app.py

import re
import gradio as gr
import torch
from transformers import DonutProcessor, VisionEncoderDecoderModel
import os
from PIL import Image

processor = DonutProcessor.from_pretrained("debu-das/donut_receipt_v1.20")
model = VisionEncoderDecoderModel.from_pretrained("debu-das/donut_receipt_v1.20")

device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)

def process_document(image):
    if isinstance(image, str):  # Si l'image est un chemin de fichier
        image = Image.open(image).convert("RGB")
    pixel_values = processor(image, return_tensors="pt").pixel_values
    task_prompt = "<s_cord-v2>"
    decoder_input_ids = processor.tokenizer(task_prompt, add_special_tokens=False, return_tensors="pt").input_ids
    
    outputs = model.generate(
        pixel_values.to(device),
        decoder_input_ids=decoder_input_ids.to(device),
        max_length=model.decoder.config.max_position_embeddings,
        early_stopping=True,
        pad_token_id=processor.tokenizer.pad_token_id,
        eos_token_id=processor.tokenizer.eos_token_id,
        use_cache=True,
        num_beams=1,
        bad_words_ids=[[processor.tokenizer.unk_token_id]],
        return_dict_in_generate=True,
    )
    
    sequence = processor.batch_decode(outputs.sequences)[0]
    sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
    sequence = re.sub(r"<.*?>", "", sequence, count=1).strip()
    
    return processor.token2json(sequence)

def process_batch(images):
    results = []
    for image in images:
        result = process_document(image)
        results.append(result)
    return results

description = "Démo Gradio pour Donut, une instance du modèle VisionEncoderDecoderModel affiné sur CORD (analyse de documents). Pour l'utiliser, téléchargez une ou plusieurs images et cliquez sur Submit, ou cliquez sur l'un des exemples pour les charger."
article = "Cette application permet maintenant de traiter plusieurs images de tickets de caisse à la fois."


demo = gr.Interface(
    fn=process_batch,  # Votre fonction de traitement par lot
    inputs=gr.File(file_count="multiple", type="filepath", label="Téléchargez vos images de tickets de caisse"),
    outputs="json",
    title="Reconnaissance des tickets de caisse en lot 🧾",
    description=description,
    article=article,
    examples=[
        [["example.jpg"]],
        [["example_1.jpg", "example_2.jpg"]],
        [["example_3.jpg", "example_4.jpg"]]
    ],
    cache_examples=False
)

port = int(os.environ.get("PORT", 7860))
demo.launch(server_name="0.0.0.0", server_port=port)