当遇到多个错误时

在尝试将值反序列化到你的类型时，Deserr允许你使用其 MergeWithError 特性累积多个错误。这是一种减少用户修复无效负载所需交互次数的好方法，从而提高用户体验。

deserr的主要部分包括

1. Deserr<E> 是主要的反序列化特质，与 Serde 不同，它非常容易手动反序列化此特质，请参阅我们的示例目录中的 implements_deserr_manually.rs 文件。
2. IntoValue 和 Value 描述了解析的序列化数据必须具有的形状
3. DeserializeError 是所有反序列化错误都必须遵守的特质
4. MergeWithError<E> 描述了如何将多个错误组合在一起。它允许 Deserr 一次返回多个反序列化错误。
5. ValuePointerRef 和 ValuePointer 指向值内部的特定位置。它们用于定位错误的起源。
6. deserialize<Ret, Val, E> 是用于反序列化值的主体函数。
   • Ret 是返回值或您想要反序列化的结构。
   • Val 是您想要从其反序列化的值类型。目前，此 crate 中只提供了一个 serde_json::Value 的实现，但您可以添加自己的！请随时查看我们的 serde_json 模块。
   • E 是在反序列化过程中发生错误时应使用的错误类型。
7. Deserr derive 程序宏

示例

使用自定义错误实现自定义类型的反序列化

在下面的示例中，我们将反序列化包含许多字段的结构，并使用一个自定义错误类型，该类型会累积在反序列化结构时遇到的全部错误。

use deserr::{deserialize, DeserializeError, Deserr, ErrorKind, errors::JsonError, Value, ValueKind, IntoValue, take_cf_content, MergeWithError, ValuePointerRef, ValuePointer};
use serde_json::json;
use std::str::FromStr;
use std::ops::ControlFlow;
use std::fmt;
use std::convert::Infallible;

/// This is our custom error type. It'll accumulate multiple `JsonError`.
#[derive(Debug)]
struct MyError(Vec<JsonError>);

impl DeserializeError for MyError {
    /// Create a new error with the custom message.
    ///
    /// Return `ControlFlow::Continue` to continue deserializing even though an error was encountered.
    /// We could return `ControlFlow::Break` as well to stop right here.
    fn error<V: IntoValue>(self_: Option<Self>, error: ErrorKind<V>, location: ValuePointerRef) -> ControlFlow<Self, Self> {
        /// The `take_cf_content` return the inner error in a `ControlFlow<E, E>`.
        let error = take_cf_content(JsonError::error(None, error, location));

        let errors = if let Some(MyError(mut errors)) = self_ {
            errors.push(error);
            errors
        } else {
            vec![error]
        };
        ControlFlow::Continue(MyError(errors))
    }
}

/// We have to implements `MergeWithError` between our error type _aaand_ our error type.
impl MergeWithError<MyError> for MyError {
    fn merge(self_: Option<Self>, mut other: MyError, _merge_location: ValuePointerRef) -> ControlFlow<Self, Self> {
        if let Some(MyError(mut errors)) = self_ {
                other.0.append(&mut errors);
        }
        ControlFlow::Continue(other)
    }
}

#[derive(Debug, Deserr, PartialEq, Eq)]
#[deserr(deny_unknown_fields)]
struct Search {
    #[deserr(default = String::new())]
    query: String,
    #[deserr(try_from(&String) = FromStr::from_str -> IndexUidError)]
    index: IndexUid,
    #[deserr(from(String) = From::from)]
    field: Wildcard,
    #[deserr(default)]
    filter: Option<serde_json::Value>,
    // Even though this field is an `Option` it IS mandatory.
    limit: Option<usize>,
    #[deserr(default)]
    offset: usize,
}

/// An `IndexUid` can only be composed of ascii characters.
#[derive(Debug, PartialEq, Eq)]
struct IndexUid(String);
/// If we encounter a non-ascii character this is the error type we're going to throw.
struct IndexUidError(char);

impl FromStr for IndexUid {
    type Err = IndexUidError;

    fn from_str(s: &str) -> Result<Self, Self::Err> {
        if let Some(c) = s.chars().find(|c| !c.is_ascii()) {
            Err(IndexUidError(c))
        } else {
            Ok(Self(s.to_string()))
        }
    }
}

impl fmt::Display for IndexUidError {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(
            f,
            "Encountered invalid character: `{}`, only ascii characters are accepted in the index",
            self.0
        )
    }
}

