| [ Index ] |
|
Code source de eZ Publish 3.9.0 |
1 <?php 2 // 3 // Definition of eZXMLInputParser class 4 // 5 // Created on: <27-Mar-2006 15:28:39 ks> 6 // 7 // SOFTWARE NAME: eZ publish 8 // SOFTWARE RELEASE: 3.9.0 9 // BUILD VERSION: 17785 10 // COPYRIGHT NOTICE: Copyright (C) 1999-2006 eZ systems AS 11 // SOFTWARE LICENSE: GNU General Public License v2.0 12 // NOTICE: > 13 // This program is free software; you can redistribute it and/or 14 // modify it under the terms of version 2.0 of the GNU General 15 // Public License as published by the Free Software Foundation. 16 // 17 // This program is distributed in the hope that it will be useful, 18 // but WITHOUT ANY WARRANTY; without even the implied warranty of 19 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 20 // GNU General Public License for more details. 21 // 22 // You should have received a copy of version 2.0 of the GNU General 23 // Public License along with this program; if not, write to the Free 24 // Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, 25 // MA 02110-1301, USA. 26 // 27 28 /* 29 Base class for the input parser. 30 The goal of the parser is XML/HTML analyzing, fixing and transforming. 31 The input is processed in 2 passes: 32 - 1st pass: Parsing input, check for syntax errors, build DOM tree. 33 - 2nd pass: Walking through DOM tree, checking validity by XML schema, 34 calling tag handlers to transform the tree. 35 36 Both passes are controlled by the arrays described bellow and user handler functions. 37 38 */ 39 40 include_once ( "lib/ezxml/classes/ezxml.php" ); 41 42 if ( !class_exists( 'eZXMLSchema' ) ) 43 include_once ( 'kernel/classes/datatypes/ezxmltext/ezxmlschema.php' ); 44 45 define( 'EZ_XMLINPUTPARSER_SHOW_NO_ERRORS', 0 ); 46 define( 'EZ_XMLINPUTPARSER_SHOW_SCHEMA_ERRORS', 1 ); 47 define( 'EZ_XMLINPUTPARSER_SHOW_ALL_ERRORS', 2 ); 48 49 class eZXMLInputParser 50 { 51 52 /* $InputTags array contains properties of elements that come from the input. 53 54 Each array element describes a tag that comes from the input. Arrays index is 55 a tag's name. Each element is an array that may contain the following members: 56 57 'name' - a string representing a new name of the tag, 58 'nameHandler' - a name of the function that returns new tag name. Function format: 59 function &tagNameHandler( $tagName, &$attributes ) 60 61 If no of those elements are defined the original tag's name is used. 62 63 'noChildren' - boolean value that determines if this tag could have child tags, 64 default value is false. 65 66 Example: 67 68 var $InputTags = array( 69 70 'old-name' => array( 'name' => 'new-name' ), 71 72 'tagname' => array( 'nameHandler' => 'tagNameHandler', 73 'noChildren' => true ), 74 75 ... 76 77 ); 78 */ 79 80 var $InputTags = array(); 81 82 /* 83 $OutputTags array contains properties of elements that are produced in the output. 84 Each array element describes a tag presented in the output. Arrays index is 85 a tag's name. Each element is an array that may contain the following members: 86 87 'parsingHandler' - "Parsing handler" called at parse pass 1 before processing tag's children. 88 'initHandler' - "Init handler" called at pass 2 before proccessing tag's children. 89 'structHandler' - "Structure handler" called at pass 2 after proccessing tag's children, 90 but before schema validity check. It can be used to implement structure 91 transformations. 92 'publishHandler' - "Publish handler" called at pass 2 after schema validity check, so it is called 93 in case the element has it's guaranteed place in the DOM tree. 94 95 'attributes' - an array that describes attributes transformations. Array's index is the 96 original name of an attribute, and the value is the new name. 97 98 'requiredInputAttributes' - attributes that are required in the input tag. If they are not presented 99 it raises invalid input flag. 100 101 Example: 102 103 var $OutputTags = array( 104 105 'custom' => array( 'parsingHandler' => 'parsingHandlerCustom', 106 'initHandler' => 'initHandlerCustom', 107 'structHandler' => 'structHandlerCustom', 108 'publishHandler' => 'publishHandlerCustom', 109 'attributes' => array( 'title' => 'name' ) ), 110 111 ... 