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          * @param $options Array: can contain 'linked' and/or 'match-whole'
 120          */
 121         function reformat( $preference, $text, $options = array('linked') ) {
 122
 123                 $linked = in_array( 'linked', $options );
 124                 $match_whole = in_array( 'match-whole', $options );
 125
 126                 if ( isset( $this->preferences[$preference] ) ) {
 127                         $preference = $this->preferences[$preference];
 128                 } else {
 129                         $preference = self::NONE;
 130                 }
 131                 for ( $i=1; $i<=self::LAST; $i++ ) {
 132                         $this->mSource = $i;
 133                         if ( isset ( $this->rules[$preference][$i] ) ) {
 134                                 # Specific rules
 135                                 $this->mTarget = $this->rules[$preference][$i];
 136                         } elseif ( isset ( $this->rules[self::ALL][$i] ) ) {
 137                                 # General rules
 138                                 $this->mTarget = $this->rules[self::ALL][$i];
 139                         } elseif ( $preference ) {
 140                                 # User preference
 141                                 $this->mTarget = $preference;
 142                         } else {
 143                                 # Default
 144                                 $this->mTarget = $i;
 145                         }
 146                         $regex = $this->regexes[$i];
 147
 148                         // Horrible hack
 149                         if (!$linked) {
 150                                 $regex = str_replace( array( '\[\[', '\]\]' ), '', $regex );
 151                         }
 152
 153                         if ($match_whole) {
 154                                 // Let's hope this works
 155                                 $regex = preg_replace( '!^/!', '/^', $regex );
 156                                 $regex = str_replace( $this->regexTrail,
 157                                         '$'.$this->regexTrail, $regex );
 158                         }
 159
 160                         // Another horrible hack
 161                         $this->mLinked = $linked;
 162                         $text = preg_replace_callback( $regex, array( &$this, 'replace' ), $text );
 163                         unset($this->mLinked);
 164                 }
 165                 return $text;
 166         }
 167
 168         /**
 169          * @param $matches
 170          */
 171         function replace( $matches ) {
 172                 # Extract information from $matches
 173                 $linked = true;
 174                 if ( isset( $this->mLinked ) )
 175                         $linked = $this->mLinked;
 176
 177                 $bits = array();
 178                 $key = $this->keys[$this->mSource];
 179                 for ( $p=0; $p < strlen($key); $p++ ) {
 180                         if ( $key{$p} != ' ' ) {
 181                                 $bits[$key{$p}] = $matches[$p+1];
 182                         }
 183                 }
 184
 185                 return $this->formatDate( $bits, $linked );
 186         }
 187
 188         function formatDate( $bits, $link = true ) {
 189                 $format = $this->targets[$this->mTarget];
 190
 191                 if (!$link) {
 192                         // strip piped links
 193                         $format = preg_replace( '/\[\[[^|]+\|([^\]]+)\]\]/', '$1', $format );
 194                         // strip remaining links
 195                         $format = str_replace( array( '[[', ']]' ), '', $format );
 196                 }
 197
 198                 # Construct new date
 199                 $text = '';
 200                 $fail = false;
 201
 202                 // Pre-generate y/Y stuff because we need the year for the <span> title.
