filterxml

filterxml

Keep it simple! Here is a Node.js module to remove unnecessary XML nodes that match given XPath expressions. It uses xpath and xmldom under the hood.

Command-line usage – Node API – Examples – Working with namespaces

Command-line usage

Install with $ npm install filterxml -g and then:

$ filterxml -e pattern -n prefix=namespaceURI input.xml output.xml

For example, remove Style and StyleMap from a Keyhole Markup Language document with:

$ filterxml --exclude kml:Style --exclude kml:StyleMap \
    --namespace kml=http://www.opengis.net/kml/2.2 \
    source.kml simplified.kml

Specify multiple patterns and namespaces with additional -e, --exclude and -n, --namespace flags. See filterxml --help for details.

Node API usage

Install with $ npm install filterxml and then:

> const filterxml = require('filterxml')
> filterxml(xmlIn, patterns, namespaces, function (err, xmlOut) { ... })

Where

• xmlIn is a string representing the input XML document.
• patterns is an array of XPath expressions, like 'book', '/bookstore/book', or '//html:title'. The matching XML nodes will be removed.
• namespaces is a map from prefixes to namespace URIs, for example { html: 'http://www.w3.org/TR/html4/' }
• xmlOut is a string representing the filtered output XML document.

Common XPath expressions to match nodes include:

• x:book to match all book nodes under a namespace associated with the x prefix in namespaces.
• x:bookstore/x:book to match all books directly under a bookstore.
• x:bookstore//x:book to match all books somewhere under a bookstore.
• x:bookstore/x:book[1] to match the first book directly under a bookstore.
• book to match all book nodes that are not under a namespace. This is a quite rare situation in real-world XML documents.

Limitations

Internally, filterxml depends on xmldom that respects the standard Web API XMLSerializer. The serializer can sometimes produce unexpected results:

• most empty-element tags like <circle /> will be converted to begin and end tags like <circle></circle>.
• some common empty-element tags like <meta /> are preserved but will lose the space before the slash like <meta/>.

Example

Let us filter out all book nodes:

const xmlIn = '<bookstore>' +
    '<book>Animal Farm</book>' +
    '<book>Nineteen Eighty-Four</book>' +
    '<essay>Reflections on Writing</essay>' +
  '</bookstore>'

filterxml(xmlIn, ['book'], {}, function (err, xmlOut) {
  if (err) { throw err }
  console.log(xmlOut)
})

Outputs:

<bookstore><essay>Reflections on Writing</essay></bookstore>

Real-world example

Let us remove Style tags from a Keyhole Markup Language (KML) file:

<?xml version="1.0" encoding="UTF-8"?>
<kml xmlns="http://www.opengis.net/kml/2.2">
  <Document>
    <name>Awesome locations</name>
    <Style id="s_ylw-pushpin060">
      <IconStyle>
        <scale>1.1</scale>
        <Icon>
          <href>http://maps.google.com/mapfiles/kml/pushpin/ylw-pushpin.png</href>
        </Icon>
        <hotSpot x="20" y="2" xunits="pixels" yunits="pixels"/>
      </IconStyle>
      <PolyStyle>
        <fill>0</fill>
      </PolyStyle>
    </Style>
    <Placemark>
      <name>Reykjavik</name>
      <Point>
        <coordinates>-21.933333,64.133333,0</coordinates>
      </Point>
    </Placemark>
  </Document>
</kml>

We read the file, filter it, and save the result. Note how we must add a namespace prefix into our pattern to match nodes under the namespace defined in kml node. Note also how we must associate any used prefix with a namespace URI.

const filterxml = require('filterxml')
const fs = require('fs')

const xmlIn = fs.readFileSync('./norway.kml')
const patterns = ['x:Style']
const namespaces = {
  x: 'http://www.opengis.net/kml/2.2'
}

filterxml(xmlIn, patterns, namespaces, function (err, xmlOut) {
  if (err) { throw err }
  fs.writeFileSync('./norway-simplified.kml', xmlOut)
})

The resulting norway-simplified.kml:

<?xml version="1.0" encoding="UTF-8"?>
<kml xmlns="http://www.opengis.net/kml/2.2">
  <Document>
    <name>Awesome locations</name>
    <Placemark>
      <name>Reykjavik</name>
      <Point>
        <coordinates>-21.933333,64.133333,0</coordinates>
      </Point>
    </Placemark>
  </Document>
</kml>

Working with namespaces

Often XML documents specify multiple namespaces. For example:

<?xml version="1.0" encoding="UTF-8"?>
<kml xmlns="http://www.opengis.net/kml/2.2" xmlns:gx="http://www.google.com/kml/ext/2.2">
  <Point>
    <gx:drawOrder>1</gx:drawOrder>
    <coordinates>25.59176188650433,45.6493071755744,0</coordinates>
  </Point>
</kml>

To match and remove the gx:drawOrder node we can not just filterxml(xmlIn, ['gx:drawOrder'], {}, callback). The callback would receive an error No namespace associated with prefix gx in gx:drawOrder. Instead, we must specify what the prefix gx in our XPath pattern means. It misleadingly looks like it has already been specified in the kml tag. We cannot blindly trust it. This is because the same prefix can map to different namespace URI in different part of the document:

<?xml version="1.0" encoding="UTF-8"?>
<mymap>
  <Places xmlns:kml="http://www.opengis.net/kml/2.2">
    <kml:Placemark>
      <kml:name>A place described with OpenGIS markup</kml:name>
    </kml:Placemark>
  </Places>
  <Cities xmlns:kml="http://www.google.com/kml/ext/2.2">
    <kml:Placemark>
      <kml:name>A place described with Google's KML markup</kml:name>
    </kml:Placemark>
  </Cities
</mymap>

Therefore we must always specify the prefixes we use in our XPath patterns. To remove gx:drawOrder the following is a valid approach. Note that we can use whatever prefix we want as long as we associate it with a correct namespace URI.

const patterns = ['foo:drawOrder']
const namespaces = { foo: 'http://www.google.com/kml/ext/2.2' }
filterxml(xmlIn, patterns, namespaces, function (err, xmlOut) {
  ...
})

The snippet above results with xmlOut equal to:

<?xml version="1.0" encoding="UTF-8"?>
<kml xmlns="http://www.opengis.net/kml/2.2" xmlns:gx="http://www.google.com/kml/ext/2.2">
  <Point>
    <coordinates>25.59176188650433,45.6493071755744,0</coordinates>
  </Point>
</kml>

So, always declare your prefixes!

Licence

MIT

Changelog

[2.0.0] - 2024-03-14

Added

• include package-lock.json in the NPM package.
• add badges for supported Node.js versions and Actions workflow. (#5)
• install dev dependency: standard (#4)
• write CHANGELOG.md (#10)

Fixed

• Breaking: preserve end of file newline characters (#9)

Changed

• Breaking: drop support for Node.js 10.x and earlier.
• Breaking: TypeErrors in serialization are not re-thrown but passed to the callback.
• convert codebase to StandardJS style. (#4)
• upgrade dependencies: commander, xpath (#7)
• upgrade dev dependencies: diff, genversion, mocha, temp (#7)
• improve documentation (#6)
• switch from TravisCI to GitHub Actions (#5)

Removed

• uninstall dependency: fs-extra (#7)
• uninstall dev dependencies: eslint, command-line-test (#4, #7)

Migration Tips

Likely you do not need to make any changes to your code. However, ensure you are running filterxml on Node.js 12.x or later.