third_party
1 year ago
IQuery.php
3 years ago
LICENSE
4 years ago
gan_formatter.php
3 years ago
gan_node_html.php
5 months ago
gan_parser_html.php
3 years ago
gan_selector_html.php
5 months ago
gan_tokenizer.php
1 year ago
gan_xml2array.php
3 years ago
ganon.php
5 months ago
index.php
3 years ago
pQuery.php
3 years ago
gan_selector_html.php
952 lines
| 1 | <?php // phpcs:ignore SlevomatCodingStandard.TypeHints.DeclareStrictTypes.DeclareStrictTypesMissing |
| 2 | /** |
| 3 | * @author Niels A.D. |
| 4 | * @author Todd Burry <todd@vanillaforums.com> |
| 5 | * @copyright 2010 Niels A.D., 2014 Todd Burry |
| 6 | * @license http://opensource.org/licenses/LGPL-2.1 LGPL-2.1 |
| 7 | * @package pQuery |
| 8 | */ |
| 9 | |
| 10 | namespace MailPoetVendor\pQuery; |
| 11 | |
| 12 | if (!defined('ABSPATH')) exit; |
| 13 | |
| 14 | |
| 15 | /** |
| 16 | * Tokenizes a css selector query |
| 17 | */ |
| 18 | class CSSQueryTokenizer extends TokenizerBase { |
| 19 | |
| 20 | /** |
| 21 | * Opening bracket token, used for "[" |
| 22 | */ |
| 23 | const TOK_BRACKET_OPEN = 100; |
| 24 | /** |
| 25 | * Closing bracket token, used for "]" |
| 26 | */ |
| 27 | const TOK_BRACKET_CLOSE = 101; |
| 28 | /** |
| 29 | * Opening brace token, used for "(" |
| 30 | */ |
| 31 | const TOK_BRACE_OPEN = 102; |
| 32 | /** |
| 33 | * Closing brace token, used for ")" |
| 34 | */ |
| 35 | const TOK_BRACE_CLOSE = 103; |
| 36 | /** |
| 37 | * String token |
| 38 | */ |
| 39 | const TOK_STRING = 104; |
| 40 | /** |
| 41 | * Colon token, used for ":" |
| 42 | */ |
| 43 | const TOK_COLON = 105; |
| 44 | /** |
| 45 | * Comma token, used for "," |
| 46 | */ |
| 47 | const TOK_COMMA = 106; |
| 48 | /** |
| 49 | * "Not" token, used for "!" |
| 50 | */ |
| 51 | const TOK_NOT = 107; |
| 52 | |
| 53 | /** |
| 54 | * "All" token, used for "*" in query |
| 55 | */ |
| 56 | const TOK_ALL = 108; |
| 57 | /** |
| 58 | * Pipe token, used for "|" |
| 59 | */ |
| 60 | const TOK_PIPE = 109; |
| 61 | /** |
| 62 | * Plus token, used for "+" |
| 63 | */ |
| 64 | const TOK_PLUS = 110; |
| 65 | /** |
| 66 | * "Sibling" token, used for "~" in query |
| 67 | */ |
| 68 | const TOK_SIBLING = 111; |
| 69 | /** |
| 70 | * Class token, used for "." in query |
| 71 | */ |
| 72 | const TOK_CLASS = 112; |
| 73 | /** |
| 74 | * ID token, used for "#" in query |
| 75 | */ |
| 76 | const TOK_ID = 113; |
| 77 | /** |
| 78 | * Child token, used for ">" in query |
| 79 | */ |
| 80 | const TOK_CHILD = 114; |
| 81 | |
| 82 | /** |
| 83 | * Attribute compare prefix token, used for "|=" |
| 84 | */ |
| 85 | const TOK_COMPARE_PREFIX = 115; |
| 86 | /** |
| 87 | * Attribute contains token, used for "*=" |
| 88 | */ |
| 89 | const TOK_COMPARE_CONTAINS = 116; |
| 90 | /** |
| 91 | * Attribute contains word token, used for "~=" |
| 92 | */ |
| 93 | const TOK_COMPARE_CONTAINS_WORD = 117; |
| 94 | /** |
| 95 | * Attribute compare end token, used for "$=" |
| 96 | */ |
| 97 | const TOK_COMPARE_ENDS = 118; |
| 98 | /** |
| 99 | * Attribute equals token, used for "=" |
| 100 | */ |
| 101 | const TOK_COMPARE_EQUALS = 119; |
| 102 | /** |
| 103 | * Attribute not equal token, used for "!=" |
| 104 | */ |
| 105 | const TOK_COMPARE_NOT_EQUAL = 120; |
| 106 | /** |
| 107 | * Attribute compare bigger than token, used for ">=" |
| 108 | */ |
| 109 | const TOK_COMPARE_BIGGER_THAN = 121; |
| 110 | /** |
| 111 | * Attribute compare smaller than token, used for "<=" |
| 112 | */ |
| 113 | const TOK_COMPARE_SMALLER_THAN = 122; |
| 114 | /** |
| 115 | * Attribute compare with regex, used for "%=" |
| 116 | */ |
| 117 | const TOK_COMPARE_REGEX = 123; |
| 118 | /** |
| 119 | * Attribute compare start token, used for "^=" |
| 120 | */ |
| 121 | const TOK_COMPARE_STARTS = 124; |
| 122 | |
| 123 | /** |
| 124 | * Sets query identifiers |
