builtin-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 "cache.h"
  10 #include "cache-tree.h"
  11 #include "quote.h"
  12 #include "blob.h"
  13 #include "delta.h"
  14 #include "builtin.h"
  15 #include "string-list.h"
  16 #include "dir.h"
  17
  18 /*
  19  *  --check turns on checking that the working tree matches the
  20  *    files that are being modified, but doesn't apply the patch
  21  *  --stat does just a diffstat, and doesn't actually apply
  22  *  --numstat does numeric diffstat, and doesn't actually apply
  23  *  --index-info shows the old and new index info for paths if available.
  24  *  --index updates the cache as well.
  25  *  --cached updates only the cache without ever touching the working tree.
  26  */
  27 static const char *prefix;
  28 static int prefix_length = -1;
  29 static int newfd = -1;
  30
  31 static int unidiff_zero;
  32 static int p_value = 1;
  33 static int p_value_known;
  34 static int check_index;
  35 static int update_index;
  36 static int cached;
  37 static int diffstat;
  38 static int numstat;
  39 static int summary;
  40 static int check;
  41 static int apply = 1;
  42 static int apply_in_reverse;
  43 static int apply_with_reject;
  44 static int apply_verbosely;
  45 static int no_add;
  46 static const char *fake_ancestor;
  47 static int line_termination = '\n';
  48 static unsigned long p_context = ULONG_MAX;
  49 static const char apply_usage[] =
  50 "git apply [--stat] [--numstat] [--summary] [--check] [--index] [--cached] [--apply] [--no-add] [--index-info] [--allow-binary-replacement] [--reverse] [--reject] [--verbose] [-z] [-pNUM] [-CNUM] [--whitespace=<nowarn|warn|fix|error|error-all>] <patch>...";
  51
  52 static enum ws_error_action {
  53         nowarn_ws_error,
  54         warn_on_ws_error,
  55         die_on_ws_error,
  56         correct_ws_error,
  57 } ws_error_action = warn_on_ws_error;
  58 static int whitespace_error;
  59 static int squelch_whitespace_errors = 5;
  60 static int applied_after_fixing_ws;
  61 static const char *patch_input_file;
  62 static const char *root;
  63 static int root_len;
  64
  65 static void parse_whitespace_option(const char *option)
  66 {
  67         if (!option) {
  68                 ws_error_action = warn_on_ws_error;
  69                 return;
  70         }
  71         if (!strcmp(option, "warn")) {
  72                 ws_error_action = warn_on_ws_error;
  73                 return;
  74         }
  75         if (!strcmp(option, "nowarn")) {
  76                 ws_error_action = nowarn_ws_error;
  77                 return;
  78         }
  79         if (!strcmp(option, "error")) {
  80                 ws_error_action = die_on_ws_error;
  81                 return;
  82         }
  83         if (!strcmp(option, "error-all")) {
  84                 ws_error_action = die_on_ws_error;
  85                 squelch_whitespace_errors = 0;
  86                 return;
  87         }
  88         if (!strcmp(option, "strip") || !strcmp(option, "fix")) {
  89                 ws_error_action = correct_ws_error;
  90                 return;
  91         }
  92         die("unrecognized whitespace option '%s'", option);
  93 }
  94
  95 static void set_default_whitespace_mode(const char *whitespace_option)
  96 {
  97         if (!whitespace_option && !apply_default_whitespace)
  98                 ws_error_action = (apply ? warn_on_ws_error : nowarn_ws_error);
  99 }
 100
 101 /*
 102  * For "diff-stat" like behaviour, we keep track of the biggest change
 103  * we've seen, and the longest filename. That allows us to do simple
 104  * scaling.
 105  */
 106 static int max_change, max_len;
 107
 108 /*
 109  * Various "current state", notably line numbers and what
 110  * file (and how) we're patching right now.. The "is_xxxx"
 111  * things are flags, where -1 means "don't know yet".
 112  */
 113 static int linenr = 1;
 114
 115 /*
 116  * This represents one "hunk" from a patch, starting with
 117  * "@@ -oldpos,oldlines +newpos,newlines @@" marker.  The
 118  * patch text is pointed at by patch, and its byte length
 119  * is stored in size.  leading and trailing are the number
 120  * of context lines.
 121  */
 122 struct fragment {
 123         unsigned long leading, trailing;
 124         unsigned long oldpos, oldlines;
 125         unsigned long newpos, newlines;
 126         const char *patch;
 127         int size;
 128         int rejected;
 129         struct fragment *next;
 130 };
 131
 132 /*
 133  * When dealing with a binary patch, we reuse "leading" field
 134  * to store the type of the binary hunk, either deflated "delta"
 135  * or deflated "literal".
 136  */
 137 #define binary_patch_method leading
 138 #define BINARY_DELTA_DEFLATED   1
 139 #define BINARY_LITERAL_DEFLATED 2
 140
 141 /*
 142  * This represents a "patch" to a file, both metainfo changes
 143  * such as creation/deletion, filemode and content changes represented
 144  * as a series of fragments.
 145  */
 146 struct patch {
 147         char *new_name, *old_name, *def_name;
 148         unsigned int old_mode, new_mode;
 149         int is_new, is_delete;  /* -1 = unknown, 0 = false, 1 = true */
 150         int rejected;
 151         unsigned ws_rule;
 152         unsigned long deflate_origlen;
 153         int lines_added, lines_deleted;
 154         int score;
 155         unsigned int is_toplevel_relative:1;
 156         unsigned int inaccurate_eof:1;
 157         unsigned int is_binary:1;
 158         unsigned int is_copy:1;
 159         unsigned int is_rename:1;
 160         unsigned int recount:1;
 161         struct fragment *fragments;
 162         char *result;
 163         size_t resultsize;
 164         char old_sha1_prefix[41];
 165         char new_sha1_prefix[41];
 166         struct patch *next;
 167 };
 168
 169 /*
 170  * A line in a file, len-bytes long (includes the terminating LF,
 171  * except for an incomplete line at the end if the file ends with
 172  * one), and its contents hashes to 'hash'.
 173  */
 174 struct line {
 175         size_t len;
 176         unsigned hash : 24;
 177         unsigned flag : 8;
 178 #define LINE_COMMON     1
 179 };
 180
 181 /*
 182  * This represents a "file", which is an array of "lines".
 183  */
 184 struct image {
 185         char *buf;
 186         size_t len;
 187         size_t nr;
 188         size_t alloc;
 189         struct line *line_allocated;
 190         struct line *line;
 191 };
 192
 193 /*
 194  * Records filenames that have been touched, in order to handle
 195  * the case where more than one patches touch the same file.
 196  */
 197
 198 static struct string_list fn_table;
 199
 200 static uint32_t hash_line(const char *cp, size_t len)
 201 {
 202         size_t i;
 203         uint32_t h;
 204         for (i = 0, h = 0; i < len; i++) {
 205                 if (!isspace(cp[i])) {
 206                         h = h * 3 + (cp[i] & 0xff);
 207                 }
 208         }
 209         return h;
 210 }
 211
 212 static void add_line_info(struct image *img, const char *bol, size_t len, unsigned flag)
 213 {
 214         ALLOC_GROW(img->line_allocated, img->nr + 1, img->alloc);
 215         img->line_allocated[img->nr].len = len;
 216         img->line_allocated[img->nr].hash = hash_line(bol, len);
 217         img->line_allocated[img->nr].flag = flag;
 218         img->nr++;
 219 }
 220
 221 static void prepare_image(struct image *image, char *buf, size_t len,
 222                           int prepare_linetable)
 223 {
 224         const char *cp, *ep;
 225
 226         memset(image, 0, sizeof(*image));
 227         image->buf = buf;
 228         image->len = len;
 229
 230         if (!prepare_linetable)
 231                 return;
 232
 233         ep = image->buf + image->len;
 234         cp = image->buf;
 235         while (cp < ep) {
 236                 const char *next;
 237                 for (next = cp; next < ep && *next != '\n'; next++)
 238                         ;
 239                 if (next < ep)
 240                         next++;
 241                 add_line_info(image, cp, next - cp, 0);
 242                 cp = next;
 243         }
 244         image->line = image->line_allocated;
 245 }
 246
 247 static void clear_image(struct image *image)
 248 {
 249         free(image->buf);
 250         image->buf = NULL;
 251         image->len = 0;
 252 }
 253
 254 static void say_patch_name(FILE *output, const char *pre,
 255                            struct patch *patch, const char *post)
 256 {
 257         fputs(pre, output);
 258         if (patch->old_name && patch->new_name &&
 259             strcmp(patch->old_name, patch->new_name)) {
 260                 quote_c_style(patch->old_name, NULL, output, 0);
 261                 fputs(" => ", output);
 262                 quote_c_style(patch->new_name, NULL, output, 0);
 263         } else {
 264                 const char *n = patch->new_name;
 265                 if (!n)
 266                         n = patch->old_name;
 267                 quote_c_style(n, NULL, output, 0);
 268         }
 269         fputs(post, output);
 270 }
 271
 272 #define CHUNKSIZE (8192)
 273 #define SLOP (16)
 274
 275 static void read_patch_file(struct strbuf *sb, int fd)
 276 {
 277         if (strbuf_read(sb, fd, 0) < 0)
 278                 die("git apply: read returned %s", strerror(errno));
 279
 280         /*
 281          * Make sure that we have some slop in the buffer
 282          * so that we can do speculative "memcmp" etc, and
 283          * see to it that it is NUL-filled.
 284          */
 285         strbuf_grow(sb, SLOP);
 286         memset(sb->buf + sb->len, 0, SLOP);
 287 }
 288
 289 static unsigned long linelen(const char *buffer, unsigned long size)
 290 {
 291         unsigned long len = 0;
 292         while (size--) {
 293                 len++;
 294                 if (*buffer++ == '\n')
 295                         break;
 296         }
 297         return len;
 298 }
 299
 300 static int is_dev_null(const char *str)
 301 {
 302         return !memcmp("/dev/null", str, 9) && isspace(str[9]);
 303 }
 304
 305 #define TERM_SPACE      1
 306 #define TERM_TAB        2
 307
 308 static int name_terminate(const char *name, int namelen, int c, int terminate)
 309 {
 310         if (c == ' ' && !(terminate & TERM_SPACE))
 311                 return 0;
 312         if (c == '\t' && !(terminate & TERM_TAB))
 313                 return 0;
 314
 315         return 1;
 316 }
 317
 318 static char *find_name(const char *line, char *def, int p_value, int terminate)
 319 {
 320         int len;
 321         const char *start = line;
 322
 323         if (*line == '"') {
 324                 struct strbuf name;
 325
 326                 /*
 327                  * Proposed "new-style" GNU patch/diff format; see
 328                  * http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
 329                  */
 330                 strbuf_init(&name, 0);
 331                 if (!unquote_c_style(&name, line, NULL)) {
 332                         char *cp;
 333
 334                         for (cp = name.buf; p_value; p_value--) {
 335                                 cp = strchr(cp, '/');
 336                                 if (!cp)
 337                                         break;
 338                                 cp++;
 339                         }
 340                         if (cp) {
 341                                 /* name can later be freed, so we need
 342                                  * to memmove, not just return cp
 343                                  */
 344                                 strbuf_remove(&name, 0, cp - name.buf);
 345                                 free(def);
 346                                 if (root)
 347                                         strbuf_insert(&name, 0, root, root_len);
 348                                 return strbuf_detach(&name, NULL);
 349                         }
 350                 }
 351                 strbuf_release(&name);
 352         }
 353
 354         for (;;) {
 355                 char c = *line;
 356
 357                 if (isspace(c)) {
 358                         if (c == '\n')
 359                                 break;
 360                         if (name_terminate(start, line-start, c, terminate))
 361                                 break;
 362                 }
 363                 line++;
 364                 if (c == '/' && !--p_value)
 365                         start = line;
 366         }
 367         if (!start)
 368                 return def;
 369         len = line - start;
 370         if (!len)
 371                 return def;
 372
 373         /*
 374          * Generally we prefer the shorter name, especially
 375          * if the other one is just a variation of that with
 376          * something else tacked on to the end (ie "file.orig"
 377          * or "file~").
 378          */
 379         if (def) {
 380                 int deflen = strlen(def);
 381                 if (deflen < len && !strncmp(start, def, deflen))
 382                         return def;
 383                 free(def);
 384         }
 385
 386         if (root) {
 387                 char *ret = xmalloc(root_len + len + 1);
 388                 strcpy(ret, root);
 389                 memcpy(ret + root_len, start, len);
 390                 ret[root_len + len] = '\0';
 391                 return ret;
 392         }
 393
 394         return xmemdupz(start, len);
 395 }
 396
 397 static int count_slashes(const char *cp)
 398 {
 399         int cnt = 0;
 400         char ch;
 401
 402         while ((ch = *cp++))
 403                 if (ch == '/')
 404                         cnt++;
 405         return cnt;
 406 }
 407
 408 /*
 409  * Given the string after "--- " or "+++ ", guess the appropriate
 410  * p_value for the given patch.
 411  */
 412 static int guess_p_value(const char *nameline)
 413 {
 414         char *name, *cp;
 415         int val = -1;
 416
 417         if (is_dev_null(nameline))
 418                 return -1;
 419         name = find_name(nameline, NULL, 0, TERM_SPACE | TERM_TAB);
 420         if (!name)
 421                 return -1;
 422         cp = strchr(name, '/');
 423         if (!cp)
 424                 val = 0;
 425         else if (prefix) {
 426                 /*
 427                  * Does it begin with "a/$our-prefix" and such?  Then this is
 428                  * very likely to apply to our directory.
 429                  */
 430                 if (!strncmp(name, prefix, prefix_length))
 431                         val = count_slashes(prefix);
 432                 else {
 433                         cp++;
 434                         if (!strncmp(cp, prefix, prefix_length))
 435                                 val = count_slashes(prefix) + 1;
 436                 }
 437         }
 438         free(name);
 439         return val;
 440 }
 441
 442 /*
 443  * Get the name etc info from the ---/+++ lines of a traditional patch header
 444  *
 445  * FIXME! The end-of-filename heuristics are kind of screwy. For existing
 446  * files, we can happily check the index for a match, but for creating a
 447  * new file we should try to match whatever "patch" does. I have no idea.
