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 "delta.h"
  12 #include "xdiff-interface.h"
  13
  14 static int use_size_cache;
  15
  16 int diff_rename_limit_default = -1;
  17
  18 int git_diff_config(const char *var, const char *value)
  19 {
  20         if (!strcmp(var, "diff.renamelimit")) {
  21                 diff_rename_limit_default = git_config_int(var, value);
  22                 return 0;
  23         }
  24
  25         return git_default_config(var, value);
  26 }
  27
  28 static char *quote_one(const char *str)
  29 {
  30         int needlen;
  31         char *xp;
  32
  33         if (!str)
  34                 return NULL;
  35         needlen = quote_c_style(str, NULL, NULL, 0);
  36         if (!needlen)
  37                 return strdup(str);
  38         xp = xmalloc(needlen + 1);
  39         quote_c_style(str, xp, NULL, 0);
  40         return xp;
  41 }
  42
  43 static char *quote_two(const char *one, const char *two)
  44 {
  45         int need_one = quote_c_style(one, NULL, NULL, 1);
  46         int need_two = quote_c_style(two, NULL, NULL, 1);
  47         char *xp;
  48
  49         if (need_one + need_two) {
  50                 if (!need_one) need_one = strlen(one);
  51                 if (!need_two) need_one = strlen(two);
  52
  53                 xp = xmalloc(need_one + need_two + 3);
  54                 xp[0] = '"';
  55                 quote_c_style(one, xp + 1, NULL, 1);
  56                 quote_c_style(two, xp + need_one + 1, NULL, 1);
  57                 strcpy(xp + need_one + need_two + 1, "\"");
  58                 return xp;
  59         }
  60         need_one = strlen(one);
  61         need_two = strlen(two);
  62         xp = xmalloc(need_one + need_two + 1);
  63         strcpy(xp, one);
  64         strcpy(xp + need_one, two);
  65         return xp;
  66 }
  67
  68 static const char *external_diff(void)
  69 {
  70         static const char *external_diff_cmd = NULL;
  71         static int done_preparing = 0;
  72
  73         if (done_preparing)
  74                 return external_diff_cmd;
  75         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  76         done_preparing = 1;
  77         return external_diff_cmd;
  78 }
  79
  80 #define TEMPFILE_PATH_LEN               50
  81
  82 static struct diff_tempfile {
  83         const char *name; /* filename external diff should read from */
  84         char hex[41];
  85         char mode[10];
  86         char tmp_path[TEMPFILE_PATH_LEN];
  87 } diff_temp[2];
  88
  89 static int count_lines(const char *data, int size)
  90 {
  91         int count, ch, completely_empty = 1, nl_just_seen = 0;
  92         count = 0;
  93         while (0 < size--) {
  94                 ch = *data++;
  95                 if (ch == '\n') {
  96                         count++;
  97                         nl_just_seen = 1;
  98                         completely_empty = 0;
  99                 }
 100                 else {
 101                         nl_just_seen = 0;
 102                         completely_empty = 0;
 103                 }
 104         }
 105         if (completely_empty)
 106                 return 0;
 107         if (!nl_just_seen)
 108                 count++; /* no trailing newline */
 109         return count;
 110 }
 111
 112 static void print_line_count(int count)
 113 {
 114         switch (count) {
 115         case 0:
 116                 printf("0,0");
 117                 break;
 118         case 1:
 119                 printf("1");
 120                 break;
 121         default:
 122                 printf("1,%d", count);
 123                 break;
 124         }
 125 }
 126
 127 static void copy_file(int prefix, const char *data, int size)
 128 {
 129         int ch, nl_just_seen = 1;
 130         while (0 < size--) {
 131                 ch = *data++;
 132                 if (nl_just_seen)
 133                         putchar(prefix);
 134                 putchar(ch);
 135                 if (ch == '\n')
 136                         nl_just_seen = 1;
 137                 else
 138                         nl_just_seen = 0;
 139         }
 140         if (!nl_just_seen)
 141                 printf("\n\\ No newline at end of file\n");
 142 }
 143
 144 static void emit_rewrite_diff(const char *name_a,
 145                               const char *name_b,
 146                               struct diff_filespec *one,
 147                               struct diff_filespec *two)
 148 {
 149         int lc_a, lc_b;
 150         diff_populate_filespec(one, 0);
 151         diff_populate_filespec(two, 0);
 152         lc_a = count_lines(one->data, one->size);
 153         lc_b = count_lines(two->data, two->size);
 154         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 155         print_line_count(lc_a);
 156         printf(" +");
 157         print_line_count(lc_b);
 158         printf(" @@\n");
 159         if (lc_a)
 160                 copy_file('-', one->data, one->size);
 161         if (lc_b)
 162                 copy_file('+', two->data, two->size);
 163 }
 164
 165 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 166 {
 167         if (!DIFF_FILE_VALID(one)) {
 168                 mf->ptr = ""; /* does not matter */
 169                 mf->size = 0;
 170                 return 0;
 171         }
 172         else if (diff_populate_filespec(one, 0))
 173                 return -1;
 174         mf->ptr = one->data;
 175         mf->size = one->size;
 176         return 0;
 177 }
 178
 179 struct emit_callback {
 180         const char **label_path;
 181 };
 182
 183 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
 184 {
 185         int i;
 186         struct emit_callback *ecbdata = priv;
 187
 188         if (ecbdata->label_path[0]) {
 189                 printf("--- %s\n", ecbdata->label_path[0]);
 190                 printf("+++ %s\n", ecbdata->label_path[1]);
 191                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 192         }
 193         for (i = 0; i < nbuf; i++)
 194                 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
 195                         return -1;
 196         return 0;
 197 }
 198
 199 static char *pprint_rename(const char *a, const char *b)
 200 {
 201         const char *old = a;
 202         const char *new = b;
 203         char *name = NULL;
 204         int pfx_length, sfx_length;
 205         int len_a = strlen(a);
 206         int len_b = strlen(b);
 207
 208         /* Find common prefix */
 209         pfx_length = 0;
 210         while (*old && *new && *old == *new) {
 211                 if (*old == '/')
 212                         pfx_length = old - a + 1;
 213                 old++;
 214                 new++;
 215         }
 216
 217         /* Find common suffix */
 218         old = a + len_a;
 219         new = b + len_b;
 220         sfx_length = 0;
 221         while (a <= old && b <= new && *old == *new) {
 222                 if (*old == '/')
 223                         sfx_length = len_a - (old - a);
 224                 old--;
 225                 new--;
 226         }
 227
 228         /*
 229          * pfx{mid-a => mid-b}sfx
 230          * {pfx-a => pfx-b}sfx
 231          * pfx{sfx-a => sfx-b}
 232          * name-a => name-b
 233          */
 234         if (pfx_length + sfx_length) {
 235                 name = xmalloc(len_a + len_b - pfx_length - sfx_length + 7);
 236                 sprintf(name, "%.*s{%.*s => %.*s}%s",
 237                         pfx_length, a,
 238                         len_a - pfx_length - sfx_length, a + pfx_length,
 239                         len_b - pfx_length - sfx_length, b + pfx_length,
 240                         a + len_a - sfx_length);
 241         }
 242         else {
 243                 name = xmalloc(len_a + len_b + 5);
 244                 sprintf(name, "%s => %s", a, b);
 245         }
 246         return name;
 247 }
 248
 249 struct diffstat_t {
 250         struct xdiff_emit_state xm;
 251
 252         int nr;
 253         int alloc;
 254         struct diffstat_file {
 255                 char *name;
 256                 unsigned is_unmerged:1;
 257                 unsigned is_binary:1;
 258                 unsigned is_renamed:1;
 259                 unsigned int added, deleted;
 260         } **files;
 261 };
 262
 263 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 264                                           const char *name_a,
 265                                           const char *name_b)
 266 {
 267         struct diffstat_file *x;
 268         x = xcalloc(sizeof (*x), 1);
 269         if (diffstat->nr == diffstat->alloc) {
 270                 diffstat->alloc = alloc_nr(diffstat->alloc);
 271                 diffstat->files = xrealloc(diffstat->files,
 272                                 diffstat->alloc * sizeof(x));
 273         }
 274         diffstat->files[diffstat->nr++] = x;
 275         if (name_b) {
