maintenance/dumpHTML.php

   1 <?php
   2 /**
   3  * @todo document
   4  * @package MediaWiki
   5  * @subpackage Maintenance
   6  */
   7
   8 /**
   9  * Usage:
  10  * php dumpHTML.php [options...]
  11  *
  12  * -d <dest>      destination directory
  13  * -s <start>     start ID
  14  * -e <end>       end ID
  15  * --images       only do image description pages
  16  * --categories   only do category pages
  17  * --special      only do miscellaneous stuff
  18  * --force-copy   copy commons instead of symlink, needed for Wikimedia
  19  * --interlang    allow interlanguage links
  20  */
  21
  22
  23 $optionsWithArgs = array( 's', 'd', 'e' );
  24
  25 require_once( "commandLine.inc" );
  26 require_once( "dumpHTML.inc" );
  27
  28 error_reporting( E_ALL & (~E_NOTICE) );
  29 define( 'CHUNK_SIZE', 50 );
  30
  31 if ( !empty( $options['s'] ) ) {
  32         $start = $options['s'];
  33 } else {
  34         $start = 1;
  35 }
  36
  37 if ( !empty( $options['e'] ) ) {
  38         $end = $options['e'];
  39 } else {
  40         $dbr =& wfGetDB( DB_SLAVE );
  41         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  42 }
  43
  44 if ( !empty( $options['d'] ) ) {
  45         $dest = $options['d'];
  46 } else {
  47         $dest = 'static';
  48 }
  49
  50 $d = new DumpHTML( array(
  51         'dest' => $dest,
  52         'forceCopy' => $options['force-copy'],
  53         'alternateScriptPath' => $options['interlang'],
  54         'interwiki' => $options['interlang'],
  55 ));
  56
  57
  58 if ( $options['special'] ) {
  59         $d->doSpecials();
  60 } elseif ( $options['images'] ) {
  61         $d->doImageDescriptions();
  62 } elseif ( $options['categories'] ) {
  63         $d->doCategories();
  64 } else {
  65         print("Creating static HTML dump in directory $dest. \n".
  66                 "Starting from page_id $start of $end.\n");
  67         $d->doArticles( $start, $end );
  68         $d->doImageDescriptions();
  69         $d->doCategories();
  70         $d->doSpecials();
  71
  72         /*
  73         if ( $end - $start > CHUNK_SIZE * 2 ) {
  74                 // Split the problem into smaller chunks, run them in different PHP instances
  75                 // This is a memory/resource leak workaround
  76                 print("Creating static HTML dump in directory $dest. \n".
  77                         "Starting from page_id $start of $end.\n");
  78
  79                 chdir( "maintenance" );
  80                 for ( $chunkStart = $start; $chunkStart < $end; $chunkStart += CHUNK_SIZE ) {
  81                         $chunkEnd = $chunkStart + CHUNK_SIZE - 1;
  82                         if ( $chunkEnd > $end ) {
  83                                 $chunkEnd = $end;
  84                         }
  85                         passthru( "php dumpHTML.php -d " . wfEscapeShellArg( $dest ) . " -s $chunkStart -e $chunkEnd" );
  86                 }
  87                 chdir( ".." );
  88                 $d->doImageDescriptions();
  89                 $d->doCategories();
  90                 $d->doMainPage( $dest );
  91         } else {
  92                 $d->doArticles( $start, $end );
  93         }
  94         */
  95 }
  96
  97 exit();
  98
  99 ?>