agno 2.1.4__py3-none-any.whl → 2.1.6__py3-none-any.whl

agno/os/routers/knowledge/knowledge.py

@@ -19,6 +19,9 @@ from agno.os.routers.knowledge.schemas import (
     ContentStatusResponse,
     ContentUpdateSchema,
     ReaderSchema,
+    VectorDbSchema,
+    VectorSearchRequestSchema,
+    VectorSearchResult,
 )
 from agno.os.schema import (
     BadRequestResponse,
@@ -303,7 +306,7 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
             }
         },
     )
-    def get_content(
+    async def get_content(
         limit: Optional[int] = Query(default=20, description="Number of content entries to return"),
         page: Optional[int] = Query(default=1, description="Page number"),
         sort_by: Optional[str] = Query(default="created_at", description="Field to sort by"),
@@ -311,7 +314,7 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
         db_id: Optional[str] = Query(default=None, description="The ID of the database to use"),
     ) -> PaginatedResponse[ContentResponseSchema]:
         knowledge = get_knowledge_instance_by_db_id(knowledge_instances, db_id)
-        contents, count = knowledge.get_content(limit=limit, page=page, sort_by=sort_by, sort_order=sort_order)
+        contents, count = await knowledge.aget_content(limit=limit, page=page, sort_by=sort_by, sort_order=sort_order)
 
         return PaginatedResponse(
             data=[
@@ -371,13 +374,13 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
             404: {"description": "Content not found", "model": NotFoundResponse},
         },
     )
-    def get_content_by_id(
+    async def get_content_by_id(
         content_id: str,
         db_id: Optional[str] = Query(default=None, description="The ID of the database to use"),
     ) -> ContentResponseSchema:
         log_info(f"Getting content by id: {content_id}")
         knowledge = get_knowledge_instance_by_db_id(knowledge_instances, db_id)
-        content = knowledge.get_content_by_id(content_id=content_id)
+        content = await knowledge.aget_content_by_id(content_id=content_id)
         if not content:
             raise HTTPException(status_code=404, detail=f"Content not found: {content_id}")
         response = ContentResponseSchema.from_dict(
@@ -411,12 +414,12 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
             500: {"description": "Failed to delete content", "model": InternalServerErrorResponse},
         },
     )
-    def delete_content_by_id(
+    async def delete_content_by_id(
         content_id: str,
         db_id: Optional[str] = Query(default=None, description="The ID of the database to use"),
     ) -> ContentResponseSchema:
         knowledge = get_knowledge_instance_by_db_id(knowledge_instances, db_id)
-        knowledge.remove_content_by_id(content_id=content_id)
+        await knowledge.aremove_content_by_id(content_id=content_id)
         log_info(f"Deleting content by id: {content_id}")
 
         return ContentResponseSchema(
@@ -443,7 +446,6 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
         knowledge = get_knowledge_instance_by_db_id(knowledge_instances, db_id)
         log_info("Deleting all content")
         knowledge.remove_all_content()
-
         return "success"
 
     @router.get(
@@ -476,13 +478,13 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
             404: {"description": "Content not found", "model": NotFoundResponse},
         },
     )
-    def get_content_status(
+    async def get_content_status(
         content_id: str,
         db_id: Optional[str] = Query(default=None, description="The ID of the database to use"),
     ) -> ContentStatusResponse:
         log_info(f"Getting content status: {content_id}")
         knowledge = get_knowledge_instance_by_db_id(knowledge_instances, db_id)
-        knowledge_status, status_message = knowledge.get_content_status(content_id=content_id)
+        knowledge_status, status_message = await knowledge.aget_content_status(content_id=content_id)
 
         # Handle the case where content is not found
         if knowledge_status is None:
@@ -513,6 +515,102 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
 
         return ContentStatusResponse(status=status, status_message=status_message or "")
 
