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 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         ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc);
  42         rename_dst_nr++;
  43         if (first < rename_dst_nr)
  44                 memmove(rename_dst + first + 1, rename_dst + first,
  45                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  46         rename_dst[first].two = alloc_filespec(two->path);
  47         fill_filespec(rename_dst[first].two, two->sha1, two->sha1_valid, two->mode);
  48         rename_dst[first].pair = NULL;
  49         return &(rename_dst[first]);
  50 }
  51
  52 /* Table of rename/copy src files */
  53 static struct diff_rename_src {
  54         struct diff_filepair *p;
  55         unsigned short score; /* to remember the break score */
  56 } *rename_src;
  57 static int rename_src_nr, rename_src_alloc;
  58
  59 static struct diff_rename_src *register_rename_src(struct diff_filepair *p)
  60 {
  61         int first, last;
  62         struct diff_filespec *one = p->one;
  63         unsigned short score = p->score;
  64
  65         first = 0;
  66         last = rename_src_nr;
  67         while (last > first) {
  68                 int next = (last + first) >> 1;
  69                 struct diff_rename_src *src = &(rename_src[next]);
  70                 int cmp = strcmp(one->path, src->p->one->path);
  71                 if (!cmp)
  72                         return src;
  73                 if (cmp < 0) {
  74                         last = next;
  75                         continue;
  76                 }
  77                 first = next+1;
  78         }
  79
  80         /* insert to make it at "first" */
  81         ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc);
  82         rename_src_nr++;
  83         if (first < rename_src_nr)
  84                 memmove(rename_src + first + 1, rename_src + first,
  85                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  86         rename_src[first].p = p;
  87         rename_src[first].score = score;
  88         return &(rename_src[first]);
  89 }
  90
  91 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
  92 {
  93         int src_len = strlen(src->path), dst_len = strlen(dst->path);
  94         while (src_len && dst_len) {
  95                 char c1 = src->path[--src_len];
  96                 char c2 = dst->path[--dst_len];
  97                 if (c1 != c2)
  98                         return 0;
  99                 if (c1 == '/')
 100                         return 1;
 101         }
 102         return (!src_len || src->path[src_len - 1] == '/') &&
 103                 (!dst_len || dst->path[dst_len - 1] == '/');
 104 }
 105
 106 struct diff_score {
 107         int src; /* index in rename_src */
 108         int dst; /* index in rename_dst */
 109         unsigned short score;
 110         short name_score;
 111 };
 112
 113 static int estimate_similarity(struct diff_filespec *src,
 114                                struct diff_filespec *dst,
 115                                int minimum_score)
 116 {
 117         /* src points at a file that existed in the original tree (or
 118          * optionally a file in the destination tree) and dst points
 119          * at a newly created file.  They may be quite similar, in which
 120          * case we want to say src is renamed to dst or src is copied into
 121          * dst, and then some edit has been applied to dst.
 122          *
 123          * Compare them and return how similar they are, representing
 124          * the score as an integer between 0 and MAX_SCORE.
 125          *
 126          * When there is an exact match, it is considered a better
 127          * match than anything else; the destination does not even
 128          * call into this function in that case.
 129          */
 130         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 131         unsigned long delta_limit;
 132         int score;
 133
 134         /* We deal only with regular files.  Symlink renames are handled
 135          * only when they are exact matches --- in other words, no edits
 136          * after renaming.
 137          */
 138         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 139                 return 0;
 140
 141         /*
 142          * Need to check that source and destination sizes are
 143          * filled in before comparing them.
 144          *
 145          * If we already have "cnt_data" filled in, we know it's
 146          * all good (avoid checking the size for zero, as that
 147          * is a possible size - we really should have a flag to
 148          * say whether the size is valid or not!)
 149          */
 150         if (!src->cnt_data &&
 151             diff_populate_filespec(src, CHECK_SIZE_ONLY))
 152                 return 0;
 153         if (!dst->cnt_data &&
 154             diff_populate_filespec(dst, CHECK_SIZE_ONLY))
 155                 return 0;
 156
 157         max_size = ((src->size > dst->size) ? src->size : dst->size);
 158         base_size = ((src->size < dst->size) ? src->size : dst->size);
 159         delta_size = max_size - base_size;
 160
 161         /* We would not consider edits that change the file size so
 162          * drastically.  delta_size must be smaller than
 163          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 164          *
 165          * Note that base_size == 0 case is handled here already
 166          * and the final score computation below would not have a
 167          * divide-by-zero issue.