112 ); 113 114 */ 115 116 var $OutputTags = array(); 117 118 var $Namespaces = array( 'image' => 'http://ez.no/namespaces/ezpublish3/image/', 119 'xhtml' => 'http://ez.no/namespaces/ezpublish3/xhtml/', 120 'custom' => 'http://ez.no/namespaces/ezpublish3/custom/' ); 121 122 /*! 123 124 The constructor. 125 126 \param $validate If true, parser quits immediately after validity flag (isInputValid) 127 set to false and function 'process' returns false. 128 129 If false, parser tries to modify and transform the input automatically 130 in order to get the valid result. 131 */ 132 133 function eZXMLInputParser( $validate = false, $errorLevel = EZ_XMLINPUTPARSER_SHOW_NO_ERRORS, $parseLineBreaks = false, 134 $removeDefaultAttrs = false ) 135 { 136 $this->quitIfInvalid = $validate; 137 $this->errorLevel = $errorLevel; 138 139 $this->removeDefaultAttrs = $removeDefaultAttrs; 140 $this->parseLineBreaks = $parseLineBreaks; 141 142 $this->XMLSchema =& eZXMLSchema::instance(); 143 //$this->getClassesList(); 144 145 include_once ( 'lib/version.php' ); 146 $this->eZPublishVersion = eZPublishSDK::majorVersion() + eZPublishSDK::minorVersion() * 0.1; 147 148 $ini =& eZINI::instance( 'ezxml.ini' ); 149 if ( $this->eZPublishVersion >= 3.8 ) 150 { 151 if ( $ini->hasVariable( 'InputSettings', 'TrimSpaces' ) ) 152 { 153 $trimSpaces = $ini->variable( 'InputSettings', 'TrimSpaces' ); 154 $this->TrimSpaces = $trimSpaces == 'true' ? true : false; 155 } 156 157 if ( $ini->hasVariable( 'InputSettings', 'AllowMultipleSpaces' ) ) 158 { 159 $allowMultipleSpaces = $ini->variable( 'InputSettings', 'AllowMultipleSpaces' ); 160 $this->AllowMultipleSpaces = $allowMultipleSpaces == 'true' ? true : false; 161 } 162 } 163 else 164 { 165 $this->TrimSpaces = true; 166 $this->AllowMultipleSpaces = false; 167 } 168 169 if ( $this->eZPublishVersion >= 3.9 ) 170 { 171 if ( $ini->hasVariable( 'InputSettings', 'AllowNumericEntities' ) ) 172 { 173 $allowNumericEntities = $ini->variable( 'InputSettings', 'AllowNumericEntities' ); 174 $this->AllowNumericEntities = $allowNumericEntities == 'true' ? true : false; 175 } 176 } 177 else 178 { 179 $this->AllowNumericEntities = false; 180 } 181 182 } 183 184 function setDOMDocumentClass( $DOMDocumentClass ) 185 { 186 $this->DOMDocumentClass = $DOMDocumentClass; 187 } 188 189 function setParseLineBreaks( $value ) 190 { 191 $this->parseLineBreaks = $value; 192 } 193 194 function setRemoveDefaultAttrs( $value ) 195 { 196 $this->removeDefaultAttrs = $value; 197 } 198 199 /*! 200 Call this function to process your input 201 */ 202 function process( $text, $createRootNode = true ) 203 { 204 $text = str_replace( "\r", '', $text); 205 $text = str_replace( "\t", ' ', $text); 206 if ( !$this->parseLineBreaks ) 207 { 208 $text = str_replace( "\n", '', $text); 209 } 210 211 if ( $createRootNode ) 212 { 213 // Creating root section with namespaces definitions 214 $this->Document = new $this->DOMDocumentClass( '', true ); 215 $mainSection =& $this->Document->createElement( 'section' ); 216 $this->Document->appendChild( $mainSection ); 217 foreach( $this->Namespaces as $prefix => $value ) 218 { 219 $mainSection->setAttributeNS( 'http://www.w3.org/2000/xmlns/', 'xmlns:' . $prefix, $value ); 220 } 221 } 222 223 // Perform pass 1 224 // Parsing the source string 225 $this->performPass1( $text ); 226 227 if ( $this->quitIfInvalid && !$this->isInputValid ) 228 return false; 229 230 // Perform pass 2 231 $this->performPass2(); 232 233 if ( $this->quitIfInvalid && !$this->isInputValid ) 234 return false; 235 236 return $this->Document; 237 } 238 239 /* 240 Pass 1: Parsing the source HTML string. 241 */ 242 243 function performPass1( &$data ) 244 { 245 $ret = true; 246 $pos = 0; 247 248 if ( $this->Document->Root ) 249 { 250 do 251 { 252 $this->parseTag( $data, $pos, $this->Document->Root ); 253 if ( $this->quitIfInvalid && !