 276                 x->name = pprint_rename(name_a, name_b);
 277                 x->is_renamed = 1;
 278         }
 279         else
 280                 x->name = strdup(name_a);
 281         return x;
 282 }
 283
 284 static void diffstat_consume(void *priv, char *line, unsigned long len)
 285 {
 286         struct diffstat_t *diffstat = priv;
 287         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 288
 289         if (line[0] == '+')
 290                 x->added++;
 291         else if (line[0] == '-')
 292                 x->deleted++;
 293 }
 294
 295 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
 296 static const char minuses[]= "----------------------------------------------------------------------";
 297
 298 static void show_stats(struct diffstat_t* data)
 299 {
 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 *prefix = "";
 322                 char *name = data->files[i]->name;
 323                 int added = data->files[i]->added;
 324                 int deleted = data->files[i]->deleted;
 325
 326                 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
 327                         char *qname = xmalloc(len + 1);
 328                         quote_c_style(name, qname, NULL, 0);
 329                         free(name);
 330                         data->files[i]->name = name = qname;
 331                 }
 332
 333                 /*
 334                  * "scale" the filename
 335                  */
 336                 len = strlen(name);
 337                 max = max_len;
 338                 if (max > 50)
 339                         max = 50;
 340                 if (len > max) {
 341                         char *slash;
 342                         prefix = "...";
 343                         max -= 3;
 344                         name += len - max;
 345                         slash = strchr(name, '/');
 346                         if (slash)
 347                                 name = slash;
 348                 }
 349                 len = max;
 350
 351                 /*
 352                  * scale the add/delete
 353                  */
 354                 max = max_change;
 355                 if (max + len > 70)
 356                         max = 70 - len;
 357
 358                 if (data->files[i]->is_binary) {
 359                         printf(" %s%-*s |  Bin\n", prefix, len, name);
 360                         goto free_diffstat_file;
 361                 }
 362                 else if (data->files[i]->is_unmerged) {
 363                         printf(" %s%-*s |  Unmerged\n", prefix, len, name);
 364                         goto free_diffstat_file;
 365                 }
 366                 else if (!data->files[i]->is_renamed &&
 367                          (added + deleted == 0)) {
 368                         total_files--;
 369                         goto free_diffstat_file;
 370                 }
 371
 372                 add = added;
 373                 del = deleted;
 374                 total = add + del;
 375                 adds += add;
 376                 dels += del;
 377
 378                 if (max_change > 0) {
 379                         total = (total * max + max_change / 2) / max_change;
 380                         add = (add * max + max_change / 2) / max_change;
 381                         del = total - add;
 382                 }
 383                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
 384                                 len, name, added + deleted,
 385                                 add, pluses, del, minuses);
 386         free_diffstat_file:
 387                 free(data->files[i]->name);
 388                 free(data->files[i]);
 389         }
 390         free(data->files);
 391         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 392                         total_files, adds, dels);
 393 }
 394
 395 static unsigned char *deflate_it(char *data,
 396                                  unsigned long size,
 397                                  unsigned long *result_size)
 398 {
 399         int bound;
 400         unsigned char *deflated;
 401         z_stream stream;
 402
 403         memset(&stream, 0, sizeof(stream));
 404         deflateInit(&stream, Z_BEST_COMPRESSION);
 405         bound = deflateBound(&stream, size);
 406         deflated = xmalloc(bound);
 407         stream.next_out = deflated;
 408         stream.avail_out = bound;
 409
 410         stream.next_in = (unsigned char *)data;
 411         stream.avail_in = size;
 412         while (deflate(&stream, Z_FINISH) == Z_OK)
 413                 ; /* nothing */
 414         deflateEnd(&stream);
 415         *result_size = stream.total_out;
 416         return deflated;
 417 }
 418
 419 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
 420 {
 421         void *cp;
 422         void *delta;
 423         void *deflated;
 424         void *data;
 425         unsigned long orig_size;
 426         unsigned long delta_size;
 427         unsigned long deflate_size;
 428         unsigned long data_size;
 429
 430         printf("GIT binary patch\n");
 431         /* We could do deflated delta, or we could do just deflated two,
 432          * whichever is smaller.
 433          */
 434         delta = NULL;
 435         deflated = deflate_it(two->ptr, two->size, &deflate_size);
 436         if (one->size && two->size) {
 437                 delta = diff_delta(one->ptr, one->size,
 438                                    two->ptr, two->size,
 439                                    &delta_size, deflate_size);
 440                 if (delta) {
 441                         void *to_free = delta;
 442                         orig_size = delta_size;
 443                         delta = deflate_it(delta, delta_size, &delta_size);
 444                         free(to_free);
 445                 }
 446         }
 447
 448         if (delta && delta_size < deflate_size) {
 449                 printf("delta %lu\n", orig_size);
 450                 free(deflated);
 451                 data = delta;
 452                 data_size = delta_size;
 453         }
 454         else {
 455                 printf("literal %lu\n", two->size);
 456                 free(delta);
 457                 data = deflated;
 458                 data_size = deflate_size;
 459         }
 460
 461         /* emit data encoded in base85 */
 462         cp = data;
 463         while (data_size) {
 464                 int bytes = (52 < data_size) ? 52 : data_size;
 465                 char line[70];
 466                 data_size -= bytes;
 467                 if (bytes <= 26)
 468                         line[0] = bytes + 'A' - 1;
 469                 else
 470                         line[0] = bytes - 26 + 'a' - 1;
 471                 encode_85(line + 1, cp, bytes);
 472                 cp += bytes;
 473                 puts(line);
 474         }
 475         printf("\n");
 476         free(data);
 477 }
 478
 479 #define FIRST_FEW_BYTES 8000
 480 static int mmfile_is_binary(mmfile_t *mf)
 481 {
 482         long sz = mf->size;
 483         if (FIRST_FEW_BYTES < sz)
 484                 sz = FIRST_FEW_BYTES;
 485         if (memchr(mf->ptr, 0, sz))
 486                 return 1;
 487         return 0;
 488 }
 489
 490 static void builtin_diff(const char *name_a,
 491                          const char *name_b,
 492                          struct diff_filespec *one,
 493                          struct diff_filespec *two,
 494                          const char *xfrm_msg,
 495                          struct diff_options *o,
 496                          int complete_rewrite)
 497 {
 498         mmfile_t mf1, mf2;
 499         const char *lbl[2];
 500         char *a_one, *b_two;
 501
 502         a_one = quote_two("a/", name_a);
 503         b_two = quote_two("b/", name_b);
 504         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
 505         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
 506         printf("diff --git %s %s\n", a_one, b_two);
 507         if (lbl[0][0] == '/') {
 508                 /* /dev/null */
 509                 printf("new file mode %06o\n", two->mode);
 510                 if (xfrm_msg && xfrm_msg[0])
 511                         puts(xfrm_msg);
 512         }
 513         else if (lbl[1][0] == '/') {
 514                 printf("deleted file mode %06o\n", one->mode);
 515                 if (xfrm_msg && xfrm_msg[0])
 516                         puts(xfrm_msg);
 517         }
 518         else {
 519                 if (one->mode != two->mode) {
 520                         printf("old mode %06o\n", one->mode);
 521                         printf("new mode %06o\n", two->mode);
 522                 }
 523                 if (xfrm_msg && xfrm_msg[0])
 524                         puts(xfrm_msg);
 525                 /*
 526                  * we do not run diff between different kind
 527                  * of objects.
