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 #include "object-store.h"
   8 #include "hashmap.h"
   9 #include "progress.h"
  10
  11 /* Table of rename/copy destinations */
  12
  13 static struct diff_rename_dst {
  14         struct diff_filespec *two;
  15         struct diff_filepair *pair;
  16 } *rename_dst;
  17 static int rename_dst_nr, rename_dst_alloc;
  18
  19 static int find_rename_dst(struct diff_filespec *two)
  20 {
  21         int first, last;
  22
  23         first = 0;
  24         last = rename_dst_nr;
  25         while (last > first) {
  26                 int next = first + ((last - first) >> 1);
  27                 struct diff_rename_dst *dst = &(rename_dst[next]);
  28                 int cmp = strcmp(two->path, dst->two->path);
  29                 if (!cmp)
  30                         return next;
  31                 if (cmp < 0) {
  32                         last = next;
  33                         continue;
  34                 }
  35                 first = next+1;
  36         }
  37         return -first - 1;
  38 }
  39
  40 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two)
  41 {
  42         int ofs = find_rename_dst(two);
  43         return ofs < 0 ? NULL : &rename_dst[ofs];
  44 }
  45
  46 /*
  47  * Returns 0 on success, -1 if we found a duplicate.
  48  */
  49 static int add_rename_dst(struct diff_filespec *two)
  50 {
  51         int first = find_rename_dst(two);
  52
  53         if (first >= 0)
  54                 return -1;
  55         first = -first - 1;
  56
  57         /* insert to make it at "first" */
  58         ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc);
  59         rename_dst_nr++;
  60         if (first < rename_dst_nr)
  61                 MOVE_ARRAY(rename_dst + first + 1, rename_dst + first,
  62                            rename_dst_nr - first - 1);
  63         rename_dst[first].two = alloc_filespec(two->path);
  64         fill_filespec(rename_dst[first].two, &two->oid, two->oid_valid,
  65                       two->mode);
  66         rename_dst[first].pair = NULL;
  67         return 0;
  68 }
  69
  70 /* Table of rename/copy src files */
  71 static struct diff_rename_src {
  72         struct diff_filepair *p;
  73         unsigned short score; /* to remember the break score */
  74 } *rename_src;
  75 static int rename_src_nr, rename_src_alloc;
  76
  77 static struct diff_rename_src *register_rename_src(struct diff_filepair *p)
  78 {
  79         int first, last;
  80         struct diff_filespec *one = p->one;
  81         unsigned short score = p->score;
  82
  83         first = 0;
  84         last = rename_src_nr;
  85         while (last > first) {
  86                 int next = first + ((last - first) >> 1);
  87                 struct diff_rename_src *src = &(rename_src[next]);
  88                 int cmp = strcmp(one->path, src->p->one->path);
  89                 if (!cmp)
  90                         return src;
  91                 if (cmp < 0) {
  92                         last = next;
  93                         continue;
  94                 }
  95                 first = next+1;
  96         }
  97
  98         /* insert to make it at "first" */
  99         ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc);
 100         rename_src_nr++;
 101         if (first < rename_src_nr)
 102                 MOVE_ARRAY(rename_src + first + 1, rename_src + first,
 103                            rename_src_nr - first - 1);
 104         rename_src[first].p = p;
 105         rename_src[first].score = score;
 106         return &(rename_src[first]);
 107 }
 108
 109 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
 110 {
 111         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 112         while (src_len && dst_len) {
 113                 char c1 = src->path[--src_len];
 114                 char c2 = dst->path[--dst_len];
 115                 if (c1 != c2)
 116                         return 0;
 117                 if (c1 == '/')
 118                         return 1;
 119         }
 120         return (!src_len || src->path[src_len - 1] == '/') &&
 121                 (!dst_len || dst->path[dst_len - 1] == '/');
 122 }
 123
 124 struct diff_score {
 125         int src; /* index in rename_src */
 126         int dst; /* index in rename_dst */
 127         unsigned short score;
 128         short name_score;
 129 };
 130
 131 static int estimate_similarity(struct repository *r,
 132                                struct diff_filespec *src,
 133                                struct diff_filespec *dst,
 134                                int minimum_score)
 135 {
 136         /* src points at a file that existed in the original tree (or
 137          * optionally a file in the destination tree) and dst points
 138          * at a newly created file.  They may be quite similar, in which
 139          * case we want to say src is renamed to dst or src is copied into
 140          * dst, and then some edit has been applied to dst.
