如何使用XSLT去除合并XML文件中的重复XML声明与根元素？

Absolutely! XSLT can absolutely solve this problem—you just need to work around the fact that your merged file isn’t well-formed XML (thanks to those duplicate declarations and root elements). Here’s a practical, self-contained solution using XSLT 3.0 (which has robust text-processing tools to handle non-well-formed input):

How it works #

We’ll break the task into four key steps:

1. Load the merged file as raw text (since we can’t parse it directly as XML).
2. Strip out all duplicate XML declarations (<?xml ...?>).
3. Split the cleaned text into individual document fragments.
4. Wrap all those fragments in a single new root element.

XSLT Code Example #

<xsl:stylesheet version="3.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:output method="xml" indent="yes" encoding="UTF-8"/>

  <!-- Main entry point (runs when no initial template is specified) -->
  <xsl:template name="xsl:initial-template">
    <!-- Load your merged input file -->
    <xsl:variable name="raw-merged-text" select="unparsed-text('your-merged-file.xml')"/>
    
    <!-- Remove all XML declarations from the text -->
    <xsl:variable name="text-without-decls" select="replace($raw-merged-text, '&lt;\?xml[^>]+?\?&gt;', '', 's')"/>
    
    <!-- Split the text into distinct document fragments (adjust regex if your roots have unique names) -->
    <xsl:variable name="document-fragments" select="tokenize($text-without-decls, '&lt;([^/][^&gt;]+)&gt;')[normalize-space(.) != '']"/>
    
    <!-- Wrap everything in a new root element -->
    <merged-documents>
      <!-- Reconstruct each original document with its root element -->
      <xsl:for-each select="tokenize($text-without-decls, '&lt;([^/][^&gt;]+)&gt;')[position() mod 2 = 0]">
        <xsl:variable name="root-tag" select="."/>
        <xsl:variable name="fragment-content" select="$document-fragments[position() = current()/position() + 1]"/>
        <xsl:value-of select="concat('&lt;', $root-tag, '&gt;', $fragment-content, '&lt;/', $root-tag, '&gt;')" disable-output-escaping="yes"/>
      </xsl:for-each>
    </merged-documents>
  </xsl:template>
</xsl:stylesheet>

Key Notes #

• XSLT 3.0 Support: This uses unparsed-text() and advanced regex, which are part of XSLT 3.0. Tools like Saxon HE/PE/EE support this natively.
• Regex Tweaks: If your original XML files all use the same root element (e.g., <record>), replace the generic split regex with something like '<record>' for better reliability.
• Disable Output Escaping: We use this to ensure the reconstructed XML elements are rendered as actual tags, not escaped text like <doc1>.

Example Input/Output #

Input (merged file):

<?xml version="1.0" encoding="UTF-8"?>
<customer>
  <id>1</id>
  <name>Alice</name>
</customer>
<?xml version="1.0" encoding="UTF-8"?>
<customer>
  <id>2</id>
  <name>Bob</name>
</customer>

Output:

<?xml version="1.0" encoding="UTF-8"?>
<merged-documents>
  <customer>
    <id>1</id>
    <name>Alice</name>
  </customer>
  <customer>
    <id>2</id>
    <name>Bob</name>
  </customer>
</merged-documents>

If you’re stuck with XSLT 1.0 (which lacks text-processing features), you’ll need to pre-process the file first (e.g., with a simple Python script or sed command) to remove XML declarations and wrap everything in a temporary root, then use basic XSLT to clean up if needed.

内容的提问来源于stack exchange，提问作者Brendon S