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