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 static const char *external_diff(void)
  17 {
  18         static const char *external_diff_cmd = NULL;
  19         static int done_preparing = 0;
  20         const char *env_diff_opts;
  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         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  34
  35         /* In case external diff fails... */
  36         env_diff_opts = getenv("GIT_DIFF_OPTS");
  37         if (env_diff_opts) diff_opts = env_diff_opts;
  38
  39         done_preparing = 1;
  40         return external_diff_cmd;
  41 }
  42
  43 #define TEMPFILE_PATH_LEN               50
  44
  45 static struct diff_tempfile {
  46         const char *name; /* filename external diff should read from */
  47         char hex[41];
  48         char mode[10];
  49         char tmp_path[TEMPFILE_PATH_LEN];
  50 } diff_temp[2];
  51
  52 static int count_lines(const char *filename)
  53 {
  54         FILE *in;
  55         int count, ch, completely_empty = 1, nl_just_seen = 0;
  56         in = fopen(filename, "r");
  57         count = 0;
  58         while ((ch = fgetc(in)) != EOF)
  59                 if (ch == '\n') {
  60                         count++;
  61                         nl_just_seen = 1;
  62                         completely_empty = 0;
  63                 }
  64                 else {
  65                         nl_just_seen = 0;
  66                         completely_empty = 0;
  67                 }
  68         fclose(in);
  69         if (completely_empty)
  70                 return 0;
  71         if (!nl_just_seen)
  72                 count++; /* no trailing newline */
  73         return count;
  74 }
  75
  76 static void print_line_count(int count)
  77 {
  78         switch (count) {
  79         case 0:
  80                 printf("0,0");
  81                 break;
  82         case 1:
  83                 printf("1");
  84                 break;
  85         default:
  86                 printf("1,%d", count);
  87                 break;
  88         }
  89 }
  90
  91 static void copy_file(int prefix, const char *filename)
  92 {
  93         FILE *in;
  94         int ch, nl_just_seen = 1;
  95         in = fopen(filename, "r");
  96         while ((ch = fgetc(in)) != EOF) {
  97                 if (nl_just_seen)
  98                         putchar(prefix);
  99                 putchar(ch);
 100                 if (ch == '\n')
 101                         nl_just_seen = 1;
 102                 else
 103                         nl_just_seen = 0;
 104         }
 105         fclose(in);
 106         if (!nl_just_seen)
 107                 printf("\n\\ No newline at end of file\n");
 108 }
 109
 110 static void emit_rewrite_diff(const char *name_a,
 111                               const char *name_b,
 112                               struct diff_tempfile *temp)
 113 {
 114         /* Use temp[i].name as input, name_a and name_b as labels */
 115         int lc_a, lc_b;
 116         lc_a = count_lines(temp[0].name);
 117         lc_b = count_lines(temp[1].name);
 118         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 119         print_line_count(lc_a);
 120         printf(" +");
 121         print_line_count(lc_b);
 122         printf(" @@\n");
 123         if (lc_a)
 124                 copy_file('-', temp[0].name);
 125         if (lc_b)
 126                 copy_file('+', temp[1].name);
 127 }
 128
 129 static void builtin_diff(const char *name_a,
 130                          const char *name_b,
 131                          struct diff_tempfile *temp,
 132                          const char *xfrm_msg,
 133                          int complete_rewrite)
 134 {
 135         int i, next_at, cmd_size;
 136         const char *const diff_cmd = "diff -L%s%s -L%s%s";
 137         const char *const diff_arg  = "%s %s||:"; /* "||:" is to return 0 */
 138         const char *input_name_sq[2];
 139         const char *path0[2];
 140         const char *path1[2];
 141         const char *name_sq[2];
 142         char *cmd;
 143
 144         name_sq[0] = sq_quote(name_a);
 145         name_sq[1] = sq_quote(name_b);
 146
 147         /* diff_cmd and diff_arg have 6 %s in total which makes
 148          * the sum of these strings 12 bytes larger than required.
 149          * we use 2 spaces around diff-opts, and we need to count
 150          * terminating NUL, so we subtract 9 here.
 151          */
 152         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 153                         strlen(diff_arg) - 9);
 154         for (i = 0; i < 2; i++) {
 155                 input_name_sq[i] = sq_quote(temp[i].name);
 156                 if (!strcmp(temp[i].name, "/dev/null")) {
 157                         path0[i] = "/dev/null";
 158                         path1[i] = "";
 159                 } else {
 160                         path0[i] = i ? "b/" : "a/";
 161                         path1[i] = name_sq[i];
 162                 }
 163                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 164                              strlen(input_name_sq[i]));
 165         }
 166
 167         cmd = xmalloc(cmd_size);
 168
 169         next_at = 0;
 170         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 171                             diff_cmd,
 172                             path0[0], path1[0], path0[1], path1[1]);
 173         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 174                             " %s ", diff_opts);
 175         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 176                             diff_arg, input_name_sq[0], input_name_sq[1]);
 177
 178         printf("diff --git a/%s b/%s\n", name_a, name_b);
 179         if (!path1[0][0]) {
 180                 printf("new file mode %s\n", temp[1].mode);
 181                 if (xfrm_msg && xfrm_msg[0])
 182                         puts(xfrm_msg);
 183         }
 184         else if (!path1[1][0]) {
 185                 printf("deleted file mode %s\n", temp[0].mode);
 186                 if (xfrm_msg && xfrm_msg[0])
 187                         puts(xfrm_msg);
 188         }
 189         else {
 190                 if (strcmp(temp[0].mode, temp[1].mode)) {
 191                         printf("old mode %s\n", temp[0].mode);
 192                         printf("new mode %s\n", temp[1].mode);
 193                 }
 194                 if (xfrm_msg && xfrm_msg[0])
 195                         puts(xfrm_msg);
 196                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 197                         /* we do not run diff between different kind
 198                          * of objects.
