修改文件上传节点

ext/video_agent_deploy.py

@@ -26,7 +26,7 @@ image = (
     .run_commands("comfy node install https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite.git")
     .run_commands("comfy node install https://github.com/WASasquatch/was-node-suite-comfyui.git")
     .run_commands("comfy node install https://github.com/cubiq/ComfyUI_essentials.git")
-    .add_local_dir(local_path=r'D:\code\ComfyUI-CustomNode',
+    .add_local_dir(local_path=r'/Users/charon/Desktop/ComfyUI-CustomNode',
                    remote_path='/root/comfy/ComfyUI/custom_nodes',
                    copy=True
                    )
@@ -58,7 +58,6 @@ vol = modal.Volume.from_name("comfy_model", environment_name="dev", create_if_mi
 @modal.web_server(8000, startup_timeout=120, label='image-video-agent-1')
 def ui_1():
     process = subprocess.Popen("comfy launch -- --cpu --listen 0.0.0.0 --port 8001", shell=True)
-    process.wait()
 
 
 @app.function(

nodes/file_upload.py

@@ -1,73 +1,198 @@
-# -*- coding: utf-8 -*-
-"""
- Author     charon
- Date       2025/9/7 13:12
- """
-import os
-import requests
 import mimetypes
+import uuid
+import boto3
+import os
+from botocore.config import Config
+import asyncio
+import torch
+import numpy as np
+from PIL import Image
 import folder_paths
 
+# 尝试导入 scipy，如果失败则给出提示
+try:
+    import scipy.io.wavfile as wavfile
+except ImportError:
+    print("------------------------------------------------------------------------------------")
+    print("[FileUploadNode] 提示: Scipy 库未安装, 如果需要处理音频输入, 请运行: pip install scipy")
+    print("------------------------------------------------------------------------------------")
+
+# --- AWS S3 配置 ---
+aws_settings = {
+    'access_key_id': 'AKIAYRH5NGRSWHN2L4M6',
+    'secret_access_key': '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',
+    'bucket_name': 'modal-media-cache',
+    'region': 'ap-northeast-2',
+    'cnd_endpoint': 'https://cdn.roasmax.cn'
+}
+
+
+# --- 核心上传逻辑 ---
+async def upload_file_s3_v2(file_path: str, remove: bool = False, perpetual: bool = False):
+    """
+    使用 boto3 客户端异步上传文件到 S3
+    """
+    resp_data = {'status': False, 'data': '', 'msg': ''}
+    if not os.path.isfile(file_path):
+        resp_data['msg'] = f'文件不存在: {file_path}'
+        print(f"[FileUploadNode ERROR] {resp_data['msg']}")
+        return resp_data
+    try:
+        s3_client = boto3.client(
+            "s3",
+            aws_access_key_id=aws_settings['access_key_id'],
+            aws_secret_access_key=aws_settings['secret_access_key'],
+            region_name=aws_settings['region'],
+            endpoint_url="https://s3-accelerate.amazonaws.com",
+            config=Config(s3={'addressing_style': 'virtual'}, signature_version='s3v4')
+        )
+
+        suffix = os.path.splitext(file_path)[-1]
+        bucket_suffix = 'material/' if perpetual else 'upload/'
+        s3_key = f"{bucket_suffix}{uuid.uuid4().hex}{suffix}"
+
+        mime_type, _ = mimetypes.guess_type(file_path)
+        extra_args = {'ContentType': mime_type if mime_type else 'application/octet-stream'}
+
+        print(
+            f"[FileUploadNode INFO] 开始上传文件 {os.path.basename(file_path)} 到 S3 bucket '{aws_settings['bucket_name']}'...")
+        s3_client.upload_file(
+            Filename=os.path.abspath(file_path),
+            Bucket=aws_settings['bucket_name'],
+            Key=s3_key,
+            ExtraArgs=extra_args
+        )
+
+        cdn_url = f"{aws_settings['cnd_endpoint'].rstrip('/')}/{s3_key.lstrip('/')}"
+        resp_data.update(status=True, data=cdn_url, msg='文件成功上传到S3')
+        print(f"[FileUploadNode INFO] 文件成功上传到S3: {cdn_url}")
+
+    except Exception as e:
+        print(f"[FileUploadNode ERROR] 上传文件到S3时出错: {e}")
+        resp_data['msg'] = f'上传文件到S3时出错: {e}'
+    finally:
+        if remove and resp_data['status'] and os.path.exists(file_path):
+            try:
+                os.remove(file_path)
+                print(f"[FileUploadNode INFO] 源文件已根据选项删除: {file_path}")
+            except Exception as e:
+                print(f"[FileUploadNode ERROR] 删除源文件失败: {e}")
+    return resp_data
+
 
