sweetrdf /
in-memory-store-sqlite
| Total Complexity | 248 |
| Total Lines | 960 |
| Duplicated Lines | 0 % |
| Coverage | 64.57% |
| Changes | 2 | ||
| Bugs | 0 | Features | 0 |
Complex classes like TurtleParser often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use TurtleParser, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 23 | class TurtleParser extends BaseParser |
||
| 24 | { |
||
| 25 | protected int $state; |
||
| 26 | protected int $max_parsing_loops; |
||
| 27 | protected string $unparsed_code; |
||
| 28 | |||
| 29 | 100 | public function __construct(Logger $logger, NamespaceHelper $namespaceHelper, StringReader $stringReader) |
|
| 30 | { |
||
| 31 | 100 | parent::__construct($logger, $namespaceHelper, $stringReader); |
|
| 32 | |||
| 33 | 100 | $this->state = 0; |
|
| 34 | 100 | $this->unparsed_code = ''; |
|
| 35 | 100 | $this->max_parsing_loops = 500; |
|
| 36 | } |
||
| 37 | |||
| 38 | 100 | protected function x($re, $v, $options = 'si') |
|
| 48 | } |
||
| 49 | |||
| 50 | 2 | private function createBnodeID(): string |
|
| 51 | { |
||
| 52 | 2 | ++$this->bnode_id; |
|
| 53 | |||
| 54 | 2 | return '_:'.$this->bnode_prefix.$this->bnode_id; |
|
| 55 | } |
||
| 56 | |||
| 57 | 2 | protected function addT(array $t): void |
|
| 58 | { |
||
| 59 | 2 | $this->triples[$this->t_count] = $t; |
|
| 60 | 2 | ++$this->t_count; |
|
| 61 | } |
||
| 62 | |||
| 63 | protected function countTriples() |
||
| 66 | } |
||
| 67 | |||
| 68 | protected function getUnparsedCode() |
||
| 69 | { |
||
| 70 | return $this->unparsed_code; |
||
| 71 | } |
||
| 72 | |||
| 73 | 2 | public function parse(string $path, string $data = ''): void |
|
| 74 | { |
||
| 75 | 2 | $this->triples = []; |
|
| 76 | 2 | $this->t_count = 0; |
|
| 77 | 2 | $this->reader->init($path, $data); |
|
| 78 | 2 | $this->base = $this->reader->getBase(); |
|
| 79 | 2 | $this->r = ['vars' => []]; |
|
| 80 | /* parse */ |
||
| 81 | 2 | $buffer = ''; |
|
| 82 | 2 | $more_triples = []; |
|
| 83 | 2 | $sub_v = ''; |
|
| 84 | 2 | $sub_v2 = ''; |
|
| 85 | 2 | $loops = 0; |
|
| 86 | 2 | $prologue_done = 0; |
|
| 87 | 2 | while ($d = $this->reader->readStream(8192)) { |
|
| 88 | 2 | $buffer .= $d; |
|
| 89 | 2 | $sub_v = $buffer; |
|
| 90 | do { |
||
| 91 | 2 | $proceed = 0; |
|
| 92 | 2 | if (!$prologue_done) { |
|
| 93 | 2 | $proceed = 1; |
|
| 94 | 2 | if ((list($sub_r, $sub_v) = $this->xPrologue($sub_v)) && $sub_r) { |
|
| 95 | 2 | $loops = 0; |
|
| 96 | 2 | $sub_v .= $this->reader->readStream(128); |
|
| 97 | /* in case we missed the final DOT in the previous prologue loop */ |
||
| 98 | 2 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 99 | $sub_v = $sub_r[1]; |
||
| 100 | } |
||
| 101 | /* more prologue to come, use outer loop */ |
||
| 102 | 2 | if ($this->x("\@?(base|prefix)", $sub_v)) { |
|
| 103 | 2 | $proceed = 0; |
|
| 104 | } |
||
| 105 | } else { |
||
| 106 | 2 | $prologue_done = 1; |
|
| 107 | } |
||
| 108 | } |
||
| 109 | if ( |
||
| 110 | 2 | $prologue_done |
|
| 111 | 2 | && (list($sub_r, $sub_v, $more_triples, $sub_v2) = $this->xTriplesBlock($sub_v)) |
|
| 112 | 2 | && \is_array($sub_r) |
|
| 113 | ) { |
||
| 114 | 2 | $proceed = 1; |
|
| 115 | 2 | $loops = 0; |
|
| 116 | 2 | foreach ($sub_r as $t) { |
|
| 117 | 2 | $this->addT($t); |
|
| 118 | } |
||
| 119 | } |
||
| 120 | 2 | } while ($proceed); |
|
| 121 | 2 | ++$loops; |
|
| 122 | 2 | $buffer = $sub_v; |
|
| 123 | 2 | if ($loops > $this->max_parsing_loops) { |
|
| 124 | $msg = 'too many loops: '.$loops.'. Could not parse "'.substr($buffer, 0, 200).'..."'; |
||
| 125 | throw new Exception($msg); |
||
| 126 | } |
||
| 127 | } |
||
| 128 | 2 | foreach ($more_triples as $t) { |
|
| 129 | $this->addT($t); |
||
