1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* This file is part of the sweetrdf/InMemoryStoreSqlite package and licensed under |
5
|
|
|
* the terms of the GPL-2 license. |
6
|
|
|
* |
7
|
|
|
* (c) Konrad Abicht <[email protected]> |
8
|
|
|
* (c) Benjamin Nowack |
9
|
|
|
* |
10
|
|
|
* For the full copyright and license information, please view the LICENSE |
11
|
|
|
* file that was distributed with this source code. |
12
|
|
|
*/ |
13
|
|
|
|
14
|
|
|
namespace sweetrdf\InMemoryStoreSqlite\Parser; |
15
|
|
|
|
16
|
|
|
use sweetrdf\InMemoryStoreSqlite\Log\Logger; |
17
|
|
|
use sweetrdf\InMemoryStoreSqlite\NamespaceHelper; |
18
|
|
|
use sweetrdf\InMemoryStoreSqlite\StringReader; |
19
|
|
|
|
20
|
|
|
use function sweetrdf\InMemoryStoreSqlite\calcBase; |
21
|
|
|
|
22
|
|
|
class SPARQLParser extends TurtleParser |
23
|
|
|
{ |
24
|
|
|
/** |
25
|
|
|
* @var array<mixed> |
26
|
|
|
*/ |
27
|
|
|
public array $bnode_pattern_index; |
28
|
|
|
|
29
|
98 |
|
public function __construct(Logger $logger, NamespaceHelper $namespaceHelper, StringReader $stringReader) |
30
|
|
|
{ |
31
|
98 |
|
parent::__construct($logger, $namespaceHelper, $stringReader); |
32
|
|
|
|
33
|
98 |
|
$this->bnode_prefix = 'arc'.substr(md5(uniqid(rand())), 0, 4).'b'; |
34
|
98 |
|
$this->bnode_id = 0; |
35
|
98 |
|
$this->bnode_pattern_index = ['patterns' => [], 'bnodes' => []]; |
36
|
|
|
} |
37
|
|
|
|
38
|
98 |
|
public function parse(string $q, string $path = ''): void |
39
|
|
|
{ |
40
|
98 |
|
$this->base = $path ? calcBase($path) : NamespaceHelper::BASE_NAMESPACE; |
41
|
98 |
|
$this->r = [ |
42
|
98 |
|
'base' => '', |
43
|
98 |
|
'vars' => [], |
44
|
98 |
|
'prefixes' => [], |
45
|
98 |
|
]; |
46
|
98 |
|
$this->unparsed_code = $q; |
47
|
98 |
|
list($r, $v) = $this->xQuery($q); |
48
|
98 |
|
if ($r) { |
49
|
97 |
|
$this->r['query'] = $r; |
50
|
97 |
|
$this->unparsed_code = trim($v); |
51
|
4 |
|
} elseif (!$this->logger->hasEntries('error') && !$this->unparsed_code) { |
52
|
|
|
$this->logger->error('Query not properly closed'); |
53
|
|
|
} |
54
|
98 |
|
$this->r['prefixes'] = $this->namespaceHelper->getNamespaces(); |
55
|
98 |
|
$this->r['base'] = $this->base; |
56
|
|
|
/* remove trailing comments */ |
57
|
98 |
|
while (preg_match('/^\s*(\#[^\xd\xa]*)(.*)$/si', $this->unparsed_code, $m)) { |
58
|
|
|
$this->unparsed_code = $m[2]; |
59
|
|
|
} |
60
|
98 |
|
if ($this->unparsed_code && !$this->logger->hasEntries('error')) { |
61
|
|
|
$rest = preg_replace('/[\x0a|\x0d]/i', ' ', substr($this->unparsed_code, 0, 30)); |
62
|
|
|
$msg = trim($rest) ? 'Could not properly handle "'.$rest.'"' : 'Syntax error, probably an incomplete pattern'; |
63
|
|
|
$this->logger->error($msg); |
64
|
|
|
} |
65
|
|
|
} |
66
|
|
|
|
67
|
|
|
/* 1 */ |
68
|
|
|
|
69
|
|
|
protected function xQuery($v) |
70
|
|
|
{ |
71
|
|
|
list($r, $v) = $this->xPrologue($v); |
72
|
|
|
foreach (['Select', 'Construct', 'Describe', 'Ask'] as $type) { |
73
|
|
|
$m = 'x'.$type.'Query'; |
74
|
|
|
if ((list($r, $v) = $this->$m($v)) && $r) { |
75
|
|
|
return [$r, $v]; |
76
|
|
|
} |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
return [0, $v]; |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/* 2 */ |
83
|
|
|
|
84
|
98 |
|
protected function xPrologue($v) |
85
|
|
|
{ |
86
|
98 |
|
$r = 0; |
87
|
98 |
|
if ((list($sub_r, $v) = $this->xBaseDecl($v)) && $sub_r) { |
88
|
|
|
$this->base = $sub_r; |
89
|
|
|
$r = 1; |
90
|
|
|
} |
91
|
98 |
|
while ((list($sub_r, $v) = $this->xPrefixDecl($v)) && $sub_r) { |
92
|
20 |
|
$this->namespaceHelper->setPrefix($sub_r['prefix'], $sub_r['uri']); |
93
|
20 |
|
$r = 1; |
94
|
|
|
} |
95
|
|
|
|
96
|
98 |
|
return [$r, $v]; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
/* 5.. */ |
100
|
|
|
|
101
|
98 |
|
protected function xSelectQuery($v) |
102
|
|
|
