combine-diff.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "diff.h"
   4 #include "diffcore.h"
   5 #include "quote.h"
   6
   7 struct path_list {
   8         struct path_list *next;
   9         int len;
  10         char *path;
  11         unsigned char sha1[20];
  12         unsigned char parent_sha1[FLEX_ARRAY][20];
  13 };
  14
  15 static int uninteresting(struct diff_filepair *p)
  16 {
  17         if (diff_unmodified_pair(p))
  18                 return 1;
  19         if (!S_ISREG(p->one->mode) || !S_ISREG(p->two->mode))
  20                 return 1;
  21         return 0;
  22 }
  23
  24 static struct path_list *intersect_paths(struct path_list *curr,
  25                                          int n, int num_parent)
  26 {
  27         struct diff_queue_struct *q = &diff_queued_diff;
  28         struct path_list *p;
  29         int i;
  30
  31         if (!n) {
  32                 struct path_list *list = NULL, **tail = &list;
  33                 for (i = 0; i < q->nr; i++) {
  34                         int len;
  35                         const char *path;
  36                         if (uninteresting(q->queue[i]))
  37                                 continue;
  38                         path = q->queue[i]->two->path;
  39                         len = strlen(path);
  40
  41                         p = xmalloc(sizeof(*p) + len + 1 + num_parent * 20);
  42                         p->path = (char*) &(p->parent_sha1[num_parent][0]);
  43                         memcpy(p->path, path, len);
  44                         p->path[len] = 0;
  45                         p->len = len;
  46                         p->next = NULL;
  47                         memcpy(p->sha1, q->queue[i]->two->sha1, 20);
  48                         memcpy(p->parent_sha1[n], q->queue[i]->one->sha1, 20);
  49                         *tail = p;
  50                         tail = &p->next;
  51                 }
  52                 return list;
  53         }
  54
  55         for (p = curr; p; p = p->next) {
  56                 int found = 0;
  57                 if (!p->len)
  58                         continue;
  59                 for (i = 0; i < q->nr; i++) {
  60                         const char *path;
  61                         int len;
  62
  63                         if (uninteresting(q->queue[i]))
  64                                 continue;
  65                         path = q->queue[i]->two->path;
  66                         len = strlen(path);
  67                         if (len == p->len && !memcmp(path, p->path, len)) {
  68                                 found = 1;
  69                                 memcpy(p->parent_sha1[n],
  70                                        q->queue[i]->one->sha1, 20);
  71                                 break;
  72                         }
  73                 }
  74                 if (!found)
  75                         p->len = 0;
  76         }
  77         return curr;
  78 }
  79
  80 struct lline {
  81         struct lline *next;
  82         int len;
  83         unsigned long parent_map;
  84         char line[FLEX_ARRAY];
  85 };
  86
  87 struct sline {
  88         struct lline *lost_head, **lost_tail;
  89         char *bol;
  90         int len;
  91         unsigned long flag;
  92 };
  93
  94 static char *grab_blob(const unsigned char *sha1, unsigned long *size)
  95 {
  96         char *blob;
  97         char type[20];
  98         if (!memcmp(sha1, null_sha1, 20)) {
  99                 /* deleted blob */
 100                 *size = 0;
 101                 return xcalloc(1, 1);
 102         }
 103         blob = read_sha1_file(sha1, type, size);
 104         if (strcmp(type, "blob"))
 105                 die("object '%s' is not a blob!", sha1_to_hex(sha1));
 106         return blob;
 107 }
 108
 109 #define TMPPATHLEN 50
 110 #define MAXLINELEN 10240
 111
 112 static void write_to_temp_file(char *tmpfile, void *blob, unsigned long size)
 113 {
 114         int fd = git_mkstemp(tmpfile, TMPPATHLEN, ".diff_XXXXXX");
 115         if (fd < 0)
 116                 die("unable to create temp-file");
 117         if (write(fd, blob, size) != size)
 118                 die("unable to write temp-file");
 119         close(fd);
 120 }
 121
 122 static void write_temp_blob(char *tmpfile, const unsigned char *sha1)
 123 {
 124         unsigned long size;
 125         void *blob;
