diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15 #include "sigchain.h"
  16 #include "submodule.h"
  17
  18 #ifdef NO_FAST_WORKING_DIRECTORY
  19 #define FAST_WORKING_DIRECTORY 0
  20 #else
  21 #define FAST_WORKING_DIRECTORY 1
  22 #endif
  23
  24 static int diff_detect_rename_default;
  25 static int diff_rename_limit_default = 200;
  26 static int diff_suppress_blank_empty;
  27 int diff_use_color_default = -1;
  28 static const char *diff_word_regex_cfg;
  29 static const char *external_diff_cmd_cfg;
  30 int diff_auto_refresh_index = 1;
  31 static int diff_mnemonic_prefix;
  32
  33 static char diff_colors[][COLOR_MAXLEN] = {
  34         GIT_COLOR_RESET,
  35         GIT_COLOR_NORMAL,       /* PLAIN */
  36         GIT_COLOR_BOLD,         /* METAINFO */
  37         GIT_COLOR_CYAN,         /* FRAGINFO */
  38         GIT_COLOR_RED,          /* OLD */
  39         GIT_COLOR_GREEN,        /* NEW */
  40         GIT_COLOR_YELLOW,       /* COMMIT */
  41         GIT_COLOR_BG_RED,       /* WHITESPACE */
  42         GIT_COLOR_NORMAL,       /* FUNCINFO */
  43 };
  44
  45 static void diff_filespec_load_driver(struct diff_filespec *one);
  46 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
  47
  48 static int parse_diff_color_slot(const char *var, int ofs)
  49 {
  50         if (!strcasecmp(var+ofs, "plain"))
  51                 return DIFF_PLAIN;
  52         if (!strcasecmp(var+ofs, "meta"))
  53                 return DIFF_METAINFO;
  54         if (!strcasecmp(var+ofs, "frag"))
  55                 return DIFF_FRAGINFO;
  56         if (!strcasecmp(var+ofs, "old"))
  57                 return DIFF_FILE_OLD;
  58         if (!strcasecmp(var+ofs, "new"))
  59                 return DIFF_FILE_NEW;
  60         if (!strcasecmp(var+ofs, "commit"))
  61                 return DIFF_COMMIT;
  62         if (!strcasecmp(var+ofs, "whitespace"))
  63                 return DIFF_WHITESPACE;
  64         if (!strcasecmp(var+ofs, "func"))
  65                 return DIFF_FUNCINFO;
  66         return -1;
  67 }
  68
  69 static int git_config_rename(const char *var, const char *value)
  70 {
  71         if (!value)
  72                 return DIFF_DETECT_RENAME;
  73         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
  74                 return  DIFF_DETECT_COPY;
  75         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
  76 }
  77
  78 /*
  79  * These are to give UI layer defaults.
  80  * The core-level commands such as git-diff-files should
  81  * never be affected by the setting of diff.renames
  82  * the user happens to have in the configuration file.
  83  */
  84 int git_diff_ui_config(const char *var, const char *value, void *cb)
  85 {
  86         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  87                 diff_use_color_default = git_config_colorbool(var, value, -1);
  88                 return 0;
  89         }
  90         if (!strcmp(var, "diff.renames")) {
  91                 diff_detect_rename_default = git_config_rename(var, value);
  92                 return 0;
  93         }
  94         if (!strcmp(var, "diff.autorefreshindex")) {
  95                 diff_auto_refresh_index = git_config_bool(var, value);
  96                 return 0;
  97         }
  98         if (!strcmp(var, "diff.mnemonicprefix")) {
  99                 diff_mnemonic_prefix = git_config_bool(var, value);
 100                 return 0;
 101         }
 102         if (!strcmp(var, "diff.external"))
 103                 return git_config_string(&external_diff_cmd_cfg, var, value);
 104         if (!strcmp(var, "diff.wordregex"))
 105                 return git_config_string(&diff_word_regex_cfg, var, value);
 106
 107         return git_diff_basic_config(var, value, cb);
 108 }
 109
 110 int git_diff_basic_config(const char *var, const char *value, void *cb)
 111 {
 112         if (!strcmp(var, "diff.renamelimit")) {
 113                 diff_rename_limit_default = git_config_int(var, value);
 114                 return 0;
 115         }
 116
 117         switch (userdiff_config(var, value)) {
 118                 case 0: break;
 119                 case -1: return -1;
 120                 default: return 0;
 121         }
 122
 123         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 124                 int slot = parse_diff_color_slot(var, 11);
 125                 if (slot < 0)
 126                         return 0;
 127                 if (!value)
 128                         return config_error_nonbool(var);
 129                 color_parse(value, var, diff_colors[slot]);
 130                 return 0;
 131         }
 132
 133         /* like GNU diff's --suppress-blank-empty option  */
 134         if (!strcmp(var, "diff.suppressblankempty") ||
 135                         /* for backwards compatibility */
 136                         !strcmp(var, "diff.suppress-blank-empty")) {
 137                 diff_suppress_blank_empty = git_config_bool(var, value);
 138                 return 0;
 139         }
 140
 141         return git_color_default_config(var, value, cb);
 142 }
 143
 144 static char *quote_two(const char *one, const char *two)
 145 {
 146         int need_one = quote_c_style(one, NULL, NULL, 1);
 147         int need_two = quote_c_style(two, NULL, NULL, 1);
 148         struct strbuf res = STRBUF_INIT;
 149
 150         if (need_one + need_two) {
 151                 strbuf_addch(&res, '"');
 152                 quote_c_style(one, &res, NULL, 1);
