git.oblomov.eu Git - git/blob - 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
   8 /* Table of rename/copy destinations */
   9
  10 static struct diff_rename_dst {
  11         struct diff_filespec *two;
  12         struct diff_filepair *pair;
  13 } *rename_dst;
  14 static int rename_dst_nr, rename_dst_alloc;
  15
  16 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
  17                                                  int insert_ok)
  18 {
  19         int first, last;
  20
  21         first = 0;
  22         last = rename_dst_nr;
  23         while (last > first) {
  24                 int next = (last + first) >> 1;
  25                 struct diff_rename_dst *dst = &(rename_dst[next]);
  26                 int cmp = strcmp(two->path, dst->two->path);
  27                 if (!cmp)
  28                         return dst;
  29                 if (cmp < 0) {
  30                         last = next;
  31                         continue;
  32                 }
  33                 first = next+1;
  34         }
  35         /* not found */
  36         if (!insert_ok)
  37                 return NULL;
  38         /* insert to make it at "first" */
  39         if (rename_dst_alloc <= rename_dst_nr) {
  40                 rename_dst_alloc = alloc_nr(rename_dst_alloc);
  41                 rename_dst = xrealloc(rename_dst,
  42                                       rename_dst_alloc * sizeof(*rename_dst));
  43         }
  44         rename_dst_nr++;
  45         if (first < rename_dst_nr)
  46                 memmove(rename_dst + first + 1, rename_dst + first,
  47                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  48         rename_dst[first].two = alloc_filespec(two->path);
  49         fill_filespec(rename_dst[first].two, two->sha1, two->mode);
  50         rename_dst[first].pair = NULL;
  51         return &(rename_dst[first]);
  52 }
  53
  54 /* Table of rename/copy src files */
  55 static struct diff_rename_src {
  56         struct diff_filespec *one;
  57         unsigned short score; /* to remember the break score */
  58 } *rename_src;
  59 static int rename_src_nr, rename_src_alloc;
  60
  61 static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
  62                                                    unsigned short score)
  63 {
  64         int first, last;
  65
  66         first = 0;
  67         last = rename_src_nr;
  68         while (last > first) {
  69                 int next = (last + first) >> 1;
  70                 struct diff_rename_src *src = &(rename_src[next]);
  71                 int cmp = strcmp(one->path, src->one->path);
  72                 if (!cmp)
  73                         return src;
  74                 if (cmp < 0) {
  75                         last = next;
  76                         continue;
  77                 }
  78                 first = next+1;
  79         }
  80
  81         /* insert to make it at "first" */
  82         if (rename_src_alloc <= rename_src_nr) {
  83                 rename_src_alloc = alloc_nr(rename_src_alloc);
  84                 rename_src = xrealloc(rename_src,
  85                                       rename_src_alloc * sizeof(*rename_src));
  86         }
  87         rename_src_nr++;
  88         if (first < rename_src_nr)
  89                 memmove(rename_src + first + 1, rename_src + first,
  90                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  91         rename_src[first].one = one;
  92         rename_src[first].score = score;
  93         return &(rename_src[first]);
  94 }
  95
  96 static int is_exact_match(struct diff_filespec *src,
  97                           struct diff_filespec *dst,
  98                           int contents_too)
  99 {
 100         if (src->sha1_valid && dst->sha1_valid &&
 101             !hashcmp(src->sha1, dst->sha1))
 102                 return 1;
 103         if (!contents_too)
 104                 return 0;
 105         if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
 106                 return 0;
 107         if (src->size != dst->size)
 108                 return 0;
 109         if (src->sha1_valid && dst->sha1_valid)
 110             return !hashcmp(src->sha1, dst->sha1);
 111         if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 112                 return 0;
 113         if (src->size == dst->size &&
 114             !memcmp(src->data, dst->data, src->size))
 115                 return 1;
 116         return 0;
 117 }
 118
 119 static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
 120 {
 121         int src_len = strlen(src->path), dst_len = strlen(dst->path);
 122         while (src_len && dst_len) {
 123                 char c1 = src->path[--src_len];
 124                 char c2 = dst->path[--dst_len];
 125                 if (c1 != c2)
 126                         return 0;
 127                 if (c1 == '/')
 128                         return 1;
 129         }
 130         return (!src_len || src->path[src_len - 1] == '/') &&
 131                 (!dst_len || dst->path[dst_len - 1] == '/');
 132 }
 133
 134 struct diff_score {
 135         int src; /* index in rename_src */
 136         int dst; /* index in rename_dst */
 137         int score;
 138         int name_score;
 139 };
 140
 141 static int estimate_similarity(struct diff_filespec *src,
 142                                struct diff_filespec *dst,
 143                                int minimum_score)
 144 {
 145         /* src points at a file that existed in the original tree (or
 146          * optionally a file in the destination tree) and dst points
 147          * at a newly created file.  They may be quite similar, in which
 148          * case we want to say src is renamed to dst or src is copied into
 149          * dst, and then some edit has been applied to dst.
