XmlStream-Saxes

XmlStream is a Node.js XML stream parser and editor, based on saxes (libexpat SAX-like parser binding).

Fork of xml-stream project which is unmaintained

$ npm install xml-stream-saxes

Rationale

When working with large XML files, it is probably a bad idea to use an XML to JavaScript object converter, or simply buffer the whole document in memory. Then again, a typical SAX parser might be too low-level for some tasks (and often a real pain).

This is why we've rolled our own stream parser that tries to address these shortcomings. It processes an XML stream chunk by chunk and fires events only for nodes of interest, matching them with CSS-like selectors.

Events

Supported events:

• data on outgoing data chunk,
• end when parsing has ended,
• startElement[: selector] on opening tag for selector match,
• updateElement[: selector] on finished node for selector match with its contents buffered,
• endElement[: selector] on closing tag for selector match,
• text[: selector] on tag text for selector match.

When adding listeners for startElement, updateElement, and text the callback can modify the provided node, before it is sent to the consumer.

Selector syntax is CSS-like and currently supports:

• ancestor descendant
• parent > child

Take a look at the examples for more information.

Element Node

Each of the four node events has a callback with one argument. When parsing, this argument is set to the current matched node. Having a chunk of XML like this:

1<item id="123" type="common">
2  <title>Item Title</title>
3  <description>Description of this item.</description>
4  (text)
5</item>

The structure of the item element node would be:

1{
2  title: 'Item Title',
3  description: 'Description of this item.',
4  '$': {
5    'id': '123',
6    'type': 'common'
7  },
8  '$name': 'item',
9  '$text': '(text)'
10}

Naturally, element text and child elements wouldn't be known until discovered in the stream, so the structure may differ across events. The complete structure as displayed should be available on updateElement. The $name is not available on endElement.

Collecting Children

It is sometimes required to select elements that have many children with one and the same name. Like this XML:

1<item id="1">
2  <subitem>one</subitem>
3  <subitem>two</subitem>
4</item>
5<item id="2">
6  <subitem>three</subitem>
7  <subitem>four</subitem>
8  <subitem>five</subitem>
9</item>

By default, parsed element node contains children as properties. In the case of several children with same names, the last one would overwrite others. To collect all of subitem elements in an array use collect:

1xml.collect('subitem');
2xml.on('endElement: item', function(item) {
3  console.log(item);
4})

Preserving Elements and Text

By default, element text is returned as one concatenated string. In this XML:

1<item>
2  one <a>1</a>
3  two <b>2</b>
4</item>

The value of $text for item would be: one 1 two 2 without any indication of the order of element a, element b, and text parts. To preserve this order:

1xml.preserve('item');
2xml.on('endElement: item', function(item) {
3  console.log(item);
4})

Pause and resume parsing

If you want parsing to pause (for example, until some asynchronous operation of yours is finished), you can pause and resume XML parsing:

1xml.pause();
2myAsyncFunction( function() {
3  xml.resume();
4});

Beware that resume() must not be called from within a handler callback.