$this->isInputValid ) 254 { 255 $ret = false; 256 break; 257 } 258 259 } 260 while( $pos < strlen( $data ) ); 261 } 262 else 263 { 264 $tmp = null; 265 $this->parseTag( $data, $pos, $tmp ); 266 if ( $this->quitIfInvalid && !$this->isInputValid ) 267 { 268 $ret = false; 269 break; 270 } 271 } 272 return $ret; 273 } 274 275 // The main recursive function for pass 1 276 277 function parseTag( &$data, &$pos, &$parent ) 278 { 279 // Find tag, determine it's type, name and attributes. 280 $initialPos = $pos; 281 282 /*if ( $this->trimSpaces ) 283 { 284 while( $pos < strlen( $data ) && $data[$pos] == ' ' ) $pos++; 285 }*/ 286 287 if ( $pos >= strlen( $data ) ) 288 { 289 return true; 290 } 291 $tagBeginPos = strpos( $data, '<', $pos ); 292 293 if ( $this->parseLineBreaks ) 294 { 295 // Regard line break as a start tag position 296 $lineBreakPos = strpos( $data, "\n", $pos ); 297 if ( $lineBreakPos !== false ) 298 { 299 if ( $tagBeginPos === false ) 300 $tagBeginPos = $lineBreakPos; 301 else 302 $tagBeginPos = min( $tagBeginPos, $lineBreakPos ); 303 } 304 } 305 306 $tagName = ''; 307 $attributes = null; 308 // If it doesn't begin with '<' then its a text node. 309 if ( $tagBeginPos != $pos || $tagBeginPos === false ) 310 { 311 $pos = $initialPos; 312 $tagName = $newTagName = '#text'; 313 $noChildren = true; 314 315 if ( !$tagBeginPos ) 316 $tagBeginPos = strlen( $data ); 317 318 $textContent = substr( $data, $pos, $tagBeginPos - $pos ); 319 320 $textContent = $this->washText( $textContent ); 321 322 $pos = $tagBeginPos; 323 if ( $textContent === '' ) 324 return false; 325 } 326 // Process closing tag. 327 elseif ( $data[$tagBeginPos] == '<' && $tagBeginPos + 1 < strlen( $data ) && 328 $data[$tagBeginPos + 1] == '/' ) 329 { 330 $tagEndPos = strpos( $data, '>', $tagBeginPos + 1 ); 331 if ( $tagEndPos === false ) 332 { 333 $pos = $tagBeginPos + 1; 334 335 $this->isInputValid = false; 336 if ( $this->errorLevel >= 2 ) 337 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong closing tag' ); 338 return false; 339 } 340 341 $pos = $tagEndPos + 1; 342 $closedTagName = strtolower( trim( substr( $data, $tagBeginPos + 2, $tagEndPos - $tagBeginPos - 2 ) ) ); 343 344 // Find matching tag in ParentStack array 345 $firstLoop = true; 346 for( $i = count( $this->ParentStack ) - 1; $i >= 0; $i-- ) 347 { 348 $parentNames = $this->ParentStack[$i]; 349 if ( $parentNames[0] == $closedTagName ) 350 { 351 array_pop( $this->ParentStack ); 352 if ( !$firstLoop ) 353 { 354 $pos = $tagBeginPos; 355 return true; 356 } 357 // If newTagName was '' we don't break children loop 358 elseif ( $parentNames[1] !== '' ) 359 return true; 360 else 361 return false; 362 } 363 $firstLoop = false; 364 } 365 366 $this->isInputValid = false; 367 if ( $this->errorLevel >= 2 ) 368 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong closing tag : </%1>.', false, array( $closedTagName ) ); 369 370 return false; 371 } 372 // Insert <br/> instead of linebreaks 373 elseif ( $this->parseLineBreaks && $data[$tagBeginPos] == "\n" ) 374 { 375 $newTagName = 'br'; 376 $noChildren = true; 377 $pos = $tagBeginPos + 1; 378 } 379 // Regular tag: get tag's name and attributes. 380 else 381 { 382 $tagEndPos = strpos( $data, '>', $tagBeginPos ); 383 if ( $tagEndPos === false ) 384 { 385 $pos = $tagBeginPos + 1; 386 387 $this->isInputValid = false; 388 if ( $this->errorLevel >= 2 ) 389 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong opening tag' ); 390 return false; 391 } 392 393 $pos = $tagEndPos + 1; 394 $tagString = substr( $data, $tagBeginPos + 1, $tagEndPos - $tagBeginPos - 1 ); 395 // Check for final backslash 396 $noChildren = substr( $tagString, -1, 1 ) == '/' ? true : false; 397 // Remove final backslash and spaces 398 $tagString = preg_replace( "/\s*\/$/", "", $tagString ); 399 400 $firstSpacePos = strpos( $tagString, ' ' ); 401 if ( $firstSpacePos === false ) 402 { 403 $tagName = strtolower( trim( $tagString ) ); 404 $attributeString = ''; 405 } 406 else 407 { 408 $tagName = strtolower( substr( $tagString, 