{ |
103
|
98 |
|
if ($sub_r = $this->x('SELECT\s+', $v)) { |
104
|
91 |
|
$r = [ |
105
|
91 |
|
'type' => 'select', |
106
|
91 |
|
'result_vars' => [], |
107
|
91 |
|
'dataset' => [], |
108
|
91 |
|
]; |
109
|
91 |
|
$all_vars = 0; |
110
|
91 |
|
$sub_v = $sub_r[1]; |
111
|
|
|
/* distinct, reduced */ |
112
|
91 |
|
if ($sub_r = $this->x('(DISTINCT|REDUCED)\s+', $sub_v)) { |
113
|
1 |
|
$r[strtolower($sub_r[1])] = 1; |
114
|
1 |
|
$sub_v = $sub_r[2]; |
115
|
|
|
} |
116
|
|
|
/* result vars */ |
117
|
91 |
|
if ($sub_r = $this->x('\*\s+', $sub_v)) { |
118
|
48 |
|
$all_vars = 1; |
119
|
48 |
|
$sub_v = $sub_r[1]; |
120
|
|
|
} else { |
121
|
45 |
|
while ((list($sub_r, $sub_v) = $this->xResultVar($sub_v)) && $sub_r) { |
122
|
43 |
|
$r['result_vars'][] = $sub_r; |
123
|
|
|
} |
124
|
|
|
} |
125
|
91 |
|
if (!$all_vars && !\count($r['result_vars'])) { |
126
|
2 |
|
$this->logger->error('No result bindings specified.'); |
127
|
|
|
} |
128
|
|
|
/* dataset */ |
129
|
91 |
|
while ((list($sub_r, $sub_v) = $this->xDatasetClause($sub_v)) && $sub_r) { |
130
|
29 |
|
$r['dataset'][] = $sub_r; |
131
|
|
|
} |
132
|
|
|
/* where */ |
133
|
91 |
|
if ((list($sub_r, $sub_v) = $this->xWhereClause($sub_v)) && $sub_r) { |
134
|
88 |
|
$r['pattern'] = $sub_r; |
135
|
|
|
} else { |
136
|
3 |
|
return [0, $v]; |
137
|
|
|
} |
138
|
|
|
/* solution modifier */ |
139
|
88 |
|
if ((list($sub_r, $sub_v) = $this->xSolutionModifier($sub_v)) && $sub_r) { |
140
|
14 |
|
$r = array_merge($r, $sub_r); |
141
|
|
|
} |
142
|
|
|
/* all vars */ |
143
|
88 |
|
if ($all_vars) { |
144
|
47 |
|
foreach ($this->r['vars'] as $var) { |
145
|
47 |
|
$r['result_vars'][] = ['var' => $var, 'aggregate' => 0, 'alias' => '']; |
146
|
|
|
} |
147
|
47 |
|
if (!$r['result_vars']) { |
148
|
|
|
$r['result_vars'][] = '*'; |
149
|
|
|
} |
150
|
|
|
} |
151
|
|
|
|
152
|
88 |
|
return [$r, $sub_v]; |
153
|
|
|
} |
154
|
|
|
|
155
|
92 |
|
return [0, $v]; |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
protected function xResultVar($v) |
159
|
|
|
{ |
160
|
|
|
return $this->xVar($v); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/* 6.. */ |
164
|
|
|
|
165
|
92 |
|
protected function xConstructQuery($v) |
166
|
|
|
{ |
167
|
92 |
|
if ($sub_r = $this->x('CONSTRUCT\s*', $v)) { |
168
|
3 |
|
$r = [ |
169
|
3 |
|
'type' => 'construct', |
170
|
3 |
|
'dataset' => [], |
171
|
3 |
|
]; |
172
|
3 |
|
$sub_v = $sub_r[1]; |
173
|
|
|
/* construct template */ |
174
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xConstructTemplate($sub_v)) && \is_array($sub_r)) { |
175
|
3 |
|
$r['construct_triples'] = $sub_r; |
176
|
|
|
} else { |
177
|
|
|
$this->logger->error('Construct Template not found'); |
178
|
|
|
|
179
|
|
|
return [0, $v]; |
180
|
|
|
} |
181
|
|
|
/* dataset */ |
182
|
3 |
|
while ((list($sub_r, $sub_v) = $this->xDatasetClause($sub_v)) && $sub_r) { |
183
|
|
|
$r['dataset'][] = $sub_r; |
184
|
|
|
} |
185
|
|
|
/* where */ |
186
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xWhereClause($sub_v)) && $sub_r) { |
187
|
3 |
|
$r['pattern'] = $sub_r; |
188
|
|
|
} else { |
189
|
|
|
return [0, $v]; |
190
|
|
|
} |
191
|
|
|
/* solution modifier */ |
192
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xSolutionModifier($sub_v)) && $sub_r) { |
193
|
|
|
$r = array_merge($r, $sub_r); |
194
|
|
|
} |
195
|
|
|
|
196
|
3 |
|
return [$r, $sub_v]; |
197
|
|
|
} |
198
|
|
|
|
199
|
92 |
|
return [0, $v]; |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/* 7.. */ |
203
|
|
|
|
204
|
92 |
|
protected function xDescribeQuery($v) |
205
|
|
|
{ |
206
|
92 |
|
if ($sub_r = $this->x('DESCRIBE\s+', $v)) { |
207
|
3 |
|
$r = [ |
208
|
3 |
|
'type' => 'describe', |
209
|
3 |
|
'result_vars' => [], |
210
|
3 |
|
'result_uris' => [], |
211
|
3 |
|
'dataset' => [], |
212
|
3 |
|
]; |
213
|
3 |
|
