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                 s->should_munmap = 1;
 381                 close(fd);
 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         strcpy(temp->tmp_path, ".diff_XXXXXX");
 423         fd = mkstemp(temp->tmp_path);
 424         if (fd < 0)
 425                 die("unable to create temp-file");
 426         if (write(fd, blob, size) != size)
 427                 die("unable to write temp-file");
 428         close(fd);
 429         temp->name = temp->tmp_path;
 430         strcpy(temp->hex, sha1_to_hex(sha1));
 431         temp->hex[40] = 0;
 432         sprintf(temp->mode, "%06o", mode);
 433 }
 434
 435 static void prepare_temp_file(const char *name,
 436                               struct diff_tempfile *temp,
 437                               struct diff_filespec *one)
 438 {
 439         if (!DIFF_FILE_VALID(one)) {
 440         not_a_valid_file:
 441                 /* A '-' entry produces this for file-2, and
 442                  * a '+' entry produces this for file-1.
 443                  */
 444                 temp->name = "/dev/null";
 445                 strcpy(temp->hex, ".");
 446                 strcpy(temp->mode, ".");
 447                 return;
 448         }
 449
 450         if (!one->sha1_valid ||
 451             work_tree_matches(name, one->sha1)) {
 452                 struct stat st;
 453                 if (lstat(name, &st) < 0) {
 454                         if (errno == ENOENT)
 455                                 goto not_a_valid_file;
 456                         die("stat(%s): %s", name, strerror(errno));
 457                 }
 458                 if (S_ISLNK(st.st_mode)) {
 459                         int ret;
 460                         char *buf, buf_[1024];
 461                         buf = ((sizeof(buf_) < st.st_size) ?
 462                                xmalloc(st.st_size) : buf_);
 463                         ret = readlink(name, buf, st.st_size);
 464                         if (ret < 0)
 465                                 die("readlink(%s)", name);
 466                         prep_temp_blob(temp, buf, st.st_size,
 467                                        (one->sha1_valid ?
 468                                         one->sha1 : null_sha1),
 469                                        (one->sha1_valid ?
 470                                         one->mode : S_IFLNK));
 471                 }
 472                 else {
 473                         /* we can borrow from the file in the work tree */
 474                         temp->name = name;
 475                         if (!one->sha1_valid)
 476                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 477                         else
 478                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 479                         /* Even though we may sometimes borrow the
 480                          * contents from the work tree, we always want
 481                          * one->mode.  mode is trustworthy even when
 482                          * !(one->sha1_valid), as long as
 483                          * DIFF_FILE_VALID(one).
 484                          */
 485                         sprintf(temp->mode, "%06o", one->mode);
 486                 }
 487                 return;
 488         }
 489         else {
 490                 if (diff_populate_filespec(one, 0))
 491                         die("cannot read data blob for %s", one->path);
 492                 prep_temp_blob(temp, one->data, one->size,
 493                                one->sha1, one->mode);
 494         }
 495 }
 496
 497 static void remove_tempfile(void)
 498 {
 499         int i;
 500
 501         for (i = 0; i < 2; i++)
 502                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 503                         unlink(diff_temp[i].name);
 504                         diff_temp[i].name = NULL;
 505                 }
 506 }
 507
 508 static void remove_tempfile_on_signal(int signo)
 509 {
 510         remove_tempfile();
 511 }
 512
 513 /* An external diff command takes:
 514  *
 515  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 516  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 517  *
 518  */
 519 static void run_external_diff(const char *pgm,
 520                               const char *name,
 521                               const char *other,
 522                               struct diff_filespec *one,
 523                               struct diff_filespec *two,
 524                               const char *xfrm_msg,
 525                               int complete_rewrite)
 526 {
 527         struct diff_tempfile *temp = diff_temp;
 528         pid_t pid;
 529         int status;
 530         static int atexit_asked = 0;
 531
 532         if (one && two) {
 533                 prepare_temp_file(name, &temp[0], one);
 534                 prepare_temp_file(other ? : name, &temp[1], two);
 535                 if (! atexit_asked &&
 536                     (temp[0].name == temp[0].tmp_path ||
 537                      temp[1].name == temp[1].tmp_path)) {
 538                         atexit_asked = 1;
 539                         atexit(remove_tempfile);
 540                 }
 541                 signal(SIGINT, remove_tempfile_on_signal);
 542         }
 543
 544         fflush(NULL);
 545         pid = fork();
 546         if (pid < 0)
 547                 die("unable to fork");
 548         if (!pid) {
 549                 if (pgm) {
 550                         if (one && two) {
 551                                 const char *exec_arg[10];
 552                                 const char **arg = &exec_arg[0];
 553                                 *arg++ = pgm;
 554                                 *arg++ = name;
 555                                 *arg++ = temp[0].name;
 556                                 *arg++ = temp[0].hex;
 557                                 *arg++ = temp[0].mode;
 558                                 *arg++ = temp[1].name;
 559                                 *arg++ = temp[1].hex;
 560                                 *arg++ = temp[1].mode;
 561                                 if (other) {
 562                                         *arg++ = other;
 563                                         *arg++ = xfrm_msg;
 564                                 }
 565                                 *arg = NULL;
 566                                 execvp(pgm, (char *const*) exec_arg);
 567                         }
 568                         else
 569                                 execlp(pgm, pgm, name, NULL);
 570                 }
 571                 /*
 572                  * otherwise we use the built-in one.
