Altova Mailing List Archives

Re: [xsl] Displaying Code Dependent on First Encounter of Specific Reference

From: Wendell Piez <wapiez@---------------->
Date: 10/14/2004 9:02:00 AM
Hi Joe,

This problem is a bit difficult, not because of any inherent difficulty in 
any of the methods you will use, but because it's both complex, and will 
require using a couple of XSLT 1.0 tricks. Disentangling it shows a way 

You actually have several problems here:

* Assigning each table or figure its correct number (based on order of 
citation, not

  order of appearance in the source)

* Citing the tables or figures where xrefs appear in line, each with its 
correct number

* Placing the tables and figures each after the paragraph where it's first 
cited, and

  not elsewhere

You will use keys for this, though not perhaps in exactly the way you're 
imagining. Likewise, it'd be nice if we could construct an array, and in 
XSLT 2.0 we could (or at least the functional equivalent thereof), but in 
XSLT 1.0 we can't. So we have to fake a couple of things. As you'll see, 
this faking may potentially get us into a bit of trouble with performance. 
The usual XSLT 1.0 approach when this happens is to split a problem into 
two or more passes, which generally gives us opportunities to optimize for 

In order to simplify this explanation I'm going to assume you have only 
tables. Figures will work just the same:

Assigning each table or figure its number ... we could either do this by 
counting the references (filtering out for repeated references) or by 
counting the tables, sorted by their first reference. While the latter 
would be nice, the former is easier in XSLT 1.0. We do this first by giving 
us a means to filter out the repeats:

<xsl:key name="tablerefs-by-rid" match="xref[@ref-type='table']" use="@rid"/>

Given $rid, we can then get all the references to any table by calling

key('tablerefs-by-rid', $rid)

and the first one only by calling

key('tablerefs-by-rid', $rid)[1]

In addition, we can get all the first references by saying, e.g.

//xref[@ref-type='table'][count(.|key('tablerefs-by-rid', @rid)[1])=1]

This XPath traverses the entire document from the root, collecting all 
xrefs that are the first reference to their table. If they aren't to a 
table, the first predicate filters them out. If they are not a first 
reference, the count of their union with the first reference will be 2 not 
1, and the second predicate will filter them out.

This uses an XPath 1.0 idiom (the count() trick) to test node identity. The 
generate-id() function is also sometimes used for this, so this would also 

//xref[generate-id()=generate-id(key('tablerefs-by-rid', @rid)[1])]

Notice here we don't need the first predicate (since xrefs not to tables 
will also be thrown out by the predicate given -- so you may prefer this.

It would be very convenient to have all these particular nodes collected 
together so we don't have to collect them over and over (an expensive 
traversal). So:

<xsl:variable name="first-table-refs"


(This is awfully close to an array, isn't it?)

Consequently, we can also get the proper number for any given 
xref[@ref-type='table'] with the expression

  [count(.|current()/preceding::xref) = count(current()/preceding::xref)]) + 1

which looks, and is, awfully obnoxious and expensive (using the costly 
preceding:: axis twice), but which can be optimized slightly as a template 

<xsl:template match="xref" mode="assign-table-number">

  <xsl:for-each select="key('tablerefs-by-rid', @rid)[1]">

    <!-- switching context to the first reference to this reference's 
table -->

    <xsl:variable name="preceding-refs" select="preceding::xref"/>

    <xsl:value-of select="count($first-table-refs

      [count(.|$preceding-refs) = count($preceding-refs)]) + 1"/>

    <!-- counting the first table references before this one, and adding 1 -->



I wish this were easier, but in XSLT 1.0 it just isn't. In 2.0, it is (and 
maybe Mike or Jeni or someone will show us how).

But it does solve problem 1, and you can see how any given 
xref[@ref-type='table'] can call <xsl:apply-templates select="." 
mode="assign-table-number"/> and get its number, thereby solving problem 2.

