diff-lib.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include <sys/types.h>
   5 #include <sys/wait.h>
   6 #include <signal.h>
   7 #include "cache.h"
   8 #include "quote.h"
   9 #include "diff.h"
  10 #include "diffcore.h"
  11 #include "xdiff-interface.h"
  12
  13 static int use_size_cache;
  14
  15 int diff_rename_limit_default = -1;
  16
  17 int git_diff_config(const char *var, const char *value)
  18 {
  19         if (!strcmp(var, "diff.renamelimit")) {
  20                 diff_rename_limit_default = git_config_int(var, value);
  21                 return 0;
  22         }
  23
  24         return git_default_config(var, value);
  25 }
  26
  27 static char *quote_one(const char *str)
  28 {
  29         int needlen;
  30         char *xp;
  31
  32         if (!str)
  33                 return NULL;
  34         needlen = quote_c_style(str, NULL, NULL, 0);
  35         if (!needlen)
  36                 return strdup(str);
  37         xp = xmalloc(needlen + 1);
  38         quote_c_style(str, xp, NULL, 0);
  39         return xp;
  40 }
  41
  42 static char *quote_two(const char *one, const char *two)
  43 {
  44         int need_one = quote_c_style(one, NULL, NULL, 1);
  45         int need_two = quote_c_style(two, NULL, NULL, 1);
  46         char *xp;
  47
  48         if (need_one + need_two) {
  49                 if (!need_one) need_one = strlen(one);
  50                 if (!need_two) need_one = strlen(two);
  51
  52                 xp = xmalloc(need_one + need_two + 3);
  53                 xp[0] = '"';
  54                 quote_c_style(one, xp + 1, NULL, 1);
  55                 quote_c_style(two, xp + need_one + 1, NULL, 1);
  56                 strcpy(xp + need_one + need_two + 1, "\"");
  57                 return xp;
  58         }
  59         need_one = strlen(one);
  60         need_two = strlen(two);
  61         xp = xmalloc(need_one + need_two + 1);
  62         strcpy(xp, one);
  63         strcpy(xp + need_one, two);
  64         return xp;
  65 }
  66
  67 static const char *external_diff(void)
  68 {
  69         static const char *external_diff_cmd = NULL;
  70         static int done_preparing = 0;
  71
  72         if (done_preparing)
  73                 return external_diff_cmd;
  74         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  75         done_preparing = 1;
  76         return external_diff_cmd;
  77 }
  78
  79 #define TEMPFILE_PATH_LEN               50
  80
  81 static struct diff_tempfile {
  82         const char *name; /* filename external diff should read from */
  83         char hex[41];
  84         char mode[10];
  85         char tmp_path[TEMPFILE_PATH_LEN];
  86 } diff_temp[2];
  87
  88 static int count_lines(const char *data, int size)
  89 {
  90         int count, ch, completely_empty = 1, nl_just_seen = 0;
  91         count = 0;
  92         while (0 < size--) {
  93                 ch = *data++;
  94                 if (ch == '\n') {
  95                         count++;
  96                         nl_just_seen = 1;
  97                         completely_empty = 0;
  98                 }
  99                 else {
 100                         nl_just_seen = 0;
 101                         completely_empty = 0;
 102                 }
 103         }
 104         if (completely_empty)
 105                 return 0;
 106         if (!nl_just_seen)
 107                 count++; /* no trailing newline */
 108         return count;
 109 }
 110
 111 static void print_line_count(int count)
 112 {
 113         switch (count) {
 114         case 0:
 115                 printf("0,0");
 116                 break;
 117         case 1:
 118                 printf("1");
 119                 break;
 120         default:
 121                 printf("1,%d", count);
 122                 break;
 123         }
 124 }
 125
 126 static void copy_file(int prefix, const char *data, int size)
 127 {
 128         int ch, nl_just_seen = 1;
 129         while (0 < size--) {
 130                 ch = *data++;
 131                 if (nl_just_seen)
 132                         putchar(prefix);
 133                 putchar(ch);
 134                 if (ch == '\n')
 135                         nl_just_seen = 1;
 136                 else
 137                         nl_just_seen = 0;
 138         }
 139         if (!nl_just_seen)
 140                 printf("\n\\ No newline at end of file\n");
 141 }
 142
 143 static void emit_rewrite_diff(const char *name_a,
 144                               const char *name_b,
 145                               struct diff_filespec *one,
 146                               struct diff_filespec *two)
 147 {
 148         int lc_a, lc_b;
 149         diff_populate_filespec(one, 0);
 150         diff_populate_filespec(two, 0);
 151         lc_a = count_lines(one->data, one->size);
 152         lc_b = count_lines(two->data, two->size);
 153         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 154         print_line_count(lc_a);
 155         printf(" +");
 156         print_line_count(lc_b);
 157         printf(" @@\n");
 158         if (lc_a)
 159                 copy_file('-', one->data, one->size);
 160         if (lc_b)
 161                 copy_file('+', two->data, two->size);
 162 }
 163
 164 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 165 {
 166         if (!DIFF_FILE_VALID(one)) {
 167                 mf->ptr = ""; /* does not matter */
 168                 mf->size = 0;
 169                 return 0;
 170         }
 171         else if (diff_populate_filespec(one, 0))
 172                 return -1;
 173         mf->ptr = one->data;
 174         mf->size = one->size;
 175         return 0;
 176 }
 177
 178 struct emit_callback {
 179         const char **label_path;
 180 };
 181
 182 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
 183 {
 184         int i;
 185         struct emit_callback *ecbdata = priv;
 186
 187         if (ecbdata->label_path[0]) {
 188                 printf("--- %s\n", ecbdata->label_path[0]);
 189                 printf("+++ %s\n", ecbdata->label_path[1]);
 190                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 191         }
 192         for (i = 0; i < nbuf; i++)
 193                 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
 194                         return -1;
 195         return 0;
 196 }
 197
 198 static char *pprint_rename(const char *a, const char *b)
 199 {
 200         const char *old = a;
 201         const char *new = b;
 202         char *name = NULL;
 203         int pfx_length, sfx_length;
 204         int len_a = strlen(a);
 205         int len_b = strlen(b);
 206
 207         /* Find common prefix */
 208         pfx_length = 0;
 209         while (*old && *new && *old == *new) {
 210                 if (*old == '/')
 211                         pfx_length = old - a + 1;
 212                 old++;
 213                 new++;
 214         }
 215
 216         /* Find common suffix */
 217         old = a + len_a;
 218         new = b + len_b;
 219         sfx_length = 0;
 220         while (a <= old && b <= new && *old == *new) {
 221                 if (*old == '/')
 222                         sfx_length = len_a - (old - a);
 223                 old--;
 224                 new--;
 225         }
 226
 227         /*
 228          * pfx{mid-a => mid-b}sfx
 229          * {pfx-a => pfx-b}sfx
 230          * pfx{sfx-a => sfx-b}
 231          * name-a => name-b
 232          */
 233         if (pfx_length + sfx_length) {
 234                 name = xmalloc(len_a + len_b - pfx_length - sfx_length + 7);
 235                 sprintf(name, "%.*s{%.*s => %.*s}%s",
 236                         pfx_length, a,
 237                         len_a - pfx_length - sfx_length, a + pfx_length,
 238                         len_b - pfx_length - sfx_length, b + pfx_length,
