[Sommaire] [Imprimer]

   1  #!/usr/bin/env php
   2  <?php
   3  //
   4  // Created on: <04-Feb-2003 11:02:34 amos>
   5  //
   6  // SOFTWARE NAME: eZ publish
   7  // SOFTWARE RELEASE: 3.9.0
   8  // BUILD VERSION: 17785
   9  // COPYRIGHT NOTICE: Copyright (C) 1999-2006 eZ systems AS
  10  // SOFTWARE LICENSE: GNU General Public License v2.0
  11  // NOTICE: >
  12  //   This program is free software; you can redistribute it and/or
  13  //   modify it under the terms of version 2.0  of the GNU General
  14  //   Public License as published by the Free Software Foundation.
  15  //
  16  //   This program is distributed in the hope that it will be useful,
  17  //   but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  //   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  //   GNU General Public License for more details.
  20  //
  21  //   You should have received a copy of version 2.0 of the GNU General
  22  //   Public License along with this program; if not, write to the Free
  23  //   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
  24  //   MA 02110-1301, USA.
  25  //
  26  //
  27  
  28  /*! \file updatexmltext.php
  29   Takes care of fixing the links in the XML format to use the url ID
  30   instead of a HREF.
  31   Will also fix all /content/download links to use the new format if any
  32   is found.
  33   Also if $fixAllAttributes is set to true it will make sure that all XML data
  34   is stored in correct charset according to site settings.
  35  */
  36  
  37  set_time_limit( 0 );
  38  
  39  include_once ( 'lib/ezutils/classes/ezcli.php' );
  40  include_once ( 'kernel/classes/ezscript.php' );
  41  
  42  $fixErrors = true;
  43  $fixAllAttributes = true;
  44  $fixAttribute = true;
  45  $fixURL = true;
  46  
  47  $cli =& eZCLI::instance();
  48  $endl = $cli->endlineString();
  49  
  50  $script =& eZScript::instance( array( 'description' => ( "eZ publish xml text field updater.\n\n".
  51                                                           "Goes trough all objects with XML fields and corrects any broken XML structures and content." .
  52                                                           "\n" .
  53                                                           "updatexmltext.php" ),
  54                                        'use-session' => true,
  55                                        'use-modules' => true,
  56                                        'use-extensions' => true ) );
  57  
  58  $script->startup();
  59  
  60  $options = $script->getOptions( "[sql]",
  61                                  "",
  62                                  array( 'sql' => "Display sql queries"
  63                                         ) );
  64  $script->initialize();
  65  
  66  $showSQL = $options['sql'] ? true : false;
  67  $siteAccess = $options['siteaccess'] ? $options['siteaccess'] : false;
  68  
  69  if ( $siteAccess )
  70  {
  71      changeSiteAccessSetting( $siteaccess, $siteAccess );
  72  }
  73  
  74  function changeSiteAccessSetting( &$siteaccess, $optionData )
  75  {
  76      global $isQuiet;
  77      $cli =& eZCLI::instance();
  78      if ( file_exists( 'settings/siteaccess/' . $optionData ) )
  79      {
  80          $siteaccess = $optionData;
  81          if ( !$isQuiet )
  82              $cli->notice( "Using siteaccess $siteaccess for xml text field update" );
  83      }
  84      else
  85      {
  86          if ( !$isQuiet )
  87              $cli->notice( "Siteaccess $optionData does not exist, using default siteaccess" );
  88      }
  89  }
  90  
  91  include_once ( 'kernel/classes/ezcontentclassattribute.php' );
  92  include_once ( 'kernel/classes/ezcontentobjectattribute.php' );
  93  include_once ( 'kernel/classes/ezcontentobject.php' );
  94  include_once ( 'kernel/classes/ezbinaryfilehandler.php' );
  95  include_once ( 'kernel/classes/datatypes/ezbinaryfile/ezbinaryfile.php' );
  96  
  97  include_once ( 'lib/ezdb/classes/ezdb.php' );
  98  include_once ( 'kernel/classes/datatypes/ezurl/ezurl.php' );
  99  
 100  $db =& eZDB::instance();
 101  $db->setIsSQLOutputEnabled( $showSQL );
 102  
 103  $xmlTypeAttributeList =& eZContentClassAttribute::fetchList( true, array( 'data_type' => 'ezxmltext',