Problem 3 is a matter of selecting, after you create a paragraph, those 
references in it that are first references to their targets (tables, 
figures, what not), which again you can do (in the case of tables) using 
this same idiom:

<xsl:template match="para">




  <xsl:apply-templates mode="get-target"


  <!-- do the same with any other keys for xrefs you have, e.g. to 
figures, perhaps

       unifying the select -->


To actually get the target you're going to need another key:

<xsl:key name="target-by-rid" match="table|figure" use="@id"/>

and then

<xsl:template match="xref" mode="get-target">
  <xsl:apply-templates select="key('target-by-rid', @rid)"

which will go apply templates to the table, figure or whatever. Note I've 
put this call also in a special mode, "show", enabling you to say in the 
default mode

<xsl:template match="table|figure"/>

so the tables only come out where you actually want them.

Whew! not bad for a bit of work, eh?

This should work fine for input at the scale of most human-readable 
documents. For higher performance (that numbering is a beast), you'll want 
to split out an analytic/sorting pass before processing, or get out the big 
rotary saw (XSLT 2.0).

Note: I just typed this up, and haven't tested, but I have used such code 
and it works. Beware particularly of missing parentheses in my XPaths, etc.


At 04:57 PM 10/13/2004, you wrote:
XSL list:

I'm developing a stylesheet that converts XML to html to display research 
articles. The articles contains three citation types, bibliographical, 
table call, and figure call. Upon encountering a table call or figure 
call, I would like to display the table or figure referred to immediately 
following the paragraph that contains the call. I want the table or figure 
to appear in the order they were referred to in the paragraph and I want 
each table or figure to only appear once in the outputted document. Tables 
and figures are numbered in order of their reference, though at any point 
you can refer to a table or figure that has been previously called.

Citations look like this:
<xref ref-type="bibr" rid="B1">1</xref>
<xref ref-type="table" rid="T1">Table 1</xref>
<xref ref-type="fig" rid="F1">Figure 1</xref>

Sample Input:

[A paragraph that includes a citation for Table 1.]

[A paragraph that includes citations for Table 2, Table 1, Figure 1, and 
Table 3.]

Sample Output:

[A paragraph that includes a citation for Table 1.]

Table 1

[A paragraph that includes citations for Table 2, Table 1, Figure 1, and 
Table 3.]

Table 2
Figure 1
Table 3

 My initial thought is to create a set of keys:
Key: Last Table Processed
Key: Last Figure Processed
Key: Last Table Encountered
Key: Last Figure Encountered

Since the tables and figures are numbered in order, a comparison of the 
two keys should be in order. This comparison should be made at the end of 
processing a paragraph. However, I'm not quite sure how I'd make such a 
comparison or even if I can use keys in that manner. I'm thinking I might 
need to generate some sort of array to keep track of the multiple 
citations encountered so that in the sample provided the output is (Table 
2, Figure 1, Table 3) and not (Table 2, Table 3, Figure 1) or (Figure 1, 
Table 2, Table 3). If I were to build an array, since at this point I 
don't need to process <xref> citations of "bibr" type, those should be 
ignored. Any suggestions would be greatly appreciated. Thanks.

> - Joe

Wendell Piez                            mailto:wapiez@xxxxxxxxxxxxxxxx
Mulberry Technologies, Inc.      
17 West Jefferson Street                    Direct Phone: 301/315-9635
Suite 207                                          Phone: 301/315-9631
Rockville, MD  20850                                 Fax: 301/315-8285
  Mulberry Technologies: A Consultancy Specializing in SGML and XML


These Archives are provided for informational purposes only and have been generated directly from the Altova mailing list archive system and are comprised of the lists set forth on Therefore, Altova does not warrant or guarantee the accuracy, reliability, completeness, usefulness, non-infringement of intellectual property rights, or quality of any content on the Altova Mailing List Archive(s), regardless of who originates that content. You expressly understand and agree that you bear all risks associated with using or relying on that content. Altova will not be liable or responsible in any way for any content posted including, but not limited to, any errors or omissions in content, or for any losses or damage of any kind incurred as a result of the use of or reliance on any content. This disclaimer and limitation on liability is in addition to the disclaimers and limitations contained in the Website Terms of Use and elsewhere on the site.