 239                         a + len_a - sfx_length);
 240         }
 241         else {
 242                 name = xmalloc(len_a + len_b + 5);
 243                 sprintf(name, "%s => %s", a, b);
 244         }
 245         return name;
 246 }
 247
 248 struct diffstat_t {
 249         struct xdiff_emit_state xm;
 250
 251         int nr;
 252         int alloc;
 253         struct diffstat_file {
 254                 char *name;
 255                 unsigned is_unmerged:1;
 256                 unsigned is_binary:1;
 257                 unsigned is_renamed:1;
 258                 unsigned int added, deleted;
 259         } **files;
 260 };
 261
 262 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 263                                           const char *name_a,
 264                                           const char *name_b)
 265 {
 266         struct diffstat_file *x;
 267         x = xcalloc(sizeof (*x), 1);
 268         if (diffstat->nr == diffstat->alloc) {
 269                 diffstat->alloc = alloc_nr(diffstat->alloc);
 270                 diffstat->files = xrealloc(diffstat->files,
 271                                 diffstat->alloc * sizeof(x));
 272         }
 273         diffstat->files[diffstat->nr++] = x;
 274         if (name_b) {
 275                 x->name = pprint_rename(name_a, name_b);
 276                 x->is_renamed = 1;
 277         }
 278         else
 279                 x->name = strdup(name_a);
 280         return x;
 281 }
 282
 283 static void diffstat_consume(void *priv, char *line, unsigned long len)
 284 {
 285         struct diffstat_t *diffstat = priv;
 286         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 287
 288         if (line[0] == '+')
 289                 x->added++;
 290         else if (line[0] == '-')
 291                 x->deleted++;
 292 }
 293
 294 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
 295 static const char minuses[]= "----------------------------------------------------------------------";
 296
 297 static void show_stats(struct diffstat_t* data)
 298 {
 299         char *prefix = "";
 300         int i, len, add, del, total, adds = 0, dels = 0;
 301         int max, max_change = 0, max_len = 0;
 302         int total_files = data->nr;
 303
 304         if (data->nr == 0)
 305                 return;
 306
 307         for (i = 0; i < data->nr; i++) {
 308                 struct diffstat_file *file = data->files[i];
 309
 310                 len = strlen(file->name);
 311                 if (max_len < len)
 312                         max_len = len;
 313
 314                 if (file->is_binary || file->is_unmerged)
 315                         continue;
 316                 if (max_change < file->added + file->deleted)
 317                         max_change = file->added + file->deleted;
 318         }
 319
 320         for (i = 0; i < data->nr; i++) {
 321                 char *name = data->files[i]->name;
 322                 int added = data->files[i]->added;
 323                 int deleted = data->files[i]->deleted;
 324
 325                 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
 326                         char *qname = xmalloc(len + 1);
 327                         quote_c_style(name, qname, NULL, 0);
 328                         free(name);
 329                         data->files[i]->name = name = qname;
 330                 }
 331
 332                 /*
 333                  * "scale" the filename
 334                  */
 335                 len = strlen(name);
 336                 max = max_len;
 337                 if (max > 50)
 338                         max = 50;
 339                 if (len > max) {
 340                         char *slash;
 341                         prefix = "...";
 342                         max -= 3;
 343                         name += len - max;
 344                         slash = strchr(name, '/');
 345                         if (slash)
 346                                 name = slash;
 347                 }
 348                 len = max;
 349
 350                 /*
 351                  * scale the add/delete
 352                  */
 353                 max = max_change;
 354                 if (max + len > 70)
 355                         max = 70 - len;
 356
 357                 if (data->files[i]->is_binary) {
 358                         printf(" %s%-*s |  Bin\n", prefix, len, name);
 359                         goto free_diffstat_file;
 360                 }
 361                 else if (data->files[i]->is_unmerged) {
 362                         printf(" %s%-*s |  Unmerged\n", prefix, len, name);
 363                         goto free_diffstat_file;
 364                 }
 365                 else if (!data->files[i]->is_renamed &&
 366                          (added + deleted == 0)) {
 367                         total_files--;
 368                         goto free_diffstat_file;
 369                 }
 370
 371                 add = added;
 372                 del = deleted;
 373                 total = add + del;
 374                 adds += add;
 375                 dels += del;
 376
 377                 if (max_change > 0) {
 378                         total = (total * max + max_change / 2) / max_change;
 379                         add = (add * max + max_change / 2) / max_change;
 380                         del = total - add;
 381                 }
 382                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
 383                                 len, name, added + deleted,
 384                                 add, pluses, del, minuses);
 385         free_diffstat_file:
 386                 free(data->files[i]->name);
 387                 free(data->files[i]);
 388         }
 389         free(data->files);
 390         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 391                         total_files, adds, dels);
 392 }
 393
 394 #define FIRST_FEW_BYTES 8000
 395 static int mmfile_is_binary(mmfile_t *mf)
 396 {
 397         long sz = mf->size;
 398         if (FIRST_FEW_BYTES < sz)
 399                 sz = FIRST_FEW_BYTES;
 400         if (memchr(mf->ptr, 0, sz))
 401                 return 1;
 402         return 0;
 403 }
 404
 405 static void builtin_diff(const char *name_a,
 406                          const char *name_b,
 407                          struct diff_filespec *one,
 408                          struct diff_filespec *two,
 409                          const char *xfrm_msg,
 410                          int complete_rewrite)
 411 {
 412         mmfile_t mf1, mf2;
 413         const char *lbl[2];
 414         char *a_one, *b_two;
 415
 416         a_one = quote_two("a/", name_a);
 417         b_two = quote_two("b/", name_b);
 418         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
 419         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
 420         printf("diff --git %s %s\n", a_one, b_two);
 421         if (lbl[0][0] == '/') {
 422                 /* /dev/null */
 423                 printf("new file mode %06o\n", two->mode);
 424                 if (xfrm_msg && xfrm_msg[0])
 425                         puts(xfrm_msg);
 426         }
 427         else if (lbl[1][0] == '/') {
 428                 printf("deleted file mode %06o\n", one->mode);
 429                 if (xfrm_msg && xfrm_msg[0])
 430                         puts(xfrm_msg);
 431         }
 432         else {
 433                 if (one->mode != two->mode) {
 434                         printf("old mode %06o\n", one->mode);
 435                         printf("new mode %06o\n", two->mode);
 436                 }
 437                 if (xfrm_msg && xfrm_msg[0])
 438                         puts(xfrm_msg);
 439                 /*
 440                  * we do not run diff between different kind
 441                  * of objects.
