wp/wp-includes/wp-diff.php
changeset 7 cf61fcea0001
parent 5 5e2f62d02dcd
child 9 177826044cd9
equal deleted inserted replaced
6:490d5cc509ed 7:cf61fcea0001
     6  *
     6  *
     7  * @package WordPress
     7  * @package WordPress
     8  * @subpackage Diff
     8  * @subpackage Diff
     9  */
     9  */
    10 
    10 
    11 if ( !class_exists( 'Text_Diff' ) ) {
    11 if ( ! class_exists( 'Text_Diff', false ) ) {
    12 	/** Text_Diff class */
    12 	/** Text_Diff class */
    13 	require( dirname(__FILE__).'/Text/Diff.php' );
    13 	require( ABSPATH . WPINC . '/Text/Diff.php' );
    14 	/** Text_Diff_Renderer class */
    14 	/** Text_Diff_Renderer class */
    15 	require( dirname(__FILE__).'/Text/Diff/Renderer.php' );
    15 	require( ABSPATH . WPINC . '/Text/Diff/Renderer.php' );
    16 	/** Text_Diff_Renderer_inline class */
    16 	/** Text_Diff_Renderer_inline class */
    17 	require( dirname(__FILE__).'/Text/Diff/Renderer/inline.php' );
    17 	require( ABSPATH . WPINC . '/Text/Diff/Renderer/inline.php' );
    18 }
    18 }
    19 
    19 
    20 /**
    20 require( ABSPATH . WPINC . '/class-wp-text-diff-renderer-table.php' );
    21  * Table renderer to display the diff lines.
    21 require( ABSPATH . WPINC . '/class-wp-text-diff-renderer-inline.php' );
    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 public
       
    32 	 * @since 2.6.0
       
    33 	 */
       
    34 	public $_leading_context_lines  = 10000;
       
    35 
       
    36 	/**
       
    37 	 * @see Text_Diff_Renderer::_trailing_context_lines
       
    38 	 * @var int
       
    39 	 * @access public
       
    40 	 * @since 2.6.0
       
    41 	 */
       
    42 	public $_trailing_context_lines = 10000;
       
    43 
       
    44 	/**
       
    45 	 * Threshold for when a diff should be saved or omitted.
       
    46 	 *
       
    47 	 * @var float
       
    48 	 * @access protected
       
    49 	 * @since 2.6.0
       
    50 	 */
       
    51 	protected $_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 	protected $inline_diff_renderer = 'WP_Text_Diff_Renderer_inline';
       
    61 
       
    62 	/**
       
    63 	 * Should we show the split view or not
       
    64 	 *
       
    65 	 * @var string
       
    66 	 * @access protected
       
    67 	 * @since 3.6.0
       
    68 	 */
       
    69 	protected $_show_split_view = true;
       
    70 
       
    71 	protected $compat_fields = array( '_show_split_view', 'inline_diff_renderer', '_diff_threshold' );
       
    72 
       
    73 	/**
       
    74 	 * Constructor - Call parent constructor with params array.
       
    75 	 *
       
    76 	 * This will set class properties based on the key value pairs in the array.
       
    77 	 *
       
    78 	 * @since 2.6.0
       
    79 	 *
       
    80 	 * @param array $params
       
    81 	 */
       
    82 	public function __construct( $params = array() ) {
       
    83 		parent::__construct( $params );
       
    84 		if ( isset( $params[ 'show_split_view' ] ) )
       
    85 			$this->_show_split_view = $params[ 'show_split_view' ];
       
    86 	}
       
    87 
       
    88 	/**
       
    89 	 * @ignore
       
    90 	 *
       
    91 	 * @param string $header
       
    92 	 * @return string
       
    93 	 */
       
    94 	public function _startBlock( $header ) {
       
    95 		return '';
       
    96 	}
       
    97 
       
    98 	/**
       
    99 	 * @ignore
       
   100 	 *
       
   101 	 * @param array $lines
       
   102 	 * @param string $prefix
       
   103 	 */
       
   104 	public function _lines( $lines, $prefix=' ' ) {
       
   105 	}
       
   106 
       
   107 	/**
       
   108 	 * @ignore
       
   109 	 *
       
   110 	 * @param string $line HTML-escape the value.
       
