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 var $_leading_context_lines = 10000;
37 * @see Text_Diff_Renderer::_trailing_context_lines
42 var $_trailing_context_lines = 10000;
45 * {@internal Missing Description}}
51 var $_diff_threshold = 0.6;
54 * Inline display helper object name.
60 var $inline_diff_renderer = 'WP_Text_Diff_Renderer_inline';
63 * Should we show the split view or not
69 var $_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 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 function _startBlock( $header ) {
100 * @param string $prefix
102 function _lines( $lines, $prefix=' ' ) {
108 * @param string $line HTML-escape the value.
111 function addedLine( $line ) {
112 return "<td class='diff-addedline'>{$line}</td>";
119 * @param string $line HTML-escape the value.
122 function deletedLine( $line ) {
123 return "<td class='diff-deletedline'>{$line}</td>";
129 * @param string $line HTML-escape the value.
132 function contextLine( $line ) {
133 return "<td class='diff-context'>{$line}</td>";
141 function emptyLine() {
142 return '<td> </td>';
149 * @param array $lines
150 * @param bool $encode
153 function _added( $lines, $encode = true ) {
155 foreach ($lines as $line) {
157 $line = htmlspecialchars( $line );
158 if ( $this->_show_split_view ) {
159 $r .= '<tr>' . $this->emptyLine() . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
161 $r .= '<tr>' . $this->addedLine( $line ) . "</tr>\n";
171 * @param array $lines
172 * @param bool $encode
175 function _deleted( $lines, $encode = true ) {
177 foreach ($lines as $line) {
179 $line = htmlspecialchars( $line );
180 if ( $this->_show_split_view ) {
181 $r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . $this->emptyLine() . "</tr>\n";
183 $r .= '<tr>' . $this->deletedLine( $line ) . "</tr>\n";
194 * @param array $lines
195 * @param bool $encode
198 function _context( $lines, $encode = true ) {
200 foreach ($lines as $line) {
202 $line = htmlspecialchars( $line );
203 if ( $this->_show_split_view ) {
204 $r .= '<tr>' . $this->contextLine( $line ) . $this->emptyLine() . $this->contextLine( $line ) . "</tr>\n";
206 $r .= '<tr>' . $this->contextLine( $line ) . "</tr>\n";
213 * Process changed lines to do word-by-word diffs for extra highlighting.
215 * (TRAC style) sometimes these lines can actually be deleted or added rows.
216 * We do additional processing to figure that out
222 * @param array $final
225 function _changed( $orig, $final ) {
228 // Does the aforementioned additional processing
229 // *_matches tell what rows are "the same" in orig and final. Those pairs will be diffed to get word changes
230 // match is numeric: an index in other column
231 // match is 'X': no match. It is a new row
232 // *_rows are column vectors for the orig column and the final column.
233 // row >= 0: an indix of the $orig or $final array
234 // row < 0: a blank row for that column
235 list($orig_matches, $final_matches, $orig_rows, $final_rows) = $this->interleave_changed_lines( $orig, $final );
237 // These will hold the word changes as determined by an inline diff
238 $orig_diffs = array();
239 $final_diffs = array();
241 // Compute word diffs for each matched pair using the inline diff
242 foreach ( $orig_matches as $o => $f ) {
243 if ( is_numeric($o) && is_numeric($f) ) {
244 $text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
245 $renderer = new $this->inline_diff_renderer;
246 $diff = $renderer->render( $text_diff );
248 // If they're too different, don't include any <ins> or <dels>
249 if ( $diff_count = preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
250 // length of all text between <ins> or <del>
251 $stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
252 // since we count lengith of text between <ins> or <del> (instead of picking just one),
253 // we double the length of chars not in those tags.
254 $stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
255 $diff_ratio = $stripped_matches / $stripped_diff;
256 if ( $diff_ratio > $this->_diff_threshold )
257 continue; // Too different. Don't save diffs.
260 // Un-inline the diffs by removing del or ins
261 $orig_diffs[$o] = preg_replace( '|<ins>.*?</ins>|', '', $diff );
262 $final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
266 foreach ( array_keys($orig_rows) as $row ) {
267 // Both columns have blanks. Ignore them.
268 if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
271 // If we have a word based diff, use it. Otherwise, use the normal line.
272 if ( isset( $orig_diffs[$orig_rows[$row]] ) )
273 $orig_line = $orig_diffs[$orig_rows[$row]];
274 elseif ( isset( $orig[$orig_rows[$row]] ) )
275 $orig_line = htmlspecialchars($orig[$orig_rows[$row]]);
279 if ( isset( $final_diffs[$final_rows[$row]] ) )
280 $final_line = $final_diffs[$final_rows[$row]];
281 elseif ( isset( $final[$final_rows[$row]] ) )
282 $final_line = htmlspecialchars($final[$final_rows[$row]]);
286 if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
287 $r .= $this->_added( array($final_line), false );
288 } elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
289 $r .= $this->_deleted( array($orig_line), false );
290 } else { // A true changed row.
