.NET Core与.NET Framework间DataTable转二进制数组数据不一致问题

我需要在.NET Core API和.NET Framework窗体应用之间通过二进制数组交换DataTable数据，但发现两者的序列化行为存在差异，导致转换后的DataTable与API发送的数据不一致。现有以下API端C#代码和客户端VB.NET代码，希望实现转换后DataTable完全一致。

API端C#代码 #

public byte[] GetDataTable(string sqlStr, string connectionString)
{
        XmlSerializer serializer = new XmlSerializer(typeof(DataTable));
        DataTable dataTable = CriaDataTable(sqlStr,connectionString);

        using (MemoryStream memoryStream = new MemoryStream())
        {
            serializer.Serialize(memoryStream, dataTable);
            return memoryStream.ToArray();
        }
}

客户端VB.NET代码 #

Public Function ApiGetDataTable(sqlStr As String, connectionString As String) As DataTable
    Dim url As String = $"{_baseUrl}GetDataTable?sqlStr={Uri.EscapeDataString(sqlStr)}&connectionString={Uri.EscapeDataString(connectionString)}"

    Dim request As New HttpRequestMessage(HttpMethod.Get, url)
    request.Headers.Add("accept", "application/octet-stream")

    Dim response = _httpClient.SendAsync(request).Result
    response.EnsureSuccessStatusCode()

    Dim content As Byte() = response.Content.ReadAsByteArrayAsync().Result

    Dim serializer As New XmlSerializer(GetType(DataTable))

    Using memoryStream As New MemoryStream(content)
        Dim dataTable As DataTable = DirectCast(serializer.Deserialize(memoryStream), DataTable)
        Return dataTable
    End Using

    Return Nothing

End Function

解决方案：改用DataTable原生XML读写方法 #

XmlSerializer对DataTable的序列化在.NET Framework和.NET Core之间存在行为差异（比如类型元数据的存储逻辑不同），导致反序列化后数据或结构不一致。使用DataTable内置的WriteXml和ReadXml方法可以保证跨框架兼容性，这是DataTable的标准化序列化实现，跨平台行为一致。

修改后的API端C#代码 #

public byte[] GetDataTable(string sqlStr, string connectionString)
{
    DataTable dataTable = CriaDataTable(sqlStr, connectionString);

    using (MemoryStream memoryStream = new MemoryStream())
    {
        // 写入DataTable的架构+数据，确保类型信息完整
        dataTable.WriteXml(memoryStream, XmlWriteMode.WriteSchema);
        return memoryStream.ToArray();
    }
}

修改后的客户端VB.NET代码 #

Public Function ApiGetDataTable(sqlStr As String, connectionString As String) As DataTable
    Dim url As String = $"{_baseUrl}GetDataTable?sqlStr={Uri.EscapeDataString(sqlStr)}&connectionString={Uri.EscapeDataString(connectionString)}"

    Dim request As New HttpRequestMessage(HttpMethod.Get, url)
    request.Headers.Add("accept", "application/octet-stream")

    Dim response = _httpClient.SendAsync(request).Result
    response.EnsureSuccessStatusCode()

    Dim content As Byte() = response.Content.ReadAsByteArrayAsync().Result

    Dim dataTable As New DataTable()
    Using memoryStream As New MemoryStream(content)
        // 先读取架构再加载数据，完整还原原DataTable结构
        dataTable.ReadXml(memoryStream, XmlReadMode.ReadSchema)
        Return dataTable
    End Using

    Return Nothing
End Function

关键说明 #

• XmlWriteMode.WriteSchema：序列化时同时写入DataTable的架构信息（列名、数据类型、约束等），保证反序列化时能精准还原结构。
• XmlReadMode.ReadSchema：读取时优先解析架构信息，再加载数据，确保列类型、结构和原DataTable完全匹配。
• 该方式完全规避了XmlSerializer的跨框架兼容性问题，因为WriteXml/ReadXml是DataTable的原生标准实现，在.NET Framework和.NET Core中行为统一。

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