git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include <sys/types.h>
   5 #include <sys/wait.h>
   6 #include <signal.h>
   7 #include "cache.h"
   8 #include "quote.h"
   9 #include "diff.h"
  10 #include "diffcore.h"
  11 #include "xdiff-interface.h"
  12
  13 static int use_size_cache;
  14
  15 int diff_rename_limit_default = -1;
  16
  17 int git_diff_config(const char *var, const char *value)
  18 {
  19         if (!strcmp(var, "diff.renamelimit")) {
  20                 diff_rename_limit_default = git_config_int(var, value);
  21                 return 0;
  22         }
  23
  24         return git_default_config(var, value);
  25 }
  26
  27 static char *quote_one(const char *str)
  28 {
  29         int needlen;
  30         char *xp;
  31
  32         if (!str)
  33                 return NULL;
  34         needlen = quote_c_style(str, NULL, NULL, 0);
  35         if (!needlen)
  36                 return strdup(str);
  37         xp = xmalloc(needlen + 1);
  38         quote_c_style(str, xp, NULL, 0);
  39         return xp;
  40 }
  41
  42 static char *quote_two(const char *one, const char *two)
  43 {
  44         int need_one = quote_c_style(one, NULL, NULL, 1);
  45         int need_two = quote_c_style(two, NULL, NULL, 1);
  46         char *xp;
  47
  48         if (need_one + need_two) {
  49                 if (!need_one) need_one = strlen(one);
  50                 if (!need_two) need_one = strlen(two);
  51
  52                 xp = xmalloc(need_one + need_two + 3);
  53                 xp[0] = '"';
  54                 quote_c_style(one, xp + 1, NULL, 1);
  55                 quote_c_style(two, xp + need_one + 1, NULL, 1);
  56                 strcpy(xp + need_one + need_two + 1, "\"");
  57                 return xp;
  58         }
  59         need_one = strlen(one);
  60         need_two = strlen(two);
  61         xp = xmalloc(need_one + need_two + 1);
  62         strcpy(xp, one);
  63         strcpy(xp + need_one, two);
  64         return xp;
  65 }
  66
  67 static const char *external_diff(void)
  68 {
  69         static const char *external_diff_cmd = NULL;
  70         static int done_preparing = 0;
  71
  72         if (done_preparing)
  73                 return external_diff_cmd;
  74         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  75         done_preparing = 1;
  76         return external_diff_cmd;
  77 }
  78
  79 #define TEMPFILE_PATH_LEN               50
  80
  81 static struct diff_tempfile {
  82         const char *name; /* filename external diff should read from */
  83         char hex[41];
  84         char mode[10];
  85         char tmp_path[TEMPFILE_PATH_LEN];
  86 } diff_temp[2];
  87
  88 static int count_lines(const char *data, int size)
  89 {
  90         int count, ch, completely_empty = 1, nl_just_seen = 0;
  91         count = 0;
  92         while (0 < size--) {
  93                 ch = *data++;
  94                 if (ch == '\n') {
  95                         count++;
  96                         nl_just_seen = 1;
  97                         completely_empty = 0;
  98                 }
  99                 else {
 100                         nl_just_seen = 0;
 101                         completely_empty = 0;
 102                 }
 103         }
 104         if (completely_empty)
 105                 return 0;
 106         if (!nl_just_seen)
 107                 count++; /* no trailing newline */
 108         return count;
 109 }
 110
 111 static void print_line_count(int count)
 112 {
 113         switch (count) {
 114         case 0:
 115                 printf("0,0");
 116                 break;
 117         case 1:
 118                 printf("1");
 119                 break;
 120         default:
 121                 printf("1,%d", count);
 122                 break;
 123         }
 124 }
 125
 126 static void copy_file(int prefix, const char *data, int size)
 127 {
 128         int ch, nl_just_seen = 1;
 129         while (0 < size--) {
 130                 ch = *data++;
 131                 if (nl_just_seen)
 132                         putchar(prefix);
 133                 putchar(ch);
 134                 if (ch == '\n')
 135                         nl_just_seen = 1;
 136                 else
 137                         nl_just_seen = 0;
 138         }
 139         if (!nl_just_seen)
 140                 printf("\n\\ No newline at end of file\n");
 141 }
 142
 143 static void emit_rewrite_diff(const char *name_a,
 144                               const char *name_b,
 145                               struct diff_filespec *one,
 146                               struct diff_filespec *two)
 147 {
 148         int lc_a, lc_b;
 149         diff_populate_filespec(one, 0);
 150         diff_populate_filespec(two, 0);
 151         lc_a = count_lines(one->data, one->size);
 152         lc_b = count_lines(two->data, two->size);
 153         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 154         print_line_count(lc_a);
 155         printf(" +");
 156         print_line_count(lc_b);
 157         printf(" @@\n");
 158         if (lc_a)
 159                 copy_file('-', one->data, one->size);
 160         if (lc_b)
 161                 copy_file('+', two->data, two->size);
 162 }
 163
 164 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 165 {
 166         if (!DIFF_FILE_VALID(one)) {
 167                 mf->ptr = ""; /* does not matter */
 168                 mf->size = 0;
 169                 return 0;
 170         }
 171         else if (diff_populate_filespec(one, 0))
 172                 return -1;
 173         mf->ptr = one->data;
 174         mf->size = one->size;
 175         return 0;
 176 }
 177
 178 struct emit_callback {
 179         const char **label_path;
 180 };
 181
 182 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
 183 {
 184         int i;
 185         struct emit_callback *ecbdata = priv;
 186
 187         if (ecbdata->label_path[0]) {
 188                 printf("--- %s\n", ecbdata->label_path[0]);
 189                 printf("+++ %s\n", ecbdata->label_path[1]);
 190                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 191         }
 192         for (i = 0; i < nbuf; i++)
 193                 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
 194                         return -1;
 195         return 0;
 196 }
 197
 198 static char *pprint_rename(const char *a, const char *b)
 199 {
 200         const char *old = a;
 201         const char *new = b;
 202         char *name = NULL;
 203         int pfx_length, sfx_length;
 204         int len_a = strlen(a);
 205         int len_b = strlen(b);
 206
 207         /* Find common prefix */
 208         pfx_length = 0;
 209         while (*old && *new && *old == *new) {
 210                 if (*old == '/')
 211                         pfx_length = old - a + 1;
 212                 old++;
 213                 new++;
 214         }
 215
 216         /* Find common suffix */
 217         old = a + len_a;
 218         new = b + len_b;
 219         sfx_length = 0;
 220         while (a <= old && b <= new && *old == *new) {
 221                 if (*old == '/')
 222                         sfx_length = len_a - (old - a);
 223                 old--;
 224                 new--;
 225         }
 226
 227         /*
 228          * pfx{mid-a => mid-b}sfx
 229          * {pfx-a => pfx-b}sfx
 230          * pfx{sfx-a => sfx-b}
 231          * name-a => name-b
 232          */
 233         if (pfx_length + sfx_length) {
 234                 name = xmalloc(len_a + len_b - pfx_length - sfx_length + 7);
 235                 sprintf(name, "%.*s{%.*s => %.*s}%s",
 236                         pfx_length, a,
 237                         len_a - pfx_length - sfx_length, a + pfx_length,
 238                         len_b - pfx_length - sfx_length, b + pfx_length,
