| Index: trunk/phase3/js2/mwEmbed/php/clrdConverter.php |
| — | — | @@ -0,0 +1,207 @@ |
| | 2 | +<?php |
| | 3 | +/* |
| | 4 | + * cldr format tries to build a structured representation of |
| | 5 | + * http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html |
| | 6 | + * |
| | 7 | + * it has structure similar to |
| | 8 | + * http://www.unicode.org/repos/cldr/tags/release-1-7-1/common/supplemental/plurals.xml |
| | 9 | + * |
| | 10 | + * It works by mapping array indexes to forms passed in via the PLURAL call |
| | 11 | + * the plural types are linear mapped to PLURAL argument index. |
| | 12 | + * |
| | 13 | + * Rules should come in the following order: |
| | 14 | + * zero, one, two, few, many, other (other is a special default case that applies to all) |
| | 15 | + * |
| | 16 | + * clrd also allows export of rule sets in json to associated javascript intepretation |
| | 17 | + */ |
| | 18 | + |
| | 19 | +class clrdConverter { |
| | 20 | + static $masterCLDR = array( |
| | 21 | + // if locale is known to have no plurals, there are no rules |
| | 22 | + array( |
| | 23 | + 'locales'=> array('az','fa','hu','ja','ko','my to','tr','vi','yo','zh', |
| | 24 | + 'bo','dz','id','jv ka','km','kn','ms','th') |
| | 25 | + ), |
| | 26 | + array( |
| | 27 | + 'locales'=> array('ar'), |
| | 28 | + 'rules' => array( |
| | 29 | + 'zero' => 0, |
| | 30 | + 'one' => 1, |
| | 31 | + 'two' => 2, |
| | 32 | + //n mod 100 in 3..10 |
| | 33 | + 'few' => array( 'mod' => 100, 'is'=>'3-10' ), |
| | 34 | + //n mod 100 in 11..99 |
| | 35 | + 'many' => array( 'mod' => 100, 'is'=>'11-99') |
| | 36 | + ) |
| | 37 | + ), |
| | 38 | + array( 'locales' => array( 'da','de','el','en','eo','es','et','fi','fo','gl', |
| | 39 | + 'he','iw','it','nb','nl','nn','no','pt_PT','sv', |
| | 40 | + 'af','bg','bn','ca','eu','fur','fy','gu','ha', |
| | 41 | + 'is','ku','lb','ml','mr','nah','ne','om','or', |
| | 42 | + 'pa','pap','ps','so','sq','sw','ta','te','tk', |
| | 43 | + 'ur','zu','mn','gsw'), |
| | 44 | + 'rules' => array( |
| | 45 | + 'one' => 1 |
| | 46 | + ) |
| | 47 | + ), |
| | 48 | + array( 'locales' => array('pt','am','bh','fil','tl','guw','hi','ln','mg','nso','ti','wa'), |
| | 49 | + 'rules'=> array( |
| | 50 | + 'one'=> '0-1' |
| | 51 | + ) |
| | 52 | + ), |
| | 53 | + array( 'locales' => array('fr'), |
| | 54 | + 'rules'=>array( |
| | 55 | + //n within 0..2 and n is not 2 |
| | 56 | + 'one' => array( 'is'=>'0-2', 'not' => 2) |
| | 57 | + ) |
| | 58 | + ), |
| | 59 | + array( 'locales' => array('lv'), |
| | 60 | + 'rules' => array( |
| | 61 | + 'zero' => 0, |
| | 62 | + //n mod 10 is 1 and n mod 100 is not 11 |
| | 63 | + 'one'=>array( |
| | 64 | + array( 'mod' => 10, 'is' => 1 ), |
| | 65 | + //AND |
| | 66 | + array( 'mod' => 100, 'not' => 11) |
| | 67 | + ) |
| | 68 | + ) |
| | 69 | + ), |
| | 70 | + array( 'locales' => array('ga','se','sma','smi','smj','smn','sms'), |
| | 71 | + 'rules' => array( |
| | 72 | + 'one' => 1, |
| | 73 | + 'two' => 2 |
| | 74 | + ) |
| | 75 | + ), |
| | 76 | + array( 'locales' => array('ro','mo'), |
| | 77 | + 'rules' => array( |
| | 78 | + 'one' => 1, |
| | 79 | + //n is 0 OR n is not 1 AND n mod 100 in 1..19 |
| | 80 | + 'few' => array( |
| | 81 | + array( 'is' => 0), |
| | 82 | + 'or', |
| | 83 | + array( |
| | 84 | + array( 'not' => 1), |
| | 85 | + array( 'mod' => 100, 'is'=>'1-19') |
| | 86 | + ) |
| | 87 | + ) |
| | 88 | + ) |
| | 89 | + ), |
| | 90 | + array( 'locales' => array( 'lt' ), |
| | 91 | + 'rules' => array( |
| | 92 | + //n mod 10 is 1 and n mod 100 not in 11..19 |
| | 93 | + 'one' => array( |
| | 94 | + array( 'mod'=>10, 'is'=> 1 ), |
| | 95 | + array( 'mod'=> 100, 'not'=> '11-19') |
| | 96 | + ), |
| | 97 | + //n mod 10 in 2..9 and n mod 100 not in 11..19 |
| | 98 | + 'few' => array( |
| | 99 | + array( 'mod'=> 10, 'is'=> '2-9' ), |
