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 "hashmap.h"
   8 #include "progress.h"
   9
  10 /* Table of rename/copy destinations */
  11
  12 static struct diff_rename_dst {
  13         struct diff_filespec *two;
  14         struct diff_filepair *pair;
  15 } *rename_dst;
  16 static int rename_dst_nr, rename_dst_alloc;
  17
  18 static int find_rename_dst(struct diff_filespec *two)
  19 {
  20         int first, last;
  21
  22         first = 0;
  23         last = rename_dst_nr;
  24         while (last > first) {
  25                 int next = (last + first) >> 1;
  26                 struct diff_rename_dst *dst = &(rename_dst[next]);
  27                 int cmp = strcmp(two->path, dst->two->path);
  28                 if (!cmp)
  29                         return next;
  30                 if (cmp < 0) {
  31                         last = next;
  32                         continue;
  33                 }
  34                 first = next+1;
  35         }
  36         return -first - 1;
  37 }
  38
  39 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two)
  40 {
  41         int ofs = find_rename_dst(two);
  42         return ofs < 0 ? NULL : &rename_dst[ofs];
  43 }
  44
  45 /*
  46  * Returns 0 on success, -1 if we found a duplicate.
  47  */
  48 static int add_rename_dst(struct diff_filespec *two)
  49 {
  50         int first = find_rename_dst(two);
  51
  52         if (first >= 0)
  53                 return -1;
  54         first = -first - 1;
  55
  56         /* insert to make it at "first" */
  57         ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc);
  58         rename_dst_nr++;
  59         if (first < rename_dst_nr)
  60                 MOVE_ARRAY(rename_dst + first + 1, rename_dst + first,
  61                            rename_dst_nr - first - 1);
  62         rename_dst[first].two = alloc_filespec(two->path);
  63         fill_filespec(rename_dst[first].two, &two->oid, two->oid_valid,
  64                       two->mode);
  65         rename_dst[first].pair = NULL;
  66         return 0;
  67 }
  68
  69 /* Table of rename/copy src files */
  70 static struct diff_rename_src {
  71         struct diff_filepair *p;
  72         unsigned short score; /* to remember the break score */
  73 } *rename_src;
  74 static int rename_src_nr, rename_src_alloc;
  75
  76 static struct diff_rename_src *register_rename_src(struct diff_filepair *p)
  77 {
  78         int first, last;
  79         struct diff_filespec *one = p->one;
  80         unsigned short score = p->score;
  81
  82         first = 0;
  83         last = rename_src_nr;
  84         while (last > first) {
  85                 int next = (last + first) >> 1;
  86                 struct diff_rename_src *src = &(rename_src[next]);
  87                 int cmp = strcmp(one->path, src->p->one->path);
  88                 if (!cmp)
  89                         return src;
  90                 if (cmp < 0) {
  91                         last = next;
  92                         continue;
  93                 }
  94                 first = next+1;
  95         }
  96
  97         /* insert to make it at "first" */
  98         ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc);
  99         rename_src_nr++;
 100         if (first < rename_src_nr)
 101                 MOVE_ARRAY(rename_src + first + 1, rename_src + first,
 102                            rename_src_nr - first - 1);
 103         rename_src[first].p = p;
 104         rename_src[first].score = score;
 105         return &(rename_src[first]);
 106 }
 107
 108 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
 109 {
 110         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 111         while (src_len && dst_len) {
 112                 char c1 = src->path[--src_len];
 113                 char c2 = dst->path[--dst_len];
 114                 if (c1 != c2)
 115                         return 0;
 116                 if (c1 == '/')
 117                         return 1;
 118         }
 119         return (!src_len || src->path[src_len - 1] == '/') &&
 120                 (!dst_len || dst->path[dst_len - 1] == '/');
 121 }
 122
 123 struct diff_score {
 124         int src; /* index in rename_src */
 125         int dst; /* index in rename_dst */
 126         unsigned short score;
 127         short name_score;
 128 };
 129
 130 static int estimate_similarity(struct diff_filespec *src,
 131                                struct diff_filespec *dst,
 132                                int minimum_score)
 133 {
 134         /* src points at a file that existed in the original tree (or
 135          * optionally a file in the destination tree) and dst points
 136          * at a newly created file.  They may be quite similar, in which
 137          * case we want to say src is renamed to dst or src is copied into
 138          * dst, and then some edit has been applied to dst.
 139          *
 140          * Compare them and return how similar they are, representing
 141          * the score as an integer between 0 and MAX_SCORE.
