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_diff_parents(struct sline *sline, unsigned long all_mask)
 275 {
 276         /*
 277          * Look at the line and see from which parents we have difference.
 278          * Lower bits of sline->flag records if the parent had this line,
 279          * so XOR with all_mask gives us on-bits for parents we have
 280          * differences with.
 281          */
 282         unsigned long parents = (sline->flag ^ all_mask);
 283         if (sline->lost_head) {
 284                 struct lline *ll;
 285                 for (ll = sline->lost_head; ll; ll = ll->next)
 286                         parents |= ll->parent_map;
 287         }
 288         return parents & all_mask;
 289 }
 290
 291 static void make_hunks(struct sline *sline, unsigned long cnt,
 292                        int num_parent, int dense)
 293 {
 294         unsigned long all_mask = (1UL<<num_parent) - 1;
 295         unsigned long mark = (1UL<<num_parent);
 296         unsigned long i;
 297
 298         i = 0;
 299         while (i < cnt) {
 300                 if (interesting(&sline[i], all_mask)) {
 301                         unsigned long j = (context < i) ? i - context : 0;
 302                         while (j <= i)
 303                                 sline[j++].flag |= mark;
 304                         while (++i < cnt) {
 305                                 if (!interesting(&sline[i], all_mask))
 306                                         break;
 307                                 sline[i].flag |= mark;
 308                         }
 309                         j = (i + context < cnt) ? i + context : cnt;
 310                         while (i < j)
 311                                 sline[i++].flag |= mark;
 312                         continue;
 313                 }
 314                 i++;
 315         }
 316         if (!dense)
 317                 return;
 318
 319         /* Look at each hunk, and if it contains changes from only
 320          * one parent, mark that uninteresting.
 321          */
 322         i = 0;
 323         while (i < cnt) {
 324                 int j, hunk_end, diffs;
 325                 unsigned long parents;
 326                 while (i < cnt && !(sline[i].flag & mark))
 327                         i++;
 328                 if (cnt <= i)
 329                         break; /* No more interesting hunks */
 330                 for (hunk_end = i + 1; hunk_end < cnt; hunk_end++)
 331                         if (!(sline[hunk_end].flag & mark))
 332                                 break;
 333                 /* [i..hunk_end) are interesting.  Now is it from
 334                  * only one parent?
 335                  * If lost lines are only from one parent and
 336                  * remaining lines existed in parents other than
 337                  * that parent, then the hunk is not that interesting.
 338                  */
 339                 parents = 0;
 340                 diffs = 0;
 341                 for (j = i; j < hunk_end; j++)
 342                         parents |= line_diff_parents(sline + j, all_mask);
 343                 /* Now, how many bits from [0..num_parent) are on? */
 344                 for (j = 0; j < num_parent; j++) {
 345                         if (parents & (1UL<<j))
 346                                 diffs++;
 347                 }
 348                 if (diffs < 2) {
 349                         /* This hunk is not that interesting after all */
 350                         for (j = i; j < hunk_end; j++)
 351                                 sline[j].flag &= ~mark;
 352                 }
 353                 i = hunk_end;
 354         }
 355 }
 356
 357 static void dump_sline(struct sline *sline, int cnt, int num_parent)
 358 {
 359         unsigned long mark = (1UL<<num_parent);
 360         int i;
 361         int lno = 0;
 362
 363         while (1) {
 364                 struct sline *sl = &sline[lno];
 365                 int hunk_end;
 366                 while (lno < cnt && !(sline[lno].flag & mark))
 367                         lno++;
 368                 if (cnt <= lno)
 369                         break;
 370                 for (hunk_end = lno + 1; hunk_end < cnt; hunk_end++)
 371                         if (!(sline[hunk_end].flag & mark))
 372                                 break;
 373                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 374                 printf(" +%d,%d ", lno+1, hunk_end-lno);
 375                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 376                 putchar('\n');
 377                 while (lno < hunk_end) {
 378                         struct lline *ll;
 379                         int j;
 380                         sl = &sline[lno++];
 381                         ll = sl->lost_head;
 382                         while (ll) {
 383                                 for (j = 0; j < num_parent; j++) {
 384                                         if (ll->parent_map & (1UL<<j))
 385                                                 putchar('-');
 386                                         else
 387                                                 putchar(' ');
 388                                 }
 389                                 puts(ll->line);