| 125 | * @see TokenizerBase::$identifiers |
| 126 | * @access private |
| 127 | */ |
| 128 | var $identifiers = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890_-?'; |
| 129 | |
| 130 | /** |
| 131 | * Map characters to match their tokens |
| 132 | * @see TokenizerBase::$custom_char_map |
| 133 | * @access private |
| 134 | */ |
| 135 | var $custom_char_map = array( |
| 136 | '.' => self::TOK_CLASS, |
| 137 | '#' => self::TOK_ID, |
| 138 | ',' => self::TOK_COMMA, |
| 139 | '>' => 'parse_gt',//self::TOK_CHILD, |
| 140 | |
| 141 | '+' => self::TOK_PLUS, |
| 142 | '~' => 'parse_sibling', |
| 143 | |
| 144 | '|' => 'parse_pipe', |
| 145 | '*' => 'parse_star', |
| 146 | '$' => 'parse_compare', |
| 147 | '=' => self::TOK_COMPARE_EQUALS, |
| 148 | '!' => 'parse_not', |
| 149 | '%' => 'parse_compare', |
| 150 | '^' => 'parse_compare', |
| 151 | '<' => 'parse_compare', |
| 152 | |
| 153 | '"' => 'parse_string', |
| 154 | "'" => 'parse_string', |
| 155 | '(' => self::TOK_BRACE_OPEN, |
| 156 | ')' => self::TOK_BRACE_CLOSE, |
| 157 | '[' => self::TOK_BRACKET_OPEN, |
| 158 | ']' => self::TOK_BRACKET_CLOSE, |
| 159 | ':' => self::TOK_COLON |
| 160 | ); |
| 161 | |
| 162 | /** |
| 163 | * Parse ">" character |
| 164 | * @internal Could be {@link TOK_CHILD} or {@link TOK_COMPARE_BIGGER_THAN} |
| 165 | * @return int |
| 166 | */ |
| 167 | protected function parse_gt() { |
| 168 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 169 | ++$this->pos; |
| 170 | return ($this->token = self::TOK_COMPARE_BIGGER_THAN); |
| 171 | } else { |
| 172 | return ($this->token = self::TOK_CHILD); |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | /** |
| 177 | * Parse "~" character |
| 178 | * @internal Could be {@link TOK_SIBLING} or {@link TOK_COMPARE_CONTAINS_WORD} |
| 179 | * @return int |
| 180 | */ |
| 181 | protected function parse_sibling() { |
| 182 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 183 | ++$this->pos; |
| 184 | return ($this->token = self::TOK_COMPARE_CONTAINS_WORD); |
| 185 | } else { |
| 186 | return ($this->token = self::TOK_SIBLING); |
| 187 | } |
| 188 | } |
| 189 | |
| 190 | /** |
| 191 | * Parse "|" character |
| 192 | * @internal Could be {@link TOK_PIPE} or {@link TOK_COMPARE_PREFIX} |
| 193 | * @return int |
| 194 | */ |
| 195 | protected function parse_pipe() { |
| 196 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 197 | ++$this->pos; |
| 198 | return ($this->token = self::TOK_COMPARE_PREFIX); |
| 199 | } else { |
| 200 | return ($this->token = self::TOK_PIPE); |
| 201 | } |
| 202 | } |
| 203 | |
| 204 | /** |
| 205 | * Parse "*" character |
| 206 | * @internal Could be {@link TOK_ALL} or {@link TOK_COMPARE_CONTAINS} |
| 207 | * @return int |
| 208 | */ |
| 209 | protected function parse_star() { |
| 210 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 211 | ++$this->pos; |
| 212 | return ($this->token = self::TOK_COMPARE_CONTAINS); |
| 213 | } else { |
| 214 | return ($this->token = self::TOK_ALL); |
| 215 | } |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * Parse "!" character |
| 220 | * @internal Could be {@link TOK_NOT} or {@link TOK_COMPARE_NOT_EQUAL} |
| 221 | * @return int |
| 222 | */ |
| 223 | protected function parse_not() { |
| 224 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 225 | ++$this->pos; |
| 226 | return ($this->token = self::TOK_COMPARE_NOT_EQUAL); |
| 227 | } else { |
| 228 | return ($this->token = self::TOK_NOT); |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | /** |
| 233 | * Parse several compare characters |
| 234 | * @return int |
| 235 | */ |
| 236 | protected function parse_compare() { |
| 237 | if ((($this->pos + 1) < $this->size) && ($this->doc[$this->pos + 1] === '=')) { |
| 238 | switch($this->doc[$this->pos++]) { |
| 239 | case '$': |
