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 #include "delta.h"
   8 #include "count-delta.h"
   9
  10 /* Table of rename/copy destinations */
  11
  12 static struct diff_rename_dst {
  13         struct diff_filespec *two;
  14         struct diff_filepair *pair;
  15 } *rename_dst;
  16 static int rename_dst_nr, rename_dst_alloc;
  17
  18 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
  19                                                  int insert_ok)
  20 {
  21         int first, last;
  22
  23         first = 0;
  24         last = rename_dst_nr;
  25         while (last > first) {
  26                 int next = (last + first) >> 1;
  27                 struct diff_rename_dst *dst = &(rename_dst[next]);
  28                 int cmp = strcmp(two->path, dst->two->path);
  29                 if (!cmp)
  30                         return dst;
  31                 if (cmp < 0) {
  32                         last = next;
  33                         continue;
  34                 }
  35                 first = next+1;
  36         }
  37         /* not found */
  38         if (!insert_ok)
  39                 return NULL;
  40         /* insert to make it at "first" */
  41         if (rename_dst_alloc <= rename_dst_nr) {
  42                 rename_dst_alloc = alloc_nr(rename_dst_alloc);
  43                 rename_dst = xrealloc(rename_dst,
  44                                       rename_dst_alloc * sizeof(*rename_dst));
  45         }
  46         rename_dst_nr++;
  47         if (first < rename_dst_nr)
  48                 memmove(rename_dst + first + 1, rename_dst + first,
  49                         (rename_dst_nr - first - 1) * sizeof(*rename_dst));
  50         rename_dst[first].two = two;
  51         rename_dst[first].pair = NULL;
  52         return &(rename_dst[first]);
  53 }
  54
  55 /* Table of rename/copy src files */
  56 static struct diff_rename_src {
  57         struct diff_filespec *one;
  58         unsigned src_stays : 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_stays)
  64 {
  65         int first, last;
  66
  67         first = 0;
  68         last = rename_src_nr;
  69         while (last > first) {
  70                 int next = (last + first) >> 1;
  71                 struct diff_rename_src *src = &(rename_src[next]);
  72                 int cmp = strcmp(one->path, src->one->path);
  73                 if (!cmp)
  74                         return src;
  75                 if (cmp < 0) {
  76                         last = next;
  77                         continue;
  78                 }
  79                 first = next+1;
  80         }
  81
  82         /* insert to make it at "first" */
  83         if (rename_src_alloc <= rename_src_nr) {
  84                 rename_src_alloc = alloc_nr(rename_src_alloc);
  85                 rename_src = xrealloc(rename_src,
  86                                       rename_src_alloc * sizeof(*rename_src));
  87         }
  88         rename_src_nr++;
  89         if (first < rename_src_nr)
  90                 memmove(rename_src + first + 1, rename_src + first,
  91                         (rename_src_nr - first - 1) * sizeof(*rename_src));
  92         rename_src[first].one = one;
  93         rename_src[first].src_stays = src_stays;
  94         return &(rename_src[first]);
  95 }
  96
  97 static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
  98 {
  99         if (src->sha1_valid && dst->sha1_valid &&
 100             !memcmp(src->sha1, dst->sha1, 20))
 101                 return 1;
 102         if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
 103                 return 0;
 104         if (src->size != dst->size)
 105                 return 0;
 106         if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 107                 return 0;
 108         if (src->size == dst->size &&
 109             !memcmp(src->data, dst->data, src->size))
 110                 return 1;
 111         return 0;
 112 }
 113
 114 struct diff_score {
 115         int src; /* index in rename_src */
 116         int dst; /* index in rename_dst */
 117         int score;
 118 };
 119
 120 static int estimate_similarity(struct diff_filespec *src,
 121                                struct diff_filespec *dst,
 122                                int minimum_score)
 123 {
 124         /* src points at a file that existed in the original tree (or
 125          * optionally a file in the destination tree) and dst points
 126          * at a newly created file.  They may be quite similar, in which
 127          * case we want to say src is renamed to dst or src is copied into
 128          * dst, and then some edit has been applied to dst.
 129          *
 130          * Compare them and return how similar they are, representing
 131          * the score as an integer between 0 and MAX_SCORE.
 132          *
 133          * When there is an exact match, it is considered a better
 134          * match than anything else; the destination does not even
 135          * call into this function in that case.
 136          */
 137         void *delta;
 138         unsigned long delta_size, base_size, src_copied, literal_added;
 139         int score;
 140
 141         /* We deal only with regular files.  Symlink renames are handled
 142          * only when they are exact matches --- in other words, no edits
 143          * after renaming.
 144          */
 145         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 146                 return 0;
 147
 148         delta_size = ((src->size < dst->size) ?
 149                       (dst->size - src->size) : (src->size - dst->size));
 150         base_size = ((src->size < dst->size) ? src->size : dst->size);
 151
 152         /* We would not consider edits that change the file size so
 153          * drastically.  delta_size must be smaller than
 154          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 155          *
 156          * Note that base_size == 0 case is handled here already
 157          * and the final score computation below would not have a
 158          * divide-by-zero issue.
