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  * -k <skin>      skin to use (defaults to dumphtml)
  16  * --images       only do image description pages
  17  * --categories   only do category pages
  18  * --redirects    only do redirects
  19  * --special      only do miscellaneous stuff
  20  * --force-copy   copy commons instead of symlink, needed for Wikimedia
  21  * --interlang    allow interlanguage links
  22  */
  23
  24
  25 $optionsWithArgs = array( 's', 'd', 'e', 'k' );
  26
  27 $profiling = false;
  28
  29 if ( $profiling ) {
  30         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  31         function wfSetupDump() {
  32                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  33                 $wgProfiling = true;
  34                 $wgProfileToDatabase = false;
  35                 $wgProfileSampleRate = 1;
  36         }
  37 }
  38
  39 require_once( "commandLine.inc" );
  40 require_once( "dumpHTML.inc" );
  41
  42 error_reporting( E_ALL & (~E_NOTICE) );
  43 define( 'CHUNK_SIZE', 50 );
  44
  45 if ( !empty( $options['s'] ) ) {
  46         $start = $options['s'];
  47 } else {
  48         $start = 1;
  49 }
  50
  51 if ( !empty( $options['e'] ) ) {
  52         $end = $options['e'];
  53 } else {
  54         $dbr =& wfGetDB( DB_SLAVE );
  55         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  56 }
  57
  58 if ( !empty( $options['d'] ) ) {
  59         $dest = $options['d'];
  60 } else {
  61         $dest = 'static';
  62 }
  63
  64 $skin = isset( $options['k'] ) ? $options['k'] : 'dumphtml';
  65
  66 $wgHTMLDump = new DumpHTML( array(
  67         'dest' => $dest,
  68         'forceCopy' => $options['force-copy'],
  69         'alternateScriptPath' => $options['interlang'],
  70         'interwiki' => $options['interlang'],
  71         'skin' => $skin,
  72 ));
  73
  74
  75 if ( $options['special'] ) {
  76         $wgHTMLDump->doSpecials();
  77 } elseif ( $options['images'] ) {
  78         $wgHTMLDump->doImageDescriptions();
  79 } elseif ( $options['categories'] ) {
  80         $wgHTMLDump->doCategories();
  81 } elseif ( $options['redirects'] ) {
  82         $wgHTMLDump->doRedirects();
  83 } else {
  84         print("Creating static HTML dump in directory $dest. \n".
  85                 "Starting from page_id $start of $end.\n");
  86
  87         $dbr =& wfGetDB( DB_SLAVE );
  88         $server = $dbr->getProperty( 'mServer' );
  89         print "Using database {$server}\n";
  90
  91         $wgHTMLDump->doArticles( $start, $end );
  92         if ( !isset( $options['e'] ) ) {
  93                 $wgHTMLDump->doImageDescriptions();
  94                 $wgHTMLDump->doCategories();
  95                 $wgHTMLDump->doSpecials();
  96         }
  97
  98         /*
  99         if ( $end - $start > CHUNK_SIZE * 2 ) {
 100                 // Split the problem into smaller chunks, run them in different PHP instances
 101                 // This is a memory/resource leak workaround
 102                 print("Creating static HTML dump in directory $dest. \n".
 103                         "Starting from page_id $start of $end.\n");
 104
 105                 chdir( "maintenance" );
 106                 for ( $chunkStart = $start; $chunkStart < $end; $chunkStart += CHUNK_SIZE ) {
 107                         $chunkEnd = $chunkStart + CHUNK_SIZE - 1;
 108                         if ( $chunkEnd > $end ) {
 109                                 $chunkEnd = $end;
 110                         }
 111                         passthru( "php dumpHTML.php -d " . wfEscapeShellArg( $dest ) . " -s $chunkStart -e $chunkEnd" );
 112                 }
 113                 chdir( ".." );
 114                 $d->doImageDescriptions();
 115                 $d->doCategories();
 116                 $d->doMainPage( $dest );
 117         } else {
 118                 $d->doArticles( $start, $end );
 119         }
 120         */
 121 }
 122
 123 if ( isset( $options['debug'] ) ) {
 124         print_r($GLOBALS);
 125 }
 126
 127 if ( $profiling ) {
 128         echo $wgProfiler->getOutput();
 129 }
 130
 131 ?>