Question

I am trying to validate an XML document against a dozen or so schemas using Nokogiri. Currently I have a root schema document that imports all the other schemas, and I validate against that.

Can I point to each schema file from the XML file itself, and have Nokogiri look in the XML file for the schemas to validate against?

Answer 1

The proper way to reference multiple schemata against which to validate an XML file is with the schemaLocation attribute:

<?xml version="1.0"?>
<foo xmlns="http://bar.com/foo"
     xmlns:bz="http://biz.biz/"
     xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
     xsi:schemaLocation="http://bar.com/foo http://www.bar.com/schemas/foo.xsd
                         http://biz.biz/ http://biz.biz/xml/ns/bz.xsd">

For each namespace in your document you list a pair of whitespace-delimited values: the namespace URI followed by a ‘hint’ as to where to find the schema for that namespace. If you provide a full URI for each hint, then you can process this with Nokogiri as such:

require 'nokogiri'
require 'open-uri'

doc = Nokogiri.XML( my_xml )
schemata_by_ns = Hash[ doc.root['schemaLocation'].scan(/(\S+)\s+(\S+)/) ]
schemata_by_ns.each do |ns,xsd_uri|
  xsd = Nokogiri::XML.Schema(open(xsd_uri))
  xsd.validate(doc).each do |error|
    puts error.message
  end
end

Disclaimer: I have never attempted to validate a single XML document using multiple namespaced schemata with Nokogiri before. As such, I have no direct experience to guarantee that the above validation will work. The validation code is based solely on Nokogiri’s schema validation documentation.