apply.c

   1 /*
   2  * apply.c
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  *
   6  * This applies patches on top of some (arbitrary) version of the SCM.
   7  *
   8  */
   9 #include <fnmatch.h>
  10 #include "cache.h"
  11 #include "quote.h"
  12 #include "blob.h"
  13 #include "delta.h"
  14
  15 //  --check turns on checking that the working tree matches the
  16 //    files that are being modified, but doesn't apply the patch
  17 //  --stat does just a diffstat, and doesn't actually apply
  18 //  --numstat does numeric diffstat, and doesn't actually apply
  19 //  --index-info shows the old and new index info for paths if available.
  20 //
  21 static const char *prefix;
  22 static int prefix_length = -1;
  23
  24 static int p_value = 1;
  25 static int allow_binary_replacement = 0;
  26 static int check_index = 0;
  27 static int write_index = 0;
  28 static int diffstat = 0;
  29 static int numstat = 0;
  30 static int summary = 0;
  31 static int check = 0;
  32 static int apply = 1;
  33 static int no_add = 0;
  34 static int show_index_info = 0;
  35 static int line_termination = '\n';
  36 static unsigned long p_context = -1;
  37 static const char apply_usage[] =
  38 "git-apply [--stat] [--numstat] [--summary] [--check] [--index] [--apply] [--no-add] [--index-info] [--allow-binary-replacement] [-z] [-pNUM] [-CNUM] [--whitespace=<nowarn|warn|error|error-all|strip>] <patch>...";
  39
  40 static enum whitespace_eol {
  41         nowarn_whitespace,
  42         warn_on_whitespace,
  43         error_on_whitespace,
  44         strip_whitespace,
  45 } new_whitespace = warn_on_whitespace;
  46 static int whitespace_error = 0;
  47 static int squelch_whitespace_errors = 5;
  48 static int applied_after_stripping = 0;
  49 static const char *patch_input_file = NULL;
  50
  51 static void parse_whitespace_option(const char *option)
  52 {
  53         if (!option) {
  54                 new_whitespace = warn_on_whitespace;
  55                 return;
  56         }
  57         if (!strcmp(option, "warn")) {
  58                 new_whitespace = warn_on_whitespace;
  59                 return;
  60         }
  61         if (!strcmp(option, "nowarn")) {
  62                 new_whitespace = nowarn_whitespace;
  63                 return;
  64         }
  65         if (!strcmp(option, "error")) {
  66                 new_whitespace = error_on_whitespace;
  67                 return;
  68         }
  69         if (!strcmp(option, "error-all")) {
  70                 new_whitespace = error_on_whitespace;
  71                 squelch_whitespace_errors = 0;
  72                 return;
  73         }
  74         if (!strcmp(option, "strip")) {
  75                 new_whitespace = strip_whitespace;
  76                 return;
  77         }
  78         die("unrecognized whitespace option '%s'", option);
  79 }
  80
  81 static void set_default_whitespace_mode(const char *whitespace_option)
  82 {
  83         if (!whitespace_option && !apply_default_whitespace) {
  84                 new_whitespace = (apply
  85                                   ? warn_on_whitespace
  86                                   : nowarn_whitespace);
  87         }
  88 }
  89
  90 /*
  91  * For "diff-stat" like behaviour, we keep track of the biggest change
  92  * we've seen, and the longest filename. That allows us to do simple
  93  * scaling.
  94  */
  95 static int max_change, max_len;
  96
  97 /*
  98  * Various "current state", notably line numbers and what
  99  * file (and how) we're patching right now.. The "is_xxxx"
 100  * things are flags, where -1 means "don't know yet".
 101  */
 102 static int linenr = 1;
 103
 104 struct fragment {
 105         unsigned long leading, trailing;
 106         unsigned long oldpos, oldlines;
 107         unsigned long newpos, newlines;
 108         const char *patch;
 109         int size;
 110         struct fragment *next;
 111 };
 112
 113 struct patch {
 114         char *new_name, *old_name, *def_name;
 115         unsigned int old_mode, new_mode;
 116         int is_rename, is_copy, is_new, is_delete, is_binary;
 117         int lines_added, lines_deleted;
 118         int score;
 119         struct fragment *fragments;
 120         char *result;
 121         unsigned long resultsize;
 122         char old_sha1_prefix[41];
 123         char new_sha1_prefix[41];
 124         struct patch *next;
 125 };
 126
 127 #define CHUNKSIZE (8192)
 128 #define SLOP (16)
 129
 130 static void *read_patch_file(int fd, unsigned long *sizep)
 131 {
 132         unsigned long size = 0, alloc = CHUNKSIZE;
 133         void *buffer = xmalloc(alloc);
 134
 135         for (;;) {
 136                 int nr = alloc - size;
 137                 if (nr < 1024) {
 138                         alloc += CHUNKSIZE;
 139                         buffer = xrealloc(buffer, alloc);
 140                         nr = alloc - size;
 141                 }
 142                 nr = xread(fd, buffer + size, nr);
 143                 if (!nr)
 144                         break;
 145                 if (nr < 0)
 146                         die("git-apply: read returned %s", strerror(errno));
 147                 size += nr;
 148         }
 149         *sizep = size;
 150
 151         /*
 152          * Make sure that we have some slop in the buffer
 153          * so that we can do speculative "memcmp" etc, and
 154          * see to it that it is NUL-filled.
 155          */
 156         if (alloc < size + SLOP)
 157                 buffer = xrealloc(buffer, size + SLOP);
 158         memset(buffer + size, 0, SLOP);
 159         return buffer;
 160 }
 161
 162 static unsigned long linelen(const char *buffer, unsigned long size)
 163 {
 164         unsigned long len = 0;
 165         while (size--) {
 166                 len++;
 167                 if (*buffer++ == '\n')
 168                         break;
 169         }
 170         return len;
 171 }
 172
 173 static int is_dev_null(const char *str)
 174 {
 175         return !memcmp("/dev/null", str, 9) && isspace(str[9]);
 176 }
 177
 178 #define TERM_SPACE      1
 179 #define TERM_TAB        2
 180
 181 static int name_terminate(const char *name, int namelen, int c, int terminate)
 182 {
 183         if (c == ' ' && !(terminate & TERM_SPACE))
 184                 return 0;
 185         if (c == '\t' && !(terminate & TERM_TAB))
 186                 return 0;
 187
 188         return 1;
 189 }
 190
 191 static char * find_name(const char *line, char *def, int p_value, int terminate)
 192 {
 193         int len;
 194         const char *start = line;
 195         char *name;
 196
 197         if (*line == '"') {
 198                 /* Proposed "new-style" GNU patch/diff format; see
 199                  * http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
 200                  */
 201                 name = unquote_c_style(line, NULL);
 202                 if (name) {
 203                         char *cp = name;
 204                         while (p_value) {
 205                                 cp = strchr(name, '/');
 206                                 if (!cp)
 207                                         break;
 208                                 cp++;
 209                                 p_value--;
 210                         }
 211                         if (cp) {
 212                                 /* name can later be freed, so we need
 213                                  * to memmove, not just return cp
 214                                  */
 215                                 memmove(name, cp, strlen(cp) + 1);
 216                                 free(def);
 217                                 return name;
 218                         }
 219                         else {
 220                                 free(name);
 221                                 name = NULL;
 222                         }
 223                 }
 224         }
 225
 226         for (;;) {
 227                 char c = *line;
 228
 229                 if (isspace(c)) {
 230                         if (c == '\n')
 231                                 break;
 232                         if (name_terminate(start, line-start, c, terminate))
 233                                 break;
 234                 }
 235                 line++;
 236                 if (c == '/' && !--p_value)
 237                         start = line;
 238         }
 239         if (!start)
 240                 return def;
 241         len = line - start;
 242         if (!len)
 243                 return def;
 244
 245         /*
 246          * Generally we prefer the shorter name, especially
 247          * if the other one is just a variation of that with
 248          * something else tacked on to the end (ie "file.orig"
 249          * or "file~").
 250          */
 251         if (def) {
 252                 int deflen = strlen(def);
 253                 if (deflen < len && !strncmp(start, def, deflen))
 254                         return def;
 255         }
 256
 257         name = xmalloc(len + 1);
 258         memcpy(name, start, len);
 259         name[len] = 0;
 260         free(def);
 261         return name;
 262 }
 263
 264 /*
 265  * Get the name etc info from the --/+++ lines of a traditional patch header
 266  *
 267  * NOTE! This hardcodes "-p1" behaviour in filename detection.
 268  *
 269  * FIXME! The end-of-filename heuristics are kind of screwy. For existing
 270  * files, we can happily check the index for a match, but for creating a
 271  * new file we should try to match whatever "patch" does. I have no idea.