 528                  */
 529                 if ((one->mode ^ two->mode) & S_IFMT)
 530                         goto free_ab_and_return;
 531                 if (complete_rewrite) {
 532                         emit_rewrite_diff(name_a, name_b, one, two);
 533                         goto free_ab_and_return;
 534                 }
 535         }
 536
 537         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 538                 die("unable to read files to diff");
 539
 540         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
 541                 /* Quite common confusing case */
 542                 if (mf1.size == mf2.size &&
 543                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
 544                         goto free_ab_and_return;
 545                 if (o->binary)
 546                         emit_binary_diff(&mf1, &mf2);
 547                 else
 548                         printf("Binary files %s and %s differ\n",
 549                                lbl[0], lbl[1]);
 550         }
 551         else {
 552                 /* Crazy xdl interfaces.. */
 553                 const char *diffopts = getenv("GIT_DIFF_OPTS");
 554                 xpparam_t xpp;
 555                 xdemitconf_t xecfg;
 556                 xdemitcb_t ecb;
 557                 struct emit_callback ecbdata;
 558
 559                 ecbdata.label_path = lbl;
 560                 xpp.flags = XDF_NEED_MINIMAL;
 561                 xecfg.ctxlen = 3;
 562                 xecfg.flags = XDL_EMIT_FUNCNAMES;
 563                 if (!diffopts)
 564                         ;
 565                 else if (!strncmp(diffopts, "--unified=", 10))
 566                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
 567                 else if (!strncmp(diffopts, "-u", 2))
 568                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
 569                 ecb.outf = fn_out;
 570                 ecb.priv = &ecbdata;
 571                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 572         }
 573
 574  free_ab_and_return:
 575         free(a_one);
 576         free(b_two);
 577         return;
 578 }
 579
 580 static void builtin_diffstat(const char *name_a, const char *name_b,
 581                              struct diff_filespec *one,
 582                              struct diff_filespec *two,
 583                              struct diffstat_t *diffstat,
 584                              int complete_rewrite)
 585 {
 586         mmfile_t mf1, mf2;
 587         struct diffstat_file *data;
 588
 589         data = diffstat_add(diffstat, name_a, name_b);
 590
 591         if (!one || !two) {
 592                 data->is_unmerged = 1;
 593                 return;
 594         }
 595         if (complete_rewrite) {
 596                 diff_populate_filespec(one, 0);
 597                 diff_populate_filespec(two, 0);
 598                 data->deleted = count_lines(one->data, one->size);
 599                 data->added = count_lines(two->data, two->size);
 600                 return;
 601         }
 602         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
 603                 die("unable to read files to diff");
 604
 605         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
 606                 data->is_binary = 1;
 607         else {
 608                 /* Crazy xdl interfaces.. */
 609                 xpparam_t xpp;
 610                 xdemitconf_t xecfg;
 611                 xdemitcb_t ecb;
 612
 613                 xpp.flags = XDF_NEED_MINIMAL;
 614                 xecfg.ctxlen = 0;
 615                 xecfg.flags = 0;
 616                 ecb.outf = xdiff_outf;
 617                 ecb.priv = diffstat;
 618                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
 619         }
 620 }
 621
 622 struct diff_filespec *alloc_filespec(const char *path)
 623 {
 624         int namelen = strlen(path);
 625         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 626
 627         memset(spec, 0, sizeof(*spec));
 628         spec->path = (char *)(spec + 1);
 629         memcpy(spec->path, path, namelen+1);
 630         return spec;
 631 }
 632
 633 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 634                    unsigned short mode)
 635 {
 636         if (mode) {
 637                 spec->mode = canon_mode(mode);
 638                 memcpy(spec->sha1, sha1, 20);
 639                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 640         }
 641 }
 642
 643 /*
 644  * Given a name and sha1 pair, if the dircache tells us the file in
 645  * the work tree has that object contents, return true, so that
 646  * prepare_temp_file() does not have to inflate and extract.
 647  */
 648 static int work_tree_matches(const char *name, const unsigned char *sha1)
 649 {
 650         struct cache_entry *ce;
 651         struct stat st;
 652         int pos, len;
 653
 654         /* We do not read the cache ourselves here, because the
 655          * benchmark with my previous version that always reads cache
 656          * shows that it makes things worse for diff-tree comparing
 657          * two linux-2.6 kernel trees in an already checked out work
 658          * tree.  This is because most diff-tree comparisons deal with
 659          * only a small number of files, while reading the cache is
 660          * expensive for a large project, and its cost outweighs the
 661          * savings we get by not inflating the object to a temporary
 662          * file.  Practically, this code only helps when we are used
 663          * by diff-cache --cached, which does read the cache before
 664          * calling us.
 665          */
 666         if (!active_cache)
 667                 return 0;
 668
 669         len = strlen(name);
 670         pos = cache_name_pos(name, len);
 671         if (pos < 0)
 672                 return 0;
 673         ce = active_cache[pos];
 674         if ((lstat(name, &st) < 0) ||
 675             !S_ISREG(st.st_mode) || /* careful! */
 676             ce_match_stat(ce, &st, 0) ||
 677             memcmp(sha1, ce->sha1, 20))
 678                 return 0;
 679         /* we return 1 only when we can stat, it is a regular file,
 680          * stat information matches, and sha1 recorded in the cache
 681          * matches.  I.e. we know the file in the work tree really is
 682          * the same as the <name, sha1> pair.
 683          */
 684         return 1;
 685 }
 686
 687 static struct sha1_size_cache {
 688         unsigned char sha1[20];
 689         unsigned long size;
 690 } **sha1_size_cache;
 691 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 692
 693 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 694                                                  int find_only,
 695                                                  unsigned long size)
 696 {
 697         int first, last;
 698         struct sha1_size_cache *e;
 699
 700         first = 0;
 701         last = sha1_size_cache_nr;
 702         while (last > first) {
 703                 int cmp, next = (last + first) >> 1;
 704                 e = sha1_size_cache[next];
 705                 cmp = memcmp(e->sha1, sha1, 20);
 706                 if (!cmp)
 707                         return e;
 708                 if (cmp < 0) {
 709                         last = next;
 710                         continue;
 711                 }
 712                 first = next+1;
 713         }
 714         /* not found */
 715         if (find_only)
 716                 return NULL;
 717         /* insert to make it at "first" */
 718         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 719                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 720                 sha1_size_cache = xrealloc(sha1_size_cache,
 721                                            sha1_size_cache_alloc *
 722                                            sizeof(*sha1_size_cache));
 723         }
 724         sha1_size_cache_nr++;
 725         if (first < sha1_size_cache_nr)
 726                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 727                         (sha1_size_cache_nr - first - 1) *
 728                         sizeof(*sha1_size_cache));
 729         e = xmalloc(sizeof(struct sha1_size_cache));
 730         sha1_size_cache[first] = e;
 731         memcpy(e->sha1, sha1, 20);
 732         e->size = size;
 733         return e;
 734 }
 735
 736 /*
 737  * While doing rename detection and pickaxe operation, we may need to
 738  * grab the data for the blob (or file) for our own in-core comparison.
 739  * diff_filespec has data and size fields for this purpose.
