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
  12 static const char *diff_opts = "-pu";
  13
  14 static int use_size_cache;
  15
  16 int diff_rename_limit_default = -1;
  17
  18 static char *quote_one(const char *str)
  19 {
  20         int needlen;
  21         char *xp;
  22
  23         if (!str)
  24                 return NULL;
  25         needlen = quote_c_style(str, NULL, NULL, 0);
  26         if (!needlen)
  27                 return strdup(str);
  28         xp = xmalloc(needlen + 1);
  29         quote_c_style(str, xp, NULL, 0);
  30         return xp;
  31 }
  32
  33 static char *quote_two(const char *one, const char *two)
  34 {
  35         int need_one = quote_c_style(one, NULL, NULL, 1);
  36         int need_two = quote_c_style(two, NULL, NULL, 1);
  37         char *xp;
  38
  39         if (need_one + need_two) {
  40                 if (!need_one) need_one = strlen(one);
  41                 if (!need_two) need_one = strlen(two);
  42
  43                 xp = xmalloc(need_one + need_two + 3);
  44                 xp[0] = '"';
  45                 quote_c_style(one, xp + 1, NULL, 1);
  46                 quote_c_style(two, xp + need_one + 1, NULL, 1);
  47                 strcpy(xp + need_one + need_two + 1, "\"");
  48                 return xp;
  49         }
  50         need_one = strlen(one);
  51         need_two = strlen(two);
  52         xp = xmalloc(need_one + need_two + 1);
  53         strcpy(xp, one);
  54         strcpy(xp + need_one, two);
  55         return xp;
  56 }
  57
  58 static const char *external_diff(void)
  59 {
  60         static const char *external_diff_cmd = NULL;
  61         static int done_preparing = 0;
  62         const char *env_diff_opts;
  63
  64         if (done_preparing)
  65                 return external_diff_cmd;
  66
  67         /*
  68          * Default values above are meant to match the
  69          * Linux kernel development style.  Examples of
  70          * alternative styles you can specify via environment
  71          * variables are:
  72          *
  73          * GIT_DIFF_OPTS="-c";
  74          */
  75         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  76
  77         /* In case external diff fails... */
  78         env_diff_opts = getenv("GIT_DIFF_OPTS");
  79         if (env_diff_opts) diff_opts = env_diff_opts;
  80
  81         done_preparing = 1;
  82         return external_diff_cmd;
  83 }
  84
  85 #define TEMPFILE_PATH_LEN               50
  86
  87 static struct diff_tempfile {
  88         const char *name; /* filename external diff should read from */
  89         char hex[41];
  90         char mode[10];
  91         char tmp_path[TEMPFILE_PATH_LEN];
  92 } diff_temp[2];
  93
  94 static int count_lines(const char *filename)
  95 {
  96         FILE *in;
  97         int count, ch, completely_empty = 1, nl_just_seen = 0;
  98         in = fopen(filename, "r");
  99         count = 0;
 100         while ((ch = fgetc(in)) != EOF)
 101                 if (ch == '\n') {
 102                         count++;
 103                         nl_just_seen = 1;
 104                         completely_empty = 0;
 105                 }
 106                 else {
 107                         nl_just_seen = 0;
 108                         completely_empty = 0;
 109                 }
 110         fclose(in);
 111         if (completely_empty)
 112                 return 0;
 113         if (!nl_just_seen)
 114                 count++; /* no trailing newline */
 115         return count;
 116 }
 117
 118 static void print_line_count(int count)
 119 {
 120         switch (count) {
 121         case 0:
 122                 printf("0,0");
 123                 break;
 124         case 1:
 125                 printf("1");
 126                 break;
 127         default:
 128                 printf("1,%d", count);
 129                 break;
 130         }
 131 }
 132
 133 static void copy_file(int prefix, const char *filename)
 134 {
 135         FILE *in;
 136         int ch, nl_just_seen = 1;
 137         in = fopen(filename, "r");
 138         while ((ch = fgetc(in)) != EOF) {
 139                 if (nl_just_seen)
 140                         putchar(prefix);
 141                 putchar(ch);
 142                 if (ch == '\n')
 143                         nl_just_seen = 1;
 144                 else
 145                         nl_just_seen = 0;
 146         }
 147         fclose(in);
 148         if (!nl_just_seen)
 149                 printf("\n\\ No newline at end of file\n");
 150 }
 151
 152 static void emit_rewrite_diff(const char *name_a,
 153                               const char *name_b,
 154                               struct diff_tempfile *temp)
 155 {
 156         /* Use temp[i].name as input, name_a and name_b as labels */
 157         int lc_a, lc_b;
 158         lc_a = count_lines(temp[0].name);
 159         lc_b = count_lines(temp[1].name);
 160         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 161         print_line_count(lc_a);
 162         printf(" +");
 163         print_line_count(lc_b);
 164         printf(" @@\n");
 165         if (lc_a)
 166                 copy_file('-', temp[0].name);
 167         if (lc_b)
 168                 copy_file('+', temp[1].name);
 169 }
 170
 171 static void builtin_diff(const char *name_a,
 172                          const char *name_b,
 173                          struct diff_tempfile *temp,
 174                          const char *xfrm_msg,
 175                          int complete_rewrite)
 176 {
 177         int i, next_at, cmd_size;
 178         const char *const diff_cmd = "diff -L%s -L%s";
 179         const char *const diff_arg  = "-- %s %s||:"; /* "||:" is to return 0 */
 180         const char *input_name_sq[2];
 181         const char *label_path[2];
 182         char *cmd;
 183
 184         /* diff_cmd and diff_arg have 4 %s in total which makes
 185          * the sum of these strings 8 bytes larger than required.