$sub_v = $sub_r[1]; |
214
|
3 |
|
$all_vars = 0; |
215
|
|
|
/* result vars/uris */ |
216
|
3 |
|
if ($sub_r = $this->x('\*\s+', $sub_v)) { |
217
|
1 |
|
$all_vars = 1; |
218
|
1 |
|
$sub_v = $sub_r[1]; |
219
|
|
|
} else { |
220
|
|
|
do { |
221
|
2 |
|
$proceed = 0; |
222
|
2 |
|
if ((list($sub_r, $sub_v) = $this->xResultVar($sub_v)) && $sub_r) { |
223
|
1 |
|
$r['result_vars'][] = $sub_r; |
224
|
1 |
|
$proceed = 1; |
225
|
|
|
} |
226
|
2 |
|
if ((list($sub_r, $sub_v) = $this->xIRIref($sub_v)) && $sub_r) { |
227
|
1 |
|
$r['result_uris'][] = $sub_r; |
228
|
1 |
|
$proceed = 1; |
229
|
|
|
} |
230
|
2 |
|
} while ($proceed); |
231
|
|
|
} |
232
|
3 |
|
if (!$all_vars && !\count($r['result_vars']) && !\count($r['result_uris'])) { |
233
|
|
|
$this->logger->error('No result bindings specified.'); |
234
|
|
|
} |
235
|
|
|
/* dataset */ |
236
|
3 |
|
while ((list($sub_r, $sub_v) = $this->xDatasetClause($sub_v)) && $sub_r) { |
237
|
|
|
$r['dataset'][] = $sub_r; |
238
|
|
|
} |
239
|
|
|
/* where */ |
240
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xWhereClause($sub_v)) && $sub_r) { |
241
|
2 |
|
$r['pattern'] = $sub_r; |
242
|
|
|
} |
243
|
|
|
/* solution modifier */ |
244
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xSolutionModifier($sub_v)) && $sub_r) { |
245
|
|
|
$r = array_merge($r, $sub_r); |
246
|
|
|
} |
247
|
|
|
/* all vars */ |
248
|
3 |
|
if ($all_vars) { |
249
|
1 |
|
foreach ($this->r['vars'] as $var) { |
250
|
1 |
|
$r['result_vars'][] = ['var' => $var, 'aggregate' => 0, 'alias' => '']; |
251
|
|
|
} |
252
|
|
|
} |
253
|
|
|
|
254
|
3 |
|
return [$r, $sub_v]; |
255
|
|
|
} |
256
|
|
|
|
257
|
92 |
|
return [0, $v]; |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
/* 8.. */ |
261
|
|
|
|
262
|
92 |
|
protected function xAskQuery($v) |
263
|
|
|
{ |
264
|
92 |
|
if ($sub_r = $this->x('ASK\s+', $v)) { |
265
|
3 |
|
$r = [ |
266
|
3 |
|
'type' => 'ask', |
267
|
3 |
|
'dataset' => [], |
268
|
3 |
|
]; |
269
|
3 |
|
$sub_v = $sub_r[1]; |
270
|
|
|
/* dataset */ |
271
|
3 |
|
while ((list($sub_r, $sub_v) = $this->xDatasetClause($sub_v)) && $sub_r) { |
272
|
2 |
|
$r['dataset'][] = $sub_r; |
273
|
|
|
} |
274
|
|
|
/* where */ |
275
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xWhereClause($sub_v)) && $sub_r) { |
276
|
3 |
|
$r['pattern'] = $sub_r; |
277
|
|
|
|
278
|
3 |
|
return [$r, $sub_v]; |
279
|
|
|
} else { |
280
|
|
|
$this->logger->error('Missing or invalid WHERE clause.'); |
281
|
|
|
} |
282
|
|
|
} |
283
|
|
|
|
284
|
92 |
|
return [0, $v]; |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
/* 9, 10, 11, 12 */ |
288
|
|
|
|
289
|
97 |
|
protected function xDatasetClause($v) |
290
|
|
|
{ |
291
|
97 |
|
if ($r = $this->x('FROM(\s+NAMED)?\s+', $v)) { |
292
|
31 |
|
$named = $r[1] ? 1 : 0; |
293
|
31 |
|
if ((list($r, $sub_v) = $this->xIRIref($r[2])) && $r) { |
294
|
31 |
|
return [['graph' => $r, 'named' => $named], $sub_v]; |
295
|
|
|
} |
296
|
|
|
} |
297
|
|
|
|
298
|
97 |
|
return [0, $v]; |
299
|
|
|
} |
300
|
|
|
|
301
|
|
|
/* 13 */ |
302
|
|
|
|
303
|
97 |
|
protected function xWhereClause($v) |
304
|
|
|
{ |
305
|
97 |
|
if ($r = $this->x('(WHERE)?', $v)) { |
306
|
97 |
|
$v = $r[2]; |
307
|
|
|
} |
308
|
97 |
|
if ((list($r, $v) = $this->xGroupGraphPattern($v)) && $r) { |
309
|
94 |
|
return [$r, $v]; |
310
|
|
|
} |
311
|
|
|
|
312
|
91 |
|
return [0, $v]; |
313
|
|
|
} |
314
|
|
|
|
315
|
|
|
/* 14, 15 */ |
316
|
|
|
|
317
|
|
|
protected function xSolutionModifier($v) |
318
|
|
|
{ |
319
|
|
|
$r = []; |
320
|
|
|
if ((list($sub_r, $sub_v) = $this->xOrderClause($v)) && $sub_r) { |
321
|
|
|
$r['order_infos'] = $sub_r; |
322
|
|
|
} |
323
|
|
|
while ((list($sub_r, $sub_v) = $this->xLimitOrOffsetClause($sub_v)) && $sub_r) { |
324
|
|
|
$r = array_merge($r, $sub_r); |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
