v2.3.16

README.md

@@ -23,23 +23,18 @@
 
 Phidata is a toolkit for building autonomous AI applications using function calling.
 
-Function calling is a powerful approach that lets LLMs take actions and dynamically choose their path based on the response,
-just like how humans solve problems. For example:
-- To build a RAG chatbot, instead of stuffing the prompt every time, give the assistant functions to search its knowledge base for relevant context, which results in better responses.
-- To build text-to-SQL, give the assistant functions to view and describe tables, search its knowledge base for context and finally, inspect and run queries to get the answer.
-- To build a customer support assistant, give the assistant functions to search the order history, product description or its knowledge base depending on the question.
-
-Phidata provides `Assistants` that come with memory, knowledge, storage and tools, making it easy to build intelligent AI applications.
+Function calling enables LLMs to achieve tasks by taking an action (function call) and intelligently choosing its next step based on the response,
+just like how humans solve problems. Phidata provides `Assistants` with built-in memory, knowledge, storage and tools, making it easy to build intelligent AI applications.
 
 ![assistants-explanation](https://github.com/phidatahq/phidata/assets/22579644/7f420011-ab8c-410a-97cc-5ad2fc0fe9d8)
 
-### Use phidata to build
+## ⭐️ Use phidata to build
 
 - **Knowledge Assistants:** Answer questions from documents (PDFs, text)
+- **Research Assistants:** Perform research and summarize findings.
 - **Data Assistants:** Analyze data by running SQL queries.
 - **Python Assistants:** Perform tasks by running python code.
 - **Customer Assistants:** Answer customer queries using product descriptions and purchase history.
-- **Research Assistants:** Perform research and summarize findings.
 - **Marketing Assistants:** Provide marketing insights, copywriting and content ideas.
 - **Travel Assistants:** Help plan travel by researching destinations, flight and hotel prices.
 - **Meal Prep Assistants:** Help plan meals by researching recipes and adding ingredients to shopping lists.

phi/knowledge/base.py

@@ -31,15 +31,18 @@ def document_lists(self) -> Iterator[List[Document]]:
 
     def search(self, query: str, num_documents: Optional[int] = None) -> List[Document]:
         """Returns relevant documents matching the query"""
-
-        if self.vector_db is None:
-            logger.warning("No vector db provided")
+        try:
+            if self.vector_db is None:
+                logger.warning("No vector db provided")
+                return []
+
+            _num_documents = num_documents or self.num_documents
+            logger.debug(f"Getting {_num_documents} relevant documents for query: {query}")
+            return self.vector_db.search(query=query, limit=_num_documents)
+        except Exception as e:
+            logger.error(f"Error searching for documents: {e}")
             return []
 
-        _num_documents = num_documents or self.num_documents
-        logger.debug(f"Getting {_num_documents} relevant documents for query: {query}")
-        return self.vector_db.search(query=query, limit=_num_documents)
-
     def load(self, recreate: bool = False) -> None:
         """Load the knowledge base to the vector db"""
 
@@ -219,3 +222,11 @@ def exists(self) -> bool:
             logger.warning("No vector db provided")
             return False
         return self.vector_db.exists()
+
+    def clear(self) -> bool:
+        """Clear the knowledge base"""
+        if self.vector_db is None:
+            logger.warning("No vector db available")
+            return True
+
+        return self.vector_db.clear()

phi/vectordb/base.py

@@ -45,3 +45,7 @@ def exists(self) -> bool:
     @abstractmethod
     def optimize(self) -> None:
         raise NotImplementedError
+
+    @abstractmethod
+    def clear(self) -> bool:
+        raise NotImplementedError

phi/vectordb/pgvector/pgvector.py

@@ -322,5 +322,13 @@ def optimize(self) -> None:
                             f"WITH (m = {self.index.m}, ef_construction = {self.index.ef_construction});"
                         )
                     )
-
         logger.debug("==== Optimized Vector DB ====")
+
+    def clear(self) -> bool:
+        from sqlalchemy import delete
+
+        with self.Session() as sess:
+            with sess.begin():
+                stmt = delete(self.table)
+                sess.execute(stmt)
+                return True

phi/vectordb/pgvector/pgvector2.py

@@ -367,5 +367,13 @@ def optimize(self) -> None:
                             f"WITH (m = {self.index.m}, ef_construction = {self.index.ef_construction});"
                         )
                     )
-
         logger.debug("==== Optimized Vector DB ====")
+
+    def clear(self) -> bool:
+        from sqlalchemy import delete
+
+        with self.Session() as sess:
+            with sess.begin():
+                stmt = delete(self.table)
+                sess.execute(stmt)
+                return True

phi/vectordb/qdrant/qdrant.py

@@ -208,3 +208,6 @@ def get_count(self) -> int:
 
     def optimize(self) -> None:
         pass
+
+    def clear(self) -> bool:
+        return False

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "phidata"
-version = "2.3.15"
+version = "2.3.16"
 description = "Build AI assistants using function calling"
 requires-python = ">=3.8"
 readme = "README.md"