 153                 quote_c_style(two, &res, NULL, 1);
 154                 strbuf_addch(&res, '"');
 155         } else {
 156                 strbuf_addstr(&res, one);
 157                 strbuf_addstr(&res, two);
 158         }
 159         return strbuf_detach(&res, NULL);
 160 }
 161
 162 static const char *external_diff(void)
 163 {
 164         static const char *external_diff_cmd = NULL;
 165         static int done_preparing = 0;
 166
 167         if (done_preparing)
 168                 return external_diff_cmd;
 169         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 170         if (!external_diff_cmd)
 171                 external_diff_cmd = external_diff_cmd_cfg;
 172         done_preparing = 1;
 173         return external_diff_cmd;
 174 }
 175
 176 static struct diff_tempfile {
 177         const char *name; /* filename external diff should read from */
 178         char hex[41];
 179         char mode[10];
 180         char tmp_path[PATH_MAX];
 181 } diff_temp[2];
 182
 183 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 184
 185 struct emit_callback {
 186         int color_diff;
 187         unsigned ws_rule;
 188         int blank_at_eof_in_preimage;
 189         int blank_at_eof_in_postimage;
 190         int lno_in_preimage;
 191         int lno_in_postimage;
 192         sane_truncate_fn truncate;
 193         const char **label_path;
 194         struct diff_words_data *diff_words;
 195         int *found_changesp;
 196         FILE *file;
 197         struct strbuf *header;
 198 };
 199
 200 static int count_lines(const char *data, int size)
 201 {
 202         int count, ch, completely_empty = 1, nl_just_seen = 0;
 203         count = 0;
 204         while (0 < size--) {
 205                 ch = *data++;
 206                 if (ch == '\n') {
 207                         count++;
 208                         nl_just_seen = 1;
 209                         completely_empty = 0;
 210                 }
 211                 else {
 212                         nl_just_seen = 0;
 213                         completely_empty = 0;
 214                 }
 215         }
 216         if (completely_empty)
 217                 return 0;
 218         if (!nl_just_seen)
 219                 count++; /* no trailing newline */
 220         return count;
 221 }
 222
 223 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 224 {
 225         if (!DIFF_FILE_VALID(one)) {
 226                 mf->ptr = (char *)""; /* does not matter */
 227                 mf->size = 0;
 228                 return 0;
 229         }
 230         else if (diff_populate_filespec(one, 0))
 231                 return -1;
 232
 233         mf->ptr = one->data;
 234         mf->size = one->size;
 235         return 0;
 236 }
 237
 238 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 239 {
 240         char *ptr = mf->ptr;
 241         long size = mf->size;
 242         int cnt = 0;
 243
 244         if (!size)
 245                 return cnt;
 246         ptr += size - 1; /* pointing at the very end */
 247         if (*ptr != '\n')
 248                 ; /* incomplete line */
 249         else
 250                 ptr--; /* skip the last LF */
 251         while (mf->ptr < ptr) {
 252                 char *prev_eol;
 253                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 254                         if (*prev_eol == '\n')
 255                                 break;
 256                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 257                         break;
 258                 cnt++;
 259                 ptr = prev_eol - 1;
 260         }
 261         return cnt;
 262 }
 263
 264 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 265                                struct emit_callback *ecbdata)
 266 {
 267         int l1, l2, at;
 268         unsigned ws_rule = ecbdata->ws_rule;
 269         l1 = count_trailing_blank(mf1, ws_rule);
 270         l2 = count_trailing_blank(mf2, ws_rule);
 271         if (l2 <= l1) {
 272                 ecbdata->blank_at_eof_in_preimage = 0;
 273                 ecbdata->blank_at_eof_in_postimage = 0;
 274                 return;
 275         }
 276         at = count_lines(mf1->ptr, mf1->size);
 277         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 278
 279         at = count_lines(mf2->ptr, mf2->size);
 280         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 281 }
 282
 283 static void emit_line_0(FILE *file, const char *set, const char *reset,
 284                         int first, const char *line, int len)
 285 {
 286         int has_trailing_newline, has_trailing_carriage_return;
 287         int nofirst;
 288
 289         if (len == 0) {
 290                 has_trailing_newline = (first == '\n');
 291                 has_trailing_carriage_return = (!has_trailing_newline &&
 292                                                 (first == '\r'));
 293                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 294         } else {
 295                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 296                 if (has_trailing_newline)
 297                         len--;
 298                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 299                 if (has_trailing_carriage_return)
 300                         len--;
 301                 nofirst = 0;
 302         }
 303
 304         if (len || !nofirst) {
 305                 fputs(set, file);