return ($v == $sub_v) ? [0, $v] : [$r, $sub_v]; |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/* 18, 19 */ |
331
|
|
|
|
332
|
97 |
|
protected function xLimitOrOffsetClause($v) |
333
|
|
|
{ |
334
|
97 |
|
if ($sub_r = $this->x('(LIMIT|OFFSET)', $v)) { |
335
|
3 |
|
$key = strtolower($sub_r[1]); |
336
|
3 |
|
$sub_v = $sub_r[2]; |
337
|
3 |
|
if ((list($sub_r, $sub_v) = $this->xINTEGER($sub_v)) && (false !== $sub_r)) { |
338
|
3 |
|
return [[$key => $sub_r], $sub_v]; |
339
|
|
|
} |
340
|
|
|
if ((list($sub_r, $sub_v) = $this->xPlaceholder($sub_v)) && (false !== $sub_r)) { |
341
|
|
|
return [[$key => $sub_r], $sub_v]; |
342
|
|
|
} |
343
|
|
|
} |
344
|
|
|
|
345
|
97 |
|
return [0, $v]; |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
/* 16 */ |
349
|
|
|
|
350
|
97 |
|
protected function xOrderClause($v) |
351
|
|
|
{ |
352
|
97 |
|
if ($sub_r = $this->x('ORDER BY\s+', $v)) { |
353
|
5 |
|
$sub_v = $sub_r[1]; |
354
|
5 |
|
$r = []; |
355
|
5 |
|
while ((list($sub_r, $sub_v) = $this->xOrderCondition($sub_v)) && $sub_r) { |
356
|
4 |
|
$r[] = $sub_r; |
357
|
|
|
} |
358
|
5 |
|
if (\count($r)) { |
359
|
4 |
|
return [$r, $sub_v]; |
360
|
|
|
} else { |
361
|
1 |
|
$this->logger->error('No order conditions specified.'); |
362
|
|
|
} |
363
|
|
|
} |
364
|
|
|
|
365
|
97 |
|
return [0, $v]; |
366
|
|
|
} |
367
|
|
|
|
368
|
|
|
/* 17, 27 */ |
369
|
|
|
|
370
|
5 |
|
protected function xOrderCondition($v) |
371
|
|
|
{ |
372
|
5 |
|
if ($sub_r = $this->x('(ASC|DESC)', $v)) { |
373
|
4 |
|
$dir = strtolower($sub_r[1]); |
374
|
4 |
|
$sub_v = $sub_r[2]; |
375
|
4 |
|
if ((list($sub_r, $sub_v) = $this->xBrackettedExpression($sub_v)) && $sub_r) { |
376
|
4 |
|
$sub_r['direction'] = $dir; |
377
|
|
|
|
378
|
4 |
|
return [$sub_r, $sub_v]; |
379
|
|
|
} |
380
|
5 |
|
} elseif ((list($sub_r, $sub_v) = $this->xVar($v)) && $sub_r) { |
381
|
|
|
$sub_r['direction'] = 'asc'; |
382
|
|
|
|
383
|
|
|
return [$sub_r, $sub_v]; |
384
|
5 |
|
} elseif ((list($sub_r, $sub_v) = $this->xBrackettedExpression($v)) && $sub_r) { |
385
|
|
|
return [$sub_r, $sub_v]; |
386
|
5 |
|
} elseif ((list($sub_r, $sub_v) = $this->xBuiltInCall($v)) && $sub_r) { |
387
|
|
|
$sub_r['direction'] = 'asc'; |
388
|
|
|
|
389
|
|
|
return [$sub_r, $sub_v]; |
390
|
5 |
|
} elseif ((list($sub_r, $sub_v) = $this->xFunctionCall($v)) && $sub_r) { |
391
|
|
|
$sub_r['direction'] = 'asc'; |
392
|
|
|
|
393
|
|
|
return [$sub_r, $sub_v]; |
394
|
|
|
} |
395
|
|
|
|
396
|
5 |
|
return [0, $v]; |
397
|
|
|
} |
398
|
|
|
|
399
|
|
|
/* 20 */ |
400
|
|
|
|
401
|
97 |
|
protected function xGroupGraphPattern($v) |
402
|
|
|
{ |
403
|
97 |
|
$pattern_id = substr(md5(uniqid(rand())), 0, 4); |
404
|
97 |
|
if ($sub_r = $this->x('\{', $v)) { |
405
|
95 |
|
$r = ['type' => 'group', 'patterns' => []]; |
406
|
95 |
|
$sub_v = $sub_r[1]; |
407
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xTriplesBlock($sub_v)) && $sub_r) { |
408
|
94 |
|
$this->indexBnodes($sub_r, $pattern_id); |
409
|
94 |
|
$r['patterns'][] = ['type' => 'triples', 'patterns' => $sub_r]; |
410
|
|
|
} |
411
|
|
|
do { |
412
|
95 |
|
$proceed = 0; |
413
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xGraphPatternNotTriples($sub_v)) && $sub_r) { |
414
|
3 |
|
$r['patterns'][] = $sub_r; |
415
|
3 |
|
$pattern_id = substr(md5(uniqid(rand())), 0, 4); |
416
|
3 |
|
$proceed = 1; |
417
|
95 |
|
} elseif ((list($sub_r, $sub_v) = $this->xFilter($sub_v)) && $sub_r) { |
418
|
23 |
|
$r['patterns'][] = ['type' => 'filter', 'constraint' => $sub_r]; |
419
|
23 |
|
$proceed = 1; |
420
|
|
|
} |
421
|
95 |
|
if ($sub_r = $this->x('\.', $sub_v)) { |
422
|
|
|
$sub_v = $sub_r[1]; |
423
|
|
|
} |
424
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xTriplesBlock($sub_v)) && $sub_r) { |
425
|
|
|
$this->indexBnodes($sub_r, $pattern_id); |
