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 * Constructor - Call parent constructor with params array.
65 * This will set class properties based on the key value pairs in the array.
69 * @param array $params
71 function __construct( $params = array() ) {
72 parent::__construct( $params );
78 * @param string $header
81 function _startBlock( $header ) {
89 * @param string $prefix
91 function _lines( $lines, $prefix=' ' ) {
97 * @param string $line HTML-escape the value.
100 function addedLine( $line ) {
101 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
107 * @param string $line HTML-escape the value.
110 function deletedLine( $line ) {
111 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
117 * @param string $line HTML-escape the value.
120 function contextLine( $line ) {
121 return "<td> </td><td class='diff-context'>{$line}</td>";
129 function emptyLine() {
130 return '<td colspan="2"> </td>';
137 * @param array $lines
138 * @param bool $encode
141 function _added( $lines, $encode = true ) {
143 foreach ($lines as $line) {
145 $line = htmlspecialchars( $line );
146 $r .= '<tr>' . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
155 * @param array $lines
156 * @param bool $encode
159 function _deleted( $lines, $encode = true ) {
161 foreach ($lines as $line) {
163 $line = htmlspecialchars( $line );
164 $r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . "</tr>\n";
173 * @param array $lines
174 * @param bool $encode
177 function _context( $lines, $encode = true ) {
179 foreach ($lines as $line) {
181 $line = htmlspecialchars( $line );
183 $this->contextLine( $line ) . $this->contextLine( $line ) . "</tr>\n";
189 * Process changed lines to do word-by-word diffs for extra highlighting.
191 * (TRAC style) sometimes these lines can actually be deleted or added rows.
192 * We do additional processing to figure that out
198 * @param array $final
201 function _changed( $orig, $final ) {
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 );
214 // These will hold the word changes as determined by an inline diff
215 $orig_diffs = array();
216 $final_diffs = array();
218 // Compute word diffs for each matched pair using the inline diff
219 foreach ( $orig_matches as $o => $f ) {
220 if ( is_numeric($o) && is_numeric($f) ) {
221 $text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
222 $renderer = new $this->inline_diff_renderer;
223 $diff = $renderer->render( $text_diff );
225 // If they're too different, don't include any <ins> or <dels>
226 if ( $diff_count = preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
227 // length of all text between <ins> or <del>
228 $stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
229 // since we count lengith of text between <ins> or <del> (instead of picking just one),
230 // we double the length of chars not in those tags.
231 $stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
232 $diff_ratio = $stripped_matches / $stripped_diff;
233 if ( $diff_ratio > $this->_diff_threshold )
234 continue; // Too different. Don't save diffs.
237 // Un-inline the diffs by removing del or ins
238 $orig_diffs[$o] = preg_replace( '|<ins>.*?</ins>|', '', $diff );
239 $final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
243 foreach ( array_keys($orig_rows) as $row ) {
244 // Both columns have blanks. Ignore them.
245 if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
248 // If we have a word based diff, use it. Otherwise, use the normal line.
249 if ( isset( $orig_diffs[$orig_rows[$row]] ) )
250 $orig_line = $orig_diffs[$orig_rows[$row]];
251 elseif ( isset( $orig[$orig_rows[$row]] ) )
252 $orig_line = htmlspecialchars($orig[$orig_rows[$row]]);
256 if ( isset( $final_diffs[$final_rows[$row]] ) )
257 $final_line = $final_diffs[$final_rows[$row]];
258 elseif ( isset( $final[$final_rows[$row]] ) )
259 $final_line = htmlspecialchars($final[$final_rows[$row]]);
263 if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
264 $r .= $this->_added( array($final_line), false );
265 } elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
266 $r .= $this->_deleted( array($orig_line), false );
267 } else { // A true changed row.
268 $r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->addedLine( $final_line ) . "</tr>\n";
276 * Takes changed blocks and matches which rows in orig turned into which rows in final.
