comfyui/ai_project.py

import json
import os
import sys
import time
import uuid
import random
from datetime import datetime
from websocket import create_connection, WebSocketTimeoutException, WebSocketConnectionClosedException
import urllib.request
import urllib.parse


def preprocess_workflow(system_prompt, width, height, batch_size, input_json='flux_work.json', output_json='processed_workflow.json'):
    """
    预处理工作流文件，更新系统提示和图像参数
    """
    try:
        with open(input_json, 'r') as f:
            workflow = json.load(f)
        
        # 更新系统提示
        workflow['31']['inputs']['system_prompt'] = system_prompt
        
        # 更新图像参数
        workflow['27']['inputs']['width'] = str(width)
        workflow['27']['inputs']['height'] = str(height)
        workflow['27']['inputs']['batch_size'] = str(batch_size)
        
        # 保存更新后的工作流
        with open(output_json, 'w') as f:
            json.dump(workflow, f, indent=2)
            
        print(f"工作流已更新并保存到: {output_json}")
        return output_json
        
    except Exception as e:
        print(f"预处理工作流出错: {str(e)}")
        sys.exit(1)

def queue_prompt(prompt, server_address, client_id):
    """向服务器队列发送提示信息"""
    p = {"prompt": prompt, "client_id": client_id}
    data = json.dumps(p).encode('utf-8')
    req = urllib.request.Request(f"http://{server_address}/prompt", data=data)
    return json.loads(urllib.request.urlopen(req).read())

def get_image(filename, subfolder, folder_type, server_address):
    """获取生成的图像"""
    data = {"filename": filename, "subfolder": subfolder, "type": folder_type}
    url_values = urllib.parse.urlencode(data)
    with urllib.request.urlopen(f"http://{server_address}/view?{url_values}") as response:
        return response.read()

def get_history(prompt_id, server_address):
    """获取历史记录"""
    with urllib.request.urlopen(f"http://{server_address}/history/{prompt_id}") as response:
        return json.loads(response.read())

def get_images(ws, prompt, server_address, client_id, timeout=600):
    """获取生成的所有图像"""
    prompt_id = queue_prompt(prompt, server_address, client_id)['prompt_id']
    print(f'提示ID: {prompt_id}')
    output_images = {}
    
    start_time = time.time()
    while True:
        if time.time() - start_time > timeout:
            print(f"超时：等待执行超过{timeout}秒")
            break
            
        try:
            out = ws.recv()
            if isinstance(out, str):
                message = json.loads(out)
                if message['type'] == 'executing':
                    data = message['data']
                    if data['node'] is None and data['prompt_id'] == prompt_id:
                        print('执行完成')
                        break
        except Exception as e:
            print(f"接收消息出错: {str(e)}")
            break

    history = get_history(prompt_id, server_address).get(prompt_id, {})
    if not history:
        print("未找到该提示的历史记录")
        return {}
    
    for node_id, node_output in history['outputs'].items():
        if 'images' in node_output:
            images_output = []
            for image in node_output['images']:
                try:
                    image_data = get_image(image['filename'], image['subfolder'], image['type'], server_address)
                    images_output.append({
                        'data': image_data,
                        'filename': image['filename'],
                        'subfolder': image['subfolder'],
                        'type': image['type']
                    })
                except Exception as e:
                    print(f"获取图像错误: {str(e)}")
            output_images[node_id] = images_output
            
    print(f'获取到 {len(output_images)} 组图像输出')
    return output_images

def generate_images(workflow_file, server_address, output_dir, client_id):
    """生成图像主函数"""
    try:
        # 加载工作流
        with open(workflow_file, 'r', encoding='utf-8') as f:
            workflow_data = json.load(f)
        
        # 使用随机种子
        seed = random.randint(1, 10**8)
        print(f'使用种子: {seed}')
        
        # 更新种子
        workflow_data['25']['inputs']['noise_seed'] = seed
        
        # 创建WebSocket连接
        ws_url = f"ws://{server_address}/ws?clientId={client_id}"
        ws = create_connection(ws_url, timeout=600)
        
        # 获取图像
        images = get_images(ws, workflow_data, server_address, client_id)
        ws.close()
        
        # 保存图像
        saved_files = []
        if images:
            for node_id, image_list in images.items():
                for i, img in enumerate(image_list):
                    timestamp = datetime.now().strftime("%Y%m%d%H%M%S")
                    filename = f"{seed}_{timestamp}_{i}.png"
                    file_path = os.path.join(output_dir, filename)
                    
                    try:
                        with open(file_path, "wb") as f:
                            f.write(img['data'])
                        saved_files.append(file_path)
                        print(f'已保存: {file_path}')
                    except Exception as e:
                        print(f"保存图像错误: {str(e)}")
        
        return saved_files
        
    except Exception as e:
        print(f"生成图像出错: {str(e)}")
        return []

if __name__ == "__main__":
    # 配置参数
    WORKING_DIR = 'output'
    COMFYUI_ENDPOINT = '127.0.0.1:8188'
    DEFAULT_WORKFLOW = './workflows/flux_work.json'
    TEMP_WORKFLOW_DIR = './workflows/temp'
    
    # 从命令行获取输入参数
    if len(sys.argv) != 5:
        print("用法: python test.py <prompt> <width> <height> <batch_size>")
        print("示例: python test.py \"南开大学图书馆，大雨天\" 2048 1024 1")
        sys.exit(1)
    
    system_prompt = sys.argv[1]
    width = int(sys.argv[2])
    height = int(sys.argv[3])
    batch_size = int(sys.argv[4])
    
    # 创建临时目录
    os.makedirs(TEMP_WORKFLOW_DIR, exist_ok=True)
    
    # 创建临时文件路径
    PROCESSED_WORKFLOW = os.path.join(TEMP_WORKFLOW_DIR, f"processed_workflow_{uuid.uuid4().hex}.json")
    
    # 1. 预处理工作流
    workflow_file = preprocess_workflow(
        system_prompt=system_prompt,
        width=width,
        height=height,
        batch_size=batch_size,
        input_json=DEFAULT_WORKFLOW,
        output_json=PROCESSED_WORKFLOW
    )
    
    # 2. 准备输出目录
    os.makedirs(WORKING_DIR, exist_ok=True)
    
    # 创建客户端ID
    client_id = str(uuid.uuid4())
    
    print(f"系统提示: {system_prompt}")
    print(f"图像尺寸: {width}x{height}")
    print(f"批次大小: {batch_size}")
    print(f"工作流文件: {workflow_file}")
    print(f"客户端ID: {client_id}")
    print(f"开始使用ComfyUI生成图像: {COMFYUI_ENDPOINT}")
    
    start_time = time.time()
    
    # 生成图像
    print(f"\n===== 开始生成图像 =====")
    saved_files = generate_images(
        workflow_file=workflow_file,
        server_address=COMFYUI_ENDPOINT,
        output_dir=WORKING_DIR,
        client_id=client_id
    )
    
    # 输出结果
    elapsed = time.time() - start_time
    print(f"\n处理完成，耗时 {elapsed:.2f} 秒")
    print(f"共生成 {len(saved_files)} 张图像")
    print(f"保存位置: {WORKING_DIR}")