426
|
|
|
$r['patterns'][] = ['type' => 'triples', 'patterns' => $sub_r]; |
427
|
|
|
$proceed = 1; |
428
|
|
|
} |
429
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xPlaceholder($sub_v)) && $sub_r) { |
430
|
|
|
$r['patterns'][] = $sub_r; |
431
|
|
|
$proceed = 1; |
432
|
|
|
} |
433
|
95 |
|
} while ($proceed); |
434
|
95 |
|
if ($sub_r = $this->x('\}', $sub_v)) { |
435
|
94 |
|
$sub_v = $sub_r[1]; |
436
|
|
|
|
437
|
94 |
|
return [$r, $sub_v]; |
438
|
|
|
} |
439
|
1 |
|
$rest = preg_replace('/[\x0a|\x0d]/i', ' ', substr($sub_v, 0, 30)); |
440
|
1 |
|
$this->logger->error('Incomplete or invalid Group Graph pattern. Could not handle "'.$rest.'"'); |
441
|
|
|
} |
442
|
|
|
|
443
|
97 |
|
return [0, $v]; |
444
|
|
|
} |
445
|
|
|
|
446
|
94 |
|
protected function indexBnodes($triples, $pattern_id) |
447
|
|
|
{ |
448
|
94 |
|
$index_id = \count($this->bnode_pattern_index['patterns']); |
|
|
|
|
449
|
94 |
|
$index_id = $pattern_id; |
450
|
94 |
|
$this->bnode_pattern_index['patterns'][] = $triples; |
451
|
94 |
|
foreach ($triples as $t) { |
452
|
94 |
|
foreach (['s', 'p', 'o'] as $term) { |
453
|
94 |
|
if ('bnode' == $t[$term.'_type']) { |
454
|
|
|
$val = $t[$term]; |
455
|
|
|
if (isset($this->bnode_pattern_index['bnodes'][$val]) && ($this->bnode_pattern_index['bnodes'][$val] != $index_id)) { |
456
|
|
|
$this->logger->error('Re-used bnode label "'.$val.'" across graph patterns'); |
457
|
|
|
} else { |
458
|
|
|
$this->bnode_pattern_index['bnodes'][$val] = $index_id; |
459
|
|
|
} |
460
|
|
|
} |
461
|
|
|
} |
462
|
|
|
} |
463
|
|
|
} |
464
|
|
|
|
465
|
|
|
/* 22.., 25.. */ |
466
|
|
|
|
467
|
95 |
|
protected function xGraphPatternNotTriples($v) |
468
|
|
|
{ |
469
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xOptionalGraphPattern($v)) && $sub_r) { |
470
|
2 |
|
return [$sub_r, $sub_v]; |
471
|
|
|
} |
472
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xGraphGraphPattern($v)) && $sub_r) { |
473
|
|
|
return [$sub_r, $sub_v]; |
474
|
|
|
} |
475
|
95 |
|
$r = ['type' => 'union', 'patterns' => []]; |
476
|
95 |
|
$sub_v = $v; |
477
|
|
|
do { |
478
|
95 |
|
$proceed = 0; |
479
|
95 |
|
if ((list($sub_r, $sub_v) = $this->xGroupGraphPattern($sub_v)) && $sub_r) { |
480
|
1 |
|
$r['patterns'][] = $sub_r; |
481
|
1 |
|
if ($sub_r = $this->x('UNION', $sub_v)) { |
482
|
1 |
|
$sub_v = $sub_r[1]; |
483
|
1 |
|
$proceed = 1; |
484
|
|
|
} |
485
|
|
|
} |
486
|
95 |
|
} while ($proceed); |
487
|
95 |
|
$pc = \count($r['patterns']); |
488
|
95 |
|
if (1 == $pc) { |
489
|
|
|
return [$r['patterns'][0], $sub_v]; |
490
|
95 |
|
} elseif ($pc > 1) { |
491
|
1 |
|
return [$r, $sub_v]; |
492
|
|
|
} |
493
|
|
|
|
494
|
95 |
|
return [0, $v]; |
495
|
|
|
} |
496
|
|
|
|
497
|
|
|
/* 23 */ |
498
|
|
|
|
499
|
95 |
|
protected function xOptionalGraphPattern($v) |
500
|
|
|
{ |
501
|
95 |
|
if ($sub_r = $this->x('OPTIONAL', $v)) { |
502
|
2 |
|
$sub_v = $sub_r[1]; |
503
|
2 |
|
if ((list($sub_r, $sub_v) = $this->xGroupGraphPattern($sub_v)) && $sub_r) { |
504
|
2 |
|
return [['type' => 'optional', 'patterns' => $sub_r['patterns']], $sub_v]; |
505
|
|
|
} |
506
|
|
|
$this->logger->error('Missing or invalid Group Graph Pattern after OPTIONAL'); |
507
|
|
|
} |
508
|
|
|
|
509
|
95 |
|
return [0, $v]; |
510
|
|
|
} |
511
|
|
|
|
512
|
|
|
/* 24.. */ |
513
|
|
|
|
514
|
95 |
|
protected function xGraphGraphPattern($v) |
515
|
|
|
{ |
516
|
95 |
|
if ($sub_r = $this->x('GRAPH', $v)) { |
517
|
|
|
$sub_v = $sub_r[1]; |
518
|
|
|
$r = ['type' => 'graph', 'var' => '', 'uri' => '', 'patterns' => []]; |
519
|
|
|
if ((list($sub_r, $sub_v) = $this->xVar($sub_v)) && $sub_r) { |
520
|
|
|
$r['var'] = $sub_r; |
521
|
|
|
} elseif ((list($sub_r, $sub_v) = $this->xIRIref($sub_v)) && $sub_r) { |
522
|
|
|
$r['uri'] = $sub_r; |
523
|
|
|
} |
524
|
|
|