 186          * we use 2 spaces around diff-opts, and we need to count
 187          * terminating NUL; we used to subtract 5 here, but we do not
 188          * care about small leaks in this subprocess that is about
 189          * to exec "diff" anymore.
 190          */
 191         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
 192                     + 128);
 193
 194         for (i = 0; i < 2; i++) {
 195                 input_name_sq[i] = sq_quote(temp[i].name);
 196                 if (!strcmp(temp[i].name, "/dev/null"))
 197                         label_path[i] = "/dev/null";
 198                 else if (!i)
 199                         label_path[i] = sq_quote(quote_two("a/", name_a));
 200                 else
 201                         label_path[i] = sq_quote(quote_two("b/", name_b));
 202                 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
 203         }
 204
 205         cmd = xmalloc(cmd_size);
 206
 207         next_at = 0;
 208         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 209                             diff_cmd, label_path[0], label_path[1]);
 210         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 211                             " %s ", diff_opts);
 212         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 213                             diff_arg, input_name_sq[0], input_name_sq[1]);
 214
 215         printf("diff --git %s %s\n",
 216                quote_two("a/", name_a), quote_two("b/", name_b));
 217         if (label_path[0][0] == '/') {
 218                 /* dev/null */
 219                 printf("new file mode %s\n", temp[1].mode);
 220                 if (xfrm_msg && xfrm_msg[0])
 221                         puts(xfrm_msg);
 222         }
 223         else if (label_path[1][0] == '/') {
 224                 printf("deleted file mode %s\n", temp[0].mode);
 225                 if (xfrm_msg && xfrm_msg[0])
 226                         puts(xfrm_msg);
 227         }
 228         else {
 229                 if (strcmp(temp[0].mode, temp[1].mode)) {
 230                         printf("old mode %s\n", temp[0].mode);
 231                         printf("new mode %s\n", temp[1].mode);
 232                 }
 233                 if (xfrm_msg && xfrm_msg[0])
 234                         puts(xfrm_msg);
 235                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 236                         /* we do not run diff between different kind
 237                          * of objects.
 238                          */
 239                         exit(0);
 240                 if (complete_rewrite) {
 241                         fflush(NULL);
 242                         emit_rewrite_diff(name_a, name_b, temp);
 243                         exit(0);
 244                 }
 245         }
 246         fflush(NULL);
 247         execlp("/bin/sh","sh", "-c", cmd, NULL);
 248 }
 249
 250 struct diff_filespec *alloc_filespec(const char *path)
 251 {
 252         int namelen = strlen(path);
 253         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 254
 255         memset(spec, 0, sizeof(*spec));
 256         spec->path = (char *)(spec + 1);
 257         memcpy(spec->path, path, namelen+1);
 258         return spec;
 259 }
 260
 261 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 262                    unsigned short mode)
 263 {
 264         if (mode) {
 265                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 266                 memcpy(spec->sha1, sha1, 20);
 267                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 268         }
 269 }
 270
 271 /*
 272  * Given a name and sha1 pair, if the dircache tells us the file in
 273  * the work tree has that object contents, return true, so that
 274  * prepare_temp_file() does not have to inflate and extract.
 275  */
 276 static int work_tree_matches(const char *name, const unsigned char *sha1)
 277 {
 278         struct cache_entry *ce;
 279         struct stat st;
 280         int pos, len;
 281
 282         /* We do not read the cache ourselves here, because the
 283          * benchmark with my previous version that always reads cache
 284          * shows that it makes things worse for diff-tree comparing
 285          * two linux-2.6 kernel trees in an already checked out work
 286          * tree.  This is because most diff-tree comparisons deal with
 287          * only a small number of files, while reading the cache is
 288          * expensive for a large project, and its cost outweighs the
 289          * savings we get by not inflating the object to a temporary
 290          * file.  Practically, this code only helps when we are used
 291          * by diff-cache --cached, which does read the cache before
 292          * calling us.
 293          */
 294         if (!active_cache)
 295                 return 0;
 296
 297         len = strlen(name);
 298         pos = cache_name_pos(name, len);
 299         if (pos < 0)
 300                 return 0;
 301         ce = active_cache[pos];
 302         if ((lstat(name, &st) < 0) ||
 303             !S_ISREG(st.st_mode) || /* careful! */
 304             ce_match_stat(ce, &st) ||
 305             memcmp(sha1, ce->sha1, 20))
 306                 return 0;
 307         /* we return 1 only when we can stat, it is a regular file,
 308          * stat information matches, and sha1 recorded in the cache
 309          * matches.  I.e. we know the file in the work tree really is
 310          * the same as the <name, sha1> pair.
 311          */
 312         return 1;
 313 }
 314
 315 static struct sha1_size_cache {
 316         unsigned char sha1[20];
 317         unsigned long size;
 318 } **sha1_size_cache;
 319 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 320
 321 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 322                                                  int find_only,
 323                                                  unsigned long size)
 324 {
 325         int first, last;
 326         struct sha1_size_cache *e;
 327
 328         first = 0;
 329         last = sha1_size_cache_nr;
 330         while (last > first) {
 331                 int cmp, next = (last + first) >> 1;
 332                 e = sha1_size_cache[next];
 333                 cmp = memcmp(e->sha1, sha1, 20);
 334                 if (!cmp)
 335                         return e;
 336                 if (cmp < 0) {
 337                         last = next;
 338                         continue;
 339                 }
 340                 first = next+1;
 341         }
 342         /* not found */
 343         if (find_only)
 344                 return NULL;
 345         /* insert to make it at "first" */
 346         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 347                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 348                 sha1_size_cache = xrealloc(sha1_size_cache,
 349                                            sha1_size_cache_alloc *
 350                                            sizeof(*sha1_size_cache));
 351         }
 352         sha1_size_cache_nr++;
 353         if (first < sha1_size_cache_nr)
 354                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 355                         (sha1_size_cache_nr - first - 1) *
 356                         sizeof(*sha1_size_cache));
 357         e = xmalloc(sizeof(struct sha1_size_cache));
 358         sha1_size_cache[first] = e;
 359         memcpy(e->sha1, sha1, 20);
 360         e->size = size;
 361         return e;
 362 }
 363
 364 /*
 365  * While doing rename detection and pickaxe operation, we may need to
 366  * grab the data for the blob (or file) for our own in-core comparison.
