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          */
 126         function reformat( $preference, $text, $options = array('linked') ) {
 127
 128                 $linked = in_array( 'linked', $options );
 129                 $match_whole = in_array( 'match-whole', $options );
 130
 131                 if ( isset( $this->preferences[$preference] ) ) {
 132                         $preference = $this->preferences[$preference];
 133                 } else {
 134                         $preference = self::NONE;
 135                 }
 136                 for ( $i=1; $i<=self::LAST; $i++ ) {
 137                         $this->mSource = $i;
 138                         if ( isset ( $this->rules[$preference][$i] ) ) {
 139                                 # Specific rules
 140                                 $this->mTarget = $this->rules[$preference][$i];
 141                         } elseif ( isset ( $this->rules[self::ALL][$i] ) ) {
 142                                 # General rules
 143                                 $this->mTarget = $this->rules[self::ALL][$i];
 144                         } elseif ( $preference ) {
 145                                 # User preference
 146                                 $this->mTarget = $preference;
 147                         } else {
 148                                 # Default
 149                                 $this->mTarget = $i;
 150                         }
 151                         $regex = $this->regexes[$i];
 152
 153                         // Horrible hack
 154                         if (!$linked) {
 155                                 $regex = str_replace( array( '\[\[', '\]\]' ), '', $regex );
 156                         }
 157
 158                         if ($match_whole) {
 159                                 // Let's hope this works
 160                                 $regex = preg_replace( '!^/!', '/^', $regex );
 161                                 $regex = str_replace( $this->regexTrail,
 162                                         '$'.$this->regexTrail, $regex );
 163                         }
 164
 165                         // Another horrible hack
 166                         $this->mLinked = $linked;
 167                         $text = preg_replace_callback( $regex, array( &$this, 'replace' ), $text );
 168                         unset($this->mLinked);
 169                 }
 170                 return $text;
 171         }
 172
 173         /**
 174          * @param $matches
 175          */
 176         function replace( $matches ) {
 177                 # Extract information from $matches
 178                 $linked = true;
 179                 if ( isset( $this->mLinked ) )
 180                         $linked = $this->mLinked;
 181
 182                 $bits = array();
 183                 $key = $this->keys[$this->mSource];
 184                 for ( $p=0; $p < strlen($key); $p++ ) {
 185                         if ( $key[$p] != ' ' ) {
 186                                 $bits[$key[$p]] = $matches[$p+1];
 187                         }
 188                 }
 189
 190                 return $this->formatDate( $bits, $linked );
 191         }
 192
 193         function formatDate( $bits, $link = true ) {
 194                 $format = $this->targets[$this->mTarget];
 195
 196                 if (!$link) {
 197                         // strip piped links
 198                         $format = preg_replace( '/\[\[[^|]+\|([^\]]+)\]\]/', '$1', $format );
 199                         // strip remaining links
 200                         $format = str_replace( array( '[[', ']]' ), '', $format );
 201                 }
 202
 203                 # Construct new date
 204                 $text = '';
 205                 $fail = false;
 206
 207                 // Pre-generate y/Y stuff because we need the year for the <span> title.
