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 "hash.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 struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
  19                                                  int insert_ok)
  20 {
  21         int first, last;
  22
  23         first = 0;
  24         last = rename_dst_nr;
  25         while (last > first) {
  26                 int next = (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 dst;
  31                 if (cmp < 0) {
  32                         last = next;
  33                         continue;
  34                 }
  35                 first = next+1;
  36         }
  37         /* not found */
  38         if (!insert_ok)
  39                 return NULL;
  40         /* insert to make it at "first" */
  41         if (rename_dst_alloc <= rename_dst_nr) {
  42                 rename_dst_alloc = alloc_nr(rename_dst_alloc);
  43                 rename_dst = xrealloc(rename_dst,
  44                                       rename_dst_alloc * sizeof(*rename_dst));
  45         }
  46         rename_dst_nr++;
  47         if (first < rename_dst_nr)
  48                 memmove(rename_dst + first + 1, rename_dst + first,
  49                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  50         rename_dst[first].two = alloc_filespec(two->path);
  51         fill_filespec(rename_dst[first].two, two->sha1, two->sha1_valid, two->mode);
  52         rename_dst[first].pair = NULL;
  53         return &(rename_dst[first]);
  54 }
  55
  56 /* Table of rename/copy src files */
  57 static struct diff_rename_src {
  58         struct diff_filepair *p;
  59         unsigned short score; /* to remember the break score */
  60 } *rename_src;
  61 static int rename_src_nr, rename_src_alloc;
  62
  63 static struct diff_rename_src *register_rename_src(struct diff_filepair *p)
  64 {
  65         int first, last;
  66         struct diff_filespec *one = p->one;
  67         unsigned short score = p->score;
  68
  69         first = 0;
  70         last = rename_src_nr;
  71         while (last > first) {
  72                 int next = (last + first) >> 1;
  73                 struct diff_rename_src *src = &(rename_src[next]);
  74                 int cmp = strcmp(one->path, src->p->one->path);
  75                 if (!cmp)
  76                         return src;
  77                 if (cmp < 0) {
  78                         last = next;
  79                         continue;
  80                 }
  81                 first = next+1;
  82         }
  83
  84         /* insert to make it at "first" */
  85         if (rename_src_alloc <= rename_src_nr) {
  86                 rename_src_alloc = alloc_nr(rename_src_alloc);
  87                 rename_src = xrealloc(rename_src,
  88                                       rename_src_alloc * sizeof(*rename_src));
  89         }
  90         rename_src_nr++;
  91         if (first < rename_src_nr)
  92                 memmove(rename_src + first + 1, rename_src + first,
  93                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  94         rename_src[first].p = p;
  95         rename_src[first].score = score;
  96         return &(rename_src[first]);
  97 }
  98
  99 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
 100 {
 101         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 102         while (src_len && dst_len) {
 103                 char c1 = src->path[--src_len];
 104                 char c2 = dst->path[--dst_len];
 105                 if (c1 != c2)
 106                         return 0;
 107                 if (c1 == '/')
 108                         return 1;
 109         }
 110         return (!src_len || src->path[src_len - 1] == '/') &&
 111                 (!dst_len || dst->path[dst_len - 1] == '/');
 112 }
 113
 114 struct diff_score {
 115         int src; /* index in rename_src */
 116         int dst; /* index in rename_dst */
 117         unsigned short score;
 118         short name_score;
 119 };
 120
 121 static int estimate_similarity(struct diff_filespec *src,
 122                                struct diff_filespec *dst,
 123                                int minimum_score)
 124 {
 125         /* src points at a file that existed in the original tree (or
 126          * optionally a file in the destination tree) and dst points
 127          * at a newly created file.  They may be quite similar, in which
 128          * case we want to say src is renamed to dst or src is copied into
 129          * dst, and then some edit has been applied to dst.
 130          *
 131          * Compare them and return how similar they are, representing
 132          * the score as an integer between 0 and MAX_SCORE.
 133          *
 134          * When there is an exact match, it is considered a better
 135          * match than anything else; the destination does not even
 136          * call into this function in that case.
 137          */
 138         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 139         unsigned long delta_limit;
 140         int score;
 141
 142         /* We deal only with regular files.  Symlink renames are handled
 143          * only when they are exact matches --- in other words, no edits
 144          * after renaming.
 145          */
 146         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 147                 return 0;
 148
 149         /*
 150          * Need to check that source and destination sizes are
 151          * filled in before comparing them.
 152          *
 153          * If we already have "cnt_data" filled in, we know it's
 154          * all good (avoid checking the size for zero, as that
 155          * is a possible size - we really should have a flag to
 156          * say whether the size is valid or not!)