 740  */
 741 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 742 {
 743         int err = 0;
 744         if (!DIFF_FILE_VALID(s))
 745                 die("internal error: asking to populate invalid file.");
 746         if (S_ISDIR(s->mode))
 747                 return -1;
 748
 749         if (!use_size_cache)
 750                 size_only = 0;
 751
 752         if (s->data)
 753                 return err;
 754         if (!s->sha1_valid ||
 755             work_tree_matches(s->path, s->sha1)) {
 756                 struct stat st;
 757                 int fd;
 758                 if (lstat(s->path, &st) < 0) {
 759                         if (errno == ENOENT) {
 760                         err_empty:
 761                                 err = -1;
 762                         empty:
 763                                 s->data = "";
 764                                 s->size = 0;
 765                                 return err;
 766                         }
 767                 }
 768                 s->size = st.st_size;
 769                 if (!s->size)
 770                         goto empty;
 771                 if (size_only)
 772                         return 0;
 773                 if (S_ISLNK(st.st_mode)) {
 774                         int ret;
 775                         s->data = xmalloc(s->size);
 776                         s->should_free = 1;
 777                         ret = readlink(s->path, s->data, s->size);
 778                         if (ret < 0) {
 779                                 free(s->data);
 780                                 goto err_empty;
 781                         }
 782                         return 0;
 783                 }
 784                 fd = open(s->path, O_RDONLY);
 785                 if (fd < 0)
 786                         goto err_empty;
 787                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 788                 close(fd);
 789                 if (s->data == MAP_FAILED)
 790                         goto err_empty;
 791                 s->should_munmap = 1;
 792         }
 793         else {
 794                 char type[20];
 795                 struct sha1_size_cache *e;
 796
 797                 if (size_only) {
 798                         e = locate_size_cache(s->sha1, 1, 0);
 799                         if (e) {
 800                                 s->size = e->size;
 801                                 return 0;
 802                         }
 803                         if (!sha1_object_info(s->sha1, type, &s->size))
 804                                 locate_size_cache(s->sha1, 0, s->size);
 805                 }
 806                 else {
 807                         s->data = read_sha1_file(s->sha1, type, &s->size);
 808                         s->should_free = 1;
 809                 }
 810         }
 811         return 0;
 812 }
 813
 814 void diff_free_filespec_data(struct diff_filespec *s)
 815 {
 816         if (s->should_free)
 817                 free(s->data);
 818         else if (s->should_munmap)
 819                 munmap(s->data, s->size);
 820         s->should_free = s->should_munmap = 0;
 821         s->data = NULL;
 822         free(s->cnt_data);
 823         s->cnt_data = NULL;
 824 }
 825
 826 static void prep_temp_blob(struct diff_tempfile *temp,
 827                            void *blob,
 828                            unsigned long size,
 829                            const unsigned char *sha1,
 830                            int mode)
 831 {
 832         int fd;
 833
 834         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 835         if (fd < 0)
 836                 die("unable to create temp-file");
 837         if (write(fd, blob, size) != size)
 838                 die("unable to write temp-file");
 839         close(fd);
 840         temp->name = temp->tmp_path;
 841         strcpy(temp->hex, sha1_to_hex(sha1));
 842         temp->hex[40] = 0;
 843         sprintf(temp->mode, "%06o", mode);
 844 }
 845
 846 static void prepare_temp_file(const char *name,
 847                               struct diff_tempfile *temp,
 848                               struct diff_filespec *one)
 849 {
 850         if (!DIFF_FILE_VALID(one)) {
 851         not_a_valid_file:
 852                 /* A '-' entry produces this for file-2, and
 853                  * a '+' entry produces this for file-1.
 854                  */
 855                 temp->name = "/dev/null";
 856                 strcpy(temp->hex, ".");
 857                 strcpy(temp->mode, ".");
 858                 return;
 859         }
 860
 861         if (!one->sha1_valid ||
 862             work_tree_matches(name, one->sha1)) {
 863                 struct stat st;
 864                 if (lstat(name, &st) < 0) {
 865                         if (errno == ENOENT)
 866                                 goto not_a_valid_file;
 867                         die("stat(%s): %s", name, strerror(errno));
 868                 }
 869                 if (S_ISLNK(st.st_mode)) {
 870                         int ret;
 871                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
 872                         if (sizeof(buf) <= st.st_size)
 873                                 die("symlink too long: %s", name);
 874                         ret = readlink(name, buf, st.st_size);
 875                         if (ret < 0)
 876                                 die("readlink(%s)", name);
 877                         prep_temp_blob(temp, buf, st.st_size,
 878                                        (one->sha1_valid ?
 879                                         one->sha1 : null_sha1),
 880                                        (one->sha1_valid ?
 881                                         one->mode : S_IFLNK));
 882                 }
 883                 else {
 884                         /* we can borrow from the file in the work tree */
 885                         temp->name = name;
 886                         if (!one->sha1_valid)
 887                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 888                         else
 889                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 890                         /* Even though we may sometimes borrow the
 891                          * contents from the work tree, we always want
 892                          * one->mode.  mode is trustworthy even when
 893                          * !(one->sha1_valid), as long as
 894                          * DIFF_FILE_VALID(one).
 895                          */
 896                         sprintf(temp->mode, "%06o", one->mode);
 897                 }
 898                 return;
 899         }
 900         else {
 901                 if (diff_populate_filespec(one, 0))
 902                         die("cannot read data blob for %s", one->path);
 903                 prep_temp_blob(temp, one->data, one->size,
 904                                one->sha1, one->mode);
 905         }
 906 }
 907
 908 static void remove_tempfile(void)
 909 {
 910         int i;
 911
 912         for (i = 0; i < 2; i++)
 913                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 914                         unlink(diff_temp[i].name);
 915                         diff_temp[i].name = NULL;
 916                 }
 917 }
 918
 919 static void remove_tempfile_on_signal(int signo)
 920 {
 921         remove_tempfile();
 922         signal(SIGINT, SIG_DFL);
 923         raise(signo);
 924 }
 925
 926 static int spawn_prog(const char *pgm, const char **arg)
 927 {
 928         pid_t pid;
 929         int status;
 930
 931         fflush(NULL);
 932         pid = fork();
 933         if (pid < 0)
 934                 die("unable to fork");
 935         if (!pid) {
 936                 execvp(pgm, (char *const*) arg);
 937                 exit(255);
 938         }
 939
 940         while (waitpid(pid, &status, 0) < 0) {
 941                 if (errno == EINTR)
 942                         continue;
 943                 return -1;
 944         }
 945
 946         /* Earlier we did not check the exit status because
 947          * diff exits non-zero if files are different, and
 948          * we are not interested in knowing that.  It was a
 949          * mistake which made it harder to quit a diff-*
 950          * session that uses the git-apply-patch-script as
 951          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 952          * should also exit non-zero only when it wants to
 953          * abort the entire diff-* session.
