translit

Go (Golang) utilities for (mostly Cyrillic) transliteration.

This project aims to provide:

• easy-to-configure transliteration golang.org/x/text/transform.Transformer
• language-specific transliterations (only Ukrainian/National and Russian/ICAO for now, PRs/issues are welcome)

This project is intended to be used with golang.org/x/text/transform - well-thought-out/convenient base for streaming text transforming.

If you a) don't need to build a custom transliterator b) are fine with custom license - take a look at essentialkaos/translit - it is fast and has plenty of standards implemented for Cyrillic.

Features

• easy to build custom transliterations with map[string]string rules - translit.Map(map[string]string{...}).Transformer()
• tries longest match first - correct multi-character transliterations like z → з and zh → ж (and not zh → зх)
• expected decent performance - implemented using tree lookup under the hood (no regexps, no brute-force multi-replacement; transformer itself should not generate any garbage on it's own)
• The Unlicense: Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by any means.

It is intentionally kept as simple as possible, but that comes at cost of a higher memory usage in some special cases:

• rules like "first word character" / "non-first word character" result in multiple tree paths (roughly, each "alphabet letter" + "non-first letter" permutations - applies mostly to uknational implementation)
• upper/lower case - each upper/lower character conversion should be a custom rule, like Б → B and б → b

Usage

go get -u github.com/mxmCherry/translit

Custom rules

package translit_test

import (
	"fmt"

	"github.com/mxmCherry/translit"
	"golang.org/x/text/transform"
)

func ExampleMap() {
	// pre-compile a transformer factory from rule map;
	// this is recommended to be a global variable in your own package:
	custom := translit.Map(
		map[string]string{
			"л":  "l",
			"Л":  "L",
			"ля": "lya",
			"Ля": "Lya",
		},
	)

	// get a "fresh" transformer (can be done for each transliteration instead of tr.Reset()):
	tr := custom.Transformer()

	var s string

	tr.Reset() // reset transformer state before usage - it is stateful and non-thread-safe
	s, _, _ = transform.String(tr, "Л - л")
	fmt.Println(s) // L - l

	tr.Reset() // reset transformer state before usage - it is stateful and non-thread-safe
	s, _, _ = transform.String(tr, "Ля-лЯ-ля")
	fmt.Println(s) // Lya-lЯ-lya - no rule for upper-case "Я", so it's not converted

	// Output:
	// L - l
	// Lya-lЯ-lya
}

Language-specific transliterator

package uknational_test

import (
	"fmt"

	"github.com/mxmCherry/translit/uknational"
	"golang.org/x/text/transform"
)

func ExampleToLatin() {
	uk := uknational.ToLatin() // this is recommended to be a global variable in your own package

	// https://uk.wikipedia.org/wiki/Панграма
	s, _, _ := transform.String(uk.Transformer(), "Десь чув, що той фраєр привіз їхньому царю грильяж та класну шубу з пір'я ґави.")
	fmt.Println(s)

	// Output:
	// Des chuv, shcho toi fraier pryviz yikhnomu tsariu hryliazh ta klasnu shubu z piria gavy.
}

Guidelines

This package aims to provide default transliterations for some languages.

Subpackage names for these transliterations should be made of ISO 639-1 language code and the standard name, for example: uknational, where uk is the language code and national is a standard (defined by national government).

One subpackage per language/standard approach is to reduce memory footprint: pay (with memory) only for what you actually use.

These subpackages should expose at least one translit.Factory constructor, ideally - two (two-way transformers, like ToLatin()/FromLatin()).

Code style

• editorconfig (recommended)
• gofmt (MUST)
• goreportcard (MUST)