Question

我正在尝试合并变更集错误。

我有一个属于用户架构的机构架构。每个字段都需要一些字段，但错误响应如下：

{
    "errors": {
        "user": {
            "password_confirmation": [
                "The password confirmation does not match the password."
            ],
            "password": [
                "This field is required."
            ],
            "name": [
                "This field is required."
            ]
        },
        "institution": {
            "web_address": [
                "is required."
            ]
        },

    }
}

如何将这些错误对象组合成一个？

我的插入内容如下：

 user_changeset =
      User.normal_user_changeset(%User{}, %{
        :email => Map.get(attrs, "email"),
        :password => Map.get(attrs, "password"),
        :password_confirmation => Map.get(attrs, "password_confirmation"),
        :name => Map.get(attrs, "name"),
        :postcode => Map.get(attrs, "postcode"),
        :dob => Map.get(attrs, "dob")
      })

    institution =
      %Institution{}
      |> Institution.changeset(%{
        :web_address => Map.get(attrs, "web_address"),
        :person_responsible => Map.get(attrs, "person_responsible"),
        :annual_turnover => Map.get(attrs, "annual_turnover")
      })
      |> Ecto.Changeset.put_assoc(:user, user_changeset)
      |> Repo.insert()

我希望错误响应是：

 {
        "errors": {
                "password_confirmation": [
                    "The password confirmation does not match the password."
                ],
                "password": [
                    "This field is required."
                ],
                "name": [
                    "This field is required."
                ]
                "web_address": [
                    "is required."
                ]
        }
    }

我的后备控制器（默认为此处）中具有此功能：

  def call(conn, {:error, %Ecto.Changeset{} = changeset}) do
    conn
    |> put_status(:unprocessable_entity)
    |> render(SfiWeb.ChangesetView, "error.json", changeset: changeset)
  end

Answer 1

您可以获取errors字段的值，并使用Enum.reduce/3合并它们：

map = Jason.decode! """
{
    "errors": {
        "user": {
            "password_confirmation": [
                "The password confirmation does not match the password."
            ],
            "password": [
                "This field is required."
            ],
            "name": [
                "This field is required."
            ]
        },
        "institution": {
            "web_address": [
                "is required."
            ]
        }
    }
}
"""

Map.update!(map, "errors", fn errors ->
  errors |> Map.values() |> Enum.reduce(%{}, &Map.merge/2)
end)
|> IO.inspect

输出：

%{
  "errors" => %{
    "name" => ["This field is required."],
    "password" => ["This field is required."],
    "password_confirmation" => ["The password confirmation does not match the password."],
    "web_address" => ["is required."]
  }
}

Answer 2

errors只是%Ecto.Changeset{}结构中的值之一。

也就是说，总是可以修改%Ecto.Changeset{}以提供自定义errors：

def fix_errors(%Ecto.Changeset{errors: errors} = ch) do
  %Ecto.Changeset{ch | errors: errors ++ [:my_custom_error]}
end

您需要的只是使用上面的函数重新映射您的输入，并将结果嵌入到变更集链中：

institution =
  %Institution{}
  |> ...
  |> Ecto.Changeset.put_assoc(:user, user_changeset)
  |> fix_errors() # see above 
  |> Repo.insert()

在此之前和之后，无论您是否为errors提供了有效的Elixir术语，我都可以展示如何自行执行转换。例如：

errors = %{
  institution: %{web_address: ["is required"]},
  user: %{
    name: ["is required."],
    password: ["is required."],
    password_confirmation: ["no match"]
  }
}

可能会展平为：

input
|> Enum.map(fn {_, v} -> v end)
|> Enum.reduce(&Map.merge/2)
#⇒ %{
#   name: ["is required."],
#   password: ["is required."],
#   password_confirmation: ["no match"],
#   web_address: ["is required"]
# }

如何合并变更集错误？

2 个答案: