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->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_filespec *one;
  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_filespec *one,
  64                                                    unsigned short score)
  65 {
  66         int first, last;
  67
  68         first = 0;
  69         last = rename_src_nr;
  70         while (last > first) {
  71                 int next = (last + first) >> 1;
  72                 struct diff_rename_src *src = &(rename_src[next]);
  73                 int cmp = strcmp(one->path, src->one->path);
  74                 if (!cmp)
  75                         return src;
  76                 if (cmp < 0) {
  77                         last = next;
  78                         continue;
  79                 }
  80                 first = next+1;
  81         }
  82
  83         /* insert to make it at "first" */
  84         if (rename_src_alloc <= rename_src_nr) {
  85                 rename_src_alloc = alloc_nr(rename_src_alloc);
  86                 rename_src = xrealloc(rename_src,
  87                                       rename_src_alloc * sizeof(*rename_src));
  88         }
  89         rename_src_nr++;
  90         if (first < rename_src_nr)
  91                 memmove(rename_src + first + 1, rename_src + first,
  92                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  93         rename_src[first].one = one;
  94         rename_src[first].score = score;
  95         return &(rename_src[first]);
  96 }
  97
  98 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
  99 {
 100         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 101         while (src_len && dst_len) {
 102                 char c1 = src->path[--src_len];
 103                 char c2 = dst->path[--dst_len];
 104                 if (c1 != c2)
 105                         return 0;
 106                 if (c1 == '/')
 107                         return 1;
 108         }
 109         return (!src_len || src->path[src_len - 1] == '/') &&
 110                 (!dst_len || dst->path[dst_len - 1] == '/');
 111 }
 112
 113 struct diff_score {
 114         int src; /* index in rename_src */
 115         int dst; /* index in rename_dst */
 116         unsigned short score;
 117         short name_score;
 118 };
 119
 120 static int estimate_similarity(struct diff_filespec *src,
 121                                struct diff_filespec *dst,
 122                                int minimum_score)
 123 {
 124         /* src points at a file that existed in the original tree (or
 125          * optionally a file in the destination tree) and dst points
 126          * at a newly created file.  They may be quite similar, in which
 127          * case we want to say src is renamed to dst or src is copied into
 128          * dst, and then some edit has been applied to dst.
 129          *
 130          * Compare them and return how similar they are, representing
 131          * the score as an integer between 0 and MAX_SCORE.
 132          *
 133          * When there is an exact match, it is considered a better
 134          * match than anything else; the destination does not even
 135          * call into this function in that case.
 136          */
 137         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 138         unsigned long delta_limit;
 139         int score;
 140
 141         /* We deal only with regular files.  Symlink renames are handled
 142          * only when they are exact matches --- in other words, no edits
 143          * after renaming.
 144          */
 145         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 146                 return 0;
 147
 148         /*
 149          * Need to check that source and destination sizes are
 150          * filled in before comparing them.
 151          *
 152          * If we already have "cnt_data" filled in, we know it's
 153          * all good (avoid checking the size for zero, as that
 154          * is a possible size - we really should have a flag to
 155          * say whether the size is valid or not!)
 156          */
 157         if (!src->cnt_data && diff_populate_filespec(src, 1))
 158                 return 0;
 159         if (!dst->cnt_data && diff_populate_filespec(dst, 1))
 160                 return 0;
 161
 162         max_size = ((src->size > dst->size) ? src->size : dst->size);
 163         base_size = ((src->size < dst->size) ? src->size : dst->size);
 164         delta_size = max_size - base_size;
 165
 166         /* We would not consider edits that change the file size so
 167          * drastically.  delta_size must be smaller than
 168          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 169          *
 170          * Note that base_size == 0 case is handled here already
 171          * and the final score computation below would not have a
 172          * divide-by-zero issue.
 173          */
 174         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 175                 return 0;
 176
 177         if (!src->cnt_data && diff_populate_filespec(src, 0))
 178                 return 0;
 179         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 180                 return 0;
 181
 182         delta_limit = (unsigned long)
 183                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 184         if (diffcore_count_changes(src, dst,
 185                                    &src->cnt_data, &dst->cnt_data,
 186                                    delta_limit,
 187                                    &src_copied, &literal_added))
 188                 return 0;
 189
 190         /* How similar are they?
 191          * what percentage of material in dst are from source?
 192          */
 193         if (!dst->size)
 194                 score = 0; /* should not happen */
 195         else
 196                 score = (int)(src_copied * MAX_SCORE / max_size);
 197         return score;
 198 }
 199
 200 static void record_rename_pair(int dst_index, int src_index, int score)
 201 {
 202         struct diff_filespec *src, *dst;
 203         struct diff_filepair *dp;
 204
 205         if (rename_dst[dst_index].pair)
 206                 die("internal error: dst already matched.");
 207
 208         src = rename_src[src_index].one;
 209         src->rename_used++;
 210         src->count++;
 211
 212         dst = rename_dst[dst_index].two;
 213         dst->count++;
 214
 215         dp = diff_queue(NULL, src, dst);
 216         dp->renamed_pair = 1;
 217         if (!strcmp(src->path, dst->path))
 218                 dp->score = rename_src[src_index].score;
 219         else
 220                 dp->score = score;
 221         rename_dst[dst_index].pair = dp;
 222 }
 223
 224 /*
 225  * We sort the rename similarity matrix with the score, in descending
 226  * order (the most similar first).
 227  */
 228 static int score_compare(const void *a_, const void *b_)
 229 {
 230         const struct diff_score *a = a_, *b = b_;
 231
 232         /* sink the unused ones to the bottom */
 233         if (a->dst < 0)
 234                 return (0 <= b->dst);
 235         else if (b->dst < 0)
 236                 return -1;
 237
 238         if (a->score == b->score)
 239                 return b->name_score - a->name_score;
 240
 241         return b->score - a->score;
 242 }
 243
 244 struct file_similarity {
 245         int src_dst, index;
 246         struct diff_filespec *filespec;
 247         struct file_similarity *next;
 248 };
 249
 250 static int find_identical_files(struct file_similarity *src,
 251                                 struct file_similarity *dst)
 252 {
 253         int renames = 0;
 254
 255         /*
 256          * Walk over all the destinations ...
 257          */
 258         do {
 259                 struct diff_filespec *target = dst->filespec;
 260                 struct file_similarity *p, *best;
 261                 int i = 100, best_score = -1;
 262
 263                 /*
 264                  * .. to find the best source match
 265                  */
 266                 best = NULL;
 267                 for (p = src; p; p = p->next) {
 268                         int score;
 269                         struct diff_filespec *source = p->filespec;
 270
 271                         /* False hash collision? */
 272                         if (hashcmp(source->sha1, target->sha1))
 273                                 continue;
 274                         /* Non-regular files? If so, the modes must match! */
 275                         if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 276                                 if (source->mode != target->mode)
 277                                         continue;
 278                         }
 279                         /* Give higher scores to sources that haven't been used already */
 280                         score = !source->rename_used;
 281                         score += basename_same(source, target);
 282                         if (score > best_score) {
 283                                 best = p;
 284                                 best_score = score;
 285                                 if (score == 2)
 286                                         break;
 287                         }
 288
 289                         /* Too many identical alternatives? Pick one */
 290                         if (!--i)
 291                                 break;
 292                 }
 293                 if (best) {
 294                         record_rename_pair(dst->index, best->index, MAX_SCORE);
 295                         renames++;
 296                 }
 297         } while ((dst = dst->next) != NULL);
 298         return renames;
 299 }
 300
 301 static void free_similarity_list(struct file_similarity *p)
 302 {
 303         while (p) {
 304                 struct file_similarity *entry = p;
 305                 p = p->next;
 306                 free(entry);
 307         }
 308 }
 309
 310 static int find_same_files(void *ptr)
 311 {
 312         int ret;
 313         struct file_similarity *p = ptr;
 314         struct file_similarity *src = NULL, *dst = NULL;
 315
 316         /* Split the hash list up into sources and destinations */
 317         do {
 318                 struct file_similarity *entry = p;
 319                 p = p->next;
 320                 if (entry->src_dst < 0) {
 321                         entry->next = src;
 322                         src = entry;
 323                 } else {
 324                         entry->next = dst;
 325                         dst = entry;
 326                 }
 327         } while (p);
 328
 329         /*
 330          * If we have both sources *and* destinations, see if
 331          * we can match them up
 332          */
 333         ret = (src && dst) ? find_identical_files(src, dst) : 0;
 334
 335         /* Free the hashes and return the number of renames found */
 336         free_similarity_list(src);
 337         free_similarity_list(dst);
 338         return ret;
 339 }
 340
 341 static unsigned int hash_filespec(struct diff_filespec *filespec)
 342 {
 343         unsigned int hash;
 344         if (!filespec->sha1_valid) {
 345                 if (diff_populate_filespec(filespec, 0))
 346                         return 0;
 347                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 348         }
 349         memcpy(&hash, filespec->sha1, sizeof(hash));
 350         return hash;
 351 }
 352
 353 static void insert_file_table(struct hash_table *table, int src_dst, int index, struct diff_filespec *filespec)
 354 {
 355         void **pos;
 356         unsigned int hash;
 357         struct file_similarity *entry = xmalloc(sizeof(*entry));
 358
 359         entry->src_dst = src_dst;
 360         entry->index = index;
 361         entry->filespec = filespec;
 362         entry->next = NULL;
 363
 364         hash = hash_filespec(filespec);
 365         pos = insert_hash(hash, entry, table);
 366
 367         /* We already had an entry there? */
 368         if (pos) {
 369                 entry->next = *pos;
 370                 *pos = entry;
 371         }
 372 }
 373
 374 /*
 375  * Find exact renames first.
 376  *
 377  * The first round matches up the up-to-date entries,
 378  * and then during the second round we try to match
 379  * cache-dirty entries as well.
 380  */
 381 static int find_exact_renames(void)
 382 {
 383         int i;
 384         struct hash_table file_table;
 385
 386         init_hash(&file_table);
 387         for (i = 0; i < rename_src_nr; i++)
 388                 insert_file_table(&file_table, -1, i, rename_src[i].one);
 389
 390         for (i = 0; i < rename_dst_nr; i++)
 391                 insert_file_table(&file_table, 1, i, rename_dst[i].two);
 392
 393         /* Find the renames */
 394         i = for_each_hash(&file_table, find_same_files);
 395
 396         /* .. and free the hash data structure */
 397         free_hash(&file_table);
 398
 399         return i;
 400 }
 401
 402 #define NUM_CANDIDATE_PER_DST 4
 403 static void record_if_better(struct diff_score m[], struct diff_score *o)
 404 {
 405         int i, worst;
 406
 407         /* find the worst one */
 408         worst = 0;
 409         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 410                 if (score_compare(&m[i], &m[worst]) > 0)
 411                         worst = i;
 412
 413         /* is it better than the worst one? */
 414         if (score_compare(&m[worst], o) > 0)
 415                 m[worst] = *o;
 416 }
 417
 418 void diffcore_rename(struct diff_options *options)
 419 {
 420         int detect_rename = options->detect_rename;
 421         int minimum_score = options->rename_score;
 422         int rename_limit = options->rename_limit;
 423         struct diff_queue_struct *q = &diff_queued_diff;
 424         struct diff_queue_struct outq;
 425         struct diff_score *mx;
 426         int i, j, rename_count;
 427         int num_create, num_src, dst_cnt;
 428         struct progress *progress = NULL;
 429
 430         if (!minimum_score)
 431                 minimum_score = DEFAULT_RENAME_SCORE;
 432
 433         for (i = 0; i < q->nr; i++) {
 434                 struct diff_filepair *p = q->queue[i];
 435                 if (!DIFF_FILE_VALID(p->one)) {
 436                         if (!DIFF_FILE_VALID(p->two))
 437                                 continue; /* unmerged */
 438                         else if (options->single_follow &&
 439                                  strcmp(options->single_follow, p->two->path))
 440                                 continue; /* not interested */
 441                         else
 442                                 locate_rename_dst(p->two, 1);
 443                 }
 444                 else if (!DIFF_FILE_VALID(p->two)) {
 445                         /*
 446                          * If the source is a broken "delete", and
 447                          * they did not really want to get broken,
 448                          * that means the source actually stays.
