agno 2.1.2__py3-none-any.whl → 2.3.13__py3-none-any.whl

agno/vectordb/singlestore/singlestore.py

@@ -1,7 +1,7 @@
 import asyncio
 import json
 from hashlib import md5
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Union
 
 try:
     from sqlalchemy.dialects import mysql
@@ -14,10 +14,11 @@ try:
 except ImportError:
     raise ImportError("`sqlalchemy` not installed")
 
+from agno.filters import FilterExpr
 from agno.knowledge.document import Document
 from agno.knowledge.embedder import Embedder
 from agno.knowledge.reranker.base import Reranker
-from agno.utils.log import log_debug, log_error, log_info
+from agno.utils.log import log_debug, log_error, log_info, log_warning
 from agno.vectordb.base import VectorDb
 from agno.vectordb.distance import Distance
 
@@ -32,6 +33,8 @@ class SingleStore(VectorDb):
         embedder: Optional[Embedder] = None,
         distance: Distance = Distance.cosine,
         reranker: Optional[Reranker] = None,
+        name: Optional[str] = None,
+        description: Optional[str] = None,
         # index: Optional[Union[Ivfflat, HNSW]] = HNSW(),
     ):
         _engine: Optional[Engine] = db_engine
@@ -44,9 +47,11 @@ class SingleStore(VectorDb):
         self.collection: str = collection
         self.schema: Optional[str] = schema
         self.db_url: Optional[str] = db_url
+        # Initialize base class with name and description
+        super().__init__(name=name, description=description)
+
         self.db_engine: Engine = _engine
         self.metadata: MetaData = MetaData(schema=self.schema)
-
         if embedder is None:
             from agno.knowledge.embedder.openai import OpenAIEmbedder
 
@@ -180,8 +185,10 @@ class SingleStore(VectorDb):
             for document in documents:
                 document.embed(embedder=self.embedder)
                 cleaned_content = document.content.replace("\x00", "\ufffd")
-                record_id = md5(cleaned_content.encode()).hexdigest()
-                _id = document.id or record_id
+                # Include content_hash in ID to ensure uniqueness across different content hashes
+                base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+                record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
+                _id = record_id
 
                 meta_data_json = json.dumps(document.meta_data)
                 usage_json = json.dumps(document.usage)
@@ -241,8 +248,10 @@ class SingleStore(VectorDb):
             for document in documents:
                 document.embed(embedder=self.embedder)
                 cleaned_content = document.content.replace("\x00", "\ufffd")
-                record_id = md5(cleaned_content.encode()).hexdigest()
-                _id = document.id or record_id
+                # Include content_hash in ID to ensure uniqueness across different content hashes
+                base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+                record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
+                _id = record_id
 
                 meta_data_json = json.dumps(document.meta_data)
                 usage_json = json.dumps(document.usage)
@@ -278,7 +287,9 @@ class SingleStore(VectorDb):
             sess.commit()
             log_debug(f"Committed {counter} documents")
 
-    def search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         """
         Search for documents based on a query and optional filters.
 
