From d01d81a6d7b184f43d4be2191ff89e3792813f71 Mon Sep 17 00:00:00 2001
From: SpudGunMan <kellykeeton@yahoo.com>
Date: Thu, 24 Oct 2024 19:21:23 -0700
Subject: [PATCH] openWebUI

---
 modules/llm.py | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/modules/llm.py b/modules/llm.py
index 138b554..4ffb113 100644
--- a/modules/llm.py
+++ b/modules/llm.py
@@ -9,8 +9,19 @@ from modules.log import *
 from ollama import Client as OllamaClient
 from googlesearch import search # pip install googlesearch-python
 
+# enahanced workflow with OpenWebUI
+openWebUI = False
+
+# This is my attempt at a simple RAG implementation it will require some setup
+# you will need to have the RAG data in a folder named rag in the data directory (../data/rag)
+# This is lighter weight and can be used in a standalone environment, needs chromadb
 ragDEV = False
 
+if openWebUI:
+    import requests
+    openWebUI_api_key = "your_api_key"
+    openWebUI_base_url = "http://localhost:3000/api/v1"
+
 if ragDEV:
     import os
     import ollama # pip install ollama
@@ -134,6 +145,18 @@ def query_collection(prompt):
     data = results['documents'][0][0]
     return data
 
+def llm_query_openWebUI(input, nodeID=0, location_name=None):
+    # passes the message-rx to the OpenWebUI API directly
+    headers = {"Authorization": f"Bearer {openWebUI_api_key}"}
+    data = {"model": llmModel, "prompt": input}
+    response = requests.post(f"{openWebUI_base_url}/chat/completions", headers=headers, json=data)
+
+    if response.status_code == 200:
+        return response.json()["choices"][0]["message"]["content"]
+    else:
+        logger.debug(f"System: LLM: {response.status_code} - {response.text}")
+        return "I am having trouble processing your request, please try again later."
+
 def llm_query(input, nodeID=0, location_name=None):
     global antiFloodLLM, llmChat_history
     googleResults = []