XML Parsing and Generation in JavaScript Applications
fast-xml-parser, libxmljs, libxmljs2, sax, xml2js, xmlbuilder, and xmldom are npm packages that handle XML processing in JavaScript environments. They fall into two main categories: parsers (which convert XML strings into structured data like JavaScript objects or DOM trees) and builders/generators (which create XML from structured input). Some support both directions, while others specialize in streaming, DOM compliance, or performance. These tools are essential when integrating with legacy systems, SOAP APIs, RSS feeds, or configuration files that rely on XML.
Npm Package Weekly Downloads Trend
Github Stars Ranking
Stat Detail
XML Processing in JavaScript: Parsing, Building, and Streaming Compared
Working with XML in modern JavaScript apps often feels like maintaining legacy integrations—SOAP APIs, enterprise data feeds, or config files—but choosing the right tool makes a big difference. The seven packages here cover parsing, building, and DOM manipulation, each with distinct trade-offs in performance, compatibility, and ease of use. Let’s break them down by real-world engineering concerns.
🧩 Core Capabilities: What Each Package Actually Does
Not all XML libraries do the same thing. First, clarify your need:
- Parsing: Convert XML string → JS object / DOM
- Building: Convert JS object / code → XML string
- DOM: Full document object model with traversal/mutation
- Streaming: Process XML incrementally (low memory)
| Package | Parse | Build | DOM | Stream |
|---|---|---|---|---|
fast-xml-parser | ✅ | ✅ | ❌ | ❌ |
libxmljs | ✅ | ✅ | ✅ | ❌ |
libxmljs2 | ✅ | ✅ | ✅ | ❌ |
sax | ✅ | ❌ | ❌ | ✅ |
xml2js | ✅ | ❌ | ❌ | ❌ |
xmlbuilder | ❌ | ✅ | ❌ | ❌ |
xmldom | ✅ | ❌ | ✅ | ❌ |
⚠️ Deprecation Note:
libxmljsis deprecated. Its npm page states: “This package is no longer maintained. Please use libxmljs2 instead.” Do not use it in new projects.
📥 Parsing XML: Object vs DOM vs Streaming
Object-Based Parsers (fast-xml-parser, xml2js)
These convert XML directly into plain JavaScript objects—ideal for quick data extraction.
fast-xml-parser prioritizes speed and simplicity:
// fast-xml-parser
import { XMLParser } from 'fast-xml-parser';
const parser = new XMLParser();
const result = parser.parse('<book><title>JS Guide</title></book>');
// { book: { title: "JS Guide" } }
xml2js uses a callback/event-based approach but hides SAX complexity:
// xml2js
import { parseString } from 'xml2js';
parseString('<book><title>JS Guide</title></book>', (err, result) => {
// result = { book: { title: ["JS Guide"] } }
});
Note: xml2js wraps values in arrays by default (to handle repeated tags), while fast-xml-parser avoids this unless configured.
DOM Parsers (libxmljs2, xmldom)
These create a tree of node objects, letting you use standard DOM methods.
libxmljs2 (Node.js-only, native binding):
// libxmljs2
import libxmljs from 'libxmljs2';
const doc = libxmljs.parseXml('<book><title>JS Guide</title></book>');
const title = doc.get('//title').text(); // "JS Guide"
xmldom (pure JS, browser-like):
// xmldom
import { DOMParser } from '@xmldom/xmldom';
const parser = new DOMParser();
const doc = parser.parseFromString('<book><title>JS Guide</title></book>', 'text/xml');
const title = doc.getElementsByTagName('title')[0].textContent; // "JS Guide"
Streaming Parser (sax)
For huge files, stream tokens as they arrive:
// sax
import sax from 'sax';
const parser = sax.createStream(true, {});
let currentTag = '';
parser.on('opentag', (node) => { currentTag = node.name; });
parser.on('text', (text) => {
if (currentTag === 'title') console.log(text); // "JS Guide"
});
parser.write('<book><title>JS Guide</title></book>').end();
You manage state manually—great for memory efficiency, poor for developer velocity on small docs.
