sweetrdf /
in-memory-store-sqlite
| Total Complexity | 248 |
| Total Lines | 953 |
| Duplicated Lines | 0 % |
| Coverage | 75.63% |
| Changes | 1 | ||
| 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 |
||
| 22 | class TurtleParser extends BaseParser |
||
| 23 | { |
||
| 24 | 88 | public function __construct(Logger $logger) |
|
| 25 | { |
||
| 26 | 88 | parent::__construct($logger); |
|
| 27 | |||
| 28 | 88 | $this->state = 0; |
|
|
Bug
Best Practice
introduced
by
|
|||
| 29 | 88 | $this->unparsed_code = ''; |
|
| 30 | 88 | $this->max_parsing_loops = 500; |
|
| 31 | 88 | } |
|
| 32 | |||
| 33 | 88 | protected function x($re, $v, $options = 'si') |
|
| 34 | { |
||
| 35 | 88 | $v = preg_replace('/^[\xA0\xC2]+/', ' ', $v); |
|
| 36 | |||
| 37 | /* comment removal */ |
||
| 38 | 88 | while (preg_match('/^\s*(\#[^\xd\xa]*)(.*)$/si', $v, $m)) { |
|
| 39 | $v = $m[2]; |
||
| 40 | } |
||
| 41 | |||
| 42 | 88 | return preg_match("/^\s*".$re.'(.*)$/'.$options, $v, $m) ? $m : false; |
|
| 43 | } |
||
| 44 | |||
| 45 | 10 | private function createBnodeID(): string |
|
| 46 | { |
||
| 47 | 10 | ++$this->bnode_id; |
|
| 48 | |||
| 49 | 10 | return '_:'.$this->bnode_prefix.$this->bnode_id; |
|
| 50 | } |
||
| 51 | |||
| 52 | 8 | protected function addT(array $t): void |
|
| 53 | { |
||
| 54 | 8 | $this->triples[$this->t_count] = $t; |
|
| 55 | 8 | ++$this->t_count; |
|
| 56 | 8 | } |
|
| 57 | |||
| 58 | protected function countTriples() |
||
| 59 | { |
||
| 60 | return $this->t_count; |
||
| 61 | } |
||
| 62 | |||
| 63 | protected function getUnparsedCode() |
||
| 64 | { |
||
| 65 | return $this->unparsed_code; |
||
| 66 | } |
||
| 67 | |||
| 68 | 9 | public function parse(string $path, string $data = ''): void |
|
| 69 | { |
||
| 70 | 9 | $this->reader = new StringReader(); |
|
| 71 | 9 | $this->reader->init($path, $data); |
|
| 72 | 9 | $this->base = $this->reader->getBase(); |
|
| 73 | 9 | $this->r = ['vars' => []]; |
|
| 74 | /* parse */ |
||
| 75 | 9 | $buffer = ''; |
|
| 76 | 9 | $more_triples = []; |
|
| 77 | 9 | $sub_v = ''; |
|
| 78 | 9 | $sub_v2 = ''; |
|
| 79 | 9 | $loops = 0; |
|
| 80 | 9 | $prologue_done = 0; |
|
| 81 | 9 | while ($d = $this->reader->readStream(8192)) { |
|
| 82 | 9 | $buffer .= $d; |
|
| 83 | 9 | $sub_v = $buffer; |
|
| 84 | do { |
||
| 85 | 9 | $proceed = 0; |
|
| 86 | 9 | if (!$prologue_done) { |
|
| 87 | 9 | $proceed = 1; |
|
| 88 | 9 | if ((list($sub_r, $sub_v) = $this->xPrologue($sub_v)) && $sub_r) { |
|
| 89 | 8 | $loops = 0; |
|
| 90 | 8 | $sub_v .= $this->reader->readStream(128); |
|
| 91 | /* in case we missed the final DOT in the previous prologue loop */ |
||
| 92 | 8 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 93 | $sub_v = $sub_r[1]; |
||
| 94 | } |
||
| 95 | /* more prologue to come, use outer loop */ |
||
| 96 | 8 | if ($this->x("\@?(base|prefix)", $sub_v)) { |
|
| 97 | 8 | $proceed = 0; |
|
| 98 | } |
||
| 99 | } else { |
||
| 100 | 9 | $prologue_done = 1; |
|
| 101 | } |
||
| 102 | } |
||
| 103 | if ( |
||
| 104 | 9 | $prologue_done |
|
| 105 | 9 | && (list($sub_r, $sub_v, $more_triples, $sub_v2) = $this->xTriplesBlock($sub_v)) |
|
| 106 | 9 | && \is_array($sub_r) |
|
| 107 | ) { |
||
| 108 | 9 | $proceed = 1; |
|
| 109 | 9 | $loops = 0; |
|
| 110 | 9 | foreach ($sub_r as $t) { |
|
| 111 | 9 | $this->addT($t); |
|
| 112 | } |
||
| 113 | } |
||
| 114 | 9 | } while ($proceed); |
|
| 115 | 9 | ++$loops; |
|
| 116 | 9 | $buffer = $sub_v; |
|
| 117 | 9 | if ($loops > $this->max_parsing_loops) { |
|
| 118 | $msg = 'too many loops: '.$loops.'. Could not parse "'.substr($buffer, 0, 200).'..."'; |
||
| 119 | throw new Exception($msg); |
||
| 120 | } |
||
| 121 | } |
||
| 122 | 9 | foreach ($more_triples as $t) { |
|
