Context Navigation

source: Dev/trunk/rdfapi/syntax/N3Parser.php @ 12

Last change on this file since 12 was 12, checked in by basvannuland, 14 years ago
Added RAP RDF API Added RDF reader writer for save and load survey
File size: 30.9 KB

Line
1	<?php
2	require_once RDFAPI_INCLUDE_DIR . 'util/Object.php';
3	require_once RDFAPI_INCLUDE_DIR . 'model/Blanknode.php';
4	require_once RDFAPI_INCLUDE_DIR . 'model/Resource.php';
5	require_once RDFAPI_INCLUDE_DIR . 'model/Literal.php';
6	require_once RDFAPI_INCLUDE_DIR . 'model/Statement.php';
7	require_once RDFAPI_INCLUDE_DIR . 'model/MemModel.php';
8	require_once RDFAPI_INCLUDE_DIR . 'constants.php';
9
10	// ----------------------------------------------------------------------------------
11	// Class: N3Parser
12	// ----------------------------------------------------------------------------------
13
14
15	/**
16	* PHP Notation3 Parser
17	*
18	* This parser can parse a subset of n3, reporting triples to a callback function
19	* or constructing a RAP Model ( http://www.wiwiss.fu-berlin.de/suhl/bizer/rdfapi )
20	*
21	* Supported N3 features:
22	* <ul>
23	* <li>Standard things, repeated triples ( ; and , ), blank nodes using [ ], self-reference ('<>')</li>
24	* <li>@prefix mappings</li>
25	* <li>= maps to owl#sameAs</li>
26	* <li>a maps to rdf-syntax-ns#type</li>
27	* <li>Literal datytype- and xmlLanguageTag support
28	* </ul>
29	* Un-supported N3 Features include:
30	* <ul>
31	* <li>Reification using { }</li>
32	* <li>. and ^ operators for tree traversal</li>
33	* <li>Any log operators, like log:forAll etc.</li>
34	* </ul>
35	*
36	* This parser is based on n3.py from Epp released 2nd March, 2002.
37	* by Sean B. Palmer
38	* ( http://infomesh.net/2002/eep/20020302-013802/n3.py )
39	*
40	* This parser is released under the GNU GPL license.
41	* ( http://www.gnu.org/licenses/gpl.txt )
42	*
43	*
44	*
45	* @author Sean B. Palmer <sean@mysterylights.com>
46	* @author Gunnar AA. Grimnes <ggrimnes@csd.abdn.ac.uk>
47	* @author Daniel Westphal <mail@d-westphal.de>
48	* @version $Id: N3Parser.php 569 2008-05-19 05:24:39Z p_frischmuth $
49	* @license GPL http://www.gnu.org/licenses/gpl.txt
50	* @package syntax
51	* @access public
52	**/
53
54	class N3Parser extends Object {
55
56
57	/* ==================== Variables ==================== */
58
59	var $Tokens;
60	var $bNode;
61	var $RDF_NS, $DAML_NS, $OWL_NS;
62	var $debug;
63	var $parseError;
64	var $parsedNamespaces = array();
65
66	/* ==================== Public Methods ==================== */
67
68	/**
69	* Constructor
70	* @access public
71	**/
72	function N3Parser() {
73	//Regular expressions:
74	$Name = '[A-Za-z0-9_@\.]+[^\.,;\[\]\s\) ]*';
75	$URI = '<[^> ]*>';
76	$bNode = '_:'.$Name;
77	$Univar = '\?'.$Name;
78	$QName = '(?:[A-Za-z][A-Za-z0-9_@\.]*)?:'.$Name;
79	$Literal = '(?:'
80	. '"(\\\"\|[^"])*"'
81	. '\|'
82	. "'(\\\'\|[^'])*'"
83	. ')';
84	# '"(?:\\"\|[^"])*"'
85	$Number = '[-+]?[0-9]+(\\.[0-9]+)?([eE][-+]?[0-9]+)?';
86	$Boolean = '@(?:true\|false)';
87	// $Literal = '"[^"\\\\](?:\\.\\[^"\\])"'; # '"(?:\\"\|[^"])"'
88	$LangTag = '@[A-Za-z\-]*[^ \^\.\;\,]';
89	$Datatype = '(\^\^)[^ ,\.;)]+';
90	$Datatype_URI = '(\^\^)'.$URI;
91	// $LLiteral = '"""[^"\\\\](?:(?:.\|"(?!""))[^"\\\\])*"""';
92	$LLiteral = '(?:'
93	. '"""[^"\\\\](?:(?:\\\\.\|"(?!""))[^"\\\\])*"""'
94	. '\|'
95	. "'''[^'\\\\](?:(?:\\\\.\|'(?!''))[^\"\\\\])*'''"
96	. ')';
97	// '"""[^"\\](?:(?:\\.\|"(?!""))[^"\\])*"""'
98	$Comment = '#.*$';
99	$Prefix = '(?:[A-Za-z][A-Za-z0-9_]*)?:';
100	$PrefixDecl = '@prefix';