 203                 if ( !isset( $bits['y'] ) && isset( $bits['Y'] ) )
 204                         $bits['y'] = $this->makeIsoYear( $bits['Y'] );
 205                 if ( !isset( $bits['Y'] ) && isset( $bits['y'] ) )
 206                         $bits['Y'] = $this->makeNormalYear( $bits['y'] );
 207
 208                 if ( !isset( $bits['m'] ) ) {
 209                         $m = $this->makeIsoMonth( $bits['F'] );
 210                         if ( !$m || $m == '00' ) {
 211                                 $fail = true;
 212                         } else {
 213                                 $bits['m'] = $m;
 214                         }
 215                 }
 216
 217                 if ( !isset($bits['d']) ) {
 218                         $bits['d'] = sprintf( '%02d', $bits['j'] );
 219                 }
 220
 221                 for ( $p=0; $p < strlen( $format ); $p++ ) {
 222                         $char = $format{$p};
 223                         switch ( $char ) {
 224                                 case 'd': # ISO day of month
 225                                         $text .= $bits['d'];
 226                                         break;
 227                                 case 'm': # ISO month
 228                                         $text .= $bits['m'];
 229                                         break;
 230                                 case 'y': # ISO year
 231                                         $text .= $bits['y'];
 232                                         break;
 233                                 case 'j': # ordinary day of month
 234                                         if ( !isset($bits['j']) ) {
 235                                                 $text .= intval( $bits['d'] );
 236                                         } else {
 237                                                 $text .= $bits['j'];
 238                                         }
 239                                         break;
 240                                 case 'F': # long month
 241                                         if ( !isset( $bits['F'] ) ) {
 242                                                 $m = intval($bits['m']);
 243                                                 if ( $m > 12 || $m < 1 ) {
 244                                                         $fail = true;
 245                                                 } else {
 246                                                         global $wgContLang;
 247                                                         $text .= $wgContLang->getMonthName( $m );
 248                                                 }
 249                                         } else {
 250                                                 $text .= ucfirst( $bits['F'] );
 251                                         }
 252                                         break;
 253                                 case 'Y': # ordinary (optional BC) year
 254                                         $text .= $bits['Y'];
 255                                         break;
 256                                 default:
 257                                         $text .= $char;
 258                         }
 259                 }
 260                 if ( $fail ) {
 261                         $text = $matches[0];
 262                 }
 263
 264                 $isoBits = array();
 265                 if ( isset($bits['y']) )
 266                         $isoBits[] = $bits['y'];
 267                 $isoBits[] = $bits['m'];
 268                 $isoBits[] = $bits['d'];
 269                 $isoDate = implode( '-', $isoBits );;
 270
 271                 // Output is not strictly HTML (it's wikitext), but <span> is whitelisted.
 272                 $text = Html::rawElement( 'span',
 273                                         array( 'class' => 'mw-formatted-date', 'title' => $isoDate ), $text );
 274
 275                 return $text;
 276         }
 277
 278         /**
 279          * @todo document
 280          */
 281         function getMonthRegex() {
 282                 global $wgContLang;
 283                 $names = array();
 284                 for( $i = 1; $i <= 12; $i++ ) {
 285                         $names[] = $wgContLang->getMonthName( $i );
 286                         $names[] = $wgContLang->getMonthAbbreviation( $i );
 287                 }
 288                 return implode( '|', $names );
 289         }
 290
 291         /**
 292          * Makes an ISO month, e.g. 02, from a month name
 293          * @param $monthName String: month name
 294          * @return string ISO month name
 295          */
 296         function makeIsoMonth( $monthName ) {
 297                 global $wgContLang;
 298
 299                 $n = $this->xMonths[$wgContLang->lc( $monthName )];
 300                 return sprintf( '%02d', $n );
 301         }
 302
 303         /**
 304          * @todo document
 305          * @param $year String: Year name
 306          * @return string ISO year name
 307          */
 308         function makeIsoYear( $year ) {
 309                 # Assumes the year is in a nice format, as enforced by the regex
 310                 if ( substr( $year, -2 ) == 'BC' ) {
 311                         $num = intval(substr( $year, 0, -3 )) - 1;
 312                         # PHP bug note: sprintf( "%04d", -1 ) fails poorly
 313                         $text = sprintf( '-%04d', $num );
 314
 315                 } else {
 316                         $text = sprintf( '%04d', $year );
 317                 }
 318                 return $text;
 319         }
 320
 321         /**
 322          * @todo document
 323          */
 324         function makeNormalYear( $iso ) {
 325                 if ( $iso{0} == '-' ) {
 326                         $text = (intval( substr( $iso, 1 ) ) + 1) . ' BC';
 327                 } else {
 328                         $text = intval( $iso );
 329                 }
 330                 return $text;
 331         }
 332 }