How to add string function to XPath in Saxon? - xpath

I want to create function which will find a word in a string like
word(#class, 'items')
which will be equals to
fn:matches(#class, "(^|\s)items(\s|$)")

There are various ways you can do this, but the simplest solution is to use XQuery, which allows you to define functions within the language:
declare function local:word($a as xs:string, $b as xs:string) as xs:boolean {
matches($a, concat("(^|\s)", $b, "(\s|$)"))
};

Here what I found for XSLT 2.0
<x:transform version="2.0"
xmlns:x="http://www.w3.org/1999/XSL/Transform"
xmlns:find="http://user.com/namespace">
<x:output method="html"/>
<x:function name="find:word">
<x:param name="src"/>
<x:param name="word"/>
<x:sequence select="matches($src, concat('(^|\s)', $word, '(\s|$)'))"/>
</x:function>
<x:template match="//*[find:word(#class, 'items')]">
<x:copy>
<x:copy-of select="#*"/>
</x:copy>
</x:template>
</x:transform>

Related

How to convert string to Xpath in Xquery function (BaseX)

I am writing Xquery function for BaseX which gets one of arguments as name of the element node. This name is then used in Xpath, but in general I cannot convert string to element.
This is how the method looks like
declare function prefix:getElementWithValue($root as document-node()?, $elem as xs:string?, $minVal as xs:float?, $maxVal as xs:float?)
as element()*
{
let $e := element {$elem} {""}
for $x in $root//SUBELEM
return if ($x//$e/#ATTRIB>=$minVal and $x//$e/#ATTRIB<=$maxVal) then ($x)
};
and the call
return prefix:getElementWithValue($db, "SomeElem", 10.0, 10.0)
and I am getting empty response from that. If I replace the $x//$e with $x//SomeElem it returns proper response. From the QueryPlan I see that the $e is treated as literal value. XPATH is not $x//SomeElem/#ATTRIB but $x//$e/#ATTRIB
So my question is how to covert string to type that can be used in XPATH?
XQuery does not have a standard function to evaluate a dynamically-constructed XPath expression.
Many XQuery processors offer some kind of extension function that does this, however. For example, BaseX offers query:eval():
https://docs.basex.org/wiki/XQuery_Module#xquery:eval
Note that variables in XQuery represent values, not fragments of expression text. Your expression $x//$e/#ATTRIB is equivalent to $x//"SomeElem"/#ATTRIB, which is quite different from $x//SomeElem/#ATTRIB.
If you know that $elem will always be an element name, then you can write $x//*[name()=$e]/#ATTRIB. But take care over namespaces.

xquery optional where statements

In Xquery 3.1 I am processing the variable parameters from a form to search for matching XML documents. The XML documents look like this:
<listBibl xml:id="TC0001" type="collection">
<bibl>
<title type="collection">Bonum universale de apibus</title>
<affiliation corresp="dominican"/>
<author nymRef="thomas_cantipratensis"/>
<location corresp="flanders"/>
<othercontent>....</othercontent>
</bibl>
</listBibl>
The user can submit optional parameters against xml:id, affiliation, author, and location, and they can be parameters with multiple values (sequences).
If the user were to submit all parameters, the query might look like:
for $c in $mycollection//listBibl[#xml:id=($params_id)]
where $c/affiliation[#corresp=($params_affil)]
and $c/author[#nymRef=($params_author)]
and $c/location[#corresp=($params_location)]
return $c
But the user may leave certain parameters empty, effectively making each where statement optional.
The only solution I can currently put together is to have a series of if...then...else statements which account for each permutation of parameters.
Is there any way in Xpath or Xquery to account for the parameters being empty with a wildcard of some sort? In pseudo code, where * represents a wished-for wildcard:
where $c/affiliation[if ($params_affil)
then #corresp=($params_affil)
else #corresp=* ]
Many thanks.
Use predicates of the form
[$params_affil=("", #corresp)]
which matches if $params_affil is either a zero-length string or equal to #corresp. And make zero-length-string (rather than empty sequence) the default if the parameter is not supplied.
Alternatively if the default for an absent parameter is (), use
[empty($params_affil) or $params_affil=#corresp)]
If that gets too repetitive, put the logic in a user-declared function.
I think you can always declare and use your own function as a predicate expression e.g.
declare function local:check-item($item as node(), $values as item()*) as xs:boolean
{
if (exists($values))
then $item = $values
else true()
};
....
where $c/affiliation[local:check-item(#corresp, $params_affil)]

