fix(server): unwrap sparse query embedding generator

mcp_plex/loader.py

@@ -10,25 +10,25 @@
 import click
 import httpx
 from fastembed import SparseTextEmbedding, TextEmbedding
-from qdrant_client.async_qdrant_client import AsyncQdrantClient
 from qdrant_client import models
+from qdrant_client.async_qdrant_client import AsyncQdrantClient
 
 from .types import (
     AggregatedItem,
     ExternalIDs,
     IMDbTitle,
+    PlexGuid,
     PlexItem,
+    PlexPerson,
     TMDBEpisode,
     TMDBItem,
     TMDBMovie,
     TMDBShow,
-    PlexGuid,
-    PlexPerson,
 )
 
 try:  # Only import plexapi when available; the sample data mode does not require it.
-    from plexapi.server import PlexServer
     from plexapi.base import PlexPartialObject
+    from plexapi.server import PlexServer
 except Exception:
     PlexServer = None  # type: ignore[assignment]
     PlexPartialObject = object  # type: ignore[assignment]

mcp_plex/server.py

@@ -2,16 +2,21 @@
 from __future__ import annotations
 
 import asyncio
-import os
 import json
+import os
 from collections import OrderedDict
-from typing import Any, Annotated
+from typing import Annotated, Any
 
+from fastembed import SparseTextEmbedding, TextEmbedding
 from fastmcp.server import FastMCP
-from qdrant_client.async_qdrant_client import AsyncQdrantClient
-from qdrant_client import models
-from fastembed import TextEmbedding, SparseTextEmbedding
 from pydantic import Field
+from qdrant_client import models
+from qdrant_client.async_qdrant_client import AsyncQdrantClient
+
+try:
+    from sentence_transformers import CrossEncoder
+except Exception:
+    CrossEncoder = None
 
 try:
     from sentence_transformers import CrossEncoder
@@ -151,7 +156,7 @@ async def search_media(
 ) -> list[dict[str, Any]]:
     """Hybrid similarity search across media items using dense and sparse vectors."""
     dense_task = asyncio.to_thread(lambda: list(_dense_model.embed([query]))[0])
-    sparse_task = asyncio.to_thread(lambda: _sparse_model.query_embed(query))
+    sparse_task = asyncio.to_thread(lambda: next(_sparse_model.query_embed(query)))
     dense_vec, sparse_vec = await asyncio.gather(dense_task, sparse_task)
     named_dense = models.NamedVector(name="dense", vector=dense_vec)
     sv = models.SparseVector(

mcp_plex/types.py

@@ -2,7 +2,7 @@
 from __future__ import annotations
 
 from dataclasses import dataclass
-from typing import List, Optional, Literal
+from typing import List, Literal, Optional
 
 from pydantic import BaseModel, Field
 

tests/test_load_from_plex.py

@@ -1,5 +1,6 @@
 import asyncio
 import types
+
 import httpx
 
 from mcp_plex import loader

tests/test_loader_cli.py

@@ -1,6 +1,7 @@
 import asyncio
-from click.testing import CliRunner
+
 import pytest
+from click.testing import CliRunner
 
 from mcp_plex import loader
 

tests/test_loader_integration.py

@@ -2,9 +2,10 @@
 from pathlib import Path
 from types import SimpleNamespace
 
-from mcp_plex import loader
 from qdrant_client import models
 
+from mcp_plex import loader
+
 
 class DummyTextEmbedding:
     def __init__(self, name: str):

tests/test_loader_unit.py

@@ -1,17 +1,17 @@
-import types
 import asyncio
-import httpx
-
+import types
 from pathlib import Path
 
+import httpx
+
 from mcp_plex.loader import (
-    _extract_external_ids,
-    _load_from_sample,
     _build_plex_item,
+    _extract_external_ids,
     _fetch_imdb,
+    _fetch_tmdb_episode,
     _fetch_tmdb_movie,
     _fetch_tmdb_show,
-    _fetch_tmdb_episode,
+    _load_from_sample,
 )
 
 

tests/test_server.py

@@ -1,15 +1,16 @@
-from typing import Any
 import asyncio
-from pathlib import Path
 import importlib
-import types
 import json
-import time
 import sys
+import time
+import types
+from pathlib import Path
+from typing import Any
+
 import pytest
+from qdrant_client import models
 
 from mcp_plex import loader
-from qdrant_client import models
 
 
 class DummyTextEmbedding:
@@ -51,7 +52,7 @@ def passage_embed(self, texts):
 
     def query_embed(self, text):
         time.sleep(0.1)
-        return DummySparseVector([0], [1.0])
+        yield DummySparseVector([0], [1.0])
 
 
 class DummyReranker: