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 "diff.h"
   9 #include "diffcore.h"
  10
  11 static const char *diff_opts = "-pu";
  12 static unsigned char null_sha1[20] = { 0, };
  13
  14 static int reverse_diff;
  15 static int use_size_cache;
  16
  17 static const char *external_diff(void)
  18 {
  19         static const char *external_diff_cmd = NULL;
  20         static int done_preparing = 0;
  21
  22         if (done_preparing)
  23                 return external_diff_cmd;
  24
  25         /*
  26          * Default values above are meant to match the
  27          * Linux kernel development style.  Examples of
  28          * alternative styles you can specify via environment
  29          * variables are:
  30          *
  31          * GIT_DIFF_OPTS="-c";
  32          */
  33         if (gitenv("GIT_EXTERNAL_DIFF"))
  34                 external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
  35
  36         /* In case external diff fails... */
  37         diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
  38
  39         done_preparing = 1;
  40         return external_diff_cmd;
  41 }
  42
  43 /* Help to copy the thing properly quoted for the shell safety.
  44  * any single quote is replaced with '\'', and the caller is
  45  * expected to enclose the result within a single quote pair.
  46  *
  47  * E.g.
  48  *  original     sq_expand     result
  49  *  name     ==> name      ==> 'name'
  50  *  a b      ==> a b       ==> 'a b'
  51  *  a'b      ==> a'\''b    ==> 'a'\''b'
  52  */
  53 static char *sq_expand(const char *src)
  54 {
  55         static char *buf = NULL;
  56         int cnt, c;
  57         const char *cp;
  58         char *bp;
  59
  60         /* count bytes needed to store the quoted string. */
  61         for (cnt = 1, cp = src; *cp; cnt++, cp++)
  62                 if (*cp == '\'')
  63                         cnt += 3;
  64
  65         buf = xmalloc(cnt);
  66         bp = buf;
  67         while ((c = *src++)) {
  68                 if (c != '\'')
  69                         *bp++ = c;
  70                 else {
  71                         bp = strcpy(bp, "'\\''");
  72                         bp += 4;
  73                 }
  74         }
  75         *bp = 0;
  76         return buf;
  77 }
  78
  79 static struct diff_tempfile {
  80         const char *name; /* filename external diff should read from */
  81         char hex[41];
  82         char mode[10];
  83         char tmp_path[50];
  84 } diff_temp[2];
  85
  86 static int count_lines(const char *filename)
  87 {
  88         FILE *in;
  89         int count, ch, completely_empty = 1, nl_just_seen = 0;
  90         in = fopen(filename, "r");
  91         count = 0;
  92         while ((ch = fgetc(in)) != EOF)
  93                 if (ch == '\n') {
  94                         count++;
  95                         nl_just_seen = 1;
  96                         completely_empty = 0;
  97                 }
  98                 else {
  99                         nl_just_seen = 0;
 100                         completely_empty = 0;
 101                 }
 102         fclose(in);
 103         if (completely_empty)
 104                 return 0;
 105         if (!nl_just_seen)
 106                 count++; /* no trailing newline */
 107         return count;
 108 }
 109
 110 static void print_line_count(int count)
 111 {
 112         switch (count) {
 113         case 0:
 114                 printf("0,0");
 115                 break;
 116         case 1:
 117                 printf("1");
 118                 break;
 119         default:
 120                 printf("1,%d", count);
 121                 break;
 122         }
 123 }
 124
 125 static void copy_file(int prefix, const char *filename)
 126 {
 127         FILE *in;
 128         int ch, nl_just_seen = 1;
 129         in = fopen(filename, "r");
 130         while ((ch = fgetc(in)) != EOF) {
 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         fclose(in);
 140         if (!nl_just_seen)
 141                 printf("\n\\ No newline at end of file\n");
 142 }
 143
 144 static void emit_rewrite_diff(const char *name_a,
 145                               const char *name_b,
 146                               struct diff_tempfile *temp)
 147 {
 148         /* Use temp[i].name as input, name_a and name_b as labels */
 149         int lc_a, lc_b;
 150         lc_a = count_lines(temp[0].name);
 151         lc_b = count_lines(temp[1].name);
 152         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 153         print_line_count(lc_a);
 154         printf(" +");
 155         print_line_count(lc_b);
 156         printf(" @@\n");
 157         if (lc_a)
 158                 copy_file('-', temp[0].name);
 159         if (lc_b)
 160                 copy_file('+', temp[1].name);
 161 }
 162
 163 static void builtin_diff(const char *name_a,
 164                          const char *name_b,
 165                          struct diff_tempfile *temp,
 166                          const char *xfrm_msg,
 167                          int complete_rewrite)
 168 {
 169         int i, next_at, cmd_size;
 170         const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
 171         const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
 172         const char *input_name_sq[2];
 173         const char *path0[2];
 174         const char *path1[2];
 175         const char *name_sq[2];
 176         char *cmd;
 177
 178         name_sq[0] = sq_expand(name_a);
 179         name_sq[1] = sq_expand(name_b);
 180
 181         /* diff_cmd and diff_arg have 6 %s in total which makes
 182          * the sum of these strings 12 bytes larger than required.
 183          * we use 2 spaces around diff-opts, and we need to count
 184          * terminating NUL, so we subtract 9 here.
 185          */
 186         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 187                         strlen(diff_arg) - 9);
 188         for (i = 0; i < 2; i++) {
 189                 input_name_sq[i] = sq_expand(temp[i].name);
 190                 if (!strcmp(temp[i].name, "/dev/null")) {
 191                         path0[i] = "/dev/null";
 192                         path1[i] = "";
 193                 } else {
 194                         path0[i] = i ? "b/" : "a/";
 195                         path1[i] = name_sq[i];
 196                 }
 197                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 198                              strlen(input_name_sq[i]));
 199         }
 200
 201         cmd = xmalloc(cmd_size);
 202
 203         next_at = 0;
 204         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 205                             diff_cmd,
 206                             path0[0], path1[0], path0[1], path1[1]);
 207         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 208                             " %s ", diff_opts);
 209         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 210                             diff_arg, input_name_sq[0], input_name_sq[1]);
 211
 212         printf("diff --git a/%s b/%s\n", name_a, name_b);
 213         if (!path1[0][0]) {
 214                 printf("new file mode %s\n", temp[1].mode);
 215                 if (xfrm_msg && xfrm_msg[0])
 216                         puts(xfrm_msg);
 217         }
 218         else if (!path1[1][0]) {
 219                 printf("deleted file mode %s\n", temp[0].mode);
 220                 if (xfrm_msg && xfrm_msg[0])
 221                         puts(xfrm_msg);
 222         }
 223         else {
 224                 if (strcmp(temp[0].mode, temp[1].mode)) {
 225                         printf("old mode %s\n", temp[0].mode);
 226                         printf("new mode %s\n", temp[1].mode);
 227                 }
 228                 if (xfrm_msg && xfrm_msg[0])
 229                         puts(xfrm_msg);
 230                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 231                         /* we do not run diff between different kind
 232                          * of objects.
 233                          */
 234                         exit(0);
 235                 if (complete_rewrite) {
 236                         fflush(NULL);
 237                         emit_rewrite_diff(name_a, name_b, temp);
 238                         exit(0);
 239                 }
 240         }
 241         fflush(NULL);
 242         execlp("/bin/sh","sh", "-c", cmd, NULL);
 243 }
 244
 245 struct diff_filespec *alloc_filespec(const char *path)
 246 {
 247         int namelen = strlen(path);
 248         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 249         spec->path = (char *)(spec + 1);
 250         strcpy(spec->path, path);
 251         spec->should_free = spec->should_munmap = 0;
 252         spec->xfrm_flags = 0;
 253         spec->size = 0;
 254         spec->data = NULL;
 255         spec->mode = 0;
 256         memset(spec->sha1, 0, 20);
 257         return spec;
 258 }
 259
 260 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 261                    unsigned short mode)
 262 {
 263         if (mode) {
 264                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 265                 memcpy(spec->sha1, sha1, 20);
 266                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 267         }
 268 }
 269
 270 /*
 271  * Given a name and sha1 pair, if the dircache tells us the file in
 272  * the work tree has that object contents, return true, so that
 273  * prepare_temp_file() does not have to inflate and extract.
 274  */
 275 static int work_tree_matches(const char *name, const unsigned char *sha1)
 276 {
 277         struct cache_entry *ce;
 278         struct stat st;
 279         int pos, len;
 280
 281         /* We do not read the cache ourselves here, because the
 282          * benchmark with my previous version that always reads cache
 283          * shows that it makes things worse for diff-tree comparing
 284          * two linux-2.6 kernel trees in an already checked out work
 285          * tree.  This is because most diff-tree comparisons deal with
 286          * only a small number of files, while reading the cache is
 287          * expensive for a large project, and its cost outweighs the
 288          * savings we get by not inflating the object to a temporary
 289          * file.  Practically, this code only helps when we are used
 290          * by diff-cache --cached, which does read the cache before
 291          * calling us.
 292          */
 293         if (!active_cache)
 294                 return 0;
 295
 296         len = strlen(name);
 297         pos = cache_name_pos(name, len);
 298         if (pos < 0)
 299                 return 0;
 300         ce = active_cache[pos];
 301         if ((lstat(name, &st) < 0) ||
 302             !S_ISREG(st.st_mode) || /* careful! */
 303             ce_match_stat(ce, &st) ||
 304             memcmp(sha1, ce->sha1, 20))
 305                 return 0;
 306         /* we return 1 only when we can stat, it is a regular file,
 307          * stat information matches, and sha1 recorded in the cache
 308          * matches.  I.e. we know the file in the work tree really is
 309          * the same as the <name, sha1> pair.
 310          */
 311         return 1;
 312 }
 313
 314 static struct sha1_size_cache {
 315         unsigned char sha1[20];
 316         unsigned long size;
 317 } **sha1_size_cache;
 318 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 319
 320 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 321                                                  int find_only,
 322                                                  unsigned long size)
 323 {
 324         int first, last;
 325         struct sha1_size_cache *e;
 326
 327         first = 0;
 328         last = sha1_size_cache_nr;
 329         while (last > first) {
 330                 int cmp, next = (last + first) >> 1;
 331                 e = sha1_size_cache[next];
 332                 cmp = memcmp(e->sha1, sha1, 20);
 333                 if (!cmp)
 334                         return e;
 335                 if (cmp < 0) {
 336                         last = next;
 337                         continue;
 338                 }
 339                 first = next+1;
 340         }
 341         /* not found */
 342         if (find_only)
 343                 return NULL;
 344         /* insert to make it at "first" */
 345         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 346                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 347                 sha1_size_cache = xrealloc(sha1_size_cache,
 348                                            sha1_size_cache_alloc *
 349                                            sizeof(*sha1_size_cache));
 350         }
 351         sha1_size_cache_nr++;
 352         if (first < sha1_size_cache_nr)
 353                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 354                         (sha1_size_cache_nr - first - 1) *
 355                         sizeof(*sha1_size_cache));
 356         e = xmalloc(sizeof(struct sha1_size_cache));
 357         sha1_size_cache[first] = e;
 358         memcpy(e->sha1, sha1, 20);
 359         e->size = size;
 360         return e;
 361 }
 362
 363 /*
 364  * While doing rename detection and pickaxe operation, we may need to
 365  * grab the data for the blob (or file) for our own in-core comparison.
 366  * diff_filespec has data and size fields for this purpose.
 367  */
 368 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 369 {
 370         int err = 0;
 371         if (!DIFF_FILE_VALID(s))
 372                 die("internal error: asking to populate invalid file.");
 373         if (S_ISDIR(s->mode))
 374                 return -1;
 375
 376         if (!use_size_cache)
 377                 size_only = 0;
 378
 379         if (s->data)
 380                 return err;
 381         if (!s->sha1_valid ||
 382             work_tree_matches(s->path, s->sha1)) {
 383                 struct stat st;
 384                 int fd;
 385                 if (lstat(s->path, &st) < 0) {
 386                         if (errno == ENOENT) {
 387                         err_empty:
 388                                 err = -1;
 389                         empty:
 390                                 s->data = "";
 391                                 s->size = 0;
 392                                 return err;
 393                         }
 394                 }
 395                 s->size = st.st_size;
 396                 if (!s->size)
 397                         goto empty;
 398                 if (size_only)
 399                         return 0;
 400                 if (S_ISLNK(st.st_mode)) {
 401                         int ret;
 402                         s->data = xmalloc(s->size);
 403                         s->should_free = 1;
 404                         ret = readlink(s->path, s->data, s->size);
 405                         if (ret < 0) {
 406                                 free(s->data);
 407                                 goto err_empty;
 408                         }
 409                         return 0;
 410                 }
 411                 fd = open(s->path, O_RDONLY);
 412                 if (fd < 0)
 413                         goto err_empty;
 414                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 415                 s->should_munmap = 1;
 416                 close(fd);
 417         }
 418         else {
 419                 char type[20];
 420                 struct sha1_size_cache *e;
 421
 422                 if (size_only) {
 423                         e = locate_size_cache(s->sha1, 1, 0);
 424                         if (e) {
 425                                 s->size = e->size;
 426                                 return 0;
 427                         }
 428                         if (!sha1_object_info(s->sha1, type, &s->size))
 429                                 locate_size_cache(s->sha1, 0, s->size);
 430                 }
 431                 else {
 432                         s->data = read_sha1_file(s->sha1, type, &s->size);
 433                         s->should_free = 1;
 434                 }
 435         }
 436         return 0;
 437 }
 438
 439 void diff_free_filespec_data(struct diff_filespec *s)
 440 {
 441         if (s->should_free)
 442                 free(s->data);
 443         else if (s->should_munmap)
 444                 munmap(s->data, s->size);
 445         s->should_free = s->should_munmap = 0;
 446         s->data = NULL;
 447 }
 448
 449 static void prep_temp_blob(struct diff_tempfile *temp,
 450                            void *blob,
 451                            unsigned long size,
 452                            unsigned char *sha1,
 453                            int mode)
 454 {
 455         int fd;
 456
 457         strcpy(temp->tmp_path, ".diff_XXXXXX");
 458         fd = mkstemp(temp->tmp_path);
 459         if (fd < 0)
 460                 die("unable to create temp-file");
 461         if (write(fd, blob, size) != size)
 462                 die("unable to write temp-file");
 463         close(fd);
 464         temp->name = temp->tmp_path;
 465         strcpy(temp->hex, sha1_to_hex(sha1));
 466         temp->hex[40] = 0;
 467         sprintf(temp->mode, "%06o", mode);
 468 }
 469
 470 static void prepare_temp_file(const char *name,
 471                               struct diff_tempfile *temp,
 472                               struct diff_filespec *one)
 473 {
 474         if (!DIFF_FILE_VALID(one)) {
 475         not_a_valid_file:
 476                 /* A '-' entry produces this for file-2, and
 477                  * a '+' entry produces this for file-1.