 367  * diff_filespec has data and size fields for this purpose.
 368  */
 369 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 370 {
 371         int err = 0;
 372         if (!DIFF_FILE_VALID(s))
 373                 die("internal error: asking to populate invalid file.");
 374         if (S_ISDIR(s->mode))
 375                 return -1;
 376
 377         if (!use_size_cache)
 378                 size_only = 0;
 379
 380         if (s->data)
 381                 return err;
 382         if (!s->sha1_valid ||
 383             work_tree_matches(s->path, s->sha1)) {
 384                 struct stat st;
 385                 int fd;
 386                 if (lstat(s->path, &st) < 0) {
 387                         if (errno == ENOENT) {
 388                         err_empty:
 389                                 err = -1;
 390                         empty:
 391                                 s->data = "";
 392                                 s->size = 0;
 393                                 return err;
 394                         }
 395                 }
 396                 s->size = st.st_size;
 397                 if (!s->size)
 398                         goto empty;
 399                 if (size_only)
 400                         return 0;
 401                 if (S_ISLNK(st.st_mode)) {
 402                         int ret;
 403                         s->data = xmalloc(s->size);
 404                         s->should_free = 1;
 405                         ret = readlink(s->path, s->data, s->size);
 406                         if (ret < 0) {
 407                                 free(s->data);
 408                                 goto err_empty;
 409                         }
 410                         return 0;
 411                 }
 412                 fd = open(s->path, O_RDONLY);
 413                 if (fd < 0)
 414                         goto err_empty;
 415                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 416                 close(fd);
 417                 if (s->data == MAP_FAILED)
 418                         goto err_empty;
 419                 s->should_munmap = 1;
 420         }
 421         else {
 422                 char type[20];
 423                 struct sha1_size_cache *e;
 424
 425                 if (size_only) {
 426                         e = locate_size_cache(s->sha1, 1, 0);
 427                         if (e) {
 428                                 s->size = e->size;
 429                                 return 0;
 430                         }
 431                         if (!sha1_object_info(s->sha1, type, &s->size))
 432                                 locate_size_cache(s->sha1, 0, s->size);
 433                 }
 434                 else {
 435                         s->data = read_sha1_file(s->sha1, type, &s->size);
 436                         s->should_free = 1;
 437                 }
 438         }
 439         return 0;
 440 }
 441
 442 void diff_free_filespec_data(struct diff_filespec *s)
 443 {
 444         if (s->should_free)
 445                 free(s->data);
 446         else if (s->should_munmap)
 447                 munmap(s->data, s->size);
 448         s->should_free = s->should_munmap = 0;
 449         s->data = NULL;
 450 }
 451
 452 static void prep_temp_blob(struct diff_tempfile *temp,
 453                            void *blob,
 454                            unsigned long size,
 455                            const unsigned char *sha1,
 456                            int mode)
 457 {
 458         int fd;
 459
 460         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 461         if (fd < 0)
 462                 die("unable to create temp-file");
 463         if (write(fd, blob, size) != size)
 464                 die("unable to write temp-file");
 465         close(fd);
 466         temp->name = temp->tmp_path;
 467         strcpy(temp->hex, sha1_to_hex(sha1));
 468         temp->hex[40] = 0;
 469         sprintf(temp->mode, "%06o", mode);
 470 }
 471
 472 static void prepare_temp_file(const char *name,
 473                               struct diff_tempfile *temp,
 474                               struct diff_filespec *one)
 475 {
 476         if (!DIFF_FILE_VALID(one)) {
 477         not_a_valid_file:
 478                 /* A '-' entry produces this for file-2, and
 479                  * a '+' entry produces this for file-1.
 480                  */
 481                 temp->name = "/dev/null";
 482                 strcpy(temp->hex, ".");
 483                 strcpy(temp->mode, ".");
 484                 return;
 485         }
 486
 487         if (!one->sha1_valid ||
 488             work_tree_matches(name, one->sha1)) {
 489                 struct stat st;
 490                 if (lstat(name, &st) < 0) {
 491                         if (errno == ENOENT)
 492                                 goto not_a_valid_file;
 493                         die("stat(%s): %s", name, strerror(errno));
 494                 }
 495                 if (S_ISLNK(st.st_mode)) {
 496                         int ret;
 497                         char *buf, buf_[1024];
 498                         buf = ((sizeof(buf_) < st.st_size) ?
 499                                xmalloc(st.st_size) : buf_);
 500                         ret = readlink(name, buf, st.st_size);
 501                         if (ret < 0)
 502                                 die("readlink(%s)", name);
 503                         prep_temp_blob(temp, buf, st.st_size,
 504                                        (one->sha1_valid ?
 505                                         one->sha1 : null_sha1),
 506                                        (one->sha1_valid ?