 573                  */
 574                 if (one && two)
 575                         builtin_diff(name, other ? : name, temp, xfrm_msg,
 576                                      complete_rewrite);
 577                 else
 578                         printf("* Unmerged path %s\n", name);
 579                 exit(0);
 580         }
 581         if (waitpid(pid, &status, 0) < 0 ||
 582             !WIFEXITED(status) || WEXITSTATUS(status)) {
 583                 /* Earlier we did not check the exit status because
 584                  * diff exits non-zero if files are different, and
 585                  * we are not interested in knowing that.  It was a
 586                  * mistake which made it harder to quit a diff-*
 587                  * session that uses the git-apply-patch-script as
 588                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 589                  * should also exit non-zero only when it wants to
 590                  * abort the entire diff-* session.
 591                  */
 592                 remove_tempfile();
 593                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 594                 exit(1);
 595         }
 596         remove_tempfile();
 597 }
 598
 599 static void run_diff(struct diff_filepair *p)
 600 {
 601         const char *pgm = external_diff();
 602         char msg_[PATH_MAX*2+200], *xfrm_msg;
 603         struct diff_filespec *one;
 604         struct diff_filespec *two;
 605         const char *name;
 606         const char *other;
 607         int complete_rewrite = 0;
 608
 609         if (DIFF_PAIR_UNMERGED(p)) {
 610                 /* unmerged */
 611                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 612                                   0);
 613                 return;
 614         }
 615
 616         name = p->one->path;
 617         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 618         one = p->one; two = p->two;
 619         switch (p->status) {
 620         case DIFF_STATUS_COPIED:
 621                 sprintf(msg_,
 622                         "similarity index %d%%\n"
 623                         "copy from %s\n"
 624                         "copy to %s",
 625                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 626                         name, other);
 627                 xfrm_msg = msg_;
 628                 break;
 629         case DIFF_STATUS_RENAMED:
 630                 sprintf(msg_,
 631                         "similarity index %d%%\n"
 632                         "rename from %s\n"
 633                         "rename to %s",
 634                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 635                         name, other);
 636                 xfrm_msg = msg_;
 637                 break;
 638         case DIFF_STATUS_MODIFIED:
 639                 if (p->score) {
 640                         sprintf(msg_,
 641                                 "dissimilarity index %d%%",
 642                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 643                         xfrm_msg = msg_;
 644                         complete_rewrite = 1;
 645                         break;
 646                 }
 647                 /* fallthru */
 648         default:
 649                 xfrm_msg = NULL;
 650         }
 651
 652         if (!pgm &&
 653             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 654             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 655                 /* a filepair that changes between file and symlink
 656                  * needs to be split into deletion and creation.
 657                  */
 658                 struct diff_filespec *null = alloc_filespec(two->path);
 659                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 660                 free(null);
 661                 null = alloc_filespec(one->path);
 662                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 663                 free(null);
 664         }
 665         else
 666                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 667                                   complete_rewrite);
 668 }
 669
 670 void diff_setup(int flags)
 671 {
 672         if (flags & DIFF_SETUP_REVERSE)
 673                 reverse_diff = 1;
 674         if (flags & DIFF_SETUP_USE_CACHE) {
 675                 if (!active_cache)
 676                         /* read-cache does not die even when it fails
 677                          * so it is safe for us to do this here.  Also
 678                          * it does not smudge active_cache or active_nr
 679                          * when it fails, so we do not have to worry about
 680                          * cleaning it up oufselves either.