 478                  */
 479                 temp->name = "/dev/null";
 480                 strcpy(temp->hex, ".");
 481                 strcpy(temp->mode, ".");
 482                 return;
 483         }
 484
 485         if (!one->sha1_valid ||
 486             work_tree_matches(name, one->sha1)) {
 487                 struct stat st;
 488                 if (lstat(name, &st) < 0) {
 489                         if (errno == ENOENT)
 490                                 goto not_a_valid_file;
 491                         die("stat(%s): %s", name, strerror(errno));
 492                 }
 493                 if (S_ISLNK(st.st_mode)) {
 494                         int ret;
 495                         char *buf, buf_[1024];
 496                         buf = ((sizeof(buf_) < st.st_size) ?
 497                                xmalloc(st.st_size) : buf_);
 498                         ret = readlink(name, buf, st.st_size);
 499                         if (ret < 0)
 500                                 die("readlink(%s)", name);
 501                         prep_temp_blob(temp, buf, st.st_size,
 502                                        (one->sha1_valid ?
 503                                         one->sha1 : null_sha1),
 504                                        (one->sha1_valid ?
 505                                         one->mode : S_IFLNK));
 506                 }
 507                 else {
 508                         /* we can borrow from the file in the work tree */
 509                         temp->name = name;
 510                         if (!one->sha1_valid)
 511                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 512                         else
 513                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 514                         /* Even though we may sometimes borrow the
 515                          * contents from the work tree, we always want
 516                          * one->mode.  mode is trustworthy even when
 517                          * !(one->sha1_valid), as long as
 518                          * DIFF_FILE_VALID(one).
 519                          */
 520                         sprintf(temp->mode, "%06o", one->mode);
 521                 }
 522                 return;
 523         }
 524         else {
 525                 if (diff_populate_filespec(one, 0))
 526                         die("cannot read data blob for %s", one->path);
 527                 prep_temp_blob(temp, one->data, one->size,
 528                                one->sha1, one->mode);
 529         }
 530 }
 531
 532 static void remove_tempfile(void)
 533 {
 534         int i;
 535
 536         for (i = 0; i < 2; i++)
 537                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 538                         unlink(diff_temp[i].name);
 539                         diff_temp[i].name = NULL;
 540                 }
 541 }
 542
 543 static void remove_tempfile_on_signal(int signo)
 544 {
 545         remove_tempfile();
 546 }
 547
 548 /* An external diff command takes:
 549  *
 550  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 551  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 552  *
 553  */
 554 static void run_external_diff(const char *pgm,
 555                               const char *name,
 556                               const char *other,
 557                               struct diff_filespec *one,
 558                               struct diff_filespec *two,
 559                               const char *xfrm_msg,
 560                               int complete_rewrite)
 561 {
 562         struct diff_tempfile *temp = diff_temp;
 563         pid_t pid;
 564         int status;
 565         static int atexit_asked = 0;
 566
 567         if (one && two) {
 568                 prepare_temp_file(name, &temp[0], one);
 569                 prepare_temp_file(other ? : name, &temp[1], two);
 570                 if (! atexit_asked &&
 571                     (temp[0].name == temp[0].tmp_path ||
 572                      temp[1].name == temp[1].tmp_path)) {
 573                         atexit_asked = 1;
 574                         atexit(remove_tempfile);
 575                 }
 576                 signal(SIGINT, remove_tempfile_on_signal);
 577         }
 578
 579         fflush(NULL);
 580         pid = fork();
 581         if (pid < 0)
 582                 die("unable to fork");
 583         if (!pid) {
 584                 if (pgm) {
 585                         if (one && two) {
 586                                 const char *exec_arg[10];
 587                                 const char **arg = &exec_arg[0];
 588                                 *arg++ = pgm;
 589                                 *arg++ = name;
 590                                 *arg++ = temp[0].name;
 591                                 *arg++ = temp[0].hex;
 592                                 *arg++ = temp[0].mode;
 593                                 *arg++ = temp[1].name;
 594                                 *arg++ = temp[1].hex;
 595                                 *arg++ = temp[1].mode;
 596                                 if (other) {
 597                                         *arg++ = other;
 598                                         *arg++ = xfrm_msg;
 599                                 }
 600                                 *arg = NULL;
 601                                 execvp(pgm, (char *const*) exec_arg);
 602                         }
 603                         else
 604                                 execlp(pgm, pgm, name, NULL);
 605                 }
 606                 /*
 607                  * otherwise we use the built-in one.