@@ -290,6 +301,8 @@ class SingleStore(VectorDb):
         Returns:
             List[Document]: List of documents that match the query.
         """
+        if filters is not None:
+            log_warning("Filters are not supported in SingleStore. No filters will be applied.")
         query_embedding = self.embedder.get_embedding(query)
         if query_embedding is None:
             log_error(f"Error getting embedding for Query: {query}")
@@ -428,9 +441,9 @@ class SingleStore(VectorDb):
         try:
             with self.Session.begin() as sess:
                 stmt = delete(self.table).where(self.table.c.id == id)
-                result = sess.execute(stmt)
-                log_info(f"Deleted {result.rowcount} records with ID {id} from table '{self.table.name}'.")
-                return result.rowcount > 0
+                result = sess.execute(stmt)  # type: ignore
+                log_info(f"Deleted {result.rowcount} records with ID {id} from table '{self.table.name}'.")  # type: ignore
+                return result.rowcount > 0  # type: ignore
         except Exception as e:
             log_error(f"Error deleting document with ID {id}: {e}")
             return False
@@ -444,11 +457,11 @@ class SingleStore(VectorDb):
         try:
             with self.Session.begin() as sess:
                 stmt = delete(self.table).where(self.table.c.content_id == content_id)
-                result = sess.execute(stmt)
+                result = sess.execute(stmt)  # type: ignore
                 log_info(
-                    f"Deleted {result.rowcount} records with content_id {content_id} from table '{self.table.name}'."
+                    f"Deleted {result.rowcount} records with content_id {content_id} from table '{self.table.name}'."  # type: ignore
                 )
-                return result.rowcount > 0
+                return result.rowcount > 0  # type: ignore
         except Exception as e:
             log_error(f"Error deleting document with content_id {content_id}: {e}")
             return False
@@ -462,9 +475,9 @@ class SingleStore(VectorDb):
         try:
             with self.Session.begin() as sess:
                 stmt = delete(self.table).where(self.table.c.name == name)
-                result = sess.execute(stmt)
-                log_info(f"Deleted {result.rowcount} records with name '{name}' from table '{self.table.name}'.")
-                return result.rowcount > 0
+                result = sess.execute(stmt)  # type: ignore
+                log_info(f"Deleted {result.rowcount} records with name '{name}' from table '{self.table.name}'.")  # type: ignore
+                return result.rowcount > 0  # type: ignore
         except Exception as e:
             log_error(f"Error deleting document with name {name}: {e}")
             return False
@@ -480,9 +493,9 @@ class SingleStore(VectorDb):
                 # Convert metadata to JSON string for comparison
                 metadata_json = json.dumps(metadata, sort_keys=True)
                 stmt = delete(self.table).where(self.table.c.meta_data == metadata_json)
-                result = sess.execute(stmt)
-                log_info(f"Deleted {result.rowcount} records with metadata {metadata} from table '{self.table.name}'.")
-                return result.rowcount > 0
+                result = sess.execute(stmt)  # type: ignore
+                log_info(f"Deleted {result.rowcount} records with metadata {metadata} from table '{self.table.name}'.")  # type: ignore
+                return result.rowcount > 0  # type: ignore
         except Exception as e:
             log_error(f"Error deleting documents with metadata {metadata}: {e}")
             return False
@@ -539,8 +552,10 @@ class SingleStore(VectorDb):
             counter = 0
             for document in documents:
                 cleaned_content = document.content.replace("\x00", "\ufffd")
-                record_id = md5(cleaned_content.encode()).hexdigest()
-                _id = document.id or record_id
+                # Include content_hash in ID to ensure uniqueness across different content hashes
+                base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+                record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
+                _id = record_id
 
                 meta_data_json = json.dumps(document.meta_data)
                 usage_json = json.dumps(document.usage)
@@ -623,8 +638,10 @@ class SingleStore(VectorDb):
             counter = 0
             for document in documents:
                 cleaned_content = document.content.replace("\x00", "\ufffd")
-                record_id = md5(cleaned_content.encode()).hexdigest()
-                _id = document.id or record_id
+                # Include content_hash in ID to ensure uniqueness across different content hashes
+                base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+                record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
+                _id = record_id
 
                 meta_data_json = json.dumps(document.meta_data)
                 usage_json = json.dumps(document.usage)
@@ -661,7 +678,7 @@ class SingleStore(VectorDb):
             log_debug(f"Committed {counter} documents")
 
     async def async_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         return self.search(query=query, limit=limit, filters=filters)
 
@@ -689,11 +706,11 @@ class SingleStore(VectorDb):
         try:
             with self.Session.begin() as sess:
                 stmt = delete(self.table).where(self.table.c.content_hash == content_hash)
-                result = sess.execute(stmt)
+                result = sess.execute(stmt)  # type: ignore
                 log_info(
-                    f"Deleted {result.rowcount} records with content_hash '{content_hash}' from table '{self.table.name}'."
+                    f"Deleted {result.rowcount} records with content_hash '{content_hash}' from table '{self.table.name}'."  # type: ignore
                 )
-                return result.rowcount > 0
+                return result.rowcount > 0  # type: ignore
         except Exception as e:
             log_error(f"Error deleting documents with content_hash {content_hash}: {e}")
             return False
@@ -712,7 +729,7 @@ class SingleStore(VectorDb):
             with self.Session.begin() as sess:
                 # Find documents with the given content_id
                 stmt = select(self.table).where(self.table.c.content_id == content_id)
-                result = sess.execute(stmt)
+                result = sess.execute(stmt)  # type: ignore
 
                 updated_count = 0
                 for row in result:
@@ -748,3 +765,7 @@ class SingleStore(VectorDb):
         except Exception as e:
             log_error(f"Error updating metadata for content_id '{content_id}': {e}")
             raise
+
+    def get_supported_search_types(self) -> List[str]:
+        """Get the supported search types for this vector database."""
+        return []  # SingleStore doesn't use SearchType enum

agno/vectordb/surrealdb/surrealdb.py

@@ -11,9 +11,10 @@ except ImportError as e:
     msg = "The `surrealdb` package is not installed. Please install it via `pip install surrealdb`."
     raise ImportError(msg) from e
 