 168          */
 169         if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 170                 return 0;
 171
 172         if (!src->cnt_data && diff_populate_filespec(src, 0))
 173                 return 0;
 174         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 175                 return 0;
 176
 177         delta_limit = (unsigned long)
 178                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 179         if (diffcore_count_changes(src, dst,
 180                                    &src->cnt_data, &dst->cnt_data,
 181                                    delta_limit,
 182                                    &src_copied, &literal_added))
 183                 return 0;
 184
 185         /* How similar are they?
 186          * what percentage of material in dst are from source?
 187          */
 188         if (!dst->size)
 189                 score = 0; /* should not happen */
 190         else
 191                 score = (int)(src_copied * MAX_SCORE / max_size);
 192         return score;
 193 }
 194
 195 static void record_rename_pair(int dst_index, int src_index, int score)
 196 {
 197         struct diff_filespec *src, *dst;
 198         struct diff_filepair *dp;
 199
 200         if (rename_dst[dst_index].pair)
 201                 die("internal error: dst already matched.");
 202
 203         src = rename_src[src_index].p->one;
 204         src->rename_used++;
 205         src->count++;
 206
 207         dst = rename_dst[dst_index].two;
 208         dst->count++;
 209
 210         dp = diff_queue(NULL, src, dst);
 211         dp->renamed_pair = 1;
 212         if (!strcmp(src->path, dst->path))
 213                 dp->score = rename_src[src_index].score;
 214         else
 215                 dp->score = score;
 216         rename_dst[dst_index].pair = dp;
 217 }
 218
 219 /*
 220  * We sort the rename similarity matrix with the score, in descending
 221  * order (the most similar first).
 222  */
 223 static int score_compare(const void *a_, const void *b_)
 224 {
 225         const struct diff_score *a = a_, *b = b_;
 226
 227         /* sink the unused ones to the bottom */
 228         if (a->dst < 0)
 229                 return (0 <= b->dst);
 230         else if (b->dst < 0)
 231                 return -1;
 232
 233         if (a->score == b->score)
 234                 return b->name_score - a->name_score;
 235
 236         return b->score - a->score;
 237 }
 238
 239 struct file_similarity {
 240         struct hashmap_entry entry;
 241         int index;
 242         struct diff_filespec *filespec;
 243 };
 244
 245 static unsigned int hash_filespec(struct diff_filespec *filespec)
 246 {
 247         if (!filespec->sha1_valid) {
 248                 if (diff_populate_filespec(filespec, 0))
 249                         return 0;
 250                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 251         }
 252         return sha1hash(filespec->sha1);
 253 }
 254
 255 static int find_identical_files(struct hashmap *srcs,
 256                                 int dst_index,
 257                                 struct diff_options *options)
 258 {
 259         int renames = 0;
 260
 261         struct diff_filespec *target = rename_dst[dst_index].two;
 262         struct file_similarity *p, *best = NULL;
 263         int i = 100, best_score = -1;
 264
 265         /*
 266          * Find the best source match for specified destination.
 267          */
 268         p = hashmap_get_from_hash(srcs, hash_filespec(target), NULL);
 269         for (; p; p = hashmap_get_next(srcs, p)) {
 270                 int score;
 271                 struct diff_filespec *source = p->filespec;
 272
 273                 /* False hash collision? */
 274                 if (hashcmp(source->sha1, target->sha1))
 275                         continue;
 276                 /* Non-regular files? If so, the modes must match! */
 277                 if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 278                         if (source->mode != target->mode)
 279                                 continue;
 280                 }
 281                 /* Give higher scores to sources that haven't been used already */
 282                 score = !source->rename_used;
 283                 if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY)
 284                         continue;
 285                 score += basename_same(source, target);
 286                 if (score > best_score) {
 287                         best = p;
 288                         best_score = score;
 289                         if (score == 2)
 290                                 break;
 291                 }
 292
 293                 /* Too many identical alternatives? Pick one */
 294                 if (!--i)
 295                         break;
 296         }
 297         if (best) {
 298                 record_rename_pair(dst_index, best->index, MAX_SCORE);
 299                 renames++;
 300         }
 301         return renames;
 302 }
 303
 304 static void insert_file_table(struct hashmap *table, int index, struct diff_filespec *filespec)
 305 {
 306         struct file_similarity *entry = xmalloc(sizeof(*entry));
 307
 308         entry->index = index;
 309         entry->filespec = filespec;
 310
 311         hashmap_entry_init(entry, hash_filespec(filespec));
 312         hashmap_add(table, entry);
 313 }
 314
 315 /*
 316  * Find exact renames first.
