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