tree-diff.c

   1 /*
   2  * Helper functions for tree diff generation
   3  */
   4 #include "cache.h"
   5 #include "diff.h"
   6 #include "diffcore.h"
   7 #include "tree.h"
   8
   9 static char *malloc_base(const char *base, int baselen, const char *path, int pathlen)
  10 {
  11         char *newbase = xmalloc(baselen + pathlen + 2);
  12         memcpy(newbase, base, baselen);
  13         memcpy(newbase + baselen, path, pathlen);
  14         memcpy(newbase + baselen + pathlen, "/", 2);
  15         return newbase;
  16 }
  17
  18 static char *malloc_fullname(const char *base, int baselen, const char *path, int pathlen)
  19 {
  20         char *fullname = xmalloc(baselen + pathlen + 1);
  21         memcpy(fullname, base, baselen);
  22         memcpy(fullname + baselen, path, pathlen);
  23         fullname[baselen + pathlen] = 0;
  24         return fullname;
  25 }
  26
  27 static void show_entry(struct diff_options *opt, const char *prefix, struct tree_desc *desc,
  28                        const char *base, int baselen);
  29
  30 static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const char *base, int baselen, struct diff_options *opt)
  31 {
  32         unsigned mode1, mode2;
  33         const char *path1, *path2;
  34         const unsigned char *sha1, *sha2;
  35         int cmp, pathlen1, pathlen2;
  36         char *fullname;
  37
  38         sha1 = tree_entry_extract(t1, &path1, &mode1);
  39         sha2 = tree_entry_extract(t2, &path2, &mode2);
  40
  41         pathlen1 = tree_entry_len(path1, sha1);
  42         pathlen2 = tree_entry_len(path2, sha2);
  43         cmp = base_name_compare(path1, pathlen1, mode1, path2, pathlen2, mode2);
  44         if (cmp < 0) {
  45                 show_entry(opt, "-", t1, base, baselen);
  46                 return -1;
  47         }
  48         if (cmp > 0) {
  49                 show_entry(opt, "+", t2, base, baselen);
  50                 return 1;
  51         }
  52         if (!DIFF_OPT_TST(opt, FIND_COPIES_HARDER) && !hashcmp(sha1, sha2) && mode1 == mode2)
  53                 return 0;
  54
  55         /*
  56          * If the filemode has changed to/from a directory from/to a regular
  57          * file, we need to consider it a remove and an add.
  58          */
  59         if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
  60                 show_entry(opt, "-", t1, base, baselen);
  61                 show_entry(opt, "+", t2, base, baselen);
  62                 return 0;
  63         }
  64
  65         if (DIFF_OPT_TST(opt, RECURSIVE) && S_ISDIR(mode1)) {
  66                 int retval;
  67                 char *newbase = malloc_base(base, baselen, path1, pathlen1);
  68                 if (DIFF_OPT_TST(opt, TREE_IN_RECURSIVE)) {
  69                         newbase[baselen + pathlen1] = 0;
  70                         opt->change(opt, mode1, mode2,
  71                                     sha1, sha2, newbase, 0, 0);
  72                         newbase[baselen + pathlen1] = '/';
  73                 }
  74                 retval = diff_tree_sha1(sha1, sha2, newbase, opt);
  75                 free(newbase);
  76                 return retval;
  77         }
  78
  79         fullname = malloc_fullname(base, baselen, path1, pathlen1);
  80         opt->change(opt, mode1, mode2, sha1, sha2, fullname, 0, 0);
  81         free(fullname);
  82         return 0;
  83 }
  84
  85 /*
  86  * Is a tree entry interesting given the pathspec we have?
  87  *
  88  * Pre-condition: baselen == 0 || base[baselen-1] == '/'
  89  *
  90  * Return:
  91  *  - 2 for "yes, and all subsequent entries will be"
  92  *  - 1 for yes
  93  *  - zero for no
  94  *  - negative for "no, and no subsequent entries will be either"
  95  */
  96 static int tree_entry_interesting(struct tree_desc *desc, const char *base, int baselen, struct diff_options *opt)
  97 {
  98         const char *path;
  99         const unsigned char *sha1;
 100         unsigned mode;
 101         int i;
 102         int pathlen;
 103         int never_interesting = -1;
 104
 105         if (!opt->nr_paths)
 106                 return 2;
 107
 108         sha1 = tree_entry_extract(desc, &path, &mode);
 109
 110         pathlen = tree_entry_len(path, sha1);
 111
 112         for (i = 0; i < opt->nr_paths; i++) {
 113                 const char *match = opt->paths[i];
 114                 int matchlen = opt->pathlens[i];
 115                 int m = -1; /* signals that we haven't called strncmp() */
 116
 117                 if (baselen >= matchlen) {
 118                         /* If it doesn't match, move along... */
 119                         if (strncmp(base, match, matchlen))
 120                                 continue;
 121
 122                         /*
 123                          * If the base is a subdirectory of a path which
 124                          * was specified, all of them are interesting.
