app.py

import gradio as gr
from transformers import AutoProcessor, Qwen2_5_VLForConditionalGeneration, TextIteratorStreamer
from transformers.image_utils import load_image
from threading import Thread
import time
import torch
import spaces
import cv2
import numpy as np
from PIL import Image
import tempfile

from parler_tts import ParlerTTSForConditionalGeneration
from transformers import AutoTokenizer
import soundfile as sf

device = "cuda:0" if torch.cuda.is_available() else "cpu"

tts_model = ParlerTTSForConditionalGeneration.from_pretrained("CONCREE/Adia_TTS").to(device)
tts_tokenizer = AutoTokenizer.from_pretrained("CONCREE/Adia_TTS")

@spaces.GPU
def tts(text):
    output_wav_path = tempfile.mktemp(suffix=".wav")
    # Description du style vocal
    description = "A clear and educational voice, with a flow adapted to learning"
    
    # Génération
    input_ids = tts_tokenizer(description, return_tensors="pt").input_ids.to(device)
    prompt_ids = tts_tokenizer(text, return_tensors="pt").input_ids.to(device)
    
    audio = tts_model.generate(
        input_ids=input_ids,
        prompt_input_ids=prompt_ids,
    )
    sf.write(output_wav_path, audio.cpu().numpy().squeeze(), tts_model.config.sampling_rate)  # Fixed: was 'model.config'

    return output_wav_path


def progress_bar_html(label: str) -> str:
    """
    Returns an HTML snippet for a thin progress bar with a label.
    The progress bar is styled as a dark animated bar.
    """
    return f'''
<div style="display: flex; align-items: center;">
    <span style="margin-right: 10px; font-size: 14px;">{label}</span>
    <div style="width: 110px; height: 5px; background-color: #9370DB; border-radius: 2px; overflow: hidden;">
        <div style="width: 100%; height: 100%; background-color: #4B0082; animation: loading 1.5s linear infinite;"></div>
    </div>
</div>
<style>
@keyframes loading {{
    0% {{ transform: translateX(-100%); }}
    100% {{ transform: translateX(100%); }}
}}
</style>
    '''

def downsample_video(video_path):
    """
    Downsamples the video to 10 evenly spaced frames.
    Each frame is converted to a PIL Image along with its timestamp.
    """
    vidcap = cv2.VideoCapture(video_path)
    total_frames = int(vidcap.get(cv2.CAP_PROP_FRAME_COUNT))
    fps = vidcap.get(cv2.CAP_PROP_FPS)
    frames = []
    if total_frames <= 0 or fps <= 0:
        vidcap.release()
        return frames
    # Sample 10 evenly spaced frames.
    frame_indices = np.linspace(0, total_frames - 1, 10, dtype=int)
    for i in frame_indices:
        vidcap.set(cv2.CAP_PROP_POS_FRAMES, i)
        success, image = vidcap.read()
        if success:
            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
            pil_image = Image.fromarray(image)
            timestamp = round(i / fps, 2)
            frames.append((pil_image, timestamp))
    vidcap.release()
    return frames

MODEL_ID = "kaamd/chtvctr" # "kaamd/chtvctr"  # Alternatively: "Qwen/Qwen2.5-VL-3B-Instruct" 
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True, min_pixels=256*28*28, max_pixels=1280*28*28)
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
    MODEL_ID,
    trust_remote_code=True,
    torch_dtype=torch.bfloat16
).to("cuda").eval()

@spaces.GPU
def model_inference(input_dict, history):
    text = input_dict["text"]
    files = input_dict["files"]

    if text.strip().lower().startswith("@video-infer"):
        # Remove the tag from the query.
        text = text[len("@video-infer"):].strip()
        if not files:
            raise gr.Error("Please upload a video file along with your @video-infer query.")  # Fixed: gr.Error syntax
        # Assume the first file is a video.
        video_path = files[0]
        frames = downsample_video(video_path)
        if not frames:
            raise gr.Error("Could not process video.")  # Fixed: gr.Error syntax
        # Build messages: start with the text prompt.
        messages = [
            {
                "role": "user",
                "content": [{"type": "text", "text": text}]
            }
        ]
        # Append each frame with a timestamp label.
        for image, timestamp in frames:
            messages[0]["content"].append({"type": "text", "text": f"Frame {timestamp}:"})
            messages[0]["content"].append({"type": "image", "image": image})
        # Collect only the images from the frames.
        video_images = [image for image, _ in frames]
        # Prepare the prompt.
        prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
        inputs = processor(
            text=[prompt],
            images=video_images,
            return_tensors="pt",
            padding=True,
        ).to("cuda")
        # Set up streaming generation.
        streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
        generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=256, temperature=2, min_p=0.8)
        thread = Thread(target=model.generate, kwargs=generation_kwargs)
        thread.start()
        buffer = ""
        for new_text in streamer:
            buffer += new_text
            time.sleep(0.001)
            yield buffer
        return  # Fixed: Added return to prevent falling through

    if len(files) > 1:
        images = [load_image(image) for image in files]
    elif len(files) == 1:
        images = [load_image(files[0])]
    else:
        images = []

    if text == "" and not images:
        raise gr.Error("Please input a query and optionally image(s).")  # Fixed: gr.Error syntax
    if text == "" and images:
        raise gr.Error("Please input a text query along with the image(s).")  # Fixed: gr.Error syntax

    messages = [
        {
            "role": "user",
            "content": [
                *[{"type": "image", "image": image} for image in images],
                {"type": "text", "text": text},
            ],
        }
    ]
    prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    inputs = processor(
        text=[prompt],
        images=images if images else None,
        return_tensors="pt",
        padding=True,
    ).to("cuda")
    streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024)
    thread = Thread(target=model.generate, kwargs=generation_kwargs)
    thread.start()
    buffer = ""
    for new_text in streamer:
        buffer += new_text
        time.sleep(0.01)
        yield buffer

    # This will only be reached after streaming is complete
    # Generate TTS for the final buffer content
    audio_path = tts(buffer)
    return audio_path  # Return the audio file path

# Option 1: Use regular Interface with streaming (recommended)
with gr.Blocks() as demo:
    gr.Markdown("# oolel-vision-experimental `@video-infer for video understanding`")
    
    chatbot = gr.Chatbot()
    msg = gr.MultimodalTextbox(
        label="Query Input", 
        file_types=["image", "video"], 
        file_count="multiple"
    )
    audio_output = gr.Audio(label="Generated Speech")
    clear = gr.Button("Clear")
    
    def respond(message, chat_history):
        # Add user message to chat history
        bot_message = ""
        chat_history.append([message["text"], ""])
        
        # Stream the response
        for response in model_inference(message, chat_history):
            bot_message = response
            chat_history[-1][1] = bot_message
            yield "", chat_history, None
        
        # Generate audio after streaming is complete
        try:
            audio_path = tts(bot_message)
            yield "", chat_history, audio_path
        except Exception as e:
            print(f"TTS Error: {e}")
            yield "", chat_history, None
    
    msg.submit(respond, [msg, chatbot], [msg, chatbot, audio_output])
    clear.click(lambda: ([], None), outputs=[chatbot, audio_output])

# Option 2: Use ChatInterface without outputs parameter (simpler but no audio)
# demo = gr.ChatInterface(
#     fn=model_inference,
#     description="# oolel-vision-experimental `@video-infer for video understanding`**",
#     fill_height=True,
#     textbox=gr.MultimodalTextbox(label="Query Input", file_types=["image", "video"], file_count="multiple"),
#     stop_btn="Stop Generation",
#     multimodal=True,
#     cache_examples=False,
# )

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