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  * --redirects    only do redirects
  18  * --special      only do miscellaneous stuff
  19  * --force-copy   copy commons instead of symlink, needed for Wikimedia
  20  * --interlang    allow interlanguage links
  21  */
  22
  23
  24 $optionsWithArgs = array( 's', 'd', 'e' );
  25
  26 $profiling = false;
  27
  28 if ( $profiling ) {
  29         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  30         function wfSetupDump() {
  31                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  32                 $wgProfiling = true;
  33                 $wgProfileToDatabase = false;
  34                 $wgProfileSampleRate = 1;
  35         }
  36 }
  37
  38 require_once( "commandLine.inc" );
  39 require_once( "dumpHTML.inc" );
  40
  41 error_reporting( E_ALL & (~E_NOTICE) );
  42 define( 'CHUNK_SIZE', 50 );
  43
  44 if ( !empty( $options['s'] ) ) {
  45         $start = $options['s'];
  46 } else {
  47         $start = 1;
  48 }
  49
  50 if ( !empty( $options['e'] ) ) {
  51         $end = $options['e'];
  52 } else {
  53         $dbr =& wfGetDB( DB_SLAVE );
  54         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  55 }
  56
  57 if ( !empty( $options['d'] ) ) {
  58         $dest = $options['d'];
  59 } else {
  60         $dest = 'static';
  61 }
  62
  63 $wgHTMLDump = new DumpHTML( array(
  64         'dest' => $dest,
  65         'forceCopy' => $options['force-copy'],
  66         'alternateScriptPath' => $options['interlang'],
  67         'interwiki' => $options['interlang'],
  68 ));
  69
  70
  71 if ( $options['special'] ) {
  72         $wgHTMLDump->doSpecials();
  73 } elseif ( $options['images'] ) {
  74         $wgHTMLDump->doImageDescriptions();
  75 } elseif ( $options['categories'] ) {
  76         $wgHTMLDump->doCategories();
  77 } elseif ( $options['redirects'] ) {
  78         $wgHTMLDump->doRedirects();
  79 } else {
  80         print("Creating static HTML dump in directory $dest. \n".
  81                 "Starting from page_id $start of $end.\n");
  82
  83         $dbr =& wfGetDB( DB_SLAVE );
  84         print "Using database {$dbr->mServer}\n";
  85
  86         $wgHTMLDump->doArticles( $start, $end );
  87         if ( !isset( $options['e'] ) ) {
  88                 $wgHTMLDump->doImageDescriptions();
  89                 $wgHTMLDump->doCategories();
  90                 $wgHTMLDump->doSpecials();
  91         }
  92
  93         /*
  94         if ( $end - $start > CHUNK_SIZE * 2 ) {
  95                 // Split the problem into smaller chunks, run them in different PHP instances
  96                 // This is a memory/resource leak workaround
  97                 print("Creating static HTML dump in directory $dest. \n".
  98                         "Starting from page_id $start of $end.\n");
  99
 100                 chdir( "maintenance" );
 101                 for ( $chunkStart = $start; $chunkStart < $end; $chunkStart += CHUNK_SIZE ) {
 102                         $chunkEnd = $chunkStart + CHUNK_SIZE - 1;
 103                         if ( $chunkEnd > $end ) {
 104                                 $chunkEnd = $end;
 105                         }
 106                         passthru( "php dumpHTML.php -d " . wfEscapeShellArg( $dest ) . " -s $chunkStart -e $chunkEnd" );
 107                 }
 108                 chdir( ".." );
 109                 $d->doImageDescriptions();
 110                 $d->doCategories();
 111                 $d->doMainPage( $dest );
 112         } else {
 113                 $d->doArticles( $start, $end );
 114         }
 115         */
 116 }
 117
 118 if ( isset( $options['debug'] ) ) {
 119         print_r($GLOBALS);
 120 }
 121
 122 if ( $profiling ) {
 123         echo $wgProfiler->getOutput();
 124 }
 125
 126 ?>