/// We need to define how the `IndexUidError` error is going to be merged with our
/// custom error type.
impl MergeWithError<IndexUidError> for MyError {
    fn merge(self_: Option<Self>, other: IndexUidError, merge_location: ValuePointerRef) -> ControlFlow<Self, Self> {
            // To be consistent with the other error and automatically get the position of the error we re-use the `JsonError`
            // type and simply define ourself as an `Unexpected` error.
        let error = take_cf_content(JsonError::error::<Infallible>(None, ErrorKind::Unexpected { msg: other.to_string() }, merge_location));
        let errors = if let Some(MyError(mut errors)) = self_ {
            errors.push(error);
            errors
        } else {
            vec![error]
        };
        ControlFlow::Continue(MyError(errors))
    }
}

/// A `Wildcard` can either contains a normal value or be a unit wildcard.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(from(String) = From::from)]
enum Wildcard {
    Wildcard,
    Value(String),
}

impl From<String> for Wildcard {
    fn from(s: String) -> Self {
        if s == "*" {
            Wildcard::Wildcard
        } else {
            Wildcard::Value(s)
        }
    }
}

// Here is an example of a typical payload we could deserialize:
let data = deserialize::<Search, _, MyError>(
    json!({ "index": "mieli", "field": "doggo", "filter": ["id = 1", ["catto = jorts"]], "limit": null }),
).unwrap();
assert_eq!(data, Search {
    query: String::new(),
    index: IndexUid(String::from("mieli")),
    field: Wildcard::Value(String::from("doggo")),
    filter: Some(json!(["id = 1", ["catto = jorts"]])),
    limit: None,
    offset: 0,
});

// And here is what happens when everything goes wrong at the same time:
let error = deserialize::<Search, _, MyError>(
    json!({ "query": 12, "index": "mieli 🍯", "field": true, "offset": "🔢"  }),
).unwrap_err();
// We're going to stringify all the error so it's easier to read
assert_eq!(error.0.into_iter().map(|error| error.to_string()).collect::<Vec<String>>().join("\n"),
"\
Invalid value type at `.query`: expected a string, but found a positive integer: `12`
Invalid value type at `.offset`: expected a positive integer, but found a string: `\"🔢\"`
Invalid value at `.index`: Encountered invalid character: `🍯`, only ascii characters are accepted in the index
Invalid value type at `.field`: expected a string, but found a boolean: `true`
Missing field `limit`\
");

支持的功能

rename_all

根据给定的案例约定重命名结构体中的所有字段。可能的值有 lowercase、camelCase。如果您需要更多的案例约定，请提出问题；添加更多是微不足道的。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(rename_all = camelCase)]
struct Search {
    query: String,
    attributes_to_retrieve: Vec<String>,
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "attributesToRetrieve": ["age", "name"] }),
)
.unwrap();
assert_eq!(data, Search {
    query: String::from("doggo"),
    attributes_to_retrieve: vec![String::from("age"), String::from("name")],
});

deny_unknown_fields

在遇到未知字段时抛出错误。当此属性不存在时，默认情况下会忽略未知字段。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug)]
#[deserr(deny_unknown_fields)]
struct Search {
    query: String,
}

let err = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "doggo": "bork" }),
)
.unwrap_err();

assert_eq!(err.to_string(), "Unknown field `doggo`: expected one of `query`");

还可以提供自定义函数来处理错误。

use deserr::{Deserr, deserialize, ErrorKind, DeserializeError, ValuePointerRef, take_cf_content, errors::JsonError};
use std::convert::Infallible;
use serde_json::json;

#[derive(Deserr, Debug)]
#[deserr(deny_unknown_fields = unknown_fields_search)]
struct Search {
    query: String,
}

fn unknown_fields_search<E: DeserializeError>(
    field: &str,
    accepted: &[&str],
    location: ValuePointerRef,
) -> E {
    // `E::error` returns a `ControlFlow<E, E>`, which returns the error and indicates
    // whether we should keep accumulating errors or not. However, here we simply
    // want to retrieve the error's value. This is what `take_cf_content` does.
    match field {
        "doggo" => take_cf_content(E::error::<Infallible>(
                None,
                ErrorKind::Unexpected {
                    msg: String::from("can I pet the doggo? uwu")
                },
                location,
            )),
        _ => take_cf_content(E::error::<Infallible>(
            None,
            deserr::ErrorKind::UnknownKey { key: field, accepted },
            location,
        )),
    }
}

let err = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "doggo": "bork" }),
)
.unwrap_err();

assert_eq!(err.to_string(), "Invalid value: can I pet the doggo? uwu");

let err = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "catto": "jorts" }),
)
.unwrap_err();

assert_eq!(err.to_string(), "Unknown field `catto`: expected one of `query`");

tag

对枚举进行外部标记。Deserr 目前不支持对枚举进行内部标记，这意味着如果您正在反序列化枚举，则始终需要使用此属性。对于完整的单元枚举，deserr 可以从字符串反序列化它们的值。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    query: Query,
}