 239                         a + len_a - sfx_length);
 240         }
 241         else {
 242                 name = xmalloc(len_a + len_b + 5);
 243                 sprintf(name, "%s => %s", a, b);
 244         }
 245         return name;
 246 }
 247
 248 struct diffstat_t {
 249         struct xdiff_emit_state xm;
 250
 251         int nr;
 252         int alloc;
 253         struct diffstat_file {
 254                 char *name;
 255                 unsigned is_unmerged:1;
 256                 unsigned is_binary:1;
 257                 unsigned is_renamed:1;
 258                 unsigned int added, deleted;
 259         } **files;
 260 };
 261
 262 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 263                                           const char *name_a,
 264                                           const char *name_b)
 265 {
 266         struct diffstat_file *x;
 267         x = xcalloc(sizeof (*x), 1);
 268         if (diffstat->nr == diffstat->alloc) {
 269                 diffstat->alloc = alloc_nr(diffstat->alloc);
 270                 diffstat->files = xrealloc(diffstat->files,
 271                                 diffstat->alloc * sizeof(x));
 272         }
 273         diffstat->files[diffstat->nr++] = x;
 274         if (name_b) {
 275                 x->name = pprint_rename(name_a, name_b);
 276                 x->is_renamed = 1;
 277         }
 278         else
 279                 x->name = strdup(name_a);
 280         return x;
 281 }
 282
 283 static void diffstat_consume(void *priv, char *line, unsigned long len)
 284 {
 285         struct diffstat_t *diffstat = priv;
 286         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 287
 288         if (line[0] == '+')
 289                 x->added++;
 290         else if (line[0] == '-')
 291                 x->deleted++;
 292 }
 293
 294 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
 295 static const char minuses[]= "----------------------------------------------------------------------";
 296
 297 static void show_stats(struct diffstat_t* data)
 298 {
 299         char *prefix = "";
 300         int i, len, add, del, total, adds = 0, dels = 0;
 301         int max, max_change = 0, max_len = 0;
 302         int total_files = data->nr;
 303
 304         if (data->nr == 0)
 305                 return;
 306
 307         for (i = 0; i < data->nr; i++) {
 308                 struct diffstat_file *file = data->files[i];
 309
 310                 len = strlen(file->name);
 311                 if (max_len < len)
 312                         max_len = len;
 313
 314                 if (file->is_binary || file->is_unmerged)
 315                         continue;
 316                 if (max_change < file->added + file->deleted)
 317                         max_change = file->added + file->deleted;
 318         }
 319
 320         for (i = 0; i < data->nr; i++) {
 321                 char *name = data->files[i]->name;
 322                 int added = data->files[i]->added;
 323                 int deleted = data->files[i]->deleted;
 324
 325                 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
 326                         char *qname = xmalloc(len + 1);
 327                         quote_c_style(name, qname, NULL, 0);
 328                         free(name);
 329                         data->files[i]->name = name = qname;
 330                 }
 331
 332                 /*
 333                  * "scale" the filename
 334                  */
 335                 len = strlen(name);
 336                 max = max_len;
 337                 if (max > 50)
 338                         max = 50;
 339                 if (len > max) {
 340                         char *slash;
 341                         prefix = "...";
 342                         max -= 3;
 343                         name += len - max;
 344                         slash = strchr(name, '/');
 345                         if (slash)
 346                                 name = slash;
 347                 }
 348                 len = max;
 349
 350                 /*
 351                  * scale the add/delete
 352                  */
 353                 max = max_change;
 354                 if (max + len > 70)
 355                         max = 70 - len;
 356
 357                 if (data->files[i]->is_binary) {
 358                         printf(" %s%-*s |  Bin\n", prefix, len, name);
 359                         goto free_diffstat_file;
 360                 }
 361                 else if (data->files[i]->is_unmerged) {
 362                         printf(" %s%-*s |  Unmerged\n", prefix, len, name);
 363                         goto free_diffstat_file;
 364                 }
 365                 else if (!data->files[i]->is_renamed &&
 366                          (added + deleted == 0)) {
 367                         total_files--;
 368                         goto free_diffstat_file;
 369                 }
 370
 371                 add = added;
 372                 del = deleted;
 373                 total = add + del;
 374                 adds += add;
 375                 dels += del;
 376
 377                 if (max_change > 0) {
 378                         total = (total * max + max_change / 2) / max_change;
 379                         add = (add * max + max_change / 2) / max_change;
 380                         del = total - add;
 381                 }
 382                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
 383                                 len, name, added + deleted,
 384                                 add, pluses, del, minuses);
 385         free_diffstat_file:
 386                 free(data->files[i]->name);
 387                 free(data->files[i]);
 388         }
 389         free(data->files);
 390         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 391                         total_files, adds, dels);
 392 }
 393
 394 #define FIRST_FEW_BYTES 8000
 395 static int mmfile_is_binary(mmfile_t *mf)
 396 {
 397         long sz = mf->size;
 398         if (FIRST_FEW_BYTES < sz)
 399                 sz = FIRST_FEW_BYTES;
 400         if (memchr(mf->ptr, 0, sz))
 401                 return 1;
 402         return 0;
 403 }
 404
 405 static void builtin_diff(const char *name_a,
 406                          const char *name_b,
 407                          struct diff_filespec *one,
 408                          struct diff_filespec *two,
 409                          const char *xfrm_msg,
 410                          int complete_rewrite)
 411 {
 412         mmfile_t mf1, mf2;
 413         const char *lbl[2];
 414         char *a_one, *b_two;
 415
 416         a_one = quote_two("a/", name_a);
 417         b_two = quote_two("b/", name_b);
 418         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
 419         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
 420         printf("diff --git %s %s\n", a_one, b_two);
 421         if (lbl[0][0] == '/') {
 422                 /* /dev/null */
 423                 printf("new file mode %06o\n", two->mode);
 424                 if (xfrm_msg && xfrm_msg[0])
 425                         puts(xfrm_msg);
 426         }
 427         else if (lbl[1][0] == '/') {
 428                 printf("deleted file mode %06o\n", one->mode);
 429                 if (xfrm_msg && xfrm_msg[0])
 430                         puts(xfrm_msg);
 431         }
 432         else {
 433                 if (one->mode != two->mode) {
 434                         printf("old mode %06o\n", one->mode);
 435                         printf("new mode %06o\n", two->mode);
 436                 }
 437                 if (xfrm_msg && xfrm_msg[0])
 438                         puts(xfrm_msg);
 439                 /*
 440                  * we do not run diff between different kind
 441                  * of objects.