101	$WS = '[ \t]';
102	$this->RDF_NS = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'; # for 'a' keyword
103	$this->DAML_NS = 'http://www.daml.org/2001/03/daml+oil#'; # for '=' keyword
104	$this->OWL_NS = 'http://www.w3.org/2002/07/owl#';
105
106	// $t = array( $LLiteral, $URI); //, $Literal, $PrefixDecl, $QName, $bNode, $Prefix,
107	// $Univar, 'a', '{', '}', '\(', '\)', '\[', '\]', ',', ';', '\.', $WS, $Comment);
108	$t = array(
109	$Datatype_URI, $Datatype, $LLiteral, $URI, $Literal,
110	$PrefixDecl, $QName, $Number, $Boolean, $bNode,
111	$Prefix, $Univar, 'a','=',
112	'{', '}', '\(', '\)', '\[', '\]', ',', ';', '\.',
113	$WS, $Comment,$LangTag
114	);
115	$this->Tokens = "/(".join($t,"\|").")/m";
116
117	$this->bNode = 0;
118	$this->debug = 0;
119	$this->bNodeMap = array();
120	$this->FixBnodes = FIX_BLANKNODES;
121	$this->parseError =false;
122	}
123
124
125	/**
126	* Sets, if BlankNode labels should be replaced by the generic label from the constants.php file
127	* default is "false" -> the used label in n3 is parsed to the model
128	* @param boolean
129	* @access public
130	**/
131	function setFixBnodes($set) {
132
133	if (($set===true) OR ($set===false)) $this->FixBnodes = $set;
134	}
135
136
137	/**
138	* This parses a N3 string and prints out the triples
139	* @param string $s
140	* @access public
141	**/
142	function parse($s) {
143	// """Get a string, tokenize, create list, convert to Eep store."""
144	$stat=$this->n3tolist($s);
145	foreach ( $stat as $t) {
146
147	if (count($t)>3) {
148	$object=$t[2];
149
150	for ($i = 3; $i < 5; $i++){
151	if ($t[$i][0]=='@')$object.=$t[$i];
152	if (substr($t[$i],0,2)=='^^')$object.=$t[$i];
153	};
154	} else {$object=$t[2];};
155
156	print '('.$t[0].', '.$t[1].', '.$object.")\n";
157
158	}
159	// return [[eep.Article(t[0]), eep.Article(t[1]), eep.Article(t[2])]
160	// for t in n3tolist(s)]
161	}
162
163
164	/**
165	* This parses a N3 string and calls func($subject, $predicate, $object) with each triple
166	* @param string $s
167	* @param string $func
168	* @access public
169	**/
170	function uparse($s,$func) {
171	// """Get a string, tokenize, create list, convert to Eep store."""
172	$stat=$this->n3tolist($s);
173	foreach ( $stat as $t) {
174
175	if (count($t)>3) {
176	$object=$t[2];
177
178	for ($i = 3; $i < 5; $i++){
179	if ($t[$i][0]=='@')$object.=$t[$i];
180	if (substr($t[$i],0,2)=='^^')$object.=$t[$i];
181	};
182	} else {$object=$t[2];};
183	// print "(".$t[0].", ".$t[1].", ".$t[2].")";
184
185	$func($t[0],$t[1],$object);
186	}
187	// return [[eep.Article(t[0]), eep.Article(t[1]), eep.Article(t[2])]
188	// for t in n3tolist(s)]
189	}
190
191
192	/**
193	* This parses a N3 string and returns a memmodel
194	* @param string $s
195	* @access public
196	* @return object Model
197	**/
198
199	function parse2model($s,$model = false) {
200	if($model == false){
201	$m=new MemModel();
202	}else{
203	$m=$model;
204	}
205	// """Get a string, tokenize, create list, convert to Eep store."""
206	$stat=$this->n3tolist($s);
207
208	foreach ( $stat as $t) {
209	$s=$this->toRDFNode($t[0],$t);
210	$p=$this->toRDFNode($t[1],$t);
211	$o=$this->toRDFNode($t[2],$t);
212
213	$new_statement= new Statement($s,$p,$o);
214
215	$m->add($new_statement);
216	// print "(".$t[0].", ".$t[1].", ".$t[2].")";
217	}
218	// return [[eep.Article(t[0]), eep.Article(t[1]), eep.Article(t[2])]
219	// for t in n3tolist(s)]
220	$m->addParsedNamespaces($this->parsedNamespaces);
221	return $m;
222	}
223
224	/**
225	* Generate a new MemModel from an URI or file.