+from agno.filters import FilterExpr
 from agno.knowledge.document import Document
 from agno.knowledge.embedder import Embedder
-from agno.utils.log import log_debug, log_error, log_info
+from agno.utils.log import log_debug, log_error, log_info, log_warning
 from agno.vectordb.base import VectorDb
 from agno.vectordb.distance import Distance
 
@@ -30,12 +31,6 @@ class SurrealDb(VectorDb):
         DEFINE INDEX IF NOT EXISTS vector_idx ON {collection} FIELDS embedding HNSW DIMENSION {dimensions} DIST {distance};
     """
 
-    DOC_EXISTS_QUERY: Final[str] = """
-        SELECT * FROM {collection}
-        WHERE content = $content
-        LIMIT 1
-    """
-
     NAME_EXISTS_QUERY: Final[str] = """
         SELECT * FROM {collection}
         WHERE meta_data.name = $name
@@ -107,11 +102,13 @@ class SurrealDb(VectorDb):
         m: int = 12,
         search_ef: int = 40,
         embedder: Optional[Embedder] = None,
+        name: Optional[str] = None,
+        description: Optional[str] = None,
+        id: Optional[str] = None,
     ):
         """Initialize SurrealDB connection.
 
         Args:
-            url: SurrealDB server URL (e.g. ws://localhost:8000/rpc)
             client: A blocking connection, either HTTP or WS
             async_client: An async connection, either HTTP or WS (default: None)
             collection: Collection name to store documents (default: documents)
@@ -122,6 +119,17 @@ class SurrealDb(VectorDb):
             embedder: Embedder instance for creating embeddings (default: OpenAIEmbedder)
 
         """
+        # Dynamic ID generation based on unique identifiers
+        if id is None:
+            from agno.utils.string import generate_id
+
+            client_info = str(client) if client else str(async_client) if async_client else "default"
+            seed = f"{client_info}#{collection}"
+            id = generate_id(seed)
+
+        # Initialize base class with name, description, and generated ID
+        super().__init__(id=id, name=name, description=description)
+
         # Embedder for embedding the document contents
         if embedder is None:
             from agno.knowledge.embedder.openai import OpenAIEmbedder
@@ -131,7 +139,6 @@ class SurrealDb(VectorDb):
         self.embedder: Embedder = embedder
         self.dimensions = self.embedder.dimensions
         self.collection = collection
-
         # Convert Distance enum to SurrealDB distance type
         self.distance = {Distance.cosine: "COSINE", Distance.l2: "EUCLIDEAN", Distance.max_inner_product: "DOT"}[
             distance
@@ -208,23 +215,6 @@ class SurrealDb(VectorDb):
             )
             self.client.query(query)
 
-    def doc_exists(self, document: Document) -> bool:
-        """Check if a document exists by its content.
-
-        Args:
-            document: The document to check.
-
-        Returns:
-            True if the document exists, False otherwise.
-
-        """
-        log_debug(f"Checking if document exists: {document.content}")
-        result = self.client.query(
-            self.DOC_EXISTS_QUERY.format(collection=self.collection),
-            {"content": document.content},
-        )
-        return bool(self._extract_result(result))
-
     def name_exists(self, name: str) -> bool:
         """Check if a document exists by its name.
 
@@ -306,7 +296,9 @@ class SurrealDb(VectorDb):
             thing = f"{self.collection}:{doc.id}" if doc.id else self.collection
             self.client.query(self.UPSERT_QUERY.format(thing=thing), data)
 
-    def search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         """Search for similar documents.
 
         Args:
@@ -318,6 +310,9 @@ class SurrealDb(VectorDb):
             A list of documents that are similar to the query.
 
