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 static unsigned char null_sha1[20] = { 0, };
  14
  15 static int reverse_diff;
  16 static int use_size_cache;
  17
  18 static const char *external_diff(void)
  19 {
  20         static const char *external_diff_cmd = NULL;
  21         static int done_preparing = 0;
  22         const char *env_diff_opts;
  23
  24         if (done_preparing)
  25                 return external_diff_cmd;
  26
  27         /*
  28          * Default values above are meant to match the
  29          * Linux kernel development style.  Examples of
  30          * alternative styles you can specify via environment
  31          * variables are:
  32          *
  33          * GIT_DIFF_OPTS="-c";
  34          */
  35         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  36
  37         /* In case external diff fails... */
  38         env_diff_opts = getenv("GIT_DIFF_OPTS");
  39         if (env_diff_opts) diff_opts = env_diff_opts;
  40
  41         done_preparing = 1;
  42         return external_diff_cmd;
  43 }
  44
  45 #define TEMPFILE_PATH_LEN               50
  46
  47 static struct diff_tempfile {
  48         const char *name; /* filename external diff should read from */
  49         char hex[41];
  50         char mode[10];
  51         char tmp_path[TEMPFILE_PATH_LEN];
  52 } diff_temp[2];
  53
  54 static int count_lines(const char *filename)
  55 {
  56         FILE *in;
  57         int count, ch, completely_empty = 1, nl_just_seen = 0;
  58         in = fopen(filename, "r");
  59         count = 0;
  60         while ((ch = fgetc(in)) != EOF)
  61                 if (ch == '\n') {
  62                         count++;
  63                         nl_just_seen = 1;
  64                         completely_empty = 0;
  65                 }
  66                 else {
  67                         nl_just_seen = 0;
  68                         completely_empty = 0;
  69                 }
  70         fclose(in);
  71         if (completely_empty)
  72                 return 0;
  73         if (!nl_just_seen)
  74                 count++; /* no trailing newline */
  75         return count;
  76 }
  77
  78 static void print_line_count(int count)
  79 {
  80         switch (count) {
  81         case 0:
  82                 printf("0,0");
  83                 break;
  84         case 1:
  85                 printf("1");
  86                 break;
  87         default:
  88                 printf("1,%d", count);
  89                 break;
  90         }
  91 }
  92
  93 static void copy_file(int prefix, const char *filename)
  94 {
  95         FILE *in;
  96         int ch, nl_just_seen = 1;
  97         in = fopen(filename, "r");
  98         while ((ch = fgetc(in)) != EOF) {
  99                 if (nl_just_seen)
 100                         putchar(prefix);
 101                 putchar(ch);
 102                 if (ch == '\n')
 103                         nl_just_seen = 1;
 104                 else
 105                         nl_just_seen = 0;
 106         }
 107         fclose(in);
 108         if (!nl_just_seen)
 109                 printf("\n\\ No newline at end of file\n");
 110 }
 111
 112 static void emit_rewrite_diff(const char *name_a,
 113                               const char *name_b,
 114                               struct diff_tempfile *temp)
 115 {
 116         /* Use temp[i].name as input, name_a and name_b as labels */
 117         int lc_a, lc_b;
 118         lc_a = count_lines(temp[0].name);
 119         lc_b = count_lines(temp[1].name);
 120         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 121         print_line_count(lc_a);
 122         printf(" +");
 123         print_line_count(lc_b);
 124         printf(" @@\n");
 125         if (lc_a)
 126                 copy_file('-', temp[0].name);
 127         if (lc_b)
 128                 copy_file('+', temp[1].name);
 129 }
 130
 131 static void builtin_diff(const char *name_a,
 132                          const char *name_b,
 133                          struct diff_tempfile *temp,
 134                          const char *xfrm_msg,
 135                          int complete_rewrite)
 136 {
 137         int i, next_at, cmd_size;
 138         const char *const diff_cmd = "diff -L%s%s -L%s%s";
 139         const char *const diff_arg  = "%s %s||:"; /* "||:" is to return 0 */
 140         const char *input_name_sq[2];
 141         const char *path0[2];
 142         const char *path1[2];
 143         const char *name_sq[2];
 144         char *cmd;
 145
 146         name_sq[0] = sq_quote(name_a);
 147         name_sq[1] = sq_quote(name_b);
 148
 149         /* diff_cmd and diff_arg have 6 %s in total which makes
 150          * the sum of these strings 12 bytes larger than required.
 151          * we use 2 spaces around diff-opts, and we need to count
 152          * terminating NUL, so we subtract 9 here.
