[xsl] unparsed-text and analyze-string

Subject: [xsl] unparsed-text and analyze-string
From: Terry Ofner <tofner@xxxxxxxxxxx>
Date: Mon, 21 Jan 2008 17:03:19 -0500
I am attempting to parse test items using unparsed-text and analyze- string as per Michael Kay's stylesheet "analyze-names.xsl" in chapter 7 of XSLT 2.0. I am trying to add elements to different parts of test items. A section of my unparsed text input looks like this (not actual items!):

	1	XSLT means--	(G6U1S01)
	A	extensible stylesheet language transformations.
	B	extremely sly lexical transformations.
	C	XML stylesheet language transformations.
	D	all the above

	2	Stem
		 display sentence (optional)
		stem part 2 (optional)	(G6U1S01)
	A	answer choice
	B	answer choice
	C	answer choice
	D	answer choice

Note that the item numbers and letters ABCD are surrounded by tabs and the the G_code in parentheses is preceded by a tab.

So far I have been able to surround the answer choices with <choice- x> elements by using this edited version of Michael's analyze-names sheet:

<?xml version="1.0"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform";

<xsl:param name="input-uri"/>
<xsl:output indent="yes"/>

<xsl:template name="main">
  <xsl:variable name="in"
              select="unparsed-text($input-uri, 'UTF-8')"/>

<xsl:analyze-string select="$in" regex="\t([A-D])\t(.*)&#10;">
  <xsl:variable name="choiceLetter" select="regex-group(1)"/>
  <xsl:variable name="choiceLower" select="lower-case($choiceLetter)"/>
    <xsl:element name="choice-{$choiceLower}">
    <xsl:value-of select="regex-group(2)"/>
     <xsl:value-of select="."/>


Output of the following sheet looks like this:

1 XSLT means-- (G6U1S01) <choice-a>extensible stylesheet language transformations.</choice-a> <choice-b>extremely sly lexical transformations.</choice-b> <choice-c>XML stylesheet language transformations.</choice-c> <choice-d>all the above</choice-d>

My problem occurs when I get greedy and add the <notes> element around the G_code in parenthesis like this:

<xsl:analyze-string select="$in" regex="\((G[3-8]U[1-7]S\d\d?)\)&#10; \t([A-D])\t(.*)&#10;">
<xsl:element name="notes">
<xsl:value-of select="regex-group(1)"/>
<xsl:variable name="choiceLetter" select="regex-group(2)"/>
<xsl:variable name="choiceLower" select="lower-case($choiceLetter)"/>
<xsl:element name="choice-{$choiceLower}">
<xsl:value-of select="regex-group(3)"/>
. . .

That regular expression produces the following output:

	1	XSLT means--	<notes>G6U1S01</notes>
	<choice-a>extensible stylesheet language transformations.</choice-a>
	B	extremely sly lexical transformations.
	C	XML stylesheet language transformations.
	D	all the above

Which makes sense, since the regex is anchored by the ((G[3-8]U[1-7]S \d\d?)\) expression.

My question: how do I design a single sheet that will iterate through various regex expressions. I would like to avoid a situation where I am piping the output of one stylesheet through another if at all possible. Also, is it possible to use unparsed-text with text that already contains some elements?

My goal is to use XSLT in the place of a manual markup process if at all possible.

Any hints would be most appreciated.


P.S. I tried adding another <xsl:analyze-string select="$in" template. That approach put two copies of the input document into the output, which, again makes sense. My regular expressions worked at least!

Current Thread