1
+ − 1
<?php
+ − 2
/**
+ − 3
* Class used internally by Diff to actually compute the diffs. This class is
+ − 4
* implemented using native PHP code.
+ − 5
*
+ − 6
* The algorithm used here is mostly lifted from the perl module
+ − 7
* Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
+ − 8
* http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
+ − 9
*
+ − 10
* More ideas are taken from:
+ − 11
* http://www.ics.uci.edu/~eppstein/161/960229.html
+ − 12
*
+ − 13
* Some ideas (and a bit of code) are taken from analyze.c, of GNU
+ − 14
* diffutils-2.7, which can be found at:
+ − 15
* ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
+ − 16
*
+ − 17
* Some ideas (subdivision by NCHUNKS > 2, and some optimizations) are from
+ − 18
* Geoffrey T. Dairiki <dairiki@dairiki.org>. The original PHP version of this
+ − 19
* code was written by him, and is used/adapted with his permission.
+ − 20
*
+ − 21
* $Horde: framework/Text_Diff/Diff/Engine/native.php,v 1.3 2006/01/06 15:56:52 jan Exp $
+ − 22
*
+ − 23
* @author Geoffrey T. Dairiki <dairiki@dairiki.org>
+ − 24
* @package Text_Diff
+ − 25
*
+ − 26
* @access private
+ − 27
*/
+ − 28
class Text_Diff_Engine_native {
+ − 29
+ − 30
function diff($from_lines, $to_lines)
+ − 31
{
+ − 32
array_walk($from_lines, array('Text_Diff', 'trimNewlines'));
+ − 33
array_walk($to_lines, array('Text_Diff', 'trimNewlines'));
+ − 34
+ − 35
$n_from = count($from_lines);
+ − 36
$n_to = count($to_lines);
+ − 37
+ − 38
$this->xchanged = $this->ychanged = array();
+ − 39
$this->xv = $this->yv = array();
+ − 40
$this->xind = $this->yind = array();
+ − 41
unset($this->seq);
+ − 42
unset($this->in_seq);
+ − 43
unset($this->lcs);
+ − 44
+ − 45
// Skip leading common lines.
+ − 46
for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
+ − 47
if ($from_lines[$skip] !== $to_lines[$skip]) {
+ − 48
break;
+ − 49
}
+ − 50
$this->xchanged[$skip] = $this->ychanged[$skip] = false;
+ − 51
}
+ − 52
+ − 53
// Skip trailing common lines.
+ − 54
$xi = $n_from; $yi = $n_to;
+ − 55
for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
+ − 56
if ($from_lines[$xi] !== $to_lines[$yi]) {
+ − 57
break;
+ − 58
}
+ − 59
$this->xchanged[$xi] = $this->ychanged[$yi] = false;
+ − 60
}
+ − 61
+ − 62
// Ignore lines which do not exist in both files.
+ − 63
for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
+ − 64
$xhash[$from_lines[$xi]] = 1;
+ − 65
}
+ − 66
for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
+ − 67
$line = $to_lines[$yi];
+ − 68
if (($this->ychanged[$yi] = empty($xhash[$line]))) {
+ − 69
continue;
+ − 70
}
+ − 71
$yhash[$line] = 1;
+ − 72
$this->yv[] = $line;
+ − 73
$this->yind[] = $yi;
+ − 74
}
+ − 75
for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
+ − 76
$line = $from_lines[$xi];
+ − 77
if (($this->xchanged[$xi] = empty($yhash[$line]))) {
+ − 78
continue;
+ − 79
}
+ − 80
$this->xv[] = $line;
+ − 81
$this->xind[] = $xi;
+ − 82
}
+ − 83
+ − 84
// Find the LCS.
+ − 85
$this->_compareseq(0, count($this->xv), 0, count($this->yv));
+ − 86
+ − 87
// Merge edits when possible.
+ − 88
$this->_shiftBoundaries($from_lines, $this->xchanged, $this->ychanged);
+ − 89
$this->_shiftBoundaries($to_lines, $this->ychanged, $this->xchanged);
+ − 90
+ − 91
// Compute the edit operations.
+ − 92
$edits = array();
+ − 93
$xi = $yi = 0;
+ − 94
while ($xi < $n_from || $yi < $n_to) {
+ − 95
assert($yi < $n_to || $this->xchanged[$xi]);
+ − 96
assert($xi < $n_from || $this->ychanged[$yi]);
+ − 97
+ − 98
// Skip matching "snake".
+ − 99
$copy = array();
+ − 100
while ($xi < $n_from && $yi < $n_to
+ − 101
&& !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
+ − 102
$copy[] = $from_lines[$xi++];
+ − 103
++$yi;
+ − 104
}
+ − 105
if ($copy) {
+ − 106
$edits[] = &new Text_Diff_Op_copy($copy);
+ − 107
}
+ − 108
+ − 109
// Find deletes & adds.