| 123 | $this->addT($t); |
||
| 124 | } |
||
| 125 | 9 | $sub_v = \count($more_triples) ? $sub_v2 : $sub_v; |
|
| 126 | 9 | $buffer = $sub_v; |
|
| 127 | 9 | $this->unparsed_code = $buffer; |
|
| 128 | |||
| 129 | /* remove trailing comments */ |
||
| 130 | 9 | while (preg_match('/^\s*(\#[^\xd\xa]*)(.*)$/si', $this->unparsed_code, $m)) { |
|
| 131 | $this->unparsed_code = $m[2]; |
||
| 132 | } |
||
| 133 | |||
| 134 | 9 | if ($this->unparsed_code && !$this->logger->hasEntries('error')) { |
|
| 135 | 9 | $rest = preg_replace('/[\x0a|\x0d]/i', ' ', substr($this->unparsed_code, 0, 30)); |
|
| 136 | 9 | if (trim($rest)) { |
|
| 137 | $this->logger->error('Could not parse "'.$rest.'"'); |
||
| 138 | } |
||
| 139 | } |
||
| 140 | 9 | } |
|
| 141 | |||
| 142 | 9 | protected function xPrologue($v) |
|
| 143 | { |
||
| 144 | 9 | $r = 0; |
|
| 145 | 9 | if (!$this->t_count) { |
|
| 146 | 9 | if ((list($sub_r, $v) = $this->xBaseDecl($v)) && $sub_r) { |
|
| 147 | $this->base = $sub_r; |
||
| 148 | $r = 1; |
||
| 149 | } |
||
| 150 | 9 | while ((list($sub_r, $v) = $this->xPrefixDecl($v)) && $sub_r) { |
|
| 151 | 8 | $this->prefixes[$sub_r['prefix']] = $sub_r['uri']; |
|
| 152 | 8 | $r = 1; |
|
| 153 | } |
||
| 154 | } |
||
| 155 | |||
| 156 | 9 | return [$r, $v]; |
|
| 157 | } |
||
| 158 | |||
| 159 | /* 3 */ |
||
| 160 | |||
| 161 | 88 | protected function xBaseDecl($v) |
|
| 162 | { |
||
| 163 | 88 | if ($r = $this->x("\@?base\s+", $v)) { |
|
| 164 | if ((list($r, $sub_v) = $this->xIRI_REF($r[1])) && $r) { |
||
| 165 | if ($sub_r = $this->x('\.', $sub_v)) { |
||
| 166 | $sub_v = $sub_r[1]; |
||
| 167 | } |
||
| 168 | |||
| 169 | return [$r, $sub_v]; |
||
| 170 | } |
||
| 171 | } |
||
| 172 | |||
| 173 | 88 | return [0, $v]; |
|
| 174 | } |
||
| 175 | |||
| 176 | /* 4 */ |
||
| 177 | |||
| 178 | 88 | protected function xPrefixDecl($v) |
|
| 179 | { |
||
| 180 | 88 | if ($r = $this->x("\@?prefix\s+", $v)) { |
|
| 181 | 12 | if ((list($r, $sub_v) = $this->xPNAME_NS($r[1])) && $r) { |
|
| 182 | 12 | $prefix = $r; |
|
| 183 | 12 | if ((list($r, $sub_v) = $this->xIRI_REF($sub_v)) && $r) { |
|
| 184 | 12 | $uri = calcURI($r, $this->base); |
|
| 185 | 12 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 186 | 8 | $sub_v = $sub_r[1]; |
|
| 187 | } |
||
| 188 | |||
| 189 | 12 | return [['prefix' => $prefix, 'uri_ref' => $r, 'uri' => $uri], $sub_v]; |
|
| 190 | } |
||
| 191 | } |
||
| 192 | } |
||
| 193 | |||
| 194 | 88 | return [0, $v]; |
|
| 195 | } |
||
| 196 | |||
| 197 | /* 21.., 32.. */ |
||
| 198 | |||
| 199 | 88 | protected function xTriplesBlock($v) |
|
| 200 | { |
||
| 201 | 88 | $pre_r = []; |
|
| 202 | 88 | $r = []; |
|
| 203 | 88 | $state = 1; |
|
| 204 | 88 | $sub_v = $v; |
|
| 205 | 88 | $buffer = $sub_v; |
|
| 206 | do { |
||
| 207 | 88 | $proceed = 0; |
|
| 208 | 88 | if (1 == $state) {/* expecting subject */ |
|
| 209 | 88 | $t = ['type' => 'triple', 's' => '', 'p' => '', 'o' => '', 's_type' => '', 'p_type' => '', 'o_type' => '', 'o_datatype' => '', 'o_lang' => '']; |
|
| 210 | 88 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 211 | 88 | $t['s'] = $sub_r['value']; |
|
| 212 | 88 | $t['s_type'] = $sub_r['type']; |
|
| 213 | 88 | $state = 2; |
|
| 214 | 88 | $proceed = 1; |
|
| 215 | 88 | if ($sub_r = $this->x('(\}|\.)', $sub_v)) { |
|
| 216 | if ('placeholder' == $t['s_type']) { |
||
| 217 | $state = 4; |
||
| 218 | } else { |
||
| 219 | 88 | $this->logger->error('"'.$sub_r[1].'" after subject found.'); |
|
| 220 | } |
||
| 221 | } |
||
| 222 | 87 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
|
| 223 | $t['s'] = $sub_r['id']; |
||
| 224 | $t['s_type'] = $sub_r['type']; |
||
| 225 | $pre_r = array_merge($pre_r, $sub_r['triples']); |
||
