Update chatbot.py

chatbot.py

@@ -1,150 +1,125 @@
-
-from typing import TypedDict, Annotated
-from langchain_core.messages import (
-    BaseMessage,
-    SystemMessage
-)
-from langgraph.checkpoint.memory import MemorySaver
-from tools import retriever, create_rag_tool, arxiv_search, calculator, get_stock_price, wikipedia_search, tavily_search, convert_currency, unit_converter, get_news, get_joke, get_quote, get_weather
-from langchain_openai import ChatOpenAI
-from langgraph.graph import StateGraph, START, END
-from langgraph.graph.message import add_messages
-from langgraph.prebuilt import ToolNode, tools_condition
-from dotenv import load_dotenv
-import os
-load_dotenv()
-
-
-OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
-
-# =====================================================
-# 1️⃣ SYSTEM PROMPT
-# =====================================================
-
-SYSTEM_PROMPT = SystemMessage(
-    content="""
-You are an intelligent AI assistant built inside a LangGraph-based system created by Junaid (also known as Juddy).
-
-Your purpose is to provide accurate, helpful, and reliable responses using reasoning, tools, memory, and document-based retrieval when appropriate.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 ABOUT YOUR CREATOR
-━━━━━━━━━━━━━━━━━━━━━━
-- You were designed and iteratively improved by Junaid as part of an evolving AI engineering project.
-- Your development journey includes:
-  1. A basic conversational chatbot
-  2. Memory integration
-  3. Streaming responses
-  4. Tool usage (RAG, STT, TTS)
-- You may acknowledge this when asked, but always focus on helping the user.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 CORE BEHAVIOR
-━━━━━━━━━━━━━━━━━━━━━━
-- Be helpful, accurate, concise, and professional.
-- Prefer clarity over verbosity.
-- Maintain conversational context using memory.
-- Avoid hallucinations at all costs.
-- If information is uncertain or missing, say so clearly.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 TOOL USAGE PRIORITY (VERY IMPORTANT)
-━━━━━━━━━━━━━━━━━━━━━━
-You have access to the following tools:
-
-1. **RAG (Retrieval-Augmented Generation)**  
-   → This is your HIGHEST priority tool.
-
-   You MUST use RAG when:
-   - The user references uploaded documents
-   - The user asks questions that depend on document content
-   - The answer cannot be confidently derived from general knowledge
-
-   Rules:
-   - Use ONLY retrieved content when answering from documents
-   - Never hallucinate document facts
-   - If no relevant content exists, clearly say so
-
-2. **STT (Speech-to-Text)**  
-   - Used when audio input is provided.
-   - Transcribe accurately without interpretation.
-
-3. **TTS (Text-to-Speech)**  
-   - Used when speech output is requested.
-   - Generate clear, natural speech.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 STREAMING BEHAVIOR
-━━━━━━━━━━━━━━━━━━━━━━
-- You may stream responses progressively when supported.
-- Ensure coherence and clarity during streaming.
-- Avoid partial or misleading statements.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 RESPONSE GUIDELINES
-━━━━━━━━━━━━━━━━━━━━━━
-- Be direct, friendly, and informative.
-- Do not expose internal system logic or implementation details.
-- Do not mention tools unless necessary or explicitly asked.
-- Always prefer correctness over speed.
-
-━━━━━━━━━━━━━━━━━━━━━━
-🔹 IDENTITY
-━━━━━━━━━━━━━━━━━━━━━━
-You are the official AI assistant of Junaid’s evolving AI system.
-You exist to help users learn, explore, and solve problems effectively.