 125                          */
 126                         if (!matchlen ||
 127                             base[matchlen] == '/' ||
 128                             match[matchlen - 1] == '/')
 129                                 return 2;
 130
 131                         /* Just a random prefix match */
 132                         continue;
 133                 }
 134
 135                 /* Does the base match? */
 136                 if (strncmp(base, match, baselen))
 137                         continue;
 138
 139                 match += baselen;
 140                 matchlen -= baselen;
 141
 142                 if (never_interesting) {
 143                         /*
 144                          * We have not seen any match that sorts later
 145                          * than the current path.
 146                          */
 147
 148                         /*
 149                          * Does match sort strictly earlier than path
 150                          * with their common parts?
 151                          */
 152                         m = strncmp(match, path,
 153                                     (matchlen < pathlen) ? matchlen : pathlen);
 154                         if (m < 0)
 155                                 continue;
 156
 157                         /*
 158                          * If we come here even once, that means there is at
 159                          * least one pathspec that would sort equal to or
 160                          * later than the path we are currently looking at.
 161                          * In other words, if we have never reached this point
 162                          * after iterating all pathspecs, it means all
 163                          * pathspecs are either outside of base, or inside the
 164                          * base but sorts strictly earlier than the current
 165                          * one.  In either case, they will never match the
 166                          * subsequent entries.  In such a case, we initialized
 167                          * the variable to -1 and that is what will be
 168                          * returned, allowing the caller to terminate early.
 169                          */
 170                         never_interesting = 0;
 171                 }
 172
 173                 if (pathlen > matchlen)
 174                         continue;
 175
 176                 if (matchlen > pathlen) {
 177                         if (match[pathlen] != '/')
 178                                 continue;
 179                         if (!S_ISDIR(mode))
 180                                 continue;
 181                 }
 182
 183                 if (m == -1)
 184                         /*
 185                          * we cheated and did not do strncmp(), so we do
 186                          * that here.
 187                          */
 188                         m = strncmp(match, path, pathlen);
 189
 190                 /*
 191                  * If common part matched earlier then it is a hit,
 192                  * because we rejected the case where path is not a
 193                  * leading directory and is shorter than match.
 194                  */
 195                 if (!m)
 196                         return 1;
 197         }
 198         return never_interesting; /* No matches */
 199 }
 200
 201 /* A whole sub-tree went away or appeared */
 202 static void show_tree(struct diff_options *opt, const char *prefix, struct tree_desc *desc, const char *base, int baselen)
 203 {
 204         int all_interesting = 0;
 205         while (desc->size) {
 206                 int show;
 207
 208                 if (all_interesting)
 209                         show = 1;
 210                 else {
 211                         show = tree_entry_interesting(desc, base, baselen,
 212                                                       opt);
 213                         if (show == 2)
 214                                 all_interesting = 1;
 215                 }
 216                 if (show < 0)
 217                         break;
 218                 if (show)
 219                         show_entry(opt, prefix, desc, base, baselen);
 220                 update_tree_entry(desc);
 221         }
 222 }
 223
 224 /* A file entry went away or appeared */
 225 static void show_entry(struct diff_options *opt, const char *prefix, struct tree_desc *desc,
 226                        const char *base, int baselen)
 227 {
 228         unsigned mode;
 229         const char *path;
 230         const unsigned char *sha1 = tree_entry_extract(desc, &path, &mode);
 231         int pathlen = tree_entry_len(path, sha1);
 232
 233         if (DIFF_OPT_TST(opt, RECURSIVE) && S_ISDIR(mode)) {
 234                 enum object_type type;
 235                 char *newbase = malloc_base(base, baselen, path, pathlen);
 236                 struct tree_desc inner;
 237                 void *tree;
 238                 unsigned long size;
 239
 240                 tree = read_sha1_file(sha1, &type, &size);
 241                 if (!tree || type != OBJ_TREE)
 242                         die("corrupt tree sha %s", sha1_to_hex(sha1));
 243
 244                 if (DIFF_OPT_TST(opt, TREE_IN_RECURSIVE)) {
