I have XML documents like:
<rootelement>
<myelement>test1</myelement>
<myelement>test2</myelement>
<myelement type='specific'>test3</myelement>
</rootelement>
I'd like to retrieve the specific myelement, and if it's not present, then the first one. So I write:
/rootelement/myelement[#type='specific' or position()=1]
The XPath spec states about the 'or expression' that:
The right operand is not evaluated if
the left operand evaluates to true
The problem is that libxml2-2.6.26 seems to apply the union of both expressions, returning a "2 Node Set" (for example using xmllint --shell).
Is it libxml2 or am I doing anything wrong ?
Short answer: your selector doesn't express what you think it does.
The or operator is a union.
The part of the spec you quoted ("The right operand is not evaluated...") is part of standard boolean logic short circuiting.
Here's why you get a 2-node set for your example input: XPath looks at every myelement that's a child of rootelement, and applies the [#type='specific' or position()=1] part to each such node to determine whether or not it matches the selector.
<myelement>test1</myelement> does not match #type='specific', but it does match position()=1, so it matches the whole selector.
<myelement>test2</myelement> does not match #type='specific', and it also does not match position()=1, so it does not match the whole selector.
<myelement type='specific'>test3</myelement> matches #type='specific' (so XPath does not have to test its position - that's the short-circuiting part) so it matches the whole selector.
The first and last <myelement>s match the whole selector, so it returns a 2-node set.
The easiest way to select elements the way you want to is to do it in two steps. Here's the pseudocode (I don't know what context you're actually using XPath in, and I'm not that familiar with writing XPath-syntax selectors):
Select elements that match /rootelement/myelement[#type='specific']
If elements is empty, select elements that match /rootelement/myelement[position()=1]
#Matt Ball explained very well the cause of your problem.
Here is an XPath one-liner selecting exactly what you want:
/*/myelement[#type='specific'] | /*[not(myelement[#type='specific'])]/myelement[1]
Related
I have an element with three occurences on the page. If i match it with Xpath expression //div[#class='col-md-9 col-xs-12'], i get all three occurences as expected.
Now i try to rework the matching element on the fly with
substring-before(//div[#class='col-md-9 col-xs-12'], 'Bewertungen'), to get the string before the word "Bewertungen",
normalize-space(//div[#class='col-md-9 col-xs-12']), to clean up redundant whitespaces,
normalize-space(substring-before(//div[#class='col-md-9 col-xs-12'] - both actions.
The problem with last three expressions is, that they extract only the first occurence of the element. It makes no difference, whether i add /text() after matching definition.
I don't understand, how an addition of normalize-space and/or substring-before influences the "main" expression in the way it stops to recognize multiple occurences of targeted element and gets only the first. Without an addition it matches everything as it should.
How is it possible to adjust the Xpath expression nr. 3 to get all occurences of an element?
Example url is https://www.provenexpert.com/de-de/jazzyshirt/
The problem is that both normalize-space() and substring-before() have a required cardinality of 1, meaning can only accept one occurrence of the element you are trying to normalize or find a substring of. Each of your expressions results in 3 sequences which these two functions cannot process. (I probably didn't express the problem properly, but I think this is the general idea).
In light of that, try:
//div[#class='col-md-9 col-xs-12']/substring-before(normalize-space(.), 'Bewertung')
Note that in XPath 1.0, functions like substring-after(), if given a set of three nodes as input, ignore all nodes except the first. XPath 2.0 changes this: it gives you an error.
In XPath 3.1 you can apply a function to each of the nodes using the apply operator, "!": //div[condition] ! substring-before(normalize-space(), 'Bewertung'). That returns a sequence of 3 strings. There's no equivalent in XPath 1.0, because there's no data type in XPath 1.0 that can represent a sequence of strings.
In XPath 2.0 you can often achieve the same effect using "/" instead of "!", but it has restrictions.
When asking questions on StackOverflow, please always mention which version of XPath you are using. We tend to assume that if people don't say, they're probably using 1.0, because 1.0 products don't generally advertise their version number.
I was just wondering if there is a shorter way of writing an XPath query to find all HREF values containing at least one of many search values?
What I currently have is the following:
//a[contains(#href, 'value1') or contains(#href, 'value2')]
But it seems quite ugly, especially if I were to have more values.
First of all, in many cases you have to live with the "ugliness" or long-windedness of expressions if only XPath 1.0 is at your disposal. Elegance is something introduced with version 2.0, I'd daresay.
But there might be ways to improve your expression: Is there a regularity to the href attributes you'd like to find? For instance, if it is sufficient as a rule to say that the said href attribute values must start with "value", then the expression could be
//a[starts-with(#href,'value')]
I know that "value1" and "value2" are most probably not your actual attribute values but there might be something else that uniquely identifies the group of a elements you're after. Post your HTML input if this is something you want us to help you with.
Personally, I do not find your expression ugly. There is just one or operator and the expression is quite short and readable. I take
if I were to have more values.
to mean that currently, there are only two attribute values you are interested in and that your question therefore is a theoretical one.