226	*
227	* @access public
228	* @param $path
229	* @throws PhpError
230	* @return object MemModel
231	*/
232	function & generateModel($path,$dummy=false,$model=false) {
233
234	$handle = fopen($path,'r') or die("N3 Parser: Could not open File: '$path' - Stopped parsing.");
235	$done=false;
236	$input="";
237	while(!$done)
238	{
239	$input .= fread( $handle, 512 );
240	$done = feof($handle);
241
242	};
243
244
245	fclose($handle);
246
247	$m = $this->parse2model($input,$model);
248	return $m;
249	}
250
251	function generateModelFromString($n3String, $model) {
252
253	$m = $this->parse2model($n3String, $model);
254	$m->addModel($m);
255	}
256
257
258	/* ==================== Private Methods from here ==================== */
259
260	// General list processing functions
261
262	/**
263	* Returns FALSE if argument is a whitespace character
264	* @access private
265	* @param string $s
266	**/
267	function isWS($s) {
268	return !preg_match('/^(#.\|\s)$/', $s);
269	}
270
271
272
273	/**
274	* Returns true if the string is not a comment
275	* @access private
276	* @param string $s
277	* @returns boolean
278	**/
279	function notComment($s) {
280	if ($s=="") return false;
281	$N3Comment = '^[ \t]*\#';
282
283	if (ereg($N3Comment,$s)) return false;
284	else return true;
285	}
286
287	/**
288	* Removes all whitespace tokens from list
289	* @access private
290	* @param array $list
291	**/
292	function filterWs($list) {
293	// var_dump($list);
294	// """Filter whitespace from a list."""
295
296	return array_filter($list, array($this,"isWS"));
297	}
298
299	/**
300	* converts a string to its unicode NFC form (e.g. \uHHHH or \UHHHHHHHH).
301	*
302	* @param String $str
303	* @return String
304	* @access private
305	*
306	*/
307	function str2unicode_nfc($str=""){
308	$result="";
309	/* try to detect encoding */
310	$tmp=str_replace("?", "", $str);
311	if(strpos(utf8_decode($tmp), "?")===false){
312	$str=utf8_decode($str);
313	}
314	for($i=0,$i_max=strlen($str);$i<$i_max;$i++){
315	$nr=0;/* unicode dec nr */
316	/* char */
317	$char=$str[$i];
318	/* utf8 binary */
319	$utf8_char=utf8_encode($char);
320	$bytes=strlen($utf8_char);
321	if($bytes==1){
322	/* 0####### (0-127) */
323	$nr=ord($utf8_char);
324	}
325	elseif($bytes==2){
326	/* 110##### 10###### = 192+x 128+x */
327	$nr=((ord($utf8_char[0])-192)*64) + (ord($utf8_char[1])-128);
328	}
329	elseif($bytes==3){
330	/* 1110#### 10###### 10###### = 224+x 128+x 128+x */
331	$nr=((ord($utf8_char[0])-224)4096) + ((ord($utf8_char[1])-128)64) + (ord($utf8_char[2])-128);
332	}
333	elseif($bytes==4){
334	/* 1111#### 10###### 10###### 10###### = 240+x 128+x 128+x 128+x */
335	$nr=((ord($utf8_char[0])-240)262144) + ((ord($utf8_char[1])-128)4096) + ((ord($utf8_char[2])-128)*64) + (ord($utf8_char[3])-128);
336	}
337	/* result (see http://www.w3.org/TR/rdf-testcases/#ntrip_strings) */
338	if($nr<9){/* #x0-#x8 (0-8) */
339	$result.="\\u".sprintf("%04X",$nr);
340	}
341	elseif($nr==9){/* #x9 (9) */
342	$result.='\t';
343	}
344	elseif($nr==10){/* #xA (10) */
345	$result.='\n';
346	}
347	elseif($nr<13){/* #xB-#xC (11-12) */
348	$result.="\\u".sprintf("%04X",$nr);
349	}
350	elseif($nr==13){/* #xD (13) */
351	$result.='\t';
352	}
353	elseif($nr<32){/* #xE-#x1F (14-31) */
354	$result.="\\u".sprintf("%04X",$nr);
355	}
356	elseif($nr<34){/* #x20-#x21 (32-33) */
357	$result.=$char;
358	}
359	elseif($nr==34){/* #x22 (34) */
360	$result.='\"';
361	}
362	elseif($nr<92){/* #x23-#x5B (35-91) */
363	$result.=$char;
364	}
365	elseif($nr==92){/* #x5C (92) */
366	$result.='\\';
367	}
368	elseif($nr<127){/* #x5D-#x7E (93-126) */
369	$result.=$char;
370	}
371	elseif($nr<65536){/* #x7F-#xFFFF (128-65535) */
372	$result.="\\u".sprintf("%04X",$nr);
373	}
374	elseif($nr<1114112){/* #x10000-#x10FFFF (65536-1114111) */
375	$result.="\\U".sprintf("%08X",$nr);
376	}
377	else{
378	/* other chars are not defined => ignore */
379	}
380	}
381	return $result;
382	}
383
384
385
386	/**
387	* Gets a slice of an array.
