| [ 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 } 270 } 271 return $ret; 272 } 273 274 // The main recursive function for pass 1 275 276 function parseTag( &$data, &$pos, &$parent ) 277 { 278 // Find tag, determine it's type, name and attributes. 279 $initialPos = $pos; 280 281 /*if ( $this->trimSpaces ) 282 { 283 while( $pos < strlen( $data ) && $data[$pos] == ' ' ) $pos++; 284 }*/ 285 286 if ( $pos >= strlen( $data ) ) 287 { 288 return true; 289 } 290 $tagBeginPos = strpos( $data, '<', $pos ); 291 292 if ( $this->parseLineBreaks ) 293 { 294 // Regard line break as a start tag position 295 $lineBreakPos = strpos( $data, "\n", $pos ); 296 if ( $lineBreakPos !== false ) 297 { 298 if ( $tagBeginPos === false ) 299 $tagBeginPos = $lineBreakPos; 300 else 301 $tagBeginPos = min( $tagBeginPos, $lineBreakPos ); 302 } 303 } 304 305 $tagName = ''; 306 $attributes = null; 307 // If it doesn't begin with '<' then its a text node. 308 if ( $tagBeginPos != $pos || $tagBeginPos === false ) 309 { 310 $pos = $initialPos; 311 $tagName = $newTagName = '#text'; 312 $noChildren = true; 313 314 if ( !$tagBeginPos ) 315 $tagBeginPos = strlen( $data ); 316 317 $textContent = substr( $data, $pos, $tagBeginPos - $pos ); 318 319 $textContent = $this->washText( $textContent ); 320 321 $pos = $tagBeginPos; 322 if ( $textContent === '' ) 323 return false; 324 } 325 // Process closing tag. 326 elseif ( $data[$tagBeginPos] == '<' && $tagBeginPos + 1 < strlen( $data ) && 327 $data[$tagBeginPos + 1] == '/' ) 328 { 329 $tagEndPos = strpos( $data, '>', $tagBeginPos + 1 ); 330 if ( $tagEndPos === false ) 331 { 332 $pos = $tagBeginPos + 1; 333 334 $this->isInputValid = false; 335 if ( $this->errorLevel >= 2 ) 336 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong closing tag' ); 337 return false; 338 } 339 340 $pos = $tagEndPos + 1; 341 $closedTagName = strtolower( trim( substr( $data, $tagBeginPos + 2, $tagEndPos - $tagBeginPos - 2 ) ) ); 342 343 // Find matching tag in ParentStack array 344 $firstLoop = true; 345 for( $i = count( $this->ParentStack ) - 1; $i >= 0; $i-- ) 346 { 347 $parentNames = $this->ParentStack[$i]; 348 if ( $parentNames[0] == $closedTagName ) 349 { 350 array_pop( $this->ParentStack ); 351 if ( !$firstLoop ) 352 { 353 $pos = $tagBeginPos; 354 return true; 355 } 356 // If newTagName was '' we don't break children loop 357 elseif ( $parentNames[1] !== '' ) 358 return true; 359 else 360 return false; 361 } 362 $firstLoop = false; 363 } 364 365 $this->isInputValid = false; 366 if ( $this->errorLevel >= 2 ) 367 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong closing tag : </%1>.', false, array( $closedTagName ) ); 368 369 return false; 370 } 371 // Insert <br/> instead of linebreaks 372 elseif ( $this->parseLineBreaks && $data[$tagBeginPos] == "\n" ) 373 { 374 $newTagName = 'br'; 375 $noChildren = true; 376 $pos = $tagBeginPos + 1; 377 } 378 // Regular tag: get tag's name and attributes. 