+    @router.post(
+        "/knowledge/search",
+        status_code=200,
+        operation_id="search_vectors",
+        summary="Search Vectors",
+        description="Search the knowledge base for relevant vectors using query, filters and search type.",
+        response_model=PaginatedResponse[VectorSearchResult],
+        responses={
+            200: {
+                "description": "Search results retrieved successfully",
+                "content": {
+                    "application/json": {
+                        "example": {
+                            "data": [
+                                {
+                                    "id": "doc_123",
+                                    "content": "Jordan Mitchell - Software Engineer with skills in JavaScript, React, Python",
+                                    "name": "cv_1",
+                                    "meta_data": {"page": 1, "chunk": 1},
+                                    "usage": {"total_tokens": 14},
+                                    "reranking_score": 0.95,
+                                    "content_id": "content_456",
+                                }
+                            ],
+                            "meta": {"page": 1, "limit": 20, "total_pages": 2, "total_count": 35},
+                        }
+                    }
+                },
+            },
+            400: {"description": "Invalid search parameters"},
+            404: {"description": "No documents found"},
+        },
+    )
+    def search_vectors(request: VectorSearchRequestSchema) -> PaginatedResponse[VectorSearchResult]:
+        import time
+
+        start_time = time.time()
+
+        knowledge = get_knowledge_instance_by_db_id(knowledge_instances, request.db_id)
+
+        # For now, validate the vector db ids exist in the knowledge base
+        # We will add more logic around this once we have multi vectordb support
+        # If vector db ids are provided, check if any of them match the knowledge's vector db
+        if request.vector_db_ids:
+            if knowledge.vector_db and knowledge.vector_db.id:
+                if knowledge.vector_db.id not in request.vector_db_ids:
+                    raise HTTPException(
+                        status_code=400,
+                        detail=f"None of the provided Vector DB IDs {request.vector_db_ids} match the knowledge base Vector DB ID {knowledge.vector_db.id}",
+                    )
+            else:
+                raise HTTPException(status_code=400, detail="Knowledge base has no vector database configured")
+
+        # Calculate pagination parameters
+        meta = request.meta
+        limit = meta.limit if meta and meta.limit is not None else 20
+        page = meta.page if meta and meta.page is not None else 1
+
+        # Use max_results if specified, otherwise use a higher limit for search then paginate
+        search_limit = request.max_results
+
+        results = knowledge.search(
+            query=request.query, max_results=search_limit, filters=request.filters, search_type=request.search_type
+        )
+
+        # Calculate pagination
+        total_results = len(results)
+        start_idx = (page - 1) * limit
+
+        # Ensure start_idx doesn't exceed the total results
+        if start_idx >= total_results and total_results > 0:
+            # If page is beyond available results, return empty results
+            paginated_results = []
+        else:
+            end_idx = min(start_idx + limit, total_results)
+            paginated_results = results[start_idx:end_idx]
+
+        search_time_ms = (time.time() - start_time) * 1000
+
+        # Convert Document objects to serializable format
+        document_results = [VectorSearchResult.from_document(doc) for doc in paginated_results]
+
+        # Calculate pagination info
+        total_pages = (total_results + limit - 1) // limit  # Ceiling division
+
+        return PaginatedResponse(
+            data=document_results,
+            meta=PaginationInfo(
+                page=page,
+                limit=limit,
+                total_pages=total_pages,
+                total_count=total_results,
+                search_time_ms=search_time_ms,
+            ),
+        )
+
     @router.get(
         "/knowledge/config",
         status_code=200,
@@ -735,6 +833,14 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
                                     "description": "A chunking strategy that splits markdown based on structure like headers, paragraphs and sections",
                                 },
                             },
+                            "vector_dbs": [
+                                {
+                                    "id": "vector_db_1",
+                                    "name": "Vector DB 1",
+                                    "description": "Vector DB 1 description",
+                                    "search_types": ["vector", "keyword", "hybrid"],
+                                }
+                            ],
                             "filters": ["filter_tag_1", "filter_tag2"],
                         }
                     }
@@ -793,8 +899,23 @@ def attach_routes(router: APIRouter, knowledge_instances: List[Knowledge]) -> AP
                     key=chunker_key, name=chunker_info.get("name"), description=chunker_info.get("description")
                 )
 
+        vector_dbs = []
+        if knowledge.vector_db:
+            search_types = knowledge.vector_db.get_supported_search_types()
+            name = knowledge.vector_db.name
+            db_id = knowledge.vector_db.id
+            vector_dbs.append(
+                VectorDbSchema(
+                    id=db_id,
+                    name=name,
+                    description=knowledge.vector_db.description,
+                    search_types=search_types,
+                )
+            )
+
         return ConfigResponseSchema(
             readers=reader_schemas,
+            vector_dbs=vector_dbs,
             readersForType=types_of_readers,
             chunkers=chunkers_dict,
             filters=knowledge.get_filters(),

agno/os/routers/knowledge/schemas.py

@@ -111,8 +111,65 @@ class ChunkerSchema(BaseModel):
     description: Optional[str] = None
 
