Update app.py

app.py

@@ -1,27 +1,17 @@
-import io
-import os
+import streamlit as st
 import requests
 import pdfplumber
 import torch
-import ffmpeg
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
-import streamlit as st
 from reportlab.lib.pagesizes import letter
-from reportlab.pdfgen import canvas
-from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
 from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer
+from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
+import os
 
-# Suppress warnings
-import warnings
-warnings.filterwarnings("ignore")
-
-# Define paths for temporary files
-temp_audio_folder = "/tmp/audios/"
-temp_pdf_path = "/tmp/uploaded_pdf.pdf"
-temp_output_pdf_path = "/tmp/response_output.pdf"
-
-# Ensure temporary directories exist
-os.makedirs(temp_audio_folder, exist_ok=True)
+# Define paths (for temporary storage)
+audio_folder_path = "./audio"  # Temporary path for uploaded files
+pdf_path = "./form.pdf"  # Temporary path for uploaded files
+output_pdf_path = "./response_output.pdf"  # Path to save the PDF
 
 # Setup models
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
@@ -40,7 +30,6 @@ whisper_pipe = pipeline(
     device=device
 )
 
-# Granite model URL and headers
 granite_url = "https://us-south.ml.cloud.ibm.com/ml/v1/text/generation?version=2023-05-29"
 granite_headers = {
     "Accept": "application/json",
@@ -128,60 +117,36 @@ def save_responses_to_pdf(responses, output_pdf_path):
     
     document.build(content)
 
-# Set up the Streamlit app
-st.title("FILL IT")
+# Streamlit UI
+st.title("Audio to Form Data Processing")
 
-# Upload multiple audio files
-uploaded_audios = st.file_uploader("Upload audio files", type=["wav", "mp3"], accept_multiple_files=True)
+# File upload
+uploaded_audio = st.file_uploader("Upload Audio File", type=["wav", "mp3"])
+uploaded_pdf = st.file_uploader("Upload PDF File", type=["pdf"])
 
-# Upload PDF file
-uploaded_pdf = st.file_uploader("Upload a PDF file with questions", type=["pdf"])
+if uploaded_audio and uploaded_pdf:
+    # Save uploaded files temporarily
+    audio_path = os.path.join(audio_folder_path, uploaded_audio.name)
+    pdf_path = os.path.join(pdf_path, uploaded_pdf.name)
 
-# Output box to display responses
-output_box = st.empty()
-
-# Button to start processing
-if st.button("Start Processing"):
-    if uploaded_audios and uploaded_pdf:
-        responses = []
-        
-        # Read uploaded PDF file
-        pdf_bytes = uploaded_pdf.read()
-        with open(temp_pdf_path, "wb") as f:
-            f.write(pdf_bytes)
-
-        # Process each uploaded audio file
-        for audio_file in uploaded_audios:
-            audio_bytes = audio_file.read()
-            audio_path = os.path.join(temp_audio_folder, audio_file.name)
-            with open(audio_path, "wb") as f:
-                f.write(audio_bytes)
-            
-            # Transcribe audio
-            transcription = transcribe_audio(audio_path)
-            
-            # Extract text and questions from PDF
-            pdf_text, questions = extract_text_from_pdf(temp_pdf_path)
-            
-            # Generate form data with Granite
-            form_data = generate_form_data(transcription, questions)
-            responses.append(form_data)
-        
-        # Display responses in output box
-        output_box.write("Processing completed. Here are the results:")
-        for index, response in enumerate(responses, start=1):
-            output_box.write(f"File {index}:\n{response}\n")
-        
-        # Save responses to PDF
-        save_responses_to_pdf(responses, temp_output_pdf_path)
-        
-        # Button to download the PDF with responses
-        with open(temp_output_pdf_path, "rb") as f:
-            st.download_button(
-                label="Download Responses as PDF",
-                data=f,
-                file_name="response_output.pdf",
-                mime="application/pdf"
-            )
-    else:
-        st.warning("Please upload both audio files and a PDF file.")
+    with open(audio_path, "wb") as f:
+        f.write(uploaded_audio.read())
+    
+    with open(pdf_path, "wb") as f:
+        f.write(uploaded_pdf.read())
+    
+    # Process files
+    transcribed_text = transcribe_audio(audio_path)
+    pdf_text, pdf_questions = extract_text_from_pdf(pdf_path)
+    form_data = generate_form_data(transcribed_text, pdf_questions)
+    
+    # Display results
+    st.write("### Extracted Form Data")
+    st.text_area("Form Data", form_data, height=300)
+    
+    # Save results to PDF
+    save_responses_to_pdf([form_data], output_pdf_path)
+    
+    # Download link for PDF
+    with open(output_pdf_path, "rb") as f:
+        st.download_button("Download Response PDF", f, file_name="response_output.pdf")