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;
 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         delta_size = count_delta(delta, delta_size);
 178         free(delta);
 179         if (delta_size == UINT_MAX)
 180                 return 0;
 181
 182         /*
 183          * Now we will give some score to it.  100% edit gets 0 points
 184          * and 0% edit gets MAX_SCORE points.
 185          */
 186         score = MAX_SCORE - (MAX_SCORE * delta_size / base_size);
 187         if (score < 0) return 0;
 188         if (MAX_SCORE < score) return MAX_SCORE;
 189         return score;
 190 }
 191
 192 static void record_rename_pair(struct diff_queue_struct *renq,
 193                                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(renq, one, two);
 210         dp->score = score ? : 1; /* make sure it is at least 1 */
 211         dp->source_stays = rename_src[src_index].src_stays;
 212         rename_dst[dst_index].pair = dp;
 213 }
 214
 215 /*
 216  * We sort the rename similarity matrix with the score, in descending
 217  * order (the most similar first).
 218  */
 219 static int score_compare(const void *a_, const void *b_)
 220 {
 221         const struct diff_score *a = a_, *b = b_;
 222         return b->score - a->score;
 223 }
 224
 225 int diff_scoreopt_parse(const char *opt)
 226 {
 227         int diglen, num, scale, i;
 228         if (opt[0] != '-' || (opt[1] != 'M' && opt[1] != 'C'))
 229                 return -1; /* that is not a -M nor -C option */
 230         diglen = strspn(opt+2, "0123456789");
 231         if (diglen == 0 || strlen(opt+2) != diglen)
 232                 return 0; /* use default */
 233         sscanf(opt+2, "%d", &num);
 234         for (i = 0, scale = 1; i < diglen; i++)
 235                 scale *= 10;
 236
 237         /* user says num divided by scale and we say internally that
 238          * is MAX_SCORE * num / scale.
 239          */
 240         return MAX_SCORE * num / scale;
 241 }
 242
 243 void diffcore_rename(int detect_rename, int minimum_score)
 244 {
 245         struct diff_queue_struct *q = &diff_queued_diff;
 246         struct diff_queue_struct renq, outq;
 247         struct diff_score *mx;
 248         int i, j;
 249         int num_create, num_src, dst_cnt;
 250
 251         if (!minimum_score)
 252                 minimum_score = DEFAULT_MINIMUM_SCORE;
 253         renq.queue = NULL;
 254         renq.nr = renq.alloc = 0;
 255
 256         for (i = 0; i < q->nr; i++) {
 257                 struct diff_filepair *p = q->queue[i];
 258                 if (!DIFF_FILE_VALID(p->one))
 259                         if (!DIFF_FILE_VALID(p->two))
 260                                 continue; /* unmerged */
 261                         else
 262                                 locate_rename_dst(p->two, 1);
 263                 else if (!DIFF_FILE_VALID(p->two))
 264                         register_rename_src(p->one, 0);
 265                 else if (detect_rename == DIFF_DETECT_COPY)
 266                         register_rename_src(p->one, 1);
 267         }
 268         if (rename_dst_nr == 0)
 269                 goto cleanup; /* nothing to do */
 270
 271         /* We really want to cull the candidates list early
 272          * with cheap tests in order to avoid doing deltas.
 273          */
 274         for (i = 0; i < rename_dst_nr; i++) {
 275                 struct diff_filespec *two = rename_dst[i].two;
 276                 for (j = 0; j < rename_src_nr; j++) {
 277                         struct diff_filespec *one = rename_src[j].one;
 278                         if (!is_exact_match(one, two))
 279                                 continue;
 280                         record_rename_pair(&renq, i, j, MAX_SCORE);
 281                         break; /* we are done with this entry */
 282                 }
 283         }
 284         diff_debug_queue("done detecting exact", &renq);
 285
 286         /* Have we run out the created file pool?  If so we can avoid
 287          * doing the delta matrix altogether.
 288          */
 289         if (renq.nr == rename_dst_nr)
 290                 goto cleanup;
 291
 292         num_create = (rename_dst_nr - renq.nr);
 293         num_src = rename_src_nr;
 294         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 295         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 296                 int base = dst_cnt * num_src;
 297                 struct diff_filespec *two = rename_dst[i].two;
 298                 if (rename_dst[i].pair)
 299                         continue; /* dealt with exact match already. */
 300                 for (j = 0; j < rename_src_nr; j++) {
 301                         struct diff_filespec *one = rename_src[j].one;
 302                         struct diff_score *m = &mx[base+j];
 303                         m->src = j;
 304                         m->dst = i;
 305                         m->score = estimate_similarity(one, two,
 306                                                        minimum_score);
 307                 }
 308                 dst_cnt++;
 309         }
 310         /* cost matrix sorted by most to least similar pair */
 311         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 312         for (i = 0; i < num_create * num_src; i++) {
 313                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 314                 if (dst->pair)
 315                         continue; /* already done, either exact or fuzzy. */
 316                 if (mx[i].score < minimum_score)
 317                         break; /* there is no more usable pair. */
 318                 record_rename_pair(&renq, mx[i].dst, mx[i].src, mx[i].score);
 319         }
 320         free(mx);
 321         diff_debug_queue("done detecting fuzzy", &renq);
 322
 323  cleanup:
 324         /* At this point, we have found some renames and copies and they
 325          * are kept in renq.  The original list is still in *q.
 326          */
 327         outq.queue = NULL;
 328         outq.nr = outq.alloc = 0;
 329         for (i = 0; i < q->nr; i++) {
 330                 struct diff_filepair *p = q->queue[i];
 331                 struct diff_rename_dst *dst = locate_rename_dst(p->two, 0);
 332                 struct diff_filepair *pair_to_free = NULL;
 333
 334                 if (dst) {
 335                         /* creation */
 336                         if (dst->pair) {
 337                                 /* renq has rename/copy to produce
 338                                  * this file already, so we do not
 339                                  * emit the creation record in the
 340                                  * output.
 341                                  */
 342                                 diff_q(&outq, dst->pair);
 343                                 pair_to_free = p;
 344                         }
 345                         else
 346                                 /* no matching rename/copy source, so record
 347                                  * this as a creation.
 348                                  */
 349                                 diff_q(&outq, p);
 350                 }
 351                 else if (!diff_unmodified_pair(p))
 352                         /* all the usual ones need to be kept */
 353                         diff_q(&outq, p);
 354                 else
 355                         /* no need to keep unmodified pairs */
 356                         pair_to_free = p;
 357
 358                 if (pair_to_free)
 359                         diff_free_filepair(pair_to_free);
 360         }
 361         diff_debug_queue("done copying original", &outq);
 362
 363         free(renq.queue);
 364         free(q->queue);
 365         *q = outq;
 366         diff_debug_queue("done collapsing", q);
 367
 368         free(rename_dst);
 369         rename_dst = NULL;
 370         rename_dst_nr = rename_dst_alloc = 0;
 371         free(rename_src);
 372         rename_src = NULL;
 373         rename_src_nr = rename_src_alloc = 0;
 374         return;
 375 }