How can I select an element with multiple classes with Xpath?

Question

In the above xml sample I would like to select all the books that belong to class foo and not in class bar by using xpath.

<?xml version="1.0" encoding="ISO-8859-1"?>
<bookstore>
  <book class="foo">
    <title lang="en">Harry Potter</title>
    <author>J K. Rowling</author>
    <year>2005</year>
    <price>29.99</price>
  </book>
  <book class="foo bar">
    <title lang="en">Harry Potter</title>
    <author>J K. Rowling</author>
    <year>2005</year>
    <price>29.99</price>
  </book>
  <book class="foo bar">
    <title lang="en">Harry Potter</title>
    <author>J K. Rowling</author>
    <year>2005</year>
    <price>29.99</price>
  </book>
</bookstore>

Good question, +1. See my answer for two different XPath 2.0 solutions of which the first might be the most efficient of them all especially with a non-optimizing XPath 2.0 engine. — Dimitre Novatchev, Apr 17 '11 at 01:11

score 39 · Accepted Answer · edited Feb 28 '15 at 17:04

39

By padding the @class value with leading and trailing spaces, you can test for the presence of " foo " and " bar " and not worry about whether it was first, middle, or last, and any false positive hits on "food" or "barren" @class values:

/bookstore/book[contains(concat(' ',@class,' '),' foo ')
        and not(contains(concat(' ',@class,' '),' bar '))]

edited Feb 28 '15 at 17:04

Shimmy Weitzhandler

101,809
122
424
632

answered Apr 14 '11 at 11:22

Mads Hansen

63,927
12
112
147

1

What if `@class` contains tab or even new-line character instead of space. Here comes handy the `normalize-space` function (XPath 1.0) that strips the leading and trailing white-space from a string, replaces sequences of whitespace characters by a single space, e.g. `concat(' ',normalize-space(@class),' ')` – Steven Pribilinskiy Mar 01 '15 at 09:27
@Steven Pribilinskiy - That should not be necessary. Due to how attribute values are normalized by the XML parser, tabs and carriage returns will have already been normalized into a space. http://www.w3.org/TR/xml/#AVNormalize – Mads Hansen Mar 01 '15 at 15:48

score 11 · Answer 2 · answered Apr 14 '11 at 11:55

11

Although I like Mads solution: Here is another approach for XPath 2.0:

/bookstore/book[
                 tokenize(@class," ")="foo" 
                 and not(tokenize(@class," ")="bar")
               ]

Please note that the following expressions are both true:

("foo","bar")="foo" -> true
("foo","bar")="bar" -> true

answered Apr 14 '11 at 11:55

Dennis Münkle

5,036
1
19
18

+1 for the XPath 2.0 solution. So many things are easier with 2.0. – Mads Hansen Apr 14 '11 at 11:58

Dimitre Novatchev · Answer 3 · 2011-04-17T01:09:19.723

4

XPath 2.0:

/*/*[for $s in concat(' ',@class,' ') 
            return 
               matches($s, ' foo ') 
             and 
              not(matches($s, ' bar '))
      ]

Here no tokenization is done and $s is calculated only once.

Or even:

/*/book[@class
          [every $t in tokenize(.,' ') satisfies $t ne 'bar']
          [some  $t in tokenize(.,' ') satisfies $t eq 'foo']
       ]

edited Apr 17 '11 at 01:09

answered Apr 17 '11 at 00:40

Dimitre Novatchev

240,661
26
293
431

How can I select an element with multiple classes with Xpath?

3 Answers3

Linked