 272  */
 273 static void parse_traditional_patch(const char *first, const char *second, struct patch *patch)
 274 {
 275         char *name;
 276
 277         first += 4;     // skip "--- "
 278         second += 4;    // skip "+++ "
 279         if (is_dev_null(first)) {
 280                 patch->is_new = 1;
 281                 patch->is_delete = 0;
 282                 name = find_name(second, NULL, p_value, TERM_SPACE | TERM_TAB);
 283                 patch->new_name = name;
 284         } else if (is_dev_null(second)) {
 285                 patch->is_new = 0;
 286                 patch->is_delete = 1;
 287                 name = find_name(first, NULL, p_value, TERM_SPACE | TERM_TAB);
 288                 patch->old_name = name;
 289         } else {
 290                 name = find_name(first, NULL, p_value, TERM_SPACE | TERM_TAB);
 291                 name = find_name(second, name, p_value, TERM_SPACE | TERM_TAB);
 292                 patch->old_name = patch->new_name = name;
 293         }
 294         if (!name)
 295                 die("unable to find filename in patch at line %d", linenr);
 296 }
 297
 298 static int gitdiff_hdrend(const char *line, struct patch *patch)
 299 {
 300         return -1;
 301 }
 302
 303 /*
 304  * We're anal about diff header consistency, to make
 305  * sure that we don't end up having strange ambiguous
 306  * patches floating around.
 307  *
 308  * As a result, gitdiff_{old|new}name() will check
 309  * their names against any previous information, just
 310  * to make sure..
 311  */
 312 static char *gitdiff_verify_name(const char *line, int isnull, char *orig_name, const char *oldnew)
 313 {
 314         if (!orig_name && !isnull)
 315                 return find_name(line, NULL, 1, 0);
 316
 317         if (orig_name) {
 318                 int len;
 319                 const char *name;
 320                 char *another;
 321                 name = orig_name;
 322                 len = strlen(name);
 323                 if (isnull)
 324                         die("git-apply: bad git-diff - expected /dev/null, got %s on line %d", name, linenr);
 325                 another = find_name(line, NULL, 1, 0);
 326                 if (!another || memcmp(another, name, len))
 327                         die("git-apply: bad git-diff - inconsistent %s filename on line %d", oldnew, linenr);
 328                 free(another);
 329                 return orig_name;
 330         }
 331         else {
 332                 /* expect "/dev/null" */
 333                 if (memcmp("/dev/null", line, 9) || line[9] != '\n')
 334                         die("git-apply: bad git-diff - expected /dev/null on line %d", linenr);
 335                 return NULL;
 336         }
 337 }
 338
 339 static int gitdiff_oldname(const char *line, struct patch *patch)
 340 {
 341         patch->old_name = gitdiff_verify_name(line, patch->is_new, patch->old_name, "old");
 342         return 0;
 343 }
 344
 345 static int gitdiff_newname(const char *line, struct patch *patch)
 346 {
 347         patch->new_name = gitdiff_verify_name(line, patch->is_delete, patch->new_name, "new");
 348         return 0;
 349 }
 350
 351 static int gitdiff_oldmode(const char *line, struct patch *patch)
 352 {
 353         patch->old_mode = strtoul(line, NULL, 8);
 354         return 0;
 355 }
 356
 357 static int gitdiff_newmode(const char *line, struct patch *patch)
 358 {
 359         patch->new_mode = strtoul(line, NULL, 8);
 360         return 0;
 361 }
 362
 363 static int gitdiff_delete(const char *line, struct patch *patch)
 364 {
 365         patch->is_delete = 1;
 366         patch->old_name = patch->def_name;
 367         return gitdiff_oldmode(line, patch);
 368 }
 369
 370 static int gitdiff_newfile(const char *line, struct patch *patch)
 371 {
 372         patch->is_new = 1;
 373         patch->new_name = patch->def_name;
 374         return gitdiff_newmode(line, patch);
 375 }
 376
 377 static int gitdiff_copysrc(const char *line, struct patch *patch)
 378 {
 379         patch->is_copy = 1;
 380         patch->old_name = find_name(line, NULL, 0, 0);
 381         return 0;
 382 }
 383
 384 static int gitdiff_copydst(const char *line, struct patch *patch)
 385 {
 386         patch->is_copy = 1;
 387         patch->new_name = find_name(line, NULL, 0, 0);
 388         return 0;
 389 }
 390
 391 static int gitdiff_renamesrc(const char *line, struct patch *patch)
 392 {
 393         patch->is_rename = 1;
 394         patch->old_name = find_name(line, NULL, 0, 0);
 395         return 0;
 396 }
 397
 398 static int gitdiff_renamedst(const char *line, struct patch *patch)
 399 {
 400         patch->is_rename = 1;
 401         patch->new_name = find_name(line, NULL, 0, 0);
 402         return 0;
 403 }
 404
 405 static int gitdiff_similarity(const char *line, struct patch *patch)
 406 {
 407         if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
 408                 patch->score = 0;
 409         return 0;
 410 }
 411
 412 static int gitdiff_dissimilarity(const char *line, struct patch *patch)
 413 {
 414         if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
 415                 patch->score = 0;
 416         return 0;
 417 }
 418
 419 static int gitdiff_index(const char *line, struct patch *patch)
 420 {
 421         /* index line is N hexadecimal, "..", N hexadecimal,
 422          * and optional space with octal mode.
 423          */
 424         const char *ptr, *eol;
 425         int len;
 426
 427         ptr = strchr(line, '.');
 428         if (!ptr || ptr[1] != '.' || 40 < ptr - line)
 429                 return 0;
 430         len = ptr - line;
 431         memcpy(patch->old_sha1_prefix, line, len);
 432         patch->old_sha1_prefix[len] = 0;
 433
 434         line = ptr + 2;
 435         ptr = strchr(line, ' ');
 436         eol = strchr(line, '\n');
 437
 438         if (!ptr || eol < ptr)
 439                 ptr = eol;
 440         len = ptr - line;
 441
 442         if (40 < len)
 443                 return 0;
 444         memcpy(patch->new_sha1_prefix, line, len);
 445         patch->new_sha1_prefix[len] = 0;
 446         if (*ptr == ' ')
 447                 patch->new_mode = patch->old_mode = strtoul(ptr+1, NULL, 8);
 448         return 0;
 449 }
 450
 451 /*
 452  * This is normal for a diff that doesn't change anything: we'll fall through
 453  * into the next diff. Tell the parser to break out.
 454  */
 455 static int gitdiff_unrecognized(const char *line, struct patch *patch)
 456 {
 457         return -1;
 458 }
 459
 460 static const char *stop_at_slash(const char *line, int llen)
 461 {
 462         int i;
 463
 464         for (i = 0; i < llen; i++) {
 465                 int ch = line[i];
 466                 if (ch == '/')
 467                         return line + i;
 468         }
 469         return NULL;
 470 }
 471
 472 /* This is to extract the same name that appears on "diff --git"
 473  * line.  We do not find and return anything if it is a rename
 474  * patch, and it is OK because we will find the name elsewhere.
 475  * We need to reliably find name only when it is mode-change only,
 476  * creation or deletion of an empty file.  In any of these cases,
 477  * both sides are the same name under a/ and b/ respectively.
 478  */
 479 static char *git_header_name(char *line, int llen)
 480 {
 481         int len;
 482         const char *name;
 483         const char *second = NULL;
 484
 485         line += strlen("diff --git ");
 486         llen -= strlen("diff --git ");
 487
 488         if (*line == '"') {
 489                 const char *cp;
 490                 char *first = unquote_c_style(line, &second);
 491                 if (!first)
 492                         return NULL;
 493
 494                 /* advance to the first slash */
 495                 cp = stop_at_slash(first, strlen(first));
 496                 if (!cp || cp == first) {
 497                         /* we do not accept absolute paths */
 498                 free_first_and_fail:
 499                         free(first);
 500                         return NULL;
 501                 }
 502                 len = strlen(cp+1);
 503                 memmove(first, cp+1, len+1); /* including NUL */
 504
 505                 /* second points at one past closing dq of name.
 506                  * find the second name.
 507                  */
 508                 while ((second < line + llen) && isspace(*second))
 509                         second++;
 510
 511                 if (line + llen <= second)
 512                         goto free_first_and_fail;
 513                 if (*second == '"') {
 514                         char *sp = unquote_c_style(second, NULL);
 515                         if (!sp)
 516                                 goto free_first_and_fail;
 517                         cp = stop_at_slash(sp, strlen(sp));
 518                         if (!cp || cp == sp) {
 519                         free_both_and_fail:
 520                                 free(sp);
 521                                 goto free_first_and_fail;
 522                         }
 523                         /* They must match, otherwise ignore */
 524                         if (strcmp(cp+1, first))
 525                                 goto free_both_and_fail;
 526                         free(sp);
 527                         return first;
 528                 }
 529
 530                 /* unquoted second */
 531                 cp = stop_at_slash(second, line + llen - second);
 532                 if (!cp || cp == second)
 533                         goto free_first_and_fail;
 534                 cp++;
 535                 if (line + llen - cp != len + 1 ||
 536                     memcmp(first, cp, len))
 537                         goto free_first_and_fail;
 538                 return first;
 539         }
 540
 541         /* unquoted first name */
 542         name = stop_at_slash(line, llen);
 543         if (!name || name == line)
 544                 return NULL;
 545
 546         name++;
 547
 548         /* since the first name is unquoted, a dq if exists must be
 549          * the beginning of the second name.