 153          */
 154         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 155                         strlen(diff_arg) - 9);
 156         for (i = 0; i < 2; i++) {
 157                 input_name_sq[i] = sq_quote(temp[i].name);
 158                 if (!strcmp(temp[i].name, "/dev/null")) {
 159                         path0[i] = "/dev/null";
 160                         path1[i] = "";
 161                 } else {
 162                         path0[i] = i ? "b/" : "a/";
 163                         path1[i] = name_sq[i];
 164                 }
 165                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 166                              strlen(input_name_sq[i]));
 167         }
 168
 169         cmd = xmalloc(cmd_size);
 170
 171         next_at = 0;
 172         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 173                             diff_cmd,
 174                             path0[0], path1[0], path0[1], path1[1]);
 175         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 176                             " %s ", diff_opts);
 177         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 178                             diff_arg, input_name_sq[0], input_name_sq[1]);
 179
 180         printf("diff --git a/%s b/%s\n", name_a, name_b);
 181         if (!path1[0][0]) {
 182                 printf("new file mode %s\n", temp[1].mode);
 183                 if (xfrm_msg && xfrm_msg[0])
 184                         puts(xfrm_msg);
 185         }
 186         else if (!path1[1][0]) {
 187                 printf("deleted file mode %s\n", temp[0].mode);
 188                 if (xfrm_msg && xfrm_msg[0])
 189                         puts(xfrm_msg);
 190         }
 191         else {
 192                 if (strcmp(temp[0].mode, temp[1].mode)) {
 193                         printf("old mode %s\n", temp[0].mode);
 194                         printf("new mode %s\n", temp[1].mode);
 195                 }
 196                 if (xfrm_msg && xfrm_msg[0])
 197                         puts(xfrm_msg);
 198                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 199                         /* we do not run diff between different kind
 200                          * of objects.
 201                          */
 202                         exit(0);
 203                 if (complete_rewrite) {
 204                         fflush(NULL);
 205                         emit_rewrite_diff(name_a, name_b, temp);
 206                         exit(0);
 207                 }
 208         }
 209         fflush(NULL);
 210         execlp("/bin/sh","sh", "-c", cmd, NULL);
 211 }
 212
 213 struct diff_filespec *alloc_filespec(const char *path)
 214 {
 215         int namelen = strlen(path);
 216         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 217         spec->path = (char *)(spec + 1);
 218         strcpy(spec->path, path);
 219         spec->should_free = spec->should_munmap = 0;
 220         spec->xfrm_flags = 0;
 221         spec->size = 0;
 222         spec->data = NULL;
 223         spec->mode = 0;
 224         memset(spec->sha1, 0, 20);
 225         return spec;
 226 }
 227
 228 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 229                    unsigned short mode)
 230 {
 231         if (mode) {
 232                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 233                 memcpy(spec->sha1, sha1, 20);
 234                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 235         }
 236 }
 237
 238 /*
 239  * Given a name and sha1 pair, if the dircache tells us the file in
 240  * the work tree has that object contents, return true, so that
 241  * prepare_temp_file() does not have to inflate and extract.
 242  */
 243 static int work_tree_matches(const char *name, const unsigned char *sha1)
 244 {
 245         struct cache_entry *ce;
 246         struct stat st;
 247         int pos, len;
 248
 249         /* We do not read the cache ourselves here, because the
 250          * benchmark with my previous version that always reads cache
 251          * shows that it makes things worse for diff-tree comparing
 252          * two linux-2.6 kernel trees in an already checked out work
 253          * tree.  This is because most diff-tree comparisons deal with
 254          * only a small number of files, while reading the cache is
 255          * expensive for a large project, and its cost outweighs the
 256          * savings we get by not inflating the object to a temporary
 257          * file.  Practically, this code only helps when we are used
 258          * by diff-cache --cached, which does read the cache before
 259          * calling us.
 260          */
 261         if (!active_cache)
 262                 return 0;
 263
 264         len = strlen(name);
 265         pos = cache_name_pos(name, len);
 266         if (pos < 0)
 267                 return 0;
 268         ce = active_cache[pos];
 269         if ((lstat(name, &st) < 0) ||
 270             !S_ISREG(st.st_mode) || /* careful! */
 271             ce_match_stat(ce, &st) ||
 272             memcmp(sha1, ce->sha1, 20))
 273                 return 0;
 274         /* we return 1 only when we can stat, it is a regular file,
 275          * stat information matches, and sha1 recorded in the cache
 276          * matches.  I.e. we know the file in the work tree really is
 277          * the same as the <name, sha1> pair.
