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