| 130 | } |
||
| 131 | 2 | $sub_v = \count($more_triples) ? $sub_v2 : $sub_v; |
|
| 132 | 2 | $buffer = $sub_v; |
|
| 133 | 2 | $this->unparsed_code = $buffer; |
|
| 134 | |||
| 135 | /* remove trailing comments */ |
||
| 136 | 2 | while (preg_match('/^\s*(\#[^\xd\xa]*)(.*)$/si', $this->unparsed_code, $m)) { |
|
| 137 | $this->unparsed_code = $m[2]; |
||
| 138 | } |
||
| 139 | |||
| 140 | 2 | if ($this->unparsed_code && !$this->logger->hasEntries('error')) { |
|
| 141 | $rest = preg_replace('/[\x0a|\x0d]/i', ' ', substr($this->unparsed_code, 0, 30)); |
||
| 142 | if (trim($rest)) { |
||
| 143 | $this->logger->error('Could not parse "'.$rest.'"'); |
||
| 144 | } |
||
| 145 | } |
||
| 146 | } |
||
| 147 | |||
| 148 | 2 | protected function xPrologue($v) |
|
| 149 | { |
||
| 150 | 2 | $r = 0; |
|
| 151 | 2 | if (!$this->t_count) { |
|
| 152 | 2 | if ((list($sub_r, $v) = $this->xBaseDecl($v)) && $sub_r) { |
|
| 153 | $this->base = $sub_r; |
||
| 154 | $r = 1; |
||
| 155 | } |
||
| 156 | 2 | while ((list($sub_r, $v) = $this->xPrefixDecl($v)) && $sub_r) { |
|
| 157 | 2 | $this->namespaceHelper->setPrefix($sub_r['prefix'], $sub_r['uri']); |
|
| 158 | 2 | $r = 1; |
|
| 159 | } |
||
| 160 | } |
||
| 161 | |||
| 162 | 2 | return [$r, $v]; |
|
| 163 | } |
||
| 164 | |||
| 165 | /* 3 */ |
||
| 166 | |||
| 167 | 100 | protected function xBaseDecl($v) |
|
| 168 | { |
||
| 169 | 100 | if ($r = $this->x("\@?base\s+", $v)) { |
|
| 170 | if ((list($r, $sub_v) = $this->xIRI_REF($r[1])) && $r) { |
||
| 171 | if ($sub_r = $this->x('\.', $sub_v)) { |
||
| 172 | $sub_v = $sub_r[1]; |
||
| 173 | } |
||
| 174 | |||
| 175 | return [$r, $sub_v]; |
||
| 176 | } |
||
| 177 | } |
||
| 178 | |||
| 179 | 100 | return [0, $v]; |
|
| 180 | } |
||
| 181 | |||
| 182 | /* 4 */ |
||
| 183 | |||
| 184 | 100 | protected function xPrefixDecl($v) |
|
| 185 | { |
||
| 186 | 100 | if ($r = $this->x("\@?prefix\s+", $v)) { |
|
| 187 | 22 | if ((list($r, $sub_v) = $this->xPNAME_NS($r[1])) && $r) { |
|
| 188 | 22 | $prefix = $r; |
|
| 189 | 22 | if ((list($r, $sub_v) = $this->xIRI_REF($sub_v)) && $r) { |
|
| 190 | 22 | $uri = calcURI($r, $this->base); |
|
| 191 | 22 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 192 | 16 | $sub_v = $sub_r[1]; |
|
| 193 | } |
||
| 194 | |||
| 195 | 22 | return [['prefix' => $prefix, 'uri_ref' => $r, 'uri' => $uri], $sub_v]; |
|
| 196 | } |
||
| 197 | } |
||
| 198 | } |
||
| 199 | |||
| 200 | 100 | return [0, $v]; |
|
| 201 | } |
||
| 202 | |||
| 203 | /* 21.., 32.. */ |
||
| 204 | |||
| 205 | 100 | protected function xTriplesBlock($v) |
|
| 206 | { |
||
| 207 | 100 | $pre_r = []; |
|
| 208 | 100 | $r = []; |
|
| 209 | 100 | $state = 1; |
|
| 210 | 100 | $sub_v = $v; |
|
| 211 | 100 | $buffer = $sub_v; |
|
| 212 | do { |
||
| 213 | 100 | $proceed = 0; |
|
| 214 | 100 | if (1 == $state) {/* expecting subject */ |
|
| 215 | 100 | $t = ['type' => 'triple', 's' => '', 'p' => '', 'o' => '', 's_type' => '', 'p_type' => '', 'o_type' => '', 'o_datatype' => '', 'o_lang' => '']; |
|
| 216 | 100 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 217 | 100 | $t['s'] = $sub_r['value']; |
|
| 218 | 100 | $t['s_type'] = $sub_r['type']; |
|
| 219 | 100 | $state = 2; |
|
| 220 | 100 | $proceed = 1; |
|
| 221 | 100 | if ($sub_r = $this->x('(\}|\.)', $sub_v)) { |
|
| 222 | if ('placeholder' == $t['s_type']) { |
||
| 223 | $state = 4; |
||
| 224 | } else { |
||
| 225 | 100 | $this->logger->error('"'.$sub_r[1].'" after subject found.'); |
|
| 226 | } |
||
| 227 | } |
||
| 228 | 98 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
|
| 229 | $t['s'] = $sub_r['id']; |
||
| 230 | $t['s_type'] = $sub_r['type']; |
||
| 231 | $pre_r = array_merge($pre_r, $sub_r['triples']); |
||