 954          */
 955         if (WIFEXITED(status) && !WEXITSTATUS(status))
 956                 return 0;
 957         return -1;
 958 }
 959
 960 /* An external diff command takes:
 961  *
 962  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 963  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 964  *
 965  */
 966 static void run_external_diff(const char *pgm,
 967                               const char *name,
 968                               const char *other,
 969                               struct diff_filespec *one,
 970                               struct diff_filespec *two,
 971                               const char *xfrm_msg,
 972                               int complete_rewrite)
 973 {
 974         const char *spawn_arg[10];
 975         struct diff_tempfile *temp = diff_temp;
 976         int retval;
 977         static int atexit_asked = 0;
 978         const char *othername;
 979         const char **arg = &spawn_arg[0];
 980
 981         othername = (other? other : name);
 982         if (one && two) {
 983                 prepare_temp_file(name, &temp[0], one);
 984                 prepare_temp_file(othername, &temp[1], two);
 985                 if (! atexit_asked &&
 986                     (temp[0].name == temp[0].tmp_path ||
 987                      temp[1].name == temp[1].tmp_path)) {
 988                         atexit_asked = 1;
 989                         atexit(remove_tempfile);
 990                 }
 991                 signal(SIGINT, remove_tempfile_on_signal);
 992         }
 993
 994         if (one && two) {
 995                 *arg++ = pgm;
 996                 *arg++ = name;
 997                 *arg++ = temp[0].name;
 998                 *arg++ = temp[0].hex;
 999                 *arg++ = temp[0].mode;
1000                 *arg++ = temp[1].name;
1001                 *arg++ = temp[1].hex;
1002                 *arg++ = temp[1].mode;
1003                 if (other) {
1004                         *arg++ = other;
1005                         *arg++ = xfrm_msg;
1006                 }
1007         } else {
1008                 *arg++ = pgm;
1009                 *arg++ = name;
1010         }
1011         *arg = NULL;
1012         retval = spawn_prog(pgm, spawn_arg);
1013         remove_tempfile();
1014         if (retval) {
1015                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1016                 exit(1);
1017         }
1018 }
1019
1020 static void run_diff_cmd(const char *pgm,
1021                          const char *name,
1022                          const char *other,
1023                          struct diff_filespec *one,
1024                          struct diff_filespec *two,
1025                          const char *xfrm_msg,
1026                          struct diff_options *o,
1027                          int complete_rewrite)
1028 {
1029         if (pgm) {
1030                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1031                                   complete_rewrite);
1032                 return;
1033         }
1034         if (one && two)
1035                 builtin_diff(name, other ? other : name,
1036                              one, two, xfrm_msg, o, complete_rewrite);
1037         else
1038                 printf("* Unmerged path %s\n", name);
1039 }
1040
1041 static void diff_fill_sha1_info(struct diff_filespec *one)
1042 {
1043         if (DIFF_FILE_VALID(one)) {
1044                 if (!one->sha1_valid) {
1045                         struct stat st;
1046                         if (lstat(one->path, &st) < 0)
1047                                 die("stat %s", one->path);
1048                         if (index_path(one->sha1, one->path, &st, 0))
1049                                 die("cannot hash %s\n", one->path);
1050                 }
1051         }
1052         else
1053                 memset(one->sha1, 0, 20);
1054 }
1055
1056 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1057 {
1058         const char *pgm = external_diff();
1059         char msg[PATH_MAX*2+300], *xfrm_msg;
1060         struct diff_filespec *one;
1061         struct diff_filespec *two;
1062         const char *name;
1063         const char *other;
1064         char *name_munged, *other_munged;
1065         int complete_rewrite = 0;
1066         int len;
1067
1068         if (DIFF_PAIR_UNMERGED(p)) {
1069                 /* unmerged */
1070                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1071                 return;
1072         }
1073
1074         name = p->one->path;
1075         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1076         name_munged = quote_one(name);
1077         other_munged = quote_one(other);
1078         one = p->one; two = p->two;
1079
1080         diff_fill_sha1_info(one);
1081         diff_fill_sha1_info(two);
1082
1083         len = 0;
1084         switch (p->status) {
1085         case DIFF_STATUS_COPIED:
1086                 len += snprintf(msg + len, sizeof(msg) - len,
1087                                 "similarity index %d%%\n"
1088                                 "copy from %s\n"
1089                                 "copy to %s\n",
1090                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1091                                 name_munged, other_munged);
1092                 break;
1093         case DIFF_STATUS_RENAMED:
1094                 len += snprintf(msg + len, sizeof(msg) - len,
1095                                 "similarity index %d%%\n"
1096                                 "rename from %s\n"
1097                                 "rename to %s\n",
1098                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1099                                 name_munged, other_munged);
1100                 break;
1101         case DIFF_STATUS_MODIFIED:
1102                 if (p->score) {
1103                         len += snprintf(msg + len, sizeof(msg) - len,
1104                                         "dissimilarity index %d%%\n",
1105                                         (int)(0.5 + p->score *
1106                                               100.0/MAX_SCORE));
1107                         complete_rewrite = 1;
1108                         break;
1109                 }
1110                 /* fallthru */
1111         default:
1112                 /* nothing */
1113                 ;
1114         }
1115
1116         if (memcmp(one->sha1, two->sha1, 20)) {
1117                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1118
1119                 len += snprintf(msg + len, sizeof(msg) - len,
1120                                 "index %.*s..%.*s",
1121                                 abbrev, sha1_to_hex(one->sha1),
1122                                 abbrev, sha1_to_hex(two->sha1));
1123                 if (one->mode == two->mode)
1124                         len += snprintf(msg + len, sizeof(msg) - len,
1125                                         " %06o", one->mode);
1126                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1127         }
1128
1129         if (len)
1130                 msg[--len] = 0;
1131         xfrm_msg = len ? msg : NULL;
1132
1133         if (!pgm &&
1134             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1135             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1136                 /* a filepair that changes between file and symlink
1137                  * needs to be split into deletion and creation.
1138                  */
1139                 struct diff_filespec *null = alloc_filespec(two->path);
1140                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1141                 free(null);
1142                 null = alloc_filespec(one->path);
1143                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1144                 free(null);
1145         }
1146         else
1147                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1148                              complete_rewrite);
1149
1150         free(name_munged);
1151         free(other_munged);
1152 }
1153
1154 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1155                          struct diffstat_t *diffstat)
1156 {
1157         const char *name;
1158         const char *other;
1159         int complete_rewrite = 0;
1160
1161         if (DIFF_PAIR_UNMERGED(p)) {
1162                 /* unmerged */
1163                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, 0);
1164                 return;
1165         }
1166
1167         name = p->one->path;
1168         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1169
1170         diff_fill_sha1_info(p->one);
1171         diff_fill_sha1_info(p->two);
1172
1173         if (p->status == DIFF_STATUS_MODIFIED && p->score)
1174                 complete_rewrite = 1;
1175         builtin_diffstat(name, other, p->one, p->two, diffstat, complete_rewrite);
1176 }
1177
1178 void diff_setup(struct diff_options *options)
1179 {
1180         memset(options, 0, sizeof(*options));
1181         options->output_format = DIFF_FORMAT_RAW;
1182         options->line_termination = '\n';
1183         options->break_opt = -1;
1184         options->rename_limit = -1;
1185
1186         options->change = diff_change;
1187         options->add_remove = diff_addremove;
1188 }
1189
1190 int diff_setup_done(struct diff_options *options)
1191 {
1192         if ((options->find_copies_harder &&
1193              options->detect_rename != DIFF_DETECT_COPY) ||
1194             (0 <= options->rename_limit && !options->detect_rename))
1195                 return -1;
1196
1197         /*
1198          * These cases always need recursive; we do not drop caller-supplied
1199          * recursive bits for other formats here.
1200          */
1201         if ((options->output_format == DIFF_FORMAT_PATCH) ||
1202             (options->output_format == DIFF_FORMAT_DIFFSTAT))
1203                 options->recursive = 1;
1204
1205         if (options->detect_rename && options->rename_limit < 0)
1206                 options->rename_limit = diff_rename_limit_default;
1207         if (options->setup & DIFF_SETUP_USE_CACHE) {
1208                 if (!active_cache)
1209                         /* read-cache does not die even when it fails
1210                          * so it is safe for us to do this here.  Also
1211                          * it does not smudge active_cache or active_nr
1212                          * when it fails, so we do not have to worry about
1213                          * cleaning it up ourselves either.