 157          */
 158         if (!src->cnt_data && diff_populate_filespec(src, 1))
 159                 return 0;
 160         if (!dst->cnt_data && diff_populate_filespec(dst, 1))
 161                 return 0;
 162
 163         max_size = ((src->size > dst->size) ? src->size : dst->size);
 164         base_size = ((src->size < dst->size) ? src->size : dst->size);
 165         delta_size = max_size - base_size;
 166
 167         /* We would not consider edits that change the file size so
 168          * drastically.  delta_size must be smaller than
 169          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 170          *
 171          * Note that base_size == 0 case is handled here already
 172          * and the final score computation below would not have a
 173          * divide-by-zero issue.
 174          */
 175         if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 176                 return 0;
 177
 178         if (!src->cnt_data && diff_populate_filespec(src, 0))
 179                 return 0;
 180         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 181                 return 0;
 182
 183         delta_limit = (unsigned long)
 184                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 185         if (diffcore_count_changes(src, dst,
 186                                    &src->cnt_data, &dst->cnt_data,
 187                                    delta_limit,
 188                                    &src_copied, &literal_added))
 189                 return 0;
 190
 191         /* How similar are they?
 192          * what percentage of material in dst are from source?
 193          */
 194         if (!dst->size)
 195                 score = 0; /* should not happen */
 196         else
 197                 score = (int)(src_copied * MAX_SCORE / max_size);
 198         return score;
 199 }
 200
 201 static void record_rename_pair(int dst_index, int src_index, int score)
 202 {
 203         struct diff_filespec *src, *dst;
 204         struct diff_filepair *dp;
 205
 206         if (rename_dst[dst_index].pair)
 207                 die("internal error: dst already matched.");
 208
 209         src = rename_src[src_index].p->one;
 210         src->rename_used++;
 211         src->count++;
 212
 213         dst = rename_dst[dst_index].two;
 214         dst->count++;
 215
 216         dp = diff_queue(NULL, src, dst);
 217         dp->renamed_pair = 1;
 218         if (!strcmp(src->path, dst->path))
 219                 dp->score = rename_src[src_index].score;
 220         else
 221                 dp->score = score;
 222         rename_dst[dst_index].pair = dp;
 223 }
 224
 225 /*
 226  * We sort the rename similarity matrix with the score, in descending
 227  * order (the most similar first).
 228  */
 229 static int score_compare(const void *a_, const void *b_)
 230 {
 231         const struct diff_score *a = a_, *b = b_;
 232
 233         /* sink the unused ones to the bottom */
 234         if (a->dst < 0)
 235                 return (0 <= b->dst);
 236         else if (b->dst < 0)
 237                 return -1;
 238
 239         if (a->score == b->score)
 240                 return b->name_score - a->name_score;
 241
 242         return b->score - a->score;
 243 }
 244
 245 struct file_similarity {
 246         int index;
 247         struct diff_filespec *filespec;
 248         struct file_similarity *next;
 249 };
 250
 251 static unsigned int hash_filespec(struct diff_filespec *filespec)
 252 {
 253         unsigned int hash;
 254         if (!filespec->sha1_valid) {
 255                 if (diff_populate_filespec(filespec, 0))
 256                         return 0;
 257                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 258         }
 259         memcpy(&hash, filespec->sha1, sizeof(hash));
 260         return hash;
 261 }
 262
 263 static int find_identical_files(struct hash_table *srcs,
 264                                 int dst_index,
 265                                 struct diff_options *options)
 266 {
 267         int renames = 0;
 268
 269         struct diff_filespec *target = rename_dst[dst_index].two;
 270         struct file_similarity *p, *best;
 271         int i = 100, best_score = -1;
 272
 273         /*
 274          * Find the best source match for specified destination.
 275          */
 276         best = NULL;
 277         for (p = lookup_hash(hash_filespec(target), srcs); p; p = p->next) {
 278                 int score;
 279                 struct diff_filespec *source = p->filespec;
 280
 281                 /* False hash collision? */
 282                 if (hashcmp(source->sha1, target->sha1))
 283                         continue;
 284                 /* Non-regular files? If so, the modes must match! */
 285                 if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 286                         if (source->mode != target->mode)
 287                                 continue;
 288                 }
 289                 /* Give higher scores to sources that haven't been used already */
 290                 score = !source->rename_used;
 291                 if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY)
 292                         continue;
 293                 score += basename_same(source, target);
 294                 if (score > best_score) {
 295                         best = p;
 296                         best_score = score;
 297                         if (score == 2)
 298                                 break;
 299                 }
 300
 301                 /* Too many identical alternatives? Pick one */
 302                 if (!--i)
 303                         break;
 304         }
 305         if (best) {
 306                 record_rename_pair(dst_index, best->index, MAX_SCORE);
 307                 renames++;
 308         }
 309         return renames;
 310 }
 311
 312 static int free_similarity_list(void *p, void *unused)
 313 {
 314         while (p) {
 315                 struct file_similarity *entry = p;
 316                 p = entry->next;
 317                 free(entry);
 318         }
 319         return 0;
 320 }
 321
 322 static void insert_file_table(struct hash_table *table, int index, struct diff_filespec *filespec)
 323 {
 324         void **pos;
 325         unsigned int hash;
 326         struct file_similarity *entry = xmalloc(sizeof(*entry));
 327
 328         entry->index = index;
 329         entry->filespec = filespec;
 330         entry->next = NULL;
 331
 332         hash = hash_filespec(filespec);
 333         pos = insert_hash(hash, entry, table);
 334
 335         /* We already had an entry there? */
 336         if (pos) {
 337                 entry->next = *pos;
 338                 *pos = entry;
 339         }
 340 }
 341
 342 /*
 343  * Find exact renames first.
