maintenance/importImages.php

   1 <?php
   2
   3 /**
   4  * Maintenance script to import one or more images from the local file system into
   5  * the wiki without using the web-based interface
   6  *
   7  * @file
   8  * @ingroup Maintenance
   9  * @author Rob Church <robchur@gmail.com>
  10  */
  11
  12 $optionsWithArgs = array( 'extensions', 'comment', 'comment-file', 'comment-ext', 'user', 'license', 'sleep', 'limit', 'from' );
  13 require_once( dirname(__FILE__) . '/commandLine.inc' );
  14 require_once( 'importImages.inc' );
  15 $processed = $added = $ignored = $skipped = $overwritten = $failed = 0;
  16
  17 echo( "Import Images\n\n" );
  18
  19 # Need a path
  20 if( count( $args ) > 0 ) {
  21
  22         $dir = $args[0];
  23
  24         # Check Protection
  25         if (isset($options['protect']) && isset($options['unprotect']))
  26                         die("Cannot specify both protect and unprotect.  Only 1 is allowed.\n");
  27
  28 if (isset($options['protect']) && $options['protect'] == 1)
  29                         die("You must specify a protection option.\n");
  30
  31         # Prepare the list of allowed extensions
  32         global $wgFileExtensions;
  33         $extensions = isset( $options['extensions'] )
  34                 ? explode( ',', strtolower( $options['extensions'] ) )
  35                 : $wgFileExtensions;
  36
  37         # Search the path provided for candidates for import
  38         $files = findFiles( $dir, $extensions );
  39
  40         # Initialise the user for this operation
  41         $user = isset( $options['user'] )
  42                 ? User::newFromName( $options['user'] )
  43                 : User::newFromName( 'Maintenance script' );
  44         if( !$user instanceof User )
  45                 $user = User::newFromName( 'Maintenance script' );
  46         $wgUser = $user;
  47
  48         # Get block check. If a value is given, this specified how often the check is performed
  49         if ( isset( $options['check-userblock'] ) ) {
  50                 if ( !$options['check-userblock'] ) $checkUserBlock = 1;
  51                 else $checkUserBlock = (int)$options['check-userblock'];
  52         } else {
  53                 $checkUserBlock = false;
  54         }
  55
  56         # Get --from
  57         $from = @$options['from'];
  58
  59         # Get sleep time.
  60         $sleep = @$options['sleep'];
  61         if ( $sleep ) $sleep = (int)$sleep;
  62
  63         # Get limit number
  64         $limit = @$options['limit'];
  65         if ( $limit ) $limit = (int)$limit;
  66
  67         # Get the upload comment
  68         $comment = 'Importing image file';
  69
  70         if ( isset( $options['comment-file'] ) ) {
  71                 $comment =  file_get_contents( $options['comment-file'] );
  72                 if ( $comment === false || $comment === NULL ) {
  73                         die( "failed to read comment file: {$options['comment-file']}\n" );
  74                 }
  75         }
  76         else if ( isset( $options['comment'] ) ) {
  77                 $comment =  $options['comment'];
  78         }
  79
  80         $commentExt = isset( $options['comment-ext'] ) ? $options['comment-ext'] : false;
  81
  82         # Get the license specifier
  83         $license = isset( $options['license'] ) ? $options['license'] : '';
  84
  85         # Batch "upload" operation
  86         if( ( $count = count( $files ) ) > 0 ) {
  87
  88                 foreach( $files as $file ) {
  89                         $base = wfBaseName( $file );
  90
  91                         # Validate a title
  92                         $title = Title::makeTitleSafe( NS_FILE, $base );
  93                         if( !is_object( $title ) ) {
  94                                 echo( "{$base} could not be imported; a valid title cannot be produced\n" );
  95                                 continue;
  96                         }
  97
  98                         if ( $from ) {
  99                                 if ( $from == $title->getDBkey() ) {
 100                                         $from = NULL;
 101                                 } else {
 102                                         $ignored++;
 103                                         continue;
 104                                 }
 105                         }
 106
 107                         if ( $checkUserBlock && ( ( $processed % $checkUserBlock ) == 0 ) ) {
 108                                 $user->clearInstanceCache( 'name' ); //reload from DB!
 109                                 if ( $user->isBlocked() ) {
 110                                         echo( $user->getName() . " was blocked! Aborting." );
 111                                         break;
 112                                 }
 113                         }
 114
 115                         # Check existence
 116                         $image = wfLocalFile( $title );
 117                         if( $image->exists() ) {
 118                                 if( isset( $options['overwrite'] ) ) {
 119                                         echo( "{$base} exists, overwriting..." );
 120                                         $svar = 'overwritten';
 121                                 } else {
 122                                         echo( "{$base} exists, skipping\n" );
 123                                         $skipped++;
 124                                         continue;
 125                                 }
 126                         } else {
 127                                 if ( isset( $options['skip-dupes'] ) ) {
 128                                         $repo = $image->getRepo();
 129                                         $sha1 = File::sha1Base36( $file ); #XXX: we end up calculating this again when actually uploading. that sucks.
