dop251/scsu

A Standard Compression Scheme for Unicode implementation in Go

SCSU

A Standard Compression Scheme for Unicode implementation in Go.

This in an implementation of SCSU as described in https://www.unicode.org/reports/tr6/tr6-4.html

Although UTF-8 is now the most commonly used and recommended encoding, in some cases the use of SCSU can be beneficial. For example when storing or transmitting short alphabetical texts (Arabic, Hebrew, Russian, etc.) where general-purpose compression algorithms are inefficient, but SCSU provides nearly 50% compression ratio over UTF-8.

The code is based on the sample Java implementation found at ftp://ftp.unicode.org/Public/PROGRAMS/SCSU/ however the encoding algorithm has been slightly modified as the implementation above contains a few bugs.

The code has been fuzz-tested using https://github.com/dvyukov/go-fuzz to ensure that random input neither crashes the Encoder nor the Decoder, and if it happens to be a valid UTF-8, an Encode/Decode cycle produces identical output.

Usage Scenarios.

Encode a string into a []byte:

b, err := scsu.Encode(s, nil) // the second argument can be an existing slice which will be appended

Decode a []byte into a string:

s, err := scsu.Decode(b)

Use a Writer:

writer := scsu.NewWriter(outWriter)
n, err := writer.WriteString(s)
n, err = writer.WriteRune(r)
n, err = writer.WriteRunes(runeSource)

Use an Encoder:

encoder := scsu.NewEncoder()
buf, err := encoder.Encode(runeSource, buf) // assuming buf has enough capacity this does zero allocs
// encoder then can be re-used

Use a Reader:

reader := scsu.NewReader(byteReader)
s, err := reader.ReadString() // read the entire string
r, size, err := reader.ReadRune() // or a single rune

scsu's Issues

File bug reports to Unicode about their encoder?

The README states:

however the encoding algorithm has been slightly modified as the implementation above contains a few bugs.

It might be helpful to other implementers to document the bugs of the encoder; updating the reference code may be even cooler. The reference code was supposedly introduced in UTS #6 version 4, so as far as Unicode the organization is concerned it's from 2005 instead of 1997.