 681                          */
 682                         read_cache();
 683         }
 684         if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 685                 use_size_cache = 1;
 686
 687 }
 688
 689 static int parse_num(const char **cp_p)
 690 {
 691         int num, scale, ch, cnt;
 692         const char *cp = *cp_p;
 693
 694         cnt = num = 0;
 695         scale = 1;
 696         while ('0' <= (ch = *cp) && ch <= '9') {
 697                 if (cnt++ < 5) {
 698                         /* We simply ignore more than 5 digits precision. */
 699                         scale *= 10;
 700                         num = num * 10 + ch - '0';
 701                 }
 702                 *cp++;
 703         }
 704         *cp_p = cp;
 705
 706         /* user says num divided by scale and we say internally that
 707          * is MAX_SCORE * num / scale.
 708          */
 709         return (MAX_SCORE * num / scale);
 710 }
 711
 712 int diff_scoreopt_parse(const char *opt)
 713 {
 714         int opt1, opt2, cmd;
 715
 716         if (*opt++ != '-')
 717                 return -1;
 718         cmd = *opt++;
 719         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 720                 return -1; /* that is not a -M, -C nor -B option */
 721
 722         opt1 = parse_num(&opt);
 723         if (cmd != 'B')
 724                 opt2 = 0;
 725         else {
 726                 if (*opt == 0)
 727                         opt2 = 0;
 728                 else if (*opt != '/')
 729                         return -1; /* we expect -B80/99 or -B80 */
 730                 else {
 731                         opt++;
 732                         opt2 = parse_num(&opt);
 733                 }
 734         }
 735         if (*opt != 0)
 736                 return -1;
 737         return opt1 | (opt2 << 16);
 738 }
 739
 740 struct diff_queue_struct diff_queued_diff;
 741
 742 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 743 {
 744         if (queue->alloc <= queue->nr) {
 745                 queue->alloc = alloc_nr(queue->alloc);
 746                 queue->queue = xrealloc(queue->queue,
 747                                         sizeof(dp) * queue->alloc);
 748         }
 749         queue->queue[queue->nr++] = dp;
 750 }
 751
 752 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 753                                  struct diff_filespec *one,
 754                                  struct diff_filespec *two)
 755 {
 756         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 757         dp->one = one;
 758         dp->two = two;
 759         dp->score = 0;
 760         dp->status = 0;
 761         dp->source_stays = 0;
 762         dp->broken_pair = 0;
 763         diff_q(queue, dp);
 764         return dp;
 765 }
 766
 767 void diff_free_filepair(struct diff_filepair *p)
 768 {
 769         diff_free_filespec_data(p->one);
 770         diff_free_filespec_data(p->two);
 771         free(p);
 772 }
 773
 774 static void diff_flush_raw(struct diff_filepair *p,
 775                            int line_termination,
 776                            int inter_name_termination)
 777 {
 778         int two_paths;
 779         char status[10];
 780
 781         if (line_termination) {
 782                 const char *err = "path %s cannot be expressed without -z";
 783                 if (strchr(p->one->path, line_termination) ||
 784                     strchr(p->one->path, inter_name_termination))
 785                         die(err, p->one->path);
 786                 if (strchr(p->two->path, line_termination) ||
 787                     strchr(p->two->path, inter_name_termination))
 788                         die(err, p->two->path);
 789         }
 790
 791         if (p->score)
 792                 sprintf(status, "%c%03d", p->status,
 793                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 794         else {
 795                 status[0] = p->status;
 796                 status[1] = 0;
 797         }
 798         switch (p->status) {
 799         case DIFF_STATUS_COPIED:
 800         case DIFF_STATUS_RENAMED:
 801                 two_paths = 1;
 802                 break;
 803         case DIFF_STATUS_ADDED:
 804         case DIFF_STATUS_DELETED:
 805                 two_paths = 0;
 806                 break;
 807         default:
 808                 two_paths = 0;
 809                 break;
 810         }
 811         printf(":%06o %06o %s ",
 812                p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 813         printf("%s %s%c%s",
 814                sha1_to_hex(p->two->sha1),
 815                status,
 816                inter_name_termination,
 817                p->one->path);
 818         if (two_paths)
 819                 printf("%c%s", inter_name_termination, p->two->path);
 820         putchar(line_termination);
 821 }
 822
 823 static void diff_flush_name(struct diff_filepair *p,
 824                             int line_termination)
 825 {
 826         printf("%s%c", p->two->path, line_termination);
 827 }
 828
 829 int diff_unmodified_pair(struct diff_filepair *p)
 830 {
 831         /* This function is written stricter than necessary to support
 832          * the currently implemented transformers, but the idea is to
 833          * let transformers to produce diff_filepairs any way they want,
 834          * and filter and clean them up here before producing the output.
