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