388	* Returns the wanted slice, as well as the remainder of the array.
389	* e.g. getSpan(['p', 'q', 'r'], 1, 2) gives (['q'], ['p', 'r'])
390	* @return array
391	* @access private
392	* @param array $list
393	* @param integer $start
394	* @param integer $end
395	**/
396	function getSpan($list, $start, $end) {
397
398	$pre=array_slice($list, 0, $start);
399	$post=array_slice($list, $end);
400
401	return array(array_slice($list, $start,$end-$start),$this->array_concat($pre,$post));
402	}
403
404
405	/**
406	* Concatenates two arrays
407	* @param array $a
408	* @param array $b
409	* @returns array
410	* @access private
411	**/
412	function array_concat($a, $b) {
413	array_splice($a,count($a),0,$b);
414	return $a;
415	}
416
417	/**
418	* Returns an array with all indexes where item appears in list
419	* @param array $list
420	* @param string $item
421	* @returns array
422	* @access private
423	**/
424	function posns($list, $item) {
425	$res=array();
426	$i=0;
427	foreach ( $list as $k=>$v) {
428	if ($v === $item ) $res[]=$i;
429	$i++;
430	}
431	$res[]=$i;
432	return $res;
433	}
434
435
436	/* More N3 specific functions */
437
438	/**
439	* Returns a list of tokens
440	* @param string $s
441	* @returns array
442	* @access private
443	**/
444	function toke($s) {
445
446	// print "$s\n";
447	// """Notation3 tokenizer. Takes in a string, returns a raw token list."""
448	if (strlen($s) == 0) die('Document has no content!');
449
450	$s=str_replace("\r\n","\n",$s);
451	$s=str_replace("\r","\n",$s);
452
453
454	//$lines=explode("\n",$s);
455
456	//$reallines=array_filter($lines, array($this, "notComment"));
457	// print "LINES: ".join($reallines, " ")." :LINES\n";
458	//array_walk($reallines, array($this, "trimLine"));
459	//$res=array();
460
461	// foreach ($reallines as $l) {
462	//preg_match_all($this->Tokens, $l, $newres);
463	//$res=$this->array_concat($res,$newres[0]);
464	//}
465
466	$res=array();
467	preg_match_all($this->Tokens, $s, $newres);
468	$res=$this->array_concat($res, array_map('trim', $newres[0]));
469	//var_dump($newres[0]);
470	return $res;
471	}
472
473	/**
474	* Returns a list with the elements between start and end as one quoted string
475	* e.g. listify(["a","b","c","d"],1,2) => ["a","b c", "d"]
476	* @param array $list
477	* @param integer $start
478	* @param integer $end
479	* @returns array
480	* @access private
481	**/
482	function listify($list, $start, $end) {
483
484	//Re-form a list, merge elements start->end into one quoted element
485	//Start and end are offsets...
486
487	$l=$end-$start;
488
489	$s=array_slice($list, 0, $start);
490	$m=array_slice($list, $start,$l);
491	$e=array_slice($list, $end);
492
493	// array_push($s,"\"".join($m," ")."\"");
494	array_push($s,$m);
495
496	return $this->array_concat($s,$e);
497	}
498
499	/**
500	* Returns an array with prefixes=>namespace mappings
501	* @param array $list
502	* @access private
503	* @returns array
504	**/
505	function getPrefixes($list) {
506
507	$prefixes=array();
508	$ns=1;
509	$name=2;
510	foreach ($list as $l) {
511	if ($l=='@prefix') {
512	// while '@prefix' in list {
513
514	$pos=current($list);
515	//pos = list.index('@prefix')
516	$r = $this->getSpan($list, $pos, ($pos+4)); # processes the prefix tokens
517	$binding=$r[0];
518	$list=$r[1];
519	$prefixes[$binding[$ns]] = substr($binding[$name],1,-1);
520	$this->parsedNamespaces[substr($binding[$name],1,-1)] = substr($binding[$ns],0,-1);
521	}
522	}
523
524	if (count($prefixes)<1) $list= array_slice($list,0);
525
526	return array($prefixes, $list);
527	}
528
529	/**
530	* Callback function for replacing "a" elements with the right RDF uri.
531	* @param string $l
532	* @access private
533	**/
534	function replace_a_type(&$l,$p) {
535	if ($l=='a') $l='<'.$this->RDF_NS.'type>';
536	}
537
538	/**
539	* Callback function for replacing "=" elements with the right DAML+OIL uri.