 
+class VectorDbSchema(BaseModel):
+    id: str
+    name: Optional[str] = None
+    description: Optional[str] = None
+    search_types: Optional[List[str]] = None
+
+
+class VectorSearchResult(BaseModel):
+    """Schema for search result documents."""
+
+    id: str
+    content: str
+    name: Optional[str] = None
+    meta_data: Optional[Dict[str, Any]] = None
+    usage: Optional[Dict[str, Any]] = None
+    reranking_score: Optional[float] = None
+    content_id: Optional[str] = None
+    content_origin: Optional[str] = None
+    size: Optional[int] = None
+
+    @classmethod
+    def from_document(cls, document) -> "VectorSearchResult":
+        """Convert a Document object to a serializable VectorSearchResult."""
+        return cls(
+            id=document.id,
+            content=document.content,
+            name=getattr(document, "name", None),
+            meta_data=getattr(document, "meta_data", None),
+            usage=getattr(document, "usage", None),
+            reranking_score=getattr(document, "reranking_score", None),
+            content_id=getattr(document, "content_id", None),
+            content_origin=getattr(document, "content_origin", None),
+            size=getattr(document, "size", None),
+        )
+
+
+class VectorSearchRequestSchema(BaseModel):
+    """Schema for vector search request."""
+
+    class Meta(BaseModel):
+        """Inline metadata schema for pagination."""
+
+        limit: Optional[int] = Field(20, description="Number of results per page", ge=1, le=100)
+        page: Optional[int] = Field(1, description="Page number", ge=1)
+
+    query: str = Field(..., description="The search query")
+    db_id: Optional[str] = Field(None, description="The content database id")
+    vector_db_ids: Optional[List[str]] = Field(None, description="List of vector database ids to search in")
+    search_type: Optional[str] = Field(None, description="The type of search to perform")
+    max_results: Optional[int] = Field(None, description="The maximum number of results to return")
+    filters: Optional[Dict[str, Any]] = Field(None, description="The filters to apply to the search")
+    meta: Optional[Meta] = Field(
+        None, description="Pagination metadata. Limit and page number to return a subset of results."
+    )
+
+
 class ConfigResponseSchema(BaseModel):
     readers: Optional[Dict[str, ReaderSchema]] = None
     readersForType: Optional[Dict[str, List[str]]] = None
     chunkers: Optional[Dict[str, ChunkerSchema]] = None
     filters: Optional[List[str]] = None
+    vector_dbs: Optional[List[VectorDbSchema]] = None

agno/os/routers/memory/memory.py

@@ -1,12 +1,12 @@
 import logging
 import math
-from typing import List, Optional
+from typing import List, Optional, Union, cast
 from uuid import uuid4
 
 from fastapi import Depends, HTTPException, Path, Query, Request
 from fastapi.routing import APIRouter
 
-from agno.db.base import BaseDb
+from agno.db.base import AsyncBaseDb, BaseDb
 from agno.db.schemas import UserMemory
 from agno.os.auth import get_authentication_dependency
 from agno.os.routers.memory.schemas import (
@@ -31,7 +31,9 @@ from agno.os.utils import get_db
 logger = logging.getLogger(__name__)
 
 
-def get_memory_router(dbs: dict[str, BaseDb], settings: AgnoAPISettings = AgnoAPISettings(), **kwargs) -> APIRouter:
+def get_memory_router(
+    dbs: dict[str, Union[BaseDb, AsyncBaseDb]], settings: AgnoAPISettings = AgnoAPISettings(), **kwargs
+) -> APIRouter:
     """Create memory router with comprehensive OpenAPI documentation for user memory management endpoints."""
     router = APIRouter(
         dependencies=[Depends(get_authentication_dependency(settings))],
@@ -47,7 +49,7 @@ def get_memory_router(dbs: dict[str, BaseDb], settings: AgnoAPISettings = AgnoAP
     return attach_routes(router=router, dbs=dbs)
 
 
-def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
+def attach_routes(router: APIRouter, dbs: dict[str, Union[BaseDb, AsyncBaseDb]]) -> APIRouter:
     @router.post(
         "/memories",
         response_model=UserMemorySchema,
@@ -92,15 +94,29 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
             raise HTTPException(status_code=400, detail="User ID is required")
 
         db = get_db(dbs, db_id)
-        user_memory = db.upsert_user_memory(
-            memory=UserMemory(
-                memory_id=str(uuid4()),
-                memory=payload.memory,
-                topics=payload.topics or [],
-                user_id=payload.user_id,
-            ),
-            deserialize=False,
-        )
+
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            user_memory = await db.upsert_user_memory(
+                memory=UserMemory(
+                    memory_id=str(uuid4()),
+                    memory=payload.memory,
+                    topics=payload.topics or [],
+                    user_id=payload.user_id,
+                ),
+                deserialize=False,
+            )
+        else:
+            user_memory = db.upsert_user_memory(
+                memory=UserMemory(
+                    memory_id=str(uuid4()),
+                    memory=payload.memory,
+                    topics=payload.topics or [],
+                    user_id=payload.user_id,
+                ),
+                deserialize=False,
+            )
+
         if not user_memory:
             raise HTTPException(status_code=500, detail="Failed to create memory")
 