 608                  */
 609                 if (one && two)
 610                         builtin_diff(name, other ? : name, temp, xfrm_msg,
 611                                      complete_rewrite);
 612                 else
 613                         printf("* Unmerged path %s\n", name);
 614                 exit(0);
 615         }
 616         if (waitpid(pid, &status, 0) < 0 ||
 617             !WIFEXITED(status) || WEXITSTATUS(status)) {
 618                 /* Earlier we did not check the exit status because
 619                  * diff exits non-zero if files are different, and
 620                  * we are not interested in knowing that.  It was a
 621                  * mistake which made it harder to quit a diff-*
 622                  * session that uses the git-apply-patch-script as
 623                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 624                  * should also exit non-zero only when it wants to
 625                  * abort the entire diff-* session.
 626                  */
 627                 remove_tempfile();
 628                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 629                 exit(1);
 630         }
 631         remove_tempfile();
 632 }
 633
 634 static void run_diff(struct diff_filepair *p)
 635 {
 636         const char *pgm = external_diff();
 637         char msg_[PATH_MAX*2+200], *xfrm_msg;
 638         struct diff_filespec *one;
 639         struct diff_filespec *two;
 640         const char *name;
 641         const char *other;
 642         int complete_rewrite = 0;
 643
 644         if (DIFF_PAIR_UNMERGED(p)) {
 645                 /* unmerged */
 646                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 647                                   0);
 648                 return;
 649         }
 650
 651         name = p->one->path;
 652         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 653         one = p->one; two = p->two;
 654         switch (p->status) {
 655         case 'C':
 656                 sprintf(msg_,
 657                         "similarity index %d%%\n"
 658                         "copy from %s\n"
 659                         "copy to %s",
 660                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 661                         name, other);
 662                 xfrm_msg = msg_;
 663                 break;
 664         case 'R':
 665                 sprintf(msg_,
 666                         "similarity index %d%%\n"
 667                         "rename from %s\n"
 668                         "rename to %s",
 669                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 670                         name, other);
 671                 xfrm_msg = msg_;
 672                 break;
 673         case 'M':
 674                 if (p->score) {
 675                         sprintf(msg_,
 676                                 "dissimilarity index %d%%",
 677                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 678                         xfrm_msg = msg_;
 679                         complete_rewrite = 1;
 680                         break;
 681                 }
 682                 /* fallthru */
 683         default:
 684                 xfrm_msg = NULL;
 685         }
 686
 687         if (!pgm &&
 688             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 689             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 690                 /* a filepair that changes between file and symlink
 691                  * needs to be split into deletion and creation.
 692                  */
 693                 struct diff_filespec *null = alloc_filespec(two->path);
 694                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 695                 free(null);
 696                 null = alloc_filespec(one->path);
 697                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 698                 free(null);
 699         }
 700         else
 701                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 702                                   complete_rewrite);
 703 }
 704
 705 void diff_setup(int flags)
 706 {
 707         if (flags & DIFF_SETUP_REVERSE)
 708                 reverse_diff = 1;
 709         if (flags & DIFF_SETUP_USE_CACHE) {
 710                 if (!active_cache)
 711                         /* read-cache does not die even when it fails
 712                          * so it is safe for us to do this here.  Also
 713                          * it does not smudge active_cache or active_nr
 714                          * when it fails, so we do not have to worry about
 715                          * cleaning it up oufselves either.
 716                          */
 717                         read_cache();
 718         }
 719         if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 720                 use_size_cache = 1;
 721
 722 }
 723
 724 static int parse_num(const char **cp_p)
 725 {
 726         int num, scale, ch, cnt;
 727         const char *cp = *cp_p;
 728
 729         cnt = num = 0;
 730         scale = 1;
 731         while ('0' <= (ch = *cp) && ch <= '9') {
 732                 if (cnt++ < 5) {
 733                         /* We simply ignore more than 5 digits precision. */
 734                         scale *= 10;
 735                         num = num * 10 + ch - '0';
 736                 }
 737                 *cp++;
 738         }
 739         *cp_p = cp;
 740
 741         /* user says num divided by scale and we say internally that
 742          * is MAX_SCORE * num / scale.
