| Index: trunk/extensions/Translate/scripts/plural-comparison.php |
| — | — | @@ -0,0 +1,204 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | + * Script for comparing different plural implementations. |
| | 5 | + * |
| | 6 | + * @author Niklas Laxstrom |
| | 7 | + * |
| | 8 | + * @copyright Copyright © 2010, Niklas Laxström |
| | 9 | + * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License 2.0 or later |
| | 10 | + * @file |
| | 11 | + */ |
| | 12 | + |
| | 13 | +// Standard boilerplate to define $IP |
| | 14 | +if ( getenv( 'MW_INSTALL_PATH' ) !== false ) { |
| | 15 | + $IP = getenv( 'MW_INSTALL_PATH' ); |
| | 16 | +} else { |
| | 17 | + $dir = dirname( __FILE__ ); $IP = "$dir/../../.."; |
| | 18 | +} |
| | 19 | +require_once( "$IP/maintenance/Maintenance.php" ); |
| | 20 | + |
| | 21 | +/// Script for comparing different plural implementations. |
| | 22 | +class PluralCompare extends Maintenance { |
| | 23 | + public function __construct() { |
| | 24 | + parent::__construct(); |
| | 25 | + $this->mDescription = 'Script for comparing different plural implementations.'; |
| | 26 | + } |
| | 27 | + |
| | 28 | + public function execute() { |
| | 29 | + $mwLanguages = $this->loadMediaWiki(); |
| | 30 | + $gtLanguages = $this->loadGettext(); |
| | 31 | + $clLanguages = $this->loadCLDR(); |
| | 32 | + |
| | 33 | + $allkeys = array_keys( $mwLanguages + $gtLanguages + $clLanguages ); |
| | 34 | + sort( $allkeys ); |
| | 35 | + |
| | 36 | + $this->output( sprintf( "%12s %3s %3s %4s\n", 'Code', 'MW', 'Get', 'CLDR' ) ); |
| | 37 | + foreach( $allkeys as $index => $code ) { |
| | 38 | + $mw = isset( $mwLanguages[$code] ) ? ( $mwLanguages[$code] === false ? '.' : '+' ) : ''; |
| | 39 | + $gt = isset( $gtLanguages[$code] ) ? ( $gtLanguages[$code] === '(n != 1)' ? '.' : '+' ) : ''; |
| | 40 | + $cl = isset( $clLanguages[$code] ) ? ( $clLanguages[$code][0] === 'Default' ? '.' : '+' ) : ''; |
| | 41 | + $this->output( sprintf( "%12s %-3s %-3s %-4s\n", $code, $mw, $gt, $cl ) ); |
| | 42 | + |
| | 43 | + if ( substr_count( sprintf( '%s%s%s', $mw, $gt, $cl ), '+' ) < 2 ) { |
| | 44 | + unset( $allkeys[$index] ); |
| | 45 | + } |
| | 46 | + } |
| | 47 | + |
| | 48 | + $this->output( "\n" ); |
| | 49 | + $c = count( $allkeys ); |
| | 50 | + $this->output( "Proceeding to test differences in $c languages\n" ); |
| | 51 | + |
| | 52 | + foreach ( $allkeys as $code ) { |
| | 53 | + $output = sprintf( "%3s %3s %3s %4s for [$code]\n", 'I', 'MW', 'Get', 'CLDR' ); |
| | 54 | + |
| | 55 | + if ( isset( $mwLanguages[$code] ) && $mwLanguages[$code] !== false ) { |
| | 56 | + $obj = Language::factory( $code ); |
| | 57 | + } else { |
| | 58 | + $obj = false; |
| | 59 | + } |
| | 60 | + |
| | 61 | + if ( isset( $gtLanguages[$code] ) ) { |
| | 62 | + $gtExp = 'return (int) ' . str_replace( 'n', '$i', $gtLanguages[$code] ) . ';'; |
| | 63 | + } else { |
| | 64 | + $gtExp = false; |
| | 65 | + } |