540	* @param string $l
541	* @access private
542	**/
543	function replace_equal(&$l,$p) {
544	if ($l=='=') $l='<'.$this->OWL_NS.'sameAs>';
545	}
546
547	/**
548	* Callback function for replacing "this" elements with the right RDF uri.
549	* @param string $l
550	* @access private
551	**/
552	function replace_this($l,$p) {
553	if ($l=='this') $l='<urn:urn-n:this>';
554	}
555
556	/**
557	* Applies stuff :)
558	* Expands namespace prefixes etc.
559	* @param array $prefixes
560	* @param array $list
561	* @returns $list
562	* @access private
563	**/
564	function applyStuff($prefixes, $list)
565	{
566	array_walk($list, array($this, 'replace_a_type'));
567	array_walk($list, array($this, 'replace_equal'));
568	array_walk($list, array($this, 'replace_this'));
569
570	for ($i = 0; $i < count($list); $i++) {
571
572	if ($list[$i]=='<>') {
573	if (!isset($path)) {
574	if (!isset($_SERVER['SERVER_ADDR'])) {
575	$_SERVER['SERVER_ADDR'] = 'localhost';
576	}
577	if (!isset($_SERVER['REQUEST_URI'])) {
578	$_SERVER['REQUEST_URI'] = '/rdfapi-php';
579	}
580	$list[$i] = '<http://'.$_SERVER['SERVER_ADDR'].$_SERVER['REQUEST_URI'].'#generate_timestamp_'.time().'>';
581	} else {
582	$list[$i] = '<'.$path.'>';
583	};
584	};
585
586
587	if (preg_match('/^[-+]?[0-9]+$/', $list[$i])) {
588	//integer
589	$list[$i] = intval($list[$i]);
590	} else if (is_numeric($list[$i])) {
591	//float or decimal
592	// After conversion we cannot distinguish between both
593	$list[$i] = floatval($list[$i]);
594	} else if ((!strstr('<_"\'?.;,{}[]()@', $list[$i]{0}))
595	&& (substr($list[$i],0,3) != '^^<')
596	) {
597	//prefix or unknown
598	$_r = explode(':', $list[$i]);
599	$ns = $_r[0] . ':';
600	$name = $_r[1];
601
602	if (isset($prefixes[$ns])) {
603	$list[$i] = '<'.$prefixes[$ns].$name.'>';
604	} else if (isset($prefixes[substr($ns, 2)])) {
605	$list[$i] = '^^' . $prefixes[substr($ns, 2)] . $name . '';
606	} else {
607	//die('Prefix not declared:'.$ns);
608	$this->parseError = true;
609	trigger_error('Prefix not declared: '.$ns, E_USER_ERROR);
610	break;
611	}
612
613	} else {
614	if ($list[$i]{0} == '"') {
615	$bLiteral = true;
616	$chBase = '"';
617	} else if ($list[$i]{0} == '\'') {
618	$bLiteral = true;
619	$chBase = '\'';
620	} else {
621	$bLiteral = false;
622	}
623	if ($bLiteral) {
624	$tripleBase = $chBase . $chBase . $chBase;
625	// Congratulations - it's a literal!
626	if (substr($list[$i], 0, 3) == $tripleBase) {
627	if (substr($list[$i],-3,3) == $tripleBase) {
628	// A big literal...
629	$lit = substr($list[$i],3,-3);
630	// print "++$lit++";
631	$lit=str_replace('\n', '\\n',$lit);
632
633	//$lit=ereg_replace("[^\\]" . $chBase, "\\" . $chBase, $lit);
634	$lit = stripslashes($lit);
635
636	$list[$i] = $chBase . $lit . $chBase;
637	} else {
638	die ('Incorrect string formatting: '.substr($list[$i],-3,3));
639	}
640	} else {
641	if (strstr($list[$i],"\n")) {
642	die('Newline in literal: ' . $list[$i]);
643	}
644	$list[$i] = stripslashes($list[$i]);
645	}
646	}
647	}
648
649	if (substr($list[$i],0,2)=='^^') {
650	if ($list[$i][2]!='<') {
651	$list[$i] = '^^<' . substr($list[$i], 2) . '>';
652	}
653	};
654
655	}//foreach list item
656
657	return $list;
658	}//function applyStuff($prefixes, $list)
659
660
661
662	/**
663	* Returns an array of triples extracted from the list of n3 tokens
664	* @param array $list
665	* @returns array
666	* @access private
667	**/
668	function getStatements($list) {
669
670
671	$statements = array();
672
673	while (in_array('.', $list)) {
674	// for($i=0;$i<count($list); $i++) {
675	// if ($list[$i]==".") {
676	// while '.' in list {
677	$pos=array_search('.',$list);
678
679	$r=$this->getSpan($list, 0, $pos+1);
680
681	$statement=$r[0];
682	$list = $r[1];
683
684	array_pop($statement);
685	$statements[]=$statement;
686	}
687
688	return $statements;
689	}
690
691	/**
692	* Gets a list of triples with same subject
693	* e.g. :Gunnar :firstname "Gunnar" ; :lastname "Grimnes.
