includes/parser/DateFormatter.php

   1 <?php
   2 /**
   3  * Date formatter
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * Date formatter, recognises dates in plain text and formats them accoding to user preferences.
  10  * @todo preferences, OutputPage
  11  * @ingroup Parser
  12  */
  13 class DateFormatter
  14 {
  15         var $mSource, $mTarget;
  16         var $monthNames = '', $rxDM, $rxMD, $rxDMY, $rxYDM, $rxMDY, $rxYMD;
  17
  18         var $regexes, $pDays, $pMonths, $pYears;
  19         var $rules, $xMonths, $preferences;
  20
  21         const ALL = -1;
  22         const NONE = 0;
  23         const MDY = 1;
  24         const DMY = 2;
  25         const YMD = 3;
  26         const ISO1 = 4;
  27         const LASTPREF = 4;
  28         const ISO2 = 5;
  29         const YDM = 6;
  30         const DM = 7;
  31         const MD = 8;
  32         const LAST = 8;
  33
  34         /**
  35          * @todo document
  36          */
  37         function __construct() {
  38                 global $wgContLang;
  39
  40                 $this->monthNames = $this->getMonthRegex();
  41                 for ( $i=1; $i<=12; $i++ ) {
  42                         $this->xMonths[$wgContLang->lc( $wgContLang->getMonthName( $i ) )] = $i;
  43                         $this->xMonths[$wgContLang->lc( $wgContLang->getMonthAbbreviation( $i ) )] = $i;
  44                 }
  45
  46                 $this->regexTrail = '(?![a-z])/iu';
  47
  48                 # Partial regular expressions
  49                 $this->prxDM = '\[\[(\d{1,2})[ _](' . $this->monthNames . ')\]\]';
  50                 $this->prxMD = '\[\[(' . $this->monthNames . ')[ _](\d{1,2})\]\]';
  51                 $this->prxY = '\[\[(\d{1,4}([ _]BC|))\]\]';
  52                 $this->prxISO1 = '\[\[(-?\d{4})]]-\[\[(\d{2})-(\d{2})\]\]';
  53                 $this->prxISO2 = '\[\[(-?\d{4})-(\d{2})-(\d{2})\]\]';
  54
  55                 # Real regular expressions
  56                 $this->regexes[self::DMY] = "/{$this->prxDM}(?: *, *| +){$this->prxY}{$this->regexTrail}";
  57                 $this->regexes[self::YDM] = "/{$this->prxY}(?: *, *| +){$this->prxDM}{$this->regexTrail}";
  58                 $this->regexes[self::MDY] = "/{$this->prxMD}(?: *, *| +){$this->prxY}{$this->regexTrail}";
  59                 $this->regexes[self::YMD] = "/{$this->prxY}(?: *, *| +){$this->prxMD}{$this->regexTrail}";
  60                 $this->regexes[self::DM] = "/{$this->prxDM}{$this->regexTrail}";
  61                 $this->regexes[self::MD] = "/{$this->prxMD}{$this->regexTrail}";
  62                 $this->regexes[self::ISO1] = "/{$this->prxISO1}{$this->regexTrail}";
  63                 $this->regexes[self::ISO2] = "/{$this->prxISO2}{$this->regexTrail}";
  64
  65                 # Extraction keys
  66                 # See the comments in replace() for the meaning of the letters
  67                 $this->keys[self::DMY] = 'jFY';
  68                 $this->keys[self::YDM] = 'Y jF';
  69                 $this->keys[self::MDY] = 'FjY';
  70                 $this->keys[self::YMD] = 'Y Fj';
  71                 $this->keys[self::DM] = 'jF';
  72                 $this->keys[self::MD] = 'Fj';
  73                 $this->keys[self::ISO1] = 'ymd'; # y means ISO year
  74                 $this->keys[self::ISO2] = 'ymd';
  75
  76                 # Target date formats
  77                 $this->targets[self::DMY] = '[[F j|j F]] [[Y]]';
  78                 $this->targets[self::YDM] = '[[Y]], [[F j|j F]]';
  79                 $this->targets[self::MDY] = '[[F j]], [[Y]]';
  80                 $this->targets[self::YMD] = '[[Y]] [[F j]]';
  81                 $this->targets[self::DM] = '[[F j|j F]]';
  82                 $this->targets[self::MD] = '[[F j]]';
  83                 $this->targets[self::ISO1] = '[[Y|y]]-[[F j|m-d]]';
  84                 $this->targets[self::ISO2] = '[[y-m-d]]';
  85
  86                 # Rules
  87                 #            pref    source       target
  88                 $this->rules[self::DMY][self::MD]       = self::DM;
  89                 $this->rules[self::ALL][self::MD]       = self::MD;
  90                 $this->rules[self::MDY][self::DM]       = self::MD;
  91                 $this->rules[self::ALL][self::DM]       = self::DM;
  92                 $this->rules[self::NONE][self::ISO2]    = self::ISO1;
  93
  94                 $this->preferences = array(
  95                         'default' => self::NONE,
  96                         'dmy' => self::DMY,
  97                         'mdy' => self::MDY,
  98                         'ymd' => self::YMD,
  99                         'ISO 8601' => self::ISO1,
 100                 );
 101         }
 102
 103         /**
 104          * Get a DateFormatter object
 105          *