| 232 | $state = 2; |
||
| 233 | $proceed = 1; |
||
| 234 | if ($sub_r = $this->x('\.', $sub_v)) { |
||
|
Unused Code
introduced
by
|
|||
| 235 | $this->logger->error('DOT after subject found.'); |
||
| 236 | } |
||
| 237 | 98 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
|
| 238 | $t['s'] = $sub_r['id']; |
||
| 239 | $t['s_type'] = $sub_r['type']; |
||
| 240 | $pre_r = array_merge($pre_r, $sub_r['triples']); |
||
| 241 | $state = 2; |
||
| 242 | $proceed = 1; |
||
| 243 | 98 | } elseif ($sub_r = $this->x('\.', $sub_v)) { |
|
| 244 | $this->logger->error('Subject expected, DOT found.'.$sub_v); |
||
| 245 | } |
||
| 246 | } |
||
| 247 | 100 | if (2 == $state) {/* expecting predicate */ |
|
| 248 | 100 | if ($sub_r = $this->x('a\s+', $sub_v)) { |
|
| 249 | 1 | $sub_v = $sub_r[1]; |
|
| 250 | 1 | $t['p'] = NamespaceHelper::NAMESPACE_RDF.'type'; |
|
| 251 | 1 | $t['p_type'] = 'uri'; |
|
| 252 | 1 | $state = 3; |
|
| 253 | 1 | $proceed = 1; |
|
| 254 | 100 | } elseif ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 255 | 100 | if ('bnode' == $sub_r['type']) { |
|
| 256 | $this->logger->error('Blank node used as triple predicate'); |
||
| 257 | } |
||
| 258 | 100 | $t['p'] = $sub_r['value']; |
|
| 259 | 100 | $t['p_type'] = $sub_r['type']; |
|
| 260 | 100 | $state = 3; |
|
| 261 | 100 | $proceed = 1; |
|
| 262 | } elseif ($sub_r = $this->x('\.', $sub_v)) { |
||
| 263 | $state = 4; |
||
| 264 | } elseif ($sub_r = $this->x('\}', $sub_v)) { |
||
| 265 | $buffer = $sub_v; |
||
| 266 | $r = array_merge($r, $pre_r); |
||
| 267 | $pre_r = []; |
||
| 268 | $proceed = 0; |
||
| 269 | } |
||
| 270 | } |
||
| 271 | 100 | if (3 == $state) {/* expecting object */ |
|
| 272 | 100 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 273 | 100 | $t['o'] = $sub_r['value']; |
|
| 274 | 100 | $t['o_type'] = $sub_r['type']; |
|
| 275 | 100 | $t['o_lang'] = $sub_r['lang'] ?? ''; |
|
| 276 | 100 | $t['o_datatype'] = $sub_r['datatype'] ?? ''; |
|
| 277 | 100 | $pre_r[] = $t; |
|
| 278 | 100 | $state = 4; |
|
| 279 | 100 | $proceed = 1; |
|
| 280 | 2 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
|
| 281 | $t['o'] = $sub_r['id']; |
||
| 282 | $t['o_type'] = $sub_r['type']; |
||
| 283 | $t['o_datatype'] = ''; |
||
| 284 | $pre_r = array_merge($pre_r, [$t], $sub_r['triples']); |
||
| 285 | $state = 4; |
||
| 286 | $proceed = 1; |
||
| 287 | 2 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
|
| 288 | 2 | $t['o'] = $sub_r['id']; |
|
| 289 | 2 | $t['o_type'] = $sub_r['type']; |
|
| 290 | 2 | $t['o_datatype'] = ''; |
|
| 291 | 2 | $pre_r = array_merge($pre_r, [$t], $sub_r['triples']); |
|
| 292 | 2 | $state = 4; |
|
| 293 | 2 | $proceed = 1; |
|
| 294 | } |
||
| 295 | } |
||
| 296 | 100 | if (4 == $state) {/* expecting . or ; or , or } */ |
|
| 297 | 100 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 298 | 98 | $sub_v = $sub_r[1]; |
|
| 299 | 98 | $buffer = $sub_v; |
|
| 300 | 98 | $r = array_merge($r, $pre_r); |
|
| 301 | 98 | $pre_r = []; |
|
| 302 | 98 | $state = 1; |
|
| 303 | 98 | $proceed = 1; |
|
| 304 | 29 | } elseif ($sub_r = $this->x('\;', $sub_v)) { |
|
| 305 | 11 | $sub_v = $sub_r[1]; |
|
| 306 | 11 | $state = 2; |
|
| 307 | 11 | $proceed = 1; |
|
| 308 | 25 | } elseif ($sub_r = $this->x('\,', $sub_v)) { |
|
| 309 | 20 | $sub_v = $sub_r[1]; |
|
| 310 | 20 | $state = 3; |
|
| 311 | 20 | $proceed = 1; |
|
| 312 | 20 | if ($sub_r = $this->x('\}', $sub_v)) { |
|
| 313 | $this->logger->error('Object expected, } found.'); |
||
| 314 | } |
||
| 315 | } |
||
| 316 | 100 | if ($sub_r = $this->x('(\}|\{|OPTIONAL|FILTER|GRAPH)', $sub_v)) { |
|
| 317 | 97 | $buffer = $sub_v; |
|
| 318 | 97 | $r = array_merge($r, $pre_r); |
|
| 319 | 97 | $pre_r = []; |
|
