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  * --shared-desc        only do shared (commons) image description pages
  21  * --no-shared-desc     don't do shared image description pages
  22  * --categories         only do category pages
  23  * --redirects          only do redirects
  24  * --special            only do miscellaneous stuff
  25  * --force-copy         copy commons instead of symlink, needed for Wikimedia
  26  * --interlang          allow interlanguage links
  27  * --image-snapshot     copy all images used to the destination directory
  28  * --compress           generate compressed version of the html pages
  29  * --udp-profile <N>    profile 1/N rendering operations using ProfilerSimpleUDP
  30  */
  31
  32
  33 $optionsWithArgs = array( 's', 'd', 'e', 'k', 'checkpoint', 'slice', 'udp-profile' );
  34
  35 $profiling = false;
  36
  37 if ( $profiling ) {
  38         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  39         function wfSetupDump() {
  40                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  41                 $wgProfiling = true;
  42                 $wgProfileToDatabase = false;
  43                 $wgProfileSampleRate = 1;
  44         }
  45 }
  46
  47 if ( in_array( '--udp-profile', $argv ) ) {
  48         define( 'MW_FORCE_PROFILE', 1 );
  49 }
  50
  51 require_once( "commandLine.inc" );
  52 require_once( "dumpHTML.inc" );
  53
  54 error_reporting( E_ALL & (~E_NOTICE) );
  55
  56 if ( !empty( $options['s'] ) ) {
  57         $start = $options['s'];
  58 } else {
  59         $start = 1;
  60 }
  61
  62 if ( !empty( $options['e'] ) ) {
  63         $end = $options['e'];
  64 } else {
  65         $dbr =& wfGetDB( DB_SLAVE );
  66         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  67 }
  68
  69 if ( !empty( $options['d'] ) ) {
  70         $dest = $options['d'];
  71 } else {
  72         $dest = "$IP/static";
  73 }
  74
  75 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  76
  77 if ( $options['slice'] ) {
  78         $bits = explode( '/', $options['slice'] );
  79         if ( count( $bits ) != 2 || $bits[0] < 1 || $bits[0] > $bits[1] ) {
  80                 print "Invalid slice specification";
  81                 exit;
  82         }
  83         $sliceNumerator = $bits[0];
  84         $sliceDenominator = $bits[1];
  85 } else {
  86         $sliceNumerator = $sliceDenominator = 1;
  87 }
  88
  89 $wgHTMLDump = new DumpHTML( array(
  90         'dest' => $dest,
  91         'forceCopy' => $options['force-copy'],
  92         'alternateScriptPath' => $options['interlang'],
  93         'interwiki' => $options['interlang'],
  94         'skin' => $skin,
  95         'makeSnapshot' => $options['image-snapshot'],
  96         'checkpointFile' => $options['checkpoint'],
  97         'startID' => $start,
  98         'endID' => $end,
  99         'sliceNumerator' => $sliceNumerator,
 100         'sliceDenominator' => $sliceDenominator,
 101         'noOverwrite' => $options['no-overwrite'],
 102         'compress' => $options['compress'],
 103         'noSharedDesc' => $options['no-shared-desc'],
 104         'udpProfile' => $options['udp-profile'],
 105 ));
 106
 107
 108 if ( $options['special'] ) {
 109         $wgHTMLDump->doSpecials();
 110 } elseif ( $options['images'] ) {
 111         $wgHTMLDump->doImageDescriptions();
 112 } elseif ( $options['categories'] ) {
 113         $wgHTMLDump->doCategories();
 114 } elseif ( $options['redirects'] ) {
 115         $wgHTMLDump->doRedirects();
 116 } elseif ( $options['shared-desc'] ) {
 117         $wgHTMLDump->doSharedImageDescriptions();
 118 } else {
 119         print "Creating static HTML dump in directory $dest. \n";
 120         $dbr =& wfGetDB( DB_SLAVE );
 121         $server = $dbr->getProperty( 'mServer' );
 122         print "Using database {$server}\n";
 123
 124         if ( !isset( $options['e'] ) ) {
 125                 $wgHTMLDump->doEverything();
 126         } else {
 127                 $wgHTMLDump->doArticles();
 128         }
 129 }
 130
 131 if ( isset( $options['debug'] ) ) {
 132         #print_r($GLOBALS);
 133         # Workaround for bug #36957
 134         $globals = array_keys( $GLOBALS );
 135         #sort( $globals );
 136         $sizes = array();
 137         foreach ( $globals as $name ) {
 138                  $sizes[$name] = strlen( serialize( $GLOBALS[$name] ) );
 139         }
 140         arsort($sizes);
 141         $sizes = array_slice( $sizes, 0, 20 );
 142         foreach ( $sizes as $name => $size ) {
 143                 printf( "%9d %s\n", $size, $name );
 144         }
 145 }
 146
 147 if ( $profiling ) {
 148         echo $wgProfiler->getOutput();
 149 }
 150
 151 ?>