| 240 | return ($this->token = self::TOK_COMPARE_ENDS); |
| 241 | case '%': |
| 242 | return ($this->token = self::TOK_COMPARE_REGEX); |
| 243 | case '^': |
| 244 | return ($this->token = self::TOK_COMPARE_STARTS); |
| 245 | case '<': |
| 246 | return ($this->token = self::TOK_COMPARE_SMALLER_THAN); |
| 247 | } |
| 248 | } |
| 249 | return false; |
| 250 | } |
| 251 | |
| 252 | /** |
| 253 | * Parse strings (" and ') |
| 254 | * @return int |
| 255 | */ |
| 256 | protected function parse_string() { |
| 257 | $char = $this->doc[$this->pos]; |
| 258 | |
| 259 | while (true) { |
| 260 | if ($this->next_search($char.'\\', false) !== self::TOK_NULL) { |
| 261 | if($this->doc[$this->pos] === $char) { |
| 262 | break; |
| 263 | } else { |
| 264 | ++$this->pos; |
| 265 | } |
| 266 | } else { |
| 267 | $this->pos = $this->size - 1; |
| 268 | break; |
| 269 | } |
| 270 | } |
| 271 | |
| 272 | return ($this->token = self::TOK_STRING); |
| 273 | } |
| 274 | |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * Performs a css select query on HTML nodes |
| 279 | */ |
| 280 | class HtmlSelector { |
| 281 | |
| 282 | /** |
| 283 | * Parser object |
| 284 | * @internal If string, then it will create a new instance as parser |
| 285 | * @var CSSQueryTokenizer |
| 286 | */ |
| 287 | var $parser = 'MailPoetVendor\\pQuery\\CSSQueryTokenizer'; |
| 288 | |
| 289 | /** |
| 290 | * Target of queries |
| 291 | * @var DomNode |
| 292 | */ |
| 293 | var $root = null; |
| 294 | |
| 295 | /** |
| 296 | * Last performed query, result in {@link $result} |
| 297 | * @var string |
| 298 | */ |
| 299 | var $query = ''; |
| 300 | |
| 301 | /** |
| 302 | * Array of matching nodes |
| 303 | * @var array |
| 304 | */ |
| 305 | var $result = array(); |
| 306 | |
| 307 | /** |
| 308 | * Include root in search, if false the only child nodes are evaluated |
| 309 | * @var bool |
| 310 | */ |
| 311 | var $search_root = false; |
| 312 | |
| 313 | /** |
| 314 | * Search recursively |
| 315 | * @var bool |
| 316 | */ |
| 317 | var $search_recursive = true; |
| 318 | |
| 319 | /** |
| 320 | * Extra function map for custom filters |
| 321 | * @var array |
| 322 | * @internal array('root' => 'filter_root') will cause the |
| 323 | * selector to call $this->filter_root at :root |
| 324 | * @see DomNode::$filter_map |
| 325 | */ |
| 326 | var $custom_filter_map = array(); |
| 327 | |
| 328 | /** |
| 329 | * Class constructor |
| 330 | * @param DomNode $root {@link $root} |
| 331 | * @param string $query |
| 332 | * @param bool $search_root {@link $search_root} |
| 333 | * @param bool $search_recursive {@link $search_recursive} |
| 334 | * @param CSSQueryTokenizer $parser If null, then default class will be used |
| 335 | */ |
| 336 | function __construct($root, $query = '*', $search_root = false, $search_recursive = true, $parser = null) { |
| 337 | if ($parser === null) { |
| 338 | $parser = new $this->parser(); |
| 339 | } |
| 340 | $this->parser = $parser; |
| 341 | $this->root =& $root; |
| 342 | |
| 343 | $this->search_root = $search_root; |
| 344 | $this->search_recursive = $search_recursive; |
| 345 | |
| 346 | $this->select($query); |
| 347 | } |
| 348 | |
| 349 | #php4 PHP4 class constructor compatibility |
| 350 | #function HtmlSelector($root, $query = '*', $search_root = false, $search_recursive = true, $parser = null) {return $this->__construct($root, $query, $search_root, $search_recursive, $parser);} |
| 351 | #php4e |
| 352 | |
| 353 | /** |
| 354 | * toString method, returns {@link $query} |
| 355 | * @return string |
| 356 | * @access private |
| 357 | */ |
| 358 | function __toString() { |
| 359 | return $this->query; |
| 360 | } |
| 361 | |
| 362 | /** |
| 363 | * Class magic invoke method, performs {@link select()} |
| 364 | * @return array |