 245                         newbase[baselen + pathlen] = 0;
 246                         opt->add_remove(opt, *prefix, mode, sha1, newbase, 0);
 247                         newbase[baselen + pathlen] = '/';
 248                 }
 249
 250                 init_tree_desc(&inner, tree, size);
 251                 show_tree(opt, prefix, &inner, newbase, baselen + 1 + pathlen);
 252
 253                 free(tree);
 254                 free(newbase);
 255         } else {
 256                 char *fullname = malloc_fullname(base, baselen, path, pathlen);
 257                 opt->add_remove(opt, prefix[0], mode, sha1, fullname, 0);
 258                 free(fullname);
 259         }
 260 }
 261
 262 static void skip_uninteresting(struct tree_desc *t, const char *base, int baselen, struct diff_options *opt, int *all_interesting)
 263 {
 264         while (t->size) {
 265                 int show = tree_entry_interesting(t, base, baselen, opt);
 266                 if (show == 2)
 267                         *all_interesting = 1;
 268                 if (!show) {
 269                         update_tree_entry(t);
 270                         continue;
 271                 }
 272                 /* Skip it all? */
 273                 if (show < 0)
 274                         t->size = 0;
 275                 return;
 276         }
 277 }
 278
 279 int diff_tree(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
 280 {
 281         int baselen = strlen(base);
 282         int all_t1_interesting = 0;
 283         int all_t2_interesting = 0;
 284
 285         for (;;) {
 286                 if (DIFF_OPT_TST(opt, QUICK) &&
 287                     DIFF_OPT_TST(opt, HAS_CHANGES))
 288                         break;
 289                 if (opt->nr_paths) {
 290                         if (!all_t1_interesting)
 291                                 skip_uninteresting(t1, base, baselen, opt,
 292                                                    &all_t1_interesting);
 293                         if (!all_t2_interesting)
 294                                 skip_uninteresting(t2, base, baselen, opt,
 295                                                    &all_t2_interesting);
 296                 }
 297                 if (!t1->size) {
 298                         if (!t2->size)
 299                                 break;
 300                         show_entry(opt, "+", t2, base, baselen);
 301                         update_tree_entry(t2);
 302                         continue;
 303                 }
 304                 if (!t2->size) {
 305                         show_entry(opt, "-", t1, base, baselen);
 306                         update_tree_entry(t1);
 307                         continue;
 308                 }
 309                 switch (compare_tree_entry(t1, t2, base, baselen, opt)) {
 310                 case -1:
 311                         update_tree_entry(t1);
 312                         continue;
 313                 case 0:
 314                         update_tree_entry(t1);
 315                         /* Fallthrough */
 316                 case 1:
 317                         update_tree_entry(t2);
 318                         continue;
 319                 }
 320                 die("git diff-tree: internal error");
 321         }
 322         return 0;
 323 }
 324
 325 /*
 326  * Does it look like the resulting diff might be due to a rename?
 327  *  - single entry
 328  *  - not a valid previous file
 329  */
 330 static inline int diff_might_be_rename(void)
 331 {
 332         return diff_queued_diff.nr == 1 &&
 333                 !DIFF_FILE_VALID(diff_queued_diff.queue[0]->one);
 334 }
 335
 336 static void try_to_follow_renames(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
 337 {
 338         struct diff_options diff_opts;
 339         struct diff_queue_struct *q = &diff_queued_diff;
 340         struct diff_filepair *choice;
 341         const char *paths[1];
 342         int i;
 343
 344         /* Remove the file creation entry from the diff queue, and remember it */
 345         choice = q->queue[0];
 346         q->nr = 0;
 347
 348         diff_setup(&diff_opts);
 349         DIFF_OPT_SET(&diff_opts, RECURSIVE);
 350         DIFF_OPT_SET(&diff_opts, FIND_COPIES_HARDER);
 351         diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 352         diff_opts.single_follow = opt->paths[0];
 353         diff_opts.break_opt = opt->break_opt;
 354         paths[0] = NULL;
 355         diff_tree_setup_paths(paths, &diff_opts);
 356         if (diff_setup_done(&diff_opts) < 0)
 357                 die("unable to set up diff options to follow renames");
 358         diff_tree(t1, t2, base, &diff_opts);
 359         diffcore_std(&diff_opts);
 360         diff_tree_release_paths(&diff_opts);
 361
 362         /* Go through the new set of filepairing, and see if we find a more interesting one */
 363         opt->found_follow = 0;
 364         for (i = 0; i < q->nr; i++) {
 365                 struct diff_filepair *p = q->queue[i];
 366
 367                 /*
 368                  * Found a source? Not only do we use that for the new
 369                  * diff_queued_diff, we will also use that as the path in
 370                  * the future!