         """
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in SurrealDB. No filters will be applied.")
+            filters = None
         query_embedding = self.embedder.get_embedding(query)
         if query_embedding is None:
             log_error(f"Error getting embedding for Query: {query}")
@@ -475,19 +470,6 @@ class SurrealDb(VectorDb):
             ),
         )
 
-    async def async_doc_exists(self, document: Document) -> bool:
-        """Check if a document exists by its content asynchronously.
-
-        Returns:
-            True if the document exists, False otherwise.
-
-        """
-        response = await self.async_client.query(
-            self.DOC_EXISTS_QUERY.format(collection=self.collection),
-            {"content": document.content},
-        )
-        return bool(self._extract_result(response))
-
     async def async_name_exists(self, name: str) -> bool:
         """Check if a document exists by its name asynchronously.
 
@@ -548,7 +530,7 @@ class SurrealDb(VectorDb):
         self,
         query: str,
         limit: int = 5,
-        filters: Optional[Dict[str, Any]] = None,
+        filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None,
     ) -> List[Document]:
         """Search for similar documents asynchronously.
 
@@ -561,6 +543,10 @@ class SurrealDb(VectorDb):
             A list of documents that are similar to the query.
 
         """
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in SurrealDB. No filters will be applied.")
+            filters = None
+
         query_embedding = self.embedder.get_embedding(query)
         if query_embedding is None:
             log_error(f"Error getting embedding for Query: {query}")
@@ -671,3 +657,7 @@ class SurrealDb(VectorDb):
         except Exception as e:
             log_error(f"Error updating metadata for content_id '{content_id}': {e}")
             raise
+
+    def get_supported_search_types(self) -> List[str]:
+        """Get the supported search types for this vector database."""
+        return []  # SurrealDb doesn't use SearchType enum

agno/vectordb/upstashdb/upstashdb.py

@@ -1,5 +1,5 @@
 import asyncio
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Union
 
 try:
     from upstash_vector import Index, Vector
@@ -9,10 +9,11 @@ except ImportError:
         "The `upstash-vector` package is not installed, please install using `pip install upstash-vector`"
     )
 
+from agno.filters import FilterExpr
 from agno.knowledge.document import Document
 from agno.knowledge.embedder import Embedder
 from agno.knowledge.reranker.base import Reranker
-from agno.utils.log import log_info, logger
+from agno.utils.log import log_info, log_warning, logger
 from agno.vectordb.base import VectorDb
 
 DEFAULT_NAMESPACE = ""
@@ -32,6 +33,8 @@ class UpstashVectorDb(VectorDb):
         embedder (Optional[Embedder], optional): The embedder to use. If None, uses Upstash hosted embedding models.
         namespace (Optional[str], optional): The namespace to use. Defaults to DEFAULT_NAMESPACE.
         reranker (Optional[Reranker], optional): The reranker to use. Defaults to None.
+        name (Optional[str], optional): The name of the vector database. Defaults to None.
+        description (Optional[str], optional): The description of the vector database. Defaults to None.
         **kwargs: Additional keyword arguments.
     """
 
@@ -45,8 +48,28 @@ class UpstashVectorDb(VectorDb):
         embedder: Optional[Embedder] = None,
         namespace: Optional[str] = DEFAULT_NAMESPACE,
         reranker: Optional[Reranker] = None,
+        name: Optional[str] = None,
+        description: Optional[str] = None,
+        id: Optional[str] = None,
         **kwargs: Any,
     ) -> None:
+        # Validate required parameters
+        if not url:
+            raise ValueError("URL must be provided.")
+        if not token:
+            raise ValueError("Token must be provided.")
+
+        # Dynamic ID generation based on unique identifiers
+        if id is None:
+            from agno.utils.string import generate_id
+
+            namespace_identifier = namespace or DEFAULT_NAMESPACE
+            seed = f"{url}#{namespace_identifier}"
+            id = generate_id(seed)
+
+        # Initialize base class with name, description, and generated ID
+        super().__init__(id=id, name=name, description=description)
+
         self._index: Optional[Index] = None
         self.url: str = url
         self.token: str = token
@@ -56,7 +79,6 @@ class UpstashVectorDb(VectorDb):
         self.namespace: str = namespace if namespace is not None else DEFAULT_NAMESPACE
         self.kwargs: Dict[str, Any] = kwargs
         self.use_upstash_embeddings: bool = embedder is None
-
         if embedder is None:
             logger.warning(
                 "You have not provided an embedder, using Upstash hosted embedding models. "
@@ -303,7 +325,7 @@ class UpstashVectorDb(VectorDb):
         self,
         query: str,
         limit: int = 5,
-        filters: Optional[Dict[str, Any]] = None,
+        filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None,
         namespace: Optional[str] = None,
     ) -> List[Document]:
         """Search for documents in the index.
@@ -316,7 +338,9 @@ class UpstashVectorDb(VectorDb):
             List[Document]: List of matching documents.
         """
         _namespace = self.namespace if namespace is None else namespace
-
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in UpstashDB. No filters will be applied.")
+            filters = None
         filter_str = "" if filters is None else str(filters)
 