 550          */
 551         for (second = name; second < line + llen; second++) {
 552                 if (*second == '"') {
 553                         const char *cp = second;
 554                         const char *np;
 555                         char *sp = unquote_c_style(second, NULL);
 556
 557                         if (!sp)
 558                                 return NULL;
 559                         np = stop_at_slash(sp, strlen(sp));
 560                         if (!np || np == sp) {
 561                         free_second_and_fail:
 562                                 free(sp);
 563                                 return NULL;
 564                         }
 565                         np++;
 566                         len = strlen(np);
 567                         if (len < cp - name &&
 568                             !strncmp(np, name, len) &&
 569                             isspace(name[len])) {
 570                                 /* Good */
 571                                 memmove(sp, np, len + 1);
 572                                 return sp;
 573                         }
 574                         goto free_second_and_fail;
 575                 }
 576         }
 577
 578         /*
 579          * Accept a name only if it shows up twice, exactly the same
 580          * form.
 581          */
 582         for (len = 0 ; ; len++) {
 583                 char c = name[len];
 584
 585                 switch (c) {
 586                 default:
 587                         continue;
 588                 case '\n':
 589                         return NULL;
 590                 case '\t': case ' ':
 591                         second = name+len;
 592                         for (;;) {
 593                                 char c = *second++;
 594                                 if (c == '\n')
 595                                         return NULL;
 596                                 if (c == '/')
 597                                         break;
 598                         }
 599                         if (second[len] == '\n' && !memcmp(name, second, len)) {
 600                                 char *ret = xmalloc(len + 1);
 601                                 memcpy(ret, name, len);
 602                                 ret[len] = 0;
 603                                 return ret;
 604                         }
 605                 }
 606         }
 607         return NULL;
 608 }
 609
 610 /* Verify that we recognize the lines following a git header */
 611 static int parse_git_header(char *line, int len, unsigned int size, struct patch *patch)
 612 {
 613         unsigned long offset;
 614
 615         /* A git diff has explicit new/delete information, so we don't guess */
 616         patch->is_new = 0;
 617         patch->is_delete = 0;
 618
 619         /*
 620          * Some things may not have the old name in the
 621          * rest of the headers anywhere (pure mode changes,
 622          * or removing or adding empty files), so we get
 623          * the default name from the header.
 624          */
 625         patch->def_name = git_header_name(line, len);
 626
 627         line += len;
 628         size -= len;
 629         linenr++;
 630         for (offset = len ; size > 0 ; offset += len, size -= len, line += len, linenr++) {
 631                 static const struct opentry {
 632                         const char *str;
 633                         int (*fn)(const char *, struct patch *);
 634                 } optable[] = {
 635                         { "@@ -", gitdiff_hdrend },
 636                         { "--- ", gitdiff_oldname },
 637                         { "+++ ", gitdiff_newname },
 638                         { "old mode ", gitdiff_oldmode },
 639                         { "new mode ", gitdiff_newmode },
 640                         { "deleted file mode ", gitdiff_delete },
 641                         { "new file mode ", gitdiff_newfile },
 642                         { "copy from ", gitdiff_copysrc },
 643                         { "copy to ", gitdiff_copydst },
 644                         { "rename old ", gitdiff_renamesrc },
 645                         { "rename new ", gitdiff_renamedst },
 646                         { "rename from ", gitdiff_renamesrc },
 647                         { "rename to ", gitdiff_renamedst },
 648                         { "similarity index ", gitdiff_similarity },
 649                         { "dissimilarity index ", gitdiff_dissimilarity },
 650                         { "index ", gitdiff_index },
 651                         { "", gitdiff_unrecognized },
 652                 };
 653                 int i;
 654
 655                 len = linelen(line, size);
 656                 if (!len || line[len-1] != '\n')
 657                         break;
 658                 for (i = 0; i < ARRAY_SIZE(optable); i++) {
 659                         const struct opentry *p = optable + i;
 660                         int oplen = strlen(p->str);
 661                         if (len < oplen || memcmp(p->str, line, oplen))
 662                                 continue;
 663                         if (p->fn(line + oplen, patch) < 0)
 664                                 return offset;
 665                         break;
 666                 }
 667         }
 668
 669         return offset;
 670 }
 671
 672 static int parse_num(const char *line, unsigned long *p)
 673 {
 674         char *ptr;
 675
 676         if (!isdigit(*line))
 677                 return 0;
 678         *p = strtoul(line, &ptr, 10);
 679         return ptr - line;
 680 }
 681
 682 static int parse_range(const char *line, int len, int offset, const char *expect,
 683                         unsigned long *p1, unsigned long *p2)
 684 {
 685         int digits, ex;
 686
 687         if (offset < 0 || offset >= len)
 688                 return -1;
 689         line += offset;
 690         len -= offset;
 691
 692         digits = parse_num(line, p1);
 693         if (!digits)
 694                 return -1;
 695
 696         offset += digits;
 697         line += digits;
 698         len -= digits;
 699
 700         *p2 = 1;
 701         if (*line == ',') {
 702                 digits = parse_num(line+1, p2);
 703                 if (!digits)
 704                         return -1;
 705
 706                 offset += digits+1;
 707                 line += digits+1;
 708                 len -= digits+1;
 709         }
 710
 711         ex = strlen(expect);
 712         if (ex > len)
 713                 return -1;
 714         if (memcmp(line, expect, ex))
 715                 return -1;
 716
 717         return offset + ex;
 718 }
 719
 720 /*
 721  * Parse a unified diff fragment header of the
 722  * form "@@ -a,b +c,d @@"
 723  */
 724 static int parse_fragment_header(char *line, int len, struct fragment *fragment)
 725 {
 726         int offset;
 727
 728         if (!len || line[len-1] != '\n')
 729                 return -1;
 730
 731         /* Figure out the number of lines in a fragment */
 732         offset = parse_range(line, len, 4, " +", &fragment->oldpos, &fragment->oldlines);
 733         offset = parse_range(line, len, offset, " @@", &fragment->newpos, &fragment->newlines);
 734
 735         return offset;
 736 }
 737
 738 static int find_header(char *line, unsigned long size, int *hdrsize, struct patch *patch)
 739 {
 740         unsigned long offset, len;
 741
 742         patch->is_rename = patch->is_copy = 0;
 743         patch->is_new = patch->is_delete = -1;
 744         patch->old_mode = patch->new_mode = 0;
 745         patch->old_name = patch->new_name = NULL;
 746         for (offset = 0; size > 0; offset += len, size -= len, line += len, linenr++) {
 747                 unsigned long nextlen;
 748
 749                 len = linelen(line, size);
 750                 if (!len)
 751                         break;
 752
 753                 /* Testing this early allows us to take a few shortcuts.. */
 754                 if (len < 6)
 755                         continue;
 756
 757                 /*
 758                  * Make sure we don't find any unconnected patch fragmants.
 759                  * That's a sign that we didn't find a header, and that a
 760                  * patch has become corrupted/broken up.
 761                  */
 762                 if (!memcmp("@@ -", line, 4)) {
 763                         struct fragment dummy;
 764                         if (parse_fragment_header(line, len, &dummy) < 0)
 765                                 continue;
 766                         error("patch fragment without header at line %d: %.*s", linenr, (int)len-1, line);
 767                 }
 768
 769                 if (size < len + 6)
 770                         break;
 771
 772                 /*
 773                  * Git patch? It might not have a real patch, just a rename
 774                  * or mode change, so we handle that specially
 775                  */
 776                 if (!memcmp("diff --git ", line, 11)) {
 777                         int git_hdr_len = parse_git_header(line, len, size, patch);
 778                         if (git_hdr_len <= len)
 779                                 continue;
 780                         if (!patch->old_name && !patch->new_name) {
 781                                 if (!patch->def_name)
 782                                         die("git diff header lacks filename information (line %d)", linenr);
 783                                 patch->old_name = patch->new_name = patch->def_name;
 784                         }
 785                         *hdrsize = git_hdr_len;
 786                         return offset;
 787                 }
 788
 789                 /** --- followed by +++ ? */
 790                 if (memcmp("--- ", line,  4) || memcmp("+++ ", line + len, 4))
 791                         continue;
 792
 793                 /*
 794                  * We only accept unified patches, so we want it to
 795                  * at least have "@@ -a,b +c,d @@\n", which is 14 chars
 796                  * minimum
 797                  */
 798                 nextlen = linelen(line + len, size - len);
 799                 if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
 800                         continue;
 801
 802                 /* Ok, we'll consider it a patch */
 803                 parse_traditional_patch(line, line+len, patch);
 804                 *hdrsize = len + nextlen;
 805                 linenr += 2;
 806                 return offset;
 807         }
 808         return -1;
 809 }
 810
 811 /*
 812  * Parse a unified diff. Note that this really needs
 813  * to parse each fragment separately, since the only
 814  * way to know the difference between a "---" that is
 815  * part of a patch, and a "---" that starts the next
 816  * patch is to look at the line counts..