   111 	 * @return string
       
   112 	 */
       
   113 	public function addedLine( $line ) {
       
   114 		return "<td class='diff-addedline'>{$line}</td>";
       
   115 
       
   116 	}
       
   117 
       
   118 	/**
       
   119 	 * @ignore
       
   120 	 *
       
   121 	 * @param string $line HTML-escape the value.
       
   122 	 * @return string
       
   123 	 */
       
   124 	public function deletedLine( $line ) {
       
   125 		return "<td class='diff-deletedline'>{$line}</td>";
       
   126 	}
       
   127 
       
   128 	/**
       
   129 	 * @ignore
       
   130 	 *
       
   131 	 * @param string $line HTML-escape the value.
       
   132 	 * @return string
       
   133 	 */
       
   134 	public function contextLine( $line ) {
       
   135 		return "<td class='diff-context'>{$line}</td>";
       
   136 	}
       
   137 
       
   138 	/**
       
   139 	 * @ignore
       
   140 	 *
       
   141 	 * @return string
       
   142 	 */
       
   143 	public function emptyLine() {
       
   144 		return '<td>&nbsp;</td>';
       
   145 	}
       
   146 
       
   147 	/**
       
   148 	 * @ignore
       
   149 	 * @access public
       
   150 	 *
       
   151 	 * @param array $lines
       
   152 	 * @param bool $encode
       
   153 	 * @return string
       
   154 	 */
       
   155 	public function _added( $lines, $encode = true ) {
       
   156 		$r = '';
       
   157 		foreach ($lines as $line) {
       
   158 			if ( $encode ) {
       
   159 				$processed_line = htmlspecialchars( $line );
       
   160 
       
   161 				/**
       
   162 				 * Contextually filter a diffed line.
       
   163 				 *
       
   164 				 * Filters TextDiff processing of diffed line. By default, diffs are processed with
       
   165 				 * htmlspecialchars. Use this filter to remove or change the processing. Passes a context
       
   166 				 * indicating if the line is added, deleted or unchanged.
       
   167 				 *
       
   168 				 * @since 4.1.0
       
   169 				 *
       
   170 				 * @param String $processed_line The processed diffed line.
       
   171 				 * @param String $line           The unprocessed diffed line.
       
   172 		 		 * @param string null            The line context. Values are 'added', 'deleted' or 'unchanged'.
       
   173 				 */
       
   174 				$line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'added' );
       
   175 			}
       
   176 
       
   177 			if ( $this->_show_split_view ) {
       
   178 				$r .= '<tr>' . $this->emptyLine() . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
       
   179 			} else {
       
   180 				$r .= '<tr>' . $this->addedLine( $line ) . "</tr>\n";
       
   181 			}
       
   182 		}
       
   183 		return $r;
       
   184 	}
       
   185 
       
   186 	/**
       
   187 	 * @ignore
       
   188 	 * @access public
       
   189 	 *
       
   190 	 * @param array $lines
       
   191 	 * @param bool $encode
       
   192 	 * @return string
       
   193 	 */
       
   194 	public function _deleted( $lines, $encode = true ) {
       
   195 		$r = '';
       
   196 		foreach ($lines as $line) {
       
   197 			if ( $encode ) {
       
   198 				$processed_line = htmlspecialchars( $line );
       
   199 
       
   200 				/** This filter is documented in wp-includes/wp-diff.php */
       
   201 				$line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'deleted' );
       
   202 			}
       
   203 			if ( $this->_show_split_view ) {
       
   204 				$r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . $this->emptyLine() . "</tr>\n";
       
   205 			} else {
       
   206 				$r .= '<tr>' . $this->deletedLine( $line ) . "</tr>\n";
       
   207 			}
       
   208 
       
   209 		}
       
   210 		return $r;
       
   211 	}
       
   212 
       
   213 	/**
       
   214 	 * @ignore
       
   215 	 * @access public
       
   216 	 *
       
   217 	 * @param array $lines
       
   218 	 * @param bool $encode
       
   219 	 * @return string
       
   220 	 */
       
   221 	public function _context( $lines, $encode = true ) {
       
   222 		$r = '';
       
   223 		foreach ($lines as $line) {
       
   224 			if ( $encode ) {
       
   225 				$processed_line = htmlspecialchars( $line );
       
   226 
       
   227 				/** This filter is documented in wp-includes/wp-diff.php */
       
   228 				$line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'unchanged' );
       
   229 			}
       
   230 			if (  $this->_show_split_view ) {
       
   231 				$r .= '<tr>' . $this->contextLine( $line ) . $this->emptyLine() . $this->contextLine( $line )  . "</tr>\n";
       
   232 			} else {
       
   233 				$r .= '<tr>' . $this->contextLine( $line ) . "</tr>\n";
       
   234 			}
       
   235 		}
       
   236 		return $r;
       
   237 	}
       
   238 
       
   239 	/**
       
   240 	 * Process changed lines to do word-by-word diffs for extra highlighting.
       