 199                          */
 200                         exit(0);
 201                 if (complete_rewrite) {
 202                         fflush(NULL);
 203                         emit_rewrite_diff(name_a, name_b, temp);
 204                         exit(0);
 205                 }
 206         }
 207         fflush(NULL);
 208         execlp("/bin/sh","sh", "-c", cmd, NULL);
 209 }
 210
 211 struct diff_filespec *alloc_filespec(const char *path)
 212 {
 213         int namelen = strlen(path);
 214         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 215
 216         memset(spec, 0, sizeof(*spec));
 217         spec->path = (char *)(spec + 1);
 218         memcpy(spec->path, path, namelen+1);
 219         return spec;
 220 }
 221
 222 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 223                    unsigned short mode)
 224 {
 225         if (mode) {
 226                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 227                 memcpy(spec->sha1, sha1, 20);
 228                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 229         }
 230 }
 231
 232 /*
 233  * Given a name and sha1 pair, if the dircache tells us the file in
 234  * the work tree has that object contents, return true, so that
 235  * prepare_temp_file() does not have to inflate and extract.
 236  */
 237 static int work_tree_matches(const char *name, const unsigned char *sha1)
 238 {
 239         struct cache_entry *ce;
 240         struct stat st;
 241         int pos, len;
 242
 243         /* We do not read the cache ourselves here, because the
 244          * benchmark with my previous version that always reads cache
 245          * shows that it makes things worse for diff-tree comparing
 246          * two linux-2.6 kernel trees in an already checked out work
 247          * tree.  This is because most diff-tree comparisons deal with
 248          * only a small number of files, while reading the cache is
 249          * expensive for a large project, and its cost outweighs the
 250          * savings we get by not inflating the object to a temporary
 251          * file.  Practically, this code only helps when we are used
 252          * by diff-cache --cached, which does read the cache before
 253          * calling us.
 254          */
 255         if (!active_cache)
 256                 return 0;
 257
 258         len = strlen(name);
 259         pos = cache_name_pos(name, len);
 260         if (pos < 0)
 261                 return 0;
 262         ce = active_cache[pos];
 263         if ((lstat(name, &st) < 0) ||
 264             !S_ISREG(st.st_mode) || /* careful! */
 265             ce_match_stat(ce, &st) ||
 266             memcmp(sha1, ce->sha1, 20))
 267                 return 0;
 268         /* we return 1 only when we can stat, it is a regular file,
 269          * stat information matches, and sha1 recorded in the cache
 270          * matches.  I.e. we know the file in the work tree really is
 271          * the same as the <name, sha1> pair.
 272          */
 273         return 1;
 274 }
 275
 276 static struct sha1_size_cache {
 277         unsigned char sha1[20];
 278         unsigned long size;
 279 } **sha1_size_cache;
 280 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 281
 282 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 283                                                  int find_only,
 284                                                  unsigned long size)
 285 {
 286         int first, last;
 287         struct sha1_size_cache *e;
 288
 289         first = 0;
 290         last = sha1_size_cache_nr;
 291         while (last > first) {
 292                 int cmp, next = (last + first) >> 1;
 293                 e = sha1_size_cache[next];
 294                 cmp = memcmp(e->sha1, sha1, 20);
 295                 if (!cmp)
 296                         return e;
 297                 if (cmp < 0) {
 298                         last = next;
 299                         continue;
 300                 }
 301                 first = next+1;
 302         }
 303         /* not found */
 304         if (find_only)
 305                 return NULL;
 306         /* insert to make it at "first" */
 307         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 308                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 309                 sha1_size_cache = xrealloc(sha1_size_cache,
 310                                            sha1_size_cache_alloc *
 311                                            sizeof(*sha1_size_cache));
 312         }
 313         sha1_size_cache_nr++;
 314         if (first < sha1_size_cache_nr)
 315                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 316                         (sha1_size_cache_nr - first - 1) *
 317                         sizeof(*sha1_size_cache));
 318         e = xmalloc(sizeof(struct sha1_size_cache));
 319         sha1_size_cache[first] = e;
 320         memcpy(e->sha1, sha1, 20);
 321         e->size = size;
 322         return e;
 323 }
 324
 325 /*
 326  * While doing rename detection and pickaxe operation, we may need to
 327  * grab the data for the blob (or file) for our own in-core comparison.
 328  * diff_filespec has data and size fields for this purpose.
