ans_generator1.py

from transformers import pipeline
import PyPDF2
import os

UPLOAD_DIR = "uploaded_pdfs"

class AnswerGenerator:
    def __init__(self):
        # ✅ Default FLAN-T5 model for question answering
        self.qa_pipeline = pipeline("question-answering", model="google/flan-t5-base")

#---------------------------------------------------------------
# updated the modal
 #self.qa_pipeline = pipeline(
           # "question-answering",
          #  model="tiiuae/falcon-7b-instruct",    # <-- Updated model here
          #  tokenizer="tiiuae/falcon-7b-instruct"  # <-- Explicitly specifying tokenizer)
#-----------------------------------------------------------------
    def extract_pdf_text(self, token):
        pdf_path = os.path.join(UPLOAD_DIR, f"{token}.pdf")
        if not os.path.exists(pdf_path):
            raise FileNotFoundError("PDF not found for given token")

        with open(pdf_path, "rb") as f:
            reader = PyPDF2.PdfReader(f)
            return [page.extract_text() or "" for page in reader.pages]  # List of page texts

    def generate_answers(self, token, questions):
        pages = self.extract_pdf_text(token)
        full_text = "\n".join(pages)  # Merge pages as context
        results = []

        for question in questions:
            try:
                # ✅ Default FLAN-T5 logic
                result = self.qa_pipeline(question=question, context=full_text)
                results.append({"question": question, "answer": result["answer"]})



            except Exception as e:
                results.append({"question": question, "answer": "Error", "error": str(e)})

        return results