Json Schema简介和Json Schema的高性能.net实现库 LateApexEarlySpeed.Json.Schema
什么是Json Schema ?
Json schema是一种声明式语言,它可以用来标识Json的结构,数据类型和数据的具体限制,它提供了描述期望Json结构的标准化方法。
利用Json Schema, 你可以定义Json结构的各种规则,以便确定Json数据在各个子系统中交互传输时保持兼容和一致的格式。
一般来说,系统可以自己实现逻辑来判断当前json是否满足接口要求,比如是否某个字段存在,是否属性值是有效的。但当验证需求变得复杂后,比如有大量嵌套json结构,属性之间的复杂关联限制等等,则容易编写出考虑不全的验证代码。另外,当系统需要动态的json数据要求,比如先由用户自己决定他需要的json结构,然后系统根据用户表达的定制化json结构需求,帮助用户验证后续的json数据。这种系统代码编译时无法确定的json结构,就需要另一种解决方案。
Json Schema就是针对这种问题的比较自然的解决方案。它可以让你或你的用户描述希望的json结构和值的内容限制,有效属性,是否是required, 还有有效值的定义,等等。。利用Json Schema, 人们可以更好的理解Json结构,而且程序也可以根据你的Json Schema验证Json数据。
Json Schema语法的学习见官方介绍。
比如下面的一个简单例子,用.net下的Json Schema实现库library LateApexEarlySpeed.Json.Schema进行Json数据的验证:
Json Schema (文件:schema.json):
{
"type": "object",
"properties": {
"propBoolean": {
"type": "boolean"
},
"propArray": {
"type": "array",
"uniqueItems": true
}
}
}
Json 数据 (文件:instance.json):
{
"propBoolean": true,
"propArray": [ 1, 2, 3, 4, 4 ]
}
C# 代码:
string jsonSchema = File.ReadAllText("schema.json");
string instance = File.ReadAllText("instance.json");
var jsonValidator = new JsonValidator(jsonSchema);
ValidationResult validationResult = jsonValidator.Validate(instance);
if (validationResult.IsValid)
{
Console.WriteLine("good");
}
else
{
Console.WriteLine($"Failed keyword: {validationResult.Keyword}");
Console.WriteLine($"ResultCode: {validationResult.ResultCode}");
Console.WriteLine($"Error message: {validationResult.ErrorMessage}");
Console.WriteLine($"Failed instance location: {validationResult.InstanceLocation}");
Console.WriteLine($"Failed relative keyword location: {validationResult.RelativeKeywordLocation}");
}
输出:
Failed keyword: uniqueItems
ResultCode: DuplicatedArrayItems
Error message: There are duplicated array items
Failed instance location: /propArray
Failed relative keyword location: /properties/propArray/uniqueItems
LateApexEarlySpeed.Json.Schema中文介绍
项目原始文档:https://github.com/lateapexearlyspeed/Lateapexearlyspeed.JsonSchema
中文文档:
LateApexEarlySpeed.Json.Schema是2023年12月发布的一个新的.net下的Json Schema实现库library,基于截止到2023年12月为止最新版的Json schema - draft 2020.12。
Json Schema验证功能经过了official json schema test-suite for draft 2020.12的测试。(部分排除的用例见下面的已知限制章节)
主要特点:
- 基于微软.net下默认的System.Text.Json而非经典的Newtonsoft.Json
- 高性能:和已有的知名且杰出的.net下的一些JsonSchema library相比,具有很好的性能 (在common case下,利用BenchmarkDotnet进行的性能测试)。用户请根据自己的使用场景进行性能验证
一些性能测试结果 (下面的测试比较都是在同样的使用方式下进行,见 性能建议 ):
12th Gen Intel Core i7-12800H, 1 CPU, 20 logical and 14 physical cores
[Host] : .NET 6.0.25 (6.0.2523.51912), X64 RyuJIT AVX2
DefaultJob : .NET 6.0.25 (6.0.2523.51912), X64 RyuJIT AVX2
有效数据用例:
Method | Mean | Error | StdDev | Gen0 | Gen1 | Allocated |
---|---|---|---|---|---|---|
ValidateByPopularSTJBasedValidator | 29.80 us | 0.584 us | 0.573 us | 4.4556 | 0.2441 | 55.1 KB |
ValidateByThisValidator | 15.99 us | 0.305 us | 0.300 us | 1.9531 | - | 24.2 KB |
无效数据用例:
Method | Mean | Error | StdDev | Median | Gen0 | Gen1 | Allocated |
---|---|---|---|---|---|---|---|