+ − 110
$delete = array();
+ − 111
while ($xi < $n_from && $this->xchanged[$xi]) {
+ − 112
$delete[] = $from_lines[$xi++];
+ − 113
}
+ − 114
+ − 115
$add = array();
+ − 116
while ($yi < $n_to && $this->ychanged[$yi]) {
+ − 117
$add[] = $to_lines[$yi++];
+ − 118
}
+ − 119
+ − 120
if ($delete && $add) {
+ − 121
$edits[] = &new Text_Diff_Op_change($delete, $add);
+ − 122
} elseif ($delete) {
+ − 123
$edits[] = &new Text_Diff_Op_delete($delete);
+ − 124
} elseif ($add) {
+ − 125
$edits[] = &new Text_Diff_Op_add($add);
+ − 126
}
+ − 127
}
+ − 128
+ − 129
return $edits;
+ − 130
}
+ − 131
+ − 132
/**
+ − 133
* Divides the Largest Common Subsequence (LCS) of the sequences (XOFF,
+ − 134
* XLIM) and (YOFF, YLIM) into NCHUNKS approximately equally sized
+ − 135
* segments.
+ − 136
*
+ − 137
* Returns (LCS, PTS). LCS is the length of the LCS. PTS is an array of
+ − 138
* NCHUNKS+1 (X, Y) indexes giving the diving points between sub
+ − 139
* sequences. The first sub-sequence is contained in (X0, X1), (Y0, Y1),
+ − 140
* the second in (X1, X2), (Y1, Y2) and so on. Note that (X0, Y0) ==
+ − 141
* (XOFF, YOFF) and (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
+ − 142
*
+ − 143
* This function assumes that the first lines of the specified portions of
+ − 144
* the two files do not match, and likewise that the last lines do not
+ − 145
* match. The caller must trim matching lines from the beginning and end
+ − 146
* of the portions it is going to specify.
+ − 147
*/
+ − 148
function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks)
+ − 149
{
+ − 150
$flip = false;
+ − 151
+ − 152
if ($xlim - $xoff > $ylim - $yoff) {
+ − 153
/* Things seems faster (I'm not sure I understand why) when the
+ − 154
* shortest sequence is in X. */
+ − 155
$flip = true;
+ − 156
list ($xoff, $xlim, $yoff, $ylim)
+ − 157
= array($yoff, $ylim, $xoff, $xlim);
+ − 158
}
+ − 159
+ − 160
if ($flip) {
+ − 161
for ($i = $ylim - 1; $i >= $yoff; $i--) {
+ − 162
$ymatches[$this->xv[$i]][] = $i;
+ − 163
}
+ − 164
} else {
+ − 165
for ($i = $ylim - 1; $i >= $yoff; $i--) {
+ − 166
$ymatches[$this->yv[$i]][] = $i;
+ − 167
}
+ − 168
}
+ − 169
+ − 170
$this->lcs = 0;
+ − 171
$this->seq[0]= $yoff - 1;
+ − 172
$this->in_seq = array();
+ − 173
$ymids[0] = array();
+ − 174
+ − 175
$numer = $xlim - $xoff + $nchunks - 1;
+ − 176
$x = $xoff;
+ − 177
for ($chunk = 0; $chunk < $nchunks; $chunk++) {
+ − 178
if ($chunk > 0) {
+ − 179
for ($i = 0; $i <= $this->lcs; $i++) {
+ − 180
$ymids[$i][$chunk - 1] = $this->seq[$i];
+ − 181
}
+ − 182
}
+ − 183
+ − 184
$x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
+ − 185
for (; $x < $x1; $x++) {
+ − 186
$line = $flip ? $this->yv[$x] : $this->xv[$x];
+ − 187
if (empty($ymatches[$line])) {
+ − 188
continue;
+ − 189
}
+ − 190
$matches = $ymatches[$line];
+ − 191
foreach ($matches as $y) {
+ − 192
if (empty($this->in_seq[$y])) {
+ − 193
$k = $this->_lcsPos($y);
+ − 194
assert($k > 0);
+ − 195
$ymids[$k] = $ymids[$k - 1];
+ − 196
break;
+ − 197
}
+ − 198
}
+ − 199
+ − 200
while (list($junk, $y) = each($matches)) {
+ − 201
if ($y > $this->seq[$k - 1]) {
+ − 202
assert($y < $this->seq[$k]);
+ − 203
/* Optimization: this is a common case: next match is
+ − 204
* just replacing previous match. */
+ − 205
$this->in_seq[$this->seq[$k]] = false;
+ − 206
$this->seq[$k] = $y;
+ − 207
$this->in_seq[$y] = 1;
+ − 208
} elseif (empty($this->in_seq[$y])) {