 278          */
 279         return 1;
 280 }
 281
 282 static struct sha1_size_cache {
 283         unsigned char sha1[20];
 284         unsigned long size;
 285 } **sha1_size_cache;
 286 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 287
 288 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 289                                                  int find_only,
 290                                                  unsigned long size)
 291 {
 292         int first, last;
 293         struct sha1_size_cache *e;
 294
 295         first = 0;
 296         last = sha1_size_cache_nr;
 297         while (last > first) {
 298                 int cmp, next = (last + first) >> 1;
 299                 e = sha1_size_cache[next];
 300                 cmp = memcmp(e->sha1, sha1, 20);
 301                 if (!cmp)
 302                         return e;
 303                 if (cmp < 0) {
 304                         last = next;
 305                         continue;
 306                 }
 307                 first = next+1;
 308         }
 309         /* not found */
 310         if (find_only)
 311                 return NULL;
 312         /* insert to make it at "first" */
 313         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 314                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 315                 sha1_size_cache = xrealloc(sha1_size_cache,
 316                                            sha1_size_cache_alloc *
 317                                            sizeof(*sha1_size_cache));
 318         }
 319         sha1_size_cache_nr++;
 320         if (first < sha1_size_cache_nr)
 321                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 322                         (sha1_size_cache_nr - first - 1) *
 323                         sizeof(*sha1_size_cache));
 324         e = xmalloc(sizeof(struct sha1_size_cache));
 325         sha1_size_cache[first] = e;
 326         memcpy(e->sha1, sha1, 20);
 327         e->size = size;
 328         return e;
 329 }
 330
 331 /*
 332  * While doing rename detection and pickaxe operation, we may need to
 333  * grab the data for the blob (or file) for our own in-core comparison.
 334  * diff_filespec has data and size fields for this purpose.
 335  */
 336 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 337 {
 338         int err = 0;
 339         if (!DIFF_FILE_VALID(s))
 340                 die("internal error: asking to populate invalid file.");
 341         if (S_ISDIR(s->mode))
 342                 return -1;
 343
 344         if (!use_size_cache)
 345                 size_only = 0;
 346
 347         if (s->data)
 348                 return err;
 349         if (!s->sha1_valid ||
 350             work_tree_matches(s->path, s->sha1)) {
 351                 struct stat st;
 352                 int fd;
 353                 if (lstat(s->path, &st) < 0) {
 354                         if (errno == ENOENT) {
 355                         err_empty:
 356                                 err = -1;
 357                         empty:
 358                                 s->data = "";
 359                                 s->size = 0;
 360                                 return err;
 361                         }
 362                 }
 363                 s->size = st.st_size;
 364                 if (!s->size)
 365                         goto empty;
 366                 if (size_only)
 367                         return 0;
 368                 if (S_ISLNK(st.st_mode)) {
 369                         int ret;
 370                         s->data = xmalloc(s->size);
 371                         s->should_free = 1;
 372                         ret = readlink(s->path, s->data, s->size);
 373                         if (ret < 0) {
 374                                 free(s->data);
 375                                 goto err_empty;
 376                         }
 377                         return 0;
 378                 }
 379                 fd = open(s->path, O_RDONLY);
 380                 if (fd < 0)
 381                         goto err_empty;
 382                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 383                 close(fd);
 384                 if (s->data == MAP_FAILED)
 385                         goto err_empty;
 386                 s->should_munmap = 1;
 387         }
 388         else {
 389                 char type[20];
 390                 struct sha1_size_cache *e;
 391
 392                 if (size_only) {
 393                         e = locate_size_cache(s->sha1, 1, 0);
 394                         if (e) {
 395                                 s->size = e->size;
 396                                 return 0;
 397                         }
 398                         if (!sha1_object_info(s->sha1, type, &s->size))
 399                                 locate_size_cache(s->sha1, 0, s->size);
 400                 }
 401                 else {
 402                         s->data = read_sha1_file(s->sha1, type, &s->size);
 403                         s->should_free = 1;
 404                 }
 405         }
 406         return 0;
 407 }
 408
 409 void diff_free_filespec(struct diff_filespec *s)
 410 {
 411         if (s->should_free)
 412                 free(s->data);
 413         else if (s->should_munmap)
 414                 munmap(s->data, s->size);
 415         free(s);
 416 }
 417
 418 static void prep_temp_blob(struct diff_tempfile *temp,
 419                            void *blob,
 420                            unsigned long size,
 421                            unsigned char *sha1,
 422                            int mode)
 423 {
 424         int fd;
 425
 426         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 427         if (fd < 0)
 428                 die("unable to create temp-file");
 429         if (write(fd, blob, size) != size)
 430                 die("unable to write temp-file");
 431         close(fd);
 432         temp->name = temp->tmp_path;
 433         strcpy(temp->hex, sha1_to_hex(sha1));
 434         temp->hex[40] = 0;
 435         sprintf(temp->mode, "%06o", mode);
 436 }
 437
 438 static void prepare_temp_file(const char *name,
 439                               struct diff_tempfile *temp,
 440                               struct diff_filespec *one)
 441 {
 442         if (!DIFF_FILE_VALID(one)) {
 443         not_a_valid_file:
 444                 /* A '-' entry produces this for file-2, and
 445                  * a '+' entry produces this for file-1.
 446                  */
 447                 temp->name = "/dev/null";
 448                 strcpy(temp->hex, ".");
 449                 strcpy(temp->mode, ".");
 450                 return;
 451         }
 452
 453         if (!one->sha1_valid ||
 454             work_tree_matches(name, one->sha1)) {
 455                 struct stat st;
 456                 if (lstat(name, &st) < 0) {
 457                         if (errno == ENOENT)
 458                                 goto not_a_valid_file;
 459                         die("stat(%s): %s", name, strerror(errno));
 460                 }
 461                 if (S_ISLNK(st.st_mode)) {
 462                         int ret;
 463                         char *buf, buf_[1024];
 464                         buf = ((sizeof(buf_) < st.st_size) ?
