diffcore-rename.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "diff.h"
   6 #include "diffcore.h"
   7
   8 /* Table of rename/copy destinations */
   9
  10 static struct diff_rename_dst {
  11         struct diff_filespec *two;
  12         struct diff_filepair *pair;
  13 } *rename_dst;
  14 static int rename_dst_nr, rename_dst_alloc;
  15
  16 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
  17                                                  int insert_ok)
  18 {
  19         int first, last;
  20
  21         first = 0;
  22         last = rename_dst_nr;
  23         while (last > first) {
  24                 int next = (last + first) >> 1;
  25                 struct diff_rename_dst *dst = &(rename_dst[next]);
  26                 int cmp = strcmp(two->path, dst->two->path);
  27                 if (!cmp)
  28                         return dst;
  29                 if (cmp < 0) {
  30                         last = next;
  31                         continue;
  32                 }
  33                 first = next+1;
  34         }
  35         /* not found */
  36         if (!insert_ok)
  37                 return NULL;
  38         /* insert to make it at "first" */
  39         if (rename_dst_alloc <= rename_dst_nr) {
  40                 rename_dst_alloc = alloc_nr(rename_dst_alloc);
  41                 rename_dst = xrealloc(rename_dst,
  42                                       rename_dst_alloc * sizeof(*rename_dst));
  43         }
  44         rename_dst_nr++;
  45         if (first < rename_dst_nr)
  46                 memmove(rename_dst + first + 1, rename_dst + first,
  47                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  48         rename_dst[first].two = alloc_filespec(two->path);
  49         fill_filespec(rename_dst[first].two, two->sha1, two->mode);
  50         rename_dst[first].pair = NULL;
  51         return &(rename_dst[first]);
  52 }
  53
  54 /* Table of rename/copy src files */
  55 static struct diff_rename_src {
  56         struct diff_filespec *one;
  57         unsigned short score; /* to remember the break score */
  58         unsigned src_path_left : 1;
  59 } *rename_src;
  60 static int rename_src_nr, rename_src_alloc;
  61
  62 static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
  63                                                    int src_path_left,
  64                                                    unsigned short score)
  65 {
  66         int first, last;
  67
  68         first = 0;
  69         last = rename_src_nr;
  70         while (last > first) {
  71                 int next = (last + first) >> 1;
  72                 struct diff_rename_src *src = &(rename_src[next]);
  73                 int cmp = strcmp(one->path, src->one->path);
  74                 if (!cmp)
  75                         return src;
  76                 if (cmp < 0) {
  77                         last = next;
  78                         continue;
  79                 }
  80                 first = next+1;
  81         }
  82
  83         /* insert to make it at "first" */
  84         if (rename_src_alloc <= rename_src_nr) {
  85                 rename_src_alloc = alloc_nr(rename_src_alloc);
  86                 rename_src = xrealloc(rename_src,
  87                                       rename_src_alloc * sizeof(*rename_src));
  88         }
  89         rename_src_nr++;
  90         if (first < rename_src_nr)
  91                 memmove(rename_src + first + 1, rename_src + first,
  92                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  93         rename_src[first].one = one;
  94         rename_src[first].score = score;
  95         rename_src[first].src_path_left = src_path_left;
  96         return &(rename_src[first]);
  97 }
  98
  99 static int is_exact_match(struct diff_filespec *src,
 100                           struct diff_filespec *dst,
 101                           int contents_too)
 102 {
 103         if (src->sha1_valid && dst->sha1_valid &&
 104             !hashcmp(src->sha1, dst->sha1))
 105                 return 1;
 106         if (!contents_too)
 107                 return 0;
 108         if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
 109                 return 0;
 110         if (src->size != dst->size)
 111                 return 0;
 112         if (src->sha1_valid && dst->sha1_valid)
 113             return !hashcmp(src->sha1, dst->sha1);
 114         if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 115                 return 0;
 116         if (src->size == dst->size &&
 117             !memcmp(src->data, dst->data, src->size))
 118                 return 1;
 119         return 0;
 120 }
 121
 122 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
 123 {
 124         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 125         while (src_len && dst_len) {
 126                 char c1 = src->path[--src_len];
 127                 char c2 = dst->path[--dst_len];
 128                 if (c1 != c2)
 129                         return 0;
 130                 if (c1 == '/')
 131                         return 1;
 132         }
 133         return (!src_len || src->path[src_len - 1] == '/') &&
 134                 (!dst_len || dst->path[dst_len - 1] == '/');
 135 }
 136
 137 struct diff_score {
 138         int src; /* index in rename_src */
 139         int dst; /* index in rename_dst */
 140         int score;
 141         int name_score;
 142 };
 143
 144 static int estimate_similarity(struct diff_filespec *src,
 145                                struct diff_filespec *dst,
 146                                int minimum_score)
 147 {
 148         /* src points at a file that existed in the original tree (or
 149          * optionally a file in the destination tree) and dst points
 150          * at a newly created file.  They may be quite similar, in which
 151          * case we want to say src is renamed to dst or src is copied into
 152          * dst, and then some edit has been applied to dst.
