app.py

import gradio as gr
import requests
import json
import time
import os

# --- D-ID API Configuration ---
D_ID_API_BASE_URL = "https://api.d-id.com"
D_ID_TALKS_ENDPOINT = f"{D_ID_API_BASE_URL}/talks"
DEFAULT_AVATAR_IMAGE_URL = "https://placehold.co/256x256/ADD8E6/000000?text=Avatar"

# --- Available Voices (expandable) ---
AVAILABLE_VOICES = [
    "en-US-JennyNeural",
    "en-US-GuyNeural",
    "en-GB-RyanNeural",
    "en-GB-SoniaNeural",
    "es-ES-ElviraNeural",
    "fr-FR-DeniseNeural"
]

# --- Function to call D-ID API ---
def generate_did_video(api_key: str, text_input: str, avatar_image_url: str, voice_id: str):
    if not api_key:
        yield None, "❌ Error: D-ID API Key is required."
        return
    if not text_input:
        yield None, "❌ Error: Text input is required."
        return

    avatar_image_url = avatar_image_url or DEFAULT_AVATAR_IMAGE_URL

    headers = {
        "Authorization": f"Basic {api_key}",
        "Content-Type": "application/json"
    }

    payload = {
        "script": {
            "type": "text",
            "input": text_input,
            "provider": {
                "type": "microsoft",
                "voice_id": voice_id
            }
        },
        "source_url": avatar_image_url
    }

    try:
        yield None, "📤 Sending request to D-ID API..."
        response = requests.post(D_ID_TALKS_ENDPOINT, headers=headers, json=payload)
        response.raise_for_status()
        talk_data = response.json()
        talk_id = talk_data.get("id")
        if not talk_id:
            yield None, "❌ Error: No talk ID returned."
            return

        yield None, f"⏳ Talk created (ID: {talk_id}). Polling status..."

        status = ""
        retries = 0
        max_retries = 30
        video_url = None

        while retries < max_retries:
            time.sleep(5)
            retries += 1
            poll_response = requests.get(f"{D_ID_TALKS_ENDPOINT}/{talk_id}", headers=headers)
            poll_response.raise_for_status()
            status_data = poll_response.json()
            status = status_data.get("status", "")

            if status == "done":
                video_url = status_data.get("result_url")
                if video_url:
                    yield video_url, "✅ Video generation complete!"
                    return
                else:
                    yield None, "❌ Error: No result URL found."
                    return
            elif status == "error":
                error_message = status_data.get("error", "Unknown error")
                yield None, f"❌ Error: {error_message}"
                return
            else:
                yield None, f"🔄 Status: {status} (Attempt {retries}/{max_retries})"

        yield None, "❌ Error: Timeout. Try again later."

    except requests.exceptions.RequestException as e:
        yield None, f"❌ Request error: {e}"
    except Exception as e:
        yield None, f"❌ Unexpected error: {e}"

# --- Gradio Interface ---
with gr.Blocks() as demo:
    gr.Markdown("# 🗣️ D-ID Talking Avatar Video Generator")

    with gr.Row():
        api_key_input = gr.Textbox(
            label="🔐 D-ID API Key (Base64)",
            type="password",
            placeholder="e.g., aGVsbG86cGFzc3dvcmQ=",
            info="Base64 of `username:password`"
        )
        avatar_image_url_input = gr.Textbox(
            label="🖼️ Avatar Image URL (Optional)",
            placeholder=DEFAULT_AVATAR_IMAGE_URL,
            info="Public image URL. Default avatar will be used if blank."
        )

    with gr.Row():
        text_input = gr.Textbox(
            label="📝 Text for Avatar",
            lines=4,
            placeholder="Type what you want your avatar to say..."
        )
        voice_dropdown = gr.Dropdown(
            label="🎤 Voice Selection",
            choices=AVAILABLE_VOICES,
            value=AVAILABLE_VOICES[0]
        )

    generate_btn = gr.Button("🚀 Generate Video")
    output_video = gr.Video(label="🎬 Output Video")
    status_output = gr.Markdown("")

    generate_btn.click(
        fn=generate_did_video,
        inputs=[api_key_input, text_input, avatar_image_url_input, voice_dropdown],
        outputs=[output_video, status_output]
    )

    gr.Examples(
        examples=[
            ["", "Hello! I'm your AI avatar.", DEFAULT_AVATAR_IMAGE_URL, "en-US-JennyNeural"],
            ["", "This is a demo of D-ID + Hugging Face!", DEFAULT_AVATAR_IMAGE_URL, "en-US-GuyNeural"]
        ],
        inputs=[api_key_input, text_input, avatar_image_url_input, voice_dropdown]
    )

demo.launch()