Re: [xsl] Hanging regex

[David Carlisle <davidc@xxxxxxxxx>]

On 17/11/2012 12:05, Ihe Onwuka wrote:
> First let me dissect the regex
>
> 	   <xsl:analyze-string select="." flags="x"
> 			       regex="(.+?)
> 			              ((-?\d*\s*)+$)"
>
> is targeted at lines of balance sheet text such as below  where we do
> not know how many amounts will occur
>
>    1. Total Quick Assets                              1,511
> 2,829          1,694          4,429
>
> (.+?)  lazily matches the non-financial half of the line  - in this
> case it will gobble up 1. Total Quick Assets
>
> ((-?\d*\s*)+$) captures the financial half - allowing for a leading
> minus sign - the inner brackets are for grouping not capture.
>
> Here is some test data - a file containing the following
>
>
>   I. Current Assets                                   1,871
> 2,829          1,694          4,429
>    1. Total Quick Assets                              1,511
> 2,829          1,694          4,429
>     Short-term financial instrument                      31
> 16             45              -
>    2. Total Inventories                                 359
>   -              -              -
>   II. Leased Housing Assets                               -
>   -              -              -
>   III. Deferred Liabilities                               -
>   -              -              -
>   III.Capital Adjustments                                 -
>   -            -28            -30
>   V. Retained Earnings                               -2,840
> -4,664         -4,363         -4,383
>
>
>
> **********************************************************************************************************
>
>
>
>
> FINANCIAL INFORMATION					             1. Financial Statements
>
> Income Statement
> ------------------
> 									    (Unit : KRW million)
> **********************************************************************************************************
>
> Here is the stylesheet
>
> <?xml version="1.0" encoding="UTF-8"?>
> <xsl:stylesheet
>          xmlns:xs="http://www.w3.org/2001/XMLSchema";
>          xmlns:xsl="http://www.w3.org/1999/XSL/Transform";
> 	exclude-result-prefixes="xs" version="2.0">
>    <xsl:output indent="yes"/>
>    <xsl:param name="input" as="xs:string" required="yes"/>
>
>    <xsl:template match="/">
>      <!-- read in text whilst removing comma punctuation from monetary
> fields -->
>      <xsl:for-each select="tokenize(replace(unparsed-text($input,
> 'iso-8859-1'),',(\d{3}\D)','$1'), '\r?\n')">
>
>         <!-- Delete lines that don't contain alphanumeric text -->	
>         <xsl:if test="matches(.,'\w') and matches(.,'(-|\d)+')">	
>           <line>
> 	   <xsl:analyze-string select="." flags="x"
> 			       regex="(.+?)
> 			              ((-?\d*\s*)+$)">

That's a wildly expensive regex. If I change it to

	   <xsl:analyze-string select="." flags="x"
			       regex="(.+?)
			              ((-|\d|\s)+$)">


I get identical output for your input and adding the extra line doesn't 
make it take appreciably longer


David