ChatGPT自定义GPT4应用OpenAPI Schema上传S3二进制文件异常求助

问题背景 #

开发自定义GPT4应用时，通过OpenAPI Schema配置PUT接口上传PDF/Word等二进制文件到S3桶，API连接成功但S3中文件为空或仅含元数据。示例请求体如下：

{"metadata": {"name": "Manual", "description": "A comprehensive manual for Capri, including important contact numbers, other information, instructions, recommendations, and FAQs."}, "fileData": "/mnt/data/test.pdf"}

用Postman测试API返回200且文件正常，但导入OpenAPI Schema后出现异常。当前使用的OpenAPI Schema代码如下：

{"openapi": "3.1.0","info": {"title": "S3 Bucket File Management API","version": "1.0.0"},"servers": [{"url": "https://HIDDEN.execute-api.us-east-1.amazonaws.com/prod"}],"paths": {"/bnb-bucket-ham/{filename}": {"put": {"summary": "Upload a file to the S3 bucket","operationId": "uploadFile","parameters": [{"name": "filename","in": "path","required": true,"schema": {"type": "string"}}],"requestBody": {"required": true,"content": {"application/json": {"schema": {"type": "object","properties": {"metadata": {"type": "object","properties": {"name": {"type": "string"},"description": {"type": "string"}},"required": ["name"]},"fileData": {"type": "string","format": "byte"}},"required": ["metadata", "fileData"]}},"application/pdf": {},"application/msword": {},"application/vnd.openxmlformats-officedocument.wordprocessingml.document": {}}},"responses": {"200": {"description": "File uploaded successfully"}}},"get": {"summary": "Retrieve a file from the S3 bucket","operationId": "getFile","parameters": [{"name": "filename","in": "path","required": true,"schema": {"type": "string"}}],"responses": {"200": {"description": "File retrieved successfully","content": {"application/pdf": {"schema": {"type": "string","format": "binary"}},"application/msword": {"schema": {"type": "string","format": "binary"}},"application/vnd.openxmlformats-officedocument.wordprocessingml.document": {"schema": {"type": "string","format": "binary"}}}},"404": {"description": "File not found"}}}}}

可行性判断 #

需求完全可行，问题出在OpenAPI Schema的定义与GPT4自定义应用的文件上传逻辑不匹配，导致文件内容未被正确编码传输。

解决方法 #

• 修正请求体内容类型与结构：原Schema混合定义JSON和二进制内容类型，GPT4调用时可能误传文件路径而非二进制内容。改用multipart/form-data格式，同时支持元数据和二进制文件上传，这是文件上传的标准流程：

  "requestBody": {
    "required": true,
    "content": {
      "multipart/form-data": {
        "schema": {
          "type": "object",
          "properties": {
            "metadata": {
              "type": "object",
              "properties": {
                "name": {"type": "string"},
                "description": {"type": "string"}
              },
              "required": ["name"]
            },
            "fileData": {
              "type": "string",
              "format": "binary"
            }
          },
          "required": ["metadata", "fileData"]
        }
      }
    }
  }
  

• 移除冗余配置：原Schema中单独定义的application/pdf、application/msword等内容类型未指定schema，属于无效配置，会导致GPT4混淆请求格式，直接删除这些条目。
• 适配后端处理逻辑：确认AWS Lambda或后端服务支持multipart/form-data请求，能正确解析元数据和二进制文件流，再将文件内容写入S3桶。若之前仅支持JSON格式，需修改后端代码处理multipart请求。
• 明确调用提示：在Schema的summary或description中补充说明：需将文件内容以二进制形式作为fileData字段上传，同时附带metadata对象，避免GPT4误传文件路径。

内容的提问来源于stack exchange，提问作者Hamoun Karami