| 320 | 97 | $proceed = 0; |
|
| 321 | } |
||
| 322 | } |
||
| 323 | 100 | } while ($proceed); |
|
| 324 | |||
| 325 | 100 | return \count($r) ? [$r, $buffer, $pre_r, $sub_v] : [0, $buffer, $pre_r, $sub_v]; |
|
| 326 | } |
||
| 327 | |||
| 328 | /* 39.. */ |
||
| 329 | |||
| 330 | 98 | protected function xBlankNodePropertyList($v) |
|
| 331 | { |
||
| 332 | 98 | if ($sub_r = $this->x('\[', $v)) { |
|
| 333 | 2 | $sub_v = $sub_r[1]; |
|
| 334 | 2 | $s = $this->createBnodeID(); |
|
| 335 | 2 | $r = ['id' => $s, 'type' => 'bnode', 'triples' => []]; |
|
| 336 | 2 | $t = ['type' => 'triple', 's' => $s, 'p' => '', 'o' => '', 's_type' => 'bnode', 'p_type' => '', 'o_type' => '', 'o_datatype' => '', 'o_lang' => '']; |
|
| 337 | 2 | $state = 2; |
|
| 338 | 2 | $closed = 0; |
|
| 339 | do { |
||
| 340 | 2 | $proceed = 0; |
|
| 341 | 2 | if (2 == $state) {/* expecting predicate */ |
|
| 342 | 2 | if ($sub_r = $this->x('a\s+', $sub_v)) { |
|
| 343 | $sub_v = $sub_r[1]; |
||
| 344 | $t['p'] = NamespaceHelper::NAMESPACE_RDF.'type'; |
||
| 345 | $t['p_type'] = 'uri'; |
||
| 346 | $state = 3; |
||
| 347 | $proceed = 1; |
||
| 348 | 2 | } elseif ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 349 | 2 | $t['p'] = $sub_r['value']; |
|
| 350 | 2 | $t['p_type'] = $sub_r['type']; |
|
| 351 | 2 | $state = 3; |
|
| 352 | 2 | $proceed = 1; |
|
| 353 | } |
||
| 354 | } |
||
| 355 | 2 | if (3 == $state) {/* expecting object */ |
|
| 356 | 2 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 357 | 2 | $t['o'] = $sub_r['value']; |
|
| 358 | 2 | $t['o_type'] = $sub_r['type']; |
|
| 359 | 2 | $t['o_lang'] = $sub_r['lang'] ?? ''; |
|
| 360 | 2 | $t['o_datatype'] = $sub_r['datatype'] ?? ''; |
|
| 361 | 2 | $r['triples'][] = $t; |
|
| 362 | 2 | $state = 4; |
|
| 363 | 2 | $proceed = 1; |
|
| 364 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
||
| 365 | $t['o'] = $sub_r['id']; |
||
| 366 | $t['o_type'] = $sub_r['type']; |
||
| 367 | $t['o_datatype'] = ''; |
||
| 368 | $r['triples'] = array_merge($r['triples'], [$t], $sub_r['triples']); |
||
| 369 | $state = 4; |
||
| 370 | $proceed = 1; |
||
| 371 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
||
| 372 | $t['o'] = $sub_r['id']; |
||
| 373 | $t['o_type'] = $sub_r['type']; |
||
| 374 | $t['o_datatype'] = ''; |
||
| 375 | $r['triples'] = array_merge($r['triples'], [$t], $sub_r['triples']); |
||
| 376 | $state = 4; |
||
| 377 | $proceed = 1; |
||
| 378 | } |
||
| 379 | } |
||
| 380 | 2 | if (4 == $state) {/* expecting . or ; or , or ] */ |
|
| 381 | 2 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 382 | $sub_v = $sub_r[1]; |
||
| 383 | $state = 1; |
||
| 384 | $proceed = 1; |
||
| 385 | } |
||
| 386 | 2 | if ($sub_r = $this->x('\;', $sub_v)) { |
|
| 387 | 1 | $sub_v = $sub_r[1]; |
|
| 388 | 1 | $state = 2; |
|
| 389 | 1 | $proceed = 1; |
|
| 390 | } |
||
| 391 | 2 | if ($sub_r = $this->x('\,', $sub_v)) { |
|
| 392 | $sub_v = $sub_r[1]; |
||
| 393 | $state = 3; |
||
| 394 | $proceed = 1; |
||
| 395 | } |
||
| 396 | 2 | if ($sub_r = $this->x('\]', $sub_v)) { |
|
| 397 | 2 | $sub_v = $sub_r[1]; |
|
| 398 | 2 | $proceed = 0; |
|
| 399 | 2 | $closed = 1; |
|
| 400 | } |
||
| 401 | } |
||
| 402 | 2 | } while ($proceed); |
|
| 403 | 2 | if ($closed) { |
|
| 404 | 2 | return [$r, $sub_v]; |
|
| 405 | } |
||
| 406 | |||
| 407 | return [0, $v]; |
||
| 408 | } |
||
| 409 | |||
| 410 | 98 | return [0, $v]; |
|
| 411 | } |
||
| 412 | |||
| 413 | /* 40.. */ |
||
| 414 | |||
| 415 | 98 | protected function xCollection($v) |
|
| 416 | { |
||
| 417 | 98 | if ($sub_r = $this->x('\(', $v)) { |
|
| 418 | $sub_v = $sub_r[1]; |
||
| 419 | $s = $this->createBnodeID(); |
||
| 420 | $r = ['id' => $s, 'type' => 'bnode', 'triples' => []]; |