 306                 if (!nofirst)
 307                         fputc(first, file);
 308                 fwrite(line, len, 1, file);
 309                 fputs(reset, file);
 310         }
 311         if (has_trailing_carriage_return)
 312                 fputc('\r', file);
 313         if (has_trailing_newline)
 314                 fputc('\n', file);
 315 }
 316
 317 static void emit_line(FILE *file, const char *set, const char *reset,
 318                       const char *line, int len)
 319 {
 320         emit_line_0(file, set, reset, line[0], line+1, len-1);
 321 }
 322
 323 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 324 {
 325         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 326               ecbdata->blank_at_eof_in_preimage &&
 327               ecbdata->blank_at_eof_in_postimage &&
 328               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 329               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 330                 return 0;
 331         return ws_blank_line(line, len, ecbdata->ws_rule);
 332 }
 333
 334 static void emit_add_line(const char *reset,
 335                           struct emit_callback *ecbdata,
 336                           const char *line, int len)
 337 {
 338         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 339         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 340
 341         if (!*ws)
 342                 emit_line_0(ecbdata->file, set, reset, '+', line, len);
 343         else if (new_blank_line_at_eof(ecbdata, line, len))
 344                 /* Blank line at EOF - paint '+' as well */
 345                 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
 346         else {
 347                 /* Emit just the prefix, then the rest. */
 348                 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
 349                 ws_check_emit(line, len, ecbdata->ws_rule,
 350                               ecbdata->file, set, reset, ws);
 351         }
 352 }
 353
 354 static void emit_hunk_header(struct emit_callback *ecbdata,
 355                              const char *line, int len)
 356 {
 357         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 358         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 359         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 360         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 361         static const char atat[2] = { '@', '@' };
 362         const char *cp, *ep;
 363
 364         /*
 365          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 366          * it always is at least 10 bytes long.
 367          */
 368         if (len < 10 ||
 369             memcmp(line, atat, 2) ||
 370             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 371                 emit_line(ecbdata->file, plain, reset, line, len);
 372                 return;
 373         }
 374         ep += 2; /* skip over @@ */
 375
 376         /* The hunk header in fraginfo color */
 377         emit_line(ecbdata->file, frag, reset, line, ep - line);
 378
 379         /* blank before the func header */
 380         for (cp = ep; ep - line < len; ep++)
 381                 if (*ep != ' ' && *ep != '\t')
 382                         break;
 383         if (ep != cp)
 384                 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
 385
 386         if (ep < line + len)
 387                 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
 388 }
 389
 390 static struct diff_tempfile *claim_diff_tempfile(void) {
 391         int i;
 392         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 393                 if (!diff_temp[i].name)
 394                         return diff_temp + i;
 395         die("BUG: diff is failing to clean up its tempfiles");
 396 }
 397
 398 static int remove_tempfile_installed;
 399
 400 static void remove_tempfile(void)
 401 {
 402         int i;
 403         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 404                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 405                         unlink_or_warn(diff_temp[i].name);
 406                 diff_temp[i].name = NULL;
 407         }
 408 }
 409
 410 static void remove_tempfile_on_signal(int signo)
 411 {
 412         remove_tempfile();
 413         sigchain_pop(signo);
 414         raise(signo);
 415 }
 416
 417 static void print_line_count(FILE *file, int count)
 418 {
 419         switch (count) {
 420         case 0:
 421                 fprintf(file, "0,0");
 422                 break;
 423         case 1:
 424                 fprintf(file, "1");
 425                 break;
 426         default:
 427                 fprintf(file, "1,%d", count);
 428                 break;
 429         }
 430 }
 431
 432 static void emit_rewrite_lines(struct emit_callback *ecb,
 433                                int prefix, const char *data, int size)
 434 {
 435         const char *endp = NULL;
 436         static const char *nneof = " No newline at end of file\n";
 437         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 438         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 439
 440         while (0 < size) {
 441                 int len;
 442
 443                 endp = memchr(data, '\n', size);
 444                 len = endp ? (endp - data + 1) : size;
 445                 if (prefix != '+') {
 446                         ecb->lno_in_preimage++;
 447                         emit_line_0(ecb->file, old, reset, '-',
