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