 317  *
 318  * The first round matches up the up-to-date entries,
 319  * and then during the second round we try to match
 320  * cache-dirty entries as well.
 321  */
 322 static int find_exact_renames(struct diff_options *options)
 323 {
 324         int i, renames = 0;
 325         struct hashmap file_table;
 326
 327         /* Add all sources to the hash table */
 328         hashmap_init(&file_table, NULL, rename_src_nr);
 329         for (i = 0; i < rename_src_nr; i++)
 330                 insert_file_table(&file_table, i, rename_src[i].p->one);
 331
 332         /* Walk the destinations and find best source match */
 333         for (i = 0; i < rename_dst_nr; i++)
 334                 renames += find_identical_files(&file_table, i, options);
 335
 336         /* Free the hash data structure and entries */
 337         hashmap_free(&file_table, 1);
 338
 339         return renames;
 340 }
 341
 342 #define NUM_CANDIDATE_PER_DST 4
 343 static void record_if_better(struct diff_score m[], struct diff_score *o)
 344 {
 345         int i, worst;
 346
 347         /* find the worst one */
 348         worst = 0;
 349         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 350                 if (score_compare(&m[i], &m[worst]) > 0)
 351                         worst = i;
 352
 353         /* is it better than the worst one? */
 354         if (score_compare(&m[worst], o) > 0)
 355                 m[worst] = *o;
 356 }
 357
 358 /*
 359  * Returns:
 360  * 0 if we are under the limit;
 361  * 1 if we need to disable inexact rename detection;
 362  * 2 if we would be under the limit if we were given -C instead of -C -C.
 363  */
 364 static int too_many_rename_candidates(int num_create,
 365                                       struct diff_options *options)
 366 {
 367         int rename_limit = options->rename_limit;
 368         int num_src = rename_src_nr;
 369         int i;
 370
 371         options->needed_rename_limit = 0;
 372
 373         /*
 374          * This basically does a test for the rename matrix not
 375          * growing larger than a "rename_limit" square matrix, ie:
 376          *
 377          *    num_create * num_src > rename_limit * rename_limit
 378          *
 379          * but handles the potential overflow case specially (and we
 380          * assume at least 32-bit integers)
 381          */
 382         if (rename_limit <= 0 || rename_limit > 32767)
 383                 rename_limit = 32767;
 384         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 385             (num_create * num_src <= rename_limit * rename_limit))
 386                 return 0;
 387
 388         options->needed_rename_limit =
 389                 num_src > num_create ? num_src : num_create;
 390
 391         /* Are we running under -C -C? */
 392         if (!DIFF_OPT_TST(options, FIND_COPIES_HARDER))
 393                 return 1;
 394
 395         /* Would we bust the limit if we were running under -C? */
 396         for (num_src = i = 0; i < rename_src_nr; i++) {
 397                 if (diff_unmodified_pair(rename_src[i].p))
 398                         continue;
 399                 num_src++;
 400         }
 401         if ((num_create <= rename_limit || num_src <= rename_limit) &&
 402             (num_create * num_src <= rename_limit * rename_limit))
 403                 return 2;
 404         return 1;
 405 }
 406
 407 static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies)
 408 {
 409         int count = 0, i;
 410
 411         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 412                 struct diff_rename_dst *dst;
 413
 414                 if ((mx[i].dst < 0) ||
 415                     (mx[i].score < minimum_score))
 416                         break; /* there is no more usable pair. */
 417                 dst = &rename_dst[mx[i].dst];
 418                 if (dst->pair)
 419                         continue; /* already done, either exact or fuzzy. */
 420                 if (!copies && rename_src[mx[i].src].p->one->rename_used)
 421                         continue;
 422                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 423                 count++;
 424         }
 425         return count;
 426 }
 427
 428 void diffcore_rename(struct diff_options *options)
 429 {
 430         int detect_rename = options->detect_rename;
 431         int minimum_score = options->rename_score;
 432         struct diff_queue_struct *q = &diff_queued_diff;
 433         struct diff_queue_struct outq;
 434         struct diff_score *mx;
 435         int i, j, rename_count, skip_unmodified = 0;
 436         int num_create, dst_cnt;
 437         struct progress *progress = NULL;
 438
 439         if (!minimum_score)
 440                 minimum_score = DEFAULT_RENAME_SCORE;
 441
 442         for (i = 0; i < q->nr; i++) {
 443                 struct diff_filepair *p = q->queue[i];