694	* @param array $list
695	* @returns array
696	* @acces private
697	**/
698	function getPovs($list) {
699	$povs = array();
700	while (in_array(';', $list)) {
701	$r=$this->posns($list,';');
702	$pos=array_slice($r,0,2);
703	$r = $this->getSpan($list, $pos[0], $pos[1]);
704	$pov=$r[0];
705	$list=$r[1];
706
707	// skip lone semicolons, e.g. "<a> <b> <c> ; ."
708	if (count($pov) == 1) continue;
709
710	$povs[]=array_slice($pov,1);
711	}
712
713	return array($list, $povs);
714	}
715
716	/**
717	* Gets a list of triples with same predicate
718	* e.g. :Gunnar :likes "Cheese", "Wine".
719	* @access private
720	* @param array $list
721	* @returns array
722	**/
723	function getObjs($list) {
724
725
726	$objs = array();
727	while (in_array(",",$list)) {
728	$pos=array_search(",",$list);
729	// for($i=0;$i<count($list); $i++) {
730	// if ($list[$i]==",") {
731	// while ',' in list {
732
733
734	$get_array_fields=2;
735	if (isset ($list[$pos+2])) {
736	if (@$list[$pos+2][0]=='@') $get_array_fields++;
737	if (@$list[$pos+2][0]=='^') $get_array_fields++;
738	};
739	if (isset ($list[$pos+3])) { if (@$list[$pos+3][0]=='^') $get_array_fields++;};
740
741
742	$r=$this->getSpan($list, $pos, ($pos+$get_array_fields));
743
744	$obj=$r[0];
745	if (!isset($obj[2])) $obj[2]=' ';
746	if (!isset($obj[3])) $obj[3]=' ';
747
748	$list=$r[1];
749
750
751	$objs[]=$obj;
752
753
754	}
755
756	return array($list, $objs);
757	}
758
759	/**
760	* Does the real work, returns a list of subject, predicate, object triples.
761	* @param array $list
762	* @returns array
763	* @access private
764	**/
765	function statementize($list) {
766
767	if (count($list) == 1 && preg_match("/_".BNODE_PREFIX."[0-9]+_/",$list[0])) {
768	if ($this->debug) print "Ignored bNode exists statement. $list\n";
769	return array();
770	}
771
772
773
774	if (count($list) == 3) return array($list);
775	if (count($list) < 3) {
776	throw new Exception(
777	'N3 statement too short,'
778	. ' only ' . count($list) . ' elements instead of 3:' . "\n"
779	. implode("\n", $list)
780	);
781	}
782
783	//Get all ;
784	$r=$this->getPovs($list);
785	$spo=$r[0];
786	$po=$r[1];
787	$all=array();
788
789
790
791	// (spo, po), all = getPovs(list), []
792	$subject = $spo[0];
793	foreach ($po as $pop) {
794	// for pop in po {
795	$r=$this->getObjs($pop);
796
797	$myPo=$r[0];
798	$obj=$r[1];
799	//myPo, obj = getObjs(pop)
800
801	if (!isset($myPo[2])) $myPo[2]=' ';
802	if (!isset($myPo[3])) $myPo[3]=' ';
803
804
805	$predicate = $myPo[0];
806	$all[]=array($subject,$predicate,$myPo[1],$myPo[2],$myPo[3]);
807	// all.append([subject, predicate, myPo[1]])
808
809
810
811	foreach ($obj as $o) $all[]=array($subject,$predicate, $o[1],$o[2],$o[3]);
812	// for x in obj: all.append([subject, predicate, x])
813
814	}
815
816
817
818	$r = $this->getObjs($spo);
819	$spo=$r[0];
820
821	$objs=$r[1];
822
823	//spo, objs = getObjs(spo)
824	$subject=$spo[0];
825	$predicate=$spo[1];
826
827
828	if(!isset($spo[3])) $spo[3]=' ';
829	if(!isset($spo[4])) $spo[4]=' ';
830
831	$all[]=array($subject, $predicate, $spo[2],$spo[3],$spo[4]);
832
833	foreach ($objs as $obj) $all[]=array($subject, $predicate, $obj[1],$obj[2],$obj[3]);
834
835	return $all;
836	}
837
838	/**
839	* Makes lists of elements in list into a seperate array element.
