Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
| Comment: | Use a more precise (though slower) diff algorithm when working with small strings whose differences are hard to detect. |
|---|---|
| Downloads: | Tarball | ZIP archive |
| Timelines: | family | ancestors | descendants | both | trunk |
| Files: | files | file ages | folders |
| SHA1: |
1e08452c6574e0afe895699776d331c5 |
| User & Date: | drh 2010-08-21 17:32:43.000 |
Context
|
2010-08-23
| ||
| 19:33 | Update SQLite to the 3.7.2 release candidate. ... (check-in: 6fd2ff1b09 user: drh tags: trunk) | |
|
2010-08-21
| ||
| 17:32 | Use a more precise (though slower) diff algorithm when working with small strings whose differences are hard to detect. ... (check-in: 1e08452c65 user: drh tags: trunk) | |
| 04:06 | The command-line "diff" does not ignore whitespace at the end of lines and it generates a well-formed patch file that can be fed directly into "patch -p 0". Ticket [a9f7b23c2e376af]. GUI diffs and the merge commands do ignore end-of-line whitespace. ... (check-in: 5ef7435ac0 user: drh tags: trunk) | |
Changes
Changes to src/diff.c.
| ︙ | ︙ | |||
280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 |
m = R[r+nr*3];
if( m>nContext ) m = nContext;
for(j=0; j<m; j++){
appendDiffLine(pOut, " ", &B[b+j]);
}
}
}
/*
** Compare two blocks of text on lines iS1 through iE1-1 of the aFrom[]
** file and lines iS2 through iE2-1 of the aTo[] file. Locate a sequence
** of lines in these two blocks that are exactly the same. Return
** the bounds of the matching sequence.
*/
static void longestCommonSequence(
DContext *p, /* Two files being compared */
int iS1, int iE1, /* Range of lines in p->aFrom[] */
int iS2, int iE2, /* Range of lines in p->aTo[] */
int *piSX, int *piEX, /* Write p->aFrom[] common segment here */
int *piSY, int *piEY /* Write p->aTo[] common segment here */
){
double bestScore = -1e30; /* Best score seen so far */
int i, j; /* Loop counters */
int iSX, iSY, iEX, iEY; /* Current match */
double score; /* Current score */
int skew; /* How lopsided is the match */
int dist; /* Distance of match from center */
int mid; /* Center of the span */
int iSXb, iSYb, iEXb, iEYb; /* Best match so far */
int iSXp, iSYp, iEXp, iEYp; /* Previous match */
iSXb = iSXp = iS1;
iEXb = iEXp = iS1;
iSYb = iSYp = iS2;
iEYb = iEYp = iS2;
mid = (iE1 + iS1)/2;
for(i=iS1; i<iE1; i++){
| > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 |
m = R[r+nr*3];
if( m>nContext ) m = nContext;
for(j=0; j<m; j++){
appendDiffLine(pOut, " ", &B[b+j]);
}
}
}
/*
** Compute the optimal longest common subsequence (LCS) using an
** exhaustive search. This version of the LCS is only used for
** shorter input strings since runtime is O(N*N) where N is the
** input string length.
*/
static void optimalLCS(
DContext *p, /* Two files being compared */
int iS1, int iE1, /* Range of lines in p->aFrom[] */
int iS2, int iE2, /* Range of lines in p->aTo[] */
int *piSX, int *piEX, /* Write p->aFrom[] common segment here */
int *piSY, int *piEY /* Write p->aTo[] common segment here */
){
int mxLength = 0; /* Length of longest common subsequence */
int i, j; /* Loop counters */
int k; /* Length of a candidate subsequence */
int iSXb = iS1; /* Best match so far */
int iSYb = iS2; /* Best match so far */
for(i=iS1; i<iE1-mxLength; i++){
for(j=iS2; j<iE2-mxLength; j++){
if( !same_dline(&p->aFrom[i], &p->aTo[j]) ) continue;
if( mxLength && !same_dline(&p->aFrom[i+mxLength], &p->aTo[j+mxLength]) ){
continue;
}
k = 1;
while( i+k<iE1 && j+k<iE2 && same_dline(&p->aFrom[i+k],&p->aTo[j+k]) ){
k++;
}
if( k>mxLength ){
iSXb = i;
iSYb = j;
mxLength = k;
}
}
}
*piSX = iSXb;
*piEX = iSXb + mxLength;
*piSY = iSYb;
*piEY = iSYb + mxLength;
