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