| 365 | * @access private |
| 366 | */ |
| 367 | function __invoke($query = '*') { |
| 368 | return $this->select($query); |
| 369 | } |
| 370 | |
| 371 | /** |
| 372 | * Perform query |
| 373 | * @param string $query |
| 374 | * @return array False on failure |
| 375 | */ |
| 376 | function select($query = '*') { |
| 377 | $this->parser->setDoc($query); |
| 378 | $this->query = $query; |
| 379 | return (($this->parse()) ? $this->result : false); |
| 380 | } |
| 381 | |
| 382 | /** |
| 383 | * Trigger error |
| 384 | * @param string $error |
| 385 | * @internal %pos% and %tok% will be replace in string with position and token(string) |
| 386 | * @access private |
| 387 | */ |
| 388 | protected function error($error) { |
| 389 | $error = htmlentities(str_replace( |
| 390 | array('%tok%', '%pos%'), |
| 391 | array($this->parser->getTokenString(), (int) $this->parser->getPos()), |
| 392 | $error |
| 393 | ), ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML401); |
| 394 | |
| 395 | // phpcs:ignore QITStandard.PHP.DebugCode.DebugFunctionFound |
| 396 | trigger_error(esc_html($error)); |
| 397 | } |
| 398 | |
| 399 | /** |
| 400 | * Get identifier (parse identifier or string) |
| 401 | * @param bool $do_error Error on failure |
| 402 | * @return string False on failure |
| 403 | * @access private |
| 404 | */ |
| 405 | protected function parse_getIdentifier($do_error = true) { |
| 406 | $p =& $this->parser; |
| 407 | $tok = $p->token; |
| 408 | |
| 409 | if ($tok === CSSQueryTokenizer::TOK_IDENTIFIER) { |
| 410 | return $p->getTokenString(); |
| 411 | } elseif($tok === CSSQueryTokenizer::TOK_STRING) { |
| 412 | return str_replace(array('\\\'', '\\"', '\\\\'), array('\'', '"', '\\'), $p->getTokenString(1, -1)); |
| 413 | } elseif ($do_error) { |
| 414 | $this->error('Expected identifier at %pos%!'); |
| 415 | } |
| 416 | return false; |
| 417 | } |
| 418 | |
| 419 | /** |
| 420 | * Get query conditions (tag, attribute and filter conditions) |
| 421 | * @return array False on failure |
| 422 | * @see DomNode::match() |
| 423 | * @access private |
| 424 | */ |
| 425 | protected function parse_conditions() { |
| 426 | $p =& $this->parser; |
| 427 | $tok = $p->token; |
| 428 | |
| 429 | if ($tok === CSSQueryTokenizer::TOK_NULL) { |
| 430 | $this->error('Invalid search pattern(1): Empty string!'); |
| 431 | return false; |
| 432 | } |
| 433 | $conditions_all = array(); |
| 434 | |
| 435 | //Tags |
| 436 | while ($tok !== CSSQueryTokenizer::TOK_NULL) { |
| 437 | $conditions = array('tags' => array(), 'attributes' => array()); |
| 438 | |
| 439 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 440 | $tok = $p->next(); |
| 441 | if (($tok === CSSQueryTokenizer::TOK_PIPE) && ($tok = $p->next()) && ($tok !== CSSQueryTokenizer::TOK_ALL)) { |
| 442 | if (($tag = $this->parse_getIdentifier()) === false) { |
| 443 | return false; |
| 444 | } |
| 445 | $conditions['tags'][] = array( |
| 446 | 'tag' => $tag, |
| 447 | 'compare' => 'name' |
| 448 | ); |
| 449 | $tok = $p->next_no_whitespace(); |
| 450 | } else { |
| 451 | $conditions['tags'][''] = array( |
| 452 | 'tag' => '', |
| 453 | 'match' => false |
| 454 | ); |
| 455 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 456 | $tok = $p->next_no_whitespace(); |
| 457 | } |
| 458 | } |
| 459 | } elseif ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 460 | $tok = $p->next(); |
| 461 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 462 | $conditions['tags'][] = array( |
| 463 | 'tag' => '', |
| 464 | 'compare' => 'namespace', |
| 465 | ); |
| 466 | } elseif (($tag = $this->parse_getIdentifier()) !== false) { |
| 467 | $conditions['tags'][] = array( |
| 468 | 'tag' => $tag, |
| 469 | 'compare' => 'total', |
| 470 | ); |
| 471 | } else { |
| 472 | return false; |
| 473 | } |