1214                          */
1215                         read_cache();
1216         }
1217         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1218                 use_size_cache = 1;
1219         if (options->abbrev <= 0 || 40 < options->abbrev)
1220                 options->abbrev = 40; /* full */
1221
1222         return 0;
1223 }
1224
1225 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1226 {
1227         const char *arg = av[0];
1228         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1229                 options->output_format = DIFF_FORMAT_PATCH;
1230         else if (!strcmp(arg, "--patch-with-raw")) {
1231                 options->output_format = DIFF_FORMAT_PATCH;
1232                 options->with_raw = 1;
1233         }
1234         else if (!strcmp(arg, "--stat"))
1235                 options->output_format = DIFF_FORMAT_DIFFSTAT;
1236         else if (!strcmp(arg, "--patch-with-stat")) {
1237                 options->output_format = DIFF_FORMAT_PATCH;
1238                 options->with_stat = 1;
1239         }
1240         else if (!strcmp(arg, "-z"))
1241                 options->line_termination = 0;
1242         else if (!strncmp(arg, "-l", 2))
1243                 options->rename_limit = strtoul(arg+2, NULL, 10);
1244         else if (!strcmp(arg, "--full-index"))
1245                 options->full_index = 1;
1246         else if (!strcmp(arg, "--binary")) {
1247                 options->output_format = DIFF_FORMAT_PATCH;
1248                 options->full_index = options->binary = 1;
1249         }
1250         else if (!strcmp(arg, "--name-only"))
1251                 options->output_format = DIFF_FORMAT_NAME;
1252         else if (!strcmp(arg, "--name-status"))
1253                 options->output_format = DIFF_FORMAT_NAME_STATUS;
1254         else if (!strcmp(arg, "-R"))
1255                 options->reverse_diff = 1;
1256         else if (!strncmp(arg, "-S", 2))
1257                 options->pickaxe = arg + 2;
1258         else if (!strcmp(arg, "-s"))
1259                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1260         else if (!strncmp(arg, "-O", 2))
1261                 options->orderfile = arg + 2;
1262         else if (!strncmp(arg, "--diff-filter=", 14))
1263                 options->filter = arg + 14;
1264         else if (!strcmp(arg, "--pickaxe-all"))
1265                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1266         else if (!strcmp(arg, "--pickaxe-regex"))
1267                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1268         else if (!strncmp(arg, "-B", 2)) {
1269                 if ((options->break_opt =
1270                      diff_scoreopt_parse(arg)) == -1)
1271                         return -1;
1272         }
1273         else if (!strncmp(arg, "-M", 2)) {
1274                 if ((options->rename_score =
1275                      diff_scoreopt_parse(arg)) == -1)
1276                         return -1;
1277                 options->detect_rename = DIFF_DETECT_RENAME;
1278         }
1279         else if (!strncmp(arg, "-C", 2)) {
1280                 if ((options->rename_score =
1281                      diff_scoreopt_parse(arg)) == -1)
1282                         return -1;
1283                 options->detect_rename = DIFF_DETECT_COPY;
1284         }
1285         else if (!strcmp(arg, "--find-copies-harder"))
1286                 options->find_copies_harder = 1;
1287         else if (!strcmp(arg, "--abbrev"))
1288                 options->abbrev = DEFAULT_ABBREV;
1289         else if (!strncmp(arg, "--abbrev=", 9)) {
1290                 options->abbrev = strtoul(arg + 9, NULL, 10);
1291                 if (options->abbrev < MINIMUM_ABBREV)
1292                         options->abbrev = MINIMUM_ABBREV;
1293                 else if (40 < options->abbrev)
1294                         options->abbrev = 40;
1295         }
1296         else
1297                 return 0;
1298         return 1;
1299 }
1300
1301 static int parse_num(const char **cp_p)
1302 {
1303         unsigned long num, scale;
1304         int ch, dot;
1305         const char *cp = *cp_p;
1306
1307         num = 0;
1308         scale = 1;
1309         dot = 0;
1310         for(;;) {
1311                 ch = *cp;
1312                 if ( !dot && ch == '.' ) {
1313                         scale = 1;
1314                         dot = 1;
1315                 } else if ( ch == '%' ) {
1316                         scale = dot ? scale*100 : 100;
1317                         cp++;   /* % is always at the end */
1318                         break;
1319                 } else if ( ch >= '0' && ch <= '9' ) {
1320                         if ( scale < 100000 ) {
1321                                 scale *= 10;
1322                                 num = (num*10) + (ch-'0');
1323                         }
1324                 } else {
1325                         break;
1326                 }
1327                 cp++;
1328         }
1329         *cp_p = cp;
1330
1331         /* user says num divided by scale and we say internally that
1332          * is MAX_SCORE * num / scale.
1333          */
1334         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1335 }
1336
1337 int diff_scoreopt_parse(const char *opt)
1338 {
1339         int opt1, opt2, cmd;
1340
1341         if (*opt++ != '-')
1342                 return -1;
1343         cmd = *opt++;
1344         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1345                 return -1; /* that is not a -M, -C nor -B option */
1346
1347         opt1 = parse_num(&opt);
1348         if (cmd != 'B')
1349                 opt2 = 0;
1350         else {
1351                 if (*opt == 0)
1352                         opt2 = 0;
1353                 else if (*opt != '/')
1354                         return -1; /* we expect -B80/99 or -B80 */
1355                 else {
1356                         opt++;
1357                         opt2 = parse_num(&opt);
1358                 }
1359         }
1360         if (*opt != 0)
1361                 return -1;
1362         return opt1 | (opt2 << 16);
1363 }
1364
1365 struct diff_queue_struct diff_queued_diff;
1366
1367 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1368 {
1369         if (queue->alloc <= queue->nr) {
1370                 queue->alloc = alloc_nr(queue->alloc);
1371                 queue->queue = xrealloc(queue->queue,
1372                                         sizeof(dp) * queue->alloc);
1373         }
1374         queue->queue[queue->nr++] = dp;
1375 }
1376
1377 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1378                                  struct diff_filespec *one,
1379                                  struct diff_filespec *two)
1380 {
1381         struct diff_filepair *dp = xmalloc(sizeof(*dp));
1382         dp->one = one;
1383         dp->two = two;
1384         dp->score = 0;
1385         dp->status = 0;
1386         dp->source_stays = 0;
1387         dp->broken_pair = 0;
1388         if (queue)
1389                 diff_q(queue, dp);
1390         return dp;
1391 }
1392
1393 void diff_free_filepair(struct diff_filepair *p)
1394 {
1395         diff_free_filespec_data(p->one);
1396         diff_free_filespec_data(p->two);
1397         free(p->one);
1398         free(p->two);
1399         free(p);
1400 }
1401
1402 /* This is different from find_unique_abbrev() in that
1403  * it stuffs the result with dots for alignment.
