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  * --images       only do image description pages
  16  * --categories   only do category pages
  17  * --special      only do miscellaneous stuff
  18  * --force-copy   copy commons instead of symlink, needed for Wikimedia
  19  * --interlang    allow interlanguage links
  20  */
  21
  22
  23 $optionsWithArgs = array( 's', 'd', 'e' );
  24
  25 $profiling = false;
  26
  27 if ( $profiling ) {
  28         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  29         function wfSetupDump() {
  30                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  31                 $wgProfiling = true;
  32                 $wgProfileToDatabase = false;
  33                 $wgProfileSampleRate = 1;
  34         }
  35 }
  36
  37 require_once( "commandLine.inc" );
  38 require_once( "dumpHTML.inc" );
  39
  40 error_reporting( E_ALL & (~E_NOTICE) );
  41 define( 'CHUNK_SIZE', 50 );
  42
  43 if ( !empty( $options['s'] ) ) {
  44         $start = $options['s'];
  45 } else {
  46         $start = 1;
  47 }
  48
  49 if ( !empty( $options['e'] ) ) {
  50         $end = $options['e'];
  51 } else {
  52         $dbr =& wfGetDB( DB_SLAVE );
  53         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  54 }
  55
  56 if ( !empty( $options['d'] ) ) {
  57         $dest = $options['d'];
  58 } else {
  59         $dest = 'static';
  60 }
  61
  62 $d = new DumpHTML( array(
  63         'dest' => $dest,
  64         'forceCopy' => $options['force-copy'],
  65         'alternateScriptPath' => $options['interlang'],
  66         'interwiki' => $options['interlang'],
  67 ));
  68
  69
  70 if ( $options['special'] ) {
  71         $d->doSpecials();
  72 } elseif ( $options['images'] ) {
  73         $d->doImageDescriptions();
  74 } elseif ( $options['categories'] ) {
  75         $d->doCategories();
  76 } else {
  77         print("Creating static HTML dump in directory $dest. \n".
  78                 "Starting from page_id $start of $end.\n");
  79
  80         $dbr =& wfGetDB( DB_SLAVE );
  81         print "Using database {$dbr->mServer}\n";
  82
  83         $d->doArticles( $start, $end );
  84         if ( !isset( $options['e'] ) ) {
  85                 $d->doImageDescriptions();
  86                 $d->doCategories();
  87                 $d->doSpecials();
  88         }
  89
  90         /*
  91         if ( $end - $start > CHUNK_SIZE * 2 ) {
  92                 // Split the problem into smaller chunks, run them in different PHP instances
  93                 // This is a memory/resource leak workaround
  94                 print("Creating static HTML dump in directory $dest. \n".
  95                         "Starting from page_id $start of $end.\n");
  96
  97                 chdir( "maintenance" );
  98                 for ( $chunkStart = $start; $chunkStart < $end; $chunkStart += CHUNK_SIZE ) {
  99                         $chunkEnd = $chunkStart + CHUNK_SIZE - 1;
 100                         if ( $chunkEnd > $end ) {
 101                                 $chunkEnd = $end;
 102                         }
 103                         passthru( "php dumpHTML.php -d " . wfEscapeShellArg( $dest ) . " -s $chunkStart -e $chunkEnd" );
 104                 }
 105                 chdir( ".." );
 106                 $d->doImageDescriptions();
 107                 $d->doCategories();
 108                 $d->doMainPage( $dest );
 109         } else {
 110                 $d->doArticles( $start, $end );
 111         }
 112         */
 113 }
 114
 115 if ( isset( $options['debug'] ) ) {
 116         print_r($GLOBALS);
 117 }
 118
 119 if ( $profiling ) {
 120         echo $wgProfiler->getOutput();
 121 }
 122
 123 ?>