379 else 380 { 381 $tagEndPos = strpos( $data, '>', $tagBeginPos ); 382 if ( $tagEndPos === false ) 383 { 384 $pos = $tagBeginPos + 1; 385 386 $this->isInputValid = false; 387 if ( $this->errorLevel >= 2 ) 388 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', 'Wrong opening tag' ); 389 return false; 390 } 391 392 $pos = $tagEndPos + 1; 393 $tagString = substr( $data, $tagBeginPos + 1, $tagEndPos - $tagBeginPos - 1 ); 394 // Check for final backslash 395 $noChildren = substr( $tagString, -1, 1 ) == '/' ? true : false; 396 // Remove final backslash and spaces 397 $tagString = preg_replace( "/\s*\/$/", "", $tagString ); 398 399 $firstSpacePos = strpos( $tagString, ' ' ); 400 if ( $firstSpacePos === false ) 401 { 402 $tagName = strtolower( trim( $tagString ) ); 403 $attributeString = ''; 404 } 405 else 406 { 407 $tagName = strtolower( substr( $tagString, 0, $firstSpacePos ) ); 408 $attributeString = substr( $tagString, $firstSpacePos + 1 ); 409 $attributeString = trim( $attributeString ); 410 // Parse attribute string 411 if ( $attributeString ) 412 $attributes = $this->parseAttributes( $attributeString ); 413 } 414 415 // Determine tag's name 416 if ( isset( $this->InputTags[$tagName] ) ) 417 { 418 $thisInputTag =& $this->InputTags[$tagName]; 419 420 if ( isset( $thisInputTag['name'] ) ) 421 $newTagName = $thisInputTag['name']; 422 else 423 $newTagName =& $this->callInputHandler( 'nameHandler', $tagName, $attributes ); 424 } 425 else 426 { 427 if ( $this->XMLSchema->exists( $tagName ) ) 428 { 429 $newTagName = $tagName; 430 } 431 else 432 { 433 $this->isInputValid = false; 434 if ( $this->errorLevel >= 2 ) 435 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Unknown tag: <%1>.", false, array( $tagName ) ); 436 437 return false; 438 } 439 } 440 441 // Check 'noChildren' property 442 if ( isset( $thisInputTag['noChildren'] ) ) 443 $noChildren = true; 444 445 $thisOutputTag =& $this->OutputTags[$newTagName]; 446 447 // Implementation of 'autoCloseOn' rule ( Handling of unclosed tags, ex.: <p>, <li> ) 448 if ( isset( $thisOutputTag['autoCloseOn'] ) && 449 $parent && 450 $parent->parentNode && 451 in_array( $parent->nodeName, $thisOutputTag['autoCloseOn'] ) ) 452 { 453 // Wrong nesting: auto-close parent and try to re-parse this tag at higher level 454 array_pop( $this->ParentStack ); 455 $pos = $tagBeginPos; 456 return true; 457 } 458 459 // Append to parent stack 460 if ( !$noChildren && $newTagName !== false ) 461 { 462 $this->ParentStack[] = array( $tagName, $newTagName, $attributeString ); 463 } 464 465 if ( !$newTagName ) 466 { 467 if ( $newTagName === false ) 468 { 469 $this->isInputValid = false; 470 if ( $this->errorLevel >= 2 ) 471 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Can't convert tag's name: <%1>.", false, array( $tagName ) ); 472 } 473 return false; 474 // TODO: return it before and don't append to ParentStack? 475 // (need to skip processing closing tag on empty tagname) 476 // LATER: no.. this is not very good for data consistance 477 } 478 479 // wordmatch.ini support 480 if ( $attributeString ) 481 { 482 $this->wordMatchSupport( $newTagName, $attributes, $attributeString ); 483 } 484 } 485 486 // Create text or normal node. 487 if ( $newTagName == '#text' ) 488 $element = $this->Document->createTextNode( $textContent ); 489 else 490 $element =& $this->Document->createElement( $newTagName ); 491 492 if ( $attributes ) 493 { 494 $this->setAttributes( $element, $attributes ); 495 } 496 497 // Append element as a child or set it as root if there is no parent. 498 if ( $parent ) 499 $parent->appendChild( $element ); 500 else 501 $this->Document->appendChild( $element ); 502 // php5 TODO : $this->Document->documentElement->appendChild( $element ); 503 504 $params = array(); 505 $params[] =& $data; 506 $params[] =& $pos; 507 $params[] =& $tagBeginPos; 508 $result =& $this->callOutputHandler( 'parsingHandler', $element, $params ); 509 510 if ( $result === false ) 511 { 512 // This tag is already parsed in handler 513 if ( !