 465                                xmalloc(st.st_size) : buf_);
 466                         ret = readlink(name, buf, st.st_size);
 467                         if (ret < 0)
 468                                 die("readlink(%s)", name);
 469                         prep_temp_blob(temp, buf, st.st_size,
 470                                        (one->sha1_valid ?
 471                                         one->sha1 : null_sha1),
 472                                        (one->sha1_valid ?
 473                                         one->mode : S_IFLNK));
 474                 }
 475                 else {
 476                         /* we can borrow from the file in the work tree */
 477                         temp->name = name;
 478                         if (!one->sha1_valid)
 479                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 480                         else
 481                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 482                         /* Even though we may sometimes borrow the
 483                          * contents from the work tree, we always want
 484                          * one->mode.  mode is trustworthy even when
 485                          * !(one->sha1_valid), as long as
 486                          * DIFF_FILE_VALID(one).
 487                          */
 488                         sprintf(temp->mode, "%06o", one->mode);
 489                 }
 490                 return;
 491         }
 492         else {
 493                 if (diff_populate_filespec(one, 0))
 494                         die("cannot read data blob for %s", one->path);
 495                 prep_temp_blob(temp, one->data, one->size,
 496                                one->sha1, one->mode);
 497         }
 498 }
 499
 500 static void remove_tempfile(void)
 501 {
 502         int i;
 503
 504         for (i = 0; i < 2; i++)
 505                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 506                         unlink(diff_temp[i].name);
 507                         diff_temp[i].name = NULL;
 508                 }
 509 }
 510
 511 static void remove_tempfile_on_signal(int signo)
 512 {
 513         remove_tempfile();
 514 }
 515
 516 /* An external diff command takes:
 517  *
 518  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 519  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 520  *
 521  */
 522 static void run_external_diff(const char *pgm,
 523                               const char *name,
 524                               const char *other,
 525                               struct diff_filespec *one,
 526                               struct diff_filespec *two,
 527                               const char *xfrm_msg,
 528                               int complete_rewrite)
 529 {
 530         struct diff_tempfile *temp = diff_temp;
 531         pid_t pid;
 532         int status;
 533         static int atexit_asked = 0;
 534         const char *othername;
 535
 536         othername = (other? other : name);
 537         if (one && two) {
 538                 prepare_temp_file(name, &temp[0], one);
 539                 prepare_temp_file(othername, &temp[1], two);
 540                 if (! atexit_asked &&
 541                     (temp[0].name == temp[0].tmp_path ||
 542                      temp[1].name == temp[1].tmp_path)) {
 543                         atexit_asked = 1;
 544                         atexit(remove_tempfile);
 545                 }
 546                 signal(SIGINT, remove_tempfile_on_signal);
 547         }
 548
 549         fflush(NULL);
 550         pid = fork();
 551         if (pid < 0)
 552                 die("unable to fork");
 553         if (!pid) {
 554                 if (pgm) {
 555                         if (one && two) {
 556                                 const char *exec_arg[10];
 557                                 const char **arg = &exec_arg[0];
 558                                 *arg++ = pgm;
 559                                 *arg++ = name;
 560                                 *arg++ = temp[0].name;
 561                                 *arg++ = temp[0].hex;
 562                                 *arg++ = temp[0].mode;
 563                                 *arg++ = temp[1].name;
 564                                 *arg++ = temp[1].hex;
 565                                 *arg++ = temp[1].mode;
 566                                 if (other) {
 567                                         *arg++ = other;
 568                                         *arg++ = xfrm_msg;
 569                                 }
 570                                 *arg = NULL;
 571                                 execvp(pgm, (char *const*) exec_arg);
 572                         }
 573                         else
 574                                 execlp(pgm, pgm, name, NULL);
 575                 }
 576                 /*
 577                  * otherwise we use the built-in one.
 578                  */
 579                 if (one && two)
 580                         builtin_diff(name, othername, temp, xfrm_msg,
 581                                      complete_rewrite);
 582                 else
 583                         printf("* Unmerged path %s\n", name);
 584                 exit(0);
 585         }
 586         if (waitpid(pid, &status, 0) < 0 ||
 587             !WIFEXITED(status) || WEXITSTATUS(status)) {
 588                 /* Earlier we did not check the exit status because
 589                  * diff exits non-zero if files are different, and
 590                  * we are not interested in knowing that.  It was a
 591                  * mistake which made it harder to quit a diff-*
 592                  * session that uses the git-apply-patch-script as
 593                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 594                  * should also exit non-zero only when it wants to
 595                  * abort the entire diff-* session.
