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         int score;
 116         int 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, 0))
 157                 return 0;
 158         if (!dst->cnt_data && diff_populate_filespec(dst, 0))
 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         delta_limit = (unsigned long)
 177                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 178         if (diffcore_count_changes(src, dst,
 179                                    &src->cnt_data, &dst->cnt_data,
 180                                    delta_limit,
 181                                    &src_copied, &literal_added))
 182                 return 0;
 183
 184         /* How similar are they?
 185          * what percentage of material in dst are from source?
 186          */
 187         if (!dst->size)
 188                 score = 0; /* should not happen */
 189         else
 190                 score = (int)(src_copied * MAX_SCORE / max_size);
 191         return score;
 192 }
 193
 194 static void record_rename_pair(int dst_index, int src_index, int score)
 195 {
 196         struct diff_filespec *src, *dst;
 197         struct diff_filepair *dp;
 198
 199         if (rename_dst[dst_index].pair)
 200                 die("internal error: dst already matched.");
 201
 202         src = rename_src[src_index].one;
 203         src->rename_used++;
 204         src->count++;
 205
 206         dst = rename_dst[dst_index].two;
 207         dst->count++;
 208
 209         dp = diff_queue(NULL, src, dst);
 210         dp->renamed_pair = 1;
 211         if (!strcmp(src->path, dst->path))
 212                 dp->score = rename_src[src_index].score;
 213         else
 214                 dp->score = score;
 215         rename_dst[dst_index].pair = dp;
 216 }
 217
 218 /*
 219  * We sort the rename similarity matrix with the score, in descending
 220  * order (the most similar first).
 221  */
 222 static int score_compare(const void *a_, const void *b_)
 223 {
 224         const struct diff_score *a = a_, *b = b_;
 225
 226         if (a->score == b->score)
 227                 return b->name_score - a->name_score;
 228
 229         return b->score - a->score;
 230 }
 231
 232 struct file_similarity {
 233         int src_dst, index;
 234         struct diff_filespec *filespec;
 235         struct file_similarity *next;
 236 };
 237
 238 static int find_identical_files(struct file_similarity *src,
 239                                 struct file_similarity *dst)
 240 {
 241         int renames = 0;
 242
 243         /*
 244          * Walk over all the destinations ...
 245          */
 246         do {
 247                 struct diff_filespec *target = dst->filespec;
 248                 struct file_similarity *p, *best;
 249                 int i = 100, best_score = -1;
 250
 251                 /*
 252                  * .. to find the best source match
 253                  */
 254                 best = NULL;
 255                 for (p = src; p; p = p->next) {
 256                         int score;
 257                         struct diff_filespec *source = p->filespec;
 258
 259                         /* False hash collission? */
 260                         if (hashcmp(source->sha1, target->sha1))
 261                                 continue;
 262                         /* Non-regular files? If so, the modes must match! */
 263                         if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
 264                                 if (source->mode != target->mode)
 265                                         continue;
 266                         }
 267                         /* Give higher scores to sources that haven't been used already */
 268                         score = !source->rename_used;
 269                         score += basename_same(source, target);
 270                         if (score > best_score) {
 271                                 best = p;
 272                                 best_score = score;
 273                                 if (score == 2)
 274                                         break;
 275                         }
 276
 277                         /* Too many identical alternatives? Pick one */
 278                         if (!--i)
 279                                 break;
 280                 }
 281                 if (best) {
 282                         record_rename_pair(dst->index, best->index, MAX_SCORE);
 283                         renames++;
 284                 }
 285         } while ((dst = dst->next) != NULL);
 286         return renames;
 287 }
 288
 289 static void free_similarity_list(struct file_similarity *p)
 290 {
 291         while (p) {
 292                 struct file_similarity *entry = p;
 293                 p = p->next;
 294                 free(entry);
 295         }
 296 }
 297
 298 static int find_same_files(void *ptr)
 299 {
 300         int ret;
 301         struct file_similarity *p = ptr;
 302         struct file_similarity *src = NULL, *dst = NULL;
 303
 304         /* Split the hash list up into sources and destinations */
 305         do {
 306                 struct file_similarity *entry = p;
 307                 p = p->next;
 308                 if (entry->src_dst < 0) {
 309                         entry->next = src;
 310                         src = entry;
 311                 } else {
 312                         entry->next = dst;
 313                         dst = entry;
 314                 }
 315         } while (p);
 316
 317         /*
 318          * If we have both sources *and* destinations, see if
 319          * we can match them up
 320          */
 321         ret = (src && dst) ? find_identical_files(src, dst) : 0;
 322
 323         /* Free the hashes and return the number of renames found */
 324         free_similarity_list(src);
 325         free_similarity_list(dst);
 326         return ret;
 327 }
 328
 329 static unsigned int hash_filespec(struct diff_filespec *filespec)
 330 {
 331         unsigned int hash;
 332         if (!filespec->sha1_valid) {
 333                 if (diff_populate_filespec(filespec, 0))
 334                         return 0;
 335                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 336         }
 337         memcpy(&hash, filespec->sha1, sizeof(hash));
 338         return hash;
 339 }
 340
 341 static void insert_file_table(struct hash_table *table, int src_dst, int index, struct diff_filespec *filespec)
 342 {
 343         void **pos;
 344         unsigned int hash;
 345         struct file_similarity *entry = xmalloc(sizeof(*entry));
 346
 347         entry->src_dst = src_dst;
 348         entry->index = index;
 349         entry->filespec = filespec;
 350         entry->next = NULL;
 351
 352         hash = hash_filespec(filespec);
 353         pos = insert_hash(hash, entry, table);
 354
 355         /* We already had an entry there? */
 356         if (pos) {
 357                 entry->next = *pos;
 358                 *pos = entry;
 359         }
 360 }
 361
 362 /*
 363  * Find exact renames first.