 507                                         one->mode : S_IFLNK));
 508                 }
 509                 else {
 510                         /* we can borrow from the file in the work tree */
 511                         temp->name = name;
 512                         if (!one->sha1_valid)
 513                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 514                         else
 515                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 516                         /* Even though we may sometimes borrow the
 517                          * contents from the work tree, we always want
 518                          * one->mode.  mode is trustworthy even when
 519                          * !(one->sha1_valid), as long as
 520                          * DIFF_FILE_VALID(one).
 521                          */
 522                         sprintf(temp->mode, "%06o", one->mode);
 523                 }
 524                 return;
 525         }
 526         else {
 527                 if (diff_populate_filespec(one, 0))
 528                         die("cannot read data blob for %s", one->path);
 529                 prep_temp_blob(temp, one->data, one->size,
 530                                one->sha1, one->mode);
 531         }
 532 }
 533
 534 static void remove_tempfile(void)
 535 {
 536         int i;
 537
 538         for (i = 0; i < 2; i++)
 539                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 540                         unlink(diff_temp[i].name);
 541                         diff_temp[i].name = NULL;
 542                 }
 543 }
 544
 545 static void remove_tempfile_on_signal(int signo)
 546 {
 547         remove_tempfile();
 548 }
 549
 550 /* An external diff command takes:
 551  *
 552  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 553  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 554  *
 555  */
 556 static void run_external_diff(const char *pgm,
 557                               const char *name,
 558                               const char *other,
 559                               struct diff_filespec *one,
 560                               struct diff_filespec *two,
 561                               const char *xfrm_msg,
 562                               int complete_rewrite)
 563 {
 564         struct diff_tempfile *temp = diff_temp;
 565         pid_t pid;
 566         int status;
 567         static int atexit_asked = 0;
 568         const char *othername;
 569
 570         othername = (other? other : name);
 571         if (one && two) {
 572                 prepare_temp_file(name, &temp[0], one);
 573                 prepare_temp_file(othername, &temp[1], two);
 574                 if (! atexit_asked &&
 575                     (temp[0].name == temp[0].tmp_path ||
 576                      temp[1].name == temp[1].tmp_path)) {
 577                         atexit_asked = 1;
 578                         atexit(remove_tempfile);
 579                 }
 580                 signal(SIGINT, remove_tempfile_on_signal);
 581         }
 582
 583         fflush(NULL);
 584         pid = fork();
 585         if (pid < 0)
 586                 die("unable to fork");
 587         if (!pid) {
 588                 if (pgm) {
 589                         if (one && two) {
 590                                 const char *exec_arg[10];
 591                                 const char **arg = &exec_arg[0];
 592                                 *arg++ = pgm;
 593                                 *arg++ = name;
 594                                 *arg++ = temp[0].name;
 595                                 *arg++ = temp[0].hex;
 596                                 *arg++ = temp[0].mode;
 597                                 *arg++ = temp[1].name;
 598                                 *arg++ = temp[1].hex;
 599                                 *arg++ = temp[1].mode;
 600                                 if (other) {
 601                                         *arg++ = other;
 602                                         *arg++ = xfrm_msg;
 603                                 }
 604                                 *arg = NULL;
 605                                 execvp(pgm, (char *const*) exec_arg);
 606                         }
 607                         else
 608                                 execlp(pgm, pgm, name, NULL);
 609                 }
 610                 /*
 611                  * otherwise we use the built-in one.
 612                  */
 613                 if (one && two)
 614                         builtin_diff(name, othername, temp, xfrm_msg,
 615                                      complete_rewrite);
 616                 else
 617                         printf("* Unmerged path %s\n", name);
 618                 exit(0);
 619         }
 620         if (waitpid(pid, &status, 0) < 0 ||
 621             !WIFEXITED(status) || WEXITSTATUS(status)) {
 622                 /* Earlier we did not check the exit status because
 623                  * diff exits non-zero if files are different, and
 624                  * we are not interested in knowing that.  It was a
 625                  * mistake which made it harder to quit a diff-*
 626                  * session that uses the git-apply-patch-script as
 627                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 628                  * should also exit non-zero only when it wants to
 629                  * abort the entire diff-* session.