 448  */
 449 static void parse_traditional_patch(const char *first, const char *second, struct patch *patch)
 450 {
 451         char *name;
 452
 453         first += 4;     /* skip "--- " */
 454         second += 4;    /* skip "+++ " */
 455         if (!p_value_known) {
 456                 int p, q;
 457                 p = guess_p_value(first);
 458                 q = guess_p_value(second);
 459                 if (p < 0) p = q;
 460                 if (0 <= p && p == q) {
 461                         p_value = p;
 462                         p_value_known = 1;
 463                 }
 464         }
 465         if (is_dev_null(first)) {
 466                 patch->is_new = 1;
 467                 patch->is_delete = 0;
 468                 name = find_name(second, NULL, p_value, TERM_SPACE | TERM_TAB);
 469                 patch->new_name = name;
 470         } else if (is_dev_null(second)) {
 471                 patch->is_new = 0;
 472                 patch->is_delete = 1;
 473                 name = find_name(first, NULL, p_value, TERM_SPACE | TERM_TAB);
 474                 patch->old_name = name;
 475         } else {
 476                 name = find_name(first, NULL, p_value, TERM_SPACE | TERM_TAB);
 477                 name = find_name(second, name, p_value, TERM_SPACE | TERM_TAB);
 478                 patch->old_name = patch->new_name = name;
 479         }
 480         if (!name)
 481                 die("unable to find filename in patch at line %d", linenr);
 482 }
 483
 484 static int gitdiff_hdrend(const char *line, struct patch *patch)
 485 {
 486         return -1;
 487 }
 488
 489 /*
 490  * We're anal about diff header consistency, to make
 491  * sure that we don't end up having strange ambiguous
 492  * patches floating around.
 493  *
 494  * As a result, gitdiff_{old|new}name() will check
 495  * their names against any previous information, just
 496  * to make sure..
 497  */
 498 static char *gitdiff_verify_name(const char *line, int isnull, char *orig_name, const char *oldnew)
 499 {
 500         if (!orig_name && !isnull)
 501                 return find_name(line, NULL, p_value, TERM_TAB);
 502
 503         if (orig_name) {
 504                 int len;
 505                 const char *name;
 506                 char *another;
 507                 name = orig_name;
 508                 len = strlen(name);
 509                 if (isnull)
 510                         die("git apply: bad git-diff - expected /dev/null, got %s on line %d", name, linenr);
 511                 another = find_name(line, NULL, p_value, TERM_TAB);
 512                 if (!another || memcmp(another, name, len))
 513                         die("git apply: bad git-diff - inconsistent %s filename on line %d", oldnew, linenr);
 514                 free(another);
 515                 return orig_name;
 516         }
 517         else {
 518                 /* expect "/dev/null" */
 519                 if (memcmp("/dev/null", line, 9) || line[9] != '\n')
 520                         die("git apply: bad git-diff - expected /dev/null on line %d", linenr);
 521                 return NULL;
 522         }
 523 }
 524
 525 static int gitdiff_oldname(const char *line, struct patch *patch)
 526 {
 527         patch->old_name = gitdiff_verify_name(line, patch->is_new, patch->old_name, "old");
 528         return 0;
 529 }
 530
 531 static int gitdiff_newname(const char *line, struct patch *patch)
 532 {
 533         patch->new_name = gitdiff_verify_name(line, patch->is_delete, patch->new_name, "new");
 534         return 0;
 535 }
 536
 537 static int gitdiff_oldmode(const char *line, struct patch *patch)
 538 {
 539         patch->old_mode = strtoul(line, NULL, 8);
 540         return 0;
 541 }
 542
 543 static int gitdiff_newmode(const char *line, struct patch *patch)
 544 {
 545         patch->new_mode = strtoul(line, NULL, 8);
 546         return 0;
 547 }
 548
 549 static int gitdiff_delete(const char *line, struct patch *patch)
 550 {
 551         patch->is_delete = 1;
 552         patch->old_name = patch->def_name;
 553         return gitdiff_oldmode(line, patch);
 554 }
 555
 556 static int gitdiff_newfile(const char *line, struct patch *patch)
 557 {
 558         patch->is_new = 1;
 559         patch->new_name = patch->def_name;
 560         return gitdiff_newmode(line, patch);
 561 }
 562
 563 static int gitdiff_copysrc(const char *line, struct patch *patch)
 564 {
 565         patch->is_copy = 1;
 566         patch->old_name = find_name(line, NULL, 0, 0);
 567         return 0;
 568 }
 569
 570 static int gitdiff_copydst(const char *line, struct patch *patch)
 571 {
 572         patch->is_copy = 1;
 573         patch->new_name = find_name(line, NULL, 0, 0);
 574         return 0;
 575 }
 576
 577 static int gitdiff_renamesrc(const char *line, struct patch *patch)
 578 {
 579         patch->is_rename = 1;
 580         patch->old_name = find_name(line, NULL, 0, 0);
 581         return 0;
 582 }
 583
 584 static int gitdiff_renamedst(const char *line, struct patch *patch)
 585 {
 586         patch->is_rename = 1;
 587         patch->new_name = find_name(line, NULL, 0, 0);
 588         return 0;
 589 }
 590
 591 static int gitdiff_similarity(const char *line, struct patch *patch)
 592 {
 593         if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
 594                 patch->score = 0;
 595         return 0;
 596 }
 597
 598 static int gitdiff_dissimilarity(const char *line, struct patch *patch)
 599 {
 600         if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
 601                 patch->score = 0;
 602         return 0;
 603 }
 604
 605 static int gitdiff_index(const char *line, struct patch *patch)
 606 {
 607         /*
 608          * index line is N hexadecimal, "..", N hexadecimal,
 609          * and optional space with octal mode.
 610          */
 611         const char *ptr, *eol;
 612         int len;
 613
 614         ptr = strchr(line, '.');
 615         if (!ptr || ptr[1] != '.' || 40 < ptr - line)
 616                 return 0;
 617         len = ptr - line;
 618         memcpy(patch->old_sha1_prefix, line, len);
 619         patch->old_sha1_prefix[len] = 0;
 620
 621         line = ptr + 2;
 622         ptr = strchr(line, ' ');
 623         eol = strchr(line, '\n');
 624
 625         if (!ptr || eol < ptr)
 626                 ptr = eol;
 627         len = ptr - line;
 628
 629         if (40 < len)
 630                 return 0;
 631         memcpy(patch->new_sha1_prefix, line, len);
 632         patch->new_sha1_prefix[len] = 0;
 633         if (*ptr == ' ')
 634                 patch->new_mode = patch->old_mode = strtoul(ptr+1, NULL, 8);
 635         return 0;
 636 }
 637
 638 /*
 639  * This is normal for a diff that doesn't change anything: we'll fall through
 640  * into the next diff. Tell the parser to break out.
 641  */
 642 static int gitdiff_unrecognized(const char *line, struct patch *patch)
 643 {
 644         return -1;
 645 }
 646
 647 static const char *stop_at_slash(const char *line, int llen)
 648 {
 649         int i;
 650
 651         for (i = 0; i < llen; i++) {
 652                 int ch = line[i];
 653                 if (ch == '/')
 654                         return line + i;
 655         }
 656         return NULL;
 657 }
 658
 659 /*
 660  * This is to extract the same name that appears on "diff --git"
 661  * line.  We do not find and return anything if it is a rename
 662  * patch, and it is OK because we will find the name elsewhere.
 663  * We need to reliably find name only when it is mode-change only,
 664  * creation or deletion of an empty file.  In any of these cases,
 665  * both sides are the same name under a/ and b/ respectively.
 666  */
 667 static char *git_header_name(char *line, int llen)
 668 {
 669         const char *name;
 670         const char *second = NULL;
 671         size_t len;
 672
 673         line += strlen("diff --git ");
 674         llen -= strlen("diff --git ");
 675
 676         if (*line == '"') {
 677                 const char *cp;
 678                 struct strbuf first;
 679                 struct strbuf sp;
 680
 681                 strbuf_init(&first, 0);
 682                 strbuf_init(&sp, 0);
 683
 684                 if (unquote_c_style(&first, line, &second))
 685                         goto free_and_fail1;
 686
 687                 /* advance to the first slash */
 688                 cp = stop_at_slash(first.buf, first.len);
 689                 /* we do not accept absolute paths */
 690                 if (!cp || cp == first.buf)
 691                         goto free_and_fail1;
 692                 strbuf_remove(&first, 0, cp + 1 - first.buf);
 693
 694                 /*
 695                  * second points at one past closing dq of name.
 696                  * find the second name.
 697                  */
 698                 while ((second < line + llen) && isspace(*second))
 699                         second++;
 700
 701                 if (line + llen <= second)
 702                         goto free_and_fail1;
 703                 if (*second == '"') {
 704                         if (unquote_c_style(&sp, second, NULL))
 705                                 goto free_and_fail1;
 706                         cp = stop_at_slash(sp.buf, sp.len);
 707                         if (!cp || cp == sp.buf)
 708                                 goto free_and_fail1;
 709                         /* They must match, otherwise ignore */
 710                         if (strcmp(cp + 1, first.buf))
 711                                 goto free_and_fail1;
 712                         strbuf_release(&sp);
 713                         return strbuf_detach(&first, NULL);
 714                 }
 715
 716                 /* unquoted second */
 717                 cp = stop_at_slash(second, line + llen - second);
 718                 if (!cp || cp == second)
 719                         goto free_and_fail1;
 720                 cp++;
 721                 if (line + llen - cp != first.len + 1 ||
 722                     memcmp(first.buf, cp, first.len))
 723                         goto free_and_fail1;
 724                 return strbuf_detach(&first, NULL);
 725
 726         free_and_fail1:
 727                 strbuf_release(&first);
 728                 strbuf_release(&sp);
 729                 return NULL;
 730         }
 731
 732         /* unquoted first name */
 733         name = stop_at_slash(line, llen);
 734         if (!name || name == line)
 735                 return NULL;
 736         name++;
 737
 738         /*
 739          * since the first name is unquoted, a dq if exists must be
 740          * the beginning of the second name.
 741          */
 742         for (second = name; second < line + llen; second++) {
 743                 if (*second == '"') {
 744                         struct strbuf sp;
 745                         const char *np;
 746
 747                         strbuf_init(&sp, 0);
 748                         if (unquote_c_style(&sp, second, NULL))
 749                                 goto free_and_fail2;
 750
 751                         np = stop_at_slash(sp.buf, sp.len);
 752                         if (!np || np == sp.buf)
 753                                 goto free_and_fail2;
 754                         np++;
 755
 756                         len = sp.buf + sp.len - np;
 757                         if (len < second - name &&
 758                             !strncmp(np, name, len) &&
 759                             isspace(name[len])) {
 760                                 /* Good */
 761                                 strbuf_remove(&sp, 0, np - sp.buf);
 762                                 return strbuf_detach(&sp, NULL);
 763                         }
 764
 765                 free_and_fail2:
 766                         strbuf_release(&sp);
 767                         return NULL;
 768                 }
 769         }
 770
 771         /*
 772          * Accept a name only if it shows up twice, exactly the same
 773          * form.
 774          */
 775         for (len = 0 ; ; len++) {
 776                 switch (name[len]) {
 777                 default:
 778                         continue;
 779                 case '\n':
 780                         return NULL;
 781                 case '\t': case ' ':
 782                         second = name+len;
 783                         for (;;) {
 784                                 char c = *second++;
 785                                 if (c == '\n')
 786                                         return NULL;
 787                                 if (c == '/')
 788                                         break;
 789                         }
 790                         if (second[len] == '\n' && !memcmp(name, second, len)) {
 791                                 return xmemdupz(name, len);
 792                         }
 793                 }
 794         }
 795 }
 796
 797 /* Verify that we recognize the lines following a git header */
 798 static int parse_git_header(char *line, int len, unsigned int size, struct patch *patch)
 799 {
 800         unsigned long offset;
 801
 802         /* A git diff has explicit new/delete information, so we don't guess */
 803         patch->is_new = 0;
 804         patch->is_delete = 0;
 805
 806         /*
 807          * Some things may not have the old name in the
 808          * rest of the headers anywhere (pure mode changes,
 809          * or removing or adding empty files), so we get
 810          * the default name from the header.