||
| 421 | $closed = 0; |
||
| 422 | do { |
||
| 423 | $proceed = 0; |
||
| 424 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
||
| 425 | $r['triples'][] = [ |
||
| 426 | 'type' => 'triple', |
||
| 427 | 's' => $s, |
||
| 428 | 's_type' => 'bnode', |
||
| 429 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
||
| 430 | 'p_type' => 'uri', |
||
| 431 | 'o' => $sub_r['value'], |
||
| 432 | 'o_type' => $sub_r['type'], |
||
| 433 | 'o_lang' => $sub_r['lang'] ?? '', |
||
| 434 | 'o_datatype' => $sub_r['datatype'] ?? '', |
||
| 435 | ]; |
||
| 436 | $proceed = 1; |
||
| 437 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
||
| 438 | $r['triples'][] = [ |
||
| 439 | 'type' => 'triple', |
||
| 440 | 's' => $s, |
||
| 441 | 's_type' => 'bnode', |
||
| 442 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
||
| 443 | 'p_type' => 'uri', |
||
| 444 | 'o' => $sub_r['id'], |
||
| 445 | 'o_type' => $sub_r['type'], |
||
| 446 | 'o_lang' => '', |
||
| 447 | 'o_datatype' => '', |
||
| 448 | ]; |
||
| 449 | $r['triples'] = array_merge($r['triples'], $sub_r['triples']); |
||
| 450 | $proceed = 1; |
||
| 451 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
||
| 452 | $r['triples'][] = [ |
||
| 453 | 'type' => 'triple', |
||
| 454 | 's' => $s, |
||
| 455 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
||
| 456 | 'o' => $sub_r['id'], |
||
| 457 | 's_type' => 'bnode', |
||
| 458 | 'p_type' => 'uri', |
||
| 459 | 'o_type' => $sub_r['type'], |
||
| 460 | 'o_lang' => '', |
||
| 461 | 'o_datatype' => '', |
||
| 462 | ]; |
||
| 463 | $r['triples'] = array_merge($r['triples'], $sub_r['triples']); |
||
| 464 | $proceed = 1; |
||
| 465 | } |
||
| 466 | if ($proceed) { |
||
| 467 | if ($sub_r = $this->x('\)', $sub_v)) { |
||
| 468 | $sub_v = $sub_r[1]; |
||
| 469 | $r['triples'][] = [ |
||
| 470 | 'type' => 'triple', |
||
| 471 | 's' => $s, |
||
| 472 | 's_type' => 'bnode', |
||
| 473 | 'p' => NamespaceHelper::NAMESPACE_RDF.'rest', |
||
| 474 | 'p_type' => 'uri', |
||
| 475 | 'o' => NamespaceHelper::NAMESPACE_RDF.'nil', |
||
| 476 | 'o_type' => 'uri', |
||
| 477 | 'o_lang' => '', |
||
| 478 | 'o_datatype' => '', |
||
| 479 | ]; |
||
| 480 | $closed = 1; |
||
| 481 | $proceed = 0; |
||
| 482 | } else { |
||
| 483 | $next_s = $this->createBnodeID(); |
||
| 484 | $r['triples'][] = [ |
||
| 485 | 'type' => 'triple', |
||
| 486 | 's' => $s, |
||
| 487 | 'p' => NamespaceHelper::NAMESPACE_RDF.'rest', |
||
| 488 | 'o' => $next_s, |
||
| 489 | 's_type' => 'bnode', |
||
| 490 | 'p_type' => 'uri', |
||
| 491 | 'o_type' => 'bnode', |
||
| 492 | 'o_lang' => '', |
||
| 493 | 'o_datatype' => '', |
||
| 494 | ]; |
||
| 495 | $s = $next_s; |
||
| 496 | } |
||
| 497 | } |
||
| 498 | } while ($proceed); |
||
| 499 | if ($closed) { |
||
| 500 | return [$r, $sub_v]; |
||
| 501 | } |
||
| 502 | } |
||
| 503 | |||
| 504 | 98 | return [0, $v]; |
|
| 505 | } |
||
| 506 | |||
| 507 | /* 42 */ |
||
| 508 | |||
| 509 | 100 | protected function xVarOrTerm($v) |
|
| 510 | { |
||
| 511 | 100 | if ((list($sub_r, $sub_v) = $this->xVar($v)) && $sub_r) { |
|
| 512 | 94 | return [$sub_r, $sub_v]; |
|
| 513 | 100 | } elseif ((list($sub_r, $sub_v) = $this->xGraphTerm($v)) && $sub_r) { |
|
| 514 | 96 | return [$sub_r, $sub_v]; |
|
| 515 | } |
||
| 516 | |||
| 517 | 98 | return [0, $v]; |
|
| 518 | } |
||
| 519 | |||
| 520 | /* 44, 74.., 75.. */ |
||
| 521 | |||
| 522 | 100 | protected function xVar($v) |
|
| 523 | { |
||
| 524 | 100 | if ($r = $this->x('(\?|\$)([^\s]+)', $v)) { |
|
| 525 | 94 | if ((list($sub_r, $sub_v) = $this->xVARNAME($r[2])) && $sub_r) { |
|
| 526 | 94 | if (!\in_array($sub_r, $this->r['vars'])) { |
|
| 527 | 94 | $this->r['vars'][] = $sub_r; |
|
| 528 | } |
||