   241 	 *
       
   242 	 * (TRAC style) sometimes these lines can actually be deleted or added rows.
       
   243 	 * We do additional processing to figure that out
       
   244 	 *
       
   245 	 * @access public
       
   246 	 * @since 2.6.0
       
   247 	 *
       
   248 	 * @param array $orig
       
   249 	 * @param array $final
       
   250 	 * @return string
       
   251 	 */
       
   252 	public function _changed( $orig, $final ) {
       
   253 		$r = '';
       
   254 
       
   255 		// Does the aforementioned additional processing
       
   256 		// *_matches tell what rows are "the same" in orig and final. Those pairs will be diffed to get word changes
       
   257 		//	match is numeric: an index in other column
       
   258 		//	match is 'X': no match. It is a new row
       
   259 		// *_rows are column vectors for the orig column and the final column.
       
   260 		//	row >= 0: an indix of the $orig or $final array
       
   261 		//	row  < 0: a blank row for that column
       
   262 		list($orig_matches, $final_matches, $orig_rows, $final_rows) = $this->interleave_changed_lines( $orig, $final );
       
   263 
       
   264 		// These will hold the word changes as determined by an inline diff
       
   265 		$orig_diffs  = array();
       
   266 		$final_diffs = array();
       
   267 
       
   268 		// Compute word diffs for each matched pair using the inline diff
       
   269 		foreach ( $orig_matches as $o => $f ) {
       
   270 			if ( is_numeric($o) && is_numeric($f) ) {
       
   271 				$text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
       
   272 				$renderer = new $this->inline_diff_renderer;
       
   273 				$diff = $renderer->render( $text_diff );
       
   274 
       
   275 				// If they're too different, don't include any <ins> or <dels>
       
   276 				if ( preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
       
   277 					// length of all text between <ins> or <del>
       
   278 					$stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
       
   279 					// since we count lengith of text between <ins> or <del> (instead of picking just one),
       
   280 					//	we double the length of chars not in those tags.
       
   281 					$stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
       
   282 					$diff_ratio = $stripped_matches / $stripped_diff;
       
   283 					if ( $diff_ratio > $this->_diff_threshold )
       
   284 						continue; // Too different. Don't save diffs.
       
   285 				}
       
   286 
       
   287 				// Un-inline the diffs by removing del or ins
       
   288 				$orig_diffs[$o]  = preg_replace( '|<ins>.*?</ins>|', '', $diff );
       
   289 				$final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
       
   290 			}
       
   291 		}
       
   292 
       
   293 		foreach ( array_keys($orig_rows) as $row ) {
       
   294 			// Both columns have blanks. Ignore them.
       
   295 			if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
       
   296 				continue;
       
   297 
       
   298 			// If we have a word based diff, use it. Otherwise, use the normal line.
       
   299 			if ( isset( $orig_diffs[$orig_rows[$row]] ) )
       
   300 				$orig_line = $orig_diffs[$orig_rows[$row]];
       
   301 			elseif ( isset( $orig[$orig_rows[$row]] ) )
       
   302 				$orig_line = htmlspecialchars($orig[$orig_rows[$row]]);
       
   303 			else
       
   304 				$orig_line = '';
       
   305 
       
   306 			if ( isset( $final_diffs[$final_rows[$row]] ) )
       
   307 				$final_line = $final_diffs[$final_rows[$row]];
       
   308 			elseif ( isset( $final[$final_rows[$row]] ) )
       
   309 				$final_line = htmlspecialchars($final[$final_rows[$row]]);
       
   310 			else
       
   311 				$final_line = '';
       
   312 
       
   313 			if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
       
   314 				$r .= $this->_added( array($final_line), false );
       
   315 			} elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
       
   316 				$r .= $this->_deleted( array($orig_line), false );
       
   317 			} else { // A true changed row.
       