 153          *
 154          * Compare them and return how similar they are, representing
 155          * the score as an integer between 0 and MAX_SCORE.
 156          *
 157          * When there is an exact match, it is considered a better
 158          * match than anything else; the destination does not even
 159          * call into this function in that case.
 160          */
 161         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 162         unsigned long delta_limit;
 163         int score;
 164
 165         /* We deal only with regular files.  Symlink renames are handled
 166          * only when they are exact matches --- in other words, no edits
 167          * after renaming.
 168          */
 169         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 170                 return 0;
 171
 172         max_size = ((src->size > dst->size) ? src->size : dst->size);
 173         base_size = ((src->size < dst->size) ? src->size : dst->size);
 174         delta_size = max_size - base_size;
 175
 176         /* We would not consider edits that change the file size so
 177          * drastically.  delta_size must be smaller than
 178          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 179          *
 180          * Note that base_size == 0 case is handled here already
 181          * and the final score computation below would not have a
 182          * divide-by-zero issue.
 183          */
 184         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 185                 return 0;
 186
 187         if ((!src->cnt_data && diff_populate_filespec(src, 0))
 188                 || (!dst->cnt_data && diff_populate_filespec(dst, 0)))
 189                 return 0; /* error but caught downstream */
 190
 191
 192         delta_limit = (unsigned long)
 193                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 194         if (diffcore_count_changes(src, dst,
 195                                    &src->cnt_data, &dst->cnt_data,
 196                                    delta_limit,
 197                                    &src_copied, &literal_added))
 198                 return 0;
 199
 200         /* How similar are they?
 201          * what percentage of material in dst are from source?
 202          */
 203         if (!dst->size)
 204                 score = 0; /* should not happen */
 205         else
 206                 score = (int)(src_copied * MAX_SCORE / max_size);
 207         return score;
 208 }
 209
 210 static void record_rename_pair(int dst_index, int src_index, int score)
 211 {
 212         struct diff_filespec *src, *dst;
 213         struct diff_filepair *dp;
 214
 215         if (rename_dst[dst_index].pair)
 216                 die("internal error: dst already matched.");
 217
 218         src = rename_src[src_index].one;
 219         src->count++;
 220
 221         dst = rename_dst[dst_index].two;
 222         dst->count++;
 223
 224         dp = diff_queue(NULL, src, dst);
 225         dp->renamed_pair = 1;
 226         if (!strcmp(src->path, dst->path))
 227                 dp->score = rename_src[src_index].score;
 228         else
 229                 dp->score = score;
 230         dp->source_stays = rename_src[src_index].src_path_left;
 231         rename_dst[dst_index].pair = dp;
 232 }
 233
 234 /*
 235  * We sort the rename similarity matrix with the score, in descending
 236  * order (the most similar first).
 237  */
 238 static int score_compare(const void *a_, const void *b_)
 239 {
 240         const struct diff_score *a = a_, *b = b_;
 241
 242         if (a->score == b->score)
 243                 return b->name_score - a->name_score;
 244
 245         return b->score - a->score;
 246 }
 247
 248 static int compute_stays(struct diff_queue_struct *q,
 249                          struct diff_filespec *one)
 250 {
 251         int i;
 252         for (i = 0; i < q->nr; i++) {
 253                 struct diff_filepair *p = q->queue[i];
 254                 if (strcmp(one->path, p->two->path))
 255                         continue;
 256                 if (DIFF_PAIR_RENAME(p)) {
 257                         return 0; /* something else is renamed into this */
 258                 }
 259         }
 260         return 1;
 261 }
 262
 263 /*
 264  * Find exact renames first.
 265  *
 266  * The first round matches up the up-to-date entries,
 267  * and then during the second round we try to match
 268  * cache-dirty entries as well.
 269  *
 270  * Note: the rest of the rename logic depends on this
 271  * phase also populating all the filespecs for any
 272  * entry that isn't matched up with an exact rename,
 273  * see "is_exact_match()".