| 529 | |||
| 530 | 94 | return [['value' => $sub_r, 'type' => 'var'], $sub_v.$r[3]]; |
|
| 531 | } |
||
| 532 | } |
||
| 533 | |||
| 534 | 100 | return [0, $v]; |
|
| 535 | } |
||
| 536 | |||
| 537 | /* 45 */ |
||
| 538 | |||
| 539 | 100 | protected function xGraphTerm($v) |
|
| 540 | { |
||
| 541 | 100 | foreach ([ |
|
| 542 | 100 | 'IRIref' => 'uri', |
|
| 543 | 100 | 'RDFLiteral' => 'literal', |
|
| 544 | 100 | 'NumericLiteral' => 'literal', |
|
| 545 | 100 | 'BooleanLiteral' => 'literal', |
|
| 546 | 100 | 'BlankNode' => 'bnode', |
|
| 547 | 100 | 'NIL' => 'uri', |
|
| 548 | 100 | 'Placeholder' => 'placeholder', |
|
| 549 | 100 | ] as $term => $type) { |
|
| 550 | 100 | $m = 'x'.$term; |
|
| 551 | 100 | if ((list($sub_r, $sub_v) = $this->$m($v)) && $sub_r) { |
|
| 552 | 96 | if (!\is_array($sub_r)) { |
|
| 553 | 96 | $sub_r = ['value' => $sub_r]; |
|
| 554 | } |
||
| 555 | 96 | $sub_r['type'] = $sub_r['type'] ?? $type; |
|
| 556 | |||
| 557 | 96 | return [$sub_r, $sub_v]; |
|
| 558 | } |
||
| 559 | } |
||
| 560 | |||
| 561 | 98 | return [0, $v]; |
|
| 562 | } |
||
| 563 | |||
| 564 | /* 60 */ |
||
| 565 | |||
| 566 | 99 | protected function xRDFLiteral($v) |
|
| 586 | } |
||
| 587 | |||
| 588 | /* 61.., 62.., 63.., 64.. */ |
||
| 589 | |||
| 590 | 98 | protected function xNumericLiteral($v) |
|
| 591 | { |
||
| 592 | 98 | $sub_r = $this->x('(\-|\+)?', $v); |
|
| 593 | 98 | $prefix = $sub_r[1]; |
|
| 594 | 98 | $sub_v = $sub_r[2]; |
|
| 595 | 98 | foreach (['DOUBLE' => 'double', 'DECIMAL' => 'decimal', 'INTEGER' => 'integer'] as $type => $xsd) { |
|
| 596 | 98 | $m = 'x'.$type; |
|
| 597 | 98 | if ((list($sub_r, $sub_v) = $this->$m($sub_v)) && (false !== $sub_r)) { |
|
| 598 | 20 | $r = [ |
|
| 599 | 20 | 'value' => $prefix.$sub_r, |
|
| 600 | 20 | 'type' => 'literal', |
|
| 601 | 20 | 'datatype' => NamespaceHelper::NAMESPACE_XSD.$xsd, |
|
| 602 | 20 | ]; |
|
| 603 | |||
| 604 | 20 | return [$r, $sub_v]; |
|
| 605 | } |
||
| 606 | } |
||
| 607 | |||
| 608 | 98 | return [0, $v]; |
|
| 609 | } |
||
| 610 | |||
| 611 | /* 65.. */ |
||
| 612 | |||
| 613 | 98 | protected function xBooleanLiteral($v) |
|
| 614 | { |
||
| 615 | 98 | if ($r = $this->x('(true|false)', $v)) { |
|
| 616 | return [$r[1], $r[2]]; |
||
| 617 | } |
||
| 618 | |||
| 619 | 98 | return [0, $v]; |
|
| 620 | } |
||
| 621 | |||
| 622 | /* 66.., 87.., 88.., 89.., 90.., 91.. */ |
||
| 623 | |||
| 624 | 99 | protected function xString($v) |
|
| 657 | } |
||
| 658 | |||
| 659 | /* 67 */ |
||
| 660 | |||
| 661 | 100 | protected function xIRIref($v) |
|
| 662 | { |
||
| 663 | 100 | if ((list($r, $v) = $this->xIRI_REF($v)) && $r) { |
|
| 664 | 95 | return [calcURI($r, $this->base), $v]; |
|
| 665 | 100 | } elseif ((list($r, $v) = $this->xPrefixedName($v)) && $r) { |
|
| 666 | 30 | return [$r, $v]; |
|
| 667 | } |
||
| 668 | |||
| 669 | 99 | return [0, $v]; |
|
| 670 | } |
||
| 671 | |||
| 672 | /* 68 */ |
||
| 673 | |||
| 674 | 100 | protected function xPrefixedName($v) |
|
| 675 | { |
||
| 676 | 100 | if ((list($r, $v) = $this->xPNAME_LN($v)) && $r) { |
|
| 677 | 30 | return [$r, $v]; |
|
| 678 | 99 | } elseif ((list($r, $sub_v) = $this->xPNAME_NS($v)) && $r) { |
|
| 679 | 8 | return $this->namespaceHelper->hasPrefix($r) |
|
| 680 | ? [$this->namespaceHelper->getNamespace($r), $sub_v] |
||
| 681 | 8 | : [0, $v]; |
|
| 682 | } |
||
| 683 | |||
| 684 | 99 | return [0, $v]; |
|
| 685 | } |
||
| 686 | |||
| 687 | /* 69.., 73.., 93, 94.. */ |
||
| 688 | |||
| 689 | 98 | protected function xBlankNode($v) |
|
| 690 | { |
||
| 691 | 98 | if (($r = $this->x('\_\:', $v)) && (list($r, $sub_v) = $this->xPN_LOCAL($r[1])) && $r) { |
|
| 692 | 5 | return [['type' => 'bnode', 'value' => '_:'.$r], $sub_v]; |
|