 159          */
 160         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 161                 return 0;
 162
 163         if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
 164                 return 0; /* error but caught downstream */
 165
 166         delta = diff_delta(src->data, src->size,
 167                            dst->data, dst->size,
 168                            &delta_size);
 169
 170         /* A delta that has a lot of literal additions would have
 171          * big delta_size no matter what else it does.
 172          */
 173         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 174                 return 0;
 175
 176         /* Estimate the edit size by interpreting delta. */
 177         if (count_delta(delta, delta_size, &src_copied, &literal_added)) {
 178                 free(delta);
 179                 return 0;
 180         }
 181         free(delta);
 182
 183         /* Extent of damage */
 184         if (src->size + literal_added < src_copied)
 185                 delta_size = 0;
 186         else
 187                 delta_size = (src->size - src_copied) + literal_added;
 188
 189         /*
 190          * Now we will give some score to it.  100% edit gets 0 points
 191          * and 0% edit gets MAX_SCORE points.
 192          */
 193         score = MAX_SCORE - (MAX_SCORE * delta_size / base_size);
 194         if (score < 0) return 0;
 195         if (MAX_SCORE < score) return MAX_SCORE;
 196         return score;
 197 }
 198
 199 static void record_rename_pair(struct diff_queue_struct *renq,
 200                                int dst_index, int src_index, int score)
 201 {
 202         struct diff_filespec *one, *two, *src, *dst;
 203         struct diff_filepair *dp;
 204
 205         if (rename_dst[dst_index].pair)
 206                 die("internal error: dst already matched.");
 207
 208         src = rename_src[src_index].one;
 209         one = alloc_filespec(src->path);
 210         fill_filespec(one, src->sha1, src->mode);
 211
 212         dst = rename_dst[dst_index].two;
 213         two = alloc_filespec(dst->path);
 214         fill_filespec(two, dst->sha1, dst->mode);
 215
 216         dp = diff_queue(renq, one, two);
 217         dp->score = score;
 218         dp->source_stays = rename_src[src_index].src_stays;
 219         rename_dst[dst_index].pair = dp;
 220 }
 221
 222 /*
 223  * We sort the rename similarity matrix with the score, in descending
 224  * order (the most similar first).
 225  */
 226 static int score_compare(const void *a_, const void *b_)
 227 {
 228         const struct diff_score *a = a_, *b = b_;
 229         return b->score - a->score;
 230 }
 231
 232 int diff_scoreopt_parse(const char *opt)
 233 {
 234         int diglen, num, scale, i;
 235         if (opt[0] != '-' || (opt[1] != 'M' && opt[1] != 'C' && opt[1] != 'B'))
 236                 return -1; /* that is not a -M, -C nor -B option */
 237         diglen = strspn(opt+2, "0123456789");
 238         if (diglen == 0 || strlen(opt+2) != diglen)
 239                 return 0; /* use default */
 240         sscanf(opt+2, "%d", &num);
 241         for (i = 0, scale = 1; i < diglen; i++)
 242                 scale *= 10;
 243
 244         /* user says num divided by scale and we say internally that
 245          * is MAX_SCORE * num / scale.
 246          */
 247         return MAX_SCORE * num / scale;
 248 }
 249
 250 void diffcore_rename(int detect_rename, int minimum_score)
 251 {
 252         struct diff_queue_struct *q = &diff_queued_diff;
 253         struct diff_queue_struct renq, outq;
 254         struct diff_score *mx;
 255         int i, j;
 256         int num_create, num_src, dst_cnt;
 257
 258         if (!minimum_score)
 259                 minimum_score = DEFAULT_RENAME_SCORE;
 260         renq.queue = NULL;
 261         renq.nr = renq.alloc = 0;
 262
 263         for (i = 0; i < q->nr; i++) {
 264                 struct diff_filepair *p = q->queue[i];
 265                 if (!DIFF_FILE_VALID(p->one))
 266                         if (!DIFF_FILE_VALID(p->two))
 267                                 continue; /* unmerged */
 268                         else
 269                                 locate_rename_dst(p->two, 1);
 270                 else if (!DIFF_FILE_VALID(p->two))
 271                         register_rename_src(p->one, 0);
 272                 else if (detect_rename == DIFF_DETECT_COPY)
 273                         register_rename_src(p->one, 1);
 274         }
 275         if (rename_dst_nr == 0)
 276                 goto cleanup; /* nothing to do */
 277
 278         /* We really want to cull the candidates list early
 279          * with cheap tests in order to avoid doing deltas.
