3 * WordPress Diff bastard child of old MediaWiki Diff Formatter.
5 * Basically all that remains is the table structure and some method names.
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' );
21 * Table renderer to display the diff lines.
24 * @uses Text_Diff_Renderer Extends
26 class WP_Text_Diff_Renderer_Table extends Text_Diff_Renderer {
29 * @see Text_Diff_Renderer::_leading_context_lines
34 public $_leading_context_lines = 10000;
37 * @see Text_Diff_Renderer::_trailing_context_lines
42 public $_trailing_context_lines = 10000;
45 * {@internal Missing Description}}
51 protected $_diff_threshold = 0.6;
54 * Inline display helper object name.
60 protected $inline_diff_renderer = 'WP_Text_Diff_Renderer_inline';
63 * Should we show the split view or not
69 protected $_show_split_view = true;
72 * Constructor - Call parent constructor with params array.
74 * This will set class properties based on the key value pairs in the array.
78 * @param array $params
80 public function __construct( $params = array() ) {
81 parent::__construct( $params );
82 if ( isset( $params[ 'show_split_view' ] ) )
83 $this->_show_split_view = $params[ 'show_split_view' ];
89 * @param string $header
92 public function _startBlock( $header ) {
100 * @param string $prefix
102 public function _lines( $lines, $prefix=' ' ) {
108 * @param string $line HTML-escape the value.
111 public function addedLine( $line ) {
112 return "<td class='diff-addedline'>{$line}</td>";
119 * @param string $line HTML-escape the value.
122 public function deletedLine( $line ) {
123 return "<td class='diff-deletedline'>{$line}</td>";
129 * @param string $line HTML-escape the value.
132 public function contextLine( $line ) {
133 return "<td class='diff-context'>{$line}</td>";
141 public function emptyLine() {
142 return '<td> </td>';
149 * @param array $lines
150 * @param bool $encode
153 public function _added( $lines, $encode = true ) {
155 foreach ($lines as $line) {
157 $processed_line = htmlspecialchars( $line );
160 * Contextually filter a diffed line.
162 * Filters TextDiff processing of diffed line. By default, diffs are processed with
163 * htmlspecialchars. Use this filter to remove or change the processing. Passes a context
164 * indicating if the line is added, deleted or unchanged.
168 * @param String $processed_line The processed diffed line.
169 * @param String $line The unprocessed diffed line.
170 * @param string null The line context. Values are 'added', 'deleted' or 'unchanged'.
172 $line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'added' );
175 if ( $this->_show_split_view ) {
176 $r .= '<tr>' . $this->emptyLine() . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
178 $r .= '<tr>' . $this->addedLine( $line ) . "</tr>\n";
188 * @param array $lines
189 * @param bool $encode
192 public function _deleted( $lines, $encode = true ) {
194 foreach ($lines as $line) {
196 $processed_line = htmlspecialchars( $line );
198 /** This filter is documented in wp-includes/wp-diff.php */
199 $line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'deleted' );
201 if ( $this->_show_split_view ) {
202 $r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . $this->emptyLine() . "</tr>\n";
204 $r .= '<tr>' . $this->deletedLine( $line ) . "</tr>\n";
215 * @param array $lines
216 * @param bool $encode
219 public function _context( $lines, $encode = true ) {
221 foreach ($lines as $line) {
223 $processed_line = htmlspecialchars( $line );
225 /** This filter is documented in wp-includes/wp-diff.php */
226 $line = apply_filters( 'process_text_diff_html', $processed_line, $line, 'unchanged' );
228 if ( $this->_show_split_view ) {
229 $r .= '<tr>' . $this->contextLine( $line ) . $this->emptyLine() . $this->contextLine( $line ) . "</tr>\n";
231 $r .= '<tr>' . $this->contextLine( $line ) . "</tr>\n";
238 * Process changed lines to do word-by-word diffs for extra highlighting.
240 * (TRAC style) sometimes these lines can actually be deleted or added rows.
241 * We do additional processing to figure that out
247 * @param array $final
250 public function _changed( $orig, $final ) {
253 // Does the aforementioned additional processing
254 // *_matches tell what rows are "the same" in orig and final. Those pairs will be diffed to get word changes
255 // match is numeric: an index in other column
256 // match is 'X': no match. It is a new row
257 // *_rows are column vectors for the orig column and the final column.
258 // row >= 0: an indix of the $orig or $final array
259 // row < 0: a blank row for that column
260 list($orig_matches, $final_matches, $orig_rows, $final_rows) = $this->interleave_changed_lines( $orig, $final );
262 // These will hold the word changes as determined by an inline diff
263 $orig_diffs = array();
264 $final_diffs = array();
266 // Compute word diffs for each matched pair using the inline diff
267 foreach ( $orig_matches as $o => $f ) {
268 if ( is_numeric($o) && is_numeric($f) ) {
269 $text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
270 $renderer = new $this->inline_diff_renderer;
271 $diff = $renderer->render( $text_diff );
273 // If they're too different, don't include any <ins> or <dels>
274 if ( $diff_count = preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
275 // length of all text between <ins> or <del>
276 $stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
277 // since we count lengith of text between <ins> or <del> (instead of picking just one),
278 // we double the length of chars not in those tags.
279 $stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
280 $diff_ratio = $stripped_matches / $stripped_diff;
281 if ( $diff_ratio > $this->_diff_threshold )
282 continue; // Too different. Don't save diffs.
285 // Un-inline the diffs by removing del or ins
286 $orig_diffs[$o] = preg_replace( '|<ins>.*?</ins>|', '', $diff );
287 $final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
291 foreach ( array_keys($orig_rows) as $row ) {
292 // Both columns have blanks. Ignore them.
293 if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
296 // If we have a word based diff, use it. Otherwise, use the normal line.