| 693 | } |
||
| 694 | 98 | if ($r = $this->x('\[[\x20\x9\xd\xa]*\]', $v)) { |
|
| 695 | return [['type' => 'bnode', 'value' => $this->createBnodeID()], $r[1]]; |
||
| 696 | } |
||
| 697 | |||
| 698 | 98 | return [0, $v]; |
|
| 699 | } |
||
| 700 | |||
| 701 | /* 70.. @@sync with SPARQLParser */ |
||
| 702 | |||
| 703 | 2 | protected function xIRI_REF($v) |
|
| 704 | { |
||
| 705 | //if ($r = $this->x('\<([^\<\>\"\{\}\|\^\'[:space:]]*)\>', $v)) { |
||
| 706 | 2 | if (($r = $this->x('\<(\$\{[^\>]*\})\>', $v)) && ($sub_r = $this->xPlaceholder($r[1]))) { |
|
| 707 | return [$r[1], $r[2]]; |
||
| 708 | 2 | } elseif ($r = $this->x('\<\>', $v)) { |
|
| 709 | return [true, $r[1]]; |
||
| 710 | 2 | } elseif ($r = $this->x('\<([^\s][^\<\>]*)\>', $v)) { |
|
| 711 | 2 | return [$r[1] ? $r[1] : true, $r[2]]; |
|
| 712 | } |
||
| 713 | |||
| 714 | 2 | return [0, $v]; |
|
| 715 | } |
||
| 716 | |||
| 717 | /* 71 */ |
||
| 718 | |||
| 719 | 100 | protected function xPNAME_NS($v) |
|
| 725 | } |
||
| 726 | |||
| 727 | /* 72 */ |
||
| 728 | |||
| 729 | 100 | protected function xPNAME_LN($v) |
|
| 730 | { |
||
| 731 | 100 | if ((list($r, $sub_v) = $this->xPNAME_NS($v)) && $r) { |
|
| 732 | 33 | if (!$this->x('\s', $sub_v) && (list($sub_r, $sub_v) = $this->xPN_LOCAL($sub_v)) && $sub_r) { |
|
| 733 | 33 | if (!$this->namespaceHelper->hasPrefix($r)) { |
|
| 734 | 8 | return [0, $v]; |
|
| 735 | } |
||
| 736 | |||
| 737 | 30 | return [$this->namespaceHelper->getNamespace($r).$sub_r, $sub_v]; |
|
| 738 | } |
||
| 739 | } |
||
| 740 | |||
| 741 | 99 | return [0, $v]; |
|
| 742 | } |
||
| 743 | |||
| 744 | /* 76 */ |
||
| 745 | |||
| 746 | 53 | protected function xLANGTAG($v) |
|
| 753 | } |
||
| 754 | |||
| 755 | /* 77.. */ |
||
| 756 | |||
| 757 | 98 | protected function xINTEGER($v) |
|
| 758 | { |
||
| 759 | 98 | if ($r = $this->x('([0-9]+)', $v)) { |
|
| 760 | 21 | return [$r[1], $r[2]]; |
|
| 761 | } |
||
| 762 | |||
| 763 | 98 | return [false, $v]; |
|
| 764 | } |
||
| 765 | |||
| 766 | /* 78.. */ |
||
| 767 | |||
| 768 | 98 | protected function xDECIMAL($v) |
|
| 769 | { |
||
| 770 | 98 | if ($r = $this->x('([0-9]+\.[0-9]*)', $v)) { |
|
| 771 | 9 | return [$r[1], $r[2]]; |
|
| 772 | } |
||
| 773 | 98 | if ($r = $this->x('(\.[0-9]+)', $v)) { |
|
| 774 | return [$r[1], $r[2]]; |
||
| 775 | } |
||
| 776 | |||
| 777 | 98 | return [false, $v]; |
|
| 778 | } |
||
| 779 | |||
| 780 | /* 79.., 86.. */ |
||
| 781 | |||
| 782 | 98 | protected function xDOUBLE($v) |
|
| 795 | } |
||
| 796 | |||
| 797 | /* 92 */ |
||
| 798 | |||
| 799 | 98 | protected function xNIL($v) |
|
| 800 | { |
||
| 801 | 98 | if ($r = $this->x('\([\x20\x9\xd\xa]*\)', $v)) { |
|
| 802 | return [['type' => 'uri', 'value' => NamespaceHelper::NAMESPACE_RDF.'nil'], $r[1]]; |
||
| 803 | } |
||
| 804 | |||
| 805 | 98 | return [0, $v]; |
|
| 806 | } |
||
| 807 | |||
| 808 | /* 95.. */ |
||
| 809 | |||
| 810 | 99 | protected function xPN_CHARS_BASE($v) |
|
| 817 | } |
||
| 818 | |||
| 819 | /* 96 */ |
||
| 820 | |||
| 821 | 95 | protected function xPN_CHARS_U($v) |
|
| 830 | } |
||
| 831 | |||
| 832 | /* 97.. */ |
||
| 833 | |||
| 834 | 94 | protected function xVARNAME($v) |
|
| 835 | { |
||
| 836 | 94 | $r = ''; |
|
| 837 | do { |
||
| 838 | 94 | $proceed = 0; |
|
| 839 | 94 | if ($sub_r = $this->x('([0-9]+)', $v)) { |
|
| 840 | 3 | $r .= $sub_r[1]; |
|
| 841 | 3 | $v = $sub_r[2]; |
|
| 842 | 3 | $proceed = 1; |
|
| 843 | 94 | } elseif ((list($sub_r, $sub_v) = $this->xPN_CHARS_U($v)) && $sub_r) { |
|
| 844 | 94 | $r .= $sub_r; |
|
| 845 | 94 | $v = $sub_v; |
|
| 846 | 94 | $proceed = 1; |
|
| 847 | 94 | } elseif ($r && ($sub_r = $this->x('([\xb7\x300-\x36f]+)', $v))) { |
|
| 848 | $r .= $sub_r[1]; |
||