 596                  */
 597                 remove_tempfile();
 598                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 599                 exit(1);
 600         }
 601         remove_tempfile();
 602 }
 603
 604 static void run_diff(struct diff_filepair *p)
 605 {
 606         const char *pgm = external_diff();
 607         char msg_[PATH_MAX*2+200], *xfrm_msg;
 608         struct diff_filespec *one;
 609         struct diff_filespec *two;
 610         const char *name;
 611         const char *other;
 612         int complete_rewrite = 0;
 613
 614         if (DIFF_PAIR_UNMERGED(p)) {
 615                 /* unmerged */
 616                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 617                                   0);
 618                 return;
 619         }
 620
 621         name = p->one->path;
 622         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 623         one = p->one; two = p->two;
 624         switch (p->status) {
 625         case DIFF_STATUS_COPIED:
 626                 sprintf(msg_,
 627                         "similarity index %d%%\n"
 628                         "copy from %s\n"
 629                         "copy to %s",
 630                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 631                         name, other);
 632                 xfrm_msg = msg_;
 633                 break;
 634         case DIFF_STATUS_RENAMED:
 635                 sprintf(msg_,
 636                         "similarity index %d%%\n"
 637                         "rename from %s\n"
 638                         "rename to %s",
 639                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 640                         name, other);
 641                 xfrm_msg = msg_;
 642                 break;
 643         case DIFF_STATUS_MODIFIED:
 644                 if (p->score) {
 645                         sprintf(msg_,
 646                                 "dissimilarity index %d%%",
 647                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 648                         xfrm_msg = msg_;
 649                         complete_rewrite = 1;
 650                         break;
 651                 }
 652                 /* fallthru */
 653         default:
 654                 xfrm_msg = NULL;
 655         }
 656
 657         if (!pgm &&
 658             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 659             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 660                 /* a filepair that changes between file and symlink
 661                  * needs to be split into deletion and creation.
 662                  */
 663                 struct diff_filespec *null = alloc_filespec(two->path);
 664                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 665                 free(null);
 666                 null = alloc_filespec(one->path);
 667                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 668                 free(null);
 669         }
 670         else
 671                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 672                                   complete_rewrite);
 673 }
 674
 675 void diff_setup(int flags)
 676 {
 677         if (flags & DIFF_SETUP_REVERSE)
 678                 reverse_diff = 1;
 679         if (flags & DIFF_SETUP_USE_CACHE) {
 680                 if (!active_cache)
 681                         /* read-cache does not die even when it fails
 682                          * so it is safe for us to do this here.  Also
 683                          * it does not smudge active_cache or active_nr
 684                          * when it fails, so we do not have to worry about
 685                          * cleaning it up oufselves either.
 686                          */
 687                         read_cache();
 688         }
 689         if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 690                 use_size_cache = 1;
 691
 692 }
 693
 694 static int parse_num(const char **cp_p)
 695 {
 696         int num, scale, ch, cnt;
 697         const char *cp = *cp_p;
 698
 699         cnt = num = 0;
 700         scale = 1;
 701         while ('0' <= (ch = *cp) && ch <= '9') {
 702                 if (cnt++ < 5) {
 703                         /* We simply ignore more than 5 digits precision. */
 704                         scale *= 10;
 705                         num = num * 10 + ch - '0';
 706                 }
 707                 cp++;
 708         }
 709         *cp_p = cp;
 710
 711         /* user says num divided by scale and we say internally that
 712          * is MAX_SCORE * num / scale.
 713          */
 714         return (MAX_SCORE * num / scale);
 715 }
 716
 717 int diff_scoreopt_parse(const char *opt)
 718 {
 719         int opt1, opt2, cmd;
 720
 721         if (*opt++ != '-')
 722                 return -1;
 723         cmd = *opt++;
 724         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 725                 return -1; /* that is not a -M, -C nor -B option */
 726
 727         opt1 = parse_num(&opt);
 728         if (cmd != 'B')
 729                 opt2 = 0;
 730         else {
 731                 if (*opt == 0)
 732                         opt2 = 0;
 733                 else if (*opt != '/')
 734                         return -1; /* we expect -B80/99 or -B80 */
 735                 else {
 736                         opt++;
 737                         opt2 = parse_num(&opt);
 738                 }
 739         }
 740         if (*opt != 0)
 741                 return -1;
 742         return opt1 | (opt2 << 16);
 743 }
 744
 745 struct diff_queue_struct diff_queued_diff;
 746
 747 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 748 {
 749         if (queue->alloc <= queue->nr) {
 750                 queue->alloc = alloc_nr(queue->alloc);
 751                 queue->queue = xrealloc(queue->queue,
 752                                         sizeof(dp) * queue->alloc);
 753         }
 754         queue->queue[queue->nr++] = dp;
 755 }
 756
 757 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 758                                  struct diff_filespec *one,
 759                                  struct diff_filespec *two)
 760 {
 761         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 762         dp->one = one;
 763         dp->two = two;
 764         dp->score = 0;
 765         dp->status = 0;
 766         dp->source_stays = 0;
 767         dp->broken_pair = 0;
 768         diff_q(queue, dp);
 769         return dp;
 770 }
 771
 772 void diff_free_filepair(struct diff_filepair *p)
 773 {
 774         diff_free_filespec(p->one);
