Ever had the urge to parse XML? And wanted to access the data in some sane, easy way? Don't want to compile a C parser, for whatever reason? Then xml2js is what you're looking for!
Simple XML to JavaScript object converter. It supports bi-directional conversion. Uses sax-js and xmlbuilder-js.
Note: If you're looking for a full DOM parser, you probably want JSDom.
Simplest way to install xml2js
is to use npm, just npm install xml2js
which will download xml2js and all dependencies.
xml2js is also available via Bower, just bower install xml2js
which will download xml2js and all dependencies.
No extensive tutorials required because you are a smart developer! The task of parsing XML should be an easy one, so let's make it so! Here's some examples.
You want to parse XML as simple and easy as possible? It's dangerous to go alone, take this:
var parseString = require('xml2js').parseString;
var xml = "<root>Hello xml2js!</root>"
parseString(xml, function (err, result) {
console.dir(result);
});
Can't get easier than this, right? This works starting with xml2js
0.2.3.
With CoffeeScript it looks like this:
{parseString} = require 'xml2js'
xml = "<root>Hello xml2js!</root>"
parseString xml, (err, result) ->
console.dir result
If you need some special options, fear not, xml2js
supports a number of
options (see below), you can specify these as second argument:
parseString(xml, {trim: true}, function (err, result) {
});
That's right, if you have been using xml-simple or a home-grown wrapper, this was added in 0.1.11 just for you:
var fs = require('fs'),
xml2js = require('xml2js');
var parser = new xml2js.Parser();
fs.readFile(__dirname + '/foo.xml', function(err, data) {
parser.parseString(data, function (err, result) {
console.dir(result);
console.log('Done');
});
});
Look ma, no event listeners!
You can also use xml2js
from
CoffeeScript, further reducing
the clutter:
fs = require 'fs',
xml2js = require 'xml2js'
parser = new xml2js.Parser()
fs.readFile __dirname + '/foo.xml', (err, data) ->
parser.parseString data, (err, result) ->
console.dir result
console.log 'Done.'
But what happens if you forget the new
keyword to create a new Parser
? In
the middle of a nightly coding session, it might get lost, after all. Worry
not, we got you covered! Starting with 0.2.8 you can also leave it out, in
which case xml2js
will helpfully add it for you, no bad surprises and
inexplicable bugs!
var xml2js = require('xml2js');
var xml = '<foo></foo>';
// With parser
var parser = new xml2js.Parser(/* options */);
parser.parseStringPromise(xml).then(function (result) {
console.dir(result);
console.log('Done');
})
.catch(function (err) {
// Failed
});
// Without parser
xml2js.parseStringPromise(xml /*, options */).then(function (result) {
console.dir(result);
console.log('Done');
})
.catch(function (err) {
// Failed
});
If you want to parse multiple files, you have multiple possibilities:
- You can create one
xml2js.Parser
per file. That's the recommended one and is promised to always just work. - You can call
reset()
on your parser object. - You can hope everything goes well anyway. This behaviour is not guaranteed work always, if ever. Use option #1 if possible. Thanks!
Just wrap the result
object in a call to JSON.stringify
like this
JSON.stringify(result)
. You get a string containing the JSON representation
of the parsed object that you can feed to JSON-hungry consumers.
You might wonder why, using console.dir
or console.log
the output at some
level is only [Object]
. Don't worry, this is not because xml2js
got lazy.
That's because Node uses util.inspect
to convert the object into strings and
that function stops after depth=2
which is a bit low for most XML.
To display the whole deal, you can use console.log(util.inspect(result, false, null))
, which displays the whole result.
So much for that, but what if you use
eyes for nice colored output and it
truncates the output with ā¦
? Don't fear, there's also a solution for that,
you just need to increase the maxLength
limit by creating a custom inspector
var inspect = require('eyes').inspector({maxLength: false})
and then you can
easily inspect(result)
.
Since 0.4.0, objects can be also be used to build XML:
var xml2js = require('xml2js');
var obj = {name: "Super", Surname: "Man", age: 23};
var builder = new xml2js.Builder();
var xml = builder.buildObject(obj);
will result in:
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<root>
<name>Super</name>
<Surname>Man</Surname>
<age>23</age>
</root>
At the moment, a one to one bi-directional conversion is guaranteed only for
default configuration, except for attrkey
, charkey
and explicitArray
options
you can redefine to your taste. Writing CDATA is supported via setting the cdata
option to true
.
To specify attributes:
var xml2js = require('xml2js');
var obj = {root: {$: {id: "my id"}, _: "my inner text"}};
var builder = new xml2js.Builder();
var xml = builder.buildObject(obj);
will result in:
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<root id="my id">my inner text</root>
You can generate XML that declares XML namespace prefix / URI pairs with xmlns attributes.