if ($r['var'] || $r['uri']) { |
525
|
|
|
if ((list($sub_r, $sub_v) = $this->xGroupGraphPattern($sub_v)) && $sub_r) { |
526
|
|
|
$r['patterns'][] = $sub_r; |
527
|
|
|
|
528
|
|
|
return [$r, $sub_v]; |
529
|
|
|
} |
530
|
|
|
$this->logger->error('Missing or invalid Graph Pattern'); |
531
|
|
|
} |
532
|
|
|
} |
533
|
|
|
|
534
|
95 |
|
return [0, $v]; |
535
|
|
|
} |
536
|
|
|
|
537
|
|
|
/* 26.., 27.. */ |
538
|
|
|
|
539
|
95 |
|
protected function xFilter($v) |
540
|
|
|
{ |
541
|
95 |
|
if ($r = $this->x('FILTER', $v)) { |
542
|
23 |
|
$sub_v = $r[1]; |
543
|
23 |
|
if ((list($r, $sub_v) = $this->xBrackettedExpression($sub_v)) && $r) { |
544
|
19 |
|
return [$r, $sub_v]; |
545
|
|
|
} |
546
|
4 |
|
if ((list($r, $sub_v) = $this->xBuiltInCall($sub_v)) && $r) { |
547
|
4 |
|
return [$r, $sub_v]; |
548
|
|
|
} |
549
|
|
|
if ((list($r, $sub_v) = $this->xFunctionCall($sub_v)) && $r) { |
550
|
|
|
return [$r, $sub_v]; |
551
|
|
|
} |
552
|
|
|
$this->logger->error('Incomplete FILTER'); |
553
|
|
|
} |
554
|
|
|
|
555
|
95 |
|
return [0, $v]; |
556
|
|
|
} |
557
|
|
|
|
558
|
|
|
/* 28.. */ |
559
|
|
|
|
560
|
5 |
|
protected function xFunctionCall($v) |
561
|
|
|
{ |
562
|
5 |
|
if ((list($r, $sub_v) = $this->xIRIref($v)) && $r) { |
563
|
|
|
if ((list($sub_r, $sub_v) = $this->xArgList($sub_v)) && $sub_r) { |
564
|
|
|
return [['type' => 'function_call', 'uri' => $r, 'args' => $sub_r], $sub_v]; |
565
|
|
|
} |
566
|
|
|
} |
567
|
|
|
|
568
|
5 |
|
return [0, $v]; |
569
|
|
|
} |
570
|
|
|
|
571
|
|
|
/* 29 */ |
572
|
|
|
|
573
|
18 |
|
protected function xArgList($v) |
574
|
|
|
{ |
575
|
18 |
|
$r = []; |
576
|
18 |
|
$sub_v = $v; |
577
|
18 |
|
$closed = 0; |
578
|
18 |
|
if ($sub_r = $this->x('\(', $sub_v)) { |
579
|
18 |
|
$sub_v = $sub_r[1]; |
580
|
|
|
do { |
581
|
18 |
|
$proceed = 0; |
582
|
18 |
|
if ((list($sub_r, $sub_v) = $this->xExpression($sub_v)) && $sub_r) { |
583
|
18 |
|
$r[] = $sub_r; |
584
|
18 |
|
if ($sub_r = $this->x('\,', $sub_v)) { |
585
|
4 |
|
$sub_v = $sub_r[1]; |
586
|
4 |
|
$proceed = 1; |
587
|
|
|
} |
588
|
|
|
} |
589
|
18 |
|
if ($sub_r = $this->x('\)', $sub_v)) { |
590
|
18 |
|
$sub_v = $sub_r[1]; |
591
|
18 |
|
$closed = 1; |
592
|
18 |
|
$proceed = 0; |
593
|
|
|
} |
594
|
18 |
|
} while ($proceed); |
595
|
|
|
} |
596
|
|
|
|
597
|
18 |
|
return $closed ? [$r, $sub_v] : [0, $v]; |
598
|
|
|
} |
599
|
|
|
|
600
|
|
|
/* 30, 31 */ |
601
|
|
|
|
602
|
92 |
|
protected function xConstructTemplate($v) |
603
|
|
|
{ |
604
|
92 |
|
if ($sub_r = $this->x('\{', $v)) { |
605
|
92 |
|
$r = []; |
606
|
92 |
|
if ((list($sub_r, $sub_v) = $this->xTriplesBlock($sub_r[1])) && \is_array($sub_r)) { |
607
|
92 |
|
$r = $sub_r; |
608
|
|
|
} |
609
|
92 |
|
if ($sub_r = $this->x('\}', $sub_v)) { |
610
|
91 |
|
return [$r, $sub_r[1]]; |
611
|
|
|
} |
612
|
|
|
} |
613
|
|
|
|
614
|
5 |
|
return [0, $v]; |
615
|
|
|
} |
616
|
|
|
|
617
|
|
|
/* 46, 47 */ |
618
|
|
|
|
619
|
27 |
|
protected function xExpression($v) |
620
|
|
|
{ |
621
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xConditionalAndExpression($v)) && $sub_r) { |
622
|
27 |
|
$r = ['type' => 'expression', 'sub_type' => 'or', 'patterns' => [$sub_r]]; |
623
|
|
|
do { |
624
|
27 |
|
$proceed = 0; |
625
|
27 |
|
if ($sub_r = $this->x('\|\|', $sub_v)) { |
626
|
|
|
$sub_v = $sub_r[1]; |
627
|
|
|
if ((list($sub_r, $sub_v) = $this->xConditionalAndExpression($sub_v)) && $sub_r) { |
628
|
|
|
$r['patterns'][] = $sub_r; |
629
|
|
|
$proceed = 1; |
630
|
|
|
} |
631
|
|
|
} |
632
|
27 |
|
} while ($proceed); |
633
|
|
|
|
634
|
27 |
|
return 1 == \count($r['patterns']) ? [$r['patterns'][0], $sub_v] : [$r, $sub_v]; |
635
|
|
|
} |
636
|
|
|
|
637
|
|
|
return [0, $v]; |
638
|
|
|
} |
639
|
|
|
|
640
|
|
|
/* 48.., 49.. */ |
641
|
|
|
|
642
|
27 |
|
protected function xConditionalAndExpression($v) |
