I'm working on scraping HTML pages and transforming the selected content from them into XML and I've been seeing a few errors here and there thrown by the Saxon XSLT parser:

WARN [2011-01-07 14:44:35 EST] (SaxonTranslator#429473) improper java log: XPTY0004: A sequence of more than one item is not allowed as the first argument of
WARN [2011-01-07 14:44:35 EST] (SaxonTranslator#429473) improper java log: matches() ("<!--//<![CDATA[ var m...", "<!--//<![CDATA[ va...", ...)
ERROR [2011-01-07 14:44:35 EST] (SaxonTranslator#429473) error while processing file 'ffd827cfd0f59a793ba99a309b392e40.substituted': Saxon error...A sequence of more than one item is not allowed as the first argument of matches() ("<!-- google_ad_client = "ca-...", "", ...) (~/sandbox/core_plugins/JavaPlugin.rb:131:in `execute')

The same with type of error with ("<!--//<![CDATA[ var m...", "<!--//<![CDATA[ va...", ...)

Also one with ("...", "", ...)

Here is the bit of the file where I believe the problem is occurring:

<?xml version="1.0" encoding="ISO-8859-1"?>
<xsl:stylesheet
  xmlns:exsl="http://exslt.org/common"
  xmlns:str="http://exslt.org/strings"
  xmlns:date="http://exslt.org/dates-and-times"
  xmlns:xs="http://www.w3.org/2001/XMLSchema"
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:html="http://www.w3.org/1999/xhtml"
  extension-element-prefixes="exsl str date"
  version="2.0">
  <xsl:import href="utils.xslt"/>
  <xsl:output method="xml" indent="yes"/> 
  <xsl:param name="feed" select="rand"/>
  
  <xsl:template match="/">

    <!-- Regular expression to find dates in the file -->
    <xsl:variable name="date_regex" 
      select="'((Jaunary|February|March|April|May|June|July|August|September|October|November|December) \d?\d, \d\d\d\d)'"/>

    <p:Report>
      <p:PublicationMetadata>
        <p:AdministrativeMetadata>

          <!-- DATE -->
          <p:DateList>

            <xsl:choose>
              <xsl:when test="matches(//body, $date_regex)">

                <!-- Sometimes the text of an article contains a date.  To prevent the extra -->
                <!-- date from being included in the output, we save the regex results to a -->
                <!-- sequence-valued variable, then use only the first entry. -->
                <xsl:variable name="regex_result" as="xs:string*">
                  <xsl:analyze-string regex="{$date_regex}" select="//body">
                    <xsl:matching-substring>
                      <xsl:sequence select="."/>
                    </xsl:matching-substring>
                  </xsl:analyze-string>
                </xsl:variable>

                <p:DatePublished>
                  <xsl:call-template name="FormatDate">
                    <xsl:with-param name="docDate" select="$regex_result[1]"/>
                  </xsl:call-template>
                </p:DatePublished>
                <p:DatePosted>
                  <xsl:call-template name="FormatDate">
                    <xsl:with-param name="docDate" select="$regex_result[1]"/>
                  </xsl:call-template>
                </p:DatePosted>

              </xsl:when>

              <xsl:otherwise>
                <!-- Use current date if the document doesn't have a date field -->
                <!-- Returns format yyyy-mm-dd -->
                <xsl:variable name="curdate">
                  <xsl:value-of select="date:date()"/>
                </xsl:variable>
                
                <p:DatePublished>
                  <xsl:value-of select="$curdate"/>
                  <xsl:text>T12:00:00Z</xsl:text>
                </p:DatePublished>
                <p:DatePosted>
                  <xsl:value-of select="$curdate"/>
                  <xsl:text>T12:00:00Z</xsl:text>
                </p:DatePosted>
                
              </xsl:otherwise>
              
            </xsl:choose>
          </p:DateList>
          
      <!-- ...Rest of the file... -->

</xsl:stylesheet>

The error is pretty self explanatory. The matches functions takes a single supplied STRING and tests if it matches a certain regex format. anytime you call the matches() functions you can only put a single string in the input. "matches(//body, $date_regex)" it's VERY unlikely that there's 1 string in the body node of an entire document. You'll have to drill down specifically to 1 and only 1 string before you use the matches() function.

Some of the input files had multiple body elements. They were doing stuff like this in the same file:

<html>
<head> ... </head>
<body>
...
<head> ... </head>
<body>
...
</body>
</html>
</td>
...
<html>
<body>
...
</body>
</html>
<script type="text/javascript"> ...
</body>
</html>

So the xpath expression was returning multiple elements which breaks the matches() function. It's just a matter of getting the right xpath to deal with that kind of html. What worked to resolve the errors was making the following change:
<xsl:when test="matches(/body[1], $date_regex)">

Edited 5 Years Ago by weloki: n/a

Yep, a predicate of [1] will get you the first "body" element that returns. That would solve the problem :) Glad you figured it out.

This question has already been answered. Start a new discussion instead.