eXist-DB / XQuery StringValue cannot be cast to AnyURIValue (using compression:zip)

In eXist 4.4/XQuery 3.1, I am building a function to compress a number of xml files into a zip using compression:zip.
I have one function which collects all the URIs for the documents to be compressed, schedule:get-document-uris-for-zip(xmlid as xs:string). This function returns lists of documents like the following:
/db/apps/deheresi/data/MS609-0001.xml
/db/apps/deheresi/data/MS609-0002.xml
/db/apps/deheresi/data/MS609-0003.xml
/db/apps/deheresi/data/MS609-0004.xml
/db/apps/deheresi/data/MS609-0005.xml
/db/apps/deheresi/data/MS609-0006.xml
/db/apps/deheresi/data/MS609-0007.xml
/db/apps/deheresi/data/MS609-0008.xml
/db/apps/deheresi/data/MS609-0009.xml
/db/apps/deheresi/data/MS609-0010.xml
This function is called by the compression function as follows
declare function schedule:create-zip-by-batch()
{
let $batch := doc(concat($globalvar:URIdocuments,"document_collections.xml"))
for $entry in $batch//collection[#compile="y"]
let $zipobject := compression:zip(schedule:get-document-uris-for-zip($entry/string(#xml:id)),false())
let $zipstore := xmldb:store("/db/apps/deheresi/documents",
"MS609_tei.zip",
$zipobject)
return $zipstore
};
This is throwing a cast error as follows, but I can't identify how to resolve this...
org.exist.xquery.value.StringValue cannot be cast to org.exist.xquery.value.AnyURIValue
Many thanks in advance.
Edit - I'm adding here the part of the function schedule:get-document-uris-for-zip(xmlid as xs:string) which outputs the list of URIs. The URIs are built through string concatenation:
(: get names of documents which meet criteria :)
let $list := xmldb:get-child-resources("/db/apps/deheresi/data")[starts-with(., $y/string(#filename)) and ends-with(., $y/string(#ext))]
(: create URI for each document :)
return
for $n in $list
return concat("/db/apps/deheresi/data/",$n)
You're right to find this function a bit confusing. The (eXist-specific) compression:zip() function $sources parameter is typed as if it is quite flexible way, as xs:anyType()+. But really it is quite strict about the two types of item it accepts: a sequence of URIs (i.e., of type xs:anyURI), or a sequence of <entry> elements:
<entry name="filename.ext"
type="collection|uri|binary|xml|text"
method="deflate|store"
>data</entry>
See https://exist-db.org/exist/apps/fundocs/view.html?uri=http://exist-db.org/xquery/compression#zip.2.
The problem with your code is that you are passing strings in your $sources parameter, and have not cast these strings as xs:anyURI.
Here is sample working code:
xquery version "3.1";
let $prepare :=
(
xmldb:create-collection("/db", "test"),
xmldb:store("/db/test", "test.xml", <test/>)
)
let $zip := compression:zip("/db/test/test.xml" cast as xs:anyURI, false())
return
xmldb:store("/db/test", "test.zip", $zip)

