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         max_size = ((src->size > dst->size) ? src->size : dst->size);
 148         base_size = ((src->size < dst->size) ? src->size : dst->size);
 149         delta_size = max_size - base_size;
 150
 151         /* We would not consider edits that change the file size so
 152          * drastically.  delta_size must be smaller than
 153          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 154          *
 155          * Note that base_size == 0 case is handled here already
 156          * and the final score computation below would not have a
 157          * divide-by-zero issue.
 158          */
 159         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 160                 return 0;
 161
 162         if ((!src->cnt_data && diff_populate_filespec(src, 0))
 163                 || (!dst->cnt_data && diff_populate_filespec(dst, 0)))
 164                 return 0; /* error but caught downstream */
 165
 166
 167         delta_limit = (unsigned long)
 168                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 169         if (diffcore_count_changes(src, dst,
 170                                    &src->cnt_data, &dst->cnt_data,
 171                                    delta_limit,
 172                                    &src_copied, &literal_added))
 173                 return 0;
 174
 175         /* How similar are they?
 176          * what percentage of material in dst are from source?
 177          */
 178         if (!dst->size)
 179                 score = 0; /* should not happen */
 180         else
 181                 score = (int)(src_copied * MAX_SCORE / max_size);
 182         return score;
 183 }
 184
 185 static void record_rename_pair(int dst_index, int src_index, int score)
 186 {
 187         struct diff_filespec *src, *dst;
 188         struct diff_filepair *dp;
 189
 190         if (rename_dst[dst_index].pair)
 191                 die("internal error: dst already matched.");
 192
 193         src = rename_src[src_index].one;
 194         src->rename_used++;
 195         src->count++;
 196
 197         dst = rename_dst[dst_index].two;
 198         dst->count++;
 199
 200         dp = diff_queue(NULL, src, dst);
 201         dp->renamed_pair = 1;
 202         if (!strcmp(src->path, dst->path))
 203                 dp->score = rename_src[src_index].score;
 204         else
 205                 dp->score = score;
 206         rename_dst[dst_index].pair = dp;
 207 }
 208
 209 /*
 210  * We sort the rename similarity matrix with the score, in descending
 211  * order (the most similar first).
 212  */
 213 static int score_compare(const void *a_, const void *b_)
 214 {
 215         const struct diff_score *a = a_, *b = b_;
 216
 217         if (a->score == b->score)
 218                 return b->name_score - a->name_score;
 219
 220         return b->score - a->score;
 221 }
 222
 223 struct file_similarity {
 224         int src_dst, index;
 225         struct diff_filespec *filespec;
 226         struct file_similarity *next;
 227 };
 228
 229 static int find_identical_files(struct file_similarity *src,
 230                                 struct file_similarity *dst)
 231 {
 232         int renames = 0;
 233
 234         /*
 235          * Walk over all the destinations ...
 236          */
 237         do {
 238                 struct diff_filespec *one = dst->filespec;
 239                 struct file_similarity *p, *best;
 240                 int i = 100;
 241
 242                 /*
 243                  * .. to find the best source match
 244                  */
 245                 best = NULL;
 246                 for (p = src; p; p = p->next) {
 247                         struct diff_filespec *two = p->filespec;
 248
 249                         /* False hash collission? */
 250                         if (hashcmp(one->sha1, two->sha1))
 251                                 continue;
 252                         /* Non-regular files? If so, the modes must match! */
 253                         if (!S_ISREG(one->mode) || !S_ISREG(two->mode)) {
 254                                 if (one->mode != two->mode)
 255                                         continue;
 256                         }
 257                         best = p;
 258                         if (basename_same(one, two))
 259                                 break;
 260
 261                         /* Too many identical alternatives? Pick one */
 262                         if (!--i)
 263                                 break;
 264                 }
 265                 if (best) {
 266                         record_rename_pair(dst->index, best->index, MAX_SCORE);
 267                         renames++;
 268                 }
 269         } while ((dst = dst->next) != NULL);
 270         return renames;
 271 }
 272
 273 /*
 274  * Note: the rest of the rename logic depends on this
 275  * phase also populating all the filespecs for any
 276  * entry that isn't matched up with an exact rename.