 743          */
 744         return (MAX_SCORE * num / scale);
 745 }
 746
 747 int diff_scoreopt_parse(const char *opt)
 748 {
 749         int opt1, opt2, cmd;
 750
 751         if (*opt++ != '-')
 752                 return -1;
 753         cmd = *opt++;
 754         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 755                 return -1; /* that is not a -M, -C nor -B option */
 756
 757         opt1 = parse_num(&opt);
 758         if (cmd != 'B')
 759                 opt2 = 0;
 760         else {
 761                 if (*opt == 0)
 762                         opt2 = 0;
 763                 else if (*opt != '/')
 764                         return -1; /* we expect -B80/99 or -B80 */
 765                 else {
 766                         opt++;
 767                         opt2 = parse_num(&opt);
 768                 }
 769         }
 770         if (*opt != 0)
 771                 return -1;
 772         return opt1 | (opt2 << 16);
 773 }
 774
 775 struct diff_queue_struct diff_queued_diff;
 776
 777 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 778 {
 779         if (queue->alloc <= queue->nr) {
 780                 queue->alloc = alloc_nr(queue->alloc);
 781                 queue->queue = xrealloc(queue->queue,
 782                                         sizeof(dp) * queue->alloc);
 783         }
 784         queue->queue[queue->nr++] = dp;
 785 }
 786
 787 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 788                                  struct diff_filespec *one,
 789                                  struct diff_filespec *two)
 790 {
 791         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 792         dp->one = one;
 793         dp->two = two;
 794         dp->score = 0;
 795         dp->status = 0;
 796         dp->source_stays = 0;
 797         dp->broken_pair = 0;
 798         diff_q(queue, dp);
 799         return dp;
 800 }
 801
 802 void diff_free_filepair(struct diff_filepair *p)
 803 {
 804         diff_free_filespec_data(p->one);
 805         diff_free_filespec_data(p->two);
 806         free(p);
 807 }
 808
 809 static void diff_flush_raw(struct diff_filepair *p,
 810                            int line_termination,
 811                            int inter_name_termination)
 812 {
 813         int two_paths;
 814         char status[10];
 815
 816         if (line_termination) {
 817                 const char *err = "path %s cannot be expressed without -z";
 818                 if (strchr(p->one->path, line_termination) ||
 819                     strchr(p->one->path, inter_name_termination))
 820                         die(err, p->one->path);
 821                 if (strchr(p->two->path, line_termination) ||
 822                     strchr(p->two->path, inter_name_termination))
 823                         die(err, p->two->path);
 824         }
 825
 826         if (p->score)
 827                 sprintf(status, "%c%03d", p->status,
 828                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 829         else {
 830                 status[0] = p->status;
 831                 status[1] = 0;
 832         }
 833         switch (p->status) {
 834         case 'C': case 'R':
 835                 two_paths = 1;
 836                 break;
 837         case 'N': case 'D':
 838                 two_paths = 0;
 839                 break;
 840         default:
 841                 two_paths = 0;
 842                 break;
 843         }
 844         printf(":%06o %06o %s ",
 845                p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 846         printf("%s %s%c%s",
 847                sha1_to_hex(p->two->sha1),
 848                status,
 849                inter_name_termination,
 850                p->one->path);
 851         if (two_paths)
 852                 printf("%c%s", inter_name_termination, p->two->path);
 853         putchar(line_termination);
 854 }
 855
 856 int diff_unmodified_pair(struct diff_filepair *p)
 857 {
 858         /* This function is written stricter than necessary to support
 859          * the currently implemented transformers, but the idea is to
 860          * let transformers to produce diff_filepairs any way they want,
 861          * and filter and clean them up here before producing the output.
 862          */
 863         struct diff_filespec *one, *two;
 864
 865         if (DIFF_PAIR_UNMERGED(p))
 866                 return 0; /* unmerged is interesting */
 867
 868         one = p->one;
 869         two = p->two;
 870
 871         /* deletion, addition, mode or type change
 872          * and rename are all interesting.
 873          */
 874         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 875             DIFF_PAIR_MODE_CHANGED(p) ||
 876             strcmp(one->path, two->path))
 877                 return 0;
 878
 879         /* both are valid and point at the same path.  that is, we are
 880          * dealing with a change.
 881          */
 882         if (one->sha1_valid && two->sha1_valid &&
 883             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 884                 return 1; /* no change */
 885         if (!one->sha1_valid && !two->sha1_valid)
 886                 return 1; /* both look at the same file on the filesystem. */
 887         return 0;
 888 }
 889
 890 static void diff_flush_patch(struct diff_filepair *p)
 891 {
 892         if (diff_unmodified_pair(p))
 893                 return;
 894
 895         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 896             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 897                 return; /* no tree diffs in patch format */
 898
 899         run_diff(p);
 900 }
 901
 902 int diff_queue_is_empty(void)
 903 {
 904         struct diff_queue_struct *q = &diff_queued_diff;
 905         int i;
 906         for (i = 0; i < q->nr; i++)
 907                 if (!diff_unmodified_pair(q->queue[i]))
 908                         return 0;
 909         return 1;
 910 }
 911
 912 #if DIFF_DEBUG
 913 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 914 {
 915         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 916                 x, one ? : "",
 917                 s->path,
 918                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 919                 s->mode,
 920                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 921         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 922                 x, one ? : "",
 923                 s->size, s->xfrm_flags);
 924 }
 925
 926 void diff_debug_filepair(const struct diff_filepair *p, int i)
 927 {
 928         diff_debug_filespec(p->one, i, "one");
 929         diff_debug_filespec(p->two, i, "two");
 930         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 931                 p->score, p->status ? : '?',
 932                 p->source_stays, p->broken_pair);
 933 }
 934
 935 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 936 {
 937         int i;
 938         if (msg)
 939                 fprintf(stderr, "%s\n", msg);
 940         fprintf(stderr, "q->nr = %d\n", q->nr);
 941         for (i = 0; i < q->nr; i++) {
 942                 struct diff_filepair *p = q->queue[i];
 943                 diff_debug_filepair(p, i);
 944         }
 945 }
 946 #endif
 947
 948 static void diff_resolve_rename_copy(void)
 949 {
 950         int i, j;
 951         struct diff_filepair *p, *pp;
 952         struct diff_queue_struct *q = &diff_queued_diff;
 953
 954         diff_debug_queue("resolve-rename-copy", q);
 955
 956         for (i = 0; i < q->nr; i++) {
 957                 p = q->queue[i];
 958                 p->status = 0; /* undecided */
 959                 if (DIFF_PAIR_UNMERGED(p))
 960                         p->status = 'U';
 961                 else if (!DIFF_FILE_VALID(p->one))
 962                         p->status = 'N';
 963                 else if (!DIFF_FILE_VALID(p->two))
 964                         p->status = 'D';
 965                 else if (DIFF_PAIR_TYPE_CHANGED(p))
 966                         p->status = 'T';
 967
 968                 /* from this point on, we are dealing with a pair
 969                  * whose both sides are valid and of the same type, i.e.
