handler.py

from typing import Dict, List, Any
from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
from modelscope import snapshot_download
from qwen_vl_utils import process_vision_info
import torch
import os
import base64
import io
from PIL import Image
import ffmpeg
import logging
import requests

class EndpointHandler():
    def __init__(self, path=""):
        self.model_dir = path
        self.model = Qwen2VLForConditionalGeneration.from_pretrained(
            self.model_dir, torch_dtype="auto", device_map="auto"
        )
        self.processor = AutoProcessor.from_pretrained(self.model_dir)

    def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
        """
        data args:
            inputs (str): The input text, including any image or video references.
            max_new_tokens (int, optional): Maximum number of tokens to generate. Defaults to 128.
        Return:
            A dictionary containing the generated text.
        """
        inputs = data.get("inputs")
        max_new_tokens = data.get("max_new_tokens", 128)

        # Construct the messages list from the input string
        messages = [{"role": "user", "content": self._parse_input(inputs)}]

        text = self.processor.apply_chat_template(
            messages, tokenize=False, add_generation_prompt=True
        )
        image_inputs, video_inputs = process_vision_info(messages)

        inputs = self.processor(
            text=[text],
            images=image_inputs,
            videos=video_inputs,
            padding=True,
            return_tensors="pt",
        )
        inputs = inputs.to("cuda" if torch.cuda.is_available() else "cpu")

        generated_ids = self.model.generate(**inputs, max_new_tokens=max_new_tokens)
        generated_ids_trimmed = [
            out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
        ]
        output_text = self.processor.batch_decode(
            generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
        )[0]  # Return a single string

        return {"generated_text": output_text}

    def _parse_input(self, input_string):
        """Parses the input string to identify image/video references and text."""
        content = []
        parts = input_string.split("<image>")
        for i, part in enumerate(parts):
            if i % 2 == 0:  # Text part
                content.append({"type": "text", "text": part.strip()})
            else:  # Image/video part
                if part.startswith("video:"):
                    video_path = part.split("video:")[1].strip()
                    video_frames = self._extract_video_frames(video_path)
                    if video_frames:
                        content.append({"type": "video", "video": video_frames, "fps": 1})  # Add fps
                else:
                    image = self._load_image(part.strip())
                    if image:
                        content.append({"type": "image", "image": image})
        return content

    def _load_image(self, image_data):
        """Loads an image from a URL or base64 encoded string."""
        if image_data.startswith("http"):
            try:
                image = Image.open(requests.get(image_data, stream=True).raw)
            except Exception as e:
                logging.error(f"Error loading image from URL: {e}")
                return None
        elif image_data.startswith("data:image"):
            try:
                image_data = image_data.split(",")[1]
                image_bytes = base64.b64decode(image_data)
                image = Image.open(io.BytesIO(image_bytes))
            except Exception as e:
                logging.error(f"Error loading image from base64: {e}")
                return None
        else:
            logging.error("Invalid image data format. Must be URL or base64 encoded.")
            return None
        return image

    def _extract_video_frames(self, video_path, fps=1):
        """Extracts frames from a video at the specified FPS."""
        try:
            probe = ffmpeg.probe(video_path)
            video_stream = next((stream for stream in probe['streams'] if stream['codec_type'] == 'video'), None)
            if not video_stream:
                logging.error(f"No video stream found in {video_path}")
                return None

            width = int(video_stream['width'])
            height = int(video_stream['height'])

            out, _ = (
                ffmpeg
                .input(video_path)
                .filter('fps', fps=fps)
                .output('pipe:', format='rawvideo', pix_fmt='rgb24')
                .run(capture_stdout=True)
            )
            frames = []
            for i in range(0, len(out), width * height * 3):
                frame_data = out[i:i + width * height * 3]
                frame = Image.frombytes('RGB', (width, height), frame_data)
                frames.append(frame)
            return frames

        except ffmpeg.Error as e:
            logging.error(f"Error extracting video frames: {e}")
            return None