$noChildren ) 514 array_pop( $this->ParentStack ); 515 return false; 516 } 517 518 if ( $this->quitIfInvalid && !$this->isInputValid ) 519 return false; 520 521 // Process children 522 if ( !$noChildren ) 523 { 524 do 525 { 526 $parseResult = $this->parseTag( $data, $pos, $element ); 527 528 if ( $this->quitIfInvalid && !$this->isInputValid ) 529 return false; 530 } 531 while( $parseResult !== true ); 532 } 533 534 return false; 535 } 536 537 /* 538 Helper functions for pass 1 539 */ 540 541 function parseAttributes( $attributeString ) 542 { 543 // Convert single quotes to double quotes 544 $attributeString = preg_replace( "/ +([a-zA-Z0-9:-_#\-]+) *\='(.*?)'/e", "' \\1'.'=\"'.'\\2'.'\"'", ' ' . $attributeString ); 545 546 // Convert no quotes to double quotes and remove extra spaces 547 $attributeString = preg_replace( "/ +([a-zA-Z0-9:-_#\-]+) *\= *([^\s'\"]+)/e", "' \\1'.'=\"'.'\\2'.'\" '", $attributeString ); 548 549 // Split by quotes followed by spaces 550 $attributeArray = preg_split( "#(?<=\") +#", $attributeString ); 551 552 $attributes = array(); 553 foreach( $attributeArray as $attrStr ) 554 { 555 if ( !$attrStr || strlen( $attrStr ) < 4 ) 556 continue; 557 558 list( $attrName, $attrValue ) = split( '="', $attrStr ); 559 560 $attrName = strtolower( trim( $attrName ) ); 561 if ( !$attrName ) 562 continue; 563 564 $attrValue = substr( $attrValue, 0, -1 ); 565 if ( $attrValue === '' || $attrValue === false ) 566 continue; 567 568 $attributes[$attrName] = $attrValue; 569 } 570 571 return $attributes; 572 } 573 574 function setAttributes( &$element, $attributes ) 575 { 576 $thisOutputTag =& $this->OutputTags[$element->nodeName]; 577 578 foreach( $attributes as $key => $value ) 579 { 580 // Convert attribute names 581 if ( isset( $thisOutputTag['attributes'] ) && 582 isset( $thisOutputTag['attributes'][$key] ) ) 583 { 584 $qualifiedName = $thisOutputTag['attributes'][$key]; 585 } 586 else 587 $qualifiedName = $key; 588 589 // Filter classes 590 if ( $qualifiedName == 'class' ) 591 { 592 $classesList = $this->XMLSchema->getClassesList( $element->nodeName ); 593 if ( !in_array( $value, $classesList ) ) 594 { 595 $this->isInputValid = false; 596 if ( $this->errorLevel >= 2 ) 597 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Class '%1' is not allowed for element <%2> (check content.ini).", false, array( $value, $element->nodeName ) ); 598 continue; 599 } 600 } 601 602 // Create attribute nodes 603 if ( $qualifiedName ) 604 { 605 if ( strpos( $qualifiedName, ':' ) ) 606 { 607 list( $prefix, $name ) = explode( ':', $qualifiedName ); 608 if ( isset( $this->Namespaces[$prefix] ) ) 609 { 610 $URI = $this->Namespaces[$prefix]; 611 $element->setAttributeNS( $URI, $qualifiedName, $value ); 612 } 613 else 614 eZDebug::writeWarning( "No namespace defined for prefix '$prefix'.", 'eZXML input parser' ); 615 } 616 else 617 { 618 $element->setAttribute( $qualifiedName, $value ); 619 } 620 } 621 } 622 623 // Check for required attrs are present 624 if ( isset( $this->OutputTags[$element->nodeName]['requiredInputAttributes'] ) ) 625 { 626 foreach( $this->OutputTags[$element->nodeName]['requiredInputAttributes'] as $reqAttrName ) 627 { 628 $presented = false; 629 foreach( $attributes as $key => $value ) 630 { 631 if ( $key == $reqAttrName ) 632 { 633 $presented = true; 634 break; 635 } 636 } 637 if ( !