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
   9 /* Table of rename/copy destinations */
  10
  11 static struct diff_rename_dst {
  12         struct diff_filespec *two;
  13         struct diff_filepair *pair;
  14 } *rename_dst;
  15 static int rename_dst_nr, rename_dst_alloc;
  16
  17 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
  18                                                  int insert_ok)
  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 dst;
  30                 if (cmp < 0) {
  31                         last = next;
  32                         continue;
  33                 }
  34                 first = next+1;
  35         }
  36         /* not found */
  37         if (!insert_ok)
  38                 return NULL;
  39         /* insert to make it at "first" */
  40         if (rename_dst_alloc <= rename_dst_nr) {
  41                 rename_dst_alloc = alloc_nr(rename_dst_alloc);
  42                 rename_dst = xrealloc(rename_dst,
  43                                       rename_dst_alloc * sizeof(*rename_dst));
  44         }
  45         rename_dst_nr++;
  46         if (first < rename_dst_nr)
  47                 memmove(rename_dst + first + 1, rename_dst + first,
  48                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  49         rename_dst[first].two = alloc_filespec(two->path);
  50         fill_filespec(rename_dst[first].two, two->sha1, two->mode);
  51         rename_dst[first].pair = NULL;
  52         return &(rename_dst[first]);
  53 }
  54
  55 /* Table of rename/copy src files */
  56 static struct diff_rename_src {
  57         struct diff_filespec *one;
  58         unsigned short score; /* to remember the break score */
  59 } *rename_src;
  60 static int rename_src_nr, rename_src_alloc;
  61
  62 static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
  63                                                    unsigned short score)
  64 {
  65         int first, last;
  66
  67         first = 0;
  68         last = rename_src_nr;
  69         while (last > first) {
  70                 int next = (last + first) >> 1;
  71                 struct diff_rename_src *src = &(rename_src[next]);
  72                 int cmp = strcmp(one->path, src->one->path);
  73                 if (!cmp)
  74                         return src;
  75                 if (cmp < 0) {
  76                         last = next;
  77                         continue;
  78                 }
  79                 first = next+1;
  80         }
  81
  82         /* insert to make it at "first" */
  83         if (rename_src_alloc <= rename_src_nr) {
  84                 rename_src_alloc = alloc_nr(rename_src_alloc);
  85                 rename_src = xrealloc(rename_src,
  86                                       rename_src_alloc * sizeof(*rename_src));
  87         }
  88         rename_src_nr++;
  89         if (first < rename_src_nr)
  90                 memmove(rename_src + first + 1, rename_src + first,
  91                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  92         rename_src[first].one = one;
  93         rename_src[first].score = score;
  94         return &(rename_src[first]);
  95 }
  96
  97 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
  98 {
  99         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 100         while (src_len && dst_len) {
 101                 char c1 = src->path[--src_len];
 102                 char c2 = dst->path[--dst_len];
 103                 if (c1 != c2)
 104                         return 0;
 105                 if (c1 == '/')
 106                         return 1;
 107         }
 108         return (!src_len || src->path[src_len - 1] == '/') &&
 109                 (!dst_len || dst->path[dst_len - 1] == '/');
 110 }
 111
 112 struct diff_score {
 113         int src; /* index in rename_src */
 114         int dst; /* index in rename_dst */
 115         unsigned short score;
 116         short name_score;
 117 };
 118
 119 static int estimate_similarity(struct diff_filespec *src,
 120                                struct diff_filespec *dst,
 121                                int minimum_score)
 122 {
 123         /* src points at a file that existed in the original tree (or
 124          * optionally a file in the destination tree) and dst points
 125          * at a newly created file.  They may be quite similar, in which
 126          * case we want to say src is renamed to dst or src is copied into
 127          * dst, and then some edit has been applied to dst.
 128          *
 129          * Compare them and return how similar they are, representing
 130          * the score as an integer between 0 and MAX_SCORE.
 131          *
 132          * When there is an exact match, it is considered a better
 133          * match than anything else; the destination does not even
 134          * call into this function in that case.
 135          */
 136         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 137         unsigned long delta_limit;
 138         int score;
 139
 140         /* We deal only with regular files.  Symlink renames are handled
 141          * only when they are exact matches --- in other words, no edits
 142          * after renaming.
 143          */
 144         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 145                 return 0;
 146
 147         /*
 148          * Need to check that source and destination sizes are
 149          * filled in before comparing them.