 811          */
 812         patch->def_name = git_header_name(line, len);
 813         if (patch->def_name && root) {
 814                 char *s = xmalloc(root_len + strlen(patch->def_name) + 1);
 815                 strcpy(s, root);
 816                 strcpy(s + root_len, patch->def_name);
 817                 free(patch->def_name);
 818                 patch->def_name = s;
 819         }
 820
 821         line += len;
 822         size -= len;
 823         linenr++;
 824         for (offset = len ; size > 0 ; offset += len, size -= len, line += len, linenr++) {
 825                 static const struct opentry {
 826                         const char *str;
 827                         int (*fn)(const char *, struct patch *);
 828                 } optable[] = {
 829                         { "@@ -", gitdiff_hdrend },
 830                         { "--- ", gitdiff_oldname },
 831                         { "+++ ", gitdiff_newname },
 832                         { "old mode ", gitdiff_oldmode },
 833                         { "new mode ", gitdiff_newmode },
 834                         { "deleted file mode ", gitdiff_delete },
 835                         { "new file mode ", gitdiff_newfile },
 836                         { "copy from ", gitdiff_copysrc },
 837                         { "copy to ", gitdiff_copydst },
 838                         { "rename old ", gitdiff_renamesrc },
 839                         { "rename new ", gitdiff_renamedst },
 840                         { "rename from ", gitdiff_renamesrc },
 841                         { "rename to ", gitdiff_renamedst },
 842                         { "similarity index ", gitdiff_similarity },
 843                         { "dissimilarity index ", gitdiff_dissimilarity },
 844                         { "index ", gitdiff_index },
 845                         { "", gitdiff_unrecognized },
 846                 };
 847                 int i;
 848
 849                 len = linelen(line, size);
 850                 if (!len || line[len-1] != '\n')
 851                         break;
 852                 for (i = 0; i < ARRAY_SIZE(optable); i++) {
 853                         const struct opentry *p = optable + i;
 854                         int oplen = strlen(p->str);
 855                         if (len < oplen || memcmp(p->str, line, oplen))
 856                                 continue;
 857                         if (p->fn(line + oplen, patch) < 0)
 858                                 return offset;
 859                         break;
 860                 }
 861         }
 862
 863         return offset;
 864 }
 865
 866 static int parse_num(const char *line, unsigned long *p)
 867 {
 868         char *ptr;
 869
 870         if (!isdigit(*line))
 871                 return 0;
 872         *p = strtoul(line, &ptr, 10);
 873         return ptr - line;
 874 }
 875
 876 static int parse_range(const char *line, int len, int offset, const char *expect,
 877                        unsigned long *p1, unsigned long *p2)
 878 {
 879         int digits, ex;
 880
 881         if (offset < 0 || offset >= len)
 882                 return -1;
 883         line += offset;
 884         len -= offset;
 885
 886         digits = parse_num(line, p1);
 887         if (!digits)
 888                 return -1;
 889
 890         offset += digits;
 891         line += digits;
 892         len -= digits;
 893
 894         *p2 = 1;
 895         if (*line == ',') {
 896                 digits = parse_num(line+1, p2);
 897                 if (!digits)
 898                         return -1;
 899
 900                 offset += digits+1;
 901                 line += digits+1;
 902                 len -= digits+1;
 903         }
 904
 905         ex = strlen(expect);
 906         if (ex > len)
 907                 return -1;
 908         if (memcmp(line, expect, ex))
 909                 return -1;
 910
 911         return offset + ex;
 912 }
 913
 914 static void recount_diff(char *line, int size, struct fragment *fragment)
 915 {
 916         int oldlines = 0, newlines = 0, ret = 0;
 917
 918         if (size < 1) {
 919                 warning("recount: ignore empty hunk");
 920                 return;
 921         }
 922
 923         for (;;) {
 924                 int len = linelen(line, size);
 925                 size -= len;
 926                 line += len;
 927
 928                 if (size < 1)
 929                         break;
 930
 931                 switch (*line) {
 932                 case ' ': case '\n':
 933                         newlines++;
 934                         /* fall through */
 935                 case '-':
 936                         oldlines++;
 937                         continue;
 938                 case '+':
 939                         newlines++;
 940                         continue;
 941                 case '\\':
 942                         continue;
 943                 case '@':
 944                         ret = size < 3 || prefixcmp(line, "@@ ");
 945                         break;
 946                 case 'd':
 947                         ret = size < 5 || prefixcmp(line, "diff ");
 948                         break;
 949                 default:
 950                         ret = -1;
 951                         break;
 952                 }
 953                 if (ret) {
 954                         warning("recount: unexpected line: %.*s",
 955                                 (int)linelen(line, size), line);
 956                         return;
 957                 }
 958                 break;
 959         }
 960         fragment->oldlines = oldlines;
 961         fragment->newlines = newlines;
 962 }
 963
 964 /*
 965  * Parse a unified diff fragment header of the
 966  * form "@@ -a,b +c,d @@"
 967  */
 968 static int parse_fragment_header(char *line, int len, struct fragment *fragment)
 969 {
 970         int offset;
 971
 972         if (!len || line[len-1] != '\n')
 973                 return -1;
 974
 975         /* Figure out the number of lines in a fragment */
 976         offset = parse_range(line, len, 4, " +", &fragment->oldpos, &fragment->oldlines);
 977         offset = parse_range(line, len, offset, " @@", &fragment->newpos, &fragment->newlines);
 978
 979         return offset;
 980 }
 981
 982 static int find_header(char *line, unsigned long size, int *hdrsize, struct patch *patch)
 983 {
 984         unsigned long offset, len;
 985
 986         patch->is_toplevel_relative = 0;
 987         patch->is_rename = patch->is_copy = 0;
 988         patch->is_new = patch->is_delete = -1;
 989         patch->old_mode = patch->new_mode = 0;
 990         patch->old_name = patch->new_name = NULL;
 991         for (offset = 0; size > 0; offset += len, size -= len, line += len, linenr++) {
 992                 unsigned long nextlen;
 993
 994                 len = linelen(line, size);
 995                 if (!len)
 996                         break;
 997
 998                 /* Testing this early allows us to take a few shortcuts.. */
 999                 if (len < 6)
1000                         continue;
1001
1002                 /*
1003                  * Make sure we don't find any unconnected patch fragments.
1004                  * That's a sign that we didn't find a header, and that a
1005                  * patch has become corrupted/broken up.
1006                  */
1007                 if (!memcmp("@@ -", line, 4)) {
1008                         struct fragment dummy;
1009                         if (parse_fragment_header(line, len, &dummy) < 0)
1010                                 continue;
1011                         die("patch fragment without header at line %d: %.*s",
1012                             linenr, (int)len-1, line);
1013                 }
1014
1015                 if (size < len + 6)
1016                         break;
1017
1018                 /*
1019                  * Git patch? It might not have a real patch, just a rename
1020                  * or mode change, so we handle that specially
1021                  */
1022                 if (!memcmp("diff --git ", line, 11)) {
1023                         int git_hdr_len = parse_git_header(line, len, size, patch);
1024                         if (git_hdr_len <= len)
1025                                 continue;
1026                         if (!patch->old_name && !patch->new_name) {
1027                                 if (!patch->def_name)
1028                                         die("git diff header lacks filename information (line %d)", linenr);
1029                                 patch->old_name = patch->new_name = patch->def_name;
1030                         }
1031                         patch->is_toplevel_relative = 1;
1032                         *hdrsize = git_hdr_len;
1033                         return offset;
1034                 }
1035
1036                 /* --- followed by +++ ? */
1037                 if (memcmp("--- ", line,  4) || memcmp("+++ ", line + len, 4))
1038                         continue;
1039
1040                 /*
1041                  * We only accept unified patches, so we want it to
1042                  * at least have "@@ -a,b +c,d @@\n", which is 14 chars
1043                  * minimum ("@@ -0,0 +1 @@\n" is the shortest).
1044                  */
1045                 nextlen = linelen(line + len, size - len);
1046                 if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
1047                         continue;
1048
1049                 /* Ok, we'll consider it a patch */
1050                 parse_traditional_patch(line, line+len, patch);
1051                 *hdrsize = len + nextlen;
1052                 linenr += 2;
1053                 return offset;
1054         }
1055         return -1;
1056 }
1057
1058 static void check_whitespace(const char *line, int len, unsigned ws_rule)
1059 {
1060         char *err;
1061         unsigned result = ws_check(line + 1, len - 1, ws_rule);
1062         if (!result)
1063                 return;
1064
1065         whitespace_error++;
1066         if (squelch_whitespace_errors &&
1067             squelch_whitespace_errors < whitespace_error)
1068                 ;
1069         else {
1070                 err = whitespace_error_string(result);
1071                 fprintf(stderr, "%s:%d: %s.\n%.*s\n",
1072                         patch_input_file, linenr, err, len - 2, line + 1);
1073                 free(err);
1074         }
1075 }
1076
1077 /*
1078  * Parse a unified diff. Note that this really needs to parse each
1079  * fragment separately, since the only way to know the difference
1080  * between a "---" that is part of a patch, and a "---" that starts
1081  * the next patch is to look at the line counts..
1082  */
1083 static int parse_fragment(char *line, unsigned long size,
1084                           struct patch *patch, struct fragment *fragment)
1085 {
1086         int added, deleted;
1087         int len = linelen(line, size), offset;
1088         unsigned long oldlines, newlines;
1089         unsigned long leading, trailing;
1090
1091         offset = parse_fragment_header(line, len, fragment);
1092         if (offset < 0)
1093                 return -1;
1094         if (offset > 0 && patch->recount)
1095                 recount_diff(line + offset, size - offset, fragment);
1096         oldlines = fragment->oldlines;
1097         newlines = fragment->newlines;
1098         leading = 0;
1099         trailing = 0;
1100
1101         /* Parse the thing.. */
1102         line += len;
1103         size -= len;
1104         linenr++;
1105         added = deleted = 0;
1106         for (offset = len;
1107              0 < size;
1108              offset += len, size -= len, line += len, linenr++) {
1109                 if (!oldlines && !newlines)
1110                         break;
1111                 len = linelen(line, size);
1112                 if (!len || line[len-1] != '\n')
1113                         return -1;
1114                 switch (*line) {
1115                 default:
1116                         return -1;
1117                 case '\n': /* newer GNU diff, an empty context line */
1118                 case ' ':
1119                         oldlines--;
1120                         newlines--;
1121                         if (!deleted && !added)
1122                                 leading++;
1123                         trailing++;
1124                         break;
1125                 case '-':
1126                         if (apply_in_reverse &&
1127                             ws_error_action != nowarn_ws_error)
1128                                 check_whitespace(line, len, patch->ws_rule);
1129                         deleted++;
1130                         oldlines--;
1131                         trailing = 0;
1132                         break;
1133                 case '+':
1134                         if (!apply_in_reverse &&
1135                             ws_error_action != nowarn_ws_error)
1136                                 check_whitespace(line, len, patch->ws_rule);
1137                         added++;
1138                         newlines--;
1139                         trailing = 0;
1140                         break;
1141
1142                 /*
1143                  * We allow "\ No newline at end of file". Depending
1144                  * on locale settings when the patch was produced we
1145                  * don't know what this line looks like. The only
1146                  * thing we do know is that it begins with "\ ".
1147                  * Checking for 12 is just for sanity check -- any
1148                  * l10n of "\ No newline..." is at least that long.
1149                  */
1150                 case '\\':
1151                         if (len < 12 || memcmp(line, "\\ ", 2))
1152                                 return -1;
1153                         break;
1154                 }
1155         }
1156         if (oldlines || newlines)
1157                 return -1;
1158         fragment->leading = leading;
1159         fragment->trailing = trailing;
1160
1161         /*
1162          * If a fragment ends with an incomplete line, we failed to include
1163          * it in the above loop because we hit oldlines == newlines == 0
1164          * before seeing it.
1165          */
1166         if (12 < size && !memcmp(line, "\\ ", 2))
1167                 offset += linelen(line, size);
1168
1169         patch->lines_added += added;
1170         patch->lines_deleted += deleted;
1171
1172         if (0 < patch->is_new && oldlines)
1173                 return error("new file depends on old contents");
1174         if (0 < patch->is_delete && newlines)
1175                 return error("deleted file still has contents");
1176         return offset;
1177 }
1178
1179 static int parse_single_patch(char *line, unsigned long size, struct patch *patch)
1180 {
1181         unsigned long offset = 0;
1182         unsigned long oldlines = 0, newlines = 0, context = 0;
1183         struct fragment **fragp = &patch->fragments;
1184
1185         while (size > 4 && !memcmp(line, "@@ -", 4)) {
1186                 struct fragment *fragment;
1187                 int len;
1188
1189                 fragment = xcalloc(1, sizeof(*fragment));
1190                 len = parse_fragment(line, size, patch, fragment);
1191                 if (len <= 0)
1192                         die("corrupt patch at line %d", linenr);
1193                 fragment->patch = line;
1194                 fragment->size = len;
1195                 oldlines += fragment->oldlines;
1196                 newlines += fragment->newlines;
1197                 context += fragment->leading + fragment->trailing;
1198
1199                 *fragp = fragment;
1200                 fragp = &fragment->next;
1201
1202                 offset += len;
1203                 line += len;
1204                 size -= len;
1205         }
1206
1207         /*
1208          * If something was removed (i.e. we have old-lines) it cannot
1209          * be creation, and if something was added it cannot be
1210          * deletion.  However, the reverse is not true; --unified=0
1211          * patches that only add are not necessarily creation even
1212          * though they do not have any old lines, and ones that only
1213          * delete are not necessarily deletion.
1214          *
1215          * Unfortunately, a real creation/deletion patch do _not_ have
1216          * any context line by definition, so we cannot safely tell it
1217          * apart with --unified=0 insanity.  At least if the patch has
1218          * more than one hunk it is not creation or deletion.
1219          */
1220         if (patch->is_new < 0 &&
1221             (oldlines || (patch->fragments && patch->fragments->next)))
1222                 patch->is_new = 0;
1223         if (patch->is_delete < 0 &&
1224             (newlines || (patch->fragments && patch->fragments->next)))
1225                 patch->is_delete = 0;
1226
1227         if (0 < patch->is_new && oldlines)
1228                 die("new file %s depends on old contents", patch->new_name);
1229         if (0 < patch->is_delete && newlines)
1230                 die("deleted file %s still has contents", patch->old_name);
1231         if (!patch->is_delete && !newlines && context)
1232                 fprintf(stderr, "** warning: file %s becomes empty but "
1233                         "is not deleted\n", patch->new_name);
1234
1235         return offset;
1236 }
1237
1238 static inline int metadata_changes(struct patch *patch)
1239 {
1240         return  patch->is_rename > 0 ||
1241                 patch->is_copy > 0 ||
1242                 patch->is_new > 0 ||
1243                 patch->is_delete ||
1244                 (patch->old_mode && patch->new_mode &&
1245                  patch->old_mode != patch->new_mode);
1246 }
1247
1248 static char *inflate_it(const void *data, unsigned long size,
1249                         unsigned long inflated_size)
1250 {
1251         z_stream stream;
1252         void *out;
1253         int st;
1254
1255         memset(&stream, 0, sizeof(stream));
1256
1257         stream.next_in = (unsigned char *)data;
1258         stream.avail_in = size;
1259         stream.next_out = out = xmalloc(inflated_size);
1260         stream.avail_out = inflated_size;
1261         git_inflate_init(&stream);
1262         st = git_inflate(&stream, Z_FINISH);
1263         git_inflate_end(&stream);
1264         if ((st != Z_STREAM_END) || stream.total_out != inflated_size) {
1265                 free(out);
1266                 return NULL;
1267         }
1268         return out;
1269 }
1270
1271 static struct fragment *parse_binary_hunk(char **buf_p,
1272                                           unsigned long *sz_p,
1273                                           int *status_p,
1274                                           int *used_p)
1275 {
1276         /*
1277          * Expect a line that begins with binary patch method ("literal"
1278          * or "delta"), followed by the length of data before deflating.
1279          * a sequence of 'length-byte' followed by base-85 encoded data
1280          * should follow, terminated by a newline.
1281          *
1282          * Each 5-byte sequence of base-85 encodes up to 4 bytes,
1283          * and we would limit the patch line to 66 characters,
1284          * so one line can fit up to 13 groups that would decode
1285          * to 52 bytes max.  The length byte 'A'-'Z' corresponds
1286          * to 1-26 bytes, and 'a'-'z' corresponds to 27-52 bytes.
