source: trunk/www.guidonia.net/wp/wp-includes/wp-diff.php@ 44

Last change on this file since 44 was 44, checked in by luciano, 15 years ago
File size: 12.1 KB
Line 
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
11if ( !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 */
26class 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 * PHP4 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 unknown
68 *
69 * @param array $params
70 */
71 function Text_Diff_Renderer_Table( $params = array() ) {
72 $parent = get_parent_class($this);
73 $this->$parent( $params );
74 }
75
76 /**
77 * @ignore
78 *
79 * @param string $header
80 * @return string
81 */
82 function _startBlock( $header ) {
83 return '';
84 }
85
86 /**
87 * @ignore
88 *
89 * @param array $lines
90 * @param string $prefix
91 */
92 function _lines( $lines, $prefix=' ' ) {
93 }
94
95 /**
96 * @ignore
97 *
98 * @param string $line HTML-escape the value.
99 * @return string
100 */
101 function addedLine( $line ) {
102 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
103 }
104
105 /**
106 * @ignore
107 *
108 * @param string $line HTML-escape the value.
109 * @return string
110 */
111 function deletedLine( $line ) {
112 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
113 }
114
115 /**
116 * @ignore
117 *
118 * @param string $line HTML-escape the value.
119 * @return string
120 */
121 function contextLine( $line ) {
122 return "<td> </td><td class='diff-context'>{$line}</td>";
123 }
124
125 /**
126 * @ignore
127 *
128 * @return string
129 */
130 function emptyLine() {
131 return '<td colspan="2">&nbsp;</td>';
132 }
133
134 /**
135 * @ignore
136 * @access private
137 *
138 * @param array $lines
139 * @param bool $encode
140 * @return string
141 */
142 function _added( $lines, $encode = true ) {
143 $r = '';
144 foreach ($lines as $line) {
145 if ( $encode )
146 $line = htmlspecialchars( $line );
147 $r .= '<tr>' . $this->emptyLine() . $this->addedLine( $line ) . "</tr>\n";
148 }
149 return $r;
150 }
151
152 /**
153 * @ignore
154 * @access private
155 *
156 * @param array $lines
157 * @param bool $encode
158 * @return string
159 */
160 function _deleted( $lines, $encode = true ) {
161 $r = '';
162 foreach ($lines as $line) {
163 if ( $encode )
164 $line = htmlspecialchars( $line );
165 $r .= '<tr>' . $this->deletedLine( $line ) . $this->emptyLine() . "</tr>\n";
166 }
167 return $r;
168 }
169
170 /**
171 * @ignore
172 * @access private
173 *
174 * @param array $lines
175 * @param bool $encode
176 * @return string
177 */
178 function _context( $lines, $encode = true ) {
179 $r = '';
180 foreach ($lines as $line) {
181 if ( $encode )
182 $line = htmlspecialchars( $line );
183 $r .= '<tr>' .
184 $this->contextLine( $line ) . $this->contextLine( $line ) . "</tr>\n";
185 }
186 return $r;
187 }
188
189 /**
190 * Process changed lines to do word-by-word diffs for extra highlighting.
191 *
192 * (TRAC style) sometimes these lines can actually be deleted or added rows.
193 * We do additional processing to figure that out
194 *
195 * @access private
196 * @since 2.6.0
197 *
198 * @param array $orig
199 * @param array $final
200 * @return string
201 */
202 function _changed( $orig, $final ) {
203 $r = '';
204
205 // Does the aforementioned additional processing
206 // *_matches tell what rows are "the same" in orig and final. Those pairs will be diffed to get word changes
207 // match is numeric: an index in other column
208 // match is 'X': no match. It is a new row
209 // *_rows are column vectors for the orig column and the final column.
210 // row >= 0: an indix of the $orig or $final array
211 // row < 0: a blank row for that column
212 list($orig_matches, $final_matches, $orig_rows, $final_rows) = $this->interleave_changed_lines( $orig, $final );
213
214
215 // These will hold the word changes as determined by an inline diff
216 $orig_diffs = array();
217 $final_diffs = array();
218
219 // Compute word diffs for each matched pair using the inline diff
220 foreach ( $orig_matches as $o => $f ) {
221 if ( is_numeric($o) && is_numeric($f) ) {
222 $text_diff = new Text_Diff( 'auto', array( array($orig[$o]), array($final[$f]) ) );
223 $renderer = new $this->inline_diff_renderer;
224 $diff = $renderer->render( $text_diff );
225
226 // If they're too different, don't include any <ins> or <dels>
227 if ( $diff_count = preg_match_all( '!(<ins>.*?</ins>|<del>.*?</del>)!', $diff, $diff_matches ) ) {
228 // length of all text between <ins> or <del>
229 $stripped_matches = strlen(strip_tags( join(' ', $diff_matches[0]) ));
230 // since we count lengith of text between <ins> or <del> (instead of picking just one),
231 // we double the length of chars not in those tags.
