
I guess somebody at caller.com forgot to test this ColdFusion script before the election started. That, or maybe they consider divide-by-zero errors in the output a feature.

<tr height=19 style='height:14.25pt'>
<td height=19 class=xl34 style='height:14.25pt'>Kay Bailey Hutchinson-R*</td>
<td class=xl35 style='border-left:none' x:num
x:fmla="='G:\DailyArt\For the Web\electionNov06\[US reps.xls]Sheet1'!$B$4">2659119</td>
<td class=xl36 style='border-left:none' x:num="0.61691000152190612"
x:fmla="='G:\DailyArt\For the Web\electionNov06\[US reps.xls]Sheet1'!$C$4">61.7%</td>
<td class=xl40 style='border-left:none'>a</td>
<td colspan=2 style='mso-ignore:colspan'></td>
</tr>
substring() and <xsl:choose>. Sure, I could have written an external function to do so much more cleanly, but then it becomes a pain when it’s time to move from one XSLT engine to another.str:split for parsing YouTube’s tags. I’m almost certain I could have accomplished my goal with a recursive template, but I decided not to let NIH syndrome get in my way. EXSLT still doesn’t completely solve portability problems, though.
Accepts BBCode with a few enhancements.
Are they calling excel from the script? Is there some way to generate an HTML report like that from within Excel? Or did they manually embed a spreadsheet inside a Word document and save that as HTML?