man RDF::Query (3): A complete SPARQL 1.1 Query and Update implementation for use with RDF::Trine.

VERSION

This document describes RDF::Query version 2.916.

SYNOPSIS

# SPARQL SELECT Query
my $query = RDF::Query->new( 'SELECT * WHERE ...' );
my $iterator = $query->execute( $model );
while (my $row = $iterator->next) {
# $row is a HASHref containing variable name -> RDF Term bindings
print $row->{ 'var' }->as_string;
}

# SPARQL CONSTRUCT/DESCRIBE Query
my $query = RDF::Query->new( 'CONSTRUCT { ... } WHERE ...' );
my $iterator = $query->execute( $model );
while (my $st = $iterator->next) {
# $st is a RDF::Trine::Statement object representing an RDF triple
print $st->as_string;
}

# SPARQL ASK Query
my $query = RDF::Query->new( 'ASK WHERE ...' );
my $iterator = $query->execute( $model );
my $bool = $iterator->get_boolean;
if ($bool) {
print "Yes!\n";
}

# RDQL Query
my $query = new RDF::Query ( $rdql, { lang => 'rdql' } );
my @rows = $query->execute( $model ); # in list context, returns all results

DESCRIPTION

RDF::Query allows SPARQL and RDQL queries to be run against an RDF model, returning rows of matching results.

See <http://www.w3.org/TR/rdf-sparql-query/> for more information on SPARQL.

See <http://www.w3.org/Submission/2004/SUBM-RDQL-20040109/> for more information on RDQL.

CHANGES IN VERSION 2.900

The 2.9xx versions of RDF::Query introduce some significant changes that will lead to a stable 3.000 release supporting SPARQL 1.1. Version 2.902 introduces the SPARQL 1.1 features up to date with the SPARQL 1.1 working drafts as of its release date. Version 2.902 also is the first version to require use of RDF::Trine for the underlying RDF store. This change means that RDF::Core is no longer supported, and while Redland is still supported, its handling of ``contexts'' (named graphs) means that existing RDF triples stored in Redland without associated contexts will not be accessible from RDF::Query. See RDF::Trine::Store for more information on supported backend stores.

CHANGES IN VERSION 2.000

There are many changes in the code between the 1.x and 2.x releases. Most of these changes will only affect queries that should have raised errors in the first place (SPARQL parsing, queries that use undefined namespaces, etc.). Beyond these changes, however, there are some significant API changes that will affect all users:

Use of RDF::Trine objects

All nodes and statements returned by RDF::Query are now RDF::Trine objects (more specifically, RDF::Trine::Node and RDF::Trine::Statement objects). This differes from RDF::Query 1.x where nodes and statements were of the same type as the underlying model (Redland nodes from a Redland model and RDF::Core nodes from an RDF::Core model).

In the past, it was possible to execute a query and not know what type of nodes were going to be returned, leading to overly verbose code that required examining all nodes and statements with the bridge object. This new API brings consistency to both the execution model and client code, greatly simplifying interaction with query results.

Binding Result Values

Binding result values returned by calling "$iterator->next" are now HASH references (instead of ARRAY references), keyed by variable name. Where prior code might use this code (modulo model definition and namespace declarations):

  my $sparql = 'SELECT ?name ?homepage WHERE { [ foaf:name ?name ; foaf:homepage ?homepage ] }';
  my $query = RDF::Query->new( $sparql );
  my $iterator = $query->execute( $model );
  while (my $row = $iterator->()) {
    my ($name, $homepage) = @$row;
    # ...
  }

New code using RDF::Query 2.000 and later should instead use:

  my $sparql = 'SELECT ?name ?homepage WHERE { [ foaf:name ?name ; foaf:homepage ?homepage ] }';
  my $query = RDF::Query->new( $sparql );
  my $iterator = $query->execute( $model );
  while (my $row = $iterator->next) {
    my $name = $row->{ name };
    my $homepage = $row->{ homepage };
    # ...
  }

(Also notice the new method calling syntax for retrieving rows.)

METHODS

"new ( $query, \%options )"

Returns a new RDF::Query object for the specified $query. The query language defaults to SPARQL 1.1, but may be set specifically with the appropriate %options value. Valid %options are:

* lang

Specifies the query language. Acceptable values are 'sparql11', 'sparql', or 'rdql'.

* base_uri

Specifies the base URI used in parsing the query.

* update

A boolean value indicating whether update operations are allowed during query execution.