| 474 | $tok = $p->next_no_whitespace(); |
| 475 | } elseif ($tok === CSSQueryTokenizer::TOK_BRACE_OPEN) { |
| 476 | $tok = $p->next_no_whitespace(); |
| 477 | $last_mode = 'or'; |
| 478 | |
| 479 | while (true) { |
| 480 | $match = true; |
| 481 | $compare = 'total'; |
| 482 | |
| 483 | if ($tok === CSSQueryTokenizer::TOK_NOT) { |
| 484 | $match = false; |
| 485 | $tok = $p->next_no_whitespace(); |
| 486 | } |
| 487 | |
| 488 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 489 | $tok = $p->next(); |
| 490 | if ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 491 | $this->next(); |
| 492 | $compare = 'name'; |
| 493 | if (($tag = $this->parse_getIdentifier()) === false) { |
| 494 | return false; |
| 495 | } |
| 496 | } |
| 497 | } elseif ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 498 | $tok = $p->next(); |
| 499 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 500 | $tag = ''; |
| 501 | $compare = 'namespace'; |
| 502 | } elseif (($tag = $this->parse_getIdentifier()) === false) { |
| 503 | return false; |
| 504 | } |
| 505 | $tok = $p->next_no_whitespace(); |
| 506 | } else { |
| 507 | if (($tag = $this->parse_getIdentifier()) === false) { |
| 508 | return false; |
| 509 | } |
| 510 | $tok = $p->next(); |
| 511 | if ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 512 | $tok = $p->next(); |
| 513 | |
| 514 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 515 | $compare = 'namespace'; |
| 516 | } elseif (($tag_name = $this->parse_getIdentifier()) !== false) { |
| 517 | $tag = $tag.':'.$tag_name; |
| 518 | } else { |
| 519 | return false; |
| 520 | } |
| 521 | |
| 522 | $tok = $p->next_no_whitespace(); |
| 523 | } |
| 524 | } |
| 525 | if ($tok === CSSQueryTokenizer::TOK_WHITESPACE) { |
| 526 | $tok = $p->next_no_whitespace(); |
| 527 | } |
| 528 | |
| 529 | $conditions['tags'][] = array( |
| 530 | 'tag' => $tag, |
| 531 | 'match' => $match, |
| 532 | 'operator' => $last_mode, |
| 533 | 'compare' => $compare |
| 534 | ); |
| 535 | switch($tok) { |
| 536 | case CSSQueryTokenizer::TOK_COMMA: |
| 537 | $tok = $p->next_no_whitespace(); |
| 538 | $last_mode = 'or'; |
| 539 | continue 2; |
| 540 | case CSSQueryTokenizer::TOK_PLUS: |
| 541 | $tok = $p->next_no_whitespace(); |
| 542 | $last_mode = 'and'; |
| 543 | continue 2; |
| 544 | case CSSQueryTokenizer::TOK_BRACE_CLOSE: |
| 545 | $tok = $p->next(); |
| 546 | break 2; |
| 547 | default: |
| 548 | $this->error('Expected closing brace or comma at pos %pos%!'); |
| 549 | return false; |
| 550 | } |
| 551 | } |
| 552 | } elseif (($tag = $this->parse_getIdentifier(false)) !== false) { |
| 553 | $tok = $p->next(); |
| 554 | if ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 555 | $tok = $p->next(); |
| 556 | |
| 557 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 558 | $conditions['tags'][] = array( |
| 559 | 'tag' => $tag, |
| 560 | 'compare' => 'namespace' |
| 561 | ); |
| 562 | } elseif (($tag_name = $this->parse_getIdentifier()) !== false) { |
| 563 | $tag = $tag.':'.$tag_name; |
| 564 | $conditions['tags'][] = array( |
| 565 | 'tag' => $tag, |
| 566 | 'match' => true |
| 567 | ); |
| 568 | } else { |
| 569 | return false; |
| 570 | } |
| 571 | |
| 572 | $tok = $p->next(); |
| 573 | } elseif ($tag === 'text' && $tok === CSSQueryTokenizer::TOK_BRACE_OPEN) { |
| 574 | $pos = $p->getPos(); |
| 575 | $tok = $p->next(); |
| 576 | if ($tok === CSSQueryTokenizer::TOK_BRACE_CLOSE) { |
| 577 | $conditions['tags'][] = array( |
| 578 | 'tag' => '~text~', |
| 579 | 'match' => true |
| 580 | ); |
| 581 | $p->next(); |
| 582 | } else { |
| 583 | $p->setPos($pos); |
| 584 | } |
| 585 | } else { |
| 586 | $conditions['tags'][] = array( |
| 587 | 'tag' => $tag, |
| 588 | 'match' => true |
| 589 | ); |
| 590 | } |
| 591 | } else { |