840	* e.g. doLists(["a","b","[","c","]","d"], "[","]")=> ["a","b", ["c"], "d"]
841	* @param array $list
842	* @param string $schar
843	* @param string $echar
844	* @returns array
845	* @access private
846	**/
847	function doLists($list, $schar, $echar) {
848
849	while (in_array($schar, $list)) {
850	// while schar in list {
851	$ndict = array();
852	$nestingLevel = 0;
853	$biggest = 0;
854	for ($i = 0; $i < count($list); $i++) {
855	if ($list[$i] == $schar) {
856	$nestingLevel += 1;
857	if (!in_array($nestingLevel, array_keys($ndict))) {
858	$ndict[$nestingLevel] = array(array($i));
859	} else {
860	$ndict[$nestingLevel][]=array($i);
861	}
862	}
863	if ($list[$i] == $echar) {
864	if (!in_array($nestingLevel, array_keys($ndict))) {
865	$ndict[$nestingLevel]=array(array($i));
866	} else {
867	$ndict[$nestingLevel][count($ndict[$nestingLevel])-1][]=$i;
868	$nestingLevel-= 1;
869	# elif type(list[i]) == type([]) {
870	# list[i] = doLists(list[i], schar, echar)
871	}
872	}
873	}
874	foreach (array_keys($ndict) as $key) {
875	if ($key > $biggest) $biggest = $key;
876	}
877
878	$tol = $ndict[$biggest][0];
879	$list = $this->listify($list, $tol[0], ($tol[1]+1));
880	}
881	return $list;
882	}
883
884	/**
885	* Apply doLists for all different types of list.
886	* @param array
887	* @returns array
888	* @access private
889	**/
890	function listStuff($list) {
891	# y, z = zip(['[', ']'], ['{', '}'], ['(', ')'])
892	# return map(doLists, [list, list, list], y, z).pop()
893	$list = $this->doLists($list, '[', ']');
894	$list = $this->doLists($list, '{', '}');
895	return $this->doLists($list, '(', ')');
896	}
897
898	/**
899	* Generates a new node id.
900	* @access private
901	* @returns string
902	**/
903	function bnodeID() {
904
905	$this->bNode++;
906	return "_".BNODE_PREFIX.$this->bNode."_";
907	}
908
909	/**
910	* This makes bNodes out of variables like _:a etc.
911	* @access private
912	* @param array $list
913	* @returns array
914	**/
915	function fixAnon($list) {
916	// $map=array();
917	for($i=0;$i<count($list);$i++) {
918	$l=$list[$i];
919	if (substr($l,0,2)=="_:") {
920	if (!isset($this->bNodeMap[$l])) {
921	$a=$this->bnodeID();
922	$this->bNodeMap[$l]=$a;
923	} else $a=$this->bNodeMap[$l];
924	$list[$i]=$a;
925	}
926	}
927	return $list;
928	}
929
930	/**
931	* This makes [ ] lists into bnodes.
932	* @access private
933	* @param array $list
934	* @return array
935	**/
936	function expandLists($list) {
937
938	for($i=0;$i<count($list);$i++) {
939	if (is_array($list[$i])) {
940	if ( $list[$i][0]=='[' ) {
941	$bnode=$this->bnodeID();
942	$prop=$list[$i];
943	$list[$i]=$bnode;
944	$list[]=$bnode;
945	$list=$this->array_concat($list, array_slice($prop,1,-1));
946	$list[]='.';
947	}elseif($list[$i][0]=='(') {
948
949	$rdfNil = '<'. RDF_NAMESPACE_URI . RDF_NIL .'>';
950	$rdfFirst = '<'. RDF_NAMESPACE_URI . RDF_FIRST .'>';
951	$rdfRest = '<'. RDF_NAMESPACE_URI . RDF_REST .'>';
952
953	// local copy of list without "(" and ")"
954	$t_list = array_slice($list[$i], 1, -1);
955
956	//prepare bnodes
957	$fromBnode = $this->bnodeID();
958	$toBnode = $this->bnodeID();
959
960	//link first bnode into graph
961	$list[$i] = $fromBnode;
962
963	$count = count($t_list);
964
965	//loop through list, convert to RDF linked list
966	for ($idx = 0; $idx < $count; $idx++){
967
968	// set rdf:first
969	$list[] = $fromBnode;
970	$list[] = $rdfFirst;
971	$list[] = $t_list[$idx];
972	$list[] = '.';
973
974	// set rdf:rest (nil or next bnode)
975	if ($idx == $count - 1) {
976	$list[] = $fromBnode;
977	$list[] = $rdfRest;
978	$list[] = $rdfNil;
979	$list[] = '.';
980	}
981	else {
982	$list[] = $fromBnode;
983	$list[] = $rdfRest;
984	$list[] = $toBnode;
985	$list[] = '.';
986
987	$fromBnode = $toBnode;
988	$toBnode = $this->bnodeID();
989	}
990	}
991	}
992	else {
993	die('Only [ ] and () lists are supported!');
994	}
995	}
996
997
998	}
999	return $list;
1000	}
1001
1002	/**
1003	* Main work-horse function. This converts a N3 string to a list of statements
1004	* @param string $s
1005	* @returns array
1006	* @access private
1007	**/
1008	function n3tolist($s) {
1009
1010	// """Convert an N3 string into a list of triples as strings."""