 208                 if ( !isset( $bits['y'] ) && isset( $bits['Y'] ) )
 209                         $bits['y'] = $this->makeIsoYear( $bits['Y'] );
 210                 if ( !isset( $bits['Y'] ) && isset( $bits['y'] ) )
 211                         $bits['Y'] = $this->makeNormalYear( $bits['y'] );
 212
 213                 if ( !isset( $bits['m'] ) ) {
 214                         $m = $this->makeIsoMonth( $bits['F'] );
 215                         if ( !$m || $m == '00' ) {
 216                                 $fail = true;
 217                         } else {
 218                                 $bits['m'] = $m;
 219                         }
 220                 }
 221
 222                 if ( !isset($bits['d']) ) {
 223                         $bits['d'] = sprintf( '%02d', $bits['j'] );
 224                 }
 225
 226                 for ( $p=0; $p < strlen( $format ); $p++ ) {
 227                         $char = $format[$p];
 228                         switch ( $char ) {
 229                                 case 'd': # ISO day of month
 230                                         $text .= $bits['d'];
 231                                         break;
 232                                 case 'm': # ISO month
 233                                         $text .= $bits['m'];
 234                                         break;
 235                                 case 'y': # ISO year
 236                                         $text .= $bits['y'];
 237                                         break;
 238                                 case 'j': # ordinary day of month
 239                                         if ( !isset($bits['j']) ) {
 240                                                 $text .= intval( $bits['d'] );
 241                                         } else {
 242                                                 $text .= $bits['j'];
 243                                         }
 244                                         break;
 245                                 case 'F': # long month
 246                                         if ( !isset( $bits['F'] ) ) {
 247                                                 $m = intval($bits['m']);
 248                                                 if ( $m > 12 || $m < 1 ) {
 249                                                         $fail = true;
 250                                                 } else {
 251                                                         global $wgContLang;
 252                                                         $text .= $wgContLang->getMonthName( $m );
 253                                                 }
 254                                         } else {
 255                                                 $text .= ucfirst( $bits['F'] );
 256                                         }
 257                                         break;
 258                                 case 'Y': # ordinary (optional BC) year
 259                                         $text .= $bits['Y'];
 260                                         break;
 261                                 default:
 262                                         $text .= $char;
 263                         }
 264                 }
 265                 if ( $fail ) {
 266                         $text = $matches[0];
 267                 }
 268
 269                 $isoBits = array();
 270                 if ( isset($bits['y']) )
 271                         $isoBits[] = $bits['y'];
 272                 $isoBits[] = $bits['m'];
 273                 $isoBits[] = $bits['d'];
 274                 $isoDate = implode( '-', $isoBits );
 275
 276                 // Output is not strictly HTML (it's wikitext), but <span> is whitelisted.
 277                 $text = Html::rawElement( 'span',
 278                                         array( 'class' => 'mw-formatted-date', 'title' => $isoDate ), $text );
 279
 280                 return $text;
 281         }
 282
 283         /**
 284          * @todo document
 285          */
 286         function getMonthRegex() {
 287                 global $wgContLang;
 288                 $names = array();
 289                 for( $i = 1; $i <= 12; $i++ ) {
 290                         $names[] = $wgContLang->getMonthName( $i );
 291                         $names[] = $wgContLang->getMonthAbbreviation( $i );
 292                 }
 293                 return implode( '|', $names );
 294         }
 295
 296         /**
 297          * Makes an ISO month, e.g. 02, from a month name
 298          * @param $monthName String: month name
 299          * @return string ISO month name
 300          */
 301         function makeIsoMonth( $monthName ) {
 302                 global $wgContLang;
 303
 304                 $n = $this->xMonths[$wgContLang->lc( $monthName )];
 305                 return sprintf( '%02d', $n );
 306         }
 307
 308         /**
 309          * @todo document
 310          * @param $year String: Year name
 311          * @return string ISO year name
 312          */
 313         function makeIsoYear( $year ) {
 314                 # Assumes the year is in a nice format, as enforced by the regex
 315                 if ( substr( $year, -2 ) == 'BC' ) {
 316                         $num = intval(substr( $year, 0, -3 )) - 1;
 317                         # PHP bug note: sprintf( "%04d", -1 ) fails poorly
 318                         $text = sprintf( '-%04d', $num );
 319
 320                 } else {
 321                         $text = sprintf( '%04d', $year );
 322                 }
 323                 return $text;
 324         }
 325
 326         /**
 327          * @todo document
 328          */
 329         function makeNormalYear( $iso ) {
 330                 if ( $iso[0] == '-' ) {
 331                         $text = (intval( substr( $iso, 1 ) ) + 1) . ' BC';
 332                 } else {
 333                         $text = intval( $iso );
 334                 }
 335                 return $text;
 336         }
 337 }