| 592 | unset($conditions['tags']); |
| 593 | } |
| 594 | |
| 595 | //Class |
| 596 | $last_mode = 'or'; |
| 597 | if ($tok === CSSQueryTokenizer::TOK_CLASS) { |
| 598 | $p->next(); |
| 599 | if (($class = $this->parse_getIdentifier()) === false) { |
| 600 | return false; |
| 601 | } |
| 602 | |
| 603 | $conditions['attributes'][] = array( |
| 604 | 'attribute' => 'class', |
| 605 | 'operator_value' => 'contains_word', |
| 606 | 'value' => $class, |
| 607 | 'operator_result' => $last_mode |
| 608 | ); |
| 609 | $last_mode = 'and'; |
| 610 | $tok = $p->next(); |
| 611 | } |
| 612 | |
| 613 | //ID |
| 614 | if ($tok === CSSQueryTokenizer::TOK_ID) { |
| 615 | $p->next(); |
| 616 | if (($id = $this->parse_getIdentifier()) === false) { |
| 617 | return false; |
| 618 | } |
| 619 | |
| 620 | $conditions['attributes'][] = array( |
| 621 | 'attribute' => 'id', |
| 622 | 'operator_value' => 'equals', |
| 623 | 'value' => $id, |
| 624 | 'operator_result' => $last_mode |
| 625 | ); |
| 626 | $last_mode = 'and'; |
| 627 | $tok = $p->next(); |
| 628 | } |
| 629 | |
| 630 | //Attributes |
| 631 | if ($tok === CSSQueryTokenizer::TOK_BRACKET_OPEN) { |
| 632 | $tok = $p->next_no_whitespace(); |
| 633 | |
| 634 | while (true) { |
| 635 | $match = true; |
| 636 | $compare = 'total'; |
| 637 | if ($tok === CSSQueryTokenizer::TOK_NOT) { |
| 638 | $match = false; |
| 639 | $tok = $p->next_no_whitespace(); |
| 640 | } |
| 641 | |
| 642 | if ($tok === CSSQueryTokenizer::TOK_ALL) { |
| 643 | $tok = $p->next(); |
| 644 | if ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 645 | $tok = $p->next(); |
| 646 | if (($attribute = $this->parse_getIdentifier()) === false) { |
| 647 | return false; |
| 648 | } |
| 649 | $compare = 'name'; |
| 650 | $tok = $p->next(); |
| 651 | } else { |
| 652 | $this->error('Expected pipe at pos %pos%!'); |
| 653 | return false; |
| 654 | } |
| 655 | } elseif ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 656 | $tok = $p->next(); |
| 657 | if (($tag = $this->parse_getIdentifier()) === false) { |
| 658 | return false; |
| 659 | } |
| 660 | $tok = $p->next_no_whitespace(); |
| 661 | } elseif (($attribute = $this->parse_getIdentifier()) !== false) { |
| 662 | $tok = $p->next(); |
| 663 | if ($tok === CSSQueryTokenizer::TOK_PIPE) { |
| 664 | $tok = $p->next(); |
| 665 | |
| 666 | if (($attribute_name = $this->parse_getIdentifier()) !== false) { |
| 667 | $attribute = $attribute.':'.$attribute_name; |
| 668 | } else { |
| 669 | return false; |
| 670 | } |
| 671 | |
| 672 | $tok = $p->next(); |
| 673 | } |
| 674 | } else { |
| 675 | return false; |
| 676 | } |
| 677 | if ($tok === CSSQueryTokenizer::TOK_WHITESPACE) { |
| 678 | $tok = $p->next_no_whitespace(); |
| 679 | } |
| 680 | |
| 681 | $operator_value = ''; |
| 682 | $val = ''; |
| 683 | switch($tok) { |
| 684 | case CSSQueryTokenizer::TOK_COMPARE_PREFIX: |
| 685 | case CSSQueryTokenizer::TOK_COMPARE_CONTAINS: |
| 686 | case CSSQueryTokenizer::TOK_COMPARE_CONTAINS_WORD: |
| 687 | case CSSQueryTokenizer::TOK_COMPARE_ENDS: |
| 688 | case CSSQueryTokenizer::TOK_COMPARE_EQUALS: |
| 689 | case CSSQueryTokenizer::TOK_COMPARE_NOT_EQUAL: |
| 690 | case CSSQueryTokenizer::TOK_COMPARE_REGEX: |
| 691 | case CSSQueryTokenizer::TOK_COMPARE_STARTS: |
| 692 | case CSSQueryTokenizer::TOK_COMPARE_BIGGER_THAN: |
| 693 | case CSSQueryTokenizer::TOK_COMPARE_SMALLER_THAN: |
| 694 | $operator_value = $p->getTokenString(($tok === CSSQueryTokenizer::TOK_COMPARE_EQUALS) ? 0 : -1); |
| 695 | $p->next_no_whitespace(); |
| 696 | |
| 697 | if (($val = $this->parse_getIdentifier()) === false) { |
| 698 | return false; |
| 699 | } |
| 700 | |
| 701 | $tok = $p->next_no_whitespace(); |
| 702 | break; |
| 703 | } |
| 704 | |
| 705 | if ($operator_value && $val) { |
| 706 | $conditions['attributes'][] = array( |