 141          *
 142          * Compare them and return how similar they are, representing
 143          * the score as an integer between 0 and MAX_SCORE.
 144          *
 145          * When there is an exact match, it is considered a better
 146          * match than anything else; the destination does not even
 147          * call into this function in that case.
 148          */
 149         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 150         int score;
 151
 152         /* We deal only with regular files.  Symlink renames are handled
 153          * only when they are exact matches --- in other words, no edits
 154          * after renaming.
 155          */
 156         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 157                 return 0;
 158
 159         /*
 160          * Need to check that source and destination sizes are
 161          * filled in before comparing them.
 162          *
 163          * If we already have "cnt_data" filled in, we know it's
 164          * all good (avoid checking the size for zero, as that
 165          * is a possible size - we really should have a flag to
 166          * say whether the size is valid or not!)
 167          */
 168         if (!src->cnt_data &&
 169             diff_populate_filespec(r, src, CHECK_SIZE_ONLY))
 170                 return 0;
 171         if (!dst->cnt_data &&
 172             diff_populate_filespec(r, dst, CHECK_SIZE_ONLY))
 173                 return 0;
 174
 175         max_size = ((src->size > dst->size) ? src->size : dst->size);
 176         base_size = ((src->size < dst->size) ? src->size : dst->size);
 177         delta_size = max_size - base_size;
 178
 179         /* We would not consider edits that change the file size so
 180          * drastically.  delta_size must be smaller than
 181          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 182          *
 183          * Note that base_size == 0 case is handled here already
 184          * and the final score computation below would not have a
 185          * divide-by-zero issue.
 186          */
 187         if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 188                 return 0;
 189
 190         if (!src->cnt_data && diff_populate_filespec(r, src, 0))
 191                 return 0;
 192         if (!dst->cnt_data && diff_populate_filespec(r, dst, 0))
 193                 return 0;
 194
 195         if (diffcore_count_changes(r, src, dst,
 196                                    &src->cnt_data, &dst->cnt_data,
 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].p->one;
 219         src->rename_used++;
 220         src->count++;
 221
 222         dst = rename_dst[dst_index].two;
 223         dst->count++;
 224
 225         dp = diff_queue(NULL, src, dst);
 226         dp->renamed_pair = 1;
 227         if (!strcmp(src->path, dst->path))
 228                 dp->score = rename_src[src_index].score;
 229         else
 230                 dp->score = score;
 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         /* sink the unused ones to the bottom */
 243         if (a->dst < 0)
 244                 return (0 <= b->dst);
 245         else if (b->dst < 0)
 246                 return -1;
 247
 248         if (a->score == b->score)
 249                 return b->name_score - a->name_score;
 250
 251         return b->score - a->score;
 252 }
 253
 254 struct file_similarity {
 255         struct hashmap_entry entry;
 256         int index;
 257         struct diff_filespec *filespec;
 258 };
 259
 260 static unsigned int hash_filespec(struct repository *r,
 261                                   struct diff_filespec *filespec)
 262 {
 263         if (!filespec->oid_valid) {
 264                 if (diff_populate_filespec(r, filespec, 0))
 265                         return 0;
 266                 hash_object_file(filespec->data, filespec->size, "blob",
 267                                  &filespec->oid);
 268         }
 269         return oidhash(&filespec->oid);
 270 }
 271
 272 static int find_identical_files(struct hashmap *srcs,
 273                                 int dst_index,
 274                                 struct diff_options *options)
 275 {
 276         int renames = 0;
 277         struct hashmap_entry *ent;
 278         struct diff_filespec *target = rename_dst[dst_index].two;
 279         struct file_similarity *p, *best = NULL;
 280         int i = 100, best_score = -1;
 281
 282         /*
 283          * Find the best source match for specified destination.
