[Summary view] [Print] [Text view]

   1  <?php
   2  
   3  /**
   4   * BookmarkletGen : converts readable Javascript code into a bookmarklet link
   5   *
   6   * Features :
   7   * - removes comments
   8   * - compresses code, not literal strings
   9   *   Example:
  10   *   function someName( param ) { alert( "this is a string" ) }
  11   *   will return:
  12   *   function%20someName(param){alert("this is a string")}
  13   * - wraps code into a self invoking function
  14   *
  15   * This is basically a slightly enhanced PHP port of the excellent Bookmarklet Crunchinator
  16   * http://ted.mielczarek.org/code/mozilla/bookmarklet.html
  17   *
  18   */
  19  namespace Ozh\Bookmarkletgen;
  20  
  21  class Bookmarkletgen {
  22  
  23      private $literal_strings = array();
  24      
  25      /**
  26       * Main function, calls all others
  27       *
  28       * @param  string $code  Javascript code to bookmarkletify
  29       * @return string        Bookmarklet link
  30       */
  31      public function crunch( $code ) {
  32          $out = "(function() {\n" . $code . "\n})();";
  33  
  34          $out = $this->replace_strings( $out );
  35          $out = $this->kill_comments( $out );
  36          $out = $this->compress_white_space( $out );
  37          $out = $this->combine_strings( $out );
  38          $out = $this->restore_strings( $out );
  39          $out = $this->encodeURIComponent( $out );
  40          $out = 'javascript:' . $out;
  41  
  42          return $out;
  43      }
  44      
  45      /**
  46       * PHP port of Javascript function encodeURIComponent
  47       *
  48       * From http://stackoverflow.com/a/1734255/36850
  49       *
  50       * @since
  51       * @param  string $str  String to encode
  52       * @return string       Encoded string
  53       */
  54      // 
  55      private function encodeURIComponent( $str ) {
  56          $revert = array(
  57              '%21'=>'!', '%2A'=>'*', '%28'=>'(', '%29'=>')',
  58          );
  59      
  60          return strtr( rawurlencode( $str ), $revert );
  61      }
  62  
  63      /**
  64       * Kill comment lines and blocks
  65       *
  66       * @param  string $code  Commented Javascript code
  67       * @return string        Commentless code
  68       */
  69      private function kill_comments( $code ) {
  70          $code = preg_replace( '!\s*//.+$!m', '', $code );
  71          $code = preg_replace( '!/\*.+?\*/!sm', '', $code ); // s modifier: dot matches new lines
  72          
  73          return $code;
  74      }
  75      
  76      /**
  77       * Compress white space
  78       *
  79       * Remove some extraneous spaces and make the whole script a one liner
  80       *
  81       * @param  string $code  Javascript code
  82       * @return string        Compressed code
  83       */
  84      private function compress_white_space( $code ) {
  85          // Tabs to space, no more than 1 consecutive space
  86          $code = preg_replace( '!\t!m', ' ', $code );
  87          $code = preg_replace( '![ ]{2,}!m', ' ', $code );
  88          
  89          // Remove uneccessary white space around operators, braces and brackets.
  90          // \xHH sequence is: !%&()*+,-/:;<=>?[]\{|}~
  91          $code = preg_replace( '/\s([\x21\x25\x26\x28\x29\x2a\x2b\x2c\x2d\x2f\x3a\x3b\x3c\x3d\x3e\x3f\x5b\x5d\x5c\x7b\x7c\x7d\x7e])/m', "$1", $code );
  92          $code = preg_replace( '/([\x21\x25\x26\x28\x29\x2a\x2b\x2c\x2d\x2f\x3a\x3b\x3c\x3d\x3e\x3f\x5b\x5d\x5c\x7b\x7c\x7d\x7e])\s/m', "$1", $code );
  93          
  94          // Split on each line, trim leading/trailing white space, kill empty lines, combine everything in one line
  95          $code = preg_split( '/\r\n|\r|\n/', $code );
  96          foreach( $code as $i => $line ) {
  97              $code[ $i ] = trim( $line );
  98          }
  99          $code = implode( '', $code );
 100  
 101          return $code;
 102      }
 103      
 104      /**
 105       * Combine any consecutive strings
 106       *
 107       * In the case we have two consecutive quoted strings (eg: "hello" + "world"), save a couple more
 108       * length and combine them
 109       *
 110       * @param  string $code  Javascript code
 111       * @return string        Javascript code
 112       */
 113      private function combine_strings( $code ) {
 114          $code = preg_replace('/"\+"/m', "", $code);
 115          $code = preg_replace("/'\+'/m", "", $code);
 116  
 117          return $code;
 118      }
 119  
 120      
 121      /**
 122       * Replace all literal strings (eg: "hello world") with a placeholder and collect them in an array
 123       *
 124       * The idea is that strings cannot be trimmed or white-space optimized: take them out first before uglifying
 125       * the code, then we'll reinject them back in later
 126       *
 127       * @param  string $code  Javascript code
 128       * @return string        Javascript code with placeholders (eg "__1__") instead of literal strings
 129       */
 130      private function replace_strings( $code ) {
 131          $return    = "";
 132          $literal   = "";
 133          $quoteChar = "";
 134          $escaped   = false;
 135  
 136          // Split script into individual lines.
 137          $lines = explode("\n", $code);
 138          $count = count( $lines );
 139          for( $i = 0; $i < $count; $i++ ) {
 140  
 141              $j = 0;
 142              $inQuote = false;
 143              while ($j < strlen( $lines[$i] ) ) {
 144                  $c = $lines[ $i ][ $j ];
 145  
 146                  // If not already in a string, look for the start of one.
 147                  if (!$inQuote) {
 148                      if ($c == '"' || $c == "'") {
 149                          $inQuote = true;
 150                          $escaped = false;
 151                          $quoteChar = $c;
 152                          $literal = $c;
 153                      }
 154                      else {
 155                          $return .= $c;
 156                      }
 157                  }
 158  
 159                  // Already in a string, look for end and copy characters.
 160                  else {
 161                      if ($c == $quoteChar && !$escaped) {
 162                          $inQuote = false;
 163                          $literal .= $quoteChar;
 164                          $return .= "__" . count( $this->literal_strings ) . "__";
 165                          $this->literal_strings[ count( $this->literal_strings ) ] = $literal;
 166                      }
 167                      else if ($c == "\\" && !$escaped) {
 168                          $escaped = true;
 169                      }
 170                      else {
 171                          $escaped = false;
 172                      }
 173                      $literal .= $c;
 174                  }
 175                  $j++;
 176              }
 177              $return .= "\n";
 178          }
 179  
 180          return $return;
 181      }
 182      
 183      /**
 184       * Restore literal strings by replacing their placeholders with actual strings
 185       *
 186       * @param  string $code  Javascript code with placeholders
 187       * @return string        Javascript code with actual strings
 188       */
 189      private function restore_strings( $code ) {
 190          foreach( $this->literal_strings as $i => $string ) {
 191              $code = preg_replace( '/__' . $i . '__/', $string, $code, 1 );
 192          }
 193          
 194          return $code;
 195      }
 196  
 197  }