php-json-ld/jsonld.php
2011-07-18 16:22:57 -04:00

2621 lines
68 KiB
PHP

<?php
/**
* PHP implementation of JSON-LD.
*
* @author Dave Longley
*
* Copyright (c) 2011 Digital Bazaar, Inc. All rights reserved.
*/
define('__S', '@subject');
define('__T', '@type');
define('JSONLD_RDF', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
define('JSONLD_RDF_TYPE', JSONLD_RDF . 'type');
define('JSONLD_XSD', 'http://www.w3.org/2001/XMLSchema#');
define('JSONLD_XSD_ANY_TYPE', JSONLD_XSD . 'anyType');
define('JSONLD_XSD_BOOLEAN', JSONLD_XSD . 'boolean');
define('JSONLD_XSD_DOUBLE', JSONLD_XSD . 'double');
define('JSONLD_XSD_INTEGER', JSONLD_XSD . 'integer');
define('JSONLD_XSD_ANY_URI', JSONLD_XSD . 'anyURI');
/**
* Creates the JSON-LD default context.
*
* @return the JSON-LD default context.
*/
function jsonld_create_default_context()
{
return (object)array(
'rdf' => JSONLD_RDF,
'rdfs' => 'http://www.w3.org/2000/01/rdf-schema#',
'owl' => 'http://www.w3.org/2002/07/owl#',
'xsd' => 'http://www.w3.org/2001/XMLSchema#',
'dcterms' => 'http://purl.org/dc/terms/',
'foaf' => 'http://xmlns.com/foaf/0.1/',
'cal' => 'http://www.w3.org/2002/12/cal/ical#',
'vcard' => 'http://www.w3.org/2006/vcard/ns#',
'geo' => 'http://www.w3.org/2003/01/geo/wgs84_pos#',
'cc' => 'http://creativecommons.org/ns#',
'sioc' => 'http://rdfs.org/sioc/ns#',
'doap' => 'http://usefulinc.com/ns/doap#',
'com' => 'http://purl.org/commerce#',
'ps' => 'http://purl.org/payswarm#',
'gr' => 'http://purl.org/goodrelations/v1#',
'sig' => 'http://purl.org/signature#',
'ccard' => 'http://purl.org/commerce/creditcard#',
'@coerce' => (object)array(
'xsd:anyURI' => array('foaf:homepage', 'foaf:member'),
'xsd:integer' => 'foaf:age'
),
'@vocab' => ''
);
}
/**
* Normalizes a JSON-LD object.
*
* @param input the JSON-LD object to normalize.
*
* @return the normalized JSON-LD object.
*/
function jsonld_normalize($input)
{
$p = new JsonLdProcessor();
return $p->normalize($input);
};
/**
* Removes the context from a JSON-LD object.
*
* @param input the JSON-LD object to remove the context from.
*
* @return the context-neutral JSON-LD object.
*/
function jsonld_remove_context($input)
{
$rval = null;
if($input !== null)
{
$ctx = jsonld_create_default_context();
$rval = _expand($ctx, null, $input, false);
}
return $rval;
};
function jsonld_expand($input)
{
return jsonld_remove_context($input);
}
/**
* Adds the given context to the given context-neutral JSON-LD object.
*
* @param ctx the new context to use.
* @param input the context-neutral JSON-LD object to add the context to.
*
* @return the JSON-LD object with the new context.
*/
function jsonld_add_context($ctx, $input)
{
$rval;
// TODO: should context simplification be optional? (ie: remove context
// entries that are not used in the output)
$ctx = jsonld_merge_contexts(jsonld_create_default_context(), $ctx);
// setup output context
$ctxOut = new stdClass();
// compact
$rval = _compact($ctx, null, $input, $ctxOut);
// add context if used
if(count(array_keys((array)$ctxOut)) > 0)
{
// add copy of context to every entry in output array
if(is_array($rval))
{
foreach($rval as $v)
{
$v->{'@context'} = _cloneContext($ctxOut);
}
}
else
{
$rval->{'@context'} = $ctxOut;
}
}
return $rval;
}
/**
* Changes the context of JSON-LD object "input" to "context", returning the
* output.
*
* @param ctx the new context to use.
* @param input the input JSON-LD object.
*
* @return the output JSON-LD object.
*/
function jsonld_change_context($ctx, $input)
{
// remove context and then add new one
return jsonld_add_context($ctx, jsonld_remove_context($input));
}
function jsonld_compact($ctx, $input)
{
return jsonld_change_context($ctx, $input);
}
/**
* Merges one context with another.
*
* @param ctx1 the context to overwrite/append to.
* @param ctx2 the new context to merge onto ctx1.
*
* @return the merged context.
*/
function jsonld_merge_contexts($ctx1, $ctx2)
{
// copy contexts
$merged = _cloneContext($ctx1);
$copy = _cloneContext($ctx2);
// if the new context contains any IRIs that are in the merged context,
// remove them from the merged context, they will be overwritten
foreach($copy as $key => $value)
{
// ignore special keys starting with '@'
if(strpos($key, '@') !== 0)
{
foreach($merged as $mkey => $mvalue)
{
if($mvalue === $value)
{
unset($merged->$mkey);
break;
}
}
}
}
// @coerce must be specially-merged, remove from contexts
$coerceExists = isset($merged->{'@coerce'}) or isset($copy->{'@coerce'});
if($coerceExists)
{
$c1 = isset($merged->{'@coerce'}) ? $merged->{'@coerce'} : new stdClass();
$c2 = isset($copy->{'@coerce'}) ? $copy->{'@coerce'} : new stdClass();
unset($merged->{'@coerce'});
unset($copy->{'@coerce'});
}
// merge contexts
foreach($copy as $key => $value)
{
$merged->$key = $value;
}
// special-merge @coerce
if($coerceExists)
{
foreach($c1 as $type => $p1)
{
// append existing-type properties that don't already exist
if(isset($c2->$type))
{
$p2 = $c2->$type;
// normalize props in c2 to array for single-code-path iterating
if(!is_array($p2))
{
$p2 = array($p2);
}
// add unique properties from p2 to p1
foreach($p2 as $i => $p)
{
if((!is_array($p1) and $p1 !== $p) or
(is_array($p1) and array_search($p, $p1) === false))
{
if(is_array($p1))
{
$p1[] = $p;
}
else
{
$p1 = array($p1, $p);
}
}
}
$c1->$type = $p1;
}
}
// add new types from new @coerce
foreach($c2 as $type => $value)
{
if(!isset($c1->$type))
{
$c1->$type = $value;
}
}
// ensure there are no property duplicates in @coerce
$unique = new stdClass();
$dups = array();
foreach($c1 as $type => $p)
{
if(is_string($p))
{
$p = array($p);
}
foreach($p as $v)
{
if(!isset($unique->$v))
{
$unique->$v = true;
}
else if(!in_array($v, $dups))
{
$dups[] = $v;
}
}
}
if(count($dups)> 0)
{
throw new Exception(
'Invalid type coercion specification. More than one ' .
'type specified for at least one property. duplicates=' .
print_r(dups, true));
}
$merged->{'@coerce'} = $c1;
}
return $merged;
}
/**
* Expands a term into an absolute IRI. The term may be a regular term, a
* CURIE, a relative IRI, or an absolute IRI. In any case, the associated
* absolute IRI will be returned.
*
* @param ctx the context to use.
* @param term the term to expand.
*
* @return the expanded term as an absolute IRI.
*/
function jsonld_expand_term($ctx, $term)
{
return _expandTerm($ctx, $term);
}
/**
* Compacts an IRI into a term or CURIE if it can be. IRIs will not be
* compacted to relative IRIs if they match the given context's default
* vocabulary.