 817  */
 818 static int parse_fragment(char *line, unsigned long size, struct patch *patch, struct fragment *fragment)
 819 {
 820         int added, deleted;
 821         int len = linelen(line, size), offset;
 822         unsigned long oldlines, newlines;
 823         unsigned long leading, trailing;
 824
 825         offset = parse_fragment_header(line, len, fragment);
 826         if (offset < 0)
 827                 return -1;
 828         oldlines = fragment->oldlines;
 829         newlines = fragment->newlines;
 830         leading = 0;
 831         trailing = 0;
 832
 833         if (patch->is_new < 0) {
 834                 patch->is_new =  !oldlines;
 835                 if (!oldlines)
 836                         patch->old_name = NULL;
 837         }
 838         if (patch->is_delete < 0) {
 839                 patch->is_delete = !newlines;
 840                 if (!newlines)
 841                         patch->new_name = NULL;
 842         }
 843
 844         if (patch->is_new && oldlines)
 845                 return error("new file depends on old contents");
 846         if (patch->is_delete != !newlines) {
 847                 if (newlines)
 848                         return error("deleted file still has contents");
 849                 fprintf(stderr, "** warning: file %s becomes empty but is not deleted\n", patch->new_name);
 850         }
 851
 852         /* Parse the thing.. */
 853         line += len;
 854         size -= len;
 855         linenr++;
 856         added = deleted = 0;
 857         for (offset = len; size > 0; offset += len, size -= len, line += len, linenr++) {
 858                 if (!oldlines && !newlines)
 859                         break;
 860                 len = linelen(line, size);
 861                 if (!len || line[len-1] != '\n')
 862                         return -1;
 863                 switch (*line) {
 864                 default:
 865                         return -1;
 866                 case ' ':
 867                         oldlines--;
 868                         newlines--;
 869                         if (!deleted && !added)
 870                                 leading++;
 871                         trailing++;
 872                         break;
 873                 case '-':
 874                         deleted++;
 875                         oldlines--;
 876                         trailing = 0;
 877                         break;
 878                 case '+':
 879                         /*
 880                          * We know len is at least two, since we have a '+' and
 881                          * we checked that the last character was a '\n' above.
 882                          * That is, an addition of an empty line would check
 883                          * the '+' here.  Sneaky...
 884                          */
 885                         if ((new_whitespace != nowarn_whitespace) &&
 886                             isspace(line[len-2])) {
 887                                 whitespace_error++;
 888                                 if (squelch_whitespace_errors &&
 889                                     squelch_whitespace_errors <
 890                                     whitespace_error)
 891                                         ;
 892                                 else {
 893                                         fprintf(stderr, "Adds trailing whitespace.\n%s:%d:%.*s\n",
 894                                                 patch_input_file,
 895                                                 linenr, len-2, line+1);
 896                                 }
 897                         }
 898                         added++;
 899                         newlines--;
 900                         trailing = 0;
 901                         break;
 902
 903                 /* We allow "\ No newline at end of file". Depending
 904                  * on locale settings when the patch was produced we
 905                  * don't know what this line looks like. The only
 906                  * thing we do know is that it begins with "\ ".
 907                  * Checking for 12 is just for sanity check -- any
 908                  * l10n of "\ No newline..." is at least that long.
 909                  */
 910                 case '\\':
 911                         if (len < 12 || memcmp(line, "\\ ", 2))
 912                                 return -1;
 913                         break;
 914                 }
 915         }
 916         if (oldlines || newlines)
 917                 return -1;
 918         fragment->leading = leading;
 919         fragment->trailing = trailing;
 920
 921         /* If a fragment ends with an incomplete line, we failed to include
 922          * it in the above loop because we hit oldlines == newlines == 0
 923          * before seeing it.
 924          */
 925         if (12 < size && !memcmp(line, "\\ ", 2))
 926                 offset += linelen(line, size);
 927
 928         patch->lines_added += added;
 929         patch->lines_deleted += deleted;
 930         return offset;
 931 }
 932
 933 static int parse_single_patch(char *line, unsigned long size, struct patch *patch)
 934 {
 935         unsigned long offset = 0;
 936         struct fragment **fragp = &patch->fragments;
 937
 938         while (size > 4 && !memcmp(line, "@@ -", 4)) {
 939                 struct fragment *fragment;
 940                 int len;
 941
 942                 fragment = xcalloc(1, sizeof(*fragment));
 943                 len = parse_fragment(line, size, patch, fragment);
 944                 if (len <= 0)
 945                         die("corrupt patch at line %d", linenr);
 946
 947                 fragment->patch = line;
 948                 fragment->size = len;
 949
 950                 *fragp = fragment;
 951                 fragp = &fragment->next;
 952
 953                 offset += len;
 954                 line += len;
 955                 size -= len;
 956         }
 957         return offset;
 958 }
 959
 960 static inline int metadata_changes(struct patch *patch)
 961 {
 962         return  patch->is_rename > 0 ||
 963                 patch->is_copy > 0 ||
 964                 patch->is_new > 0 ||
 965                 patch->is_delete ||
 966                 (patch->old_mode && patch->new_mode &&
 967                  patch->old_mode != patch->new_mode);
 968 }
 969
 970 static int parse_binary(char *buffer, unsigned long size, struct patch *patch)
 971 {
 972         /* We have read "GIT binary patch\n"; what follows is a
 973          * sequence of 'length-byte' followed by base-85 encoded
 974          * delta data.
 975          *
 976          * Each 5-byte sequence of base-85 encodes up to 4 bytes,
 977          * and we would limit the patch line to 66 characters,
 978          * so one line can fit up to 13 groups that would decode
 979          * to 52 bytes max.  The length byte 'A'-'Z' corresponds
 980          * to 1-26 bytes, and 'a'-'z' corresponds to 27-52 bytes.
 981          * The end of binary is signalled with an empty line.
 982          */
 983         int llen, used;
 984         struct fragment *fragment;
 985         char *delta = NULL;
 986
 987         patch->is_binary = 1;
 988         patch->fragments = fragment = xcalloc(1, sizeof(*fragment));
 989         used = 0;
 990         while (1) {
 991                 int byte_length, max_byte_length, newsize;
 992                 llen = linelen(buffer, size);
 993                 used += llen;
 994                 linenr++;
 995                 if (llen == 1)
 996                         break;
 997                 /* Minimum line is "A00000\n" which is 7-byte long,
 998                  * and the line length must be multiple of 5 plus 2.
 999                  */
1000                 if ((llen < 7) || (llen-2) % 5)
1001                         goto corrupt;
1002                 max_byte_length = (llen - 2) / 5 * 4;
1003                 byte_length = *buffer;
1004                 if ('A' <= byte_length && byte_length <= 'Z')
1005                         byte_length = byte_length - 'A' + 1;
1006                 else if ('a' <= byte_length && byte_length <= 'z')
1007                         byte_length = byte_length - 'a' + 27;
1008                 else
1009                         goto corrupt;
1010                 /* if the input length was not multiple of 4, we would
1011                  * have filler at the end but the filler should never
1012                  * exceed 3 bytes
1013                  */
1014                 if (max_byte_length < byte_length ||
1015                     byte_length <= max_byte_length - 4)
1016                         goto corrupt;
1017                 newsize = fragment->size + byte_length;
1018                 delta = xrealloc(delta, newsize);
1019                 if (decode_85(delta + fragment->size,
1020                               buffer + 1,
1021                               byte_length))
1022                         goto corrupt;
1023                 fragment->size = newsize;
1024                 buffer += llen;
1025                 size -= llen;
1026         }
1027         fragment->patch = delta;
1028         return used;
1029  corrupt:
1030         return error("corrupt binary patch at line %d: %.*s",
1031                      linenr-1, llen-1, buffer);
1032 }
1033
1034 static int parse_chunk(char *buffer, unsigned long size, struct patch *patch)
1035 {
1036         int hdrsize, patchsize;
1037         int offset = find_header(buffer, size, &hdrsize, patch);
1038
1039         if (offset < 0)
1040                 return offset;
1041
1042         patchsize = parse_single_patch(buffer + offset + hdrsize, size - offset - hdrsize, patch);
1043
1044         if (!patchsize) {
1045                 static const char *binhdr[] = {
1046                         "Binary files ",
1047                         "Files ",
1048                         NULL,
1049                 };
1050                 static const char git_binary[] = "GIT binary patch\n";
1051                 int i;
1052                 int hd = hdrsize + offset;
1053                 unsigned long llen = linelen(buffer + hd, size - hd);
1054
1055                 if (llen == sizeof(git_binary) - 1 &&
1056                     !memcmp(git_binary, buffer + hd, llen)) {
1057                         int used;
1058                         linenr++;
1059                         used = parse_binary(buffer + hd + llen,
1060                                             size - hd - llen, patch);
1061                         if (used)
1062                                 patchsize = used + llen;
1063                         else
1064                                 patchsize = 0;
1065                 }
1066                 else if (!memcmp(" differ\n", buffer + hd + llen - 8, 8)) {
1067                         for (i = 0; binhdr[i]; i++) {
1068                                 int len = strlen(binhdr[i]);
1069                                 if (len < size - hd &&
1070                                     !memcmp(binhdr[i], buffer + hd, len)) {
1071                                         linenr++;
1072                                         patch->is_binary = 1;
1073                                         patchsize = llen;
1074                                         break;
1075                                 }
1076                         }
1077                 }
1078
1079                 /* Empty patch cannot be applied if:
1080                  * - it is a binary patch and we do not do binary_replace, or
1081                  * - text patch without metadata change
1082                  */
1083                 if ((apply || check) &&
1084                     (patch->is_binary
1085                      ? !allow_binary_replacement
1086                      : !metadata_changes(patch)))
1087                         die("patch with only garbage at line %d", linenr);
1088         }