$presented ) 638 { 639 $this->isInputValid = false; 640 if ( $this->errorLevel >= 2 ) 641 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Required attribute '%1' is not presented in tag <%2>.", 642 false, array( $reqAttrName, $element->nodeName ) ); 643 } 644 } 645 } 646 } 647 648 function washText( $textContent ) 649 { 650 $textContent = $this->entitiesDecode( $textContent ); 651 652 if ( !$this->AllowNumericEntities ) 653 $textContent = $this->convertNumericEntities( $textContent ); 654 655 if ( !$this->AllowMultipleSpaces ) 656 $textContent = preg_replace( "/ {2,}/", " ", $textContent ); 657 658 return $textContent; 659 } 660 661 function entitiesDecode( $text ) 662 { 663 //$text = str_replace( "&", "&", $text ); 664 $text = str_replace( "'", "'", $text ); 665 666 $text = str_replace( ">", ">", $text ); 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 return $text; 673 } 674 675 function convertNumericEntities( $text ) 676 { 677 if ( strlen( $text ) < 4 ) 678 { 679 return $text; 680 } 681 // Convert other HTML entities to the current charset characters. 682 include_once ( 'lib/ezi18n/classes/eztextcodec.php' ); 683 $codec = eZTextCodec::instance( 'unicode', false ); 684 $pos = 0; 685 $domString = ""; 686 while ( $pos < strlen( $text ) - 1 ) 687 { 688 $startPos = $pos; 689 while( !( $text[$pos] == '&' && $text[$pos + 1] == '#' ) && $pos < strlen( $text ) - 1 ) 690 $pos++; 691 692 $domString .= substr( $text, $startPos, $pos - $startPos ); 693 694 if ( $pos < strlen( $text ) - 1 ) 695 { 696 $endPos = strpos( $text, ";", $pos + 2 ); 697 if ( $endPos === false ) 698 { 699 $convertedText .= '&#'; 700 $pos += 2; 701 continue; 702 } 703 704 $code = substr( $text, $pos + 2, $endPos - ( $pos + 2 ) ); 705 $char = $codec->convertString( array( $code ) ); 706 707 $pos = $endPos + 1; 708 $domString .= $char; 709 } 710 else 711 { 712 $domString .= substr( $text, $pos, 2 ); 713 } 714 } 715 return $domString; 716 } 717 718 /*function getClassesList() 719 { 720 $ini =& eZINI::instance( 'content.ini' ); 721 foreach( array_keys( $this->OutputTags ) as $tagName ) 722 { 723 if ( $ini->hasVariable( $tagName, 'AvailableClasses' ) ) 724 { 725 $avail = $ini->variable( $tagName, 'AvailableClasses' ); 726 if ( is_array( $avail ) && count( $avail ) ) 727 $this->OutputTags[$tagName]['classesList'] = $avail; 728 else 729 $this->OutputTags[$tagName]['classesList'] = array(); 730 } 731 else 732 $this->OutputTags[$tagName]['classesList'] = array(); 733 } 734 }*/ 735 736 function wordMatchSupport( $newTagName, &$attributes, $attributeString ) 737 { 738 $ini =& eZINI::instance( 'wordmatch.ini' ); 739 if ( $ini->hasVariable( $newTagName, 'MatchString' ) ) 740 { 741 $matchArray = $ini->variable( $newTagName, 'MatchString' ); 742 if ( $matchArray ) 743 { 744 foreach ( array_keys( $matchArray ) as $key ) 745 { 746 $matchString = $matchArray[$key]; 747 if ( preg_match( "/$matchString/i", $attributeString ) ) 748 { 749 $attributes['class'] = $key; 750 unset( $attributes['style'] ); 751 } 752 } 753 } 754 } 755 } 756 757 758 /* 759 Pass 2: Process the tree, run handlers, rebuild and validate. 