| 707 | 'attribute' => $attribute, |
| 708 | 'operator_value' => $operator_value, |
| 709 | 'value' => $val, |
| 710 | 'match' => $match, |
| 711 | 'operator_result' => $last_mode, |
| 712 | 'compare' => $compare |
| 713 | ); |
| 714 | } else { |
| 715 | $conditions['attributes'][] = array( |
| 716 | 'attribute' => $attribute, |
| 717 | 'value' => $match, |
| 718 | 'operator_result' => $last_mode, |
| 719 | 'compare' => $compare |
| 720 | ); |
| 721 | } |
| 722 | |
| 723 | switch($tok) { |
| 724 | case CSSQueryTokenizer::TOK_COMMA: |
| 725 | $tok = $p->next_no_whitespace(); |
| 726 | $last_mode = 'or'; |
| 727 | continue 2; |
| 728 | case CSSQueryTokenizer::TOK_PLUS: |
| 729 | $tok = $p->next_no_whitespace(); |
| 730 | $last_mode = 'and'; |
| 731 | continue 2; |
| 732 | case CSSQueryTokenizer::TOK_BRACKET_CLOSE: |
| 733 | $tok = $p->next(); |
| 734 | break 2; |
| 735 | default: |
| 736 | $this->error('Expected closing bracket or comma at pos %pos%!'); |
| 737 | return false; |
| 738 | } |
| 739 | } |
| 740 | } |
| 741 | |
| 742 | if (count($conditions['attributes']) < 1) { |
| 743 | unset($conditions['attributes']); |
| 744 | } |
| 745 | |
| 746 | while($tok === CSSQueryTokenizer::TOK_COLON) { |
| 747 | if (count($conditions) < 1) { |
| 748 | $conditions['tags'] = array(array( |
| 749 | 'tag' => '', |
| 750 | 'match' => false |
| 751 | )); |
| 752 | } |
| 753 | |
| 754 | $tok = $p->next(); |
| 755 | if (($filter = $this->parse_getIdentifier()) === false) { |
| 756 | return false; |
| 757 | } |
| 758 | |
| 759 | if (($tok = $p->next()) === CSSQueryTokenizer::TOK_BRACE_OPEN) { |
| 760 | $start = $p->pos; |
| 761 | $count = 1; |
| 762 | while ((($tok = $p->next()) !== CSSQueryTokenizer::TOK_NULL) && !(($tok === CSSQueryTokenizer::TOK_BRACE_CLOSE) && (--$count === 0))) { |
| 763 | if ($tok === CSSQueryTokenizer::TOK_BRACE_OPEN) { |
| 764 | ++$count; |
| 765 | } |
| 766 | } |
| 767 | |
| 768 | |
| 769 | if ($tok !== CSSQueryTokenizer::TOK_BRACE_CLOSE) { |
| 770 | $this->error('Expected closing brace at pos %pos%!'); |
| 771 | return false; |
| 772 | } |
| 773 | $len = $p->pos - 1 - $start; |
| 774 | $params = (($len > 0) ? substr($p->doc, $start + 1, $len) : ''); |
| 775 | $tok = $p->next(); |
| 776 | } else { |
| 777 | $params = ''; |
| 778 | } |
| 779 | |
| 780 | $conditions['filters'][] = array('filter' => $filter, 'params' => $params); |
| 781 | } |
| 782 | if (count($conditions) < 1) { |
| 783 | $this->error('Invalid search pattern(2): No conditions found!'); |
| 784 | return false; |
| 785 | } |
| 786 | $conditions_all[] = $conditions; |
| 787 | |
| 788 | if ($tok === CSSQueryTokenizer::TOK_WHITESPACE) { |
| 789 | $tok = $p->next_no_whitespace(); |
| 790 | } |
| 791 | |
| 792 | if ($tok === CSSQueryTokenizer::TOK_COMMA) { |
| 793 | $tok = $p->next_no_whitespace(); |
| 794 | continue; |
| 795 | } else { |
| 796 | break; |
| 797 | } |
| 798 | } |
| 799 | |
| 800 | return $conditions_all; |
| 801 | } |
| 802 | |
| 803 | |
| 804 | /** |
| 805 | * Evaluate root node using custom callback |
| 806 | * @param array $conditions {@link parse_conditions()} |
| 807 | * @param bool|int $recursive |
| 808 | * @param bool $check_root |
| 809 | * @return array |
| 810 | * @access private |
| 811 | */ |
| 812 | protected function parse_callback($conditions, $recursive = true, $check_root = false) { |
| 813 | return ($this->result = $this->root->getChildrenByMatch( |
| 814 | $conditions, |
| 815 | $recursive, |
| 816 | $check_root, |
| 817 | $this->custom_filter_map |
| 818 | )); |
| 819 | } |
| 820 | |
| 821 | /** |
| 822 | * Parse first bit of query, only root node has to be evaluated now |
| 823 | * @param bool|int $recursive |
| 824 | * @return bool |
| 825 | * @internal Result of query is set in {@link $result} |
| 826 | * @access private |