 class FileUploadNode:
-    ENVIRONMENT_URL_MAP = {
-        "prod": "https://bowongai-prod--text-video-agent-fastapi-app.modal.run",
-        "test": "https://bowongai-test--text-video-agent-fastapi-app.modal.run",
-        "dev": "https://bowongai-dev--text-video-agent-fastapi-app.modal.run"
-    }
-    API_ENDPOINT = "/api/file/upload/s3"
 
     @classmethod
     def INPUT_TYPES(s):
         return {
             "required": {
-                "file_to_upload": ("*", {"file_upload": True}),
+                "perpetual": ("BOOLEAN", {"default": False}),
-                "environment": (list(s.ENVIRONMENT_URL_MAP.keys()), {"default": "dev"}),
+                "remove_source_file": ("BOOLEAN", {"default": False}),
-                "form_field_name": ("STRING", {"multiline": False, "default": "file"}),
             },
+            "optional": {
+                "video": ("*",),
+                "image": ("IMAGE",),
+                "audio": ("AUDIO",),
+            }
         }
 
-    CATEGORY = "不忘科技-自定义节点🚩/utils/文件上传"
     RETURN_TYPES = ("STRING",)
-    RETURN_NAMES = ("url",)
+    RETURN_NAMES = ("file_url",)
-    FUNCTION = "handler_file_upload"
+    FUNCTION = "upload_file"
+    CATEGORY = "不忘科技-自定义节点🚩/utils/通用文件上传"
 
-    def handler_file_upload(self, file_to_upload, environment, form_field_name):
+    def tensor_to_pil(self, tensor: torch.Tensor) -> Image.Image:
-        if not file_to_upload or not isinstance(file_to_upload, str):
+        image_np = tensor[0].cpu().numpy()
-            return (f"Error: Invalid file input. Please upload a file.",)
+        image_np = (image_np * 255).astype(np.uint8)
+        return Image.fromarray(image_np)
 
-        base_url = self.ENVIRONMENT_URL_MAP.get(environment)
+    def save_pil_to_temp(self, pil_image: Image.Image) -> str:
-        if not base_url:
+        output_dir = folder_paths.get_temp_directory()
-            error_message = f"Error: Invalid environment '{environment}' selected."
+        (full_output_folder, filename, _, _, _) = folder_paths.get_save_image_path("uploader_temp", output_dir)
-            return (error_message,)
+        filepath = os.path.join(full_output_folder, f"{filename}.png")
-        full_api_url = base_url + self.API_ENDPOINT
+        pil_image.save(filepath, 'PNG')
+        return filepath
 
-        local_filepath = folder_paths.get_annotated_filepath(file_to_upload)
+    def save_audio_tensor_to_temp(self, waveform_tensor: torch.Tensor, sample_rate: int) -> str:
+        if 'wavfile' not in globals():
+            raise ImportError("Scipy 库未安装。请在您的 ComfyUI 环境中运行 'pip install scipy' 来启用音频处理功能。")
 
-        if not os.path.exists(local_filepath):
+        waveform_np = waveform_tensor.cpu().numpy()
-            return (f"Error: File not found at path: {local_filepath}",)
+        if waveform_np.ndim == 3: waveform_np = waveform_np[0]
+        if waveform_np.shape[0] < waveform_np.shape[1]: waveform_np = waveform_np.T
 