 284          */
 285         ent = hashmap_get_from_hash(srcs,
 286                                   hash_filespec(options->repo, target),
 287                                   NULL);
 288         for (; ent; ent = hashmap_get_next(srcs, ent)) {
 289                 int score;
 290                 struct diff_filespec *source;
 291
 292                 p = container_of(ent, struct file_similarity, entry);
 293                 source = p->filespec;
 294
 295                 /* False hash collision? */
 296                 if (!oideq(&source->oid, &target->oid))
 297                         continue;
 298                 /* Non-regular files? If so, the modes must match! */
 299                 if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 300                         if (source->mode != target->mode)
 301                                 continue;
 302                 }
 303                 /* Give higher scores to sources that haven't been used already */
 304                 score = !source->rename_used;
 305                 if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY)
 306                         continue;
 307                 score += basename_same(source, target);
 308                 if (score > best_score) {
 309                         best = p;
 310                         best_score = score;
 311                         if (score == 2)
 312                                 break;
 313                 }
 314
 315                 /* Too many identical alternatives? Pick one */
 316                 if (!--i)
 317                         break;
 318         }
 319         if (best) {
 320                 record_rename_pair(dst_index, best->index, MAX_SCORE);
 321                 renames++;
 322         }
 323         return renames;
 324 }
 325
 326 static void insert_file_table(struct repository *r,
 327                               struct hashmap *table, int index,
 328                               struct diff_filespec *filespec)
 329 {
 330         struct file_similarity *entry = xmalloc(sizeof(*entry));
 331
 332         entry->index = index;
 333         entry->filespec = filespec;
 334
 335         hashmap_entry_init(&entry->entry, hash_filespec(r, filespec));
 336         hashmap_add(table, &entry->entry);
 337 }
 338
 339 /*
 340  * Find exact renames first.
 341  *
 342  * The first round matches up the up-to-date entries,
 343  * and then during the second round we try to match
 344  * cache-dirty entries as well.
 345  */
 346 static int find_exact_renames(struct diff_options *options)
 347 {
 348         int i, renames = 0;
 349         struct hashmap file_table;
 350
 351         /* Add all sources to the hash table in reverse order, because
 352          * later on they will be retrieved in LIFO order.
 353          */
 354         hashmap_init(&file_table, NULL, NULL, rename_src_nr);
 355         for (i = rename_src_nr-1; i >= 0; i--)
 356                 insert_file_table(options->repo,
 357                                   &file_table, i,
 358                                   rename_src[i].p->one);
 359
 360         /* Walk the destinations and find best source match */
 361         for (i = 0; i < rename_dst_nr; i++)
 362                 renames += find_identical_files(&file_table, i, options);
 363
 364         /* Free the hash data structure and entries */
 365         hashmap_free(&file_table, 1);
 366
 367         return renames;
 368 }
 369
 370 #define NUM_CANDIDATE_PER_DST 4
 371 static void record_if_better(struct diff_score m[], struct diff_score *o)
 372 {
 373         int i, worst;
 374
 375         /* find the worst one */
 376         worst = 0;
 377         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 378                 if (score_compare(&m[i], &m[worst]) > 0)
 379                         worst = i;
 380
 381         /* is it better than the worst one? */
 382         if (score_compare(&m[worst], o) > 0)
 383                 m[worst] = *o;
 384 }
 385
 386 /*
 387  * Returns:
 388  * 0 if we are under the limit;
 389  * 1 if we need to disable inexact rename detection;
 390  * 2 if we would be under the limit if we were given -C instead of -C -C.
 391  */
 392 static int too_many_rename_candidates(int num_create,
 393                                       struct diff_options *options)
 394 {
 395         int rename_limit = options->rename_limit;
 396         int num_src = rename_src_nr;
 397         int i;
 398
 399         options->needed_rename_limit = 0;
 400
 401         /*
 402          * This basically does a test for the rename matrix not
 403          * growing larger than a "rename_limit" square matrix, ie:
 404          *
 405          *    num_create * num_src > rename_limit * rename_limit
 406          */
 407         if (rename_limit <= 0)
 408                 rename_limit = 32767;
 409         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 410             ((uint64_t)num_create * (uint64_t)num_src
 411              <= (uint64_t)rename_limit * (uint64_t)rename_limit))
 412                 return 0;
 413
 414         options->needed_rename_limit =
 415                 num_src > num_create ? num_src : num_create;
 416
 417         /* Are we running under -C -C? */
 418         if (!options->flags.find_copies_harder)
 419                 return 1;
 420
 421         /* Would we bust the limit if we were running under -C? */
 422         for (num_src = i = 0; i < rename_src_nr; i++) {
 423                 if (diff_unmodified_pair(rename_src[i].p))
 424                         continue;
 425                 num_src++;
 426         }
 427         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 428             ((uint64_t)num_create * (uint64_t)num_src
 429              <= (uint64_t)rename_limit * (uint64_t)rename_limit))
 430                 return 2;
 431         return 1;
 432 }
 433
 434 static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies)
 435 {
 436         int count = 0, i;
 437
 438         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 439                 struct diff_rename_dst *dst;
 440
 441                 if ((mx[i].dst < 0) ||
 442                     (mx[i].score < minimum_score))
 443                         break; /* there is no more usable pair. */
 444                 dst = &rename_dst[mx[i].dst];