 106          * @return DateFormatter object
 107          */
 108         public static function &getInstance() {
 109                 global $wgMemc;
 110                 static $dateFormatter = false;
 111                 if ( !$dateFormatter ) {
 112                         $dateFormatter = $wgMemc->get( wfMemcKey( 'dateformatter' ) );
 113                         if ( !$dateFormatter ) {
 114                                 $dateFormatter = new DateFormatter;
 115                                 $wgMemc->set( wfMemcKey( 'dateformatter' ), $dateFormatter, 3600 );
 116                         }
 117                 }
 118                 return $dateFormatter;
 119         }
 120
 121         /**
 122          * @param $preference String: User preference
 123          * @param $text String: Text to reformat
 124          * @param $options Array: can contain 'linked' and/or 'match-whole'
 125          * @return mixed|String
 126          */
 127         function reformat( $preference, $text, $options = array('linked') ) {
 128
 129                 $linked = in_array( 'linked', $options );
 130                 $match_whole = in_array( 'match-whole', $options );
 131
 132                 if ( isset( $this->preferences[$preference] ) ) {
 133                         $preference = $this->preferences[$preference];
 134                 } else {
 135                         $preference = self::NONE;
 136                 }
 137                 for ( $i=1; $i<=self::LAST; $i++ ) {
 138                         $this->mSource = $i;
 139                         if ( isset ( $this->rules[$preference][$i] ) ) {
 140                                 # Specific rules
 141                                 $this->mTarget = $this->rules[$preference][$i];
 142                         } elseif ( isset ( $this->rules[self::ALL][$i] ) ) {
 143                                 # General rules
 144                                 $this->mTarget = $this->rules[self::ALL][$i];
 145                         } elseif ( $preference ) {
 146                                 # User preference
 147                                 $this->mTarget = $preference;
 148                         } else {
 149                                 # Default
 150                                 $this->mTarget = $i;
 151                         }
 152                         $regex = $this->regexes[$i];
 153
 154                         // Horrible hack
 155                         if (!$linked) {
 156                                 $regex = str_replace( array( '\[\[', '\]\]' ), '', $regex );
 157                         }
 158
 159                         if ($match_whole) {
 160                                 // Let's hope this works
 161                                 $regex = preg_replace( '!^/!', '/^', $regex );
 162                                 $regex = str_replace( $this->regexTrail,
 163                                         '$'.$this->regexTrail, $regex );
 164                         }
 165
 166                         // Another horrible hack
 167                         $this->mLinked = $linked;
 168                         $text = preg_replace_callback( $regex, array( &$this, 'replace' ), $text );
 169                         unset($this->mLinked);
 170                 }
 171                 return $text;
 172         }
 173
 174         /**
 175          * @param $matches
 176          * @return string
 177          */
 178         function replace( $matches ) {
 179                 # Extract information from $matches
 180                 $linked = true;
 181                 if ( isset( $this->mLinked ) )
 182                         $linked = $this->mLinked;
 183
 184                 $bits = array();
 185                 $key = $this->keys[$this->mSource];
 186                 for ( $p=0; $p < strlen($key); $p++ ) {
 187                         if ( $key[$p] != ' ' ) {
 188                                 $bits[$key[$p]] = $matches[$p+1];
 189                         }
 190                 }
 191
 192                 return $this->formatDate( $bits, $linked );
 193         }
 194
 195         function formatDate( $bits, $link = true ) {
 196                 $format = $this->targets[$this->mTarget];
 197
 198                 if (!$link) {
 199                         // strip piped links
 200                         $format = preg_replace( '/\[\[[^|]+\|([^\]]+)\]\]/', '$1', $format );
 201                         // strip remaining links
 202                         $format = str_replace( array( '[[', ']]' ), '', $format );
 203                 }
 204
 205                 # Construct new date
 206                 $text = '';
 207                 $fail = false;
 208
 209                 // Pre-generate y/Y stuff because we need the year for the <span> title.