🏗️ Building XML: Code-Driven vs Template-Driven
Only fast-xml-parser and xmlbuilder generate XML, but differently.
xmlbuilder uses a fluent, method-chaining API:
// xmlbuilder
import { create } from 'xmlbuilder';
const xml = create({ book: { title: 'JS Guide' } }).end({ pretty: true });
// <book>
// <title>JS Guide</title>
// </book>
Or programmatically:
const root = create('book');
root.ele('title').txt('JS Guide');
console.log(root.end());
fast-xml-parser includes a builder that works from JS objects:
// fast-xml-parser builder
import { XMLBuilder } from 'fast-xml-parser';
const builder = new XMLBuilder({ format: true });
const xml = builder.build({ book: { title: 'JS Guide' } });
xmlbuilder offers more control (namespaces, attributes, CDATA), while fast-xml-parser’s builder is simpler but less feature-rich.
🖥️ Environment Compatibility: Browser, Node, Edge
- Browser-safe (pure JS):
fast-xml-parser,sax,xml2js,xmlbuilder,xmldom - Node.js-only (native deps):
libxmljs2(and deprecatedlibxmljs)
If you’re targeting Vercel Edge Functions, Cloudflare Workers, or any non-Node runtime, avoid libxmljs2. All others work anywhere JavaScript runs.
🔍 Advanced Features: XPath, Validation, Namespaces
Need XPath queries? Only libxmljs2 and xmldom support them natively:
// libxmljs2 XPath
const nodes = doc.find('//book[price > 30]');
// xmldom + xpath package (separate install)
import xpath from 'xpath';
const nodes = xpath.select('//book/title', doc);
XSD validation? Only libxmljs2 provides it out of the box.
Namespace handling is best in libxmljs2 and xmlbuilder; others have limited or inconsistent support.
🧪 Error Handling and Robustness
saxandxmldomthrow detailed parse errors for malformed XML.fast-xml-parsercan be configured to ignore or report errors via options.xml2jspasses errors to its callback.libxmljs2throws C-level errors that can crash Node if unhandled—wrap in try/catch.
🔄 Round-Trip Example: Parse → Modify → Serialize
Suppose you need to read an XML config, change a value, and write it back.
With fast-xml-parser (simplest round-trip):
import { XMLParser, XMLBuilder } from 'fast-xml-parser';
const xml = '<config><timeout>30</timeout></config>';
const parser = new XMLParser();
const builder = new XMLBuilder();
let obj = parser.parse(xml);
obj.config.timeout = 60;
const newXml = builder.build(obj);
With xmldom (DOM-style mutation):
import { DOMParser } from '@xmldom/xmldom';
import { XMLSerializer } from '@xmldom/xmldom';
const parser = new DOMParser();
const serializer = new XMLSerializer();
const doc = parser.parseFromString('<config><timeout>30</timeout></config>', 'text/xml');
doc.getElementsByTagName('timeout')[0].textContent = '60';
const newXml = serializer.serializeToString(doc);
The DOM approach is more verbose but mirrors browser XML handling—useful if your team already knows DOM APIs.
📊 When to Use Which: Decision Matrix
| Scenario | Best Choice(s) |
|---|---|
| Quick XML ↔ JSON in browser or Node | fast-xml-parser |
| Generate XML from code with full control | xmlbuilder |
| Parse massive XML files with low memory | sax |
| Need XPath, XSD, or full DOM in Node.js | libxmljs2 |
| Simulate browser XML handling in tests | xmldom |
| Simple one-off parsing (legacy codebases) | xml2js |
| New project requiring native XML features | Never libxmljs (deprecated) |
💡 Final Guidance
Start with fast-xml-parser if you just need to get data in and out of XML quickly—it’s fast, dependency-free, and works everywhere. If you’re generating complex XML documents, pair it with xmlbuilder.
