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 if( isset($options['help']) ) { usage(); die(); }
  17 // default output is WikiText
  18 if( !isset($options['output']) ) { $options['output']='wiki'; }
  19
  20
  21 /** Print a usage message*/
  22 function usage() {
  23 print <<<END
  24 Usage: php transstat.php [--help] [--output:csv|text|wiki] [--noredundant]
  25          --help : this helpful message
  26        --output : select an output engine one of:
  27                     * 'csv'  : Comma Separated Values.
  28                     * 'wiki' : MediaWiki syntax.
  29                     * 'text' : Text with tabs.
  30                       Default output is 'wiki'
  31   --noredundant : do NOT calculate redundant (takes some time)
  32
  33
  34 END;
  35 }
  36
  37
  38 /** A general output object. Need to be overriden */
  39 class statsOutput {
  40         var $output; // buffer that contain the text
  41         function statsOutput() { $this->output='';}
  42         function getContent() { return $this->output;}
  43
  44         function formatPercent($subset, $total, $revert=false, $accuracy=2) {
  45                 return @sprintf( '%.' . $accuracy . 'f%%', 100 * $subset / $total );
  46         }
  47
  48         // Override the next methods
  49         function heading() {}
  50         function footer() {}
  51         function blockstart() {}
  52         function blockend() {}
  53         function element($in, $heading=false) {}
  54 }
  55
  56
  57 /** Outputs WikiText */
  58 class wikiStatsOutput extends statsOutput {
  59         function heading() {
  60                 $this->output .= "{| border=2 cellpadding=4 cellspacing=0 style=\"background: #f9f9f9; border: 1px #aaa solid; border-collapse: collapse;\" width=100%\n";
  61         }
  62         function footer()     { $this->output .= "|}\n"; }
  63         function blockstart() { $this->output .= "|-\n"; }
  64         function blockend()   { $this->output .= ''; }
  65         function element($in, $heading = false) {
  66                 $this->output .= ($heading ? '!' : '|') . " $in\n";
  67         }
  68         function formatPercent($subset, $total, $revert=false, $accuracy=2) {
  69                 $v = @round(255 * $subset / $total);
  70                 if($revert) $v = 255 - $v;
  71                 if($v < 128) {
  72                         // red to yellow
  73                         $red = 'FF';
  74                         $green = sprintf('%02X', 2*$v);
  75                 } else {
  76                         // yellow to green
  77                         $red   = sprintf('%02X', 2*(255 -$v) );
  78                         $green = 'FF';
  79                 }
  80                 $blue  = '00';
  81                 $color = $red.$green.$blue;
  82
  83                 $percent = statsOutput::formatPercent($subset, $total, $revert, $accuracy);
  84                 return 'bgcolor="#'.$color.'" | '.$percent;
  85         }
  86 }
  87
  88 /** Output text. To be used on a terminal for example. */
  89 class textStatsOutput extends statsOutput {
  90         function element($in, $heading = false) {
  91                 $this->output .= $in."\t";
  92         }
  93         function blockend(){ $this->output .="\n";}
  94 }
  95
  96 /** csv output. Some people love excel */
  97 class csvStatsOutput extends statsOutput {
  98         function element($in, $heading = false) {
  99                 $this->output .= $in.";";
 100         }
 101         function blockend(){ $this->output .="\n";}
 102 }
 103
 104
 105 /** FIXME: This takes an obscene amount of time */
 106 if(isset($options['noredundant'])) {
 107         function redundant(&$arr) { return 'NC'; }
 108 } else {
 109         function redundant(&$arr) {
 110                 global $wgAllMessagesEn;
 111
 112                 $redundant = 0;
 113                 foreach(array_keys($arr) as $key) {
 114                         if ( ! array_key_exists( $key, $wgAllMessagesEn) )
 115                                 ++$redundant;
 116                 }
 117                 return $redundant;
 118         }
 119 }
 120
 121 // Select an output engine
 122 switch ($options['output']) {
 123         case 'csv':
 124                 $out = new csvStatsOutput(); break;
 125         case 'text':
 126                 $out = new textStatsOutput(); break;
 127         case 'wiki':
 128                 $out = new wikiStatsOutput(); break;
 129         default:
 130                 usage(); die();
 131         break;
 132 }
 133
 134 // available language files
 135 $langs = array();
 136 $dir = opendir("$IP/languages");
 137 while ($file = readdir($dir)) {
 138         if (preg_match("/Language(.*?)\.php$/", $file, $m)) {
 139                 $langs[] = $m[1];
 140         }
 141 }
 142 sort($langs);
 143
 144 // Cleanup file list
 145 foreach($langs as $key => $lang) {
 146         if ($lang == 'Utf8' || $lang == '' || $lang == 'Converter')
 147                 unset($langs[$key]);
 148 }
 149
 150 //  Load message and compute stuff
 151 $msgs = array();
 152 foreach($langs as $lang) {
 153         // Since they aren't loaded by default..
 154         require_once( 'languages/Language' . $lang . '.php' );
 155         $arr = 'wgAllMessages' . $lang;
 156         if (@is_array($$arr)) { // Some of them don't have a message array
 157                 $msgs[$wgContLang->lcfirst($lang)] = array(
 158                         'total' => count($$arr),
 159                         'redundant' => redundant($$arr),
 160                 );
 161         } else {
 162                 $msgs[$wgContLang->lcfirst($lang)] = array(
 163                         'total' => 0,
 164                         'redundant' => 0,
 165                 );
 166         }
 167 }
 168
 169 // Top entry
 170 $out->heading();
 171 $out->blockstart();
 172 $out->element('Language', true);
 173 $out->element('Translated', true);
 174 $out->element('%', true);
 175 $out->element('Untranslated', true);
 176 $out->element('%', true);
 177 $out->element('Redundant', true);
 178 $out->element('%', true);
 179 $out->blockend();
 180
 181 // Generate rows
 182 foreach($msgs as $lang => $stats) {
 183         $out->blockstart();
 184         $out->element($wgContLang->getLanguageName(strtr($lang, '_', '-')) . " ($lang)"); // Language
 185         $out->element($stats['total'] . '/' . $msgs['en']['total']); // Translated
 186         $out->element($out->formatPercent($stats['total'], $msgs['en']['total'])); // % Translated
 187         $out->element($msgs['en']['total'] - $stats['total']); // Untranslated
 188         $out->element($out->formatPercent($msgs['en']['total'] - $stats['total'], $msgs['en']['total'], true)); // % Untranslated
 189         if($stats['redundant'] =='NC') {
 190                 $out->element('NC');
 191                 $out->element('NC');
 192         } else {
 193                 $out->element($stats['redundant'] . '/' . $stats['total']); // Redundant
 194                 $out->element($out->formatPercent($stats['redundant'],  $stats['total'],true)); // % Redundant
 195         }
 196         $out->blockend();
 197 }
 198 $out->footer();
 199
 200 // Final output
 201 echo $out->getContent();
 202 ?>