 442                  */
 443                 if ((one->mode ^ two->mode) & S_IFMT)
 444                         goto free_ab_and_return;
 445                 if (complete_rewrite) {
 446                         emit_rewrite_diff(name_a, name_b, one, two);
 447                         goto free_ab_and_return;
 448                 }
 449         }
 450
 451         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 452                 die("unable to read files to diff");
 453
 454         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 455                 printf("Binary files %s and %s differ\n", lbl[0], lbl[1]);
 456         else {
 457                 /* Crazy xdl interfaces.. */
 458                 const char *diffopts = getenv("GIT_DIFF_OPTS");
 459                 xpparam_t xpp;
 460                 xdemitconf_t xecfg;
 461                 xdemitcb_t ecb;
 462                 struct emit_callback ecbdata;
 463
 464                 ecbdata.label_path = lbl;
 465                 xpp.flags = XDF_NEED_MINIMAL;
 466                 xecfg.ctxlen = 3;
 467                 xecfg.flags = XDL_EMIT_FUNCNAMES;
 468                 if (!diffopts)
 469                         ;
 470                 else if (!strncmp(diffopts, "--unified=", 10))
 471                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
 472                 else if (!strncmp(diffopts, "-u", 2))
 473                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
 474                 ecb.outf = fn_out;
 475                 ecb.priv = &ecbdata;
 476                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 477         }
 478
 479  free_ab_and_return:
 480         free(a_one);
 481         free(b_two);
 482         return;
 483 }
 484
 485 static void builtin_diffstat(const char *name_a, const char *name_b,
 486                              struct diff_filespec *one,
 487                              struct diff_filespec *two,
 488                              struct diffstat_t *diffstat)
 489 {
 490         mmfile_t mf1, mf2;
 491         struct diffstat_file *data;
 492
 493         data = diffstat_add(diffstat, name_a, name_b);
 494
 495         if (!one || !two) {
 496                 data->is_unmerged = 1;
 497                 return;
 498         }
 499
 500         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 501                 die("unable to read files to diff");
 502
 503         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 504                 data->is_binary = 1;
 505         else {
 506                 /* Crazy xdl interfaces.. */
 507                 xpparam_t xpp;
 508                 xdemitconf_t xecfg;
 509                 xdemitcb_t ecb;
 510
 511                 xpp.flags = XDF_NEED_MINIMAL;
 512                 xecfg.ctxlen = 0;
 513                 xecfg.flags = 0;
 514                 ecb.outf = xdiff_outf;
 515                 ecb.priv = diffstat;
 516                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 517         }
 518 }
 519
 520 struct diff_filespec *alloc_filespec(const char *path)
 521 {
 522         int namelen = strlen(path);
 523         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 524
 525         memset(spec, 0, sizeof(*spec));
 526         spec->path = (char *)(spec + 1);
 527         memcpy(spec->path, path, namelen+1);
 528         return spec;
 529 }
 530
 531 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 532                    unsigned short mode)
 533 {
 534         if (mode) {
 535                 spec->mode = canon_mode(mode);
 536                 memcpy(spec->sha1, sha1, 20);
 537                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 538         }
 539 }
 540
 541 /*
 542  * Given a name and sha1 pair, if the dircache tells us the file in
 543  * the work tree has that object contents, return true, so that
 544  * prepare_temp_file() does not have to inflate and extract.
 545  */
 546 static int work_tree_matches(const char *name, const unsigned char *sha1)
 547 {
 548         struct cache_entry *ce;
 549         struct stat st;
 550         int pos, len;
 551
 552         /* We do not read the cache ourselves here, because the
 553          * benchmark with my previous version that always reads cache
 554          * shows that it makes things worse for diff-tree comparing
 555          * two linux-2.6 kernel trees in an already checked out work
 556          * tree.  This is because most diff-tree comparisons deal with
 557          * only a small number of files, while reading the cache is
 558          * expensive for a large project, and its cost outweighs the
 559          * savings we get by not inflating the object to a temporary
 560          * file.  Practically, this code only helps when we are used
 561          * by diff-cache --cached, which does read the cache before
 562          * calling us.
 563          */
 564         if (!active_cache)
 565                 return 0;
 566
 567         len = strlen(name);
 568         pos = cache_name_pos(name, len);
 569         if (pos < 0)
 570                 return 0;
 571         ce = active_cache[pos];
 572         if ((lstat(name, &st) < 0) ||
 573             !S_ISREG(st.st_mode) || /* careful! */
 574             ce_match_stat(ce, &st, 0) ||
 575             memcmp(sha1, ce->sha1, 20))
 576                 return 0;
 577         /* we return 1 only when we can stat, it is a regular file,
 578          * stat information matches, and sha1 recorded in the cache
 579          * matches.  I.e. we know the file in the work tree really is
 580          * the same as the <name, sha1> pair.
 581          */
 582         return 1;
 583 }
 584
 585 static struct sha1_size_cache {
 586         unsigned char sha1[20];
 587         unsigned long size;
 588 } **sha1_size_cache;
 589 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 590
 591 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 592                                                  int find_only,
 593                                                  unsigned long size)
 594 {
 595         int first, last;
 596         struct sha1_size_cache *e;
 597
 598         first = 0;
 599         last = sha1_size_cache_nr;
 600         while (last > first) {
 601                 int cmp, next = (last + first) >> 1;
 602                 e = sha1_size_cache[next];
 603                 cmp = memcmp(e->sha1, sha1, 20);
 604                 if (!cmp)
 605                         return e;
 606                 if (cmp < 0) {
 607                         last = next;
 608                         continue;
 609                 }
 610                 first = next+1;
 611         }
 612         /* not found */
 613         if (find_only)
 614                 return NULL;
 615         /* insert to make it at "first" */
 616         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 617                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 618                 sha1_size_cache = xrealloc(sha1_size_cache,
 619                                            sha1_size_cache_alloc *
 620                                            sizeof(*sha1_size_cache));
 621         }
 622         sha1_size_cache_nr++;
 623         if (first < sha1_size_cache_nr)
 624                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 625                         (sha1_size_cache_nr - first - 1) *
 626                         sizeof(*sha1_size_cache));
 627         e = xmalloc(sizeof(struct sha1_size_cache));
 628         sha1_size_cache[first] = e;
 629         memcpy(e->sha1, sha1, 20);
 630         e->size = size;
 631         return e;
 632 }
 633
 634 /*
 635  * While doing rename detection and pickaxe operation, we may need to
 636  * grab the data for the blob (or file) for our own in-core comparison.
 637  * diff_filespec has data and size fields for this purpose.