291 if ( $this->_show_split_view ) {
292 $r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->emptyLine() . $this->addedLine( $final_line ) . "</tr>\n";
294 $r .= '<tr>' . $this->deletedLine( $orig_line ) . "</tr><tr>" . $this->addedLine( $final_line ) . "</tr>\n";
303 * Takes changed blocks and matches which rows in orig turned into which rows in final.
306 * *_matches ( which rows match with which )
307 * *_rows ( order of rows in each column interleaved with blank rows as
312 * @param unknown_type $orig
313 * @param unknown_type $final
316 function interleave_changed_lines( $orig, $final ) {
318 // Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
320 foreach ( array_keys($orig) as $o ) {
321 foreach ( array_keys($final) as $f ) {
322 $matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
325 asort($matches); // Order by string distance.
327 $orig_matches = array();
328 $final_matches = array();
330 foreach ( $matches as $keys => $difference ) {
331 list($o, $f) = explode(',', $keys);
335 // Already have better matches for these guys
336 if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
339 // First match for these guys. Must be best match
340 if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
341 $orig_matches[$o] = $f;
342 $final_matches[$f] = $o;
346 // Best match of this final is already taken? Must mean this final is a new row.
347 if ( isset($orig_matches[$o]) )
348 $final_matches[$f] = 'x';
350 // Best match of this orig is already taken? Must mean this orig is a deleted row.
351 elseif ( isset($final_matches[$f]) )
352 $orig_matches[$o] = 'x';
355 // We read the text in this order
356 ksort($orig_matches);
357 ksort($final_matches);
359 // Stores rows and blanks for each column.
360 $orig_rows = $orig_rows_copy = array_keys($orig_matches);
361 $final_rows = array_keys($final_matches);
363 // Interleaves rows with blanks to keep matches aligned.
364 // We may end up with some extraneous blank rows, but we'll just ignore them later.
365 foreach ( $orig_rows_copy as $orig_row ) {
366 $final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
367 $orig_pos = (int) array_search($orig_row, $orig_rows, true);
369 if ( false === $final_pos ) { // This orig is paired with a blank final.
370 array_splice( $final_rows, $orig_pos, 0, -1 );
371 } elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
372 $diff_pos = $final_pos - $orig_pos;
373 while ( $diff_pos < 0 )
374 array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
375 } elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
376 $diff_pos = $orig_pos - $final_pos;
377 while ( $diff_pos < 0 )
378 array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
382 // Pad the ends with blank rows if the columns aren't the same length
383 $diff_count = count($orig_rows) - count($final_rows);
384 if ( $diff_count < 0 ) {
385 while ( $diff_count < 0 )
386 array_push($orig_rows, $diff_count++);
387 } elseif ( $diff_count > 0 ) {
388 $diff_count = -1 * $diff_count;
389 while ( $diff_count < 0 )
390 array_push($final_rows, $diff_count++);
393 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
399 echo "-- DEBUG Matches: Orig -> Final --";
401 foreach ( $orig_matches as $o => $f ) {
403 echo "ORIG: $o, FINAL: $f\n";
404 var_dump($orig[$o],$final[$f]);
408 echo "-- DEBUG Matches: Final -> Orig --";
410 foreach ( $final_matches as $f => $o ) {
412 echo "FINAL: $f, ORIG: $o\n";
413 var_dump($final[$f],$orig[$o]);
417 echo "-- DEBUG Rows: Orig -- Final --";
420 foreach ( $orig_rows as $row => $o ) {
423 $f = $final_rows[$row];
430 echo "-- END DEBUG --";
434 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
439 * Computes a number that is intended to reflect the "distance" between two strings.
443 * @param string $string1
444 * @param string $string2
447 function compute_string_distance( $string1, $string2 ) {
448 // Vectors containing character frequency for all chars in each string
449 $chars1 = count_chars($string1);
450 $chars2 = count_chars($string2);
452 // L1-norm of difference vector.
453 $difference = array_sum( array_map( array($this, 'difference'), $chars1, $chars2 ) );
455 // $string1 has zero length? Odd. Give huge penalty by not dividing.
459 // Return distance per charcter (of string1)
460 return $difference / strlen($string1);
471 function difference( $a, $b ) {
472 return abs( $a - $b );
478 * Better word splitting than the PEAR package provides.
481 * @uses Text_Diff_Renderer_inline Extends
483 class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
489 * @param string $string
490 * @param string $newlineEscape
493 function _splitOnWords($string, $newlineEscape = "\n") {
494 $string = str_replace("\0", '', $string);
495 $words = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
496 $words = str_replace( "\n", $newlineEscape, $words );