 150          *
 151          * If we already have "cnt_data" filled in, we know it's
 152          * all good (avoid checking the size for zero, as that
 153          * is a possible size - we really should have a flag to
 154          * say whether the size is valid or not!)
 155          */
 156         if (!src->cnt_data && diff_populate_filespec(src, 1))
 157                 return 0;
 158         if (!dst->cnt_data && diff_populate_filespec(dst, 1))
 159                 return 0;
 160
 161         max_size = ((src->size > dst->size) ? src->size : dst->size);
 162         base_size = ((src->size < dst->size) ? src->size : dst->size);
 163         delta_size = max_size - base_size;
 164
 165         /* We would not consider edits that change the file size so
 166          * drastically.  delta_size must be smaller than
 167          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 168          *
 169          * Note that base_size == 0 case is handled here already
 170          * and the final score computation below would not have a
 171          * divide-by-zero issue.
 172          */
 173         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 174                 return 0;
 175
 176         if (!src->cnt_data && diff_populate_filespec(src, 0))
 177                 return 0;
 178         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 179                 return 0;
 180
 181         delta_limit = (unsigned long)
 182                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 183         if (diffcore_count_changes(src, dst,
 184                                    &src->cnt_data, &dst->cnt_data,
 185                                    delta_limit,
 186                                    &src_copied, &literal_added))
 187                 return 0;
 188
 189         /* How similar are they?
 190          * what percentage of material in dst are from source?
 191          */
 192         if (!dst->size)
 193                 score = 0; /* should not happen */
 194         else
 195                 score = (int)(src_copied * MAX_SCORE / max_size);
 196         return score;
 197 }
 198
 199 static void record_rename_pair(int dst_index, int src_index, int score)
 200 {
 201         struct diff_filespec *src, *dst;
 202         struct diff_filepair *dp;
 203
 204         if (rename_dst[dst_index].pair)
 205                 die("internal error: dst already matched.");
 206
 207         src = rename_src[src_index].one;
 208         src->rename_used++;
 209         src->count++;
 210
 211         dst = rename_dst[dst_index].two;
 212         dst->count++;
 213
 214         dp = diff_queue(NULL, src, dst);
 215         dp->renamed_pair = 1;
 216         if (!strcmp(src->path, dst->path))
 217                 dp->score = rename_src[src_index].score;
 218         else
 219                 dp->score = score;
 220         rename_dst[dst_index].pair = dp;
 221 }
 222
 223 /*
 224  * We sort the rename similarity matrix with the score, in descending
 225  * order (the most similar first).
 226  */
 227 static int score_compare(const void *a_, const void *b_)
 228 {
 229         const struct diff_score *a = a_, *b = b_;
 230
 231         /* sink the unused ones to the bottom */
 232         if (a->dst < 0)
 233                 return (0 <= b->dst);
 234         else if (b->dst < 0)
 235                 return -1;
 236
 237         if (a->score == b->score)
 238                 return b->name_score - a->name_score;
 239
 240         return b->score - a->score;
 241 }
 242
 243 struct file_similarity {
 244         int src_dst, index;
 245         struct diff_filespec *filespec;
 246         struct file_similarity *next;
 247 };
 248
 249 static int find_identical_files(struct file_similarity *src,
 250                                 struct file_similarity *dst,
 251                                 struct diff_options *options)
 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, void *data)
 311 {
 312         int ret;
 313         struct file_similarity *p = ptr;
 314         struct file_similarity *src = NULL, *dst = NULL;
 315         struct diff_options *options = data;
 316
 317         /* Split the hash list up into sources and destinations */
 318         do {
 319                 struct file_similarity *entry = p;
 320                 p = p->next;
 321                 if (entry->src_dst < 0) {
 322                         entry->next = src;
 323                         src = entry;
 324                 } else {
 325                         entry->next = dst;
 326                         dst = entry;
 327                 }
 328         } while (p);
 329
 330         /*
 331          * If we have both sources *and* destinations, see if
 332          * we can match them up
 333          */
 334         ret = (src && dst) ? find_identical_files(src, dst, options) : 0;
 335
 336         /* Free the hashes and return the number of renames found */
 337         free_similarity_list(src);
 338         free_similarity_list(dst);
 339         return ret;
 340 }
 341
 342 static unsigned int hash_filespec(struct diff_filespec *filespec)
 343 {
 344         unsigned int hash;
 345         if (!filespec->sha1_valid) {
 346                 if (diff_populate_filespec(filespec, 0))
 347                         return 0;
 348                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 349         }
 350         memcpy(&hash, filespec->sha1, sizeof(hash));
 351         return hash;
 352 }
 353
 354 static void insert_file_table(struct hash_table *table, int src_dst, int index, struct diff_filespec *filespec)
 355 {
 356         void **pos;
 357         unsigned int hash;
 358         struct file_similarity *entry = xmalloc(sizeof(*entry));
 359
 360         entry->src_dst = src_dst;
 361         entry->index = index;
 362         entry->filespec = filespec;
 363         entry->next = NULL;
 364
 365         hash = hash_filespec(filespec);
 366         pos = insert_hash(hash, entry, table);
 367
 368         /* We already had an entry there? */
 369         if (pos) {
 370                 entry->next = *pos;
 371                 *pos = entry;
 372         }
 373 }
 374
 375 /*
 376  * Find exact renames first.