 638  */
 639 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 640 {
 641         int err = 0;
 642         if (!DIFF_FILE_VALID(s))
 643                 die("internal error: asking to populate invalid file.");
 644         if (S_ISDIR(s->mode))
 645                 return -1;
 646
 647         if (!use_size_cache)
 648                 size_only = 0;
 649
 650         if (s->data)
 651                 return err;
 652         if (!s->sha1_valid ||
 653             work_tree_matches(s->path, s->sha1)) {
 654                 struct stat st;
 655                 int fd;
 656                 if (lstat(s->path, &st) < 0) {
 657                         if (errno == ENOENT) {
 658                         err_empty:
 659                                 err = -1;
 660                         empty:
 661                                 s->data = "";
 662                                 s->size = 0;
 663                                 return err;
 664                         }
 665                 }
 666                 s->size = st.st_size;
 667                 if (!s->size)
 668                         goto empty;
 669                 if (size_only)
 670                         return 0;
 671                 if (S_ISLNK(st.st_mode)) {
 672                         int ret;
 673                         s->data = xmalloc(s->size);
 674                         s->should_free = 1;
 675                         ret = readlink(s->path, s->data, s->size);
 676                         if (ret < 0) {
 677                                 free(s->data);
 678                                 goto err_empty;
 679                         }
 680                         return 0;
 681                 }
 682                 fd = open(s->path, O_RDONLY);
 683                 if (fd < 0)
 684                         goto err_empty;
 685                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 686                 close(fd);
 687                 if (s->data == MAP_FAILED)
 688                         goto err_empty;
 689                 s->should_munmap = 1;
 690         }
 691         else {
 692                 char type[20];
 693                 struct sha1_size_cache *e;
 694
 695                 if (size_only) {
 696                         e = locate_size_cache(s->sha1, 1, 0);
 697                         if (e) {
 698                                 s->size = e->size;
 699                                 return 0;
 700                         }
 701                         if (!sha1_object_info(s->sha1, type, &s->size))
 702                                 locate_size_cache(s->sha1, 0, s->size);
 703                 }
 704                 else {
 705                         s->data = read_sha1_file(s->sha1, type, &s->size);
 706                         s->should_free = 1;
 707                 }
 708         }
 709         return 0;
 710 }
 711
 712 void diff_free_filespec_data(struct diff_filespec *s)
 713 {
 714         if (s->should_free)
 715                 free(s->data);
 716         else if (s->should_munmap)
 717                 munmap(s->data, s->size);
 718         s->should_free = s->should_munmap = 0;
 719         s->data = NULL;
 720         free(s->cnt_data);
 721         s->cnt_data = NULL;
 722 }
 723
 724 static void prep_temp_blob(struct diff_tempfile *temp,
 725                            void *blob,
 726                            unsigned long size,
 727                            const unsigned char *sha1,
 728                            int mode)
 729 {
 730         int fd;
 731
 732         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 733         if (fd < 0)
 734                 die("unable to create temp-file");
 735         if (write(fd, blob, size) != size)
 736                 die("unable to write temp-file");
 737         close(fd);
 738         temp->name = temp->tmp_path;
 739         strcpy(temp->hex, sha1_to_hex(sha1));
 740         temp->hex[40] = 0;
 741         sprintf(temp->mode, "%06o", mode);
 742 }
 743
 744 static void prepare_temp_file(const char *name,
 745                               struct diff_tempfile *temp,
 746                               struct diff_filespec *one)
 747 {
 748         if (!DIFF_FILE_VALID(one)) {
 749         not_a_valid_file:
 750                 /* A '-' entry produces this for file-2, and
 751                  * a '+' entry produces this for file-1.
 752                  */
 753                 temp->name = "/dev/null";
 754                 strcpy(temp->hex, ".");
 755                 strcpy(temp->mode, ".");
 756                 return;
 757         }
 758
 759         if (!one->sha1_valid ||
 760             work_tree_matches(name, one->sha1)) {
 761                 struct stat st;
 762                 if (lstat(name, &st) < 0) {
 763                         if (errno == ENOENT)
 764                                 goto not_a_valid_file;
 765                         die("stat(%s): %s", name, strerror(errno));
 766                 }
 767                 if (S_ISLNK(st.st_mode)) {
 768                         int ret;
 769                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
 770                         if (sizeof(buf) <= st.st_size)
 771                                 die("symlink too long: %s", name);
 772                         ret = readlink(name, buf, st.st_size);
 773                         if (ret < 0)
 774                                 die("readlink(%s)", name);
 775                         prep_temp_blob(temp, buf, st.st_size,
 776                                        (one->sha1_valid ?
 777                                         one->sha1 : null_sha1),
 778                                        (one->sha1_valid ?
 779                                         one->mode : S_IFLNK));
 780                 }
 781                 else {
 782                         /* we can borrow from the file in the work tree */
 783                         temp->name = name;
 784                         if (!one->sha1_valid)
 785                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 786                         else
 787                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 788                         /* Even though we may sometimes borrow the
 789                          * contents from the work tree, we always want
 790                          * one->mode.  mode is trustworthy even when
 791                          * !(one->sha1_valid), as long as
 792                          * DIFF_FILE_VALID(one).
 793                          */
 794                         sprintf(temp->mode, "%06o", one->mode);
 795                 }
 796                 return;
 797         }
 798         else {
 799                 if (diff_populate_filespec(one, 0))
 800                         die("cannot read data blob for %s", one->path);
 801                 prep_temp_blob(temp, one->data, one->size,
 802                                one->sha1, one->mode);
 803         }
 804 }
 805
 806 static void remove_tempfile(void)
 807 {
 808         int i;
 809
 810         for (i = 0; i < 2; i++)
 811                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 812                         unlink(diff_temp[i].name);
 813                         diff_temp[i].name = NULL;
 814                 }
 815 }
 816
 817 static void remove_tempfile_on_signal(int signo)
 818 {
 819         remove_tempfile();
 820         signal(SIGINT, SIG_DFL);
 821         raise(signo);
 822 }
 823
 824 static int spawn_prog(const char *pgm, const char **arg)
 825 {
 826         pid_t pid;
 827         int status;
 828
 829         fflush(NULL);
 830         pid = fork();
 831         if (pid < 0)
 832                 die("unable to fork");
 833         if (!pid) {
 834                 execvp(pgm, (char *const*) arg);
 835                 exit(255);
 836         }
 837
 838         while (waitpid(pid, &status, 0) < 0) {
 839                 if (errno == EINTR)
 840                         continue;
 841                 return -1;
 842         }
 843
 844         /* Earlier we did not check the exit status because
 845          * diff exits non-zero if files are different, and
 846          * we are not interested in knowing that.  It was a
 847          * mistake which made it harder to quit a diff-*
 848          * session that uses the git-apply-patch-script as
 849          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 850          * should also exit non-zero only when it wants to
 851          * abort the entire diff-* session.
 852          */
 853         if (WIFEXITED(status) && !WEXITSTATUS(status))
 854                 return 0;
 855         return -1;
 856 }
 857
 858 /* An external diff command takes:
 859  *
 860  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 861  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 862  *
 863  */
 864 static void run_external_diff(const char *pgm,
 865                               const char *name,
 866                               const char *other,
 867                               struct diff_filespec *one,
 868                               struct diff_filespec *two,
 869                               const char *xfrm_msg,
 870                               int complete_rewrite)
 871 {
 872         const char *spawn_arg[10];
 873         struct diff_tempfile *temp = diff_temp;
 874         int retval;
 875         static int atexit_asked = 0;
 876         const char *othername;
 877         const char **arg = &spawn_arg[0];