1089
1090         return offset + hdrsize + patchsize;
1091 }
1092
1093 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
1094 static const char minuses[]= "----------------------------------------------------------------------";
1095
1096 static void show_stats(struct patch *patch)
1097 {
1098         const char *prefix = "";
1099         char *name = patch->new_name;
1100         char *qname = NULL;
1101         int len, max, add, del, total;
1102
1103         if (!name)
1104                 name = patch->old_name;
1105
1106         if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
1107                 qname = xmalloc(len + 1);
1108                 quote_c_style(name, qname, NULL, 0);
1109                 name = qname;
1110         }
1111
1112         /*
1113          * "scale" the filename
1114          */
1115         len = strlen(name);
1116         max = max_len;
1117         if (max > 50)
1118                 max = 50;
1119         if (len > max) {
1120                 char *slash;
1121                 prefix = "...";
1122                 max -= 3;
1123                 name += len - max;
1124                 slash = strchr(name, '/');
1125                 if (slash)
1126                         name = slash;
1127         }
1128         len = max;
1129
1130         /*
1131          * scale the add/delete
1132          */
1133         max = max_change;
1134         if (max + len > 70)
1135                 max = 70 - len;
1136
1137         add = patch->lines_added;
1138         del = patch->lines_deleted;
1139         total = add + del;
1140
1141         if (max_change > 0) {
1142                 total = (total * max + max_change / 2) / max_change;
1143                 add = (add * max + max_change / 2) / max_change;
1144                 del = total - add;
1145         }
1146         if (patch->is_binary)
1147                 printf(" %s%-*s |  Bin\n", prefix, len, name);
1148         else
1149                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
1150                        len, name, patch->lines_added + patch->lines_deleted,
1151                        add, pluses, del, minuses);
1152         if (qname)
1153                 free(qname);
1154 }
1155
1156 static int read_old_data(struct stat *st, const char *path, void *buf, unsigned long size)
1157 {
1158         int fd;
1159         unsigned long got;
1160
1161         switch (st->st_mode & S_IFMT) {
1162         case S_IFLNK:
1163                 return readlink(path, buf, size);
1164         case S_IFREG:
1165                 fd = open(path, O_RDONLY);
1166                 if (fd < 0)
1167                         return error("unable to open %s", path);
1168                 got = 0;
1169                 for (;;) {
1170                         int ret = xread(fd, buf + got, size - got);
1171                         if (ret <= 0)
1172                                 break;
1173                         got += ret;
1174                 }
1175                 close(fd);
1176                 return got;
1177
1178         default:
1179                 return -1;
1180         }
1181 }
1182
1183 static int find_offset(const char *buf, unsigned long size, const char *fragment, unsigned long fragsize, int line, int *lines)
1184 {
1185         int i;
1186         unsigned long start, backwards, forwards;
1187
1188         if (fragsize > size)
1189                 return -1;
1190
1191         start = 0;
1192         if (line > 1) {
1193                 unsigned long offset = 0;
1194                 i = line-1;
1195                 while (offset + fragsize <= size) {
1196                         if (buf[offset++] == '\n') {
1197                                 start = offset;
1198                                 if (!--i)
1199                                         break;
1200                         }
1201                 }
1202         }
1203
1204         /* Exact line number? */
1205         if (!memcmp(buf + start, fragment, fragsize))
1206                 return start;
1207
1208         /*
1209          * There's probably some smart way to do this, but I'll leave
1210          * that to the smart and beautiful people. I'm simple and stupid.
1211          */
1212         backwards = start;
1213         forwards = start;
1214         for (i = 0; ; i++) {
1215                 unsigned long try;
1216                 int n;
1217
1218                 /* "backward" */
1219                 if (i & 1) {
1220                         if (!backwards) {
1221                                 if (forwards + fragsize > size)
1222                                         break;
1223                                 continue;
1224                         }
1225                         do {
1226                                 --backwards;
1227                         } while (backwards && buf[backwards-1] != '\n');
1228                         try = backwards;
1229                 } else {
1230                         while (forwards + fragsize <= size) {
1231                                 if (buf[forwards++] == '\n')
1232                                         break;
1233                         }
1234                         try = forwards;
1235                 }
1236
1237                 if (try + fragsize > size)
1238                         continue;
1239                 if (memcmp(buf + try, fragment, fragsize))
1240                         continue;
1241                 n = (i >> 1)+1;
1242                 if (i & 1)
1243                         n = -n;
1244                 *lines = n;
1245                 return try;
1246         }
1247
1248         /*
1249          * We should start searching forward and backward.
1250          */
1251         return -1;
1252 }
1253
1254 static void remove_first_line(const char **rbuf, int *rsize)
1255 {
1256         const char *buf = *rbuf;
1257         int size = *rsize;
1258         unsigned long offset;
1259         offset = 0;
1260         while (offset <= size) {
1261                 if (buf[offset++] == '\n')
1262                         break;
1263         }
1264         *rsize = size - offset;
1265         *rbuf = buf + offset;
1266 }
1267
1268 static void remove_last_line(const char **rbuf, int *rsize)
1269 {
1270         const char *buf = *rbuf;
1271         int size = *rsize;
1272         unsigned long offset;
1273         offset = size - 1;
1274         while (offset > 0) {
1275                 if (buf[--offset] == '\n')
1276                         break;
1277         }
1278         *rsize = offset + 1;
1279 }
1280
1281 struct buffer_desc {
1282         char *buffer;
1283         unsigned long size;
1284         unsigned long alloc;
1285 };
1286
1287 static int apply_line(char *output, const char *patch, int plen)
1288 {
1289         /* plen is number of bytes to be copied from patch,
1290          * starting at patch+1 (patch[0] is '+').  Typically
1291          * patch[plen] is '\n'.
1292          */
1293         int add_nl_to_tail = 0;
1294         if ((new_whitespace == strip_whitespace) &&
1295             1 < plen && isspace(patch[plen-1])) {
1296                 if (patch[plen] == '\n')
1297                         add_nl_to_tail = 1;
1298                 plen--;
1299                 while (0 < plen && isspace(patch[plen]))
1300                         plen--;
1301                 applied_after_stripping++;
1302         }
1303         memcpy(output, patch + 1, plen);
1304         if (add_nl_to_tail)
1305                 output[plen++] = '\n';
1306         return plen;
1307 }
1308
1309 static int apply_one_fragment(struct buffer_desc *desc, struct fragment *frag)
1310 {
1311         char *buf = desc->buffer;
1312         const char *patch = frag->patch;
1313         int offset, size = frag->size;
1314         char *old = xmalloc(size);
1315         char *new = xmalloc(size);
1316         const char *oldlines, *newlines;
1317         int oldsize = 0, newsize = 0;
1318         unsigned long leading, trailing;
1319         int pos, lines;
1320
1321         while (size > 0) {
1322                 int len = linelen(patch, size);
1323                 int plen;
1324
1325                 if (!len)
1326                         break;
1327
1328                 /*
1329                  * "plen" is how much of the line we should use for
1330                  * the actual patch data. Normally we just remove the
1331                  * first character on the line, but if the line is
1332                  * followed by "\ No newline", then we also remove the
1333                  * last one (which is the newline, of course).
1334                  */
1335                 plen = len-1;
1336                 if (len < size && patch[len] == '\\')
1337                         plen--;
1338                 switch (*patch) {
1339                 case ' ':
1340                 case '-':
1341                         memcpy(old + oldsize, patch + 1, plen);
1342                         oldsize += plen;
1343                         if (*patch == '-')
1344                                 break;
1345                 /* Fall-through for ' ' */
1346                 case '+':
1347                         if (*patch != '+' || !no_add)
1348                                 newsize += apply_line(new + newsize, patch,
1349                                                       plen);
1350                         break;
1351                 case '@': case '\\':
1352                         /* Ignore it, we already handled it */
1353                         break;
1354                 default:
1355                         return -1;
1356                 }
1357                 patch += len;
1358                 size -= len;
1359         }
1360
1361 #ifdef NO_ACCURATE_DIFF
1362         if (oldsize > 0 && old[oldsize - 1] == '\n' &&
1363                         newsize > 0 && new[newsize - 1] == '\n') {
1364                 oldsize--;
1365                 newsize--;
1366         }
1367 #endif
1368
1369         oldlines = old;
1370         newlines = new;
1371         leading = frag->leading;
1372         trailing = frag->trailing;
1373         lines = 0;
1374         pos = frag->newpos;
1375         for (;;) {
1376                 offset = find_offset(buf, desc->size, oldlines, oldsize, pos, &lines);
1377                 if (offset >= 0) {
1378                         int diff = newsize - oldsize;
1379                         unsigned long size = desc->size + diff;
1380                         unsigned long alloc = desc->alloc;
1381
1382                         /* Warn if it was necessary to reduce the number
1383                          * of context lines.
1384                          */
1385                         if ((leading != frag->leading) || (trailing != frag->trailing))
1386                                 fprintf(stderr, "Context reduced to (%ld/%ld) to apply fragment at %d\n",
1387                                         leading, trailing, pos + lines);
1388
1389                         if (size > alloc) {
1390                                 alloc = size + 8192;
1391                                 desc->alloc = alloc;
1392                                 buf = xrealloc(buf, alloc);
1393                                 desc->buffer = buf;
1394                         }
1395                         desc->size = size;
1396                         memmove(buf + offset + newsize, buf + offset + oldsize, size - offset - newsize);
1397                         memcpy(buf + offset, newlines, newsize);
1398                         offset = 0;
1399
1400                         break;
1401                 }
1402
1403                 /* Am I at my context limits? */
1404                 if ((leading <= p_context) && (trailing <= p_context))
1405                         break;
1406                 /* Reduce the number of context lines
1407                  * Reduce both leading and trailing if they are equal
1408                  * otherwise just reduce the larger context.
