Add AI search query expansion

app/main.py

@@ -5,12 +5,12 @@ from fastapi import Depends, FastAPI, File, Form, HTTPException, Request, Upload
 from fastapi.responses import FileResponse, RedirectResponse, Response
 from fastapi.staticfiles import StaticFiles
 from fastapi.templating import Jinja2Templates
-from sqlalchemy import func, or_
+from sqlalchemy import func, false, or_
 from sqlalchemy.orm import Session
 
 from app.db import get_db, init_db
 from app.images import process_upload
-from app.llm import test_connection
+from app.llm import expand_query, is_configured, test_connection
 from app.llm import LLMResult
 from app.models import Box, Item, SubItem
 from app.settings_store import LLMConfig, get_app_settings, save_app_settings
@@ -160,100 +160,130 @@ def _build_boxes_overview_summary(db: Session) -> dict[str, int | str]:
     }
 
 
-def _build_search_results(db: Session, query: str) -> list[dict]:
-    keyword = f"%{query.lower()}%"
+def _build_search_results(db: Session, query: str | list[str]) -> list[dict]:
+    """Search Box / Item / SubItem by name and note using case-insensitive LIKE.
+
+    Accepts either a single query string or a list of keywords.
+    When multiple keywords are given, they are combined with OR — a match on
+    *any* keyword is sufficient.
+    """
+    keywords = [query] if isinstance(query, str) else query
+    patterns = [f"%{kw.lower()}%" for kw in keywords]
+
+    def _or_like(column, note_column):
+        """Build an OR filter that matches any pattern on either column."""
+        conditions = []
+        for pat in patterns:
+            conditions.append(func.lower(column).like(pat))
+            conditions.append(func.lower(func.coalesce(note_column, "")).like(pat))
+        return or_(false(), *conditions) if conditions else false()
+
     results: list[dict] = []
+    seen_ids: set[tuple[str, int]] = set()
+
+    def _add(result_type: str, obj_id: int, entry: dict) -> None:
+        key = (result_type, obj_id)
+        if key not in seen_ids:
+            seen_ids.add(key)
+            results.append(entry)
 
     box_matches = (
         db.query(Box)
-        .filter(
-            or_(
-                func.lower(Box.name).like(keyword),
-                func.lower(func.coalesce(Box.note, "")).like(keyword),
-            )
-        )
+        .filter(_or_like(Box.name, Box.note))
         .order_by(Box.id.desc())
         .all()
     )
     for box in box_matches:
-        results.append(
-            {
-                "type": "Box",
-                "name": box.name,
-                "note": box.note,
-                "detail_url": f"/boxes/{box.id}",
-                "detail_label": "查看箱子",
-                "secondary_url": None,
-                "secondary_label": None,
-                "path": "顶层箱子",
-                "is_container": None,
-                "image_url": f"/boxes/{box.id}/image" if box.image_blob else None,
-            }
-        )
+        _add("Box", box.id, {
+            "type": "Box",
+            "name": box.name,
+            "note": box.note,
+            "detail_url": f"/boxes/{box.id}",
+            "detail_label": "查看箱子",
+            "secondary_url": None,
+            "secondary_label": None,
+            "path": "顶层箱子",
+            "is_container": None,
+            "image_url": f"/boxes/{box.id}/image" if box.image_blob else None,
+        })
 