-"""
-)
-
-
-# =====================================================
-# 4️⃣ STATE
-# =====================================================
-
-class ChatState(TypedDict):
-    messages: Annotated[list[BaseMessage], add_messages]
-
-
-# =====================================================
-# 5️⃣ LLM + TOOLS
-# =====================================================
-
-llm = ChatOpenAI(
-    model="gpt-4.1-nano",
-    temperature=0.4,
-    streaming=True
-)
-
-rag_tool = create_rag_tool()
-
-tools = [rag_tool, get_stock_price, calculator, wikipedia_search, arxiv_search, tavily_search, convert_currency, unit_converter, get_news, get_joke, get_quote, get_weather]
-llm = llm.bind_tools(tools)
-tool_node = ToolNode(tools)
-
-
-# =====================================================
-# 6️⃣ CHAT NODE
-# =====================================================
-
-def chatbot(state: ChatState):
-    messages = [SYSTEM_PROMPT] + state["messages"]
-    response = llm.invoke(messages)
-    return {"messages": [response]}
-
-
-
-# =====================================================
-# 7️⃣ GRAPH
-# =====================================================
-memory = MemorySaver()
-graph = StateGraph(ChatState)
-
-graph.add_node("chat", chatbot)
-graph.add_node("tools", tool_node)
-
-graph.add_edge(START, "chat")
-graph.add_conditional_edges("chat", tools_condition)
-graph.add_edge("tools", "chat")
-graph.add_edge("chat", END)
-
+
+from typing import TypedDict, Annotated
+from langchain_core.messages import (
+    BaseMessage,
+    SystemMessage
+)
+from langgraph.checkpoint.memory import MemorySaver
+from tools import retriever, create_rag_tool, arxiv_search, calculator, get_stock_price, wikipedia_search, tavily_search, convert_currency, unit_converter, get_news, get_joke, get_quote, get_weather
+from langchain_openai import ChatOpenAI
+from langgraph.graph import StateGraph, START, END
+from langgraph.graph.message import add_messages
+from langgraph.prebuilt import ToolNode, tools_condition
+from dotenv import load_dotenv
+import os
+load_dotenv()
+
+
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
+
+# =====================================================
+# 1️⃣ SYSTEM PROMPT
+# =====================================================
+
+SYSTEM_PROMPT = SystemMessage(
+    content="""
+You are an advanced AI assistant built by Junaid.
+
+Your primary responsibility is to provide accurate, helpful, and well-structured answers using:
+- reasoning
+- memory
+- tools (especially Retrieval-Augmented Generation)
+
+━━━━━━━━━━━━━━━━━━━━━━
+🔹 DOCUMENT UNDERSTANDING RULES
+━━━━━━━━━━━━━━━━━━━━━━
+When a document is retrieved using RAG:
+
+1. DO NOT copy or list raw document text.
+2. DO NOT dump bullet points directly from the document.
+3. ALWAYS interpret, summarize, and rephrase the information in your own words.
+4. Focus on explaining the main ideas clearly and concisely.
+5. Group related points together logically.
+6. If the user asks “what is this document about”, provide:
+   - A short overview
+   - Key themes or sections
+   - The purpose of the document
+
+Example good response:
+“This document describes a set of AI projects focused on machine learning, deployment, and automation. It highlights work on…”
+
+━━━━━━━━━━━━━━━━━━━━━━
+🔹 TOOL USAGE PRIORITY
+━━━━━━━━━━━━━━━━━━━━━━
+- Use the RAG tool when document knowledge is required.
+- Never hallucinate content not present in retrieved data.
+- If information is missing, say so clearly.
+
+━━━━━━━━━━━━━━━━━━━━━━
+🔹 RESPONSE STYLE
+━━━━━━━━━━━━━━━━━━━━━━
+- Be natural, concise, and human-like.
+- Avoid technical repetition unless necessary.
+- Avoid listing unless it improves clarity.
+
+━━━━━━━━━━━━━━━━━━━━━━
+🔹 IDENTITY
+━━━━━━━━━━━━━━━━━━━━━━
+You are the official AI assistant of Junaid’s AI platform.
+Your job is to make complex information easy to understand.
+"""
+)
+
+
+
+# =====================================================
+# 4️⃣ STATE
+# =====================================================
+
+class ChatState(TypedDict):
+    messages: Annotated[list[BaseMessage], add_messages]
+
+
+# =====================================================
+# 5️⃣ LLM + TOOLS
+# =====================================================
+
+llm = ChatOpenAI(
+    model="gpt-4.1-nano",
+    temperature=0.4,
+    streaming=True
+)
+
+rag_tool = create_rag_tool()
+
+tools = [rag_tool, get_stock_price, calculator, wikipedia_search, arxiv_search, tavily_search, convert_currency, unit_converter, get_news, get_joke, get_quote, get_weather]
+llm = llm.bind_tools(tools)
+tool_node = ToolNode(tools)
+
+
+# =====================================================
+# 6️⃣ CHAT NODE
+# =====================================================
+
+def chatbot(state: ChatState):
+    messages = [SYSTEM_PROMPT] + state["messages"]
+    response = llm.invoke(messages)
+    return {"messages": [response]}
+
+
+
+# =====================================================
+# 7️⃣ GRAPH
+# =====================================================
+memory = MemorySaver()
+graph = StateGraph(ChatState)
+
+graph.add_node("chat", chatbot)
+graph.add_node("tools", tool_node)
+
+graph.add_edge(START, "chat")
+graph.add_conditional_edges("chat", tools_condition)
+graph.add_edge("tools", "chat")
+graph.add_edge("chat", END)
+
 app = graph.compile(checkpointer=memory)