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) || diff_populate_filespec(dst))
 103                 /* this is an error but will be caught downstream */
 104                 return 0;
 105         if (src->size == dst->size &&
 106             !memcmp(src->data, dst->data, src->size))
 107                 return 1;
 108         return 0;
 109 }
 110
 111 struct diff_score {
 112         int src; /* index in rename_src */
 113         int dst; /* index in rename_dst */
 114         int score;
 115 };
 116
 117 static int estimate_similarity(struct diff_filespec *src,
 118                                struct diff_filespec *dst,
 119                                int minimum_score)
 120 {
 121         /* src points at a file that existed in the original tree (or
 122          * optionally a file in the destination tree) and dst points
 123          * at a newly created file.  They may be quite similar, in which
 124          * case we want to say src is renamed to dst or src is copied into
 125          * dst, and then some edit has been applied to dst.
 126          *
 127          * Compare them and return how similar they are, representing
 128          * the score as an integer between 0 and 10000, except
 129          * where they match exactly it is considered better than anything
 130          * else.
 131          */
 132         void *delta;
 133         unsigned long delta_size, base_size;
 134         int score;
 135
 136         /* We deal only with regular files.  Symlink renames are handled
 137          * only when they are exact matches --- in other words, no edits
 138          * after renaming.
 139          */
 140         if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
 141                 return 0;
 142
 143         delta_size = ((src->size < dst->size) ?
 144                       (dst->size - src->size) : (src->size - dst->size));
 145         base_size = ((src->size < dst->size) ? src->size : dst->size);
 146
 147         /* We would not consider edits that change the file size so
 148          * drastically.  delta_size must be smaller than
 149          * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
 150          * Note that base_size == 0 case is handled here already
 151          * and the final score computation below would not have a
 152          * divide-by-zero issue.
 153          */
 154         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 155                 return 0;
 156
 157         delta = diff_delta(src->data, src->size,
 158                            dst->data, dst->size,
 159                            &delta_size);
 160
 161         /* A delta that has a lot of literal additions would have
 162          * big delta_size no matter what else it does.
 163          */
 164         if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
 165                 return 0;
 166
 167         /* Estimate the edit size by interpreting delta. */
 168         delta_size = count_delta(delta, delta_size);
 169         free(delta);
 170         if (delta_size == UINT_MAX)
 171                 return 0;
 172
 173         /*
 174          * Now we will give some score to it.  100% edit gets 0 points
 175          * and 0% edit gets MAX_SCORE points.
 176          */
 177         score = MAX_SCORE - (MAX_SCORE * delta_size / base_size);
 178         if (score < 0) return 0;
 179         if (MAX_SCORE < score) return MAX_SCORE;
 180         return score;
 181 }
 182
 183 static void record_rename_pair(struct diff_queue_struct *renq,
 184                                int dst_index, int src_index, int score)
 185 {
 186         struct diff_filespec *one, *two, *src, *dst;
 187         struct diff_filepair *dp;
 188
 189         if (rename_dst[dst_index].pair)
 190                 die("internal error: dst already matched.");
 191
 192         src = rename_src[src_index].one;
 193         one = alloc_filespec(src->path);
 194         fill_filespec(one, src->sha1, src->mode);
 195
 196         dst = rename_dst[dst_index].two;
 197         two = alloc_filespec(dst->path);
 198         fill_filespec(two, dst->sha1, dst->mode);
 199
 200         dp = diff_queue(renq, one, two);
 201         dp->score = score ? : 1; /* make sure it is at least 1 */
 202         dp->source_stays = rename_src[src_index].src_stays;
 203         rename_dst[dst_index].pair = dp;
 204 }
 205
 206 /*
 207  * We sort the rename similarity matrix with the score, in descending
 208  * order (the most similar first).
 209  */
 210 static int score_compare(const void *a_, const void *b_)
 211 {
 212         const struct diff_score *a = a_, *b = b_;
 213         return b->score - a->score;
 214 }
 215
 216 int diff_scoreopt_parse(const char *opt)
 217 {
 218         int diglen, num, scale, i;
 219         if (opt[0] != '-' || (opt[1] != 'M' && opt[1] != 'C'))
 220                 return -1; /* that is not a -M nor -C option */
 221         diglen = strspn(opt+2, "0123456789");
 222         if (diglen == 0 || strlen(opt+2) != diglen)
 223                 return 0; /* use default */
 224         sscanf(opt+2, "%d", &num);
 225         for (i = 0, scale = 1; i < diglen; i++)
 226                 scale *= 10;
 227
 228         /* user says num divided by scale and we say internally that
 229          * is MAX_SCORE * num / scale.