 445                 if (dst->pair)
 446                         continue; /* already done, either exact or fuzzy. */
 447                 if (!copies && rename_src[mx[i].src].p->one->rename_used)
 448                         continue;
 449                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 450                 count++;
 451         }
 452         return count;
 453 }
 454
 455 void diffcore_rename(struct diff_options *options)
 456 {
 457         int detect_rename = options->detect_rename;
 458         int minimum_score = options->rename_score;
 459         struct diff_queue_struct *q = &diff_queued_diff;
 460         struct diff_queue_struct outq;
 461         struct diff_score *mx;
 462         int i, j, rename_count, skip_unmodified = 0;
 463         int num_create, dst_cnt;
 464         struct progress *progress = NULL;
 465
 466         if (!minimum_score)
 467                 minimum_score = DEFAULT_RENAME_SCORE;
 468
 469         for (i = 0; i < q->nr; i++) {
 470                 struct diff_filepair *p = q->queue[i];
 471                 if (!DIFF_FILE_VALID(p->one)) {
 472                         if (!DIFF_FILE_VALID(p->two))
 473                                 continue; /* unmerged */
 474                         else if (options->single_follow &&
 475                                  strcmp(options->single_follow, p->two->path))
 476                                 continue; /* not interested */
 477                         else if (!options->flags.rename_empty &&
 478                                  is_empty_blob_oid(&p->two->oid))
 479                                 continue;
 480                         else if (add_rename_dst(p->two) < 0) {
 481                                 warning("skipping rename detection, detected"
 482                                         " duplicate destination '%s'",
 483                                         p->two->path);
 484                                 goto cleanup;
 485                         }
 486                 }
 487                 else if (!options->flags.rename_empty &&
 488                          is_empty_blob_oid(&p->one->oid))
 489                         continue;
 490                 else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) {
 491                         /*
 492                          * If the source is a broken "delete", and
 493                          * they did not really want to get broken,
 494                          * that means the source actually stays.
 495                          * So we increment the "rename_used" score
 496                          * by one, to indicate ourselves as a user
 497                          */
 498                         if (p->broken_pair && !p->score)
 499                                 p->one->rename_used++;
 500                         register_rename_src(p);
 501                 }
 502                 else if (detect_rename == DIFF_DETECT_COPY) {
 503                         /*
 504                          * Increment the "rename_used" score by
 505                          * one, to indicate ourselves as a user.
 506                          */
 507                         p->one->rename_used++;
 508                         register_rename_src(p);
 509                 }
 510         }
 511         if (rename_dst_nr == 0 || rename_src_nr == 0)
 512                 goto cleanup; /* nothing to do */
 513
 514         /*
 515          * We really want to cull the candidates list early
 516          * with cheap tests in order to avoid doing deltas.
 517          */
 518         rename_count = find_exact_renames(options);
 519
 520         /* Did we only want exact renames? */
 521         if (minimum_score == MAX_SCORE)
 522                 goto cleanup;
 523
 524         /*
 525          * Calculate how many renames are left (but all the source
 526          * files still remain as options for rename/copies!)
 527          */
 528         num_create = (rename_dst_nr - rename_count);
 529
 530         /* All done? */
 531         if (!num_create)
 532                 goto cleanup;
 533
 534         switch (too_many_rename_candidates(num_create, options)) {
 535         case 1:
 536                 goto cleanup;
 537         case 2:
 538                 options->degraded_cc_to_c = 1;
 539                 skip_unmodified = 1;
 540                 break;
 541         default:
 542                 break;
 543         }
 544
 545         if (options->show_rename_progress) {
 546                 progress = start_delayed_progress(
 547                                 _("Performing inexact rename detection"),
 548                                 (uint64_t)rename_dst_nr * (uint64_t)rename_src_nr);
 549         }
 550
 551         mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_create), sizeof(*mx));
 552         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 553                 struct diff_filespec *two = rename_dst[i].two;
 554                 struct diff_score *m;
 555
 556                 if (rename_dst[i].pair)
 557                         continue; /* dealt with exact match already. */
 558
 559                 m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
 560                 for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
 561                         m[j].dst = -1;
 562
 563                 for (j = 0; j < rename_src_nr; j++) {
 564                         struct diff_filespec *one = rename_src[j].p->one;
 565                         struct diff_score this_src;
 566
 567                         if (skip_unmodified &&
 568                             diff_unmodified_pair(rename_src[j].p))
 569                                 continue;
 570
 571                         this_src.score = estimate_similarity(options->repo,
 572                                                              one, two,
 573                                                              minimum_score);
 574                         this_src.name_score = basename_same(one, two);
 575                         this_src.dst = i;
 576                         this_src.src = j;
 577                         record_if_better(m, &this_src);
 578                         /*
 579                          * Once we run estimate_similarity,
 580                          * We do not need the text anymore.