 442                  */
 443                 if ((one->mode ^ two->mode) & S_IFMT)
 444                         goto free_ab_and_return;
 445                 if (complete_rewrite) {
 446                         emit_rewrite_diff(name_a, name_b, one, two);
 447                         goto free_ab_and_return;
 448                 }
 449         }
 450
 451         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 452                 die("unable to read files to diff");
 453
 454         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 455                 printf("Binary files %s and %s differ\n", lbl[0], lbl[1]);
 456         else {
 457                 /* Crazy xdl interfaces.. */
 458                 const char *diffopts = getenv("GIT_DIFF_OPTS");
 459                 xpparam_t xpp;
 460                 xdemitconf_t xecfg;
 461                 xdemitcb_t ecb;
 462                 struct emit_callback ecbdata;
 463
 464                 ecbdata.label_path = lbl;
 465                 xpp.flags = XDF_NEED_MINIMAL;
 466                 xecfg.ctxlen = 3;
 467                 xecfg.flags = XDL_EMIT_FUNCNAMES;
 468                 if (!diffopts)
 469                         ;
 470                 else if (!strncmp(diffopts, "--unified=", 10))
 471                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
 472                 else if (!strncmp(diffopts, "-u", 2))
 473                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
 474                 ecb.outf = fn_out;
 475                 ecb.priv = &ecbdata;
 476                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 477         }
 478
 479  free_ab_and_return:
 480         free(a_one);
 481         free(b_two);
 482         return;
 483 }
 484
 485 static void builtin_diffstat(const char *name_a, const char *name_b,
 486                              struct diff_filespec *one,
 487                              struct diff_filespec *two,
 488                              struct diffstat_t *diffstat,
 489                              int complete_rewrite)
 490 {
 491         mmfile_t mf1, mf2;
 492         struct diffstat_file *data;
 493
 494         data = diffstat_add(diffstat, name_a, name_b);
 495
 496         if (!one || !two) {
 497                 data->is_unmerged = 1;
 498                 return;
 499         }
 500         if (complete_rewrite) {
 501                 diff_populate_filespec(one, 0);
 502                 diff_populate_filespec(two, 0);
 503                 data->deleted = count_lines(one->data, one->size);
 504                 data->added = count_lines(two->data, two->size);
 505                 return;
 506         }
 507         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 508                 die("unable to read files to diff");
 509
 510         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 511                 data->is_binary = 1;
 512         else {
 513                 /* Crazy xdl interfaces.. */
 514                 xpparam_t xpp;
 515                 xdemitconf_t xecfg;
 516                 xdemitcb_t ecb;
 517
 518                 xpp.flags = XDF_NEED_MINIMAL;
 519                 xecfg.ctxlen = 0;
 520                 xecfg.flags = 0;
 521                 ecb.outf = xdiff_outf;
 522                 ecb.priv = diffstat;
 523                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 524         }
 525 }
 526
 527 struct diff_filespec *alloc_filespec(const char *path)
 528 {
 529         int namelen = strlen(path);
 530         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 531
 532         memset(spec, 0, sizeof(*spec));
 533         spec->path = (char *)(spec + 1);
 534         memcpy(spec->path, path, namelen+1);
 535         return spec;
 536 }
 537
 538 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 539                    unsigned short mode)
 540 {
 541         if (mode) {
 542                 spec->mode = canon_mode(mode);
 543                 memcpy(spec->sha1, sha1, 20);
 544                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 545         }
 546 }
 547
 548 /*
 549  * Given a name and sha1 pair, if the dircache tells us the file in
 550  * the work tree has that object contents, return true, so that
 551  * prepare_temp_file() does not have to inflate and extract.
 552  */
 553 static int work_tree_matches(const char *name, const unsigned char *sha1)
 554 {
 555         struct cache_entry *ce;
 556         struct stat st;
 557         int pos, len;
 558
 559         /* We do not read the cache ourselves here, because the
 560          * benchmark with my previous version that always reads cache
 561          * shows that it makes things worse for diff-tree comparing
 562          * two linux-2.6 kernel trees in an already checked out work
 563          * tree.  This is because most diff-tree comparisons deal with
 564          * only a small number of files, while reading the cache is
 565          * expensive for a large project, and its cost outweighs the
 566          * savings we get by not inflating the object to a temporary
 567          * file.  Practically, this code only helps when we are used
 568          * by diff-cache --cached, which does read the cache before
 569          * calling us.
 570          */
 571         if (!active_cache)
 572                 return 0;
 573
 574         len = strlen(name);
 575         pos = cache_name_pos(name, len);
 576         if (pos < 0)
 577                 return 0;
 578         ce = active_cache[pos];
 579         if ((lstat(name, &st) < 0) ||
 580             !S_ISREG(st.st_mode) || /* careful! */
 581             ce_match_stat(ce, &st, 0) ||
 582             memcmp(sha1, ce->sha1, 20))
 583                 return 0;
 584         /* we return 1 only when we can stat, it is a regular file,
 585          * stat information matches, and sha1 recorded in the cache
 586          * matches.  I.e. we know the file in the work tree really is
 587          * the same as the <name, sha1> pair.
 588          */
 589         return 1;
 590 }
 591
 592 static struct sha1_size_cache {
 593         unsigned char sha1[20];
 594         unsigned long size;
 595 } **sha1_size_cache;
 596 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 597
 598 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 599                                                  int find_only,
 600                                                  unsigned long size)
 601 {
 602         int first, last;
 603         struct sha1_size_cache *e;
 604
 605         first = 0;
 606         last = sha1_size_cache_nr;
 607         while (last > first) {
 608                 int cmp, next = (last + first) >> 1;
 609                 e = sha1_size_cache[next];
 610                 cmp = memcmp(e->sha1, sha1, 20);
 611                 if (!cmp)
 612                         return e;
 613                 if (cmp < 0) {
 614                         last = next;
 615                         continue;
 616                 }
 617                 first = next+1;
 618         }
 619         /* not found */
 620         if (find_only)
 621                 return NULL;
 622         /* insert to make it at "first" */
 623         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 624                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 625                 sha1_size_cache = xrealloc(sha1_size_cache,
 626                                            sha1_size_cache_alloc *
 627                                            sizeof(*sha1_size_cache));
 628         }
 629         sha1_size_cache_nr++;
 630         if (first < sha1_size_cache_nr)
 631                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 632                         (sha1_size_cache_nr - first - 1) *
 633                         sizeof(*sha1_size_cache));
 634         e = xmalloc(sizeof(struct sha1_size_cache));
 635         sha1_size_cache[first] = e;
 636         memcpy(e->sha1, sha1, 20);
 637         e->size = size;
 638         return e;
 639 }
 640
 641 /*
 642  * While doing rename detection and pickaxe operation, we may need to
 643  * grab the data for the blob (or file) for our own in-core comparison.
 644  * diff_filespec has data and size fields for this purpose.