 329  */
 330 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 331 {
 332         int err = 0;
 333         if (!DIFF_FILE_VALID(s))
 334                 die("internal error: asking to populate invalid file.");
 335         if (S_ISDIR(s->mode))
 336                 return -1;
 337
 338         if (!use_size_cache)
 339                 size_only = 0;
 340
 341         if (s->data)
 342                 return err;
 343         if (!s->sha1_valid ||
 344             work_tree_matches(s->path, s->sha1)) {
 345                 struct stat st;
 346                 int fd;
 347                 if (lstat(s->path, &st) < 0) {
 348                         if (errno == ENOENT) {
 349                         err_empty:
 350                                 err = -1;
 351                         empty:
 352                                 s->data = "";
 353                                 s->size = 0;
 354                                 return err;
 355                         }
 356                 }
 357                 s->size = st.st_size;
 358                 if (!s->size)
 359                         goto empty;
 360                 if (size_only)
 361                         return 0;
 362                 if (S_ISLNK(st.st_mode)) {
 363                         int ret;
 364                         s->data = xmalloc(s->size);
 365                         s->should_free = 1;
 366                         ret = readlink(s->path, s->data, s->size);
 367                         if (ret < 0) {
 368                                 free(s->data);
 369                                 goto err_empty;
 370                         }
 371                         return 0;
 372                 }
 373                 fd = open(s->path, O_RDONLY);
 374                 if (fd < 0)
 375                         goto err_empty;
 376                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 377                 close(fd);
 378                 if (s->data == MAP_FAILED)
 379                         goto err_empty;
 380                 s->should_munmap = 1;
 381         }
 382         else {
 383                 char type[20];
 384                 struct sha1_size_cache *e;
 385
 386                 if (size_only) {
 387                         e = locate_size_cache(s->sha1, 1, 0);
 388                         if (e) {
 389                                 s->size = e->size;
 390                                 return 0;
 391                         }
 392                         if (!sha1_object_info(s->sha1, type, &s->size))
 393                                 locate_size_cache(s->sha1, 0, s->size);
 394                 }
 395                 else {
 396                         s->data = read_sha1_file(s->sha1, type, &s->size);
 397                         s->should_free = 1;
 398                 }
 399         }
 400         return 0;
 401 }
 402
 403 void diff_free_filespec_data(struct diff_filespec *s)
 404 {
 405         if (s->should_free)
 406                 free(s->data);
 407         else if (s->should_munmap)
 408                 munmap(s->data, s->size);
 409         s->should_free = s->should_munmap = 0;
 410         s->data = NULL;
 411 }
 412
 413 static void prep_temp_blob(struct diff_tempfile *temp,
 414                            void *blob,
 415                            unsigned long size,
 416                            const unsigned char *sha1,
 417                            int mode)
 418 {
 419         int fd;
 420
 421         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 422         if (fd < 0)
 423                 die("unable to create temp-file");
 424         if (write(fd, blob, size) != size)
 425                 die("unable to write temp-file");
 426         close(fd);
 427         temp->name = temp->tmp_path;
 428         strcpy(temp->hex, sha1_to_hex(sha1));
 429         temp->hex[40] = 0;
 430         sprintf(temp->mode, "%06o", mode);
 431 }
 432
 433 static void prepare_temp_file(const char *name,
 434                               struct diff_tempfile *temp,
 435                               struct diff_filespec *one)
 436 {
 437         if (!DIFF_FILE_VALID(one)) {
 438         not_a_valid_file:
 439                 /* A '-' entry produces this for file-2, and
 440                  * a '+' entry produces this for file-1.
 441                  */
 442                 temp->name = "/dev/null";
 443                 strcpy(temp->hex, ".");
 444                 strcpy(temp->mode, ".");
 445                 return;
 446         }
 447
 448         if (!one->sha1_valid ||
 449             work_tree_matches(name, one->sha1)) {
 450                 struct stat st;
 451                 if (lstat(name, &st) < 0) {
 452                         if (errno == ENOENT)
 453                                 goto not_a_valid_file;
 454                         die("stat(%s): %s", name, strerror(errno));
 455                 }
 456                 if (S_ISLNK(st.st_mode)) {
 457                         int ret;
 458                         char *buf, buf_[1024];
 459                         buf = ((sizeof(buf_) < st.st_size) ?
 460                                xmalloc(st.st_size) : buf_);
 461                         ret = readlink(name, buf, st.st_size);
 462                         if (ret < 0)
 463                                 die("readlink(%s)", name);
 464                         prep_temp_blob(temp, buf, st.st_size,
 465                                        (one->sha1_valid ?
 466                                         one->sha1 : null_sha1),
 467                                        (one->sha1_valid ?
 468                                         one->mode : S_IFLNK));
 469                 }
 470                 else {
 471                         /* we can borrow from the file in the work tree */
 472                         temp->name = name;
 473                         if (!one->sha1_valid)
 474                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 475                         else
 476                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 477                         /* Even though we may sometimes borrow the
 478                          * contents from the work tree, we always want
 479                          * one->mode.  mode is trustworthy even when
 480                          * !(one->sha1_valid), as long as
 481                          * DIFF_FILE_VALID(one).