 377  *
 378  * The first round matches up the up-to-date entries,
 379  * and then during the second round we try to match
 380  * cache-dirty entries as well.
 381  */
 382 static int find_exact_renames(struct diff_options *options)
 383 {
 384         int i;
 385         struct hash_table file_table;
 386
 387         init_hash(&file_table);
 388         for (i = 0; i < rename_src_nr; i++)
 389                 insert_file_table(&file_table, -1, i, rename_src[i].one);
 390
 391         for (i = 0; i < rename_dst_nr; i++)
 392                 insert_file_table(&file_table, 1, i, rename_dst[i].two);
 393
 394         /* Find the renames */
 395         i = for_each_hash(&file_table, find_same_files, options);
 396
 397         /* .. and free the hash data structure */
 398         free_hash(&file_table);
 399
 400         return i;
 401 }
 402
 403 #define NUM_CANDIDATE_PER_DST 4
 404 static void record_if_better(struct diff_score m[], struct diff_score *o)
 405 {
 406         int i, worst;
 407
 408         /* find the worst one */
 409         worst = 0;
 410         for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
 411                 if (score_compare(&m[i], &m[worst]) > 0)
 412                         worst = i;
 413
 414         /* is it better than the worst one? */
 415         if (score_compare(&m[worst], o) > 0)
 416                 m[worst] = *o;
 417 }
 418
 419 void diffcore_rename(struct diff_options *options)
 420 {
 421         int detect_rename = options->detect_rename;
 422         int minimum_score = options->rename_score;
 423         int rename_limit = options->rename_limit;
 424         struct diff_queue_struct *q = &diff_queued_diff;
 425         struct diff_queue_struct outq;
 426         struct diff_score *mx;
 427         int i, j, rename_count;
 428         int num_create, num_src, dst_cnt;
 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(options);
 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         if (rename_limit <= 0 || rename_limit > 32767)
 499                 rename_limit = 32767;
 500         if ((num_create > rename_limit && num_src > rename_limit) ||
 501             (num_create * num_src > rename_limit * rename_limit)) {
 502                 if (options->warn_on_too_large_rename)
 503                         warning("too many files (created: %d deleted: %d), skipping inexact rename detection", num_create, num_src);
 504                 goto cleanup;
 505         }
 506
 507         mx = xcalloc(num_create * NUM_CANDIDATE_PER_DST, sizeof(*mx));
 508         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 509                 struct diff_filespec *two = rename_dst[i].two;
 510                 struct diff_score *m;
 511
 512                 if (rename_dst[i].pair)
 513                         continue; /* dealt with exact match already. */
 514
 515                 m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
 516                 for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
 517                         m[j].dst = -1;
 518
 519                 for (j = 0; j < rename_src_nr; j++) {
 520                         struct diff_filespec *one = rename_src[j].one;
 521                         struct diff_score this_src;
 522                         this_src.score = estimate_similarity(one, two,
 523                                                              minimum_score);
 524                         this_src.name_score = basename_same(one, two);
 525                         this_src.dst = i;
 526                         this_src.src = j;
 527                         record_if_better(m, &this_src);
 528                         /*
 529                          * Once we run estimate_similarity,
 530                          * We do not need the text anymore.