Reach for sax only when file size forces streaming. Use xmldom if you’re writing XML unit tests or need DOM parity. Reserve libxmljs2 for heavy-duty enterprise scenarios where you absolutely need XPath or validation—and accept the native dependency cost.
And remember: never start a new project with libxmljs. Its successor exists for good reason.
How to Choose: xml2js vs fast-xml-parser vs libxmljs vs libxmljs2 vs sax vs xmlbuilder vs xmldom
- xml2js:
Choose
xml2jsif you want a straightforward, widely used parser that converts XML to JavaScript objects with minimal setup. It’s built onsaxunder the hood but abstracts away event handling. It’s suitable for moderate-sized documents and offers decent customization, though it’s slower thanfast-xml-parserand lacks streaming output. - fast-xml-parser:
Choose
fast-xml-parserif you need a fast, pure JavaScript parser that converts XML to JSON without native dependencies. It’s ideal for browser-compatible applications or lightweight Node.js services where bundle size and startup time matter. It supports basic validation and can preserve order via array output, but doesn’t offer full DOM APIs or streaming. - libxmljs:
Avoid
libxmljsin new projects — it is officially deprecated per its npm page and GitHub repository. The maintainers recommend migrating tolibxmljs2. It wraps the native libxml2 library, requiring compilation and limiting browser use, but historically offered strong standards compliance and XPath support. - libxmljs2:
Choose
libxmljs2if you require full DOM Level 2 compliance, XPath queries, XSD validation, or high-performance parsing of large documents in Node.js. It depends on native bindings (via node-gyp), so it won’t work in browsers or serverless edge runtimes. Use it only when you need features beyond basic parsing and can manage native dependency overhead. - sax:
Choose
saxif you’re processing very large XML files or need low-memory, streaming parsing. It’s a pure JavaScript SAX-style parser that emits events as it reads tokens, giving you fine control over memory usage. However, you must manually reconstruct object structure, making it more complex for simple use cases. - xmlbuilder:
Choose
xmlbuilderif your primary need is generating well-formed XML from JavaScript objects or programmatic calls. It provides a fluent API for building XML trees and supports namespaces, CDATA, and pretty printing. It does not parse XML, so pair it with a parser likefast-xml-parserif you need round-trip functionality. - xmldom:
Choose
xmldomif you need a browser-like DOM implementation for XML in Node.js or other non-browser environments. It parses XML into a standards-compliant DOM tree (withdocument,element, etc.), enabling traversal and manipulation similar to browser XML handling. However, it’s heavier than object-based parsers and doesn’t support streaming.
Popular Comparisons
README for xml2js
node-xml2js
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!
Description
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.
Installation
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.
Usage
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.
Shoot-and-forget usage
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) {
});
Simple as pie usage
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!
Promise usage
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
});
Parsing multiple files
If you want to parse multiple files, you have multiple possibilities:
- You can create one
xml2js.Parserper 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!
So you wanna some JSON?
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.
Displaying results
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 …? 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).
XML builder usage
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>
Adding xmlns attributes
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>
Processing attribute, tag names and values
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.normalizeis 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.: thexmlnsprefix 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
Options
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 acharkeyprefix 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 ifignoreAttrsistrue.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 ifexplicitChildrenis set totrue. Added in 0.2.5.preserveChildrenOrder(defaultfalse): Modifies the behavior ofexplicitChildrenso that the value of the "children" property becomes an ordered array. When this istrue, every node will also get a#namefield 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 ifexplicitChildrenis 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 ofxml2jsmight 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 totruewhich 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 anArrayof 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 anArrayof 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 anArrayof 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 anArrayof functions with following signature:
Added in 0.4.6function (value, name){ //do something with `name` return name }
Options for the Builder class
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(defaultrootor the root key name): root element name to be used in caseexplicitRootisfalseor 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.versionA version number string, e.g. 1.0xmldec.encodingEncoding declaration, e.g. UTF-8xmldec.standalonestandalone 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.
Updating to new version
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.
Running tests, development
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!
Getting support
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.