 344  *
 345  * The first round matches up the up-to-date entries,
 346  * and then during the second round we try to match
 347  * cache-dirty entries as well.
 348  */
 349 static int find_exact_renames(struct diff_options *options)
 350 {
 351         int i, renames = 0;
 352         struct hash_table file_table;
 353
 354         /* Add all sources to the hash table */
 355         init_hash(&file_table);
 356         preallocate_hash(&file_table, rename_src_nr);
 357         for (i = 0; i < rename_src_nr; i++)
 358                 insert_file_table(&file_table, i, 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 source file_similarity chains */
 365         for_each_hash(&file_table, free_similarity_list, options);
 366
 367         /* .. and free the hash data structure */
 368         free_hash(&file_table);
 369
 370         return renames;
 371 }
 372
 373 #define NUM_CANDIDATE_PER_DST 4
 374 static void record_if_better(struct diff_score m[], struct diff_score *o)
 375 {
 376         int i, worst;
 377
 378         /* find the worst one */
 379         worst = 0;
 380         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 381                 if (score_compare(&m[i], &m[worst]) > 0)
 382                         worst = i;
 383
 384         /* is it better than the worst one? */
 385         if (score_compare(&m[worst], o) > 0)
 386                 m[worst] = *o;
 387 }
 388
 389 /*
 390  * Returns:
 391  * 0 if we are under the limit;
 392  * 1 if we need to disable inexact rename detection;
 393  * 2 if we would be under the limit if we were given -C instead of -C -C.
 394  */
 395 static int too_many_rename_candidates(int num_create,
 396                                       struct diff_options *options)
 397 {
 398         int rename_limit = options->rename_limit;
 399         int num_src = rename_src_nr;
 400         int i;
 401
 402         options->needed_rename_limit = 0;
 403
 404         /*
 405          * This basically does a test for the rename matrix not
 406          * growing larger than a "rename_limit" square matrix, ie:
 407          *
 408          *    num_create * num_src > rename_limit * rename_limit
 409          *
 410          * but handles the potential overflow case specially (and we
 411          * assume at least 32-bit integers)
 412          */
 413         if (rename_limit <= 0 || rename_limit > 32767)
 414                 rename_limit = 32767;
 415         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 416             (num_create * num_src <= rename_limit * rename_limit))
 417                 return 0;
 418
 419         options->needed_rename_limit =
 420                 num_src > num_create ? num_src : num_create;
 421
 422         /* Are we running under -C -C? */
 423         if (!DIFF_OPT_TST(options, FIND_COPIES_HARDER))
 424                 return 1;
 425
 426         /* Would we bust the limit if we were running under -C? */
 427         for (num_src = i = 0; i < rename_src_nr; i++) {
 428                 if (diff_unmodified_pair(rename_src[i].p))
 429                         continue;
 430                 num_src++;
 431         }
 432         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 433             (num_create * num_src <= rename_limit * rename_limit))
 434                 return 2;
 435         return 1;
 436 }
 437
 438 static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies)
 439 {
 440         int count = 0, i;
 441
 442         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 443                 struct diff_rename_dst *dst;
 444
 445                 if ((mx[i].dst < 0) ||
 446                     (mx[i].score < minimum_score))
 447                         break; /* there is no more usable pair. */
 448                 dst = &rename_dst[mx[i].dst];
 449                 if (dst->pair)
 450                         continue; /* already done, either exact or fuzzy. */
 451                 if (!copies && rename_src[mx[i].src].p->one->rename_used)
 452                         continue;
 453                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 454                 count++;
 455         }
 456         return count;
 457 }
 458
 459 void diffcore_rename(struct diff_options *options)
 460 {
 461         int detect_rename = options->detect_rename;
 462         int minimum_score = options->rename_score;
 463         struct diff_queue_struct *q = &diff_queued_diff;
 464         struct diff_queue_struct outq;
 465         struct diff_score *mx;
 466         int i, j, rename_count, skip_unmodified = 0;
 467         int num_create, dst_cnt;
 468         struct progress *progress = NULL;
 469
 470         if (!minimum_score)
 471                 minimum_score = DEFAULT_RENAME_SCORE;
 472
 473         for (i = 0; i < q->nr; i++) {
 474                 struct diff_filepair *p = q->queue[i];
 475                 if (!DIFF_FILE_VALID(p->one)) {
 476                         if (!DIFF_FILE_VALID(p->two))
 477                                 continue; /* unmerged */