0, $firstSpacePos ) ); 409 $attributeString = substr( $tagString, $firstSpacePos + 1 ); 410 $attributeString = trim( $attributeString ); 411 // Parse attribute string 412 if ( $attributeString ) 413 $attributes = $this->parseAttributes( $attributeString ); 414 } 415 416 // Determine tag's name 417 if ( isset( $this->InputTags[$tagName] ) ) 418 { 419 $thisInputTag =& $this->InputTags[$tagName]; 420 421 if ( isset( $thisInputTag['name'] ) ) 422 $newTagName = $thisInputTag['name']; 423 else 424 $newTagName =& $this->callInputHandler( 'nameHandler', $tagName, $attributes ); 425 } 426 else 427 { 428 if ( $this->XMLSchema->exists( $tagName ) ) 429 { 430 $newTagName = $tagName; 431 } 432 else 433 { 434 $this->isInputValid = false; 435 if ( $this->errorLevel >= 2 ) 436 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Unknown tag: <%1>.", false, array( $tagName ) ); 437 438 return false; 439 } 440 } 441 442 // Check 'noChildren' property 443 if ( isset( $thisInputTag['noChildren'] ) ) 444 $noChildren = true; 445 446 $thisOutputTag =& $this->OutputTags[$newTagName]; 447 448 // Implementation of 'autoCloseOn' rule ( Handling of unclosed tags, ex.: <p>, <li> ) 449 if ( isset( $thisOutputTag['autoCloseOn'] ) && 450 $parent && 451 $parent->parentNode && 452 in_array( $parent->nodeName, $thisOutputTag['autoCloseOn'] ) ) 453 { 454 // Wrong nesting: auto-close parent and try to re-parse this tag at higher level 455 array_pop( $this->ParentStack ); 456 $pos = $tagBeginPos; 457 return true; 458 } 459 460 // Append to parent stack 461 if ( !$noChildren && $newTagName !== false ) 462 { 463 $this->ParentStack[] = array( $tagName, $newTagName, $attributeString ); 464 } 465 466 if ( !$newTagName ) 467 { 468 if ( $newTagName === false ) 469 { 470 $this->isInputValid = false; 471 if ( $this->errorLevel >= 2 ) 472 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Can't convert tag's name: <%1>.", false, array( $tagName ) ); 473 } 474 return false; 475 // TODO: return it before and don't append to ParentStack? 476 // (need to skip processing closing tag on empty tagname) 477 // LATER: no.. this is not very good for data consistance 478 } 479 480 // wordmatch.ini support 481 if ( $attributeString ) 482 { 483 $this->wordMatchSupport( $newTagName, $attributes, $attributeString ); 484 } 485 } 486 487 // Create text or normal node. 488 if ( $newTagName == '#text' ) 489 $element = $this->Document->createTextNode( $textContent ); 490 else 491 $element =& $this->Document->createElement( $newTagName ); 492 493 if ( $attributes ) 494 { 495 $this->setAttributes( $element, $attributes ); 496 } 497 498 // Append element as a child or set it as root if there is no parent. 499 if ( $parent ) 500 $parent->appendChild( $element ); 501 else 502 $this->Document->appendChild( $element ); 503 // php5 TODO : $this->Document->documentElement->appendChild( $element ); 504 505 $params = array(); 506 $params[] =& $data; 507 $params[] =& $pos; 508 $params[] =& $tagBeginPos; 509 $result =& $this->callOutputHandler( 'parsingHandler', $element, $params ); 510 511 if ( $result === false ) 512 { 513 // This tag is already parsed in handler 514 if ( !$noChildren ) 515 array_pop( $this->ParentStack ); 516 return false; 517 } 518 519 if ( $this->quitIfInvalid && !$this->isInputValid ) 520 return false; 521 522 // Process children 523 if ( !$noChildren ) 524 { 525 do 526 { 527 $parseResult = $this->parseTag( $data, $pos, $element ); 528 529 if ( $this->quitIfInvalid && !$this->isInputValid ) 530 return false; 531 } 532 while( $parseResult !== true ); 533 } 534 535 return false; 536 } 537 538 /* 539 Helper functions for pass 1 540 */ 541 542 function parseAttributes( $attributeString ) 543 { 544 // Convert single quotes to double quotes 545 $attributeString = preg_replace( "/ +([a-zA-Z0-9:-_#\-]+) *\='(.*?)'/e", "' \\1'.'=\"'.'\\2'.'\"'", ' ' . $attributeString ); 546 547 // Convert no quotes to double quotes and remove extra spaces 548 $attributeString = preg_replace( "/ +([a-zA-Z0-9:-_#\-]+) *\= *([^\s'\"]+)/e", "' \\1'.'