 142          *
 143          * When there is an exact match, it is considered a better
 144          * match than anything else; the destination does not even
 145          * call into this function in that case.
 146          */
 147         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 148         int score;
 149
 150         /* We deal only with regular files.  Symlink renames are handled
 151          * only when they are exact matches --- in other words, no edits
 152          * after renaming.
 153          */
 154         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 155                 return 0;
 156
 157         /*
 158          * Need to check that source and destination sizes are
 159          * filled in before comparing them.
 160          *
 161          * If we already have "cnt_data" filled in, we know it's
 162          * all good (avoid checking the size for zero, as that
 163          * is a possible size - we really should have a flag to
 164          * say whether the size is valid or not!)
 165          */
 166         if (!src->cnt_data &&
 167             diff_populate_filespec(src, CHECK_SIZE_ONLY))
 168                 return 0;
 169         if (!dst->cnt_data &&
 170             diff_populate_filespec(dst, CHECK_SIZE_ONLY))
 171                 return 0;
 172
 173         max_size = ((src->size > dst->size) ? src->size : dst->size);
 174         base_size = ((src->size < dst->size) ? src->size : dst->size);
 175         delta_size = max_size - base_size;
 176
 177         /* We would not consider edits that change the file size so
 178          * drastically.  delta_size must be smaller than
 179          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 180          *
 181          * Note that base_size == 0 case is handled here already
 182          * and the final score computation below would not have a
 183          * divide-by-zero issue.
 184          */
 185         if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 186                 return 0;
 187
 188         if (!src->cnt_data && diff_populate_filespec(src, 0))
 189                 return 0;
 190         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 191                 return 0;
 192
 193         if (diffcore_count_changes(src, dst,
 194                                    &src->cnt_data, &dst->cnt_data,
 195                                    &src_copied, &literal_added))
 196                 return 0;
 197
 198         /* How similar are they?
 199          * what percentage of material in dst are from source?
 200          */
 201         if (!dst->size)
 202                 score = 0; /* should not happen */
 203         else
 204                 score = (int)(src_copied * MAX_SCORE / max_size);
 205         return score;
 206 }
 207
 208 static void record_rename_pair(int dst_index, int src_index, int score)
 209 {
 210         struct diff_filespec *src, *dst;
 211         struct diff_filepair *dp;
 212
 213         if (rename_dst[dst_index].pair)
 214                 die("internal error: dst already matched.");
 215
 216         src = rename_src[src_index].p->one;
 217         src->rename_used++;
 218         src->count++;
 219
 220         dst = rename_dst[dst_index].two;
 221         dst->count++;
 222
 223         dp = diff_queue(NULL, src, dst);
 224         dp->renamed_pair = 1;
 225         if (!strcmp(src->path, dst->path))
 226                 dp->score = rename_src[src_index].score;
 227         else
 228                 dp->score = score;
 229         rename_dst[dst_index].pair = dp;
 230 }
 231
 232 /*
 233  * We sort the rename similarity matrix with the score, in descending
 234  * order (the most similar first).
 235  */
 236 static int score_compare(const void *a_, const void *b_)
 237 {
 238         const struct diff_score *a = a_, *b = b_;
 239
 240         /* sink the unused ones to the bottom */
 241         if (a->dst < 0)
 242                 return (0 <= b->dst);
 243         else if (b->dst < 0)
 244                 return -1;
 245
 246         if (a->score == b->score)
 247                 return b->name_score - a->name_score;
 248
 249         return b->score - a->score;
 250 }
 251
 252 struct file_similarity {
 253         struct hashmap_entry entry;
 254         int index;
 255         struct diff_filespec *filespec;
 256 };
 257
 258 static unsigned int hash_filespec(struct diff_filespec *filespec)
 259 {
 260         if (!filespec->oid_valid) {
 261                 if (diff_populate_filespec(filespec, 0))
 262                         return 0;
 263                 hash_sha1_file(filespec->data, filespec->size, "blob",
 264                                filespec->oid.hash);
 265         }
 266         return sha1hash(filespec->oid.hash);
 267 }
 268
 269 static int find_identical_files(struct hashmap *srcs,
 270                                 int dst_index,
 271                                 struct diff_options *options)
 272 {
 273         int renames = 0;
 274
 275         struct diff_filespec *target = rename_dst[dst_index].two;
 276         struct file_similarity *p, *best = NULL;
 277         int i = 100, best_score = -1;
 278
 279         /*
 280          * Find the best source match for specified destination.