 645  */
 646 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 647 {
 648         int err = 0;
 649         if (!DIFF_FILE_VALID(s))
 650                 die("internal error: asking to populate invalid file.");
 651         if (S_ISDIR(s->mode))
 652                 return -1;
 653
 654         if (!use_size_cache)
 655                 size_only = 0;
 656
 657         if (s->data)
 658                 return err;
 659         if (!s->sha1_valid ||
 660             work_tree_matches(s->path, s->sha1)) {
 661                 struct stat st;
 662                 int fd;
 663                 if (lstat(s->path, &st) < 0) {
 664                         if (errno == ENOENT) {
 665                         err_empty:
 666                                 err = -1;
 667                         empty:
 668                                 s->data = "";
 669                                 s->size = 0;
 670                                 return err;
 671                         }
 672                 }
 673                 s->size = st.st_size;
 674                 if (!s->size)
 675                         goto empty;
 676                 if (size_only)
 677                         return 0;
 678                 if (S_ISLNK(st.st_mode)) {
 679                         int ret;
 680                         s->data = xmalloc(s->size);
 681                         s->should_free = 1;
 682                         ret = readlink(s->path, s->data, s->size);
 683                         if (ret < 0) {
 684                                 free(s->data);
 685                                 goto err_empty;
 686                         }
 687                         return 0;
 688                 }
 689                 fd = open(s->path, O_RDONLY);
 690                 if (fd < 0)
 691                         goto err_empty;
 692                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 693                 close(fd);
 694                 if (s->data == MAP_FAILED)
 695                         goto err_empty;
 696                 s->should_munmap = 1;
 697         }
 698         else {
 699                 char type[20];
 700                 struct sha1_size_cache *e;
 701
 702                 if (size_only) {
 703                         e = locate_size_cache(s->sha1, 1, 0);
 704                         if (e) {
 705                                 s->size = e->size;
 706                                 return 0;
 707                         }
 708                         if (!sha1_object_info(s->sha1, type, &s->size))
 709                                 locate_size_cache(s->sha1, 0, s->size);
 710                 }
 711                 else {
 712                         s->data = read_sha1_file(s->sha1, type, &s->size);
 713                         s->should_free = 1;
 714                 }
 715         }
 716         return 0;
 717 }
 718
 719 void diff_free_filespec_data(struct diff_filespec *s)
 720 {
 721         if (s->should_free)
 722                 free(s->data);
 723         else if (s->should_munmap)
 724                 munmap(s->data, s->size);
 725         s->should_free = s->should_munmap = 0;
 726         s->data = NULL;
 727         free(s->cnt_data);
 728         s->cnt_data = NULL;
 729 }
 730
 731 static void prep_temp_blob(struct diff_tempfile *temp,
 732                            void *blob,
 733                            unsigned long size,
 734                            const unsigned char *sha1,
 735                            int mode)
 736 {
 737         int fd;
 738
 739         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 740         if (fd < 0)
 741                 die("unable to create temp-file");
 742         if (write(fd, blob, size) != size)
 743                 die("unable to write temp-file");
 744         close(fd);
 745         temp->name = temp->tmp_path;
 746         strcpy(temp->hex, sha1_to_hex(sha1));
 747         temp->hex[40] = 0;
 748         sprintf(temp->mode, "%06o", mode);
 749 }
 750
 751 static void prepare_temp_file(const char *name,
 752                               struct diff_tempfile *temp,
 753                               struct diff_filespec *one)
 754 {
 755         if (!DIFF_FILE_VALID(one)) {
 756         not_a_valid_file:
 757                 /* A '-' entry produces this for file-2, and
 758                  * a '+' entry produces this for file-1.
 759                  */
 760                 temp->name = "/dev/null";
 761                 strcpy(temp->hex, ".");
 762                 strcpy(temp->mode, ".");
 763                 return;
 764         }
 765
 766         if (!one->sha1_valid ||
 767             work_tree_matches(name, one->sha1)) {
 768                 struct stat st;
 769                 if (lstat(name, &st) < 0) {
 770                         if (errno == ENOENT)
 771                                 goto not_a_valid_file;
 772                         die("stat(%s): %s", name, strerror(errno));
 773                 }
 774                 if (S_ISLNK(st.st_mode)) {
 775                         int ret;
 776                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
 777                         if (sizeof(buf) <= st.st_size)
 778                                 die("symlink too long: %s", name);
 779                         ret = readlink(name, buf, st.st_size);
 780                         if (ret < 0)
 781                                 die("readlink(%s)", name);
 782                         prep_temp_blob(temp, buf, st.st_size,
 783                                        (one->sha1_valid ?
 784                                         one->sha1 : null_sha1),
 785                                        (one->sha1_valid ?
 786                                         one->mode : S_IFLNK));
 787                 }
 788                 else {
 789                         /* we can borrow from the file in the work tree */
 790                         temp->name = name;
 791                         if (!one->sha1_valid)
 792                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 793                         else
 794                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 795                         /* Even though we may sometimes borrow the
 796                          * contents from the work tree, we always want
 797                          * one->mode.  mode is trustworthy even when
 798                          * !(one->sha1_valid), as long as
 799                          * DIFF_FILE_VALID(one).
 800                          */
 801                         sprintf(temp->mode, "%06o", one->mode);
 802                 }
 803                 return;
 804         }
 805         else {
 806                 if (diff_populate_filespec(one, 0))
 807                         die("cannot read data blob for %s", one->path);
 808                 prep_temp_blob(temp, one->data, one->size,
 809                                one->sha1, one->mode);
 810         }
 811 }
 812
 813 static void remove_tempfile(void)
 814 {
 815         int i;
 816
 817         for (i = 0; i < 2; i++)
 818                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 819                         unlink(diff_temp[i].name);
 820                         diff_temp[i].name = NULL;
 821                 }
 822 }
 823
 824 static void remove_tempfile_on_signal(int signo)
 825 {
 826         remove_tempfile();
 827         signal(SIGINT, SIG_DFL);
 828         raise(signo);
 829 }
 830
 831 static int spawn_prog(const char *pgm, const char **arg)
 832 {
 833         pid_t pid;
 834         int status;
 835
 836         fflush(NULL);
 837         pid = fork();
 838         if (pid < 0)
 839                 die("unable to fork");
 840         if (!pid) {
 841                 execvp(pgm, (char *const*) arg);
 842                 exit(255);
 843         }
 844
 845         while (waitpid(pid, &status, 0) < 0) {
 846                 if (errno == EINTR)
 847                         continue;
 848                 return -1;
 849         }
 850
 851         /* Earlier we did not check the exit status because
 852          * diff exits non-zero if files are different, and
 853          * we are not interested in knowing that.  It was a
 854          * mistake which made it harder to quit a diff-*
 855          * session that uses the git-apply-patch-script as
 856          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 857          * should also exit non-zero only when it wants to
 858          * abort the entire diff-* session.