 126         blob = grab_blob(sha1, &size);
 127         write_to_temp_file(tmpfile, blob, size);
 128         free(blob);
 129 }
 130
 131 static int parse_num(char **cp_p, unsigned int *num_p)
 132 {
 133         char *cp = *cp_p;
 134         unsigned int num = 0;
 135         int read_some;
 136
 137         while ('0' <= *cp && *cp <= '9')
 138                 num = num * 10 + *cp++ - '0';
 139         if (!(read_some = cp - *cp_p))
 140                 return -1;
 141         *cp_p = cp;
 142         *num_p = num;
 143         return 0;
 144 }
 145
 146 static int parse_hunk_header(char *line, int len,
 147                              unsigned int *ob, unsigned int *on,
 148                              unsigned int *nb, unsigned int *nn)
 149 {
 150         char *cp;
 151         cp = line + 4;
 152         if (parse_num(&cp, ob)) {
 153         bad_line:
 154                 return error("malformed diff output: %s", line);
 155         }
 156         if (*cp == ',') {
 157                 cp++;
 158                 if (parse_num(&cp, on))
 159                         goto bad_line;
 160         }
 161         else
 162                 *on = 1;
 163         if (*cp++ != ' ' || *cp++ != '+')
 164                 goto bad_line;
 165         if (parse_num(&cp, nb))
 166                 goto bad_line;
 167         if (*cp == ',') {
 168                 cp++;
 169                 if (parse_num(&cp, nn))
 170                         goto bad_line;
 171         }
 172         else
 173                 *nn = 1;
 174         return -!!memcmp(cp, " @@", 3);
 175 }
 176
 177 static void append_lost(struct sline *sline, int n, const char *line)
 178 {
 179         struct lline *lline;
 180         int len = strlen(line);
 181         unsigned long this_mask = (1UL<<n);
 182         if (line[len-1] == '\n')
 183                 len--;
 184
 185         /* Check to see if we can squash things */
 186         if (sline->lost_head) {
 187                 struct lline *last_one = NULL;
 188                 /* We cannot squash it with earlier one */
 189                 for (lline = sline->lost_head;
 190                      lline;
 191                      lline = lline->next)
 192                         if (lline->parent_map & this_mask)
 193                                 last_one = lline;
 194                 lline = last_one ? last_one->next : sline->lost_head;
 195                 while (lline) {
 196                         if (lline->len == len &&
 197                             !memcmp(lline->line, line, len)) {
 198                                 lline->parent_map |= this_mask;
 199                                 return;
 200                         }
 201                         lline = lline->next;
 202                 }
 203         }
 204
 205         lline = xmalloc(sizeof(*lline) + len + 1);
 206         lline->len = len;
 207         lline->next = NULL;
 208         lline->parent_map = this_mask;
 209         memcpy(lline->line, line, len);
 210         lline->line[len] = 0;
 211         *sline->lost_tail = lline;
 212         sline->lost_tail = &lline->next;
 213 }
 214
 215 static void combine_diff(const unsigned char *parent, const char *ourtmp,
 216                          struct sline *sline, int cnt, int n)
 217 {
 218         FILE *in;
 219         char parent_tmp[TMPPATHLEN];
 220         char cmd[TMPPATHLEN * 2 + 1024];
 221         char line[MAXLINELEN];
 222         unsigned int lno, ob, on, nb, nn;
 223         unsigned long pmask = ~(1UL << n);
 224         struct sline *lost_bucket = NULL;
 225
 226         write_temp_blob(parent_tmp, parent);
 227         sprintf(cmd, "diff --unified=0 -La/x -Lb/x '%s' '%s'",
 228                 parent_tmp, ourtmp);
 229         in = popen(cmd, "r");
 230         if (!in)
 231                 return;
 232
 233         lno = 1;
 234         while (fgets(line, sizeof(line), in) != NULL) {
 235                 int len = strlen(line);
 236                 if (5 < len && !memcmp("@@ -", line, 4)) {
 237                         if (parse_hunk_header(line, len,
 238                                               &ob, &on, &nb, &nn))
 239                                 break;
 240                         lno = nb;
 241                         if (!nb) {
 242                                 /* @@ -1,2 +0,0 @@ to remove the
 243                                  * first two lines...
