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