| 226 | $state = 2; |
||
| 227 | $proceed = 1; |
||
| 228 | if ($sub_r = $this->x('\.', $sub_v)) { |
||
| 229 | $this->logger->error('DOT after subject found.'); |
||
| 230 | } |
||
| 231 | 87 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
|
| 232 | $t['s'] = $sub_r['id']; |
||
| 233 | $t['s_type'] = $sub_r['type']; |
||
| 234 | $pre_r = array_merge($pre_r, $sub_r['triples']); |
||
| 235 | $state = 2; |
||
| 236 | $proceed = 1; |
||
| 237 | 87 | } elseif ($sub_r = $this->x('\.', $sub_v)) { |
|
| 238 | $this->logger->error('Subject expected, DOT found.'.$sub_v); |
||
| 239 | } |
||
| 240 | } |
||
| 241 | 88 | if (2 == $state) {/* expecting predicate */ |
|
| 242 | 88 | if ($sub_r = $this->x('a\s+', $sub_v)) { |
|
| 243 | $sub_v = $sub_r[1]; |
||
| 244 | $t['p'] = NamespaceHelper::NAMESPACE_RDF.'type'; |
||
| 245 | $t['p_type'] = 'uri'; |
||
| 246 | $state = 3; |
||
| 247 | $proceed = 1; |
||
| 248 | 88 | } elseif ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 249 | 88 | if ('bnode' == $sub_r['type']) { |
|
| 250 | $this->logger->error('Blank node used as triple predicate'); |
||
| 251 | } |
||
| 252 | 88 | $t['p'] = $sub_r['value']; |
|
| 253 | 88 | $t['p_type'] = $sub_r['type']; |
|
| 254 | 88 | $state = 3; |
|
| 255 | 88 | $proceed = 1; |
|
| 256 | 1 | } elseif ($sub_r = $this->x('\.', $sub_v)) { |
|
| 257 | $state = 4; |
||
| 258 | 1 | } elseif ($sub_r = $this->x('\}', $sub_v)) { |
|
| 259 | $buffer = $sub_v; |
||
| 260 | $r = array_merge($r, $pre_r); |
||
| 261 | $pre_r = []; |
||
| 262 | $proceed = 0; |
||
| 263 | } |
||
| 264 | } |
||
| 265 | 88 | if (3 == $state) {/* expecting object */ |
|
| 266 | 88 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 267 | 88 | $t['o'] = $sub_r['value']; |
|
| 268 | 88 | $t['o_type'] = $sub_r['type']; |
|
| 269 | 88 | $t['o_lang'] = $sub_r['lang'] ?? ''; |
|
| 270 | 88 | $t['o_datatype'] = $sub_r['datatype'] ?? ''; |
|
| 271 | 88 | $pre_r[] = $t; |
|
| 272 | 88 | $state = 4; |
|
| 273 | 88 | $proceed = 1; |
|
| 274 | 11 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
|
| 275 | 8 | $t['o'] = $sub_r['id']; |
|
| 276 | 8 | $t['o_type'] = $sub_r['type']; |
|
| 277 | 8 | $t['o_datatype'] = ''; |
|
| 278 | 8 | $pre_r = array_merge($pre_r, [$t], $sub_r['triples']); |
|
| 279 | 8 | $state = 4; |
|
| 280 | 8 | $proceed = 1; |
|
| 281 | 11 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
|
| 282 | 10 | $t['o'] = $sub_r['id']; |
|
| 283 | 10 | $t['o_type'] = $sub_r['type']; |
|
| 284 | 10 | $t['o_datatype'] = ''; |
|
| 285 | 10 | $pre_r = array_merge($pre_r, [$t], $sub_r['triples']); |
|
| 286 | 10 | $state = 4; |
|
| 287 | 10 | $proceed = 1; |
|
| 288 | } |
||
| 289 | } |
||
| 290 | 88 | if (4 == $state) {/* expecting . or ; or , or } */ |
|
| 291 | 88 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 292 | 88 | $sub_v = $sub_r[1]; |
|
| 293 | 88 | $buffer = $sub_v; |
|
| 294 | 88 | $r = array_merge($r, $pre_r); |
|
| 295 | 88 | $pre_r = []; |
|
| 296 | 88 | $state = 1; |
|
| 297 | 88 | $proceed = 1; |
|
| 298 | 17 | } elseif ($sub_r = $this->x('\;', $sub_v)) { |
|
| 299 | 11 | $sub_v = $sub_r[1]; |
|
| 300 | 11 | $state = 2; |
|
| 301 | 11 | $proceed = 1; |
|
| 302 | 14 | } elseif ($sub_r = $this->x('\,', $sub_v)) { |
|
| 303 | 12 | $sub_v = $sub_r[1]; |
|
| 304 | 12 | $state = 3; |
|
| 305 | 12 | $proceed = 1; |
|
| 306 | 12 | if ($sub_r = $this->x('\}', $sub_v)) { |
|
| 307 | $this->logger->error('Object expected, } found.'); |
||
| 308 | } |
||
| 309 | } |
||
| 310 | 88 | if ($sub_r = $this->x('(\}|\{|OPTIONAL|FILTER|GRAPH)', $sub_v)) { |
|
| 311 | 87 | $buffer = $sub_v; |
|
| 312 | 87 | $r = array_merge($r, $pre_r); |
|
| 313 | 87 | $pre_r = []; |
|
| 314 | 87 | $proceed = 0; |
|
| 315 | } |
||
