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