232 $stripped_diff = strlen(strip_tags( $diff )) * 2 - $stripped_matches;
233 $diff_ratio = $stripped_matches / $stripped_diff;
234 if ( $diff_ratio > $this->_diff_threshold )
235 continue; // Too different. Don't save diffs.
236 }
237
238 // Un-inline the diffs by removing del or ins
239 $orig_diffs[$o] = preg_replace( '|<ins>.*?</ins>|', '', $diff );
240 $final_diffs[$f] = preg_replace( '|<del>.*?</del>|', '', $diff );
241 }
242 }
243
244 foreach ( array_keys($orig_rows) as $row ) {
245 // Both columns have blanks. Ignore them.
246 if ( $orig_rows[$row] < 0 && $final_rows[$row] < 0 )
247 continue;
248
249 // If we have a word based diff, use it. Otherwise, use the normal line.
250 $orig_line = isset($orig_diffs[$orig_rows[$row]])
251 ? $orig_diffs[$orig_rows[$row]]
252 : htmlspecialchars($orig[$orig_rows[$row]]);
253 $final_line = isset($final_diffs[$final_rows[$row]])
254 ? $final_diffs[$final_rows[$row]]
255 : htmlspecialchars($final[$final_rows[$row]]);
256
257 if ( $orig_rows[$row] < 0 ) { // Orig is blank. This is really an added row.
258 $r .= $this->_added( array($final_line), false );
259 } elseif ( $final_rows[$row] < 0 ) { // Final is blank. This is really a deleted row.
260 $r .= $this->_deleted( array($orig_line), false );
261 } else { // A true changed row.
262 $r .= '<tr>' . $this->deletedLine( $orig_line ) . $this->addedLine( $final_line ) . "</tr>\n";
263 }
264 }
265
266 return $r;
267 }
268
269 /**
270 * Takes changed blocks and matches which rows in orig turned into which rows in final.
271 *
272 * Returns
273 * *_matches ( which rows match with which )
274 * *_rows ( order of rows in each column interleaved with blank rows as
275 * necessary )
276 *
277 * @since 2.6.0
278 *
279 * @param unknown_type $orig
280 * @param unknown_type $final
281 * @return unknown
282 */
283 function interleave_changed_lines( $orig, $final ) {
284
285 // Contains all pairwise string comparisons. Keys are such that this need only be a one dimensional array.
286 $matches = array();
287 foreach ( array_keys($orig) as $o ) {
288 foreach ( array_keys($final) as $f ) {
289 $matches["$o,$f"] = $this->compute_string_distance( $orig[$o], $final[$f] );
290 }
291 }
292 asort($matches); // Order by string distance.
293
294 $orig_matches = array();
295 $final_matches = array();
296
297 foreach ( $matches as $keys => $difference ) {
298 list($o, $f) = explode(',', $keys);
299 $o = (int) $o;
300 $f = (int) $f;
301
302 // Already have better matches for these guys
303 if ( isset($orig_matches[$o]) && isset($final_matches[$f]) )
304 continue;
305
306 // First match for these guys. Must be best match
307 if ( !isset($orig_matches[$o]) && !isset($final_matches[$f]) ) {
308 $orig_matches[$o] = $f;
309 $final_matches[$f] = $o;
310 continue;
311 }
312
313 // Best match of this final is already taken? Must mean this final is a new row.
314 if ( isset($orig_matches[$o]) )
315 $final_matches[$f] = 'x';
316
317 // Best match of this orig is already taken? Must mean this orig is a deleted row.
318 elseif ( isset($final_matches[$f]) )
319 $orig_matches[$o] = 'x';
320 }
321
322 // We read the text in this order
323 ksort($orig_matches);
324 ksort($final_matches);
325
326
327 // Stores rows and blanks for each column.
328 $orig_rows = $orig_rows_copy = array_keys($orig_matches);
329 $final_rows = array_keys($final_matches);
330
331 // Interleaves rows with blanks to keep matches aligned.
332 // We may end up with some extraneous blank rows, but we'll just ignore them later.
333 foreach ( $orig_rows_copy as $orig_row ) {
334 $final_pos = array_search($orig_matches[$orig_row], $final_rows, true);
335 $orig_pos = (int) array_search($orig_row, $orig_rows, true);
336
337 if ( false === $final_pos ) { // This orig is paired with a blank final.