 281          */
 282         p = hashmap_get_from_hash(srcs, hash_filespec(target), NULL);
 283         for (; p; p = hashmap_get_next(srcs, p)) {
 284                 int score;
 285                 struct diff_filespec *source = p->filespec;
 286
 287                 /* False hash collision? */
 288                 if (oidcmp(&source->oid, &target->oid))
 289                         continue;
 290                 /* Non-regular files? If so, the modes must match! */
 291                 if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 292                         if (source->mode != target->mode)
 293                                 continue;
 294                 }
 295                 /* Give higher scores to sources that haven't been used already */
 296                 score = !source->rename_used;
 297                 if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY)
 298                         continue;
 299                 score += basename_same(source, target);
 300                 if (score > best_score) {
 301                         best = p;
 302                         best_score = score;
 303                         if (score == 2)
 304                                 break;
 305                 }
 306
 307                 /* Too many identical alternatives? Pick one */
 308                 if (!--i)
 309                         break;
 310         }
 311         if (best) {
 312                 record_rename_pair(dst_index, best->index, MAX_SCORE);
 313                 renames++;
 314         }
 315         return renames;
 316 }
 317
 318 static void insert_file_table(struct hashmap *table, int index, struct diff_filespec *filespec)
 319 {
 320         struct file_similarity *entry = xmalloc(sizeof(*entry));
 321
 322         entry->index = index;
 323         entry->filespec = filespec;
 324
 325         hashmap_entry_init(entry, hash_filespec(filespec));
 326         hashmap_add(table, entry);
 327 }
 328
 329 /*
 330  * Find exact renames first.
 331  *
 332  * The first round matches up the up-to-date entries,
 333  * and then during the second round we try to match
 334  * cache-dirty entries as well.
 335  */
 336 static int find_exact_renames(struct diff_options *options)
 337 {
 338         int i, renames = 0;
 339         struct hashmap file_table;
 340
 341         /* Add all sources to the hash table in reverse order, because
 342          * later on they will be retrieved in LIFO order.
 343          */
 344         hashmap_init(&file_table, NULL, NULL, rename_src_nr);
 345         for (i = rename_src_nr-1; i >= 0; i--)
 346                 insert_file_table(&file_table, i, rename_src[i].p->one);
 347
 348         /* Walk the destinations and find best source match */
 349         for (i = 0; i < rename_dst_nr; i++)
 350                 renames += find_identical_files(&file_table, i, options);
 351
 352         /* Free the hash data structure and entries */
 353         hashmap_free(&file_table, 1);
 354
 355         return renames;
 356 }
 357
 358 #define NUM_CANDIDATE_PER_DST 4
 359 static void record_if_better(struct diff_score m[], struct diff_score *o)
 360 {
 361         int i, worst;
 362
 363         /* find the worst one */
 364         worst = 0;
 365         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 366                 if (score_compare(&m[i], &m[worst]) > 0)
 367                         worst = i;
 368
 369         /* is it better than the worst one? */
 370         if (score_compare(&m[worst], o) > 0)
 371                 m[worst] = *o;
 372 }
 373
 374 /*
 375  * Returns:
 376  * 0 if we are under the limit;
 377  * 1 if we need to disable inexact rename detection;
 378  * 2 if we would be under the limit if we were given -C instead of -C -C.
 379  */
 380 static int too_many_rename_candidates(int num_create,
 381                                       struct diff_options *options)
 382 {
 383         int rename_limit = options->rename_limit;
 384         int num_src = rename_src_nr;
 385         int i;
 386
 387         options->needed_rename_limit = 0;
 388
 389         /*
 390          * This basically does a test for the rename matrix not
 391          * growing larger than a "rename_limit" square matrix, ie:
 392          *
 393          *    num_create * num_src > rename_limit * rename_limit
 394          */
 395         if (rename_limit <= 0)
 396                 rename_limit = 32767;
 397         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 398             ((uint64_t)num_create * (uint64_t)num_src
 399              <= (uint64_t)rename_limit * (uint64_t)rename_limit))
 400                 return 0;
 401
 402         options->needed_rename_limit =
 403                 num_src > num_create ? num_src : num_create;
 404
 405         /* Are we running under -C -C? */
 406         if (!options->flags.find_copies_harder)
 407                 return 1;
 408
 409         /* Would we bust the limit if we were running under -C? */
 410         for (num_src = i = 0; i < rename_src_nr; i++) {
 411                 if (diff_unmodified_pair(rename_src[i].p))
 412                         continue;
 413                 num_src++;
 414         }
 415         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 416             ((uint64_t)num_create * (uint64_t)num_src
 417              <= (uint64_t)rename_limit * (uint64_t)rename_limit))
 418                 return 2;
 419         return 1;
 420 }
 421
 422 static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies)
 423 {
 424         int count = 0, i;
 425
 426         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 427                 struct diff_rename_dst *dst;
 428
 429                 if ((mx[i].dst < 0) ||
 430                     (mx[i].score < minimum_score))
 431                         break; /* there is no more usable pair. */
 432                 dst = &rename_dst[mx[i].dst];
 433                 if (dst->pair)
 434                         continue; /* already done, either exact or fuzzy. */
 435                 if (!copies && rename_src[mx[i].src].p->one->rename_used)