760 */ 761 762 function performPass2() 763 { 764 $tmp = null; 765 766 //php5 TODO: $this->Document->documentElement; 767 $this->processSubtree( $this->Document->Root, $tmp ); 768 } 769 770 // main recursive function for pass 2 771 772 function &processSubtree( &$element, &$lastHandlerResult ) 773 { 774 $ret = null; 775 $tmp = null; 776 777 //eZDOMNode::writeDebugStr( $element, '$element' ); 778 //eZDOMNode::writeDebugStr( $this->Document->Root, 'root' ); 779 780 // Call "Init handler" 781 $this->callOutputHandler( 'initHandler', $element, $tmp ); 782 783 // Process children 784 if ( $element->hasChildNodes() ) 785 { 786 // Make another copy of children to save primary structure 787 // php5 TODO: childNodes->item(), childNodes->length() 788 $childrenCount = count( $element->Children ); 789 $children = array(); 790 foreach( array_keys( $element->Children ) as $child_key ) 791 { 792 $children[] =& $element->Children[$child_key]; 793 } 794 $lastResult = null; 795 $newElements = array(); 796 for( $i = 0; $i < $childrenCount; $i++ ) 797 { 798 $childReturn =& $this->processSubtree( $children[$i], $lastResult ); 799 if ( isset( $childReturn['result'] ) ) 800 $lastResult =& $childReturn['result']; 801 else 802 unset( $lastResult ); 803 804 if ( isset( $childReturn['new_elements'] ) ) 805 $newElements = array_merge( $newElements, $childReturn['new_elements'] ); 806 807 unset( $childReturn ); 808 if ( $this->quitIfInvalid && !$this->isInputValid ) 809 { 810 return $ret; 811 } 812 } 813 814 // process elements created in children handlers 815 $this->processNewElements( $newElements ); 816 } 817 818 // Call "Structure handler" 819 $ret =& $this->callOutputHandler( 'structHandler', $element, $lastHandlerResult ); 820 821 // Process by schema and fix tree 822 if ( !$this->processElementBySchema( $element ) ) 823 { 824 return $ret; 825 } 826 827 $tmp = null; 828 // Call "Publish handler" 829 $this->callOutputHandler( 'publishHandler', $element, $tmp ); 830 831 // Process attributes according to the schema 832 if( $element->hasAttributes() ) 833 { 834 if ( !$this->XMLSchema->hasAttributes( $element ) ) 835 { 836 $element->removeAttributes(); 837 } 838 else 839 { 840 $this->processAttributesBySchema( $element ); 841 } 842 } 843 return $ret; 844 } 845 /* 846 Helper functions for pass 2 847 */ 848 849 // Check element's schema and fix subtree if needed 850 function processElementBySchema( &$element, $verbose = true ) 851 { 852 $parent =& $element->parentNode; 853 if ( $parent ) 854 { 855 // If this is a foreign element, remove it 856 if ( !$this->XMLSchema->exists( $element ) ) 857 { 858 if ( $element->nodeName == 'custom' ) 859 { 860 $this->isInputValid = false; 861 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Custom tag '%1' is not allowed.", 862 false, array( $element->getAttribute( 'name' ) ) ); 863 } 864 $parent->removeChild( $element ); 865 return false; 866 } 867 868 // Delete if children required and no children 869 if ( ( $this->XMLSchema->childrenRequired( $element ) || $element->getAttribute( 'children_required' ) ) 870 && !$element->hasChildNodes() ) 871 { 872 $this->isInputValid = false; 873 if ( $verbose && $this->errorLevel >= 1 ) 874 { 875 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "<%1> tag can't be empty.", 876 false, array( $element->nodeName ) ); 877 } 878 $parent->removeChild( $element ); 879 return false; 880 } 881 882 // Check schema and remove wrong elements 883 $schemaCheckResult = $this->XMLSchema->check( $parent, $element ); 884 if ( !