 104                                                                            'version' => 0 ) );
 105  $classAttributeIDList = array();
 106  for ( $i = 0; $i < count( $xmlTypeAttributeList ); ++$i )
 107  {
 108      $xmlTypeAttribute =& $xmlTypeAttributeList[$i];
 109      $classAttributeIDList[] = $xmlTypeAttribute->attribute( 'id' );
 110  }
 111  unset( $xmlTypeAttributeList );
 112  
 113  $urlCount = eZURL::fetchListCount();
 114  
 115  $attributeCount = eZContentObjectAttribute::fetchListByClassID( $classAttributeIDList, false, array( 'offset' => 0,
 116                                                                                                       'length' => 3 ),
 117                                                                  false, true );
 118  $urlList = eZURL::fetchList();
 119  
 120  $urlRefMap = array();
 121  $urlIDMap = array();
 122  for ( $i = 0; $i < count( $urlList ); ++$i )
 123  {
 124      $url =& $urlList[$i];
 125      $urlRefMap[$url->attribute( 'url' )] =& $url;
 126      $urlIDMap[$url->attribute( 'id' )] =& $url;
 127  }
 128  
 129  function findAndReplaceLinks( &$doc, &$node )
 130  {
 131      global $urlRefMap;
 132      global $urlIDMap;
 133      global $showDebug;
 134      global $fixErrors;
 135      $foundLinks = false;
 136      unset( $children );
 137      if ( !$node )
 138          return $foundLinks;
 139      $children =& $node->Children;
 140      for ( $i = 0; $i < count( $children ); ++$i )
 141      {
 142          unset( $child );
 143          $child =& $children[$i];
 144          if ( $child->name() == 'link' )
 145          {
 146              unset( $linkAttributes );
 147              $linkAttributes =& $child->attributes();
 148              unset( $hrefAttribute );
 149              unset( $idAttribute );
 150              unset( $targetAttribute );
 151              $hrefAttribute = null;
 152              $idAttribute = null;
 153              $targetAttribute = null;
 154              for ( $j = 0; $j < count( $linkAttributes ); ++$j )
 155              {
 156                  $linkAttribute =& $linkAttributes[$j];
 157                  if ( $linkAttribute->name() == 'href' )
 158                      $hrefAttribute =& $linkAttributes[$j];
 159                  else if ( $linkAttribute->name() == 'id' )
 160                      $idAttribute =& $linkAttributes[$j];
 161                  else if ( $linkAttribute->name() == 'target' )
 162                      $targetAttribute =& $linkAttributes[$j];
 163              }
 164              if ( $idAttribute === null and
 165                   $hrefAttribute !== null )
 166              {
 167                  $href = $hrefAttribute->content();
 168                  if ( array_key_exists( $href, $urlRefMap ) )
 169                  {
 170                      if ( $showDebug )
 171                          print( "Found '$href'\n" );
 172                      $url =& $urlRefMap[$href];
 173                  }
 174                  else
 175                  {
 176                      if ( $showDebug )
 177                          print( "Found new '$href'\n" );
 178                      $urlID = eZURL::registerURL( $href );
 179                      $url = eZURL::fetch( $urlID );
 180                      $urlRefMap[$href] =& $url;
 181                      $urlIDMap[$urlID] =& $url;
 182                  }
 183                  $idAttribute = $doc->createAttributeNode( 'id', $url->attribute( 'id' ) );
 184                  $child->appendAttribute( $idAttribute );
 185                  $child->removeNamedAttribute( 'href' );
 186                  $foundLinks = true;
 187              }
 188              if ( $targetAttribute !== null )
 189              {
 190                  $target = $targetAttribute->content();
 191                  if ( $target == '_self' )
 192                  {
 193                      if ( $showDebug )
 194                          print( "Found '$target'\n" );
 195                      $child->removeNamedAttribute( 'target' );
 196                      $foundLinks = true;
 197                  }
 198              }
 199          }
 200          if ( findAndReplaceLinks( $doc, $child ) )
 201              $foundLinks = true;
