app.py

import streamlit as st
import torch
from peft import PeftModel, PeftConfig
from transformers import AutoModelForSeq2SeqLM,AutoTokenizer,AutoTokenizer
import re


# Process text fun
def process_text(text):
    # Remove extra white space from the text
    text = " ".join(text.split())

    # Remove special characters
    text = re.sub(r"[^a-zA-Z0-9.]+", " ", text)
    
    # Remove multiple full stops
    text = re.sub(r"\.{2,}", ".", text)
    
    return text

# Load model
def load_model():
    config = PeftConfig.from_pretrained("sami606713/medical_report_summarizer")
    base_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/bart-large-cnn")
    model = PeftModel.from_pretrained(base_model, "sami606713/medical_report_summarizer")
    return model

# Load tokenizer
def load_tokenizer():
    tokenizer=AutoTokenizer.from_pretrained("sami606713/medical_report_summarizer")
    return tokenizer

# Summerize text
def summerize_text(text):
    # process the text
    text=process_text(text)
    # tokenize the text
    tokenizer=load_tokenizer()
    inputs = tokenizer(text, max_length=1000, truncation=True, padding='max_length', return_tensors='pt')

    # Generate the output idx
    model=load_model()
    summary_ids = model.generate(**inputs, max_length=300)
    predicted_summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
    return predicted_summary

# set the page layout
st.set_page_config(page_title="Medical Report Summerization", 
                   page_icon="🩺", layout="wide")

# Set the title of the app
st.title('Medical Report Summerization')

# Add a pdf file uploader
# uploaded_file = st.file_uploader("Upload a PDF file", type=["pdf"])
# if uploaded_file is not None:
#     st.write(uploaded_file)

text=st.text_area("Enter report here:\n",height=250)
if st.button("Summarize"):
    with st.spinner("Summarizing..."):  # Changed to st.spinner for a spinner effect
        result = summerize_text(text)
        st.write(f"*{result}*")