279 * *_matches ( which rows match with which )
280 * *_rows ( order of rows in each column interleaved with blank rows as
285 * @param unknown_type $orig
286 * @param unknown_type $final
289 function interleave_changed_lines( $orig, $final ) {
291 // Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
293 foreach ( array_keys($orig) as $o ) {
294 foreach ( array_keys($final) as $f ) {
295 $matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
298 asort($matches); // Order by string distance.
300 $orig_matches = array();
301 $final_matches = array();
303 foreach ( $matches as $keys => $difference ) {
304 list($o, $f) = explode(',', $keys);
308 // Already have better matches for these guys
309 if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
312 // First match for these guys. Must be best match
313 if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
314 $orig_matches[$o] = $f;
315 $final_matches[$f] = $o;
319 // Best match of this final is already taken? Must mean this final is a new row.
320 if ( isset($orig_matches[$o]) )
321 $final_matches[$f] = 'x';
323 // Best match of this orig is already taken? Must mean this orig is a deleted row.
324 elseif ( isset($final_matches[$f]) )
325 $orig_matches[$o] = 'x';
328 // We read the text in this order
329 ksort($orig_matches);
330 ksort($final_matches);
333 // Stores rows and blanks for each column.
334 $orig_rows = $orig_rows_copy = array_keys($orig_matches);
335 $final_rows = array_keys($final_matches);
337 // Interleaves rows with blanks to keep matches aligned.
338 // We may end up with some extraneous blank rows, but we'll just ignore them later.
339 foreach ( $orig_rows_copy as $orig_row ) {
340 $final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
341 $orig_pos = (int) array_search($orig_row, $orig_rows, true);
343 if ( false === $final_pos ) { // This orig is paired with a blank final.
344 array_splice( $final_rows, $orig_pos, 0, -1 );
345 } elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
346 $diff_pos = $final_pos - $orig_pos;
347 while ( $diff_pos < 0 )
348 array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
349 } elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
350 $diff_pos = $orig_pos - $final_pos;
351 while ( $diff_pos < 0 )
352 array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
357 // Pad the ends with blank rows if the columns aren't the same length
358 $diff_count = count($orig_rows) - count($final_rows);
359 if ( $diff_count < 0 ) {
360 while ( $diff_count < 0 )
361 array_push($orig_rows, $diff_count++);
362 } elseif ( $diff_count > 0 ) {
363 $diff_count = -1 * $diff_count;
364 while ( $diff_count < 0 )
365 array_push($final_rows, $diff_count++);
368 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
374 echo "-- DEBUG Matches: Orig -> Final --";
376 foreach ( $orig_matches as $o => $f ) {
378 echo "ORIG: $o, FINAL: $f\n";
379 var_dump($orig[$o],$final[$f]);
383 echo "-- DEBUG Matches: Final -> Orig --";
385 foreach ( $final_matches as $f => $o ) {
387 echo "FINAL: $f, ORIG: $o\n";
388 var_dump($final[$f],$orig[$o]);
392 echo "-- DEBUG Rows: Orig -- Final --";
395 foreach ( $orig_rows as $row => $o ) {
398 $f = $final_rows[$row];
405 echo "-- END DEBUG --";
409 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
414 * Computes a number that is intended to reflect the "distance" between two strings.
418 * @param string $string1
419 * @param string $string2
422 function compute_string_distance( $string1, $string2 ) {
423 // Vectors containing character frequency for all chars in each string
424 $chars1 = count_chars($string1);
425 $chars2 = count_chars($string2);
427 // L1-norm of difference vector.
428 $difference = array_sum( array_map( array(&$this, 'difference'), $chars1, $chars2 ) );
430 // $string1 has zero length? Odd. Give huge penalty by not dividing.
434 // Return distance per charcter (of string1)
435 return $difference / strlen($string1);
446 function difference( $a, $b ) {
447 return abs( $a - $b );
453 * Better word splitting than the PEAR package provides.
456 * @uses Text_Diff_Renderer_inline Extends
458 class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
464 * @param string $string
465 * @param string $newlineEscape
468 function _splitOnWords($string, $newlineEscape = "\n") {
469 $string = str_replace("\0", '', $string);
470 $words = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
471 $words = str_replace( "\n", $newlineEscape, $words );