app.py

import gradio as gr
from urllib.request import urlopen
import certifi
import json
import pandas as pd
import os
import warnings
warnings.filterwarnings("ignore")

from langchain.document_loaders import CSVLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.chains import RetrievalQA
from langchain.prompts import PromptTemplate
from langchain_community.llms import HuggingFacePipeline
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

# Hugging Face API Key
os.environ["HUGGINGFACEHUB_API_TOKEN"] = ""

# Load Falcon-7B
model_id = "tiiuae/falcon-7b-instruct"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    max_new_tokens=512,
    do_sample=True,
    temperature=0.1,
    top_k=50,
    top_p=0.95,
    eos_token_id=tokenizer.eos_token_id,
)

llm = HuggingFacePipeline(pipeline=pipe)

# Prompt Template
template = """
You are a Financial Market Expert. Get the Market Economic Data and Market News about Company and Build the Financial Report.
Understand this Market Information: {context}
Answer the Query: {question}
Output must be in tabular form.
"""
PROMPT = PromptTemplate(input_variables=["context", "question"], template=template)

# === Core Function ===

def generate_report(ticker: str, exchange: str):
    api_key = "C1HRSweTniWdBuLmTTse9w8KpkoiouM5"
    try:
        # Fetch Data
        if exchange == "NSE":
            url = f"https://financialmodelingprep.com/api/v3/search?query={ticker}&exchange=NSE&apikey={api_key}"
        else:
            url = f"https://financialmodelingprep.com/api/v3/quote/{ticker}?apikey={api_key}"
        
        response = urlopen(url, cafile=certifi.where())
        data = json.loads(response.read().decode("utf-8"))

        # Save to CSV
        df = pd.DataFrame(data)
        if 'timestamp' in df.columns:
            df['timestamp'] = pd.to_datetime(df['timestamp'], errors='coerce')
        if 'earningsAnnouncement' in df.columns:
            df['earningsAnnouncement'] = pd.to_datetime(df['earningsAnnouncement'], errors='coerce')
        df.to_csv("eco_ind.csv", index=False)

        # Chroma Vector Store
        loader = CSVLoader("eco_ind.csv")
        documents = loader.load()
        splitter = RecursiveCharacterTextSplitter(chunk_size=100, chunk_overlap=10)
        texts = splitter.split_documents(documents)
        embeddings = HuggingFaceEmbeddings()

        chroma = Chroma.from_documents(
            documents=texts,
            collection_name="economic_data",
            embedding=embeddings,
            persist_directory="docs/chroma_rag"
        )

        retriever = chroma.as_retriever(search_kwargs={"k": 2})
        qa = RetrievalQA.from_chain_type(
            llm=llm,
            chain_type="stuff",
            chain_type_kwargs={"prompt": PROMPT},
            retriever=retriever,
            return_source_documents=False
        )

        # Ask LLM
        query = f"{ticker} Financial Report"
        result = qa({"query": query})
        return result["result"]

    except Exception as e:
        return f"❌ Error occurred: {e}"

# === Gradio UI ===

iface = gr.Interface(
    fn=generate_report,
    inputs=[
        gr.Textbox(label="Enter Ticker (e.g., MSFT)"),
        gr.Radio(choices=["US", "NSE"], label="Select Exchange")
    ],
    outputs=gr.Textbox(label="Generated Financial Report"),
    title="📊 Financial Report Generator using Falcon + Chroma + LangChain",
    description="Enter a stock ticker and choose an exchange to get an LLM-generated tabular financial report."
)

if __name__ == "__main__":
    iface.launch(share=True)