| | 66 | + |
| | 67 | + if ( isset( $clLanguages[$code] ) ) { |
| | 68 | + $cldrExp = $clLanguages[$code][1]; |
| | 69 | + } else { |
| | 70 | + $cldrExp = false; |
| | 71 | + } |
| | 72 | + |
| | 73 | + $cldrmap = array(); |
| | 74 | + $error = false; |
| | 75 | + |
| | 76 | + for( $i = 0; $i <= 200; $i++ ) { |
| | 77 | + $mw = $obj ? $obj->convertPlural( $i, array( 0, 1, 2, 3, 4, 5 ) ) : '?'; |
| | 78 | + $gt = $gtExp ? eval( $gtExp ) : '?'; |
| | 79 | + $cldr = $cldrExp !== false ? $this->evalCLDRRule( $i, $cldrExp ) : '?'; |
| | 80 | + |
| | 81 | + if ( self::comp( $mw, $gt ) ) { |
| | 82 | + $value = $gt !== '?' ? $gt : $mw; |
| | 83 | + if ( !isset( $cldrmap[$cldr] ) ) { |
| | 84 | + $cldrmap[$cldr] = $value; |
| | 85 | + if ( $cldr !== '?' ) { |
| | 86 | + $output .= sprintf( "%3s %-3s %-3s %-6s # Established that %-6s == $mw\n", $i, $mw, $gt, $cldr, $cldr ); |
| | 87 | + } |
| | 88 | + continue; |
| | 89 | + } elseif ( self::comp( $cldrmap[$cldr], $value ) ) { |
| | 90 | + continue; |
| | 91 | + } elseif ( $i > 4 && $value === 1 && self::comp( $cldr, 'other' ) ) { |
| | 92 | + if ( $i === 5 ) { |
| | 93 | + $output .= "Supressing further output for this language.\n"; |
| | 94 | + } |
| | 95 | + continue; |
| | 96 | + } |
| | 97 | + } |
| | 98 | + $error = true; |
| | 99 | + $output .= sprintf( "%3s %-3s %-3s %-6s\n", $i, $mw, $gt, $cldr ); |
| | 100 | + } |
| | 101 | + |
| | 102 | + if ( $error ) { |
| | 103 | + $this->output( "$output\n" ); |
| | 104 | + } |
| | 105 | + } |
| | 106 | + |
| | 107 | + } |
| | 108 | + |
| | 109 | + public static function comp( $a, $b ) { |
| | 110 | + return $a === '?' || $b === '?' || $a === $b; |
| | 111 | + } |
| | 112 | + |
| | 113 | + public function loadCLDR() { |
| | 114 | + $filename = dirname( __FILE__ ) . '/../data/plural-cldr.yaml'; |
| | 115 | + $data = TranslateYaml::load( $filename ); |
| | 116 | + $languages = array(); |
| | 117 | + $ruleExps = array(); |
| | 118 | + foreach ( $data['rulesets'] as $name => $rules ) { |
| | 119 | + $ruleExps[$name] = array(); |
| | 120 | + foreach( $rules as $rulename => $rule ) { |
| | 121 | + $ruleExps[$name][$rulename] = $this->parseCLDRRule( $rule ); |
| | 122 | + } |
| | 123 | + } |
| | 124 | + |
| | 125 | + foreach ( $data['locales'] as $code => $rulename ) { |
| | 126 | + $languages[$code] = array( $rulename, $ruleExps[$rulename] ); |
| | 127 | + } |
| | 128 | + |
| | 129 | + return $languages; |
| | 130 | + } |
| | 131 | + |
| | 132 | + public function loadMediaWiki() { |
| | 133 | + $mwLanguages = Language::getLanguageNames( true ); |
| | 134 | + foreach ( $mwLanguages as $code => $name ) { |
| | 135 | + $obj = Language::factory( $code ); |
| | 136 | + $method = new ReflectionMethod( $obj, 'convertPlural' ); |
| | 137 | + if ( $method->getDeclaringClass()->name === 'Language' ) { |
| | 138 | + $mwLanguages[$code] = false; |
| | 139 | + } |
| | 140 | + } |
| | 141 | + return $mwLanguages; |
