如何用XPath选取位于w:ins/w:del之间的指定w:r元素？

Let's break down what's off with your original XPath and build the correct one step by step, tailored for your Nokogiri workflow.

What's Wrong with Your Current Expression #

Your existing XPath //w:r[. = ' ' and preceding-sibling::w:ins and following-sibling::w:del] has two critical gaps:

• It uses preceding-sibling::w:ins and following-sibling::w:del, which match any preceding w:ins or following w:del—not the immediately adjacent sibling elements you need.
• It doesn't account for the "either/or" rule (previous/next can be w:ins or w:del), and checking . = ' ' might not reliably catch whitespace-only w:t nodes inside the w:r.

The Correct XPath Expression #

Here's the adjusted query that meets all your requirements:

//w:r[
  w:t[normalize-space(.) = '' and string-length(.) > 0]
  and preceding-sibling::*[1][self::w:ins or self::w:del]
  and following-sibling::*[1][self::w:ins or self::w:del]
]

Let's Break Down Each Part #

• w:t[normalize-space(.) = '' and string-length(.) > 0]: Ensures the w:r contains a w:t element with nothing but whitespace (handles single spaces, multiple spaces, tabs, etc.). The string-length check ensures we don't match empty w:t nodes with no content at all.
• preceding-sibling::*[1][self::w:ins or self::w:del]: Targets the immediately preceding sibling element (using *[1] to grab the closest one) and verifies it's either w:ins or w:del.
• following-sibling::*[1][self::w:ins or self::w:del]: Does the same for the immediately following sibling element.

Using This in Nokogiri #

Don't forget to register the WordprocessingML namespace when querying—Nokogiri doesn't auto-detect it by default:

require 'nokogiri'

# Load your XML content (file or string)
xml_content = File.read('your_word_doc.xml')
doc = Nokogiri::XML(xml_content)

# Register the 'w' namespace for WordprocessingML
namespace = { 'w' => 'http://schemas.openxmlformats.org/wordprocessingml/2006/main' }

# Run the XPath query
target_elements = doc.xpath("//w:r[w:t[normalize-space(.) = '' and string-length(.) > 0] and preceding-sibling::*[1][self::w:ins or self::w:del] and following-sibling::*[1][self::w:ins or self::w:del]]", namespace)

Testing Against Your Examples #

• Example 1: This will select the first and third whitespace-only w:r elements (the ones you want), and skip the second one (since its previous sibling is a regular w:r, not w:ins/w:del).
• Example 2: It won't select the whitespace w:r because its next sibling is a regular w:r, not w:ins or w:del.

内容的提问来源于stack exchange，提问作者chell