 482                          */
 483                         sprintf(temp->mode, "%06o", one->mode);
 484                 }
 485                 return;
 486         }
 487         else {
 488                 if (diff_populate_filespec(one, 0))
 489                         die("cannot read data blob for %s", one->path);
 490                 prep_temp_blob(temp, one->data, one->size,
 491                                one->sha1, one->mode);
 492         }
 493 }
 494
 495 static void remove_tempfile(void)
 496 {
 497         int i;
 498
 499         for (i = 0; i < 2; i++)
 500                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 501                         unlink(diff_temp[i].name);
 502                         diff_temp[i].name = NULL;
 503                 }
 504 }
 505
 506 static void remove_tempfile_on_signal(int signo)
 507 {
 508         remove_tempfile();
 509 }
 510
 511 /* An external diff command takes:
 512  *
 513  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 514  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 515  *
 516  */
 517 static void run_external_diff(const char *pgm,
 518                               const char *name,
 519                               const char *other,
 520                               struct diff_filespec *one,
 521                               struct diff_filespec *two,
 522                               const char *xfrm_msg,
 523                               int complete_rewrite)
 524 {
 525         struct diff_tempfile *temp = diff_temp;
 526         pid_t pid;
 527         int status;
 528         static int atexit_asked = 0;
 529         const char *othername;
 530
 531         othername = (other? other : name);
 532         if (one && two) {
 533                 prepare_temp_file(name, &temp[0], one);
 534                 prepare_temp_file(othername, &temp[1], two);
 535                 if (! atexit_asked &&
 536                     (temp[0].name == temp[0].tmp_path ||
 537                      temp[1].name == temp[1].tmp_path)) {
 538                         atexit_asked = 1;
 539                         atexit(remove_tempfile);
 540                 }
 541                 signal(SIGINT, remove_tempfile_on_signal);
 542         }
 543
 544         fflush(NULL);
 545         pid = fork();
 546         if (pid < 0)
 547                 die("unable to fork");
 548         if (!pid) {
 549                 if (pgm) {
 550                         if (one && two) {
 551                                 const char *exec_arg[10];
 552                                 const char **arg = &exec_arg[0];
 553                                 *arg++ = pgm;
 554                                 *arg++ = name;
 555                                 *arg++ = temp[0].name;
 556                                 *arg++ = temp[0].hex;
 557                                 *arg++ = temp[0].mode;
 558                                 *arg++ = temp[1].name;
 559                                 *arg++ = temp[1].hex;
 560                                 *arg++ = temp[1].mode;
 561                                 if (other) {
 562                                         *arg++ = other;
 563                                         *arg++ = xfrm_msg;
 564                                 }
 565                                 *arg = NULL;
 566                                 execvp(pgm, (char *const*) exec_arg);
 567                         }
 568                         else
 569                                 execlp(pgm, pgm, name, NULL);
 570                 }
 571                 /*
 572                  * otherwise we use the built-in one.
 573                  */
 574                 if (one && two)
 575                         builtin_diff(name, othername, temp, xfrm_msg,
 576                                      complete_rewrite);
 577                 else
 578                         printf("* Unmerged path %s\n", name);
 579                 exit(0);
 580         }
 581         if (waitpid(pid, &status, 0) < 0 ||
 582             !WIFEXITED(status) || WEXITSTATUS(status)) {
 583                 /* Earlier we did not check the exit status because
 584                  * diff exits non-zero if files are different, and
 585                  * we are not interested in knowing that.  It was a
 586                  * mistake which made it harder to quit a diff-*
 587                  * session that uses the git-apply-patch-script as
 588                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 589                  * should also exit non-zero only when it wants to
 590                  * abort the entire diff-* session.
 591                  */
 592                 remove_tempfile();
 593                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 594                 exit(1);
 595         }
 596         remove_tempfile();
 597 }
 598
 599 static void run_diff(struct diff_filepair *p)
 600 {
 601         const char *pgm = external_diff();
 602         char msg_[PATH_MAX*2+200], *xfrm_msg;
 603         struct diff_filespec *one;
 604         struct diff_filespec *two;
 605         const char *name;
 606         const char *other;
 607         int complete_rewrite = 0;
 608
 609         if (DIFF_PAIR_UNMERGED(p)) {
 610                 /* unmerged */
 611                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 612                                   0);
 613                 return;
 614         }
 615
 616         name = p->one->path;
 617         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 618         one = p->one; two = p->two;
 619         switch (p->status) {
 620         case DIFF_STATUS_COPIED:
 621                 sprintf(msg_,
 622                         "similarity index %d%%\n"
 623                         "copy from %s\n"
 624                         "copy to %s",
 625                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 626                         name, other);
 627                 xfrm_msg = msg_;
 628                 break;
 629         case DIFF_STATUS_RENAMED:
 630                 sprintf(msg_,
 631                         "similarity index %d%%\n"
 632                         "rename from %s\n"
 633                         "rename to %s",
 634                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 635                         name, other);
 636                 xfrm_msg = msg_;
 637                 break;
 638         case DIFF_STATUS_MODIFIED:
 639                 if (p->score) {
 640                         sprintf(msg_,
 641                                 "dissimilarity index %d%%",
 642                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 643                         xfrm_msg = msg_;
 644                         complete_rewrite = 1;
 645                         break;
 646                 }
 647                 /* fallthru */
 648         default:
 649                 xfrm_msg = NULL;
 650         }
 651
 652         if (!pgm &&
 653             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 654             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 655                 /* a filepair that changes between file and symlink
 656                  * needs to be split into deletion and creation.
