A blazingly fast JSON serializing & deserializing library, accelerated by JIT (just-in-time compiling) and SIMD (single-instruction-multiple-data).
## Requirement
- Go 1.15~1.20
- Linux/MacOS/Windows
- Amd64 ARCH
## Features
- Runtime object binding without code generation
- Complete APIs for JSON value manipulation
- Fast, fast, fast!
## Benchmarks
For **all sizes** of json and **all scenarios** of usage, **Sonic performs best**.
Default behaviors are mostly consistent with `encoding/json`, except HTML escaping form (see [Escape HTML](https://github.com/bytedance/sonic/blob/main/README.md#escape-html)) and `SortKeys` feature (optional support see [Sort Keys](https://github.com/bytedance/sonic/blob/main/README.md#sort-keys)) that is **NOT** in conformity to [RFC8259](https://datatracker.ietf.org/doc/html/rfc8259).
```go
import "github.com/bytedance/sonic"
var data YourSchema
// Marshal
output, err := sonic.Marshal(&data)
// Unmarshal
err := sonic.Unmarshal(output, &data)
```
### Streaming IO
Sonic supports decoding json from `io.Reader` or encoding objects into `io.`Writer`, aims at handling multiple values as well as reducing memory consumption.
jm := root.InterfaceUseNumber().(json.Number) // jn == jm
// Get float64
fn := root.Float64()
fm := root.Interface().(float64) // jn == jm
```
### Sort Keys
On account of the performance loss from sorting (roughly 10%), sonic doesn't enable this feature by default. If your component depends on it to work (like [zstd](https://github.com/facebook/zstd)), Use it like this:
```go
import "github.com/bytedance/sonic"
import "github.com/bytedance/sonic/encoder"
// Binding map only
m := map[string]interface{}{}
v, err := encoder.Encode(m, encoder.SortMapKeys)
// Or ast.Node.SortKeys() before marshal
var root := sonic.Get(JSON)
err := root.SortKeys()
```
### Escape HTML
On account of the performance loss (roughly 15%), sonic doesn't enable this feature by default. You can use `encoder.EscapeHTML` option to open this feature (align with `encoding/json.HTMLEscape`).
```go
import "github.com/bytedance/sonic"
v := map[string]string{"&&":"<>"}
ret, err := Encode(v, EscapeHTML) // ret == `{"\u0026\u0026":{"X":"\u003c\u003e"}}`
```
### Compact Format
Sonic encodes primitive objects (struct/map...) as compact-format JSON by default, except marshaling `json.RawMessage` or `json.Marshaler`: sonic ensures validating their output JSON but **DONOT** compacting them for performance concerns. We provide the option `encoder.CompactMarshaler` to add compacting process.
### Print Error
If there invalid syntax in input JSON, sonic will return `decoder.SyntaxError`, which supports pretty-printing of error position
```go
import "github.com/bytedance/sonic"
import "github.com/bytedance/sonic/decoder"
var data interface{}
err := sonic.UnmarshalString("[[[}]]", &data)
if err != nil {
/* One line by default */
println(e.Error()) // "Syntax error at index 3: invalid char\n\n\t[[[}]]\n\t...^..\n"
/* Pretty print */
if e, ok := err.(decoder.SyntaxError); ok {
/*Syntax error at index 3: invalid char
[[[}]]
...^..
*/
print(e.Description())
} else if me, ok := err.(*decoder.MismatchTypeError); ok {
// decoder.MismatchTypeError is new to Sonic v1.6.0
print(me.Description())
}
}
```
#### Mismatched Types [Sonic v1.6.0]
If there a **mismatch-typed** value for a given key, sonic will report `decoder.MismatchTypeError` (if there are many, report the last one), but still skip wrong the value and keep decoding next JSON.
```go
import "github.com/bytedance/sonic"
import "github.com/bytedance/sonic/decoder"
var data = struct{
A int
B int
}{}
err := UnmarshalString(`{"A":"1","B":1}`, &data)
println(err.Error()) // Mismatch type int with value string "at index 5: mismatched type with value\n\n\t{\"A\":\"1\",\"B\":1}\n\t.....^.........\n"
fmt.Printf("%+v", data) // {A:0 B:1}
```
### Ast.Node
Sonic/ast.Node is a completely self-contained AST for JSON. It implements serialization and deserialization both and provides robust APIs for obtaining and modification of generic data.
#### Get/Index
Search partial JSON by given paths, which must be non-negative integer or string, or nil
**Tip**: since `Index()` uses offset to locate data, which is much faster than scanning like `Get()`, we suggest you use it as much as possible. And sonic also provides another API `IndexOrGet()` to underlying use offset as well as ensure the key is matched.
Sonic **DOES NOT** ensure to support all environments, due to the difficulty of developing high-performance codes. For developers who use sonic to build their applications in different environments, we have the following suggestions:
- Developing on **Mac M1**: Make sure you have Rosetta 2 installed on your machine, and set `GOARCH=amd64` when building your application. Rosetta 2 can automatically translate x86 binaries to arm64 binaries and run x86 applications on Mac M1.
