1 | <?php
|
---|
2 |
|
---|
3 | // ----------------------------------------------------------------------------------
|
---|
4 | // Class: N3Serializer
|
---|
5 | // ----------------------------------------------------------------------------------
|
---|
6 |
|
---|
7 | /**
|
---|
8 | * PHP Notation3 Serializer
|
---|
9 | *
|
---|
10 | * This class serialises models to N3 Syntax.
|
---|
11 | *
|
---|
12 | * Supported N3 features:
|
---|
13 | * <ul>
|
---|
14 | * <li>Using [ ] for blank nodes, or _: if necessary</li>
|
---|
15 | * <li>Literal datatype- and xmlLanguageTag support</li>
|
---|
16 | * </ul>
|
---|
17 | *
|
---|
18 | * Un-supported N3 Features include:
|
---|
19 | * <ul>
|
---|
20 | * <li>Reification</li>
|
---|
21 | * </ul>
|
---|
22 | *
|
---|
23 | *
|
---|
24 | * @author Gunnar AA. Grimnes <ggrimnes@csd.abdn.ac.uk>
|
---|
25 | * @author Daniel Westphal <mail@d-westphal.de>
|
---|
26 | * @author Christian Weiske <cweiske@cweiske.de>
|
---|
27 | * @version $Id: N3Serializer.php 556 2008-01-22 10:52:48Z fusel2k $
|
---|
28 | * @package syntax
|
---|
29 | * @access public
|
---|
30 | **/
|
---|
31 |
|
---|
32 | define('MAGIC_STRING', '~~~');
|
---|
33 |
|
---|
34 | class N3Serializer extends Object
|
---|
35 | {
|
---|
36 |
|
---|
37 | var $debug = false;
|
---|
38 |
|
---|
39 | var $prefixes = array();
|
---|
40 | var $noPrefixes = array();
|
---|
41 |
|
---|
42 | var $done; // keeps track of already serialized resources
|
---|
43 | var $resourcetext;
|
---|
44 | var $resourcetext_taken;
|
---|
45 | var $model;
|
---|
46 | var $res;
|
---|
47 | var $anon;
|
---|
48 |
|
---|
49 | var $styleCompress = false;
|
---|
50 | var $stylePretty = false;
|
---|
51 | var $styleNest = false;
|
---|
52 |
|
---|
53 | /**
|
---|
54 | * Constructor
|
---|
55 | *
|
---|
56 | * @access public
|
---|
57 | */
|
---|
58 | function N3Serializer()
|
---|
59 | {
|
---|
60 | $this->debug = false;
|
---|
61 | }
|
---|
62 |
|
---|
63 |
|
---|
64 | /**
|
---|
65 | * Adds a new namespace prefix to use.
|
---|
66 | * Unknown namespaces will become ns0, ns1 etc.
|
---|
67 | * @access public
|
---|
68 | * @param string $s
|
---|
69 | * @returns void
|
---|
70 | **/
|
---|
71 | function addNSPrefix($ns, $prefix)
|
---|
72 | {
|
---|
73 | $this->prefixes[$ns] = $prefix;
|
---|
74 | }
|
---|
75 |
|
---|
76 |
|
---|
77 |
|
---|
78 | /**
|
---|
79 | * Clears all previously set namespace prefixes
|
---|
80 | */
|
---|
81 | function clearNSPrefixes()
|
---|
82 | {
|
---|
83 | $this->prefixes = array();
|
---|
84 | }
|
---|
85 |
|
---|
86 |
|
---|
87 |
|
---|
88 | /**
|
---|
89 | * Add a namespace that shall not get shortened by using a prefix.
|
---|
90 | *
|
---|
91 | * @param string $ns Namespace URI like "http://example.com/"
|
---|
92 | */
|
---|
93 | function addNoNSPrefix($ns)
|
---|
94 | {
|
---|
95 | $this->noPrefixes[$ns] = true;
|
---|
96 | }
|
---|
97 |
|
---|
98 |
|
---|
99 |
|
---|
100 | /**
|
---|
101 | * Clears all previously set noNamespace prefixes
|
---|
102 | */
|
---|
103 | function clearNoNSPrefixes()
|
---|
104 | {
|
---|
105 | $this->noPrefixes = array();
|
---|
106 | }
|
---|
107 |
|
---|
108 |
|
---|
109 |
|
---|
110 | /**
|
---|
111 | * Serializes a model to N3 syntax.