 657                  */
 658                 struct diff_filespec *null = alloc_filespec(two->path);
 659                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 660                 free(null);
 661                 null = alloc_filespec(one->path);
 662                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 663                 free(null);
 664         }
 665         else
 666                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 667                                   complete_rewrite);
 668 }
 669
 670 void diff_setup(struct diff_options *options)
 671 {
 672         memset(options, 0, sizeof(*options));
 673         options->output_format = DIFF_FORMAT_RAW;
 674         options->line_termination = '\n';
 675         options->break_opt = -1;
 676         options->rename_limit = -1;
 677 }
 678
 679 int diff_setup_done(struct diff_options *options)
 680 {
 681         if ((options->find_copies_harder || 0 <= options->rename_limit) &&
 682             options->detect_rename != DIFF_DETECT_COPY)
 683                 return -1;
 684         if (options->setup & DIFF_SETUP_USE_CACHE) {
 685                 if (!active_cache)
 686                         /* read-cache does not die even when it fails
 687                          * so it is safe for us to do this here.  Also
 688                          * it does not smudge active_cache or active_nr
 689                          * when it fails, so we do not have to worry about
 690                          * cleaning it up oufselves either.
 691                          */
 692                         read_cache();
 693         }
 694         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
 695                 use_size_cache = 1;
 696
 697         return 0;
 698 }
 699
 700 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 701 {
 702         const char *arg = av[0];
 703         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
 704                 options->output_format = DIFF_FORMAT_PATCH;
 705         else if (!strcmp(arg, "-z"))
 706                 options->line_termination = 0;
 707         else if (!strncmp(arg, "-l", 2))
 708                 options->rename_limit = strtoul(arg+2, NULL, 10);
 709         else if (!strcmp(arg, "--name-only"))
 710                 options->output_format = DIFF_FORMAT_NAME;
 711         else if (!strcmp(arg, "--name-status"))
 712                 options->output_format = DIFF_FORMAT_NAME_STATUS;
 713         else if (!strcmp(arg, "-R"))
 714                 options->reverse_diff = 1;
 715         else if (!strncmp(arg, "-S", 2))
 716                 options->pickaxe = arg + 2;
 717         else if (!strcmp(arg, "-s"))
 718                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
 719         else if (!strncmp(arg, "-O", 2))
 720                 options->orderfile = arg + 2;
 721         else if (!strncmp(arg, "--diff-filter=", 14))
 722                 options->filter = arg + 14;
 723         else if (!strcmp(arg, "--pickaxe-all"))
 724                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
 725         else if (!strncmp(arg, "-B", 2)) {
 726                 if ((options->break_opt =
 727                      diff_scoreopt_parse(arg)) == -1)
 728                         return -1;
 729         }
 730         else if (!strncmp(arg, "-M", 2)) {
 731                 if ((options->rename_score =
 732                      diff_scoreopt_parse(arg)) == -1)
 733                         return -1;
 734                 options->detect_rename = DIFF_DETECT_RENAME;
 735         }
 736         else if (!strncmp(arg, "-C", 2)) {
 737                 if ((options->rename_score =
 738                      diff_scoreopt_parse(arg)) == -1)
 739                         return -1;
 740                 options->detect_rename = DIFF_DETECT_COPY;
 741         }
 742         else if (!strcmp(arg, "--find-copies-harder"))
 743                 options->find_copies_harder = 1;
 744         else
 745                 return 0;
 746         return 1;
 747 }
 748
 749 static int parse_num(const char **cp_p)
 750 {
 751         int num, scale, ch, cnt;
 752         const char *cp = *cp_p;
 753
 754         cnt = num = 0;
 755         scale = 1;
 756         while ('0' <= (ch = *cp) && ch <= '9') {
 757                 if (cnt++ < 5) {
 758                         /* We simply ignore more than 5 digits precision. */
 759                         scale *= 10;
 760                         num = num * 10 + ch - '0';
 761                 }
 762                 cp++;
 763         }
 764         *cp_p = cp;
 765
 766         /* user says num divided by scale and we say internally that
 767          * is MAX_SCORE * num / scale.