   318 				if ( $this->_show_split_view ) {
       
   319 					$r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->emptyLine() . $this->addedLine( $final_line ) . "</tr>\n";
       
   320 				} else {
       
   321 					$r .= '<tr>' . $this->deletedLine( $orig_line ) . "</tr><tr>" . $this->addedLine( $final_line ) . "</tr>\n";
       
   322 				}
       
   323 			}
       
   324 		}
       
   325 
       
   326 		return $r;
       
   327 	}
       
   328 
       
   329 	/**
       
   330 	 * Takes changed blocks and matches which rows in orig turned into which rows in final.
       
   331 	 *
       
   332 	 * Returns
       
   333 	 *	*_matches ( which rows match with which )
       
   334 	 *	*_rows ( order of rows in each column interleaved with blank rows as
       
   335 	 *		necessary )
       
   336 	 *
       
   337 	 * @since 2.6.0
       
   338 	 *
       
   339 	 * @param array $orig
       
   340 	 * @param array $final
       
   341 	 * @return array
       
   342 	 */
       
   343 	public function interleave_changed_lines( $orig, $final ) {
       
   344 
       
   345 		// Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
       
   346 		$matches = array();
       
   347 		foreach ( array_keys($orig) as $o ) {
       
   348 			foreach ( array_keys($final) as $f ) {
       
   349 				$matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
       
   350 			}
       
   351 		}
       
   352 		asort($matches); // Order by string distance.
       
   353 
       
   354 		$orig_matches  = array();
       
   355 		$final_matches = array();
       
   356 
       
   357 		foreach ( $matches as $keys => $difference ) {
       
   358 			list($o, $f) = explode(',', $keys);
       
   359 			$o = (int) $o;
       
   360 			$f = (int) $f;
       
   361 
       
   362 			// Already have better matches for these guys
       
   363 			if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
       
   364 				continue;
       
   365 
       
   366 			// First match for these guys. Must be best match
       
   367 			if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
       
   368 				$orig_matches[$o] = $f;
       
   369 				$final_matches[$f] = $o;
       
   370 				continue;
       
   371 			}
       
   372 
       
   373 			// Best match of this final is already taken?  Must mean this final is a new row.
       
   374 			if ( isset($orig_matches[$o]) )
       
   375 				$final_matches[$f] = 'x';
       
   376 
       
   377 			// Best match of this orig is already taken?  Must mean this orig is a deleted row.
       
   378 			elseif ( isset($final_matches[$f]) )
       
   379 				$orig_matches[$o] = 'x';
       
   380 		}
       
   381 
       
   382 		// We read the text in this order
       
   383 		ksort($orig_matches);
       
   384 		ksort($final_matches);
       
   385 
       
   386 		// Stores rows and blanks for each column.
       
   387 		$orig_rows = $orig_rows_copy = array_keys($orig_matches);
       
   388 		$final_rows = array_keys($final_matches);
       
   389 
       
   390 		// Interleaves rows with blanks to keep matches aligned.
       
   391 		// We may end up with some extraneous blank rows, but we'll just ignore them later.
       
   392 		foreach ( $orig_rows_copy as $orig_row ) {
       
   393 			$final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
       
   394 			$orig_pos = (int) array_search($orig_row, $orig_rows, true);
       
   395 
       
   396 			if ( false === $final_pos ) { // This orig is paired with a blank final.
       
   397 				array_splice( $final_rows, $orig_pos, 0, -1 );
       
   398 			} elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
       
   399 				$diff_pos = $final_pos - $orig_pos;
       
   400 				while ( $diff_pos < 0 )
       
   401 					array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
       
   402 			} elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
       
   403 				$diff_pos = $orig_pos - $final_pos;
       
   404 				while ( $diff_pos < 0 )
       
   405 					array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
       
   406 			}
       
   407 		}
       
   408 
       
   409 		// Pad the ends with blank rows if the columns aren't the same length
       
   410 		$diff_count = count($orig_rows) - count($final_rows);
       
   411 		if ( $diff_count < 0 ) {
       
   412 			while ( $diff_count < 0 )
       
   413 				array_push($orig_rows, $diff_count++);
       
   414 		} elseif ( $diff_count > 0 ) {
       
   415 			$diff_count = -1 * $diff_count;
       
   416 			while ( $diff_count < 0 )
       
   417 				array_push($final_rows, $diff_count++);
       
   418 		}
       
   419 
       
   420 		return array($orig_matches, $final_matches, $orig_rows, $final_rows);
       
   421 	}
       
   422 
       
   423 	/**
       
   424 	 * Computes a number that is intended to reflect the "distance" between two strings.
       