*
* @param ctx the context to use.
* @param iri the IRI to compact.
*
* @return the compacted IRI as a term or CURIE or the original IRI.
*/
function jsonld_compact_iri($ctx, $iri)
{
return _compactIri($ctx, $iri, null);
}
/**
* Frames JSON-LD input.
*
* @param input the JSON-LD input.
* @param frame the frame to use.
* @param options framing options to use.
*
* @return the framed output.
*/
function jsonld_frame($input, $frame, $options=null)
{
$rval;
// normalize input
$input = jsonld_normalize($input);
// save frame context
$ctx = null;
if(isset($frame->{'@context'}))
{
$ctx = jsonld_merge_contexts(
jsonld_create_default_context(), $frame->{'@context'});
}
// remove context from frame
$frame = jsonld_remove_context($frame);
// create framing options
// TODO: merge in options from function parameter
$options = new stdClass();
$options->defaults = new stdClass();
$options->defaults->embedOn = true;
$options->defaults->explicitOn = false;
// build map of all subjects
$subjects = new stdClass();
foreach($input as $i)
{
$subjects->{$i->{__S}->{'@iri'}} = $i;
}
// frame input
$rval = _frame($subjects, $input, $frame, new stdClass(), $options);
// apply context
if($ctx !== null and $rval !== null)
{
$rval = jsonld_add_context($ctx, $rval);
}
return $rval;
}
/**
* Compacts an IRI into a term or CURIE if it can be. IRIs will not be
* compacted to relative IRIs if they match the given context's default
* vocabulary.
*
* @param ctx the context to use.
* @param iri the IRI to compact.
* @param usedCtx a context to update if a value was used from "ctx".
*
* @return the compacted IRI as a term or CURIE or the original IRI.
*/
function _compactIri($ctx, $iri, $usedCtx)
{
$rval = null;
// check the context for a term that could shorten the IRI
// (give preference to terms over CURIEs)
foreach($ctx as $key => $value)
{
// skip special context keys (start with '@')
if(strlen($key) > 0 and $key[0] !== '@')
{
// compact to a term
if($iri === $ctx->$key)
{
$rval = $key;
if($usedCtx !== null)
{
$usedCtx->$key = $ctx->$key;
}
break;
}
}
}
// term not found, if term is rdf type, use built-in keyword
if($rval === null and $iri === JSONLD_RDF_TYPE)
{
$rval = __T;
}
// term not found, check the context for a CURIE prefix
if($rval === null)
{
foreach($ctx as $key => $value)
{
// skip special context keys (start with '@')
if(strlen($key) > 0 and $key[0] !== '@')
{
// see if IRI begins with the next IRI from the context
$ctxIri = $ctx->$key;
$idx = strpos($iri, $ctxIri);
// compact to a CURIE
if($idx === 0 and strlen($iri) > strlen($ctxIri))
{
$rval = $key . ':' . substr($iri, strlen($ctxIri));
if($usedCtx !== null)
{
$usedCtx->$key = $ctxIri;
}
break;
}
}
}
}
// could not compact IRI
if($rval === null)
{
$rval = $iri;
}
return $rval;
}
/**
* Expands a term into an absolute IRI. The term may be a regular term, a
* CURIE, a relative IRI, or an absolute IRI. In any case, the associated
* absolute IRI will be returned.
*
* @param ctx the context to use.
* @param term the term to expand.
* @param usedCtx a context to update if a value was used from "ctx".
*
* @return the expanded term as an absolute IRI.
*/
function _expandTerm($ctx, $term, $usedCtx)
{
$rval;
// 1. If the property has a colon, then it is a CURIE or an absolute IRI:
$idx = strpos($term, ':');
if($idx !== false)
{
// get the potential CURIE prefix
$prefix = substr($term, 0, $idx);
// 1.1. See if the prefix is in the context:
if(isset($ctx->$prefix))
{
// prefix found, expand property to absolute IRI
$rval = $ctx->$prefix . substr($term, $idx + 1);
if($usedCtx !== null)
{
$usedCtx->$prefix = $ctx->$prefix;
}
}
// 1.2. Prefix is not in context, property is already an absolute IRI:
else
{
$rval = $term;
}
}
// 2. If the property is in the context, then it's a term.
else if(isset($ctx->$term))
{
$rval = $ctx->$term;
if($usedCtx !== null)
{
$usedCtx->$term = $rval;
}
}
// 3. The property is the special-case subject.
else if($term === __S)
{
$rval = __S;
}
// 4. The property is the special-case rdf type.
else if($term === __T)
{
$rval = JSONLD_RDF_TYPE;
}
// 5. The property is a relative IRI, prepend the default vocab.
else
{
$rval = $ctx->{'@vocab'} . $term;
if($usedCtx !== null)
{
$usedCtx->{'@vocab'} = $ctx->{'@vocab'};
}
}
return $rval;
}
/**
* Sets a subject's property to the given object value. If a value already
* exists, it will be appended to an array.
*
* @param s the subject.
* @param p the property.
* @param o the object.
*/
function _setProperty($s, $p, $o)
{
if(isset($s->$p))
{
if(is_array($s->$p))
{
array_push($s->$p, $o);
}
else
{
$s->$p = array($s->$p, $o);
}
}
else
{
$s->$p = $o;
}
}
/**
* Clones a string/number or an object and sorts the keys. Deep clone
* is not performed. This function will deep copy arrays, but that feature
* isn't needed in this implementation at present. If it is needed in the
* future, it will have to be implemented here.
*
* @param value the value to clone.
*
* @return the cloned value.
*/
function _clone($value)
{
$rval;
if(is_object($value))
{
$rval = new stdClass();
$keys = array_keys((array)$value);
sort($keys);
foreach($keys as $key)
{
$rval->$key = $value->$key;
}
}
else
{
$rval = $value;
}
return $rval;
}
/**
* Clones a context.
*
* @param ctx the context to clone.
*
* @return the clone of the context.
*/
function _cloneContext($ctx)
{
$rval = new stdClass();
foreach($ctx as $key => $value)
{
// deep-copy @coerce
if($key === '@coerce')
{
$rval->{'@coerce'} = new stdClass();
foreach($ctx->{'@coerce'} as $type => $p)
{
$rval->{'@coerce'}->$type = $p;
}
}
else
{
$rval->$key = $ctx->$key;
}
}
return $rval;
}
/**
* Gets the coerce type for the given property.
*
* @param ctx the context to use.
* @param property the property to get the coerced type for.
* @param usedCtx a context to update if a value was used from "ctx".
*
* @return the coerce type, null for none.
*/
function _getCoerceType($ctx, $property, $usedCtx)
{
$rval = null;
// get expanded property
$p = _expandTerm($ctx, $property, null);
// built-in type coercion JSON-LD-isms
if($p === __S or $p === JSONLD_RDF_TYPE)
{
$rval = JSONLD_XSD_ANY_URI;
}
// check type coercion for property
else
{
// force compacted property
$p = _compactIri($ctx, $p, null);
foreach($ctx->{'@coerce'} as $type => $props)
{
// get coerced properties (normalize to an array)
if(!is_array($props))
{
$props = array($props);
}
// look for the property in the array
foreach($props as $prop)
{
// property found
if($prop === $p)
{
$rval = _expandTerm($ctx, $type, $usedCtx);
if($usedCtx !== null)
{
if(!isset($usedCtx->{'@coerce'}))
{
$usedCtx->{'@coerce'} = new stdClass();
}
if(!isset($usedCtx->{'@coerce'}->$type))
{
$usedCtx->{'@coerce'}->$type = $p;
}
else
{
$c = $usedCtx->{'@coerce'}->$type;
if(is_array($c) and in_array($p, $c) or
is_string($c) and $c !== $p)
{
_setProperty($usedCtx->{'@coerce'}, $type, $p);
}
}
}
break;
}
}
}
}
return $rval;
}
/**
* Recursively compacts a value. This method will compact IRIs to CURIEs or
* terms and do reverse type coercion to compact a value.