xquery parameter query

I'm trying to query an exist-db with xquery by taking parameters from the URL and building up seach parameters
xquery version "1.0";
declare namespace request="http://exist-db.org/xquery/request";
declare namespace xs="http://www.w3.org/2001/XMLSchema";
declare option exist:serialize "method=xml media-type=text/xml omit-xml-declaration=no indent=yes";
let $param1:= request:get-parameter("param1",'0')
let $person :=
if($param1 = '0')
then "'*'"
else concat('contributions/person/#val="',$param1,'"')
return
<xml>
{
for $x in subsequence(//foo/bar[$person],1,3)
return $x
}
</xml>
The code above shows that I get the parameter from the url $param1.
variable $person checks to see if there was a parameter and based on that creates a query parameter. This variable works fine, from testing it prints out either '*' for no param or
contributions/person/#val='hello, world'
When I run the query it prints out as if the value is '*'. In the for $x part, can I pass a variable like that? I've tried putting concat($person,'') with the same results. Hardcoding the full path gives me the results I'm looking for, but I'm looking to create something more dynamic.
To note: there is only one variable, $person, but there will be others once I get it to work
I think ideally you would avoid dynamic string evaluation. In this example, some pretty simple reorganization would solve the problem without it:
<xml>
{
for $x in subsequence(//foo/bar[
if ($param1 = '0')
then *
else (contributions/person/#val = $param1)
],1,3)
return $x
}
</xml>
However, you can use eval(), but keep in mind there are security risks:
<xml>
{
for $x in subsequence(eval(
concat('//foo/bar[',$person,']')
),1,3)
return $x
}
</xml>

Passing a subpath as a variable (type of node) or either eval() an string containing a relative XPATH

I want to succed with following functionality, and it works if you harcoded it:
declare function local:sort($collection as node()*, $filter as xs:SUBPATH?) as node()*{
for $element in $collection
order by
if ($filter) then ($element/$filter) (: OR SOME KIND OF fn:eval($filter) IF WE DEFINED $filter AS AN xs:string :)
else ($element/name()) (: Default :)
descending
return $element
};
And it could be called like that:
for $element in local:sort(doc('data')/Data/*,'/#myAttr')
return $element
or
for $element in local:sort(doc('data')/Data/*,'/subnode/subnode/name()')
return $element
or
for $element in local:sort(doc('data')/Data/*,()) (: This is the default of function = own elmentĀ“s name :)
return $element
Mx problem is passing the subpath. Either I need to know some kind of way to send a relative XPATH as an argument and type of node, or I need some kind of eval to pass from xs:string to a runtime valid code
Any help?
You could consider (a) generating a query in which the filter is hard-coded, or (b) using an eval() function specific to your XQuery vendor, or (c) using XQuery 3.0 higher-order functions if your chosen XQuery engine supports them yet.
I ended up doing the following, thanks to this idea How can I solve this autoncremental var case in XQUERY 1.0 FLOWR?:
(: This is a workaround solution as xquery:eval() is not working with var bindings until BaseX 7.3 version :)
(: It evals $context node + literal subpath with a pattern of '(/)lit/lit/lit/#attr' or '(/)lit/lit/lit' representing (/)lit/lit/lit/name(). If subpath null, it returns $context/name().:)
declare function u:eval-path($context as node()*, $subnodes as xs:string*) as item()* {
if(empty($subnodes)) then $context/name()
else(
if (count($subnodes) eq 1) then ( (: Last Element :)
if (starts-with($subnodes[1],'#')) then $context/#*[name()=substring-after($subnodes[1],'#')]
else $context/*[name()=$subnodes[1]]/name()
)
else if ($subnodes[1] eq '') then u:eval-path($context, $subnodes[position() gt 1])
else u:eval-path($context/*[name()=$subnodes[1]],$subnodes[position() gt 1])
)
};
(: Sorts the given collection by given criteria, which should be a pattern '(/)lit/lit/lit/#attr' or '(/)lit/lit/lit' representing (/)lit/lit/lit/name() :)
(: If criteria is null, everything is ordered by $elements/name(). Theres no way to filter intermediate nodes as in /lit/*[name()='X']/lit :)
declare function u:sort($collection as node()*, $criteria as xs:string?) as node()*{
for $element in $collection
order by u:eval-path($element,tokenize($criteria,'/'))
ascending
return $element
};

Resources