| 316 | } |
||
| 317 | 88 | } while ($proceed); |
|
| 318 | |||
| 319 | 88 | return \count($r) ? [$r, $buffer, $pre_r, $sub_v] : [0, $buffer, $pre_r, $sub_v]; |
|
| 320 | } |
||
| 321 | |||
| 322 | /* 39.. */ |
||
| 323 | |||
| 324 | 87 | protected function xBlankNodePropertyList($v) |
|
| 325 | { |
||
| 326 | 87 | if ($sub_r = $this->x('\[', $v)) { |
|
| 327 | 10 | $sub_v = $sub_r[1]; |
|
| 328 | 10 | $s = $this->createBnodeID(); |
|
| 329 | 10 | $r = ['id' => $s, 'type' => 'bnode', 'triples' => []]; |
|
| 330 | 10 | $t = ['type' => 'triple', 's' => $s, 'p' => '', 'o' => '', 's_type' => 'bnode', 'p_type' => '', 'o_type' => '', 'o_datatype' => '', 'o_lang' => '']; |
|
| 331 | 10 | $state = 2; |
|
| 332 | 10 | $closed = 0; |
|
| 333 | do { |
||
| 334 | 10 | $proceed = 0; |
|
| 335 | 10 | if (2 == $state) {/* expecting predicate */ |
|
| 336 | 10 | if ($sub_r = $this->x('a\s+', $sub_v)) { |
|
| 337 | $sub_v = $sub_r[1]; |
||
| 338 | $t['p'] = NamespaceHelper::NAMESPACE_RDF.'type'; |
||
| 339 | $t['p_type'] = 'uri'; |
||
| 340 | $state = 3; |
||
| 341 | $proceed = 1; |
||
| 342 | 10 | } elseif ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 343 | 10 | $t['p'] = $sub_r['value']; |
|
| 344 | 10 | $t['p_type'] = $sub_r['type']; |
|
| 345 | 10 | $state = 3; |
|
| 346 | 10 | $proceed = 1; |
|
| 347 | } |
||
| 348 | } |
||
| 349 | 10 | if (3 == $state) {/* expecting object */ |
|
| 350 | 10 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 351 | 10 | $t['o'] = $sub_r['value']; |
|
| 352 | 10 | $t['o_type'] = $sub_r['type']; |
|
| 353 | 10 | $t['o_lang'] = $sub_r['lang'] ?? ''; |
|
| 354 | 10 | $t['o_datatype'] = $sub_r['datatype'] ?? ''; |
|
| 355 | 10 | $r['triples'][] = $t; |
|
| 356 | 10 | $state = 4; |
|
| 357 | 10 | $proceed = 1; |
|
| 358 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
||
| 359 | $t['o'] = $sub_r['id']; |
||
| 360 | $t['o_type'] = $sub_r['type']; |
||
| 361 | $t['o_datatype'] = ''; |
||
| 362 | $r['triples'] = array_merge($r['triples'], [$t], $sub_r['triples']); |
||
| 363 | $state = 4; |
||
| 364 | $proceed = 1; |
||
| 365 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
||
| 366 | $t['o'] = $sub_r['id']; |
||
| 367 | $t['o_type'] = $sub_r['type']; |
||
| 368 | $t['o_datatype'] = ''; |
||
| 369 | $r['triples'] = array_merge($r['triples'], [$t], $sub_r['triples']); |
||
| 370 | $state = 4; |
||
| 371 | $proceed = 1; |
||
| 372 | } |
||
| 373 | } |
||
| 374 | 10 | if (4 == $state) {/* expecting . or ; or , or ] */ |
|
| 375 | 10 | if ($sub_r = $this->x('\.', $sub_v)) { |
|
| 376 | $sub_v = $sub_r[1]; |
||
| 377 | $state = 1; |
||
| 378 | $proceed = 1; |
||
| 379 | } |
||
| 380 | 10 | if ($sub_r = $this->x('\;', $sub_v)) { |
|
| 381 | 9 | $sub_v = $sub_r[1]; |
|
| 382 | 9 | $state = 2; |
|
| 383 | 9 | $proceed = 1; |
|
| 384 | } |
||
| 385 | 10 | if ($sub_r = $this->x('\,', $sub_v)) { |
|
| 386 | $sub_v = $sub_r[1]; |
||
| 387 | $state = 3; |
||
| 388 | $proceed = 1; |
||
| 389 | } |
||
| 390 | 10 | if ($sub_r = $this->x('\]', $sub_v)) { |
|
| 391 | 10 | $sub_v = $sub_r[1]; |
|
| 392 | 10 | $proceed = 0; |
|
| 393 | 10 | $closed = 1; |
|
| 394 | } |
||
| 395 | } |
||
| 396 | 10 | } while ($proceed); |
|
| 397 | 10 | if ($closed) { |
|
| 398 | 10 | return [$r, $sub_v]; |
|
| 399 | } |
||
| 400 | |||
| 401 | 8 | return [0, $v]; |
|
| 402 | } |
||
| 403 | |||
| 404 | 87 | return [0, $v]; |
|
| 405 | } |
||
| 406 | |||
| 407 | /* 40.. */ |
||
| 408 | |||
| 409 | 87 | protected function xCollection($v) |
|
| 410 | { |
||
| 411 | 87 | if ($sub_r = $this->x('\(', $v)) { |
|
| 412 | 8 | $sub_v = $sub_r[1]; |
|
| 413 | 8 | $s = $this->createBnodeID(); |
|
| 414 | 8 | $r = ['id' => $s, 'type' => 'bnode', 'triples' => []]; |
|