*
* @param ctx the context to use.
* @param property the property that points to the value, NULL for none.
* @param value the value to compact.
* @param usedCtx a context to update if a value was used from "ctx".
*
* @return the compacted value.
*/
function _compact($ctx, $property, $value, $usedCtx)
{
$rval;
if($value === null)
{
$rval = null;
}
else if(is_array($value))
{
// recursively add compacted values to array
$rval = array();
foreach($value as $v)
{
$rval[] = _compact($ctx, $property, $v, $usedCtx);
}
}
// graph literal/disjoint graph
else if(
is_object($value) and
isset($value->{__S}) and is_array($value->{__S}))
{
$rval = new stdClass();
$rval->{__S} = _compact($ctx, $property, $value->{__S}, $usedCtx);
}
// value has sub-properties if it doesn't define a literal or IRI value
else if(
is_object($value) and
!isset($value->{'@literal'}) and !isset($value->{'@iri'}))
{
// recursively handle sub-properties that aren't a sub-context
$rval = new stdClass();
foreach($value as $key => $v)
{
if($v !== '@context')
{
// set object to compacted property
_setProperty(
$rval, _compactIri($ctx, $key, $usedCtx),
_compact($ctx, $key, $v, $usedCtx));
}
}
}
else
{
// get coerce type
$coerce = _getCoerceType($ctx, $property, $usedCtx);
// get type from value, to ensure coercion is valid
$type = null;
if(is_object($value))
{
// type coercion can only occur if language is not specified
if(!isset($value->{'@language'}))
{
// datatype must match coerce type if specified
if(isset($value->{'@datatype'}))
{
$type = $value->{'@datatype'};
}
// datatype is IRI
else if(isset($value->{'@iri'}))
{
$type = JSONLD_XSD_ANY_URI;
}
// can be coerced to any type
else
{
$type = $coerce;
}
}
}
// type can be coerced to anything
else if(is_string($value))
{
$type = $coerce;
}
// types that can be auto-coerced from a JSON-builtin
if($coerce === null and
($type === JSONLD_XSD_BOOLEAN or $type === JSONLD_XSD_INTEGER or
$type === JSONLD_XSD_DOUBLE))
{
$coerce = $type;
}
// do reverse type-coercion
if($coerce !== null)
{
// type is only null if a language was specified, which is an error
// if type coercion is specified
if($type === null)
{
throw new Exception(
'Cannot coerce type when a language is specified. ' .
'The language information would be lost.');
}
// if the value type does not match the coerce type, it is an error
else if($type !== $coerce)
{
throw new Exception(
'Cannot coerce type because the datatype does not match.');
}
// do reverse type-coercion
else
{
if(is_object($value))
{
if(isset($value->{'@iri'}))
{
$rval = $value->{'@iri'};
}
else if(isset($value->{'@literal'}))
{
$rval = $value->{'@literal'};
}
}
else
{
$rval = $value;
}
// do basic JSON types conversion
if($coerce === JSONLD_XSD_BOOLEAN)
{
$rval = ($rval === 'true' or $rval != 0);
}
else if($coerce === JSONLD_XSD_DOUBLE)
{
$rval = floatval($rval);
}
else if($coerce === JSONLD_XSD_INTEGER)
{
$rval = intval($rval);
}
}
}
// no type-coercion, just copy value
else
{
$rval = _clone($value);
}
// compact IRI
if($type === JSONLD_XSD_ANY_URI)
{
if(is_object($rval))
{
$rval->{'@iri'} = _compactIri($ctx, $rval->{'@iri'}, $usedCtx);
}
else
{
$rval = _compactIri($ctx, $rval, $usedCtx);
}
}
}
return $rval;
}
/**
* Recursively expands a value using the given context. Any context in
* the value will be removed.
*
* @param ctx the context.
* @param property the property that points to the value, NULL for none.
* @param value the value to expand.
* @param expandSubjects true to expand subjects (normalize), false not to.
*
* @return the expanded value.
*/
function _expand($ctx, $property, $value, $expandSubjects)
{
$rval;
// TODO: add data format error detection?
// if no property is specified and the value is a string (this means the
// value is a property itself), expand to an IRI
if($property === null and is_string($value))
{
$rval = _expandTerm($ctx, $value, null);
}
else if(is_array($value))
{
// recursively add expanded values to array
$rval = array();
foreach($value as $v)
{
$rval[] = _expand($ctx, $property, $v, $expandSubjects);
}
}
else if(is_object($value))
{
// value has sub-properties if it doesn't define a literal or IRI value
if(!(isset($value->{'@literal'}) or isset($value->{'@iri'})))
{
// if value has a context, use it
if(isset($value->{'@context'}))
{
$ctx = jsonld_merge_contexts($ctx, $value->{'@context'});
}
// recursively handle sub-properties that aren't a sub-context
$rval = new stdClass();
foreach($value as $key => $v)
{
// preserve frame keywords
if($key === '@embed' or $key === '@explicit')
{
_setProperty($rval, $key, _clone($v));
}
else if($key !== '@context')
{
// set object to expanded property
_setProperty(
$rval, _expandTerm($ctx, $key, null),
_expand($ctx, $key, $v, $expandSubjects));
}
}
}
// value is already expanded
else
{
$rval = _clone($value);
}
}
else
{
// do type coercion
$coerce = _getCoerceType($ctx, $property, null);
// automatic coercion for basic JSON types
if($coerce === null and !is_string($value) &&
(is_numeric($value) or is_bool($value)))
{
if(is_bool($value))
{
$coerce = JSONLD_XSD_BOOLEAN;
}
else if(is_int($value))
{
$coerce = JSONLD_XSD_INTEGER;
}
else
{
$coerce = JSONLD_XSD_DOUBLE;
}
}
// coerce to appropriate datatype, only expand subjects if requested
if($coerce !== null and ($property !== __S or $expandSubjects))
{
$rval = new stdClass();
// expand IRI
if($coerce === JSONLD_XSD_ANY_URI)
{
$rval->{'@iri'} = _expandTerm($ctx, $value, null);
}
// other datatype
else
{
$rval->{'@datatype'} = $coerce;
if($coerce === JSONLD_XSD_DOUBLE)
{
// do special JSON-LD double format
$value = preg_replace(
'/(e(?:\+|-))([0-9])$/', '${1}0${2}',
sprintf('%1.6e', $value));
}
else if($coerce === JSONLD_XSD_BOOLEAN)
{
$value = $value ? 'true' : 'false';
}
$rval->{'@literal'} = '' . $value;
}
}
// nothing to coerce
else
{
$rval = '' . $value;
}
}
return $rval;
}
function _isBlankNodeIri($v)
{
return strpos($v, '_:') === 0;
}
function _isNamedBlankNode($v)
{
// look for "_:" at the beginning of the subject
return (
is_object($v) and isset($v->{__S}) and
isset($v->{__S}->{'@iri'}) and _isBlankNodeIri($v->{__S}->{'@iri'}));
}
function _isBlankNode($v)
{
// look for no subject or named blank node
return (
is_object($v) and
!(isset($v->{'@iri'}) or isset($v->{'@literal'})) and
(!isset($v->{__S}) or _isNamedBlankNode($v)));
}
/**
* Compares two values.