 150          *
 151          * Compare them and return how similar they are, representing
 152          * the score as an integer between 0 and MAX_SCORE.
 153          *
 154          * When there is an exact match, it is considered a better
 155          * match than anything else; the destination does not even
 156          * call into this function in that case.
 157          */
 158         unsigned long max_size, delta_size, base_size, src_copied, literal_added;
 159         unsigned long delta_limit;
 160         int score;
 161
 162         /* We deal only with regular files.  Symlink renames are handled
 163          * only when they are exact matches --- in other words, no edits
 164          * after renaming.
 165          */
 166         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 167                 return 0;
 168
 169         max_size = ((src->size > dst->size) ? src->size : dst->size);
 170         base_size = ((src->size < dst->size) ? src->size : dst->size);
 171         delta_size = max_size - base_size;
 172
 173         /* We would not consider edits that change the file size so
 174          * drastically.  delta_size must be smaller than
 175          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 176          *
 177          * Note that base_size == 0 case is handled here already
 178          * and the final score computation below would not have a
 179          * divide-by-zero issue.
 180          */
 181         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 182                 return 0;
 183
 184         if ((!src->cnt_data && diff_populate_filespec(src, 0))
 185                 || (!dst->cnt_data && diff_populate_filespec(dst, 0)))
 186                 return 0; /* error but caught downstream */
 187
 188
 189         delta_limit = (unsigned long)
 190                 (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE);
 191         if (diffcore_count_changes(src, dst,
 192                                    &src->cnt_data, &dst->cnt_data,
 193                                    delta_limit,
 194                                    &src_copied, &literal_added))
 195                 return 0;
 196
 197         /* How similar are they?
 198          * what percentage of material in dst are from source?
 199          */
 200         if (!dst->size)
 201                 score = 0; /* should not happen */
 202         else
 203                 score = (int)(src_copied * MAX_SCORE / max_size);
 204         return score;
 205 }
 206
 207 static void record_rename_pair(int dst_index, int src_index, int score)
 208 {
 209         struct diff_filespec *src, *dst;
 210         struct diff_filepair *dp;
 211
 212         if (rename_dst[dst_index].pair)
 213                 die("internal error: dst already matched.");
 214
 215         src = rename_src[src_index].one;
 216         src->rename_used++;
 217         src->count++;
 218
 219         dst = rename_dst[dst_index].two;
 220         dst->count++;
 221
 222         dp = diff_queue(NULL, src, dst);
 223         dp->renamed_pair = 1;
 224         if (!strcmp(src->path, dst->path))
 225                 dp->score = rename_src[src_index].score;
 226         else
 227                 dp->score = score;
 228         rename_dst[dst_index].pair = dp;
 229 }
 230
 231 /*
 232  * We sort the rename similarity matrix with the score, in descending
 233  * order (the most similar first).
 234  */
 235 static int score_compare(const void *a_, const void *b_)
 236 {
 237         const struct diff_score *a = a_, *b = b_;
 238
 239         if (a->score == b->score)
 240                 return b->name_score - a->name_score;
 241
 242         return b->score - a->score;
 243 }
 244
 245 /*
 246  * Find exact renames first.
 247  *
 248  * The first round matches up the up-to-date entries,
 249  * and then during the second round we try to match
 250  * cache-dirty entries as well.
 251  *
 252  * Note: the rest of the rename logic depends on this
 253  * phase also populating all the filespecs for any
 254  * entry that isn't matched up with an exact rename,
 255  * see "is_exact_match()".