 202          unset( $child );
 203      }
 204      unset( $children );
 205      return $foundLinks;
 206  }
 207  
 208  $wrongURLCount = 0;
 209  $fixedURLCount = 0;
 210  $wrongLinkCount = 0;
 211  
 212  $attributeOffset = 0;
 213  $attributeLimit = 140;
 214  
 215  $dotCount = 0;
 216  $dotTotalCount = 0;
 217  $dotMax = 70;
 218  
 219  include_once ( 'lib/ezxml/classes/ezxml.php' );
 220  $xml = new eZXML();
 221  
 222  if ( $fixAttribute )
 223  {
 224      print( "Fixing bad xml text links\n" );
 225      $badXMLArray = array();
 226      while ( $attributeOffset < $attributeCount )
 227      {
 228          $percent = ( $dotTotalCount * 100.0 ) / ( $attributeCount - 1 );
 229  //         if ( $percent > 27.76 )
 230  //             print( "cd " . $attributeOffset . ", " . $attributeLimit );
 231          unset( $objectAttributeList );
 232          $objectAttributeList =& eZContentObjectAttribute::fetchListByClassID( $classAttributeIDList, false, array( 'offset' => $attributeOffset,
 233                                                                                                                     'length' => $attributeLimit ),
 234                                                                                true, false );
 235  //         if ( $percent > 27.76 )
 236  //             print( "ef" );
 237          $lastID = false;
 238          for ( $i = 0; $i < count( $objectAttributeList ); ++$i )
 239          {
 240              $percent = ( $dotTotalCount * 100.0 ) / ( $attributeCount - 1 );
 241              $objectAttribute =& $objectAttributeList[$i];
 242  //             if ( $percent > 27.76 )
 243  //             {
 244  //                 print( $objectAttribute->attribute( 'id' ) . " (" . $objectAttribute->attribute( 'version' ) . ")\n" );
 245  //             }
 246  //             if ( $lastID !== false and
 247  //                  $objectAttribute->attribute( 'id' ) == $lastID)
 248  //                 print( "Found duplicate " . $objectAttribute->attribute( 'id' ) . "\n" );
 249              $lastID = $objectAttribute->attribute( 'id' );
 250              $dataType = $objectAttribute->dataType();
 251              $handleAttribute = true;
 252              $badDataType = false;
 253              if ( !$dataType or get_class( $dataType ) != 'ezxmltexttype' )
 254              {
 255                  $handleAttribute = false;
 256                  $badDataType = true;
 257              }
 258              unset( $content );
 259              $content = null;
 260              if ( $handleAttribute )
 261              {
 262                  $content =& $objectAttribute->content();
 263                  if ( !$content or !is_object( $content ) )
 264                      $handleAttribute = false;
 265              }
 266              unset( $xmlData );
 267              $xmlData = null;
 268              if ( $handleAttribute )
 269              {
 270  //                 if ( !is_object( $content ) )
 271  //                     print( get_class( $dataType ) . ", " . gettype( $content ) . " [$content]" . "\n" );
 272  //                 if ( is_object( $content ) )
 273  //                 {
 274                  $xmlData = $content->attribute( 'xml_data' );
 275                  if ( !$xmlData )
 276                      $handleAttribute = false;
 277  //                 }
 278  //                 else
 279  //                     $handleAttribute = false;
 280              }
 281              unset( $doc );
 282              $doc = null;
 283              if ( $handleAttribute )
 284              {
 285                  $doc =& $xml->domTree( $xmlData );
 286                  if ( $doc )
 287                  {
 288                      if ( findAndReplaceLinks( $doc, $doc->root() ) or
 289                           $objectAttribute->attribute( 'data_int' ) < EZ_XMLTEXT_VERSION_TIMESTAMP or
 290                           $fixAllAttributes )
 291                      {
 292                          if ( $showDebug )
 293                              print( "Links found and replaced\n" );
 294  //                 print( $doc->toString() . "\n" );
 295                          $docString = eZXMLTextType::domString( $doc );
