This is an implementation of YAML, a human friendly data serialization language.
Previously, it was a native port of PyYAML. Now, it is rewritten from scratch, directly from version 1.2 of the specification.
- The last argument of loader functions (
load
,loadAll
,safeLoad
, andsafeLoadAll
) was changed. Now, it is an optionalsettings
plain object. See the API listing onload
function for details. scan
,parse
,compose
,addConstructor
functions and all of the classes likeLoader
,Constructor
,Resolver
was dropped because of complete architecture overhaul.- The parsing process consists of only one stage now. The loader constructs the resulting strings, arrays, and objects (mappings) without any interim representation objects. So, tag interpreters receive native JavaScript objects directly.
If your code does not use neither custom tags nor explicitly specified Loader
class (the last argument of load
function), you are not required to change
anything.
Otherwise, you should rewrite your tag constructors and custom Loader classes to conform the new schema-based API. It consists of two classes: Schema and Type. The both are described below in the API listing. Here is an example of the difference:
JS-YAML 1.x.x
var yaml = require('js-yaml');
yaml.addConstructor('!cookies', function (node) {
var array, index, length;
array = this.constructSequence(node);
for (index = 0, length = array.length; index < length; index += 1) {
array[index] = 'A ' + array[index] + ' with some cookies!';
}
return array;
});
result = yaml.load(data);
JS-YAML 2.0.x
var yaml = require('js-yaml');
var cookiesType = new yaml.Type('!cookies', {
loader: {
kind: 'array',
resolver: function (array, explicit) {
var index, length;
for (index = 0, length = array.length; index < length; index += 1) {
array[index] = 'A ' + array[index] + ' with some cookies!';
}
return array;
}
}
});
var COOKIES_SCHEMA = new yaml.Schema({
include: [ yaml.DEFAULT_SCHEMA ],
explicit: [ cookiesType ]
});
result = yaml.load(data, { schema: COOKIES_SCHEMA });
y
,yes
,n
,no
,on
,off
are not converted to Booleans anymore. Decision to drop support of such "magic" was made after speaking with YAML core developers: from now on we try to keep as minimal subset of rules as possible to keep things obvious. Booleans are following YAML 1.2 core schema now: http://www.yaml.org/spec/1.2/spec.html#id2804923require('file.yml')
now returns a single document (was array of documents) and throws an error when file contains multiple documents. That should improve switching between YAML <-> JSON. Sorequire('file.yml')
will give the same result as if it wasrequire('file.json')
now.- CLI tool
js-yaml
become part ofjs-yaml
again.
npm install js-yaml
If you want to inspect your YAML files from CLI, install js-yaml globally:
npm install js-yaml -g
usage: js-yaml [-h] [-v] [-c] [-j] [-t] file
Positional arguments:
file File with YAML document(s)
Optional arguments:
-h, --help Show this help message and exit.
-v, --version Show program's version number and exit.
-c, --compact Display errors in compact mode
-j, --to-json Output a non-funky boring JSON
-t, --trace Show stack trace on error
<script src="js-yaml.min.js"></script>
<script type="text/javascript">
var doc = jsyaml.load('greeting: hello\nname: world');
</script>
Browser support is still buggy, and mostly done to run online demo. If you can help to improve browser compatibility and AMD support - rise pull request.
Support of oldIEs and some other prehistoric browsers is possible using es5-shims. Just include shims before jsyaml to use it with outdated browsers.
JS-YAML automatically registers handlers for .yml
and .yaml
files. You can
load them just with require
. That's mostly equivalent to calling load()
on
fetched content of a file. Just with one string!
require('js-yaml');
// Get document, or throw exception on error
var doc = require('/home/ixti/example.yml');
console.log(doc);
Parses string
as single YAML document. Returns JS object or throws
YAMLException
on error.