+ − 209
$k = $this->_lcsPos($y);
+ − 210
assert($k > 0);
+ − 211
$ymids[$k] = $ymids[$k - 1];
+ − 212
}
+ − 213
}
+ − 214
}
+ − 215
}
+ − 216
+ − 217
$seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
+ − 218
$ymid = $ymids[$this->lcs];
+ − 219
for ($n = 0; $n < $nchunks - 1; $n++) {
+ − 220
$x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
+ − 221
$y1 = $ymid[$n] + 1;
+ − 222
$seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
+ − 223
}
+ − 224
$seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
+ − 225
+ − 226
return array($this->lcs, $seps);
+ − 227
}
+ − 228
+ − 229
function _lcsPos($ypos)
+ − 230
{
+ − 231
$end = $this->lcs;
+ − 232
if ($end == 0 || $ypos > $this->seq[$end]) {
+ − 233
$this->seq[++$this->lcs] = $ypos;
+ − 234
$this->in_seq[$ypos] = 1;
+ − 235
return $this->lcs;
+ − 236
}
+ − 237
+ − 238
$beg = 1;
+ − 239
while ($beg < $end) {
+ − 240
$mid = (int)(($beg + $end) / 2);
+ − 241
if ($ypos > $this->seq[$mid]) {
+ − 242
$beg = $mid + 1;
+ − 243
} else {
+ − 244
$end = $mid;
+ − 245
}
+ − 246
}
+ − 247
+ − 248
assert($ypos != $this->seq[$end]);
+ − 249
+ − 250
$this->in_seq[$this->seq[$end]] = false;
+ − 251
$this->seq[$end] = $ypos;
+ − 252
$this->in_seq[$ypos] = 1;
+ − 253
return $end;
+ − 254
}
+ − 255
+ − 256
/**
+ − 257
* Finds LCS of two sequences.
+ − 258
*
+ − 259
* The results are recorded in the vectors $this->{x,y}changed[], by
+ − 260
* storing a 1 in the element for each line that is an insertion or
+ − 261
* deletion (ie. is not in the LCS).
+ − 262
*
+ − 263
* The subsequence of file 0 is (XOFF, XLIM) and likewise for file 1.
+ − 264
*
+ − 265
* Note that XLIM, YLIM are exclusive bounds. All line numbers are
+ − 266
* origin-0 and discarded lines are not counted.
+ − 267
*/
+ − 268
function _compareseq ($xoff, $xlim, $yoff, $ylim)
+ − 269
{
+ − 270
/* Slide down the bottom initial diagonal. */
+ − 271
while ($xoff < $xlim && $yoff < $ylim
+ − 272
&& $this->xv[$xoff] == $this->yv[$yoff]) {
+ − 273
++$xoff;
+ − 274
++$yoff;
+ − 275
}
+ − 276
+ − 277
/* Slide up the top initial diagonal. */
+ − 278
while ($xlim > $xoff && $ylim > $yoff
+ − 279
&& $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
+ − 280
--$xlim;
+ − 281
--$ylim;
+ − 282
}
+ − 283
+ − 284
if ($xoff == $xlim || $yoff == $ylim) {
+ − 285
$lcs = 0;
+ − 286
} else {
+ − 287
/* This is ad hoc but seems to work well. $nchunks =
+ − 288
* sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5); $nchunks =
+ − 289
* max(2,min(8,(int)$nchunks)); */
+ − 290
$nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
+ − 291
list($lcs, $seps)
+ − 292
= $this->_diag($xoff, $xlim, $yoff, $ylim, $nchunks);
+ − 293
}
+ − 294
+ − 295
if ($lcs == 0) {
+ − 296
/* X and Y sequences have no common subsequence: mark all
+ − 297
* changed. */
+ − 298
while ($yoff < $ylim) {
+ − 299
$this->ychanged[$this->yind[$yoff++]] = 1;
+ − 300
}
+ − 301
while ($xoff < $xlim) {
+ − 302
$this->xchanged[$this->xind[$xoff++]] = 1;
+ − 303
}
+ − 304
} else {
+ − 305
/* Use the partitions to split this problem into subproblems. */
+ − 306
reset($seps);
+ − 307
$pt1 = $seps[0];
+ − 308
while ($pt2 = next($seps)) {
+ − 309
$this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
+ − 310
$pt1 = $pt2;
+ − 311
}
+ − 312
}
+ − 313
}
+ − 314
+ − 315
/**
+ − 316
* Adjusts inserts/deletes of identical lines to join changes as much as
+ − 317
* possible.