 256  */
 257 static int find_exact_renames(void)
 258 {
 259         int rename_count = 0;
 260         int contents_too;
 261
 262         for (contents_too = 0; contents_too < 2; contents_too++) {
 263                 int i;
 264
 265                 for (i = 0; i < rename_dst_nr; i++) {
 266                         struct diff_filespec *two = rename_dst[i].two;
 267                         int j;
 268
 269                         if (rename_dst[i].pair)
 270                                 continue; /* dealt with an earlier round */
 271                         for (j = 0; j < rename_src_nr; j++) {
 272                                 int k;
 273                                 struct diff_filespec *one = rename_src[j].one;
 274                                 if (!is_exact_match(one, two, contents_too))
 275                                         continue;
 276
 277                                 /* see if there is a basename match, too */
 278                                 for (k = j; k < rename_src_nr; k++) {
 279                                         one = rename_src[k].one;
 280                                         if (basename_same(one, two) &&
 281                                                 is_exact_match(one, two,
 282                                                         contents_too)) {
 283                                                 j = k;
 284                                                 break;
 285                                         }
 286                                 }
 287
 288                                 record_rename_pair(i, j, (int)MAX_SCORE);
 289                                 rename_count++;
 290                                 break; /* we are done with this entry */
 291                         }
 292                 }
 293         }
 294         return rename_count;
 295 }
 296
 297 void diffcore_rename(struct diff_options *options)
 298 {
 299         int detect_rename = options->detect_rename;
 300         int minimum_score = options->rename_score;
 301         int rename_limit = options->rename_limit;
 302         struct diff_queue_struct *q = &diff_queued_diff;
 303         struct diff_queue_struct outq;
 304         struct diff_score *mx;
 305         int i, j, rename_count;
 306         int num_create, num_src, dst_cnt;
 307
 308         if (!minimum_score)
 309                 minimum_score = DEFAULT_RENAME_SCORE;
 310
 311         for (i = 0; i < q->nr; i++) {
 312                 struct diff_filepair *p = q->queue[i];
 313                 if (!DIFF_FILE_VALID(p->one)) {
 314                         if (!DIFF_FILE_VALID(p->two))
 315                                 continue; /* unmerged */
 316                         else if (options->single_follow &&
 317                                  strcmp(options->single_follow, p->two->path))
 318                                 continue; /* not interested */
 319                         else
 320                                 locate_rename_dst(p->two, 1);
 321                 }
 322                 else if (!DIFF_FILE_VALID(p->two)) {
 323                         /*
 324                          * If the source is a broken "delete", and
 325                          * they did not really want to get broken,
 326                          * that means the source actually stays.
 327                          * So we increment the "rename_used" score
 328                          * by one, to indicate ourselves as a user
 329                          */
 330                         if (p->broken_pair && !p->score)
 331                                 p->one->rename_used++;
 332                         register_rename_src(p->one, p->score);
 333                 }
 334                 else if (detect_rename == DIFF_DETECT_COPY) {
 335                         /*
 336                          * Increment the "rename_used" score by
 337                          * one, to indicate ourselves as a user.
 338                          */
 339                         p->one->rename_used++;
 340                         register_rename_src(p->one, p->score);
 341                 }
 342         }
 343         if (rename_dst_nr == 0 || rename_src_nr == 0)
 344                 goto cleanup; /* nothing to do */
 345
 346         /*
 347          * This basically does a test for the rename matrix not
 348          * growing larger than a "rename_limit" square matrix, ie:
 349          *
 350          *    rename_dst_nr * rename_src_nr > rename_limit * rename_limit
 351          *
 352          * but handles the potential overflow case specially (and we
 353          * assume at least 32-bit integers)
 354          */
 355         if (rename_limit <= 0 || rename_limit > 32767)
 356                 rename_limit = 32767;
 357         if (rename_dst_nr > rename_limit && rename_src_nr > rename_limit)
 358                 goto cleanup;
 359         if (rename_dst_nr * rename_src_nr > rename_limit * rename_limit)
 360                 goto cleanup;
 361
 362         /*
 363          * We really want to cull the candidates list early
 364          * with cheap tests in order to avoid doing deltas.
 365          */
 366         rename_count = find_exact_renames();
 367
 368         /* Have we run out the created file pool?  If so we can avoid
 369          * doing the delta matrix altogether.