 448                                     data, len);
 449                 } else {
 450                         ecb->lno_in_postimage++;
 451                         emit_add_line(reset, ecb, data, len);
 452                 }
 453                 size -= len;
 454                 data += len;
 455         }
 456         if (!endp) {
 457                 const char *plain = diff_get_color(ecb->color_diff,
 458                                                    DIFF_PLAIN);
 459                 emit_line_0(ecb->file, plain, reset, '\\',
 460                             nneof, strlen(nneof));
 461         }
 462 }
 463
 464 static void emit_rewrite_diff(const char *name_a,
 465                               const char *name_b,
 466                               struct diff_filespec *one,
 467                               struct diff_filespec *two,
 468                               const char *textconv_one,
 469                               const char *textconv_two,
 470                               struct diff_options *o)
 471 {
 472         int lc_a, lc_b;
 473         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 474         const char *name_a_tab, *name_b_tab;
 475         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 476         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 477         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 478         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 479         const char *a_prefix, *b_prefix;
 480         const char *data_one, *data_two;
 481         size_t size_one, size_two;
 482         struct emit_callback ecbdata;
 483
 484         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 485                 a_prefix = o->b_prefix;
 486                 b_prefix = o->a_prefix;
 487         } else {
 488                 a_prefix = o->a_prefix;
 489                 b_prefix = o->b_prefix;
 490         }
 491
 492         name_a += (*name_a == '/');
 493         name_b += (*name_b == '/');
 494         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 495         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 496
 497         strbuf_reset(&a_name);
 498         strbuf_reset(&b_name);
 499         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 500         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 501
 502         diff_populate_filespec(one, 0);
 503         diff_populate_filespec(two, 0);
 504         if (textconv_one) {
 505                 data_one = run_textconv(textconv_one, one, &size_one);
 506                 if (!data_one)
 507                         die("unable to read files to diff");
 508         }
 509         else {
 510                 data_one = one->data;
 511                 size_one = one->size;
 512         }
 513         if (textconv_two) {
 514                 data_two = run_textconv(textconv_two, two, &size_two);
 515                 if (!data_two)
 516                         die("unable to read files to diff");
 517         }
 518         else {
 519                 data_two = two->data;
 520                 size_two = two->size;
 521         }
 522
 523         memset(&ecbdata, 0, sizeof(ecbdata));
 524         ecbdata.color_diff = color_diff;
 525         ecbdata.found_changesp = &o->found_changes;
 526         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
 527         ecbdata.file = o->file;
 528         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 529                 mmfile_t mf1, mf2;
 530                 mf1.ptr = (char *)data_one;
 531                 mf2.ptr = (char *)data_two;
 532                 mf1.size = size_one;
 533                 mf2.size = size_two;
 534                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 535         }
 536         ecbdata.lno_in_preimage = 1;
 537         ecbdata.lno_in_postimage = 1;
 538
 539         lc_a = count_lines(data_one, size_one);
 540         lc_b = count_lines(data_two, size_two);
 541         fprintf(o->file,
 542                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 543                 metainfo, a_name.buf, name_a_tab, reset,
 544                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 545         print_line_count(o->file, lc_a);
 546         fprintf(o->file, " +");
 547         print_line_count(o->file, lc_b);
 548         fprintf(o->file, " @@%s\n", reset);
 549         if (lc_a)
 550                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 551         if (lc_b)
 552                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 553         if (textconv_one)
 554                 free(data_one);
 555         if (textconv_two)
 556                 free(data_two);
 557 }
 558
 559 struct diff_words_buffer {
 560         mmfile_t text;
 561         long alloc;
 562         struct diff_words_orig {
 563                 const char *begin, *end;
 564         } *orig;
 565         int orig_nr, orig_alloc;
 566 };
 567
 568 static void diff_words_append(char *line, unsigned long len,
 569                 struct diff_words_buffer *buffer)
 570 {
 571         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 572         line++;
 573         len--;
 574         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 575         buffer->text.size += len;
 576         buffer->text.ptr[buffer->text.size] = '\0';
 577 }
 578
 579 struct diff_words_data {
 580         struct diff_words_buffer minus, plus;
 581         const char *current_plus;