 274  */
 275 static int find_exact_renames(void)
 276 {
 277         int rename_count = 0;
 278         int contents_too;
 279
 280         for (contents_too = 0; contents_too < 2; contents_too++) {
 281                 int i;
 282
 283                 for (i = 0; i < rename_dst_nr; i++) {
 284                         struct diff_filespec *two = rename_dst[i].two;
 285                         int j;
 286
 287                         if (rename_dst[i].pair)
 288                                 continue; /* dealt with an earlier round */
 289                         for (j = 0; j < rename_src_nr; j++) {
 290                                 int k;
 291                                 struct diff_filespec *one = rename_src[j].one;
 292                                 if (!is_exact_match(one, two, contents_too))
 293                                         continue;
 294
 295                                 /* see if there is a basename match, too */
 296                                 for (k = j; k < rename_src_nr; k++) {
 297                                         one = rename_src[k].one;
 298                                         if (basename_same(one, two) &&
 299                                                 is_exact_match(one, two,
 300                                                         contents_too)) {
 301                                                 j = k;
 302                                                 break;
 303                                         }
 304                                 }
 305
 306                                 record_rename_pair(i, j, (int)MAX_SCORE);
 307                                 rename_count++;
 308                                 break; /* we are done with this entry */
 309                         }
 310                 }
 311         }
 312         return rename_count;
 313 }
 314
 315 void diffcore_rename(struct diff_options *options)
 316 {
 317         int detect_rename = options->detect_rename;
 318         int minimum_score = options->rename_score;
 319         int rename_limit = options->rename_limit;
 320         struct diff_queue_struct *q = &diff_queued_diff;
 321         struct diff_queue_struct outq;
 322         struct diff_score *mx;
 323         int i, j, rename_count;
 324         int num_create, num_src, dst_cnt;
 325
 326         if (!minimum_score)
 327                 minimum_score = DEFAULT_RENAME_SCORE;
 328
 329         for (i = 0; i < q->nr; i++) {
 330                 struct diff_filepair *p = q->queue[i];
 331                 if (!DIFF_FILE_VALID(p->one)) {
 332                         if (!DIFF_FILE_VALID(p->two))
 333                                 continue; /* unmerged */
 334                         else if (options->single_follow &&
 335                                  strcmp(options->single_follow, p->two->path))
 336                                 continue; /* not interested */
 337                         else
 338                                 locate_rename_dst(p->two, 1);
 339                 }
 340                 else if (!DIFF_FILE_VALID(p->two)) {
 341                         /* If the source is a broken "delete", and
 342                          * they did not really want to get broken,
 343                          * that means the source actually stays.
 344                          */
 345                         int stays = (p->broken_pair && !p->score);
 346                         register_rename_src(p->one, stays, p->score);
 347                 }
 348                 else if (detect_rename == DIFF_DETECT_COPY)
 349                         register_rename_src(p->one, 1, p->score);
 350         }
 351         if (rename_dst_nr == 0 || rename_src_nr == 0)
 352                 goto cleanup; /* nothing to do */
 353
 354         /*
 355          * This basically does a test for the rename matrix not
 356          * growing larger than a "rename_limit" square matrix, ie:
 357          *
 358          *    rename_dst_nr * rename_src_nr > rename_limit * rename_limit
 359          *
 360          * but handles the potential overflow case specially (and we
 361          * assume at least 32-bit integers)
 362          */
 363         if (rename_limit <= 0 || rename_limit > 32767)
 364                 rename_limit = 32767;
 365         if (rename_dst_nr > rename_limit && rename_src_nr > rename_limit)
 366                 goto cleanup;
 367         if (rename_dst_nr * rename_src_nr > rename_limit * rename_limit)
 368                 goto cleanup;
 369
 370         /*
 371          * We really want to cull the candidates list early
 372          * with cheap tests in order to avoid doing deltas.
 373          */
 374         rename_count = find_exact_renames();
 375
 376         /* Have we run out the created file pool?  If so we can avoid
 377          * doing the delta matrix altogether.
 378          */
 379         if (rename_count == rename_dst_nr)
 380                 goto cleanup;
 381
 382         if (minimum_score == MAX_SCORE)
 383                 goto cleanup;
 384
 385         num_create = (rename_dst_nr - rename_count);
 386         num_src = rename_src_nr;
 387         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 388         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 389                 int base = dst_cnt * num_src;
 390                 struct diff_filespec *two = rename_dst[i].two;
 391                 if (rename_dst[i].pair)
 392                         continue; /* dealt with exact match already. */
 393                 for (j = 0; j < rename_src_nr; j++) {
 394                         struct diff_filespec *one = rename_src[j].one;
 395                         struct diff_score *m = &mx[base+j];
 396                         m->src = j;
 397                         m->dst = i;
 398                         m->score = estimate_similarity(one, two,
 399                                                        minimum_score);
 400                         m->name_score = basename_same(one, two);
 401                         diff_free_filespec_blob(one);
 402                 }
 403                 /* We do not need the text anymore */
 404                 diff_free_filespec_blob(two);
 405                 dst_cnt++;
 406         }
 407         /* cost matrix sorted by most to least similar pair */
 408         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 409         for (i = 0; i < num_create * num_src; i++) {
 410                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 411                 if (dst->pair)
 412                         continue; /* already done, either exact or fuzzy. */
 413                 if (mx[i].score < minimum_score)
 414                         break; /* there is no more usable pair. */
 415                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 416                 rename_count++;
 417         }
 418         free(mx);
 419
 420  cleanup:
 421         /* At this point, we have found some renames and copies and they
 422          * are recorded in rename_dst.  The original list is still in *q.