 449                          * So we increment the "rename_used" score
 450                          * by one, to indicate ourselves as a user
 451                          */
 452                         if (p->broken_pair && !p->score)
 453                                 p->one->rename_used++;
 454                         register_rename_src(p->one, p->score);
 455                 }
 456                 else if (detect_rename == DIFF_DETECT_COPY) {
 457                         /*
 458                          * Increment the "rename_used" score by
 459                          * one, to indicate ourselves as a user.
 460                          */
 461                         p->one->rename_used++;
 462                         register_rename_src(p->one, p->score);
 463                 }
 464         }
 465         if (rename_dst_nr == 0 || rename_src_nr == 0)
 466                 goto cleanup; /* nothing to do */
 467
 468         /*
 469          * We really want to cull the candidates list early
 470          * with cheap tests in order to avoid doing deltas.
 471          */
 472         rename_count = find_exact_renames();
 473
 474         /* Did we only want exact renames? */
 475         if (minimum_score == MAX_SCORE)
 476                 goto cleanup;
 477
 478         /*
 479          * Calculate how many renames are left (but all the source
 480          * files still remain as options for rename/copies!)
 481          */
 482         num_create = (rename_dst_nr - rename_count);
 483         num_src = rename_src_nr;
 484
 485         /* All done? */
 486         if (!num_create)
 487                 goto cleanup;
 488
 489         /*
 490          * This basically does a test for the rename matrix not
 491          * growing larger than a "rename_limit" square matrix, ie:
 492          *
 493          *    num_create * num_src > rename_limit * rename_limit
 494          *
 495          * but handles the potential overflow case specially (and we
 496          * assume at least 32-bit integers)
 497          */
 498         options->needed_rename_limit = 0;
 499         if (rename_limit <= 0 || rename_limit > 32767)
 500                 rename_limit = 32767;
 501         if ((num_create > rename_limit && num_src > rename_limit) ||
 502             (num_create * num_src > rename_limit * rename_limit)) {
 503                 options->needed_rename_limit =
 504                         num_src > num_create ? num_src : num_create;
 505                 goto cleanup;
 506         }
 507
 508         if (options->show_rename_progress) {
 509                 progress = start_progress_delay(
 510                                 "Performing inexact rename detection",
 511                                 rename_dst_nr * rename_src_nr, 50, 1);
 512         }
 513
 514         mx = xcalloc(num_create * NUM_CANDIDATE_PER_DST, sizeof(*mx));
 515         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 516                 struct diff_filespec *two = rename_dst[i].two;
 517                 struct diff_score *m;
 518
 519                 if (rename_dst[i].pair)
 520                         continue; /* dealt with exact match already. */
 521
 522                 m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
 523                 for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
 524                         m[j].dst = -1;
 525
 526                 for (j = 0; j < rename_src_nr; j++) {
 527                         struct diff_filespec *one = rename_src[j].one;
 528                         struct diff_score this_src;
 529                         this_src.score = estimate_similarity(one, two,
 530                                                              minimum_score);
 531                         this_src.name_score = basename_same(one, two);
 532                         this_src.dst = i;
 533                         this_src.src = j;
 534                         record_if_better(m, &this_src);
 535                         /*
 536                          * Once we run estimate_similarity,
 537                          * We do not need the text anymore.