| 415 | 8 | $closed = 0; |
|
| 416 | do { |
||
| 417 | 8 | $proceed = 0; |
|
| 418 | 8 | if ((list($sub_r, $sub_v) = $this->xVarOrTerm($sub_v)) && $sub_r) { |
|
| 419 | 8 | $r['triples'][] = [ |
|
| 420 | 8 | 'type' => 'triple', |
|
| 421 | 8 | 's' => $s, |
|
| 422 | 8 | 's_type' => 'bnode', |
|
| 423 | 8 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
|
| 424 | 8 | 'p_type' => 'uri', |
|
| 425 | 8 | 'o' => $sub_r['value'], |
|
| 426 | 8 | 'o_type' => $sub_r['type'], |
|
| 427 | 8 | 'o_lang' => $sub_r['lang'] ?? '', |
|
| 428 | 8 | 'o_datatype' => $sub_r['datatype'] ?? '', |
|
| 429 | ]; |
||
| 430 | 8 | $proceed = 1; |
|
| 431 | } elseif ((list($sub_r, $sub_v) = $this->xCollection($sub_v)) && $sub_r) { |
||
| 432 | $r['triples'][] = [ |
||
| 433 | 'type' => 'triple', |
||
| 434 | 's' => $s, |
||
| 435 | 's_type' => 'bnode', |
||
| 436 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
||
| 437 | 'p_type' => 'uri', |
||
| 438 | 'o' => $sub_r['id'], |
||
| 439 | 'o_type' => $sub_r['type'], |
||
| 440 | 'o_lang' => '', |
||
| 441 | 'o_datatype' => '', |
||
| 442 | ]; |
||
| 443 | $r['triples'] = array_merge($r['triples'], $sub_r['triples']); |
||
| 444 | $proceed = 1; |
||
| 445 | } elseif ((list($sub_r, $sub_v) = $this->xBlankNodePropertyList($sub_v)) && $sub_r) { |
||
| 446 | $r['triples'][] = [ |
||
| 447 | 'type' => 'triple', |
||
| 448 | 's' => $s, |
||
| 449 | 'p' => NamespaceHelper::NAMESPACE_RDF.'first', |
||
| 450 | 'o' => $sub_r['id'], |
||
| 451 | 's_type' => 'bnode', |
||
| 452 | 'p_type' => 'uri', |
||
| 453 | 'o_type' => $sub_r['type'], |
||
| 454 | 'o_lang' => '', |
||
| 455 | 'o_datatype' => '', |
||
| 456 | ]; |
||
| 457 | $r['triples'] = array_merge($r['triples'], $sub_r['triples']); |
||
| 458 | $proceed = 1; |
||
| 459 | } |
||
| 460 | 8 | if ($proceed) { |
|
| 461 | 8 | if ($sub_r = $this->x('\)', $sub_v)) { |
|
| 462 | 8 | $sub_v = $sub_r[1]; |
|
| 463 | 8 | $r['triples'][] = [ |
|
| 464 | 8 | 'type' => 'triple', |
|
| 465 | 8 | 's' => $s, |
|
| 466 | 8 | 's_type' => 'bnode', |
|
| 467 | 8 | 'p' => NamespaceHelper::NAMESPACE_RDF.'rest', |
|
| 468 | 8 | 'p_type' => 'uri', |
|
| 469 | 8 | 'o' => NamespaceHelper::NAMESPACE_RDF.'nil', |
|
| 470 | 8 | 'o_type' => 'uri', |
|
| 471 | 8 | 'o_lang' => '', |
|
| 472 | 8 | 'o_datatype' => '', |
|
| 473 | ]; |
||
| 474 | 8 | $closed = 1; |
|
| 475 | 8 | $proceed = 0; |
|
| 476 | } else { |
||
| 477 | 8 | $next_s = $this->createBnodeID(); |
|
| 478 | 8 | $r['triples'][] = [ |
|
| 479 | 8 | 'type' => 'triple', |
|
| 480 | 8 | 's' => $s, |
|
| 481 | 8 | 'p' => NamespaceHelper::NAMESPACE_RDF.'rest', |
|
| 482 | 8 | 'o' => $next_s, |
|
| 483 | 8 | 's_type' => 'bnode', |
|
| 484 | 8 | 'p_type' => 'uri', |
|
| 485 | 8 | 'o_type' => 'bnode', |
|
| 486 | 8 | 'o_lang' => '', |
|
| 487 | 8 | 'o_datatype' => '', |
|
| 488 | ]; |
||
| 489 | 8 | $s = $next_s; |
|
| 490 | } |
||
| 491 | } |
||
| 492 | 8 | } while ($proceed); |
|
| 493 | 8 | if ($closed) { |
|
| 494 | 8 | return [$r, $sub_v]; |
|
| 495 | } |
||
| 496 | } |
||
| 497 | |||
| 498 | 87 | return [0, $v]; |
|
| 499 | } |
||
| 500 | |||
| 501 | /* 42 */ |
||
| 502 | |||
| 503 | 88 | protected function xVarOrTerm($v) |
|
| 504 | { |
||
| 505 | 88 | if ((list($sub_r, $sub_v) = $this->xVar($v)) && $sub_r) { |
|
| 506 | 85 | return [$sub_r, $sub_v]; |
|
| 507 | 88 | } elseif ((list($sub_r, $sub_v) = $this->xGraphTerm($v)) && $sub_r) { |
|
| 508 | 87 | return [$sub_r, $sub_v]; |
|
| 509 | } |
||
| 510 | |||
| 511 | 87 | return [0, $v]; |
|
| 512 | } |
||
| 513 | |||
| 514 | /* 44, 74.., 75.. */ |
||
| 515 | |||
| 516 | 88 | protected function xVar($v) |
|
| 517 | { |
||
| 518 | 88 | if ($r = $this->x('(\?|\$)([^\s]+)', $v)) { |
|
