December 15, 2004

Accessing raw text files using XSLT 1.0

Every once and a while someone will post a question to XSL-List in regards to accessing non-XML files and using them in a transformation. While its impossible in XSLT 1.0 to directly access non-XML text files heres a trick that David Carlisle recently posted to yet another question on this subject:

--

> I need my xsl reads a txt file, like this:

XSL1 can only read XML files, although you can easily make an xml file wrapper around your text if you have a file.txt that you know doesn't have any < or or > & characters in it then you can make a file.xml that looks like

<!DOCTYPE vector [
<!ENTITY x SYSTEM "file.txt">
]]>
<vector>&x;</vector>

then in your xslt you can go

<xsl:copy-of select="document('file.xml')"/>

and will get

<vector>
10101010101010101111010101010
10101011100001010101010101010
00011010101010101010101011011
</vector>

as the result.

-- In follow up Michael Kay adds:

In 2.0 it's simply:

<vector>
 <xsl:value-of select="unparsed-text('file.txt')"/>
</vector>

--

Related to this topic, special character handling in XSLT is something that is not very well understood. A fantastic reference on this subject was created by Dave Pawson and can be found within his excellent XSL FAQ

And to close up this post: Encoding happens to be one of the current topics over on UnderstandingXSLT.com in which Jeni Tennison has posted a lot of valuable insite and related links. If you haven't had a chance to add the XML feed of your choice to your reader here is the Atom, RSS 1.0(RDF), and RSS 2.0 feed.