[Summary view] [Print] [Text view]

   1  <?php
   2  /**
   3   * WordPress Diff bastard child of old MediaWiki Diff Formatter.
   4   *
   5   * Basically all that remains is the table structure and some method names.
   6   *
   7   * @package WordPress
   8   * @subpackage Diff
   9   */
  10  
  11  if ( !class_exists( 'Text_Diff' ) ) {
  12      /** Text_Diff class */
  13      require( dirname(__FILE__).'/Text/Diff.php' );
  14      /** Text_Diff_Renderer class */
  15      require( dirname(__FILE__).'/Text/Diff/Renderer.php' );
  16      /** Text_Diff_Renderer_inline class */
  17      require( dirname(__FILE__).'/Text/Diff/Renderer/inline.php' );
  18  }
  19  
  20  /**
  21   * Table renderer to display the diff lines.
  22   *
  23   * @since 2.6.0
  24   * @uses Text_Diff_Renderer Extends
  25   */
  26  class WP_Text_Diff_Renderer_Table extends Text_Diff_Renderer {
  27  
  28      /**
  29       * @see Text_Diff_Renderer::_leading_context_lines
  30       * @var int
  31       * @access protected
  32       * @since 2.6.0
  33       */
  34      var $_leading_context_lines  = 10000;
  35  
  36      /**
  37       * @see Text_Diff_Renderer::_trailing_context_lines
  38       * @var int
  39       * @access protected
  40       * @since 2.6.0
  41       */
  42      var $_trailing_context_lines = 10000;
  43  
  44      /**
  45       * {@internal Missing Description}}
  46       *
  47       * @var float
  48       * @access protected
  49       * @since 2.6.0
  50       */
  51      var $_diff_threshold = 0.6;
  52  
  53      /**
  54       * Inline display helper object name.
  55       *
  56       * @var string
  57       * @access protected
  58       * @since 2.6.0
  59       */
  60      var $inline_diff_renderer = 'WP_Text_Diff_Renderer_inline';
  61  
  62      /**
  63       * Constructor - Call parent constructor with params array.
  64       *
  65       * This will set class properties based on the key value pairs in the array.
  66       *
  67       * @since 2.6.0
  68       *
  69       * @param array $params
  70       */
  71  	function __construct( $params = array() ) {
  72          parent::__construct( $params );
  73      }
  74  
  75      /**
  76       * @ignore
  77       *
  78       * @param string $header
  79       * @return string
  80       */
  81  	function _startBlock( $header ) {
  82          return '';
  83      }
  84  
  85      /**
  86       * @ignore
  87       *
  88       * @param array $lines
  89       * @param string $prefix
  90       */
  91  	function _lines( $lines, $prefix=' ' ) {
  92      }
  93  
  94      /**
  95       * @ignore
  96       *
  97       * @param string $line HTML-escape the value.
  98       * @return string
  99       */
 100  	function addedLine( $line ) {
 101          return "<td>+</td><td class='diff-addedline'>{$line}</td>";
 102      }
 103  
 104      /**
 105       * @ignore
 106       *
 107       * @param string $line HTML-escape the value.
 108       * @return string
 109       */
 110  	function deletedLine( $line ) {
 111          return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
 112      }
 113  
 114      /**
 115       * @ignore
 116       *
 117       * @param string $line HTML-escape the value.
 118       * @return string
 119       */
 120  	function contextLine( $line ) {
 121          return "<td> </td><td class='diff-context'>{$line}</td>";
 122      }
 123  
 124      /**
 125       * @ignore
 126       *
 127       * @return string
 128       */
 129  	function emptyLine() {
 130          return '<td colspan="2">&nbsp;</td>';
 131      }
 132  
 133      /**
 134       * @ignore
 135       * @access private
 136       *
 137       * @param array $lines
 138       * @param bool $encode
 139       * @return string
 140       */
 141  	function _added( $lines, $encode = true ) {
 142          $r = '';
 143          foreach ($lines as $line) {
 144              if ( $encode )
 145                  $line = htmlspecialchars( $line );
 146              $r .= '<tr>' . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
 147          }
 148          return $r;
 149      }
 150  
 151      /**
 152       * @ignore
 153       * @access private
 154       *
 155       * @param array $lines
 156       * @param bool $encode
 157       * @return string
 158       */
 159  	function _deleted( $lines, $encode = true ) {
 160          $r = '';
 161          foreach ($lines as $line) {
 162              if ( $encode )
 163                  $line = htmlspecialchars( $line );
 164              $r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . "</tr>\n";
 165          }
 166          return $r;
 167      }
 168  
 169      /**
 170       * @ignore
 171       * @access private
 172       *
 173       * @param array $lines
 174       * @param bool $encode
 175       * @return string
 176       */
 177  	function _context( $lines, $encode = true ) {
 178          $r = '';
 179          foreach ($lines as $line) {
 180              if ( $encode )
 181                  $line = htmlspecialchars( $line );
 182              $r .= '<tr>' .
 183                  $this->contextLine( $line ) . $this->contextLine( $line ) . "</tr>\n";
 184          }
 185          return $r;
 186      }
 187  
 188      /**
 189       * Process changed lines to do word-by-word diffs for extra highlighting.