 878
 879         othername = (other? other : name);
 880         if (one && two) {
 881                 prepare_temp_file(name, &temp[0], one);
 882                 prepare_temp_file(othername, &temp[1], two);
 883                 if (! atexit_asked &&
 884                     (temp[0].name == temp[0].tmp_path ||
 885                      temp[1].name == temp[1].tmp_path)) {
 886                         atexit_asked = 1;
 887                         atexit(remove_tempfile);
 888                 }
 889                 signal(SIGINT, remove_tempfile_on_signal);
 890         }
 891
 892         if (one && two) {
 893                 *arg++ = pgm;
 894                 *arg++ = name;
 895                 *arg++ = temp[0].name;
 896                 *arg++ = temp[0].hex;
 897                 *arg++ = temp[0].mode;
 898                 *arg++ = temp[1].name;
 899                 *arg++ = temp[1].hex;
 900                 *arg++ = temp[1].mode;
 901                 if (other) {
 902                         *arg++ = other;
 903                         *arg++ = xfrm_msg;
 904                 }
 905         } else {
 906                 *arg++ = pgm;
 907                 *arg++ = name;
 908         }
 909         *arg = NULL;
 910         retval = spawn_prog(pgm, spawn_arg);
 911         remove_tempfile();
 912         if (retval) {
 913                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 914                 exit(1);
 915         }
 916 }
 917
 918 static void run_diff_cmd(const char *pgm,
 919                          const char *name,
 920                          const char *other,
 921                          struct diff_filespec *one,
 922                          struct diff_filespec *two,
 923                          const char *xfrm_msg,
 924                          int complete_rewrite)
 925 {
 926         if (pgm) {
 927                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 928                                   complete_rewrite);
 929                 return;
 930         }
 931         if (one && two)
 932                 builtin_diff(name, other ? other : name,
 933                              one, two, xfrm_msg, complete_rewrite);
 934         else
 935                 printf("* Unmerged path %s\n", name);
 936 }
 937
 938 static void diff_fill_sha1_info(struct diff_filespec *one)
 939 {
 940         if (DIFF_FILE_VALID(one)) {
 941                 if (!one->sha1_valid) {
 942                         struct stat st;
 943                         if (lstat(one->path, &st) < 0)
 944                                 die("stat %s", one->path);
 945                         if (index_path(one->sha1, one->path, &st, 0))
 946                                 die("cannot hash %s\n", one->path);
 947                 }
 948         }
 949         else
 950                 memset(one->sha1, 0, 20);
 951 }
 952
 953 static void run_diff(struct diff_filepair *p, struct diff_options *o)
 954 {
 955         const char *pgm = external_diff();
 956         char msg[PATH_MAX*2+300], *xfrm_msg;
 957         struct diff_filespec *one;
 958         struct diff_filespec *two;
 959         const char *name;
 960         const char *other;
 961         char *name_munged, *other_munged;
 962         int complete_rewrite = 0;
 963         int len;
 964
 965         if (DIFF_PAIR_UNMERGED(p)) {
 966                 /* unmerged */
 967                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, 0);
 968                 return;
 969         }
 970
 971         name = p->one->path;
 972         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 973         name_munged = quote_one(name);
 974         other_munged = quote_one(other);
 975         one = p->one; two = p->two;
 976
 977         diff_fill_sha1_info(one);
 978         diff_fill_sha1_info(two);
 979
 980         len = 0;
 981         switch (p->status) {
 982         case DIFF_STATUS_COPIED:
 983                 len += snprintf(msg + len, sizeof(msg) - len,
 984                                 "similarity index %d%%\n"
 985                                 "copy from %s\n"
 986                                 "copy to %s\n",
 987                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 988                                 name_munged, other_munged);
 989                 break;
 990         case DIFF_STATUS_RENAMED:
 991                 len += snprintf(msg + len, sizeof(msg) - len,
 992                                 "similarity index %d%%\n"
 993                                 "rename from %s\n"
 994                                 "rename to %s\n",
 995                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 996                                 name_munged, other_munged);
 997                 break;
 998         case DIFF_STATUS_MODIFIED:
 999                 if (p->score) {
1000                         len += snprintf(msg + len, sizeof(msg) - len,
1001                                         "dissimilarity index %d%%\n",
1002                                         (int)(0.5 + p->score *
1003                                               100.0/MAX_SCORE));
1004                         complete_rewrite = 1;
1005                         break;
1006                 }
1007                 /* fallthru */
1008         default:
1009                 /* nothing */
1010                 ;
1011         }
1012
1013         if (memcmp(one->sha1, two->sha1, 20)) {
1014                 char one_sha1[41];
1015                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1016                 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
1017
1018                 len += snprintf(msg + len, sizeof(msg) - len,
1019                                 "index %.*s..%.*s",
1020                                 abbrev, one_sha1, abbrev,
1021                                 sha1_to_hex(two->sha1));
1022                 if (one->mode == two->mode)
1023                         len += snprintf(msg + len, sizeof(msg) - len,
1024                                         " %06o", one->mode);
1025                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1026         }
1027
1028         if (len)
1029                 msg[--len] = 0;
1030         xfrm_msg = len ? msg : NULL;
1031
1032         if (!pgm &&
1033             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1034             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1035                 /* a filepair that changes between file and symlink
1036                  * needs to be split into deletion and creation.
1037                  */
1038                 struct diff_filespec *null = alloc_filespec(two->path);
1039                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, 0);
1040                 free(null);
1041                 null = alloc_filespec(one->path);
1042                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, 0);
1043                 free(null);
1044         }
1045         else
1046                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg,
1047                              complete_rewrite);
1048
1049         free(name_munged);
1050         free(other_munged);
1051 }
1052
1053 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1054                          struct diffstat_t *diffstat)
1055 {
1056         const char *name;
1057         const char *other;
1058
1059         if (DIFF_PAIR_UNMERGED(p)) {
1060                 /* unmerged */
1061                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat);
1062                 return;
1063         }
1064
1065         name = p->one->path;
1066         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1067
1068         diff_fill_sha1_info(p->one);
1069         diff_fill_sha1_info(p->two);
1070
1071         builtin_diffstat(name, other, p->one, p->two, diffstat);
1072 }
1073
1074 void diff_setup(struct diff_options *options)
1075 {
1076         memset(options, 0, sizeof(*options));
1077         options->output_format = DIFF_FORMAT_RAW;
1078         options->line_termination = '\n';
1079         options->break_opt = -1;
1080         options->rename_limit = -1;
1081
1082         options->change = diff_change;
1083         options->add_remove = diff_addremove;
1084 }
1085
1086 int diff_setup_done(struct diff_options *options)
1087 {
1088         if ((options->find_copies_harder &&
1089              options->detect_rename != DIFF_DETECT_COPY) ||
1090             (0 <= options->rename_limit && !options->detect_rename))
1091                 return -1;
1092
1093         /*
1094          * These cases always need recursive; we do not drop caller-supplied
1095          * recursive bits for other formats here.
1096          */
1097         if ((options->output_format == DIFF_FORMAT_PATCH) ||
1098             (options->output_format == DIFF_FORMAT_DIFFSTAT))
1099                 options->recursive = 1;
1100
1101         if (options->detect_rename && options->rename_limit < 0)
1102                 options->rename_limit = diff_rename_limit_default;
1103         if (options->setup & DIFF_SETUP_USE_CACHE) {
1104                 if (!active_cache)
1105                         /* read-cache does not die even when it fails
1106                          * so it is safe for us to do this here.  Also
1107                          * it does not smudge active_cache or active_nr
1108                          * when it fails, so we do not have to worry about
1109                          * cleaning it up ourselves either.
