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