         if not self.use_upstash_embeddings and self.embedder is not None:
@@ -602,7 +626,7 @@ class UpstashVectorDb(VectorDb):
         self.index.upsert(vectors, namespace=_namespace)
 
     async def async_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         raise NotImplementedError(f"Async not supported on {self.__class__.__name__}.")
 
@@ -688,3 +712,7 @@ class UpstashVectorDb(VectorDb):
         except Exception as e:
             logger.error(f"Error updating metadata for content_id '{content_id}': {e}")
             raise
+
+    def get_supported_search_types(self) -> List[str]:
+        """Get the supported search types for this vector database."""
+        return []  # UpstashVectorDb doesn't use SearchType enum

agno/vectordb/weaviate/weaviate.py

@@ -3,7 +3,7 @@ import json
 import uuid
 from hashlib import md5
 from os import getenv
-from typing import Any, Dict, List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Tuple, Union
 
 try:
     from warnings import filterwarnings
@@ -18,10 +18,11 @@ try:
 except ImportError:
     raise ImportError("Weaviate is not installed. Install using 'pip install weaviate-client'.")
 
+from agno.filters import FilterExpr
 from agno.knowledge.document import Document
 from agno.knowledge.embedder import Embedder
 from agno.knowledge.reranker.base import Reranker
-from agno.utils.log import log_debug, log_info, logger
+from agno.utils.log import log_debug, log_info, log_warning, logger
 from agno.vectordb.base import VectorDb
 from agno.vectordb.search import SearchType
 from agno.vectordb.weaviate.index import Distance, VectorIndex
@@ -41,6 +42,9 @@ class Weaviate(VectorDb):
         local: bool = False,
         # Collection params
         collection: str = "default",
+        name: Optional[str] = None,
+        description: Optional[str] = None,
+        id: Optional[str] = None,
         vector_index: VectorIndex = VectorIndex.HNSW,
         distance: Distance = Distance.COSINE,
         # Search/Embedding params
@@ -49,6 +53,17 @@ class Weaviate(VectorDb):
         reranker: Optional[Reranker] = None,
         hybrid_search_alpha: float = 0.5,
     ):
+        # Dynamic ID generation based on unique identifiers
+        if id is None:
+            from agno.utils.string import generate_id
+
+            connection_identifier = wcd_url or "local" if local else "default"
+            seed = f"{connection_identifier}#{collection}"
+            id = generate_id(seed)
+
+        # Initialize base class with name, description, and generated ID
+        super().__init__(id=id, name=name, description=description)
+
         # Connection setup
         self.wcd_url = wcd_url or getenv("WCD_URL")
         self.wcd_api_key = wcd_api_key or getenv("WCD_API_KEY")
@@ -232,7 +247,9 @@ class Weaviate(VectorDb):
                 continue
 
             cleaned_content = document.content.replace("\x00", "\ufffd")
-            record_id = md5(cleaned_content.encode()).hexdigest()
+            # Include content_hash in ID to ensure uniqueness across different content hashes
+            base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+            record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
             doc_uuid = uuid.UUID(hex=record_id[:32])
 
             # Merge filters with metadata
@@ -323,7 +340,9 @@ class Weaviate(VectorDb):
 
                     # Clean content and generate UUID
                     cleaned_content = document.content.replace("\x00", "\ufffd")
-                    record_id = md5(cleaned_content.encode()).hexdigest()
+                    # Include content_hash in ID to ensure uniqueness across different content hashes
+                    base_id = document.id or md5(cleaned_content.encode()).hexdigest()
+                    record_id = md5(f"{base_id}_{content_hash}".encode()).hexdigest()
                     doc_uuid = uuid.UUID(hex=record_id[:32])
 
                     # Serialize meta_data to JSON string
@@ -378,7 +397,9 @@ class Weaviate(VectorDb):
         await self.async_insert(content_hash=content_hash, documents=documents, filters=filters)
         return
 
-    def search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         """
         Perform a search based on the configured search type.
 