 370          */
 371         if (rename_count == rename_dst_nr)
 372                 goto cleanup;
 373
 374         if (minimum_score == MAX_SCORE)
 375                 goto cleanup;
 376
 377         num_create = (rename_dst_nr - rename_count);
 378         num_src = rename_src_nr;
 379         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 380         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 381                 int base = dst_cnt * num_src;
 382                 struct diff_filespec *two = rename_dst[i].two;
 383                 if (rename_dst[i].pair)
 384                         continue; /* dealt with exact match already. */
 385                 for (j = 0; j < rename_src_nr; j++) {
 386                         struct diff_filespec *one = rename_src[j].one;
 387                         struct diff_score *m = &mx[base+j];
 388                         m->src = j;
 389                         m->dst = i;
 390                         m->score = estimate_similarity(one, two,
 391                                                        minimum_score);
 392                         m->name_score = basename_same(one, two);
 393                         diff_free_filespec_blob(one);
 394                 }
 395                 /* We do not need the text anymore */
 396                 diff_free_filespec_blob(two);
 397                 dst_cnt++;
 398         }
 399         /* cost matrix sorted by most to least similar pair */
 400         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 401         for (i = 0; i < num_create * num_src; i++) {
 402                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 403                 if (dst->pair)
 404                         continue; /* already done, either exact or fuzzy. */
 405                 if (mx[i].score < minimum_score)
 406                         break; /* there is no more usable pair. */
 407                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 408                 rename_count++;
 409         }
 410         free(mx);
 411
 412  cleanup:
 413         /* At this point, we have found some renames and copies and they
 414          * are recorded in rename_dst.  The original list is still in *q.
 415          */
 416         outq.queue = NULL;
 417         outq.nr = outq.alloc = 0;
 418         for (i = 0; i < q->nr; i++) {
 419                 struct diff_filepair *p = q->queue[i];
 420                 struct diff_filepair *pair_to_free = NULL;
 421
 422                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 423                         /*
 424                          * Creation
 425                          *
 426                          * We would output this create record if it has
 427                          * not been turned into a rename/copy already.
 428                          */
 429                         struct diff_rename_dst *dst =
 430                                 locate_rename_dst(p->two, 0);
 431                         if (dst && dst->pair) {
 432                                 diff_q(&outq, dst->pair);
 433                                 pair_to_free = p;
 434                         }
 435                         else
 436                                 /* no matching rename/copy source, so
 437                                  * record this as a creation.
 438                                  */
 439                                 diff_q(&outq, p);
 440                 }
 441                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 442                         /*
 443                          * Deletion
 444                          *
 445                          * We would output this delete record if:
 446                          *
 447                          * (1) this is a broken delete and the counterpart
 448                          *     broken create remains in the output; or
 449                          * (2) this is not a broken delete, and rename_dst
 450                          *     does not have a rename/copy to move p->one->path
 451                          *     out of existence.
 452                          *
 453                          * Otherwise, the counterpart broken create
 454                          * has been turned into a rename-edit; or
 455                          * delete did not have a matching create to
 456                          * begin with.
 457                          */
 458                         if (DIFF_PAIR_BROKEN(p)) {
 459                                 /* broken delete */
 460                                 struct diff_rename_dst *dst =
 461                                         locate_rename_dst(p->one, 0);
 462                                 if (dst && dst->pair)
 463                                         /* counterpart is now rename/copy */
 464                                         pair_to_free = p;
 465                         }
 466                         else {
 467                                 if (p->one->rename_used)
 468                                         /* this path remains */
 469                                         pair_to_free = p;
 470                         }
 471
 472                         if (pair_to_free)
 473                                 ;
 474                         else
 475                                 diff_q(&outq, p);
 476                 }
 477                 else if (!diff_unmodified_pair(p))
 478                         /* all the usual ones need to be kept */
 479                         diff_q(&outq, p);
 480                 else
 481                         /* no need to keep unmodified pairs */
 482                         pair_to_free = p;
 483
 484                 if (pair_to_free)
 485                         diff_free_filepair(pair_to_free);
 486         }
 487         diff_debug_queue("done copying original", &outq);
 488
 489         free(q->queue);
 490         *q = outq;
 491         diff_debug_queue("done collapsing", q);
 492
 493         for (i = 0; i < rename_dst_nr; i++)
 494                 free_filespec(rename_dst[i].two);
 495
 496         free(rename_dst);
 497         rename_dst = NULL;
 498         rename_dst_nr = rename_dst_alloc = 0;
 499         free(rename_src);
 500         rename_src = NULL;
 501         rename_src_nr = rename_src_alloc = 0;
 502         return;
 503 }