maintenance/dumpHTML.php

   1 <?php
   2 /**
   3  * @todo document
   4  * @addtogroup Maintenance
   5  */
   6
   7 $usage = <<<ENDS
   8 Usage:
   9 php dumpHTML.php [options...]
  10
  11         --help               show this message
  12
  13         -d <dest>            destination directory
  14         -s <start>           start ID
  15         -e <end>             end ID
  16         -k <skin>            skin to use (defaults to htmldump)
  17         --no-overwrite       skip existing HTML files
  18         --checkpoint <file>  use a checkpoint file to allow restarting of interrupted dumps
  19         --slice <n/m>        split the job into m segments and do the n'th one
  20         --images             only do image description pages
  21         --shared-desc        only do shared (commons) image description pages
  22         --no-shared-desc     don't do shared image description pages
  23         --categories         only do category pages
  24         --redirects          only do redirects
  25         --special            only do miscellaneous stuff
  26         --force-copy         copy commons instead of symlink, needed for Wikimedia
  27         --interlang          allow interlanguage links
  28         --image-snapshot     copy all images used to the destination directory
  29         --compress           generate compressed version of the html pages
  30         --udp-profile <N>    profile 1/N rendering operations using ProfilerSimpleUDP
  31
  32 ENDS;
  33
  34 $optionsWithArgs = array( 's', 'd', 'e', 'k', 'checkpoint', 'slice', 'udp-profile' );
  35 $options = array( 'help' );
  36 $profiling = false;
  37
  38 if ( $profiling ) {
  39         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  40         function wfSetupDump() {
  41                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  42                 $wgProfiling = true;
  43                 $wgProfileToDatabase = false;
  44                 $wgProfileSampleRate = 1;
  45         }
  46 }
  47
  48 if ( in_array( '--udp-profile', $argv ) ) {
  49         define( 'MW_FORCE_PROFILE', 1 );
  50 }
  51
  52 require_once( "commandLine.inc" );
  53 require_once( "dumpHTML.inc" );
  54
  55 error_reporting( E_ALL & (~E_NOTICE) );
  56
  57 if( isset( $options['help'] ) ) {
  58         echo $usage;
  59         exit;
  60 }
  61
  62 if ( !empty( $options['s'] ) ) {
  63         $start = $options['s'];
  64 } else {
  65         $start = 1;
  66 }
  67
  68 if ( !empty( $options['e'] ) ) {
  69         $end = $options['e'];
  70 } else {
  71         $dbr = wfGetDB( DB_SLAVE );
  72         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  73 }
  74
  75 if ( !empty( $options['d'] ) ) {
  76         $dest = $options['d'];
  77 } else {
  78         $dest = "$IP/static";
  79 }
  80
  81 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  82
  83 if ( $options['slice'] ) {
  84         $bits = explode( '/', $options['slice'] );
  85         if ( count( $bits ) != 2 || $bits[0] < 1 || $bits[0] > $bits[1] ) {
  86                 print "Invalid slice specification";
  87                 exit;
  88         }
  89         $sliceNumerator = $bits[0];
  90         $sliceDenominator = $bits[1];
  91 } else {
  92         $sliceNumerator = $sliceDenominator = 1;
  93 }
  94
  95 $wgHTMLDump = new DumpHTML( array(
  96         'dest' => $dest,
  97         'forceCopy' => $options['force-copy'],
  98         'alternateScriptPath' => $options['interlang'],
  99         'interwiki' => $options['interlang'],
 100         'skin' => $skin,
 101         'makeSnapshot' => $options['image-snapshot'],
 102         'checkpointFile' => $options['checkpoint'],
 103         'startID' => $start,
 104         'endID' => $end,
 105         'sliceNumerator' => $sliceNumerator,
 106         'sliceDenominator' => $sliceDenominator,
 107         'noOverwrite' => $options['no-overwrite'],
 108         'compress' => $options['compress'],
 109         'noSharedDesc' => $options['no-shared-desc'],
 110         'udpProfile' => $options['udp-profile'],
 111 ));
 112
 113
 114 if ( $options['special'] ) {
 115         $wgHTMLDump->doSpecials();
 116 } elseif ( $options['images'] ) {
 117         $wgHTMLDump->doImageDescriptions();
 118 } elseif ( $options['categories'] ) {
 119         $wgHTMLDump->doCategories();
 120 } elseif ( $options['redirects'] ) {
 121         $wgHTMLDump->doRedirects();
 122 } elseif ( $options['shared-desc'] ) {
 123         $wgHTMLDump->doSharedImageDescriptions();
 124 } else {
 125         print "Creating static HTML dump in directory $dest. \n";
 126         $dbr = wfGetDB( DB_SLAVE );
 127         $server = $dbr->getProperty( 'mServer' );
 128         print "Using database {$server}\n";
 129
 130         if ( !isset( $options['e'] ) ) {
 131                 $wgHTMLDump->doEverything();
 132         } else {
 133                 $wgHTMLDump->doArticles();
 134         }
 135 }
 136
 137 if ( isset( $options['debug'] ) ) {
 138         #print_r($GLOBALS);
 139         # Workaround for bug #36957
 140         $globals = array_keys( $GLOBALS );
 141         #sort( $globals );
 142         $sizes = array();
 143         foreach ( $globals as $name ) {
 144                  $sizes[$name] = strlen( serialize( $GLOBALS[$name] ) );
 145         }
 146         arsort($sizes);
 147         $sizes = array_slice( $sizes, 0, 20 );
 148         foreach ( $sizes as $name => $size ) {
 149                 printf( "%9d %s\n", $size, $name );
 150         }
 151 }
 152
 153 if ( $profiling ) {
 154         echo $wgProfiler->getOutput();
 155 }
 156
 157