tannin
v1.2.0
Published
gettext localization library compatible with Jed-formatted locale data
Downloads
279,244
Maintainers
Readme
Tannin
Tannin is a gettext localization library.
Inspired by Jed, it is built to be largely compatible with Jed-formatted locale data, and even offers a Jed drop-in replacement compatibility shim to easily convert an existing project. Contrasted with Jed, it is more heavily optimized for performance and bundle size. While Jed works well with one-off translations, it suffers in single-page applications with repeated rendering of elements. Using Tannin, you can expect a bundle size 20% that of Jed (980 bytes gzipped) and upwards of 330x better performance (see benchmarks). It does so without sacrificing the safety of plural forms evaluation, using a hand-crafted expression parser in place of the verbose compiled grammar included in Jed.
Furthermore, the project is architected as a mono-repo, published on npm under the @tannin
scope. These modules can be used standalone, with or without Tannin. For example, you may find value in @tannin/compile
for creating an expression evaluator, or @tannin/sprintf
as a minimal printf string formatter.
The following modules are available:
tannin
@tannin/compat
@tannin/compile
@tannin/evaluate
@tannin/plural-forms
@tannin/compat
@tannin/postfix
@tannin/sprintf
Installation
Using npm as a package manager:
npm install tannin
Otherwise, download a pre-built copy from unpkg:
https://unpkg.com/tannin/dist/tannin.min.js
Usage
Construct a new instance of Tannin
, passing locale data in the form of a Jed-formatted JSON object.
The returned Tannin
instance includes the fully-qualified dcnpgettext
function to retrieve a translated string.
import Tannin from 'tannin';
const i18n = new Tannin( {
the_domain: {
'': {
domain: 'the_domain',
lang: 'en',
plural_forms: 'nplurals=2; plural=(n != 1);',
},
example: [ 'singular translation', 'plural translation' ],
},
} );
i18n.dcnpgettext( 'the_domain', undefined, 'example' );
// ⇒ 'singular translation'
Tannin accepts plural_forms
both as a standard gettext plural forms string or as a function which, given a number, should return the (zero-based) plural form index. Providing plural_forms
as a function can yield a performance gain of approximately 8x for plural evaluation.
For example, consider the following "default" English (untranslated) initialization:
const i18n = new Tannin( {
messages: {
'': {
domain: 'messages',
plural_forms: ( n ) => n === 1 ? 0 : 1,
},
},
} );
i18n.dcnpgettext( 'messages', undefined, 'example', 'examples', 1 );
// ⇒ 'example'
i18n.dcnpgettext( 'messages', undefined, 'example', 'examples', 2 );
// ⇒ 'examples'
Jed Compatibility
For a more human-friendly API, or to more easily transition an existing project, consider using @tannin/compat
as a drop-in replacement for Jed.
import Jed from '@tannin/compat';
const i18n = new Jed( {
locale_data: {
the_domain: {
'': {
domain: 'the_domain',
lang: 'en',
plural_forms: 'nplurals=2; plural=(n != 1);',
},
example: [ 'singular translation', 'plural translation' ],
},
},
domain: 'the_domain',
} );
i18n.translate( 'example' ).fetch();
// ⇒ 'singular translation'
Benchmarks
The following benchmarks are performed in Node 10.16.0 on a MacBook Pro (2019), 2.4 GHz 8-Core Intel Core i9, 32 GB 2400 MHz DDR4 RAM.
Singular
---
Tannin x 216,670,213 ops/sec ±0.73% (90 runs sampled)
Tannin (Optimized Default) x 219,477,869 ops/sec ±0.32% (96 runs sampled)
Jed x 58,730,499 ops/sec ±0.34% (96 runs sampled)
Singular (Untranslated)
---
Tannin x 75,835,743 ops/sec ±1.26% (96 runs sampled)
Tannin (Optimized Default) x 76,474,169 ops/sec ±0.61% (92 runs sampled)
Jed x 241,632 ops/sec ±0.73% (96 runs sampled)
Plural
---
Tannin x 7,108,006 ops/sec ±0.96% (95 runs sampled)
Tannin (Optimized Default) x 51,658,190 ops/sec ±1.25% (94 runs sampled)
Jed x 236,797 ops/sec ±0.98% (97 runs sampled)
To run benchmarks on your own machine:
git clone https://github.com/aduth/tannin.git
cd tannin
npm install
node packages/tannin/benchmark
License
Copyright 2019-2020 Andrew Duthie
Released under the MIT License.