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 src_path_left : 1;
  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                                                    int src_path_left)
  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].src_path_left = src_path_left;
  93         return &(rename_src[first]);
  94 }
  95
  96 static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
  97 {
  98         if (src->sha1_valid && dst->sha1_valid &&
  99             !memcmp(src->sha1, dst->sha1, 20))
 100                 return 1;
 101         if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
 102                 return 0;
 103         if (src->size != dst->size)
 104                 return 0;
 105         if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 106                 return 0;
 107         if (src->size == dst->size &&
 108             !memcmp(src->data, dst->data, src->size))
 109                 return 1;
 110         return 0;
 111 }
 112
 113 struct diff_score {
 114         int src; /* index in rename_src */
 115         int dst; /* index in rename_dst */
 116         int 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 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         delta_size = ((src->size < dst->size) ?
 148                       (dst->size - src->size) : (src->size - dst->size));
 149         base_size = ((src->size < dst->size) ? src->size : dst->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 (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 163                 return 0; /* error but caught downstream */
 164
 165
 166         delta_limit = base_size * (MAX_SCORE-minimum_score) / MAX_SCORE;
 167         if (diffcore_count_changes(src->data, src->size,
 168                                    dst->data, dst->size,
 169                                    delta_limit,
 170                                    &src_copied, &literal_added))
 171                 return 0;
 172
 173         /* How similar are they?
 174          * what percentage of material in dst are from source?
 175          */
 176         if (dst->size < src_copied)
 177                 score = MAX_SCORE;
 178         else if (!dst->size)
 179                 score = 0; /* should not happen */
 180         else
 181                 score = src_copied * MAX_SCORE / dst->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 *one, *two, *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         one = alloc_filespec(src->path);
 195         fill_filespec(one, src->sha1, src->mode);
 196
 197         dst = rename_dst[dst_index].two;
 198         two = alloc_filespec(dst->path);
 199         fill_filespec(two, dst->sha1, dst->mode);
 200
 201         dp = diff_queue(NULL, one, two);
 202         dp->score = score;
 203         dp->source_stays = rename_src[src_index].src_path_left;
 204         rename_dst[dst_index].pair = dp;
 205 }
 206
 207 /*
 208  * We sort the rename similarity matrix with the score, in descending
 209  * order (the most similar first).
 210  */
 211 static int score_compare(const void *a_, const void *b_)
 212 {
 213         const struct diff_score *a = a_, *b = b_;
 214         return b->score - a->score;
 215 }
 216
 217 static int compute_stays(struct diff_queue_struct *q,
 218                          struct diff_filespec *one)
 219 {
 220         int i;
 221         for (i = 0; i < q->nr; i++) {
 222                 struct diff_filepair *p = q->queue[i];
 223                 if (strcmp(one->path, p->two->path))
 224                         continue;
 225                 if (DIFF_PAIR_RENAME(p)) {
 226                         return 0; /* something else is renamed into this */
 227                 }
 228         }
 229         return 1;
 230 }
 231
 232 void diffcore_rename(struct diff_options *options)
 233 {
 234         int detect_rename = options->detect_rename;
 235         int minimum_score = options->rename_score;
 236         int rename_limit = options->rename_limit;
 237         struct diff_queue_struct *q = &diff_queued_diff;
 238         struct diff_queue_struct outq;
 239         struct diff_score *mx;
 240         int i, j, rename_count;
 241         int num_create, num_src, dst_cnt;
 242
 243         if (!minimum_score)
 244                 minimum_score = DEFAULT_RENAME_SCORE;
 245         rename_count = 0;
 246
 247         for (i = 0; i < q->nr; i++) {
 248                 struct diff_filepair *p = q->queue[i];
 249                 if (!DIFF_FILE_VALID(p->one))
 250                         if (!DIFF_FILE_VALID(p->two))
 251                                 continue; /* unmerged */
 252                         else
 253                                 locate_rename_dst(p->two, 1);
 254                 else if (!DIFF_FILE_VALID(p->two)) {
 255                         /* If the source is a broken "delete", and
 256                          * they did not really want to get broken,
 257                          * that means the source actually stays.
 258                          */
 259                         int stays = (p->broken_pair && !p->score);
 260                         register_rename_src(p->one, stays);
 261                 }
 262                 else if (detect_rename == DIFF_DETECT_COPY)
 263                         register_rename_src(p->one, 1);
 264         }
 265         if (rename_dst_nr == 0 || rename_src_nr == 0 ||
 266             (0 < rename_limit && rename_limit < rename_dst_nr))
 267                 goto cleanup; /* nothing to do */
 268
 269         /* We really want to cull the candidates list early
 270          * with cheap tests in order to avoid doing deltas.
 271          */
 272         for (i = 0; i < rename_dst_nr; i++) {
 273                 struct diff_filespec *two = rename_dst[i].two;
 274                 for (j = 0; j < rename_src_nr; j++) {
 275                         struct diff_filespec *one = rename_src[j].one;
 276                         if (!is_exact_match(one, two))
 277                                 continue;
 278                         record_rename_pair(i, j, MAX_SCORE);
 279                         rename_count++;
 280                         break; /* we are done with this entry */
 281                 }
 282         }
 283
 284         /* Have we run out the created file pool?  If so we can avoid
 285          * doing the delta matrix altogether.