 582         FILE *file;
 583         regex_t *word_regex;
 584 };
 585
 586 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 587 {
 588         struct diff_words_data *diff_words = priv;
 589         int minus_first, minus_len, plus_first, plus_len;
 590         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 591
 592         if (line[0] != '@' || parse_hunk_header(line, len,
 593                         &minus_first, &minus_len, &plus_first, &plus_len))
 594                 return;
 595
 596         /* POSIX requires that first be decremented by one if len == 0... */
 597         if (minus_len) {
 598                 minus_begin = diff_words->minus.orig[minus_first].begin;
 599                 minus_end =
 600                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 601         } else
 602                 minus_begin = minus_end =
 603                         diff_words->minus.orig[minus_first].end;
 604
 605         if (plus_len) {
 606                 plus_begin = diff_words->plus.orig[plus_first].begin;
 607                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 608         } else
 609                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 610
 611         if (diff_words->current_plus != plus_begin)
 612                 fwrite(diff_words->current_plus,
 613                                 plus_begin - diff_words->current_plus, 1,
 614                                 diff_words->file);
 615         if (minus_begin != minus_end)
 616                 color_fwrite_lines(diff_words->file,
 617                                 diff_get_color(1, DIFF_FILE_OLD),
 618                                 minus_end - minus_begin, minus_begin);
 619         if (plus_begin != plus_end)
 620                 color_fwrite_lines(diff_words->file,
 621                                 diff_get_color(1, DIFF_FILE_NEW),
 622                                 plus_end - plus_begin, plus_begin);
 623
 624         diff_words->current_plus = plus_end;
 625 }
 626
 627 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 628 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 629                 int *begin, int *end)
 630 {
 631         if (word_regex && *begin < buffer->size) {
 632                 regmatch_t match[1];
 633                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 634                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 635                                         '\n', match[0].rm_eo - match[0].rm_so);
 636                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 637                         *begin += match[0].rm_so;
 638                         return *begin >= *end;
 639                 }
 640                 return -1;
 641         }
 642
 643         /* find the next word */
 644         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 645                 (*begin)++;
 646         if (*begin >= buffer->size)
 647                 return -1;
 648
 649         /* find the end of the word */
 650         *end = *begin + 1;
 651         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 652                 (*end)++;
 653
 654         return 0;
 655 }
 656
 657 /*
 658  * This function splits the words in buffer->text, stores the list with
 659  * newline separator into out, and saves the offsets of the original words
 660  * in buffer->orig.
 661  */
 662 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 663                 regex_t *word_regex)
 664 {
 665         int i, j;
 666         long alloc = 0;
 667
 668         out->size = 0;
 669         out->ptr = NULL;
 670
 671         /* fake an empty "0th" word */
 672         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 673         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 674         buffer->orig_nr = 1;
 675
 676         for (i = 0; i < buffer->text.size; i++) {
 677                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 678                         return;
 679
 680                 /* store original boundaries */
 681                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 682                                 buffer->orig_alloc);
 683                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 684                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 685                 buffer->orig_nr++;
 686
 687                 /* store one word */
 688                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 689                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 690                 out->ptr[out->size + j - i] = '\n';
 691                 out->size += j - i + 1;
 692
 693                 i = j - 1;
 694         }
 695 }
 696
 697 /* this executes the word diff on the accumulated buffers */
 698 static void diff_words_show(struct diff_words_data *diff_words)
 699 {
 700         xpparam_t xpp;
 701         xdemitconf_t xecfg;
 702         xdemitcb_t ecb;
 703         mmfile_t minus, plus;
 704
 705         /* special case: only removal */
 706         if (!diff_words->plus.text.size) {
 707                 color_fwrite_lines(diff_words->file,
 708                         diff_get_color(1, DIFF_FILE_OLD),
 709                         diff_words->minus.text.size, diff_words->minus.text.ptr);
 710                 diff_words->minus.text.size = 0;
 711                 return;
 712         }
 713
 714         diff_words->current_plus = diff_words->plus.text.ptr;
 715