=\"'.'\\2'.'\" '", $attributeString ); 549 550 // Split by quotes followed by spaces 551 $attributeArray = preg_split( "#(?<=\") +#", $attributeString ); 552 553 $attributes = array(); 554 foreach( $attributeArray as $attrStr ) 555 { 556 if ( !$attrStr || strlen( $attrStr ) < 4 ) 557 continue; 558 559 list( $attrName, $attrValue ) = split( '="', $attrStr ); 560 561 $attrName = strtolower( trim( $attrName ) ); 562 if ( !$attrName ) 563 continue; 564 565 $attrValue = substr( $attrValue, 0, -1 ); 566 if ( $attrValue === '' || $attrValue === false ) 567 continue; 568 569 $attributes[$attrName] = $attrValue; 570 } 571 572 return $attributes; 573 } 574 575 function setAttributes( &$element, $attributes ) 576 { 577 $thisOutputTag =& $this->OutputTags[$element->nodeName]; 578 579 foreach( $attributes as $key => $value ) 580 { 581 // Convert attribute names 582 if ( isset( $thisOutputTag['attributes'] ) && 583 isset( $thisOutputTag['attributes'][$key] ) ) 584 { 585 $qualifiedName = $thisOutputTag['attributes'][$key]; 586 } 587 else 588 $qualifiedName = $key; 589 590 // Filter classes 591 if ( $qualifiedName == 'class' ) 592 { 593 $classesList = $this->XMLSchema->getClassesList( $element->nodeName ); 594 if ( !in_array( $value, $classesList ) ) 595 { 596 $this->isInputValid = false; 597 if ( $this->errorLevel >= 2 ) 598 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Class '%1' is not allowed for element <%2> (check content.ini).", false, array( $value, $element->nodeName ) ); 599 continue; 600 } 601 } 602 603 // Create attribute nodes 604 if ( $qualifiedName ) 605 { 606 if ( strpos( $qualifiedName, ':' ) ) 607 { 608 list( $prefix, $name ) = explode( ':', $qualifiedName ); 609 if ( isset( $this->Namespaces[$prefix] ) ) 610 { 611 $URI = $this->Namespaces[$prefix]; 612 $element->setAttributeNS( $URI, $qualifiedName, $value ); 613 } 614 else 615 eZDebug::writeWarning( "No namespace defined for prefix '$prefix'.", 'eZXML input parser' ); 616 } 617 else 618 { 619 $element->setAttribute( $qualifiedName, $value ); 620 } 621 } 622 } 623 624 // Check for required attrs are present 625 if ( isset( $this->OutputTags[$element->nodeName]['requiredInputAttributes'] ) ) 626 { 627 foreach( $this->OutputTags[$element->nodeName]['requiredInputAttributes'] as $reqAttrName ) 628 { 629 $presented = false; 630 foreach( $attributes as $key => $value ) 631 { 632 if ( $key == $reqAttrName ) 633 { 634 $presented = true; 635 break; 636 } 637 } 638 if ( !$presented ) 639 { 640 $this->isInputValid = false; 641 if ( $this->errorLevel >= 2 ) 642 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Required attribute '%1' is not presented in tag <%2>.", 643 false, array( $reqAttrName, $element->nodeName ) ); 644 } 645 } 646 } 647 } 648 649 function washText( $textContent ) 650 { 651 $textContent = $this->entitiesDecode( $textContent ); 652 653 if ( !$this->AllowNumericEntities ) 654 $textContent = $this->convertNumericEntities( $textContent ); 655 656 if ( !$this->AllowMultipleSpaces ) 657 $textContent = preg_replace( "/ {2,}/", " ", $textContent ); 658 659 return $textContent; 660 } 661 662 function entitiesDecode( $text ) 663 { 664 //$text = str_replace( "&", "&", $text ); 665 $text = str_replace( "'", "'", $text ); 666 667 $text = str_replace( ">", ">", $text ); 668 $text = str_replace( "<", "<", $text ); 669 $text = str_replace( "'", "'", $text ); 670 $text = str_replace( """, '"', $text ); 671 $text = str_replace( "&", "&", $text ); 672 $text = str_replace( " ", " ", $text ); 673 return $text; 674 } 675 676 function convertNumericEntities( $text ) 677 { 678 if ( strlen( $text ) < 4 ) 679 { 680 return $text; 681 } 682 // Convert other HTML entities to the current charset characters. 683 include_once ( 'lib/ezi18n/classes/eztextcodec.php' ); 684 $codec = eZTextCodec::instance( 'unicode', false ); 685 $pos = 0; 686 $domString = ""; 687 while ( $pos < strlen( $text ) - 1 ) 688 { 689 $startPos = $pos; 690 while( !( $text[$pos] == '&' && $text[$pos + 1] == '#' ) && $pos < strlen( $text ) - 1 ) 691 $pos++; 692 693 $domString .