 581                          */
 582                         diff_free_filespec_blob(one);
 583                         diff_free_filespec_blob(two);
 584                 }
 585                 dst_cnt++;
 586                 display_progress(progress, (uint64_t)(i+1)*(uint64_t)rename_src_nr);
 587         }
 588         stop_progress(&progress);
 589
 590         /* cost matrix sorted by most to least similar pair */
 591         QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare);
 592
 593         rename_count += find_renames(mx, dst_cnt, minimum_score, 0);
 594         if (detect_rename == DIFF_DETECT_COPY)
 595                 rename_count += find_renames(mx, dst_cnt, minimum_score, 1);
 596         free(mx);
 597
 598  cleanup:
 599         /* At this point, we have found some renames and copies and they
 600          * are recorded in rename_dst.  The original list is still in *q.
 601          */
 602         DIFF_QUEUE_CLEAR(&outq);
 603         for (i = 0; i < q->nr; i++) {
 604                 struct diff_filepair *p = q->queue[i];
 605                 struct diff_filepair *pair_to_free = NULL;
 606
 607                 if (DIFF_PAIR_UNMERGED(p)) {
 608                         diff_q(&outq, p);
 609                 }
 610                 else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 611                         /*
 612                          * Creation
 613                          *
 614                          * We would output this create record if it has
 615                          * not been turned into a rename/copy already.
 616                          */
 617                         struct diff_rename_dst *dst = locate_rename_dst(p->two);
 618                         if (dst && dst->pair) {
 619                                 diff_q(&outq, dst->pair);
 620                                 pair_to_free = p;
 621                         }
 622                         else
 623                                 /* no matching rename/copy source, so
 624                                  * record this as a creation.
 625                                  */
 626                                 diff_q(&outq, p);
 627                 }
 628                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 629                         /*
 630                          * Deletion
 631                          *
 632                          * We would output this delete record if:
 633                          *
 634                          * (1) this is a broken delete and the counterpart
 635                          *     broken create remains in the output; or
 636                          * (2) this is not a broken delete, and rename_dst
 637                          *     does not have a rename/copy to move p->one->path
 638                          *     out of existence.
 639                          *
 640                          * Otherwise, the counterpart broken create
 641                          * has been turned into a rename-edit; or
 642                          * delete did not have a matching create to
 643                          * begin with.
 644                          */
 645                         if (DIFF_PAIR_BROKEN(p)) {
 646                                 /* broken delete */
 647                                 struct diff_rename_dst *dst = locate_rename_dst(p->one);
 648                                 if (dst && dst->pair)
 649                                         /* counterpart is now rename/copy */
 650                                         pair_to_free = p;
 651                         }
 652                         else {
 653                                 if (p->one->rename_used)
 654                                         /* this path remains */
 655                                         pair_to_free = p;
 656                         }
 657
 658                         if (pair_to_free)
 659                                 ;
 660                         else
 661                                 diff_q(&outq, p);
 662                 }
 663                 else if (!diff_unmodified_pair(p))
 664                         /* all the usual ones need to be kept */
 665                         diff_q(&outq, p);
 666                 else
 667                         /* no need to keep unmodified pairs */
 668                         pair_to_free = p;
 669
 670                 if (pair_to_free)
 671                         diff_free_filepair(pair_to_free);
 672         }
 673         diff_debug_queue("done copying original", &outq);
 674
 675         free(q->queue);
 676         *q = outq;
 677         diff_debug_queue("done collapsing", q);
 678
 679         for (i = 0; i < rename_dst_nr; i++)
 680                 free_filespec(rename_dst[i].two);
 681
 682         FREE_AND_NULL(rename_dst);
 683         rename_dst_nr = rename_dst_alloc = 0;
 684         FREE_AND_NULL(rename_src);
 685         rename_src_nr = rename_src_alloc = 0;
 686         return;
 687 }