parse XSLT::Stylesheet using the libxslt-preferred options

[menafkasap]

The content in CDATA isn't affected by disable-output-escaping

Two files shared below don't produce the same output with xsltproc and nokogiri.

FILES

test.xsl

<?xml version="1.0"?>
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:output method="text" omit-xml-declaration="yes" />
    <xsl:template match="/">
        <xsl:text disable-output-escaping="yes"><![CDATA[<>]]>
</xsl:text>
    </xsl:template>
</xsl:stylesheet>

---

test.xml

<t></t>

---

test.rb

#!/usr/bin/env ruby

require 'nokogiri'

xslt  = Nokogiri::XSLT(File.read(ARGV[0]))
xml = Nokogiri::XML(File.read(ARGV[1]))

puts xslt.transform(xml)

OUTPUTS

xsltproc test.xsl test.xml

<>

---

./test.rb test.xsl test.xml

<?xml version="1.0"?>
&lt;&gt;

[flavorjones]

@menafkasap Thanks for reporting this issue, and apologies for my slow reply.

I've tracked this down to the parse options used for the XSL file. xsltproc uses the C constant XSLT_PARSE_OPTIONS (which is NOENT | DTDLOAD | DTDATTR | NOCDATA), while Nokogiri uses XML::ParseOptions::DEFAULT_XML (which is RECOVER | NONET). Specifically, the behavioral difference you're spotting is due to the presence/absence of NOCDATA.

I think what we should be doing here is using these same options for the XSL files within Nokogiri (e.g., in the XSLT.parse method).

[flavorjones]

PR submitted at #2221. Will land in v1.12.x.