 244                                  */
 245                                 nb = 1;
 246                         }
 247                         lost_bucket = &sline[nb-1]; /* sline is 0 based */
 248                         continue;
 249                 }
 250                 if (!lost_bucket)
 251                         continue;
 252                 switch (line[0]) {
 253                 case '-':
 254                         append_lost(lost_bucket, n, line+1);
 255                         break;
 256                 case '+':
 257                         sline[lno-1].flag &= pmask;
 258                         lno++;
 259                         break;
 260                 }
 261         }
 262         fclose(in);
 263         unlink(parent_tmp);
 264 }
 265
 266 static unsigned long context = 3;
 267 static char combine_marker = '@';
 268
 269 static int interesting(struct sline *sline, unsigned long all_mask)
 270 {
 271         return ((sline->flag & all_mask) != all_mask || sline->lost_head);
 272 }
 273
 274 static unsigned long line_common_diff(struct sline *sline, unsigned long all_mask)
 275 {
 276         /*
 277          * Look at the line and see from which parents we have the
 278          * same difference.
 279          */
 280
 281         /* Lower bits of sline->flag records if the parent had this
 282          * line, so XOR with all_mask gives us on-bits for parents we
 283          * have differences with.
 284          */
 285         unsigned long common_adds = (sline->flag ^ all_mask) & all_mask;
 286         unsigned long common_removes = all_mask;
 287
 288         /* If all the parents have this line, that also counts as
 289          * having the same difference.
 290          */
 291         if (!common_adds)
 292                 common_adds = all_mask;
 293
 294         if (sline->lost_head) {
 295                 /* Lost head list records the lines removed from
 296                  * the parents, and parent_map records from which
 297                  * parent the line was removed.
 298                  */
 299                 struct lline *ll;
 300                 for (ll = sline->lost_head; ll; ll = ll->next) {
 301                         common_removes &= ll->parent_map;
 302                 }
 303         }
 304         return common_adds & common_removes;
 305 }
 306
 307 static unsigned long line_all_diff(struct sline *sline, unsigned long all_mask)
 308 {
 309         /*
 310          * Look at the line and see from which parents we have some difference.
 311          */
 312         unsigned long different = (sline->flag ^ all_mask) & all_mask;
 313         if (sline->lost_head) {
 314                 /* Lost head list records the lines removed from
 315                  * the parents, and parent_map records from which
 316                  * parent the line was removed.
 317                  */
 318                 struct lline *ll;
 319                 for (ll = sline->lost_head; ll; ll = ll->next) {
 320                         different |= ll->parent_map;
 321                 }
 322         }
 323         return different;
 324 }
 325
 326 static int make_hunks(struct sline *sline, unsigned long cnt,
 327                        int num_parent, int dense)
 328 {
 329         unsigned long all_mask = (1UL<<num_parent) - 1;
 330         unsigned long mark = (1UL<<num_parent);
 331         unsigned long i;
 332         int has_interesting = 0;
 333
 334         i = 0;
 335         while (i < cnt) {
 336                 if (interesting(&sline[i], all_mask)) {
 337                         unsigned long j = (context < i) ? i - context : 0;
 338                         while (j <= i)
 339                                 sline[j++].flag |= mark;
 340                         while (++i < cnt) {
 341                                 if (!interesting(&sline[i], all_mask))
 342                                         break;
 343                                 sline[i].flag |= mark;
 344                         }
 345                         j = (i + context < cnt) ? i + context : cnt;
 346                         while (i < j)
 347                                 sline[i++].flag |= mark;
 348                         has_interesting = 1;
 349                         continue;
 350                 }
 351                 i++;
 352         }
 353         if (!dense)
 354                 return has_interesting;
 355
 356         /* Look at each hunk, and if we have changes from only one
 357          * parent, or the changes are the same from all but one
 358          * parent, mark that uninteresting.