@@ -390,6 +411,9 @@ class Weaviate(VectorDb):
         Returns:
             List[Document]: List of matching documents.
         """
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in Weaviate. No filters will be applied.")
+            filters = None
         if self.search_type == SearchType.vector:
             return self.vector_search(query, limit, filters)
         elif self.search_type == SearchType.keyword:
@@ -401,7 +425,7 @@ class Weaviate(VectorDb):
             return []
 
     async def async_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         """
         Perform a search based on the configured search type asynchronously.
@@ -414,6 +438,9 @@ class Weaviate(VectorDb):
         Returns:
             List[Document]: List of matching documents.
         """
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in Weaviate. No filters will be applied.")
+            filters = None
         if self.search_type == SearchType.vector:
             return await self.async_vector_search(query, limit, filters)
         elif self.search_type == SearchType.keyword:
@@ -424,7 +451,9 @@ class Weaviate(VectorDb):
             logger.error(f"Invalid search type '{self.search_type}'.")
             return []
 
-    def vector_search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def vector_search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         try:
             query_embedding = self.embedder.get_embedding(query)
             if query_embedding is None:
@@ -459,7 +488,7 @@ class Weaviate(VectorDb):
             self.get_client().close()
 
     async def async_vector_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         """
         Perform a vector search in Weaviate asynchronously.
@@ -504,7 +533,9 @@ class Weaviate(VectorDb):
             logger.error(f"Error searching for documents: {e}")
             return []
 
-    def keyword_search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def keyword_search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         try:
             collection = self.get_client().collections.get(self.collection)
             filter_expr = self._build_filter_expression(filters)
@@ -535,7 +566,7 @@ class Weaviate(VectorDb):
             self.get_client().close()
 
     async def async_keyword_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         """
         Perform a keyword search in Weaviate asynchronously.
@@ -576,7 +607,9 @@ class Weaviate(VectorDb):
             logger.error(f"Error searching for documents: {e}")
             return []
 
-    def hybrid_search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:
+    def hybrid_search(
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
+    ) -> List[Document]:
         try:
             query_embedding = self.embedder.get_embedding(query)
             if query_embedding is None:
@@ -614,7 +647,7 @@ class Weaviate(VectorDb):
             self.get_client().close()
 
     async def async_hybrid_search(
-        self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None
+        self, query: str, limit: int = 5, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None
     ) -> List[Document]:
         """
         Perform a hybrid search combining vector and keyword search in Weaviate asynchronously.
@@ -835,7 +868,7 @@ class Weaviate(VectorDb):
         """Indicate that upsert functionality is available."""
         return True
 
-    def _build_filter_expression(self, filters: Optional[Dict[str, Any]]):
+    def _build_filter_expression(self, filters: Optional[Union[Dict[str, Any], List[FilterExpr]]]):
         """
         Build a filter expression for Weaviate queries.
 
@@ -847,7 +880,9 @@ class Weaviate(VectorDb):
         """
         if not filters:
             return None
-
+        if isinstance(filters, List):
+            log_warning("Filters Expressions are not supported in Weaviate. No filters will be applied.")
+            return None
         try:
             # Create a filter for each key-value pair
             filter_conditions = []
@@ -968,3 +1003,7 @@ class Weaviate(VectorDb):
         except Exception as e:
             logger.error(f"Error deleting documents by content_hash '{content_hash}': {e}")
             return False
+
+    def get_supported_search_types(self) -> List[str]:
+        """Get the supported search types for this vector database."""
+        return [SearchType.vector, SearchType.keyword, SearchType.hybrid]

agno/workflow/__init__.py

@@ -1,3 +1,4 @@
+from agno.workflow.agent import WorkflowAgent
 from agno.workflow.condition import Condition
 from agno.workflow.loop import Loop
 from agno.workflow.parallel import Parallel
@@ -9,6 +10,7 @@ from agno.workflow.workflow import Workflow
 
 __all__ = [
     "Workflow",
+    "WorkflowAgent",
     "Steps",
     "Step",
     "Loop",