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