ValidateByPopularSTJBasedValidator | 65.04 us | 2.530 us | 7.341 us | 66.87 us | 4.5776 | 0.1221 | 56.42 KB |
ValidateByThisValidator | 15.47 us | 1.160 us | 3.421 us | 17.14 us | 1.4954 | - | 18.45 KB |
Note: "STJ"是"System.Text.Json"的缩写,这个是.net sdk里默认带的json基础包, LateApexEarlySpeed.Json.Schema这个库也是基于这个STJ编写的。
基准Schema:
{
"$id": "http://main",
"type": "object",
"additionalProperties": false,
"patternProperties": {
"propB*lean": {
"type": "boolean"
}
},
"dependentRequired": {
"propNull": [ "propBoolean", "propArray" ]
},
"dependentSchemas": {
"propNull": {
"type": "object"
}
},
"propertyNames": true,
"required": [ "propNull", "propBoolean" ],
"maxProperties": 100,
"minProperties": 0,
"properties": {
"propNull": {
"type": "null"
},
"propBoolean": {
"type": "boolean",
"allOf": [
true,
{ "type": "boolean" }
]
},
"propArray": {
"type": "array",
"anyOf": [ false, true ],
"contains": { "type": "integer" },
"maxContains": 100,
"minContains": 2,
"maxItems": 100,
"minItems": 1,
"prefixItems": [
{ "type": "integer" }
],
"items": { "type": "integer" },
"uniqueItems": true
},
"propNumber": {
"type": "number",
"if": {
"const": 1.5
},
"then": true,
"else": true,
"enum": [ 1.5, 0, 1 ]
},
"propString": {
"type": "string",
"maxLength": 100,
"minLength": 0,
"not": false,
"pattern": "abcde"
},
"propInteger": {
"$ref": "#/$defs/typeIsInteger",
"exclusiveMaximum": 100,
"exclusiveMinimum": 0,
"maximum": 100,
"minimum": 0,
"multipleOf": 0.5,
"oneOf": [ true, false ]
}
},
"$defs": {
"typeIsInteger": { "$ref": "http://inside#/$defs/typeIsInteger" },
"toTestAnchor": {
"$anchor": "test-anchor"
},
"toTestAnotherResourceRef": {
"$id": "http://inside",
"$defs": {
"typeIsInteger": { "type": "integer" }
}
}
}
}
有效基准数据:
{
"propNull": null,
"propBoolean": true,
"propArray": [ 1, 2, 3, 4, 5 ],
"propNumber": 1.5,
"propString": "abcde",
"propInteger": 1
}
无效基准数据:
{
"propNull": null,
"propBoolean": true,
"propArray": [ 1, 2, 3, 4, 4 ], // Two '4', duplicated
"propNumber": 1.5,
"propString": "abcde",
"propInteger": 1
}
基础用法
安装Nuget package
Install-Package LateApexEarlySpeed.Json.Schema
string jsonSchema = File.ReadAllText("schema.json");
string instance = File.ReadAllText("instance.json");
var jsonValidator = new JsonValidator(jsonSchema);
ValidationResult validationResult = jsonValidator.Validate(instance);
if (validationResult.IsValid)
{
Console.WriteLine("good");
}
else
{
Console.WriteLine($"Failed keyword: {validationResult.Keyword}");
Console.WriteLine($"ResultCode: {validationResult.ResultCode}");
Console.WriteLine($"Error message: {validationResult.ErrorMessage}");
Console.WriteLine($"Failed instance location: {validationResult.InstanceLocation}");
Console.WriteLine($"Failed relative keyword location: {validationResult.RelativeKeywordLocation}");
Console.WriteLine($"Failed schema resource base uri: {validationResult.SchemaResourceBaseUri}");
}
输出信息
当json数据验证失败后,可以查看错误数据的具体信息:
-
IsValid: As summary indicator for passed validation or failed validation.
-
ResultCode: The specific error type when validation failed.