 835          */
 836         struct diff_filespec *one, *two;
 837
 838         if (DIFF_PAIR_UNMERGED(p))
 839                 return 0; /* unmerged is interesting */
 840
 841         one = p->one;
 842         two = p->two;
 843
 844         /* deletion, addition, mode or type change
 845          * and rename are all interesting.
 846          */
 847         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 848             DIFF_PAIR_MODE_CHANGED(p) ||
 849             strcmp(one->path, two->path))
 850                 return 0;
 851
 852         /* both are valid and point at the same path.  that is, we are
 853          * dealing with a change.
 854          */
 855         if (one->sha1_valid && two->sha1_valid &&
 856             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 857                 return 1; /* no change */
 858         if (!one->sha1_valid && !two->sha1_valid)
 859                 return 1; /* both look at the same file on the filesystem. */
 860         return 0;
 861 }
 862
 863 static void diff_flush_patch(struct diff_filepair *p)
 864 {
 865         if (diff_unmodified_pair(p))
 866                 return;
 867
 868         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 869             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 870                 return; /* no tree diffs in patch format */
 871
 872         run_diff(p);
 873 }
 874
 875 int diff_queue_is_empty(void)
 876 {
 877         struct diff_queue_struct *q = &diff_queued_diff;
 878         int i;
 879         for (i = 0; i < q->nr; i++)
 880                 if (!diff_unmodified_pair(q->queue[i]))
 881                         return 0;
 882         return 1;
 883 }
 884
 885 #if DIFF_DEBUG
 886 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 887 {
 888         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 889                 x, one ? : "",
 890                 s->path,
 891                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 892                 s->mode,
 893                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 894         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 895                 x, one ? : "",
 896                 s->size, s->xfrm_flags);
 897 }
 898
 899 void diff_debug_filepair(const struct diff_filepair *p, int i)
 900 {
 901         diff_debug_filespec(p->one, i, "one");
 902         diff_debug_filespec(p->two, i, "two");
 903         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 904                 p->score, p->status ? : '?',
 905                 p->source_stays, p->broken_pair);
 906 }
 907
 908 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 909 {
 910         int i;
 911         if (msg)
 912                 fprintf(stderr, "%s\n", msg);
 913         fprintf(stderr, "q->nr = %d\n", q->nr);
 914         for (i = 0; i < q->nr; i++) {
 915                 struct diff_filepair *p = q->queue[i];
 916                 diff_debug_filepair(p, i);
 917         }
 918 }
 919 #endif
 920
 921 static void diff_resolve_rename_copy(void)
 922 {
 923         int i, j;
 924         struct diff_filepair *p, *pp;
 925         struct diff_queue_struct *q = &diff_queued_diff;
 926
 927         diff_debug_queue("resolve-rename-copy", q);
 928
 929         for (i = 0; i < q->nr; i++) {
 930                 p = q->queue[i];
 931                 p->status = 0; /* undecided */
 932                 if (DIFF_PAIR_UNMERGED(p))
 933                         p->status = DIFF_STATUS_UNMERGED;
 934                 else if (!DIFF_FILE_VALID(p->one))
 935                         p->status = DIFF_STATUS_ADDED;
 936                 else if (!DIFF_FILE_VALID(p->two))
 937                         p->status = DIFF_STATUS_DELETED;
 938                 else if (DIFF_PAIR_TYPE_CHANGED(p))
 939                         p->status = DIFF_STATUS_TYPE_CHANGED;
 940
 941                 /* from this point on, we are dealing with a pair
 942                  * whose both sides are valid and of the same type, i.e.
 943                  * either in-place edit or rename/copy edit.
 944                  */
 945                 else if (DIFF_PAIR_RENAME(p)) {
 946                         if (p->source_stays) {
 947                                 p->status = DIFF_STATUS_COPIED;
 948                                 continue;
 949                         }
 950                         /* See if there is some other filepair that
 951                          * copies from the same source as us.  If so
 952                          * we are a copy.  Otherwise we are a rename.