 277  */
 278 static void free_similarity_list(struct file_similarity *p)
 279 {
 280         while (p) {
 281                 struct file_similarity *entry = p;
 282                 p = p->next;
 283
 284                 /* Stupid special case, see note above! */
 285                 diff_populate_filespec(entry->filespec, 0);
 286                 free(entry);
 287         }
 288 }
 289
 290 static int find_same_files(void *ptr)
 291 {
 292         int ret;
 293         struct file_similarity *p = ptr;
 294         struct file_similarity *src = NULL, *dst = NULL;
 295
 296         /* Split the hash list up into sources and destinations */
 297         do {
 298                 struct file_similarity *entry = p;
 299                 p = p->next;
 300                 if (entry->src_dst < 0) {
 301                         entry->next = src;
 302                         src = entry;
 303                 } else {
 304                         entry->next = dst;
 305                         dst = entry;
 306                 }
 307         } while (p);
 308
 309         /*
 310          * If we have both sources *and* destinations, see if
 311          * we can match them up
 312          */
 313         ret = (src && dst) ? find_identical_files(src, dst) : 0;
 314
 315         /* Free the hashes and return the number of renames found */
 316         free_similarity_list(src);
 317         free_similarity_list(dst);
 318         return ret;
 319 }
 320
 321 static unsigned int hash_filespec(struct diff_filespec *filespec)
 322 {
 323         unsigned int hash;
 324         if (!filespec->sha1_valid) {
 325                 if (diff_populate_filespec(filespec, 0))
 326                         return 0;
 327                 hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
 328         }
 329         memcpy(&hash, filespec->sha1, sizeof(hash));
 330         return hash;
 331 }
 332
 333 static void insert_file_table(struct hash_table *table, int src_dst, int index, struct diff_filespec *filespec)
 334 {
 335         void **pos;
 336         unsigned int hash;
 337         struct file_similarity *entry = xmalloc(sizeof(*entry));
 338
 339         entry->src_dst = src_dst;
 340         entry->index = index;
 341         entry->filespec = filespec;
 342         entry->next = NULL;
 343
 344         hash = hash_filespec(filespec);
 345         pos = insert_hash(hash, entry, table);
 346
 347         /* We already had an entry there? */
 348         if (pos) {
 349                 entry->next = *pos;
 350                 *pos = entry;
 351         }
 352 }
 353
 354 /*
 355  * Find exact renames first.
 356  *
 357  * The first round matches up the up-to-date entries,
 358  * and then during the second round we try to match
 359  * cache-dirty entries as well.
 360  */
 361 static int find_exact_renames(void)
 362 {
 363         int i;
 364         struct hash_table file_table;
 365
 366         init_hash(&file_table);
 367         for (i = 0; i < rename_src_nr; i++)
 368                 insert_file_table(&file_table, -1, i, rename_src[i].one);
 369
 370         for (i = 0; i < rename_dst_nr; i++)
 371                 insert_file_table(&file_table, 1, i, rename_dst[i].two);
 372
 373         /* Find the renames */
 374         i = for_each_hash(&file_table, find_same_files);
 375
 376         /* .. and free the hash data structure */
 377         free_hash(&file_table);
 378
 379         return i;
 380 }
 381
 382 void diffcore_rename(struct diff_options *options)
 383 {
 384         int detect_rename = options->detect_rename;
 385         int minimum_score = options->rename_score;
 386         int rename_limit = options->rename_limit;
 387         struct diff_queue_struct *q = &diff_queued_diff;
 388         struct diff_queue_struct outq;
 389         struct diff_score *mx;
 390         int i, j, rename_count;
 391         int num_create, num_src, dst_cnt;
 392
 393         if (!minimum_score)
 394                 minimum_score = DEFAULT_RENAME_SCORE;
 395
 396         for (i = 0; i < q->nr; i++) {
 397                 struct diff_filepair *p = q->queue[i];
 398                 if (!DIFF_FILE_VALID(p->one)) {
 399                         if (!DIFF_FILE_VALID(p->two))
 400                                 continue; /* unmerged */
 401                         else if (options->single_follow &&
 402                                  strcmp(options->single_follow, p->two->path))
 403                                 continue; /* not interested */
 404                         else
 405                                 locate_rename_dst(p->two, 1);
 406                 }
 407                 else if (!DIFF_FILE_VALID(p->two)) {
 408                         /*
 409                          * If the source is a broken "delete", and
 410                          * they did not really want to get broken,
 411                          * that means the source actually stays.
 412                          * So we increment the "rename_used" score
 413                          * by one, to indicate ourselves as a user
 414                          */
 415                         if (p->broken_pair && !p->score)
 416                                 p->one->rename_used++;
 417                         register_rename_src(p->one, p->score);
 418                 }
 419                 else if (detect_rename == DIFF_DETECT_COPY) {
 420                         /*
 421                          * Increment the "rename_used" score by
 422                          * one, to indicate ourselves as a user.
 423                          */
 424                         p->one->rename_used++;
 425                         register_rename_src(p->one, p->score);
 426                 }
 427         }
 428         if (rename_dst_nr == 0 || rename_src_nr == 0)
 429                 goto cleanup; /* nothing to do */
 430
 431         /*
 432          * This basically does a test for the rename matrix not
 433          * growing larger than a "rename_limit" square matrix, ie:
 434          *
 435          *    rename_dst_nr * rename_src_nr > rename_limit * rename_limit
 436          *
 437          * but handles the potential overflow case specially (and we
 438          * assume at least 32-bit integers)
 439          */
 440         if (rename_limit <= 0 || rename_limit > 32767)
 441                 rename_limit = 32767;
 442         if (rename_dst_nr > rename_limit && rename_src_nr > rename_limit)
 443                 goto cleanup;
 444         if (rename_dst_nr * rename_src_nr > rename_limit * rename_limit)
 445                 goto cleanup;
 446
 447         /*
 448          * We really want to cull the candidates list early
 449          * with cheap tests in order to avoid doing deltas.
