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  * --no-overwrite       skip existing HTML files
  17  * --checkpoint <file>  use a checkpoint file to allow restarting of interrupted dumps
  18  * --slice <n/m>        split the job into m segments and do the n'th one
  19  * --images             only do image description pages
  20  * --categories         only do category pages
  21  * --redirects          only do redirects
  22  * --special            only do miscellaneous stuff
  23  * --force-copy         copy commons instead of symlink, needed for Wikimedia
  24  * --interlang          allow interlanguage links
  25  * --image-snapshot     copy all images used to the destination directory
  26  */
  27
  28
  29 $optionsWithArgs = array( 's', 'd', 'e', 'k', 'checkpoint', 'slice' );
  30
  31 $profiling = false;
  32
  33 if ( $profiling ) {
  34         define( 'MW_CMDLINE_CALLBACK', 'wfSetupDump' );
  35         function wfSetupDump() {
  36                 global $wgProfiling, $wgProfileToDatabase, $wgProfileSampleRate;
  37                 $wgProfiling = true;
  38                 $wgProfileToDatabase = false;
  39                 $wgProfileSampleRate = 1;
  40         }
  41 }
  42
  43 require_once( "commandLine.inc" );
  44 require_once( "dumpHTML.inc" );
  45
  46 error_reporting( E_ALL & (~E_NOTICE) );
  47
  48 if ( !empty( $options['s'] ) ) {
  49         $start = $options['s'];
  50 } else {
  51         $start = 1;
  52 }
  53
  54 if ( !empty( $options['e'] ) ) {
  55         $end = $options['e'];
  56 } else {
  57         $dbr =& wfGetDB( DB_SLAVE );
  58         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  59 }
  60
  61 if ( !empty( $options['d'] ) ) {
  62         $dest = $options['d'];
  63 } else {
  64         $dest = "$IP/static";
  65 }
  66
  67 $skin = isset( $options['k'] ) ? $options['k'] : 'htmldump';
  68
  69 if ( $options['slice'] ) {
  70         $bits = explode( '/', $options['slice'] );
  71         if ( count( $bits ) != 2 || $bits[0] < 1 || $bits[0] > $bits[1] ) {
  72                 print "Invalid slice specification";
  73                 exit;
  74         }
  75         $sliceNumerator = $bits[0];
  76         $sliceDenominator = $bits[1];
  77 } else {
  78         $sliceNumerator = $sliceDenominator = 1;
  79 }
  80
  81 $wgHTMLDump = new DumpHTML( array(
  82         'dest' => $dest,
  83         'forceCopy' => $options['force-copy'],
  84         'alternateScriptPath' => $options['interlang'],
  85         'interwiki' => $options['interlang'],
  86         'skin' => $skin,
  87         'makeSnapshot' => $options['image-snapshot'],
  88         'checkpointFile' => $options['checkpoint'],
  89         'startID' => $start,
  90         'endID' => $end,
  91         'sliceNumerator' => $sliceNumerator,
  92         'sliceDenominator' => $sliceDenominator,
  93         'noOverwrite' => $options['no-overwrite'],
  94 ));
  95
  96
  97 if ( $options['special'] ) {
  98         $wgHTMLDump->doSpecials();
  99 } elseif ( $options['images'] ) {
 100         $wgHTMLDump->doImageDescriptions();
 101 } elseif ( $options['categories'] ) {
 102         $wgHTMLDump->doCategories();
 103 } elseif ( $options['redirects'] ) {
 104         $wgHTMLDump->doRedirects();
 105 } else {
 106         print "Creating static HTML dump in directory $dest. \n";
 107         $dbr =& wfGetDB( DB_SLAVE );
 108         $server = $dbr->getProperty( 'mServer' );
 109         print "Using database {$server}\n";
 110
 111         if ( !isset( $options['e'] ) ) {
 112                 $wgHTMLDump->doEverything();
 113         } else {
 114                 $wgHTMLDump->doArticles();
 115         }
 116 }
 117
 118 if ( isset( $options['debug'] ) ) {
 119         print_r($GLOBALS);
 120 }
 121
 122 if ( $profiling ) {
 123         echo $wgProfiler->getOutput();
 124 }
 125
 126 ?>