 210                 if ( !isset( $bits['y'] ) && isset( $bits['Y'] ) )
 211                         $bits['y'] = $this->makeIsoYear( $bits['Y'] );
 212                 if ( !isset( $bits['Y'] ) && isset( $bits['y'] ) )
 213                         $bits['Y'] = $this->makeNormalYear( $bits['y'] );
 214
 215                 if ( !isset( $bits['m'] ) ) {
 216                         $m = $this->makeIsoMonth( $bits['F'] );
 217                         if ( !$m || $m == '00' ) {
 218                                 $fail = true;
 219                         } else {
 220                                 $bits['m'] = $m;
 221                         }
 222                 }
 223
 224                 if ( !isset($bits['d']) ) {
 225                         $bits['d'] = sprintf( '%02d', $bits['j'] );
 226                 }
 227
 228                 for ( $p=0; $p < strlen( $format ); $p++ ) {
 229                         $char = $format[$p];
 230                         switch ( $char ) {
 231                                 case 'd': # ISO day of month
 232                                         $text .= $bits['d'];
 233                                         break;
 234                                 case 'm': # ISO month
 235                                         $text .= $bits['m'];
 236                                         break;
 237                                 case 'y': # ISO year
 238                                         $text .= $bits['y'];
 239                                         break;
 240                                 case 'j': # ordinary day of month
 241                                         if ( !isset($bits['j']) ) {
 242                                                 $text .= intval( $bits['d'] );
 243                                         } else {
 244                                                 $text .= $bits['j'];
 245                                         }
 246                                         break;
 247                                 case 'F': # long month
 248                                         if ( !isset( $bits['F'] ) ) {
 249                                                 $m = intval($bits['m']);
 250                                                 if ( $m > 12 || $m < 1 ) {
 251                                                         $fail = true;
 252                                                 } else {
 253                                                         global $wgContLang;
 254                                                         $text .= $wgContLang->getMonthName( $m );
 255                                                 }
 256                                         } else {
 257                                                 $text .= ucfirst( $bits['F'] );
 258                                         }
 259                                         break;
 260                                 case 'Y': # ordinary (optional BC) year
 261                                         $text .= $bits['Y'];
 262                                         break;
 263                                 default:
 264                                         $text .= $char;
 265                         }
 266                 }
 267                 if ( $fail ) {
 268                         $text = $matches[0];
 269                 }
 270
 271                 $isoBits = array();
 272                 if ( isset($bits['y']) )
 273                         $isoBits[] = $bits['y'];
 274                 $isoBits[] = $bits['m'];
 275                 $isoBits[] = $bits['d'];
 276                 $isoDate = implode( '-', $isoBits );
 277
 278                 // Output is not strictly HTML (it's wikitext), but <span> is whitelisted.
 279                 $text = Html::rawElement( 'span',
 280                                         array( 'class' => 'mw-formatted-date', 'title' => $isoDate ), $text );
 281
 282                 return $text;
 283         }
 284
 285         /**
 286          * @todo document
 287          * @return string
 288          */
 289         function getMonthRegex() {
 290                 global $wgContLang;
 291                 $names = array();
 292                 for( $i = 1; $i <= 12; $i++ ) {
 293                         $names[] = $wgContLang->getMonthName( $i );
 294                         $names[] = $wgContLang->getMonthAbbreviation( $i );
 295                 }
 296                 return implode( '|', $names );
 297         }
 298
 299         /**
 300          * Makes an ISO month, e.g. 02, from a month name
 301          * @param $monthName String: month name
 302          * @return string ISO month name
 303          */
 304         function makeIsoMonth( $monthName ) {
 305                 global $wgContLang;
 306
 307                 $n = $this->xMonths[$wgContLang->lc( $monthName )];
 308                 return sprintf( '%02d', $n );
 309         }
 310
 311         /**
 312          * @todo document
 313          * @param $year String: Year name
 314          * @return string ISO year name
 315          */
 316         function makeIsoYear( $year ) {
 317                 # Assumes the year is in a nice format, as enforced by the regex
 318                 if ( substr( $year, -2 ) == 'BC' ) {
 319                         $num = intval(substr( $year, 0, -3 )) - 1;
 320                         # PHP bug note: sprintf( "%04d", -1 ) fails poorly
 321                         $text = sprintf( '-%04d', $num );
 322
 323                 } else {
 324                         $text = sprintf( '%04d', $year );
 325                 }
 326                 return $text;
 327         }
 328
 329         /**
 330          * @todo document
 331          * @return int|string
 332          */
 333         function makeNormalYear( $iso ) {
 334                 if ( $iso[0] == '-' ) {
 335                         $text = (intval( substr( $iso, 1 ) ) + 1) . ' BC';
 336                 } else {
 337                         $text = intval( $iso );
 338                 }
 339                 return $text;
 340         }
 341 }