* load_data

A boolean value indicating whether URIs used in SPARQL FROM and FROM NAMED clauses should be dereferenced and the resulting RDF content used to construct the dataset against which the query is run.

"get ( $model )"

Executes the query using the specified model, and returns the first matching row as a LIST of values.

"prepare ( $model )"

Prepares the query, constructing a query execution plan, and returns a list containing ($plan, $context). To execute the plan, call "execute_plan( $plan, $context )".

"execute ( $model, %args )"

Executes the query using the specified RDF $model. If called in a list context, returns an array of rows, otherwise returns an RDF::Trine::Iterator object. The iterator returned may be an instance of several subclasses of RDF::Trine::Iterator:

* A RDF::Trine::Iterator::Bindings object is returned for query forms producing variable binding results (SELECT queries).

* A RDF::Trine::Iterator::Graph object is returned for query forms producing in an RDF graph result (DESCRIBE and CONSTRUCT queries).

* A RDF::Trine::Iterator::Boolean object is returned for query forms producing a true/false result (ASK queries).

"execute_plan ( $plan, $context )"

Executes the query plan generated by the "<prepare"> method using the supplied RDF::Query::ExecutionContext object. Return value(s) are the same as for the "<execute"> method.

"prepare_with_named_graphs ( $model, @uris )"

"execute_with_named_graphs ( $model, @uris )"

Executes the query using the specified RDF $model, loading the contents of the specified @uris into named graphs immediately prior to matching the query. Otherwise, acts just like "execute".

"pattern"

Returns the RDF::Query::Algebra::GroupGraphPattern algebra pattern for this query.

"is_update"

"as_sparql"

Returns the query as a string in the SPARQL syntax.

"as_hash"

Returns the query as a nested set of plain data structures (no objects).

"sse"

Returns the query as a string in the SSE syntax.

"dateparser"

Returns the DateTime::Format::W3CDTF object associated with this query object.

"specifies_update_dataset"

Returns true if the query specifies a custom update dataset via the WITH or USING keywords, false otherwise.

"add_function ( $uri, $function )"

Associates the custom function $function (a CODE reference) with the specified URI, allowing the function to be called by query FILTERs.

"supported_extensions"

Returns a list of URLs representing extensions to SPARQL that are supported by the query engine.

"supported_functions"

Returns a list URLs that may be used as functions in FILTER clauses (and the SELECT clause if the SPARQL 1.1 parser is used).

"add_computed_statement_generator ( $predicate => \&generator )"

Adds a statement generator for the given $predicate to the query object. This statement generator will be called as "$generator->( $query, $model, \%bound, $s, $p, $o, $c )" and is expected to return an RDF::Trine::Iterator::Graph object containing statements with $predicate.

"get_computed_statement_generators ( [ $predicate ] )"

Returns an ARRAY reference of computed statement generator closures.

"add_hook_once ( $hook_uri, $function, $token )"

Calls "add_hook" adding the supplied $function only once based on the $token identifier. This may be useful if the only code that is able to add a hook is called many times (in an extension function, for example).

"add_hook ( $hook_uri, $function )"

Associates the custom function $function (a CODE reference) with the RDF::Query code hook specified by $uri. Each function that has been associated with a particular hook will be called (in the order they were registered as hooks) when the hook event occurs. See ``Defined Hooks'' for more information.

"parsed ()"

Returns the parse tree.

"model"

Returns the RDF::Trine::Model object for this query.

"useragent"

Returns the LWP::UserAgent object used for retrieving web content.

"log ( $key [, $value ] )"

If no logger object is associated with this query object, does nothing. Otherwise, return or set the corresponding value depending on whether a $value is specified.

"logger"

Returns the logger object associated with this query object (if present).

"error ()"

Returns the last error the parser experienced.

DEFINED HOOKS

The following hook URIs are defined and may be used to extend the query engine functionality using the "add_hook" method:

http://kasei.us/code/rdf-query/hooks/post-create-model

Called after loading all external files to a temporary model in queries that use FROM and FROM NAMED.

Args: ( $query, $model )

$query is the RDF::Query object. $model is the RDF::Trine::Model object.

http://kasei.us/code/rdf-query/hooks/post-execute

Called immediately before returning a result iterator from the execute method.

Args: ( $query, $model, $iterator )

$query is the RDF::Query object. $model is the RDF::Trine::Model object. $iterator is a RDF::Trine::Iterator object.

AUTHOR

 Gregory Todd Williams <[email protected]>