 953                          */
 954                         for (j = i + 1; j < q->nr; j++) {
 955                                 pp = q->queue[j];
 956                                 if (strcmp(pp->one->path, p->one->path))
 957                                         continue; /* not us */
 958                                 if (!DIFF_PAIR_RENAME(pp))
 959                                         continue; /* not a rename/copy */
 960                                 /* pp is a rename/copy from the same source */
 961                                 p->status = DIFF_STATUS_COPIED;
 962                                 break;
 963                         }
 964                         if (!p->status)
 965                                 p->status = DIFF_STATUS_RENAMED;
 966                 }
 967                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 968                          p->one->mode != p->two->mode)
 969                         p->status = DIFF_STATUS_MODIFIED;
 970                 else {
 971                         /* This is a "no-change" entry and should not
 972                          * happen anymore, but prepare for broken callers.
 973                          */
 974                         error("feeding unmodified %s to diffcore",
 975                               p->one->path);
 976                         p->status = DIFF_STATUS_UNKNOWN;
 977                 }
 978         }
 979         diff_debug_queue("resolve-rename-copy done", q);
 980 }
 981
 982 void diff_flush(int diff_output_style, int line_termination)
 983 {
 984         struct diff_queue_struct *q = &diff_queued_diff;
 985         int i;
 986         int inter_name_termination = '\t';
 987
 988         if (!line_termination)
 989                 inter_name_termination = 0;
 990
 991         for (i = 0; i < q->nr; i++) {
 992                 struct diff_filepair *p = q->queue[i];
 993                 if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
 994                     (p->status == DIFF_STATUS_UNKNOWN))
 995                         continue;
 996                 if (p->status == 0)
 997                         die("internal error in diff-resolve-rename-copy");
 998                 switch (diff_output_style) {
 999                 case DIFF_FORMAT_PATCH:
1000                         diff_flush_patch(p);
1001                         break;
1002                 case DIFF_FORMAT_RAW:
1003                         diff_flush_raw(p, line_termination,
1004                                        inter_name_termination);
1005                         break;
1006                 case DIFF_FORMAT_NAME:
1007                         diff_flush_name(p, line_termination);
1008                         break;
1009                 }
1010         }
1011         for (i = 0; i < q->nr; i++)
1012                 diff_free_filepair(q->queue[i]);
1013         free(q->queue);
1014         q->queue = NULL;
1015         q->nr = q->alloc = 0;
1016 }
1017
1018 static void diffcore_apply_filter(const char *filter)
1019 {
1020         int i;
1021         struct diff_queue_struct *q = &diff_queued_diff;
1022         struct diff_queue_struct outq;
1023         outq.queue = NULL;
1024         outq.nr = outq.alloc = 0;
1025
1026         if (!filter)
1027                 return;
1028
1029         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1030                 int found;
1031                 for (i = found = 0; !found && i < q->nr; i++) {
1032                         struct diff_filepair *p = q->queue[i];
1033                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1034                              ((p->score &&
1035                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1036                               (!p->score &&
1037                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1038                             ((p->status != DIFF_STATUS_MODIFIED) &&
1039                              strchr(filter, p->status)))
1040                                 found++;
1041                 }
1042                 if (found)
1043                         return;
1044
1045                 /* otherwise we will clear the whole queue
1046                  * by copying the empty outq at the end of this
1047                  * function, but first clear the current entries
1048                  * in the queue.