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(tag = "type")]
enum Query {
    Single {
        search: String,
    },
    Multi {
        searches: Vec<String>,
    }
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": { "type": "Single", "search": "bork" } }),
)
.unwrap();
assert_eq!(data, Search {
    query: Query::Single {
        search: String::from("bork"),
    },
});

from

从函数而不是从 Value 反序列化类型。您需要提供以下信息；

1. 函数的输入类型（这里 &String）
2. 函数的路径（这里，我们简单地使用 std 的 FromStr 实现）

Deserr 将首先尝试使用其 Deserr<E> 实现来反序列化给定的类型。这意味着 from 的输入类型可能很复杂。然后 Deserr 将调用您的函数。

如果您的函数可能失败，请参阅 try_from。

它可以用作容器属性

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(from(String) = From::from)]
enum Wildcard {
    Wildcard,
    Value(String),
}

impl From<String> for Wildcard {
    fn from(s: String) -> Self {
        if s == "*" {
            Wildcard::Wildcard
        } else {
            Wildcard::Value(s)
        }
    }
}

let data = deserialize::<Wildcard, _, JsonError>(
    json!("doggo"),
)
.unwrap();
assert_eq!(data, Wildcard::Value(String::from("doggo")));

let data = deserialize::<Wildcard, _, JsonError>(
    json!("*"),
)
.unwrap();
assert_eq!(data, Wildcard::Wildcard);

或字段属性

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(from(String) = From::from)]
enum Wildcard {
    Wildcard,
    Value(String),
}

impl From<String> for Wildcard {
    fn from(s: String) -> Self {
        if s == "*" {
            Wildcard::Wildcard
        } else {
            Wildcard::Value(s)
        }
    }
}

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    query: String,
    #[deserr(from(String) = From::from)]
    field: Wildcard,
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "field": "catto" }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), field: Wildcard::Value(String::from("catto")) });

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "field": "*" }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), field: Wildcard::Wildcard });

try_from

尝试从函数而不是从 Value 反序列化类型。您需要提供以下信息；

1. 函数的输入类型（这里 &String）
2. 函数的路径（这里，我们简单地使用 std 的 FromStr 实现）
3. 此函数可能返回的错误类型（这里 Infallible）

deserr 首先会尝试使用其 Deserr<E> 实现来反序列化给定的类型。这意味着 try_from 的输入类型可能很复杂。然后 deserr 会调用您的函数，并将指定的错误累积到调用者的错误类型中。

它可以用作容器属性

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;
use std::str::FromStr;
use std::fmt;

// Notice how the `try_from` allows us to leverage the deserr limitation on tuple struct.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(try_from(&String) = FromStr::from_str -> AsciiStringError)]
struct AsciiString(String);

#[derive(Debug)]
struct AsciiStringError(char);

impl fmt::Display for AsciiStringError {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(
            f,
            "Encountered invalid character: `{}`, only ascii characters are accepted",
            self.0
        )
    }
}
impl std::error::Error for AsciiStringError {}

impl FromStr for AsciiString {
    type Err = AsciiStringError;

    fn from_str(s: &str) -> Result<Self, Self::Err> {
        if let Some(c) = s.chars().find(|c| !c.is_ascii()) {
            Err(AsciiStringError(c))
        } else {
            Ok(Self(s.to_string()))
        }
    }
}

let data = deserialize::<AsciiString, _, JsonError>(
    json!("doggo"),
)
.unwrap();
assert_eq!(data, AsciiString(String::from("doggo")));

let error = deserialize::<AsciiString, _, JsonError>(
    json!("👉👈"),
)
.unwrap_err();
assert_eq!(error.to_string(), "Invalid value: Encountered invalid character: `👉`, only ascii characters are accepted");

或字段属性

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;
use std::convert::Infallible;
use std::str::FromStr;
use std::num::ParseIntError;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    query: String,
    #[deserr(try_from(&String) = FromStr::from_str -> ParseIntError)]
    limit: usize,

}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "limit": "12" }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), limit: 12 });

let error = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "limit": 12 }),
)
.unwrap_err();
assert_eq!(error.to_string(), "Invalid value type at `.limit`: expected a string, but found a positive integer: `12`");

validate

在结构体反序列化后对其进行验证。这在您的验证逻辑需要考虑多个字段时非常有用。

use deserr::{Deserr, DeserializeError, ErrorKind, ValuePointerRef, deserialize, errors::JsonError};
use serde_json::json;
use std::convert::Infallible;

