Split node value with XPath

Question

Is there some kind of split() function in XPath? Say I have this XML:

<root>
   <path>C:\folder\filename</path>
</root>

And I want to retrieve filename, how can I do this? I know I can get the node value like this:

//path/text()

How can I get only the filename? (I know there is a concat() function, so maybe there is a split() function?)

Good question, +1. Besides the good answers you've got, see my answer for two other possible solutions. — Dimitre Novatchev, Nov 23 '10 at 17:00

score 9 · Accepted Answer · edited Nov 24 '19 at 13:46

9

If you have an xpath-2.0 capable API, you can solve this in two ways:

replace technique

Try using:

fn:replace(string,pattern,replace)

e.g.

fn:replace(//path/text(),".*/","")

tokenize technique

You may get some mileage from tokenize:

fn:tokenize(string,pattern)

e.g. (thanks to Martin)

tokenize(/root/path, '\\')[last()]

w3schools xml processing "xsl functions" documentation

edited Nov 24 '19 at 13:46

Vogel612

5,620
5
48
73

answered Nov 23 '10 at 10:47

Alex Brown

41,819
10
94
108

3

Yes, `tokenize(/root/path, '\\')[last()]` allows that but note that both replace and tokenize are XPath 2.0 and not available in XPath 1.0 so you need an XPath 2.0 implementation like Saxon 9 (http://saxon.sourceforge.net/). – Martin Honnen Nov 23 '10 at 11:37
Thanks, I couldn't work out whether last applied to dynamically generated collections. – Alex Brown Nov 23 '10 at 11:39

score 1 · Answer 2 · answered Nov 23 '10 at 16:56

While I would use:

tokenize(/*/*, '\\')[last()]

there are also numerous other ways to obtain the desired string:

  codepoints-to-string
    (reverse
      (string-to-codepoints
         (substring-before
            (codepoints-to-string
                 (reverse
                    (string-to-codepoints(/*/*)
                  )
              ),
              '\'
            )
          )
       )
     )

Or:

  substring(/*/*,
            index-of(string-to-codepoints(/*/*),
            string-to-codepoints('\')
            )
            [last()]
          + 1
           )

Split node value with XPath

2 Answers2