|
---|
112 | *
|
---|
113 | * @param object Model $model
|
---|
114 | * @return string
|
---|
115 | * @access public
|
---|
116 | */
|
---|
117 | function &serialize(&$m)
|
---|
118 | {
|
---|
119 | if (is_a($m, 'DbModel')) {
|
---|
120 | $m=$m->getMemModel();
|
---|
121 | }
|
---|
122 |
|
---|
123 | $this->reset();
|
---|
124 | $this->model = $m;
|
---|
125 | $this->res = "";
|
---|
126 |
|
---|
127 | // copy default namespaces
|
---|
128 | global $default_prefixes;
|
---|
129 | foreach($default_prefixes as $prefix => $namespace) {
|
---|
130 | $this->addNSPrefix($namespace,$prefix);
|
---|
131 | }
|
---|
132 |
|
---|
133 | $nps= $this->model->getParsedNamespaces();
|
---|
134 | if ($nps!=false) {
|
---|
135 | foreach ($nps as $uri => $prefix){
|
---|
136 | $this->addNSPrefix($uri,$prefix);
|
---|
137 | }
|
---|
138 | }
|
---|
139 |
|
---|
140 | $namespaces = array();
|
---|
141 | $count = array();
|
---|
142 | $resources = array();
|
---|
143 | foreach ($this->model->triples as $t) {
|
---|
144 | $s = $t->getSubject();
|
---|
145 | if (is_a($s, 'Resource')) {
|
---|
146 | $namespaces[$s->getNamespace()] = 1;
|
---|
147 | }
|
---|
148 | $p = $t->getPredicate();
|
---|
149 | if (is_a($p, 'Resource')) {
|
---|
150 | $namespaces[$p->getNamespace()] = 1;
|
---|
151 | }
|
---|
152 | $o = $t->getObject();
|
---|
153 | if (is_a($o, 'Resource')) {
|
---|
154 | $namespaces[$o->getNamespace()] = 1;
|
---|
155 | }
|
---|
156 | $uri = $s->getURI();
|
---|
157 |
|
---|
158 | if (isset($count[$uri])) {
|
---|
159 | $count[$uri]++;
|
---|
160 | } else {
|
---|
161 | $count[$uri] = 0;
|
---|
162 | $resources[$uri] = $s;
|
---|
163 | }
|
---|
164 |
|
---|
165 | if ($this->styleNest && is_a($s, 'BlankNode')) {
|
---|
166 | //make sure blank nodes are sorted *after* normal nodes
|
---|
167 | //so that they can be included
|
---|
168 | $count[$uri] -= 0.00001;
|
---|
169 | }
|
---|
170 | }
|
---|
171 |
|
---|
172 | if (!HIDE_ADVERTISE) {
|
---|
173 | $this->res .= '# Generated by N3Serializer.php from RDF RAP.'.LINEFEED
|
---|
174 | .'# http://www.wiwiss.fu-berlin.de/suhl/bizer/rdfapi/index.html'
|
---|
175 | .LINEFEED.LINEFEED;
|
---|
176 | }
|
---|
177 |
|
---|
178 | $this->doNamespaces($namespaces);
|
---|
179 |
|
---|
180 | $this->res .= LINEFEED.LINEFEED;
|
---|
181 |
|
---|
182 | arsort($count);
|
---|
183 |
|
---|
184 | foreach ( $count as $k => $v) {
|
---|
185 | $this->doResource($resources[$k]);
|
---|
186 | // $this->res.=" .\n";
|
---|
187 | }
|
---|
188 |
|
---|
189 | //make all replacements
|
---|
190 | do {
|
---|
191 | $bReplacements = false;
|
---|
192 | foreach ($this->resourcetext as $r => $t) {
|
---|
193 | if (preg_match_all('/'.MAGIC_STRING.'([^ ]+)'.MAGIC_STRING.'/', $t, $ms, PREG_SET_ORDER)) {
|
---|
194 | foreach ($ms as $mseach) {
|
---|
195 | $rp = $this->resourcetext[$mseach[1]];
|
---|
196 | $this->resourcetext[$r] = preg_replace('/'.MAGIC_STRING.$mseach[1].MAGIC_STRING.'/', $rp, $t);
|
---|
197 | $bReplacements = true;
|
---|
198 | }
|
---|
199 | }
|
---|
200 |
|
---|
201 | }
|
---|
202 | } while ($bReplacements);
|
---|
203 |
|
---|
204 | //after all replacements took place, put the lines out
|
---|
205 | $c = 0;
|
---|
206 | foreach ($this->resourcetext as $r => $t) {
|
---|
207 | if ($this->debug) {
|
---|
208 | $this->res .= $c . ': ';
|
---|
209 | }
|
---|
210 | if (!(isset($this->resourcetext_taken[$r]) && $this->resourcetext_taken[$r]>0)) {
|
---|
211 | $this->res .= $t . ' .' . LINEFEED;
|
---|
212 | if ($this->stylePretty) {
|
---|
213 | $this->res .= LINEFEED;
|
---|
214 | }
|
---|
215 | } else if ($this->debug ) {
|
---|
216 | $this->res.=' Skipping : '.$t.LINEFEED;
|
---|
217 | }
|
---|
218 | $c++;
|
---|
219 | }
|
---|
220 |
|
---|
221 | // $max=-1111;
|
---|
222 | // $maxkey="";
|
---|
223 | // foreach ($count as $k=>$c) {
|
---|
224 | // if ( $c>$max) { $maxkey=$k; $max=$c; }
|
---|
225 | // }
|
---|
226 |
|
---|
227 | // if ($this->debug) {
|
---|
228 | // print "$maxkey is subject of most triples! ($max) \n";
|
---|
229 | // }
|
---|
230 |
|
---|
231 | return $this->res;
|
---|
232 | }//function &serialize(&$m)
|
---|
233 |
|
---|
234 |
|
---|
235 |
|
---|
236 | /**
|
---|
237 | * Serializes a model and saves it into a file.