 359          */
 360         has_interesting = 0;
 361         i = 0;
 362         while (i < cnt) {
 363                 int j, hunk_end, same, diff;
 364                 unsigned long same_diff, all_diff;
 365                 while (i < cnt && !(sline[i].flag & mark))
 366                         i++;
 367                 if (cnt <= i)
 368                         break; /* No more interesting hunks */
 369                 for (hunk_end = i + 1; hunk_end < cnt; hunk_end++)
 370                         if (!(sline[hunk_end].flag & mark))
 371                                 break;
 372                 /* [i..hunk_end) are interesting.  Now does it have
 373                  * the same change with all but one parent?
 374                  */
 375                 same_diff = all_mask;
 376                 all_diff = 0;
 377                 for (j = i; j < hunk_end; j++) {
 378                         same_diff &= line_common_diff(sline + j, all_mask);
 379                         all_diff |= line_all_diff(sline + j, all_mask);
 380                 }
 381                 diff = same = 0;
 382                 for (j = 0; j < num_parent; j++) {
 383                         if (same_diff & (1UL<<j))
 384                                 same++;
 385                         if (all_diff & (1UL<<j))
 386                                 diff++;
 387                 }
 388                 if ((num_parent - 1 <= same) || (diff == 1)) {
 389                         /* This hunk is not that interesting after all */
 390                         for (j = i; j < hunk_end; j++)
 391                                 sline[j].flag &= ~mark;
 392                 }
 393                 else
 394                         has_interesting = 1;
 395                 i = hunk_end;
 396         }
 397         return has_interesting;
 398 }
 399
 400 static void dump_sline(struct sline *sline, int cnt, int num_parent)
 401 {
 402         unsigned long mark = (1UL<<num_parent);
 403         int i;
 404         int lno = 0;
 405
 406         while (1) {
 407                 struct sline *sl = &sline[lno];
 408                 int hunk_end;
 409                 while (lno < cnt && !(sline[lno].flag & mark))
 410                         lno++;
 411                 if (cnt <= lno)
 412                         break;
 413                 for (hunk_end = lno + 1; hunk_end < cnt; hunk_end++)
 414                         if (!(sline[hunk_end].flag & mark))
 415                                 break;
 416                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 417                 printf(" +%d,%d ", lno+1, hunk_end-lno);
 418                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 419                 putchar('\n');
 420                 while (lno < hunk_end) {
 421                         struct lline *ll;
 422                         int j;
 423                         sl = &sline[lno++];
 424                         ll = sl->lost_head;
 425                         while (ll) {
 426                                 for (j = 0; j < num_parent; j++) {
 427                                         if (ll->parent_map & (1UL<<j))
 428                                                 putchar('-');
 429                                         else
 430                                                 putchar(' ');
 431                                 }
 432                                 puts(ll->line);
 433                                 ll = ll->next;
 434                         }
 435                         for (j = 0; j < num_parent; j++) {
 436                                 if ((1UL<<j) & sl->flag)
 437                                         putchar(' ');
 438                                 else
 439                                         putchar('+');
 440                         }
 441                         printf("%.*s\n", sl->len, sl->bol);
 442                 }
 443         }
 444 }
 445
 446 static int show_combined_diff(struct path_list *elem, int num_parent,
 447                               int dense, const char *header, int show_empty)
 448 {
 449         unsigned long size, cnt, lno;
 450         char *result, *cp, *ep;
 451         struct sline *sline; /* survived lines */
 452         int i, show_hunks, shown_header = 0;
 453         char ourtmp[TMPPATHLEN];
 454
 455         /* Read the result of merge first */
 456         result = grab_blob(elem->sha1, &size);
 457         write_to_temp_file(ourtmp, result, size);
 458
 459         for (cnt = 0, cp = result; cp - result < size; cp++) {