| 849 | $v = $sub_r[2]; |
||
| 850 | $proceed = 1; |
||
| 851 | } |
||
| 852 | 94 | } while ($proceed); |
|
| 853 | |||
| 854 | 94 | return [$r, $v]; |
|
| 855 | } |
||
| 856 | |||
| 857 | /* 98.. */ |
||
| 858 | |||
| 859 | 6 | protected function xPN_CHARS($v) |
|
| 860 | { |
||
| 861 | 6 | if ((list($r, $sub_v) = $this->xPN_CHARS_U($v)) && $r) { |
|
| 862 | return [$r, $sub_v]; |
||
| 863 | 6 | } elseif ($r = $this->x('([\-0-9\xb7\x300-\x36f])', $v)) { |
|
| 864 | return [$r[1], $r[2]]; |
||
| 865 | } |
||
| 866 | |||
| 867 | 6 | return [false, $v]; |
|
| 868 | } |
||
| 869 | |||
| 870 | /* 99 */ |
||
| 871 | |||
| 872 | 100 | protected function xPN_PREFIX($v) |
|
| 873 | { |
||
| 874 | 100 | if ($sub_r = $this->x("([^\s\:\(\)\{\}\;\,]+)", $v, 's')) {/* accelerator */ |
|
| 875 | 78 | return [$sub_r[1], $sub_r[2]]; /* @@testing */ |
|
| 876 | } |
||
| 877 | 99 | if ((list($r, $sub_v) = $this->xPN_CHARS_BASE($v)) && $r) { |
|
| 878 | do { |
||
| 879 | $proceed = 0; |
||
| 880 | list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v); |
||
| 881 | if (false !== $sub_r) { |
||
| 882 | $r .= $sub_r; |
||
| 883 | $proceed = 1; |
||
| 884 | } elseif ($sub_r = $this->x("\.", $sub_v)) { |
||
| 885 | $r .= '.'; |
||
| 886 | $sub_v = $sub_r[1]; |
||
| 887 | $proceed = 1; |
||
| 888 | } |
||
| 889 | } while ($proceed); |
||
| 890 | list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v); |
||
| 891 | $r .= $sub_r ?: ''; |
||
| 892 | } |
||
| 893 | |||
| 894 | 99 | return [$r, $sub_v]; |
|
| 895 | } |
||
| 896 | |||
| 897 | /* 100 */ |
||
| 898 | |||
| 899 | 33 | protected function xPN_LOCAL($v) |
|
| 900 | { |
||
| 901 | 33 | if (($sub_r = $this->x("([^\s\(\)\{\}\[\]\;\,\.]+)", $v, 's')) && !preg_match('/^\./', $sub_r[2])) {/* accelerator */ |
|
| 902 | 33 | return [$sub_r[1], $sub_r[2]]; /* @@testing */ |
|
| 903 | } |
||
| 904 | 6 | $r = ''; |
|
| 905 | 6 | $sub_v = $v; |
|
| 906 | do { |
||
| 907 | 6 | $proceed = 0; |
|
| 908 | 6 | if ($this->x('\s', $sub_v)) { |
|
| 909 | return [$r, $sub_v]; |
||
| 910 | } |
||
| 911 | 6 | if ($sub_r = $this->x('([0-9])', $sub_v)) { |
|
| 912 | 6 | $r .= $sub_r[1]; |
|
| 913 | 6 | $sub_v = $sub_r[2]; |
|
| 914 | 6 | $proceed = 1; |
|
| 915 | 6 | } elseif ((list($sub_r, $sub_v) = $this->xPN_CHARS_U($sub_v)) && $sub_r) { |
|
| 916 | 6 | $r .= $sub_r; |
|
| 917 | 6 | $proceed = 1; |
|
| 918 | 6 | } elseif ($r) { |
|
| 919 | 6 | if (($sub_r = $this->x('(\.)', $sub_v)) && !preg_match('/^[\s\}]/s', $sub_r[2])) { |
|
| 920 | $r .= $sub_r[1]; |
||
| 921 | $sub_v = $sub_r[2]; |
||
| 922 | } |
||
| 923 | 6 | if ((list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v)) && $sub_r) { |
|
| 924 | $r .= $sub_r; |
||
| 925 | $proceed = 1; |
||
| 926 | } |
||
| 927 | } |
||
| 928 | 6 | } while ($proceed); |
|
| 929 | |||
| 930 | 6 | return [$r, $sub_v]; |
|
| 931 | } |
||
| 932 | |||
| 933 | 53 | protected function unescapeNtripleUTF($v) |
|
| 962 | } |
||
| 963 | |||
| 964 | 98 | protected function xPlaceholder($v) |
|
| 965 | { |
||
| 966 | //if ($r = $this->x('(\?|\$)\{([^\}]+)\}', $v)) { |
||
| 967 | 98 | if ($r = $this->x('(\?|\$)', $v)) { |
|
| 968 | if (preg_match('/(\{(?:[^{}]+|(?R))*\})/', $r[2], $m) && 0 === strpos(trim($r[2]), $m[1])) { |
||
| 969 | $ph = substr($m[1], 1, -1); |
||
| 970 | $rest = substr(trim($r[2]), \strlen($m[1])); |
||
| 971 | if (!isset($this->r['placeholders'])) { |
||
| 972 | $this->r['placeholders'] = []; |
||
| 973 | } |
||
| 974 | if (!\in_array($ph, $this->r['placeholders'])) { |
||
| 975 | $this->r['placeholders'][] = $ph; |
||
| 976 | } |
||
| 977 | |||
| 978 | return [['value' => $ph, 'type' => 'placeholder'], $rest]; |
||
| 979 | } |
||
| 983 | } |
||
| 984 | } |
||
| 985 |
Loading history...