 364  *
 365  * The first round matches up the up-to-date entries,
 366  * and then during the second round we try to match
 367  * cache-dirty entries as well.
 368  */
 369 static int find_exact_renames(void)
 370 {
 371         int i;
 372         struct hash_table file_table;
 373
 374         init_hash(&file_table);
 375         for (i = 0; i < rename_src_nr; i++)
 376                 insert_file_table(&file_table, -1, i, rename_src[i].one);
 377
 378         for (i = 0; i < rename_dst_nr; i++)
 379                 insert_file_table(&file_table, 1, i, rename_dst[i].two);
 380
 381         /* Find the renames */
 382         i = for_each_hash(&file_table, find_same_files);
 383
 384         /* .. and free the hash data structure */
 385         free_hash(&file_table);
 386
 387         return i;
 388 }
 389
 390 void diffcore_rename(struct diff_options *options)
 391 {
 392         int detect_rename = options->detect_rename;
 393         int minimum_score = options->rename_score;
 394         int rename_limit = options->rename_limit;
 395         struct diff_queue_struct *q = &diff_queued_diff;
 396         struct diff_queue_struct outq;
 397         struct diff_score *mx;
 398         int i, j, rename_count;
 399         int num_create, num_src, dst_cnt;
 400
 401         if (!minimum_score)
 402                 minimum_score = DEFAULT_RENAME_SCORE;
 403
 404         for (i = 0; i < q->nr; i++) {
 405                 struct diff_filepair *p = q->queue[i];
 406                 if (!DIFF_FILE_VALID(p->one)) {
 407                         if (!DIFF_FILE_VALID(p->two))
 408                                 continue; /* unmerged */
 409                         else if (options->single_follow &&
 410                                  strcmp(options->single_follow, p->two->path))
 411                                 continue; /* not interested */
 412                         else
 413                                 locate_rename_dst(p->two, 1);
 414                 }
 415                 else if (!DIFF_FILE_VALID(p->two)) {
 416                         /*
 417                          * If the source is a broken "delete", and
 418                          * they did not really want to get broken,
 419                          * that means the source actually stays.
 420                          * So we increment the "rename_used" score
 421                          * by one, to indicate ourselves as a user
 422                          */
 423                         if (p->broken_pair && !p->score)
 424                                 p->one->rename_used++;
 425                         register_rename_src(p->one, p->score);
 426                 }
 427                 else if (detect_rename == DIFF_DETECT_COPY) {
 428                         /*
 429                          * Increment the "rename_used" score by
 430                          * one, to indicate ourselves as a user.
 431                          */
 432                         p->one->rename_used++;
 433                         register_rename_src(p->one, p->score);
 434                 }
 435         }
 436         if (rename_dst_nr == 0 || rename_src_nr == 0)
 437                 goto cleanup; /* nothing to do */
 438
 439         /*
 440          * We really want to cull the candidates list early
 441          * with cheap tests in order to avoid doing deltas.
 442          */
 443         rename_count = find_exact_renames();
 444
 445         /* Did we only want exact renames? */
 446         if (minimum_score == MAX_SCORE)
 447                 goto cleanup;
 448
 449         /*
 450          * Calculate how many renames are left (but all the source
 451          * files still remain as options for rename/copies!)