// `__Deserr_E` represents the Error returned by the generated `Deserr` implementation.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(validate = validate_range -> __Deserr_E)]
struct Range {
    min: u8,
    max: u8,
}

fn validate_range<E: DeserializeError>(
    range: Range,
    location: ValuePointerRef,
) -> Result<Range, E> {
    if range.min > range.max {
        Err(deserr::take_cf_content(E::error::<Infallible>(
            None,
            ErrorKind::Unexpected {
                msg: format!(
                    "`max` (`{}`) should be greater than `min` (`{}`)",
                    range.max, range.min
                ),
            },
            location,
        )))
    } else {
        Ok(range)
    }
}

let data = deserialize::<Range, _, JsonError>(
    json!({ "min": 2, "max": 4 }),
)
.unwrap();
assert_eq!(data, Range { min: 2, max: 4 });

let error = deserialize::<Range, _, JsonError>(
    json!({ "min": 4, "max": 2 }),
)
.unwrap_err();
assert_eq!(error.to_string(), "Invalid value: `max` (`2`) should be greater than `min` (`4`)");

default

允许您为字段指定默认值。

请注意，与 serde 不同，默认情况下，Option 不自动使用此属性。在这里，您需要显式定义您的类型是否可以获取默认值。这使得它更不容易出错，并更容易将可选字段强制为必填。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    #[deserr(default)]
    query: Option<String>,
    #[deserr(default = 20)]
    limit: usize,
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "limit": 4 }),
)
.unwrap();
assert_eq!(data, Search { query: Some(String::from("doggo")), limit: 4 });

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo" }),
)
.unwrap();
assert_eq!(data, Search { query: Some(String::from("doggo")), limit: 20 });

skip

允许您跳过字段的反序列化。它不会出现在由 deny_unknown_fields 生成的字段列表中，也不会出现在 ErrorKind 类型的 UnknownKey 变体中。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    query: String,
    // A field can be skipped if it implements `Default` or if the `default` attribute is specified.
    #[deserr(skip)]
    hidden: usize,
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo" }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), hidden: 0 });

// if you try to specify the field, it is ignored
let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "hidden": 2 }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), hidden: 0 });

// Here, we're going to see how skip interacts with `deny_unknown_fields`

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(deny_unknown_fields)]
struct Search2 {
    query: String,
    // A field can be skipped if it implements `Default`.
    #[deserr(skip)]
    hidden: usize,
}

let error = deserialize::<Search2, _, JsonError>(
    json!({ "query": "doggo", "hidden": 1 }),
)
.unwrap_err();
// NOTE: `hidden` isn't in the list of expected fields + `hidden` is effectively considered as a non-existing field.
assert_eq!(error.to_string(), "Unknown field `hidden`: expected one of `query`");

map

在反序列化后映射字段。

use deserr::{Deserr, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    query: String,
    #[deserr(map = add_one)]
    limit: usize,
}

fn add_one(n: usize) -> usize {
    n.saturating_add(1)
}

let data = deserialize::<Search, _, JsonError>(
    json!({ "query": "doggo", "limit": 0 }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), limit: 1 });

// Let's see how `map` interacts with the `default` attributes.
#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search2 {
    query: String,
    #[deserr(default, map = add_one)]
    limit: usize,
}

let data = deserialize::<Search2, _, JsonError>(
    json!({ "query": "doggo" }),
)
.unwrap();
// As we can see, the `map` attribute is applied AFTER the `default`.
assert_eq!(data, Search2 { query: String::from("doggo"), limit: 1 });

missing_field_error

如果您想为缺失的特定字段自定义错误信息，这将提供机会。

use deserr::{Deserr, DeserializeError, ValuePointerRef, ErrorKind, deserialize, errors::JsonError};
use serde_json::json;
use std::convert::Infallible;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search {
    #[deserr(missing_field_error = missing_query_field)]
    query: String,
    limit: usize,
}

fn missing_query_field<E: DeserializeError>(_field_name: &str, location: ValuePointerRef) -> E {
    deserr::take_cf_content(E::error::<Infallible>(
        None,
        ErrorKind::Unexpected {
            msg: String::from("I really need the query field, please give it to me uwu"),
        },
        location,
    ))
}

let error = deserialize::<Search, _, JsonError>(
    json!({ "limit": 0 }),
)
.unwrap_err();
assert_eq!(error.to_string(), "Invalid value: I really need the query field, please give it to me uwu");

error

自定义反序列化此结构时可以返回的错误类型，而不是保持其通用性。

use deserr::{Deserr, DeserializeError, ValuePointerRef, ErrorKind, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(error = JsonError)]
struct Search {
    query: String,
    limit: usize,
}

