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