 296                          $objectAttribute->setAttribute( 'data_text', $docString );
 297                          $objectAttribute->setAttribute( 'data_int', EZ_XMLTEXT_VERSION_TIMESTAMP );
 298                          if ( findAndReplaceLinks( $doc, $doc->root() ) or
 299                               $objectAttribute->attribute( 'data_int' ) < EZ_XMLTEXT_VERSION_TIMESTAMP )
 300                          {
 301                              ++$wrongLinkCount;
 302                              print( '*' );
 303                          }
 304                          else
 305                              print( '@' );
 306                      }
 307                      else
 308                          print( '.' );
 309                  }
 310                  else
 311                  {
 312                      if ( $showDebug )
 313                          print( "Invalid XML data: $xmlData\n" );
 314                      if ( trim( $xmlData ) == '' )
 315                      {
 316                          unset( $doc );
 317                          $doc = new eZDOMDocument();
 318                          $doc->setRoot( $doc->createElementNode( 'section' ) );
 319                          $docString = eZXMLTextType::domString( $doc );
 320                          $objectAttribute->setAttribute( 'data_text', $docString );
 321                          $objectAttribute->setAttribute( 'data_int', EZ_XMLTEXT_VERSION_TIMESTAMP );
 322                          ++$wrongLinkCount;
 323                          print( '0' );
 324                      }
 325                      else
 326                      {
 327                          unset( $doc );
 328                          $doc = new eZDOMDocument();
 329                          $doc->setRoot( $doc->createElementNode( 'section' ) );
 330                          $docString = eZXMLTextType::domString( $doc );
 331                          $objectAttribute->setAttribute( 'data_text', $docString );
 332                          $objectAttribute->setAttribute( 'data_int', EZ_XMLTEXT_VERSION_TIMESTAMP );
 333                          ++$wrongLinkCount;
 334                          $badXMLArray[] = array( 'id' => $objectAttribute->attribute( 'id' ),
 335                                                  'version' => $objectAttribute->attribute( 'version' ) );
 336                          print( '-' );
 337                      }
 338                  }
 339                  if ( $fixErrors )
 340                      $objectAttribute->sync();
 341              }
 342              else
 343              {
 344                  if ( $badDataType )
 345                      print( 'x' );
 346                  else
 347                      print( '.' );
 348              }
 349              ++$dotCount;
 350              ++$dotTotalCount;
 351              if ( $dotCount >= $dotMax or $dotTotalCount >= $attributeCount )
 352              {
 353                  $percent = number_format( ( $dotTotalCount * 100.0 ) / ( $attributeCount ), 2 );
 354                  $dotSpace = '';
 355                  if ( $dotTotalCount > $dotMax )
 356                      $dotSpace = str_repeat( ' ', $dotMax - $dotCount );
 357                  print( $dotSpace . " " . $percent . "% ( $dotTotalCount )\n" );
 358                  $dotCount = 0;
 359              }
 360  //             if ( $percent > 27.76 )
 361  //                 print( "ab" );
 362          }
 363  
 364          $attributeOffset += $attributeLimit;
 365      }
 366      print( "\n" );
 367      print( ". Ignored\n" );
 368      print( "@ XML data upgrade\n" );
 369      print( "* Fixed url usage\n" );
 370      print( "- Invalid XML data\n" );
 371      print( "0 Empty XML data\n" );
 372      print( "x Wrong datatype, should be ezxmltext\n" );
 373      if ( count( $badXMLArray ) > 0 )
 374      {
 375          print( "The following attributes had bad XML\n" );
 376          $len = 0;
 377          $i = 0;
 378          foreach ( $badXMLArray as $badXML )
 379          {
 380              $text = '';
 381              if ( $len > 70 )
 382              {
 383                  $len = 0;
 384                  if ( $i > 0 )
 385                      print( ',' );
 386                  print( "\n" );