// As we can see, rust is able to infer the error type.
let data = deserialize::<Search, _, _>(
    json!({ "query": "doggo", "limit": 1 }),
)
.unwrap();
assert_eq!(data, Search { query: String::from("doggo"), limit: 1 });

它也可以用作字段属性；

use deserr::{Deserr, DeserializeError, ValuePointerRef, ErrorKind, deserialize, errors::JsonError};
use serde_json::json;

// Since the error returned by the `Search` structure needs to implements `MergeWithError<JsonError>`
// we also need to specify the `error` attribute as a `JsonError`. But as you will see later there are
// other solutions.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(error = JsonError)]
struct Search<A> {
    #[deserr(error = JsonError)]
    query: A,
    limit: usize,
}

where_predicate

允许您将 where 子句添加到 deserr 将生成的 Deserr 实现中。

use deserr::{Deserr, DeserializeError, MergeWithError, deserialize, errors::JsonError};
use serde_json::json;

// Here we can constraint the generic `__Deserr_E` type used by deserr to implements `MergeWithError`.
// Now instead of constraining the final error type it stays generic if it's able to accumulate with
// with a `JsonError`.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(where_predicate = __Deserr_E: MergeWithError<JsonError>, where_predicate = A: Deserr<JsonError>)]
struct Search<A> {
    #[deserr(error = JsonError)]
    query: A,
    limit: usize,
}

needs_predicate

自动为具有此属性的每个字段添加 where_predicate = FieldType: Deserr<ErrType>。

use deserr::{Deserr, DeserializeError, MergeWithError, deserialize, errors::JsonError};
use serde_json::json;

#[derive(Deserr, Debug, PartialEq, Eq)]
struct Search<A> {
    #[deserr(needs_predicate)]
    query: A,
    limit: usize,
}

它与以下内容严格等价

use deserr::{Deserr, DeserializeError, MergeWithError, deserialize, errors::JsonError};
use serde_json::json;

// `__Deserr_E` represents the Error returned by the generated `Deserr` implementation.
#[derive(Deserr, Debug, PartialEq, Eq)]
#[deserr(where_predicate = A: Deserr<__Deserr_E>)]
struct Search<A> {
    query: A,
    limit: usize,
}

与 serde 的比较

由于 deserr 需要先反序列化负载到一个通用的 Value，这会在创建您的结构之前分配大量内存，所以它比 serde 慢得多。

例如，在 Meilisearch 的搜索路由中，在有效负载的情况下，我们观察到 400% 的减速（慢 4 倍）。这使得我们的搜索请求的反序列化时间从 500ns 降低到 2µs。这对于大多数用例来说已经足够快，但如果大部分时间都花在反序列化上，可能会成为问题。

数据结构支持

容器属性

字段属性

功能标志

serde-json

导入 serde_json 并提供；

• 为 deserr::IntoValue 实现了 serde_json::Value，使得这两个crate可以方便地一起使用。
• 为 JsonError 类型提供了一个默认实现，提供了尽可能好的泛型错误信息。

serde-cs

导入 serde-cs 并提供；

• 为 Deserr 实现了 serde_cs::CS<R>。

actix-web

导入 actix-web 和 futures 并提供；

• 如果使用 serde-json 功能，则实现了一个 json actix-web 提取器。
• 如果使用 serde-json 功能，则实现了 ResponseError 对 JsonError 类型的实现。

常见问题解答

但是为什么？

在 Meilisearch，我们希望在失败反序列化特定字段时自定义返回的错误代码。一些错误信息也非常不明确，无法编辑。

维护方面如何？

在 Meilisearch，我们已经在生产中使用 deserr，因此它得到了良好的维护。

在哪里可以看到此crate的使用示例？

目前，您可以在本仓库的 examples 目录中阅读我们的示例。您还可以查看我们的集成测试；每个属性都有一个简单易读的测试。

当然，您还可以阅读 Meilisearch 的代码，其中在所有路由上使用了 deserr。

我的问题没有列出来

请，如果您认为这个库中存在错误或想添加新功能，请打开一个问题或讨论。如果您想更直接地与我们聊天，您可以在 discord（https://discord.com/invite/meilisearch）上加入我们。

许可证