*
* @param v1 the first value.
* @param v2 the second value.
*
* @return -1 if v1 < v2, 0 if v1 == v2, 1 if v1 > v2.
*/
function _compare($v1, $v2)
{
$rval = 0;
if(is_array($v1) and is_array($v2))
{
$length = count($v1);
for($i = 0; $i < $length and $rval === 0; ++$i)
{
$rval = _compare($v1[$i], $v2[$i]);
}
}
else
{
$rval = ($v1 < $v2 ? -1 : ($v1 > $v2 ? 1 : 0));
}
return $rval;
}
/**
* Compares two keys in an object. If the key exists in one object
* and not the other, that object is less. If the key exists in both objects,
* then the one with the lesser value is less.
*
* @param o1 the first object.
* @param o2 the second object.
* @param key the key.
*
* @return -1 if o1 < o2, 0 if o1 == o2, 1 if o1 > o2.
*/
function _compareObjectKeys($o1, $o2, $key)
{
$rval = 0;
if(isset($o1->$key))
{
if(isset($o2->$key))
{
$rval = _compare($o1->$key, $o2->$key);
}
else
{
$rval = -1;
}
}
else if(isset($o2->$key))
{
$rval = 1;
}
return $rval;
}
/**
* Compares two object values.
*
* @param o1 the first object.
* @param o2 the second object.
*
* @return -1 if o1 < o2, 0 if o1 == o2, 1 if o1 > o2.
*/
function _compareObjects($o1, $o2)
{
$rval = 0;
if(is_string($o1))
{
if(!is_string($o2))
{
$rval = -1;
}
else
{
$rval = _compare($o1, $o2);
}
}
else if(is_string($o2))
{
$rval = 1;
}
else
{
$rval = _compareObjectKeys($o1, $o2, '@literal');
if($rval === 0)
{
if(isset($o1->{'@literal'}))
{
$rval = _compareObjectKeys($o1, $o2, '@datatype');
if($rval === 0)
{
$rval = _compareObjectKeys($o1, $o2, '@language');
}
}
// both are '@iri' objects
else
{
$rval = _compare($o1->{'@iri'}, $o2->{'@iri'});
}
}
}
return $rval;
}
/**
* Filter function for bnodes.
*
* @param e the array element.
*
* @return true to return the element in the filter results, false not to.
*/
function _filterBlankNodes($e)
{
return (is_string($e) or
!(isset($e->{'@iri'}) and _isBlankNodeIri($e->{'@iri'})));
}
/**
* Compares the object values between two bnodes.
*
* @param a the first bnode.
* @param b the second bnode.
*
* @return -1 if a < b, 0 if a == b, 1 if a > b.
*/
function _compareBlankNodeObjects($a, $b)
{
$rval = 0;
/*
3. For each property, compare sorted object values.
3.1. The bnode with fewer objects is first.
3.2. For each object value, compare only literals and non-bnodes.
3.2.1. The bnode with fewer non-bnodes is first.
3.2.2. The bnode with a string object is first.
3.2.3. The bnode with the alphabetically-first string is first.
3.2.4. The bnode with a @literal is first.
3.2.5. The bnode with the alphabetically-first @literal is first.
3.2.6. The bnode with the alphabetically-first @datatype is first.
3.2.7. The bnode with a @language is first.
3.2.8. The bnode with the alphabetically-first @language is first.
3.2.9. The bnode with the alphabetically-first @iri is first.
*/
foreach($a as $p => $value)
{
// step #3.1
$lenA = is_array($a->$p) ? count($a->$p) : 1;
$lenB = is_array($b->$p) ? count($b->$p) : 1;
$rval = _compare($lenA, $lenB);
// step #3.2.1
if($rval === 0)
{
// normalize objects to an array
$objsA = $a->$p;
$objsB = $b->$p;
if(!is_array($objsA))
{
$objsA = array($objsA);
$objsB = array($objsB);
}
// filter non-bnodes (remove bnodes from comparison)
$objsA = array_filter($objsA, '_filterBlankNodes');
$objsB = array_filter($objsB, '_filterBlankNodes');
$objsALen = count($objsA);
$rval = _compare($objsALen, count($objsB));
}
// steps #3.2.2-3.2.9
if($rval === 0)
{
usort($objsA, '_compareObjects');
usort($objsB, '_compareObjects');
for($i = 0; $i < $objsALen and $rval === 0; ++$i)
{
$rval = _compareObjects($objsA[$i], $objsB[$i]);
}
}
if($rval !== 0)
{
break;
}
}
return $rval;
}
/**
* A blank node name generator that uses a prefix and counter.
*/
class NameGenerator
{
public function __construct($prefix)
{
$this->count = -1;
$this->base = '_:' . $prefix;
}
public function next()
{
$this->count += 1;
return $this->current();
}
public function current()
{
return $this->base . $this->count;
}
public function inNamespace($iri)
{
return strpos($iri, $this->base) === 0;
}
}
/**
* Populates a map of all named subjects from the given input and an array
* of all unnamed bnodes (includes embedded ones).
*
* @param input the input (must be expanded, no context).
* @param subjects the subjects map to populate.
* @param bnodes the bnodes array to populate.
*/
function _collectSubjects($input, $subjects, $bnodes)
{
if(is_array($input))
{
foreach($input as $value)
{
_collectSubjects($value, $subjects, $bnodes);
}
}
else if(is_object($input))
{
if(isset($input->{__S}))
{
// graph literal
if(is_array($input->{__S}))
{
_collectSubjects($input->{__S}, $subjects, $bnodes);
}
// named subject
else
{
$subjects->{$input->{__S}->{'@iri'}} = $input;
}
}
// unnamed blank node
else if(_isBlankNode($input))
{
$bnodes[] = $input;
}
// recurse through subject properties
foreach($input as $value)
{
_collectSubjects($value, $subjects, $bnodes);
}
}
}
/**
* Filters duplicate IRIs.
*/
class DuplicateIriFilter
{
public function __construct($iri)
{
$this->iri = $iri;
}
public function filter($e)
{
return (is_object($e) and isset($e->{'@iri'}) and
$e->{'@iri'} === $this->iri);
}
}
/**
* Flattens the given value into a map of unique subjects. It is assumed that
* all blank nodes have been uniquely named before this call. Array values for
* properties will be sorted.
*
* @param parent the value's parent, NULL for none.
* @param parentProperty the property relating the value to the parent.
* @param value the value to flatten.
* @param subjects the map of subjects to write to.