 775         diff_free_filespec(p->two);
 776         free(p);
 777 }
 778
 779 static void diff_flush_raw(struct diff_filepair *p,
 780                            int line_termination,
 781                            int inter_name_termination)
 782 {
 783         int two_paths;
 784         char status[10];
 785
 786         if (line_termination) {
 787                 const char *const err =
 788                         "path %s cannot be expressed without -z";
 789                 if (strchr(p->one->path, line_termination) ||
 790                     strchr(p->one->path, inter_name_termination))
 791                         die(err, p->one->path);
 792                 if (strchr(p->two->path, line_termination) ||
 793                     strchr(p->two->path, inter_name_termination))
 794                         die(err, p->two->path);
 795         }
 796
 797         if (p->score)
 798                 sprintf(status, "%c%03d", p->status,
 799                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 800         else {
 801                 status[0] = p->status;
 802                 status[1] = 0;
 803         }
 804         switch (p->status) {
 805         case DIFF_STATUS_COPIED:
 806         case DIFF_STATUS_RENAMED:
 807                 two_paths = 1;
 808                 break;
 809         case DIFF_STATUS_ADDED:
 810         case DIFF_STATUS_DELETED:
 811                 two_paths = 0;
 812                 break;
 813         default:
 814                 two_paths = 0;
 815                 break;
 816         }
 817         printf(":%06o %06o %s ",
 818                p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 819         printf("%s %s%c%s",
 820                sha1_to_hex(p->two->sha1),
 821                status,
 822                inter_name_termination,
 823                p->one->path);
 824         if (two_paths)
 825                 printf("%c%s", inter_name_termination, p->two->path);
 826         putchar(line_termination);
 827 }
 828
 829 static void diff_flush_name(struct diff_filepair *p,
 830                             int line_termination)
 831 {
 832         printf("%s%c", p->two->path, line_termination);
 833 }
 834
 835 int diff_unmodified_pair(struct diff_filepair *p)
 836 {
 837         /* This function is written stricter than necessary to support
 838          * the currently implemented transformers, but the idea is to
 839          * let transformers to produce diff_filepairs any way they want,
 840          * and filter and clean them up here before producing the output.
 841          */
 842         struct diff_filespec *one, *two;
 843
 844         if (DIFF_PAIR_UNMERGED(p))
 845                 return 0; /* unmerged is interesting */
 846
 847         one = p->one;
 848         two = p->two;
 849
 850         /* deletion, addition, mode or type change
 851          * and rename are all interesting.
 852          */
 853         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 854             DIFF_PAIR_MODE_CHANGED(p) ||
 855             strcmp(one->path, two->path))
 856                 return 0;
 857
 858         /* both are valid and point at the same path.  that is, we are
 859          * dealing with a change.
 860          */
 861         if (one->sha1_valid && two->sha1_valid &&
 862             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 863                 return 1; /* no change */
 864         if (!one->sha1_valid && !two->sha1_valid)
 865                 return 1; /* both look at the same file on the filesystem. */
 866         return 0;
 867 }
 868
 869 static void diff_flush_patch(struct diff_filepair *p)
 870 {
 871         if (diff_unmodified_pair(p))
 872                 return;
 873
 874         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 875             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 876                 return; /* no tree diffs in patch format */
 877
 878         run_diff(p);
 879 }
 880
 881 int diff_queue_is_empty(void)
 882 {
 883         struct diff_queue_struct *q = &diff_queued_diff;
 884         int i;
 885         for (i = 0; i < q->nr; i++)
 886                 if (!diff_unmodified_pair(q->queue[i]))
 887                         return 0;
 888         return 1;
 889 }
 890
 891 #if DIFF_DEBUG
 892 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 893 {
 894         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 895                 x, one ? one : "",
 896                 s->path,
 897                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 898                 s->mode,
 899                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 900         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 901                 x, one ? one : "",
 902                 s->size, s->xfrm_flags);
 903 }
 904
 905 void diff_debug_filepair(const struct diff_filepair *p, int i)
 906 {
 907         diff_debug_filespec(p->one, i, "one");
 908         diff_debug_filespec(p->two, i, "two");
 909         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 910                 p->score, p->status ? p->status : '?',
 911                 p->source_stays, p->broken_pair);
 912 }
 913
 914 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 915 {
 916         int i;
 917         if (msg)
 918                 fprintf(stderr, "%s\n", msg);
 919         fprintf(stderr, "q->nr = %d\n", q->nr);
 920         for (i = 0; i < q->nr; i++) {
 921                 struct diff_filepair *p = q->queue[i];
 922                 diff_debug_filepair(p, i);
 923         }
 924 }
 925 #endif
 926
 927 static void diff_resolve_rename_copy(void)
 928 {
 929         int i, j;
 930         struct diff_filepair *p, *pp;
 931         struct diff_queue_struct *q = &diff_queued_diff;
 932
 933         diff_debug_queue("resolve-rename-copy", q);
 934
 935         for (i = 0; i < q->nr; i++) {
 936                 p = q->queue[i];
 937                 p->status = 0; /* undecided */
 938                 if (DIFF_PAIR_UNMERGED(p))
 939                         p->status = DIFF_STATUS_UNMERGED;
 940                 else if (!DIFF_FILE_VALID(p->one))
 941                         p->status = DIFF_STATUS_ADDED;
 942                 else if (!DIFF_FILE_VALID(p->two))
 943                         p->status = DIFF_STATUS_DELETED;
 944                 else if (DIFF_PAIR_TYPE_CHANGED(p))
 945                         p->status = DIFF_STATUS_TYPE_CHANGED;
 946
 947                 /* from this point on, we are dealing with a pair
 948                  * whose both sides are valid and of the same type, i.e.