= substr( $text, $startPos, $pos - $startPos ); 694 695 if ( $pos < strlen( $text ) - 1 ) 696 { 697 $endPos = strpos( $text, ";", $pos + 2 ); 698 if ( $endPos === false ) 699 { 700 $convertedText .= '&#'; 701 $pos += 2; 702 continue; 703 } 704 705 $code = substr( $text, $pos + 2, $endPos - ( $pos + 2 ) ); 706 $char = $codec->convertString( array( $code ) ); 707 708 $pos = $endPos + 1; 709 $domString .= $char; 710 } 711 else 712 { 713 $domString .= substr( $text, $pos, 2 ); 714 } 715 } 716 return $domString; 717 } 718 719 /*function getClassesList() 720 { 721 $ini =& eZINI::instance( 'content.ini' ); 722 foreach( array_keys( $this->OutputTags ) as $tagName ) 723 { 724 if ( $ini->hasVariable( $tagName, 'AvailableClasses' ) ) 725 { 726 $avail = $ini->variable( $tagName, 'AvailableClasses' ); 727 if ( is_array( $avail ) && count( $avail ) ) 728 $this->OutputTags[$tagName]['classesList'] = $avail; 729 else 730 $this->OutputTags[$tagName]['classesList'] = array(); 731 } 732 else 733 $this->OutputTags[$tagName]['classesList'] = array(); 734 } 735 }*/ 736 737 function wordMatchSupport( $newTagName, &$attributes, $attributeString ) 738 { 739 $ini =& eZINI::instance( 'wordmatch.ini' ); 740 if ( $ini->hasVariable( $newTagName, 'MatchString' ) ) 741 { 742 $matchArray = $ini->variable( $newTagName, 'MatchString' ); 743 if ( $matchArray ) 744 { 745 foreach ( array_keys( $matchArray ) as $key ) 746 { 747 $matchString = $matchArray[$key]; 748 if ( preg_match( "/$matchString/i", $attributeString ) ) 749 { 750 $attributes['class'] = $key; 751 unset( $attributes['style'] ); 752 } 753 } 754 } 755 } 756 } 757 758 759 /* 760 Pass 2: Process the tree, run handlers, rebuild and validate. 761 */ 762 763 function performPass2() 764 { 765 $tmp = null; 766 767 //php5 TODO: $this->Document->documentElement; 768 $this->processSubtree( $this->Document->Root, $tmp ); 769 } 770 771 // main recursive function for pass 2 772 773 function &processSubtree( &$element, &$lastHandlerResult ) 774 { 775 $ret = null; 776 $tmp = null; 777 778 //eZDOMNode::writeDebugStr( $element, '$element' ); 779 //eZDOMNode::writeDebugStr( $this->Document->Root, 'root' ); 780 781 // Call "Init handler" 782 $this->callOutputHandler( 'initHandler', $element, $tmp ); 783 784 // Process children 785 if ( $element->hasChildNodes() ) 786 { 787 // Make another copy of children to save primary structure 788 // php5 TODO: childNodes->item(), childNodes->length() 789 $childrenCount = count( $element->Children ); 790 $children = array(); 791 foreach( array_keys( $element->Children ) as $child_key ) 792 { 793 $children[] =& $element->Children[$child_key]; 794 } 795 $lastResult = null; 796 $newElements = array(); 797 for( $i = 0; $i < $childrenCount; $i++ ) 798 { 799 $childReturn =& $this->processSubtree( $children[$i], $lastResult ); 800 if ( isset( $childReturn['result'] ) ) 801 $lastResult =& $childReturn['result']; 802 else 803 unset( $lastResult ); 804 805 if ( isset( $childReturn['new_elements'] ) ) 806 $newElements = array_merge( $newElements, $childReturn['new_elements'] ); 807 808 unset( $childReturn ); 809 if ( $this->quitIfInvalid && !$this->isInputValid ) 810 { 811 return $ret; 812 } 813 } 814 815 // process elements created in children handlers 816 $this->processNewElements( $newElements ); 817 } 818 819 // Call "Structure handler" 820 $ret =& $this->callOutputHandler( 'structHandler', $element, $lastHandlerResult ); 821 822 // Process by schema and fix tree 823 if ( !$this->processElementBySchema( $element ) ) 824 { 825 return $ret; 826 } 827 828 $tmp = null; 829 // Call "Publish handler" 830 $this->callOutputHandler( 'publishHandler', $element, $tmp ); 831 832 // Process attributes according to the schema 833 if( $element->hasAttributes() ) 834 { 835 if ( !