1404  */
1405 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1406 {
1407         int abblen;
1408         const char *abbrev;
1409         if (len == 40)
1410                 return sha1_to_hex(sha1);
1411
1412         abbrev = find_unique_abbrev(sha1, len);
1413         if (!abbrev)
1414                 return sha1_to_hex(sha1);
1415         abblen = strlen(abbrev);
1416         if (abblen < 37) {
1417                 static char hex[41];
1418                 if (len < abblen && abblen <= len + 2)
1419                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1420                 else
1421                         sprintf(hex, "%s...", abbrev);
1422                 return hex;
1423         }
1424         return sha1_to_hex(sha1);
1425 }
1426
1427 static void diff_flush_raw(struct diff_filepair *p,
1428                            int line_termination,
1429                            int inter_name_termination,
1430                            struct diff_options *options,
1431                            int output_format)
1432 {
1433         int two_paths;
1434         char status[10];
1435         int abbrev = options->abbrev;
1436         const char *path_one, *path_two;
1437
1438         path_one = p->one->path;
1439         path_two = p->two->path;
1440         if (line_termination) {
1441                 path_one = quote_one(path_one);
1442                 path_two = quote_one(path_two);
1443         }
1444
1445         if (p->score)
1446                 sprintf(status, "%c%03d", p->status,
1447                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1448         else {
1449                 status[0] = p->status;
1450                 status[1] = 0;
1451         }
1452         switch (p->status) {
1453         case DIFF_STATUS_COPIED:
1454         case DIFF_STATUS_RENAMED:
1455                 two_paths = 1;
1456                 break;
1457         case DIFF_STATUS_ADDED:
1458         case DIFF_STATUS_DELETED:
1459                 two_paths = 0;
1460                 break;
1461         default:
1462                 two_paths = 0;
1463                 break;
1464         }
1465         if (output_format != DIFF_FORMAT_NAME_STATUS) {
1466                 printf(":%06o %06o %s ",
1467                        p->one->mode, p->two->mode,
1468                        diff_unique_abbrev(p->one->sha1, abbrev));
1469                 printf("%s ",
1470                        diff_unique_abbrev(p->two->sha1, abbrev));
1471         }
1472         printf("%s%c%s", status, inter_name_termination, path_one);
1473         if (two_paths)
1474                 printf("%c%s", inter_name_termination, path_two);
1475         putchar(line_termination);
1476         if (path_one != p->one->path)
1477                 free((void*)path_one);
1478         if (path_two != p->two->path)
1479                 free((void*)path_two);
1480 }
1481
1482 static void diff_flush_name(struct diff_filepair *p,
1483                             int inter_name_termination,
1484                             int line_termination)
1485 {
1486         char *path = p->two->path;
1487
1488         if (line_termination)
1489                 path = quote_one(p->two->path);
1490         else
1491                 path = p->two->path;
1492         printf("%s%c", path, line_termination);
1493         if (p->two->path != path)
1494                 free(path);
1495 }
1496
1497 int diff_unmodified_pair(struct diff_filepair *p)
1498 {
1499         /* This function is written stricter than necessary to support
1500          * the currently implemented transformers, but the idea is to
1501          * let transformers to produce diff_filepairs any way they want,
1502          * and filter and clean them up here before producing the output.
1503          */
1504         struct diff_filespec *one, *two;
1505
1506         if (DIFF_PAIR_UNMERGED(p))
1507                 return 0; /* unmerged is interesting */
1508
1509         one = p->one;
1510         two = p->two;
1511
1512         /* deletion, addition, mode or type change
1513          * and rename are all interesting.
1514          */
1515         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1516             DIFF_PAIR_MODE_CHANGED(p) ||
1517             strcmp(one->path, two->path))
1518                 return 0;
1519
1520         /* both are valid and point at the same path.  that is, we are
1521          * dealing with a change.
1522          */
1523         if (one->sha1_valid && two->sha1_valid &&
1524             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1525                 return 1; /* no change */
1526         if (!one->sha1_valid && !two->sha1_valid)
1527                 return 1; /* both look at the same file on the filesystem. */
1528         return 0;
1529 }
1530
1531 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1532 {
1533         if (diff_unmodified_pair(p))
1534                 return;
1535
1536         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1537             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1538                 return; /* no tree diffs in patch format */
1539
1540         run_diff(p, o);
1541 }
1542
1543 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1544                             struct diffstat_t *diffstat)
1545 {
1546         if (diff_unmodified_pair(p))
1547                 return;
1548
1549         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1550             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1551                 return; /* no tree diffs in patch format */
1552
1553         run_diffstat(p, o, diffstat);
1554 }
1555
1556 int diff_queue_is_empty(void)
1557 {
1558         struct diff_queue_struct *q = &diff_queued_diff;
1559         int i;
1560         for (i = 0; i < q->nr; i++)
1561                 if (!diff_unmodified_pair(q->queue[i]))
1562                         return 0;
1563         return 1;
1564 }
1565
1566 #if DIFF_DEBUG
1567 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1568 {
1569         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1570                 x, one ? one : "",
1571                 s->path,
1572                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1573                 s->mode,
1574                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1575         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1576                 x, one ? one : "",
1577                 s->size, s->xfrm_flags);
1578 }
1579
1580 void diff_debug_filepair(const struct diff_filepair *p, int i)
1581 {
1582         diff_debug_filespec(p->one, i, "one");
1583         diff_debug_filespec(p->two, i, "two");
1584         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1585                 p->score, p->status ? p->status : '?',
1586                 p->source_stays, p->broken_pair);
1587 }
1588
1589 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1590 {
1591         int i;
1592         if (msg)
1593                 fprintf(stderr, "%s\n", msg);
1594         fprintf(stderr, "q->nr = %d\n", q->nr);
1595         for (i = 0; i < q->nr; i++) {
1596                 struct diff_filepair *p = q->queue[i];
1597                 diff_debug_filepair(p, i);
1598         }
1599 }
1600 #endif
1601
1602 static void diff_resolve_rename_copy(void)
1603 {
1604         int i, j;
1605         struct diff_filepair *p, *pp;
1606         struct diff_queue_struct *q = &diff_queued_diff;
1607
1608         diff_debug_queue("resolve-rename-copy", q);
1609
1610         for (i = 0; i < q->nr; i++) {
1611                 p = q->queue[i];
1612                 p->status = 0; /* undecided */
1613                 if (DIFF_PAIR_UNMERGED(p))
1614                         p->status = DIFF_STATUS_UNMERGED;
1615                 else if (!DIFF_FILE_VALID(p->one))
1616                         p->status = DIFF_STATUS_ADDED;
1617                 else if (!DIFF_FILE_VALID(p->two))
1618                         p->status = DIFF_STATUS_DELETED;
1619                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1620                         p->status = DIFF_STATUS_TYPE_CHANGED;
1621
1622                 /* from this point on, we are dealing with a pair
1623                  * whose both sides are valid and of the same type, i.e.
1624                  * either in-place edit or rename/copy edit.
1625                  */
1626                 else if (DIFF_PAIR_RENAME(p)) {
1627                         if (p->source_stays) {
1628                                 p->status = DIFF_STATUS_COPIED;
1629                                 continue;
1630                         }
1631                         /* See if there is some other filepair that
1632                          * copies from the same source as us.  If so
1633                          * we are a copy.  Otherwise we are either a
1634                          * copy if the path stays, or a rename if it
1635                          * does not, but we already handled "stays" case.
1636                          */
1637                         for (j = i + 1; j < q->nr; j++) {
1638                                 pp = q->queue[j];
1639                                 if (strcmp(pp->one->path, p->one->path))
1640                                         continue; /* not us */
1641                                 if (!DIFF_PAIR_RENAME(pp))
1642                                         continue; /* not a rename/copy */
1643                                 /* pp is a rename/copy from the same source */
1644                                 p->status = DIFF_STATUS_COPIED;
1645                                 break;
1646                         }
1647                         if (!p->status)
1648                                 p->status = DIFF_STATUS_RENAMED;
1649                 }
1650                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1651                          p->one->mode != p->two->mode)
1652                         p->status = DIFF_STATUS_MODIFIED;
1653                 else {
1654                         /* This is a "no-change" entry and should not
1655                          * happen anymore, but prepare for broken callers.