 280          */
 281         for (i = 0; i < rename_dst_nr; i++) {
 282                 struct diff_filespec *two = rename_dst[i].two;
 283                 for (j = 0; j < rename_src_nr; j++) {
 284                         struct diff_filespec *one = rename_src[j].one;
 285                         if (!is_exact_match(one, two))
 286                                 continue;
 287                         record_rename_pair(&renq, i, j, MAX_SCORE);
 288                         break; /* we are done with this entry */
 289                 }
 290         }
 291         diff_debug_queue("done detecting exact", &renq);
 292
 293         /* Have we run out the created file pool?  If so we can avoid
 294          * doing the delta matrix altogether.
 295          */
 296         if (renq.nr == rename_dst_nr)
 297                 goto cleanup;
 298
 299         num_create = (rename_dst_nr - renq.nr);
 300         num_src = rename_src_nr;
 301         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 302         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 303                 int base = dst_cnt * num_src;
 304                 struct diff_filespec *two = rename_dst[i].two;
 305                 if (rename_dst[i].pair)
 306                         continue; /* dealt with exact match already. */
 307                 for (j = 0; j < rename_src_nr; j++) {
 308                         struct diff_filespec *one = rename_src[j].one;
 309                         struct diff_score *m = &mx[base+j];
 310                         m->src = j;
 311                         m->dst = i;
 312                         m->score = estimate_similarity(one, two,
 313                                                        minimum_score);
 314                 }
 315                 dst_cnt++;
 316         }
 317         /* cost matrix sorted by most to least similar pair */
 318         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 319         for (i = 0; i < num_create * num_src; i++) {
 320                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 321                 if (dst->pair)
 322                         continue; /* already done, either exact or fuzzy. */
 323                 if (mx[i].score < minimum_score)
 324                         break; /* there is no more usable pair. */
 325                 record_rename_pair(&renq, mx[i].dst, mx[i].src, mx[i].score);
 326         }
 327         free(mx);
 328         diff_debug_queue("done detecting fuzzy", &renq);
 329
 330  cleanup:
 331         /* At this point, we have found some renames and copies and they
 332          * are kept in renq.  The original list is still in *q.
 333          */
 334         outq.queue = NULL;
 335         outq.nr = outq.alloc = 0;
 336         for (i = 0; i < q->nr; i++) {
 337                 struct diff_filepair *p = q->queue[i];
 338                 struct diff_filepair *pair_to_free = NULL;
 339
 340                 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
 341                         /*
 342                          * Creation
 343                          *
 344                          * We would output this create record if it has
 345                          * not been turned into a rename/copy already.
 346                          */
 347                         struct diff_rename_dst *dst =
 348                                 locate_rename_dst(p->two, 0);
 349                         if (dst && dst->pair) {
 350                                 diff_q(&outq, dst->pair);
 351                                 pair_to_free = p;
 352                         }
 353                         else
 354                                 /* no matching rename/copy source, so
 355                                  * record this as a creation.
 356                                  */
 357                                 diff_q(&outq, p);
 358                 }
 359                 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
 360                         /*
 361                          * Deletion
 362                          *
 363                          * We would output this delete record if:
 364                          *
 365                          * (1) this is a broken delete and the counterpart
 366                          *     broken create remains in the output; or
 367                          * (2) this is not a broken delete, and renq does
 368                          *     not have a rename/copy to move p->one->path
 369                          *     out.
 370                          *
 371                          * Otherwise, the counterpart broken create
 372                          * has been turned into a rename-edit; or
 373                          * delete did not have a matching create to
 374                          * begin with.
 375                          */
 376                         if (DIFF_PAIR_BROKEN(p)) {
 377                                 /* broken delete */
 378                                 struct diff_rename_dst *dst =
 379                                         locate_rename_dst(p->one, 0);
 380                                 if (dst && dst->pair)
 381                                         /* counterpart is now rename/copy */
 382                                         pair_to_free = p;
 383                         }
 384                         else {
 385                                 for (j = 0; j < renq.nr; j++)
 386                                         if (!strcmp(renq.queue[j]->one->path,
 387                                                     p->one->path))
 388                                                 break;
 389                                 if (j < renq.nr)
 390                                         /* this path remains */
 391                                         pair_to_free = p;
 392                         }
 393
 394                         if (pair_to_free)
 395                                 ;
 396                         else
 397                                 diff_q(&outq, p);
 398                 }
 399                 else if (!diff_unmodified_pair(p))
 400                         /* all the usual ones need to be kept */
 401                         diff_q(&outq, p);
 402                 else
 403                         /* no need to keep unmodified pairs */
 404                         pair_to_free = p;
 405
 406                 if (pair_to_free)
 407                         diff_free_filepair(pair_to_free);
 408         }
 409         diff_debug_queue("done copying original", &outq);
 410
 411         free(renq.queue);
 412         free(q->queue);
 413         *q = outq;
 414         diff_debug_queue("done collapsing", q);
 415
 416         free(rename_dst);
 417         rename_dst = NULL;
 418         rename_dst_nr = rename_dst_alloc = 0;
 419         free(rename_src);
 420         rename_src = NULL;
 421         rename_src_nr = rename_src_alloc = 0;
 422         return;
 423 }