| 519 | 85 | if ((list($sub_r, $sub_v) = $this->xVARNAME($r[2])) && $sub_r) { |
|
| 520 | 85 | if (!\in_array($sub_r, $this->r['vars'])) { |
|
| 521 | 85 | $this->r['vars'][] = $sub_r; |
|
| 522 | } |
||
| 523 | |||
| 524 | 85 | return [['value' => $sub_r, 'type' => 'var'], $sub_v.$r[3]]; |
|
| 525 | } |
||
| 526 | } |
||
| 527 | |||
| 528 | 88 | return [0, $v]; |
|
| 529 | } |
||
| 530 | |||
| 531 | /* 45 */ |
||
| 532 | |||
| 533 | 88 | protected function xGraphTerm($v) |
|
| 534 | { |
||
| 535 | foreach ([ |
||
| 536 | 88 | 'IRIref' => 'uri', |
|
| 537 | 'RDFLiteral' => 'literal', |
||
| 538 | 'NumericLiteral' => 'literal', |
||
| 539 | 'BooleanLiteral' => 'literal', |
||
| 540 | 'BlankNode' => 'bnode', |
||
| 541 | 'NIL' => 'uri', |
||
| 542 | 'Placeholder' => 'placeholder', |
||
| 543 | 88 | ] as $term => $type) { |
|
| 544 | 88 | $m = 'x'.$term; |
|
| 545 | 88 | if ((list($sub_r, $sub_v) = $this->$m($v)) && $sub_r) { |
|
| 546 | 87 | if (!\is_array($sub_r)) { |
|
| 547 | 87 | $sub_r = ['value' => $sub_r]; |
|
| 548 | } |
||
| 549 | 87 | $sub_r['type'] = $sub_r['type'] ?? $type; |
|
| 550 | |||
| 551 | 87 | return [$sub_r, $sub_v]; |
|
| 552 | } |
||
| 553 | } |
||
| 554 | |||
| 555 | 87 | return [0, $v]; |
|
| 556 | } |
||
| 557 | |||
| 558 | /* 60 */ |
||
| 559 | |||
| 560 | 88 | protected function xRDFLiteral($v) |
|
| 580 | } |
||
| 581 | |||
| 582 | /* 61.., 62.., 63.., 64.. */ |
||
| 583 | |||
| 584 | 87 | protected function xNumericLiteral($v) |
|
| 585 | { |
||
| 586 | 87 | $sub_r = $this->x('(\-|\+)?', $v); |
|
| 587 | 87 | $prefix = $sub_r[1]; |
|
| 588 | 87 | $sub_v = $sub_r[2]; |
|
| 589 | 87 | foreach (['DOUBLE' => 'double', 'DECIMAL' => 'decimal', 'INTEGER' => 'integer'] as $type => $xsd) { |
|
| 590 | 87 | $m = 'x'.$type; |
|
| 591 | 87 | if ((list($sub_r, $sub_v) = $this->$m($sub_v)) && (false !== $sub_r)) { |
|
| 592 | 12 | $r = [ |
|
| 593 | 12 | 'value' => $prefix.$sub_r, |
|
| 594 | 12 | 'type' => 'literal', |
|
| 595 | 12 | 'datatype' => NamespaceHelper::NAMESPACE_XSD.$xsd, |
|
| 596 | ]; |
||
| 597 | |||
| 598 | 12 | return [$r, $sub_v]; |
|
| 599 | } |
||
| 600 | } |
||
| 601 | |||
| 602 | 87 | return [0, $v]; |
|
| 603 | } |
||
| 604 | |||
| 605 | /* 65.. */ |
||
| 606 | |||
| 607 | 87 | protected function xBooleanLiteral($v) |
|
| 608 | { |
||
| 609 | 87 | if ($r = $this->x('(true|false)', $v)) { |
|
| 610 | return [$r[1], $r[2]]; |
||
| 611 | } |
||
| 612 | |||
| 613 | 87 | return [0, $v]; |
|
| 614 | } |
||
| 615 | |||
| 616 | /* 66.., 87.., 88.., 89.., 90.., 91.. */ |
||
| 617 | |||
| 618 | 88 | protected function xString($v) |
|
| 651 | } |
||
| 652 | |||
| 653 | /* 67 */ |
||
| 654 | |||
| 655 | 88 | protected function xIRIref($v) |
|
| 656 | { |
||
| 657 | 88 | if ((list($r, $v) = $this->xIRI_REF($v)) && $r) { |
|
| 658 | 87 | return [calcURI($r, $this->base), $v]; |
|
| 659 | 88 | } elseif ((list($r, $v) = $this->xPrefixedName($v)) && $r) { |
|
| 660 | 19 | return [$r, $v]; |
|
| 661 | } |
||
| 662 | |||
| 663 | 88 | return [0, $v]; |
|
| 664 | } |
||
| 665 | |||
| 666 | /* 68 */ |
||
| 667 | |||
| 668 | 88 | protected function xPrefixedName($v) |
|
| 669 | { |
||
| 670 | 88 | if ((list($r, $v) = $this->xPNAME_LN($v)) && $r) { |
|
| 671 | 19 | return [$r, $v]; |
|
| 672 | 88 | } elseif ((list($r, $sub_v) = $this->xPNAME_NS($v)) && $r) { |
|
| 673 | 16 | return isset($this->prefixes[$r]) ? [$this->prefixes[$r], $sub_v] : [0, $v]; |
|
| 674 | } |
||
| 675 | |||
| 676 | 88 | return [0, $v]; |
|
| 677 | } |
||
| 678 | |||
| 679 | /* 69.., 73.., 93, 94.. */ |
||
| 680 | |||
| 681 | 87 | protected function xBlankNode($v) |
|
| 682 | { |
||
| 683 | 87 | if (($r = $this->x('\_\:', $v)) && (list($r, $sub_v) = $this->xPN_LOCAL($r[1])) && $r) { |
|