| | 100 | + array( 'mod'=> 100, 'not' => '11-19') |
| | 101 | + ), |
| | 102 | + ) |
| | 103 | + ), |
| | 104 | + array( 'locales' => array( 'hr','ru','sr','uk','be','bs','sh' ), |
| | 105 | + 'rules' => array( |
| | 106 | + //n mod 10 is 1 and n mod 100 is not 11 |
| | 107 | + 'one' => array( |
| | 108 | + array( 'mod' => 10, 'is' => 1), |
| | 109 | + array( 'mod' => 100, 'not' => 11) |
| | 110 | + ), |
| | 111 | + //n mod 10 in 2..4 and n mod 100 not in 12..14 |
| | 112 | + 'few' => array( |
| | 113 | + array( 'mod' => 10, 'is' => '2-4'), |
| | 114 | + array( 'mod' => 100, 'not' => '12-14') |
| | 115 | + ), |
| | 116 | + //n mod 10 is 0 or n mod 10 in 5..9 or n mod 100 in 11..14 |
| | 117 | + 'many' => array( |
| | 118 | + array( 'mod'=> 10, 'is' => 0), |
| | 119 | + 'or', |
| | 120 | + array( 'mod'=> 10, 'is' => '5-9'), |
| | 121 | + 'or', |
| | 122 | + array( 'mod'=> 100, 'is' => '11-14') |
| | 123 | + ), |
| | 124 | + ) |
| | 125 | + ), |
| | 126 | + array( 'locales' => array('cs','sk'), |
| | 127 | + 'rules' => array( |
| | 128 | + 'one' => 1, |
| | 129 | + 'few'=> '2-4' |
| | 130 | + ) |
| | 131 | + ), |
| | 132 | + array( 'locales' => array('pl'), |
| | 133 | + 'rules '=> array( |
| | 134 | + 'one' => 1, |
| | 135 | + 'few' => array( |
| | 136 | + //n mod 10 in 2..4 |
| | 137 | + array( 'mod' => 10, 'is' => '2-4'), |
| | 138 | + //and n mod 100 not in 12..14 |
| | 139 | + array( 'mod' => 100, 'not'=> '12-14'), |
| | 140 | + //and n mod 100 not in 22..24 |
| | 141 | + array( 'mod' => 100, 'in' => '22-24') |
| | 142 | + ) |
| | 143 | + ) |
| | 144 | + ), |
| | 145 | + array( 'locales' => array('sl'), |
| | 146 | + 'rules' => array( |
| | 147 | + 'one' => array( 'mod'=>100, 'is' => 1 ), |
| | 148 | + 'two' => array( 'mod'=>100, 'is' => 2 ), |
| | 149 | + 'few' => array( 'mod'=>100, 'is' => '3-4') |
| | 150 | + ) |
| | 151 | + ), |
| | 152 | + array( 'locales' => array('mt'), |
| | 153 | + 'rules' => array( |
| | 154 | + 'one' => 1, |
| | 155 | + //n is 0 or n mod 100 in 2..10 |
| | 156 | + 'few' => array( |
| | 157 | + array( 'is' => 0 ), |
| | 158 | + 'or', |
| | 159 | + array( 'mod' => 100, 'is' => '2-10') |
| | 160 | + ), |
| | 161 | + //n mod 100 in 11..19 |
| | 162 | + 'many' => array( 'mod'=>100, 'is' => '11-19') |
| | 163 | + ) |
| | 164 | + ), |
| | 165 | + array( 'locales' => array( 'mk' ), |
| | 166 | + 'rules' => array( |
| | 167 | + 'one' => array('mod' => 10, 'is' => '1') |
| | 168 | + ) |
| | 169 | + ), |
| | 170 | + array( 'locales' => array( 'cy' ), |
| | 171 | + 'rules' => array( |
| | 172 | + 'one' => 1, |
| | 173 | + 'two' => 2, |
| | 174 | + //n is 8 or n is 11 |
| | 175 | + 'many' => array( |
| | 176 | + array( 'is' => 8 ), |
| | 177 | + array( 'is' => 11 ) |
| | 178 | + ) |
| | 179 | + ) |
| | 180 | + ) |
| | 181 | + ); |
| | 182 | + //takes the cldr representation and returns the proper form |
| | 183 | + function cldrConvertPlural($count, $forms){ |
| | 184 | + if ( !count($forms) ) { return ''; } |
| | 185 | + //get the rule set |
| | 186 | + $ruleSet = $this->getCldrRuleSet(); |
| | 187 | + //get the number of forms (ruleSet Count + 1 for 'other' ) |
| | 188 | + $fomsCount = count( $ruleSet ) + 1; |
| | 189 | + |
| | 190 | + //if count is 1 no plurals for this language: |
| | 191 | + if( count( $forms ) == 1) |
| | 192 | + return $forms[0]; |
| | 193 | + |
| | 194 | + $forms = $this->preConvertPlural( $forms, $fomsCount ); |
| | 195 | + |
| | 196 | + } |
| | 197 | + |
| | 198 | + function getCldrRuleSet(){ |
| | 199 | + $code = $this->getCode(); |
| | 200 | + foreach($this->masterCLDR as $ruleSet){ |
| | 201 | + if( in_array($code, $ruleSet['locales']) ){ |
| | 202 | + return $ruleSet['rules']; |
| | 203 | + } |
| | 204 | + } |
| | 205 | + //could not find the language code |
| | 206 | + return false; |
| | 207 | + } |
| | 208 | +} |
| \ No newline at end of file |