-
ErrorMessage: the specific wording for human readable message
-
Keyword: current keyword when validation failed
-
InstanceLocation: The location of the JSON value within the instance being validated. The value is a JSON Pointer.
-
RelativeKeywordLocation: The relative location of the validating keyword that follows the validation path. The value is a JSON Pointer, and it includes any by-reference applicators such as "$ref" or "$dynamicRef". Eg:
/properties/width/$ref/minimum
-
SubSchemaRefFullUri: The absolute, dereferenced location of the validating keyword when validation failed. The value is a full URI using the canonical URI of the relevant schema resource with a JSON Pointer fragment, and it doesn't include by-reference applicators such as "$ref" or "$dynamicRef" as non-terminal path components. Eg:
https://example.com/schemas/common#/$defs/count/minimum
-
SchemaResourceBaseUri: The absolute base URI of referenced json schema resource when validation failed. Eg:
https://example.com/schemas/common
性能建议
尽可能的重用已实例化的JsonValidator实例(JsonValidator可以简单理解为代表一个json schema验证文档)来验证json数据,以便获得更高性能
外部json schema依赖的支持
除了自动支持当前schema文档内的引用关系,还支持外部json schema依赖:
- 本地schema依赖文本
var jsonValidator = new JsonValidator(jsonSchema);
string externalJsonSchema = File.ReadAllText("schema2.json");
jsonValidator.AddExternalDocument(externalJsonSchema);
ValidationResult validationResult = jsonValidator.Validate(instance);
- 远程schema url (实现库将访问网络来获得远程的schema)
var jsonValidator = new JsonValidator(jsonSchema);
await jsonValidator.AddHttpDocumentAsync(new Uri("http://this-is-json-schema-document"));
ValidationResult validationResult = jsonValidator.Validate(instance);
自定义keyword的支持
除了json schema specification中的标准keywords之外,还支持用户创建自定义keyword来实现额外的验证需求:
{
"type": "object",
"properties": {
"prop1": {
"customKeyword": "Expected value"
}
}
}
ValidationKeywordRegistry.AddKeyword<CustomKeyword>();
[Keyword("customKeyword")] // It is your custom keyword name
[JsonConverter(typeof(CustomKeywordJsonConverter))] // Use 'CustomKeywordJsonConverter' to deserialize to 'CustomKeyword' instance out from json schema text
internal class CustomKeyword : KeywordBase
{
private readonly string _customValue; // Simple example value
public CustomKeyword(string customValue)
{
_customValue = customValue;
}
// Do your custom validation work here
protected override ValidationResult ValidateCore(JsonInstanceElement instance, JsonSchemaOptions options)
{
if (instance.ValueKind != JsonValueKind.String)
{
return ValidationResult.ValidResult;
}
return instance.GetString() == _customValue
? ValidationResult.ValidResult
: ValidationResult.CreateFailedResult(ResultCode.UnexpectedValue, "It is not my expected value.", options.ValidationPathStack, Name, instance.Location);
}
}
internal class CustomKeywordJsonConverter : JsonConverter<CustomKeyword>
{
// Library will input json value of your custom keyword: "customKeyword" to this method.
public override CustomKeyword? Read(ref Utf8JsonReader reader, Type typeToConvert, JsonSerializerOptions options)
{
// Briefly:
return new CustomKeyword(reader.GetString()!);
}
public override void Write(Utf8JsonWriter writer, CustomKeyword value, JsonSerializerOptions options)
{
throw new NotImplementedException();
}
}
Format支持
目前library支持如下format:
- uri
- uri-reference
- date
- time
- date-time
- uuid
- hostname
- ipv4
- ipv6
- json-pointer
- regex
如果需要自定义format验证,可以实现一个FormatValidator子类并注册:
[Format("custom_format")] // this is your custom format name in json schema
public class TestCustomFormatValidator : FormatValidator
{
public override bool Validate(string content)
{
// custom format validation logic here...
}
}
// register it globally
FormatRegistry.AddFormatType<TestCustomFormatValidator>();
Other extension usage doc is to be continued .
限制
- 目前library关注于验证,暂不支持annotation
- 因为暂不支持annotation, 所以不支持如下keywords: unevaluatedProperties, unevaluatedItems
- 目前不支持 content-encoded string
问题报告
欢迎把使用过程中遇到的问题和希望增加的功能发到github repo issue中