 970                  * either in-place edit or rename/copy edit.
 971                  */
 972                 else if (DIFF_PAIR_RENAME(p)) {
 973                         if (p->source_stays) {
 974                                 p->status = 'C';
 975                                 continue;
 976                         }
 977                         /* See if there is some other filepair that
 978                          * copies from the same source as us.  If so
 979                          * we are a copy.  Otherwise we are a rename.
 980                          */
 981                         for (j = i + 1; j < q->nr; j++) {
 982                                 pp = q->queue[j];
 983                                 if (strcmp(pp->one->path, p->one->path))
 984                                         continue; /* not us */
 985                                 if (!DIFF_PAIR_RENAME(pp))
 986                                         continue; /* not a rename/copy */
 987                                 /* pp is a rename/copy from the same source */
 988                                 p->status = 'C';
 989                                 break;
 990                         }
 991                         if (!p->status)
 992                                 p->status = 'R';
 993                 }
 994                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 995                          p->one->mode != p->two->mode)
 996                         p->status = 'M';
 997                 else {
 998                         /* This is a "no-change" entry and should not
 999                          * happen anymore, but prepare for broken callers.
1000                          */
1001                         error("feeding unmodified %s to diffcore",
1002                               p->one->path);
1003                         p->status = 'X';
1004                 }
1005         }
1006         diff_debug_queue("resolve-rename-copy done", q);
1007 }
1008
1009 void diff_flush(int diff_output_style)
1010 {
1011         struct diff_queue_struct *q = &diff_queued_diff;
1012         int i;
1013         int line_termination = '\n';
1014         int inter_name_termination = '\t';
1015
1016         if (diff_output_style == DIFF_FORMAT_MACHINE)
1017                 line_termination = inter_name_termination = 0;
1018
1019         for (i = 0; i < q->nr; i++) {
1020                 struct diff_filepair *p = q->queue[i];
1021                 if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
1022                     (p->status == 'X'))
1023                         continue;
1024                 if (p->status == 0)
1025                         die("internal error in diff-resolve-rename-copy");
1026                 switch (diff_output_style) {
1027                 case DIFF_FORMAT_PATCH:
1028                         diff_flush_patch(p);
1029                         break;
1030                 case DIFF_FORMAT_HUMAN:
1031                 case DIFF_FORMAT_MACHINE:
1032                         diff_flush_raw(p, line_termination,
1033                                        inter_name_termination);
1034                         break;
1035                 }
1036         }
1037         for (i = 0; i < q->nr; i++)
1038                 diff_free_filepair(q->queue[i]);
1039         free(q->queue);
1040         q->queue = NULL;
1041         q->nr = q->alloc = 0;
1042 }
1043
1044 static void diffcore_apply_filter(const char *filter)
1045 {
1046         int i;
1047         struct diff_queue_struct *q = &diff_queued_diff;
1048         struct diff_queue_struct outq;
1049         outq.queue = NULL;
1050         outq.nr = outq.alloc = 0;
1051
1052         if (!filter)
1053                 return;
1054
1055         if (strchr(filter, 'A')) {
1056                 /* All-or-none */
1057                 int found;
1058                 for (i = found = 0; !found && i < q->nr; i++) {
1059                         struct diff_filepair *p = q->queue[i];
1060                         if (((p->status == 'M') &&
1061                              ((p->score && strchr(filter, 'B')) ||
1062                               (!p->score && strchr(filter, 'M')))) ||
1063                             ((p->status != 'M') && strchr(filter, p->status)))
1064                                 found++;
1065                 }
1066                 if (found)
1067                         return;
1068
1069                 /* otherwise we will clear the whole queue
1070                  * by copying the empty outq at the end of this
1071                  * function, but first clear the current entries
1072                  * in the queue.
