W3C home > Mailing lists > Public > www-ql@w3.org > April to June 2004

Re: XQuery Predicate syntax

From: Jeni Tennison <jeni@jenitennison.com>
Date: Fri, 23 Apr 2004 15:44:43 +0100
Message-ID: <472407647.20040423154443@jenitennison.com>
To: Murali Mani <mani@CS.UCLA.EDU>
Cc: www-ql@w3.org, Philippe Michiels <philippe.michiels@ua.ac.be>

Hi Murali,

> Thanks for the responses.. Thanks especially for illustrating text () and
> string () functions..

Note that text() isn't a function, it's a node test. When you do
something like:

  text() = "Adam"

this is short for:

  child::text() = "Adam"

The "child::text()" is a location path that selects all the text nodes
that are children of the context node.

On the other hand, string() is a function that takes a node as an
argument and returns its string value. When you do:

  string(.) = "Adam"

this isn't short for anything (in XPath 2.0; in XPath 1.0 it's short
for string(self::node()) = "Adam").

>> Note: The original question used the query
>> //student["Adam"]/advisor. It's not clear what XML structure that
>> is supposed to query. If it's:
>>   <student>Adam<advisor>Benny</advisor></student>
>> then you need //student[text()[1] = "Adam"]/advisor (test the first
>> child text node against the string), but this is badly designed XML
>> (mixed content for data-oriented information), so I assume that
>> this isn't what was wanted.
> The above was the example in mind, even I believe it is bad design
> for "data oriented applications"; but that was not the concern..
> Do we need text ()[1] = "Adam", isn't text () = "Adam" sufficient?

It depends what you want to test and how much you can guarantee about
the structure of your document. If you do text() = "Adam" then this
will test all the text node children of the context node against the
string "Adam". If you know (somehow) that there's only one text node
child of your <student> element then obviously this will always give
exactly the same result as text()[1] = "Adam". On the other hand,
text()[1] might give you unanticipated results sometimes, for example
if someone does:

  <student> <?editor cursor?>Adam<advisor>Benny</advisor></student>

then text()[1] will give you the space before the processing

These problems are precisely why it isn't a good idea to use mixed
content for data like this.
> I thought a predicate expression such as
> (x op y) is true, if (x op y') is true for some y' in y... Am I right??

You're right that general comparisons (=, !=, <, <=, > and >=) are
existential: they do all possible pairs of comparisons between the
items in the operand sequences and return true if any one of those
comparisons are true. This isn't the case for all operators: the value
comparisons (eq, ne, lt, le, gt and ge) give you an error if either
argument is a sequence containing more than one item.

> what is the semantics of (x op y) if x is a collection? is it valid
> for x to be a collection?

Do you mean "sequence"? Yes, x can be a sequence.



Jeni Tennison
Received on Friday, 23 April 2004 11:19:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:17:16 UTC