size=(256, 256)):
+    path.parent.mkdir(parents=True, exist_ok=True)
+    w, h = size
+    im = Image.new("L", size, 0)
+    px = im.load()
+    for y in range(h):
+        for x in range(w):
+            if (x / w if orient == "v" else y / h) >= 0.5:
+                px[x, y] = 255
+    im.convert("RGB").save(path, "JPEG")
+
+
+def _sidecar(media: Path, payload: dict):
+    media.with_suffix(".json").write_text(json.dumps(payload))
+
+
+def test_apply_sidecar_idempotent_on_deep_rederive(
+    importer, deep_importer, import_layout,
+):
+    """Normal-flow path: deep rederive on an already-imported image finds
+    the existing provenance row via .scalar_one_or_none() and skips the
+    insert. Exactly one ImageProvenance row after both runs."""
+    import_root, _ = import_layout
+    m = import_root / "Alice" / "a.jpg"
+    _split(m, "v")
+    _sidecar(m, {
+        "category": "patreon", "id": 555,
+        "url": "https://patreon.com/posts/555", "title": "Set 1",
+    })
+    r = importer.import_one(m)
+    assert r.status == "imported"
+
+    # Re-import via deep mode — sha matches → _deep_rederive → _apply_sidecar.
+    r2 = deep_importer.import_one(m)
+    assert r2.status == "refreshed"
+
+    count = importer.session.execute(
+        select(func.count()).select_from(ImageProvenance)
+    ).scalar_one()
+    assert count == 1
+
+
+def test_apply_sidecar_recovers_from_integrity_error(
+    importer, deep_importer, import_layout, db_sync, monkeypatch,
+):
+    """Race recovery: a row already exists for (image, post). We force the
+    importer's existence-check SELECT to return None for one call, mimicking
+    the race window where two workers both saw no row. The savepoint INSERT
+    then trips uq_image_provenance_image_post; the helper rolls the
+    savepoint back, no exception escapes, and the row count stays at 1.
+    """
+    import_root, _ = import_layout
+    m = import_root / "Bob" / "b.jpg"
+    _split(m, "v")
+    _sidecar(m, {
+        "category": "patreon", "id": 777,
+        "url": "https://patreon.com/posts/777", "title": "Set 2",
+    })
+    # First import lays the canonical provenance row.
+    r = importer.import_one(m)
+    assert r.status == "imported"
+    rec = importer.session.get(ImageRecord, r.image_id)
+    src = importer.session.execute(select(Source)).scalar_one()
+    assert rec is not None
+    assert src is not None
+
+    # Monkeypatch session.execute so the FIRST select inside _apply_sidecar's
+    # existence-check returns a "no row" wrapper. Subsequent selects (e.g.
+    # the find_or_create_source / find_or_create_post existence checks
+    # earlier in _apply_sidecar) all run normally; we intercept only the
+    # ImageProvenance lookup, identified by the SELECT's target columns
+    # mentioning image_provenance.
+    real_execute = db_sync.execute
+    intercepted = [False]
+
+    def _intercepting_execute(stmt, *args, **kwargs):
+        text = str(stmt)
+        if (
+            not intercepted[0]
+            and "image_provenance" in text
+            and "image_record_id" in text
+            and "post_id" in text
+        ):
+            intercepted[0] = True
+
+            class _ForcedMiss:
+                def scalar_one_or_none(self):
+                    return None
+            return _ForcedMiss()
+        return real_execute(stmt, *args, **kwargs)
+
+    monkeypatch.setattr(db_sync, "execute", _intercepting_execute)
+
+    # Re-import via deep mode → _deep_rederive → _apply_sidecar. With the
+    # provenance-SELECT forced to miss, the helper will attempt the INSERT,
+    # trip uq_image_provenance_image_post, catch IntegrityError, and recover.
+    r2 = deep_importer.import_one(m)
+    assert r2.status == "refreshed"
+
+    # Lift the intercept, then verify the row count.
+    monkeypatch.undo()
+    count = db_sync.execute(
+        select(func.count()).select_from(ImageProvenance)
+    ).scalar_one()
+    assert count == 1

tests/test_importer_upsert_helpers.py

@@ -145,3 +145,68 @@ def test_find_or_create_source_recovers_from_integrity_error(
         artist_id=artist_row.id, platform="patreon", url=canonical_url,
     )
     assert recovered.id == pre_existing.id
+
+
+def test_source_for_sidecar_reuses_existing_subscription(
+    importer, artist_row, db_sync,
+):
+    """The filesystem-import sidecar resolver should attach to whatever
+    Source already exists for (artist, platform) — the canonical subscription
+    Source — regardless of its URL. Without this, every imported post
+    spawned its own Source row.
+    """
+    canonical = Source(
+        artist_id=artist_row.id, platform="patreon",
+        url="https://www.patreon.com/cw/testartist", enabled=True,
+    )
+    db_sync.add(canonical)
+    db_sync.flush()
+
+    resolved = importer._source_for_sidecar(
+        artist_id=artist_row.id, platform="patreon",
+        artist_slug=artist_row.slug,
+    )
+    assert resolved.id == canonical.id
+
+
+def test_source_for_sidecar_creates_synthetic_anchor_when_none_exists(
+    importer, artist_row, db_sync,
+):
+    """No subscription Source for this (artist, platform) yet. The helper
+    creates one synthetic anchor (enabled=False, url='sidecar:<plat>:<slug>')
+    so subsequent imports reuse it instead of spawning per-post Sources.
+    """
+    resolved = importer._source_for_sidecar(
+        artist_id=artist_row.id, platform="pixiv",
+        artist_slug=artist_row.slug,
+    )
+    assert resolved.url == f"sidecar:pixiv:{artist_row.slug}"
+    assert resolved.enabled is False
+    assert resolved.artist_id == artist_row.id
+    assert resolved.platform == "pixiv"
+
+    # Second call returns the same row (no new Source spawned).
+    again = importer._source_for_sidecar(
+        artist_id=artist_row.id, platform="pixiv",
+        artist_slug=artist_row.slug,
+    )
+    assert again.id == resolved.id
+
+
+def test_source_for_sidecar_distinct_platforms_distinct_anchors(
+    importer, artist_row, db_sync,
+):
+    """One synthetic anchor per (artist, platform). Different platforms get
+    different anchors even when no campaign Source exists for either.
+    """
+    p = importer._source_for_sidecar(
+        artist_id=artist_row.id, platform="patreon",
+        artist_slug=artist_row.slug,
+    )
+    x = importer._source_for_sidecar(
+        artist_id=artist_row.id, platform="pixiv",
+        artist_slug=artist_row.slug,
+    )
+    assert p.id != x.id
+    assert p.platform == "patreon"
+    assert x.platform == "pixiv"

tests/test_tasks_register.py

@@ -23,3 +23,7 @@ def test_import_media_file_registered():
 
 def test_generate_thumbnail_registered():
     assert "backend.app.tasks.thumbnail.generate_thumbnail" in celery.tasks
+
+
+def test_backfill_thumbnails_registered():
+    assert "backend.app.tasks.thumbnail.backfill_thumbnails" in celery.tasks