1110                          */
1111                         read_cache();
1112         }
1113         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1114                 use_size_cache = 1;
1115         if (options->abbrev <= 0 || 40 < options->abbrev)
1116                 options->abbrev = 40; /* full */
1117
1118         return 0;
1119 }
1120
1121 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1122 {
1123         const char *arg = av[0];
1124         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1125                 options->output_format = DIFF_FORMAT_PATCH;
1126         else if (!strcmp(arg, "--patch-with-raw")) {
1127                 options->output_format = DIFF_FORMAT_PATCH;
1128                 options->with_raw = 1;
1129         }
1130         else if (!strcmp(arg, "--stat"))
1131                 options->output_format = DIFF_FORMAT_DIFFSTAT;
1132         else if (!strcmp(arg, "--patch-with-stat")) {
1133                 options->output_format = DIFF_FORMAT_PATCH;
1134                 options->with_stat = 1;
1135         }
1136         else if (!strcmp(arg, "-z"))
1137                 options->line_termination = 0;
1138         else if (!strncmp(arg, "-l", 2))
1139                 options->rename_limit = strtoul(arg+2, NULL, 10);
1140         else if (!strcmp(arg, "--full-index"))
1141                 options->full_index = 1;
1142         else if (!strcmp(arg, "--name-only"))
1143                 options->output_format = DIFF_FORMAT_NAME;
1144         else if (!strcmp(arg, "--name-status"))
1145                 options->output_format = DIFF_FORMAT_NAME_STATUS;
1146         else if (!strcmp(arg, "-R"))
1147                 options->reverse_diff = 1;
1148         else if (!strncmp(arg, "-S", 2))
1149                 options->pickaxe = arg + 2;
1150         else if (!strcmp(arg, "-s"))
1151                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1152         else if (!strncmp(arg, "-O", 2))
1153                 options->orderfile = arg + 2;
1154         else if (!strncmp(arg, "--diff-filter=", 14))
1155                 options->filter = arg + 14;
1156         else if (!strcmp(arg, "--pickaxe-all"))
1157                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1158         else if (!strcmp(arg, "--pickaxe-regex"))
1159                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1160         else if (!strncmp(arg, "-B", 2)) {
1161                 if ((options->break_opt =
1162                      diff_scoreopt_parse(arg)) == -1)
1163                         return -1;
1164         }
1165         else if (!strncmp(arg, "-M", 2)) {
1166                 if ((options->rename_score =
1167                      diff_scoreopt_parse(arg)) == -1)
1168                         return -1;
1169                 options->detect_rename = DIFF_DETECT_RENAME;
1170         }
1171         else if (!strncmp(arg, "-C", 2)) {
1172                 if ((options->rename_score =
1173                      diff_scoreopt_parse(arg)) == -1)
1174                         return -1;
1175                 options->detect_rename = DIFF_DETECT_COPY;
1176         }
1177         else if (!strcmp(arg, "--find-copies-harder"))
1178                 options->find_copies_harder = 1;
1179         else if (!strcmp(arg, "--abbrev"))
1180                 options->abbrev = DEFAULT_ABBREV;
1181         else if (!strncmp(arg, "--abbrev=", 9)) {
1182                 options->abbrev = strtoul(arg + 9, NULL, 10);
1183                 if (options->abbrev < MINIMUM_ABBREV)
1184                         options->abbrev = MINIMUM_ABBREV;
1185                 else if (40 < options->abbrev)
1186                         options->abbrev = 40;
1187         }
1188         else
1189                 return 0;
1190         return 1;
1191 }
1192
1193 static int parse_num(const char **cp_p)
1194 {
1195         unsigned long num, scale;
1196         int ch, dot;
1197         const char *cp = *cp_p;
1198
1199         num = 0;
1200         scale = 1;
1201         dot = 0;
1202         for(;;) {
1203                 ch = *cp;
1204                 if ( !dot && ch == '.' ) {
1205                         scale = 1;
1206                         dot = 1;
1207                 } else if ( ch == '%' ) {
1208                         scale = dot ? scale*100 : 100;
1209                         cp++;   /* % is always at the end */
1210                         break;
1211                 } else if ( ch >= '0' && ch <= '9' ) {
1212                         if ( scale < 100000 ) {
1213                                 scale *= 10;
1214                                 num = (num*10) + (ch-'0');
1215                         }
1216                 } else {
1217                         break;
1218                 }
1219                 cp++;
1220         }
1221         *cp_p = cp;
1222
1223         /* user says num divided by scale and we say internally that
1224          * is MAX_SCORE * num / scale.
1225          */
1226         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1227 }
1228
1229 int diff_scoreopt_parse(const char *opt)
1230 {
1231         int opt1, opt2, cmd;
1232
1233         if (*opt++ != '-')
1234                 return -1;
1235         cmd = *opt++;
1236         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1237                 return -1; /* that is not a -M, -C nor -B option */
1238
1239         opt1 = parse_num(&opt);
1240         if (cmd != 'B')
1241                 opt2 = 0;
1242         else {
1243                 if (*opt == 0)
1244                         opt2 = 0;
1245                 else if (*opt != '/')
1246                         return -1; /* we expect -B80/99 or -B80 */
1247                 else {
1248                         opt++;
1249                         opt2 = parse_num(&opt);
1250                 }
1251         }
1252         if (*opt != 0)
1253                 return -1;
1254         return opt1 | (opt2 << 16);
1255 }
1256
1257 struct diff_queue_struct diff_queued_diff;
1258
1259 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1260 {
1261         if (queue->alloc <= queue->nr) {
1262                 queue->alloc = alloc_nr(queue->alloc);
1263                 queue->queue = xrealloc(queue->queue,
1264                                         sizeof(dp) * queue->alloc);
1265         }
1266         queue->queue[queue->nr++] = dp;
1267 }
1268
1269 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1270                                  struct diff_filespec *one,
1271                                  struct diff_filespec *two)
1272 {
1273         struct diff_filepair *dp = xmalloc(sizeof(*dp));
1274         dp->one = one;
1275         dp->two = two;
1276         dp->score = 0;
1277         dp->status = 0;
1278         dp->source_stays = 0;
1279         dp->broken_pair = 0;
1280         if (queue)
1281                 diff_q(queue, dp);
1282         return dp;
1283 }
1284
1285 void diff_free_filepair(struct diff_filepair *p)
1286 {
1287         diff_free_filespec_data(p->one);
1288         diff_free_filespec_data(p->two);
1289         free(p->one);
1290         free(p->two);
1291         free(p);
1292 }
1293
1294 /* This is different from find_unique_abbrev() in that
1295  * it stuffs the result with dots for alignment.
1296  */
1297 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1298 {
1299         int abblen;
1300         const char *abbrev;
1301         if (len == 40)
1302                 return sha1_to_hex(sha1);
1303
1304         abbrev = find_unique_abbrev(sha1, len);
1305         if (!abbrev)
1306                 return sha1_to_hex(sha1);
1307         abblen = strlen(abbrev);
1308         if (abblen < 37) {
1309                 static char hex[41];
1310                 if (len < abblen && abblen <= len + 2)
1311                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1312                 else
1313                         sprintf(hex, "%s...", abbrev);
1314                 return hex;
1315         }
1316         return sha1_to_hex(sha1);
1317 }
1318
1319 static void diff_flush_raw(struct diff_filepair *p,
1320                            int line_termination,
1321                            int inter_name_termination,
1322                            struct diff_options *options,
1323                            int output_format)
1324 {
1325         int two_paths;
1326         char status[10];
1327         int abbrev = options->abbrev;
1328         const char *path_one, *path_two;
1329
1330         path_one = p->one->path;
1331         path_two = p->two->path;
1332         if (line_termination) {
1333                 path_one = quote_one(path_one);
1334                 path_two = quote_one(path_two);
1335         }
1336
1337         if (p->score)
1338                 sprintf(status, "%c%03d", p->status,
1339                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1340         else {
1341                 status[0] = p->status;
1342                 status[1] = 0;
1343         }
1344         switch (p->status) {
1345         case DIFF_STATUS_COPIED:
1346         case DIFF_STATUS_RENAMED:
1347                 two_paths = 1;
1348                 break;
1349         case DIFF_STATUS_ADDED:
1350         case DIFF_STATUS_DELETED:
1351                 two_paths = 0;
1352                 break;
1353         default:
1354                 two_paths = 0;
1355                 break;
1356         }
1357         if (output_format != DIFF_FORMAT_NAME_STATUS) {
1358                 printf(":%06o %06o %s ",
1359                        p->one->mode, p->two->mode,
1360                        diff_unique_abbrev(p->one->sha1, abbrev));
1361                 printf("%s ",
1362                        diff_unique_abbrev(p->two->sha1, abbrev));
1363         }
1364         printf("%s%c%s", status, inter_name_termination, path_one);
1365         if (two_paths)
1366                 printf("%c%s", inter_name_termination, path_two);
1367         putchar(line_termination);
1368         if (path_one != p->one->path)
1369                 free((void*)path_one);
1370         if (path_two != p->two->path)
1371                 free((void*)path_two);
1372 }
1373
1374 static void diff_flush_name(struct diff_filepair *p,
1375                             int inter_name_termination,
1376                             int line_termination)
1377 {
1378         char *path = p->two->path;
1379
1380         if (line_termination)
1381                 path = quote_one(p->two->path);
1382         else
1383                 path = p->two->path;
1384         printf("%s%c", path, line_termination);
1385         if (p->two->path != path)
1386                 free(path);
1387 }
1388
1389 int diff_unmodified_pair(struct diff_filepair *p)
1390 {
1391         /* This function is written stricter than necessary to support
1392          * the currently implemented transformers, but the idea is to
1393          * let transformers to produce diff_filepairs any way they want,
1394          * and filter and clean them up here before producing the output.
