includes/StringUtils.php

   1 <?php
   2
   3 class StringUtils {
   4         /**
   5          * Perform an operation equivalent to
   6          *
   7          *     preg_replace( "!$startDelim(.*?)$endDelim!", $replace, $subject );
   8          *
   9          * except that it's worst-case O(N) instead of O(N^2)
  10          *
  11          * Compared to delimiterReplace(), this implementation is fast but memory-
  12          * hungry and inflexible. The memory requirements are such that I don't
  13          * recommend using it on anything but guaranteed small chunks of text.
  14          */
  15         static function hungryDelimiterReplace( $startDelim, $endDelim, $replace, $subject ) {
  16                 $segments = explode( $startDelim, $subject );
  17                 $output = array_shift( $segments );
  18                 foreach ( $segments as $s ) {
  19                         $endDelimPos = strpos( $s, $endDelim );
  20                         if ( $endDelimPos === false ) {
  21                                 $output .= $startDelim . $s;
  22                         } else {
  23                                 $output .= $replace . substr( $s, $endDelimPos + strlen( $endDelim ) );
  24                         }
  25                 }
  26                 return $output;
  27         }
  28
  29         /**
  30          * Perform an operation equivalent to
  31          *
  32          *   preg_replace_callback( "!$startDelim(.*)$endDelim!s$flags", $callback, $subject )
  33          *
  34          * This implementation is slower than staticDelimiterReplace but uses far less
  35          * memory and allows regular expression delimiters.
  36          *
  37          * @param string $flags Regular expression flags
  38          */
  39         static function delimiterReplaceCallback( $startDelim, $endDelim, $callback, $subject, $flags = '' ) {
  40                 $inputPos = 0;
  41                 $outputPos = 0;
  42                 $output = '';
  43                 $foundStart = false;
  44
  45                 while ( $inputPos < strlen( $subject ) &&
  46                   preg_match( "!($startDelim)|($endDelim)!$flags", $subject, $m, PREG_OFFSET_CAPTURE, $inputPos ) )
  47                 {
  48                         if ( $m[1][0] != '' ) {
  49                                 # Found start
  50                                 # Write out the non-matching section
  51                                 $output .= substr( $subject, $outputPos, $m[1][1] - $outputPos );
  52                                 $foundStart = true;
  53                                 $inputPos = $contentPos = $m[1][1] + strlen( $m[1][0] );
  54                                 $outputPos = $m[1][1];
  55                         } elseif ( $m[2][0] != '' ) {
  56                                 # Found end
  57                                 if ( $foundStart ) {
  58                                         # Found match
  59                                         $output .= call_user_func( $callback, array(
  60                                                 substr( $subject, $outputPos, $m[2][1] + strlen( $m[2][0] ) - $outputPos ),
  61                                                 substr( $subject, $contentPos, $m[2][1] - $contentPos )
  62                                         ));
  63                                         $foundStart = false;
  64                                 } else {
  65                                         # Non-matching end, write it out
  66                                         $output .= substr( $subject, $inputPos, $m[2][1] + strlen( $m[2][0] ) - $outputPos );
  67                                 }
  68                                 $inputPos = $outputPos = $m[2][1] + strlen( $m[2][0] );
  69                         } else {
  70                                 throw new MWException( 'Invalid delimiter given to ' . __METHOD__ );
  71                         }
  72                 }
  73                 if ( $outputPos < strlen( $subject ) ) {
  74                         $output .= substr( $subject, $outputPos );
  75                 }
  76                 return $output;
  77         }
  78
  79         /*
  80          * Perform an operation equivalent to
  81          *
  82          *   preg_replace( "!$startDelim(.*)$endDelim!$flags", $replace, $subject )
  83          *
  84          * @param string $startDelim Start delimiter regular expression
  85          * @param string $endDelim End delimiter regular expression
  86          * @param string $replace Replacement string. May contain $1, which will be
  87          *               replaced by the text between the delimiters
  88          * @param string $subject String to search
  89          * @return string The string with the matches replaced
  90          */
  91         static function delimiterReplace( $startDelim, $endDelim, $replace, $subject, $flags = '' ) {
  92                 $replacer = new RegexlikeReplacer( $replace );
  93                 return self::delimiterReplaceCallback( $startDelim, $endDelim,
  94                         $replacer->cb(), $subject, $flags );
  95         }
  96
  97         /**
  98          * More or less "markup-safe" explode()
  99          * Ignores any instances of the separator inside <...>
 100          * @param string $separator
 101          * @param string $text
 102          * @return array
 103          */
 104         static function explodeMarkup( $separator, $text ) {
 105                 $placeholder = "\x00";
 106
 107                 // Remove placeholder instances
 108                 $text = str_replace( $placeholder, '', $text );
 109
 110                 // Replace instances of the separator inside HTML-like tags with the placeholder
 111                 $replacer = new DoubleReplacer( $separator, $placeholder );
 112                 $cleaned = StringUtils::delimiterReplaceCallback( '<', '>', $replacer->cb(), $text );
 113
 114                 // Explode, then put the replaced separators back in
 115                 $items = explode( $separator, $cleaned );
 116                 foreach( $items as $i => $str ) {
 117                         $items[$i] = str_replace( $placeholder, $separator, $str );
 118                 }
 119
 120                 return $items;
 121         }
 122
 123         /**
 124          * Escape a string to make it suitable for inclusion in a preg_replace()
 125          * replacement parameter.
