From 2b7688bd752bc76736016dafee6070515f35c4d6 Mon Sep 17 00:00:00 2001
From: Codex <codex@localhost>
Date: Thu, 23 Jul 2026 02:18:11 +0000
Subject: [PATCH] fix(file-index): prune removed active assets on rebuild

Treat the source registry as authoritative during file-index rebuilds while preserving staged uploads. This prevents deleted archived sources from remaining searchable.

Tests: python3 -m unittest tests.test_file_assets tests.test_business_facts
---
 obsidian_wiki/file_assets.py | 34 ++++++++++++++++++++++------------
 tests/test_file_assets.py    | 26 ++++++++++++++++++++++++++
 2 files changed, 48 insertions(+), 12 deletions(-)

diff --git a/obsidian_wiki/file_assets.py b/obsidian_wiki/file_assets.py
index 74c261c..dafdca0 100644
--- a/obsidian_wiki/file_assets.py
+++ b/obsidian_wiki/file_assets.py
@@ -185,21 +185,31 @@ def rebuild_from_registry(vault_path: str | Path, *, db_path: Path | None = None
     if not registry_path.is_file():
         raise FileNotFoundError(f"source registry not found: {registry_path}")
 
-    indexed = 0
+    records: list[dict[str, Any]] = []
     skipped = 0
+    for raw in registry_path.read_text(encoding="utf-8").splitlines():
+        if not raw.strip():
+            continue
+        try:
+            record = json.loads(raw)
+        except json.JSONDecodeError:
+            skipped += 1
+            continue
+        if record.get("record_type") == "source_archive":
+            records.append(record)
+
     with connect(db_path) as conn:
-        for raw in registry_path.read_text(encoding="utf-8").splitlines():
-            if not raw.strip():
-                continue
-            try:
-                record = json.loads(raw)
-            except json.JSONDecodeError:
-                skipped += 1
-                continue
-            if record.get("record_type") != "source_archive":
-                continue
+        # A rebuild is authoritative for archived sources. Preserve uploaded
+        # staging rows, but discard active rows that no longer have a registry
+        # record so deleted sources cannot remain searchable.
+        conn.execute(
+            "DELETE FROM file_assets_fts WHERE file_id IN (SELECT file_id FROM file_assets WHERE status = 'active')"
+        )
+        conn.execute("DELETE FROM file_assets WHERE status = 'active'")
+        for record in records:
             upsert_asset(conn, asset_from_registry_record(record), status="active")
-            indexed += 1
+        conn.execute("DELETE FROM file_assets_fts WHERE file_id NOT IN (SELECT file_id FROM file_assets)")
+    indexed = len(records)
     return RebuildResult(db_path=db_path, registry_path=registry_path, indexed=indexed, skipped=skipped)
 
 
diff --git a/tests/test_file_assets.py b/tests/test_file_assets.py
index 3ee4e57..27c76dd 100644
--- a/tests/test_file_assets.py
+++ b/tests/test_file_assets.py
@@ -55,6 +55,32 @@ class FileAssetIndexTests(unittest.TestCase):
         self.assertEqual(json.loads(rows[0]["pitch"]), ["0.5"])
         self.assertEqual(json.loads(rows[0]["pin_count"]), ["18"])
 
+    def test_rebuild_removes_active_records_removed_from_registry(self):
+        retained_id = "sha256:" + "c" * 64
+        removed_id = "sha256:" + "d" * 64
+        self._write_registry_record(
+            source_id=retained_id,
+            archive_path="_sources/archive/bjd/pdf/retained.pdf",
+            original_name="retained.pdf",
+            title="Retained source",
+        )
+        self._write_registry_record(
+            source_id=removed_id,
+            archive_path="_sources/archive/bjd/xlsx/removed.xlsx",
+            original_name="removed.xlsx",
+            title="Removed source",
+        )
+        rebuild_from_registry(self.vault, db_path=self.db_path)
+
+        retained_record = json.loads(self.registry_path.read_text(encoding="utf-8").splitlines()[0])
+        self.registry_path.write_text(json.dumps(retained_record) + "\n", encoding="utf-8")
+        rebuild_from_registry(self.vault, db_path=self.db_path)
+
+        remaining = search_assets(self.vault, "retained", db_path=self.db_path, limit=5, read_only=True)
+        removed = search_assets(self.vault, "removed", db_path=self.db_path, limit=5, read_only=True)
+        self.assertEqual([row["file_id"] for row in remaining], [retained_id])
+        self.assertEqual(removed, [])
+
     def _write_registry_record(self, **fields):
         record = {
             "record_type": "source_archive",