 768          */
 769         return (MAX_SCORE * num / scale);
 770 }
 771
 772 int diff_scoreopt_parse(const char *opt)
 773 {
 774         int opt1, opt2, cmd;
 775
 776         if (*opt++ != '-')
 777                 return -1;
 778         cmd = *opt++;
 779         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 780                 return -1; /* that is not a -M, -C nor -B option */
 781
 782         opt1 = parse_num(&opt);
 783         if (cmd != 'B')
 784                 opt2 = 0;
 785         else {
 786                 if (*opt == 0)
 787                         opt2 = 0;
 788                 else if (*opt != '/')
 789                         return -1; /* we expect -B80/99 or -B80 */
 790                 else {
 791                         opt++;
 792                         opt2 = parse_num(&opt);
 793                 }
 794         }
 795         if (*opt != 0)
 796                 return -1;
 797         return opt1 | (opt2 << 16);
 798 }
 799
 800 struct diff_queue_struct diff_queued_diff;
 801
 802 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 803 {
 804         if (queue->alloc <= queue->nr) {
 805                 queue->alloc = alloc_nr(queue->alloc);
 806                 queue->queue = xrealloc(queue->queue,
 807                                         sizeof(dp) * queue->alloc);
 808         }
 809         queue->queue[queue->nr++] = dp;
 810 }
 811
 812 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 813                                  struct diff_filespec *one,
 814                                  struct diff_filespec *two)
 815 {
 816         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 817         dp->one = one;
 818         dp->two = two;
 819         dp->score = 0;
 820         dp->status = 0;
 821         dp->source_stays = 0;
 822         dp->broken_pair = 0;
 823         if (queue)
 824                 diff_q(queue, dp);
 825         return dp;
 826 }
 827
 828 void diff_free_filepair(struct diff_filepair *p)
 829 {
 830         diff_free_filespec_data(p->one);
 831         diff_free_filespec_data(p->two);
 832         free(p->one);
 833         free(p->two);
 834         free(p);
 835 }
 836
 837 static void diff_flush_raw(struct diff_filepair *p,
 838                            int line_termination,
 839                            int inter_name_termination,
 840                            int output_format)
 841 {
 842         int two_paths;
 843         char status[10];
 844
 845         if (line_termination) {
 846                 const char *const err =
 847                         "path %s cannot be expressed without -z";
 848                 if (strchr(p->one->path, line_termination) ||
 849                     strchr(p->one->path, inter_name_termination))
 850                         die(err, p->one->path);
 851                 if (strchr(p->two->path, line_termination) ||
 852                     strchr(p->two->path, inter_name_termination))
 853                         die(err, p->two->path);
 854         }
 855
 856         if (p->score)
 857                 sprintf(status, "%c%03d", p->status,
 858                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 859         else {
 860                 status[0] = p->status;
 861                 status[1] = 0;
 862         }
 863         switch (p->status) {
 864         case DIFF_STATUS_COPIED:
 865         case DIFF_STATUS_RENAMED:
 866                 two_paths = 1;
 867                 break;
 868         case DIFF_STATUS_ADDED:
 869         case DIFF_STATUS_DELETED:
 870                 two_paths = 0;
 871                 break;
 872         default:
 873                 two_paths = 0;
 874                 break;
 875         }
 876         if (output_format != DIFF_FORMAT_NAME_STATUS) {
 877                 printf(":%06o %06o %s ",
 878                        p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 879                 printf("%s ", sha1_to_hex(p->two->sha1));
 880         }
 881         printf("%s%c%s",status, inter_name_termination, p->one->path);
 882         if (two_paths)
 883                 printf("%c%s", inter_name_termination, p->two->path);
 884         putchar(line_termination);
 885 }
 886
 887 static void diff_flush_name(struct diff_filepair *p,
 888                             int line_termination)
 889 {
 890         printf("%s%c", p->two->path, line_termination);
 891 }
 892
 893 int diff_unmodified_pair(struct diff_filepair *p)
 894 {
 895         /* This function is written stricter than necessary to support
 896          * the currently implemented transformers, but the idea is to
 897          * let transformers to produce diff_filepairs any way they want,
 898          * and filter and clean them up here before producing the output.
 899          */
 900         struct diff_filespec *one, *two;
 901
 902         if (DIFF_PAIR_UNMERGED(p))
 903                 return 0; /* unmerged is interesting */
 904
 905         one = p->one;
 906         two = p->two;
 907
 908         /* deletion, addition, mode or type change
 909          * and rename are all interesting.
 910          */
 911         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 912             DIFF_PAIR_MODE_CHANGED(p) ||
 913             strcmp(one->path, two->path))
 914                 return 0;
 915
 916         /* both are valid and point at the same path.  that is, we are
 917          * dealing with a change.