1409                  */
1410                 if (leading >= trailing) {
1411                         remove_first_line(&oldlines, &oldsize);
1412                         remove_first_line(&newlines, &newsize);
1413                         pos--;
1414                         leading--;
1415                 }
1416                 if (trailing > leading) {
1417                         remove_last_line(&oldlines, &oldsize);
1418                         remove_last_line(&newlines, &newsize);
1419                         trailing--;
1420                 }
1421         }
1422
1423         free(old);
1424         free(new);
1425         return offset;
1426 }
1427
1428 static int apply_binary(struct buffer_desc *desc, struct patch *patch)
1429 {
1430         const char *name = patch->old_name ? patch->old_name : patch->new_name;
1431         unsigned char sha1[20];
1432         unsigned char hdr[50];
1433         int hdrlen;
1434
1435         if (!allow_binary_replacement)
1436                 return error("cannot apply binary patch to '%s' "
1437                              "without --allow-binary-replacement",
1438                              name);
1439
1440         /* For safety, we require patch index line to contain
1441          * full 40-byte textual SHA1 for old and new, at least for now.
1442          */
1443         if (strlen(patch->old_sha1_prefix) != 40 ||
1444             strlen(patch->new_sha1_prefix) != 40 ||
1445             get_sha1_hex(patch->old_sha1_prefix, sha1) ||
1446             get_sha1_hex(patch->new_sha1_prefix, sha1))
1447                 return error("cannot apply binary patch to '%s' "
1448                              "without full index line", name);
1449
1450         if (patch->old_name) {
1451                 /* See if the old one matches what the patch
1452                  * applies to.
1453                  */
1454                 write_sha1_file_prepare(desc->buffer, desc->size,
1455                                         blob_type, sha1, hdr, &hdrlen);
1456                 if (strcmp(sha1_to_hex(sha1), patch->old_sha1_prefix))
1457                         return error("the patch applies to '%s' (%s), "
1458                                      "which does not match the "
1459                                      "current contents.",
1460                                      name, sha1_to_hex(sha1));
1461         }
1462         else {
1463                 /* Otherwise, the old one must be empty. */
1464                 if (desc->size)
1465                         return error("the patch applies to an empty "
1466                                      "'%s' but it is not empty", name);
1467         }
1468
1469         if (desc->buffer) {
1470                 free(desc->buffer);
1471                 desc->alloc = desc->size = 0;
1472         }
1473         get_sha1_hex(patch->new_sha1_prefix, sha1);
1474         if (!memcmp(sha1, null_sha1, 20))
1475                 return 0; /* deletion patch */
1476
1477         if (has_sha1_file(sha1)) {
1478                 char type[10];
1479                 unsigned long size;
1480
1481                 desc->buffer = read_sha1_file(sha1, type, &size);
1482                 if (!desc->buffer)
1483                         return error("the necessary postimage %s for "
1484                                      "'%s' cannot be read",
1485                                      patch->new_sha1_prefix, name);
1486                 desc->alloc = desc->size = size;
1487         }
1488         else {
1489                 char type[10];
1490                 unsigned long src_size, dst_size;
1491                 void *src;
1492
1493                 get_sha1_hex(patch->old_sha1_prefix, sha1);
1494                 src = read_sha1_file(sha1, type, &src_size);
1495                 if (!src)
1496                         return error("the necessary preimage %s for "
1497                                      "'%s' cannot be read",
1498                                      patch->old_sha1_prefix, name);
1499
1500                 /* patch->fragment->patch has the delta data and
1501                  * we should apply it to the preimage.
1502                  */
1503                 desc->buffer = patch_delta(src, src_size,
1504                                            (void*) patch->fragments->patch,
1505                                            patch->fragments->size,
1506                                            &dst_size);
1507                 if (!desc->buffer)
1508                         return error("binary patch does not apply to '%s'",
1509                                      name);
1510                 desc->size = desc->alloc = dst_size;
1511
1512                 /* verify that the result matches */
1513                 write_sha1_file_prepare(desc->buffer, desc->size, blob_type,
1514                                         sha1, hdr, &hdrlen);
1515                 if (strcmp(sha1_to_hex(sha1), patch->new_sha1_prefix))
1516                         return error("binary patch to '%s' creates incorrect result", name);
1517         }
1518
1519         return 0;
1520 }
1521
1522 static int apply_fragments(struct buffer_desc *desc, struct patch *patch)
1523 {
1524         struct fragment *frag = patch->fragments;
1525         const char *name = patch->old_name ? patch->old_name : patch->new_name;
1526
1527         if (patch->is_binary)
1528                 return apply_binary(desc, patch);
1529
1530         while (frag) {
1531                 if (apply_one_fragment(desc, frag) < 0)
1532                         return error("patch failed: %s:%ld",
1533                                      name, frag->oldpos);
1534                 frag = frag->next;
1535         }
1536         return 0;
1537 }
1538
1539 static int apply_data(struct patch *patch, struct stat *st)
1540 {
1541         char *buf;
1542         unsigned long size, alloc;
1543         struct buffer_desc desc;
1544
1545         size = 0;
1546         alloc = 0;
1547         buf = NULL;
1548         if (patch->old_name) {
1549                 size = st->st_size;
1550                 alloc = size + 8192;
1551                 buf = xmalloc(alloc);
1552                 if (read_old_data(st, patch->old_name, buf, alloc) != size)
1553                         return error("read of %s failed", patch->old_name);
1554         }
1555
1556         desc.size = size;
1557         desc.alloc = alloc;
1558         desc.buffer = buf;
1559         if (apply_fragments(&desc, patch) < 0)
1560                 return -1;
1561         patch->result = desc.buffer;
1562         patch->resultsize = desc.size;
1563
1564         if (patch->is_delete && patch->resultsize)
1565                 return error("removal patch leaves file contents");
1566
1567         return 0;
1568 }
1569
1570 static int check_patch(struct patch *patch)
1571 {
1572         struct stat st;
1573         const char *old_name = patch->old_name;
1574         const char *new_name = patch->new_name;
1575         const char *name = old_name ? old_name : new_name;
1576
1577         if (old_name) {
1578                 int changed;
1579                 int stat_ret = lstat(old_name, &st);
1580
1581                 if (check_index) {
1582                         int pos = cache_name_pos(old_name, strlen(old_name));
1583                         if (pos < 0)
1584                                 return error("%s: does not exist in index",
1585                                              old_name);
1586                         if (stat_ret < 0) {
1587                                 struct checkout costate;
1588                                 if (errno != ENOENT)
1589                                         return error("%s: %s", old_name,
1590                                                      strerror(errno));
1591                                 /* checkout */
1592                                 costate.base_dir = "";
1593                                 costate.base_dir_len = 0;
1594                                 costate.force = 0;
1595                                 costate.quiet = 0;
1596                                 costate.not_new = 0;
1597                                 costate.refresh_cache = 1;
1598                                 if (checkout_entry(active_cache[pos],
1599                                                    &costate,
1600                                                    NULL) ||
1601                                     lstat(old_name, &st))
1602                                         return -1;
1603                         }
1604
1605                         changed = ce_match_stat(active_cache[pos], &st, 1);
1606                         if (changed)
1607                                 return error("%s: does not match index",
1608                                              old_name);
1609                 }
1610                 else if (stat_ret < 0)
1611                         return error("%s: %s", old_name, strerror(errno));
1612
1613                 if (patch->is_new < 0)
1614                         patch->is_new = 0;
1615                 st.st_mode = ntohl(create_ce_mode(st.st_mode));
1616                 if (!patch->old_mode)
1617                         patch->old_mode = st.st_mode;
1618                 if ((st.st_mode ^ patch->old_mode) & S_IFMT)
1619                         return error("%s: wrong type", old_name);
1620                 if (st.st_mode != patch->old_mode)
1621                         fprintf(stderr, "warning: %s has type %o, expected %o\n",
1622                                 old_name, st.st_mode, patch->old_mode);
1623         }
1624
1625         if (new_name && (patch->is_new | patch->is_rename | patch->is_copy)) {
1626                 if (check_index && cache_name_pos(new_name, strlen(new_name)) >= 0)
1627                         return error("%s: already exists in index", new_name);
1628                 if (!lstat(new_name, &st))
1629                         return error("%s: already exists in working directory", new_name);
1630                 if (errno != ENOENT)
1631                         return error("%s: %s", new_name, strerror(errno));
1632                 if (!patch->new_mode) {
1633                         if (patch->is_new)
1634                                 patch->new_mode = S_IFREG | 0644;
1635                         else
1636                                 patch->new_mode = patch->old_mode;
1637                 }
1638         }
1639
1640         if (new_name && old_name) {
1641                 int same = !strcmp(old_name, new_name);
1642                 if (!patch->new_mode)
1643                         patch->new_mode = patch->old_mode;
1644                 if ((patch->old_mode ^ patch->new_mode) & S_IFMT)
1645                         return error("new mode (%o) of %s does not match old mode (%o)%s%s",
1646                                 patch->new_mode, new_name, patch->old_mode,
1647                                 same ? "" : " of ", same ? "" : old_name);
1648         }
1649
1650         if (apply_data(patch, &st) < 0)
1651                 return error("%s: patch does not apply", name);
1652         return 0;
1653 }
1654
1655 static int check_patch_list(struct patch *patch)
1656 {
1657         int error = 0;
1658
1659         for (;patch ; patch = patch->next)
1660                 error |= check_patch(patch);
1661         return error;
1662 }
1663
1664 static inline int is_null_sha1(const unsigned char *sha1)
1665 {
1666         return !memcmp(sha1, null_sha1, 20);
1667 }
1668
1669 static void show_index_list(struct patch *list)
1670 {
1671         struct patch *patch;
1672
1673         /* Once we start supporting the reverse patch, it may be
1674          * worth showing the new sha1 prefix, but until then...