643
|
|
|
{ |
644
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xRelationalExpression($v)) && $sub_r) { |
645
|
27 |
|
$r = ['type' => 'expression', 'sub_type' => 'and', 'patterns' => [$sub_r]]; |
646
|
|
|
do { |
647
|
27 |
|
$proceed = 0; |
648
|
27 |
|
if ($sub_r = $this->x('\&\&', $sub_v)) { |
649
|
1 |
|
$sub_v = $sub_r[1]; |
650
|
1 |
|
if ((list($sub_r, $sub_v) = $this->xRelationalExpression($sub_v)) && $sub_r) { |
651
|
1 |
|
$r['patterns'][] = $sub_r; |
652
|
1 |
|
$proceed = 1; |
653
|
|
|
} |
654
|
|
|
} |
655
|
27 |
|
} while ($proceed); |
656
|
|
|
|
657
|
27 |
|
return 1 == \count($r['patterns']) ? [$r['patterns'][0], $sub_v] : [$r, $sub_v]; |
658
|
|
|
} |
659
|
|
|
|
660
|
|
|
return [0, $v]; |
661
|
|
|
} |
662
|
|
|
|
663
|
|
|
/* 50, 51 */ |
664
|
|
|
|
665
|
27 |
|
protected function xRelationalExpression($v) |
666
|
|
|
{ |
667
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xAdditiveExpression($v)) && $sub_r) { |
668
|
27 |
|
$r = ['type' => 'expression', 'sub_type' => 'relational', 'patterns' => [$sub_r]]; |
669
|
|
|
do { |
670
|
27 |
|
$proceed = 0; |
671
|
|
|
/* don't mistake '<' + uriref with '<'-operator ("longest token" rule) */ |
672
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xIRI_REF($sub_v)) && $sub_r) { |
673
|
|
|
$this->logger->error('Expected operator, found IRIref: "'.$sub_r.'".'); |
674
|
|
|
} |
675
|
27 |
|
if ($sub_r = $this->x('(\!\=|\=\=|\=|\<\=|\>\=|\<|\>)', $sub_v)) { |
676
|
9 |
|
$op = $sub_r[1]; |
677
|
9 |
|
$sub_v = $sub_r[2]; |
678
|
9 |
|
$r['operator'] = $op; |
679
|
9 |
|
if ((list($sub_r, $sub_v) = $this->xAdditiveExpression($sub_v)) && $sub_r) { |
680
|
|
|
//$sub_r['operator'] = $op; |
681
|
9 |
|
$r['patterns'][] = $sub_r; |
682
|
9 |
|
$proceed = 1; |
683
|
|
|
} |
684
|
|
|
} |
685
|
27 |
|
} while ($proceed); |
686
|
|
|
|
687
|
27 |
|
return 1 == \count($r['patterns']) ? [$r['patterns'][0], $sub_v] : [$r, $sub_v]; |
688
|
|
|
} |
689
|
|
|
|
690
|
|
|
return [0, $v]; |
691
|
|
|
} |
692
|
|
|
|
693
|
|
|
/* 52 */ |
694
|
|
|
|
695
|
27 |
|
protected function xAdditiveExpression($v) |
696
|
|
|
{ |
697
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xMultiplicativeExpression($v)) && $sub_r) { |
698
|
27 |
|
$r = ['type' => 'expression', 'sub_type' => 'additive', 'patterns' => [$sub_r]]; |
699
|
|
|
do { |
700
|
27 |
|
$proceed = 0; |
701
|
27 |
|
if ($sub_r = $this->x('(\+|\-)', $sub_v)) { |
702
|
|
|
$op = $sub_r[1]; |
703
|
|
|
$sub_v = $sub_r[2]; |
704
|
|
|
if ((list($sub_r, $sub_v) = $this->xMultiplicativeExpression($sub_v)) && $sub_r) { |
705
|
|
|
$sub_r['operator'] = $op; |
706
|
|
|
$r['patterns'][] = $sub_r; |
707
|
|
|
$proceed = 1; |
708
|
|
|
} elseif ((list($sub_r, $sub_v) = $this->xNumericLiteral($sub_v)) && $sub_r) { |
709
|
|
|
$r['patterns'][] = ['type' => 'numeric', 'operator' => $op, 'value' => $sub_r]; |
710
|
|
|
$proceed = 1; |
711
|
|
|
} |
712
|
|
|
} |
713
|
27 |
|
} while ($proceed); |
714
|
|
|
//return array($r, $sub_v); |
715
|
27 |
|
return 1 == \count($r['patterns']) ? [$r['patterns'][0], $sub_v] : [$r, $sub_v]; |
716
|
|
|
} |
717
|
|
|
|
718
|
|
|
return [0, $v]; |
719
|
|
|
} |
720
|
|
|
|
721
|
|
|
/* 53 */ |
722
|
|
|
|
723
|
27 |
|
protected function xMultiplicativeExpression($v) |
724
|
|
|
{ |
725
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xUnaryExpression($v)) && $sub_r) { |
726
|
27 |
|
$r = ['type' => 'expression', 'sub_type' => 'multiplicative', 'patterns' => [$sub_r]]; |
727
|
|
|
do { |
728
|
27 |
|
$proceed = 0; |
729
|
27 |
|
if ($sub_r = $this->x('(\*|\/)', $sub_v)) { |
730
|
|
|
$op = $sub_r[1]; |
731
|
|
|
$sub_v = $sub_r[2]; |
732
|
|
|
if ((list($sub_r, $sub_v) = $this->xUnaryExpression($sub_v)) && $sub_r) { |
733
|
|
|
$sub_r['operator'] = $op; |
734
|
|
|
$r['patterns'][] = $sub_r; |
735
|
|
|
$proceed = 1; |
736