297 if ( isset( $orig_diffs[$orig_rows[$row]] ) )
298 $orig_line = $orig_diffs[$orig_rows[$row]];
299 elseif ( isset( $orig[$orig_rows[$row]] ) )
300 $orig_line = htmlspecialchars($orig[$orig_rows[$row]]);
304 if ( isset( $final_diffs[$final_rows[$row]] ) )
305 $final_line = $final_diffs[$final_rows[$row]];
306 elseif ( isset( $final[$final_rows[$row]] ) )
307 $final_line = htmlspecialchars($final[$final_rows[$row]]);
311 if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
312 $r .= $this->_added( array($final_line), false );
313 } elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
314 $r .= $this->_deleted( array($orig_line), false );
315 } else { // A true changed row.
316 if ( $this->_show_split_view ) {
317 $r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->emptyLine() . $this->addedLine( $final_line ) . "</tr>\n";
319 $r .= '<tr>' . $this->deletedLine( $orig_line ) . "</tr><tr>" . $this->addedLine( $final_line ) . "</tr>\n";
328 * Takes changed blocks and matches which rows in orig turned into which rows in final.
331 * *_matches ( which rows match with which )
332 * *_rows ( order of rows in each column interleaved with blank rows as
338 * @param array $final
341 public function interleave_changed_lines( $orig, $final ) {
343 // Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
345 foreach ( array_keys($orig) as $o ) {
346 foreach ( array_keys($final) as $f ) {
347 $matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
350 asort($matches); // Order by string distance.
352 $orig_matches = array();
353 $final_matches = array();
355 foreach ( $matches as $keys => $difference ) {
356 list($o, $f) = explode(',', $keys);
360 // Already have better matches for these guys
361 if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
364 // First match for these guys. Must be best match
365 if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
366 $orig_matches[$o] = $f;
367 $final_matches[$f] = $o;
371 // Best match of this final is already taken? Must mean this final is a new row.
372 if ( isset($orig_matches[$o]) )
373 $final_matches[$f] = 'x';
375 // Best match of this orig is already taken? Must mean this orig is a deleted row.
376 elseif ( isset($final_matches[$f]) )
377 $orig_matches[$o] = 'x';
380 // We read the text in this order
381 ksort($orig_matches);
382 ksort($final_matches);
384 // Stores rows and blanks for each column.
385 $orig_rows = $orig_rows_copy = array_keys($orig_matches);
386 $final_rows = array_keys($final_matches);
388 // Interleaves rows with blanks to keep matches aligned.
389 // We may end up with some extraneous blank rows, but we'll just ignore them later.
390 foreach ( $orig_rows_copy as $orig_row ) {
391 $final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
392 $orig_pos = (int) array_search($orig_row, $orig_rows, true);
394 if ( false === $final_pos ) { // This orig is paired with a blank final.
395 array_splice( $final_rows, $orig_pos, 0, -1 );
396 } elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
397 $diff_pos = $final_pos - $orig_pos;
398 while ( $diff_pos < 0 )
399 array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
400 } elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
401 $diff_pos = $orig_pos - $final_pos;
402 while ( $diff_pos < 0 )
403 array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
407 // Pad the ends with blank rows if the columns aren't the same length
408 $diff_count = count($orig_rows) - count($final_rows);
409 if ( $diff_count < 0 ) {
410 while ( $diff_count < 0 )
411 array_push($orig_rows, $diff_count++);
412 } elseif ( $diff_count > 0 ) {
413 $diff_count = -1 * $diff_count;
414 while ( $diff_count < 0 )
415 array_push($final_rows, $diff_count++);
418 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
422 * Computes a number that is intended to reflect the "distance" between two strings.
426 * @param string $string1
427 * @param string $string2
430 public function compute_string_distance( $string1, $string2 ) {
431 // Vectors containing character frequency for all chars in each string
432 $chars1 = count_chars($string1);
433 $chars2 = count_chars($string2);
435 // L1-norm of difference vector.
436 $difference = array_sum( array_map( array($this, 'difference'), $chars1, $chars2 ) );
438 // $string1 has zero length? Odd. Give huge penalty by not dividing.
442 // Return distance per character (of string1).
443 return $difference / strlen($string1);
454 public function difference( $a, $b ) {
455 return abs( $a - $b );
459 * Make private properties readable for backwards compatibility.
464 * @param string $name Property to get.
465 * @return mixed Property.
467 public function __get( $name ) {
472 * Make private properties settable for backwards compatibility.
477 * @param string $name Property to check if set.
478 * @param mixed $value Property value.
479 * @return mixed Newly-set property.
481 public function __set( $name, $value ) {
482 return $this->$name = $value;
486 * Make private properties checkable for backwards compatibility.
491 * @param string $name Property to check if set.
492 * @return bool Whether the property is set.
494 public function __isset( $name ) {
495 return isset( $this->$name );
499 * Make private properties un-settable for backwards compatibility.
504 * @param string $name Property to unset.
506 public function __unset( $name ) {
507 unset( $this->$name );
511 * Make private/protected methods readable for backwards compatibility.
516 * @param callable $name Method to call.
517 * @param array $arguments Arguments to pass when calling.
518 * @return mixed|bool Return value of the callback, false otherwise.
520 public function __call( $name, $arguments ) {
521 return call_user_func_array( array( $this, $name ), $arguments );
526 * Better word splitting than the PEAR package provides.
529 * @uses Text_Diff_Renderer_inline Extends
531 class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
537 * @param string $string
538 * @param string $newlineEscape
541 public function _splitOnWords($string, $newlineEscape = "\n") {
542 $string = str_replace("\0", '', $string);
543 $words = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
544 $words = str_replace( "\n", $newlineEscape, $words );