fix: user-agent for scrape

CHANGELOG.md

@@ -5,6 +5,11 @@
     *Andrea Sponziello* 
 ### **Copyrigth**: *Tiledesk SRL*
 
+## [2024-07-09]
+### 0.2.6
+- add: DELETE /api/chunk/<chunk_id>/namespace/<namespace>
+- add: search_type parameter similarity|mmr 
+
 ## [2024-07-01]
 ### 0.2.5
 - fix: user-agent for scrape

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "tilellm"
-version = "0.2.5"
+version = "0.2.6"
 description = "tiledesk for RAG"
 authors = ["Gianluca Lorenzo <[email protected]>"]
 repository = "https://github.com/Tiledesk/tiledesk-llm"

tilellm/__main__.py

@@ -30,11 +30,13 @@
                                            add_pc_item,
                                            delete_namespace,
                                            delete_id_from_namespace,
+                                           delete_chunk_id_from_namespace,
                                            get_ids_namespace,
                                            get_listitems_namespace,
                                            get_desc_namespace,
                                            get_list_namespace,
-                                           get_sources_namespace, ask_to_llm)
+                                           get_sources_namespace,
+                                           ask_to_llm)
 
 import logging
 
@@ -475,6 +477,26 @@ async def delete_namespace_main(namespace: str):
         raise HTTPException(status_code=400, detail=repr(ex))
 
 
+@app.delete("/api/chunk/{chunk_id}/namespace/{namespace}")
+async def delete_item_chunk_id_namespace_main(chunk_id: str, namespace: str):
+    """
+    Delete items from namespace identified by id and namespace
+    :param chunk_id:
+    :param namespace:
+    :return:
+    """
+    try:
+
+        logger.info(f"delete id {chunk_id} dal namespace {namespace}")
+        result = await delete_chunk_id_from_namespace(chunk_id, namespace)
+
+        return JSONResponse(content={"message": f"ids {chunk_id} in Namespace {namespace} deleted"})
+    except Exception as ex:
+        print(repr(ex))
+        logger.error(ex)
+        raise HTTPException(status_code=400, detail=repr(ex))
+
+
 @app.post("/api/delete/namespace")
 async def delete_namespace_main(namespace_to_delete: PineconeNamespaceToDelete):
     """

tilellm/controller/controller.py

@@ -240,7 +240,7 @@ async def ask_with_memory(question_answer, repo=None) -> RetrievalResult:
 
         vector_store = await repo.create_pc_index(oai_embeddings, emb_dimension)
 
-        retriever = vector_store.as_retriever(search_type='similarity',
+        retriever = vector_store.as_retriever(search_type=question_answer.search_type,
                                               search_kwargs={'k': question_answer.top_k,
                                                              'namespace': question_answer.namespace}
                                               )
@@ -596,6 +596,20 @@ async def delete_id_from_namespace(metadata_id: str, namespace: str, repo=None):
         logger.error(ex)
         raise ex
 
+@inject_repo
+async def delete_chunk_id_from_namespace(chunk_id:str, namespace: str, repo=None):
+    """
+    Delete chunk by id from namespace
+    :param chunk_id:
+    :param namespace:
+    :param repo:
+    :return:
+    """
+    try:
+        return await repo.delete_pc_chunk_id_namespace(chunk_id=chunk_id, namespace=namespace)
+    except Exception as ex:
+        logger.error(ex)
+        raise ex
 
 @inject_repo
 async def get_list_namespace(repo=None) -> PineconeNamespaceResult:

tilellm/models/item_model.py

@@ -59,6 +59,7 @@ class QuestionAnswer(BaseModel):
     embedding: str = Field(default_factory=lambda: "text-embedding-ada-002")
     debug: bool = Field(default_factory=lambda: False)
     system_context: Optional[str] = None
+    search_type: str = Field(default_factory=lambda: "similarity")
     chat_history_dict: Optional[Dict[str, ChatEntry]] = None
 
     @field_validator("temperature")

tilellm/store/pinecone/pinecone_repository_base.py

@@ -50,6 +50,34 @@ async def delete_pc_namespace(namespace: str):
     async def delete_pc_ids_namespace(self, metadata_id: str, namespace: str):
         pass
 
+    @staticmethod
+    async def delete_pc_chunk_id_namespace(chunk_id: str, namespace: str):
+        """
+        delete chunk from pinecone
+        :param chunk_id:
+        :param namespace:
+        :return:
+        """
+        """
+                Delete namespace from Pinecone index
+                :param namespace:
+                :return:
+                """
+        import pinecone
+        try:
+            pc = pinecone.Pinecone(
+                api_key=const.PINECONE_API_KEY
+            )
+            host = pc.describe_index(const.PINECONE_INDEX).host
+            index = pc.Index(name=const.PINECONE_INDEX, host=host)
+            # vector_store = Pinecone.from_existing_index(const.PINECONE_INDEX,)
+            delete_response = index.delete(ids=[chunk_id], namespace=namespace)
+        except Exception as ex:
+
+            logger.error(ex)
+
+            raise ex
+
     @staticmethod
     async def get_pc_ids_namespace( metadata_id: str, namespace: str) -> PineconeItems:
         """