 949                  * either in-place edit or rename/copy edit.
 950                  */
 951                 else if (DIFF_PAIR_RENAME(p)) {
 952                         if (p->source_stays) {
 953                                 p->status = DIFF_STATUS_COPIED;
 954                                 continue;
 955                         }
 956                         /* See if there is some other filepair that
 957                          * copies from the same source as us.  If so
 958                          * we are a copy.  Otherwise we are either a
 959                          * copy if the path stays, or a rename if it
 960                          * does not, but we already handled "stays" case.
 961                          */
 962                         for (j = i + 1; j < q->nr; j++) {
 963                                 pp = q->queue[j];
 964                                 if (strcmp(pp->one->path, p->one->path))
 965                                         continue; /* not us */
 966                                 if (!DIFF_PAIR_RENAME(pp))
 967                                         continue; /* not a rename/copy */
 968                                 /* pp is a rename/copy from the same source */
 969                                 p->status = DIFF_STATUS_COPIED;
 970                                 break;
 971                         }
 972                         if (!p->status)
 973                                 p->status = DIFF_STATUS_RENAMED;
 974                 }
 975                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 976                          p->one->mode != p->two->mode)
 977                         p->status = DIFF_STATUS_MODIFIED;
 978                 else {
 979                         /* This is a "no-change" entry and should not
 980                          * happen anymore, but prepare for broken callers.
 981                          */
 982                         error("feeding unmodified %s to diffcore",
 983                               p->one->path);
 984                         p->status = DIFF_STATUS_UNKNOWN;
 985                 }
 986         }
 987         diff_debug_queue("resolve-rename-copy done", q);
 988 }
 989
 990 void diff_flush(int diff_output_style, int line_termination)
 991 {
 992         struct diff_queue_struct *q = &diff_queued_diff;
 993         int i;
 994         int inter_name_termination = '\t';
 995
 996         if (!line_termination)
 997                 inter_name_termination = 0;
 998
 999         for (i = 0; i < q->nr; i++) {
1000                 struct diff_filepair *p = q->queue[i];
1001                 if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
1002                     (p->status == DIFF_STATUS_UNKNOWN))
1003                         continue;
1004                 if (p->status == 0)
1005                         die("internal error in diff-resolve-rename-copy");
1006                 switch (diff_output_style) {
1007                 case DIFF_FORMAT_PATCH:
1008                         diff_flush_patch(p);
1009                         break;
1010                 case DIFF_FORMAT_RAW:
1011                         diff_flush_raw(p, line_termination,
1012                                        inter_name_termination);
1013                         break;
1014                 case DIFF_FORMAT_NAME:
1015                         diff_flush_name(p, line_termination);
1016                         break;
1017                 }
1018                 diff_free_filepair(q->queue[i]);
1019         }
1020         free(q->queue);
1021         q->queue = NULL;
1022         q->nr = q->alloc = 0;
1023 }
1024
1025 static void diffcore_apply_filter(const char *filter)
1026 {
1027         int i;
1028         struct diff_queue_struct *q = &diff_queued_diff;
1029         struct diff_queue_struct outq;
1030         outq.queue = NULL;
1031         outq.nr = outq.alloc = 0;
1032
1033         if (!filter)
1034                 return;
1035
1036         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1037                 int found;
1038                 for (i = found = 0; !found && i < q->nr; i++) {
1039                         struct diff_filepair *p = q->queue[i];
1040                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1041                              ((p->score &&
1042                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1043                               (!p->score &&
1044                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1045                             ((p->status != DIFF_STATUS_MODIFIED) &&
1046                              strchr(filter, p->status)))
1047                                 found++;
1048                 }
1049                 if (found)
1050                         return;
1051
1052                 /* otherwise we will clear the whole queue
1053                  * by copying the empty outq at the end of this
1054                  * function, but first clear the current entries
1055                  * in the queue.
