Enable search prompt using ddg

README.md

@@ -1,16 +1,26 @@
 # The Llama4U Project
 [![Python application](https://github.com/virajmalia/llama4u/actions/workflows/CI.yml/badge.svg)](https://github.com/virajmalia/llama4u/actions/workflows/CI.yml)
 
-## Vision
-Develop a free and open source, fully-featured AI solution with agents.
+Llama4U is a privacy-focused AI assistant developed using [Ollama][1], [LangChain][2] and [Llama3][3]. A completely free AI solution that can be hosted locally, while providing online capabilities in a responsible and user-controllable way.
+
+#### *APIs that have usage limitations or require keys to be registered with an online account won't be added to this project.*
+
+## Steps to run
+1. Host `llama3` model from [Ollama][1] on your computer.
+2. Clone this repository.
+3. `pip install -e .`
+4. `llama4u`
+
+`llama4u --help` for full CLI.
+
+## List of chat commands
+
+- `/search`: Perform online search using DuckDuckGo
 
 ## Current motivations for feature-set
 - Perplexity AI
 - ChatGPT/GPT4o
 
-## Rule
-- APIs that have usage limitations or require keys to be registered with an online account won't be added to this project.
-
 ## System requirements
 - Powerful CPU or Nvidia GPU (>=8G VRAM)
 - Ubuntu 22.04
@@ -32,16 +42,6 @@ fi
 echo $CUDACXX && $CUDACXX --version
 ```
 
-## Steps to run
-1. Host `llama3` model from [Ollama][1] on your computer
-2. `pip install -e .`
-3. `llama4u`
-
-`llama4u --help` for full CLI
-
-## Description
-Llama4U is an AI assistant developed using [Ollama][1], [LangChain][2] and [Llama3][3]. A completely free AI solution that can be hosted locally, while providing online capabilities in a responsible and user-controllable way.
-
 ## Credits
 - Meta, for the open source Llama models
 - Ollama

pyproject.toml

@@ -23,7 +23,6 @@ dependencies = [
     "langchain-core",
     "langchain-community",
     "langchain-chroma",
-    "duckduckgo_search",
     "termcolor"
 ]
 

src/llama4u.py

@@ -3,6 +3,7 @@
 from termcolor import colored
 from langchain_community.chat_models.ollama import ChatOllama
 from langchain_community.chat_message_histories.in_memory import ChatMessageHistory
+from langchain_community.tools.ddg_search.tool import DuckDuckGoSearchRun
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 from langchain_core.runnables.history import RunnableWithMessageHistory
 from input.input import parse_arguments
@@ -46,9 +47,17 @@ def get_session_history(self, session_id):
     async def chat_session(self):
         """ Chat session with history """
         while True:
+            # Get input
             print(colored('>>> ', 'yellow'), end="")
             user_prompt = input()
 
+            # Redirect search queries
+            if user_prompt.startswith("/search"):
+                search_results = DuckDuckGoSearchRun().run(user_prompt.replace("/search", ""))
+                user_prompt = \
+                f"Summarize the following search results as if you are answering:{search_results}"
+
+            # Invoke chain
             response = self.with_msg_history.invoke(
                 {"input": user_prompt},
                 config={"configurable": {"session_id": "abc123"}},