 452          */
 453         num_create = (rename_dst_nr - rename_count);
 454         num_src = rename_src_nr;
 455
 456         /* All done? */
 457         if (!num_create)
 458                 goto cleanup;
 459
 460         /*
 461          * This basically does a test for the rename matrix not
 462          * growing larger than a "rename_limit" square matrix, ie:
 463          *
 464          *    num_create * num_src > rename_limit * rename_limit
 465          *
 466          * but handles the potential overflow case specially (and we
 467          * assume at least 32-bit integers)
 468          */
 469         if (rename_limit <= 0 || rename_limit > 32767)
 470                 rename_limit = 32767;
 471         if ((num_create > rename_limit && num_src > rename_limit) ||
 472             (num_create * num_src > rename_limit * rename_limit)) {
 473                 warning("too many files, skipping inexact rename detection");
 474                 goto cleanup;
 475         }
 476
 477         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 478         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 479                 int base = dst_cnt * num_src;
 480                 struct diff_filespec *two = rename_dst[i].two;
 481                 if (rename_dst[i].pair)
 482                         continue; /* dealt with exact match already. */
 483                 for (j = 0; j < rename_src_nr; j++) {
 484                         struct diff_filespec *one = rename_src[j].one;
 485                         struct diff_score *m = &mx[base+j];
 486                         m->src = j;
 487                         m->dst = i;
 488                         m->score = estimate_similarity(one, two,
 489                                                        minimum_score);
 490                         m->name_score = basename_same(one, two);
 491                         diff_free_filespec_blob(one);
 492                 }
 493                 /* We do not need the text anymore */
 494                 diff_free_filespec_blob(two);
 495                 dst_cnt++;
 496         }
 497         /* cost matrix sorted by most to least similar pair */
 498         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 499         for (i = 0; i < num_create * num_src; i++) {
 500                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 501                 struct diff_filespec *src;
 502                 if (dst->pair)
 503                         continue; /* already done, either exact or fuzzy. */
 504                 if (mx[i].score < minimum_score)
 505                         break; /* there is no more usable pair. */
 506                 src = rename_src[mx[i].src].one;
 507                 if (src->rename_used)
 508                         continue;
 509                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 510                 rename_count++;
 511         }
 512         for (i = 0; i < num_create * num_src; i++) {
 513                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 514                 if (dst->pair)
 515                         continue; /* already done, either exact or fuzzy. */
 516                 if (mx[i].score < minimum_score)
 517                         break; /* there is no more usable pair. */
 518                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 519                 rename_count++;
 520         }
 521         free(mx);
 522
 523  cleanup:
 524         /* At this point, we have found some renames and copies and they
 525          * are recorded in rename_dst.  The original list is still in *q.
 526          */
 527         outq.queue = NULL;
 528         outq.nr = outq.alloc = 0;
 529         for (i = 0; i < q->nr; i++) {
 530                 struct diff_filepair *p = q->queue[i];
 531                 struct diff_filepair *pair_to_free = NULL;
 532
 533                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 534                         /*
 535                          * Creation
 536                          *
 537                          * We would output this create record if it has
 538                          * not been turned into a rename/copy already.
 539                          */
 540                         struct diff_rename_dst *dst =
 541                                 locate_rename_dst(p->two, 0);
 542                         if (dst && dst->pair) {
 543                                 diff_q(&outq, dst->pair);
 544                                 pair_to_free = p;
 545                         }
 546                         else
 547                                 /* no matching rename/copy source, so
 548                                  * record this as a creation.
 549                                  */
 550                                 diff_q(&outq, p);
 551                 }
 552                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 553                         /*
 554                          * Deletion
 555                          *
 556                          * We would output this delete record if:
 557                          *
 558                          * (1) this is a broken delete and the counterpart
 559                          *     broken create remains in the output; or
 560                          * (2) this is not a broken delete, and rename_dst
 561                          *     does not have a rename/copy to move p->one->path
 562                          *     out of existence.
 563                          *
 564                          * Otherwise, the counterpart broken create
 565                          * has been turned into a rename-edit; or
 566                          * delete did not have a matching create to
 567                          * begin with.
 568                          */
 569                         if (DIFF_PAIR_BROKEN(p)) {
 570                                 /* broken delete */
 571                                 struct diff_rename_dst *dst =
 572                                         locate_rename_dst(p->one, 0);
 573                                 if (dst && dst->pair)
 574                                         /* counterpart is now rename/copy */
 575                                         pair_to_free = p;
 576                         }
 577                         else {
 578                                 if (p->one->rename_used)
 579                                         /* this path remains */
 580                                         pair_to_free = p;
 581                         }
 582
 583                         if (pair_to_free)
 584                                 ;
 585                         else
 586                                 diff_q(&outq, p);
 587                 }
 588                 else if (!diff_unmodified_pair(p))
 589                         /* all the usual ones need to be kept */
 590                         diff_q(&outq, p);
 591                 else
 592                         /* no need to keep unmodified pairs */
 593                         pair_to_free = p;
 594
 595                 if (pair_to_free)
 596                         diff_free_filepair(pair_to_free);
 597         }
 598         diff_debug_queue("done copying original", &outq);
 599
 600         free(q->queue);
 601         *q = outq;
 602         diff_debug_queue("done collapsing", q);
 603
 604         for (i = 0; i < rename_dst_nr; i++)
 605                 free_filespec(rename_dst[i].two);
 606
 607         free(rename_dst);
 608         rename_dst = NULL;
 609         rename_dst_nr = rename_dst_alloc = 0;
 610         free(rename_src);
 611         rename_src = NULL;
 612         rename_src_nr = rename_src_alloc = 0;
 613         return;
 614 }