|
---|
238 | * Returns FALSE if the model couldn't be saved to the file.
|
---|
239 | *
|
---|
240 | * @param object MemModel $model
|
---|
241 | * @param string $filename
|
---|
242 | * @return boolean
|
---|
243 | * @access public
|
---|
244 | */
|
---|
245 | function saveAs(&$model, $filename)
|
---|
246 | {
|
---|
247 | // serialize model
|
---|
248 | $n3 = $this->serialize($model);
|
---|
249 |
|
---|
250 | // write serialized model to file
|
---|
251 | $file_handle = @fopen($filename, 'w');
|
---|
252 | if ($file_handle) {
|
---|
253 | fwrite($file_handle, $n3);
|
---|
254 | fclose($file_handle);
|
---|
255 | return true;
|
---|
256 | } else {
|
---|
257 | return false;
|
---|
258 | };
|
---|
259 | }
|
---|
260 |
|
---|
261 |
|
---|
262 |
|
---|
263 | /**
|
---|
264 | * Set to true, if the N3 serializer should try to compress the blank node
|
---|
265 | * syntax using [] whereever possible.
|
---|
266 | */
|
---|
267 | function setCompress($compress)
|
---|
268 | {
|
---|
269 | $this->styleCompress = $compress;
|
---|
270 | }
|
---|
271 |
|
---|
272 |
|
---|
273 |
|
---|
274 | /**
|
---|
275 | * Enables pretty printing in semicolon delimited sentences.
|
---|
276 | */
|
---|
277 | function setPrettyPrint($prettyPrint)
|
---|
278 | {
|
---|
279 | $this->stylePretty = $prettyPrint;
|
---|
280 | }
|
---|
281 |
|
---|
282 |
|
---|
283 |
|
---|
284 | /**
|
---|
285 | * Enables nesting of blank nodes with [] if
|
---|
286 | * compression is activated via @see setCompress
|
---|
287 | */
|
---|
288 | function setNest($nest)
|
---|
289 | {
|
---|
290 | $this->styleNest = $nest;
|
---|
291 | }
|
---|
292 |
|
---|
293 | /* ==================== Private Methods from here ==================== */
|
---|
294 |
|
---|
295 |
|
---|
296 | /**
|
---|
297 | * Readies this object for serializing another model
|
---|
298 | * @access private
|
---|
299 | * @param void
|
---|
300 | * @returns void
|
---|
301 | **/
|
---|
302 | function reset()
|
---|
303 | {
|
---|
304 | $this->anon = 0;
|
---|
305 | $this->done = array();
|
---|
306 | $this->resourcetext_taken = array();
|
---|
307 | $this->resourcetext = array();
|
---|
308 | $this->res = '';
|
---|
309 | $this->model = null;
|
---|
310 | }
|
---|
311 |
|
---|
312 |
|
---|
313 |
|
---|
314 | /**
|
---|
315 | * Makes ns0, ns1 etc. prefixes for unknown prefixes.
|
---|
316 | * Outputs @prefix lines.