*/
function _flatten($parent, $parentProperty, $value, $subjects)
{
$flattened = null;
if(is_array($value))
{
// list of objects or a disjoint graph
foreach($value as $v)
{
_flatten($parent, $parentProperty, $v, $subjects);
}
// if value is a list of objects, sort them
if(count($value) > 0 and
(is_string($value[0]) or
(is_object($value[0]) and
(isset($value[0]->{'@literal'}) or isset($value[0]->{'@iri'})))))
{
// sort values
usort($value, '_compareObjects');
}
}
else if(is_object($value))
{
// graph literal/disjoint graph
if(isset($value->{__S}) and is_array($value->{__S}))
{
// cannot flatten embedded graph literals
if($parent !== null)
{
throw new Exception('Embedded graph literals cannot be flattened.');
}
// top-level graph literal
foreach($value->{__S} as $k => $v)
{
_flatten($parent, $parentProperty, $v, $subjects);
}
}
// already-expanded value
else if(isset($value->{'@literal'}) or isset($value->{'@iri'}))
{
$flattened = _clone($value);
}
// subject
else
{
// create or fetch existing subject
if(isset($subjects->{$value->{__S}->{'@iri'}}))
{
// FIXME: __S might be a graph literal (as {})
$subject = $subjects->{$value->{__S}->{'@iri'}};
}
else
{
$subject = new stdClass();
if(isset($value->{__S}))
{
// FIXME: __S might be a graph literal (as {})
$subjects->{$value->{__S}->{'@iri'}} = $subject;
}
}
$flattened = $subject;
// flatten embeds
foreach($value as $key => $v)
{
if(is_array($v))
{
$subject->$key = new ArrayObject();
_flatten($subject->$key, null, $value->$key, $subjects);
$subject->$key = (array)$subject->$key;
if(count($subject->$key) === 1)
{
// convert subject[key] to object if only 1 value was added
$arr = $subject->$key;
$subject->$key = $arr[0];
}
}
else
{
_flatten($subject, $key, $value->$key, $subjects);
}
}
}
}
// string value
else
{
$flattened = $value;
}
// add flattened value to parent
if($flattened !== null and $parent !== null)
{
// remove top-level __s for subjects
// 'http://mypredicate': {'@subject': {'@iri': 'http://mysubject'}}
// becomes
// 'http://mypredicate': {'@iri': 'http://mysubject'}
if(is_object($flattened) and isset($flattened->{__S}))
{
$flattened = $flattened->{__S};
}
if($parent instanceof ArrayObject)
{
// do not add duplicate IRIs for the same property
$duplicate = false;
if(is_object($flattened) and isset($flattened->{'@iri'}))
{
$duplicate = count(array_filter(
(array)$parent, array(
new DuplicateIriFilter($flattened->{'@iri'}), 'filter'))) > 0;
}
if(!$duplicate)
{
$parent[] = $flattened;
}
}
else
{
$parent->$parentProperty = $flattened;
}
}
};
/**
* A MappingBuilder is used to build a mapping of existing blank node names
* to a form for serialization. The serialization is used to compare blank
* nodes against one another to determine a sort order.
*/
class MappingBuilder
{
/**
* Constructs a new MappingBuilder.
*/
public function __construct()
{
$this->count = 1;
$this->mapped = new stdClass();
$this->mapping = new stdClass();
$this->output = new stdClass();
}
/**
* Copies this MappingBuilder.
*
* @return the MappingBuilder copy.
*/
public function copy()
{
$rval = new MappingBuilder();
$rval->count = $this->count;
$rval->mapped = _clone($this->mapped);
$rval->mapping = _clone($this->mapping);
$rval->output = _clone($this->output);
return $rval;
}
/**
* Maps the next name to the given bnode IRI if the bnode IRI isn't already
* in the mapping. If the given bnode IRI is canonical, then it will be
* given a shortened form of the same name.
*
* @param iri the blank node IRI to map the next name to.
*
* @return the mapped name.
*/
public function mapNode($iri)
{
if(!isset($this->mapping->$iri))
{
if(strpos($iri, '_:c14n') === 0)
{
$this->mapping->$iri = 'c' . substr($iri, 6);
}
else
{
$this->mapping->$iri = 's' . $this->count++;
}
}
return $this->mapping->$iri;
}
}
/**
* Rotates the elements in an array one position.
*
* @param a the array.
*/
function _rotate(&$a)
{
$e = array_shift($a);
array_push($a, $e);
return $e;
}
/**
* Compares two serializations for the same blank node. If the two
* serializations aren't complete enough to determine if they are equal (or if
* they are actually equal), 0 is returned.
*
* @param s1 the first serialization.
* @param s2 the second serialization.
*
* @return -1 if s1 < s2, 0 if s1 == s2 (or indeterminate), 1 if s1 > v2.
*/
function _compareSerializations($s1, $s2)
{
$rval = 0;
$s1Len = strlen($s1);
$s2Len = strlen($s2);
if($s1Len == $s2Len)
{
$rval = strcmp($s1, $s2);
}
else
{
$rval = strncmp($s1, $s2, ($s1Len > $s2Len) ? $s2Len : $s1Len);
}
return $rval;
}
/**
* Compares two nodes based on their iris.
*
* @param a the first node.
* @param b the second node.
*
* @return -1 if iriA < iriB, 0 if iriA == iriB, 1 if iriA > iriB.
*/
function _compareIris($a, $b)
{
return _compare($a->{__S}->{'@iri'}, $b->{__S}->{'@iri'});
}
/**
* Filters blank node edges.
*
* @param e the edge to check.
*
* @return true if the edge is a blank node edge.
*/
function _filterBlankNodeEdges($e)
{
return _isBlankNodeIri($e->s);
}
/**
* Sorts mapping keys based on their associated mapping values.
*/
class MappingKeySorter
{
public function __construct($mapping)
{
$this->mapping = $mapping;
}
public function compare($a, $b)
{
return _compare($this->mapping->$a, $this->mapping->$b);
}
}
/**
* A container for JSON-LD processing state.
*/
class JsonLdProcessor
{
/**
* Constructs a JSON-LD processor.
*/
public function __construct()
{
$this->ng = new stdClass();
$this->ng->tmp = null;
$this->ng->c14n = null;
}
/**
* Normalizes a JSON-LD object.
*
* @param input the JSON-LD object to normalize.
*
* @return the normalized JSON-LD object.
*/
public function normalize($input)
{
$rval = array();
// TODO: validate context
if($input !== null)
{
// get default context
$ctx = jsonld_create_default_context();
// expand input
$expanded = _expand($ctx, null, $input, true);
// assign names to unnamed bnodes
$this->nameBlankNodes($expanded);
// flatten
$subjects = new stdClass();
_flatten(null, null, $expanded, $subjects);
// append subjects with sorted properties to array
foreach($subjects as $key => $s)
{
$sorted = new stdClass();
$keys = array_keys((array)$s);
sort($keys);
foreach($keys as $i => $k)
{
$sorted->$k = $s->$k;
}
$rval[] = $sorted;
}
// canonicalize blank nodes
$this->canonicalizeBlankNodes($rval);
// sort output
usort($rval, '_compareIris');
}
return $rval;
}
/**
* Assigns unique names to blank nodes that are unnamed in the given input.
*
* @param input the input to assign names to.
*/
public function nameBlankNodes($input)
{
// create temporary blank node name generator
$ng = $this->ng->tmp = new NameGenerator('tmp');
// collect subjects and unnamed bnodes
$subjects = new stdClass();
$bnodes = new ArrayObject();
_collectSubjects($input, $subjects, $bnodes);
// uniquely name all unnamed bnodes
foreach($bnodes as $i => $bnode)
{
if(!isset($bnode->{__S}))
{
// generate names until one is unique
while(isset($subjects->{$ng->next()}));
$bnode->{__S} = new stdClass();
$bnode->{__S}->{'@iri'} = $ng->current();
$subjects->{$ng->current()} = $bnode;
}
}
}
/**
* Renames a blank node, changing its references, etc. The method assumes
* that the given name is unique.
*
* @param b the blank node to rename.
* @param id the new name to use.
