Built in Functions

Working with Scoped Datasets

Motivating example from S.Bichof's MS thesis, p50 (72 of 207); also see presentation:

• Fixes unintended behavior of nested queries.
• Extends the scope of a dataset over subqueries and allows blank node joins See corrected p54 (76 of 207)

xQuery callable Functions:

• xsparql:createScopedDataset($query,$id): evaluate a SPARQL query, storing the bindings to be reused later.
• xsparql:sparqlScopedDataset($query,$id,$joinVars,$pos): evaluate a SPARQL query, using previously stored dataset $id. $joinVars: comma-separated joining variables that will be put in the initial binding of $query $pos is the current iteration
• _xsparql:deleteScopedDataset($id): delete a scoped dataset
• xsparql:scopedDatasetPopResults($id): delete the last results from the stack

SAXON Extension Functions Classes

The classes bind Java functions to xQuery ones

_xsparql Namespace ("http://xsparql.deri.org/demo/xquery/xsparql.xquery")

• createScopedDatasetExtFunction - binding for the createScopedDataset as above
• deleteScopedDatasetExtFunction - binding for the deleteScopedDataset
• scopedDatasetPopResultsExtFunction - delete the last results from the stack
• sparqlQueryExtFunction - binding for the sparqlQuery function
• sparqlScopedDatasetExtFunction - binding for the sparqlScopedDataset function
• sqlQueryExtFunction - binding for the _sqlQuery function
• turtleGraphToURIExtFunction - Saves temp graph to file
• scopedDatasetPopResultsExtFunction - binding for the scopedDatasetPopResults function
• doPostQueryExtFunction - Can execute SQL queries as sqlQuery(), not sure what is the difference.
• createNamedGraphExtFunction - binding for the createNamedGraph function
• deleteNamedGraphExtFunction - binding for the deleteNamedGraph function
• sparqlQueryTDBExtFunction - binding for the _sparqlQueryTDB function

xsparql Namespace ("http://xsparql.deri.org/demo/xquery/sparql-functions.xquery")

• getRDBTableAttributesExtFunction - receives the information about attributes from a specific table

let $table := xsparql:getRDBTableAttributes("EMP")
return $table

result:

<metadata>
   <foreignKeys/>
   <columns>
      <column name="&#34;EMPNO&#34;" type="INT"/>
      <column name="&#34;ENAME&#34;" type="VARCHAR"/>
      <column name="&#34;JOB&#34;" type="VARCHAR"/>
      <column name="&#34;DEPTNO&#34;" type="INT"/>
   </columns>
</metadata>

• getRDBTablesExtFunction - receives the names of all tables in the current db

let $tables := xsparql:getRDBTables()
return $tables

result:

<relations>
   <relation>EMP</relation>
   <relation>test1</relation>
</relations>

• jsonDocExtFunction - convert JSON to XML
• sqlQuery

let $res := "
       Select EMPNO
            , ENAME
            , JOB
            , DEPTNO
         from EMP
       "

for row $row from sqlQuery($res)
return $row

result:

<result>
   <SQLbinding type="INT" name="&#34;EMPNO&#34;">7369</SQLbinding>
   <SQLbinding type="VARCHAR" name="&#34;ENAME&#34;">SMITH</SQLbinding>
   <SQLbinding type="VARCHAR" name="&#34;JOB&#34;">CLERK</SQLbinding>
   <SQLbinding type="INT" name="&#34;DEPTNO&#34;">10</SQLbinding>
</result>

No namespace

• xqueryEvaluatorSaxon - registers the Saxon extention functions

Skolemisation Functions

These replace blank nodes to a URL using a specific prefix, to guarantee sameness between datasets. See Skolemisation in Wikipedia. They leave other terms unmolested.

• xsparql:skolem($term,$prefix): skolemize a term using the provided prefix
• xsparql:skolem($term): skolemise a term using default prefix http://xsparql.deri.org/skolem#
• xsparql:skolemise($terms): skolemise sequence of terms
• xsparql:skolemiseGraph($graph,$prefix): skolemise all blank nodes in a graph using the provided prefix
• xsparql:skolemiseGraph($graph): skolemise all blank nodes in a graph using default prefix http://xsparql.deri.org/skolem#

Functions Implemented in XQuery

https://github.com/semantalytics/xsparql/tree/master/xsparql-rewriter/src/main/resources/xquery/sparql-functions.xquery

Term Accessors

• xsparql:isBlank($term)
• xsparql:isIRI($term)
• xsparql:isLiteral($term)
• xsparql:bound($term)
• xsparql:datatype($term)
• xsparql:lang($term)

Examples:

• test_datatype_lang.xsparql

{fn:concat($o," : ",xsparql:datatype($o)," : ",xsparql:lang($o)) }

• query1_json-ld.xsparql

Create RDF Terms

• xsparql:createURI($uri): creates URI from string
• xsparql:createBNode($name): creates blank node using provided string name
• xsparql:createLiteral($str): creates plain-text literal. Is it equivalent to one with datatype xsd:string as per RDF 1.1?
• xsparql:createLiteral($str,$lang,$datatype): creates literal with given value, lang and datatype. Allows both lang and datatype but shouldn't.
• xsparql:createTerm($item): creates RDF term, guessing the type from the form of input: literal if it has XML type xs:integer (but that's not carried over to the datatype), blank node if it matches _:\w+, URI if it has XML type xs:anyURI, URI if it starts with one of 4-5 fixed schemes (see bug https://github.com/semantalytics/xsparql/issues/4), literal otherwise
• xsparql:createTerm($item,$lang,$datatype): creates RDF term, guessing the type from the form of input. If blank node or URI is guessed, the lang and datatype are ignored.

keepRDFTerms

From test_datatype_lang.xsparql:

declare option xsparql:keepRDFTerms "true";

This was added on 2014-09-08 but is now removed. I tested the query and it returns fine, so I guess this is now always enabled