XPath doesn't match when desired element contains child elements?

Question

I have the following XPath expression:

//a[@attribute='my-attribute']

When I have the following element in the HTML that XPath is searching, it matches as expected:

<a attribute="my-attribute">Some text</a>

But if there is an <svg> tag under that element, XPath returns no match:

<a attribute="my-attribute">
    <svg xmlns="http://www.w3.org/2000/svg" width="100%" height="100%"
        viewBox="0 0 24 24" focusable="false"></svg>
</a>

Why doesn't XPath match in this case? Is there a way I can modify my expression to make it match?

EDIT:

Apparently it has to do with the namespace on the <svg> element. Using the local-name() function makes it match in the XPath tester I'm using:

//*[local-name()='a' and @attribute='my-attribute']

However, this still doesn't match when running through Selenium WebDriver. Any idea of how to get this working with Selenium?

Is there any specific question for us to answer or you are looking for generic ideas? — undetected Selenium, Mar 27 '19 at 21:16
What is the selenium outcome with `//*[local-name()='a' and @attribute='my-attribute']` when you have only `a` tag without `svg`. — supputuri, Mar 27 '19 at 21:22
It has nothing to do with *namespaces* and *child `svg` element* as long as you're trying to select anchor node — JaSON, Mar 27 '19 at 21:40
@JaSON I would have thought the same thing, but removing the `xmlns` attribute from the `svg` tag causes the `//a[@attribute='my-attribute']` expression to match. — Andrew Mairose, Mar 28 '19 at 13:07
As you can see, your problem cannot be reproduced in http://www.xpathtester.com/xpath/91e66f48ea100183e9e3b1958ceed7b5 — Alejandro, Mar 28 '19 at 22:23

kjhughes · Answer 1 · 2019-03-27T22:40:35.523

3

You may be confused by how the XPath hosting environment is presenting the selected a elements.

Adding an svg element to the a element will not affect what's selected by

//a[@attribute='my-attribute']

In the case of

<a attribute="my-attribute">Some text</a>

the a element has a string value consisting of more than just white space characters, but with

<a attribute="my-attribute">
    <svg xmlns="http://www.w3.org/2000/svg" width="100%" height="100%"
        viewBox="0 0 24 24" focusable="false"></svg>
</a>

the a element has a string value that consists only of whites space, so for text results of the selection, you wouldn't see anything selected.

If you evaluate count(//a[@attribute='my-attribute']), you'll likely see the same results for both cases.

edited Mar 27 '19 at 22:40

answered Mar 27 '19 at 22:30

kjhughes

106,133
27
181
240

From a generic perspective this answer should address OP's concerns. – undetected Selenium Mar 28 '19 at 06:13
@kjhughes I'm not trying to select the text of the `a` element. I'm trying to select the `a` element itself. Adding the `svg` element on its own doesn't affect what is selected by the `//a[@attribute='my-attribute']`, but it is the `xmlns` attribute on the `svg` element. Removing the `xmlns` attribute causes the xpath expression to match the `a` tag, but with the `xmlns` on the `svg` element, the expression yields no results. – Andrew Mairose Mar 28 '19 at 13:03
Adding an `svg` element ***does*** affect what's selected by that expression if the `svg` element has a `xmlns` attribute. I have verified this in an XPath tester. That is the entire reason for me asking this question. – Andrew Mairose Mar 28 '19 at 13:11
**No,** that would not happen. You've omitted some important detail, and the burden is on you to provide a true [mcve] that exhibits the impossible results you're describing. You will likely find your mistake in the course of producing such an MCVE because you won't be able to construct a ***complete*** and verifiable example that behaves as you describe. – kjhughes Mar 28 '19 at 13:19

score 0 · Answer 2 · answered Sep 10 '19 at 16:46

Following is a possible solution in vb.net.

Public Class XmlNodeListWithNamespace
    ' see https://stackoverflow.com/questions/55385520/xpath-doesnt-match-when-desired-element-contains-child-elements
    ' @JaSON I would have thought the same thing,
    ' but removing the xmlns attribute from the svg tag
    ' causes the //a[@attribute='my-attribute'] expression to match.
    ' – Andrew Mairose
    ' Mar 28 at 13:07 "Asked 5 months ago  Active 5 months ago" implies 2019-03-28 13:07.

    ' Therefore, I first considered deleting all occurrences of
    '       xmlns="" and xmlns="http://www.w3.org/1999/xhtml"
    ' I did this using the following Replacement.
    ' gstrHtml = Regex.Replace(
    '            input:=gstrHtml,
    '            pattern:=" *xmlns=""[^""]*""",
    '            replacement:="",
    '            options:=RegexOptions.IgnoreCase
    '        )

    ' However, the solution below retains the namespace, while avoiding unsightly xpath strings.

    ''' <summary>
    ''' For a given xpath, returns an XmlNodeList, taking account of the xmlns namespace.
    ''' </summary>
    ''' <param name="oXmlDocument">The current XML document.</param>
    ''' <param name="xpath">A normal xpath string, without any namespace qualifier.</param>
    ''' <returns>The XmlNodeList for the given xpath.</returns>
    Public Shared Function NodeList(
        oXmlDocument As XmlDocument,
        xpath As String
    ) As XmlNodeList

        Dim strXpath As String = xpath

        ' Insert Namespace Qualifier.  For example, 
        '    "//pre"                                            becomes "//x:pre"
        '    "/html/body/form/div/pre"                          becomes "/x:html/x:body/x:form/x:div/x:pre"
        '    "//div[@id='nv_bot_contents']/pre"                 becomes "//x:div[@id='nv_bot_contents']/x:pre"
        '    "//div[@id='nv_bot_contents']/pre[@data-xxx='X2']" becomes "//x:div[@id='nv_bot_contents']/x:pre[@data-xxx='X2']"
        '    "//div[@id='nv_bot_contents']/pre[@data-xxx]"      becomes "//x:div[@id='nv_bot_contents']/x:pre[@data-xxx]"
        '    "//pre[@data-xxx]"                                 becomes "//x:pre[@data-xxx]"
        strXpath = Regex.Replace(
                        input:=strXpath,
                        pattern:="(/)(\w+)",
                        replacement:="$1x:$2"
                    )

        ' See https://stackoverflow.com/questions/40796231/how-does-xpath-deal-with-xml-namespaces/40796315#40796315
        Dim oXmlNamespaceManager As New XmlNamespaceManager(nameTable:=oXmlDocument.NameTable)
        oXmlNamespaceManager.AddNamespace("x", "http://www.w3.org/1999/xhtml")

        Dim oXmlNodeList As XmlNodeList = oXmlDocument.SelectNodes(
            xpath:=strXpath,
            nsmgr:=oXmlNamespaceManager
        )

        Return oXmlNodeList

    End Function

End Class

Sample invocation:

Dim oXmlNodeList As XmlNodeList =
            XmlNodeListWithNamespace.NodeList(
                oXmlDocument:=oXmlDocument,
                xpath:="//pre"
            )

XPath doesn't match when desired element contains child elements?

2 Answers2