$this->XMLSchema->hasAttributes( $element ) ) 836 { 837 $element->removeAttributes(); 838 } 839 else 840 { 841 $this->processAttributesBySchema( $element ); 842 } 843 } 844 return $ret; 845 } 846 /* 847 Helper functions for pass 2 848 */ 849 850 // Check element's schema and fix subtree if needed 851 function processElementBySchema( &$element, $verbose = true ) 852 { 853 $parent =& $element->parentNode; 854 if ( $parent ) 855 { 856 // If this is a foreign element, remove it 857 if ( !$this->XMLSchema->exists( $element ) ) 858 { 859 if ( $element->nodeName == 'custom' ) 860 { 861 $this->isInputValid = false; 862 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Custom tag '%1' is not allowed.", 863 false, array( $element->getAttribute( 'name' ) ) ); 864 } 865 $parent->removeChild( $element ); 866 return false; 867 } 868 869 // Delete if children required and no children 870 if ( ( $this->XMLSchema->childrenRequired( $element ) || $element->getAttribute( 'children_required' ) ) 871 && !$element->hasChildNodes() ) 872 { 873 $parent->removeChild( $element ); 874 return false; 875 } 876 877 // Check schema and remove wrong elements 878 $schemaCheckResult = $this->XMLSchema->check( $parent, $element ); 879 if ( !$schemaCheckResult ) 880 { 881 if ( $schemaCheckResult === false ) 882 { 883 // Remove indenting spaces 884 if ( $element->Type == EZ_XML_NODE_TEXT && !trim( $element->content() ) ) 885 { 886 $parent->removeChild( $element ); 887 return false; 888 } 889 890 $this->isInputValid = false; 891 if ( $verbose && $this->errorLevel >= 1 ) 892 { 893 $elementName = $element->nodeName == '#text' ? $element->nodeName : '<' . $element->nodeName . '>'; 894 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "%1 is not allowed to be a child of <%2>.", 895 false, array( $elementName, $parent->nodeName ) ); 896 } 897 } 898 $this->fixSubtree( $element, $element ); 899 return false; 900 } 901 } 902 // TODO: break processing of any node that doesn't have parent 903 // and is not a root node. 904 elseif ( $element->nodeName != 'section' ) 905 { 906 return false; 907 } 908 return true; 909 } 910 911 // Remove only nodes that don't match schema (recursively) 912 function fixSubtree( &$element, &$mainChild ) 913 { 914 $parent =& $element->parentNode; 915 $mainParent =& $mainChild->parentNode; 916 if ( $element->hasChildNodes() ) 917 { 918 foreach( array_keys( $element->Children ) as $child_key ) 919 { 920 $child =& $element->Children[$child_key]; 921 922 $element->removeChild( $child ); 923 // php5 TODO: use child returned by insertBefore (php dom manual) 924 $mainParent->insertBefore( $child, $mainChild ); 925 926 if ( !$this->XMLSchema->check( $mainParent, $child ) ) 927 $this->fixSubtree( $child, $mainChild ); 928 } 929 } 930 $parent->removeChild( $element ); 931 } 932 933 function processAttributesBySchema( &$element, $verbose = true ) 934 { 935 // Remove attributes that don't match schema 936 $schemaAttributes = $this->XMLSchema->attributes( $element ); 937 if ( $this->eZPublishVersion >= 3.9 ) 938 { 939 $schemaCustomAttributes = $this->XMLSchema->customAttributes( $element ); 940 } 941 942 $attributes = $element->attributes(); 943 foreach( $attributes as $attr ) 944 { 945 $allowed = false; 946 $removeAttr = false; 947 948 // php5 TODO: small letters 949 if ( $attr->Prefix ) 950 $fullName = $attr->Prefix . ':' . $attr->LocalName; 951 else 952 $fullName = $attr->LocalName; 953 954 if ( $this->eZPublishVersion >= 3.9 ) 955 { 956 // check for allowed custom attributes (3.9) 957 if ( $attr->Prefix == 'custom' && in_array( $attr->LocalName, $schemaCustomAttributes ) ) 958 { 959 $allowed = true; 960 } 961 else 962 { 963 if ( in_array( $fullName, $schemaAttributes ) ) 964 { 965 $allowed = true; 966 } 967 elseif ( in_array( $fullName, $schemaCustomAttributes ) ) 968 { 969 // add 'custom' prefix if it is not given 970 $allowed = true; 971 $removeAttr = true; 972 $element->setAttributeNS( $this->Namespaces['custom'], 'custom:' . $fullName, $attr->content() ); 973 } 974 } 975 } 976 else 977 { 978 if ( $attr->Prefix == 'custom' || 979 in_array( $fullName, $schemaAttributes ) ) 980 { 981 $allowed = true; 982 } 983 } 984 985 if ( !