NOTE: This function does not understands multi-document sources, it throws exception on those.
settings
is an optional hash-like object allows to change the loader's
behaviour. It may contain the following keys:
schema
specifies a schema to use. It'sDEFAULT_SCHEMA
by default. See below for more information about the schemas.validate
(default true) enables/disables validation of the input stream according to YAML rules. If you are sure about the input, you can set it to false and (maybe) gain some additional performance.strict
(default false) makes the loader to throw errors instead of warnings.legacy
(default false) makes the loader to expect YAML 1.1 documents if such documents have no explicit %YAML directive.name
(default null) is a string to be used as a file path in error/warning messages.resolve
(default true) enables/disables resolving of nodes with non-specific'?'
tag. That are plain scalars without any explicit tag. The result of switching it to false is that all of the plain scalars will loaded as strings. This may significantly increase parsing performance.
var yaml = require('js-yaml');
// pass the string
fs.readFile('/home/ixti/example.yml', 'utf8', function (err, data) {
if (err) {
// handle error
return;
}
try {
console.log( yaml.load(data) );
} catch(e) {
console.log(e);
}
});
Same as load()
, but understands multi-doc sources and apply iterator
to each
document.
var yaml = require('js-yaml');
// pass the string
fs.readFile('/home/ixti/example.yml', 'utf8', function (err, data) {
if (err) {
// handle error
return;
}
try {
yaml.loadAll(data, function (doc) {
console.log(doc);
});
} catch(e) {
console.log(e);
}
});
Same as load()
but uses SAFE_SCHEMA by default - only recommended tags of YAML
specification (no JavaScript-specific tags, e.g. !!js/regexp
).
Same as loadAll()
but uses SAFE_SCHEMA by default - only recommended tags of
YAML specification (no JavaScript-specific tags, e.g. !!js/regexp
).
Serializes object
as single, bare YAML document. settings
is an optional
hash-like object allows to change the dumper's behaviour. It may contain the
following keys:
schema
specifies a schema to use. It'sDEFAULT_SCHEMA
by default. See below for more information about the schemas.indent
(default 2) indentation width to use (in spaces).flowLevel
(default -1) specifies level of nesting on which the dumper must switch to the flow style (i.e. JSON-like) of collection nodes.styles
is a "tag" => "style" map. Each tag may have own set of styles. See below for full listing standard tag styles.
var yaml = require('js-yaml');
var object = {
name: 'Wizzard',
level: 17,
sanity: null,
inventory: [
{
name: 'Hat',
features: [ 'magic', 'pointed' ],
traits: {}
},
{
name: 'Staff',
features: [],
traits: { damage: 10 }
},
{
name: 'Cloak',
features: [ 'old' ],
traits: { defence: 0, comfort: 3 }
}
]
};
console.log(yaml.dump(object, {
flowLevel: 3,
styles: {
'!!int' : 'hexadecimal',
'!!null' : 'camelcase'
}
}));
"name": "Wizzard"
"level": 0x11
"sanity": Null
"inventory":
- "name": "Hat"
"features": ["magic", "pointed"]
"traits": {}
- "name": "Staff"
"features": []
"traits": {"damage": 0xA}
- "name": "Cloak"
"features": ["old"]
"traits": {"defence": 0x0, "comfort": 0x3}
Same as dump()
but uses SAFE_SCHEMA by default - only recommended tags of YAML
specification (no JavaScript-specific tags, e.g. !!js/regexp
).
Constructs an object to use by the loader via the schema
setting described
above. Schemas are collections of YAML type objects collected in implicit
and explicit
arrays. The loader will try to resolve each plain scalar in a
document using the resolver function associeted with each type in the implicit
list. If a node has an explicit tag, the loader will look for it in the both
lists. include
is an array of super schemas. When compiling a schema, the
loader will take types in bottom-top order; the specified schema comes first,
and all of super schemas come next in order of they are placed in the include
list. Recursively.
NOTE: Schemas must be immutable. So, it's better to use CONSTANT_LIKE_NAMES for them.
There are predifined schemas in JS-YAML: MINIMAL_SCHEMA
, SAFE_SCHEMA
, and
DEFAULT_SCHEMA
.
