includes/parser/DateFormatter.php

   1 <?php
   2
   3 /**
   4  * Date formatter, recognises dates in plain text and formats them accoding to user preferences.
   5  * @todo preferences, OutputPage
   6  * @ingroup Parser
   7  */
   8 class DateFormatter
   9 {
  10         var $mSource, $mTarget;
  11         var $monthNames = '', $rxDM, $rxMD, $rxDMY, $rxYDM, $rxMDY, $rxYMD;
  12
  13         var $regexes, $pDays, $pMonths, $pYears;
  14         var $rules, $xMonths, $preferences;
  15
  16         const ALL = -1;
  17         const NONE = 0;
  18         const MDY = 1;
  19         const DMY = 2;
  20         const YMD = 3;
  21         const ISO1 = 4;
  22         const LASTPREF = 4;
  23         const ISO2 = 5;
  24         const YDM = 6;
  25         const DM = 7;
  26         const MD = 8;
  27         const LAST = 8;
  28
  29         /**
  30          * @todo document
  31          */
  32         function DateFormatter() {
  33                 global $wgContLang;
  34
  35                 $this->monthNames = $this->getMonthRegex();
  36                 for ( $i=1; $i<=12; $i++ ) {
  37                         $this->xMonths[$wgContLang->lc( $wgContLang->getMonthName( $i ) )] = $i;
  38                         $this->xMonths[$wgContLang->lc( $wgContLang->getMonthAbbreviation( $i ) )] = $i;
  39                 }
  40
  41                 $this->regexTrail = '(?![a-z])/iu';
  42
  43                 # Partial regular expressions
  44                 $this->prxDM = '\[\[(\d{1,2})[ _](' . $this->monthNames . ')\]\]';
  45                 $this->prxMD = '\[\[(' . $this->monthNames . ')[ _](\d{1,2})\]\]';
  46                 $this->prxY = '\[\[(\d{1,4}([ _]BC|))\]\]';
  47                 $this->prxISO1 = '\[\[(-?\d{4})]]-\[\[(\d{2})-(\d{2})\]\]';
  48                 $this->prxISO2 = '\[\[(-?\d{4})-(\d{2})-(\d{2})\]\]';
  49
  50                 # Real regular expressions
  51                 $this->regexes[self::DMY] = "/{$this->prxDM}(?: *, *| +){$this->prxY}{$this->regexTrail}";
  52                 $this->regexes[self::YDM] = "/{$this->prxY}(?: *, *| +){$this->prxDM}{$this->regexTrail}";
  53                 $this->regexes[self::MDY] = "/{$this->prxMD}(?: *, *| +){$this->prxY}{$this->regexTrail}";
  54                 $this->regexes[self::YMD] = "/{$this->prxY}(?: *, *| +){$this->prxMD}{$this->regexTrail}";
  55                 $this->regexes[self::DM] = "/{$this->prxDM}{$this->regexTrail}";
  56                 $this->regexes[self::MD] = "/{$this->prxMD}{$this->regexTrail}";
  57                 $this->regexes[self::ISO1] = "/{$this->prxISO1}{$this->regexTrail}";
  58                 $this->regexes[self::ISO2] = "/{$this->prxISO2}{$this->regexTrail}";
  59
  60                 # Extraction keys
  61                 # See the comments in replace() for the meaning of the letters
  62                 $this->keys[self::DMY] = 'jFY';
  63                 $this->keys[self::YDM] = 'Y jF';
  64                 $this->keys[self::MDY] = 'FjY';
  65                 $this->keys[self::YMD] = 'Y Fj';
  66                 $this->keys[self::DM] = 'jF';
  67                 $this->keys[self::MD] = 'Fj';
  68                 $this->keys[self::ISO1] = 'ymd'; # y means ISO year
  69                 $this->keys[self::ISO2] = 'ymd';
  70
  71                 # Target date formats
  72                 $this->targets[self::DMY] = '[[F j|j F]] [[Y]]';
  73                 $this->targets[self::YDM] = '[[Y]], [[F j|j F]]';
  74                 $this->targets[self::MDY] = '[[F j]], [[Y]]';
  75                 $this->targets[self::YMD] = '[[Y]] [[F j]]';
  76                 $this->targets[self::DM] = '[[F j|j F]]';
  77                 $this->targets[self::MD] = '[[F j]]';
  78                 $this->targets[self::ISO1] = '[[Y|y]]-[[F j|m-d]]';
  79                 $this->targets[self::ISO2] = '[[y-m-d]]';
  80
  81                 # Rules
  82                 #            pref    source       target
  83                 $this->rules[self::DMY][self::MD]       = self::DM;
  84                 $this->rules[self::ALL][self::MD]       = self::MD;
  85                 $this->rules[self::MDY][self::DM]       = self::MD;
  86                 $this->rules[self::ALL][self::DM]       = self::DM;
  87                 $this->rules[self::NONE][self::ISO2]    = self::ISO1;
  88
  89                 $this->preferences = array(
  90                         'default' => self::NONE,
  91                         'dmy' => self::DMY,
  92                         'mdy' => self::MDY,
  93                         'ymd' => self::YMD,
  94                         'ISO 8601' => self::ISO1,
  95                 );
  96         }
  97
  98         /**
