@ffz/icu-msgparser
v2.0.0
Published
A tiny, basic parser for ICU MessageFormat strings that generates a basic AST.
Downloads
241
Readme
@ffz/icu-msgparser
A simple JavaScript parser for the ICU Message Format.
- Under 2 Kilobytes, Minified and Gzipped
- Decently Fast
- Only does what it needs to.
Include the Thing
import Parser from '@ffz/icu-msgparser';
Parse the Stuff...
const parser = new Parser(),
ast = parser.parse(`Hello, <b>{name}</b>! You have {messages, plural,
=0 {no messages}
one {one message}
other {# messages}
} and you're {completion, number, percentage} done.`);
ast === [
"Hello, ",
{
n: "b",
c: [
{v: "name"}
]
},
"! You have ",
{
v: "messages",
t: "plural",
o: {
"=0": ["no messages"],
one: ["one message"],
other: [
{v: "messages", t: "number"},
" messages"
]
}
},
" and you're ",
{
v: "completion",
t: "number",
f: "percentage"
},
" done."
]
Make Mistakes?
> parser.parse('Hello, {name{!');
SyntaxError: expected , or } at position 12 but found {
Interpret the AST
// ASTs are a simple array of strings and placeholder objects.
type MessageAST = MessageNode[];
type MessageNode = string | MessagePlaceholder;
// Placeholder objects are either tags or variables.
type MessagePlaceholder = MessageTag | MessageVariable;
type MessageTag = {
// The name of the tag. For `<link>` this would be `link`.
n: string;
// If the tag has children, the child AST will be stored in this
// variable. Otherwise, it will be undefined.
c?: MessageAST;
};
type MessageVariable = {
// n is never defined on variables and can be used to discriminate
// between variables and tags.
n: undefined;
// The name of the variable. For `{count,number}` this would be `count`.
v: string;
// The type of the variable, if a type is included. Given the example
// `{count,number}` this would be `number`.
t?: string;
// The format of the variable, if a format is included. Given the example
// `{count,number,::currency/USD}` this would be `::currency/USD`.
// For subnumeric types, this will be the `offset:` if one was provided
// and the type will be a number.
f?: string | number;
// For submessage types, this will contain all of the separate submessage
// ASTs, with their rules as unprocessed strings.
o?: MessageSubmessages;
}
type MessageSubmessages = {
[rule: string]: MessageAST;
};
Strings are just strings.
Placeholders can be tags or variables.
All tags will have a n
, or name, which tells the interpreter what tag
handler to use for them. c
, or children/content, is only included for a tag with
contents.
All variables will have a v
, or value, which tells the interpreter what
value to use for them. t
, or type, is only included for variables with
types given. This would be plural
, number
, etc.
f
is the optionally provided format. For plural
, selectordinal
, as well
as any custom types that use offset numbers, f
will be the offset number
if provided.
Finally, o
is an object with containing all the parsed sub-messages for a
variables with sub-messages.
Know the API
const parser = new Parser(/* options: */ {
// Symbols (Single Character Only)
OPEN: '{',
CLOSE: '}',
SEP: ',',
ESCAPE: "'",
SUB_VAR: '#',
TAG_OPEN: '<',
TAG_CLOSE: '>',
TAG_CLOSING: '/',
// Offset
OFFSET: 'offset:',
// Types that support offset:
subnumeric_types: ['plural', 'selectordinal'],
// Types that support sub-messages:
submessage_types: ['plural', 'selectordinal', 'select'],
// Config Flags
allowTags: true,
requireOther: true,
// or
requireOther: ['select']
});
const ast = parser.parse(message);
Tags can be completely disabled by setting allowTags
to false
.
By default, types with submessages are required to have an other
type. You
can set requireOther
to false
to disable this behavior entirely, or set
it to an array of specific types that should require other
while all other
types should not.
Make Sure It Still Works
Run tests using npm test
.
Contribute the Stuff
Please submit all issues and pull requests to the FrankerFaceZ/icu-msgparser repository.