 716         memset(&xpp, 0, sizeof(xpp));
 717         memset(&xecfg, 0, sizeof(xecfg));
 718         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 719         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
 720         xpp.flags = XDF_NEED_MINIMAL;
 721         /* as only the hunk header will be parsed, we need a 0-context */
 722         xecfg.ctxlen = 0;
 723         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 724                       &xpp, &xecfg, &ecb);
 725         free(minus.ptr);
 726         free(plus.ptr);
 727         if (diff_words->current_plus != diff_words->plus.text.ptr +
 728                         diff_words->plus.text.size)
 729                 fwrite(diff_words->current_plus,
 730                         diff_words->plus.text.ptr + diff_words->plus.text.size
 731                         - diff_words->current_plus, 1,
 732                         diff_words->file);
 733         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 734 }
 735
 736 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
 737 static void diff_words_flush(struct emit_callback *ecbdata)
 738 {
 739         if (ecbdata->diff_words->minus.text.size ||
 740             ecbdata->diff_words->plus.text.size)
 741                 diff_words_show(ecbdata->diff_words);
 742 }
 743
 744 static void free_diff_words_data(struct emit_callback *ecbdata)
 745 {
 746         if (ecbdata->diff_words) {
 747                 diff_words_flush(ecbdata);
 748                 free (ecbdata->diff_words->minus.text.ptr);
 749                 free (ecbdata->diff_words->minus.orig);
 750                 free (ecbdata->diff_words->plus.text.ptr);
 751                 free (ecbdata->diff_words->plus.orig);
 752                 free(ecbdata->diff_words->word_regex);
 753                 free(ecbdata->diff_words);
 754                 ecbdata->diff_words = NULL;
 755         }
 756 }
 757
 758 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 759 {
 760         if (diff_use_color)
 761                 return diff_colors[ix];
 762         return "";
 763 }
 764
 765 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 766 {
 767         const char *cp;
 768         unsigned long allot;
 769         size_t l = len;
 770
 771         if (ecb->truncate)
 772                 return ecb->truncate(line, len);
 773         cp = line;
 774         allot = l;
 775         while (0 < l) {
 776                 (void) utf8_width(&cp, &l);
 777                 if (!cp)
 778                         break; /* truncated in the middle? */
 779         }
 780         return allot - l;
 781 }
 782
 783 static void find_lno(const char *line, struct emit_callback *ecbdata)
 784 {
 785         const char *p;
 786         ecbdata->lno_in_preimage = 0;
 787         ecbdata->lno_in_postimage = 0;
 788         p = strchr(line, '-');
 789         if (!p)
 790                 return; /* cannot happen */
 791         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
 792         p = strchr(p, '+');
 793         if (!p)
 794                 return; /* cannot happen */
 795         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
 796 }
 797
 798 static void fn_out_consume(void *priv, char *line, unsigned long len)
 799 {
 800         struct emit_callback *ecbdata = priv;
 801         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 802         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 803         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 804
 805         if (ecbdata->header) {
 806                 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
 807                 strbuf_reset(ecbdata->header);
 808                 ecbdata->header = NULL;
 809         }
 810         *(ecbdata->found_changesp) = 1;
 811
 812         if (ecbdata->label_path[0]) {
 813                 const char *name_a_tab, *name_b_tab;
 814
 815                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 816                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 817
 818                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 819                         meta, ecbdata->label_path[0], reset, name_a_tab);
 820                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 821                         meta, ecbdata->label_path[1], reset, name_b_tab);
 822                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 823         }
 824
 825         if (diff_suppress_blank_empty
 826             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 827                 line[0] = '\n';
 828                 len = 1;
 829         }
 830
 831         if (line[0] == '@') {
 832                 if (ecbdata->diff_words)
 833                         diff_words_flush(ecbdata);
 834                 len = sane_truncate_line(ecbdata, line, len);
 835                 find_lno(line, ecbdata);
 836                 emit_hunk_header(ecbdata, line, len);
 837                 if (line[len-1] != '\n')
 838                         putc('\n', ecbdata->file);
 839                 return;
 840         }
 841
 842         if (len < 1) {
 843                 emit_line(ecbdata->file, reset, reset, line, len);
 844                 return;
 845         }
 846
 847         if (ecbdata->diff_words) {
 848                 if (line[0] == '-') {
 849                         diff_words_append(line, len,
 850                                           &ecbdata->diff_words->minus);
 851                         return;