1287          */
1288         int llen, used;
1289         unsigned long size = *sz_p;
1290         char *buffer = *buf_p;
1291         int patch_method;
1292         unsigned long origlen;
1293         char *data = NULL;
1294         int hunk_size = 0;
1295         struct fragment *frag;
1296
1297         llen = linelen(buffer, size);
1298         used = llen;
1299
1300         *status_p = 0;
1301
1302         if (!prefixcmp(buffer, "delta ")) {
1303                 patch_method = BINARY_DELTA_DEFLATED;
1304                 origlen = strtoul(buffer + 6, NULL, 10);
1305         }
1306         else if (!prefixcmp(buffer, "literal ")) {
1307                 patch_method = BINARY_LITERAL_DEFLATED;
1308                 origlen = strtoul(buffer + 8, NULL, 10);
1309         }
1310         else
1311                 return NULL;
1312
1313         linenr++;
1314         buffer += llen;
1315         while (1) {
1316                 int byte_length, max_byte_length, newsize;
1317                 llen = linelen(buffer, size);
1318                 used += llen;
1319                 linenr++;
1320                 if (llen == 1) {
1321                         /* consume the blank line */
1322                         buffer++;
1323                         size--;
1324                         break;
1325                 }
1326                 /*
1327                  * Minimum line is "A00000\n" which is 7-byte long,
1328                  * and the line length must be multiple of 5 plus 2.
1329                  */
1330                 if ((llen < 7) || (llen-2) % 5)
1331                         goto corrupt;
1332                 max_byte_length = (llen - 2) / 5 * 4;
1333                 byte_length = *buffer;
1334                 if ('A' <= byte_length && byte_length <= 'Z')
1335                         byte_length = byte_length - 'A' + 1;
1336                 else if ('a' <= byte_length && byte_length <= 'z')
1337                         byte_length = byte_length - 'a' + 27;
1338                 else
1339                         goto corrupt;
1340                 /* if the input length was not multiple of 4, we would
1341                  * have filler at the end but the filler should never
1342                  * exceed 3 bytes
1343                  */
1344                 if (max_byte_length < byte_length ||
1345                     byte_length <= max_byte_length - 4)
1346                         goto corrupt;
1347                 newsize = hunk_size + byte_length;
1348                 data = xrealloc(data, newsize);
1349                 if (decode_85(data + hunk_size, buffer + 1, byte_length))
1350                         goto corrupt;
1351                 hunk_size = newsize;
1352                 buffer += llen;
1353                 size -= llen;
1354         }
1355
1356         frag = xcalloc(1, sizeof(*frag));
1357         frag->patch = inflate_it(data, hunk_size, origlen);
1358         if (!frag->patch)
1359                 goto corrupt;
1360         free(data);
1361         frag->size = origlen;
1362         *buf_p = buffer;
1363         *sz_p = size;
1364         *used_p = used;
1365         frag->binary_patch_method = patch_method;
1366         return frag;
1367
1368  corrupt:
1369         free(data);
1370         *status_p = -1;
1371         error("corrupt binary patch at line %d: %.*s",
1372               linenr-1, llen-1, buffer);
1373         return NULL;
1374 }
1375
1376 static int parse_binary(char *buffer, unsigned long size, struct patch *patch)
1377 {
1378         /*
1379          * We have read "GIT binary patch\n"; what follows is a line
1380          * that says the patch method (currently, either "literal" or
1381          * "delta") and the length of data before deflating; a
1382          * sequence of 'length-byte' followed by base-85 encoded data
1383          * follows.
1384          *
1385          * When a binary patch is reversible, there is another binary
1386          * hunk in the same format, starting with patch method (either
1387          * "literal" or "delta") with the length of data, and a sequence
1388          * of length-byte + base-85 encoded data, terminated with another
1389          * empty line.  This data, when applied to the postimage, produces
1390          * the preimage.
1391          */
1392         struct fragment *forward;
1393         struct fragment *reverse;
1394         int status;
1395         int used, used_1;
1396
1397         forward = parse_binary_hunk(&buffer, &size, &status, &used);
1398         if (!forward && !status)
1399                 /* there has to be one hunk (forward hunk) */
1400                 return error("unrecognized binary patch at line %d", linenr-1);
1401         if (status)
1402                 /* otherwise we already gave an error message */
1403                 return status;
1404
1405         reverse = parse_binary_hunk(&buffer, &size, &status, &used_1);
1406         if (reverse)
1407                 used += used_1;
1408         else if (status) {
1409                 /*
1410                  * Not having reverse hunk is not an error, but having
1411                  * a corrupt reverse hunk is.
1412                  */
1413                 free((void*) forward->patch);
1414                 free(forward);
1415                 return status;
1416         }
1417         forward->next = reverse;
1418         patch->fragments = forward;
1419         patch->is_binary = 1;
1420         return used;
1421 }
1422
1423 static int parse_chunk(char *buffer, unsigned long size, struct patch *patch)
1424 {
1425         int hdrsize, patchsize;
1426         int offset = find_header(buffer, size, &hdrsize, patch);
1427
1428         if (offset < 0)
1429                 return offset;
1430
1431         patch->ws_rule = whitespace_rule(patch->new_name
1432                                          ? patch->new_name
1433                                          : patch->old_name);
1434
1435         patchsize = parse_single_patch(buffer + offset + hdrsize,
1436                                        size - offset - hdrsize, patch);
1437
1438         if (!patchsize) {
1439                 static const char *binhdr[] = {
1440                         "Binary files ",
1441                         "Files ",
1442                         NULL,
1443                 };
1444                 static const char git_binary[] = "GIT binary patch\n";
1445                 int i;
1446                 int hd = hdrsize + offset;
1447                 unsigned long llen = linelen(buffer + hd, size - hd);
1448
1449                 if (llen == sizeof(git_binary) - 1 &&
1450                     !memcmp(git_binary, buffer + hd, llen)) {
1451                         int used;
1452                         linenr++;
1453                         used = parse_binary(buffer + hd + llen,
1454                                             size - hd - llen, patch);
1455                         if (used)
1456                                 patchsize = used + llen;
1457                         else
1458                                 patchsize = 0;
1459                 }
1460                 else if (!memcmp(" differ\n", buffer + hd + llen - 8, 8)) {
1461                         for (i = 0; binhdr[i]; i++) {
1462                                 int len = strlen(binhdr[i]);
1463                                 if (len < size - hd &&
1464                                     !memcmp(binhdr[i], buffer + hd, len)) {
1465                                         linenr++;
1466                                         patch->is_binary = 1;
1467                                         patchsize = llen;
1468                                         break;
1469                                 }
1470                         }
1471                 }
1472
1473                 /* Empty patch cannot be applied if it is a text patch
1474                  * without metadata change.  A binary patch appears
1475                  * empty to us here.
1476                  */
1477                 if ((apply || check) &&
1478                     (!patch->is_binary && !metadata_changes(patch)))
1479                         die("patch with only garbage at line %d", linenr);
1480         }
1481
1482         return offset + hdrsize + patchsize;
1483 }
1484
1485 #define swap(a,b) myswap((a),(b),sizeof(a))
1486
1487 #define myswap(a, b, size) do {         \
1488         unsigned char mytmp[size];      \
1489         memcpy(mytmp, &a, size);                \
1490         memcpy(&a, &b, size);           \
1491         memcpy(&b, mytmp, size);                \
1492 } while (0)
1493
1494 static void reverse_patches(struct patch *p)
1495 {
1496         for (; p; p = p->next) {
1497                 struct fragment *frag = p->fragments;
1498
1499                 swap(p->new_name, p->old_name);
1500                 swap(p->new_mode, p->old_mode);
1501                 swap(p->is_new, p->is_delete);
1502                 swap(p->lines_added, p->lines_deleted);
1503                 swap(p->old_sha1_prefix, p->new_sha1_prefix);
1504
1505                 for (; frag; frag = frag->next) {
1506                         swap(frag->newpos, frag->oldpos);
1507                         swap(frag->newlines, frag->oldlines);
1508                 }
1509         }
1510 }
1511
1512 static const char pluses[] =
1513 "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
1514 static const char minuses[]=
1515 "----------------------------------------------------------------------";
1516
1517 static void show_stats(struct patch *patch)
1518 {
1519         struct strbuf qname;
1520         char *cp = patch->new_name ? patch->new_name : patch->old_name;
1521         int max, add, del;
1522
1523         strbuf_init(&qname, 0);
1524         quote_c_style(cp, &qname, NULL, 0);
1525
1526         /*
1527          * "scale" the filename
1528          */
1529         max = max_len;
1530         if (max > 50)
1531                 max = 50;
1532
1533         if (qname.len > max) {
1534                 cp = strchr(qname.buf + qname.len + 3 - max, '/');
1535                 if (!cp)
1536                         cp = qname.buf + qname.len + 3 - max;
1537                 strbuf_splice(&qname, 0, cp - qname.buf, "...", 3);
1538         }
1539
1540         if (patch->is_binary) {
1541                 printf(" %-*s |  Bin\n", max, qname.buf);
1542                 strbuf_release(&qname);
1543                 return;
1544         }
1545
1546         printf(" %-*s |", max, qname.buf);
1547         strbuf_release(&qname);
1548
1549         /*
1550          * scale the add/delete
1551          */
1552         max = max + max_change > 70 ? 70 - max : max_change;
1553         add = patch->lines_added;
1554         del = patch->lines_deleted;
1555
1556         if (max_change > 0) {
1557                 int total = ((add + del) * max + max_change / 2) / max_change;
1558                 add = (add * max + max_change / 2) / max_change;
1559                 del = total - add;
1560         }
1561         printf("%5d %.*s%.*s\n", patch->lines_added + patch->lines_deleted,
1562                 add, pluses, del, minuses);
1563 }
1564
1565 static int read_old_data(struct stat *st, const char *path, struct strbuf *buf)
1566 {
1567         switch (st->st_mode & S_IFMT) {
1568         case S_IFLNK:
1569                 strbuf_grow(buf, st->st_size);
1570                 if (readlink(path, buf->buf, st->st_size) != st->st_size)
1571                         return -1;
1572                 strbuf_setlen(buf, st->st_size);
1573                 return 0;
1574         case S_IFREG:
1575                 if (strbuf_read_file(buf, path, st->st_size) != st->st_size)
1576                         return error("unable to open or read %s", path);
1577                 convert_to_git(path, buf->buf, buf->len, buf, 0);
1578                 return 0;
1579         default:
1580                 return -1;
1581         }
1582 }
1583
1584 static void update_pre_post_images(struct image *preimage,
1585                                    struct image *postimage,
1586                                    char *buf,
1587                                    size_t len)
1588 {
1589         int i, ctx;
1590         char *new, *old, *fixed;
1591         struct image fixed_preimage;
1592
1593         /*
1594          * Update the preimage with whitespace fixes.  Note that we
1595          * are not losing preimage->buf -- apply_one_fragment() will
1596          * free "oldlines".
1597          */
1598         prepare_image(&fixed_preimage, buf, len, 1);
1599         assert(fixed_preimage.nr == preimage->nr);
1600         for (i = 0; i < preimage->nr; i++)
1601                 fixed_preimage.line[i].flag = preimage->line[i].flag;
1602         free(preimage->line_allocated);
1603         *preimage = fixed_preimage;
1604
1605         /*
1606          * Adjust the common context lines in postimage, in place.
1607          * This is possible because whitespace fixing does not make
1608          * the string grow.
1609          */
1610         new = old = postimage->buf;
1611         fixed = preimage->buf;
1612         for (i = ctx = 0; i < postimage->nr; i++) {
1613                 size_t len = postimage->line[i].len;
1614                 if (!(postimage->line[i].flag & LINE_COMMON)) {
1615                         /* an added line -- no counterparts in preimage */
1616                         memmove(new, old, len);
1617                         old += len;
1618                         new += len;
1619                         continue;
1620                 }
1621
1622                 /* a common context -- skip it in the original postimage */
1623                 old += len;
1624
1625                 /* and find the corresponding one in the fixed preimage */
1626                 while (ctx < preimage->nr &&
1627                        !(preimage->line[ctx].flag & LINE_COMMON)) {
1628                         fixed += preimage->line[ctx].len;
1629                         ctx++;
1630                 }
1631                 if (preimage->nr <= ctx)
1632                         die("oops");
1633
1634                 /* and copy it in, while fixing the line length */
1635                 len = preimage->line[ctx].len;
1636                 memcpy(new, fixed, len);
1637                 new += len;
1638                 fixed += len;
1639                 postimage->line[i].len = len;
1640                 ctx++;
1641         }
1642
1643         /* Fix the length of the whole thing */
1644         postimage->len = new - postimage->buf;
1645 }
1646
1647 static int match_fragment(struct image *img,
1648                           struct image *preimage,
1649                           struct image *postimage,
1650                           unsigned long try,
1651                           int try_lno,
1652                           unsigned ws_rule,
1653                           int match_beginning, int match_end)
1654 {
1655         int i;
1656         char *fixed_buf, *buf, *orig, *target;
1657
1658         if (preimage->nr + try_lno > img->nr)
1659                 return 0;
1660
1661         if (match_beginning && try_lno)
1662                 return 0;
1663
1664         if (match_end && preimage->nr + try_lno != img->nr)
1665                 return 0;
1666
1667         /* Quick hash check */
1668         for (i = 0; i < preimage->nr; i++)
1669                 if (preimage->line[i].hash != img->line[try_lno + i].hash)
1670                         return 0;
1671
1672         /*
1673          * Do we have an exact match?  If we were told to match
1674          * at the end, size must be exactly at try+fragsize,
1675          * otherwise try+fragsize must be still within the preimage,
1676          * and either case, the old piece should match the preimage
1677          * exactly.
1678          */
1679         if ((match_end
1680              ? (try + preimage->len == img->len)
1681              : (try + preimage->len <= img->len)) &&
1682             !memcmp(img->buf + try, preimage->buf, preimage->len))
1683                 return 1;
1684
1685         if (ws_error_action != correct_ws_error)
1686                 return 0;
1687
1688         /*
1689          * The hunk does not apply byte-by-byte, but the hash says
1690          * it might with whitespace fuzz.
