fix: M4 Phase 1 bugs - dev.chunks refs, search_path, uuid column

Bug fixes from M4 report:
- 4 remaining dev.chunks → dev.chunk in SQL queries
- search_path includes public for pgvector extension
- get_chunk_by_chunk_id_and_uuid: uuid → file_uuid
- New endpoint: GET /api/v1/file/:uuid/chunk/:chunk_id

src/api/five_w1h_agent_api.rs

@@ -776,7 +776,7 @@ pub async fn run_5w1h_agent(db: &PostgresDb, file_uuid: &str) -> anyhow::Result<
         match embedder.embed_document(text).await {
             Ok(vector) => {
                 if let Err(e) = sqlx::query(
-                    "UPDATE dev.chunks SET embedding = $1::vector WHERE chunk_id = $2 AND file_uuid = $3"
+                    "UPDATE dev.chunk SET embedding = $1::vector WHERE chunk_id = $2 AND file_uuid = $3"
                 )
                 .bind(&vector as &[f32])
                 .bind(chunk_id)

src/api/server.rs

@@ -1306,6 +1306,25 @@ async fn trigger_processing(
     })))
 }
 
+async fn get_chunk_by_path(
+    Path((file_uuid, chunk_id)): Path<(String, String)>,
+    State(state): State<AppState>,
+) -> Result<Json<Chunk>, StatusCode> {
+    let chunk = state
+        .db
+        .get_chunk_by_chunk_id_and_uuid(&chunk_id, &file_uuid)
+        .await
+        .map_err(|_| {
+            tracing::error!("[get_chunk_by_path] DB error: {}:{}", file_uuid, chunk_id);
+            StatusCode::INTERNAL_SERVER_ERROR
+        })?
+        .ok_or_else(|| {
+            tracing::warn!("[get_chunk_by_path] Not found: {}:{}", file_uuid, chunk_id);
+            StatusCode::NOT_FOUND
+        })?;
+    Ok(Json(chunk))
+}
+
 async fn get_asset_status(
     State(state): State<AppState>,
     Path(uuid): Path<String>,
@@ -2508,6 +2527,7 @@ pub async fn start_server(host: &str, port: u16) -> anyhow::Result<()> {
         .route("/api/v1/files/scan", get(scan_files))
         .route("/api/v1/file/:file_uuid/probe", get(probe_by_uuid))
         .route("/api/v1/file/:file_uuid/process", post(trigger_processing))
+        .route("/api/v1/file/:file_uuid/chunk/:chunk_id", get(get_chunk_by_path))
 
         .route("/api/v1/progress/:uuid", get(get_progress))
         .route("/api/v1/jobs", get(list_jobs))

src/core/db/postgres_db.rs

@@ -741,8 +741,8 @@ impl PostgresDb {
                 let schema = schema.to_string();
                 tracing::debug!("after_connect: setting search_path to {}", schema);
                 Box::pin(async move {
-                    // Always set search_path explicitly to avoid using default "dev, public"
-                    sqlx::query(&format!("SET search_path TO {}", schema))
+                    // Include public schema for pgvector extension
+                    sqlx::query(&format!("SET search_path TO {}, public", schema))
                         .execute(conn)
                         .await?;
                     Ok(())
@@ -2774,7 +2774,7 @@ impl PostgresDb {
     ) -> Result<Option<Chunk>> {
         let table = "dev.chunk";
         let row = sqlx::query(&format!(
-            "SELECT COALESCE(file_id, 0) as file_id, uuid, chunk_id, chunk_type, COALESCE(fps, 24.0) as fps, COALESCE(start_frame, 0) as start_frame, COALESCE(end_frame, 0) as end_frame, text_content, content, metadata, vector_id, COALESCE(frame_count, 0) as frame_count, pre_chunk_ids, parent_chunk_id, child_chunk_ids, visual_stats FROM {} WHERE chunk_id = $1 AND uuid = $2",
+            "SELECT COALESCE(file_id, 0) as file_id, file_uuid, chunk_id, chunk_type, COALESCE(fps, 24.0) as fps, COALESCE(start_frame, 0) as start_frame, COALESCE(end_frame, 0) as end_frame, text_content, content, metadata, vector_id, COALESCE(frame_count, 0) as frame_count, pre_chunk_ids, parent_chunk_id, child_chunk_ids, visual_stats FROM {} WHERE chunk_id = $1 AND file_uuid = $2",
             table
         ))
             .bind(chunk_id)
@@ -2821,7 +2821,7 @@ impl PostgresDb {
 
             Ok(Some(Chunk {
                 file_id,
-                uuid: r.get("uuid"),
+                uuid: r.get("file_uuid"),
                 chunk_id: r.get("chunk_id"),
 
                 chunk_type,
@@ -4623,7 +4623,7 @@ impl PostgresDb {
                 COALESCE(summary_text, text_content, '') as summary,
                 metadata,
                 (1 - (embedding <=> $1::vector)) as similarity
-            FROM dev.chunks
+            FROM dev.chunk
             WHERE file_uuid = $2 AND chunk_type = 'cut' AND embedding IS NOT NULL
             ORDER BY embedding <=> $1::vector
             LIMIT $3

src/worker/job_worker.rs

@@ -1002,7 +1002,7 @@ impl JobWorker {
         let pool = db.pool();
 
         let rows = sqlx::query_as::<_, (String, String, String, f64, f64, String)>(
-            "SELECT chunk_id, chunk_type, text_content, start_time, end_time, content::text FROM dev.chunks WHERE file_uuid = $1 AND chunk_type = 'sentence' AND embedding IS NULL AND (text_content IS NOT NULL AND text_content != '') ORDER BY chunk_index",
+            "SELECT chunk_id, chunk_type, text_content, start_time, end_time, content::text FROM dev.chunk WHERE file_uuid = $1 AND chunk_type = 'sentence' AND embedding IS NULL AND (text_content IS NOT NULL AND text_content != '') ORDER BY id",
         )
         .bind(uuid)
         .fetch_all(pool)

src/worker/processor.rs

@@ -1080,7 +1080,7 @@ impl ProcessorPool {
                 "top_5": scene.top_5,
             });
             let _ = sqlx::query(
-                "UPDATE dev.chunks SET metadata = metadata || $1::jsonb WHERE file_uuid=$2 AND chunk_id=$3"
+                "UPDATE dev.chunk SET metadata = metadata || $1::jsonb WHERE file_uuid=$2 AND chunk_id=$3"
             )
             .bind(&meta)
             .bind(uuid)