 190       *
 191       * (TRAC style) sometimes these lines can actually be deleted or added rows.
 192       * We do additional processing to figure that out
 193       *
 194       * @access private
 195       * @since 2.6.0
 196       *
 197       * @param array $orig
 198       * @param array $final
 199       * @return string
 200       */
 201  	function _changed( $orig, $final ) {
 202          $r = '';
 203  
 204          // Does the aforementioned additional processing
 205          // *_matches tell what rows are "the same" in orig and final. Those pairs will be diffed to get word changes
 206          //    match is numeric: an index in other column
 207          //    match is 'X': no match. It is a new row
 208          // *_rows are column vectors for the orig column and the final column.
 209          //    row >= 0: an indix of the $orig or $final array
 210          //    row  < 0: a blank row for that column
 211          list($orig_matches, $final_matches, $orig_rows, $final_rows) = $this->interleave_changed_lines( $orig, $final );
 212  
 213          // These will hold the word changes as determined by an inline diff
 214          $orig_diffs  = array();
 215          $final_diffs = array();
 216  
 217          // Compute word diffs for each matched pair using the inline diff
 218          foreach ( $orig_matches as $o => $f ) {
 219              if ( is_numeric($o) && is_numeric($f) ) {
 220                  $text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
 221                  $renderer = new $this->inline_diff_renderer;
 222                  $diff = $renderer->render( $text_diff );
 223  
 224                  // If they're too different, don't include any <ins> or <dels>
 225                  if ( $diff_count = preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
 226                      // length of all text between <ins> or <del>
 227                      $stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
 228                      // since we count lengith of text between <ins> or <del> (instead of picking just one),
 229                      //    we double the length of chars not in those tags.
 230                      $stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
 231                      $diff_ratio = $stripped_matches / $stripped_diff;
 232                      if ( $diff_ratio > $this->_diff_threshold )
 233                          continue; // Too different. Don't save diffs.
 234                  }
 235  
 236                  // Un-inline the diffs by removing del or ins
 237                  $orig_diffs[$o]  = preg_replace( '|<ins>.*?</ins>|', '', $diff );
 238                  $final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
 239              }
 240          }
 241  
 242          foreach ( array_keys($orig_rows) as $row ) {
 243              // Both columns have blanks. Ignore them.
 244              if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
 245                  continue;
 246  
 247              // If we have a word based diff, use it. Otherwise, use the normal line.
 248              if ( isset( $orig_diffs[$orig_rows[$row]] ) )
 249                  $orig_line = $orig_diffs[$orig_rows[$row]];
 250              elseif ( isset( $orig[$orig_rows[$row]] ) )
 251                  $orig_line = htmlspecialchars($orig[$orig_rows[$row]]);
 252              else
 253                  $orig_line = '';
 254  
 255              if ( isset( $final_diffs[$final_rows[$row]] ) )
 256                  $final_line = $final_diffs[$final_rows[$row]];
 257              elseif ( isset( $final[$final_rows[$row]] ) )
 258                  $final_line = htmlspecialchars($final[$final_rows[$row]]);
 259              else
 260                  $final_line = '';
 261  
 262              if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
 263                  $r .= $this->_added( array($final_line), false );
 264              } elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
 265                  $r .= $this->_deleted( array($orig_line), false );
 266              } else { // A true changed row.
 267                  $r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->addedLine( $final_line ) . "</tr>\n";
 268              }
 269          }
 270  
 271          return $r;
 272      }
 273  
 274      /**
 275       * Takes changed blocks and matches which rows in orig turned into which rows in final.
 276       *
 277       * Returns
 278       *    *_matches ( which rows match with which )
 279       *    *_rows ( order of rows in each column interleaved with blank rows as
 280       *        necessary )
 281       *
 282       * @since 2.6.0
 283       *
 284       * @param unknown_type $orig
 285       * @param unknown_type $final
 286       * @return unknown
 287       */
 288  	function interleave_changed_lines( $orig, $final ) {
 289  
 290          // Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
 291          $matches = array();
 292          foreach ( array_keys($orig) as $o ) {
 293              foreach ( array_keys($final) as $f ) {
 294                  $matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
 295              }
 296          }
 297          asort($matches); // Order by string distance.
 298  
 299          $orig_matches  = array();
 300          $final_matches = array();
 301  
 302          foreach ( $matches as $keys => $difference ) {
 303              list($o, $f) = explode(',', $keys);
 304              $o = (int) $o;
 305              $f = (int) $f;
 306  
 307              // Already have better matches for these guys
 308              if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
 309                  continue;
 310  
 311              // First match for these guys. Must be best match
 312              if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
 313                  $orig_matches[$o] = $f;
 314                  $final_matches[$f] = $o;
 315                  continue;
 316              }
 317  
 318              // Best match of this final is already taken?  Must mean this final is a new row.
 319              if ( isset($orig_matches[$o]) )
 320                  $final_matches[$f] = 'x';
 321  
 322              // Best match of this orig is already taken?  Must mean this orig is a deleted row.