1691          */
1692         fixed_buf = xmalloc(preimage->len + 1);
1693         buf = fixed_buf;
1694         orig = preimage->buf;
1695         target = img->buf + try;
1696         for (i = 0; i < preimage->nr; i++) {
1697                 size_t fixlen; /* length after fixing the preimage */
1698                 size_t oldlen = preimage->line[i].len;
1699                 size_t tgtlen = img->line[try_lno + i].len;
1700                 size_t tgtfixlen; /* length after fixing the target line */
1701                 char tgtfixbuf[1024], *tgtfix;
1702                 int match;
1703
1704                 /* Try fixing the line in the preimage */
1705                 fixlen = ws_fix_copy(buf, orig, oldlen, ws_rule, NULL);
1706
1707                 /* Try fixing the line in the target */
1708                 if (sizeof(tgtfixbuf) > tgtlen)
1709                         tgtfix = tgtfixbuf;
1710                 else
1711                         tgtfix = xmalloc(tgtlen);
1712                 tgtfixlen = ws_fix_copy(tgtfix, target, tgtlen, ws_rule, NULL);
1713
1714                 /*
1715                  * If they match, either the preimage was based on
1716                  * a version before our tree fixed whitespace breakage,
1717                  * or we are lacking a whitespace-fix patch the tree
1718                  * the preimage was based on already had (i.e. target
1719                  * has whitespace breakage, the preimage doesn't).
1720                  * In either case, we are fixing the whitespace breakages
1721                  * so we might as well take the fix together with their
1722                  * real change.
1723                  */
1724                 match = (tgtfixlen == fixlen && !memcmp(tgtfix, buf, fixlen));
1725
1726                 if (tgtfix != tgtfixbuf)
1727                         free(tgtfix);
1728                 if (!match)
1729                         goto unmatch_exit;
1730
1731                 orig += oldlen;
1732                 buf += fixlen;
1733                 target += tgtlen;
1734         }
1735
1736         /*
1737          * Yes, the preimage is based on an older version that still
1738          * has whitespace breakages unfixed, and fixing them makes the
1739          * hunk match.  Update the context lines in the postimage.
1740          */
1741         update_pre_post_images(preimage, postimage,
1742                                fixed_buf, buf - fixed_buf);
1743         return 1;
1744
1745  unmatch_exit:
1746         free(fixed_buf);
1747         return 0;
1748 }
1749
1750 static int find_pos(struct image *img,
1751                     struct image *preimage,
1752                     struct image *postimage,
1753                     int line,
1754                     unsigned ws_rule,
1755                     int match_beginning, int match_end)
1756 {
1757         int i;
1758         unsigned long backwards, forwards, try;
1759         int backwards_lno, forwards_lno, try_lno;
1760
1761         if (preimage->nr > img->nr)
1762                 return -1;
1763
1764         /*
1765          * If match_begining or match_end is specified, there is no
1766          * point starting from a wrong line that will never match and
1767          * wander around and wait for a match at the specified end.
1768          */
1769         if (match_beginning)
1770                 line = 0;
1771         else if (match_end)
1772                 line = img->nr - preimage->nr;
1773
1774         if (line > img->nr)
1775                 line = img->nr;
1776
1777         try = 0;
1778         for (i = 0; i < line; i++)
1779                 try += img->line[i].len;
1780
1781         /*
1782          * There's probably some smart way to do this, but I'll leave
1783          * that to the smart and beautiful people. I'm simple and stupid.
1784          */
1785         backwards = try;
1786         backwards_lno = line;
1787         forwards = try;
1788         forwards_lno = line;
1789         try_lno = line;
1790
1791         for (i = 0; ; i++) {
1792                 if (match_fragment(img, preimage, postimage,
1793                                    try, try_lno, ws_rule,
1794                                    match_beginning, match_end))
1795                         return try_lno;
1796
1797         again:
1798                 if (backwards_lno == 0 && forwards_lno == img->nr)
1799                         break;
1800
1801                 if (i & 1) {
1802                         if (backwards_lno == 0) {
1803                                 i++;
1804                                 goto again;
1805                         }
1806                         backwards_lno--;
1807                         backwards -= img->line[backwards_lno].len;
1808                         try = backwards;
1809                         try_lno = backwards_lno;
1810                 } else {
1811                         if (forwards_lno == img->nr) {
1812                                 i++;
1813                                 goto again;
1814                         }
1815                         forwards += img->line[forwards_lno].len;
1816                         forwards_lno++;
1817                         try = forwards;
1818                         try_lno = forwards_lno;
1819                 }
1820
1821         }
1822         return -1;
1823 }
1824
1825 static void remove_first_line(struct image *img)
1826 {
1827         img->buf += img->line[0].len;
1828         img->len -= img->line[0].len;
1829         img->line++;
1830         img->nr--;
1831 }
1832
1833 static void remove_last_line(struct image *img)
1834 {
1835         img->len -= img->line[--img->nr].len;
1836 }
1837
1838 static void update_image(struct image *img,
1839                          int applied_pos,
1840                          struct image *preimage,
1841                          struct image *postimage)
1842 {
1843         /*
1844          * remove the copy of preimage at offset in img
1845          * and replace it with postimage
1846          */
1847         int i, nr;
1848         size_t remove_count, insert_count, applied_at = 0;
1849         char *result;
1850
1851         for (i = 0; i < applied_pos; i++)
1852                 applied_at += img->line[i].len;
1853
1854         remove_count = 0;
1855         for (i = 0; i < preimage->nr; i++)
1856                 remove_count += img->line[applied_pos + i].len;
1857         insert_count = postimage->len;
1858
1859         /* Adjust the contents */
1860         result = xmalloc(img->len + insert_count - remove_count + 1);
1861         memcpy(result, img->buf, applied_at);
1862         memcpy(result + applied_at, postimage->buf, postimage->len);
1863         memcpy(result + applied_at + postimage->len,
1864                img->buf + (applied_at + remove_count),
1865                img->len - (applied_at + remove_count));
1866         free(img->buf);
1867         img->buf = result;
1868         img->len += insert_count - remove_count;
1869         result[img->len] = '\0';
1870
1871         /* Adjust the line table */
1872         nr = img->nr + postimage->nr - preimage->nr;
1873         if (preimage->nr < postimage->nr) {
1874                 /*
1875                  * NOTE: this knows that we never call remove_first_line()
1876                  * on anything other than pre/post image.
1877                  */
1878                 img->line = xrealloc(img->line, nr * sizeof(*img->line));
1879                 img->line_allocated = img->line;
1880         }
1881         if (preimage->nr != postimage->nr)
1882                 memmove(img->line + applied_pos + postimage->nr,
1883                         img->line + applied_pos + preimage->nr,
1884                         (img->nr - (applied_pos + preimage->nr)) *
1885                         sizeof(*img->line));
1886         memcpy(img->line + applied_pos,
1887                postimage->line,
1888                postimage->nr * sizeof(*img->line));
1889         img->nr = nr;
1890 }
1891
1892 static int apply_one_fragment(struct image *img, struct fragment *frag,
1893                               int inaccurate_eof, unsigned ws_rule)
1894 {
1895         int match_beginning, match_end;
1896         const char *patch = frag->patch;
1897         int size = frag->size;
1898         char *old, *new, *oldlines, *newlines;
1899         int new_blank_lines_at_end = 0;
1900         unsigned long leading, trailing;
1901         int pos, applied_pos;
1902         struct image preimage;
1903         struct image postimage;
1904
1905         memset(&preimage, 0, sizeof(preimage));
1906         memset(&postimage, 0, sizeof(postimage));
1907         oldlines = xmalloc(size);
1908         newlines = xmalloc(size);
1909
1910         old = oldlines;
1911         new = newlines;
1912         while (size > 0) {
1913                 char first;
1914                 int len = linelen(patch, size);
1915                 int plen, added;
1916                 int added_blank_line = 0;
1917
1918                 if (!len)
1919                         break;
1920
1921                 /*
1922                  * "plen" is how much of the line we should use for
1923                  * the actual patch data. Normally we just remove the
1924                  * first character on the line, but if the line is
1925                  * followed by "\ No newline", then we also remove the
1926                  * last one (which is the newline, of course).
1927                  */
1928                 plen = len - 1;
1929                 if (len < size && patch[len] == '\\')
1930                         plen--;
1931                 first = *patch;
1932                 if (apply_in_reverse) {
1933                         if (first == '-')
1934                                 first = '+';
1935                         else if (first == '+')
1936                                 first = '-';
1937                 }
1938
1939                 switch (first) {
1940                 case '\n':
1941                         /* Newer GNU diff, empty context line */
1942                         if (plen < 0)
1943                                 /* ... followed by '\No newline'; nothing */
1944                                 break;
1945                         *old++ = '\n';
1946                         *new++ = '\n';
1947                         add_line_info(&preimage, "\n", 1, LINE_COMMON);
1948                         add_line_info(&postimage, "\n", 1, LINE_COMMON);
1949                         break;
1950                 case ' ':
1951                 case '-':
1952                         memcpy(old, patch + 1, plen);
1953                         add_line_info(&preimage, old, plen,
1954                                       (first == ' ' ? LINE_COMMON : 0));
1955                         old += plen;
1956                         if (first == '-')
1957                                 break;
1958                 /* Fall-through for ' ' */
1959                 case '+':
1960                         /* --no-add does not add new lines */
1961                         if (first == '+' && no_add)
1962                                 break;
1963
1964                         if (first != '+' ||
1965                             !whitespace_error ||
1966                             ws_error_action != correct_ws_error) {
1967                                 memcpy(new, patch + 1, plen);
1968                                 added = plen;
1969                         }
1970                         else {
1971                                 added = ws_fix_copy(new, patch + 1, plen, ws_rule, &applied_after_fixing_ws);
1972                         }
1973                         add_line_info(&postimage, new, added,
1974                                       (first == '+' ? 0 : LINE_COMMON));
1975                         new += added;
1976                         if (first == '+' &&
1977                             added == 1 && new[-1] == '\n')
1978                                 added_blank_line = 1;
1979                         break;
1980                 case '@': case '\\':
1981                         /* Ignore it, we already handled it */
1982                         break;
1983                 default:
1984                         if (apply_verbosely)
1985                                 error("invalid start of line: '%c'", first);
1986                         return -1;
1987                 }
1988                 if (added_blank_line)
1989                         new_blank_lines_at_end++;
1990                 else
1991                         new_blank_lines_at_end = 0;
1992                 patch += len;
1993                 size -= len;
1994         }
1995         if (inaccurate_eof &&
1996             old > oldlines && old[-1] == '\n' &&
1997             new > newlines && new[-1] == '\n') {
1998                 old--;
1999                 new--;
2000         }
2001
2002         leading = frag->leading;
2003         trailing = frag->trailing;
2004
2005         /*
2006          * A hunk to change lines at the beginning would begin with
2007          * @@ -1,L +N,M @@
2008          * but we need to be careful.  -U0 that inserts before the second
2009          * line also has this pattern.
2010          *
2011          * And a hunk to add to an empty file would begin with
2012          * @@ -0,0 +N,M @@
2013          *
2014          * In other words, a hunk that is (frag->oldpos <= 1) with or
2015          * without leading context must match at the beginning.
2016          */
2017         match_beginning = (!frag->oldpos ||
2018                            (frag->oldpos == 1 && !unidiff_zero));
2019
2020         /*
2021          * A hunk without trailing lines must match at the end.
2022          * However, we simply cannot tell if a hunk must match end
2023          * from the lack of trailing lines if the patch was generated
2024          * with unidiff without any context.
2025          */
2026         match_end = !unidiff_zero && !trailing;
2027
2028         pos = frag->newpos ? (frag->newpos - 1) : 0;
2029         preimage.buf = oldlines;
2030         preimage.len = old - oldlines;
2031         postimage.buf = newlines;
2032         postimage.len = new - newlines;
2033         preimage.line = preimage.line_allocated;
2034         postimage.line = postimage.line_allocated;
2035
2036         for (;;) {
2037
2038                 applied_pos = find_pos(img, &preimage, &postimage, pos,
2039                                        ws_rule, match_beginning, match_end);
2040
2041                 if (applied_pos >= 0)
2042                         break;
2043
2044                 /* Am I at my context limits? */
2045                 if ((leading <= p_context) && (trailing <= p_context))
2046                         break;
2047                 if (match_beginning || match_end) {
2048                         match_beginning = match_end = 0;
2049                         continue;
2050                 }
2051
2052                 /*
2053                  * Reduce the number of context lines; reduce both
2054                  * leading and trailing if they are equal otherwise
2055                  * just reduce the larger context.
2056                  */
2057                 if (leading >= trailing) {
2058                         remove_first_line(&preimage);
2059                         remove_first_line(&postimage);
2060                         pos--;
2061                         leading--;
2062                 }
2063                 if (trailing > leading) {
2064                         remove_last_line(&preimage);
2065                         remove_last_line(&postimage);
2066                         trailing--;
2067                 }
2068         }
2069
2070         if (applied_pos >= 0) {
2071                 if (ws_error_action == correct_ws_error &&
2072                     new_blank_lines_at_end &&
2073                     postimage.nr + applied_pos == img->nr) {
2074                         /*
2075                          * If the patch application adds blank lines
2076                          * at the end, and if the patch applies at the
2077                          * end of the image, remove those added blank
2078                          * lines.
2079                          */
2080                         while (new_blank_lines_at_end--)
2081                                 remove_last_line(&postimage);
2082                 }
2083
2084                 /*
2085                  * Warn if it was necessary to reduce the number
2086                  * of context lines.
