bupa1018's picture
Update app.py
df2b26b
raw
history blame
4.15 kB
import os
import json
import gradio as gr
import streamlit as st
from huggingface_hub import HfApi, login
from dotenv import load_dotenv
from download_repo import download_gitlab_repo_to_hfspace
from process_repo import extract_repo_files
from chunking import chunk_pythoncode_and_add_metadata, chunk_text_and_add_metadata
from vectorstore import setup_vectorstore
from llm import get_groq_llm
from kadi_apy_bot import KadiAPYBot
from repo_versions import store_message_from_json
# Load environment variables from .env file
load_dotenv()
# Load configuration from JSON file
with open("config.json", "r") as file:
config = json.load(file)
GROQ_API_KEY = os.environ["GROQ_API_KEY"]
HF_TOKEN = os.environ["HF_Token"]
VECTORSTORE_DIRECTORY = config["vectorstore_directory"]
CHUNK_SIZE = config["chunking"]["chunk_size"]
CHUNK_OVERLAP = config["chunking"]["chunk_overlap"]
EMBEDDING_MODEL_NAME = config["embedding_model"]["name"]
EMBEDDING_MODEL_VERSION = config["embedding_model"]["version"]
LLM_MODEL_NAME = config["llm_model"]["name"]
LLM_MODEL_TEMPERATURE = config["llm_model"]["temperature"]
GITLAB_API_URL = config["gitlab"]["api_url"]
GITLAB_PROJECT_ID = config["gitlab"]["project id"]
GITLAB_PROJECT_VERSION = config["gitlab"]["project version"]
DATA_DIR = config["data_dir"]
HF_SPACE_NAME = config["hf_space_name"]
login(HF_TOKEN)
hf_api = HfApi()
def initialize():
global kadiAPY_bot
download_gitlab_repo_to_hfspace(GITLAB_API_URL, GITLAB_PROJECT_ID, GITLAB_PROJECT_VERSION, DATA_DIR, hf_api, HF_SPACE_NAME)
code_texts, code_references = extract_repo_files(DATA_DIR, ['kadi_apy'], [])
doc_texts, doc_references = extract_repo_files(DATA_DIR, ['docs'], [])
print("Length of code_texts: ", len(code_texts))
print("Length of doc_files: ", len(doc_texts))
code_chunks = chunk_pythoncode_and_add_metadata(code_texts, code_references)
doc_chunks = chunk_text_and_add_metadata(doc_texts, doc_references, CHUNK_SIZE, CHUNK_OVERLAP)
print(f"Total number of code_chunks: {len(code_chunks)}")
print(f"Total number of doc_chunks: {len(doc_chunks)}")
vectorstore = setup_vectorstore(doc_chunks + code_chunks, EMBEDDING_MODEL_NAME, VECTORSTORE_DIRECTORY)
llm = get_groq_llm(LLM_MODEL_NAME, LLM_MODEL_TEMPERATURE, GROQ_API_KEY)
kadiAPY_bot = KadiAPYBot(llm, vectorstore)
initialize()
import streamlit as st
def main():
st.set_page_config(page_title="KadiAPY - AI Coding-Assistant", layout="wide")
st.title("KadiAPY - AI Coding-Assistant")
st.markdown("AI assistant for KadiAPY based on RAG architecture powered by LLM")
tab1, tab2 = st.tabs(["KadiAPY - AI Assistant", "Try Asking"])
with tab1:
st.write("### Kadi Bot")
if "chat_history" not in st.session_state:
st.session_state.chat_history = []
# Chatbot
chatbot_placeholder = st.empty()
question = st.text_input("Question", placeholder="Type in your question and press Enter")
if st.button("Submit"):
if question:
st.session_state.chat_history.append({"User": question})
response = bot_kadi(question) # Replace `bot_kadi` with your response generation function
st.session_state.chat_history.append({"Kadi Bot": response})
if st.button("Clear"):
st.session_state.chat_history = []
for exchange in st.session_state.chat_history:
for speaker, text in exchange.items():
st.write(f"**{speaker}:** {text}")
with tab2:
st.write("### Try Asking...")
examples = [
"Write me a python script which can convert plain JSON to a Kadi4Mat-compatible extra metadata structure",
"I need a method to upload a file to a record. The id of the record is 3",
]
for example in examples:
st.write(f"- {example}")
def bot_kadi(question):
# Dummy function for the AI response, replace with your logic
return "This is a placeholder response from Kadi Bot."
if __name__ == "__main__":
main()