 852                 } else if (line[0] == '+') {
 853                         diff_words_append(line, len,
 854                                           &ecbdata->diff_words->plus);
 855                         return;
 856                 }
 857                 diff_words_flush(ecbdata);
 858                 line++;
 859                 len--;
 860                 emit_line(ecbdata->file, plain, reset, line, len);
 861                 return;
 862         }
 863
 864         if (line[0] != '+') {
 865                 const char *color =
 866                         diff_get_color(ecbdata->color_diff,
 867                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
 868                 ecbdata->lno_in_preimage++;
 869                 if (line[0] == ' ')
 870                         ecbdata->lno_in_postimage++;
 871                 emit_line(ecbdata->file, color, reset, line, len);
 872         } else {
 873                 ecbdata->lno_in_postimage++;
 874                 emit_add_line(reset, ecbdata, line + 1, len - 1);
 875         }
 876 }
 877
 878 static char *pprint_rename(const char *a, const char *b)
 879 {
 880         const char *old = a;
 881         const char *new = b;
 882         struct strbuf name = STRBUF_INIT;
 883         int pfx_length, sfx_length;
 884         int len_a = strlen(a);
 885         int len_b = strlen(b);
 886         int a_midlen, b_midlen;
 887         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 888         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 889
 890         if (qlen_a || qlen_b) {
 891                 quote_c_style(a, &name, NULL, 0);
 892                 strbuf_addstr(&name, " => ");
 893                 quote_c_style(b, &name, NULL, 0);
 894                 return strbuf_detach(&name, NULL);
 895         }
 896
 897         /* Find common prefix */
 898         pfx_length = 0;
 899         while (*old && *new && *old == *new) {
 900                 if (*old == '/')
 901                         pfx_length = old - a + 1;
 902                 old++;
 903                 new++;
 904         }
 905
 906         /* Find common suffix */
 907         old = a + len_a;
 908         new = b + len_b;
 909         sfx_length = 0;
 910         while (a <= old && b <= new && *old == *new) {
 911                 if (*old == '/')
 912                         sfx_length = len_a - (old - a);
 913                 old--;
 914                 new--;
 915         }
 916
 917         /*
 918          * pfx{mid-a => mid-b}sfx
 919          * {pfx-a => pfx-b}sfx
 920          * pfx{sfx-a => sfx-b}
 921          * name-a => name-b
 922          */
 923         a_midlen = len_a - pfx_length - sfx_length;
 924         b_midlen = len_b - pfx_length - sfx_length;
 925         if (a_midlen < 0)
 926                 a_midlen = 0;
 927         if (b_midlen < 0)
 928                 b_midlen = 0;
 929
 930         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 931         if (pfx_length + sfx_length) {
 932                 strbuf_add(&name, a, pfx_length);
 933                 strbuf_addch(&name, '{');
 934         }
 935         strbuf_add(&name, a + pfx_length, a_midlen);
 936         strbuf_addstr(&name, " => ");
 937         strbuf_add(&name, b + pfx_length, b_midlen);
 938         if (pfx_length + sfx_length) {
 939                 strbuf_addch(&name, '}');
 940                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 941         }
 942         return strbuf_detach(&name, NULL);
 943 }
 944
 945 struct diffstat_t {
 946         int nr;
 947         int alloc;
 948         struct diffstat_file {
 949                 char *from_name;
 950                 char *name;
 951                 char *print_name;
 952                 unsigned is_unmerged:1;
 953                 unsigned is_binary:1;
 954                 unsigned is_renamed:1;
 955                 unsigned int added, deleted;
 956         } **files;
 957 };
 958
 959 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 960                                           const char *name_a,
 961                                           const char *name_b)
 962 {
 963         struct diffstat_file *x;
 964         x = xcalloc(sizeof (*x), 1);
 965         if (diffstat->nr == diffstat->alloc) {
 966                 diffstat->alloc = alloc_nr(diffstat->alloc);
 967                 diffstat->files = xrealloc(diffstat->files,
 968                                 diffstat->alloc * sizeof(x));
 969         }
 970         diffstat->files[diffstat->nr++] = x;
 971         if (name_b) {
 972                 x->from_name = xstrdup(name_a);
 973                 x->name = xstrdup(name_b);
 974                 x->is_renamed = 1;
 975         }
 976         else {
 977                 x->from_name = NULL;
 978                 x->name = xstrdup(name_a);
 979         }
 980         return x;
 981 }
 982
 983 static void diffstat_consume(void *priv, char *line, unsigned long len)
 984 {
 985         struct diffstat_t *diffstat = priv;
 986         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 987
 988         if (line[0] == '+')
 989                 x->added++;
 990         else if (line[0] == '-')
 991                 x->deleted++;
 992 }
 993
 994 const char mime_boundary_leader[] = "------------";
 995
 996 static int scale_linear(int it, int width, int max_change)
 997 {
 998         /*
 999          * make sure that at least one '-' is printed if there were deletions,
1000          * and likewise for '+'.