2087                  */
2088                 if ((leading != frag->leading) ||
2089                     (trailing != frag->trailing))
2090                         fprintf(stderr, "Context reduced to (%ld/%ld)"
2091                                 " to apply fragment at %d\n",
2092                                 leading, trailing, applied_pos+1);
2093                 update_image(img, applied_pos, &preimage, &postimage);
2094         } else {
2095                 if (apply_verbosely)
2096                         error("while searching for:\n%.*s",
2097                               (int)(old - oldlines), oldlines);
2098         }
2099
2100         free(oldlines);
2101         free(newlines);
2102         free(preimage.line_allocated);
2103         free(postimage.line_allocated);
2104
2105         return (applied_pos < 0);
2106 }
2107
2108 static int apply_binary_fragment(struct image *img, struct patch *patch)
2109 {
2110         struct fragment *fragment = patch->fragments;
2111         unsigned long len;
2112         void *dst;
2113
2114         /* Binary patch is irreversible without the optional second hunk */
2115         if (apply_in_reverse) {
2116                 if (!fragment->next)
2117                         return error("cannot reverse-apply a binary patch "
2118                                      "without the reverse hunk to '%s'",
2119                                      patch->new_name
2120                                      ? patch->new_name : patch->old_name);
2121                 fragment = fragment->next;
2122         }
2123         switch (fragment->binary_patch_method) {
2124         case BINARY_DELTA_DEFLATED:
2125                 dst = patch_delta(img->buf, img->len, fragment->patch,
2126                                   fragment->size, &len);
2127                 if (!dst)
2128                         return -1;
2129                 clear_image(img);
2130                 img->buf = dst;
2131                 img->len = len;
2132                 return 0;
2133         case BINARY_LITERAL_DEFLATED:
2134                 clear_image(img);
2135                 img->len = fragment->size;
2136                 img->buf = xmalloc(img->len+1);
2137                 memcpy(img->buf, fragment->patch, img->len);
2138                 img->buf[img->len] = '\0';
2139                 return 0;
2140         }
2141         return -1;
2142 }
2143
2144 static int apply_binary(struct image *img, struct patch *patch)
2145 {
2146         const char *name = patch->old_name ? patch->old_name : patch->new_name;
2147         unsigned char sha1[20];
2148
2149         /*
2150          * For safety, we require patch index line to contain
2151          * full 40-byte textual SHA1 for old and new, at least for now.
2152          */
2153         if (strlen(patch->old_sha1_prefix) != 40 ||
2154             strlen(patch->new_sha1_prefix) != 40 ||
2155             get_sha1_hex(patch->old_sha1_prefix, sha1) ||
2156             get_sha1_hex(patch->new_sha1_prefix, sha1))
2157                 return error("cannot apply binary patch to '%s' "
2158                              "without full index line", name);
2159
2160         if (patch->old_name) {
2161                 /*
2162                  * See if the old one matches what the patch
2163                  * applies to.
2164                  */
2165                 hash_sha1_file(img->buf, img->len, blob_type, sha1);
2166                 if (strcmp(sha1_to_hex(sha1), patch->old_sha1_prefix))
2167                         return error("the patch applies to '%s' (%s), "
2168                                      "which does not match the "
2169                                      "current contents.",
2170                                      name, sha1_to_hex(sha1));
2171         }
2172         else {
2173                 /* Otherwise, the old one must be empty. */
2174                 if (img->len)
2175                         return error("the patch applies to an empty "
2176                                      "'%s' but it is not empty", name);
2177         }
2178
2179         get_sha1_hex(patch->new_sha1_prefix, sha1);
2180         if (is_null_sha1(sha1)) {
2181                 clear_image(img);
2182                 return 0; /* deletion patch */
2183         }
2184
2185         if (has_sha1_file(sha1)) {
2186                 /* We already have the postimage */
2187                 enum object_type type;
2188                 unsigned long size;
2189                 char *result;
2190
2191                 result = read_sha1_file(sha1, &type, &size);
2192                 if (!result)
2193                         return error("the necessary postimage %s for "
2194                                      "'%s' cannot be read",
2195                                      patch->new_sha1_prefix, name);
2196                 clear_image(img);
2197                 img->buf = result;
2198                 img->len = size;
2199         } else {
2200                 /*
2201                  * We have verified buf matches the preimage;
2202                  * apply the patch data to it, which is stored
2203                  * in the patch->fragments->{patch,size}.
2204                  */
2205                 if (apply_binary_fragment(img, patch))
2206                         return error("binary patch does not apply to '%s'",
2207                                      name);
2208
2209                 /* verify that the result matches */
2210                 hash_sha1_file(img->buf, img->len, blob_type, sha1);
2211                 if (strcmp(sha1_to_hex(sha1), patch->new_sha1_prefix))
2212                         return error("binary patch to '%s' creates incorrect result (expecting %s, got %s)",
2213                                 name, patch->new_sha1_prefix, sha1_to_hex(sha1));
2214         }
2215
2216         return 0;
2217 }
2218
2219 static int apply_fragments(struct image *img, struct patch *patch)
2220 {
2221         struct fragment *frag = patch->fragments;
2222         const char *name = patch->old_name ? patch->old_name : patch->new_name;
2223         unsigned ws_rule = patch->ws_rule;
2224         unsigned inaccurate_eof = patch->inaccurate_eof;
2225
2226         if (patch->is_binary)
2227                 return apply_binary(img, patch);
2228
2229         while (frag) {
2230                 if (apply_one_fragment(img, frag, inaccurate_eof, ws_rule)) {
2231                         error("patch failed: %s:%ld", name, frag->oldpos);
2232                         if (!apply_with_reject)
2233                                 return -1;
2234                         frag->rejected = 1;
2235                 }
2236                 frag = frag->next;
2237         }
2238         return 0;
2239 }
2240
2241 static int read_file_or_gitlink(struct cache_entry *ce, struct strbuf *buf)
2242 {
2243         if (!ce)
2244                 return 0;
2245
2246         if (S_ISGITLINK(ce->ce_mode)) {
2247                 strbuf_grow(buf, 100);
2248                 strbuf_addf(buf, "Subproject commit %s\n", sha1_to_hex(ce->sha1));
2249         } else {
2250                 enum object_type type;
2251                 unsigned long sz;
2252                 char *result;
2253
2254                 result = read_sha1_file(ce->sha1, &type, &sz);
2255                 if (!result)
2256                         return -1;
2257                 /* XXX read_sha1_file NUL-terminates */
2258                 strbuf_attach(buf, result, sz, sz + 1);
2259         }
2260         return 0;
2261 }
2262
2263 static struct patch *in_fn_table(const char *name)
2264 {
2265         struct string_list_item *item;
2266
2267         if (name == NULL)
2268                 return NULL;
2269
2270         item = string_list_lookup(name, &fn_table);
2271         if (item != NULL)
2272                 return (struct patch *)item->util;
2273
2274         return NULL;
2275 }
2276
2277 static void add_to_fn_table(struct patch *patch)
2278 {
2279         struct string_list_item *item;
2280
2281         /*
2282          * Always add new_name unless patch is a deletion
2283          * This should cover the cases for normal diffs,
2284          * file creations and copies
2285          */
2286         if (patch->new_name != NULL) {
2287                 item = string_list_insert(patch->new_name, &fn_table);
2288                 item->util = patch;
2289         }
2290
2291         /*
2292          * store a failure on rename/deletion cases because
2293          * later chunks shouldn't patch old names
2294          */
2295         if ((patch->new_name == NULL) || (patch->is_rename)) {
2296                 item = string_list_insert(patch->old_name, &fn_table);
2297                 item->util = (struct patch *) -1;
2298         }
2299 }
2300
2301 static int apply_data(struct patch *patch, struct stat *st, struct cache_entry *ce)
2302 {
2303         struct strbuf buf;
2304         struct image image;
2305         size_t len;
2306         char *img;
2307         struct patch *tpatch;
2308
2309         strbuf_init(&buf, 0);
2310
2311         if (!(patch->is_copy || patch->is_rename) &&
2312             ((tpatch = in_fn_table(patch->old_name)) != NULL)) {
2313                 if (tpatch == (struct patch *) -1) {
2314                         return error("patch %s has been renamed/deleted",
2315                                 patch->old_name);
2316                 }
2317                 /* We have a patched copy in memory use that */
2318                 strbuf_add(&buf, tpatch->result, tpatch->resultsize);
2319         } else if (cached) {
2320                 if (read_file_or_gitlink(ce, &buf))
2321                         return error("read of %s failed", patch->old_name);
2322         } else if (patch->old_name) {
2323                 if (S_ISGITLINK(patch->old_mode)) {
2324                         if (ce) {
2325                                 read_file_or_gitlink(ce, &buf);
2326                         } else {
2327                                 /*
2328                                  * There is no way to apply subproject
2329                                  * patch without looking at the index.
2330                                  */
2331                                 patch->fragments = NULL;
2332                         }
2333                 } else {
2334                         if (read_old_data(st, patch->old_name, &buf))
2335                                 return error("read of %s failed", patch->old_name);
2336                 }
2337         }
2338
2339         img = strbuf_detach(&buf, &len);
2340         prepare_image(&image, img, len, !patch->is_binary);
2341
2342         if (apply_fragments(&image, patch) < 0)
2343                 return -1; /* note with --reject this succeeds. */
2344         patch->result = image.buf;
2345         patch->resultsize = image.len;
2346         add_to_fn_table(patch);
2347         free(image.line_allocated);
2348
2349         if (0 < patch->is_delete && patch->resultsize)
2350                 return error("removal patch leaves file contents");
2351
2352         return 0;
2353 }
2354
2355 static int check_to_create_blob(const char *new_name, int ok_if_exists)
2356 {
2357         struct stat nst;
2358         if (!lstat(new_name, &nst)) {
2359                 if (S_ISDIR(nst.st_mode) || ok_if_exists)
2360                         return 0;
2361                 /*
2362                  * A leading component of new_name might be a symlink
2363                  * that is going to be removed with this patch, but
2364                  * still pointing at somewhere that has the path.
2365                  * In such a case, path "new_name" does not exist as
2366                  * far as git is concerned.
2367                  */
2368                 if (has_symlink_leading_path(strlen(new_name), new_name))
2369                         return 0;
2370
2371                 return error("%s: already exists in working directory", new_name);
2372         }
2373         else if ((errno != ENOENT) && (errno != ENOTDIR))
2374                 return error("%s: %s", new_name, strerror(errno));
2375         return 0;
2376 }
2377
2378 static int verify_index_match(struct cache_entry *ce, struct stat *st)
2379 {
2380         if (S_ISGITLINK(ce->ce_mode)) {
2381                 if (!S_ISDIR(st->st_mode))
2382                         return -1;
2383                 return 0;
2384         }
2385         return ce_match_stat(ce, st, CE_MATCH_IGNORE_VALID);
2386 }
2387
2388 static int check_preimage(struct patch *patch, struct cache_entry **ce, struct stat *st)
2389 {
2390         const char *old_name = patch->old_name;
2391         struct patch *tpatch = NULL;
2392         int stat_ret = 0;
2393         unsigned st_mode = 0;
2394
2395         /*
2396          * Make sure that we do not have local modifications from the
2397          * index when we are looking at the index.  Also make sure
2398          * we have the preimage file to be patched in the work tree,
2399          * unless --cached, which tells git to apply only in the index.
2400          */
2401         if (!old_name)
2402                 return 0;
2403
2404         assert(patch->is_new <= 0);
2405
2406         if (!(patch->is_copy || patch->is_rename) &&
2407             (tpatch = in_fn_table(old_name)) != NULL) {
2408                 if (tpatch == (struct patch *) -1) {
2409                         return error("%s: has been deleted/renamed", old_name);
2410                 }
2411                 st_mode = tpatch->new_mode;
2412         } else if (!cached) {
2413                 stat_ret = lstat(old_name, st);
2414                 if (stat_ret && errno != ENOENT)
2415                         return error("%s: %s", old_name, strerror(errno));
2416         }
2417
2418         if (check_index && !tpatch) {
2419                 int pos = cache_name_pos(old_name, strlen(old_name));
2420                 if (pos < 0) {
2421                         if (patch->is_new < 0)
2422                                 goto is_new;
2423                         return error("%s: does not exist in index", old_name);
2424                 }
2425                 *ce = active_cache[pos];
2426                 if (stat_ret < 0) {
2427                         struct checkout costate;
2428                         /* checkout */
2429                         costate.base_dir = "";
2430                         costate.base_dir_len = 0;
2431                         costate.force = 0;
2432                         costate.quiet = 0;
2433                         costate.not_new = 0;
2434                         costate.refresh_cache = 1;
2435                         if (checkout_entry(*ce, &costate, NULL) ||
2436                             lstat(old_name, st))
2437                                 return -1;
2438                 }
2439                 if (!cached && verify_index_match(*ce, st))
2440                         return error("%s: does not match index", old_name);
2441                 if (cached)
2442                         st_mode = (*ce)->ce_mode;
2443         } else if (stat_ret < 0) {
2444                 if (patch->is_new < 0)
2445                         goto is_new;
2446                 return error("%s: %s", old_name, strerror(errno));
2447         }
2448
2449         if (!cached)
2450                 st_mode = ce_mode_from_stat(*ce, st->st_mode);
2451
2452         if (patch->is_new < 0)
2453                 patch->is_new = 0;
2454         if (!patch->old_mode)
2455                 patch->old_mode = st_mode;
2456         if ((st_mode ^ patch->old_mode) & S_IFMT)
2457                 return error("%s: wrong type", old_name);
2458         if (st_mode != patch->old_mode)
2459                 fprintf(stderr, "warning: %s has type %o, expected %o\n",
2460                         old_name, st_mode, patch->old_mode);
2461         return 0;
2462
2463  is_new:
2464         patch->is_new = 1;
2465         patch->is_delete = 0;
2466         patch->old_name = NULL;
2467         return 0;
2468 }
2469
2470 static int check_patch(struct patch *patch)
2471 {
2472         struct stat st;
2473         const char *old_name = patch->old_name;
2474         const char *new_name = patch->new_name;
2475         const char *name = old_name ? old_name : new_name;
2476         struct cache_entry *ce = NULL;
2477         int ok_if_exists;
2478         int status;
2479
2480         patch->rejected = 1; /* we will drop this after we succeed */
2481
2482         status = check_preimage(patch, &ce, &st);
2483         if (status)
2484                 return status;
2485         old_name = patch->old_name;
2486
2487         if (in_fn_table(new_name) == (struct patch *) -1)
2488                 /*
2489                  * A type-change diff is always split into a patch to
2490                  * delete old, immediately followed by a patch to
2491                  * create new (see diff.c::run_diff()); in such a case
2492                  * it is Ok that the entry to be deleted by the
2493                  * previous patch is still in the working tree and in
2494                  * the index.