*/
public function renameBlankNode($b, $id)
{
$old = $b->{__S}->{'@iri'};
// update bnode IRI
$b->{__S}->{'@iri'} = $id;
// update subjects map
$subjects = $this->subjects;
$subjects->$id = $subjects->$old;
unset($subjects->$old);
// update reference and property lists
$this->edges->refs->$id = $this->edges->refs->$old;
$this->edges->props->$id = $this->edges->props->$old;
unset($this->edges->refs->$old);
unset($this->edges->props->$old);
// update references to this bnode
$refs = $this->edges->refs->$id->all;
foreach($refs as $i => $r)
{
$iri = $r->s;
if($iri === $old)
{
$iri = $id;
}
$ref = $subjects->$iri;
$props = $this->edges->props->$iri->all;
foreach($props as $prop)
{
if($prop->s === $old)
{
$prop->s = $id;
// normalize property to array for single code-path
$p = $prop->p;
$tmp = is_object($ref->$p) ? array($ref->$p) :
(is_array($ref->$p) ? $ref->$p : array());
$length = count($tmp);
for($n = 0; $n < $length; ++$n)
{
if(is_object($tmp[$n]) and
isset($tmp[$n]->{'@iri'}) and $tmp[$n]->{'@iri'} === $old)
{
$tmp[$n]->{'@iri'} = $id;
}
}
}
}
}
// update references from this bnode
$props = $this->edges->props->$id->all;
foreach($props as $prop)
{
$iri = $prop->s;
$refs = $this->edges->refs->$iri->all;
foreach($refs as $r)
{
if($r->s === $old)
{
$r->s = $id;
}
}
}
}
/**
* Canonically names blank nodes in the given input.
*
* @param input the flat input graph to assign names to.
*/
public function canonicalizeBlankNodes($input)
{
// create serialization state
$this->renamed = new stdClass();
$this->mappings = new stdClass();
$this->serializations = new stdClass();
// collect subjects and bnodes from flat input graph
$edges = $this->edges = new stdClass();
$edges->refs = new stdClass();
$edges->props = new stdClass();
$subjects = $this->subjects = new stdClass();
$bnodes = array();
foreach($input as $v)
{
$iri = $v->{__S}->{'@iri'};
$subjects->$iri = $v;
$edges->refs->$iri = new stdClass();
$edges->refs->$iri->all = array();
$edges->refs->$iri->bnodes = array();
$edges->props->$iri = new stdClass();
$edges->props->$iri->all = array();
$edges->props->$iri->bnodes = array();
if(_isBlankNodeIri($iri))
{
$bnodes[] = $v;
}
}
// collect edges in the graph
$this->collectEdges();
// create canonical blank node name generator
$c14n = $this->ng->c14n = new NameGenerator('c14n');
$ngTmp = $this->ng->tmp;
// rename all bnodes that happen to be in the c14n namespace
// and initialize serializations
foreach($bnodes as $i => $bnode)
{
$iri = $bnode->{__S}->{'@iri'};
if($c14n->inNamespace($iri))
{
// generate names until one is unique
while(isset($subjects->{$ngTmp->next()}));
$this->renameBlankNode($bnode, $ngTmp->current());
$iri = $bnode->{__S}->{'@iri'};
}
$this->serializations->$iri = new stdClass();
$this->serializations->$iri->props = null;
$this->serializations->$iri->refs = null;
}
// keep sorting and naming blank nodes until they are all named
while(count($bnodes) > 0)
{
usort($bnodes, array($this, 'deepCompareBlankNodes'));
// name all bnodes according to the first bnode's relation mappings
$bnode = array_shift($bnodes);
$iri = $bnode->{__S}->{'@iri'};
$dirs = array('props', 'refs');
foreach($dirs as $dir)
{
// if no serialization has been computed, name only the first node
if($this->serializations->$iri->$dir === null)
{
$mapping = new stdClass();
$mapping->$iri = 's1';
}
else
{
$mapping = $this->serializations->$iri->$dir->m;
}
// sort keys by value to name them in order
$keys = array_keys((array)$mapping);
usort($keys, array(new MappingKeySorter($mapping), 'compare'));
// name bnodes in mapping
$renamed = array();
foreach($keys as $i => $iriK)
{
if(!$c14n->inNamespace($iri) and isset($subjects->$iriK))
{
$this->renameBlankNode($subjects->$iriK, $c14n->next());
$renamed[] = $iriK;
}
}
// only keep non-canonically named bnodes
$tmp = $bnodes;
$bnodes = array();
foreach($tmp as $i => $b)
{
$iriB = $b->{__S}->{'@iri'};
if(!$c14n->inNamespace($iriB))
{
// mark serializations related to the named bnodes as dirty
foreach($renamed as $r)
{
$this->markSerializationDirty($iriB, $r, $dir);
}
$bnodes[] = $b;
}
}
}
}
// sort property lists that now have canonically-named bnodes
foreach($edges->props as $key => $value)
{
if(count($value->bnodes) > 0)
{
$bnode = $subjects->$key;
foreach($bnode as $p => $v)
{
if(strpos($p, '@') !== 0 and is_array($v))
{
usort($v, '_compareObjects');
$bnode->$p = $v;
}
}
}
}
}
/**
* Marks a relation serialization as dirty if necessary.
*
* @param iri the IRI of the bnode to check.
* @param changed the old IRI of the bnode that changed.
* @param dir the direction to check ('props' or 'refs').
*/
public function markSerializationDirty($iri, $changed, $dir)
{
$s = $this->serializations->$iri;
if($s->$dir !== null and isset($s->$dir->m->$changed))
{
$s->$dir = null;
}
}
/**
* Serializes the properties of the given bnode for its relation
* serialization.
*
* @param b the blank node.
*
* @return the serialized properties.
*/
public function serializeProperties($b)
{
$rval = '';
foreach($b as $p => $o)
{
if($p !== '@subject')
{
$first = true;
$objs = is_array($o) ? $o : array($o);
foreach($objs as $obj)
{
if($first)
{
$first = false;
}
else
{
$rval .= '|';
}
if(is_object($obj) and isset($obj->{'@iri'}) &&
_isBlankNodeIri($obj->{'@iri'}))
{
$rval .= '_:';
}
else
{
$rval .= str_replace('\\/', '/', json_encode($obj));
}
}
}
}
return $rval;
}
/**
* Recursively creates a relation serialization (partial or full).
*
* @param keys the keys to serialize in the current output.
* @param output the current mapping builder output.
* @param done the already serialized keys.
*
* @return the relation serialization.
*/
public function recursiveSerializeMapping($keys, $output, $done)
{
$rval = '';
foreach($keys as $k)
{
if(!isset($output->$k))
{
break;
}
if(isset($done->$k))
{
// mark cycle
$rval .= '_' . $k;
}
else
{
$done->$k = true;
$tmp = $output->$k;
foreach($tmp->k as $s)
{
$rval .= $s;
$iri = $tmp->m->$s;
if(isset($this->subjects->$iri))
{
$b = $this->subjects->$iri;
// serialize properties
$rval .= '<';
$rval .= $this->serializeProperties($b);
$rval .= '>';
// serialize references
$rval .= '<';
$first = true;
$refs = $this->edges->refs->$iri->all;
foreach($refs as $r)
{
if($first)
{
$first = false;
}
else
{
$rval .= '|';
}
$rval .= _isBlankNodeIri($r->s) ? '_:' : $refs->s;
}
$rval .= '>';
}
}
$rval .= $this->recursiveSerializeMapping($tmp->k, $output, $done);
}
}
return $rval;
}
/**
* Creates a relation serialization (partial or full).
*
* @param output the current mapping builder output.
*
* @return the relation serialization.
*/
public function serializeMapping($output)
{
return $this->recursiveSerializeMapping(
array('s1'), $output, new stdClass());
}
/**
* Recursively serializes adjacent bnode combinations.
*
* @param s the serialization to update.
* @param top the top of the serialization.
* @param mb the MappingBuilder to use.
* @param dir the edge direction to use ('props' or 'refs').
* @param mapped all of the already-mapped adjacent bnodes.
* @param notMapped all of the not-yet mapped adjacent bnodes.