In case you're using XPath 2 and would like to have exact matches instead of also matches only containing part of a search value, you can shorten with
//a[#href = ('value1', 'value2')]
For contains() this syntax wouldn't work as the second argument of contains() is only allowed to be 0 or 1 value.
In XPath 2 you could also use
//a[some $s in ('value1', 'value2') satisfies contains(#href, $s)]
or
//a[matches(#href, "value1|value2")]
If I have two XPath queries where the second one is meant to further drill down the result of the first, can I safely let my script combine them into a single query by...
placing parenthesis around the first query,
prefixing the second query with with a slash, and then
simply concatenating the two strings ?
Context
The concrete usecase that sparked this question involves extracting information from XML/XHTML documents according to externally supplied pairs of "CSS selector + attribute name" using XPath behind the scenes.
For example the script may get the following as input:
selector: a#home, a.chapter
attribute: href
It then compiles the selector to an XPath query using the HTML::Selector::XPath Perl module, and the attribute by simply prefixing a # ... which in this case would yield:
XPath query 1: //a[#id='home'] | //a[contains(concat(' ', #class, ' '), ' chapter ')]
XPath query 2: #href
And then it repeatedly passes those queries to libxml2's XPath engine to extract the requested information (in this example, a list of URLs) from the XML documents in question.
It works, but I would prefer to combine the two queries into a single one, which would simplify the code for invoking them and reduce the performance overhead:
XPath query: (//a[#id='home'] | //a[contains(concat(' ', #class, ' '), ' chapter ')])/#href
(note the added parenthesis and slash)
But is this safe to do programmatically, for arbitrary input queries?
In general, no, you can't concatenate two arbitrary XPath expressions in this way, especially not in XPath 1.0. It's easy to find counter-examples: in XPath 1.0 you can't even have a union expression on the RHS of '/', so concatenating "/a" and "(b|c)" would fail.
In XPath 2.0, the result will always be syntactically valid, but in may contain type errors, e.g. if the expressions are "count(a)" and "b". The LHS operand of "/" must evaluate to a sequence of nodes.
Sure, this should work. However, you will always have to respect the correct context. If the elements in your example in the first query have no href attribute, you will get an empty result set.
Also, you will have to take care of e.g. a leading slash in front of your second query, so that you don't end up with a descendant-or-self axis step, which might not be what you want. Apart from that, this should always work - The worst that can happen that it is not logical correct (i.e. you don't get the expected result), but it should always be valid XPath.
I am developing an application that accepts user-defined XPath expressions and employs them as part of its runtime operation.
However, I would like to be able to infer some additional data by programmatically manipulating the expression, and I am curious to know whether there are any situations in which this approach might fail.
Given any user-defined XPath expression that returns a node set, is it safe to wrap it in the XPath count() function to determine the number of nodes in the set:
count(user_defined_expression)
Similarly, is it safe to append an array index to the expression to extract one of the nodes in the set:
user_defined_expression[1]
Well an XPath expression (in XPath 1.0) can yield a node-set or a string or a number or a boolean and doing count(expression) only makes sense on any expression yielding a node-set.
As for adding a positional predicate, I think you might want to use parentheses around your expression i.e. to change /root/foo/bar into (/root/foo/bar)[1] as that way you select the first bar element in the node-set selected by /root/foo/bar while without them you would get /root/foo/bar[1] which would select the first bar child element of any foo child element of the root element.
Are you checking that such user-defined expressions always evaluate to node-set?
If yes, first Expr is ok. Datatype will be correct for fn:count
Second one is a lot trickier, with a lot of situations there predicate will overweight axis, for example. Check this answer for a simple analysis. It will be difficult to say, what a user really meant.
A more robust approach would be to convert the XPath expression to XQueryX, which is an XML representation of the abstract syntax tree; you can then do XQuery or XSLT transformations on this XML representation, and then convert back to a modified XPath (or XQuery) for evaluation.
However, this will still only give you the syntactic structure of the expression; if you want semantic information, such as the inferred static type of the result, you will probably have to poke inside an XPath process that exposes this information.
<bits>
<thing>Match this please</thing>
<thing>Don't match this</thing>
<thing>Match <b>this</b> please</thing>
</bits>
An expression like this:
//thing[text()='Match this please']
will locate the first 'thing' but not the third, because the phrase is distributed through a child node.
Is there an expression that would match the first and the third 'thing' in my example?
Try:
//thing[string()='Match this please']
jsfiddle:
http://jsfiddle.net/ZG9n3/2/
Please check the reference to see if this is going to work for you:
http://www.w3.org/TR/xpath/#function-string
Is there an expression that would
match the first and the third 'thing'
in my example?
You mean: Is there an expression that would select the first and the third element named thing, based on their string value.
Use:
/*/thing[. = 'Match this please']
The predicate compares the string value of the context node to the string "Match this please".
By definition the string value of an element is the concatenation (in document order) of all of its text-nodes descendents.
Note: Always try to avoid the // abbreviation because its use may incur big inefficiency. Whenever the structure of an XML document is known, use a chain of specific location steps.