$allowed ) 986 { 987 $removeAttr = true; 988 $this->isInputValid = false; 989 if ( $verbose && $this->errorLevel >= 1 ) 990 { 991 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Attribute '%1' is not allowed in <%2> element.", 992 false, array( $fullName, $element->nodeName ) ); 993 } 994 } 995 elseif ( $this->removeDefaultAttrs ) 996 { 997 // Remove attributes having default values 998 $default = $this->XMLSchema->attrDefaultValue( $element->nodeName, $fullName ); 999 if ( $attr->Content == $default ) 1000 { 1001 $removeAttr = true; 1002 } 1003 } 1004 1005 if ( $removeAttr ) 1006 { 1007 if ( $attr->Prefix ) 1008 $element->removeAttributeNS( $attr->NamespaceURI, $attr->LocalName ); 1009 else 1010 $element->removeAttribute( $attr->nodeName ); 1011 } 1012 } 1013 } 1014 1015 function &callInputHandler( $handlerName, $tagName, &$attributes ) 1016 { 1017 $result = null; 1018 $thisInputTag =& $this->InputTags[$tagName]; 1019 if ( isset( $thisInputTag[$handlerName] ) ) 1020 { 1021 if ( is_callable( array( $this, $thisInputTag[$handlerName] ) ) ) 1022 eval( '$result =& $this->' . $thisInputTag[$handlerName] . '( $tagName, $attributes );' ); 1023 else 1024 eZDebug::writeWarning( "'$handlerName' input handler for tag <$tagName> doesn't exist: '" . $thisInputTag[$handlerName] . "'.", 'eZXML input parser' ); 1025 } 1026 return $result; 1027 } 1028 1029 function &callOutputHandler( $handlerName, &$element, &$params ) 1030 { 1031 $result = null; 1032 $thisOutputTag =& $this->OutputTags[$element->nodeName]; 1033 if ( isset( $thisOutputTag[$handlerName] ) ) 1034 { 1035 if ( is_callable( array( $this, $thisOutputTag[$handlerName] ) ) ) 1036 eval( '$result =& $this->' . $thisOutputTag[$handlerName] . '( $element, $params );' ); 1037 else 1038 eZDebug::writeWarning( "'$handlerName' output handler for tag <$element->nodeName> doesn't exist: '" . $thisOutputTag[$handlerName] . "'.", 'eZXML input parser' ); 1039 } 1040 return $result; 1041 } 1042 1043 // Creates new element and adds it to array for further post-processing. 1044 // Use this function if you need to process newly created element (check it by schema 1045 // and call 'structure' and 'publish' handlers) 1046 function &createAndPublishElement( $elementName, &$ret ) 1047 { 1048 $element =& $this->Document->createElement( $elementName ); 1049 //$element->setAttribute( 'ezparser-new-element', 'true' ); 1050 1051 if ( !isset( $ret['new_elements'] ) ) 1052 $ret['new_elements'] = array(); 1053 1054 $ret['new_elements'][] =& $element; 1055 return $element; 1056 } 1057 1058 function processNewElements( &$createdElements ) 1059 { 1060 // Call publish handlers for newly created elements 1061 foreach( array_keys( $createdElements ) as $key ) 1062 { 1063 $element =& $createdElements[$key]; 1064 1065 $tmp = null; 1066 // Call "Structure handler" 1067 $this->callOutputHandler( 'structHandler', $element, $tmp ); 1068 1069 if ( !$this->processElementBySchema( $element ) ) 1070 continue; 1071 1072 $tmp2 = null; 1073 // Call "Publish handler" 1074 $this->callOutputHandler( 'publishHandler', $element, $tmp2 ); 1075 //$element->removeAttribute( 'ezparser-new-element' ); 1076 } 1077 } 1078 1079 function getMessages() 1080 { 1081 return $this->Messages; 1082 } 1083 1084 function isValid() 1085 { 1086 return $this->isInputValid; 1087 } 1088 1089 var $DOMDocumentClass = 'eZDOMDocument'; 1090 1091 var $XMLSchema; 1092 var $Document; 1093 var $Messages = array(); 1094 var $eZPublishVersion; 1095 1096 var $ParentStack = array(); 1097 1098 var $errorLevel = 0; 1099 1100 var $isInputValid = true; 1101 var $quitIfInvalid = false; 1102 1103 // options that depend on settings 1104 var $TrimSpaces = true; 1105 var $AllowMultipleSpaces = false; 1106 var $AllowNumericEntities = false; 1107 1108 // options that depend on parameters passed 1109 var $parseLineBreaks = false; 1110 var $removeDefaultAttrs = false; 1111 1112 var $createdElements = array(); 1113 } 1114 ?>
titre
Description
Corps
titre
Description
Corps
titre
Description
Corps
titre
Corps
| Généré le : Sat Feb 24 10:30:04 2007 | par Balluche grâce à PHPXref 0.7 |