 130
 131                                         $dupes = $repo->findBySha1( $sha1 );
 132
 133                                         if ( $dupes ) {
 134                                                 echo( "{$base} already exists as " . $dupes[0]->getName() . ", skipping\n" );
 135                                                 $skipped++;
 136                                                 continue;
 137                                         }
 138                                 }
 139
 140                                 echo( "Importing {$base}..." );
 141                                 $svar = 'added';
 142                         }
 143
 144                         # Find comment text
 145                         $commentText = false;
 146
 147                         if ( $commentExt ) {
 148                                 $f = findAuxFile( $file, $commentExt );
 149                                 if ( !$f ) {
 150                                         echo( " No comment file with extension {$commentExt} found for {$file}, using default comment. " );
 151                                 } else {
 152                                         $commentText = file_get_contents( $f );
 153                                         if ( !$f ) {
 154                                                 echo( " Failed to load comment file {$f}, using default comment. " );
 155                                         }
 156                                 }
 157
 158                                 if ( $commentText && $comment ) {
 159                                         $commentText = trim( $commentText ) . "\n\n" . trim( $comment );
 160                                 }
 161                         }
 162
 163                         if ( !$commentText ) {
 164                                 $commentText = $comment;
 165                         }
 166
 167                         # Import the file
 168                         if ( isset( $options['dry'] ) ) {
 169                                 echo( " publishing {$file}... " );
 170                         } else {
 171                                 $archive = $image->publish( $file );
 172                                 if( WikiError::isError( $archive ) || !$archive->isGood() ) {
 173                                         echo( "failed.\n" );
 174                                         $failed++;
 175                                         continue;
 176                                 }
 177                         }
 178
 179                         $doProtect = false;
 180                         $restrictions = array();
 181
 182                         global $wgRestrictionLevels;
 183
 184                         $protectLevel = isset($options['protect']) ? $options['protect'] : null;
 185
 186                         if ( $protectLevel && in_array( $protectLevel, $wgRestrictionLevels ) ) {
 187                                         $restrictions['move'] = $protectLevel;
 188                                         $restrictions['edit'] = $protectLevel;
 189                                         $doProtect = true;
 190                         }
 191                         if (isset($options['unprotect'])) {
 192                                         $restrictions['move'] = '';
 193                                         $restrictions['edit'] = '';
 194                                         $doProtect = true;
 195                         }
 196
 197
 198                         if ( isset( $options['dry'] ) ) {
 199                                 echo( "done.\n" );
 200                         } else if ( $image->recordUpload( $archive->value, $commentText, $license ) ) {
 201                                 # We're done!
 202                                 echo( "done.\n" );
 203                                 if ($doProtect) {
 204                                                 # Protect the file
 205                                                 $article = new Article( $title );
 206                                                 echo "\nWaiting for slaves...\n";
 207                                                 // Wait for slaves.
 208                                                 sleep(2.0);
 209                                                 wfWaitForSlaves( 1.0 );
 210
 211                                                 echo( "\nSetting image restrictions ... " );
 212                                                 if ( $article->updateRestrictions($restrictions) )
 213                                                                 echo( "done.\n" );
 214                                                 else
 215                                                                 echo( "failed.\n" );
 216                                 }
 217
 218                         } else {
 219                                 echo( "failed.\n" );
 220                                 $svar = 'failed';
 221                         }
 222
 223                         $$svar++;
 224                         $processed++;
 225
 226                         if ( $limit && $processed >= $limit )
 227                                 break;
 228
 229                         if ( $sleep )
 230                                 sleep( $sleep );
 231                 }
 232
 233                 # Print out some statistics
 234                 echo( "\n" );
 235                 foreach( array( 'count' => 'Found', 'limit' => 'Limit', 'ignored' => 'Ignored',
 236                         'added' => 'Added', 'skipped' => 'Skipped', 'overwritten' => 'Overwritten',
 237                         'failed' => 'Failed' ) as $var => $desc ) {
 238                         if( $$var > 0 )
 239                                 echo( "{$desc}: {$$var}\n" );
 240                 }
 241
 242         } else {
 243                 echo( "No suitable files could be found for import.\n" );
 244         }
 245
 246 } else {
 247         showUsage();
 248 }
 249
 250 exit(0);
 251
 252 function showUsage( $reason = false ) {
 253         if( $reason ) {
 254                 echo( $reason . "\n" );
 255         }
 256
 257         echo <<<END
 258 Imports images and other media files into the wiki
 259 USAGE: php importImages.php [options] <dir>
 260
 261 <dir> : Path to the directory containing images to be imported
 262
 263 Options:
 264 --extensions=<exts>     Comma-separated list of allowable extensions, defaults to \$wgFileExtensions
 265 --overwrite             Overwrite existing images with the same name (default is to skip them)
 266 --limit=<num>           Limit the number of images to process. Ignored or skipped images are not counted.
 267 --from=<name>           Ignore all files until the one with the given name. Useful for resuming
 268                         aborted imports. <name> should be the file's canonical database form.
 269 --skip-dupes            Skip images that were already uploaded under a different name (check SHA1)
 270 --sleep=<sec>           Sleep between files. Useful mostly for debugging.
 271 --user=<username>       Set username of uploader, default 'Maintenance script'
 272 --check-userblock       Check if the user got blocked during import.
 273 --comment=<text>        Set upload summary comment, default 'Importing image file'.
 274 --comment-file=<file>   Set upload summary comment the the content of <file>.
 275 --comment-ext=<ext>     Causes the comment for each file to be loaded from a file with the same name
 276                         but the extension <ext>. If a global comment is also given, it is appended.
 277 --license=<code>        Use an optional license template
 278 --dry                   Dry run, don't import anything
 279 --protect=<protect>     Specify the protect value (autoconfirmed,sysop)
 280 --unprotect             Unprotects all uploaded images
 281
 282 END;
 283         exit(1);
 284 }