$schemaCheckResult ) 885 { 886 if ( $schemaCheckResult === false ) 887 { 888 // Remove indenting spaces 889 if ( $element->Type == EZ_XML_NODE_TEXT && !trim( $element->content() ) ) 890 { 891 $parent->removeChild( $element ); 892 return false; 893 } 894 895 $this->isInputValid = false; 896 if ( $verbose && $this->errorLevel >= 1 ) 897 { 898 $elementName = $element->nodeName == '#text' ? $element->nodeName : '<' . $element->nodeName . '>'; 899 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "%1 is not allowed to be a child of <%2>.", 900 false, array( $elementName, $parent->nodeName ) ); 901 } 902 } 903 $this->fixSubtree( $element, $element ); 904 return false; 905 } 906 } 907 // TODO: break processing of any node that doesn't have parent 908 // and is not a root node. 909 elseif ( $element->nodeName != 'section' ) 910 { 911 return false; 912 } 913 return true; 914 } 915 916 // Remove only nodes that don't match schema (recursively) 917 function fixSubtree( &$element, &$mainChild ) 918 { 919 $parent =& $element->parentNode; 920 $mainParent =& $mainChild->parentNode; 921 if ( $element->hasChildNodes() ) 922 { 923 foreach( array_keys( $element->Children ) as $child_key ) 924 { 925 $child =& $element->Children[$child_key]; 926 927 $element->removeChild( $child ); 928 // php5 TODO: use child returned by insertBefore (php dom manual) 929 $mainParent->insertBefore( $child, $mainChild ); 930 931 if ( !$this->XMLSchema->check( $mainParent, $child ) ) 932 $this->fixSubtree( $child, $mainChild ); 933 } 934 } 935 $parent->removeChild( $element ); 936 } 937 938 function processAttributesBySchema( &$element, $verbose = true ) 939 { 940 // Remove attributes that don't match schema 941 $schemaAttributes = $this->XMLSchema->attributes( $element ); 942 if ( $this->eZPublishVersion >= 3.9 ) 943 { 944 $schemaCustomAttributes = $this->XMLSchema->customAttributes( $element ); 945 } 946 947 $attributes = $element->attributes(); 948 foreach( $attributes as $attr ) 949 { 950 $allowed = false; 951 $removeAttr = false; 952 953 // php5 TODO: small letters 954 if ( $attr->Prefix ) 955 $fullName = $attr->Prefix . ':' . $attr->LocalName; 956 else 957 $fullName = $attr->LocalName; 958 959 if ( $this->eZPublishVersion >= 3.9 ) 960 { 961 // check for allowed custom attributes (3.9) 962 if ( $attr->Prefix == 'custom' && in_array( $attr->LocalName, $schemaCustomAttributes ) ) 963 { 964 $allowed = true; 965 } 966 else 967 { 968 if ( in_array( $fullName, $schemaAttributes ) ) 969 { 970 $allowed = true; 971 } 972 elseif ( in_array( $fullName, $schemaCustomAttributes ) ) 973 { 974 // add 'custom' prefix if it is not given 975 $allowed = true; 976 $removeAttr = true; 977 $element->setAttributeNS( $this->Namespaces['custom'], 'custom:' . $fullName, $attr->content() ); 978 } 979 } 980 } 981 else 982 { 983 if ( $attr->Prefix == 'custom' || 984 in_array( $fullName, $schemaAttributes ) ) 985 { 986 $allowed = true; 987 } 988 } 989 990 if ( !$allowed ) 991 { 992 $removeAttr = true; 993 $this->isInputValid = false; 994 if ( $verbose && $this->errorLevel >= 1 ) 995 { 996 $this->Messages[] = ezi18n( 'kernel/classes/datatypes/ezxmltext', "Attribute '%1' is not allowed in <%2> element.", 997 false, array( $fullName, $element->nodeName ) ); 998 } 999 } 1000 elseif ( $this->removeDefaultAttrs ) 1001 { 1002 // Remove attributes having default values 1003 $default = $this->XMLSchema->attrDefaultValue( $element->nodeName, $fullName ); 1004 if ( $attr->Content == $default ) 1005 { 1006 $removeAttr = true; 1007 } 1008 } 1009 1010 if ( $removeAttr ) 1011 { 1012 if ( $attr->Prefix ) 1013 $element->removeAttributeNS( $attr->NamespaceURI, $attr->LocalName ); 1014 else 1015 $element->removeAttribute( $attr->nodeName ); 1016 } 1017 } 1018 } 1019 1020 function &callInputHandler( $handlerName, $tagName, &$attributes ) 1021 { 1022 $result = null; 1023 $thisInputTag =& $this->InputTags[$tagName]; 1024 if ( isset( $thisInputTag[$handlerName] ) ) 1025 { 1026 if ( is_callable( array( $this, $thisInputTag[$handlerName] ) ) ) 1027 eval( '$result =& $this->' . $thisInputTag[$handlerName] . '( $tagName, $attributes );' ); 1028 else 1029 eZDebug::writeWarning( "'$handlerName' input handler for tag <$tagName> doesn't exist: '" . $thisInputTag[$handlerName] . "'.", 'eZXML input parser' ); 1030 } 1031 return $result; 1032 } 1033 1034 function &callOutputHandler( $handlerName, &$element, &$params ) 1035 { 1036 $result = null; 1037 $thisOutputTag =& $this->OutputTags[$element->nodeName]; 1038 if ( isset( $thisOutputTag[$handlerName] ) ) 1039 { 1040 if ( is_callable( array( $this, $thisOutputTag[$handlerName] ) ) ) 1041 eval( '$result =& $this->' . $thisOutputTag[$handlerName] . '( $element, $params );' ); 1042 else 1043 eZDebug::writeWarning( "'$handlerName' output handler for tag <$element->nodeName> doesn't exist: '" . $thisOutputTag[$handlerName] . "'.", 'eZXML input parser' ); 1044 } 1045 return $result; 1046 } 1047 1048 // Creates new element and adds it to array for further post-processing. 1049 // Use this function if you need to process newly created element (check it by schema 1050 // and call 'structure' and 'publish' handlers) 1051 function &createAndPublishElement( $elementName, &$ret ) 1052 { 1053 $element =& $this->Document->createElement( $elementName ); 1054 //$element->setAttribute( 'ezparser-new-element', 'true' ); 1055 1056 if ( !isset( $ret['new_elements'] ) ) 1057 $ret['new_elements'] = array(); 1058 1059 $ret['new_elements'][] =& $element; 1060 return $element; 1061 } 1062 1063 function processNewElements( &$createdElements ) 1064 { 1065 // Call publish handlers for newly created elements 1066 foreach( array_keys( $createdElements ) as $key ) 1067 { 1068 $element =& $createdElements[$key]; 1069 1070 $tmp = null; 1071 // Call "Structure handler" 1072 $this->callOutputHandler( 'structHandler', $element, $tmp ); 1073 1074 if ( !$this->processElementBySchema( $element ) ) 1075 continue; 1076 1077 $tmp2 = null; 1078 // Call "Publish handler" 1079 $this->callOutputHandler( 'publishHandler', $element, $tmp2 ); 1080 //$element->removeAttribute( 'ezparser-new-element' ); 1081 } 1082 } 1083 1084 function getMessages() 1085 { 1086 return $this->Messages; 1087 } 1088 1089 function isValid() 1090 { 1091 return $this->isInputValid; 1092 } 1093 1094 var $DOMDocumentClass = 'eZDOMDocument'; 1095 1096 var $XMLSchema; 1097 var $Document; 1098 var $Messages = array(); 1099 var $eZPublishVersion; 1100 1101 var $ParentStack = array(); 1102 1103 var $errorLevel = 0; 1104 1105 var $isInputValid = true; 1106 var $quitIfInvalid = false; 1107 1108 // options that depend on settings 1109 var $TrimSpaces = true; 1110 var $AllowMultipleSpaces = false; 1111 var $AllowNumericEntities = false; 1112 1113 // options that depend on parameters passed 1114 var $parseLineBreaks = false; 1115 var $removeDefaultAttrs = false; 1116 1117 var $createdElements = array(); 1118 } 1119 ?>
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 |