 630                  */
 631                 remove_tempfile();
 632                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 633                 exit(1);
 634         }
 635         remove_tempfile();
 636 }
 637
 638 static void diff_fill_sha1_info(struct diff_filespec *one)
 639 {
 640         if (DIFF_FILE_VALID(one)) {
 641                 if (!one->sha1_valid) {
 642                         struct stat st;
 643                         if (stat(one->path, &st) < 0)
 644                                 die("stat %s", one->path);
 645                         if (index_path(one->sha1, one->path, &st, 0))
 646                                 die("cannot hash %s\n", one->path);
 647                 }
 648         }
 649         else
 650                 memset(one->sha1, 0, 20);
 651 }
 652
 653 static void run_diff(struct diff_filepair *p, struct diff_options *o)
 654 {
 655         const char *pgm = external_diff();
 656         char msg[PATH_MAX*2+300], *xfrm_msg;
 657         struct diff_filespec *one;
 658         struct diff_filespec *two;
 659         const char *name;
 660         const char *other;
 661         char *name_munged, *other_munged;
 662         int complete_rewrite = 0;
 663         int len;
 664
 665         if (DIFF_PAIR_UNMERGED(p)) {
 666                 /* unmerged */
 667                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 668                                   0);
 669                 return;
 670         }
 671
 672         name = p->one->path;
 673         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 674         name_munged = quote_one(name);
 675         other_munged = quote_one(other);
 676         one = p->one; two = p->two;
 677
 678         diff_fill_sha1_info(one);
 679         diff_fill_sha1_info(two);
 680
 681         len = 0;
 682         switch (p->status) {
 683         case DIFF_STATUS_COPIED:
 684                 len += snprintf(msg + len, sizeof(msg) - len,
 685                                 "similarity index %d%%\n"
 686                                 "copy from %s\n"
 687                                 "copy to %s\n",
 688                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 689                                 name_munged, other_munged);
 690                 break;
 691         case DIFF_STATUS_RENAMED:
 692                 len += snprintf(msg + len, sizeof(msg) - len,
 693                                 "similarity index %d%%\n"
 694                                 "rename from %s\n"
 695                                 "rename to %s\n",
 696                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 697                                 name_munged, other_munged);
 698                 break;
 699         case DIFF_STATUS_MODIFIED:
 700                 if (p->score) {
 701                         len += snprintf(msg + len, sizeof(msg) - len,
 702                                         "dissimilarity index %d%%\n",
 703                                         (int)(0.5 + p->score *
 704                                               100.0/MAX_SCORE));
 705                         complete_rewrite = 1;
 706                         break;
 707                 }
 708                 /* fallthru */
 709         default:
 710                 /* nothing */
 711                 ;
 712         }
 713
 714         if (memcmp(one->sha1, two->sha1, 20)) {
 715                 char one_sha1[41];
 716                 const char *index_fmt = o->full_index ? "index %s..%s" : "index %.7s..%.7s";
 717                 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
 718
 719                 len += snprintf(msg + len, sizeof(msg) - len,
 720                                 index_fmt, one_sha1, sha1_to_hex(two->sha1));
 721                 if (one->mode == two->mode)
 722                         len += snprintf(msg + len, sizeof(msg) - len,
 723                                         " %06o", one->mode);
 724                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
 725         }
 726
 727         if (len)
 728                 msg[--len] = 0;
 729         xfrm_msg = len ? msg : NULL;
 730
 731         if (!pgm &&
 732             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 733             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 734                 /* a filepair that changes between file and symlink
 735                  * needs to be split into deletion and creation.
 736                  */
 737                 struct diff_filespec *null = alloc_filespec(two->path);
 738                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 739                 free(null);
 740                 null = alloc_filespec(one->path);
 741                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 742                 free(null);
 743         }
 744         else
 745                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 746                                   complete_rewrite);
 747
 748         free(name_munged);
 749         free(other_munged);
 750 }
 751
 752 void diff_setup(struct diff_options *options)
 753 {
 754         memset(options, 0, sizeof(*options));
 755         options->output_format = DIFF_FORMAT_RAW;
 756         options->line_termination = '\n';
 757         options->break_opt = -1;
 758         options->rename_limit = -1;
 759
 760         options->change = diff_change;
 761         options->add_remove = diff_addremove;
 762 }
 763
 764 int diff_setup_done(struct diff_options *options)
 765 {
 766         if ((options->find_copies_harder &&
 767              options->detect_rename != DIFF_DETECT_COPY) ||
 768             (0 <= options->rename_limit && !options->detect_rename))
 769                 return -1;
 770         if (options->detect_rename && options->rename_limit < 0)
 771                 options->rename_limit = diff_rename_limit_default;
 772         if (options->setup & DIFF_SETUP_USE_CACHE) {
 773                 if (!active_cache)
 774                         /* read-cache does not die even when it fails
 775                          * so it is safe for us to do this here.  Also
 776                          * it does not smudge active_cache or active_nr
 777                          * when it fails, so we do not have to worry about
 778                          * cleaning it up oufselves either.
 779                          */
 780                         read_cache();
 781         }
 782         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
 783                 use_size_cache = 1;
 784
 785         return 0;
 786 }
 787
 788 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 789 {
 790         const char *arg = av[0];
 791         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
 792                 options->output_format = DIFF_FORMAT_PATCH;
 793         else if (!strcmp(arg, "-z"))
 794                 options->line_termination = 0;
 795         else if (!strncmp(arg, "-l", 2))
 796                 options->rename_limit = strtoul(arg+2, NULL, 10);
 797         else if (!strcmp(arg, "--full-index"))
 798                 options->full_index = 1;
 799         else if (!strcmp(arg, "--name-only"))
 800                 options->output_format = DIFF_FORMAT_NAME;
 801         else if (!strcmp(arg, "--name-status"))
 802                 options->output_format = DIFF_FORMAT_NAME_STATUS;
 803         else if (!strcmp(arg, "-R"))
 804                 options->reverse_diff = 1;
 805         else if (!strncmp(arg, "-S", 2))
 806                 options->pickaxe = arg + 2;
 807         else if (!strcmp(arg, "-s"))
 808                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
 809         else if (!strncmp(arg, "-O", 2))
 810                 options->orderfile = arg + 2;
 811         else if (!strncmp(arg, "--diff-filter=", 14))
 812                 options->filter = arg + 14;
 813         else if (!strcmp(arg, "--pickaxe-all"))
 814                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
 815         else if (!strncmp(arg, "-B", 2)) {
 816                 if ((options->break_opt =
 817                      diff_scoreopt_parse(arg)) == -1)
 818                         return -1;
 819         }
 820         else if (!strncmp(arg, "-M", 2)) {
 821                 if ((options->rename_score =
 822                      diff_scoreopt_parse(arg)) == -1)
 823                         return -1;
 824                 options->detect_rename = DIFF_DETECT_RENAME;
 825         }
 826         else if (!strncmp(arg, "-C", 2)) {
 827                 if ((options->rename_score =
 828                      diff_scoreopt_parse(arg)) == -1)
 829                         return -1;
 830                 options->detect_rename = DIFF_DETECT_COPY;
 831         }
 832         else if (!strcmp(arg, "--find-copies-harder"))
 833                 options->find_copies_harder = 1;
 834         else
 835                 return 0;
 836         return 1;
 837 }
 838
 839 static int parse_num(const char **cp_p)
 840 {
 841         int num, scale, ch, cnt;
 842         const char *cp = *cp_p;
 843
 844         cnt = num = 0;
 845         scale = 1;
 846         while ('0' <= (ch = *cp) && ch <= '9') {
 847                 if (cnt++ < 5) {
 848                         /* We simply ignore more than 5 digits precision. */
 849                         scale *= 10;
 850                         num = num * 10 + ch - '0';
 851                 }
 852                 cp++;
 853         }
 854         *cp_p = cp;
 855
 856         /* special case: -M100 would mean 1.0 not 0.1 */
 857         if (num == 100 && scale == 1000)
 858                 return MAX_SCORE;
 859
 860         /* user says num divided by scale and we say internally that
 861          * is MAX_SCORE * num / scale.