+ − 318
*
+ − 319
* We do something when a run of changed lines include a line at one end
+ − 320
* and has an excluded, identical line at the other. We are free to
+ − 321
* choose which identical line is included. `compareseq' usually chooses
+ − 322
* the one at the beginning, but usually it is cleaner to consider the
+ − 323
* following identical line to be the "change".
+ − 324
*
+ − 325
* This is extracted verbatim from analyze.c (GNU diffutils-2.7).
+ − 326
*/
+ − 327
function _shiftBoundaries($lines, &$changed, $other_changed)
+ − 328
{
+ − 329
$i = 0;
+ − 330
$j = 0;
+ − 331
+ − 332
assert('count($lines) == count($changed)');
+ − 333
$len = count($lines);
+ − 334
$other_len = count($other_changed);
+ − 335
+ − 336
while (1) {
+ − 337
/* Scan forward to find the beginning of another run of
+ − 338
* changes. Also keep track of the corresponding point in the
+ − 339
* other file.
+ − 340
*
+ − 341
* Throughout this code, $i and $j are adjusted together so that
+ − 342
* the first $i elements of $changed and the first $j elements of
+ − 343
* $other_changed both contain the same number of zeros (unchanged
+ − 344
* lines).
+ − 345
*
+ − 346
* Furthermore, $j is always kept so that $j == $other_len or
+ − 347
* $other_changed[$j] == false. */
+ − 348
while ($j < $other_len && $other_changed[$j]) {
+ − 349
$j++;
+ − 350
}
+ − 351
+ − 352
while ($i < $len && ! $changed[$i]) {
+ − 353
assert('$j < $other_len && ! $other_changed[$j]');
+ − 354
$i++; $j++;
+ − 355
while ($j < $other_len && $other_changed[$j]) {
+ − 356
$j++;
+ − 357
}
+ − 358
}
+ − 359
+ − 360
if ($i == $len) {
+ − 361
break;
+ − 362
}
+ − 363
+ − 364
$start = $i;
+ − 365
+ − 366
/* Find the end of this run of changes. */
+ − 367
while (++$i < $len && $changed[$i]) {
+ − 368
continue;
+ − 369
}
+ − 370
+ − 371
do {
+ − 372
/* Record the length of this run of changes, so that we can
+ − 373
* later determine whether the run has grown. */
+ − 374
$runlength = $i - $start;
+ − 375
+ − 376
/* Move the changed region back, so long as the previous
+ − 377
* unchanged line matches the last changed one. This merges
+ − 378
* with previous changed regions. */
+ − 379
while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
+ − 380
$changed[--$start] = 1;
+ − 381
$changed[--$i] = false;
+ − 382
while ($start > 0 && $changed[$start - 1]) {
+ − 383
$start--;
+ − 384
}
+ − 385
assert('$j > 0');
+ − 386
while ($other_changed[--$j]) {
+ − 387
continue;
+ − 388
}
+ − 389
assert('$j >= 0 && !$other_changed[$j]');
+ − 390
}
+ − 391
+ − 392
/* Set CORRESPONDING to the end of the changed run, at the
+ − 393
* last point where it corresponds to a changed run in the
+ − 394
* other file. CORRESPONDING == LEN means no such point has
+ − 395
* been found. */
+ − 396
$corresponding = $j < $other_len ? $i : $len;
+ − 397
+ − 398
/* Move the changed region forward, so long as the first
+ − 399
* changed line matches the following unchanged one. This
+ − 400
* merges with following changed regions. Do this second, so
+ − 401
* that if there are no merges, the changed region is moved
+ − 402
* forward as far as possible. */
+ − 403
while ($i < $len && $lines[$start] == $lines[$i]) {
+ − 404
$changed[$start++] = false;
+ − 405
$changed[$i++] = 1;
+ − 406
while ($i < $len && $changed[$i]) {
+ − 407
$i++;
+ − 408
}
+ − 409
+ − 410
assert('$j < $other_len && ! $other_changed[$j]');
+ − 411
$j++;
+ − 412
if ($j < $other_len && $other_changed[$j]) {
+ − 413
$corresponding = $i;
+ − 414
while ($j < $other_len && $other_changed[$j]) {
+ − 415
$j++;
+ − 416
}
+ − 417
}
+ − 418
}
+ − 419
} while ($runlength != $i - $start);
+ − 420
+ − 421
/* If possible, move the fully-merged run of changes back to a
+ − 422
* corresponding run in the other file. */
+ − 423
while ($corresponding < $i) {
+ − 424
$changed[--$start] = 1;
+ − 425
$changed[--$i] = 0;
+ − 426
assert('$j > 0');
+ − 427
while ($other_changed[--$j]) {
+ − 428
continue;
+ − 429
}
+ − 430
assert('$j >= 0 && !$other_changed[$j]');
+ − 431
}
+ − 432
}
+ − 433
}
+ − 434
+ − 435
}