- Developing on **Linux arm64**: You can install qemu and use the `qemu-x86_64 -cpu max` command to convert x86 binaries to amr64 binaries for applications built with sonic. The qemu can achieve a similar transfer effect to Rosetta 2 on Mac M1.
For developers who want to use sonic on Linux arm64 without qemu, or those who want to handle JSON strictly consistent with `encoding/json`, we provide some compatible APIs as `sonic.API`
-`ConfigDefault`: the sonic's default config (`EscapeHTML=false`,`SortKeys=false`...) to run on sonic-supporting environment. It will fall back to `encoding/json` with the corresponding config, and some options like `SortKeys=false` will be invalid.
-`ConfigStd`: the std-compatible config (`EscapeHTML=true`,`SortKeys=true`...) to run on sonic-supporting environment. It will fall back to `encoding/json`.
-`ConfigFastest`: the fastest config (`NoQuoteTextMarshaler=true`) to run on sonic-supporting environment. It will fall back to `encoding/json` with the corresponding config, and some options will be invalid.
## Tips
### Pretouch
Since Sonic uses [golang-asm](https://github.com/twitchyliquid64/golang-asm) as a JIT assembler, which is NOT very suitable for runtime compiling, first-hit running of a huge schema may cause request-timeout or even process-OOM. For better stability, we advise **using `Pretouch()` for huge-schema or compact-memory applications** before `Marshal()/Unmarshal()`.
```go
import (
"reflect"
"github.com/bytedance/sonic"
"github.com/bytedance/sonic/option"
)
func init() {
var v HugeStruct
// For most large types (nesting depth <= option.DefaultMaxInlineDepth)
err := sonic.Pretouch(reflect.TypeOf(v))
// with more CompileOption...
err := sonic.Pretouch(reflect.TypeOf(v),
// If the type is too deep nesting (nesting depth > option.DefaultMaxInlineDepth),
// you can set compile recursive loops in Pretouch for better stability in JIT.
option.WithCompileRecursiveDepth(loop),
// For a large nested struct, try to set a smaller depth to reduce compiling time.
option.WithCompileMaxInlineDepth(depth),
)
}
```
### Copy string
When decoding **string values without any escaped characters**, sonic references them from the origin JSON buffer instead of mallocing a new buffer to copy. This helps a lot for CPU performance but may leave the whole JSON buffer in memory as long as the decoded objects are being used. In practice, we found the extra memory introduced by referring JSON buffer is usually 20% ~ 80% of decoded objects. Once an application holds these objects for a long time (for example, cache the decoded objects for reusing), its in-use memory on the server may go up. We provide the option `decoder.CopyString()` for users to choose not to reference the JSON buffer, which may cause a decline in CPU performance to some degree.
### Pass string or []byte?
For alignment to `encoding/json`, we provide API to pass `[]byte` as an argument, but the string-to-bytes copy is conducted at the same time considering safety, which may lose performance when the origin JSON is huge. Therefore, you can use `UnmarshalString()` and `GetFromString()` to pass a string, as long as your origin data is a string or **nocopy-cast** is safe for your []byte. We also provide API `MarshalString()` for convenient **nocopy-cast** of encoded JSON []byte, which is safe since sonic's output bytes is always duplicated and unique.
### Accelerate `encoding.TextMarshaler`
To ensure data security, sonic.Encoder quotes and escapes string values from `encoding.TextMarshaler` interfaces by default, which may degrade performance much if most of your data is in form of them. We provide `encoder.NoQuoteTextMarshaler` to skip these operations, which means you **MUST** ensure their output string escaped and quoted following [RFC8259](https://datatracker.ietf.org/doc/html/rfc8259).
### Better performance for generic data
In **fully-parsed** scenario, `Unmarshal()` performs better than `Get()`+`Node.Interface()`. But if you only have a part of the schema for specific json, you can combine `Get()` and `Unmarshal()` together:
Even if you don't have any schema, use `ast.Node` as the container of generic values instead of `map` or `interface`:
```go
import "github.com/bytedance/sonic"
root, err := sonic.GetFromString(_TwitterJson)
user := root.GetByPath("statuses", 3, "user") // === root.Get("status").Index(3).Get("user")
err = user.Check()
// err = user.LoadAll() // only call this when you want to use 'user' concurrently...
go someFunc(user)
```
Why? Because `ast.Node` stores its children using `array`:
-`Array`'s performance is **much better** than `Map` when Inserting (Deserialize) and Scanning (Serialize) data;
- **Hashing** (`map[x]`) is not as efficient as **Indexing** (`array[x]`), which `ast.Node` can conduct on **both array and object**;
- Using `Interface()`/`Map()` means Sonic must parse all the underlying values, while `ast.Node` can parse them **on demand**.
**CAUTION:** `ast.Node`**DOESN'T** ensure concurrent security directly, due to its **lazy-load** design. However, you can call `Node.Load()`/`Node.LoadAll()` to achieve that, which may bring performance reduction while it still works faster than converting to `map` or `interface{}`
## Community
Sonic is a subproject of [CloudWeGo](https://www.cloudwego.io/). We are committed to building a cloud native ecosystem.