2495                  */
2496                 ok_if_exists = 1;
2497         else
2498                 ok_if_exists = 0;
2499
2500         if (new_name &&
2501             ((0 < patch->is_new) | (0 < patch->is_rename) | patch->is_copy)) {
2502                 if (check_index &&
2503                     cache_name_pos(new_name, strlen(new_name)) >= 0 &&
2504                     !ok_if_exists)
2505                         return error("%s: already exists in index", new_name);
2506                 if (!cached) {
2507                         int err = check_to_create_blob(new_name, ok_if_exists);
2508                         if (err)
2509                                 return err;
2510                 }
2511                 if (!patch->new_mode) {
2512                         if (0 < patch->is_new)
2513                                 patch->new_mode = S_IFREG | 0644;
2514                         else
2515                                 patch->new_mode = patch->old_mode;
2516                 }
2517         }
2518
2519         if (new_name && old_name) {
2520                 int same = !strcmp(old_name, new_name);
2521                 if (!patch->new_mode)
2522                         patch->new_mode = patch->old_mode;
2523                 if ((patch->old_mode ^ patch->new_mode) & S_IFMT)
2524                         return error("new mode (%o) of %s does not match old mode (%o)%s%s",
2525                                 patch->new_mode, new_name, patch->old_mode,
2526                                 same ? "" : " of ", same ? "" : old_name);
2527         }
2528
2529         if (apply_data(patch, &st, ce) < 0)
2530                 return error("%s: patch does not apply", name);
2531         patch->rejected = 0;
2532         return 0;
2533 }
2534
2535 static int check_patch_list(struct patch *patch)
2536 {
2537         int err = 0;
2538
2539         while (patch) {
2540                 if (apply_verbosely)
2541                         say_patch_name(stderr,
2542                                        "Checking patch ", patch, "...\n");
2543                 err |= check_patch(patch);
2544                 patch = patch->next;
2545         }
2546         return err;
2547 }
2548
2549 /* This function tries to read the sha1 from the current index */
2550 static int get_current_sha1(const char *path, unsigned char *sha1)
2551 {
2552         int pos;
2553
2554         if (read_cache() < 0)
2555                 return -1;
2556         pos = cache_name_pos(path, strlen(path));
2557         if (pos < 0)
2558                 return -1;
2559         hashcpy(sha1, active_cache[pos]->sha1);
2560         return 0;
2561 }
2562
2563 /* Build an index that contains the just the files needed for a 3way merge */
2564 static void build_fake_ancestor(struct patch *list, const char *filename)
2565 {
2566         struct patch *patch;
2567         struct index_state result = { 0 };
2568         int fd;
2569
2570         /* Once we start supporting the reverse patch, it may be
2571          * worth showing the new sha1 prefix, but until then...
2572          */
2573         for (patch = list; patch; patch = patch->next) {
2574                 const unsigned char *sha1_ptr;
2575                 unsigned char sha1[20];
2576                 struct cache_entry *ce;
2577                 const char *name;
2578
2579                 name = patch->old_name ? patch->old_name : patch->new_name;
2580                 if (0 < patch->is_new)
2581                         continue;
2582                 else if (get_sha1(patch->old_sha1_prefix, sha1))
2583                         /* git diff has no index line for mode/type changes */
2584                         if (!patch->lines_added && !patch->lines_deleted) {
2585                                 if (get_current_sha1(patch->new_name, sha1) ||
2586                                     get_current_sha1(patch->old_name, sha1))
2587                                         die("mode change for %s, which is not "
2588                                                 "in current HEAD", name);
2589                                 sha1_ptr = sha1;
2590                         } else
2591                                 die("sha1 information is lacking or useless "
2592                                         "(%s).", name);
2593                 else
2594                         sha1_ptr = sha1;
2595
2596                 ce = make_cache_entry(patch->old_mode, sha1_ptr, name, 0, 0);
2597                 if (!ce)
2598                         die("make_cache_entry failed for path '%s'", name);
2599                 if (add_index_entry(&result, ce, ADD_CACHE_OK_TO_ADD))
2600                         die ("Could not add %s to temporary index", name);
2601         }
2602
2603         fd = open(filename, O_WRONLY | O_CREAT, 0666);
2604         if (fd < 0 || write_index(&result, fd) || close(fd))
2605                 die ("Could not write temporary index to %s", filename);
2606
2607         discard_index(&result);
2608 }
2609
2610 static void stat_patch_list(struct patch *patch)
2611 {
2612         int files, adds, dels;
2613
2614         for (files = adds = dels = 0 ; patch ; patch = patch->next) {
2615                 files++;
2616                 adds += patch->lines_added;
2617                 dels += patch->lines_deleted;
2618                 show_stats(patch);
2619         }
2620
2621         printf(" %d files changed, %d insertions(+), %d deletions(-)\n", files, adds, dels);
2622 }
2623
2624 static void numstat_patch_list(struct patch *patch)
2625 {
2626         for ( ; patch; patch = patch->next) {
2627                 const char *name;
2628                 name = patch->new_name ? patch->new_name : patch->old_name;
2629                 if (patch->is_binary)
2630                         printf("-\t-\t");
2631                 else
2632                         printf("%d\t%d\t", patch->lines_added, patch->lines_deleted);
2633                 write_name_quoted(name, stdout, line_termination);
2634         }
2635 }
2636
2637 static void show_file_mode_name(const char *newdelete, unsigned int mode, const char *name)
2638 {
2639         if (mode)
2640                 printf(" %s mode %06o %s\n", newdelete, mode, name);
2641         else
2642                 printf(" %s %s\n", newdelete, name);
2643 }
2644
2645 static void show_mode_change(struct patch *p, int show_name)
2646 {
2647         if (p->old_mode && p->new_mode && p->old_mode != p->new_mode) {
2648                 if (show_name)
2649                         printf(" mode change %06o => %06o %s\n",
2650                                p->old_mode, p->new_mode, p->new_name);
2651                 else
2652                         printf(" mode change %06o => %06o\n",
2653                                p->old_mode, p->new_mode);
2654         }
2655 }
2656
2657 static void show_rename_copy(struct patch *p)
2658 {
2659         const char *renamecopy = p->is_rename ? "rename" : "copy";
2660         const char *old, *new;
2661
2662         /* Find common prefix */
2663         old = p->old_name;
2664         new = p->new_name;
2665         while (1) {
2666                 const char *slash_old, *slash_new;
2667                 slash_old = strchr(old, '/');
2668                 slash_new = strchr(new, '/');
2669                 if (!slash_old ||
2670                     !slash_new ||
2671                     slash_old - old != slash_new - new ||
2672                     memcmp(old, new, slash_new - new))
2673                         break;
2674                 old = slash_old + 1;
2675                 new = slash_new + 1;
2676         }
2677         /* p->old_name thru old is the common prefix, and old and new
2678          * through the end of names are renames
2679          */
2680         if (old != p->old_name)
2681                 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2682                        (int)(old - p->old_name), p->old_name,
2683                        old, new, p->score);
2684         else
2685                 printf(" %s %s => %s (%d%%)\n", renamecopy,
2686                        p->old_name, p->new_name, p->score);
2687         show_mode_change(p, 0);
2688 }
2689
2690 static void summary_patch_list(struct patch *patch)
2691 {
2692         struct patch *p;
2693
2694         for (p = patch; p; p = p->next) {
2695                 if (p->is_new)
2696                         show_file_mode_name("create", p->new_mode, p->new_name);
2697                 else if (p->is_delete)
2698                         show_file_mode_name("delete", p->old_mode, p->old_name);
2699                 else {
2700                         if (p->is_rename || p->is_copy)
2701                                 show_rename_copy(p);
2702                         else {
2703                                 if (p->score) {
2704                                         printf(" rewrite %s (%d%%)\n",
2705                                                p->new_name, p->score);
2706                                         show_mode_change(p, 0);
2707                                 }
2708                                 else
2709                                         show_mode_change(p, 1);
2710                         }
2711                 }
2712         }
2713 }
2714
2715 static void patch_stats(struct patch *patch)
2716 {
2717         int lines = patch->lines_added + patch->lines_deleted;
2718
2719         if (lines > max_change)
2720                 max_change = lines;
2721         if (patch->old_name) {
2722                 int len = quote_c_style(patch->old_name, NULL, NULL, 0);
2723                 if (!len)
2724                         len = strlen(patch->old_name);
2725                 if (len > max_len)
2726                         max_len = len;
2727         }
2728         if (patch->new_name) {
2729                 int len = quote_c_style(patch->new_name, NULL, NULL, 0);
2730                 if (!len)
2731                         len = strlen(patch->new_name);
2732                 if (len > max_len)
2733                         max_len = len;
2734         }
2735 }
2736
2737 static void remove_file(struct patch *patch, int rmdir_empty)
2738 {
2739         if (update_index) {
2740                 if (remove_file_from_cache(patch->old_name) < 0)
2741                         die("unable to remove %s from index", patch->old_name);
2742         }
2743         if (!cached) {
2744                 if (S_ISGITLINK(patch->old_mode)) {
2745                         if (rmdir(patch->old_name))
2746                                 warning("unable to remove submodule %s",
2747                                         patch->old_name);
2748                 } else if (!unlink(patch->old_name) && rmdir_empty) {
2749                         remove_path(patch->old_name);
2750                 }
2751         }
2752 }
2753
2754 static void add_index_file(const char *path, unsigned mode, void *buf, unsigned long size)
2755 {
2756         struct stat st;
2757         struct cache_entry *ce;
2758         int namelen = strlen(path);
2759         unsigned ce_size = cache_entry_size(namelen);
2760
2761         if (!update_index)
2762                 return;
2763
2764         ce = xcalloc(1, ce_size);
2765         memcpy(ce->name, path, namelen);
2766         ce->ce_mode = create_ce_mode(mode);
2767         ce->ce_flags = namelen;
2768         if (S_ISGITLINK(mode)) {
2769                 const char *s = buf;
2770
2771                 if (get_sha1_hex(s + strlen("Subproject commit "), ce->sha1))
2772                         die("corrupt patch for subproject %s", path);
2773         } else {
2774                 if (!cached) {
2775                         if (lstat(path, &st) < 0)
2776                                 die("unable to stat newly created file %s",
2777                                     path);
2778                         fill_stat_cache_info(ce, &st);
2779                 }
2780                 if (write_sha1_file(buf, size, blob_type, ce->sha1) < 0)
2781                         die("unable to create backing store for newly created file %s", path);
2782         }
2783         if (add_cache_entry(ce, ADD_CACHE_OK_TO_ADD) < 0)
2784                 die("unable to add cache entry for %s", path);
2785 }
2786
2787 static int try_create_file(const char *path, unsigned int mode, const char *buf, unsigned long size)
2788 {
2789         int fd;
2790         struct strbuf nbuf;
2791
2792         if (S_ISGITLINK(mode)) {
2793                 struct stat st;
2794                 if (!lstat(path, &st) && S_ISDIR(st.st_mode))
2795                         return 0;
2796                 return mkdir(path, 0777);
2797         }
2798
2799         if (has_symlinks && S_ISLNK(mode))
2800                 /* Although buf:size is counted string, it also is NUL
2801                  * terminated.
2802                  */
2803                 return symlink(buf, path);
2804
2805         fd = open(path, O_CREAT | O_EXCL | O_WRONLY, (mode & 0100) ? 0777 : 0666);
2806         if (fd < 0)
2807                 return -1;
2808
2809         strbuf_init(&nbuf, 0);
2810         if (convert_to_working_tree(path, buf, size, &nbuf)) {
2811                 size = nbuf.len;
2812                 buf  = nbuf.buf;
2813         }
2814         write_or_die(fd, buf, size);
2815         strbuf_release(&nbuf);
2816
2817         if (close(fd) < 0)
2818                 die("closing file %s: %s", path, strerror(errno));
2819         return 0;
2820 }
2821
2822 /*
2823  * We optimistically assume that the directories exist,
2824  * which is true 99% of the time anyway. If they don't,
2825  * we create them and try again.
2826  */
2827 static void create_one_file(char *path, unsigned mode, const char *buf, unsigned long size)
2828 {
2829         if (cached)
2830                 return;
2831         if (!try_create_file(path, mode, buf, size))
2832                 return;
2833
2834         if (errno == ENOENT) {
2835                 if (safe_create_leading_directories(path))
2836                         return;
2837                 if (!try_create_file(path, mode, buf, size))
2838                         return;
2839         }
2840
2841         if (errno == EEXIST || errno == EACCES) {
2842                 /* We may be trying to create a file where a directory
2843                  * used to be.
2844                  */
2845                 struct stat st;
2846                 if (!lstat(path, &st) && (!S_ISDIR(st.st_mode) || !rmdir(path)))
2847                         errno = EEXIST;
2848         }
2849
2850         if (errno == EEXIST) {
2851                 unsigned int nr = getpid();
2852
2853                 for (;;) {
2854                         char newpath[PATH_MAX];
2855                         mksnpath(newpath, sizeof(newpath), "%s~%u", path, nr);
2856                         if (!try_create_file(newpath, mode, buf, size)) {
2857                                 if (!rename(newpath, path))
2858                                         return;
2859                                 unlink(newpath);
2860                                 break;
2861                         }
2862                         if (errno != EEXIST)
2863                                 break;
2864                         ++nr;
2865                 }
2866         }
2867         die("unable to write file %s mode %o", path, mode);
2868 }
2869
2870 static void create_file(struct patch *patch)
2871 {
2872         char *path = patch->new_name;
2873         unsigned mode = patch->new_mode;
2874         unsigned long size = patch->resultsize;
2875         char *buf = patch->result;
2876
2877         if (!mode)
2878                 mode = S_IFREG | 0644;
2879         create_one_file(path, mode, buf, size);
2880         add_index_file(path, mode, buf, size);
2881 }
2882
2883 /* phase zero is to remove, phase one is to create */
2884 static void write_out_one_result(struct patch *patch, int phase)
2885 {
2886         if (patch->is_delete > 0) {
2887                 if (phase == 0)
2888                         remove_file(patch, 1);
2889                 return;
2890         }
2891         if (patch->is_new > 0 || patch->is_copy) {
2892                 if (phase == 1)
2893                         create_file(patch);
2894                 return;
2895         }
2896         /*
2897          * Rename or modification boils down to the same
2898          * thing: remove the old, write the new
2899          */
2900         if (phase == 0)
2901                 remove_file(patch, patch->is_rename);
2902         if (phase == 1)
2903                 create_file(patch);
2904 }
2905
2906 static int write_out_one_reject(struct patch *patch)
2907 {
2908         FILE *rej;
2909         char namebuf[PATH_MAX];
2910         struct fragment *frag;
2911         int cnt = 0;
2912
2913         for (cnt = 0, frag = patch->fragments; frag; frag = frag->next) {
2914                 if (!frag->rejected)
2915                         continue;
2916                 cnt++;
2917         }
2918
2919         if (!cnt) {
2920                 if (apply_verbosely)
2921                         say_patch_name(stderr,
2922                                        "Applied patch ", patch, " cleanly.\n");
2923                 return 0;
2924         }
2925
2926         /* This should not happen, because a removal patch that leaves
2927          * contents are marked "rejected" at the patch level.