1395          */
1396         struct diff_filespec *one, *two;
1397
1398         if (DIFF_PAIR_UNMERGED(p))
1399                 return 0; /* unmerged is interesting */
1400
1401         one = p->one;
1402         two = p->two;
1403
1404         /* deletion, addition, mode or type change
1405          * and rename are all interesting.
1406          */
1407         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1408             DIFF_PAIR_MODE_CHANGED(p) ||
1409             strcmp(one->path, two->path))
1410                 return 0;
1411
1412         /* both are valid and point at the same path.  that is, we are
1413          * dealing with a change.
1414          */
1415         if (one->sha1_valid && two->sha1_valid &&
1416             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1417                 return 1; /* no change */
1418         if (!one->sha1_valid && !two->sha1_valid)
1419                 return 1; /* both look at the same file on the filesystem. */
1420         return 0;
1421 }
1422
1423 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1424 {
1425         if (diff_unmodified_pair(p))
1426                 return;
1427
1428         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1429             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1430                 return; /* no tree diffs in patch format */
1431
1432         run_diff(p, o);
1433 }
1434
1435 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1436                             struct diffstat_t *diffstat)
1437 {
1438         if (diff_unmodified_pair(p))
1439                 return;
1440
1441         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1442             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1443                 return; /* no tree diffs in patch format */
1444
1445         run_diffstat(p, o, diffstat);
1446 }
1447
1448 int diff_queue_is_empty(void)
1449 {
1450         struct diff_queue_struct *q = &diff_queued_diff;
1451         int i;
1452         for (i = 0; i < q->nr; i++)
1453                 if (!diff_unmodified_pair(q->queue[i]))
1454                         return 0;
1455         return 1;
1456 }
1457
1458 #if DIFF_DEBUG
1459 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1460 {
1461         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1462                 x, one ? one : "",
1463                 s->path,
1464                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1465                 s->mode,
1466                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1467         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1468                 x, one ? one : "",
1469                 s->size, s->xfrm_flags);
1470 }
1471
1472 void diff_debug_filepair(const struct diff_filepair *p, int i)
1473 {
1474         diff_debug_filespec(p->one, i, "one");
1475         diff_debug_filespec(p->two, i, "two");
1476         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1477                 p->score, p->status ? p->status : '?',
1478                 p->source_stays, p->broken_pair);
1479 }
1480
1481 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1482 {
1483         int i;
1484         if (msg)
1485                 fprintf(stderr, "%s\n", msg);
1486         fprintf(stderr, "q->nr = %d\n", q->nr);
1487         for (i = 0; i < q->nr; i++) {
1488                 struct diff_filepair *p = q->queue[i];
1489                 diff_debug_filepair(p, i);
1490         }
1491 }
1492 #endif
1493
1494 static void diff_resolve_rename_copy(void)
1495 {
1496         int i, j;
1497         struct diff_filepair *p, *pp;
1498         struct diff_queue_struct *q = &diff_queued_diff;
1499
1500         diff_debug_queue("resolve-rename-copy", q);
1501
1502         for (i = 0; i < q->nr; i++) {
1503                 p = q->queue[i];
1504                 p->status = 0; /* undecided */
1505                 if (DIFF_PAIR_UNMERGED(p))
1506                         p->status = DIFF_STATUS_UNMERGED;
1507                 else if (!DIFF_FILE_VALID(p->one))
1508                         p->status = DIFF_STATUS_ADDED;
1509                 else if (!DIFF_FILE_VALID(p->two))
1510                         p->status = DIFF_STATUS_DELETED;
1511                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1512                         p->status = DIFF_STATUS_TYPE_CHANGED;
1513
1514                 /* from this point on, we are dealing with a pair
1515                  * whose both sides are valid and of the same type, i.e.
1516                  * either in-place edit or rename/copy edit.
1517                  */
1518                 else if (DIFF_PAIR_RENAME(p)) {
1519                         if (p->source_stays) {
1520                                 p->status = DIFF_STATUS_COPIED;
1521                                 continue;
1522                         }
1523                         /* See if there is some other filepair that
1524                          * copies from the same source as us.  If so
1525                          * we are a copy.  Otherwise we are either a
1526                          * copy if the path stays, or a rename if it
1527                          * does not, but we already handled "stays" case.
1528                          */
1529                         for (j = i + 1; j < q->nr; j++) {
1530                                 pp = q->queue[j];
1531                                 if (strcmp(pp->one->path, p->one->path))
1532                                         continue; /* not us */
1533                                 if (!DIFF_PAIR_RENAME(pp))
1534                                         continue; /* not a rename/copy */
1535                                 /* pp is a rename/copy from the same source */
1536                                 p->status = DIFF_STATUS_COPIED;
1537                                 break;
1538                         }
1539                         if (!p->status)
1540                                 p->status = DIFF_STATUS_RENAMED;
1541                 }
1542                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1543                          p->one->mode != p->two->mode)
1544                         p->status = DIFF_STATUS_MODIFIED;
1545                 else {
1546                         /* This is a "no-change" entry and should not
1547                          * happen anymore, but prepare for broken callers.