| 827 | */ |
| 828 | protected function parse_single($recursive = true) { |
| 829 | if (($c = $this->parse_conditions()) === false) { |
| 830 | return false; |
| 831 | } |
| 832 | |
| 833 | $this->parse_callback($c, $recursive, $this->search_root); |
| 834 | return true; |
| 835 | } |
| 836 | |
| 837 | /** |
| 838 | * Evaluate sibling nodes |
| 839 | * @return bool |
| 840 | * @internal Result of query is set in {@link $result} |
| 841 | * @access private |
| 842 | */ |
| 843 | protected function parse_adjacent() { |
| 844 | $tmp = $this->result; |
| 845 | $this->result = array(); |
| 846 | if (($c = $this->parse_conditions()) === false) { |
| 847 | return false; |
| 848 | } |
| 849 | |
| 850 | foreach($tmp as $t) { |
| 851 | if (($sibling = $t->getNextSibling()) !== false) { |
| 852 | if ($sibling->match($c, true, $this->custom_filter_map)) { |
| 853 | $this->result[] = $sibling; |
| 854 | } |
| 855 | } |
| 856 | } |
| 857 | |
| 858 | return true; |
| 859 | } |
| 860 | |
| 861 | /** |
| 862 | * Evaluate {@link $result} |
| 863 | * @param bool $parent Evaluate parent nodes |
| 864 | * @param bool|int $recursive |
| 865 | * @return bool |
| 866 | * @internal Result of query is set in {@link $result} |
| 867 | * @access private |
| 868 | */ |
| 869 | protected function parse_result($parent = false, $recursive = true) { |
| 870 | $tmp = $this->result; |
| 871 | $tmp_res = array(); |
| 872 | if (($c = $this->parse_conditions()) === false) { |
| 873 | return false; |
| 874 | } |
| 875 | |
| 876 | foreach(array_keys($tmp) as $t) { |
| 877 | $this->root = (($parent) ? $tmp[$t]->parent : $tmp[$t]); |
| 878 | $this->parse_callback($c, $recursive); |
| 879 | foreach(array_keys($this->result) as $r) { |
| 880 | if (!in_array($this->result[$r], $tmp_res, true)) { |
| 881 | $tmp_res[] = $this->result[$r]; |
| 882 | } |
| 883 | } |
| 884 | } |
| 885 | $this->result = $tmp_res; |
| 886 | return true; |
| 887 | } |
| 888 | |
| 889 | /** |
| 890 | * Parse full query |
| 891 | * @return bool |
| 892 | * @internal Result of query is set in {@link $result} |
| 893 | * @access private |
| 894 | */ |
| 895 | protected function parse() { |
| 896 | $p =& $this->parser; |
| 897 | $p->setPos(0); |
| 898 | $this->result = array(); |
| 899 | |
| 900 | if (!$this->parse_single()) { |
| 901 | return false; |
| 902 | } |
| 903 | |
| 904 | while (count($this->result) > 0) { |
| 905 | switch($p->token) { |
| 906 | case CSSQueryTokenizer::TOK_CHILD: |
| 907 | $this->parser->next_no_whitespace(); |
| 908 | if (!$this->parse_result(false, 1)) { |
| 909 | return false; |
| 910 | } |
| 911 | break; |
| 912 | |
| 913 | case CSSQueryTokenizer::TOK_SIBLING: |
| 914 | $this->parser->next_no_whitespace(); |
| 915 | if (!$this->parse_result(true, 1)) { |
| 916 | return false; |
| 917 | } |
| 918 | break; |
| 919 | |
| 920 | case CSSQueryTokenizer::TOK_PLUS: |
| 921 | $this->parser->next_no_whitespace(); |
| 922 | if (!$this->parse_adjacent()) { |
| 923 | return false; |
| 924 | } |
| 925 | break; |
| 926 | |
| 927 | case CSSQueryTokenizer::TOK_ALL: |
| 928 | case CSSQueryTokenizer::TOK_IDENTIFIER: |
| 929 | case CSSQueryTokenizer::TOK_STRING: |
| 930 | case CSSQueryTokenizer::TOK_BRACE_OPEN: |
| 931 | case CSSQueryTokenizer::TOK_BRACKET_OPEN: |
| 932 | case CSSQueryTokenizer::TOK_ID: |
| 933 | case CSSQueryTokenizer::TOK_CLASS: |
| 934 | case CSSQueryTokenizer::TOK_COLON: |
| 935 | if (!$this->parse_result()) { |
| 936 | return false; |
| 937 | } |
| 938 | break; |
| 939 | |
| 940 | case CSSQueryTokenizer::TOK_NULL: |
| 941 | break 2; |
| 942 | |
| 943 | default: |
| 944 | $this->error('Invalid search pattern(3): No result modifier found!'); |
| 945 | return false; |
| 946 | } |
| 947 | } |
| 948 | |
| 949 | return true; |
| 950 | } |
| 951 | } |
| 952 |