 538                          */
 539                         diff_free_filespec_blob(one);
 540                         diff_free_filespec_blob(two);
 541                 }
 542                 dst_cnt++;
 543                 display_progress(progress, (i+1)*rename_src_nr);
 544         }
 545         stop_progress(&progress);
 546
 547         /* cost matrix sorted by most to least similar pair */
 548         qsort(mx, dst_cnt * NUM_CANDIDATE_PER_DST, sizeof(*mx), score_compare);
 549
 550         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 551                 struct diff_rename_dst *dst;
 552
 553                 if ((mx[i].dst < 0) ||
 554                     (mx[i].score < minimum_score))
 555                         break; /* there is no more usable pair. */
 556                 dst = &rename_dst[mx[i].dst];
 557                 if (dst->pair)
 558                         continue; /* already done, either exact or fuzzy. */
 559                 if (rename_src[mx[i].src].one->rename_used)
 560                         continue;
 561                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 562                 rename_count++;
 563         }
 564
 565         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 566                 struct diff_rename_dst *dst;
 567
 568                 if ((mx[i].dst < 0) ||
 569                     (mx[i].score < minimum_score))
 570                         break; /* there is no more usable pair. */
 571                 dst = &rename_dst[mx[i].dst];
 572                 if (dst->pair)
 573                         continue; /* already done, either exact or fuzzy. */
 574                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 575                 rename_count++;
 576         }
 577         free(mx);
 578
 579  cleanup:
 580         /* At this point, we have found some renames and copies and they
 581          * are recorded in rename_dst.  The original list is still in *q.
 582          */
 583         DIFF_QUEUE_CLEAR(&outq);
 584         for (i = 0; i < q->nr; i++) {
 585                 struct diff_filepair *p = q->queue[i];
 586                 struct diff_filepair *pair_to_free = NULL;
 587
 588                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 589                         /*
 590                          * Creation
 591                          *
 592                          * We would output this create record if it has
 593                          * not been turned into a rename/copy already.
 594                          */
 595                         struct diff_rename_dst *dst =
 596                                 locate_rename_dst(p->two, 0);
 597                         if (dst && dst->pair) {
 598                                 diff_q(&outq, dst->pair);
 599                                 pair_to_free = p;
 600                         }
 601                         else
 602                                 /* no matching rename/copy source, so
 603                                  * record this as a creation.
 604                                  */
 605                                 diff_q(&outq, p);
 606                 }
 607                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 608                         /*
 609                          * Deletion
 610                          *
 611                          * We would output this delete record if:
 612                          *
 613                          * (1) this is a broken delete and the counterpart
 614                          *     broken create remains in the output; or
 615                          * (2) this is not a broken delete, and rename_dst
 616                          *     does not have a rename/copy to move p->one->path
 617                          *     out of existence.
 618                          *
 619                          * Otherwise, the counterpart broken create
 620                          * has been turned into a rename-edit; or
 621                          * delete did not have a matching create to
 622                          * begin with.
 623                          */
 624                         if (DIFF_PAIR_BROKEN(p)) {
 625                                 /* broken delete */
 626                                 struct diff_rename_dst *dst =
 627                                         locate_rename_dst(p->one, 0);
 628                                 if (dst && dst->pair)
 629                                         /* counterpart is now rename/copy */
 630                                         pair_to_free = p;
 631                         }
 632                         else {
 633                                 if (p->one->rename_used)
 634                                         /* this path remains */
 635                                         pair_to_free = p;
 636                         }
 637
 638                         if (pair_to_free)
 639                                 ;
 640                         else
 641                                 diff_q(&outq, p);
 642                 }
 643                 else if (!diff_unmodified_pair(p))
 644                         /* all the usual ones need to be kept */
 645                         diff_q(&outq, p);
 646                 else
 647                         /* no need to keep unmodified pairs */
 648                         pair_to_free = p;
 649
 650                 if (pair_to_free)
 651                         diff_free_filepair(pair_to_free);
 652         }
 653         diff_debug_queue("done copying original", &outq);
 654
 655         free(q->queue);
 656         *q = outq;
 657         diff_debug_queue("done collapsing", q);
 658
 659         for (i = 0; i < rename_dst_nr; i++)
 660                 free_filespec(rename_dst[i].two);
 661
 662         free(rename_dst);
 663         rename_dst = NULL;
 664         rename_dst_nr = rename_dst_alloc = 0;
 665         free(rename_src);
 666         rename_src = NULL;
 667         rename_src_nr = rename_src_alloc = 0;
 668         return;
 669 }