  99          * Get a DateFormatter object
 100          *
 101          * @return DateFormatter object
 102          */
 103         public static function &getInstance() {
 104                 global $wgMemc;
 105                 static $dateFormatter = false;
 106                 if ( !$dateFormatter ) {
 107                         $dateFormatter = $wgMemc->get( wfMemcKey( 'dateformatter' ) );
 108                         if ( !$dateFormatter ) {
 109                                 $dateFormatter = new DateFormatter;
 110                                 $wgMemc->set( wfMemcKey( 'dateformatter' ), $dateFormatter, 3600 );
 111                         }
 112                 }
 113                 return $dateFormatter;
 114         }
 115
 116         /**
 117          * @param $preference String: User preference
 118          * @param $text String: Text to reformat
 119          */
 120         function reformat( $preference, $text, $options = array('linked') ) {
 121
 122                 $linked = in_array( 'linked', $options );
 123                 $match_whole = in_array( 'match-whole', $options );
 124
 125                 if ( isset( $this->preferences[$preference] ) ) {
 126                         $preference = $this->preferences[$preference];
 127                 } else {
 128                         $preference = self::NONE;
 129                 }
 130                 for ( $i=1; $i<=self::LAST; $i++ ) {
 131                         $this->mSource = $i;
 132                         if ( isset ( $this->rules[$preference][$i] ) ) {
 133                                 # Specific rules
 134                                 $this->mTarget = $this->rules[$preference][$i];
 135                         } elseif ( isset ( $this->rules[self::ALL][$i] ) ) {
 136                                 # General rules
 137                                 $this->mTarget = $this->rules[self::ALL][$i];
 138                         } elseif ( $preference ) {
 139                                 # User preference
 140                                 $this->mTarget = $preference;
 141                         } else {
 142                                 # Default
 143                                 $this->mTarget = $i;
 144                         }
 145                         $regex = $this->regexes[$i];
 146
 147                         // Horrible hack
 148                         if (!$linked) {
 149                                 $regex = str_replace( array( '\[\[', '\]\]' ), '', $regex );
 150                         }
 151
 152                         if ($match_whole) {
 153                                 // Let's hope this works
 154                                 $regex = preg_replace( '!^/!', '/^', $regex );
 155                                 $regex = str_replace( $this->regexTrail,
 156                                         '$'.$this->regexTrail, $regex );
 157                         }
 158
 159                         // Another horrible hack
 160                         $this->mLinked = $linked;
 161                         $text = preg_replace_callback( $regex, array( &$this, 'replace' ), $text );
 162                         unset($this->mLinked);
 163                 }
 164                 return $text;
 165         }
 166
 167         /**
 168          * @param $matches
 169          */
 170         function replace( $matches ) {
 171                 # Extract information from $matches
 172                 $linked = true;
 173                 if ( isset( $this->mLinked ) )
 174                         $linked = $this->mLinked;
 175
 176                 $bits = array();
 177                 $key = $this->keys[$this->mSource];
 178                 for ( $p=0; $p < strlen($key); $p++ ) {
 179                         if ( $key{$p} != ' ' ) {
 180                                 $bits[$key{$p}] = $matches[$p+1];
 181                         }
 182                 }
 183
 184                 return $this->formatDate( $bits, $linked );
 185         }
 186
 187         function formatDate( $bits, $link = true ) {
 188                 $format = $this->targets[$this->mTarget];
 189
 190                 if (!$link) {
 191                         // strip piped links
 192                         $format = preg_replace( '/\[\[[^|]+\|([^\]]+)\]\]/', '$1', $format );
 193                         // strip remaining links
 194                         $format = str_replace( array( '[[', ']]' ), '', $format );
 195                 }
 196
 197                 # Construct new date
 198                 $text = '';
 199                 $fail = false;
 200
 201                 // Pre-generate y/Y stuff because we need the year for the <span> title.