 918          */
 919         if (one->sha1_valid && two->sha1_valid &&
 920             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 921                 return 1; /* no change */
 922         if (!one->sha1_valid && !two->sha1_valid)
 923                 return 1; /* both look at the same file on the filesystem. */
 924         return 0;
 925 }
 926
 927 static void diff_flush_patch(struct diff_filepair *p)
 928 {
 929         if (diff_unmodified_pair(p))
 930                 return;
 931
 932         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 933             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 934                 return; /* no tree diffs in patch format */
 935
 936         run_diff(p);
 937 }
 938
 939 int diff_queue_is_empty(void)
 940 {
 941         struct diff_queue_struct *q = &diff_queued_diff;
 942         int i;
 943         for (i = 0; i < q->nr; i++)
 944                 if (!diff_unmodified_pair(q->queue[i]))
 945                         return 0;
 946         return 1;
 947 }
 948
 949 #if DIFF_DEBUG
 950 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 951 {
 952         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 953                 x, one ? one : "",
 954                 s->path,
 955                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 956                 s->mode,
 957                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 958         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 959                 x, one ? one : "",
 960                 s->size, s->xfrm_flags);
 961 }
 962
 963 void diff_debug_filepair(const struct diff_filepair *p, int i)
 964 {
 965         diff_debug_filespec(p->one, i, "one");
 966         diff_debug_filespec(p->two, i, "two");
 967         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 968                 p->score, p->status ? p->status : '?',
 969                 p->source_stays, p->broken_pair);
 970 }
 971
 972 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 973 {
 974         int i;
 975         if (msg)
 976                 fprintf(stderr, "%s\n", msg);
 977         fprintf(stderr, "q->nr = %d\n", q->nr);
 978         for (i = 0; i < q->nr; i++) {
 979                 struct diff_filepair *p = q->queue[i];
 980                 diff_debug_filepair(p, i);
 981         }
 982 }
 983 #endif
 984
 985 static void diff_resolve_rename_copy(void)
 986 {
 987         int i, j;
 988         struct diff_filepair *p, *pp;
 989         struct diff_queue_struct *q = &diff_queued_diff;
 990
 991         diff_debug_queue("resolve-rename-copy", q);
 992
 993         for (i = 0; i < q->nr; i++) {
 994                 p = q->queue[i];
 995                 p->status = 0; /* undecided */
 996                 if (DIFF_PAIR_UNMERGED(p))
 997                         p->status = DIFF_STATUS_UNMERGED;
 998                 else if (!DIFF_FILE_VALID(p->one))
 999                         p->status = DIFF_STATUS_ADDED;
1000                 else if (!DIFF_FILE_VALID(p->two))
1001                         p->status = DIFF_STATUS_DELETED;
1002                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1003                         p->status = DIFF_STATUS_TYPE_CHANGED;
1004
1005                 /* from this point on, we are dealing with a pair
1006                  * whose both sides are valid and of the same type, i.e.
1007                  * either in-place edit or rename/copy edit.
1008                  */
1009                 else if (DIFF_PAIR_RENAME(p)) {
1010                         if (p->source_stays) {
1011                                 p->status = DIFF_STATUS_COPIED;
1012                                 continue;
1013                         }
1014                         /* See if there is some other filepair that
1015                          * copies from the same source as us.  If so
1016                          * we are a copy.  Otherwise we are either a
1017                          * copy if the path stays, or a rename if it
1018                          * does not, but we already handled "stays" case.
1019                          */
1020                         for (j = i + 1; j < q->nr; j++) {
1021                                 pp = q->queue[j];
1022                                 if (strcmp(pp->one->path, p->one->path))
1023                                         continue; /* not us */
1024                                 if (!DIFF_PAIR_RENAME(pp))
1025                                         continue; /* not a rename/copy */
1026                                 /* pp is a rename/copy from the same source */
1027                                 p->status = DIFF_STATUS_COPIED;
1028                                 break;
1029                         }
1030                         if (!p->status)
1031                                 p->status = DIFF_STATUS_RENAMED;
1032                 }
1033                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1034                          p->one->mode != p->two->mode)
1035                         p->status = DIFF_STATUS_MODIFIED;
1036                 else {
1037                         /* This is a "no-change" entry and should not
1038                          * happen anymore, but prepare for broken callers.
1039                          */
1040                         error("feeding unmodified %s to diffcore",
1041                               p->one->path);
1042                         p->status = DIFF_STATUS_UNKNOWN;
1043                 }
1044         }
1045         diff_debug_queue("resolve-rename-copy done", q);
1046 }
1047
1048 void diff_flush(struct diff_options *options)
1049 {
1050         struct diff_queue_struct *q = &diff_queued_diff;
1051         int i;
1052         int inter_name_termination = '\t';
1053         int diff_output_format = options->output_format;
1054         int line_termination = options->line_termination;
1055
1056         if (!line_termination)
1057                 inter_name_termination = 0;
1058
1059         for (i = 0; i < q->nr; i++) {
1060                 struct diff_filepair *p = q->queue[i];
1061                 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1062                     (p->status == DIFF_STATUS_UNKNOWN))
1063                         continue;
1064                 if (p->status == 0)
1065                         die("internal error in diff-resolve-rename-copy");
1066                 switch (diff_output_format) {
1067                 case DIFF_FORMAT_PATCH:
1068                         diff_flush_patch(p);
1069                         break;
1070                 case DIFF_FORMAT_RAW:
1071                 case DIFF_FORMAT_NAME_STATUS:
1072                         diff_flush_raw(p, line_termination,
1073                                        inter_name_termination,
1074                                        diff_output_format);
1075                         break;
1076                 case DIFF_FORMAT_NAME:
1077                         diff_flush_name(p, line_termination);
1078                         break;
1079                 }
1080                 diff_free_filepair(q->queue[i]);
1081         }
1082         free(q->queue);
1083         q->queue = NULL;
1084         q->nr = q->alloc = 0;
1085 }
1086
1087 static void diffcore_apply_filter(const char *filter)
1088 {
1089         int i;
1090         struct diff_queue_struct *q = &diff_queued_diff;
1091         struct diff_queue_struct outq;
1092         outq.queue = NULL;
1093         outq.nr = outq.alloc = 0;
1094
1095         if (!filter)
1096                 return;
1097
1098         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1099                 int found;
1100                 for (i = found = 0; !found && i < q->nr; i++) {
1101                         struct diff_filepair *p = q->queue[i];
1102                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1103                              ((p->score &&
1104                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1105                               (!p->score &&
1106                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1107                             ((p->status != DIFF_STATUS_MODIFIED) &&
1108                              strchr(filter, p->status)))
1109                                 found++;
1110                 }
1111                 if (found)
1112                         return;
1113
1114                 /* otherwise we will clear the whole queue
1115                  * by copying the empty outq at the end of this
1116                  * function, but first clear the current entries
1117                  * in the queue.
