app.py

import gradio as gr
import boto3
import re
import os

# Credenciales desde Secrets
textract = boto3.client(
    "textract",
    region_name="us-east-1",
    aws_access_key_id="ASIA4LHG3BULWEOMKU4Z",
    aws_secret_access_key="C3fjhuRYrATatQPxt3V1/9S/30OegS1XhBFHgEq4",
    aws_session_token="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"
)

total_acumulat = 0.0

def obtenir_total_tiquet(response):
    paraules_clau = [
        "TOTAL", "TOTAL DUE", "IMPORT", "IMPORT TOTAL",
        "A PAGAR", "TOTAL A PAGAR", "AMOUNT", "AMOUNT DUE"
    ]

    linies = []
    for bloc in response["Blocks"]:
        if bloc["BlockType"] == "LINE":
            linies.append(bloc["Text"])

    for i, linia in enumerate(linies):
        text = linia.upper()

        if "SUBTOTAL" in text:
            continue

        for paraula in paraules_clau:
            if paraula in text:

                numeros = re.findall(r"\d+[.,]\d{2}", linia)
                if numeros:
                    return numeros[-1]

                for j in range(1, 4):
                    if i + j < len(linies):
                        numeros = re.findall(r"\d+[.,]\d{2}", linies[i + j])
                        if numeros:
                            return numeros[-1]

    return None


def formatar_total(total_text):
    if total_text is None:
        return None

    total_text = total_text.replace(",", ".")

    try:
        return round(float(total_text), 2)
    except:
        return None


def processar_tiquet(ruta_imatge):
    global total_acumulat

    try:
        with open(ruta_imatge, "rb") as f:
            imatge_bytes = f.read()

        response = textract.analyze_document(
            Document={"Bytes": imatge_bytes},
            FeatureTypes=["FORMS"]
        )

        total_text = obtenir_total_tiquet(response)
        total_num = formatar_total(total_text)

        if total_num is not None:
            total_acumulat += total_num
            return f"{total_num:.2f}", f"{total_acumulat:.2f}", "OK"
        else:
            return "No detectat", f"{total_acumulat:.2f}", "No s'ha pogut detectar"

    except Exception as e:
        return "Error", f"{total_acumulat:.2f}", str(e)


with gr.Blocks() as demo:
    gr.Markdown("# Extractor de totals de tiquets Toni Vivas")
    gr.Markdown(
        "Puja una imatge d'un tiquet de compra i l'aplicació extraura l'import total "
        "i anira acumulant els imports dels diferents tiquets."
    )

    with gr.Row():
        with gr.Column(scale=1):
            imatge_input = gr.Image(
                label="Puja una imatge del tiquet",
                type="filepath",
                height=420
            )

        with gr.Column(scale=1):
            import_detectat = gr.Textbox(
                label="Import detectat",
                interactive=False
            )

            import_acumulat = gr.Textbox(
                label="Import acumulat",
                interactive=False
            )

            estat = gr.Textbox(
                label="Estat",
                interactive=False
            )

            boto = gr.Button("Analitzar")

    boto.click(
        fn=processar_tiquet,
        inputs=imatge_input,
        outputs=[import_detectat, import_acumulat, estat]
    )

demo.launch()