app.py

import gradio as gr
import pdfplumber
from transformers import pipeline

# Lżejszy model NER (publicznie dostępny)
extractor = pipeline("ner", model="dbmdz/bert-large-cased-finetuned-conll03-english", aggregation_strategy="simple")

def extract_seller(pdf_file):
    with pdfplumber.open(pdf_file) as pdf:
        # Pobranie tekstu z PDF
        full_text = "\n".join(page.extract_text() for page in pdf.pages if page.extract_text())

    # Podział tekstu na krótkie fragmenty (maks. 512 znaków)
    chunks = [full_text[i:i+512] for i in range(0, len(full_text), 512)]

    seller_name = None

    for chunk in chunks:
        entities = extractor(chunk)

        for entity in entities:
            if "ORG" in entity["entity_group"]:  #