*/
public function serializeCombos(
$s, $top, $mb, $dir, $mapped, $notMapped)
{
// copy mapped nodes
$mapped = _clone($mapped);
// handle recursion
if(count($notMapped) > 0)
{
// map first bnode in list
$mapped->{$mb->mapNode($notMapped[0]->s)} = $notMapped[0]->s;
// recurse into remaining possible combinations
$original = $mb->copy();
$notMapped = array_slice($notMapped, 1);
$rotations = max(1, count($notMapped));
for($r = 0; $r < $rotations; ++$r)
{
$m = ($r === 0) ? $mb : $original->copy();
$this->serializeCombos($s, $top, $m, $dir, $mapped, $notMapped);
// rotate not-mapped for next combination
_rotate($notMapped);
}
}
// handle final adjacent node in current combination
else
{
$keys = array_keys((array)$mapped);
sort($keys);
$mb->output->$top = new stdClass();
$mb->output->$top->k = $keys;
$mb->output->$top->m = $mapped;
// optimize away mappings that are already too large
$_s = $this->serializeMapping($mb->output);
if($s->$dir === null or _compareSerializations($_s, $s->$dir->s) <= 0)
{
$oldCount = $mb->count;
// recurse into adjacent values
foreach($keys as $i => $k)
{
$this->serializeBlankNode($s, $mapped->$k, $mb, $dir);
}
// reserialize if more nodes were mapped
if($mb->count > $oldCount)
{
$_s = $this->serializeMapping($mb->output);
}
// update least serialization if new one has been found
if($s->$dir === null or
(_compareSerializations($_s, $s->$dir->s) <= 0 and
count($_s) >= count($s->$dir->s)))
{
$s->$dir = new stdClass();
$s->$dir->s = $_s;
$s->$dir->m = $mb->mapping;
}
}
}
}
/**
* Computes the relation serialization for the given blank node IRI.
*
* @param s the serialization to update.
* @param iri the current bnode IRI to be mapped.
* @param mb the MappingBuilder to use.
* @param dir the edge direction to use ('props' or 'refs').
*/
public function serializeBlankNode($s, $iri, $mb, $dir)
{
// only do mapping if iri not already mapped
if(!isset($mb->mapped->$iri))
{
// iri now mapped
$mb->mapped->$iri = true;
$top = $mb->mapNode($iri);
// copy original mapping builder
$original = $mb->copy();
// split adjacent bnodes on mapped and not-mapped
$adjs = $this->edges->$dir->$iri->bnodes;
$mapped = new stdClass();
$notMapped = array();
foreach($adjs as $adj)
{
if(isset($mb->mapping->{$adj->s}))
{
$mapped->{$mb->mapping->{$adj->s}} = $adj->s;
}
else
{
$notMapped[] = $adj;
}
}
// TODO: ensure this optimization does not alter canonical order
// if the current bnode already has a serialization, reuse it
/*$hint = isset($this->serializations->$iri) ?
$this->serializations->$iri->$dir : null;
if($hint !== null)
{
$hm = $hint->m;
usort(notMapped,
{
return _compare(hm[a.s], hm[b.s]);
});
for($i in notMapped)
{
mapped[mb.mapNode(notMapped[i].s)] = notMapped[i].s;
}
notMapped = array();
}*/
// loop over possible combinations
$combos = max(1, count($notMapped));
for($i = 0; $i < $combos; ++$i)
{
$m = ($i === 0) ? $mb : $original->copy();
$this->serializeCombos($s, $top, $mb, $dir, $mapped, $notMapped);
}
}
}
/**
* Compares two blank nodes for equivalence.
*
* @param a the first blank node.
* @param b the second blank node.
*
* @return -1 if a < b, 0 if a == b, 1 if a > b.
*/
public function deepCompareBlankNodes($a, $b)
{
$rval = 0;
// compare IRIs
$iriA = $a->{__S}->{'@iri'};
$iriB = $b->{__S}->{'@iri'};
if($iriA === $iriB)
{
$rval = 0;
}
else
{
// do shallow compare first
$rval = $this->shallowCompareBlankNodes($a, $b);
// deep comparison is necessary
if($rval === 0)
{
// compare property edges and then reference edges
$dirs = array('props', 'refs');
for($i = 0; $rval === 0 and $i < 2; ++$i)
{
// recompute 'a' and 'b' serializations as necessary
$dir = $dirs[$i];
$sA = $this->serializations->$iriA;
$sB = $this->serializations->$iriB;
if($sA->$dir === null)
{
$mb = new MappingBuilder();
if($dir === 'refs')
{
// keep same mapping and count from 'props' serialization
$mb->mapping = _clone($sA->props->m);
$mb->count = count(array_keys((array)$mb->mapping)) + 1;
}
$this->serializeBlankNode($sA, $iriA, $mb, $dir);
}
if($sB->$dir === null)
{
$mb = new MappingBuilder();
if($dir === 'refs')
{
// keep same mapping and count from 'props' serialization
$mb->mapping = _clone($sB->props->m);
$mb->count = count(array_keys((array)$mb->mapping)) + 1;
}
$this->serializeBlankNode($sB, $iriB, $mb, $dir);
}
// compare serializations
$rval = _compare($sA->$dir->s, $sB->$dir->s);
}
}
}
return $rval;
}
/**
* Performs a shallow sort comparison on the given bnodes.
*
* @param a the first bnode.
* @param b the second bnode.
*
* @return -1 if a < b, 0 if a == b, 1 if a > b.
*/
public function shallowCompareBlankNodes($a, $b)
{
$rval = 0;
/* ShallowSort Algorithm (when comparing two bnodes):
1. Compare the number of properties.
1.1. The bnode with fewer properties is first.
2. Compare alphabetically sorted-properties.
2.1. The bnode with the alphabetically-first property is first.
3. For each property, compare object values.
4. Compare the number of references.
4.1. The bnode with fewer references is first.
5. Compare sorted references.
5.1. The bnode with the reference iri (vs. bnode) is first.
5.2. The bnode with the alphabetically-first reference iri is first.
5.3. The bnode with the alphabetically-first reference property is
first.
*/
$pA = array_keys((array)$a);
$pB = array_keys((array)$b);
// step #1
$rval = _compare(count($pA), count($pB));
// step #2
if($rval === 0)
{
sort($pA);
sort($pB);
$rval = _compare($pA, $pB);
}
// step #3
if($rval === 0)
{
$rval = _compareBlankNodeObjects($a, $b);
}
// step #4
if($rval === 0)
{
$edgesA = $this->edges->refs->{$a->{__S}->{'@iri'}}->all;
$edgesB = $this->edges->refs->{$b->{__S}->{'@iri'}}->all;
$edgesALen = count($edgesA);
$rval = _compare($edgesALen, count($edgesB));
}
// step #5
if($rval === 0)
{
for($i = 0; $i < $edgesALen and $rval === 0; ++$i)
{
$rval = $this->compareEdges($edgesA[$i], $edgesB[$i]);
}
}
return $rval;
}
/**
* Compares two edges. Edges with an IRI (vs. a bnode ID) come first, then
* alphabetically-first IRIs, then alphabetically-first properties. If a
* blank node has been canonically named, then blank nodes will be compared
* after properties (with a preference for canonically named over
* non-canonically named), otherwise they won't be.
*
* @param a the first edge.
* @param b the second edge.
*
* @return -1 if a < b, 0 if a == b, 1 if a > b.