 460                 if (*cp == '\n')
 461                         cnt++;
 462         }
 463         if (result[size-1] != '\n')
 464                 cnt++; /* incomplete line */
 465
 466         sline = xcalloc(cnt, sizeof(*sline));
 467         ep = result;
 468         sline[0].bol = result;
 469         for (lno = 0, cp = result; cp - result < size; cp++) {
 470                 if (*cp == '\n') {
 471                         sline[lno].lost_tail = &sline[lno].lost_head;
 472                         sline[lno].len = cp - sline[lno].bol;
 473                         sline[lno].flag = (1UL<<num_parent) - 1;
 474                         lno++;
 475                         if (lno < cnt)
 476                                 sline[lno].bol = cp + 1;
 477                 }
 478         }
 479         if (result[size-1] != '\n') {
 480                 sline[cnt-1].lost_tail = &sline[cnt-1].lost_head;
 481                 sline[cnt-1].len = size - (sline[cnt-1].bol - result);
 482                 sline[cnt-1].flag = (1UL<<num_parent) - 1;
 483         }
 484
 485         for (i = 0; i < num_parent; i++)
 486                 combine_diff(elem->parent_sha1[i], ourtmp, sline, cnt, i);
 487
 488         show_hunks = make_hunks(sline, cnt, num_parent, dense);
 489
 490         if (header && (show_hunks || show_empty)) {
 491                 shown_header++;
 492                 puts(header);
 493         }
 494         if (show_hunks) {
 495                 printf("diff --%s ", dense ? "cc" : "combined");
 496                 if (quote_c_style(elem->path, NULL, NULL, 0))
 497                         quote_c_style(elem->path, NULL, stdout, 0);
 498                 else
 499                         printf("%s", elem->path);
 500                 putchar('\n');
 501                 dump_sline(sline, cnt, num_parent);
 502         }
 503         unlink(ourtmp);
 504         free(result);
 505
 506         for (i = 0; i < cnt; i++) {
 507                 if (sline[i].lost_head) {
 508                         struct lline *ll = sline[i].lost_head;
 509                         while (ll) {
 510                                 struct lline *tmp = ll;
 511                                 ll = ll->next;
 512                                 free(tmp);
 513                         }
 514                 }
 515         }
 516         free(sline);
 517         return shown_header;
 518 }
 519
 520 int diff_tree_combined_merge(const unsigned char *sha1,
 521                              const char *header,
 522                              int show_empty_merge, int dense)
 523 {
 524         struct commit *commit = lookup_commit(sha1);
 525         struct diff_options diffopts;
 526         struct commit_list *parents;
 527         struct path_list *p, *paths = NULL;
 528         int num_parent, i, num_paths;
 529
 530         diff_setup(&diffopts);
 531         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
 532         diffopts.recursive = 1;
 533
 534         /* count parents */
 535         for (parents = commit->parents, num_parent = 0;
 536              parents;
 537              parents = parents->next, num_parent++)
 538                 ; /* nothing */
 539
 540         /* find set of paths that everybody touches */
 541         for (parents = commit->parents, i = 0;
 542              parents;
 543              parents = parents->next, i++) {
 544                 struct commit *parent = parents->item;
 545                 diff_tree_sha1(parent->object.sha1, commit->object.sha1, "",
 546                                &diffopts);
 547                 paths = intersect_paths(paths, i, num_parent);
 548                 diff_flush(&diffopts);
 549         }
 550
 551         /* find out surviving paths */
 552         for (num_paths = 0, p = paths; p; p = p->next) {
 553                 if (p->len)
 554                         num_paths++;
 555         }
 556         if (num_paths || show_empty_merge) {
 557                 for (p = paths; p; p = p->next) {
 558                         if (!p->len)
 559                                 continue;
 560                         if (show_combined_diff(p, num_parent, dense, header,
 561                                                show_empty_merge))
 562                                 header = NULL;
 563                 }
 564         }
 565
 566         /* Clean things up */
 567         while (paths) {
 568                 struct path_list *tmp = paths;
 569                 paths = paths->next;
 570                 free(tmp);
 571         }
 572         return 0;
 573 }