 478                         else if (options->single_follow &&
 479                                  strcmp(options->single_follow, p->two->path))
 480                                 continue; /* not interested */
 481                         else if (!DIFF_OPT_TST(options, RENAME_EMPTY) &&
 482                                  is_empty_blob_sha1(p->two->sha1))
 483                                 continue;
 484                         else
 485                                 locate_rename_dst(p->two, 1);
 486                 }
 487                 else if (!DIFF_OPT_TST(options, RENAME_EMPTY) &&
 488                          is_empty_blob_sha1(p->one->sha1))
 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_progress_delay(
 547                                 "Performing inexact rename detection",
 548                                 rename_dst_nr * rename_src_nr, 50, 1);
 549         }
 550
 551         mx = xcalloc(num_create * NUM_CANDIDATE_PER_DST, 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(one, two,
 572                                                              minimum_score);
 573                         this_src.name_score = basename_same(one, two);
 574                         this_src.dst = i;
 575                         this_src.src = j;
 576                         record_if_better(m, &this_src);
 577                         /*
 578                          * Once we run estimate_similarity,
 579                          * We do not need the text anymore.
 580                          */
 581                         diff_free_filespec_blob(one);
 582                         diff_free_filespec_blob(two);
 583                 }
 584                 dst_cnt++;
 585                 display_progress(progress, (i+1)*rename_src_nr);
 586         }
 587         stop_progress(&progress);
 588
 589         /* cost matrix sorted by most to least similar pair */
 590         qsort(mx, dst_cnt * NUM_CANDIDATE_PER_DST, sizeof(*mx), score_compare);
 591
 592         rename_count += find_renames(mx, dst_cnt, minimum_score, 0);
 593         if (detect_rename == DIFF_DETECT_COPY)
 594                 rename_count += find_renames(mx, dst_cnt, minimum_score, 1);
 595         free(mx);
 596
 597  cleanup:
 598         /* At this point, we have found some renames and copies and they
 599          * are recorded in rename_dst.  The original list is still in *q.
 600          */
 601         DIFF_QUEUE_CLEAR(&outq);
 602         for (i = 0; i < q->nr; i++) {
 603                 struct diff_filepair *p = q->queue[i];
 604                 struct diff_filepair *pair_to_free = NULL;
 605
 606                 if (DIFF_PAIR_UNMERGED(p)) {
 607                         diff_q(&outq, p);
 608                 }
 609                 else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 610                         /*
 611                          * Creation
 612                          *
 613                          * We would output this create record if it has
 614                          * not been turned into a rename/copy already.
 615                          */
 616                         struct diff_rename_dst *dst =
 617                                 locate_rename_dst(p->two, 0);
 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 =
 648                                         locate_rename_dst(p->one, 0);
 649                                 if (dst && dst->pair)
 650                                         /* counterpart is now rename/copy */
 651                                         pair_to_free = p;
 652                         }
 653                         else {
 654                                 if (p->one->rename_used)
 655                                         /* this path remains */
 656                                         pair_to_free = p;
 657                         }
 658
 659                         if (pair_to_free)
 660                                 ;
 661                         else
 662                                 diff_q(&outq, p);
 663                 }
 664                 else if (!diff_unmodified_pair(p))
 665                         /* all the usual ones need to be kept */
 666                         diff_q(&outq, p);
 667                 else
 668                         /* no need to keep unmodified pairs */
 669                         pair_to_free = p;
 670
 671                 if (pair_to_free)
 672                         diff_free_filepair(pair_to_free);
 673         }
 674         diff_debug_queue("done copying original", &outq);
 675
 676         free(q->queue);
 677         *q = outq;
 678         diff_debug_queue("done collapsing", q);
 679
 680         for (i = 0; i < rename_dst_nr; i++)
 681                 free_filespec(rename_dst[i].two);
 682
 683         free(rename_dst);
 684         rename_dst = NULL;
 685         rename_dst_nr = rename_dst_alloc = 0;
 686         free(rename_src);
 687         rename_src = NULL;
 688         rename_src_nr = rename_src_alloc = 0;
 689         return;
 690 }