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  * -k <skin>            skin to use (defaults to htmldump)
  16  * --no-overwrite       skip existing HTML files
  17  * --checkpoint <file>  use a checkpoint file to allow restarting of interrupted dumps
  18  * --slice <n/m>        split the job into m segments and do the n'th one
  19  * --images             only do image description pages
  20  * --categories         only do category pages
  21  * --redirects          only do redirects
  22  * --special            only do miscellaneous stuff
  23  * --force-copy         copy commons instead of symlink, needed for Wikimedia
  24  * --interlang          allow interlanguage links
  25  * --image-snapshot     copy all images used to the destination directory
  26  * --compress           generate compressed version of the html pages
  27  */
  28
  29
  30 $optionsWithArgs = array( 's', 'd', 'e', 'k', 'checkpoint', 'slice' );
  31
  32 $profiling = false;
  33
  34 if ( $profiling ) {
  35         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  36         function wfSetupDump() {
  37                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  38                 $wgProfiling = true;
  39                 $wgProfileToDatabase = false;
  40                 $wgProfileSampleRate = 1;
  41         }
  42 }
  43
  44 require_once( "commandLine.inc" );
  45 require_once( "dumpHTML.inc" );
  46
  47 error_reporting( E_ALL & (~E_NOTICE) );
  48
  49 if ( !empty( $options['s'] ) ) {
  50         $start = $options['s'];
  51 } else {
  52         $start = 1;
  53 }
  54
  55 if ( !empty( $options['e'] ) ) {
  56         $end = $options['e'];
  57 } else {
  58         $dbr =& wfGetDB( DB_SLAVE );
  59         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  60 }
  61
  62 if ( !empty( $options['d'] ) ) {
  63         $dest = $options['d'];
  64 } else {
  65         $dest = "$IP/static";
  66 }
  67
  68 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  69
  70 if ( $options['slice'] ) {
  71         $bits = explode( '/', $options['slice'] );
  72         if ( count( $bits ) != 2 || $bits[0] < 1 || $bits[0] > $bits[1] ) {
  73                 print "Invalid slice specification";
  74                 exit;
  75         }
  76         $sliceNumerator = $bits[0];
  77         $sliceDenominator = $bits[1];
  78 } else {
  79         $sliceNumerator = $sliceDenominator = 1;
  80 }
  81
  82 $wgHTMLDump = new DumpHTML( array(
  83         'dest' => $dest,
  84         'forceCopy' => $options['force-copy'],
  85         'alternateScriptPath' => $options['interlang'],
  86         'interwiki' => $options['interlang'],
  87         'skin' => $skin,
  88         'makeSnapshot' => $options['image-snapshot'],
  89         'checkpointFile' => $options['checkpoint'],
  90         'startID' => $start,
  91         'endID' => $end,
  92         'sliceNumerator' => $sliceNumerator,
  93         'sliceDenominator' => $sliceDenominator,
  94         'noOverwrite' => $options['no-overwrite'],
  95         'compress' => $options['compress'],
  96 ));
  97
  98
  99 if ( $options['special'] ) {
 100         $wgHTMLDump->doSpecials();
 101 } elseif ( $options['images'] ) {
 102         $wgHTMLDump->doImageDescriptions();
 103 } elseif ( $options['categories'] ) {
 104         $wgHTMLDump->doCategories();
 105 } elseif ( $options['redirects'] ) {
 106         $wgHTMLDump->doRedirects();
 107 } else {
 108         print "Creating static HTML dump in directory $dest. \n";
 109         $dbr =& wfGetDB( DB_SLAVE );
 110         $server = $dbr->getProperty( 'mServer' );
 111         print "Using database {$server}\n";
 112
 113         if ( !isset( $options['e'] ) ) {
 114                 $wgHTMLDump->doEverything();
 115         } else {
 116                 $wgHTMLDump->doArticles();
 117         }
 118 }
 119
 120 if ( isset( $options['debug'] ) ) {
 121         #print_r($GLOBALS);
 122         # Workaround for bug #36957
 123         $globals = array_keys( $GLOBALS );
 124         #sort( $globals );
 125         $sizes = array();
 126         foreach ( $globals as $name ) {
 127                  $sizes[$name] = strlen( serialize( $GLOBALS[$name] ) );
 128         }
 129         arsort($sizes);
 130         $sizes = array_slice( $sizes, 0, 20 );
 131         foreach ( $sizes as $name => $size ) {
 132                 printf( "%9d %s\n", $size, $name );
 133         }
 134 }
 135
 136 if ( $profiling ) {
 137         echo $wgProfiler->getOutput();
 138 }
 139
 140 ?>