}
/*
** Compare two blocks of text on lines iS1 through iE1-1 of the aFrom[]
** file and lines iS2 through iE2-1 of the aTo[] file. Locate a sequence
** of lines in these two blocks that are exactly the same. Return
** the bounds of the matching sequence.
**
** If there are two or more possible answers of the same length, the
** returned sequence should be the one closest to the center of the
** input range.
**
** Ideally, the common sequence should be the longest possible common
** sequence. However, an exact computation of LCS is O(N*N) which is
** way too slow for larger files. So this routine uses an O(N)
** heuristic approximation based on hashing that usually works about
** as well. But if the O(N) algorithm doesn't get a good solution
** and N is not too large, we fall back to an exact solution by
** calling optimalLCS().
*/
static void longestCommonSequence(
DContext *p, /* Two files being compared */
int iS1, int iE1, /* Range of lines in p->aFrom[] */
int iS2, int iE2, /* Range of lines in p->aTo[] */
int *piSX, int *piEX, /* Write p->aFrom[] common segment here */
int *piSY, int *piEY /* Write p->aTo[] common segment here */
){
double bestScore = -1e30; /* Best score seen so far */
int i, j; /* Loop counters */
int iSX, iSY, iEX, iEY; /* Current match */
double score; /* Current score */
int skew; /* How lopsided is the match */
int dist; /* Distance of match from center */
int mid; /* Center of the span */
int iSXb, iSYb, iEXb, iEYb; /* Best match so far */
int iSXp, iSYp, iEXp, iEYp; /* Previous match */
iSXb = iSXp = iS1;
iEXb = iEXp = iS1;
iSYb = iSYp = iS2;
iEYb = iEYp = iS2;
mid = (iE1 + iS1)/2;
for(i=iS1; i<iE1; i++){
|
| ︙ | ︙ | |||
355 356 357 358 359 360 361 |
}else{
iSXp = iSX;
iSYp = iSY;
iEXp = iEX;
iEYp = iEY;
}
}
| > > > > > | | | | > | 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 |
}else{
iSXp = iSX;
iSYp = iSY;
iEXp = iEX;
iEYp = iEY;
}
}
if( iSXb==iEXb && (iE1-iS1)*(iE2-iS2)<400 ){
/* If no common sequence is found using the hashing heuristic and
** the input is not too big, use the expensive exact solution */
optimalLCS(p, iS1, iE1, iS2, iE2, piSX, piEX, piSY, piEY);
}else{
*piSX = iSXb;
*piSY = iSYb;
*piEX = iEXb;
*piEY = iEYb;
}
/* printf("LCS(%d..%d/%d..%d) = %d..%d/%d..%d\n",
iS1, iE1, iS2, iE2, *piSX, *piEX, *piSY, *piEY); */
}
/*
** Do a single step in the difference. Compute a sequence of
** copy/delete/insert steps that will convert lines iS1 through iE1-1 of
|
| ︙ | ︙ |
Changes to src/diffcmd.c.
| ︙ | ︙ | |||
403 404 405 406 407 408 409 |
isInternDiff = find_option("internal","i",0)!=0;
zFrom = find_option("from", "r", 1);
zTo = find_option("to", 0, 1);
if( zTo==0 ){
db_must_be_within_tree();
verify_all_options();
| | | 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 |
isInternDiff = find_option("internal","i",0)!=0;
zFrom = find_option("from", "r", 1);
zTo = find_option("to", 0, 1);
if( zTo==0 ){
db_must_be_within_tree();
verify_all_options();
if( !isInternDiff ){
zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
}
if( g.argc==3 ){
diff_one_against_disk(zFrom, zDiffCmd, 0);
}else{
diff_all_against_disk(zFrom, zDiffCmd, 0);
}
|
| ︙ | ︙ |