|
|
|
} |
737
|
|
|
} |
738
|
27 |
|
} while ($proceed); |
739
|
|
|
|
740
|
27 |
|
return 1 == \count($r['patterns']) ? [$r['patterns'][0], $sub_v] : [$r, $sub_v]; |
741
|
|
|
} |
742
|
|
|
|
743
|
|
|
return [0, $v]; |
744
|
|
|
} |
745
|
|
|
|
746
|
|
|
/* 54 */ |
747
|
|
|
|
748
|
27 |
|
protected function xUnaryExpression($v) |
749
|
|
|
{ |
750
|
27 |
|
$sub_v = $v; |
751
|
27 |
|
$op = ''; |
752
|
27 |
|
if ($sub_r = $this->x('(\!|\+|\-)', $sub_v)) { |
753
|
|
|
$op = $sub_r[1]; |
754
|
|
|
$sub_v = $sub_r[2]; |
755
|
|
|
} |
756
|
27 |
|
if ((list($sub_r, $sub_v) = $this->xPrimaryExpression($sub_v)) && $sub_r) { |
757
|
27 |
|
if (!\is_array($sub_r)) { |
758
|
|
|
$sub_r = ['type' => 'unary', 'expression' => $sub_r]; |
759
|
27 |
|
} elseif ($sub_op = $sub_r['operator'] ?? '') { |
760
|
|
|
$ops = ['!!' => '', '++' => '+', '--' => '+', '+-' => '-', '-+' => '-']; |
761
|
|
|
$op = isset($ops[$op.$sub_op]) ? $ops[$op.$sub_op] : $op.$sub_op; |
762
|
|
|
} |
763
|
27 |
|
$sub_r['operator'] = $op; |
764
|
|
|
|
765
|
27 |
|
return [$sub_r, $sub_v]; |
766
|
|
|
} |
767
|
|
|
|
768
|
|
|
return [0, $v]; |
769
|
|
|
} |
770
|
|
|
|
771
|
|
|
/* 55 */ |
772
|
|
|
|
773
|
27 |
|
protected function xPrimaryExpression($v) |
774
|
|
|
{ |
775
|
27 |
|
foreach (['BrackettedExpression', 'BuiltInCall', 'IRIrefOrFunction', 'RDFLiteral', 'NumericLiteral', 'BooleanLiteral', 'Var', 'Placeholder'] as $type) { |
776
|
27 |
|
$m = 'x'.$type; |
777
|
27 |
|
if ((list($sub_r, $sub_v) = $this->$m($v)) && $sub_r) { |
778
|
27 |
|
return [$sub_r, $sub_v]; |
779
|
|
|
} |
780
|
|
|
} |
781
|
|
|
|
782
|
|
|
return [0, $v]; |
783
|
|
|
} |
784
|
|
|
|
785
|
|
|
/* 56 */ |
786
|
|
|
|
787
|
28 |
|
protected function xBrackettedExpression($v) |
788
|
|
|
{ |
789
|
28 |
|
if ($r = $this->x('\(', $v)) { |
790
|
23 |
|
if ((list($r, $sub_v) = $this->xExpression($r[1])) && $r) { |
791
|
23 |
|
if ($sub_r = $this->x('\)', $sub_v)) { |
792
|
23 |
|
return [$r, $sub_r[1]]; |
793
|
|
|
} |
794
|
|
|
} |
795
|
|
|
} |
796
|
|
|
|
797
|
28 |
|
return [0, $v]; |
798
|
|
|
} |
799
|
|
|
|
800
|
|
|
/* 57.., 58.. */ |
801
|
|
|
|
802
|
28 |
|
protected function xBuiltInCall($v) |
803
|
|
|
{ |
804
|
28 |
|
if ($sub_r = $this->x('(str|lang|langmatches|datatype|bound|sameterm|isiri|isuri|isblank|isliteral|regex)\s*\(', $v)) { |
805
|
18 |
|
$r = ['type' => 'built_in_call', 'call' => strtolower($sub_r[1])]; |
806
|
18 |
|
if ((list($sub_r, $sub_v) = $this->xArgList('('.$sub_r[2])) && \is_array($sub_r)) { |
807
|
18 |
|
$r['args'] = $sub_r; |
808
|
|
|
|
809
|
18 |
|
return [$r, $sub_v]; |
810
|
|
|
} |
811
|
|
|
} |
812
|
|
|
|
813
|
28 |
|
return [0, $v]; |
814
|
|
|
} |
815
|
|
|
|
816
|
|
|
/* 59.. */ |
817
|
|
|
|
818
|
27 |
|
protected function xIRIrefOrFunction($v) |
819
|
|
|
{ |
820
|
27 |
|
if ((list($r, $v) = $this->xIRIref($v)) && $r) { |
821
|
1 |
|
if ((list($sub_r, $sub_v) = $this->xArgList($v)) && \is_array($sub_r)) { |
822
|
|
|
return [['type' => 'function', 'uri' => $r, 'args' => $sub_r], $sub_v]; |
823
|
|
|
} |
824
|
|
|
|
825
|
1 |
|
return [['type' => 'uri', 'uri' => $r], $sub_v]; |
826
|
|
|
} |
827
|
|
|
} |
828
|
|
|
|
829
|
|
|
/* 70.. @@sync with TurtleParser */ |
830
|
|
|
|
831
|
98 |
|
protected function xIRI_REF($v) |
832
|
|
|
{ |
833
|
98 |
|
if (($r = $this->x('\<(\$\{[^\>]*\})\>', $v)) && ($sub_r = $this->xPlaceholder($r[1]))) { |
|
|
|
|
834
|
|
|
return [$r[1], $r[2]]; |
835
|
98 |
|
} elseif ($r = $this->x('\<([^\<\>\s\"\|\^`]*)\>', $v)) { |
836
|
96 |
|
return [$r[1] ? $r[1] : true, $r[2]]; |
837
|
|
|
} |
838
|
98 |
|
/* allow reserved chars in obvious IRIs */ elseif ($r = $this->x('\<(https?\:[^\s][^\<\>]*)\>', $v)) { |
839
|
|
|
return [$r[1] ? $r[1] : true, $r[2]]; |
840
|
|
|
} |
841
|
|
|
|
842
|
98 |
|
return [0, $v]; |
843
|
|
|
} |
844
|
|
|
} |
845
|
|
|
|