 387              }
 388              else if ( $i > 0 )
 389                  $text = ', ';
 390              $text .= sprintf( "%5.d(%d)",
 391                               $badXML['id'],
 392                               $badXML['version'] );
 393              print( $text );
 394              $len += strlen( $text );
 395              ++$i;
 396          }
 397          print( "\n" );
 398      }
 399      print( "\n" );
 400  }
 401  
 402  $dotCount = 0;
 403  $dotTotalCount = 0;
 404  
 405  if ( $fixURL )
 406  {
 407      $failedURLArray = array();
 408      print( "Fixing bad download urls\n" );
 409      for ( $i = 0; $i < count( $urlList ); ++$i )
 410      {
 411          $url =& $urlList[$i];
 412          $urlText = $url->attribute( 'url' );
 413          if ( preg_match( "#/?content/download/[0-9]+/[0-9]+/[a-z_-]+/.+$#", $urlText, $matches ) )
 414          {
 415              if ( $showDebug )
 416                  print( "matched correct download url: '$urlText'\n" );
 417              print( '+' );
 418          }
 419          else if ( preg_match( "#/?content/download/([0-9]+)(/([0-9]*))?#", $urlText, $matches ) )
 420          {
 421              ++$wrongURLCount;
 422              $attributeID = $matches[1];
 423              $version = $matches[3];
 424              if ( $version == false )
 425                  $version = 1;
 426              if ( $showDebug )
 427                  print( "matched faulty download url: '$urlText' $attributeID @ $version\n" );
 428              $contentObjectAttribute = eZContentObjectAttribute::fetch( $attributeID, $version );
 429              if ( $contentObjectAttribute )
 430              {
 431                  $contentObjectID = $contentObjectAttribute->attribute( 'contentobject_id' );
 432                  $contentObject =& eZContentObject::fetch( $contentObjectID );
 433                  $downloadURL = eZBinaryFileHandler::downloadURL( $contentObject, $contentObjectAttribute );
 434                  $url->setAttribute( 'url', $downloadURL );
 435                  $url->setModified();
 436                  if ( $showDebug )
 437                      print( "correct download url: '$downloadURL/'\n" );
 438                  print( '*' );
 439              }
 440              else
 441              {
 442                  $url->setAttribute( 'is_valid', false );
 443                  $url->setModified();
 444                  print( '-' );
 445                  $failedURLArray[] = $url;
 446              }
 447              if ( $fixErrors )
 448              {
 449                  $url->sync();
 450                  ++$fixedURLCount;
 451              }
 452          }
 453          else
 454              print( '.' );
 455          ++$dotCount;
 456          ++$dotTotalCount;
 457          if ( $dotCount >= $dotMax or $dotTotalCount >= $urlCount )
 458          {
 459              $percent = number_format( ( $dotTotalCount * 100.0 ) / ( $urlCount ), 2 );
 460              $dotSpace = '';
 461              if ( $dotTotalCount > $dotMax )
 462                  $dotSpace = str_repeat( ' ', $dotMax - $dotCount );
 463              print( $dotSpace . " " . $percent . "%\n" );
 464              $dotCount = 0;
 465          }
 466      }
 467      print( "\n" );
 468      print( "+ Correct download url\n" );
 469      print( "* Fixed download url\n" );
 470      print( "- Failed to fix\n" );
 471      print( ". Ignored\n" );
 472      if ( count( $failedURLArray ) > 0 )
 473      {
 474          print( "The following urls could not be fixed\n" );
 475          foreach ( $failedURLArray as $failedURL )
 476          {
 477              print( sprintf( "%5.d: %s\n",
 478                              $failedURL->attribute( 'id' ),
 479                              $failedURL->attribute( 'url' ) ) );
 480          }
 481      }
 482      print( "\n" );
 483  }
 484  
 485  print( "Number of attributes      : $attributeCount\n" );
 486  print( "Number of fixed attributes: $wrongLinkCount\n\n" );
 487  
 488  print( "Number of urls      : $urlCount\n" );
 489  print( "Number of bad urls  : $wrongURLCount\n" );
 490  print( "Number of fixed urls: $fixedURLCount\n" );
 491  
 492  $script->shutdown();
 493  
 494  ?>