Example declaring a default namespace on the root element:
let obj = {
Foo: {
$: {
"xmlns": "http://foo.com"
}
}
};
Result of buildObject(obj)
:
<Foo xmlns="http://foo.com"/>
Example declaring non-default namespaces on non-root elements:
let obj = {
'foo:Foo': {
$: {
'xmlns:foo': 'http://foo.com'
},
'bar:Bar': {
$: {
'xmlns:bar': 'http://bar.com'
}
}
}
}
Result of buildObject(obj)
:
<foo:Foo xmlns:foo="http://foo.com">
<bar:Bar xmlns:bar="http://bar.com"/>
</foo:Foo>
Since 0.4.1 you can optionally provide the parser with attribute name and tag name processors as well as element value processors (Since 0.4.14, you can also optionally provide the parser with attribute value processors):
function nameToUpperCase(name){
return name.toUpperCase();
}
//transform all attribute and tag names and values to uppercase
parseString(xml, {
tagNameProcessors: [nameToUpperCase],
attrNameProcessors: [nameToUpperCase],
valueProcessors: [nameToUpperCase],
attrValueProcessors: [nameToUpperCase]},
function (err, result) {
// processed data
});
The tagNameProcessors
and attrNameProcessors
options
accept an Array
of functions with the following signature:
function (name){
//do something with `name`
return name
}
The attrValueProcessors
and valueProcessors
options
accept an Array
of functions with the following signature:
function (value, name) {
//`name` will be the node name or attribute name
//do something with `value`, (optionally) dependent on the node/attr name
return value
}
Some processors are provided out-of-the-box and can be found in lib/processors.js
:
-
normalize
: transforms the name to lowercase. (Automatically used whenoptions.normalize
is set totrue
) -
firstCharLowerCase
: transforms the first character to lower case. E.g. 'MyTagName' becomes 'myTagName' -
stripPrefix
: strips the xml namespace prefix. E.g<foo:Bar/>
will become 'Bar'. (N.B.: thexmlns
prefix is NOT stripped.) -
parseNumbers
: parses integer-like strings as integers and float-like strings as floats E.g. "0" becomes 0 and "15.56" becomes 15.56 -
parseBooleans
: parses boolean-like strings to booleans E.g. "true" becomes true and "False" becomes false
Apart from the default settings, there are a number of options that can be
specified for the parser. Options are specified by new Parser({optionName: value})
. Possible options are:
-
attrkey
(default:$
): Prefix that is used to access the attributes. Version 0.1 default was@
. -
charkey
(default:_
): Prefix that is used to access the character content. Version 0.1 default was#
. -
explicitCharkey
(default:false
) Determines whether or not to use acharkey
prefix for elements with no attributes. -
trim
(default:false
): Trim the whitespace at the beginning and end of text nodes. -
normalizeTags
(default:false
): Normalize all tag names to lowercase. -
normalize
(default:false
): Trim whitespaces inside text nodes. -
explicitRoot
(default:true
): Set this if you want to get the root node in the resulting object. -
emptyTag
(default:''
): what will the value of empty nodes be. In case you want to use an empty object as a default value, it is better to provide a factory function() => ({})
instead. Without this function a plain object would become a shared reference across all occurrences with unwanted behavior. -
explicitArray
(default:true
): Always put child nodes in an array if true; otherwise an array is created only if there is more than one. -
ignoreAttrs
(default:false
): Ignore all XML attributes and only create text nodes. -
mergeAttrs
(default:false
): Merge attributes and child elements as properties of the parent, instead of keying attributes off a child attribute object. This option is ignored ifignoreAttrs
istrue
. -
validator
(defaultnull
): You can specify a callable that validates the resulting structure somehow, however you want. See unit tests for an example. -
xmlns
(defaultfalse
): Give each element a field usually called '$ns' (the first character is the same as attrkey) that contains its local name and namespace URI. -
explicitChildren
(defaultfalse
): Put child elements to separate property. Doesn't work withmergeAttrs = true
. If element has no children then "children" won't be created. Added in 0.2.5. -
childkey
(default$$
): Prefix that is used to access child elements ifexplicitChildren
is set totrue
. Added in 0.2.5. -
preserveChildrenOrder
(defaultfalse
): Modifies the behavior ofexplicitChildren
so that the value of the "children" property becomes an ordered array. When this istrue
, every node will also get a#name
field whose value will correspond to the XML nodeName, so that you may iterate the "children" array and still be able to determine node names. The named (and potentially unordered) properties are also retained in this configuration at the same level as the ordered "children" array. Added in 0.4.9. -
charsAsChildren
(defaultfalse
): Determines whether chars should be considered children ifexplicitChildren
is on. Added in 0.2.5. -
includeWhiteChars
(defaultfalse
): Determines whether whitespace-only text nodes should be included. Added in 0.4.17. -
async
(defaultfalse
): Should the callbacks be async? This might be an incompatible change if your code depends on sync execution of callbacks. Future versions ofxml2js
might change this default, so the recommendation is to not depend on sync execution anyway. Added in 0.2.6. -
strict
(defaulttrue
): Set sax-js to strict or non-strict parsing mode. Defaults totrue
which is highly recommended, since parsing HTML which is not well-formed XML might yield just about anything. Added in 0.2.7. -
attrNameProcessors
(default:null
): Allows the addition of attribute name processing functions. Accepts anArray
of functions with following signature:Added in 0.4.14function (name){ //do something with `name` return name }
-
attrValueProcessors
(default:null
): Allows the addition of attribute value processing functions. Accepts anArray
of functions with following signature:Added in 0.4.1function (value, name){ //do something with `name` return name }
-
tagNameProcessors
(default:null
): Allows the addition of tag name processing functions. Accepts anArray
of functions with following signature:Added in 0.4.1function (name){ //do something with `name` return name }
-
valueProcessors
(default:null
): Allows the addition of element value processing functions. Accepts anArray
of functions with following signature:Added in 0.4.6function (value, name){ //do something with `name` return name }
These options are specified by new Builder({optionName: value})
.
