Re: XPath Search

[Nigel S. <nig...@us...>]

Returns a list of features:
//feature[contains(.,'search string')]
=20
Returns a list of components
//feature[contains(.,'search string')]/..

The trailing /.. is no accident.  It filters duplicates too

Nigel
  ----- Original Message -----=20
  From: Tod McIntyre=20
  To: Nigel Swinson=20
  Cc: php...@li...=20
  Sent: Friday, February 10, 2006 8:07 PM
  Subject: Re: XPath Search


  excellent...    very thorough!  Now, my only question is...    once I =
have a list of all the features that match a particular phrase, along =
with keywords etc.  I want to then get the component that that feature =
belongs to.  I'm not exactly sure how to get the parent.  I tried some =
of XPath the axis fcn's such as "parent::feature", to no avail.  I'm not =
even really sure what the exact syntax is in order to get this to work =
anyway.  Can someone tell me how to get the component element of the =
feature that matches.  Then all I need to do is eliminate duplicates and =
I'm set!

  Thanks again for all the help,

  Brad


  On 2/7/06, Nigel Swinson <nig...@us...> wrote:
    I've narrowed the problem down.  When I search through keywords in =
the xml file (each component has one keyword tag with multiple words) it =
works fine using:
    =
$components=3D$xPath->match("//component[contains(keyword,'string')]");
    but when I search through features (each component has multiple =
features) using:
    =
$components=3D$xPath->match("//component[contains(feature,'string')]");
    I get an empty array.

    Do I have to do something different if there are multiple tags such =
as feature?

    Worked it out.  When you search for =
//component[contains(feature,'string')], then it first of all evaluates =
//component to get a list of context nodes, which will be something =
like:

    XPathSet:Array
    (
        [0] =3D> /product[1]/component[1]
        [1] =3D> /product[1]/component[6]
    )


    Then it filters the list based on the predicate, which is the =
contains(feature,'string') bit.  But to do this, it needs to evaluate =
the "feature" for each of the context nodes.  But for the first context =
node this returns something like:

    Array
    (
        [0] =3D> /product[1]/component[1]/feature[1]
        [1] =3D> /product[1]/component[1]/feature[2]
        [2] =3D> /product[1]/component[1]/feature[3]
        [3] =3D> /product[1]/component[1]/feature[4]
    )


    So the LHS is an array, and the RHS is a string.  According to =
http://www.w3.org/TR/xpath#function-string, the contains function will =
convert the LHS to a string by "A node-set is converted to a string by =
returning the string-value of the node in the node-set that is first in =
document order . If the node-set is empty, an empty string is =
returned.".  And thus it will take the text part of =
/product[1]/component[1]/feature[1], and ignore the rest.

    This explains why it works for the keyword, as you only have one =
keyword.

    Instead of finding all components, and then filtering the list, =
instead we should find all features, and then go back up a level to get =
their owning components.  So we do:

    //feature[contains(.,'search string')]/..

    Nigel