 450          */
 451         rename_count = find_exact_renames();
 452
 453         /* Have we run out the created file pool?  If so we can avoid
 454          * doing the delta matrix altogether.
 455          */
 456         if (rename_count == rename_dst_nr)
 457                 goto cleanup;
 458
 459         if (minimum_score == MAX_SCORE)
 460                 goto cleanup;
 461
 462         num_create = (rename_dst_nr - rename_count);
 463         num_src = rename_src_nr;
 464         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 465         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 466                 int base = dst_cnt * num_src;
 467                 struct diff_filespec *two = rename_dst[i].two;
 468                 if (rename_dst[i].pair)
 469                         continue; /* dealt with exact match already. */
 470                 for (j = 0; j < rename_src_nr; j++) {
 471                         struct diff_filespec *one = rename_src[j].one;
 472                         struct diff_score *m = &mx[base+j];
 473                         m->src = j;
 474                         m->dst = i;
 475                         m->score = estimate_similarity(one, two,
 476                                                        minimum_score);
 477                         m->name_score = basename_same(one, two);
 478                         diff_free_filespec_blob(one);
 479                 }
 480                 /* We do not need the text anymore */
 481                 diff_free_filespec_blob(two);
 482                 dst_cnt++;
 483         }
 484         /* cost matrix sorted by most to least similar pair */
 485         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 486         for (i = 0; i < num_create * num_src; i++) {
 487                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 488                 if (dst->pair)
 489                         continue; /* already done, either exact or fuzzy. */
 490                 if (mx[i].score < minimum_score)
 491                         break; /* there is no more usable pair. */
 492                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 493                 rename_count++;
 494         }
 495         free(mx);
 496
 497  cleanup:
 498         /* At this point, we have found some renames and copies and they
 499          * are recorded in rename_dst.  The original list is still in *q.
 500          */
 501         outq.queue = NULL;
 502         outq.nr = outq.alloc = 0;
 503         for (i = 0; i < q->nr; i++) {
 504                 struct diff_filepair *p = q->queue[i];
 505                 struct diff_filepair *pair_to_free = NULL;
 506
 507                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 508                         /*
 509                          * Creation
 510                          *
 511                          * We would output this create record if it has
 512                          * not been turned into a rename/copy already.
 513                          */
 514                         struct diff_rename_dst *dst =
 515                                 locate_rename_dst(p->two, 0);
 516                         if (dst && dst->pair) {
 517                                 diff_q(&outq, dst->pair);
 518                                 pair_to_free = p;
 519                         }
 520                         else
 521                                 /* no matching rename/copy source, so
 522                                  * record this as a creation.
 523                                  */
 524                                 diff_q(&outq, p);
 525                 }
 526                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 527                         /*
 528                          * Deletion
 529                          *
 530                          * We would output this delete record if:
 531                          *
 532                          * (1) this is a broken delete and the counterpart
 533                          *     broken create remains in the output; or
 534                          * (2) this is not a broken delete, and rename_dst
 535                          *     does not have a rename/copy to move p->one->path
 536                          *     out of existence.
 537                          *
 538                          * Otherwise, the counterpart broken create
 539                          * has been turned into a rename-edit; or
 540                          * delete did not have a matching create to
 541                          * begin with.
 542                          */
 543                         if (DIFF_PAIR_BROKEN(p)) {
 544                                 /* broken delete */
 545                                 struct diff_rename_dst *dst =
 546                                         locate_rename_dst(p->one, 0);
 547                                 if (dst && dst->pair)
 548                                         /* counterpart is now rename/copy */
 549                                         pair_to_free = p;
 550                         }
 551                         else {
 552                                 if (p->one->rename_used)
 553                                         /* this path remains */
 554                                         pair_to_free = p;
 555                         }
 556
 557                         if (pair_to_free)
 558                                 ;
 559                         else
 560                                 diff_q(&outq, p);
 561                 }
 562                 else if (!diff_unmodified_pair(p))
 563                         /* all the usual ones need to be kept */
 564                         diff_q(&outq, p);
 565                 else
 566                         /* no need to keep unmodified pairs */
 567                         pair_to_free = p;
 568
 569                 if (pair_to_free)
 570                         diff_free_filepair(pair_to_free);
 571         }
 572         diff_debug_queue("done copying original", &outq);
 573
 574         free(q->queue);
 575         *q = outq;
 576         diff_debug_queue("done collapsing", q);
 577
 578         for (i = 0; i < rename_dst_nr; i++)
 579                 free_filespec(rename_dst[i].two);
 580
 581         free(rename_dst);
 582         rename_dst = NULL;
 583         rename_dst_nr = rename_dst_alloc = 0;
 584         free(rename_src);
 585         rename_src = NULL;
 586         rename_src_nr = rename_src_alloc = 0;
 587         return;
 588 }