*/
public function compareEdges($a, $b)
{
$rval = 0;
$bnodeA = _isBlankNodeIri($a->s);
$bnodeB = _isBlankNodeIri($b->s);
$c14n = $this->ng->c14n;
// if not both bnodes, one that is a bnode is greater
if($bnodeA != $bnodeB)
{
$rval = $bnodeA ? 1 : -1;
}
else
{
if(!$bnodeA)
{
$rval = _compare($a->s, $b->s);
}
if($rval === 0)
{
$rval = _compare($a->p, $b->p);
}
// do bnode IRI comparison if canonical naming has begun
if($rval === 0 and $c14n !== null)
{
$c14nA = $c14n->inNamespace($a->s);
$c14nB = $c14n->inNamespace($b->s);
if($c14nA != $c14nB)
{
$rval = $c14nA ? 1 : -1;
}
else if($c14nA)
{
$rval = _compare($a->s, $b->s);
}
}
}
return $rval;
}
/**
* Populates the given reference map with all of the subject edges in the
* graph. The references will be categorized by the direction of the edges,
* where 'props' is for properties and 'refs' is for references to a subject
* as an object. The edge direction categories for each IRI will be sorted
* into groups 'all' and 'bnodes'.
*/
public function collectEdges()
{
$refs = $this->edges->refs;
$props = $this->edges->props;
// collect all references and properties
foreach($this->subjects as $iri => $subject)
{
foreach($subject as $key => $object)
{
if($key !== __S)
{
// normalize to array for single codepath
$tmp = !is_array($object) ? array($object) : $object;
foreach($tmp as $o)
{
if(is_object($o) and isset($o->{'@iri'}) and
isset($this->subjects->{$o->{'@iri'}}))
{
$objIri = $o->{'@iri'};
// map object to this subject
$e = new stdClass();
$e->s = $iri;
$e->p = $key;
$refs->$objIri->all[] = $e;
// map this subject to object
$e = new stdClass();
$e->s = $objIri;
$e->p = $key;
$props->$iri->all[] = $e;
}
}
}
}
}
// create sorted categories
foreach($refs as $iri => $ref)
{
usort($ref->all, array($this, 'compareEdges'));
$ref->bnodes = array_filter($ref->all, '_filterBlankNodeEdges');
}
foreach($props as $iri => $prop)
{
usort($prop->all, array($this, 'compareEdges'));
$prop->bnodes = array_filter($prop->all, '_filterBlankNodeEdges');
}
}
}
/**
* Returns true if the given input is a subject and has one of the given types
* in the given frame.
*
* @param input the input.
* @param frame the frame with types to look for.
*
* @return true if the input has one of the given types.
*/
function _isType($input, $frame)
{
$rval = false;
// check if type(s) are specified in frame and input
$type = JSONLD_RDF_TYPE;
if(isset($frame->{JSONLD_RDF_TYPE}) and
is_object($input) and isset($input->{__S}) and
isset($input->{JSONLD_RDF_TYPE}))
{
$tmp = is_array($input->{JSONLD_RDF_TYPE}) ?
$input->{JSONLD_RDF_TYPE} : array($input->{JSONLD_RDF_TYPE});
$types = is_array($frame->$type) ?
$frame->{JSONLD_RDF_TYPE} : array($frame->{JSONLD_RDF_TYPE});
$length = count($types);
for($t = 0; $t < $length and !$rval; ++$t)
{
$type = $types[$t]->{'@iri'};
foreach($tmp as $e)
{
if($e->{'@iri'} === $type)
{
$rval = true;
break;
}
}
}
}
return $rval;
}
/**
* Returns true if the given input matches the given frame via duck-typing.
*
* @param input the input.
* @param frame the frame to check against.
*
* @return true if the input matches the frame.
*/
function _isDuckType($input, $frame)
{
$rval = false;
// frame must not have a specific type
if(!isset($frame->{JSONLD_RDF_TYPE}))
{
// get frame properties that must exist on input
$props = array_keys((array)$frame);
if(count($props) === 0)
{
// input always matches if there are no properties
$rval = true;
}
// input must be a subject with all the given properties
else if(is_object($input) and isset($input->{__S}))
{
$rval = true;
foreach($props as $prop)
{
if(!isset($input->$prop))
{
$rval = false;
break;
}
}
}
}
return $rval;
}
/**
* Recursively frames the given input according to the given frame.
*
* @param subjects a map of subjects in the graph.
* @param input the input to frame.
* @param frame the frame to use.
* @param embeds a map of previously embedded subjects, used to prevent cycles.
* @param options the framing options.
*
* @return the framed input.
*/
function _frame($subjects, $input, $frame, $embeds, $options)
{
$rval = null;
// prepare output, set limit, get array of frames
$limit = -1;
if(is_array($frame))
{
$rval = array();
$frames = $frame;
}
else
{
$frames = array($frame);
$limit = 1;
}
// iterate over frames adding input matches to list
$frameLen = count($frames);
$values = array();
for($i = 0; $i < $frameLen and $limit !== 0; ++$i)
{
// get next frame
$frame = $frames[$i];
if(!is_object($frame))
{
throw new Exception(
'Invalid JSON-LD frame. Frame type is not a map or array.');
}
// create array of values for each frame
$inLen = count($input);
$v = array();
for($n = 0; $n < $inLen and $limit !== 0; ++$n)
{
// add input to list if it matches frame specific type or duck-type
if(_isType($input[$n], $frame) or _isDuckType($input[$n], $frame))
{
$v[] = $input[$n];
--$limit;
}
}
$values[$i] = $v;
}
// for each matching value, add it to the output
$vaLen = count($values);
for($i1 = 0; $i1 < $vaLen; ++$i1)
{
foreach($values[$i1] as $value)
{
$frame = $frames[$i1];
// determine if value should be embedded or referenced
$embedOn = isset($frame->{'@embed'}) ?
$frame->{'@embed'} : $options->defaults->embedOn;
if(!$embedOn)
{
// if value is a subject, only use subject IRI as reference
if(is_object($value) and isset($value->{__S}))
{
$value = $value->{__S};
}
}
else if(
is_object($value) and isset($value->{__S}) and
isset($embeds->{$value->{__S}->{'@iri'}}))
{
// TODO: possibly support multiple embeds in the future ... and
// instead only prevent cycles?
throw new Exception(
'Multiple embeds of the same subject is not supported. ' .
'subject=' . $value->{__S}->{'@iri'});
}
// if value is a subject, do embedding and subframing
else if(is_object($value) and isset($value->{__S}))
{
$embeds->{$value->{__S}->{'@iri'}} = true;
// if explicit is on, remove keys from value that aren't in frame
$explicitOn = isset($frame->{'@explicit'}) ?
$frame->{'@explicit'} : $options->defaults->explicitOn;
if($explicitOn)
{
foreach($value as $key => $v)
{
// always include subject
if($key !== __S and !isset($frame->$key))
{
unset($value->$key);
}
}
}
// iterate over frame keys to do subframing
foreach($frame as $key => $v)
{
// skip keywords and type query
if(strpos($key, '@') !== 0 and $key !== JSONLD_RDF_TYPE)
{
if(isset($value->$key))
{
// build input and do recursion
$input = is_array($value->$key) ?
$value->$key : array($value->$key);
$length = count($input);
for($n = 0; $n < $length; ++$n)
{
// replace reference to subject w/subject
if(is_object($input[$n]) and
isset($input[$n]->{'@iri'}) and
isset($subjects->{$input[$n]->{'@iri'}}))
{
$input[$n] = $subjects->{$input[$n]->{'@iri'}};
}
}
$value->$key = _frame(
$subjects, $input, $frame->$key, $embeds, $options);
}
else
{
// add null property to value
$value->$key = null;
}
}
}
}
// add value to output
if($rval === null)
{
$rval = $value;
}
else
{
$rval[] = $value;
}
}
}
return $rval;
}
?>