最新哦，分词策略放到cli了

src/web/server.py

@@ -27,13 +27,21 @@ if str(_SRC) not in sys.path:
 from agent import chatbot_agent as ca  # noqa: E402
 from web import ui_session as ui  # noqa: E402
 from agent.utils.image_utils import IMAGE_FILE_EXTENSIONS  # noqa: E402
-from agent.utils.agent_utils import DOC_EXT_MAP  # noqa: E402
 
 _WEB_DIR = Path(__file__).resolve().parent
 _IMAGE_DIR = _WEB_DIR / "image"
 
 _AUDIO_EXT = frozenset({".wav", ".mp3", ".m4a", ".flac", ".ogg", ".webm"})
-_FILE_EXT = frozenset(DOC_EXT_MAP.keys())
+_IMAGE_EXT = frozenset(IMAGE_FILE_EXTENSIONS)
+
+
+def _infer_upload_kind(fp: str) -> str | None:
+    ext = os.path.splitext(fp)[1].lower()
+    if ext in _IMAGE_EXT:
+        return "image"
+    if ext in _AUDIO_EXT:
+        return "audio"
+    return "file"
 
 
 def _data_url_for_file(fp: str) -> str | None:
@@ -84,6 +92,7 @@ def _serialize_content(content: Any, *, meta: dict[str, Any] | None = None, role
             parts.append({"type": "text", "text": piece.get("text") or ""})
             continue
         fp = piece.get("path") or (piece.get("file") or {}).get("path")
+        display_name = piece.get("name")
         if piece.get("type") == "image_url":
             fp = fp or (piece.get("image_url") or {}).get("url")
         if piece.get("type") == "audio":
@@ -95,19 +104,19 @@ def _serialize_content(content: Any, *, meta: dict[str, Any] | None = None, role
             continue
         ext = os.path.splitext(fp)[1].lower()
         if ext in IMAGE_FILE_EXTENSIONS:
-            parts.append({"type": "image", "url": data_url, "path": fp, "name": os.path.basename(fp)})
+            parts.append({"type": "image", "url": data_url, "path": fp, "name": display_name or os.path.basename(fp)})
         elif ext in _AUDIO_EXT:
             audio_meta = meta.get("audio") if isinstance(meta.get("audio"), dict) else {}
             parts.append({
                 "type": "audio",
                 "url": data_url,
                 "path": fp,
-                "name": os.path.basename(fp),
+                "name": display_name or os.path.basename(fp),
                 "autoplay": bool(meta.get("autoplay") or audio_meta.get("autoplay")) if role == "assistant" else False,
                 "duration_seconds": audio_meta.get("duration_seconds"),
             })
         else:
-            parts.append({"type": "file", "url": data_url, "path": fp, "name": os.path.basename(fp)})
+            parts.append({"type": "file", "url": data_url, "path": fp, "name": display_name or os.path.basename(fp)})
     return parts
 
 
@@ -244,6 +253,7 @@ def _materialize_uploads_to_temp(
     """将上传文件落盘到临时目录，返回 ``(query_dict, tmp_root)``；调用方须在 finally 中删除 ``tmp_root``。"""
     tmp_root = tempfile.mkdtemp(prefix="web_mm_")
     file_paths: list[str] = []
+    file_names: list[str] = []
     for uf in images or []:
         if uf is None or not uf.filename:
             continue
@@ -252,6 +262,7 @@ def _materialize_uploads_to_temp(
         with open(dest, "wb") as f:
             shutil.copyfileobj(uf.file, f)
         file_paths.append(dest)
+        file_names.append(uf.filename)
 
     for uf in files or []:
         if uf is None or not uf.filename:
@@ -261,6 +272,7 @@ def _materialize_uploads_to_temp(
         with open(dest, "wb") as f:
             shutil.copyfileobj(uf.file, f)
         file_paths.append(dest)
+        file_names.append(uf.filename)
 
     if audio is not None and audio.filename:
         ext = Path(audio.filename).suffix or ".bin"
@@ -268,8 +280,9 @@ def _materialize_uploads_to_temp(
         with open(dest, "wb") as f:
             shutil.copyfileobj(audio.file, f)
         file_paths.append(dest)
+        file_names.append(audio.filename)
 
-    query = {"text": (text or "").strip(), "files": file_paths}
+    query = {"text": (text or "").strip(), "files": file_paths, "file_names": file_names}
     return query, tmp_root
 
 
@@ -417,4 +430,4 @@ def serve_css() -> FileResponse:
 
 @app.get("/app.js")
 def serve_js() -> FileResponse:
-    return FileResponse(_WEB_DIR / "app.js")
+    return FileResponse(_WEB_DIR / "app.js")