app.py

import os
import subprocess

# Function to install a package if it is not already installed
def install(package):
    subprocess.check_call([os.sys.executable, "-m", "pip", "install", package])

# Ensure the necessary packages are installed
install("transformers")
install("torch")
install("pandas")
install("gradio")

import pandas as pd
import gradio as gr
from transformers import AutoModel, AutoTokenizer
import torch

# Load the dataset containing PEC numbers and names
def load_dataset(file_path='PEC_Numbers_and_Names.xlsx'):
    if not os.path.exists(file_path):
        raise FileNotFoundError(f"File not found: {file_path}")
    df = pd.read_excel(file_path)
    return df

# Debugging function to get PEC number based on the name
def get_pec_number(name, df):
    print("Column names in DataFrame:", df.columns.tolist())  # Print the column names
    print(f"Looking for Name: '{name}'")
    
    # Normalize the input and dataset
    df['Name'] = df['Name'].str.strip().str.lower()
    name = name.strip().str.lower()
    
    result = df[df['Name'] == name]
    
    if not result.empty:
        print(f"Found PEC Number: {result.iloc[0]['PEC No.']}")
        return result.iloc[0]['PEC No.']
    else:
        print("Name not found.")
        return "Name not found."

# Function to process the name using the Hugging Face model
def process_with_model(name):
    inputs = tokenizer(name, return_tensors="pt")
    with torch.no_grad():
        outputs = model(**inputs)
    return outputs.last_hidden_state.mean(dim=1).squeeze().tolist()

# Combine both functions to create a prediction
def predict(name, file):
    try:
        # Load the dataset from the uploaded file if provided
        if file is not None:
            df = pd.read_excel(file.name)
        else:
            df = load_dataset()

        pec_number = get_pec_number(name, df)
        model_output = process_with_model(name)
        return f"PEC Number: {pec_number}\nModel Output: {model_output}"
    except FileNotFoundError as e:
        return str(e)

# Load the model and tokenizer from Hugging Face
tokenizer = AutoTokenizer.from_pretrained("Alibaba-NLP/gte-multilingual-base", trust_remote_code=True)
model = AutoModel.from_pretrained("Alibaba-NLP/gte-multilingual-base", trust_remote_code=True)

# Build the Gradio interface with file upload option
iface = gr.Interface(
    fn=predict,
    inputs=[
        gr.Textbox(lines=1, placeholder="Enter Name..."),
        gr.File(label="Upload PEC Numbers and Names file (optional)")
    ],
    outputs="text",
    title="Name to PEC Number Lookup with Model Integration",
    description="Enter a name to retrieve the corresponding PEC number and process it with a Hugging Face model. Optionally, upload the Excel file if not found."
)

# Run the Gradio interface
if __name__ == "__main__":
    iface.launch()