1011	$result = array();
1012
1013	$t = $this->filterWs($this->toke($s)); # tokenize the stream, and filter whitespace tokens
1014
1015	if ($this->debug) {
1016	print "Filter WS:\n";
1017	var_dump($t);
1018	}
1019	$r=$this->getPrefixes($t); # get the prefix directives, and add to a dict
1020	$prefixes=$r[0];
1021	$t=$r[1];
1022	if ($this->debug) {
1023	print "Prefixes:\n";
1024	var_dump($prefixes);
1025	print "***\n";
1026	var_dump($t);
1027	}
1028	$t=$this->applyStuff($prefixes, $t);#apply prefixes, keywords, and string formatting
1029	if ($this->debug) {
1030	print "Stuff applied:\n";
1031	var_dump($t);
1032	}
1033
1034	$t=$this->fixAnon($t); # fix _:a anons
1035	if ($this->debug) {
1036	print "Fix anon:\n";
1037	var_dump($t);
1038	}
1039
1040	$t = $this->listStuff($t); # apply list stuff: todo
1041	if ($this->debug) {
1042	print "Lists done:\n";
1043	var_dump($t);
1044	}
1045	$t=$this->expandLists($t);
1046	if ($this->debug) {
1047	print "Lists applied:\n";
1048	var_dump($t);
1049	}
1050	$t = $this->getStatements($t); # get all of the "statements" from the stream
1051
1052	foreach ($t as $stat) {
1053	$stats = $this->statementize($stat);
1054
1055	foreach ($stats as $y) {
1056	$result[]=$y;
1057	}
1058	}
1059
1060	// for x in [statementize(stat) for stat in t] {
1061	// for y in x: result.append(y)
1062	return $result;
1063	}
1064
1065
1066
1067	/**
1068	* Constructs a RAP RDFNode from URI/Literal/Bnode
1069	* @access private
1070	* @param string $s
1071	* @returns object RDFNode
1072	**/
1073	function toRDFNode($s, $state)
1074	{
1075	$ins = substr($s, 1, -1);
1076	if ($s{0} == '"' \|\| $s{0} == '\'') {
1077	$lang = NULL;
1078
1079	if (count($state)>3) {
1080	for ($i = 3; $i < count($state); $i++) {
1081	if ($state[$i][0]=='@') {
1082	$lang = substr($state[3], 1);
1083	}
1084	if (substr($state[$i],0,2) == '^^') {
1085	$dtype = substr($state[$i],2);
1086	if ($dtype[0]=='<') {
1087	$dtype = substr($dtype,1,-1);
1088	}
1089	}
1090	}
1091	}
1092
1093
1094	if (UNIC_RDF) {
1095	$ins = $this->str2unicode_nfc($ins);
1096	}
1097	$new_Literal = new Literal($ins, $lang);
1098	if (isset($dtype)) {
1099	$new_Literal->setDatatype($dtype);
1100	}
1101	return $new_Literal;
1102	} else if (is_int($s)) {
1103	$value = new Literal($s);
1104	$value->setDatatype(XML_SCHEMA . 'integer');
1105	return $value;
1106	} else if (is_float($s)) {
1107	$value = new Literal($s);
1108	$value->setDatatype(XML_SCHEMA . 'double');
1109	return $value;
1110	} else if ($s == '@true') {
1111	$value = new Literal(true);
1112	$value->setDatatype(XML_SCHEMA . 'boolean');
1113	return $value;
1114	} else if ($s == '@false') {
1115	$value = new Literal(false);
1116	$value->setDatatype(XML_SCHEMA . 'boolean');
1117	return $value;
1118	}
1119
1120	if (strstr($s, '_' . BNODE_PREFIX)) {
1121	if (($this->FixBnodes) \|\| (!array_search($s,$this->bNodeMap))) {
1122	return new BlankNode($ins);
1123	} else {
1124	return new BlankNode(
1125	trim(
1126	substr(
1127	array_search($s, $this->bNodeMap),
1128	2
1129	)
1130	)
1131	);
1132	};
1133	}
1134
1135	return new Resource($ins);
1136	}//function toRDFNode($s, $state)
1137
1138
1139
1140
1141	} //end: N3Parser
1142
1143	?>

Note: See TracBrowser for help on using the repository browser.

Download in other formats: