Integrate LangChain and Save Chat History

Closes #11 (closed)

backend/artguide/ollama_views.py

 # ollama_views.py
 
 from django.http import JsonResponse
-import requests
 import json
 import os
 from django.views.decorators.csrf import csrf_exempt
+from langchain_core.messages import HumanMessage
+from langchain_community.chat_models import ChatOllama
+from langchain_community.chat_message_histories import ChatMessageHistory
+from langchain_core.chat_history import BaseChatMessageHistory
+from langchain_core.runnables.history import RunnableWithMessageHistory
 
-OLLAMA_URL = os.environ.get("OLLAMA_ENDPOINT") + '/api/chat'
+OLLAMA_ENDPOINT = os.environ.get("OLLAMA_ENDPOINT")
+OLLAMA_MODEL = os.environ.get("OLLAMA_MODEL")
+
+store = {}
+def get_session_history(session_id: str) -> BaseChatMessageHistory:
+    if session_id not in store:
+        store[session_id] = ChatMessageHistory()
+    return store[session_id]
+
+chat = ChatOllama(base_url=OLLAMA_ENDPOINT, model=OLLAMA_MODEL)
+with_message_history = RunnableWithMessageHistory(chat, get_session_history)
 
 @csrf_exempt
 def ollama_query(request):
@@ -21,41 +35,15 @@ def ollama_query(request):
         if not message:
             return JsonResponse({"error": "No message provided"}, status=400)
 
-        # Prepare the payload for Ollama API
-        payload = {
-            "model": "phi3:mini",
-            "messages": [
-                { "role": "user", "content": message }
-            ]
-        }
-
-        headers = {'Content-Type': 'application/json'}
-
-        # Send the request to Ollama API
-        response = requests.post(OLLAMA_URL, json=payload, headers=headers)
-
-        # Log the request and response for debugging
-        print(f"Request to Ollama: {payload}")
-        print(f"Response status code: {response.status_code}")
-        print(f"Response content: {response.content}")
-
-        # Check for successful response
-        if response.status_code != 200:
-            return JsonResponse({"error": "Failed to fetch response from Ollama"}, status=response.status_code)
-
-        # Parse the response from Ollama
-        response_content = response.content.decode('utf-8')
-
-        # Log the parsed response
-        print(f"Parsed response content: {response_content}")
+        session_id = "TODO: include in request"
+        config = {"configurable": {"session_id": session_id}}
 
-        # Split the response into lines and process each line
         complete_message = ""
-        for line in response_content.split('\n'):
-            if line.strip():
-                json_line = json.loads(line)
-                complete_message += json_line['message']['content']
+        for chunks in with_message_history.stream([HumanMessage(message)], config=config):
+            complete_message += chunks.content
+            print(chunks.content)
+            # TODO: Send each chunk to the frontend as it is received
 
-        return JsonResponse({"message": complete_message.strip()}, safe=False)
+        return JsonResponse({"message": complete_message}, safe=False)
     except Exception as e:
         return JsonResponse({"error": str(e)}, status=500)