@@ -124,7 +140,11 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         db_id: Optional[str] = Query(default=None, description="Database ID to use for deletion"),
     ) -> None:
         db = get_db(dbs, db_id)
-        db.delete_user_memory(memory_id=memory_id, user_id=user_id)
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            await db.delete_user_memory(memory_id=memory_id, user_id=user_id)
+        else:
+            db.delete_user_memory(memory_id=memory_id, user_id=user_id)
 
     @router.delete(
         "/memories",
@@ -146,7 +166,11 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         db_id: Optional[str] = Query(default=None, description="Database ID to use for deletion"),
     ) -> None:
         db = get_db(dbs, db_id)
-        db.delete_user_memories(memory_ids=request.memory_ids, user_id=request.user_id)
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            await db.delete_user_memories(memory_ids=request.memory_ids, user_id=request.user_id)
+        else:
+            db.delete_user_memories(memory_ids=request.memory_ids, user_id=request.user_id)
 
     @router.get(
         "/memories",
@@ -199,18 +223,34 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         if hasattr(request.state, "user_id"):
             user_id = request.state.user_id
 
-        user_memories, total_count = db.get_user_memories(
-            limit=limit,
-            page=page,
-            user_id=user_id,
-            agent_id=agent_id,
-            team_id=team_id,
-            topics=topics,
-            search_content=search_content,
-            sort_by=sort_by,
-            sort_order=sort_order,
-            deserialize=False,
-        )
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            user_memories, total_count = await db.get_user_memories(
+                limit=limit,
+                page=page,
+                user_id=user_id,
+                agent_id=agent_id,
+                team_id=team_id,
+                topics=topics,
+                search_content=search_content,
+                sort_by=sort_by,
+                sort_order=sort_order,
+                deserialize=False,
+            )
+        else:
+            user_memories, total_count = db.get_user_memories(  # type: ignore
+                limit=limit,
+                page=page,
+                user_id=user_id,
+                agent_id=agent_id,
+                team_id=team_id,
+                topics=topics,
+                search_content=search_content,
+                sort_by=sort_by,
+                sort_order=sort_order,
+                deserialize=False,
+            )
+
         return PaginatedResponse(
             data=[UserMemorySchema.from_dict(user_memory) for user_memory in user_memories],  # type: ignore
             meta=PaginationInfo(
@@ -249,12 +289,21 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         },
     )
     async def get_memory(
+        request: Request,
         memory_id: str = Path(description="Memory ID to retrieve"),
         user_id: Optional[str] = Query(default=None, description="User ID to query memory for"),
         db_id: Optional[str] = Query(default=None, description="Database ID to query memory from"),
     ) -> UserMemorySchema:
         db = get_db(dbs, db_id)
-        user_memory = db.get_user_memory(memory_id=memory_id, user_id=user_id, deserialize=False)
+
+        if hasattr(request.state, "user_id"):
+            user_id = request.state.user_id
+
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            user_memory = await db.get_user_memory(memory_id=memory_id, user_id=user_id, deserialize=False)
+        else:
+            user_memory = db.get_user_memory(memory_id=memory_id, user_id=user_id, deserialize=False)
         if not user_memory:
             raise HTTPException(status_code=404, detail=f"Memory with ID {memory_id} not found")
 
@@ -295,7 +344,11 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         db_id: Optional[str] = Query(default=None, description="Database ID to query topics from"),
     ) -> List[str]:
         db = get_db(dbs, db_id)
-        return db.get_all_memory_topics()
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            return await db.get_all_memory_topics()
+        else:
+            return db.get_all_memory_topics()
 
     @router.patch(
         "/memories/{memory_id}",
@@ -336,8 +389,6 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         memory_id: str = Path(description="Memory ID to update"),
         db_id: Optional[str] = Query(default=None, description="Database ID to use for update"),
     ) -> UserMemorySchema:
-        db = get_db(dbs, db_id)
-
         if hasattr(request.state, "user_id"):
             user_id = request.state.user_id
             payload.user_id = user_id
@@ -345,15 +396,29 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
         if payload.user_id is None:
             raise HTTPException(status_code=400, detail="User ID is required")
 