   425 	 *
       
   426 	 * @since 2.6.0
       
   427 	 *
       
   428 	 * @param string $string1
       
   429 	 * @param string $string2
       
   430 	 * @return int
       
   431 	 */
       
   432 	public function compute_string_distance( $string1, $string2 ) {
       
   433 		// Vectors containing character frequency for all chars in each string
       
   434 		$chars1 = count_chars($string1);
       
   435 		$chars2 = count_chars($string2);
       
   436 
       
   437 		// L1-norm of difference vector.
       
   438 		$difference = array_sum( array_map( array($this, 'difference'), $chars1, $chars2 ) );
       
   439 
       
   440 		// $string1 has zero length? Odd. Give huge penalty by not dividing.
       
   441 		if ( !$string1 )
       
   442 			return $difference;
       
   443 
       
   444 		// Return distance per character (of string1).
       
   445 		return $difference / strlen($string1);
       
   446 	}
       
   447 
       
   448 	/**
       
   449 	 * @ignore
       
   450 	 * @since 2.6.0
       
   451 	 *
       
   452 	 * @param int $a
       
   453 	 * @param int $b
       
   454 	 * @return int
       
   455 	 */
       
   456 	public function difference( $a, $b ) {
       
   457 		return abs( $a - $b );
       
   458 	}
       
   459 
       
   460 	/**
       
   461 	 * Make private properties readable for backwards compatibility.
       
   462 	 *
       
   463 	 * @since 4.0.0
       
   464 	 * @access public
       
   465 	 *
       
   466 	 * @param string $name Property to get.
       
   467 	 * @return mixed Property.
       
   468 	 */
       
   469 	public function __get( $name ) {
       
   470 		if ( in_array( $name, $this->compat_fields ) ) {
       
   471 			return $this->$name;
       
   472 		}
       
   473 	}
       
   474 
       
   475 	/**
       
   476 	 * Make private properties settable for backwards compatibility.
       
   477 	 *
       
   478 	 * @since 4.0.0
       
   479 	 * @access public
       
   480 	 *
       
   481 	 * @param string $name  Property to check if set.
       
   482 	 * @param mixed  $value Property value.
       
   483 	 * @return mixed Newly-set property.
       
   484 	 */
       
   485 	public function __set( $name, $value ) {
       
   486 		if ( in_array( $name, $this->compat_fields ) ) {
       
   487 			return $this->$name = $value;
       
   488 		}
       
   489 	}
       
   490 
       
   491 	/**
       
   492 	 * Make private properties checkable for backwards compatibility.
       
   493 	 *
       
   494 	 * @since 4.0.0
       
   495 	 * @access public
       
   496 	 *
       
   497 	 * @param string $name Property to check if set.
       
   498 	 * @return bool Whether the property is set.
       
   499 	 */
       
   500 	public function __isset( $name ) {
       
   501 		if ( in_array( $name, $this->compat_fields ) ) {
       
   502 			return isset( $this->$name );
       
   503 		}
       
   504 	}
       
   505 
       
   506 	/**
       
   507 	 * Make private properties un-settable for backwards compatibility.
       
   508 	 *
       
   509 	 * @since 4.0.0
       
   510 	 * @access public
       
   511 	 *
       
   512 	 * @param string $name Property to unset.
       
   513 	 */
       
   514 	public function __unset( $name ) {
       
   515 		if ( in_array( $name, $this->compat_fields ) ) {
       
   516 			unset( $this->$name );
       
   517 		}
       
   518 	}
       
   519 }
       
   520 
       
   521 /**
       
   522  * Better word splitting than the PEAR package provides.
       
   523  *
       
   524  * @since 2.6.0
       
   525  * @uses Text_Diff_Renderer_inline Extends
       
   526  */
       
   527 class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
       
   528 
       
   529 	/**
       
   530 	 * @ignore
       
   531 	 * @since 2.6.0
       
   532 	 *
       
   533 	 * @param string $string
       
   534 	 * @param string $newlineEscape
       
   535 	 * @return string
       
   536 	 */
       
   537 	public function _splitOnWords($string, $newlineEscape = "\n") {
       
   538 		$string = str_replace("\0", '', $string);
       
   539 		$words  = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
       
   540 		$words  = str_replace( "\n", $newlineEscape, $words );
       
   541 		return $words;
       
   542 	}
       
   543 
       
   544 }