maintenance/dumpHTML.php

   1 <?php
   2 /**
   3  * @todo document
   4  * @addtogroup Maintenance
   5  */
   6
   7 function ShowUsage() {
   8 echo <<<END
   9 Usage:
  10 php dumpHTML.php --help
  11 php dumpHTML.php [options...]
  12
  13         --help               show this message
  14
  15         -d <dest>            destination directory
  16         -s <start>           start ID
  17         -e <end>             end ID
  18         -k <skin>            skin to use (defaults to htmldump)
  19         --no-overwrite       skip existing HTML files
  20         --checkpoint <file>  use a checkpoint file to allow restarting of interrupted dumps
  21         --slice <n/m>        split the job into m segments and do the n'th one
  22         --images             only do image description pages
  23         --shared-desc        only do shared (commons) image description pages
  24         --no-shared-desc     don't do shared image description pages
  25         --categories         only do category pages
  26         --redirects          only do redirects
  27         --special            only do miscellaneous stuff
  28         --force-copy         copy commons instead of symlink, needed for Wikimedia
  29         --interlang          allow interlanguage links
  30         --image-snapshot     copy all images used to the destination directory
  31         --compress           generate compressed version of the html pages
  32         --udp-profile <N>    profile 1/N rendering operations using ProfilerSimpleUDP
  33
  34 END;
  35 }
  36
  37 $optionsWithArgs = array( 's', 'd', 'e', 'k', 'checkpoint', 'slice', 'udp-profile' );
  38 $options = array( 'help' );
  39 $profiling = false;
  40
  41 if ( $profiling ) {
  42         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  43         function wfSetupDump() {
  44                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  45                 $wgProfiling = true;
  46                 $wgProfileToDatabase = false;
  47                 $wgProfileSampleRate = 1;
  48         }
  49 }
  50
  51 if ( in_array( '--udp-profile', $argv ) ) {
  52         define( 'MW_FORCE_PROFILE', 1 );
  53 }
  54
  55 require_once( "commandLine.inc" );
  56 require_once( "dumpHTML.inc" );
  57
  58 error_reporting( E_ALL & (~E_NOTICE) );
  59
  60 if( isset( $options['help'] ) ) {
  61         ShowUsage();
  62         exit();
  63 }
  64
  65 if ( !empty( $options['s'] ) ) {
  66         $start = $options['s'];
  67 } else {
  68         $start = 1;
  69 }
  70
  71 if ( !empty( $options['e'] ) ) {
  72         $end = $options['e'];
  73 } else {
  74         $dbr = wfGetDB( DB_SLAVE );
  75         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  76 }
  77
  78 if ( !empty( $options['d'] ) ) {
  79         $dest = $options['d'];
  80 } else {
  81         $dest = "$IP/static";
  82 }
  83
  84 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  85
  86 if ( $options['slice'] ) {
  87         $bits = explode( '/', $options['slice'] );
  88         if ( count( $bits ) != 2 || $bits[0] < 1 || $bits[0] > $bits[1] ) {
  89                 print "Invalid slice specification";
  90                 exit;
  91         }
  92         $sliceNumerator = $bits[0];
  93         $sliceDenominator = $bits[1];
  94 } else {
  95         $sliceNumerator = $sliceDenominator = 1;
  96 }
  97
  98 $wgHTMLDump = new DumpHTML( array(
  99         'dest' => $dest,
 100         'forceCopy' => $options['force-copy'],
 101         'alternateScriptPath' => $options['interlang'],
 102         'interwiki' => $options['interlang'],
 103         'skin' => $skin,
 104         'makeSnapshot' => $options['image-snapshot'],
 105         'checkpointFile' => $options['checkpoint'],
 106         'startID' => $start,
 107         'endID' => $end,
 108         'sliceNumerator' => $sliceNumerator,
 109         'sliceDenominator' => $sliceDenominator,
 110         'noOverwrite' => $options['no-overwrite'],
 111         'compress' => $options['compress'],
 112         'noSharedDesc' => $options['no-shared-desc'],
 113         'udpProfile' => $options['udp-profile'],
 114 ));
 115
 116
 117 if ( $options['special'] ) {
 118         $wgHTMLDump->doSpecials();
 119 } elseif ( $options['images'] ) {
 120         $wgHTMLDump->doImageDescriptions();
 121 } elseif ( $options['categories'] ) {
 122         $wgHTMLDump->doCategories();
 123 } elseif ( $options['redirects'] ) {
 124         $wgHTMLDump->doRedirects();
 125 } elseif ( $options['shared-desc'] ) {
 126         $wgHTMLDump->doSharedImageDescriptions();
 127 } else {
 128         print "Creating static HTML dump in directory $dest. \n";
 129         $dbr = wfGetDB( DB_SLAVE );
 130         $server = $dbr->getProperty( 'mServer' );
 131         print "Using database {$server}\n";
 132
 133         if ( !isset( $options['e'] ) ) {
 134                 $wgHTMLDump->doEverything();
 135         } else {
 136                 $wgHTMLDump->doArticles();
 137         }
 138 }
 139
 140 if ( isset( $options['debug'] ) ) {
 141         #print_r($GLOBALS);
 142         # Workaround for bug #36957
 143         $globals = array_keys( $GLOBALS );
 144         #sort( $globals );
 145         $sizes = array();
 146         foreach ( $globals as $name ) {
 147                  $sizes[$name] = strlen( serialize( $GLOBALS[$name] ) );
 148         }
 149         arsort($sizes);
 150         $sizes = array_slice( $sizes, 0, 20 );
 151         foreach ( $sizes as $name => $size ) {
 152                 printf( "%9d %s\n", $size, $name );
 153         }
 154 }
 155
 156 if ( $profiling ) {
 157         echo $wgProfiler->getOutput();
 158 }
 159
 160 ?>