 444                 if (!DIFF_FILE_VALID(p->one)) {
 445                         if (!DIFF_FILE_VALID(p->two))
 446                                 continue; /* unmerged */
 447                         else if (options->single_follow &&
 448                                  strcmp(options->single_follow, p->two->path))
 449                                 continue; /* not interested */
 450                         else if (!DIFF_OPT_TST(options, RENAME_EMPTY) &&
 451                                  is_empty_blob_sha1(p->two->sha1))
 452                                 continue;
 453                         else
 454                                 locate_rename_dst(p->two, 1);
 455                 }
 456                 else if (!DIFF_OPT_TST(options, RENAME_EMPTY) &&
 457                          is_empty_blob_sha1(p->one->sha1))
 458                         continue;
 459                 else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) {
 460                         /*
 461                          * If the source is a broken "delete", and
 462                          * they did not really want to get broken,
 463                          * that means the source actually stays.
 464                          * So we increment the "rename_used" score
 465                          * by one, to indicate ourselves as a user
 466                          */
 467                         if (p->broken_pair && !p->score)
 468                                 p->one->rename_used++;
 469                         register_rename_src(p);
 470                 }
 471                 else if (detect_rename == DIFF_DETECT_COPY) {
 472                         /*
 473                          * Increment the "rename_used" score by
 474                          * one, to indicate ourselves as a user.
 475                          */
 476                         p->one->rename_used++;
 477                         register_rename_src(p);
 478                 }
 479         }
 480         if (rename_dst_nr == 0 || rename_src_nr == 0)
 481                 goto cleanup; /* nothing to do */
 482
 483         /*
 484          * We really want to cull the candidates list early
 485          * with cheap tests in order to avoid doing deltas.
 486          */
 487         rename_count = find_exact_renames(options);
 488
 489         /* Did we only want exact renames? */
 490         if (minimum_score == MAX_SCORE)
 491                 goto cleanup;
 492
 493         /*
 494          * Calculate how many renames are left (but all the source
 495          * files still remain as options for rename/copies!)
 496          */
 497         num_create = (rename_dst_nr - rename_count);
 498
 499         /* All done? */
 500         if (!num_create)
 501                 goto cleanup;
 502
 503         switch (too_many_rename_candidates(num_create, options)) {
 504         case 1:
 505                 goto cleanup;
 506         case 2:
 507                 options->degraded_cc_to_c = 1;
 508                 skip_unmodified = 1;
 509                 break;
 510         default:
 511                 break;
 512         }
 513
 514         if (options->show_rename_progress) {
 515                 progress = start_progress_delay(
 516                                 _("Performing inexact rename detection"),
 517                                 rename_dst_nr * rename_src_nr, 50, 1);
 518         }
 519
 520         mx = xcalloc(num_create * NUM_CANDIDATE_PER_DST, sizeof(*mx));
 521         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 522                 struct diff_filespec *two = rename_dst[i].two;
 523                 struct diff_score *m;
 524
 525                 if (rename_dst[i].pair)
 526                         continue; /* dealt with exact match already. */
 527
 528                 m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
 529                 for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
 530                         m[j].dst = -1;
 531
 532                 for (j = 0; j < rename_src_nr; j++) {
 533                         struct diff_filespec *one = rename_src[j].p->one;
 534                         struct diff_score this_src;
 535
 536                         if (skip_unmodified &&
 537                             diff_unmodified_pair(rename_src[j].p))
 538                                 continue;
 539
 540                         this_src.score = estimate_similarity(one, two,
 541                                                              minimum_score);
 542                         this_src.name_score = basename_same(one, two);
 543                         this_src.dst = i;
 544                         this_src.src = j;
 545                         record_if_better(m, &this_src);
 546                         /*
 547                          * Once we run estimate_similarity,
 548                          * We do not need the text anymore.