 286          */
 287         if (rename_count == rename_dst_nr)
 288                 goto cleanup;
 289
 290         if (minimum_score == MAX_SCORE)
 291                 goto cleanup;
 292
 293         num_create = (rename_dst_nr - rename_count);
 294         num_src = rename_src_nr;
 295         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 296         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 297                 int base = dst_cnt * num_src;
 298                 struct diff_filespec *two = rename_dst[i].two;
 299                 if (rename_dst[i].pair)
 300                         continue; /* dealt with exact match already. */
 301                 for (j = 0; j < rename_src_nr; j++) {
 302                         struct diff_filespec *one = rename_src[j].one;
 303                         struct diff_score *m = &mx[base+j];
 304                         m->src = j;
 305                         m->dst = i;
 306                         m->score = estimate_similarity(one, two,
 307                                                        minimum_score);
 308                 }
 309                 dst_cnt++;
 310         }
 311         /* cost matrix sorted by most to least similar pair */
 312         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 313         for (i = 0; i < num_create * num_src; i++) {
 314                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 315                 if (dst->pair)
 316                         continue; /* already done, either exact or fuzzy. */
 317                 if (mx[i].score < minimum_score)
 318                         break; /* there is no more usable pair. */
 319                 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
 320                 rename_count++;
 321         }
 322         free(mx);
 323
 324  cleanup:
 325         /* At this point, we have found some renames and copies and they
 326          * are recorded in rename_dst.  The original list is still in *q.
 327          */
 328         outq.queue = NULL;
 329         outq.nr = outq.alloc = 0;
 330         for (i = 0; i < q->nr; i++) {
 331                 struct diff_filepair *p = q->queue[i];
 332                 struct diff_filepair *pair_to_free = NULL;
 333
 334                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 335                         /*
 336                          * Creation
 337                          *
 338                          * We would output this create record if it has
 339                          * not been turned into a rename/copy already.
 340                          */
 341                         struct diff_rename_dst *dst =
 342                                 locate_rename_dst(p->two, 0);
 343                         if (dst && dst->pair) {
 344                                 diff_q(&outq, dst->pair);
 345                                 pair_to_free = p;
 346                         }
 347                         else
 348                                 /* no matching rename/copy source, so
 349                                  * record this as a creation.
 350                                  */
 351                                 diff_q(&outq, p);
 352                 }
 353                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 354                         /*
 355                          * Deletion
 356                          *
 357                          * We would output this delete record if:
 358                          *
 359                          * (1) this is a broken delete and the counterpart
 360                          *     broken create remains in the output; or
 361                          * (2) this is not a broken delete, and rename_dst
 362                          *     does not have a rename/copy to move p->one->path
 363                          *     out of existence.
 364                          *
 365                          * Otherwise, the counterpart broken create
 366                          * has been turned into a rename-edit; or
 367                          * delete did not have a matching create to
 368                          * begin with.
 369                          */
 370                         if (DIFF_PAIR_BROKEN(p)) {
 371                                 /* broken delete */
 372                                 struct diff_rename_dst *dst =
 373                                         locate_rename_dst(p->one, 0);
 374                                 if (dst && dst->pair)
 375                                         /* counterpart is now rename/copy */
 376                                         pair_to_free = p;
 377                         }
 378                         else {
 379                                 for (j = 0; j < rename_dst_nr; j++) {
 380                                         if (!rename_dst[j].pair)
 381                                                 continue;
 382                                         if (strcmp(rename_dst[j].pair->
 383                                                    one->path,
 384                                                    p->one->path))
 385                                                 continue;
 386                                         break;
 387                                 }
 388                                 if (j < rename_dst_nr)
 389                                         /* this path remains */
 390                                         pair_to_free = p;
 391                         }
 392
 393                         if (pair_to_free)
 394                                 ;
 395                         else
 396                                 diff_q(&outq, p);
 397                 }
 398                 else if (!diff_unmodified_pair(p))
 399                         /* all the usual ones need to be kept */
 400                         diff_q(&outq, p);
 401                 else
 402                         /* no need to keep unmodified pairs */
 403                         pair_to_free = p;
 404
 405                 if (pair_to_free)
 406                         diff_free_filepair(pair_to_free);
 407         }
 408         diff_debug_queue("done copying original", &outq);
 409
 410         free(q->queue);
 411         *q = outq;
 412         diff_debug_queue("done collapsing", q);
 413
 414         /* We need to see which rename source really stays here;
 415          * earlier we only checked if the path is left in the result,
 416          * but even if a path remains in the result, if that is coming
 417          * from copying something else on top of it, then the original
 418          * source is lost and does not stay.
 419          */
 420         for (i = 0; i < q->nr; i++) {
 421                 struct diff_filepair *p = q->queue[i];
 422                 if (DIFF_PAIR_RENAME(p) && p->source_stays) {
 423                         /* If one appears as the target of a rename-copy,
 424                          * then mark p->source_stays = 0; otherwise
 425                          * leave it as is.
 426                          */
 427                         p->source_stays = compute_stays(q, p->one);
 428                 }
 429         }
 430
 431         for (i = 0; i < rename_dst_nr; i++) {
 432                 diff_free_filespec_data(rename_dst[i].two);
 433                 free(rename_dst[i].two);
 434         }
 435
 436         free(rename_dst);
 437         rename_dst = NULL;
 438         rename_dst_nr = rename_dst_alloc = 0;
 439         free(rename_src);
 440         rename_src = NULL;
 441         rename_src_nr = rename_src_alloc = 0;
 442         return;
 443 }