2928          */
2929         if (!patch->new_name)
2930                 die("internal error");
2931
2932         /* Say this even without --verbose */
2933         say_patch_name(stderr, "Applying patch ", patch, " with");
2934         fprintf(stderr, " %d rejects...\n", cnt);
2935
2936         cnt = strlen(patch->new_name);
2937         if (ARRAY_SIZE(namebuf) <= cnt + 5) {
2938                 cnt = ARRAY_SIZE(namebuf) - 5;
2939                 fprintf(stderr,
2940                         "warning: truncating .rej filename to %.*s.rej",
2941                         cnt - 1, patch->new_name);
2942         }
2943         memcpy(namebuf, patch->new_name, cnt);
2944         memcpy(namebuf + cnt, ".rej", 5);
2945
2946         rej = fopen(namebuf, "w");
2947         if (!rej)
2948                 return error("cannot open %s: %s", namebuf, strerror(errno));
2949
2950         /* Normal git tools never deal with .rej, so do not pretend
2951          * this is a git patch by saying --git nor give extended
2952          * headers.  While at it, maybe please "kompare" that wants
2953          * the trailing TAB and some garbage at the end of line ;-).
2954          */
2955         fprintf(rej, "diff a/%s b/%s\t(rejected hunks)\n",
2956                 patch->new_name, patch->new_name);
2957         for (cnt = 1, frag = patch->fragments;
2958              frag;
2959              cnt++, frag = frag->next) {
2960                 if (!frag->rejected) {
2961                         fprintf(stderr, "Hunk #%d applied cleanly.\n", cnt);
2962                         continue;
2963                 }
2964                 fprintf(stderr, "Rejected hunk #%d.\n", cnt);
2965                 fprintf(rej, "%.*s", frag->size, frag->patch);
2966                 if (frag->patch[frag->size-1] != '\n')
2967                         fputc('\n', rej);
2968         }
2969         fclose(rej);
2970         return -1;
2971 }
2972
2973 static int write_out_results(struct patch *list, int skipped_patch)
2974 {
2975         int phase;
2976         int errs = 0;
2977         struct patch *l;
2978
2979         if (!list && !skipped_patch)
2980                 return error("No changes");
2981
2982         for (phase = 0; phase < 2; phase++) {
2983                 l = list;
2984                 while (l) {
2985                         if (l->rejected)
2986                                 errs = 1;
2987                         else {
2988                                 write_out_one_result(l, phase);
2989                                 if (phase == 1 && write_out_one_reject(l))
2990                                         errs = 1;
2991                         }
2992                         l = l->next;
2993                 }
2994         }
2995         return errs;
2996 }
2997
2998 static struct lock_file lock_file;
2999
3000 static struct excludes {
3001         struct excludes *next;
3002         const char *path;
3003 } *excludes;
3004
3005 static int use_patch(struct patch *p)
3006 {
3007         const char *pathname = p->new_name ? p->new_name : p->old_name;
3008         struct excludes *x = excludes;
3009         while (x) {
3010                 if (fnmatch(x->path, pathname, 0) == 0)
3011                         return 0;
3012                 x = x->next;
3013         }
3014         if (0 < prefix_length) {
3015                 int pathlen = strlen(pathname);
3016                 if (pathlen <= prefix_length ||
3017                     memcmp(prefix, pathname, prefix_length))
3018                         return 0;
3019         }
3020         return 1;
3021 }
3022
3023 static void prefix_one(char **name)
3024 {
3025         char *old_name = *name;
3026         if (!old_name)
3027                 return;
3028         *name = xstrdup(prefix_filename(prefix, prefix_length, *name));
3029         free(old_name);
3030 }
3031
3032 static void prefix_patches(struct patch *p)
3033 {
3034         if (!prefix || p->is_toplevel_relative)
3035                 return;
3036         for ( ; p; p = p->next) {
3037                 if (p->new_name == p->old_name) {
3038                         char *prefixed = p->new_name;
3039                         prefix_one(&prefixed);
3040                         p->new_name = p->old_name = prefixed;
3041                 }
3042                 else {
3043                         prefix_one(&p->new_name);
3044                         prefix_one(&p->old_name);
3045                 }
3046         }
3047 }
3048
3049 #define INACCURATE_EOF  (1<<0)
3050 #define RECOUNT         (1<<1)
3051
3052 static int apply_patch(int fd, const char *filename, int options)
3053 {
3054         size_t offset;
3055         struct strbuf buf;
3056         struct patch *list = NULL, **listp = &list;
3057         int skipped_patch = 0;
3058
3059         /* FIXME - memory leak when using multiple patch files as inputs */
3060         memset(&fn_table, 0, sizeof(struct string_list));
3061         strbuf_init(&buf, 0);
3062         patch_input_file = filename;
3063         read_patch_file(&buf, fd);
3064         offset = 0;
3065         while (offset < buf.len) {
3066                 struct patch *patch;
3067                 int nr;
3068
3069                 patch = xcalloc(1, sizeof(*patch));
3070                 patch->inaccurate_eof = !!(options & INACCURATE_EOF);
3071                 patch->recount =  !!(options & RECOUNT);
3072                 nr = parse_chunk(buf.buf + offset, buf.len - offset, patch);
3073                 if (nr < 0)
3074                         break;
3075                 if (apply_in_reverse)
3076                         reverse_patches(patch);
3077                 if (prefix)
3078                         prefix_patches(patch);
3079                 if (use_patch(patch)) {
3080                         patch_stats(patch);
3081                         *listp = patch;
3082                         listp = &patch->next;
3083                 }
3084                 else {
3085                         /* perhaps free it a bit better? */
3086                         free(patch);
3087                         skipped_patch++;
3088                 }
3089                 offset += nr;
3090         }
3091
3092         if (whitespace_error && (ws_error_action == die_on_ws_error))
3093                 apply = 0;
3094
3095         update_index = check_index && apply;
3096         if (update_index && newfd < 0)
3097                 newfd = hold_locked_index(&lock_file, 1);
3098
3099         if (check_index) {
3100                 if (read_cache() < 0)
3101                         die("unable to read index file");
3102         }
3103
3104         if ((check || apply) &&
3105             check_patch_list(list) < 0 &&
3106             !apply_with_reject)
3107                 exit(1);
3108
3109         if (apply && write_out_results(list, skipped_patch))
3110                 exit(1);
3111
3112         if (fake_ancestor)
3113                 build_fake_ancestor(list, fake_ancestor);
3114
3115         if (diffstat)
3116                 stat_patch_list(list);
3117
3118         if (numstat)
3119                 numstat_patch_list(list);
3120
3121         if (summary)
3122                 summary_patch_list(list);
3123
3124         strbuf_release(&buf);
3125         return 0;
3126 }
3127
3128 static int git_apply_config(const char *var, const char *value, void *cb)
3129 {
3130         if (!strcmp(var, "apply.whitespace"))
3131                 return git_config_string(&apply_default_whitespace, var, value);
3132         return git_default_config(var, value, cb);
3133 }
3134
3135
3136 int cmd_apply(int argc, const char **argv, const char *unused_prefix)
3137 {
3138         int i;
3139         int read_stdin = 1;
3140         int options = 0;
3141         int errs = 0;
3142         int is_not_gitdir;
3143
3144         const char *whitespace_option = NULL;
3145
3146         prefix = setup_git_directory_gently(&is_not_gitdir);
3147         prefix_length = prefix ? strlen(prefix) : 0;
3148         git_config(git_apply_config, NULL);
3149         if (apply_default_whitespace)
3150                 parse_whitespace_option(apply_default_whitespace);
3151
3152         for (i = 1; i < argc; i++) {
3153                 const char *arg = argv[i];
3154                 char *end;
3155                 int fd;
3156
3157                 if (!strcmp(arg, "-")) {
3158                         errs |= apply_patch(0, "<stdin>", options);
3159                         read_stdin = 0;
3160                         continue;
3161                 }
3162                 if (!prefixcmp(arg, "--exclude=")) {
3163                         struct excludes *x = xmalloc(sizeof(*x));
3164                         x->path = arg + 10;
3165                         x->next = excludes;
3166                         excludes = x;
3167                         continue;
3168                 }
3169                 if (!prefixcmp(arg, "-p")) {
3170                         p_value = atoi(arg + 2);
3171                         p_value_known = 1;
3172                         continue;
3173                 }
3174                 if (!strcmp(arg, "--no-add")) {
3175                         no_add = 1;
3176                         continue;
3177                 }
3178                 if (!strcmp(arg, "--stat")) {
3179                         apply = 0;
3180                         diffstat = 1;
3181                         continue;
3182                 }
3183                 if (!strcmp(arg, "--allow-binary-replacement") ||
3184                     !strcmp(arg, "--binary")) {
3185                         continue; /* now no-op */
3186                 }
3187                 if (!strcmp(arg, "--numstat")) {
3188                         apply = 0;
3189                         numstat = 1;
3190                         continue;
3191                 }
3192                 if (!strcmp(arg, "--summary")) {
3193                         apply = 0;
3194                         summary = 1;
3195                         continue;
3196                 }
3197                 if (!strcmp(arg, "--check")) {
3198                         apply = 0;
3199                         check = 1;
3200                         continue;
3201                 }
3202                 if (!strcmp(arg, "--index")) {
3203                         if (is_not_gitdir)
3204                                 die("--index outside a repository");
3205                         check_index = 1;
3206                         continue;
3207                 }
3208                 if (!strcmp(arg, "--cached")) {
3209                         if (is_not_gitdir)
3210                                 die("--cached outside a repository");
3211                         check_index = 1;
3212                         cached = 1;
3213                         continue;
3214                 }
3215                 if (!strcmp(arg, "--apply")) {
3216                         apply = 1;
3217                         continue;
3218                 }
3219                 if (!strcmp(arg, "--build-fake-ancestor")) {
3220                         apply = 0;
3221                         if (++i >= argc)
3222                                 die ("need a filename");
3223                         fake_ancestor = argv[i];
3224                         continue;
3225                 }
3226                 if (!strcmp(arg, "-z")) {
3227                         line_termination = 0;
3228                         continue;
3229                 }
3230                 if (!prefixcmp(arg, "-C")) {
3231                         p_context = strtoul(arg + 2, &end, 0);
3232                         if (*end != '\0')
3233                                 die("unrecognized context count '%s'", arg + 2);
3234                         continue;
3235                 }
3236                 if (!prefixcmp(arg, "--whitespace=")) {
3237                         whitespace_option = arg + 13;
3238                         parse_whitespace_option(arg + 13);
3239                         continue;
3240                 }
3241                 if (!strcmp(arg, "-R") || !strcmp(arg, "--reverse")) {
3242                         apply_in_reverse = 1;
3243                         continue;
3244                 }
3245                 if (!strcmp(arg, "--unidiff-zero")) {
3246                         unidiff_zero = 1;
3247                         continue;
3248                 }
3249                 if (!strcmp(arg, "--reject")) {
3250                         apply = apply_with_reject = apply_verbosely = 1;
3251                         continue;
3252                 }
3253                 if (!strcmp(arg, "-v") || !strcmp(arg, "--verbose")) {
3254                         apply_verbosely = 1;
3255                         continue;
3256                 }
3257                 if (!strcmp(arg, "--inaccurate-eof")) {
3258                         options |= INACCURATE_EOF;
3259                         continue;
3260                 }
3261                 if (!strcmp(arg, "--recount")) {
3262                         options |= RECOUNT;
3263                         continue;
3264                 }
3265                 if (!prefixcmp(arg, "--directory=")) {
3266                         arg += strlen("--directory=");
3267                         root_len = strlen(arg);
3268                         if (root_len && arg[root_len - 1] != '/') {
3269                                 char *new_root;
3270                                 root = new_root = xmalloc(root_len + 2);
3271                                 strcpy(new_root, arg);
3272                                 strcpy(new_root + root_len++, "/");
3273                         } else
3274                                 root = arg;
3275                         continue;
3276                 }
3277                 if (0 < prefix_length)
3278                         arg = prefix_filename(prefix, prefix_length, arg);
3279
3280                 fd = open(arg, O_RDONLY);
3281                 if (fd < 0)
3282                         die("can't open patch '%s': %s", arg, strerror(errno));
3283                 read_stdin = 0;
3284                 set_default_whitespace_mode(whitespace_option);
3285                 errs |= apply_patch(fd, arg, options);
3286                 close(fd);
3287         }
3288         set_default_whitespace_mode(whitespace_option);
3289         if (read_stdin)
3290                 errs |= apply_patch(0, "<stdin>", options);
3291         if (whitespace_error) {
3292                 if (squelch_whitespace_errors &&
3293                     squelch_whitespace_errors < whitespace_error) {
3294                         int squelched =
3295                                 whitespace_error - squelch_whitespace_errors;
3296                         fprintf(stderr, "warning: squelched %d "
3297                                 "whitespace error%s\n",
3298                                 squelched,
3299                                 squelched == 1 ? "" : "s");
3300                 }
3301                 if (ws_error_action == die_on_ws_error)
3302                         die("%d line%s add%s whitespace errors.",
3303                             whitespace_error,
3304                             whitespace_error == 1 ? "" : "s",
3305                             whitespace_error == 1 ? "s" : "");
3306                 if (applied_after_fixing_ws && apply)
3307                         fprintf(stderr, "warning: %d line%s applied after"
3308                                 " fixing whitespace errors.\n",
3309                                 applied_after_fixing_ws,
3310                                 applied_after_fixing_ws == 1 ? "" : "s");
3311                 else if (whitespace_error)
3312                         fprintf(stderr, "warning: %d line%s add%s whitespace errors.\n",
3313                                 whitespace_error,
3314                                 whitespace_error == 1 ? "" : "s",
3315                                 whitespace_error == 1 ? "s" : "");
3316         }
3317
3318         if (update_index) {
3319                 if (write_cache(newfd, active_cache, active_nr) ||
3320                     commit_locked_index(&lock_file))
3321                         die("Unable to write new index file");
3322         }
3323
3324         return !!errs;
3325 }