| 684 | 12 | return [['type' => 'bnode', 'value' => '_:'.$r], $sub_v]; |
|
| 685 | } |
||
| 686 | 87 | if ($r = $this->x('\[[\x20\x9\xd\xa]*\]', $v)) { |
|
| 687 | return [['type' => 'bnode', 'value' => $this->createBnodeID()], $r[1]]; |
||
| 688 | } |
||
| 689 | |||
| 690 | 87 | return [0, $v]; |
|
| 691 | } |
||
| 692 | |||
| 693 | /* 70.. @@sync with SPARQLParser */ |
||
| 694 | |||
| 695 | 9 | protected function xIRI_REF($v) |
|
| 696 | { |
||
| 697 | //if ($r = $this->x('\<([^\<\>\"\{\}\|\^\'[:space:]]*)\>', $v)) { |
||
| 698 | 9 | if (($r = $this->x('\<(\$\{[^\>]*\})\>', $v)) && ($sub_r = $this->xPlaceholder($r[1]))) { |
|
| 699 | return [$r[1], $r[2]]; |
||
| 700 | 9 | } elseif ($r = $this->x('\<\>', $v)) { |
|
| 701 | 8 | return [true, $r[1]]; |
|
| 702 | 9 | } elseif ($r = $this->x('\<([^\s][^\<\>]*)\>', $v)) { |
|
| 703 | 9 | return [$r[1] ? $r[1] : true, $r[2]]; |
|
| 704 | } |
||
| 705 | |||
| 706 | 9 | return [0, $v]; |
|
| 707 | } |
||
| 708 | |||
| 709 | /* 71 */ |
||
| 710 | |||
| 711 | 88 | protected function xPNAME_NS($v) |
|
| 717 | } |
||
| 718 | |||
| 719 | /* 72 */ |
||
| 720 | |||
| 721 | 88 | protected function xPNAME_LN($v) |
|
| 722 | { |
||
| 723 | 88 | if ((list($r, $sub_v) = $this->xPNAME_NS($v)) && $r) { |
|
| 724 | 23 | if (!$this->x('\s', $sub_v) && (list($sub_r, $sub_v) = $this->xPN_LOCAL($sub_v)) && $sub_r) { |
|
| 725 | 23 | if (!isset($this->prefixes[$r])) { |
|
| 726 | 16 | return [0, $v]; |
|
| 727 | } |
||
| 728 | |||
| 729 | 19 | return [$this->prefixes[$r].$sub_r, $sub_v]; |
|
| 730 | } |
||
| 731 | } |
||
| 732 | |||
| 733 | 88 | return [0, $v]; |
|
| 734 | } |
||
| 735 | |||
| 736 | /* 76 */ |
||
| 737 | |||
| 738 | 62 | protected function xLANGTAG($v) |
|
| 745 | } |
||
| 746 | |||
| 747 | /* 77.. */ |
||
| 748 | |||
| 749 | 87 | protected function xINTEGER($v) |
|
| 750 | { |
||
| 751 | 87 | if ($r = $this->x('([0-9]+)', $v)) { |
|
| 752 | 15 | return [$r[1], $r[2]]; |
|
| 753 | } |
||
| 754 | |||
| 755 | 87 | return [false, $v]; |
|
| 756 | } |
||
| 757 | |||
| 758 | /* 78.. */ |
||
| 759 | |||
| 760 | 87 | protected function xDECIMAL($v) |
|
| 761 | { |
||
| 762 | 87 | if ($r = $this->x('([0-9]+\.[0-9]*)', $v)) { |
|
| 763 | 2 | return [$r[1], $r[2]]; |
|
| 764 | } |
||
| 765 | 87 | if ($r = $this->x('(\.[0-9]+)', $v)) { |
|
| 766 | return [$r[1], $r[2]]; |
||
| 767 | } |
||
| 768 | |||
| 769 | 87 | return [false, $v]; |
|
| 770 | } |
||
| 771 | |||
| 772 | /* 79.., 86.. */ |
||
| 773 | |||
| 774 | 87 | protected function xDOUBLE($v) |
|
| 787 | } |
||
| 788 | |||
| 789 | /* 92 */ |
||
| 790 | |||
| 791 | 87 | protected function xNIL($v) |
|
| 792 | { |
||
| 793 | 87 | if ($r = $this->x('\([\x20\x9\xd\xa]*\)', $v)) { |
|
| 794 | return [['type' => 'uri', 'value' => NamespaceHelper::NAMESPACE_RDF.'nil'], $r[1]]; |
||
| 795 | } |
||
| 796 | |||
| 797 | 87 | return [0, $v]; |
|
| 798 | } |
||
| 799 | |||
| 800 | /* 95.. */ |
||
| 801 | |||
| 802 | 87 | protected function xPN_CHARS_BASE($v) |
|
| 809 | } |
||
| 810 | |||
| 811 | /* 96 */ |
||
| 812 | |||
| 813 | 85 | protected function xPN_CHARS_U($v) |
|
| 822 | } |
||
| 823 | |||
| 824 | /* 97.. */ |
||
| 825 | |||
| 826 | 85 | protected function xVARNAME($v) |
|
| 827 | { |
||
| 828 | 85 | $r = ''; |
|
| 829 | do { |
||
| 830 | 85 | $proceed = 0; |
|
| 831 | 85 | if ($sub_r = $this->x('([0-9]+)', $v)) { |
|
| 832 | 2 | $r .= $sub_r[1]; |
|
| 833 | 2 | $v = $sub_r[2]; |
|
| 834 | 2 | $proceed = 1; |
|
| 835 | 85 | } elseif ((list($sub_r, $sub_v) = $this->xPN_CHARS_U($v)) && $sub_r) { |
|
| 836 | 85 | $r .= $sub_r; |
|
| 837 | 85 | $v = $sub_v; |
|
| 838 | 85 | $proceed = 1; |
|
| 839 | 85 | } elseif ($r && ($sub_r = $this->x('([\xb7\x300-\x36f]+)', $v))) { |
|
| 840 | $r .= $sub_r[1]; |