 436                         continue;
 437                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 438                 count++;
 439         }
 440         return count;
 441 }
 442
 443 void diffcore_rename(struct diff_options *options)
 444 {
 445         int detect_rename = options->detect_rename;
 446         int minimum_score = options->rename_score;
 447         struct diff_queue_struct *q = &diff_queued_diff;
 448         struct diff_queue_struct outq;
 449         struct diff_score *mx;
 450         int i, j, rename_count, skip_unmodified = 0;
 451         int num_create, dst_cnt;
 452         struct progress *progress = NULL;
 453
 454         if (!minimum_score)
 455                 minimum_score = DEFAULT_RENAME_SCORE;
 456
 457         for (i = 0; i < q->nr; i++) {
 458                 struct diff_filepair *p = q->queue[i];
 459                 if (!DIFF_FILE_VALID(p->one)) {
 460                         if (!DIFF_FILE_VALID(p->two))
 461                                 continue; /* unmerged */
 462                         else if (options->single_follow &&
 463                                  strcmp(options->single_follow, p->two->path))
 464                                 continue; /* not interested */
 465                         else if (!options->flags.rename_empty &&
 466                                  is_empty_blob_oid(&p->two->oid))
 467                                 continue;
 468                         else if (add_rename_dst(p->two) < 0) {
 469                                 warning("skipping rename detection, detected"
 470                                         " duplicate destination '%s'",
 471                                         p->two->path);
 472                                 goto cleanup;
 473                         }
 474                 }
 475                 else if (!options->flags.rename_empty &&
 476                          is_empty_blob_oid(&p->one->oid))
 477                         continue;
 478                 else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) {
 479                         /*
 480                          * If the source is a broken "delete", and
 481                          * they did not really want to get broken,
 482                          * that means the source actually stays.
 483                          * So we increment the "rename_used" score
 484                          * by one, to indicate ourselves as a user
 485                          */
 486                         if (p->broken_pair && !p->score)
 487                                 p->one->rename_used++;
 488                         register_rename_src(p);
 489                 }
 490                 else if (detect_rename == DIFF_DETECT_COPY) {
 491                         /*
 492                          * Increment the "rename_used" score by
 493                          * one, to indicate ourselves as a user.
 494                          */
 495                         p->one->rename_used++;
 496                         register_rename_src(p);
 497                 }
 498         }
 499         if (rename_dst_nr == 0 || rename_src_nr == 0)
 500                 goto cleanup; /* nothing to do */
 501
 502         /*
 503          * We really want to cull the candidates list early
 504          * with cheap tests in order to avoid doing deltas.
 505          */
 506         rename_count = find_exact_renames(options);
 507
 508         /* Did we only want exact renames? */
 509         if (minimum_score == MAX_SCORE)
 510                 goto cleanup;
 511
 512         /*
 513          * Calculate how many renames are left (but all the source
 514          * files still remain as options for rename/copies!)
 515          */
 516         num_create = (rename_dst_nr - rename_count);
 517
 518         /* All done? */
 519         if (!num_create)
 520                 goto cleanup;
 521
 522         switch (too_many_rename_candidates(num_create, options)) {
 523         case 1:
 524                 goto cleanup;
 525         case 2:
 526                 options->degraded_cc_to_c = 1;
 527                 skip_unmodified = 1;
 528                 break;
 529         default:
 530                 break;
 531         }
 532
 533         if (options->show_rename_progress) {
 534                 progress = start_delayed_progress(
 535                                 _("Performing inexact rename detection"),
 536                                 (uint64_t)rename_dst_nr * (uint64_t)rename_src_nr);
 537         }
 538
 539         mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_create), sizeof(*mx));
 540         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 541                 struct diff_filespec *two = rename_dst[i].two;
 542                 struct diff_score *m;
 543
 544                 if (rename_dst[i].pair)
 545                         continue; /* dealt with exact match already. */
 546
 547                 m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
 548                 for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
 549                         m[j].dst = -1;
 550
 551                 for (j = 0; j < rename_src_nr; j++) {
 552                         struct diff_filespec *one = rename_src[j].p->one;
 553                         struct diff_score this_src;
 554
 555                         if (skip_unmodified &&
 556                             diff_unmodified_pair(rename_src[j].p))
 557                                 continue;
 558
 559                         this_src.score = estimate_similarity(one, two,
 560                                                              minimum_score);
 561                         this_src.name_score = basename_same(one, two);
 562                         this_src.dst = i;
 563                         this_src.src = j;
 564                         record_if_better(m, &this_src);
 565                         /*
 566                          * Once we run estimate_similarity,
 567                          * We do not need the text anymore.