 549                          */
 550                         diff_free_filespec_blob(one);
 551                         diff_free_filespec_blob(two);
 552                 }
 553                 dst_cnt++;
 554                 display_progress(progress, (i+1)*rename_src_nr);
 555         }
 556         stop_progress(&progress);
 557
 558         /* cost matrix sorted by most to least similar pair */
 559         qsort(mx, dst_cnt * NUM_CANDIDATE_PER_DST, sizeof(*mx), score_compare);
 560
 561         rename_count += find_renames(mx, dst_cnt, minimum_score, 0);
 562         if (detect_rename == DIFF_DETECT_COPY)
 563                 rename_count += find_renames(mx, dst_cnt, minimum_score, 1);
 564         free(mx);
 565
 566  cleanup:
 567         /* At this point, we have found some renames and copies and they
 568          * are recorded in rename_dst.  The original list is still in *q.
 569          */
 570         DIFF_QUEUE_CLEAR(&outq);
 571         for (i = 0; i < q->nr; i++) {
 572                 struct diff_filepair *p = q->queue[i];
 573                 struct diff_filepair *pair_to_free = NULL;
 574
 575                 if (DIFF_PAIR_UNMERGED(p)) {
 576                         diff_q(&outq, p);
 577                 }
 578                 else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 579                         /*
 580                          * Creation
 581                          *
 582                          * We would output this create record if it has
 583                          * not been turned into a rename/copy already.
 584                          */
 585                         struct diff_rename_dst *dst =
 586                                 locate_rename_dst(p->two, 0);
 587                         if (dst && dst->pair) {
 588                                 diff_q(&outq, dst->pair);
 589                                 pair_to_free = p;
 590                         }
 591                         else
 592                                 /* no matching rename/copy source, so
 593                                  * record this as a creation.
 594                                  */
 595                                 diff_q(&outq, p);
 596                 }
 597                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 598                         /*
 599                          * Deletion
 600                          *
 601                          * We would output this delete record if:
 602                          *
 603                          * (1) this is a broken delete and the counterpart
 604                          *     broken create remains in the output; or
 605                          * (2) this is not a broken delete, and rename_dst
 606                          *     does not have a rename/copy to move p->one->path
 607                          *     out of existence.
 608                          *
 609                          * Otherwise, the counterpart broken create
 610                          * has been turned into a rename-edit; or
 611                          * delete did not have a matching create to
 612                          * begin with.
 613                          */
 614                         if (DIFF_PAIR_BROKEN(p)) {
 615                                 /* broken delete */
 616                                 struct diff_rename_dst *dst =
 617                                         locate_rename_dst(p->one, 0);
 618                                 if (dst && dst->pair)
 619                                         /* counterpart is now rename/copy */
 620                                         pair_to_free = p;
 621                         }
 622                         else {
 623                                 if (p->one->rename_used)
 624                                         /* this path remains */
 625                                         pair_to_free = p;
 626                         }
 627
 628                         if (pair_to_free)
 629                                 ;
 630                         else
 631                                 diff_q(&outq, p);
 632                 }
 633                 else if (!diff_unmodified_pair(p))
 634                         /* all the usual ones need to be kept */
 635                         diff_q(&outq, p);
 636                 else
 637                         /* no need to keep unmodified pairs */
 638                         pair_to_free = p;
 639
 640                 if (pair_to_free)
 641                         diff_free_filepair(pair_to_free);
 642         }
 643         diff_debug_queue("done copying original", &outq);
 644
 645         free(q->queue);
 646         *q = outq;
 647         diff_debug_queue("done collapsing", q);
 648
 649         for (i = 0; i < rename_dst_nr; i++)
 650                 free_filespec(rename_dst[i].two);
 651
 652         free(rename_dst);
 653         rename_dst = NULL;
 654         rename_dst_nr = rename_dst_alloc = 0;
 655         free(rename_src);
 656         rename_src = NULL;
 657         rename_src_nr = rename_src_alloc = 0;
 658         return;
 659 }