||
| 841 | $v = $sub_r[2]; |
||
| 842 | $proceed = 1; |
||
| 843 | } |
||
| 844 | 85 | } while ($proceed); |
|
| 845 | |||
| 846 | 85 | return [$r, $v]; |
|
| 847 | } |
||
| 848 | |||
| 849 | /* 98.. */ |
||
| 850 | |||
| 851 | 4 | protected function xPN_CHARS($v) |
|
| 852 | { |
||
| 853 | 4 | if ((list($r, $sub_v) = $this->xPN_CHARS_U($v)) && $r) { |
|
| 854 | return [$r, $sub_v]; |
||
| 855 | 4 | } elseif ($r = $this->x('([\-0-9\xb7\x300-\x36f])', $v)) { |
|
| 856 | return [$r[1], $r[2]]; |
||
| 857 | } |
||
| 858 | |||
| 859 | 4 | return [false, $v]; |
|
| 860 | } |
||
| 861 | |||
| 862 | /* 99 */ |
||
| 863 | |||
| 864 | 88 | protected function xPN_PREFIX($v) |
|
| 865 | { |
||
| 866 | 88 | if ($sub_r = $this->x("([^\s\:\(\)\{\}\;\,]+)", $v, 's')) {/* accelerator */ |
|
| 867 | 77 | return [$sub_r[1], $sub_r[2]]; /* @@testing */ |
|
| 868 | } |
||
| 869 | 87 | if ((list($r, $sub_v) = $this->xPN_CHARS_BASE($v)) && $r) { |
|
| 870 | do { |
||
| 871 | $proceed = 0; |
||
| 872 | list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v); |
||
| 873 | if (false !== $sub_r) { |
||
| 874 | $r .= $sub_r; |
||
| 875 | $proceed = 1; |
||
| 876 | } elseif ($sub_r = $this->x("\.", $sub_v)) { |
||
| 877 | $r .= '.'; |
||
| 878 | $sub_v = $sub_r[1]; |
||
| 879 | $proceed = 1; |
||
| 880 | } |
||
| 881 | } while ($proceed); |
||
| 882 | list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v); |
||
| 883 | $r .= $sub_r ?: ''; |
||
| 884 | } |
||
| 885 | |||
| 886 | 87 | return [$r, $sub_v]; |
|
| 887 | } |
||
| 888 | |||
| 889 | /* 100 */ |
||
| 890 | |||
| 891 | 23 | protected function xPN_LOCAL($v) |
|
| 892 | { |
||
| 893 | 23 | if (($sub_r = $this->x("([^\s\(\)\{\}\[\]\;\,\.]+)", $v, 's')) && !preg_match('/^\./', $sub_r[2])) {/* accelerator */ |
|
| 894 | 23 | return [$sub_r[1], $sub_r[2]]; /* @@testing */ |
|
| 895 | } |
||
| 896 | 9 | $r = ''; |
|
| 897 | 9 | $sub_v = $v; |
|
| 898 | do { |
||
| 899 | 9 | $proceed = 0; |
|
| 900 | 9 | if ($this->x('\s', $sub_v)) { |
|
| 901 | return [$r, $sub_v]; |
||
| 902 | } |
||
| 903 | 9 | if ($sub_r = $this->x('([0-9])', $sub_v)) { |
|
| 904 | 4 | $r .= $sub_r[1]; |
|
| 905 | 4 | $sub_v = $sub_r[2]; |
|
| 906 | 4 | $proceed = 1; |
|
| 907 | 9 | } elseif ((list($sub_r, $sub_v) = $this->xPN_CHARS_U($sub_v)) && $sub_r) { |
|
| 908 | 4 | $r .= $sub_r; |
|
| 909 | 4 | $proceed = 1; |
|
| 910 | 9 | } elseif ($r) { |
|
| 911 | 4 | if (($sub_r = $this->x('(\.)', $sub_v)) && !preg_match('/^[\s\}]/s', $sub_r[2])) { |
|
| 912 | $r .= $sub_r[1]; |
||
| 913 | $sub_v = $sub_r[2]; |
||
| 914 | } |
||
| 915 | 4 | if ((list($sub_r, $sub_v) = $this->xPN_CHARS($sub_v)) && $sub_r) { |
|
| 916 | $r .= $sub_r; |
||
| 917 | $proceed = 1; |
||
| 918 | } |
||
| 919 | } |
||
| 920 | 9 | } while ($proceed); |
|
| 921 | |||
| 922 | 9 | return [$r, $sub_v]; |
|
| 923 | } |
||
| 924 | |||
| 925 | 62 | protected function unescapeNtripleUTF($v) |
|
| 954 | } |
||
| 955 | |||
| 956 | 87 | protected function xPlaceholder($v) |
|
| 957 | { |
||
| 958 | //if ($r = $this->x('(\?|\$)\{([^\}]+)\}', $v)) { |
||
| 959 | 87 | if ($r = $this->x('(\?|\$)', $v)) { |
|
| 960 | if (preg_match('/(\{(?:[^{}]+|(?R))*\})/', $r[2], $m) && 0 === strpos(trim($r[2]), $m[1])) { |
||
| 961 | $ph = substr($m[1], 1, -1); |
||
| 962 | $rest = substr(trim($r[2]), \strlen($m[1])); |
||
| 963 | if (!isset($this->r['placeholders'])) { |
||
| 964 | $this->r['placeholders'] = []; |
||
| 965 | } |
||
| 966 | if (!\in_array($ph, $this->r['placeholders'])) { |
||
| 967 | $this->r['placeholders'][] = $ph; |
||
| 968 | } |
||
| 969 | |||
| 970 | return [['value' => $ph, 'type' => 'placeholder'], $rest]; |
||
| 971 | } |
||
| 972 | } |
||
| 973 | |||
| 977 |
Loading history...