1675          */
1676         for (patch = list; patch; patch = patch->next) {
1677                 const unsigned char *sha1_ptr;
1678                 unsigned char sha1[20];
1679                 const char *name;
1680
1681                 name = patch->old_name ? patch->old_name : patch->new_name;
1682                 if (patch->is_new)
1683                         sha1_ptr = null_sha1;
1684                 else if (get_sha1(patch->old_sha1_prefix, sha1))
1685                         die("sha1 information is lacking or useless (%s).",
1686                             name);
1687                 else
1688                         sha1_ptr = sha1;
1689
1690                 printf("%06o %s ",patch->old_mode, sha1_to_hex(sha1_ptr));
1691                 if (line_termination && quote_c_style(name, NULL, NULL, 0))
1692                         quote_c_style(name, NULL, stdout, 0);
1693                 else
1694                         fputs(name, stdout);
1695                 putchar(line_termination);
1696         }
1697 }
1698
1699 static void stat_patch_list(struct patch *patch)
1700 {
1701         int files, adds, dels;
1702
1703         for (files = adds = dels = 0 ; patch ; patch = patch->next) {
1704                 files++;
1705                 adds += patch->lines_added;
1706                 dels += patch->lines_deleted;
1707                 show_stats(patch);
1708         }
1709
1710         printf(" %d files changed, %d insertions(+), %d deletions(-)\n", files, adds, dels);
1711 }
1712
1713 static void numstat_patch_list(struct patch *patch)
1714 {
1715         for ( ; patch; patch = patch->next) {
1716                 const char *name;
1717                 name = patch->old_name ? patch->old_name : patch->new_name;
1718                 printf("%d\t%d\t", patch->lines_added, patch->lines_deleted);
1719                 if (line_termination && quote_c_style(name, NULL, NULL, 0))
1720                         quote_c_style(name, NULL, stdout, 0);
1721                 else
1722                         fputs(name, stdout);
1723                 putchar('\n');
1724         }
1725 }
1726
1727 static void show_file_mode_name(const char *newdelete, unsigned int mode, const char *name)
1728 {
1729         if (mode)
1730                 printf(" %s mode %06o %s\n", newdelete, mode, name);
1731         else
1732                 printf(" %s %s\n", newdelete, name);
1733 }
1734
1735 static void show_mode_change(struct patch *p, int show_name)
1736 {
1737         if (p->old_mode && p->new_mode && p->old_mode != p->new_mode) {
1738                 if (show_name)
1739                         printf(" mode change %06o => %06o %s\n",
1740                                p->old_mode, p->new_mode, p->new_name);
1741                 else
1742                         printf(" mode change %06o => %06o\n",
1743                                p->old_mode, p->new_mode);
1744         }
1745 }
1746
1747 static void show_rename_copy(struct patch *p)
1748 {
1749         const char *renamecopy = p->is_rename ? "rename" : "copy";
1750         const char *old, *new;
1751
1752         /* Find common prefix */
1753         old = p->old_name;
1754         new = p->new_name;
1755         while (1) {
1756                 const char *slash_old, *slash_new;
1757                 slash_old = strchr(old, '/');
1758                 slash_new = strchr(new, '/');
1759                 if (!slash_old ||
1760                     !slash_new ||
1761                     slash_old - old != slash_new - new ||
1762                     memcmp(old, new, slash_new - new))
1763                         break;
1764                 old = slash_old + 1;
1765                 new = slash_new + 1;
1766         }
1767         /* p->old_name thru old is the common prefix, and old and new
1768          * through the end of names are renames
1769          */
1770         if (old != p->old_name)
1771                 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
1772                        (int)(old - p->old_name), p->old_name,
1773                        old, new, p->score);
1774         else
1775                 printf(" %s %s => %s (%d%%)\n", renamecopy,
1776                        p->old_name, p->new_name, p->score);
1777         show_mode_change(p, 0);
1778 }
1779
1780 static void summary_patch_list(struct patch *patch)
1781 {
1782         struct patch *p;
1783
1784         for (p = patch; p; p = p->next) {
1785                 if (p->is_new)
1786                         show_file_mode_name("create", p->new_mode, p->new_name);
1787                 else if (p->is_delete)
1788                         show_file_mode_name("delete", p->old_mode, p->old_name);
1789                 else {
1790                         if (p->is_rename || p->is_copy)
1791                                 show_rename_copy(p);
1792                         else {
1793                                 if (p->score) {
1794                                         printf(" rewrite %s (%d%%)\n",
1795                                                p->new_name, p->score);
1796                                         show_mode_change(p, 0);
1797                                 }
1798                                 else
1799                                         show_mode_change(p, 1);
1800                         }
1801                 }
1802         }
1803 }
1804
1805 static void patch_stats(struct patch *patch)
1806 {
1807         int lines = patch->lines_added + patch->lines_deleted;
1808
1809         if (lines > max_change)
1810                 max_change = lines;
1811         if (patch->old_name) {
1812                 int len = quote_c_style(patch->old_name, NULL, NULL, 0);
1813                 if (!len)
1814                         len = strlen(patch->old_name);
1815                 if (len > max_len)
1816                         max_len = len;
1817         }
1818         if (patch->new_name) {
1819                 int len = quote_c_style(patch->new_name, NULL, NULL, 0);
1820                 if (!len)
1821                         len = strlen(patch->new_name);
1822                 if (len > max_len)
1823                         max_len = len;
1824         }
1825 }
1826
1827 static void remove_file(struct patch *patch)
1828 {
1829         if (write_index) {
1830                 if (remove_file_from_cache(patch->old_name) < 0)
1831                         die("unable to remove %s from index", patch->old_name);
1832         }
1833         unlink(patch->old_name);
1834 }
1835
1836 static void add_index_file(const char *path, unsigned mode, void *buf, unsigned long size)
1837 {
1838         struct stat st;
1839         struct cache_entry *ce;
1840         int namelen = strlen(path);
1841         unsigned ce_size = cache_entry_size(namelen);
1842
1843         if (!write_index)
1844                 return;
1845
1846         ce = xcalloc(1, ce_size);
1847         memcpy(ce->name, path, namelen);
1848         ce->ce_mode = create_ce_mode(mode);
1849         ce->ce_flags = htons(namelen);
1850         if (lstat(path, &st) < 0)
1851                 die("unable to stat newly created file %s", path);
1852         fill_stat_cache_info(ce, &st);
1853         if (write_sha1_file(buf, size, blob_type, ce->sha1) < 0)
1854                 die("unable to create backing store for newly created file %s", path);
1855         if (add_cache_entry(ce, ADD_CACHE_OK_TO_ADD) < 0)
1856                 die("unable to add cache entry for %s", path);
1857 }
1858
1859 static int try_create_file(const char *path, unsigned int mode, const char *buf, unsigned long size)
1860 {
1861         int fd;
1862
1863         if (S_ISLNK(mode))
1864                 return symlink(buf, path);
1865         fd = open(path, O_CREAT | O_EXCL | O_WRONLY, (mode & 0100) ? 0777 : 0666);
1866         if (fd < 0)
1867                 return -1;
1868         while (size) {
1869                 int written = xwrite(fd, buf, size);
1870                 if (written < 0)
1871                         die("writing file %s: %s", path, strerror(errno));
1872                 if (!written)
1873                         die("out of space writing file %s", path);
1874                 buf += written;
1875                 size -= written;
1876         }
1877         if (close(fd) < 0)
1878                 die("closing file %s: %s", path, strerror(errno));
1879         return 0;
1880 }
1881
1882 /*
1883  * We optimistically assume that the directories exist,
1884  * which is true 99% of the time anyway. If they don't,
1885  * we create them and try again.
