解决Logic Apps ParseJson中“Invalid type”错误：Schema问题排查

遇到的错误 #

[
  {
    "message": "Invalid type. Expected Array but got Object.",
    "lineNumber": 0,
    "linePosition": 0,
    "path": "[0]",
    "schemaId": "#/items",
    "errorType": "type",
    "childErrors": []
  }
]

当前使用的ParseJson Schema #

{
    "items": {
        "items": {
            "properties": {
                "AttachmentName": {
                    "type": "string"
                },
                "CopyFileFormat": {
                    "type": "string"
                },
                "EmailSubject": {
                    "type": "string"
                },
                "FromEmailDomain": {
                    "type": "string"
                },
                "copypath": {
                    "type": "string"
                }
            },
            "required": [
                "EmailSubject",
                "FromEmailDomain",
                "AttachmentName",
                "copypath",
                "CopyFileFormat"
            ],
            "type": "array"
        },
        "type": "array"
    },
    "type": "array"
}

存储过程执行输出结果 #

{
  "ResultSets": {
    "Table1": [
      {
        "EmailSubject": "test Value",
        "FromEmailDomain": "gmail.com",
        "AttachmentName": "test0804.PDF",
        "copypath": "test",
        "CopyFileFormat": "202304.pdf"
      }
    ]
  },
  "ReturnCode": 0,
  "OutputParameters": {}
}

问题原因 #

你的ParseJson Schema与存储过程输出的JSON结构完全不匹配：

• 存储过程输出是顶层对象，包含ResultSets、ReturnCode、OutputParameters三个属性，但你的Schema定义顶层为数组，这是核心错误来源。
• 你嵌套了多层数组类型，但实际有效数据在ResultSets.Table1下的数组中，每个元素是单个对象，而非数组。

修正后的完整ParseJson Schema #

直接匹配存储过程的完整输出结构即可，正确Schema如下：

{
    "type": "object",
    "properties": {
        "ResultSets": {
            "type": "object",
            "properties": {
                "Table1": {
                    "type": "array",
                    "items": {
                        "type": "object",
                        "properties": {
                            "EmailSubject": {
                                "type": "string"
                            },
                            "FromEmailDomain": {
                                "type": "string"
                            },
                            "AttachmentName": {
                                "type": "string"
                            },
                            "copypath": {
                                "type": "string"
                            },
                            "CopyFileFormat": {
                                "type": "string"
                            }
                        },
                        "required": [
                            "EmailSubject",
                            "FromEmailDomain",
                            "AttachmentName",
                            "copypath",
                            "CopyFileFormat"
                        ]
                    }
                }
            },
            "required": [
                "Table1"
            ]
        },
        "ReturnCode": {
            "type": "integer"
        },
        "OutputParameters": {
            "type": "object"
        }
    },
    "required": [
        "ResultSets",
        "ReturnCode",
        "OutputParameters"
    ]
}

简化方案（仅处理有效数据） #

如果Logic Apps只需要Table1中的数据，可先通过Compose或Select操作提取ResultSets.Table1，再对该数组使用ParseJson，此时Schema可简化为：

{
    "type": "array",
    "items": {
        "type": "object",
        "properties": {
            "EmailSubject": {
                "type": "string"
            },
            "FromEmailDomain": {
                "type": "string"
            },
            "AttachmentName": {
                "type": "string"
            },
            "copypath": {
                "type": "string"
            },
            "CopyFileFormat": {
                "type": "string"
            }
        },
        "required": [
            "EmailSubject",
            "FromEmailDomain",
            "AttachmentName",
            "copypath",
            "CopyFileFormat"
        ]
    }
}

内容的提问来源于stack exchange，提问作者Gopal Kanjolia