338 array_splice( $final_rows, $orig_pos, 0, -1 );
339 } elseif ( $final_pos < $orig_pos ) { // This orig's match is up a ways. Pad final with blank rows.
340 $diff_pos = $final_pos - $orig_pos;
341 while ( $diff_pos < 0 )
342 array_splice( $final_rows, $orig_pos, 0, $diff_pos++ );
343 } elseif ( $final_pos > $orig_pos ) { // This orig's match is down a ways. Pad orig with blank rows.
344 $diff_pos = $orig_pos - $final_pos;
345 while ( $diff_pos < 0 )
346 array_splice( $orig_rows, $orig_pos, 0, $diff_pos++ );
347 }
348 }
349
350
351 // Pad the ends with blank rows if the columns aren't the same length
352 $diff_count = count($orig_rows) - count($final_rows);
353 if ( $diff_count < 0 ) {
354 while ( $diff_count < 0 )
355 array_push($orig_rows, $diff_count++);
356 } elseif ( $diff_count > 0 ) {
357 $diff_count = -1 * $diff_count;
358 while ( $diff_count < 0 )
359 array_push($final_rows, $diff_count++);
360 }
361
362 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
363
364/*
365 // Debug
366 echo "\n\n\n\n\n";
367
368 echo "-- DEBUG Matches: Orig -> Final --";
369
370 foreach ( $orig_matches as $o => $f ) {
371 echo "\n\n\n\n\n";
372 echo "ORIG: $o, FINAL: $f\n";
373 var_dump($orig[$o],$final[$f]);
374 }
375 echo "\n\n\n\n\n";
376
377 echo "-- DEBUG Matches: Final -> Orig --";
378
379 foreach ( $final_matches as $f => $o ) {
380 echo "\n\n\n\n\n";
381 echo "FINAL: $f, ORIG: $o\n";
382 var_dump($final[$f],$orig[$o]);
383 }
384 echo "\n\n\n\n\n";
385
386 echo "-- DEBUG Rows: Orig -- Final --";
387
388 echo "\n\n\n\n\n";
389 foreach ( $orig_rows as $row => $o ) {
390 if ( $o < 0 )
391 $o = 'X';
392 $f = $final_rows[$row];
393 if ( $f < 0 )
394 $f = 'X';
395 echo "$o -- $f\n";
396 }
397 echo "\n\n\n\n\n";
398
399 echo "-- END DEBUG --";
400
401 echo "\n\n\n\n\n";
402
403 return array($orig_matches, $final_matches, $orig_rows, $final_rows);
404*/
405 }
406
407 /**
408 * Computes a number that is intended to reflect the "distance" between two strings.
409 *
410 * @since 2.6.0
411 *
412 * @param string $string1
413 * @param string $string2
414 * @return int
415 */
416 function compute_string_distance( $string1, $string2 ) {
417 // Vectors containing character frequency for all chars in each string
418 $chars1 = count_chars($string1);
419 $chars2 = count_chars($string2);
420
421 // L1-norm of difference vector.
422 $difference = array_sum( array_map( array(&$this, 'difference'), $chars1, $chars2 ) );
423
424 // $string1 has zero length? Odd. Give huge penalty by not dividing.
425 if ( !$string1 )
426 return $difference;
427
428 // Return distance per charcter (of string1)
429 return $difference / strlen($string1);
430 }
431
432 /**
433 * @ignore
434 * @since 2.6.0
435 *
436 * @param int $a
437 * @param int $b
438 * @return int
439 */
440 function difference( $a, $b ) {
441 return abs( $a - $b );
442 }
443
444}
445
446/**
447 * Better word splitting than the PEAR package provides.
448 *
449 * @since 2.6.0
450 * @uses Text_Diff_Renderer_inline Extends
451 */
452class WP_Text_Diff_Renderer_inline extends Text_Diff_Renderer_inline {
453
454 /**
455 * @ignore
456 * @since 2.6.0
457 *
458 * @param string $string
459 * @param string $newlineEscape
460 * @return string
461 */
462 function _splitOnWords($string, $newlineEscape = "\n") {
463 $string = str_replace("\0", '', $string);
464 $words = preg_split( '/([^\w])/u', $string, -1, PREG_SPLIT_DELIM_CAPTURE );
465 $words = str_replace( "\n", $newlineEscape, $words );
466 return $words;
467 }
468
469}
470
471?>
Note: See TracBrowser for help on using the repository browser.