| | 142 | + } |
| | 143 | + |
| | 144 | + public function loadGettext() { |
| | 145 | + $gtData = file_get_contents( dirname( __FILE__ ) . '/../data/plural-gettext.txt' ); |
| | 146 | + $gtLanguages = array(); |
| | 147 | + foreach ( preg_split( '/\n|\r/', $gtData, -1, PREG_SPLIT_NO_EMPTY ) as $line ) { |
| | 148 | + list( $code, $rule ) = explode( "\t", $line ); |
| | 149 | + $rule = preg_replace( '/^.*?plural=/', '', $rule ); |
| | 150 | + $gtLanguages[$code] = $rule; |
| | 151 | + } |
| | 152 | + return $gtLanguages; |
| | 153 | + } |
| | 154 | + |
| | 155 | + public function evalCLDRRule( $i, $rules ) { |
| | 156 | + foreach ( $rules as $name => $rule ) { |
| | 157 | + if ( eval( "return $rule;" ) ) { |
| | 158 | + return $name; |
| | 159 | + } |
| | 160 | + } |
| | 161 | + |
| | 162 | + return "other"; |
| | 163 | + } |
| | 164 | + |
| | 165 | + public function parseCLDRRule( $rule ) { |
| | 166 | + $rule = preg_replace( '/\bn\b/', '$i', $rule ); |
| | 167 | + $rule = preg_replace( '/([^ ]+) mod (\d+)/', 'self::mod(\1,\2)', $rule ); |
| | 168 | + $rule = preg_replace( '/([^ ]+) is not (\d+)/' , '\1!==\2', $rule ); |
| | 169 | + $rule = preg_replace( '/([^ ]+) is (\d+)/', '\1===\2', $rule ); |
| | 170 | + $rule = preg_replace( '/([^ ]+) not in (\d+)\.\.(\d+)/', '!self::in(\1,\2,\3)', $rule ); |
| | 171 | + $rule = preg_replace( '/([^ ]+) not within (\d+)\.\.(\d+)/', '!self::within(\1,\2,\3)', $rule ); |
| | 172 | + $rule = preg_replace( '/([^ ]+) in (\d+)\.\.(\d+)/', 'self::in(\1,\2,\3)', $rule ); |
| | 173 | + $rule = preg_replace( '/([^ ]+) within (\d+)\.\.(\d+)/', 'self::within(\1,\2,\3)', $rule ); |
| | 174 | + // AND takes precedence over OR |
| | 175 | + $andrule = '/([^ ]+) and ([^ ]+)/i'; |
| | 176 | + while( preg_match( $andrule, $rule ) ) { |
| | 177 | + $rule = preg_replace( $andrule, '(\1&&\2)', $rule ); |
| | 178 | + } |
| | 179 | + $orrule = '/([^ ]+) or ([^ ]+)/i'; |
| | 180 | + while( preg_match( $orrule, $rule ) ) { |
| | 181 | + $rule = preg_replace( $orrule, '(\1||\2)', $rule ); |
| | 182 | + } |
| | 183 | + |
| | 184 | + return $rule; |
| | 185 | + } |
| | 186 | + |
| | 187 | + public static function in( $num, $low, $high ) { |
| | 188 | + return is_int( $num ) && $num >= $low && $num <= $high; |
| | 189 | + } |
| | 190 | + |
| | 191 | + public static function within( $num, $low, $high ) { |
| | 192 | + return $num >= $low && $num <= $high; |
| | 193 | + } |
| | 194 | + |
| | 195 | + public static function mod( $num, $mod ) { |
| | 196 | + if ( is_int( $num ) ) { |
| | 197 | + return (int) fmod( $num, $mod ); |
| | 198 | + } |
| | 199 | + return fmod( $num, $mod ); |
| | 200 | + } |
| | 201 | + |
| | 202 | +} |
| | 203 | + |
| | 204 | +$maintClass = 'PluralCompare'; |
| | 205 | +require_once( DO_MAINTENANCE ); |
| Property changes on: trunk/extensions/Translate/scripts/plural-comparison.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 206 | + native |