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 htmldump)
  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  * --image-snapshot   copy all images used to the destination directory
  23  */
  24
  25
  26 $optionsWithArgs = array( 's', 'd', 'e', 'k' );
  27
  28 $profiling = false;
  29
  30 if ( $profiling ) {
  31         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  32         function wfSetupDump() {
  33                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  34                 $wgProfiling = true;
  35                 $wgProfileToDatabase = false;
  36                 $wgProfileSampleRate = 1;
  37         }
  38 }
  39
  40 require_once( "commandLine.inc" );
  41 require_once( "dumpHTML.inc" );
  42
  43 error_reporting( E_ALL & (~E_NOTICE) );
  44 define( 'CHUNK_SIZE', 50 );
  45
  46 if ( !empty( $options['s'] ) ) {
  47         $start = $options['s'];
  48 } else {
  49         $start = 1;
  50 }
  51
  52 if ( !empty( $options['e'] ) ) {
  53         $end = $options['e'];
  54 } else {
  55         $dbr =& wfGetDB( DB_SLAVE );
  56         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  57 }
  58
  59 if ( !empty( $options['d'] ) ) {
  60         $dest = $options['d'];
  61 } else {
  62         $dest = 'static';
  63 }
  64
  65 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  66
  67 $wgHTMLDump = new DumpHTML( array(
  68         'dest' => $dest,
  69         'forceCopy' => $options['force-copy'],
  70         'alternateScriptPath' => $options['interlang'],
  71         'interwiki' => $options['interlang'],
  72         'skin' => $skin,
  73         'makeSnapshot' => $options['image-snapshot'],
  74 ));
  75
  76
  77 if ( $options['special'] ) {
  78         $wgHTMLDump->doSpecials();
  79 } elseif ( $options['images'] ) {
  80         $wgHTMLDump->doImageDescriptions();
  81 } elseif ( $options['categories'] ) {
  82         $wgHTMLDump->doCategories();
  83 } elseif ( $options['redirects'] ) {
  84         $wgHTMLDump->doRedirects();
  85 } else {
  86         print("Creating static HTML dump in directory $dest. \n".
  87                 "Starting from page_id $start of $end.\n");
  88
  89         $dbr =& wfGetDB( DB_SLAVE );
  90         $server = $dbr->getProperty( 'mServer' );
  91         print "Using database {$server}\n";
  92
  93         $wgHTMLDump->doArticles( $start, $end );
  94         if ( !isset( $options['e'] ) ) {
  95                 $wgHTMLDump->doImageDescriptions();
  96                 $wgHTMLDump->doCategories();
  97                 $wgHTMLDump->doSpecials();
  98         }
  99
 100         /*
 101         if ( $end - $start > CHUNK_SIZE * 2 ) {
 102                 // Split the problem into smaller chunks, run them in different PHP instances
 103                 // This is a memory/resource leak workaround
 104                 print("Creating static HTML dump in directory $dest. \n".
 105                         "Starting from page_id $start of $end.\n");
 106
 107                 chdir( "maintenance" );
 108                 for ( $chunkStart = $start; $chunkStart < $end; $chunkStart += CHUNK_SIZE ) {
 109                         $chunkEnd = $chunkStart + CHUNK_SIZE - 1;
 110                         if ( $chunkEnd > $end ) {
 111                                 $chunkEnd = $end;
 112                         }
 113                         passthru( "php dumpHTML.php -d " . wfEscapeShellArg( $dest ) . " -s $chunkStart -e $chunkEnd" );
 114                 }
 115                 chdir( ".." );
 116                 $d->doImageDescriptions();
 117                 $d->doCategories();
 118                 $d->doMainPage( $dest );
 119         } else {
 120                 $d->doArticles( $start, $end );
 121         }
 122         */
 123 }
 124
 125 if ( isset( $options['debug'] ) ) {
 126         print_r($GLOBALS);
 127 }
 128
 129 if ( $profiling ) {
 130         echo $wgProfiler->getOutput();
 131 }
 132
 133 ?>