 568                          */
 569                         diff_free_filespec_blob(one);
 570                         diff_free_filespec_blob(two);
 571                 }
 572                 dst_cnt++;
 573                 display_progress(progress, (uint64_t)(i+1)*(uint64_t)rename_src_nr);
 574         }
 575         stop_progress(&progress);
 576
 577         /* cost matrix sorted by most to least similar pair */
 578         QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare);
 579
 580         rename_count += find_renames(mx, dst_cnt, minimum_score, 0);
 581         if (detect_rename == DIFF_DETECT_COPY)
 582                 rename_count += find_renames(mx, dst_cnt, minimum_score, 1);
 583         free(mx);
 584
 585  cleanup:
 586         /* At this point, we have found some renames and copies and they
 587          * are recorded in rename_dst.  The original list is still in *q.
 588          */
 589         DIFF_QUEUE_CLEAR(&outq);
 590         for (i = 0; i < q->nr; i++) {
 591                 struct diff_filepair *p = q->queue[i];
 592                 struct diff_filepair *pair_to_free = NULL;
 593
 594                 if (DIFF_PAIR_UNMERGED(p)) {
 595                         diff_q(&outq, p);
 596                 }
 597                 else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 598                         /*
 599                          * Creation
 600                          *
 601                          * We would output this create record if it has
 602                          * not been turned into a rename/copy already.
 603                          */
 604                         struct diff_rename_dst *dst = locate_rename_dst(p->two);
 605                         if (dst && dst->pair) {
 606                                 diff_q(&outq, dst->pair);
 607                                 pair_to_free = p;
 608                         }
 609                         else
 610                                 /* no matching rename/copy source, so
 611                                  * record this as a creation.
 612                                  */
 613                                 diff_q(&outq, p);
 614                 }
 615                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 616                         /*
 617                          * Deletion
 618                          *
 619                          * We would output this delete record if:
 620                          *
 621                          * (1) this is a broken delete and the counterpart
 622                          *     broken create remains in the output; or
 623                          * (2) this is not a broken delete, and rename_dst
 624                          *     does not have a rename/copy to move p->one->path
 625                          *     out of existence.
 626                          *
 627                          * Otherwise, the counterpart broken create
 628                          * has been turned into a rename-edit; or
 629                          * delete did not have a matching create to
 630                          * begin with.
 631                          */
 632                         if (DIFF_PAIR_BROKEN(p)) {
 633                                 /* broken delete */
 634                                 struct diff_rename_dst *dst = locate_rename_dst(p->one);
 635                                 if (dst && dst->pair)
 636                                         /* counterpart is now rename/copy */
 637                                         pair_to_free = p;
 638                         }
 639                         else {
 640                                 if (p->one->rename_used)
 641                                         /* this path remains */
 642                                         pair_to_free = p;
 643                         }
 644
 645                         if (pair_to_free)
 646                                 ;
 647                         else
 648                                 diff_q(&outq, p);
 649                 }
 650                 else if (!diff_unmodified_pair(p))
 651                         /* all the usual ones need to be kept */
 652                         diff_q(&outq, p);
 653                 else
 654                         /* no need to keep unmodified pairs */
 655                         pair_to_free = p;
 656
 657                 if (pair_to_free)
 658                         diff_free_filepair(pair_to_free);
 659         }
 660         diff_debug_queue("done copying original", &outq);
 661
 662         free(q->queue);
 663         *q = outq;
 664         diff_debug_queue("done collapsing", q);
 665
 666         for (i = 0; i < rename_dst_nr; i++)
 667                 free_filespec(rename_dst[i].two);
 668
 669         FREE_AND_NULL(rename_dst);
 670         rename_dst_nr = rename_dst_alloc = 0;
 671         FREE_AND_NULL(rename_src);
 672         rename_src_nr = rename_src_alloc = 0;
 673         return;
 674 }