1049                  */
1050                 for (i = 0; i < q->nr; i++)
1051                         diff_free_filepair(q->queue[i]);
1052         }
1053         else {
1054                 /* Only the matching ones */
1055                 for (i = 0; i < q->nr; i++) {
1056                         struct diff_filepair *p = q->queue[i];
1057
1058                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1059                              ((p->score &&
1060                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1061                               (!p->score &&
1062                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1063                             ((p->status != DIFF_STATUS_MODIFIED) &&
1064                              strchr(filter, p->status)))
1065                                 diff_q(&outq, p);
1066                         else
1067                                 diff_free_filepair(p);
1068                 }
1069         }
1070         free(q->queue);
1071         *q = outq;
1072 }
1073
1074 void diffcore_std(const char **paths,
1075                   int detect_rename, int rename_score,
1076                   const char *pickaxe, int pickaxe_opts,
1077                   int break_opt,
1078                   const char *orderfile,
1079                   const char *filter)
1080 {
1081         if (paths && paths[0])
1082                 diffcore_pathspec(paths);
1083         if (break_opt != -1)
1084                 diffcore_break(break_opt);
1085         if (detect_rename)
1086                 diffcore_rename(detect_rename, rename_score);
1087         if (break_opt != -1)
1088                 diffcore_merge_broken();
1089         if (pickaxe)
1090                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1091         if (orderfile)
1092                 diffcore_order(orderfile);
1093         diff_resolve_rename_copy();
1094         diffcore_apply_filter(filter);
1095 }
1096
1097
1098 void diffcore_std_no_resolve(const char **paths,
1099                              const char *pickaxe, int pickaxe_opts,
1100                              const char *orderfile,
1101                              const char *filter)
1102 {
1103         if (paths && paths[0])
1104                 diffcore_pathspec(paths);
1105         if (pickaxe)
1106                 diffcore_pickaxe(pickaxe, pickaxe_opts);
1107         if (orderfile)
1108                 diffcore_order(orderfile);
1109         diffcore_apply_filter(filter);
1110 }
1111
1112 void diff_addremove(int addremove, unsigned mode,
1113                     const unsigned char *sha1,
1114                     const char *base, const char *path)
1115 {
1116         char concatpath[PATH_MAX];
1117         struct diff_filespec *one, *two;
1118
1119         /* This may look odd, but it is a preparation for
1120          * feeding "there are unchanged files which should
1121          * not produce diffs, but when you are doing copy
1122          * detection you would need them, so here they are"
1123          * entries to the diff-core.  They will be prefixed
1124          * with something like '=' or '*' (I haven't decided
1125          * which but should not make any difference).
1126          * Feeding the same new and old to diff_change()
1127          * also has the same effect.
1128          * Before the final output happens, they are pruned after
1129          * merged into rename/copy pairs as appropriate.
1130          */
1131         if (reverse_diff)
1132                 addremove = (addremove == '+' ? '-' :
1133                              addremove == '-' ? '+' : addremove);
1134
1135         if (!path) path = "";
1136         sprintf(concatpath, "%s%s", base, path);
1137         one = alloc_filespec(concatpath);
1138         two = alloc_filespec(concatpath);
1139
1140         if (addremove != '+')
1141                 fill_filespec(one, sha1, mode);
1142         if (addremove != '-')
1143                 fill_filespec(two, sha1, mode);
1144
1145         diff_queue(&diff_queued_diff, one, two);
1146 }
1147
1148 void diff_helper_input(unsigned old_mode,
1149                        unsigned new_mode,
1150                        const unsigned char *old_sha1,
1151                        const unsigned char *new_sha1,
1152                        const char *old_path,
1153                        int status,
1154                        int score,
1155                        const char *new_path)
1156 {
1157         struct diff_filespec *one, *two;
1158         struct diff_filepair *dp;
1159
1160         one = alloc_filespec(old_path);
1161         two = alloc_filespec(new_path);
1162         if (old_mode)
1163                 fill_filespec(one, old_sha1, old_mode);
1164         if (new_mode)
1165                 fill_filespec(two, new_sha1, new_mode);
1166         dp = diff_queue(&diff_queued_diff, one, two);
1167         dp->score = score * MAX_SCORE / 100;
1168         dp->status = status;
1169 }
1170
1171 void diff_change(unsigned old_mode, unsigned new_mode,
1172                  const unsigned char *old_sha1,
1173                  const unsigned char *new_sha1,
1174                  const char *base, const char *path)
1175 {
1176         char concatpath[PATH_MAX];
1177         struct diff_filespec *one, *two;
1178
1179         if (reverse_diff) {
1180                 unsigned tmp;
1181                 const unsigned char *tmp_c;
1182                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1183                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1184         }
1185         if (!path) path = "";
1186         sprintf(concatpath, "%s%s", base, path);
1187         one = alloc_filespec(concatpath);
1188         two = alloc_filespec(concatpath);
1189         fill_filespec(one, old_sha1, old_mode);
1190         fill_filespec(two, new_sha1, new_mode);
1191
1192         diff_queue(&diff_queued_diff, one, two);
1193 }
1194
1195 void diff_unmerge(const char *path)
1196 {
1197         struct diff_filespec *one, *two;
1198         one = alloc_filespec(path);
1199         two = alloc_filespec(path);
1200         diff_queue(&diff_queued_diff, one, two);
1201 }