1073                  */
1074                 for (i = 0; i < q->nr; i++)
1075                         diff_free_filepair(q->queue[i]);
1076         }
1077         else {
1078                 /* Only the matching ones */
1079                 for (i = 0; i < q->nr; i++) {
1080                         struct diff_filepair *p = q->queue[i];
1081                         if (((p->status == 'M') &&
1082                              ((p->score && strchr(filter, 'B')) ||
1083                               (!p->score && strchr(filter, 'M')))) ||
1084                             ((p->status != 'M') && strchr(filter, p->status)))
1085                                 diff_q(&outq, p);
1086                         else
1087                                 diff_free_filepair(p);
1088                 }
1089         }
1090         free(q->queue);
1091         *q = outq;
1092 }
1093
1094 void diffcore_std(const char **paths,
1095                   int detect_rename, int rename_score,
1096                   const char *pickaxe, int pickaxe_opts,
1097                   int break_opt,
1098                   const char *orderfile,
1099                   const char *filter)
1100 {
1101         if (paths && paths[0])
1102                 diffcore_pathspec(paths);
1103         if (break_opt != -1)
1104                 diffcore_break(break_opt);
1105         if (detect_rename)
1106                 diffcore_rename(detect_rename, rename_score);
1107         if (break_opt != -1)
1108                 diffcore_merge_broken();
1109         if (pickaxe)
1110                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1111         if (orderfile)
1112                 diffcore_order(orderfile);
1113         diff_resolve_rename_copy();
1114         diffcore_apply_filter(filter);
1115 }
1116
1117
1118 void diffcore_std_no_resolve(const char **paths,
1119                              const char *pickaxe, int pickaxe_opts,
1120                              const char *orderfile,
1121                              const char *filter)
1122 {
1123         if (paths && paths[0])
1124                 diffcore_pathspec(paths);
1125         if (pickaxe)
1126                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1127         if (orderfile)
1128                 diffcore_order(orderfile);
1129         diffcore_apply_filter(filter);
1130 }
1131
1132 void diff_addremove(int addremove, unsigned mode,
1133                     const unsigned char *sha1,
1134                     const char *base, const char *path)
1135 {
1136         char concatpath[PATH_MAX];
1137         struct diff_filespec *one, *two;
1138
1139         /* This may look odd, but it is a preparation for
1140          * feeding "there are unchanged files which should
1141          * not produce diffs, but when you are doing copy
1142          * detection you would need them, so here they are"
1143          * entries to the diff-core.  They will be prefixed
1144          * with something like '=' or '*' (I haven't decided
1145          * which but should not make any difference).
1146          * Feeding the same new and old to diff_change()
1147          * also has the same effect.
1148          * Before the final output happens, they are pruned after
1149          * merged into rename/copy pairs as appropriate.
1150          */
1151         if (reverse_diff)
1152                 addremove = (addremove == '+' ? '-' :
1153                              addremove == '-' ? '+' : addremove);
1154
1155         if (!path) path = "";
1156         sprintf(concatpath, "%s%s", base, path);
1157         one = alloc_filespec(concatpath);
1158         two = alloc_filespec(concatpath);
1159
1160         if (addremove != '+')
1161                 fill_filespec(one, sha1, mode);
1162         if (addremove != '-')
1163                 fill_filespec(two, sha1, mode);
1164
1165         diff_queue(&diff_queued_diff, one, two);
1166 }
1167
1168 void diff_helper_input(unsigned old_mode,
1169                        unsigned new_mode,
1170                        const unsigned char *old_sha1,
1171                        const unsigned char *new_sha1,
1172                        const char *old_path,
1173                        int status,
1174                        int score,
1175                        const char *new_path)
1176 {
1177         struct diff_filespec *one, *two;
1178         struct diff_filepair *dp;
1179
1180         one = alloc_filespec(old_path);
1181         two = alloc_filespec(new_path);
1182         if (old_mode)
1183                 fill_filespec(one, old_sha1, old_mode);
1184         if (new_mode)
1185                 fill_filespec(two, new_sha1, new_mode);
1186         dp = diff_queue(&diff_queued_diff, one, two);
1187         dp->score = score * MAX_SCORE / 100;
1188         dp->status = status;
1189 }
1190
1191 void diff_change(unsigned old_mode, unsigned new_mode,
1192                  const unsigned char *old_sha1,
1193                  const unsigned char *new_sha1,
1194                  const char *base, const char *path)
1195 {
1196         char concatpath[PATH_MAX];
1197         struct diff_filespec *one, *two;
1198
1199         if (reverse_diff) {
1200                 unsigned tmp;
1201                 const unsigned char *tmp_c;
1202                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1203                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1204         }
1205         if (!path) path = "";
1206         sprintf(concatpath, "%s%s", base, path);
1207         one = alloc_filespec(concatpath);
1208         two = alloc_filespec(concatpath);
1209         fill_filespec(one, old_sha1, old_mode);
1210         fill_filespec(two, new_sha1, new_mode);
1211
1212         diff_queue(&diff_queued_diff, one, two);
1213 }
1214
1215 void diff_unmerge(const char *path)
1216 {
1217         struct diff_filespec *one, *two;
1218         one = alloc_filespec(path);
1219         two = alloc_filespec(path);
1220         diff_queue(&diff_queued_diff, one, two);
1221 }