请求优化XSLT实现文本到XML的结构化转换

Alright, let's refine your XSLT 2.0 code to hit all three of your requirements perfectly. Here's the optimized solution with breakdowns of how each part works:

<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xs="http://www.w3.org/2001/XMLSchema" exclude-result-prefixes="xs">
    <xsl:output indent="yes" method="xml"/>
    <xsl:strip-space elements="*"/>

    <xsl:param name="txt-encoding" as="xs:string" select="'iso-8859-1'"/>
    <xsl:param name="txt-uri" as="xs:string" select="'linktofile'"/>

    <!-- 字段映射：源字段名 → 目标XML元素名（按需扩展） -->
    <xsl:variable name="field-mappings" as="element()*">
        <map source="REF FOU" target="REF_VEN"/>
        <map source="GENCOD" target="EAN"/>
    </xsl:variable>

    <xsl:template match="/" name="text2xml">
        <xsl:variable name="txt" select="unparsed-text($txt-uri, $txt-encoding)"/>
        
        <!-- 拆分整个文本为单个订单块（匹配[ENTETE]到[FIN]的完整订单内容） -->
        <xsl:analyze-string select="$txt" regex="\[ENTETE\](.*?)(\[LIGNE\].*?)*\[FIN\]" flags="s">
            <xsl:matching-substring>
                <ORDER>
                    <!-- 处理ENTETE部分的键值对，生成顶层XML元素 -->
                    <xsl:analyze-string select="regex-group(1)" regex="(\S+(\s\S+)*)\s*=\s*([^=]+?)(?=\s+\S+\s*=|\s*$)">
                        <xsl:matching-substring>
                            <xsl:variable name="field-name" select="normalize-space(regex-group(1))"/>
                            <!-- 把字段名里的空格替换成下划线 -->
                            <xsl:variable name="element-name" select="replace($field-name, ' ', '_')"/>
                            <xsl:element name="{$element-name}">
                                <xsl:value-of select="normalize-space(regex-group(3))"/>
                            </xsl:element>
                        </xsl:matching-substring>
                    </xsl:analyze-string>

                    <!-- 处理所有LIGNE段落，转换为<LINE>子元素 -->
                    <xsl:analyze-string select="." regex="\[LIGNE\](.*?)(?=\[LIGNE\]|\[FIN\])" flags="s">
                        <xsl:matching-substring>
                            <LINE>
                                <xsl:analyze-string select="regex-group(1)" regex="(\S+(\s\S+)*)\s*=\s*([^=]+?)(?=\s+\S+\s*=|\s*$)">
                                    <xsl:matching-substring>
                                        <xsl:variable name="source-field" select="normalize-space(regex-group(1))"/>
                                        <!-- 优先使用字段映射，没有映射则用空格转下划线的名称 -->
                                        <xsl:variable name="target-element" select="$field-mappings[@source = $source-field]/@target"/>
                                        <xsl:variable name="element-name" select="if ($target-element) then $target-element else replace($source-field, ' ', '_')"/>
                                        <xsl:element name="{$element-name}">
                                            <!-- 清理REF CLI字段的引号和多余空格 -->
                                            <xsl:value-of select="normalize-space(replace(regex-group(3), '&quot;', ''))"/>
                                        </xsl:element>
                                    </xsl:matching-substring>
                                </xsl:analyze-string>
                            </LINE>
                        </xsl:matching-substring>
                    </xsl:analyze-string>
                </ORDER>
            </xsl:matching-substring>
        </xsl:analyze-string>
    </xsl:template>
</xsl:stylesheet>

How each requirement is fulfilled: #

1. Replace spaces in header fields with underscores

   • We use replace($field-name, ' ', '_') to swap out spaces for underscores in field names (e.g., CODE VEND becomes CODE_VEND).
   • For LIGNE-specific fields that need a full rename (like REF FOU → REF_VEN), we added a field-mappings variable to handle those custom mappings cleanly.

2. Group header data by NO_CDE

   • First, we split the entire input text into individual order blocks using the regex \[ENTETE\](.*?)(\[LIGNE\].*?)*\[FIN\]. Each block represents one order, naturally grouped by its unique NO_CDE value.
   • Each order block gets its own <ORDER> element, containing all the matching header fields and LINE items.

3. Convert multiple [LIGNE] sections to child elements

   • We use another regex \[LIGNE\](.*?)(?=\[LIGNE\]|\[FIN\]) to capture each standalone LIGNE block.
   • Each captured block is converted into a <LINE> element, with its internal fields processed the same way as header fields (plus cleaning up quotes from REF CLI values).

Extra tweaks: #

• Improved regex matching to avoid cross-field value capture, making the key-value parsing more reliable.
• Added quote removal for REF CLI to match your target XML format exactly.
• The field mapping variable is easy to extend if you need to rename more fields later.

内容的提问来源于stack exchange，提问作者Komagain