 859          */
 860         if (WIFEXITED(status) && !WEXITSTATUS(status))
 861                 return 0;
 862         return -1;
 863 }
 864
 865 /* An external diff command takes:
 866  *
 867  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 868  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 869  *
 870  */
 871 static void run_external_diff(const char *pgm,
 872                               const char *name,
 873                               const char *other,
 874                               struct diff_filespec *one,
 875                               struct diff_filespec *two,
 876                               const char *xfrm_msg,
 877                               int complete_rewrite)
 878 {
 879         const char *spawn_arg[10];
 880         struct diff_tempfile *temp = diff_temp;
 881         int retval;
 882         static int atexit_asked = 0;
 883         const char *othername;
 884         const char **arg = &spawn_arg[0];
 885
 886         othername = (other? other : name);
 887         if (one && two) {
 888                 prepare_temp_file(name, &temp[0], one);
 889                 prepare_temp_file(othername, &temp[1], two);
 890                 if (! atexit_asked &&
 891                     (temp[0].name == temp[0].tmp_path ||
 892                      temp[1].name == temp[1].tmp_path)) {
 893                         atexit_asked = 1;
 894                         atexit(remove_tempfile);
 895                 }
 896                 signal(SIGINT, remove_tempfile_on_signal);
 897         }
 898
 899         if (one && two) {
 900                 *arg++ = pgm;
 901                 *arg++ = name;
 902                 *arg++ = temp[0].name;
 903                 *arg++ = temp[0].hex;
 904                 *arg++ = temp[0].mode;
 905                 *arg++ = temp[1].name;
 906                 *arg++ = temp[1].hex;
 907                 *arg++ = temp[1].mode;
 908                 if (other) {
 909                         *arg++ = other;
 910                         *arg++ = xfrm_msg;
 911                 }
 912         } else {
 913                 *arg++ = pgm;
 914                 *arg++ = name;
 915         }
 916         *arg = NULL;
 917         retval = spawn_prog(pgm, spawn_arg);
 918         remove_tempfile();
 919         if (retval) {
 920                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 921                 exit(1);
 922         }
 923 }
 924
 925 static void run_diff_cmd(const char *pgm,
 926                          const char *name,
 927                          const char *other,
 928                          struct diff_filespec *one,
 929                          struct diff_filespec *two,
 930                          const char *xfrm_msg,
 931                          int complete_rewrite)
 932 {
 933         if (pgm) {
 934                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 935                                   complete_rewrite);
 936                 return;
 937         }
 938         if (one && two)
 939                 builtin_diff(name, other ? other : name,
 940                              one, two, xfrm_msg, complete_rewrite);
 941         else
 942                 printf("* Unmerged path %s\n", name);
 943 }
 944
 945 static void diff_fill_sha1_info(struct diff_filespec *one)
 946 {
 947         if (DIFF_FILE_VALID(one)) {
 948                 if (!one->sha1_valid) {
 949                         struct stat st;
 950                         if (lstat(one->path, &st) < 0)
 951                                 die("stat %s", one->path);
 952                         if (index_path(one->sha1, one->path, &st, 0))
 953                                 die("cannot hash %s\n", one->path);
 954                 }
 955         }
 956         else
 957                 memset(one->sha1, 0, 20);
 958 }
 959
 960 static void run_diff(struct diff_filepair *p, struct diff_options *o)
 961 {
 962         const char *pgm = external_diff();
 963         char msg[PATH_MAX*2+300], *xfrm_msg;
 964         struct diff_filespec *one;
 965         struct diff_filespec *two;
 966         const char *name;
 967         const char *other;
 968         char *name_munged, *other_munged;
 969         int complete_rewrite = 0;
 970         int len;
 971
 972         if (DIFF_PAIR_UNMERGED(p)) {
 973                 /* unmerged */
 974                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, 0);
 975                 return;
 976         }
 977
 978         name = p->one->path;
 979         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 980         name_munged = quote_one(name);
 981         other_munged = quote_one(other);
 982         one = p->one; two = p->two;
 983
 984         diff_fill_sha1_info(one);
 985         diff_fill_sha1_info(two);
 986
 987         len = 0;
 988         switch (p->status) {
 989         case DIFF_STATUS_COPIED:
 990                 len += snprintf(msg + len, sizeof(msg) - len,
 991                                 "similarity index %d%%\n"
 992                                 "copy from %s\n"
 993                                 "copy to %s\n",
 994                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 995                                 name_munged, other_munged);
 996                 break;
 997         case DIFF_STATUS_RENAMED:
 998                 len += snprintf(msg + len, sizeof(msg) - len,
 999                                 "similarity index %d%%\n"
1000                                 "rename from %s\n"
1001                                 "rename to %s\n",
1002                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1003                                 name_munged, other_munged);
1004                 break;
1005         case DIFF_STATUS_MODIFIED:
1006                 if (p->score) {
1007                         len += snprintf(msg + len, sizeof(msg) - len,
1008                                         "dissimilarity index %d%%\n",
1009                                         (int)(0.5 + p->score *
1010                                               100.0/MAX_SCORE));
1011                         complete_rewrite = 1;
1012                         break;
1013                 }
1014                 /* fallthru */
1015         default:
1016                 /* nothing */
1017                 ;
1018         }
1019
1020         if (memcmp(one->sha1, two->sha1, 20)) {
1021                 char one_sha1[41];
1022                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1023                 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
1024
1025                 len += snprintf(msg + len, sizeof(msg) - len,
1026                                 "index %.*s..%.*s",
1027                                 abbrev, one_sha1, abbrev,
1028                                 sha1_to_hex(two->sha1));
1029                 if (one->mode == two->mode)
1030                         len += snprintf(msg + len, sizeof(msg) - len,
1031                                         " %06o", one->mode);
1032                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1033         }
1034
1035         if (len)
1036                 msg[--len] = 0;
1037         xfrm_msg = len ? msg : NULL;
1038
1039         if (!pgm &&
1040             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1041             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1042                 /* a filepair that changes between file and symlink
1043                  * needs to be split into deletion and creation.
1044                  */
1045                 struct diff_filespec *null = alloc_filespec(two->path);
1046                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, 0);
1047                 free(null);
1048                 null = alloc_filespec(one->path);
1049                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, 0);
1050                 free(null);
1051         }
1052         else
1053                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg,
1054                              complete_rewrite);
1055
1056         free(name_munged);
1057         free(other_munged);
1058 }
1059
1060 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1061                          struct diffstat_t *diffstat)
1062 {
1063         const char *name;
1064         const char *other;
1065         int complete_rewrite = 0;
1066
1067         if (DIFF_PAIR_UNMERGED(p)) {
1068                 /* unmerged */
1069                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, 0);
1070                 return;
1071         }
1072
1073         name = p->one->path;
1074         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1075
1076         diff_fill_sha1_info(p->one);
1077         diff_fill_sha1_info(p->two);
1078
1079         if (p->status == DIFF_STATUS_MODIFIED && p->score)
1080                 complete_rewrite = 1;
1081         builtin_diffstat(name, other, p->one, p->two, diffstat, complete_rewrite);
1082 }
1083
1084 void diff_setup(struct diff_options *options)
1085 {
1086         memset(options, 0, sizeof(*options));
1087         options->output_format = DIFF_FORMAT_RAW;
1088         options->line_termination = '\n';
1089         options->break_opt = -1;
1090         options->rename_limit = -1;
1091
1092         options->change = diff_change;
1093         options->add_remove = diff_addremove;
1094 }
1095
1096 int diff_setup_done(struct diff_options *options)
1097 {
1098         if ((options->find_copies_harder &&
1099              options->detect_rename != DIFF_DETECT_COPY) ||
1100             (0 <= options->rename_limit && !options->detect_rename))
1101                 return -1;
1102
1103         /*
1104          * These cases always need recursive; we do not drop caller-supplied
1105          * recursive bits for other formats here.
1106          */
1107         if ((options->output_format == DIFF_FORMAT_PATCH) ||
1108             (options->output_format == DIFF_FORMAT_DIFFSTAT))
1109                 options->recursive = 1;
1110
1111         if (options->detect_rename && options->rename_limit < 0)
1112                 options->rename_limit = diff_rename_limit_default;
1113         if (options->setup & DIFF_SETUP_USE_CACHE) {
1114                 if (!active_cache)
1115                         /* read-cache does not die even when it fails
1116                          * so it is safe for us to do this here.  Also
1117                          * it does not smudge active_cache or active_nr
1118                          * when it fails, so we do not have to worry about
1119                          * cleaning it up ourselves either.
