#byte #avro #serde

serde-avro-bytes

Efficiently store Rust idiomatic bytes related types in Avro encoding

2 unstable releases

0.2.0 Mar 28, 2024
0.1.0 Mar 27, 2024

#504 in Encoding

Download history 184/week @ 2024-03-22 103/week @ 2024-03-29 37/week @ 2024-04-05

324 downloads per month

MIT/Apache

29KB
747 lines

Serde Avro Bytes

Avro is a binary encoding format which provides a "bytes" type optimized to store &[u8] data like.

Unfortunately the apache_avro encodes Vec<u8> as an array of integers thus the encoded data are twice bigger than using the bytes.

#[derive(Serialize)]
struct Record {
    data: Vec<u8>
}

fn playground() {
    let record = Record {
        data: vec![1,2]
    };
    // data field
    // => encoded as in int array : [4,1,1,1,2,0]
    // => encoded as bytes : [4,1,2]
}   

This crate provided a set of module to handle idiomatic Rust types and encode its component as "bytes".

#[derive(Serialize, Deserialize)]
struct Record {
    #[serde(with = "serde_avro_bytes::bytes")]
    key: Vec<u8>,
    #[serde(with = "serde_avro_bytes::bytes::option")]
    key2: Option<Vec<u8>>,
    #[serde(with = "serde_avro_bytes::hashmap")]
    key3: HashMap<Vec<u8>, Vec<u8>>,
    #[serde(with = "serde_avro_bytes::btreemap::option")]
    key4: Option<BTreeMap<Vec<u8>, Vec<u8>>>,
    #[serde(with = "serde_avro_bytes::list")]
    key5: Vec<Vec<u8>>,
    #[serde(with = "serde_avro_bytes::list::option")]
    key6: Option<Vec<Vec<u8>>>,
}

Features

  • bstr: adds support for working with BStrings which are convenient wrappers for partially valid UTF-8 bytes sequences provided by the bst crate. See examples/bstr.rs.

Dependencies

~6MB
~113K SLoC