 862          */
 863         return (MAX_SCORE * num / scale);
 864 }
 865
 866 int diff_scoreopt_parse(const char *opt)
 867 {
 868         int opt1, opt2, cmd;
 869
 870         if (*opt++ != '-')
 871                 return -1;
 872         cmd = *opt++;
 873         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 874                 return -1; /* that is not a -M, -C nor -B option */
 875
 876         opt1 = parse_num(&opt);
 877         if (cmd != 'B')
 878                 opt2 = 0;
 879         else {
 880                 if (*opt == 0)
 881                         opt2 = 0;
 882                 else if (*opt != '/')
 883                         return -1; /* we expect -B80/99 or -B80 */
 884                 else {
 885                         opt++;
 886                         opt2 = parse_num(&opt);
 887                 }
 888         }
 889         if (*opt != 0)
 890                 return -1;
 891         return opt1 | (opt2 << 16);
 892 }
 893
 894 struct diff_queue_struct diff_queued_diff;
 895
 896 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 897 {
 898         if (queue->alloc <= queue->nr) {
 899                 queue->alloc = alloc_nr(queue->alloc);
 900                 queue->queue = xrealloc(queue->queue,
 901                                         sizeof(dp) * queue->alloc);
 902         }
 903         queue->queue[queue->nr++] = dp;
 904 }
 905
 906 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 907                                  struct diff_filespec *one,
 908                                  struct diff_filespec *two)
 909 {
 910         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 911         dp->one = one;
 912         dp->two = two;
 913         dp->score = 0;
 914         dp->status = 0;
 915         dp->source_stays = 0;
 916         dp->broken_pair = 0;
 917         if (queue)
 918                 diff_q(queue, dp);
 919         return dp;
 920 }
 921
 922 void diff_free_filepair(struct diff_filepair *p)
 923 {
 924         diff_free_filespec_data(p->one);
 925         diff_free_filespec_data(p->two);
 926         free(p->one);
 927         free(p->two);
 928         free(p);
 929 }
 930
 931 static void diff_flush_raw(struct diff_filepair *p,
 932                            int line_termination,
 933                            int inter_name_termination,
 934                            int output_format)
 935 {
 936         int two_paths;
 937         char status[10];
 938         const char *path_one, *path_two;
 939
 940         path_one = p->one->path;
 941         path_two = p->two->path;
 942         if (line_termination) {
 943                 path_one = quote_one(path_one);
 944                 path_two = quote_one(path_two);
 945         }
 946
 947         if (p->score)
 948                 sprintf(status, "%c%03d", p->status,
 949                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 950         else {
 951                 status[0] = p->status;
 952                 status[1] = 0;
 953         }
 954         switch (p->status) {
 955         case DIFF_STATUS_COPIED:
 956         case DIFF_STATUS_RENAMED:
 957                 two_paths = 1;
 958                 break;
 959         case DIFF_STATUS_ADDED:
 960         case DIFF_STATUS_DELETED:
 961                 two_paths = 0;
 962                 break;
 963         default:
 964                 two_paths = 0;
 965                 break;
 966         }
 967         if (output_format != DIFF_FORMAT_NAME_STATUS) {
 968                 printf(":%06o %06o %s ",
 969                        p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 970                 printf("%s ", sha1_to_hex(p->two->sha1));
 971         }
 972         printf("%s%c%s", status, inter_name_termination, path_one);
 973         if (two_paths)
 974                 printf("%c%s", inter_name_termination, path_two);
 975         putchar(line_termination);
 976         if (path_one != p->one->path)
 977                 free((void*)path_one);
 978         if (path_two != p->two->path)
 979                 free((void*)path_two);
 980 }
 981
 982 static void diff_flush_name(struct diff_filepair *p,
 983                             int inter_name_termination,
 984                             int line_termination)
 985 {
 986         char *path = p->two->path;
 987
 988         if (line_termination)
 989                 path = quote_one(p->two->path);
 990         else
 991                 path = p->two->path;
 992         printf("%s%c", path, line_termination);
 993         if (p->two->path != path)
 994                 free(path);
 995 }
 996
 997 int diff_unmodified_pair(struct diff_filepair *p)
 998 {
 999         /* This function is written stricter than necessary to support
1000          * the currently implemented transformers, but the idea is to
1001          * let transformers to produce diff_filepairs any way they want,
1002          * and filter and clean them up here before producing the output.
1003          */
1004         struct diff_filespec *one, *two;
1005
1006         if (DIFF_PAIR_UNMERGED(p))
1007                 return 0; /* unmerged is interesting */
1008
1009         one = p->one;
1010         two = p->two;
1011
1012         /* deletion, addition, mode or type change
1013          * and rename are all interesting.