 371                  */
 372                 if ((p->status == 'R' || p->status == 'C') && !strcmp(p->two->path, opt->paths[0])) {
 373                         /* Switch the file-pairs around */
 374                         q->queue[i] = choice;
 375                         choice = p;
 376
 377                         /* Update the path we use from now on.. */
 378                         diff_tree_release_paths(opt);
 379                         opt->paths[0] = xstrdup(p->one->path);
 380                         diff_tree_setup_paths(opt->paths, opt);
 381
 382                         /*
 383                          * The caller expects us to return a set of vanilla
 384                          * filepairs to let a later call to diffcore_std()
 385                          * it makes to sort the renames out (among other
 386                          * things), but we already have found renames
 387                          * ourselves; signal diffcore_std() not to muck with
 388                          * rename information.
 389                          */
 390                         opt->found_follow = 1;
 391                         break;
 392                 }
 393         }
 394
 395         /*
 396          * Then, discard all the non-relevant file pairs...
 397          */
 398         for (i = 0; i < q->nr; i++) {
 399                 struct diff_filepair *p = q->queue[i];
 400                 diff_free_filepair(p);
 401         }
 402
 403         /*
 404          * .. and re-instate the one we want (which might be either the
 405          * original one, or the rename/copy we found)
 406          */
 407         q->queue[0] = choice;
 408         q->nr = 1;
 409 }
 410
 411 int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base, struct diff_options *opt)
 412 {
 413         void *tree1, *tree2;
 414         struct tree_desc t1, t2;
 415         unsigned long size1, size2;
 416         int retval;
 417
 418         tree1 = read_object_with_reference(old, tree_type, &size1, NULL);
 419         if (!tree1)
 420                 die("unable to read source tree (%s)", sha1_to_hex(old));
 421         tree2 = read_object_with_reference(new, tree_type, &size2, NULL);
 422         if (!tree2)
 423                 die("unable to read destination tree (%s)", sha1_to_hex(new));
 424         init_tree_desc(&t1, tree1, size1);
 425         init_tree_desc(&t2, tree2, size2);
 426         retval = diff_tree(&t1, &t2, base, opt);
 427         if (!*base && DIFF_OPT_TST(opt, FOLLOW_RENAMES) && diff_might_be_rename()) {
 428                 init_tree_desc(&t1, tree1, size1);
 429                 init_tree_desc(&t2, tree2, size2);
 430                 try_to_follow_renames(&t1, &t2, base, opt);
 431         }
 432         free(tree1);
 433         free(tree2);
 434         return retval;
 435 }
 436
 437 int diff_root_tree_sha1(const unsigned char *new, const char *base, struct diff_options *opt)
 438 {
 439         int retval;
 440         void *tree;
 441         unsigned long size;
 442         struct tree_desc empty, real;
 443
 444         tree = read_object_with_reference(new, tree_type, &size, NULL);
 445         if (!tree)
 446                 die("unable to read root tree (%s)", sha1_to_hex(new));
 447         init_tree_desc(&real, tree, size);
 448
 449         init_tree_desc(&empty, "", 0);
 450         retval = diff_tree(&empty, &real, base, opt);
 451         free(tree);
 452         return retval;
 453 }
 454
 455 static int count_paths(const char **paths)
 456 {
 457         int i = 0;
 458         while (*paths++)
 459                 i++;
 460         return i;
 461 }
 462
 463 void diff_tree_release_paths(struct diff_options *opt)
 464 {
 465         free(opt->pathlens);
 466 }
 467
 468 void diff_tree_setup_paths(const char **p, struct diff_options *opt)
 469 {
 470         opt->nr_paths = 0;
 471         opt->pathlens = NULL;
 472         opt->paths = NULL;
 473
 474         if (p) {
 475                 int i;
 476
 477                 opt->paths = p;
 478                 opt->nr_paths = count_paths(p);
 479                 if (opt->nr_paths == 0) {
 480                         opt->pathlens = NULL;
 481                         return;
 482                 }
 483                 opt->pathlens = xmalloc(opt->nr_paths * sizeof(int));
 484                 for (i=0; i < opt->nr_paths; i++)
 485                         opt->pathlens[i] = strlen(p[i]);
 486         }
 487 }