 531                          */
 532                         diff_free_filespec_blob(one);
 533                         diff_free_filespec_blob(two);
 534                 }
 535                 dst_cnt++;
 536         }
 537
 538         /* cost matrix sorted by most to least similar pair */
 539         qsort(mx, dst_cnt * NUM_CANDIDATE_PER_DST, sizeof(*mx), score_compare);
 540
 541         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 542                 struct diff_rename_dst *dst;
 543
 544                 if ((mx[i].dst < 0) ||
 545                     (mx[i].score < minimum_score))
 546                         break; /* there is no more usable pair. */
 547                 dst = &rename_dst[mx[i].dst];
 548                 if (dst->pair)
 549                         continue; /* already done, either exact or fuzzy. */
 550                 if (rename_src[mx[i].src].one->rename_used)
 551                         continue;
 552                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 553                 rename_count++;
 554         }
 555
 556         for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
 557                 struct diff_rename_dst *dst;
 558
 559                 if ((mx[i].dst < 0) ||
 560                     (mx[i].score < minimum_score))
 561                         break; /* there is no more usable pair. */
 562                 dst = &rename_dst[mx[i].dst];
 563                 if (dst->pair)
 564                         continue; /* already done, either exact or fuzzy. */
 565                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 566                 rename_count++;
 567         }
 568         free(mx);
 569
 570  cleanup:
 571         /* At this point, we have found some renames and copies and they
 572          * are recorded in rename_dst.  The original list is still in *q.
 573          */
 574         DIFF_QUEUE_CLEAR(&outq);
 575         for (i = 0; i < q->nr; i++) {
 576                 struct diff_filepair *p = q->queue[i];
 577                 struct diff_filepair *pair_to_free = NULL;
 578
 579                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 580                         /*
 581                          * Creation
 582                          *
 583                          * We would output this create record if it has
 584                          * not been turned into a rename/copy already.
 585                          */
 586                         struct diff_rename_dst *dst =
 587                                 locate_rename_dst(p->two, 0);
 588                         if (dst && dst->pair) {
 589                                 diff_q(&outq, dst->pair);
 590                                 pair_to_free = p;
 591                         }
 592                         else
 593                                 /* no matching rename/copy source, so
 594                                  * record this as a creation.
 595                                  */
 596                                 diff_q(&outq, p);
 597                 }
 598                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 599                         /*
 600                          * Deletion
 601                          *
 602                          * We would output this delete record if:
 603                          *
 604                          * (1) this is a broken delete and the counterpart
 605                          *     broken create remains in the output; or
 606                          * (2) this is not a broken delete, and rename_dst
 607                          *     does not have a rename/copy to move p->one->path
 608                          *     out of existence.
 609                          *
 610                          * Otherwise, the counterpart broken create
 611                          * has been turned into a rename-edit; or
 612                          * delete did not have a matching create to
 613                          * begin with.
 614                          */
 615                         if (DIFF_PAIR_BROKEN(p)) {
 616                                 /* broken delete */
 617                                 struct diff_rename_dst *dst =
 618                                         locate_rename_dst(p->one, 0);
 619                                 if (dst && dst->pair)
 620                                         /* counterpart is now rename/copy */
 621                                         pair_to_free = p;
 622                         }
 623                         else {
 624                                 if (p->one->rename_used)
 625                                         /* this path remains */
 626                                         pair_to_free = p;
 627                         }
 628
 629                         if (pair_to_free)
 630                                 ;
 631                         else
 632                                 diff_q(&outq, p);
 633                 }
 634                 else if (!diff_unmodified_pair(p))
 635                         /* all the usual ones need to be kept */
 636                         diff_q(&outq, p);
 637                 else
 638                         /* no need to keep unmodified pairs */
 639                         pair_to_free = p;
 640
 641                 if (pair_to_free)
 642                         diff_free_filepair(pair_to_free);
 643         }
 644         diff_debug_queue("done copying original", &outq);
 645
 646         free(q->queue);
 647         *q = outq;
 648         diff_debug_queue("done collapsing", q);
 649
 650         for (i = 0; i < rename_dst_nr; i++)
 651                 free_filespec(rename_dst[i].two);
 652
 653         free(rename_dst);
 654         rename_dst = NULL;
 655         rename_dst_nr = rename_dst_alloc = 0;
 656         free(rename_src);
 657         rename_src = NULL;
 658         rename_src_nr = rename_src_alloc = 0;
 659         return;
 660 }