Introduction to XSLT

As a tool for humanities computing

8 Sorting and grouping

ALLC/ACH 2002, Tübingen

21-22 July 2002

C. M. Sperberg-McQueen

Wendell Piez

TOC | First

I.1. Overview
I.2. Sorting and grouping
I.3. Sorting
I.4. Group exercises
I.5. Grouping elsewhere
I.6. Grouping in XSLT
I.7. Grouping
I.8. A simple optimization
I.9. Grouping
I.10. Muenchian grouping: keys
I.11. Group exercise: reformatting census data
I.12. Individual exercises

I. Monday afternoon 2: sorting and grouping

Overview

1 of 12

Sunday a.m.: introductions
Sunday a.m.: basics (simple transformations, if, choose, selection by attribute values)
Sunday p.m.: modes (e.g. tables of contents)
Sunday p.m.: XPath
Monday a.m.: functions, numbering
Monday a.m.: near-identity transformations
Monday p.m.: named templates, recursion
Monday p.m.: sorting and grouping

Sorting and grouping

2 of 12

sorting (easy)
grouping (takes some thought)

Sorting

3 of 12

The xsl:sort element occurs as child of apply-templates

    <xsl:apply-templates select="//w" mode="conc">
     <xsl:sort select="."/>
    </xsl:apply-templates>

Attributes:

select

data-type (text, number)

order (ascending, descending)

Group exercises

4 of 12

Sort the Gorbals data by surname.
... by occupation.
... by town and county of birth.
... by age and marital status.

Grouping elsewhere

5 of 12

A common idiom:

sortedWords = words.sort("lemma");
wordPrev = null;
write("<word-entry>");
for each w in sortedWords {
    wordThis = w.getAttribute("lemma");
    if (wordThis != wordPrev) {
       write("</word-entry>\n<word-entry>");
    }
    ...
    wordPrev = wordThis;
}
write("</word-entry>\n");

Grouping in XSLT

6 of 12

Not a common idiom in functional languages!

assignment statements
we output a tree, not a stream

Three alternatives:

make list of unique items by checking for duplicates
```
[not(@lemma=preceding::w/@lemma)]
```
at cost of ca. n × (n - 1) / 2 (ouch?)

Grouping

7 of 12

Basically: sort, and then test for equality with preceding value:

 <xsl:template match="w" mode="conc">
  <xsl:variable name="type" select="."/>

  <xsl:if test="not(preceding::w[. = $type])">
   <xsl:element name="h3">
    <xsl:value-of select="."/>
   </xsl:element>
  </xsl:if>

  <!--* ... *-->
 </xsl:template>

A simple optimization

8 of 12

sort first, then check using
```
[not(@lemma=preceding-sibling::w[1]/@lemma)]
```
You probably want them sorted anyway.

Grouping

9 of 12

Or: sort, test for equality with preceding, select:

<xsl:element name="li">
 <xsl:value-of select="."/>
 <xsl:call-template name="get-scraps-by-file">
  <xsl:with-param name="fn">
   <xsl:value-of select="."/>
  </xsl:with-param>
 </xsl:call-template>
</xsl:element>

Muenchian grouping: keys

10 of 12

Called for Steve Muench (Oracle)

Define a key:

<xsl:key name="lemma-key" 
         match="w" 
         use="@lemma"/>

Test on

generate-id(.) = 
    generate-id(key('lemma-key', @lemma)[1])

or on

1 = count(. | key('lemma-key',@lemma)[1])

Group exercise: reformatting census data

11 of 12

Define key on household number.
Define unique-households variable.
For each household, select all individuals in it and process them.
Write sanity-checker to ensure we have not lost information.

Individual exercises

12 of 12

If there is time, consult the exercises page.