maintenance/transstat.php

   1 <?php
   2 /**
   3  * @package MediaWiki
   4  * @subpackage Maintenance
   5  *
   6  * @author Ævar Arnfjörð Bjarmason <avarab@gmail.com>
   7  * @author Ashar Voultoiz <thoane@altern.org>
   8  * @bug 2499
   9  *
  10  * Output is posted from time to time on:
  11  * http://meta.wikimedia.org/wiki/Localization_statistics
  12  */
  13
  14 /** */
  15 require_once('commandLine.inc');
  16 require_once('languages.inc');
  17
  18 if( isset($options['help']) ) { usage(); die(); }
  19 // default output is WikiText
  20 if( !isset($options['output']) ) { $options['output']='wiki'; }
  21
  22
  23 /** Print a usage message*/
  24 function usage() {
  25 print <<<END
  26 Usage: php transstat.php [--help] [--output:csv|text|wiki] [--noredundant]
  27          --help : this helpful message
  28        --output : select an output engine one of:
  29                     * 'csv'  : Comma Separated Values.
  30                     * 'wiki' : MediaWiki syntax.
  31                     * 'text' : Text with tabs.
  32                       Default output is 'wiki'
  33   --noredundant : do NOT calculate redundant (takes some time)
  34
  35
  36 END;
  37 }
  38
  39
  40 /** A general output object. Need to be overriden */
  41 class statsOutput {
  42         var $output; // buffer that contain the text
  43         function statsOutput() { $this->output='';}
  44         function getContent() { return $this->output;}
  45
  46         function formatPercent($subset, $total, $revert=false, $accuracy=2) {
  47                 return @sprintf( '%.' . $accuracy . 'f%%', 100 * $subset / $total );
  48         }
  49
  50         // Override the next methods
  51         function heading() {}
  52         function footer() {}
  53         function blockstart() {}
  54         function blockend() {}
  55         function element($in, $heading=false) {}
  56 }
  57
  58
  59 /** Outputs WikiText */
  60 class wikiStatsOutput extends statsOutput {
  61         function heading() {
  62                 $this->output .= "{| border=2 cellpadding=4 cellspacing=0 style=\"background: #f9f9f9; border: 1px #aaa solid; border-collapse: collapse;\" width=100%\n";
  63         }
  64         function footer()     { $this->output .= "|}\n"; }
  65         function blockstart() { $this->output .= "|-\n"; }
  66         function blockend()   { $this->output .= ''; }
  67         function element($in, $heading = false) {
  68                 $this->output .= ($heading ? '!' : '|') . " $in\n";
  69         }
  70         function formatPercent($subset, $total, $revert=false, $accuracy=2) {
  71                 $v = @round(255 * $subset / $total);
  72                 if($revert) $v = 255 - $v;
  73                 if($v < 128) {
  74                         // red to yellow
  75                         $red = 'FF';
  76                         $green = sprintf('%02X', 2*$v);
  77                 } else {
  78                         // yellow to green
  79                         $red   = sprintf('%02X', 2*(255 -$v) );
  80                         $green = 'FF';
  81                 }
  82                 $blue  = '00';
  83                 $color = $red.$green.$blue;
  84
  85                 $percent = statsOutput::formatPercent($subset, $total, $revert, $accuracy);
  86                 return 'bgcolor="#'.$color.'" | '.$percent;
  87         }
  88 }
  89
  90 /** Output text. To be used on a terminal for example. */
  91 class textStatsOutput extends statsOutput {
  92         function element($in, $heading = false) {
  93                 $this->output .= $in."\t";
  94         }
  95         function blockend(){ $this->output .="\n";}
  96 }
  97
  98 /** csv output. Some people love excel */
  99 class csvStatsOutput extends statsOutput {
 100         function element($in, $heading = false) {
 101                 $this->output .= $in.";";
 102         }
 103         function blockend(){ $this->output .="\n";}
 104 }
 105
 106
 107 /** FIXME: This takes an obscene amount of time */
 108 if(isset($options['noredundant'])) {
 109         function redundant(&$arr) { return 'NC'; }
 110 } else {
 111         function redundant(&$arr) {
 112                 global $wgAllMessagesEn;
 113
 114                 $redundant = 0;
 115                 foreach(array_keys($arr) as $key) {
 116                         if ( ! array_key_exists( $key, $wgAllMessagesEn) )
 117                                 ++$redundant;
 118                 }
 119                 return $redundant;
 120         }
 121 }
 122
 123 // Select an output engine
 124 switch ($options['output']) {
 125         case 'csv':
 126                 $out = new csvStatsOutput(); break;
 127         case 'text':
 128                 $out = new textStatsOutput(); break;
 129         case 'wiki':
 130                 $out = new wikiStatsOutput(); break;
 131         default:
 132                 usage(); die();
 133         break;
 134 }
 135
 136 $langTool = new languages();
 137
 138 //  Load message and compute stuff
 139 $msgs = array();
 140 foreach($langTool->getList() as $langcode) {
 141         // Since they aren't loaded by default..
 142         require_once( 'languages/Language' . $langcode . '.php' );
 143         $arr = 'wgAllMessages'.$langcode;
 144         if(@is_array($$arr)) {
 145                 $msgs[$wgContLang->lcfirst($langcode)] = array(
 146                         'total' => count($$arr),
 147                         'redundant' => redundant($$arr),
 148                 );
 149         } else {
 150                 $msgs[$wgContLang->lcfirst($langcode)] = array(
 151                         'total' => 0,
 152                         'redundant' => 0,
 153                 );
 154         }
 155 }
 156
 157 // Top entry
 158 $out->heading();
 159 $out->blockstart();
 160 $out->element('Language', true);
 161 $out->element('Translated', true);
 162 $out->element('%', true);
 163 $out->element('Untranslated', true);
 164 $out->element('%', true);
 165 $out->element('Redundant', true);
 166 $out->element('%', true);
 167 $out->blockend();
 168
 169 // Generate rows
 170 foreach($msgs as $lang => $stats) {
 171         $out->blockstart();
 172         // Language
 173         $out->element($wgContLang->getLanguageName(strtr($lang, '_', '-')) . " ($lang)");
 174         // Translated
 175         $out->element($stats['total'] . '/' . $msgs['en']['total']);
 176         // % Translated
 177         $out->element($out->formatPercent($stats['total'], $msgs['en']['total']));
 178         // Untranslated
 179         $out->element($msgs['en']['total'] - $stats['total']);
 180         // % Untranslated
 181         $out->element($out->formatPercent($msgs['en']['total'] - $stats['total'], $msgs['en']['total'], true));
 182         // Redundant & % Redundant
 183         if($stats['redundant'] =='NC') {
 184                 $out->element('NC');
 185                 $out->element('NC');
 186         } else {
 187                 $out->element($stats['redundant'] . '/' . $stats['total']);
 188                 $out->element($out->formatPercent($stats['redundant'],  $stats['total'],true));
 189         }
 190         $out->blockend();
 191 }
 192 $out->footer();
 193
 194 // Final output
 195 echo $out->getContent();
 196 ?>