Possible options are:
-
attrkey
(default:$
): Prefix that is used to access the attributes. Version 0.1 default was@
. -
charkey
(default:_
): Prefix that is used to access the character content. Version 0.1 default was#
. -
rootName
(defaultroot
or the root key name): root element name to be used in caseexplicitRoot
isfalse
or to override the root element name. -
renderOpts
(default{ 'pretty': true, 'indent': ' ', 'newline': '\n' }
): Rendering options for xmlbuilder-js.- pretty: prettify generated XML
- indent: whitespace for indentation (only when pretty)
- newline: newline char (only when pretty)
-
xmldec
(default{ 'version': '1.0', 'encoding': 'UTF-8', 'standalone': true }
: XML declaration attributes.-
xmldec.version
A version number string, e.g. 1.0 -
xmldec.encoding
Encoding declaration, e.g. UTF-8 -
xmldec.standalone
standalone document declaration: true or false
-
-
doctype
(defaultnull
): optional DTD. Eg.{'ext': 'hello.dtd'}
-
headless
(default:false
): omit the XML header. Added in 0.4.3. -
allowSurrogateChars
(default:false
): allows using characters from the Unicode surrogate blocks. -
cdata
(default:false
): wrap text nodes in<![CDATA[ ... ]]>
instead of escaping when necessary. Does not add<![CDATA[ ... ]]>
if it is not required. Added in 0.4.5.
renderOpts
, xmldec
,doctype
and headless
pass through to
xmlbuilder-js.
Version 0.2 changed the default parsing settings, but version 0.1.14 introduced the default settings for version 0.2, so these settings can be tried before the migration.
var xml2js = require('xml2js');
var parser = new xml2js.Parser(xml2js.defaults["0.2"]);
To get the 0.1 defaults in version 0.2 you can just use
xml2js.defaults["0.1"]
in the same place. This provides you with enough time
to migrate to the saner way of parsing in xml2js
0.2. We try to make the
migration as simple and gentle as possible, but some breakage cannot be
avoided.
So, what exactly did change and why? In 0.2 we changed some defaults to parse
the XML in a more universal and sane way. So we disabled normalize
and trim
so xml2js
does not cut out any text content. You can reenable this at will of
course. A more important change is that we return the root tag in the resulting
JavaScript structure via the explicitRoot
setting, so you need to access the
first element. This is useful for anybody who wants to know what the root node
is and preserves more information. The last major change was to enable
explicitArray
, so everytime it is possible that one might embed more than one
sub-tag into a tag, xml2js >= 0.2 returns an array even if the array just
includes one element. This is useful when dealing with APIs that return
variable amounts of subtags.
The development requirements are handled by npm, you just need to install them.
We also have a number of unit tests, they can be run using npm test
directly
from the project root. This runs zap to discover all the tests and execute
them.
If you like to contribute, keep in mind that xml2js
is written in
CoffeeScript, so don't develop on the JavaScript files that are checked into
the repository for convenience reasons. Also, please write some unit test to
check your behaviour and if it is some user-facing thing, add some
documentation to this README, so people will know it exists. Thanks in advance!
Please, if you have a problem with the library, first make sure you read this
README. If you read this far, thanks, you're good. Then, please make sure your
problem really is with xml2js
. It is? Okay, then I'll look at it. Send me a
mail and we can talk. Please don't open issues, as I don't think that is the
proper forum for support problems. Some problems might as well really be bugs
in xml2js
, if so I'll let you know to open an issue instead :)
But if you know you really found a bug, feel free to open an issue instead.