 323              elseif ( isset($final_matches[$f]) )
 324                  $orig_matches[$o] = 'x';
 325          }
 326  
 327          // We read the text in this order
 328          ksort($orig_matches);
 329          ksort($final_matches);
 330  
 331          // Stores rows and blanks for each column.
 332          $orig_rows = $orig_rows_copy = array_keys($orig_matches);
 333          $final_rows = array_keys($final_matches);
 334  
 335          // Interleaves rows with blanks to keep matches aligned.
 336          // We may end up with some extraneous blank rows, but we'll just ignore them later.
 337          foreach ( $orig_rows_copy as $orig_row ) {
 338              $final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
 339              $orig_pos = (int) array_search($orig_row, $orig_rows, true);
 340  
 341              if ( false === $final_pos ) { // This orig is paired with a blank final.
 342                  array_splice( $final_rows, $orig_pos, 0, -1 );
 343              } elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
 344                  $diff_pos = $final_pos - $orig_pos;
 345                  while ( $diff_pos < 0 )
 346                      array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
 347              } elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
 348                  $diff_pos = $orig_pos - $final_pos;
 349                  while ( $diff_pos < 0 )
 350                      array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
 351              }
 352          }
 353  
 354          // Pad the ends with blank rows if the columns aren't the same length
 355          $diff_count = count($orig_rows) - count($final_rows);
 356          if ( $diff_count < 0 ) {
 357              while ( $diff_count < 0 )
 358                  array_push($orig_rows, $diff_count++);
 359          } elseif ( $diff_count > 0 ) {
 360              $diff_count = -1 * $diff_count;
 361              while ( $diff_count < 0 )
 362                  array_push($final_rows, $diff_count++);
 363          }
 364  
 365          return array($orig_matches, $final_matches, $orig_rows, $final_rows);
 366  
 367  /*
 368          // Debug
 369          echo "\n\n\n\n\n";
 370  
 371          echo "-- DEBUG Matches: Orig -> Final --";
 372  
 373          foreach ( $orig_matches as $o => $f ) {
 374              echo "\n\n\n\n\n";
 375              echo "ORIG: $o, FINAL: $f\n";
 376              var_dump($orig[$o],$final[$f]);
 377          }
 378          echo "\n\n\n\n\n";
 379  
 380          echo "-- DEBUG Matches: Final -> Orig --";
 381  
 382          foreach ( $final_matches as $f => $o ) {
 383              echo "\n\n\n\n\n";
 384              echo "FINAL: $f, ORIG: $o\n";
 385              var_dump($final[$f],$orig[$o]);
 386          }
 387          echo "\n\n\n\n\n";
 388  
 389          echo "-- DEBUG Rows: Orig -- Final --";
 390  
 391          echo "\n\n\n\n\n";
 392          foreach ( $orig_rows as $row => $o ) {
 393              if ( $o < 0 )
 394                  $o = 'X';
 395              $f = $final_rows[$row];
 396              if ( $f < 0 )
 397                  $f = 'X';
 398              echo "$o -- $f\n";
 399          }
 400          echo "\n\n\n\n\n";
 401  
 402          echo "-- END DEBUG --";
 403  
 404          echo "\n\n\n\n\n";
 405  
 406          return array($orig_matches, $final_matches, $orig_rows, $final_rows);
 407  */
 408      }
 409  
 410      /**
 411       * Computes a number that is intended to reflect the "distance" between two strings.
 412       *
 413       * @since 2.6.0
 414       *
 415       * @param string $string1
 416       * @param string $string2
 417       * @return int
 418       */
 419  	function compute_string_distance( $string1, $string2 ) {
 420          // Vectors containing character frequency for all chars in each string
 421          $chars1 = count_chars($string1);
 422          $chars2 = count_chars($string2);
 423  
 424          // L1-norm of difference vector.
 425          $difference = array_sum( array_map( array(&$this, 'difference'), $chars1, $chars2 ) );
 426  
 427          // $string1 has zero length? Odd. Give huge penalty by not dividing.
 428          if ( !$string1 )
 429              return $difference;
 430  
 431          // Return distance per charcter (of string1)
 432          return $difference / strlen($string1);
 433      }
 434  
 435      /**
 436       * @ignore
 437       * @since 2.6.0
 438       *
 439       * @param int $a
 440       * @param int $b
 441       * @return int
 442       */
 443  	function difference( $a, $b ) {
 444          return abs( $a - $b );
 445      }
 446  
 447  }
 448  
 449  /**
 450   * Better word splitting than the PEAR package provides.
 451   *
 452   * @since 2.6.0
 453   * @uses Text_Diff_Renderer_inline Extends
 454   */
 455  class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
 456  
 457      /**
 458       * @ignore
 459       * @since 2.6.0
 460       *
 461       * @param string $string
 462       * @param string $newlineEscape
 463       * @return string
 464       */
 465  	function _splitOnWords($string, $newlineEscape = "\n") {
 466          $string = str_replace("\0", '', $string);
 467          $words  = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
 468          $words  = str_replace( "\n", $newlineEscape, $words );
 469          return $words;
 470      }
 471  
 472  }