maintenance/dumpHTML.php

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