 390                                 ll = ll->next;
 391                         }
 392                         for (j = 0; j < num_parent; j++) {
 393                                 if ((1UL<<j) & sl->flag)
 394                                         putchar(' ');
 395                                 else
 396                                         putchar('+');
 397                         }
 398                         printf("%.*s\n", sl->len, sl->bol);
 399                 }
 400         }
 401 }
 402
 403 static void show_combined_diff(struct path_list *elem, int num_parent,
 404                                int dense)
 405 {
 406         unsigned long size, cnt, lno;
 407         char *result, *cp, *ep;
 408         struct sline *sline; /* survived lines */
 409         int i;
 410         char ourtmp[TMPPATHLEN];
 411
 412         /* Read the result of merge first */
 413         result = grab_blob(elem->sha1, &size);
 414         write_to_temp_file(ourtmp, result, size);
 415
 416         for (cnt = 0, cp = result; cp - result < size; cp++) {
 417                 if (*cp == '\n')
 418                         cnt++;
 419         }
 420         if (result[size-1] != '\n')
 421                 cnt++; /* incomplete line */
 422
 423         sline = xcalloc(cnt, sizeof(*sline));
 424         ep = result;
 425         sline[0].bol = result;
 426         for (lno = 0, cp = result; cp - result < size; cp++) {
 427                 if (*cp == '\n') {
 428                         sline[lno].lost_tail = &sline[lno].lost_head;
 429                         sline[lno].len = cp - sline[lno].bol;
 430                         sline[lno].flag = (1UL<<num_parent) - 1;
 431                         lno++;
 432                         if (lno < cnt)
 433                                 sline[lno].bol = cp + 1;
 434                 }
 435         }
 436         if (result[size-1] != '\n') {
 437                 sline[cnt-1].lost_tail = &sline[cnt-1].lost_head;
 438                 sline[cnt-1].len = size - (sline[cnt-1].bol - result);
 439                 sline[cnt-1].flag = (1UL<<num_parent) - 1;
 440         }
 441
 442         for (i = 0; i < num_parent; i++)
 443                 combine_diff(elem->parent_sha1[i], ourtmp, sline, cnt, i);
 444
 445         make_hunks(sline, cnt, num_parent, dense);
 446
 447         dump_sline(sline, cnt, num_parent);
 448         unlink(ourtmp);
 449         free(result);
 450
 451         for (i = 0; i < cnt; i++) {
 452                 if (sline[i].lost_head) {
 453                         struct lline *ll = sline[i].lost_head;
 454                         while (ll) {
 455                                 struct lline *tmp = ll;
 456                                 ll = ll->next;
 457                                 free(tmp);
 458                         }
 459                 }
 460         }
 461         free(sline);
 462 }
 463
 464 int diff_tree_combined_merge(const unsigned char *sha1,
 465                              const char *header,
 466                              int show_empty_merge, int dense)
 467 {
 468         struct commit *commit = lookup_commit(sha1);
 469         struct diff_options diffopts;
 470         struct commit_list *parents;
 471         struct path_list *p, *paths = NULL;
 472         int num_parent, i, num_paths;
 473
 474         diff_setup(&diffopts);
 475         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
 476         diffopts.recursive = 1;
 477
 478         /* count parents */
 479         for (parents = commit->parents, num_parent = 0;
 480              parents;
 481              parents = parents->next, num_parent++)
 482                 ; /* nothing */
 483
 484         /* find set of paths that everybody touches */
 485         for (parents = commit->parents, i = 0;
 486              parents;
 487              parents = parents->next, i++) {
 488                 struct commit *parent = parents->item;
 489                 diff_tree_sha1(parent->object.sha1, commit->object.sha1, "",
 490                                &diffopts);
 491                 paths = intersect_paths(paths, i, num_parent);
 492                 diff_flush(&diffopts);
 493         }
 494
 495         /* find out surviving paths */
 496         for (num_paths = 0, p = paths; p; p = p->next) {
 497                 if (p->len)
 498                         num_paths++;
 499         }
 500         if (num_paths || show_empty_merge) {
 501                 puts(header);
 502                 for (p = paths; p; p = p->next) {
 503                         if (!p->len)
 504                                 continue;
 505                         printf("diff --%s ", dense ? "cc" : "combined");
 506                         if (quote_c_style(p->path, NULL, NULL, 0))
 507                                 quote_c_style(p->path, NULL, stdout, 0);
 508                         else
 509                                 printf("%s", p->path);
 510                         putchar('\n');
 511                         show_combined_diff(p, num_parent, dense);
 512                 }
 513         }
 514
 515         /* Clean things up */
 516         while (paths) {
 517                 struct path_list *tmp = paths;
 518                 paths = paths->next;
 519                 free(tmp);
 520         }
 521         return 0;
 522 }