 423          */
 424         outq.queue = NULL;
 425         outq.nr = outq.alloc = 0;
 426         for (i = 0; i < q->nr; i++) {
 427                 struct diff_filepair *p = q->queue[i];
 428                 struct diff_filepair *pair_to_free = NULL;
 429
 430                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 431                         /*
 432                          * Creation
 433                          *
 434                          * We would output this create record if it has
 435                          * not been turned into a rename/copy already.
 436                          */
 437                         struct diff_rename_dst *dst =
 438                                 locate_rename_dst(p->two, 0);
 439                         if (dst && dst->pair) {
 440                                 diff_q(&outq, dst->pair);
 441                                 pair_to_free = p;
 442                         }
 443                         else
 444                                 /* no matching rename/copy source, so
 445                                  * record this as a creation.
 446                                  */
 447                                 diff_q(&outq, p);
 448                 }
 449                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 450                         /*
 451                          * Deletion
 452                          *
 453                          * We would output this delete record if:
 454                          *
 455                          * (1) this is a broken delete and the counterpart
 456                          *     broken create remains in the output; or
 457                          * (2) this is not a broken delete, and rename_dst
 458                          *     does not have a rename/copy to move p->one->path
 459                          *     out of existence.
 460                          *
 461                          * Otherwise, the counterpart broken create
 462                          * has been turned into a rename-edit; or
 463                          * delete did not have a matching create to
 464                          * begin with.
 465                          */
 466                         if (DIFF_PAIR_BROKEN(p)) {
 467                                 /* broken delete */
 468                                 struct diff_rename_dst *dst =
 469                                         locate_rename_dst(p->one, 0);
 470                                 if (dst && dst->pair)
 471                                         /* counterpart is now rename/copy */
 472                                         pair_to_free = p;
 473                         }
 474                         else {
 475                                 for (j = 0; j < rename_dst_nr; j++) {
 476                                         if (!rename_dst[j].pair)
 477                                                 continue;
 478                                         if (strcmp(rename_dst[j].pair->
 479                                                    one->path,
 480                                                    p->one->path))
 481                                                 continue;
 482                                         break;
 483                                 }
 484                                 if (j < rename_dst_nr)
 485                                         /* this path remains */
 486                                         pair_to_free = p;
 487                         }
 488
 489                         if (pair_to_free)
 490                                 ;
 491                         else
 492                                 diff_q(&outq, p);
 493                 }
 494                 else if (!diff_unmodified_pair(p))
 495                         /* all the usual ones need to be kept */
 496                         diff_q(&outq, p);
 497                 else
 498                         /* no need to keep unmodified pairs */
 499                         pair_to_free = p;
 500
 501                 if (pair_to_free)
 502                         diff_free_filepair(pair_to_free);
 503         }
 504         diff_debug_queue("done copying original", &outq);
 505
 506         free(q->queue);
 507         *q = outq;
 508         diff_debug_queue("done collapsing", q);
 509
 510         /* We need to see which rename source really stays here;
 511          * earlier we only checked if the path is left in the result,
 512          * but even if a path remains in the result, if that is coming
 513          * from copying something else on top of it, then the original
 514          * source is lost and does not stay.
 515          */
 516         for (i = 0; i < q->nr; i++) {
 517                 struct diff_filepair *p = q->queue[i];
 518                 if (DIFF_PAIR_RENAME(p) && p->source_stays) {
 519                         /* If one appears as the target of a rename-copy,
 520                          * then mark p->source_stays = 0; otherwise
 521                          * leave it as is.
 522                          */
 523                         p->source_stays = compute_stays(q, p->one);
 524                 }
 525         }
 526
 527         for (i = 0; i < rename_dst_nr; i++)
 528                 free_filespec(rename_dst[i].two);
 529
 530         free(rename_dst);
 531         rename_dst = NULL;
 532         rename_dst_nr = rename_dst_alloc = 0;
 533         free(rename_src);
 534         rename_src = NULL;
 535         rename_src_nr = rename_src_alloc = 0;
 536         return;
 537 }