1656                          */
1657                         error("feeding unmodified %s to diffcore",
1658                               p->one->path);
1659                         p->status = DIFF_STATUS_UNKNOWN;
1660                 }
1661         }
1662         diff_debug_queue("resolve-rename-copy done", q);
1663 }
1664
1665 static void flush_one_pair(struct diff_filepair *p,
1666                            int diff_output_format,
1667                            struct diff_options *options,
1668                            struct diffstat_t *diffstat)
1669 {
1670         int inter_name_termination = '\t';
1671         int line_termination = options->line_termination;
1672         if (!line_termination)
1673                 inter_name_termination = 0;
1674
1675         switch (p->status) {
1676         case DIFF_STATUS_UNKNOWN:
1677                 break;
1678         case 0:
1679                 die("internal error in diff-resolve-rename-copy");
1680                 break;
1681         default:
1682                 switch (diff_output_format) {
1683                 case DIFF_FORMAT_DIFFSTAT:
1684                         diff_flush_stat(p, options, diffstat);
1685                         break;
1686                 case DIFF_FORMAT_PATCH:
1687                         diff_flush_patch(p, options);
1688                         break;
1689                 case DIFF_FORMAT_RAW:
1690                 case DIFF_FORMAT_NAME_STATUS:
1691                         diff_flush_raw(p, line_termination,
1692                                        inter_name_termination,
1693                                        options, diff_output_format);
1694                         break;
1695                 case DIFF_FORMAT_NAME:
1696                         diff_flush_name(p,
1697                                         inter_name_termination,
1698                                         line_termination);
1699                         break;
1700                 case DIFF_FORMAT_NO_OUTPUT:
1701                         break;
1702                 }
1703         }
1704 }
1705
1706 void diff_flush(struct diff_options *options)
1707 {
1708         struct diff_queue_struct *q = &diff_queued_diff;
1709         int i;
1710         int diff_output_format = options->output_format;
1711         struct diffstat_t *diffstat = NULL;
1712
1713         if (diff_output_format == DIFF_FORMAT_DIFFSTAT || options->with_stat) {
1714                 diffstat = xcalloc(sizeof (struct diffstat_t), 1);
1715                 diffstat->xm.consume = diffstat_consume;
1716         }
1717
1718         if (options->with_raw) {
1719                 for (i = 0; i < q->nr; i++) {
1720                         struct diff_filepair *p = q->queue[i];
1721                         flush_one_pair(p, DIFF_FORMAT_RAW, options, NULL);
1722                 }
1723                 putchar(options->line_termination);
1724         }
1725         if (options->with_stat) {
1726                 for (i = 0; i < q->nr; i++) {
1727                         struct diff_filepair *p = q->queue[i];
1728                         flush_one_pair(p, DIFF_FORMAT_DIFFSTAT, options,
1729                                        diffstat);
1730                 }
1731                 show_stats(diffstat);
1732                 free(diffstat);
1733                 diffstat = NULL;
1734                 putchar(options->line_termination);
1735         }
1736         for (i = 0; i < q->nr; i++) {
1737                 struct diff_filepair *p = q->queue[i];
1738                 flush_one_pair(p, diff_output_format, options, diffstat);
1739                 diff_free_filepair(p);
1740         }
1741
1742         if (diffstat) {
1743                 show_stats(diffstat);
1744                 free(diffstat);
1745         }
1746
1747         free(q->queue);
1748         q->queue = NULL;
1749         q->nr = q->alloc = 0;
1750 }
1751
1752 static void diffcore_apply_filter(const char *filter)
1753 {
1754         int i;
1755         struct diff_queue_struct *q = &diff_queued_diff;
1756         struct diff_queue_struct outq;
1757         outq.queue = NULL;
1758         outq.nr = outq.alloc = 0;
1759
1760         if (!filter)
1761                 return;
1762
1763         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1764                 int found;
1765                 for (i = found = 0; !found && i < q->nr; i++) {
1766                         struct diff_filepair *p = q->queue[i];
1767                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1768                              ((p->score &&
1769                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1770                               (!p->score &&
1771                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1772                             ((p->status != DIFF_STATUS_MODIFIED) &&
1773                              strchr(filter, p->status)))
1774                                 found++;
1775                 }
1776                 if (found)
1777                         return;
1778
1779                 /* otherwise we will clear the whole queue
1780                  * by copying the empty outq at the end of this
1781                  * function, but first clear the current entries
1782                  * in the queue.
1783                  */
1784                 for (i = 0; i < q->nr; i++)
1785                         diff_free_filepair(q->queue[i]);
1786         }
1787         else {
1788                 /* Only the matching ones */
1789                 for (i = 0; i < q->nr; i++) {
1790                         struct diff_filepair *p = q->queue[i];
1791
1792                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1793                              ((p->score &&
1794                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1795                               (!p->score &&
1796                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1797                             ((p->status != DIFF_STATUS_MODIFIED) &&
1798                              strchr(filter, p->status)))
1799                                 diff_q(&outq, p);
1800                         else
1801                                 diff_free_filepair(p);
1802                 }
1803         }
1804         free(q->queue);
1805         *q = outq;
1806 }
1807
1808 void diffcore_std(struct diff_options *options)
1809 {
1810         if (options->break_opt != -1)
1811                 diffcore_break(options->break_opt);
1812         if (options->detect_rename)
1813                 diffcore_rename(options);
1814         if (options->break_opt != -1)
1815                 diffcore_merge_broken();
1816         if (options->pickaxe)
1817                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1818         if (options->orderfile)
1819                 diffcore_order(options->orderfile);
1820         diff_resolve_rename_copy();
1821         diffcore_apply_filter(options->filter);
1822 }
1823
1824
1825 void diffcore_std_no_resolve(struct diff_options *options)
1826 {
1827         if (options->pickaxe)
1828                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1829         if (options->orderfile)
1830                 diffcore_order(options->orderfile);
1831         diffcore_apply_filter(options->filter);
1832 }
1833
1834 void diff_addremove(struct diff_options *options,
1835                     int addremove, unsigned mode,
1836                     const unsigned char *sha1,
1837                     const char *base, const char *path)
1838 {
1839         char concatpath[PATH_MAX];
1840         struct diff_filespec *one, *two;
1841
1842         /* This may look odd, but it is a preparation for
1843          * feeding "there are unchanged files which should
1844          * not produce diffs, but when you are doing copy
1845          * detection you would need them, so here they are"
1846          * entries to the diff-core.  They will be prefixed
1847          * with something like '=' or '*' (I haven't decided
1848          * which but should not make any difference).
1849          * Feeding the same new and old to diff_change()
1850          * also has the same effect.
1851          * Before the final output happens, they are pruned after
1852          * merged into rename/copy pairs as appropriate.
1853          */
1854         if (options->reverse_diff)
1855                 addremove = (addremove == '+' ? '-' :
1856                              addremove == '-' ? '+' : addremove);
1857
1858         if (!path) path = "";
1859         sprintf(concatpath, "%s%s", base, path);
1860         one = alloc_filespec(concatpath);
1861         two = alloc_filespec(concatpath);
1862
1863         if (addremove != '+')
1864                 fill_filespec(one, sha1, mode);
1865         if (addremove != '-')
1866                 fill_filespec(two, sha1, mode);
1867
1868         diff_queue(&diff_queued_diff, one, two);
1869 }
1870
1871 void diff_change(struct diff_options *options,
1872                  unsigned old_mode, unsigned new_mode,
1873                  const unsigned char *old_sha1,
1874                  const unsigned char *new_sha1,
1875                  const char *base, const char *path)
1876 {
1877         char concatpath[PATH_MAX];
1878         struct diff_filespec *one, *two;
1879
1880         if (options->reverse_diff) {
1881                 unsigned tmp;
1882                 const unsigned char *tmp_c;
1883                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1884                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1885         }
1886         if (!path) path = "";
1887         sprintf(concatpath, "%s%s", base, path);
1888         one = alloc_filespec(concatpath);
1889         two = alloc_filespec(concatpath);
1890         fill_filespec(one, old_sha1, old_mode);
1891         fill_filespec(two, new_sha1, new_mode);
1892
1893         diff_queue(&diff_queued_diff, one, two);
1894 }
1895
1896 void diff_unmerge(struct diff_options *options,
1897                   const char *path)
1898 {
1899         struct diff_filespec *one, *two;
1900         one = alloc_filespec(path);
1901         two = alloc_filespec(path);
1902         diff_queue(&diff_queued_diff, one, two);
1903 }