 202                 if ( !isset( $bits['y'] ) && isset( $bits['Y'] ) )
 203                         $bits['y'] = $this->makeIsoYear( $bits['Y'] );
 204                 if ( !isset( $bits['Y'] ) && isset( $bits['y'] ) )
 205                         $bits['Y'] = $this->makeNormalYear( $bits['y'] );
 206
 207                 if ( !isset( $bits['m'] ) ) {
 208                         $m = $this->makeIsoMonth( $bits['F'] );
 209                         if ( !$m || $m == '00' ) {
 210                                 $fail = true;
 211                         } else {
 212                                 $bits['m'] = $m;
 213                         }
 214                 }
 215
 216                 if ( !isset($bits['d']) ) {
 217                         $bits['d'] = sprintf( '%02d', $bits['j'] );
 218                 }
 219
 220                 for ( $p=0; $p < strlen( $format ); $p++ ) {
 221                         $char = $format{$p};
 222                         switch ( $char ) {
 223                                 case 'd': # ISO day of month
 224                                         $text .= $bits['d'];
 225                                         break;
 226                                 case 'm': # ISO month
 227                                         $text .= $bits['m'];
 228                                         break;
 229                                 case 'y': # ISO year
 230                                         $text .= $bits['y'];
 231                                         break;
 232                                 case 'j': # ordinary day of month
 233                                         if ( !isset($bits['j']) ) {
 234                                                 $text .= intval( $bits['d'] );
 235                                         } else {
 236                                                 $text .= $bits['j'];
 237                                         }
 238                                         break;
 239                                 case 'F': # long month
 240                                         if ( !isset( $bits['F'] ) ) {
 241                                                 $m = intval($bits['m']);
 242                                                 if ( $m > 12 || $m < 1 ) {
 243                                                         $fail = true;
 244                                                 } else {
 245                                                         global $wgContLang;
 246                                                         $text .= $wgContLang->getMonthName( $m );
 247                                                 }
 248                                         } else {
 249                                                 $text .= ucfirst( $bits['F'] );
 250                                         }
 251                                         break;
 252                                 case 'Y': # ordinary (optional BC) year
 253                                         $text .= $bits['Y'];
 254                                         break;
 255                                 default:
 256                                         $text .= $char;
 257                         }
 258                 }
 259                 if ( $fail ) {
 260                         $text = $matches[0];
 261                 }
 262
 263                 $isoBits = array();
 264                 if ( isset($bits['y']) )
 265                         $isoBits[] = $bits['y'];
 266                 $isoBits[] = $bits['m'];
 267                 $isoBits[] = $bits['d'];
 268                 $isoDate = implode( '-', $isoBits );;
 269
 270                 // Output is not strictly HTML (it's wikitext), but <span> is whitelisted.
 271                 $text = Html::rawElement( 'span',
 272                                         array( 'class' => 'mw-formatted-date', 'title' => $isoDate ), $text );
 273
 274                 return $text;
 275         }
 276
 277         /**
 278          * @todo document
 279          */
 280         function getMonthRegex() {
 281                 global $wgContLang;
 282                 $names = array();
 283                 for( $i = 1; $i <= 12; $i++ ) {
 284                         $names[] = $wgContLang->getMonthName( $i );
 285                         $names[] = $wgContLang->getMonthAbbreviation( $i );
 286                 }
 287                 return implode( '|', $names );
 288         }
 289
 290         /**
 291          * Makes an ISO month, e.g. 02, from a month name
 292          * @param $monthName String: month name
 293          * @return string ISO month name
 294          */
 295         function makeIsoMonth( $monthName ) {
 296                 global $wgContLang;
 297
 298                 $n = $this->xMonths[$wgContLang->lc( $monthName )];
 299                 return sprintf( '%02d', $n );
 300         }
 301
 302         /**
 303          * @todo document
 304          * @param $year String: Year name
 305          * @return string ISO year name
 306          */
 307         function makeIsoYear( $year ) {
 308                 # Assumes the year is in a nice format, as enforced by the regex
 309                 if ( substr( $year, -2 ) == 'BC' ) {
 310                         $num = intval(substr( $year, 0, -3 )) - 1;
 311                         # PHP bug note: sprintf( "%04d", -1 ) fails poorly
 312                         $text = sprintf( '-%04d', $num );
 313
 314                 } else {
 315                         $text = sprintf( '%04d', $year );
 316                 }
 317                 return $text;
 318         }
 319
 320         /**
 321          * @todo document
 322          */
 323         function makeNormalYear( $iso ) {
 324                 if ( $iso{0} == '-' ) {
 325                         $text = (intval( substr( $iso, 1 ) ) + 1) . ' BC';
 326                 } else {
 327                         $text = intval( $iso );
 328                 }
 329                 return $text;
 330         }
 331 }