1120                          */
1121                         read_cache();
1122         }
1123         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1124                 use_size_cache = 1;
1125         if (options->abbrev <= 0 || 40 < options->abbrev)
1126                 options->abbrev = 40; /* full */
1127
1128         return 0;
1129 }
1130
1131 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1132 {
1133         const char *arg = av[0];
1134         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1135                 options->output_format = DIFF_FORMAT_PATCH;
1136         else if (!strcmp(arg, "--patch-with-raw")) {
1137                 options->output_format = DIFF_FORMAT_PATCH;
1138                 options->with_raw = 1;
1139         }
1140         else if (!strcmp(arg, "--stat"))
1141                 options->output_format = DIFF_FORMAT_DIFFSTAT;
1142         else if (!strcmp(arg, "--patch-with-stat")) {
1143                 options->output_format = DIFF_FORMAT_PATCH;
1144                 options->with_stat = 1;
1145         }
1146         else if (!strcmp(arg, "-z"))
1147                 options->line_termination = 0;
1148         else if (!strncmp(arg, "-l", 2))
1149                 options->rename_limit = strtoul(arg+2, NULL, 10);
1150         else if (!strcmp(arg, "--full-index"))
1151                 options->full_index = 1;
1152         else if (!strcmp(arg, "--name-only"))
1153                 options->output_format = DIFF_FORMAT_NAME;
1154         else if (!strcmp(arg, "--name-status"))
1155                 options->output_format = DIFF_FORMAT_NAME_STATUS;
1156         else if (!strcmp(arg, "-R"))
1157                 options->reverse_diff = 1;
1158         else if (!strncmp(arg, "-S", 2))
1159                 options->pickaxe = arg + 2;
1160         else if (!strcmp(arg, "-s"))
1161                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1162         else if (!strncmp(arg, "-O", 2))
1163                 options->orderfile = arg + 2;
1164         else if (!strncmp(arg, "--diff-filter=", 14))
1165                 options->filter = arg + 14;
1166         else if (!strcmp(arg, "--pickaxe-all"))
1167                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1168         else if (!strcmp(arg, "--pickaxe-regex"))
1169                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1170         else if (!strncmp(arg, "-B", 2)) {
1171                 if ((options->break_opt =
1172                      diff_scoreopt_parse(arg)) == -1)
1173                         return -1;
1174         }
1175         else if (!strncmp(arg, "-M", 2)) {
1176                 if ((options->rename_score =
1177                      diff_scoreopt_parse(arg)) == -1)
1178                         return -1;
1179                 options->detect_rename = DIFF_DETECT_RENAME;
1180         }
1181         else if (!strncmp(arg, "-C", 2)) {
1182                 if ((options->rename_score =
1183                      diff_scoreopt_parse(arg)) == -1)
1184                         return -1;
1185                 options->detect_rename = DIFF_DETECT_COPY;
1186         }
1187         else if (!strcmp(arg, "--find-copies-harder"))
1188                 options->find_copies_harder = 1;
1189         else if (!strcmp(arg, "--abbrev"))
1190                 options->abbrev = DEFAULT_ABBREV;
1191         else if (!strncmp(arg, "--abbrev=", 9)) {
1192                 options->abbrev = strtoul(arg + 9, NULL, 10);
1193                 if (options->abbrev < MINIMUM_ABBREV)
1194                         options->abbrev = MINIMUM_ABBREV;
1195                 else if (40 < options->abbrev)
1196                         options->abbrev = 40;
1197         }
1198         else
1199                 return 0;
1200         return 1;
1201 }
1202
1203 static int parse_num(const char **cp_p)
1204 {
1205         unsigned long num, scale;
1206         int ch, dot;
1207         const char *cp = *cp_p;
1208
1209         num = 0;
1210         scale = 1;
1211         dot = 0;
1212         for(;;) {
1213                 ch = *cp;
1214                 if ( !dot && ch == '.' ) {
1215                         scale = 1;
1216                         dot = 1;
1217                 } else if ( ch == '%' ) {
1218                         scale = dot ? scale*100 : 100;
1219                         cp++;   /* % is always at the end */
1220                         break;
1221                 } else if ( ch >= '0' && ch <= '9' ) {
1222                         if ( scale < 100000 ) {
1223                                 scale *= 10;
1224                                 num = (num*10) + (ch-'0');
1225                         }
1226                 } else {
1227                         break;
1228                 }
1229                 cp++;
1230         }
1231         *cp_p = cp;
1232
1233         /* user says num divided by scale and we say internally that
1234          * is MAX_SCORE * num / scale.
1235          */
1236         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1237 }
1238
1239 int diff_scoreopt_parse(const char *opt)
1240 {
1241         int opt1, opt2, cmd;
1242
1243         if (*opt++ != '-')
1244                 return -1;
1245         cmd = *opt++;
1246         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1247                 return -1; /* that is not a -M, -C nor -B option */
1248
1249         opt1 = parse_num(&opt);
1250         if (cmd != 'B')
1251                 opt2 = 0;
1252         else {
1253                 if (*opt == 0)
1254                         opt2 = 0;
1255                 else if (*opt != '/')
1256                         return -1; /* we expect -B80/99 or -B80 */
1257                 else {
1258                         opt++;
1259                         opt2 = parse_num(&opt);
1260                 }
1261         }
1262         if (*opt != 0)
1263                 return -1;
1264         return opt1 | (opt2 << 16);
1265 }
1266
1267 struct diff_queue_struct diff_queued_diff;
1268
1269 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1270 {
1271         if (queue->alloc <= queue->nr) {
1272                 queue->alloc = alloc_nr(queue->alloc);
1273                 queue->queue = xrealloc(queue->queue,
1274                                         sizeof(dp) * queue->alloc);
1275         }
1276         queue->queue[queue->nr++] = dp;
1277 }
1278
1279 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1280                                  struct diff_filespec *one,
1281                                  struct diff_filespec *two)
1282 {
1283         struct diff_filepair *dp = xmalloc(sizeof(*dp));
1284         dp->one = one;
1285         dp->two = two;
1286         dp->score = 0;
1287         dp->status = 0;
1288         dp->source_stays = 0;
1289         dp->broken_pair = 0;
1290         if (queue)
1291                 diff_q(queue, dp);
1292         return dp;
1293 }
1294
1295 void diff_free_filepair(struct diff_filepair *p)
1296 {
1297         diff_free_filespec_data(p->one);
1298         diff_free_filespec_data(p->two);
1299         free(p->one);
1300         free(p->two);
1301         free(p);
1302 }
1303
1304 /* This is different from find_unique_abbrev() in that
1305  * it stuffs the result with dots for alignment.
1306  */
1307 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1308 {
1309         int abblen;
1310         const char *abbrev;
1311         if (len == 40)
1312                 return sha1_to_hex(sha1);
1313
1314         abbrev = find_unique_abbrev(sha1, len);
1315         if (!abbrev)
1316                 return sha1_to_hex(sha1);
1317         abblen = strlen(abbrev);
1318         if (abblen < 37) {
1319                 static char hex[41];
1320                 if (len < abblen && abblen <= len + 2)
1321                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1322                 else
1323                         sprintf(hex, "%s...", abbrev);
1324                 return hex;
1325         }
1326         return sha1_to_hex(sha1);
1327 }
1328
1329 static void diff_flush_raw(struct diff_filepair *p,
1330                            int line_termination,
1331                            int inter_name_termination,
1332                            struct diff_options *options,
1333                            int output_format)
1334 {
1335         int two_paths;
1336         char status[10];
1337         int abbrev = options->abbrev;
1338         const char *path_one, *path_two;
1339
1340         path_one = p->one->path;
1341         path_two = p->two->path;
1342         if (line_termination) {
1343                 path_one = quote_one(path_one);
1344                 path_two = quote_one(path_two);
1345         }
1346
1347         if (p->score)
1348                 sprintf(status, "%c%03d", p->status,
1349                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1350         else {
1351                 status[0] = p->status;
1352                 status[1] = 0;
1353         }
1354         switch (p->status) {
1355         case DIFF_STATUS_COPIED:
1356         case DIFF_STATUS_RENAMED:
1357                 two_paths = 1;
1358                 break;
1359         case DIFF_STATUS_ADDED:
1360         case DIFF_STATUS_DELETED:
1361                 two_paths = 0;
1362                 break;
1363         default:
1364                 two_paths = 0;
1365                 break;
1366         }
1367         if (output_format != DIFF_FORMAT_NAME_STATUS) {
1368                 printf(":%06o %06o %s ",
1369                        p->one->mode, p->two->mode,
1370                        diff_unique_abbrev(p->one->sha1, abbrev));
1371                 printf("%s ",
1372                        diff_unique_abbrev(p->two->sha1, abbrev));
1373         }
1374         printf("%s%c%s", status, inter_name_termination, path_one);
1375         if (two_paths)
1376                 printf("%c%s", inter_name_termination, path_two);
1377         putchar(line_termination);
1378         if (path_one != p->one->path)
1379                 free((void*)path_one);
1380         if (path_two != p->two->path)
1381                 free((void*)path_two);
1382 }
1383
1384 static void diff_flush_name(struct diff_filepair *p,
1385                             int inter_name_termination,
1386                             int line_termination)
1387 {
1388         char *path = p->two->path;
1389
1390         if (line_termination)
1391                 path = quote_one(p->two->path);
1392         else
1393                 path = p->two->path;
1394         printf("%s%c", path, line_termination);
1395         if (p->two->path != path)
1396                 free(path);
1397 }
1398
1399 int diff_unmodified_pair(struct diff_filepair *p)
1400 {
1401         /* This function is written stricter than necessary to support
1402          * the currently implemented transformers, but the idea is to
1403          * let transformers to produce diff_filepairs any way they want,
1404          * and filter and clean them up here before producing the output.