-        user_memory = db.upsert_user_memory(
-            memory=UserMemory(
-                memory_id=memory_id,
-                memory=payload.memory,
-                topics=payload.topics or [],
-                user_id=payload.user_id,
-            ),
-            deserialize=False,
-        )
+        db = get_db(dbs, db_id)
+
+        if isinstance(db, AsyncBaseDb):
+            db = cast(AsyncBaseDb, db)
+            user_memory = await db.upsert_user_memory(
+                memory=UserMemory(
+                    memory_id=memory_id,
+                    memory=payload.memory,
+                    topics=payload.topics or [],
+                    user_id=payload.user_id,
+                ),
+                deserialize=False,
+            )
+        else:
+            user_memory = db.upsert_user_memory(
+                memory=UserMemory(
+                    memory_id=memory_id,
+                    memory=payload.memory,
+                    topics=payload.topics or [],
+                    user_id=payload.user_id,
+                ),
+                deserialize=False,
+            )
         if not user_memory:
             raise HTTPException(status_code=500, detail="Failed to update memory")
 
@@ -396,10 +461,17 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
     ) -> PaginatedResponse[UserStatsSchema]:
         db = get_db(dbs, db_id)
         try:
-            user_stats, total_count = db.get_user_memory_stats(
-                limit=limit,
-                page=page,
-            )
+            if isinstance(db, AsyncBaseDb):
+                db = cast(AsyncBaseDb, db)
+                user_stats, total_count = await db.get_user_memory_stats(
+                    limit=limit,
+                    page=page,
+                )
+            else:
+                user_stats, total_count = db.get_user_memory_stats(
+                    limit=limit,
+                    page=page,
+                )
             return PaginatedResponse(
                 data=[UserStatsSchema.from_dict(stats) for stats in user_stats],
                 meta=PaginationInfo(

agno/os/routers/metrics/metrics.py

@@ -1,11 +1,11 @@
 import logging
 from datetime import date, datetime, timezone
-from typing import List, Optional
+from typing import List, Optional, Union, cast
 
 from fastapi import Depends, HTTPException, Query
 from fastapi.routing import APIRouter
 
-from agno.db.base import BaseDb
+from agno.db.base import AsyncBaseDb, BaseDb
 from agno.os.auth import get_authentication_dependency
 from agno.os.routers.metrics.schemas import DayAggregatedMetrics, MetricsResponse
 from agno.os.schema import (
@@ -21,7 +21,9 @@ from agno.os.utils import get_db
 logger = logging.getLogger(__name__)
 
 
-def get_metrics_router(dbs: dict[str, BaseDb], settings: AgnoAPISettings = AgnoAPISettings(), **kwargs) -> APIRouter:
+def get_metrics_router(
+    dbs: dict[str, Union[BaseDb, AsyncBaseDb]], settings: AgnoAPISettings = AgnoAPISettings(), **kwargs
+) -> APIRouter:
     """Create metrics router with comprehensive OpenAPI documentation for system metrics and analytics endpoints."""
     router = APIRouter(
         dependencies=[Depends(get_authentication_dependency(settings))],
@@ -37,7 +39,7 @@ def get_metrics_router(dbs: dict[str, BaseDb], settings: AgnoAPISettings = AgnoA
     return attach_routes(router=router, dbs=dbs)
 
 
-def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
+def attach_routes(router: APIRouter, dbs: dict[str, Union[BaseDb, AsyncBaseDb]]) -> APIRouter:
     @router.get(
         "/metrics",
         response_model=MetricsResponse,
@@ -100,7 +102,11 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
     ) -> MetricsResponse:
         try:
             db = get_db(dbs, db_id)
-            metrics, latest_updated_at = db.get_metrics(starting_date=starting_date, ending_date=ending_date)
+            if isinstance(db, AsyncBaseDb):
+                db = cast(AsyncBaseDb, db)
+                metrics, latest_updated_at = await db.get_metrics(starting_date=starting_date, ending_date=ending_date)
+            else:
+                metrics, latest_updated_at = db.get_metrics(starting_date=starting_date, ending_date=ending_date)
 
             return MetricsResponse(
                 metrics=[DayAggregatedMetrics.from_dict(metric) for metric in metrics],
@@ -166,7 +172,11 @@ def attach_routes(router: APIRouter, dbs: dict[str, BaseDb]) -> APIRouter:
     ) -> List[DayAggregatedMetrics]:
         try:
             db = get_db(dbs, db_id)
-            result = db.calculate_metrics()
+            if isinstance(db, AsyncBaseDb):
+                db = cast(AsyncBaseDb, db)
+                result = await db.calculate_metrics()
+            else:
+                result = db.calculate_metrics()
             if result is None:
                 return []