Constructs a YAML type definition object. Such objects are used for validation,
resolving, interpreting, and representing of primitive YAML nodes: scalars
(strings), sequences (arrays), and mappings (objects). The second argument is an
object of two keys: loader
and dumper
. At least one of these must be
specified. Both of the keys are objects too.
kind
(required) is a string identifier ("string", "array", or "object")
restricts type of acceptable nodes.
resolver
(optional) is a function of two arguments: object
is a primitive
YAML node to resolve and explicit
is a boolean value. When a type is contained
in the implicit list of a schema, and a node has no explicit tag on it,
explicit
will be false. Otherwise, it will be true.
kind
(required) is a string identifier restricts type of acceptable objects.
Allowed values are: "undefined", "null", "boolean", "integer", "float",
"string", "array", "object", and "function".
instanceOf
(optional) allows to restrict acceptable objects with exactly one
class. i.e. constructor function.
predicate
(optional) is a function of one argument. It takes an object and
returns true to accept and false to discard.
representer
(optional) is a function intended to convert objects to simple,
"dumpable" form. That is a string, an array, or a plain object.
Special object used in type resolvers/representers to report failure of the resolving/representing process. If your type handler cannot to process the given object, it should return NIL.
var yaml = require('js-yaml');
var cookiesType = new yaml.Type('!cookies', {
loader: {
kind: 'string',
resolver: function (object, explicit) {
return 'A ' + object + ' with some cookies!';
}
},
dumper: {
kind: 'string',
representer: function (object, style) {
var match = /^A (.+?) with some cookies!$/.exec(object);
if (null !== match) {
return match[1];
} else {
return yaml.NIL;
}
}
}
});
var COOKIES_SCHEMA = new yaml.Schema({
include: [ yaml.DEFAULT_SCHEMA ],
explicit: [ cookiesType ]
});
var loaded = yaml.load('!cookies coffee', { schema: COOKIES_SCHEMA });
var dumped = yaml.dump(loaded, { schema: COOKIES_SCHEMA });
console.log(loaded);
console.log(dumped);
A coffee with some cookies!
!<!cookies> "coffee"
The list of standard YAML tags and corresponding JavaScipt types. See also YAML tag discussion and YAML types repository.
!!null '' # null
!!bool 'yes' # bool
!!int '3...' # number
!!float '3.14...' # number
!!binary '...base64...' # buffer
!!timestamp 'YYYY-...' # date
!!omap [ ... ] # array of key-value pairs
!!pairs [ ... ] # array or array pairs
!!set { ... } # array of objects with given keys and null values
!!str '...' # string
!!seq [ ... ] # array
!!map { ... } # object
JavaScript-specific tags
!!js/regexp /pattern/gim # RegExp
!!js/undefined '' # Undefined
!!js/function 'function () {...}' # Function
!!null
"canonical" => "~"
!!int
"binary" => "0b1", "0b101010", "0b1110001111010"
"octal" => "01", "052", "016172"
"decimal" => "1", "42", "7290"
"hexadecimal" => "0x1", "0x2A", "0x1C7A"
!!null, !!bool, !!float
"lowercase" => "null", "true", "false", ".nan", '.inf'
"uppercase" => "NULL", "TRUE", "FALSE", ".NAN", '.INF'
"camelcase" => "Null", "True", "False", ".NaN", '.Inf'
By default, !!int uses decimal
, and !!null, !!bool, !!float use lowercase
.
Note, that you use arrays or objects as key in JS-YAML. JS do not allows objects or array as keys, and stringifies (by calling .toString method) them at the moment of adding them.
---
? [ foo, bar ]
: - baz
? { foo: bar }
: - baz
- baz
{ "foo,bar": ["baz"], "[object Object]": ["baz", "baz"] }
Also, reading of properties on implicit block mapping keys is not supported yet. So, the following YAML document cannot be loaded.
&anchor foo:
foo: bar
*anchor: duplicate key
baz: bat
*anchor: duplicate key
View the LICENSE file (MIT).