1405          */
1406         struct diff_filespec *one, *two;
1407
1408         if (DIFF_PAIR_UNMERGED(p))
1409                 return 0; /* unmerged is interesting */
1410
1411         one = p->one;
1412         two = p->two;
1413
1414         /* deletion, addition, mode or type change
1415          * and rename are all interesting.
1416          */
1417         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1418             DIFF_PAIR_MODE_CHANGED(p) ||
1419             strcmp(one->path, two->path))
1420                 return 0;
1421
1422         /* both are valid and point at the same path.  that is, we are
1423          * dealing with a change.
1424          */
1425         if (one->sha1_valid && two->sha1_valid &&
1426             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1427                 return 1; /* no change */
1428         if (!one->sha1_valid && !two->sha1_valid)
1429                 return 1; /* both look at the same file on the filesystem. */
1430         return 0;
1431 }
1432
1433 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1434 {
1435         if (diff_unmodified_pair(p))
1436                 return;
1437
1438         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1439             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1440                 return; /* no tree diffs in patch format */
1441
1442         run_diff(p, o);
1443 }
1444
1445 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1446                             struct diffstat_t *diffstat)
1447 {
1448         if (diff_unmodified_pair(p))
1449                 return;
1450
1451         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1452             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1453                 return; /* no tree diffs in patch format */
1454
1455         run_diffstat(p, o, diffstat);
1456 }
1457
1458 int diff_queue_is_empty(void)
1459 {
1460         struct diff_queue_struct *q = &diff_queued_diff;
1461         int i;
1462         for (i = 0; i < q->nr; i++)
1463                 if (!diff_unmodified_pair(q->queue[i]))
1464                         return 0;
1465         return 1;
1466 }
1467
1468 #if DIFF_DEBUG
1469 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1470 {
1471         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1472                 x, one ? one : "",
1473                 s->path,
1474                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1475                 s->mode,
1476                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1477         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1478                 x, one ? one : "",
1479                 s->size, s->xfrm_flags);
1480 }
1481
1482 void diff_debug_filepair(const struct diff_filepair *p, int i)
1483 {
1484         diff_debug_filespec(p->one, i, "one");
1485         diff_debug_filespec(p->two, i, "two");
1486         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1487                 p->score, p->status ? p->status : '?',
1488                 p->source_stays, p->broken_pair);
1489 }
1490
1491 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1492 {
1493         int i;
1494         if (msg)
1495                 fprintf(stderr, "%s\n", msg);
1496         fprintf(stderr, "q->nr = %d\n", q->nr);
1497         for (i = 0; i < q->nr; i++) {
1498                 struct diff_filepair *p = q->queue[i];
1499                 diff_debug_filepair(p, i);
1500         }
1501 }
1502 #endif
1503
1504 static void diff_resolve_rename_copy(void)
1505 {
1506         int i, j;
1507         struct diff_filepair *p, *pp;
1508         struct diff_queue_struct *q = &diff_queued_diff;
1509
1510         diff_debug_queue("resolve-rename-copy", q);
1511
1512         for (i = 0; i < q->nr; i++) {
1513                 p = q->queue[i];
1514                 p->status = 0; /* undecided */
1515                 if (DIFF_PAIR_UNMERGED(p))
1516                         p->status = DIFF_STATUS_UNMERGED;
1517                 else if (!DIFF_FILE_VALID(p->one))
1518                         p->status = DIFF_STATUS_ADDED;
1519                 else if (!DIFF_FILE_VALID(p->two))
1520                         p->status = DIFF_STATUS_DELETED;
1521                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1522                         p->status = DIFF_STATUS_TYPE_CHANGED;
1523
1524                 /* from this point on, we are dealing with a pair
1525                  * whose both sides are valid and of the same type, i.e.
1526                  * either in-place edit or rename/copy edit.
1527                  */
1528                 else if (DIFF_PAIR_RENAME(p)) {
1529                         if (p->source_stays) {
1530                                 p->status = DIFF_STATUS_COPIED;
1531                                 continue;
1532                         }
1533                         /* See if there is some other filepair that
1534                          * copies from the same source as us.  If so
1535                          * we are a copy.  Otherwise we are either a
1536                          * copy if the path stays, or a rename if it
1537                          * does not, but we already handled "stays" case.
1538                          */
1539                         for (j = i + 1; j < q->nr; j++) {
1540                                 pp = q->queue[j];
1541                                 if (strcmp(pp->one->path, p->one->path))
1542                                         continue; /* not us */
1543                                 if (!DIFF_PAIR_RENAME(pp))
1544                                         continue; /* not a rename/copy */
1545                                 /* pp is a rename/copy from the same source */
1546                                 p->status = DIFF_STATUS_COPIED;
1547                                 break;
1548                         }
1549                         if (!p->status)
1550                                 p->status = DIFF_STATUS_RENAMED;
1551                 }
1552                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1553                          p->one->mode != p->two->mode)
1554                         p->status = DIFF_STATUS_MODIFIED;
1555                 else {
1556                         /* This is a "no-change" entry and should not
1557                          * happen anymore, but prepare for broken callers.
