app.py

import gradio as gr
import base64
import requests
from PIL import Image
import io

API_KEY = "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"
IMAGE_MODEL = "OpenGVLab/InternVL3-14B"

def extract_passport_info(images, document_type):
    results = []

    for image in images:
        # Convert image to base64
        buffered = io.BytesIO()
        image.save(buffered, format="JPEG")
        encoded_image = base64.b64encode(buffered.getvalue()).decode("utf-8")
        data_url = f"data:image/jpeg;base64,{encoded_image}"

        # Prompt to extract full passport data
        prompt = (
            f"Extract all passport information from the uploaded {document_type} image. "
            "Include MRZ (if present), full name, passport number, nationality, gender, "
            "date of birth, date of issue, expiry date, issuing country, and any other text or labels in other languages. "
            "Return the result in a JSON format."
        )

        # OpenRouter Payload
        payload = {
            "model": IMAGE_MODEL,
            "messages": [
                {
                    "role": "user",
                    "content": [
                        {"type": "text", "text": prompt},
                        {"type": "image_url", "image_url": {"url": data_url}},
                    ],
                }
            ],
        }

        headers = {
            "Authorization": f"Bearer {API_KEY}",
            "Content-Type": "application/json"
        }

        try:
            response = requests.post("https://openrouter.ai/api/v1/chat/completions", headers=headers, json=payload)
            result = response.json()
            print("📡 Status:", response.status_code)
            print("📡 Raw Result:", result)

            if "choices" in result:
                extracted = result["choices"][0]["message"]["content"]
                results.append({
                    "document_type": document_type,
                    "extracted_info": extracted
                })
            else:
                results.append({
                    "document_type": document_type,
                    "extracted_info": "❌ No data extracted"
                })

        except Exception as e:
            results.append({
                "document_type": document_type,
                "extracted_info": f"⚠️ Error: {str(e)}"
            })

    return results


# Gradio UI
demo = gr.Interface(
    fn=extract_passport_info,
    inputs=[
        gr.Image(type="pil", label="Upload Passport/Document Images", multiple=True),
        gr.Dropdown(
            choices=["passport_front", "passport_back", "photo", "hotel_reservation"],
            label="Document Type",
            value="passport_front",
        )
    ],
    outputs="json",
    title="Passport & Document Info Extractor",
    description="Upload one or more document images. Extracted information will include MRZ and all available text, structured in JSON format.",
)

if __name__ == "__main__":
    demo.launch()