XML Namespaces

This post covers the importance of namespaces in XML. Anyone doing XML transformations or data aggregation from these documents needs to be aware of them. This document assumes familiarity with XML to at least a minor degree.

I am not claiming this document is complete but it should hopefully serve as a quick primer for anyone wanting to wrap their heads around XML namespaces.

The Skinny

XML Namespaces, simply put, are just URIs. Full stop.

You can find the spec on namespaces at https://www.w3.org/TR/xml-names/.

Depth

Namespaces

URNs can appear which don’t look much like typical URIs, but are URIs nonetheless.

It might be best to think of the namespaces as arbitrary strings. The important takeaway here is that the XML namespace is not the prefix you see in tags and attributes. We’ll get into that later in this document.

Prefixes

A prefix is an arbitrary identifier used to refer to a namespace. It is not the namespace itself. Document producers can produce perfectly valid documents without agreeing on what prefixes to use, if any (default namespaces remove the need for prefixes).

A prefix is established with the xmlns:[prefix] attribute, where [prefix] is the name of the prefix. The namespace is the value given to the attribute.

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<i:IntuitResponse xmlns:i="http://schema.intuit.com/finance/v3"
                time="2016-10-14T10:48:39.109-07:00">
    <i:QueryResponse startPosition="1" maxResults="79" totalCount="79">
    </i:QueryResponse>
</i:IntuitResponse>

In this document, the uses i as the prefix. The namespace is http://schema.intuit.com/finance/v3. For all of these nodes, the namespace is http://schema.intuit.com/finance/v3.

Another document producer could crank this out as well:

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<foo:IntuitResponse xmlns:foo="http://schema.intuit.com/finance/v3"
                time="2016-10-14T10:48:39.109-07:00">
    <foo:QueryResponse startPosition="1" maxResults="79" totalCount="79">
    </foo:QueryResponse>
</foo:IntuitResponse>

These two documents are semantically the same, because the namespaces attached to the nodes are the same. Any sort of transformation or aggregation you do with XML data must account for this, or any variation in the prefix will run you afoul of XML validators. Some validators might be less picky in this regard - probably because the authors of said validators didn’t have a firm grasp on namespaces. The problem with ignoring them is that downstream consumers of the document will be looking at invalid documents - data will have been removed.

Another peculiar bit about namespace prefixes is that they are applied to the node which contains the attribute, not just the children.

Default Namespaces

A default namespace is essentially an implicit namespace. It is declared with the xmlns attribute, and like prefixed namespaces, the value is namespace itself (a URI).

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<IntuitResponse xmlns="http://schema.intuit.com/finance/v3"
                time="2016-10-14T10:48:39.109-07:00">
    <QueryResponse startPosition="1" maxResults="79" totalCount="79">
    </QueryResponse>
</IntuitResponse>

In this case the xmlns attribute declares the default namespace as http://schema.intuit.com/finance/v3 and it is applied to the IntuitResponse element as well as QueryResponse. This would descend down the entire document hierarchy until another default namespace were declared. The most immediate parent declaration wins when determining a child’s namespace.

It’s also noteworthy that this document is semantically the same as those found in Prefixes.

Querying an XML Document with Namespaces

I have found that tools that use XPath and even other tools will provide a mechanism to declare a namespace alias, since queries become unwieldy when the full namespace is used. These aliases look an awful lot like prefixes.

Here is what the nokogiri Ruby gem looks like when using XPath:

puts doc.xpath('/i:IntuitResponse/i:QueryResponse',
                'i' => "http://schema.intuit.com/finance/v3")

And this will get the QueryResponse contents.

Similarly, here is a C# API:

XmlNamespaceManager nsmgr = new XmlNamespaceManager(doc.NameTable);
nsmgr.AddNamespace("i", "http://schema.intuit.com/finance/v3");
XmlNodeList nodes = el.SelectNodes(@"/i:IntuitResponse/i:QueryResponse", nsmgr);

Tools such as oq, which do not use XPath, have chosen a similar approach:

oq \
  --xmlns \
  --xml-namespace-alias 'i=http://schema.intuit.com/finance/v3' \
  -i xml \
  -o xml \
  <<< <EOF

EOF