|
---|
317 | * @access private
|
---|
318 | * @param array $n
|
---|
319 | * @returns void
|
---|
320 | **/
|
---|
321 | function doNamespaces(&$n)
|
---|
322 | {
|
---|
323 | $c = 0;
|
---|
324 | foreach ($n as $ns => $nonsense) {
|
---|
325 | if (!$ns || isset($this->noPrefixes[$ns])) {
|
---|
326 | continue;
|
---|
327 | }
|
---|
328 | if (isset($this->prefixes[$ns])) {
|
---|
329 | $p = $this->prefixes[$ns];
|
---|
330 | } else {
|
---|
331 | $p = 'ns' . $c;
|
---|
332 | $this->prefixes[$ns] = $p;
|
---|
333 | $c++;
|
---|
334 | }
|
---|
335 | $this->res .= "@prefix $p: <".$ns.'> .'.LINEFEED;
|
---|
336 | }
|
---|
337 | }
|
---|
338 |
|
---|
339 |
|
---|
340 |
|
---|
341 | /**
|
---|
342 | * Fill in $resourcetext for a single resource.
|
---|
343 | * Will recurse into Objects of triples, but should never look ? (really?)
|
---|
344 | * @param object Resource $r
|
---|
345 | * @returns boolean
|
---|
346 | * @access private
|
---|
347 | **/
|
---|
348 | function doResource(&$r, $bEmbedded = false, $strIndent = ' ')
|
---|
349 | {
|
---|
350 | //var_dump($r->getURI());
|
---|
351 |
|
---|
352 | $ts = $this->model->find($r, null, null);
|
---|
353 | if (count($ts->triples) == 0) {
|
---|
354 | if ($bEmbedded) {
|
---|
355 | $this->resourcetext[$r->getURI()] = '_:' . $r->getLabel();
|
---|
356 | }
|
---|
357 | return;
|
---|
358 | }
|
---|
359 |
|
---|
360 | $out = '';
|
---|
361 |
|
---|
362 | if (isset($this->done[$r->getURI()]) && $this->done[$r->getURI()]) {
|
---|
363 | if (!$this->styleNest && is_a($r, 'BlankNode')) {
|
---|
364 | if ($this->resourcetext_taken[$r->getURI()] == 1) {
|
---|
365 | //Oh bother, we must use the _:blah construct.
|
---|
366 | $a = $this->resourcetext[$r->getURI()];
|
---|
367 | $this->resourcetext[$r->getURI()]='_:anon'.$this->anon;
|
---|
368 | $this->resourcetext['_:anon'.$this->anon]=$this->fixAnon($a, '_:anon'.$this->anon);
|
---|
369 | $this->resourcetext_taken[$r->getURI()]=2;
|
---|
370 | $this->anon++;
|
---|
371 | }
|
---|
372 | }
|
---|
373 | return false;
|
---|
374 | }
|
---|
375 |
|
---|
376 | $this->done[$r->getURI()] = true;
|
---|
377 | $compress = false;
|
---|
378 |
|
---|
379 | if (is_a($r, 'Resource')) {
|
---|
380 | if (is_a($r, 'BlankNode')) {
|
---|
381 | //test if this blanknode is referenced somewhere
|
---|
382 | $rbn = $this->model->find(null, null, $r);
|
---|
383 | $compress = (N3SER_BNODE_SHORT || $this->styleCompress)
|
---|
384 | && (
|
---|
385 | count($rbn->triples) == 0
|
---|
386 | || (count($rbn->triples) == 1 && $bEmbedded)
|
---|
387 | );
|
---|
388 | if ($compress) {
|
---|
389 | $out.='[';
|
---|
390 | } else {
|
---|
391 | $out.='_:'.$r->getLabel();
|
---|
392 | }
|
---|
393 | } else {
|
---|
394 | $this->doURI($r, $out);
|
---|
395 | }
|
---|
396 | }
|
---|
397 |
|
---|
398 | usort($ts->triples, 'statementsorter');
|
---|
399 | $lastp = '';
|
---|
400 | $out .= ' ';
|
---|
401 |
|
---|
402 | foreach ($ts->triples as $t) {
|
---|
403 | $p = $t->getPredicate();
|
---|
404 |
|
---|
405 | if ($p === $lastp) {
|
---|
406 | $out .= ' , ';
|
---|
407 | } else {
|
---|
408 | if ($lastp!='') {
|
---|
409 | if ($this->stylePretty) {
|
---|
410 | $out .= ";\n" . $strIndent;
|
---|
411 | } else {
|
---|
412 | $out .= ' ; ';
|
---|
413 | }
|
---|
414 | }
|
---|
415 | $this->doURI($p, $out);
|
---|
416 | $lastp = $p;
|
---|
417 | }
|
---|
418 |
|
---|
419 | $out .= ' ';
|
---|
420 |
|
---|