1014          */
1015         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1016             DIFF_PAIR_MODE_CHANGED(p) ||
1017             strcmp(one->path, two->path))
1018                 return 0;
1019
1020         /* both are valid and point at the same path.  that is, we are
1021          * dealing with a change.
1022          */
1023         if (one->sha1_valid && two->sha1_valid &&
1024             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1025                 return 1; /* no change */
1026         if (!one->sha1_valid && !two->sha1_valid)
1027                 return 1; /* both look at the same file on the filesystem. */
1028         return 0;
1029 }
1030
1031 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1032 {
1033         if (diff_unmodified_pair(p))
1034                 return;
1035
1036         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1037             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1038                 return; /* no tree diffs in patch format */
1039
1040         run_diff(p, o);
1041 }
1042
1043 int diff_queue_is_empty(void)
1044 {
1045         struct diff_queue_struct *q = &diff_queued_diff;
1046         int i;
1047         for (i = 0; i < q->nr; i++)
1048                 if (!diff_unmodified_pair(q->queue[i]))
1049                         return 0;
1050         return 1;
1051 }
1052
1053 #if DIFF_DEBUG
1054 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1055 {
1056         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1057                 x, one ? one : "",
1058                 s->path,
1059                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1060                 s->mode,
1061                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1062         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1063                 x, one ? one : "",
1064                 s->size, s->xfrm_flags);
1065 }
1066
1067 void diff_debug_filepair(const struct diff_filepair *p, int i)
1068 {
1069         diff_debug_filespec(p->one, i, "one");
1070         diff_debug_filespec(p->two, i, "two");
1071         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1072                 p->score, p->status ? p->status : '?',
1073                 p->source_stays, p->broken_pair);
1074 }
1075
1076 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1077 {
1078         int i;
1079         if (msg)
1080                 fprintf(stderr, "%s\n", msg);
1081         fprintf(stderr, "q->nr = %d\n", q->nr);
1082         for (i = 0; i < q->nr; i++) {
1083                 struct diff_filepair *p = q->queue[i];
1084                 diff_debug_filepair(p, i);
1085         }
1086 }
1087 #endif
1088
1089 static void diff_resolve_rename_copy(void)
1090 {
1091         int i, j;
1092         struct diff_filepair *p, *pp;
1093         struct diff_queue_struct *q = &diff_queued_diff;
1094
1095         diff_debug_queue("resolve-rename-copy", q);
1096
1097         for (i = 0; i < q->nr; i++) {
1098                 p = q->queue[i];
1099                 p->status = 0; /* undecided */
1100                 if (DIFF_PAIR_UNMERGED(p))
1101                         p->status = DIFF_STATUS_UNMERGED;
1102                 else if (!DIFF_FILE_VALID(p->one))
1103                         p->status = DIFF_STATUS_ADDED;
1104                 else if (!DIFF_FILE_VALID(p->two))
1105                         p->status = DIFF_STATUS_DELETED;
1106                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1107                         p->status = DIFF_STATUS_TYPE_CHANGED;
1108
1109                 /* from this point on, we are dealing with a pair
1110                  * whose both sides are valid and of the same type, i.e.
1111                  * either in-place edit or rename/copy edit.
1112                  */
1113                 else if (DIFF_PAIR_RENAME(p)) {
1114                         if (p->source_stays) {
1115                                 p->status = DIFF_STATUS_COPIED;
1116                                 continue;
1117                         }
1118                         /* See if there is some other filepair that
1119                          * copies from the same source as us.  If so
1120                          * we are a copy.  Otherwise we are either a
1121                          * copy if the path stays, or a rename if it
1122                          * does not, but we already handled "stays" case.
1123                          */
1124                         for (j = i + 1; j < q->nr; j++) {
1125                                 pp = q->queue[j];
1126                                 if (strcmp(pp->one->path, p->one->path))
1127                                         continue; /* not us */
1128                                 if (!DIFF_PAIR_RENAME(pp))
1129                                         continue; /* not a rename/copy */
1130                                 /* pp is a rename/copy from the same source */
1131                                 p->status = DIFF_STATUS_COPIED;
1132                                 break;
1133                         }
1134                         if (!p->status)
1135                                 p->status = DIFF_STATUS_RENAMED;
1136                 }
1137                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1138                          p->one->mode != p->two->mode)
1139                         p->status = DIFF_STATUS_MODIFIED;
1140                 else {
1141                         /* This is a "no-change" entry and should not
1142                          * happen anymore, but prepare for broken callers.