1886  */
1887 static void create_one_file(char *path, unsigned mode, const char *buf, unsigned long size)
1888 {
1889         if (!try_create_file(path, mode, buf, size))
1890                 return;
1891
1892         if (errno == ENOENT) {
1893                 if (safe_create_leading_directories(path))
1894                         return;
1895                 if (!try_create_file(path, mode, buf, size))
1896                         return;
1897         }
1898
1899         if (errno == EEXIST) {
1900                 unsigned int nr = getpid();
1901
1902                 for (;;) {
1903                         const char *newpath;
1904                         newpath = mkpath("%s~%u", path, nr);
1905                         if (!try_create_file(newpath, mode, buf, size)) {
1906                                 if (!rename(newpath, path))
1907                                         return;
1908                                 unlink(newpath);
1909                                 break;
1910                         }
1911                         if (errno != EEXIST)
1912                                 break;
1913                         ++nr;
1914                 }
1915         }
1916         die("unable to write file %s mode %o", path, mode);
1917 }
1918
1919 static void create_file(struct patch *patch)
1920 {
1921         char *path = patch->new_name;
1922         unsigned mode = patch->new_mode;
1923         unsigned long size = patch->resultsize;
1924         char *buf = patch->result;
1925
1926         if (!mode)
1927                 mode = S_IFREG | 0644;
1928         create_one_file(path, mode, buf, size);
1929         add_index_file(path, mode, buf, size);
1930 }
1931
1932 static void write_out_one_result(struct patch *patch)
1933 {
1934         if (patch->is_delete > 0) {
1935                 remove_file(patch);
1936                 return;
1937         }
1938         if (patch->is_new > 0 || patch->is_copy) {
1939                 create_file(patch);
1940                 return;
1941         }
1942         /*
1943          * Rename or modification boils down to the same
1944          * thing: remove the old, write the new
1945          */
1946         remove_file(patch);
1947         create_file(patch);
1948 }
1949
1950 static void write_out_results(struct patch *list, int skipped_patch)
1951 {
1952         if (!list && !skipped_patch)
1953                 die("No changes");
1954
1955         while (list) {
1956                 write_out_one_result(list);
1957                 list = list->next;
1958         }
1959 }
1960
1961 static struct cache_file cache_file;
1962
1963 static struct excludes {
1964         struct excludes *next;
1965         const char *path;
1966 } *excludes;
1967
1968 static int use_patch(struct patch *p)
1969 {
1970         const char *pathname = p->new_name ? p->new_name : p->old_name;
1971         struct excludes *x = excludes;
1972         while (x) {
1973                 if (fnmatch(x->path, pathname, 0) == 0)
1974                         return 0;
1975                 x = x->next;
1976         }
1977         if (0 < prefix_length) {
1978                 int pathlen = strlen(pathname);
1979                 if (pathlen <= prefix_length ||
1980                     memcmp(prefix, pathname, prefix_length))
1981                         return 0;
1982         }
1983         return 1;
1984 }
1985
1986 static int apply_patch(int fd, const char *filename)
1987 {
1988         int newfd;
1989         unsigned long offset, size;
1990         char *buffer = read_patch_file(fd, &size);
1991         struct patch *list = NULL, **listp = &list;
1992         int skipped_patch = 0;
1993
1994         patch_input_file = filename;
1995         if (!buffer)
1996                 return -1;
1997         offset = 0;
1998         while (size > 0) {
1999                 struct patch *patch;
2000                 int nr;
2001
2002                 patch = xcalloc(1, sizeof(*patch));
2003                 nr = parse_chunk(buffer + offset, size, patch);
2004                 if (nr < 0)
2005                         break;
2006                 if (use_patch(patch)) {
2007                         patch_stats(patch);
2008                         *listp = patch;
2009                         listp = &patch->next;
2010                 } else {
2011                         /* perhaps free it a bit better? */
2012                         free(patch);
2013                         skipped_patch++;
2014                 }
2015                 offset += nr;
2016                 size -= nr;
2017         }
2018
2019         newfd = -1;
2020         if (whitespace_error && (new_whitespace == error_on_whitespace))
2021                 apply = 0;
2022
2023         write_index = check_index && apply;
2024         if (write_index)
2025                 newfd = hold_index_file_for_update(&cache_file, get_index_file());
2026         if (check_index) {
2027                 if (read_cache() < 0)
2028                         die("unable to read index file");
2029         }
2030
2031         if ((check || apply) && check_patch_list(list) < 0)
2032                 exit(1);
2033
2034         if (apply)
2035                 write_out_results(list, skipped_patch);
2036
2037         if (write_index) {
2038                 if (write_cache(newfd, active_cache, active_nr) ||
2039                     commit_index_file(&cache_file))
2040                         die("Unable to write new cachefile");
2041         }
2042
2043         if (show_index_info)
2044                 show_index_list(list);
2045
2046         if (diffstat)
2047                 stat_patch_list(list);
2048
2049         if (numstat)
2050                 numstat_patch_list(list);
2051
2052         if (summary)
2053                 summary_patch_list(list);
2054
2055         free(buffer);
2056         return 0;
2057 }
2058
2059 static int git_apply_config(const char *var, const char *value)
2060 {
2061         if (!strcmp(var, "apply.whitespace")) {
2062                 apply_default_whitespace = strdup(value);
2063                 return 0;
2064         }
2065         return git_default_config(var, value);
2066 }
2067
2068
2069 int main(int argc, char **argv)
2070 {
2071         int i;
2072         int read_stdin = 1;
2073         const char *whitespace_option = NULL;
2074
2075         for (i = 1; i < argc; i++) {
2076                 const char *arg = argv[i];
2077                 char *end;
2078                 int fd;
2079
2080                 if (!strcmp(arg, "-")) {
2081                         apply_patch(0, "<stdin>");
2082                         read_stdin = 0;
2083                         continue;
2084                 }
2085                 if (!strncmp(arg, "--exclude=", 10)) {
2086                         struct excludes *x = xmalloc(sizeof(*x));
2087                         x->path = arg + 10;
2088                         x->next = excludes;
2089                         excludes = x;
2090                         continue;
2091                 }
2092                 if (!strncmp(arg, "-p", 2)) {
2093                         p_value = atoi(arg + 2);
2094                         continue;
2095                 }
2096                 if (!strcmp(arg, "--no-add")) {
2097                         no_add = 1;
2098                         continue;
2099                 }
2100                 if (!strcmp(arg, "--stat")) {
2101                         apply = 0;
2102                         diffstat = 1;
2103                         continue;
2104                 }
2105                 if (!strcmp(arg, "--allow-binary-replacement")) {
2106                         allow_binary_replacement = 1;
2107                         continue;
2108                 }
2109                 if (!strcmp(arg, "--numstat")) {
2110                         apply = 0;
2111                         numstat = 1;
2112                         continue;
2113                 }
2114                 if (!strcmp(arg, "--summary")) {
2115                         apply = 0;
2116                         summary = 1;
2117                         continue;
2118                 }
2119                 if (!strcmp(arg, "--check")) {
2120                         apply = 0;
2121                         check = 1;
2122                         continue;
2123                 }
2124                 if (!strcmp(arg, "--index")) {
2125                         check_index = 1;
2126                         continue;
2127                 }
2128                 if (!strcmp(arg, "--apply")) {
2129                         apply = 1;
2130                         continue;
2131                 }
2132                 if (!strcmp(arg, "--index-info")) {
2133                         apply = 0;
2134                         show_index_info = 1;
2135                         continue;
2136                 }
2137                 if (!strcmp(arg, "-z")) {
2138                         line_termination = 0;
2139                         continue;
2140                 }
2141                 if (!strncmp(arg, "-C", 2)) {
2142                         p_context = strtoul(arg + 2, &end, 0);
2143                         if (*end != '\0')
2144                                 die("unrecognized context count '%s'", arg + 2);
2145                         continue;
2146                 }
2147                 if (!strncmp(arg, "--whitespace=", 13)) {
2148                         whitespace_option = arg + 13;
2149                         parse_whitespace_option(arg + 13);
2150                         continue;
2151                 }
2152
2153                 if (check_index && prefix_length < 0) {
2154                         prefix = setup_git_directory();
2155                         prefix_length = prefix ? strlen(prefix) : 0;
2156                         git_config(git_apply_config);
2157                         if (!whitespace_option && apply_default_whitespace)
2158                                 parse_whitespace_option(apply_default_whitespace);
2159                 }
2160                 if (0 < prefix_length)
2161                         arg = prefix_filename(prefix, prefix_length, arg);
2162
2163                 fd = open(arg, O_RDONLY);
2164                 if (fd < 0)
2165                         usage(apply_usage);
2166                 read_stdin = 0;
2167                 set_default_whitespace_mode(whitespace_option);
2168                 apply_patch(fd, arg);
2169                 close(fd);
2170         }
2171         set_default_whitespace_mode(whitespace_option);
2172         if (read_stdin)
2173                 apply_patch(0, "<stdin>");
2174         if (whitespace_error) {
2175                 if (squelch_whitespace_errors &&
2176                     squelch_whitespace_errors < whitespace_error) {
2177                         int squelched =
2178                                 whitespace_error - squelch_whitespace_errors;
2179                         fprintf(stderr, "warning: squelched %d whitespace error%s\n",
2180                                 squelched,
2181                                 squelched == 1 ? "" : "s");
2182                 }
2183                 if (new_whitespace == error_on_whitespace)
2184                         die("%d line%s add%s trailing whitespaces.",
2185                             whitespace_error,
2186                             whitespace_error == 1 ? "" : "s",
2187                             whitespace_error == 1 ? "s" : "");
2188                 if (applied_after_stripping)
2189                         fprintf(stderr, "warning: %d line%s applied after"
2190                                 " stripping trailing whitespaces.\n",
2191                                 applied_after_stripping,
2192                                 applied_after_stripping == 1 ? "" : "s");
2193                 else if (whitespace_error)
2194                         fprintf(stderr, "warning: %d line%s add%s trailing"
2195                                 " whitespaces.\n",
2196                                 whitespace_error,
2197                                 whitespace_error == 1 ? "" : "s",
2198                                 whitespace_error == 1 ? "s" : "");
2199         }
2200         return 0;
2201 }