 230          */
 231         return MAX_SCORE * num / scale;
 232 }
 233
 234 void diffcore_rename(int detect_rename, int minimum_score)
 235 {
 236         struct diff_queue_struct *q = &diff_queued_diff;
 237         struct diff_queue_struct renq, outq;
 238         struct diff_score *mx;
 239         int i, j;
 240         int num_create, num_src, dst_cnt;
 241
 242         if (!minimum_score)
 243                 minimum_score = DEFAULT_MINIMUM_SCORE;
 244         renq.queue = NULL;
 245         renq.nr = renq.alloc = 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                         register_rename_src(p->one, 0);
 256                 else if (detect_rename == DIFF_DETECT_COPY)
 257                         register_rename_src(p->one, 1);
 258         }
 259         if (rename_dst_nr == 0)
 260                 goto cleanup; /* nothing to do */
 261
 262         /* We really want to cull the candidates list early
 263          * with cheap tests in order to avoid doing deltas.
 264          */
 265         for (i = 0; i < rename_dst_nr; i++) {
 266                 struct diff_filespec *two = rename_dst[i].two;
 267                 for (j = 0; j < rename_src_nr; j++) {
 268                         struct diff_filespec *one = rename_src[j].one;
 269                         if (!is_exact_match(one, two))
 270                                 continue;
 271                         record_rename_pair(&renq, i, j, MAX_SCORE);
 272                         break; /* we are done with this entry */
 273                 }
 274         }
 275         diff_debug_queue("done detecting exact", &renq);
 276
 277         /* Have we run out the created file pool?  If so we can avoid
 278          * doing the delta matrix altogether.
 279          */
 280         if (renq.nr == rename_dst_nr)
 281                 goto cleanup;
 282
 283         num_create = (rename_dst_nr - renq.nr);
 284         num_src = rename_src_nr;
 285         mx = xmalloc(sizeof(*mx) * num_create * num_src);
 286         for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
 287                 int base = dst_cnt * num_src;
 288                 struct diff_filespec *two = rename_dst[i].two;
 289                 if (rename_dst[i].pair)
 290                         continue; /* dealt with exact match already. */
 291                 for (j = 0; j < rename_src_nr; j++) {
 292                         struct diff_filespec *one = rename_src[j].one;
 293                         struct diff_score *m = &mx[base+j];
 294                         m->src = j;
 295                         m->dst = i;
 296                         m->score = estimate_similarity(one, two,
 297                                                        minimum_score);
 298                 }
 299                 dst_cnt++;
 300         }
 301         /* cost matrix sorted by most to least similar pair */
 302         qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
 303         for (i = 0; i < num_create * num_src; i++) {
 304                 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
 305                 if (dst->pair)
 306                         continue; /* already done, either exact or fuzzy. */
 307                 if (mx[i].score < minimum_score)
 308                         break; /* there is no more usable pair. */
 309                 record_rename_pair(&renq, mx[i].dst, mx[i].src, mx[i].score);
 310         }
 311         free(mx);
 312         diff_debug_queue("done detecting fuzzy", &renq);
 313
 314  cleanup:
 315         /* At this point, we have found some renames and copies and they
 316          * are kept in renq.  The original list is still in *q.
 317          */
 318         outq.queue = NULL;
 319         outq.nr = outq.alloc = 0;
 320         for (i = 0; i < q->nr; i++) {
 321                 struct diff_filepair *p = q->queue[i];
 322                 struct diff_rename_dst *dst = locate_rename_dst(p->two, 0);
 323                 struct diff_filepair *pair_to_free = NULL;
 324
 325                 if (dst) {
 326                         /* creation */
 327                         if (dst->pair) {
 328                                 /* renq has rename/copy to produce
 329                                  * this file already, so we do not
 330                                  * emit the creation record in the
 331                                  * output.
 332                                  */
 333                                 diff_q(&outq, dst->pair);
 334                                 pair_to_free = p;
 335                         }
 336                         else
 337                                 /* no matching rename/copy source, so record
 338                                  * this as a creation.
 339                                  */
 340                                 diff_q(&outq, p);
 341                 }
 342                 else if (!diff_unmodified_pair(p))
 343                         /* all the usual ones need to be kept */
 344                         diff_q(&outq, p);
 345                 else
 346                         /* no need to keep unmodified pairs */
 347                         pair_to_free = p;
 348
 349                 if (pair_to_free)
 350                         diff_free_filepair(pair_to_free);
 351         }
 352         diff_debug_queue("done copying original", &outq);
 353
 354         free(renq.queue);
 355         free(q->queue);
 356         *q = outq;
 357         diff_debug_queue("done collapsing", q);
 358
 359         free(rename_dst);
 360         rename_dst = NULL;
 361         rename_dst_nr = rename_dst_alloc = 0;
 362         free(rename_src);
 363         rename_src = NULL;
 364         rename_src_nr = rename_src_alloc = 0;
 365         return;
 366 }