基于XPath值遍历跨多祖先子节点链表及XML数据遍历求助

Got it, let's break down how to tackle this linked-list XML problem. It's super common when vendors structure data as chains of parent-child references spread across different sections, and it can feel like hunting for a needle in a haystack at first.

First, let's restate your XML snippet clearly (I'll fill in the truncated part for context):

<bibliographic-data>
  <reference>
    <document-id>
      <doc-number>15492293</doc-number>
    </document-id>
  </reference>
</bibliographic-data>
<related-documents>
  <divider>
    <relation>
      <parent-doc>
        <document-id>
          <code>UC12345</code>
        </document-id>
      </parent-doc>
      <!-- Assume this relation links back to another document, forming the chain -->
    </relation>
  </divider>
</related-documents>

The core challenge here is recursively following the reference chain: starting from your initial document ID, finding its parent/related documents, then those documents' parents, and so on until you hit the end of the chain.

Solution 1: XPath 2.0+ (Recursive Function) #

If you're using an XPath 2.0+ compliant tool (like Saxon, or modern XML processors), you can define a custom recursive function to traverse the entire chain in one go:

declare function local:traverse-linked-docs($start-id as xs:string, $visited-ids as xs:string* = ()) as element()* {
  <!-- Find all nodes associated with the current ID -->
  let $current-nodes := //document-id[*[text() = $start-id]]/ancestor-or-self::*[local-name() = 'reference' or local-name() = 'relation']
  <!-- Get the parent document ID from the current nodes -->
  let $parent-id := $current-nodes//parent-doc/document-id/*[text()]/text()
  <!-- Avoid infinite loops from circular references -->
  return
    if ($parent-id and not($parent-id = $visited-ids)) then
      ($current-nodes, local:traverse-linked-docs($parent-id, ($visited-ids, $start-id)))
    else
      $current-nodes
};

<!-- Call the function with your starting document number -->
local:traverse-linked-docs('15492293')

What this does: #

• Starts with your initial ID (15492293) and finds all related <reference> or <relation> nodes
• Grabs the parent document ID from those nodes
• Recursively calls itself with the parent ID, keeping track of visited IDs to avoid circular loops
• Returns every node in the entire linked chain

Solution 2: XPath 1.0 + XSLT 1.0 (Recursive Template) #

If you're stuck with XPath 1.0 (which doesn't support custom functions), you can use an XSLT 1.0 recursive template to achieve the same result:

<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:output method="xml" indent="yes"/>

  <xsl:template name="traverse-linked-chain">
    <xsl:param name="current-id"/>
    <xsl:param name="visited-ids"/>

    <!-- Find nodes linked to the current ID -->
    <xsl:variable name="current-nodes" select="//document-id[*[text() = $current-id]]/ancestor-or-self::*[local-name()='reference' or local-name()='relation']"/>
    
    <!-- Output the current nodes -->
    <xsl:copy-of select="$current-nodes"/>

    <!-- Get the parent ID -->
    <xsl:variable name="parent-id" select="$current-nodes//parent-doc/document-id/*[text()]/text()"/>

    <!-- Recurse only if parent ID exists and hasn't been visited -->
    <xsl:if test="$parent-id and not(contains(concat(' ', $visited-ids, ' '), concat(' ', $parent-id, ' ')))">
      <xsl:call-template name="traverse-linked-chain">
        <xsl:with-param name="current-id" select="$parent-id"/>
        <xsl:with-param name="visited-ids" select="concat($visited-ids, ' ', $current-id)"/>
      </xsl:call-template>
    </xsl:if>
  </xsl:template>

  <!-- Entry point: start with your initial ID -->
  <xsl:template match="/">
    <linked-documents>
      <xsl:call-template name="traverse-linked-chain">
        <xsl:with-param name="current-id" select="'15492293'"/>
        <xsl:with-param name="visited-ids" select="''"/>
      </xsl:call-template>
    </linked-documents>
  </xsl:template>
</xsl:stylesheet>

Key Notes: #

• The visited-ids parameter is critical here—without it, you could end up in an infinite loop if the XML has circular references
• The selector //document-id[*[text() = $current-id]] is flexible enough to match either <doc-number> or <code> elements inside <document-id>, which handles variations in how your vendor stores IDs

Quick Tips for Debugging #

• Test small parts first: start by writing an XPath to just find the initial document's parent ID, then build up the recursion
• Use an XML editor with XPath testing (like Oxygen XML or VS Code with XML extensions) to validate each step
• If your XML has namespaces, don't forget to declare them in your XPath/XSLT—missing namespaces is a super common gotcha

内容的提问来源于stack exchange，提问作者Hobbyist