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