Crate codee

source ·
Expand description

Easy and flexible way of encoding and decoding data into either strings or bytes.

This crate provides generic traits for Encoders and Decoders as well as several implementations for commonly used (de)serializer crates.

This makes it easily possible to abstract away the serialization and deserialization independent of the concrete crate used. You can write a function like this:

use codee::{CodecError, Decoder, Encoder};

fn store_value<T, Codec>(value: T) -> Result<(), CodecError<<Codec as Encoder<T>>::Error, <Codec as Decoder<T>>::Error>>
where
    Codec: Encoder<T, Encoded = String> + Decoder<T, Encoded = str>,
{
    let encoded = Codec::encode(&value).map_err(CodecError::Encode)?;
    let decoded = Codec::decode(&encoded).map_err(CodecError::Decode)?;

    Ok(())
}

// Then we can use it like this:

use codee::string::{JsonSerdeCodec, FromToStringCodec};

#[derive(serde::Serialize, serde::Deserialize)]
struct MyStruct {
    field: usize,
}

store_value::<i32, FromToStringCodec>(42);
store_value::<MyStruct, JsonSerdeCodec>(MyStruct { field: 42 });

§Available Codecs

There are two types of codecs: One that encodes as binary data (Vec[u8]) in the module binary and another type that encodes as strings (String) in the module string. There is also an adapter Base64 that can be used to wrap a binary codec and make it a string codec by representing the binary data as a base64 string.

Please note that many of the codecs need a feature flag to be enabled. Check the docs of the respective codec to be sure.

§String Codecs

Please have look at the module string.

§Adapters
  • string::Base64 — Wraps a binary codec and make it a string codec by representing the binary data as a base64 string.
  • string::OptionCodec — Wraps a string codec that encodes T to create a codec that encodes Option<T>.

§Binary Codecs

Please have look at the module binary.

§Custom Codecs

If you don’t find a suitable codecs for your needs, you can implement your own; it’s straightforward! If you want to create a string codec, you can look at [JsonSerdeCodec] as a starting point. In case it’s a binary codec, have a look at [BincodeSerdeCodec].

§Versioning

Versioning is the process of handling long-term data that can outlive our code.

For example, we could have a settings struct whose members change over time. We might eventually add timezone support, and we might then remove support for a thousands separator for numbers. Each change results in a new possible version of the stored data. If we stored these settings in browser storage, we would need to handle all possible versions of the data format that can occur. If we don’t offer versioning, then all settings could revert to the default every time we encounter an old format.

How best to handle versioning depends on the codec involved:

  • The FromToStringCodec can avoid versioning entirely by keeping to primitive types. In our example above, we could have decomposed the settings struct into separate timezone and number separator fields. These would be encoded as strings and stored as two separate key-value fields in the browser rather than a single field. If a field is missing, then the value intentionally would fall back to the default without interfering with the other field.

  • The ProstCodec uses Protocol buffers designed to solve the problem of long-term storage. It provides semantics for versioning that are not present in JSON or other formats.

  • The codecs that use serde under the hood can rely on serde or by providing their own manual version handling. See the next sections for more details.

§Rely on serde

A simple way to avoid complex versioning is to rely on serde’s field attributes such as serde(default) and serde(rename = "...").

§Manual Version Handling

We look at the example of the JsonSerdeCodec in this section.

To implement version handling, we parse the JSON generically then transform the resulting serde_json::Value before decoding it into our struct again.

Let’s look at an example.

use serde::{Deserialize, Serialize};
use serde_json::json;
use codee::{Encoder, Decoder};

#[derive(Serialize, Deserialize, Clone, Default, PartialEq)]
pub struct MyState {
    pub hello: String,
    // This field was added in a later version
    pub greeting: String,
}

pub struct MyStateCodec;

impl Encoder<MyState> for MyStateCodec {
    type Error = serde_json::Error;
    type Encoded = String;

    fn encode(val: &MyState) -> Result<Self::Encoded, Self::Error> {
        serde_json::to_string(val)
    }
}

impl Decoder<MyState> for MyStateCodec {
    type Error = serde_json::Error;
    type Encoded = str;

    fn decode(stored_value: &Self::Encoded) -> Result<MyState, Self::Error> {
        let mut val: serde_json::Value = serde_json::from_str(stored_value)?;
        // add "greeting": "Hello" to the object if it's missing
        if let Some(obj) = val.as_object_mut() {
            if !obj.contains_key("greeting") {
               obj.insert("greeting".to_string(), json!("Hello"));
            }
            serde_json::from_value(val)
        } else {
            Ok(MyState::default())
        }
    }
}

// Then use it just like any other codec.

§Hybrid Codecs

In case you want to write code that can be used with both, binary and string codecs, there are the HybridDecoder, HybridEncoder and IsBinary traits that are implemented automatically for all the codecs.

To see them in action, you can have a look at leptos_use::use_websocket.

Modules§

Structs§

  • A wrapper codec that relies on serde_lite. With this, you can wrap serde based codecs that also work with serde-lite like the [JsonSerdeCodec] or the [MsgpackSerdeCodec].

Enums§

Traits§