     item_matches = (
         db.query(Item)
         .join(Item.box)
-        .filter(
-            or_(
-                func.lower(Item.name).like(keyword),
-                func.lower(func.coalesce(Item.note, "")).like(keyword),
-            )
-        )
+        .filter(_or_like(Item.name, Item.note))
         .order_by(Item.id.desc())
         .all()
     )
     for item in item_matches:
-        results.append(
-            {
-                "type": "Item",
-                "name": item.name,
-                "note": item.note,
-                "detail_url": f"/items/{item.id}",
-                "detail_label": "查看物品",
-                "secondary_url": f"/boxes/{item.box.id}",
-                "secondary_label": "查看所属箱子",
-                "path": f"位于箱子：{item.box.name}",
-                "is_container": item.is_container,
-                "image_url": f"/items/{item.id}/image" if item.image_blob else None,
-            }
-        )
+        _add("Item", item.id, {
+            "type": "Item",
+            "name": item.name,
+            "note": item.note,
+            "detail_url": f"/items/{item.id}",
+            "detail_label": "查看物品",
+            "secondary_url": f"/boxes/{item.box.id}",
+            "secondary_label": "查看所属箱子",
+            "path": f"位于箱子：{item.box.name}",
+            "is_container": item.is_container,
+            "image_url": f"/items/{item.id}/image" if item.image_blob else None,
+        })
 
     subitem_matches = (
         db.query(SubItem)
         .join(SubItem.parent_item)
         .join(Item.box)
-        .filter(
-            or_(
-                func.lower(SubItem.name).like(keyword),
-                func.lower(func.coalesce(SubItem.note, "")).like(keyword),
-            )
-        )
+        .filter(_or_like(SubItem.name, SubItem.note))
         .order_by(SubItem.id.desc())
         .all()
     )
     for subitem in subitem_matches:
-        results.append(
-            {
-                "type": "SubItem",
-                "name": subitem.name,
-                "note": subitem.note,
-                "detail_url": f"/items/{subitem.parent_item.id}",
-                "detail_label": "查看所属物品",
-                "secondary_url": f"/boxes/{subitem.parent_item.box.id}",
-                "secondary_label": "查看所属箱子",
-                "path": (
-                    f"位于物品：{subitem.parent_item.name} / "
-                    f"箱子：{subitem.parent_item.box.name}"
-                ),
-                "is_container": None,
-                "image_url": f"/subitems/{subitem.id}/image" if subitem.image_blob else None,
-            }
-        )
+        _add("SubItem", subitem.id, {
+            "type": "SubItem",
+            "name": subitem.name,
+            "note": subitem.note,
+            "detail_url": f"/items/{subitem.parent_item.id}",
+            "detail_label": "查看所属物品",
+            "secondary_url": f"/boxes/{subitem.parent_item.box.id}",
+            "secondary_label": "查看所属箱子",
+            "path": (
+                f"位于物品：{subitem.parent_item.name} / "
+                f"箱子：{subitem.parent_item.box.name}"
+            ),
+            "is_container": None,
+            "image_url": f"/subitems/{subitem.id}/image" if subitem.image_blob else None,
+        })
 
     return results
 
 
+def _ai_search(db: Session, cfg: "LLMConfig", query: str) -> tuple[list[str], list[dict], str | None]:
+    """Swappable AI search seam.
+
+    Returns ``(expanded_terms, results, error_message)``.
+    - On success: expanded terms + broadened results, ``error_message`` is ``None``.
+    - On failure (timeout, network error, HTTP error): empty terms + normal LIKE
+      results + friendly error message.
+    - On empty expansion (model returned ``[]`` legitimately): empty terms + normal
+      results, ``error_message`` is ``None``.
+    """
+    result = expand_query(cfg, query, extra_hints=cfg.ai_search_extra_hints)
+
+    if result.error:
+        # Real failure (timeout / network / HTTP) → show error + fallback
+        results = _build_search_results(db, query)
+        return [], results, result.error
+
+    if not result.terms:
+        # Legitimate empty expansion → normal results, no error
+        results = _build_search_results(db, query)
+        return [], results, None
+
+    # Deduplicate: original query + expanded terms
+    all_terms = list(dict.fromkeys([query] + result.terms))
+    results = _build_search_results(db, all_terms)
+    return result.terms, results, None
+
+
 def create_app() -> FastAPI:
     @asynccontextmanager
     async def lifespan(app: FastAPI):
@@ -285,10 +315,28 @@ def create_app() -> FastAPI:
     def search_page(
         request: Request,
         q: str | None = None,
+        ai: str | None = None,
         db: Session = Depends(get_db),
     ):
         query = (q or "").strip()
-        results = _build_search_results(db, query) if query else []
+        cfg = get_app_settings(db)
+        ai_requested = ai == "1"
+        ai_available = cfg.ai_search_enabled and is_configured(cfg)
+        expanded_terms: list[str] = []
+        ai_error: str | None = None
+
+        if query:
+            if ai_requested and ai_available:
+                try:
+                    expanded_terms, results, ai_error = _ai_search(db, cfg, query)
+                except Exception:  # noqa: BLE001 — graceful degradation
+                    ai_error = "AI 搜索暂时不可用，已回退到普通搜索。"
+                    results = _build_search_results(db, query)
+            else:
+                results = _build_search_results(db, query)
+        else:
+            results = []
+
         return templates.TemplateResponse(
             request=request,
             name="search/index.html",
@@ -297,6 +345,10 @@ def create_app() -> FastAPI:
                 "query": query,
                 "results": results,
                 "searched": bool(query),
+                "ai_activated": ai_requested and ai_available and bool(query),
+                "expanded_terms": expanded_terms,
+                "ai_error": ai_error,
+                "ai_available": ai_available,
             },
         )
 
