Maxi0811/pid-interactive-demo

1import gradio as gr2import pytesseract3from PIL import Image4from PyPDF2 import PdfReader5import os6import tempfile7from transformers import pipeline8 9# Load a small language model for Q&A10qa_pipeline = pipeline("question-answering", model="deepset/roberta-base-squad2")11 12def extract_text_from_image(image):13    if image is None:14        return ""15    text = pytesseract.image_to_string(image)16    return text17 18def extract_text_from_pdf(pdf_file):19    if pdf_file is None:20        return ""21    reader = PdfReader(pdf_file)22    text = ""23    for page in reader.pages:24        text += page.extract_text() or ""25    return text26 27def combine_text(image, doc):28    image_text = extract_text_from_image(image)29    doc_text = extract_text_from_pdf(doc) if doc.name.endswith('.pdf') else doc.read().decode('utf-8')30    return f"{image_text}\n\n{doc_text}"31 32def answer_question(image, doc, question):33    context = combine_text(image, doc)34    if not context.strip():35        return "No valid text found in the image or manual."36    result = qa_pipeline(question=question, context=context)37    return result['answer']38 39iface = gr.Interface(40    fn=answer_question,41    inputs=[42        gr.Image(type="pil", label="Upload P&ID Image"),43        gr.File(label="Upload Working Manual (.txt or .pdf)"),44        gr.Textbox(label="Ask a Question")45    ],46    outputs="text",47    title="P&ID + Manual Interactive QA",48    description="Upload a P&ID image and an operational manual, and ask questions about them."49)50 51iface.launch()52