1143                          */
1144                         error("feeding unmodified %s to diffcore",
1145                               p->one->path);
1146                         p->status = DIFF_STATUS_UNKNOWN;
1147                 }
1148         }
1149         diff_debug_queue("resolve-rename-copy done", q);
1150 }
1151
1152 void diff_flush(struct diff_options *options)
1153 {
1154         struct diff_queue_struct *q = &diff_queued_diff;
1155         int i;
1156         int inter_name_termination = '\t';
1157         int diff_output_format = options->output_format;
1158         int line_termination = options->line_termination;
1159
1160         if (!line_termination)
1161                 inter_name_termination = 0;
1162
1163         for (i = 0; i < q->nr; i++) {
1164                 struct diff_filepair *p = q->queue[i];
1165                 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1166                     (p->status == DIFF_STATUS_UNKNOWN))
1167                         continue;
1168                 if (p->status == 0)
1169                         die("internal error in diff-resolve-rename-copy");
1170                 switch (diff_output_format) {
1171                 case DIFF_FORMAT_PATCH:
1172                         diff_flush_patch(p, options);
1173                         break;
1174                 case DIFF_FORMAT_RAW:
1175                 case DIFF_FORMAT_NAME_STATUS:
1176                         diff_flush_raw(p, line_termination,
1177                                        inter_name_termination,
1178                                        diff_output_format);
1179                         break;
1180                 case DIFF_FORMAT_NAME:
1181                         diff_flush_name(p,
1182                                         inter_name_termination,
1183                                         line_termination);
1184                         break;
1185                 }
1186                 diff_free_filepair(q->queue[i]);
1187         }
1188         free(q->queue);
1189         q->queue = NULL;
1190         q->nr = q->alloc = 0;
1191 }
1192
1193 static void diffcore_apply_filter(const char *filter)
1194 {
1195         int i;
1196         struct diff_queue_struct *q = &diff_queued_diff;
1197         struct diff_queue_struct outq;
1198         outq.queue = NULL;
1199         outq.nr = outq.alloc = 0;
1200
1201         if (!filter)
1202                 return;
1203
1204         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1205                 int found;
1206                 for (i = found = 0; !found && i < q->nr; i++) {
1207                         struct diff_filepair *p = q->queue[i];
1208                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1209                              ((p->score &&
1210                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1211                               (!p->score &&
1212                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1213                             ((p->status != DIFF_STATUS_MODIFIED) &&
1214                              strchr(filter, p->status)))
1215                                 found++;
1216                 }
1217                 if (found)
1218                         return;
1219
1220                 /* otherwise we will clear the whole queue
1221                  * by copying the empty outq at the end of this
1222                  * function, but first clear the current entries
1223                  * in the queue.
1224                  */
1225                 for (i = 0; i < q->nr; i++)
1226                         diff_free_filepair(q->queue[i]);
1227         }
1228         else {
1229                 /* Only the matching ones */
1230                 for (i = 0; i < q->nr; i++) {
1231                         struct diff_filepair *p = q->queue[i];
1232
1233                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1234                              ((p->score &&
1235                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1236                               (!p->score &&
1237                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1238                             ((p->status != DIFF_STATUS_MODIFIED) &&
1239                              strchr(filter, p->status)))
1240                                 diff_q(&outq, p);
1241                         else
1242                                 diff_free_filepair(p);
1243                 }
1244         }
1245         free(q->queue);
1246         *q = outq;
1247 }
1248
1249 void diffcore_std(struct diff_options *options)
1250 {
1251         if (options->paths && options->paths[0])
1252                 diffcore_pathspec(options->paths);
1253         if (options->break_opt != -1)
1254                 diffcore_break(options->break_opt);
1255         if (options->detect_rename)
1256                 diffcore_rename(options);
1257         if (options->break_opt != -1)
1258                 diffcore_merge_broken();
1259         if (options->pickaxe)
1260                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1261         if (options->orderfile)
1262                 diffcore_order(options->orderfile);
1263         diff_resolve_rename_copy();
1264         diffcore_apply_filter(options->filter);
1265 }
1266
1267
1268 void diffcore_std_no_resolve(struct diff_options *options)
1269 {
1270         if (options->pickaxe)
1271                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1272         if (options->orderfile)
1273                 diffcore_order(options->orderfile);
1274         diffcore_apply_filter(options->filter);
1275 }
1276
1277 void diff_addremove(struct diff_options *options,
1278                     int addremove, unsigned mode,
1279                     const unsigned char *sha1,
1280                     const char *base, const char *path)
1281 {
1282         char concatpath[PATH_MAX];
1283         struct diff_filespec *one, *two;
1284
1285         /* This may look odd, but it is a preparation for
1286          * feeding "there are unchanged files which should
1287          * not produce diffs, but when you are doing copy
1288          * detection you would need them, so here they are"
1289          * entries to the diff-core.  They will be prefixed
1290          * with something like '=' or '*' (I haven't decided
1291          * which but should not make any difference).
1292          * Feeding the same new and old to diff_change()
1293          * also has the same effect.
1294          * Before the final output happens, they are pruned after
1295          * merged into rename/copy pairs as appropriate.
1296          */
1297         if (options->reverse_diff)
1298                 addremove = (addremove == '+' ? '-' :
1299                              addremove == '-' ? '+' : addremove);
1300
1301         if (!path) path = "";
1302         sprintf(concatpath, "%s%s", base, path);
1303         one = alloc_filespec(concatpath);
1304         two = alloc_filespec(concatpath);
1305
1306         if (addremove != '+')
1307                 fill_filespec(one, sha1, mode);
1308         if (addremove != '-')
1309                 fill_filespec(two, sha1, mode);
1310
1311         diff_queue(&diff_queued_diff, one, two);
1312 }
1313
1314 void diff_change(struct diff_options *options,
1315                  unsigned old_mode, unsigned new_mode,
1316                  const unsigned char *old_sha1,
1317                  const unsigned char *new_sha1,
1318                  const char *base, const char *path)
1319 {
1320         char concatpath[PATH_MAX];
1321         struct diff_filespec *one, *two;
1322
1323         if (options->reverse_diff) {
1324                 unsigned tmp;
1325                 const unsigned char *tmp_c;
1326                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1327                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1328         }
1329         if (!path) path = "";
1330         sprintf(concatpath, "%s%s", base, path);
1331         one = alloc_filespec(concatpath);
1332         two = alloc_filespec(concatpath);
1333         fill_filespec(one, old_sha1, old_mode);
1334         fill_filespec(two, new_sha1, new_mode);
1335
1336         diff_queue(&diff_queued_diff, one, two);
1337 }
1338
1339 void diff_unmerge(struct diff_options *options,
1340                   const char *path)
1341 {
1342         struct diff_filespec *one, *two;
1343         one = alloc_filespec(path);
1344         two = alloc_filespec(path);
1345         diff_queue(&diff_queued_diff, one, two);
1346 }