1558                          */
1559                         error("feeding unmodified %s to diffcore",
1560                               p->one->path);
1561                         p->status = DIFF_STATUS_UNKNOWN;
1562                 }
1563         }
1564         diff_debug_queue("resolve-rename-copy done", q);
1565 }
1566
1567 static void flush_one_pair(struct diff_filepair *p,
1568                            int diff_output_format,
1569                            struct diff_options *options,
1570                            struct diffstat_t *diffstat)
1571 {
1572         int inter_name_termination = '\t';
1573         int line_termination = options->line_termination;
1574         if (!line_termination)
1575                 inter_name_termination = 0;
1576
1577         switch (p->status) {
1578         case DIFF_STATUS_UNKNOWN:
1579                 break;
1580         case 0:
1581                 die("internal error in diff-resolve-rename-copy");
1582                 break;
1583         default:
1584                 switch (diff_output_format) {
1585                 case DIFF_FORMAT_DIFFSTAT:
1586                         diff_flush_stat(p, options, diffstat);
1587                         break;
1588                 case DIFF_FORMAT_PATCH:
1589                         diff_flush_patch(p, options);
1590                         break;
1591                 case DIFF_FORMAT_RAW:
1592                 case DIFF_FORMAT_NAME_STATUS:
1593                         diff_flush_raw(p, line_termination,
1594                                        inter_name_termination,
1595                                        options, diff_output_format);
1596                         break;
1597                 case DIFF_FORMAT_NAME:
1598                         diff_flush_name(p,
1599                                         inter_name_termination,
1600                                         line_termination);
1601                         break;
1602                 case DIFF_FORMAT_NO_OUTPUT:
1603                         break;
1604                 }
1605         }
1606 }
1607
1608 void diff_flush(struct diff_options *options)
1609 {
1610         struct diff_queue_struct *q = &diff_queued_diff;
1611         int i;
1612         int diff_output_format = options->output_format;
1613         struct diffstat_t *diffstat = NULL;
1614
1615         if (diff_output_format == DIFF_FORMAT_DIFFSTAT || options->with_stat) {
1616                 diffstat = xcalloc(sizeof (struct diffstat_t), 1);
1617                 diffstat->xm.consume = diffstat_consume;
1618         }
1619
1620         if (options->with_raw) {
1621                 for (i = 0; i < q->nr; i++) {
1622                         struct diff_filepair *p = q->queue[i];
1623                         flush_one_pair(p, DIFF_FORMAT_RAW, options, NULL);
1624                 }
1625                 putchar(options->line_termination);
1626         }
1627         if (options->with_stat) {
1628                 for (i = 0; i < q->nr; i++) {
1629                         struct diff_filepair *p = q->queue[i];
1630                         flush_one_pair(p, DIFF_FORMAT_DIFFSTAT, options,
1631                                        diffstat);
1632                 }
1633                 show_stats(diffstat);
1634                 free(diffstat);
1635                 diffstat = NULL;
1636                 putchar(options->line_termination);
1637         }
1638         for (i = 0; i < q->nr; i++) {
1639                 struct diff_filepair *p = q->queue[i];
1640                 flush_one_pair(p, diff_output_format, options, diffstat);
1641                 diff_free_filepair(p);
1642         }
1643
1644         if (diffstat) {
1645                 show_stats(diffstat);
1646                 free(diffstat);
1647         }
1648
1649         free(q->queue);
1650         q->queue = NULL;
1651         q->nr = q->alloc = 0;
1652 }
1653
1654 static void diffcore_apply_filter(const char *filter)
1655 {
1656         int i;
1657         struct diff_queue_struct *q = &diff_queued_diff;
1658         struct diff_queue_struct outq;
1659         outq.queue = NULL;
1660         outq.nr = outq.alloc = 0;
1661
1662         if (!filter)
1663                 return;
1664
1665         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1666                 int found;
1667                 for (i = found = 0; !found && i < q->nr; i++) {
1668                         struct diff_filepair *p = q->queue[i];
1669                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1670                              ((p->score &&
1671                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1672                               (!p->score &&
1673                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1674                             ((p->status != DIFF_STATUS_MODIFIED) &&
1675                              strchr(filter, p->status)))
1676                                 found++;
1677                 }
1678                 if (found)
1679                         return;
1680
1681                 /* otherwise we will clear the whole queue
1682                  * by copying the empty outq at the end of this
1683                  * function, but first clear the current entries
1684                  * in the queue.
1685                  */
1686                 for (i = 0; i < q->nr; i++)
1687                         diff_free_filepair(q->queue[i]);
1688         }
1689         else {
1690                 /* Only the matching ones */
1691                 for (i = 0; i < q->nr; i++) {
1692                         struct diff_filepair *p = q->queue[i];
1693
1694                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1695                              ((p->score &&
1696                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1697                               (!p->score &&
1698                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1699                             ((p->status != DIFF_STATUS_MODIFIED) &&
1700                              strchr(filter, p->status)))
1701                                 diff_q(&outq, p);
1702                         else
1703                                 diff_free_filepair(p);
1704                 }
1705         }
1706         free(q->queue);
1707         *q = outq;
1708 }
1709
1710 void diffcore_std(struct diff_options *options)
1711 {
1712         if (options->break_opt != -1)
1713                 diffcore_break(options->break_opt);
1714         if (options->detect_rename)
1715                 diffcore_rename(options);
1716         if (options->break_opt != -1)
1717                 diffcore_merge_broken();
1718         if (options->pickaxe)
1719                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1720         if (options->orderfile)
1721                 diffcore_order(options->orderfile);
1722         diff_resolve_rename_copy();
1723         diffcore_apply_filter(options->filter);
1724 }
1725
1726
1727 void diffcore_std_no_resolve(struct diff_options *options)
1728 {
1729         if (options->pickaxe)
1730                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1731         if (options->orderfile)
1732                 diffcore_order(options->orderfile);
1733         diffcore_apply_filter(options->filter);
1734 }
1735
1736 void diff_addremove(struct diff_options *options,
1737                     int addremove, unsigned mode,
1738                     const unsigned char *sha1,
1739                     const char *base, const char *path)
1740 {
1741         char concatpath[PATH_MAX];
1742         struct diff_filespec *one, *two;
1743
1744         /* This may look odd, but it is a preparation for
1745          * feeding "there are unchanged files which should
1746          * not produce diffs, but when you are doing copy
1747          * detection you would need them, so here they are"
1748          * entries to the diff-core.  They will be prefixed
1749          * with something like '=' or '*' (I haven't decided
1750          * which but should not make any difference).
1751          * Feeding the same new and old to diff_change()
1752          * also has the same effect.
1753          * Before the final output happens, they are pruned after
1754          * merged into rename/copy pairs as appropriate.
1755          */
1756         if (options->reverse_diff)
1757                 addremove = (addremove == '+' ? '-' :
1758                              addremove == '-' ? '+' : addremove);
1759
1760         if (!path) path = "";
1761         sprintf(concatpath, "%s%s", base, path);
1762         one = alloc_filespec(concatpath);
1763         two = alloc_filespec(concatpath);
1764
1765         if (addremove != '+')
1766                 fill_filespec(one, sha1, mode);
1767         if (addremove != '-')
1768                 fill_filespec(two, sha1, mode);
1769
1770         diff_queue(&diff_queued_diff, one, two);
1771 }
1772
1773 void diff_change(struct diff_options *options,
1774                  unsigned old_mode, unsigned new_mode,
1775                  const unsigned char *old_sha1,
1776                  const unsigned char *new_sha1,
1777                  const char *base, const char *path)
1778 {
1779         char concatpath[PATH_MAX];
1780         struct diff_filespec *one, *two;
1781
1782         if (options->reverse_diff) {
1783                 unsigned tmp;
1784                 const unsigned char *tmp_c;
1785                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1786                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1787         }
1788         if (!path) path = "";
1789         sprintf(concatpath, "%s%s", base, path);
1790         one = alloc_filespec(concatpath);
1791         two = alloc_filespec(concatpath);
1792         fill_filespec(one, old_sha1, old_mode);
1793         fill_filespec(two, new_sha1, new_mode);
1794
1795         diff_queue(&diff_queued_diff, one, two);
1796 }
1797
1798 void diff_unmerge(struct diff_options *options,
1799                   const char *path)
1800 {
1801         struct diff_filespec *one, *two;
1802         one = alloc_filespec(path);
1803         two = alloc_filespec(path);
1804         diff_queue(&diff_queued_diff, one, two);
1805 }