📃 XML parser and stringifier

✨ Features

Based on quick-xml rust package (compiled to WASM)
Support XML.parse and XML.stringify
Support  comments
Support XML entities (&, &, &, …)
Support mixed content (text and nodes)
Large output transformation options
- Auto-flattening of nodes with a single child, text or attributes
- Auto-revival of boolean, number, etc.
- Auto-group same-named nodes into arrays
- Formatting options (indentation, break lines, etc.)
- Custom reviver and replacer functions
Metadata stored into non-enumerable properties for advanced usage

🕊️ Migrating from `4.x.x` to `5.x.x`

Starting from version 5.0.0, this library use a WASM-compiled binding of the quick-xml rust package. It provides better performances while allowing to support more features.

Internal API changes

The $XML internal symbol has been replaced by a set of non-enumerable properties:

Parent node can now be accessed through "~parent" property (it’ll be null for the XML document node)
Tag name can now be accessed through "~name" property
Children nodes can now be accessed through "~children" property
- CDATA can now be tested by checking whether a node has a "~name": "~cdata" (if flattened, you’ll need to check from the parent node using ~children property)

<root>
  <node><![CDATA[hello <world>]]></node>
</root>

  <ref *1> {
-   [$XML]: { cdata: [ "root", "node" ] },
+   "~parent": null,
+   "~name": "~xml",
    root: {
      node: "hello <world>",
-     [$XML]: { name: "root", parent: null },
+     "~parent": [Circular *1],
+     "~name": "root",
+     "~children": [ { "~name": "~cdata", "#text": "hello <world>" } ],
    }
  }

XML document changes

XML document properties have been moved directly to top-level rather than being stored in xml property.

Doctype is now stored in "#doctype" property, and attributes values are set to "" rather than true.

Processing instructions (like XML stylesheets) are now parsed the same way as regular nodes but have been moved into "#instructions" property.

<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet href="styles.xsl" type="text/xsl"?>
<!DOCTYPE attribute>
<root/>

  {
-   xml: {
-     "@version": "1.0",
-     "@encoding": "UTF-8",
-   },
+   "@version": "1.0",
+   "@encoding": "UTF-8",
-   "$stylesheets": [ { "@href": "styles.xsl", "@type": "text/xsl" } ]
+   "#instructions": {
+     "xml-stylesheet": { "@href": "styles.xsl", "@type": "text/xsl" }
+   },
-   doctype: { "@attribute": true },
+   "#doctype": { "@attribute": "" },
    root: null
  }

Mixed content support

This breaks any existing code that was expecting mixed content to always be a string. Now mixed content nodes will be parsed as usual, and the #text property will contain the “inner text” of the node.

Note that #text is actually a getter that recursively gets the #text of children nodes (ignoring comment nodes), so it’ll also handle nested mixed content correctly.

<root>some <b>bold</b> text</root>

  {
-   root: "some <b>bold</b> text",
+   root: {
+     "#text": "some bold text",
+     b: "bold",
+   }
  }

Comments

Comments have been moved into "#comments" property. Note that this property is now always an array, even if there is only one comment.

Additionally, you can find comments into the ~children property by searching for nodes with "~name": "~comment". If you call the #text getter on a parent node containing comments, it will return the inner text without comments.

<root><!--some comment--></root>

  {
    root: {
-     "#comment": "some comment",
+     "#comments": [ "some comment" ],
    }
  }

Parsing

Options

Parse options are categorized into 4 groups:

clean, which can remove attributes, comments, xml doctype and instructions from the output
flatten, which can flatten nodes with only a text node, empty ones or transform attributes only nodes into objects without the @ prefix
revive, which can trim content (unless xml:space="preserve"), unescape xml entities, revive booleans and numbers
- You can also provide a custom reviver function (applied after other revivals) that will be called on each attribute and node
- Note that signature of the reviver function has changed
mode, which can be either xml or html. Choosing the latter will be more permissive than the former.

  const options = {
-   reviveBooleans: true,
-   reviveNumbers: true,
-   reviver:() => {},
+   revive: { booleans: true, numbers: true, custom: () => {} },
-   emptyToNull: true,
-   flatten: true,
+   flatten: { text: true, empty: true },
-   debug: false,
-   progress: () => null,
  }

Please refer to the documentation for more information.

Parsing streams

The parse() function supports any ReaderSync, which means you can pass directly a file reader for example.

import { parse } from "./parse.ts"
parse(await Deno.readTextFile("example.xml"))

Async parsing is not supported yet, but might be added in the future.

Stringifying

Options

Stringify options are now categorized into 2 groups:

format, which can configure the indent string and automatically breakline when a text node is too long
- Since you pass a string rather than a number for indent, it means that you can also use tabs instead of space too
replace, which can forcefully escape xml entities
- You can also provide a custom replacer function that will be called on each attribute and node
- Note that signature of the replacer function has changed

  const options = {
-   indentSize: 2,
+   format: { indent: "  " },
-   escapeAllEntities: true,
-   replacer: () => {},
+   replace: { entities: true, custom: () => {} },
-   nullToEmpty: false,
-   debug: false,
-   progress: () => null,
  }

Please refer to the documentation for more information.

Stringifying content

Please refer to the above section about API changes. If you were handling XML document properties, using the $XML symbol or #comment property, or dealing with mixed nodes content, you’ll most likely need to update your code.

This library now provides comment() and cdata() helpers to respectively create comment and CDATA nodes.

import { cdata, comment, stringify } from "./stringify.ts"
stringify({
  "@version": "1.0",
  "@encoding": "UTF-8",
  root: {
    comment: comment("hello world"),
    cdata: cdata("bonjour <le monde>"),
    text: "hello world",
    node: {
      foo: true,
      bar: 42,
      baz: {
        "@attribute": "value",
      },
    },
  },
})

<?xml version="1.0" encoding="UTF-8"?>
<root>
  <comment><!--hello world--></comment>
  <cdata><![CDATA[bonjour <le monde>]]></cdata>
  <text>hello world</text>
  <node>
    <foo>true</foo>
    <bar>42</bar>
    <baz attribute="value"/>
  </node>
</root>

Note that while you can theorethically use internal API properties, we strongly advise against it currently. Supporting ~children might be added in the future (#57) for mixed content, but its behavior is not well defined yet. Setting ~name manually might lead to unexpected behaviors, especially if it differs from the parent key.

📜 License and credits

Copyright (c) Lecoq Simon <@lowlighter>. (MIT License)
https://github.com/lowlighter/libs/blob/main/LICENSE

This library used to be published at deno.land/x/xml and jsr.io/@lowlighter/xml. It was moved into jsr.io/@libs/xml starting version 5.0.0.

Version prior to 5.0.0 used to be fully written in TypeScript but it was rewritten in Rust to improve performances and support more features.

📃 XML parser and stringifier

✨ Features

🕊️ Migrating from 4.x.x to 5.x.x

Internal API changes

XML document changes

Mixed content support

Comments

Parsing

Options

Parsing streams

Stringifying

Options

Stringifying content

📜 License and credits

🕊️ Migrating from `4.x.x` to `5.x.x`