421 | $o = $t->getObject();
|
---|
422 |
|
---|
423 | if (is_a($o, 'Literal')) {
|
---|
424 | $l = $o->getLabel();
|
---|
425 | if (strpos($l, LINEFEED) === false) {
|
---|
426 | $long = false;
|
---|
427 | } else {
|
---|
428 | $long = true;
|
---|
429 | }
|
---|
430 |
|
---|
431 | //try to be intelligent
|
---|
432 | $quoteSingle = strpos($l, '\'') !== false;
|
---|
433 | $quoteDouble = strpos($l, '"') !== false;
|
---|
434 | if ($quoteSingle && !$quoteDouble) {
|
---|
435 | $quoteChar = $long ? '"""' : '"';
|
---|
436 | } else if ($quoteDouble && !$quoteSingle) {
|
---|
437 | $quoteChar = $long ? '\'\'\'' : '\'';
|
---|
438 | } else if ($quoteDouble && $quoteSingle) {
|
---|
439 | //both quotation chars inside
|
---|
440 | $quoteChar = $long ? '"""' : '"';
|
---|
441 | $l = addslashes($l);
|
---|
442 | } else {
|
---|
443 | //no quotation chars
|
---|
444 | $quoteChar = $long ? '"""' : '"';
|
---|
445 | }
|
---|
446 | $out .= $quoteChar . $l . $quoteChar;
|
---|
447 |
|
---|
448 | if ( $o->getLanguage()!='' ) {
|
---|
449 | $out.='@'.$o->getLanguage();
|
---|
450 | }
|
---|
451 | if ( $o->getDatatype()!='' ) {
|
---|
452 | $out.='^^<'.$o->getDatatype().'>';
|
---|
453 | }
|
---|
454 | }
|
---|
455 |
|
---|
456 | if (is_a($o, 'Resource')) {
|
---|
457 | if ($this->debug) {
|
---|
458 | print 'Doing object: '.$o->getURI().LINEFEED;
|
---|
459 | }
|
---|
460 | if (is_a($o, 'BlankNode')) {
|
---|
461 | if ($this->styleNest && $this->styleCompress
|
---|
462 | && !isset($this->done[$o->getURI()])
|
---|
463 | ) {
|
---|
464 | $this->doResource($o, true, $strIndent . ' ');
|
---|
465 | $out .= MAGIC_STRING . $o->getURI() . MAGIC_STRING;
|
---|
466 | //$out .= $this->resourcetext[$o->getURI()];
|
---|
467 | $this->resourcetext_taken[$o->getURI()] = 1;
|
---|
468 | } else {
|
---|
469 | $out .= '_:'.$o->getLabel();
|
---|
470 | }
|
---|
471 | } else {
|
---|
472 | $this->doURI($o, $out);
|
---|
473 | }
|
---|
474 | }
|
---|
475 | }
|
---|
476 |
|
---|
477 | if ($compress) {
|
---|
478 | $out .= ' ]';
|
---|
479 | };
|
---|
480 |
|
---|
481 | $this->resourcetext[$r->getURI()]=$out;
|
---|
482 |
|
---|
483 | return true;
|
---|
484 | }//function doResource(&$r)
|
---|
485 |
|
---|
486 |
|
---|
487 |
|
---|
488 | /**
|
---|
489 | * Format a single URI
|
---|
490 | * @param string $s
|
---|
491 | * @return void
|
---|
492 | * @access private
|
---|
493 | **/
|
---|
494 | function doURI(&$r, &$out)
|
---|
495 | {
|
---|
496 | if ($r->getURI() == 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type') {
|
---|
497 | $out .= 'a';
|
---|
498 | return;
|
---|
499 | }
|
---|
500 | $ns = $r->getNamespace();
|
---|
501 | if ($ns != '' && !isset($this->noPrefixes[$ns])) {
|
---|
502 | $out .= $this->prefixes[$ns].':'.$r->getLocalName();
|
---|
503 | } else {
|
---|
504 | //Will this ever happen? It does, now.
|
---|
505 | $out .= '<' . $r->getURI() . '>';
|
---|
506 | }
|
---|
507 | }
|
---|
508 |
|
---|
509 |
|
---|
510 |
|
---|
511 | /**
|
---|
512 | * Fix the resourcetext for a blanknode where the _: construct was used
|
---|
513 | * @param string $s
|
---|
514 | * @param string $a
|
---|
515 | * @access private
|
---|
516 | * @return void
|
---|
517 | **/
|
---|
518 | function fixAnon($t,$a)
|
---|
519 | {
|
---|
520 | $t = preg_replace("/( \] $|^\[ )/", '', $t);
|
---|
521 | return $a . $t;
|
---|
522 | }
|
---|
523 |
|
---|
524 | }
|
---|
525 |
|
---|
526 | ?> |
---|