1118                  */
1119                 for (i = 0; i < q->nr; i++)
1120                         diff_free_filepair(q->queue[i]);
1121         }
1122         else {
1123                 /* Only the matching ones */
1124                 for (i = 0; i < q->nr; i++) {
1125                         struct diff_filepair *p = q->queue[i];
1126
1127                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1128                              ((p->score &&
1129                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1130                               (!p->score &&
1131                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1132                             ((p->status != DIFF_STATUS_MODIFIED) &&
1133                              strchr(filter, p->status)))
1134                                 diff_q(&outq, p);
1135                         else
1136                                 diff_free_filepair(p);
1137                 }
1138         }
1139         free(q->queue);
1140         *q = outq;
1141 }
1142
1143 void diffcore_std(struct diff_options *options)
1144 {
1145         if (options->paths && options->paths[0])
1146                 diffcore_pathspec(options->paths);
1147         if (options->break_opt != -1)
1148                 diffcore_break(options->break_opt);
1149         if (options->detect_rename)
1150                 diffcore_rename(options);
1151         if (options->break_opt != -1)
1152                 diffcore_merge_broken();
1153         if (options->pickaxe)
1154                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1155         if (options->orderfile)
1156                 diffcore_order(options->orderfile);
1157         diff_resolve_rename_copy();
1158         diffcore_apply_filter(options->filter);
1159 }
1160
1161
1162 void diffcore_std_no_resolve(struct diff_options *options)
1163 {
1164         if (options->pickaxe)
1165                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1166         if (options->orderfile)
1167                 diffcore_order(options->orderfile);
1168         diffcore_apply_filter(options->filter);
1169 }
1170
1171 void diff_addremove(struct diff_options *options,
1172                     int addremove, unsigned mode,
1173                     const unsigned char *sha1,
1174                     const char *base, const char *path)
1175 {
1176         char concatpath[PATH_MAX];
1177         struct diff_filespec *one, *two;
1178
1179         /* This may look odd, but it is a preparation for
1180          * feeding "there are unchanged files which should
1181          * not produce diffs, but when you are doing copy
1182          * detection you would need them, so here they are"
1183          * entries to the diff-core.  They will be prefixed
1184          * with something like '=' or '*' (I haven't decided
1185          * which but should not make any difference).
1186          * Feeding the same new and old to diff_change()
1187          * also has the same effect.
1188          * Before the final output happens, they are pruned after
1189          * merged into rename/copy pairs as appropriate.
1190          */
1191         if (options->reverse_diff)
1192                 addremove = (addremove == '+' ? '-' :
1193                              addremove == '-' ? '+' : addremove);
1194
1195         if (!path) path = "";
1196         sprintf(concatpath, "%s%s", base, path);
1197         one = alloc_filespec(concatpath);
1198         two = alloc_filespec(concatpath);
1199
1200         if (addremove != '+')
1201                 fill_filespec(one, sha1, mode);
1202         if (addremove != '-')
1203                 fill_filespec(two, sha1, mode);
1204
1205         diff_queue(&diff_queued_diff, one, two);
1206 }
1207
1208 void diff_change(struct diff_options *options,
1209                  unsigned old_mode, unsigned new_mode,
1210                  const unsigned char *old_sha1,
1211                  const unsigned char *new_sha1,
1212                  const char *base, const char *path)
1213 {
1214         char concatpath[PATH_MAX];
1215         struct diff_filespec *one, *two;
1216
1217         if (options->reverse_diff) {
1218                 unsigned tmp;
1219                 const unsigned char *tmp_c;
1220                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1221                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1222         }
1223         if (!path) path = "";
1224         sprintf(concatpath, "%s%s", base, path);
1225         one = alloc_filespec(concatpath);
1226         two = alloc_filespec(concatpath);
1227         fill_filespec(one, old_sha1, old_mode);
1228         fill_filespec(two, new_sha1, new_mode);
1229
1230         diff_queue(&diff_queued_diff, one, two);
1231 }
1232
1233 void diff_unmerge(struct diff_options *options,
1234                   const char *path)
1235 {
1236         struct diff_filespec *one, *two;
1237         one = alloc_filespec(path);
1238         two = alloc_filespec(path);
1239         diff_queue(&diff_queued_diff, one, two);
1240 }