@messageformat/parser

What is @messageformat/parser?

@messageformat/parser is a library for parsing ICU MessageFormat strings. It allows developers to handle complex internationalization (i18n) scenarios by parsing and formatting messages with pluralization, gender, and other language-specific rules.

What are @messageformat/parser's main functionalities?

Parsing MessageFormat strings

This feature allows you to parse a MessageFormat string into an abstract syntax tree (AST). The example demonstrates parsing a message that handles pluralization based on the count.

const { parse } = require('@messageformat/parser');
const message = 'You have {count, plural, one {# message} other {# messages}}.';
const parsed = parse(message);
console.log(JSON.stringify(parsed, null, 2));

Handling nested messages

This feature allows you to parse messages with nested structures, such as gender-based selection combined with pluralization. The example demonstrates parsing a message that handles both gender and pluralization.

const { parse } = require('@messageformat/parser');
const message = '{gender, select, male {He} female {She} other {They}} liked {count, plural, one {# message} other {# messages}}.';
const parsed = parse(message);
console.log(JSON.stringify(parsed, null, 2));

Custom formatting

This feature allows you to parse messages with custom formatting options, such as currency formatting. The example demonstrates parsing a message that formats a number as a currency.

const { parse } = require('@messageformat/parser');
const message = 'The price is {price, number, ::currency/USD}.';
const parsed = parse(message);
console.log(JSON.stringify(parsed, null, 2));

Other packages similar to @messageformat/parser

intl-messageformat

intl-messageformat is a library for formatting ICU MessageFormat strings. It provides similar functionality to @messageformat/parser but includes additional features for formatting messages directly, without the need for a separate parser.

messageformat

messageformat is a comprehensive library for handling ICU MessageFormat strings. It includes both parsing and formatting capabilities, making it a more complete solution compared to @messageformat/parser, which focuses solely on parsing.

fluent

fluent is a localization library developed by Mozilla. It offers a different approach to internationalization compared to ICU MessageFormat, focusing on natural language syntax and flexibility. While it serves a similar purpose, its syntax and usage differ significantly from @messageformat/parser.

README

@messageformat/parser

An AST parser for ICU MessageFormat strings – part of messageformat.

The parse(src, [options]) function takes two parameters, first the string to be parsed, and a second optional parameter options, an object with the following possible keys:

• cardinal and ordinal – Arrays of valid plural categories for the current locale, used to validate plural and selectordinal keys. If these are missing or set to false, the full set of valid Unicode CLDR keys is used: 'zero', 'one', 'two', 'few', 'many', 'other'. To disable this check, pass in an empty array.

• strict – By default, the parsing applies a few relaxations to the ICU MessageFormat spec. Setting strict: true will disable these relaxations:

  • The argType of simpleArg formatting functions will be restricted to the set of number, date, time, spellout, ordinal, and duration, rather than accepting any lower-case identifier that does not start with a number.
  • The optional argStyle of simpleArg formatting functions will not be parsed as any other text, but instead as the spec requires: "In argStyleText, every single ASCII apostrophe begins and ends quoted literal text, and unquoted {curly braces} must occur in matched pairs."
  • Inside a plural or selectordinal statement, a pound symbol (#) is replaced with the input number. By default, # is also parsed as a special character in nested statements too, and can be escaped using apostrophes ('#'). In strict mode # will be parsed as a special character only directly inside a plural or selectordinal statement. Outside those, # and '#' will be parsed as literal text.

The parser only supports the default DOUBLE_OPTIONAL apostrophe mode, in which a single apostrophe only starts quoted literal text if it immediately precedes a curly brace {}, or a pound symbol # if inside a plural format. A literal apostrophe ' is represented by either a single ' or a doubled '' apostrophe character.

This package was previously named messageformat-parser.

Installation

npm install @messageformat/parser

Usage

> const { parse } = require('@messageformat/parser')
// For clarity, the examples below do not show the ctx object included for each token

> parse('So {wow}.')
[ { type: 'content', value: 'So ' },
  { type: 'argument', arg: 'wow' },
  { type: 'content', value: '.' } ]

> parse('Such { thing }. { count, selectordinal, one {First} two {Second}' +
        '                  few {Third} other {#th} } word.')
[ { type: 'content', value: 'Such ' },
  { type: 'argument', arg: 'thing' },
  { type: 'content', value: '. ' },
  { type: 'selectordinal',
    arg: 'count',
    cases: [
      { key: 'one', tokens: [ { type: 'content', value: 'First' } ] },
      { key: 'two', tokens: [ { type: 'content', value: 'Second' } ] },
      { key: 'few', tokens: [ { type: 'content', value: 'Third' } ] },
      { key: 'other',
        tokens: [ { type: 'octothorpe' }, { type: 'content', value: 'th' } ] }
    ] },
  { type: 'content', value: ' word.' } ]

> parse('Many{type,select,plural{ numbers}selectordinal{ counting}' +
                         'select{ choices}other{ some {type}}}.')
[ { type: 'content', value: 'Many' },
  { type: 'select',
    arg: 'type',
    cases: [
      { key: 'plural', tokens: [ { type: 'content', value: 'numbers' } ] },
      { key: 'selectordinal', tokens: [ { type: 'content', value: 'counting' } ] },
      { key: 'select', tokens: [ { type: 'content', value: 'choices' } ] },
      { key: 'other',
        tokens: [ { type: 'content', value: 'some ' }, { type: 'argument', arg: 'type' } ] }
    ] },
  { type: 'content', value: '.' } ]

> parse('{Such compliance')
// ParseError: invalid syntax at line 1 col 7:
//
//  {Such compliance
//        ^

> const msg = '{words, plural, zero{No words} one{One word} other{# words}}'
> parse(msg)
[ { type: 'plural',
    arg: 'words',
    cases: [
      { key: 'zero', tokens: [ { type: 'content', value: 'No words' } ] },
      { key: 'one', tokens: [ { type: 'content', value: 'One word' } ] },
      { key: 'other',
        tokens: [ { type: 'octothorpe' }, { type: 'content', value: ' words' } ] }
    ] } ]

> parse(msg, { cardinal: [ 'one', 'other' ], ordinal: [ 'one', 'two', 'few', 'other' ] })
// ParseError: The plural case zero is not valid in this locale at line 1 col 17:
//
//   {words, plural, zero{
//                   ^

For more example usage, please take a look at our test suite.

Structure

The output of parse() is an array of tokens, Array<Content | PlainArg | FunctionArg | Select>:

interface Content {
  type: 'content'
  value: string
  ctx: Context
}

interface PlainArg {
  type: 'argument'
  arg: string
  ctx: Context
}

interface FunctionArg {
  type: 'function'
  arg: string
  key: string
  param?: Array<Content | PlainArg | FunctionArg | Select | Octothorpe>
  ctx: Context
}

interface Select {
  type: 'plural' | 'select' | 'selectordinal'
  arg: string
  cases: Array<SelectCase>
  pluralOffset?: number
  ctx: Context
}

interface SelectCase {
  key: string
  tokens: Array<Content | PlainArg | FunctionArg | Select | Octothorpe>
  ctx: Context
}

interface Octothorpe {
  type: 'octothorpe'
  ctx: Context
}

interface Context {
  offset: number
  line: number
  col: number
  text: string
  lineBreaks: number
}

Messageformat is an OpenJS Foundation project, and we follow its Code of Conduct.