 126          *
 127          * @param string $string
 128          * @return string
 129          */
 130         static function escapeRegexReplacement( $string ) {
 131                 $string = str_replace( '\\', '\\\\', $string );
 132                 $string = str_replace( '$', '\\$', $string );
 133                 return $string;
 134         }
 135 }
 136
 137 /**
 138  * Base class for "replacers", objects used in preg_replace_callback() and
 139  * StringUtils::delimiterReplaceCallback()
 140  */
 141 class Replacer {
 142         function cb() {
 143                 return array( &$this, 'replace' );
 144         }
 145 }
 146
 147 /**
 148  * Class to replace regex matches with a string similar to that used in preg_replace()
 149  */
 150 class RegexlikeReplacer extends Replacer {
 151         var $r;
 152         function __construct( $r ) {
 153                 $this->r = $r;
 154         }
 155
 156         function replace( $matches ) {
 157                 $pairs = array();
 158                 foreach ( $matches as $i => $match ) {
 159                         $pairs["\$$i"] = $match;
 160                 }
 161                 return strtr( $this->r, $pairs );
 162         }
 163
 164 }
 165
 166 /**
 167  * Class to perform secondary replacement within each replacement string
 168  */
 169 class DoubleReplacer extends Replacer {
 170         function __construct( $from, $to, $index = 0 ) {
 171                 $this->from = $from;
 172                 $this->to = $to;
 173                 $this->index = $index;
 174         }
 175
 176         function replace( $matches ) {
 177                 return str_replace( $this->from, $this->to, $matches[$this->index] );
 178         }
 179 }
 180
 181 /**
 182  * Class to perform replacement based on a simple hashtable lookup
 183  */
 184 class HashtableReplacer extends Replacer {
 185         var $table, $index;
 186
 187         function __construct( $table, $index = 0 ) {
 188                 $this->table = $table;
 189                 $this->index = $index;
 190         }
 191
 192         function replace( $matches ) {
 193                 return $this->table[$matches[$this->index]];
 194         }
 195 }
 196
 197 /**
 198  * Replacement array for FSS with fallback to strtr()
 199  * Supports lazy initialisation of FSS resource
 200  */
 201 class ReplacementArray {
 202         /*mostly private*/ var $data = false;
 203         /*mostly private*/ var $fss = false;
 204
 205         /**
 206          * Create an object with the specified replacement array
 207          * The array should have the same form as the replacement array for strtr()
 208          */
 209         function __construct( $data = array() ) {
 210                 $this->data = $data;
 211         }
 212
 213         function __sleep() {
 214                 return array( 'data' );
 215         }
 216
 217         function __wakeup() {
 218                 $this->fss = false;
 219         }
 220
 221         /**
 222          * Set the whole replacement array at once
 223          */
 224         function setArray( $data ) {
 225                 $this->data = $data;
 226                 $this->fss = false;
 227         }
 228
 229         function getArray() {
 230                 return $this->data;
 231         }
 232
 233         /**
 234          * Set an element of the replacement array
 235          */
 236         function setPair( $from, $to ) {
 237                 $this->data[$from] = $to;
 238                 $this->fss = false;
 239         }
 240
 241         function mergeArray( $data ) {
 242                 $this->data = array_merge( $this->data, $data );
 243                 $this->fss = false;
 244         }
 245
 246         function merge( $other ) {
 247                 $this->data = array_merge( $this->data, $other->data );
 248                 $this->fss = false;
 249         }
 250
 251         function replace( $subject ) {
 252                 if ( function_exists( 'fss_prep_replace' ) ) {
 253                         if ( $this->fss === false ) {
 254                                 $this->fss = fss_prep_replace( $this->data );
 255                         }
 256                         return fss_exec_replace( $this->fss, $subject );
 257                 } else {
 258                         return strtr( $subject, $this->data );
 259                 }
 260         }
 261 }
 262
 263 ?>