-        headers = {'accept': 'application/json'}
+        waveform_int16 = np.int16(waveform_np * 32767)
-        filename = os.path.basename(local_filepath)
+        output_dir = folder_paths.get_temp_directory()
-        mime_type, _ = mimetypes.guess_type(local_filepath)
+        (full_output_folder, filename, _, _, _) = folder_paths.get_save_image_path("uploader_temp_audio", output_dir)
-        if mime_type is None:
+        filepath = os.path.join(full_output_folder, f"{filename}.wav")
-            mime_type = 'application/octet-stream'
+        wavfile.write(filepath, sample_rate, waveform_int16)
+        return filepath
 
+    def upload_file(self, perpetual, remove_source_file, image=None, audio=None, video=None):
+        resolved_path = None
+        if video is not None:
+            print('[FileUploadNode INFO] 检测到视频输入...')
+            unwrapped_input = video[0] if isinstance(video, (list, tuple)) and video else video
+            if hasattr(unwrapped_input, 'save_to'):
                 try:
-            with open(local_filepath, 'rb') as f:
+                    output_dir = folder_paths.get_temp_directory()
-                files = {form_field_name: (filename, f, mime_type)}
+                    (full_output_folder, filename, _, _, _) = folder_paths.get_save_image_path("uploader_temp_video",
-                print(f"Uploading '{filename}' to '{full_api_url}'...")
+                                                                                               output_dir)
-                response = requests.post(full_api_url, headers=headers, files=files)
+                    temp_video_path = os.path.join(full_output_folder, f"{filename}.mp4")
-                response.raise_for_status()
+                    unwrapped_input.save_to(temp_video_path)
-                json_response = response.json()
+                    resolved_path = temp_video_path
-
-                if json_response.get("status"):
-                    result_url = json_response.get("data")
-                    print(f"Success! URL: {result_url}")
-                    return (result_url,)
-                else:
-                    api_msg = json_response.get("msg", "Unknown API error")
-                    error_message = f"Error from API: '{api_msg}'"
-                    return (error_message,)
                 except Exception as e:
-            raise ValueError(f'File upload process failed: {e}')
+                    return (f"ERROR: 保存视频时出错: {e}",)
+            else:
+                return (f"ERROR: 不支持的视频输入格式，无法找到 save_to() 方法。",)
+
+        elif image is not None:
+            print('[FileUploadNode INFO] 检测到图像输入，正在保存为临时文件...')
+            pil_image = self.tensor_to_pil(image)
+            resolved_path = self.save_pil_to_temp(pil_image)
+
+        elif audio is not None:
+            print('[FileUploadNode INFO] 检测到音频输入...')
+            audio_info = audio[0] if isinstance(audio, (list, tuple)) and audio else audio
+            if isinstance(audio_info, dict) and 'waveform' in audio_info and 'sample_rate' in audio_info:
+                print('[FileUploadNode INFO] 正在从 waveform 数据保存为临时 .wav 文件...')
+                try:
+                    resolved_path = self.save_audio_tensor_to_temp(audio_info['waveform'], audio_info['sample_rate'])
+                except Exception as e:
+                    return (f"ERROR: 保存音频张量时出错: {e}",)
+            else:
+                return (f"ERROR: 不支持的音频输入格式，需要包含 'waveform' 和 'sample_rate' 的字典。",)
+
+        else:
+            raise ValueError("ERROR: 没有提供有效的媒体输入 (视频/图像/音频)。")
+            # return ("ERROR: 没有提供有效的媒体输入 (视频/图像/音频)。",)
+
+        if not resolved_path or not os.path.exists(resolved_path):
+            return (f"ERROR: 解析后的文件路径无效或文件不存在: {resolved_path}",)
+
+        print(f"[FileUploadNode INFO] 最终待上传文件: {resolved_path}")
+        loop = asyncio.new_event_loop()
+        asyncio.set_event_loop(loop)
+        result = loop.run_until_complete(
+            upload_file_s3_v2(
+                file_path=resolved_path,
+                remove=remove_source_file,
+                perpetual=perpetual
+            )
+        )
+        loop.close()
+
+        if result['status']:
+            return (result['data'],)
+        else:
+            error_message = f"上传失败: {result['msg']}"
+            raise ValueError(error_message)
+            # return (f"ERROR: {error_message}",)
+
+
+NODE_CLASS_MAPPINGS = {
+    "FileUploadNode": FileUploadNode
+}
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "FileUploadNode": "文件上传(s3,gs)"
+}