@@ -335,6 +387,8 @@ def create_app() -> FastAPI:
         base_url: str | None = Form(default=None),
         model: str | None = Form(default=None),
         api_key: str | None = Form(default=None),
+        ai_search_enabled: str | None = Form(default=None),
+        ai_search_extra_hints: str | None = Form(default=None),
         db: Session = Depends(get_db),
     ) -> Response:
         # Origin/Referer check for browser requests
@@ -370,6 +424,8 @@ def create_app() -> FastAPI:
                         base_url=resolved_base_url,
                         model=resolved_model,
                         api_key=existing_cfg.api_key,
+                        ai_search_enabled=ai_search_enabled == "on",
+                        ai_search_extra_hints=_clean_text(ai_search_extra_hints) or "",
                     ),
                     "api_key_configured": bool(existing_cfg.api_key),
                     "test_result": LLMResult(
@@ -382,12 +438,16 @@ def create_app() -> FastAPI:
         # submitted_key is None → keep old key; str (including "") → use new value
         resolved_api_key = submitted_key
 
+        resolved_extra_hints = _clean_text(ai_search_extra_hints) or ""
+
         save_app_settings(
             db,
             enabled=enabled == "on",
             base_url=resolved_base_url,
             model=resolved_model,
             api_key=resolved_api_key,
+            ai_search_enabled=ai_search_enabled == "on",
+            ai_search_extra_hints=resolved_extra_hints,
         )
         return RedirectResponse(url="/settings", status_code=status.HTTP_303_SEE_OTHER)
 
@@ -398,6 +458,8 @@ def create_app() -> FastAPI:
         base_url: str | None = Form(default=None),
         model: str | None = Form(default=None),
         api_key: str | None = Form(default=None),
+        ai_search_enabled: str | None = Form(default=None),
+        ai_search_extra_hints: str | None = Form(default=None),
         db: Session = Depends(get_db),
     ):
         # Origin/Referer check for browser requests
@@ -436,6 +498,8 @@ def create_app() -> FastAPI:
                         base_url=resolved_base_url,
                         model=resolved_model,
                         api_key="",
+                        ai_search_enabled=ai_search_enabled == "on",
+                        ai_search_extra_hints=_clean_text(ai_search_extra_hints) or "",
                     ),
                     "api_key_configured": bool(existing_cfg.api_key),
                     "test_result": LLMResult(
@@ -450,6 +514,8 @@ def create_app() -> FastAPI:
             base_url=resolved_base_url,
             model=resolved_model,
             api_key=resolved_api_key or "",
+            ai_search_enabled=ai_search_enabled == "on",
+            ai_search_extra_hints=_clean_text(ai_search_extra_hints) or "",
         )
 
         result = test_connection(test_cfg)