1548                          */
1549                         error("feeding unmodified %s to diffcore",
1550                               p->one->path);
1551                         p->status = DIFF_STATUS_UNKNOWN;
1552                 }
1553         }
1554         diff_debug_queue("resolve-rename-copy done", q);
1555 }
1556
1557 static void flush_one_pair(struct diff_filepair *p,
1558                            int diff_output_format,
1559                            struct diff_options *options,
1560                            struct diffstat_t *diffstat)
1561 {
1562         int inter_name_termination = '\t';
1563         int line_termination = options->line_termination;
1564         if (!line_termination)
1565                 inter_name_termination = 0;
1566
1567         switch (p->status) {
1568         case DIFF_STATUS_UNKNOWN:
1569                 break;
1570         case 0:
1571                 die("internal error in diff-resolve-rename-copy");
1572                 break;
1573         default:
1574                 switch (diff_output_format) {
1575                 case DIFF_FORMAT_DIFFSTAT:
1576                         diff_flush_stat(p, options, diffstat);
1577                         break;
1578                 case DIFF_FORMAT_PATCH:
1579                         diff_flush_patch(p, options);
1580                         break;
1581                 case DIFF_FORMAT_RAW:
1582                 case DIFF_FORMAT_NAME_STATUS:
1583                         diff_flush_raw(p, line_termination,
1584                                        inter_name_termination,
1585                                        options, diff_output_format);
1586                         break;
1587                 case DIFF_FORMAT_NAME:
1588                         diff_flush_name(p,
1589                                         inter_name_termination,
1590                                         line_termination);
1591                         break;
1592                 case DIFF_FORMAT_NO_OUTPUT:
1593                         break;
1594                 }
1595         }
1596 }
1597
1598 void diff_flush(struct diff_options *options)
1599 {
1600         struct diff_queue_struct *q = &diff_queued_diff;
1601         int i;
1602         int diff_output_format = options->output_format;
1603         struct diffstat_t *diffstat = NULL;
1604
1605         if (diff_output_format == DIFF_FORMAT_DIFFSTAT || options->with_stat) {
1606                 diffstat = xcalloc(sizeof (struct diffstat_t), 1);
1607                 diffstat->xm.consume = diffstat_consume;
1608         }
1609
1610         if (options->with_raw) {
1611                 for (i = 0; i < q->nr; i++) {
1612                         struct diff_filepair *p = q->queue[i];
1613                         flush_one_pair(p, DIFF_FORMAT_RAW, options, NULL);
1614                 }
1615                 putchar(options->line_termination);
1616         }
1617         if (options->with_stat) {
1618                 for (i = 0; i < q->nr; i++) {
1619                         struct diff_filepair *p = q->queue[i];
1620                         flush_one_pair(p, DIFF_FORMAT_DIFFSTAT, options,
1621                                        diffstat);
1622                 }
1623                 show_stats(diffstat);
1624                 free(diffstat);
1625                 diffstat = NULL;
1626                 putchar(options->line_termination);
1627         }
1628         for (i = 0; i < q->nr; i++) {
1629                 struct diff_filepair *p = q->queue[i];
1630                 flush_one_pair(p, diff_output_format, options, diffstat);
1631                 diff_free_filepair(p);
1632         }
1633
1634         if (diffstat) {
1635                 show_stats(diffstat);
1636                 free(diffstat);
1637         }
1638
1639         free(q->queue);
1640         q->queue = NULL;
1641         q->nr = q->alloc = 0;
1642 }
1643
1644 static void diffcore_apply_filter(const char *filter)
1645 {
1646         int i;
1647         struct diff_queue_struct *q = &diff_queued_diff;
1648         struct diff_queue_struct outq;
1649         outq.queue = NULL;
1650         outq.nr = outq.alloc = 0;
1651
1652         if (!filter)
1653                 return;
1654
1655         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1656                 int found;
1657                 for (i = found = 0; !found && i < q->nr; i++) {
1658                         struct diff_filepair *p = q->queue[i];
1659                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1660                              ((p->score &&
1661                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1662                               (!p->score &&
1663                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1664                             ((p->status != DIFF_STATUS_MODIFIED) &&
1665                              strchr(filter, p->status)))
1666                                 found++;
1667                 }
1668                 if (found)
1669                         return;
1670
1671                 /* otherwise we will clear the whole queue
1672                  * by copying the empty outq at the end of this
1673                  * function, but first clear the current entries
1674                  * in the queue.
1675                  */
1676                 for (i = 0; i < q->nr; i++)
1677                         diff_free_filepair(q->queue[i]);
1678         }
1679         else {
1680                 /* Only the matching ones */
1681                 for (i = 0; i < q->nr; i++) {
1682                         struct diff_filepair *p = q->queue[i];
1683
1684                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1685                              ((p->score &&
1686                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1687                               (!p->score &&
1688                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1689                             ((p->status != DIFF_STATUS_MODIFIED) &&
1690                              strchr(filter, p->status)))
1691                                 diff_q(&outq, p);
1692                         else
1693                                 diff_free_filepair(p);
1694                 }
1695         }
1696         free(q->queue);
1697         *q = outq;
1698 }
1699
1700 void diffcore_std(struct diff_options *options)
1701 {
1702         if (options->break_opt != -1)
1703                 diffcore_break(options->break_opt);
1704         if (options->detect_rename)
1705                 diffcore_rename(options);
1706         if (options->break_opt != -1)
1707                 diffcore_merge_broken();
1708         if (options->pickaxe)
1709                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1710         if (options->orderfile)
1711                 diffcore_order(options->orderfile);
1712         diff_resolve_rename_copy();
1713         diffcore_apply_filter(options->filter);
1714 }
1715
1716
1717 void diffcore_std_no_resolve(struct diff_options *options)
1718 {
1719         if (options->pickaxe)
1720                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1721         if (options->orderfile)
1722                 diffcore_order(options->orderfile);
1723         diffcore_apply_filter(options->filter);
1724 }
1725
1726 void diff_addremove(struct diff_options *options,
1727                     int addremove, unsigned mode,
1728                     const unsigned char *sha1,
1729                     const char *base, const char *path)
1730 {
1731         char concatpath[PATH_MAX];
1732         struct diff_filespec *one, *two;
1733
1734         /* This may look odd, but it is a preparation for
1735          * feeding "there are unchanged files which should
1736          * not produce diffs, but when you are doing copy
1737          * detection you would need them, so here they are"
1738          * entries to the diff-core.  They will be prefixed
1739          * with something like '=' or '*' (I haven't decided
1740          * which but should not make any difference).
1741          * Feeding the same new and old to diff_change()
1742          * also has the same effect.
1743          * Before the final output happens, they are pruned after
1744          * merged into rename/copy pairs as appropriate.
1745          */
1746         if (options->reverse_diff)
1747                 addremove = (addremove == '+' ? '-' :
1748                              addremove == '-' ? '+' : addremove);
1749
1750         if (!path) path = "";
1751         sprintf(concatpath, "%s%s", base, path);
1752         one = alloc_filespec(concatpath);
1753         two = alloc_filespec(concatpath);
1754
1755         if (addremove != '+')
1756                 fill_filespec(one, sha1, mode);
1757         if (addremove != '-')
1758                 fill_filespec(two, sha1, mode);
1759
1760         diff_queue(&diff_queued_diff, one, two);
1761 }
1762
1763 void diff_change(struct diff_options *options,
1764                  unsigned old_mode, unsigned new_mode,
1765                  const unsigned char *old_sha1,
1766                  const unsigned char *new_sha1,
1767                  const char *base, const char *path)
1768 {
1769         char concatpath[PATH_MAX];
1770         struct diff_filespec *one, *two;
1771
1772         if (options->reverse_diff) {
1773                 unsigned tmp;
1774                 const unsigned char *tmp_c;
1775                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1776                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1777         }
1778         if (!path) path = "";
1779         sprintf(concatpath, "%s%s", base, path);
1780         one = alloc_filespec(concatpath);
1781         two = alloc_filespec(concatpath);
1782         fill_filespec(one, old_sha1, old_mode);
1783         fill_filespec(two, new_sha1, new_mode);
1784
1785         diff_queue(&diff_queued_diff, one, two);
1786 }
1787
1788 void diff_unmerge(struct diff_options *options,
1789                   const char *path)
1790 {
1791         struct diff_filespec *one, *two;
1792         one = alloc_filespec(path);
1793         two = alloc_filespec(path);
1794         diff_queue(&diff_queued_diff, one, two);
1795 }