1056                  */
1057                 for (i = 0; i < q->nr; i++)
1058                         diff_free_filepair(q->queue[i]);
1059         }
1060         else {
1061                 /* Only the matching ones */
1062                 for (i = 0; i < q->nr; i++) {
1063                         struct diff_filepair *p = q->queue[i];
1064
1065                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1066                              ((p->score &&
1067                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1068                               (!p->score &&
1069                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1070                             ((p->status != DIFF_STATUS_MODIFIED) &&
1071                              strchr(filter, p->status)))
1072                                 diff_q(&outq, p);
1073                         else
1074                                 diff_free_filepair(p);
1075                 }
1076         }
1077         free(q->queue);
1078         *q = outq;
1079 }
1080
1081 void diffcore_std(const char **paths,
1082                   int detect_rename, int rename_score,
1083                   const char *pickaxe, int pickaxe_opts,
1084                   int break_opt,
1085                   const char *orderfile,
1086                   const char *filter)
1087 {
1088         if (paths && paths[0])
1089                 diffcore_pathspec(paths);
1090         if (break_opt != -1)
1091                 diffcore_break(break_opt);
1092         if (detect_rename)
1093                 diffcore_rename(detect_rename, rename_score);
1094         if (break_opt != -1)
1095                 diffcore_merge_broken();
1096         if (pickaxe)
1097                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1098         if (orderfile)
1099                 diffcore_order(orderfile);
1100         diff_resolve_rename_copy();
1101         diffcore_apply_filter(filter);
1102 }
1103
1104
1105 void diffcore_std_no_resolve(const char **paths,
1106                              const char *pickaxe, int pickaxe_opts,
1107                              const char *orderfile,
1108                              const char *filter)
1109 {
1110         if (paths && paths[0])
1111                 diffcore_pathspec(paths);
1112         if (pickaxe)
1113                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1114         if (orderfile)
1115                 diffcore_order(orderfile);
1116         diffcore_apply_filter(filter);
1117 }
1118
1119 void diff_addremove(int addremove, unsigned mode,
1120                     const unsigned char *sha1,
1121                     const char *base, const char *path)
1122 {
1123         char concatpath[PATH_MAX];
1124         struct diff_filespec *one, *two;
1125
1126         /* This may look odd, but it is a preparation for
1127          * feeding "there are unchanged files which should
1128          * not produce diffs, but when you are doing copy
1129          * detection you would need them, so here they are"
1130          * entries to the diff-core.  They will be prefixed
1131          * with something like '=' or '*' (I haven't decided
1132          * which but should not make any difference).
1133          * Feeding the same new and old to diff_change()
1134          * also has the same effect.
1135          * Before the final output happens, they are pruned after
1136          * merged into rename/copy pairs as appropriate.
1137          */
1138         if (reverse_diff)
1139                 addremove = (addremove == '+' ? '-' :
1140                              addremove == '-' ? '+' : addremove);
1141
1142         if (!path) path = "";
1143         sprintf(concatpath, "%s%s", base, path);
1144         one = alloc_filespec(concatpath);
1145         two = alloc_filespec(concatpath);
1146
1147         if (addremove != '+')
1148                 fill_filespec(one, sha1, mode);
1149         if (addremove != '-')
1150                 fill_filespec(two, sha1, mode);
1151
1152         diff_queue(&diff_queued_diff, one, two);
1153 }
1154
1155 void diff_helper_input(unsigned old_mode,
1156                        unsigned new_mode,
1157                        const unsigned char *old_sha1,
1158                        const unsigned char *new_sha1,
1159                        const char *old_path,
1160                        int status,
1161                        int score,
1162                        const char *new_path)
1163 {
1164         struct diff_filespec *one, *two;
1165         struct diff_filepair *dp;
1166
1167         one = alloc_filespec(old_path);
1168         two = alloc_filespec(new_path);
1169         if (old_mode)
1170                 fill_filespec(one, old_sha1, old_mode);
1171         if (new_mode)
1172                 fill_filespec(two, new_sha1, new_mode);
1173         dp = diff_queue(&diff_queued_diff, one, two);
1174         dp->score = score * MAX_SCORE / 100;
1175         dp->status = status;
1176 }
1177
1178 void diff_change(unsigned old_mode, unsigned new_mode,
1179                  const unsigned char *old_sha1,
1180                  const unsigned char *new_sha1,
1181                  const char *base, const char *path)
1182 {
1183         char concatpath[PATH_MAX];
1184         struct diff_filespec *one, *two;
1185
1186         if (reverse_diff) {
1187                 unsigned tmp;
1188                 const unsigned char *tmp_c;
1189                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1190                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1191         }
1192         if (!path) path = "";
1193         sprintf(concatpath, "%s%s", base, path);
1194         one = alloc_filespec(concatpath);
1195         two = alloc_filespec(concatpath);
1196         fill_filespec(one, old_sha1, old_mode);
1197         fill_filespec(two, new_sha1, new_mode);
1198
1199         diff_queue(&diff_queued_diff, one, two);
1200 }
1201
1202 void diff_unmerge(const char *path)
1203 {
1204         struct diff_filespec *one, *two;
1205         one = alloc_filespec(path);
1206         two = alloc_filespec(path);
1207         diff_queue(&diff_queued_diff, one, two);
1208 }