1001          */
1002         if (max_change < 2)
1003                 return it;
1004         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1005 }
1006
1007 static void show_name(FILE *file,
1008                       const char *prefix, const char *name, int len)
1009 {
1010         fprintf(file, " %s%-*s |", prefix, len, name);
1011 }
1012
1013 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1014 {
1015         if (cnt <= 0)
1016                 return;
1017         fprintf(file, "%s", set);
1018         while (cnt--)
1019                 putc(ch, file);
1020         fprintf(file, "%s", reset);
1021 }
1022
1023 static void fill_print_name(struct diffstat_file *file)
1024 {
1025         char *pname;
1026
1027         if (file->print_name)
1028                 return;
1029
1030         if (!file->is_renamed) {
1031                 struct strbuf buf = STRBUF_INIT;
1032                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1033                         pname = strbuf_detach(&buf, NULL);
1034                 } else {
1035                         pname = file->name;
1036                         strbuf_release(&buf);
1037                 }
1038         } else {
1039                 pname = pprint_rename(file->from_name, file->name);
1040         }
1041         file->print_name = pname;
1042 }
1043
1044 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1045 {
1046         int i, len, add, del, adds = 0, dels = 0;
1047         int max_change = 0, max_len = 0;
1048         int total_files = data->nr;
1049         int width, name_width;
1050         const char *reset, *set, *add_c, *del_c;
1051
1052         if (data->nr == 0)
1053                 return;
1054
1055         width = options->stat_width ? options->stat_width : 80;
1056         name_width = options->stat_name_width ? options->stat_name_width : 50;
1057
1058         /* Sanity: give at least 5 columns to the graph,
1059          * but leave at least 10 columns for the name.
1060          */
1061         if (width < 25)
1062                 width = 25;
1063         if (name_width < 10)
1064                 name_width = 10;
1065         else if (width < name_width + 15)
1066                 name_width = width - 15;
1067
1068         /* Find the longest filename and max number of changes */
1069         reset = diff_get_color_opt(options, DIFF_RESET);
1070         set   = diff_get_color_opt(options, DIFF_PLAIN);
1071         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1072         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1073
1074         for (i = 0; i < data->nr; i++) {
1075                 struct diffstat_file *file = data->files[i];
1076                 int change = file->added + file->deleted;
1077                 fill_print_name(file);
1078                 len = strlen(file->print_name);
1079                 if (max_len < len)
1080                         max_len = len;
1081
1082                 if (file->is_binary || file->is_unmerged)
1083                         continue;
1084                 if (max_change < change)
1085                         max_change = change;
1086         }
1087
1088         /* Compute the width of the graph part;
1089          * 10 is for one blank at the beginning of the line plus
1090          * " | count " between the name and the graph.
1091          *
1092          * From here on, name_width is the width of the name area,
1093          * and width is the width of the graph area.
1094          */
1095         name_width = (name_width < max_len) ? name_width : max_len;
1096         if (width < (name_width + 10) + max_change)
1097                 width = width - (name_width + 10);
1098         else
1099                 width = max_change;
1100
1101         for (i = 0; i < data->nr; i++) {
1102                 const char *prefix = "";
1103                 char *name = data->files[i]->print_name;
1104                 int added = data->files[i]->added;
1105                 int deleted = data->files[i]->deleted;
1106                 int name_len;
1107
1108                 /*
1109                  * "scale" the filename
1110                  */
1111                 len = name_width;
1112                 name_len = strlen(name);
1113                 if (name_width < name_len) {
1114                         char *slash;
1115                         prefix = "...";
1116                         len -= 3;
1117                         name += name_len - len;
1118                         slash = strchr(name, '/');
1119                         if (slash)
1120                                 name = slash;
1121                 }
1122
1123                 if (data->files[i]->is_binary) {
1124                         show_name(options->file, prefix, name, len);
1125                         fprintf(options->file, "  Bin ");
1126                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
1127                         fprintf(options->file, " -> ");
1128                         fprintf(options->file, "%s%d%s", add_c, added, reset);
1129                         fprintf(options->file, " bytes");
1130                         fprintf(options->file, "\n");
1131                         continue;
1132                 }
1133                 else if (data->files[i]->is_unmerged) {
1134                         show_name(options->file, prefix, name, len);
1135                         fprintf(options->file, "  Unmerged\n");
1136                         continue;
1137                 }
1138                 else if (!data->files[i]->is_renamed &&
1139                          (added + deleted == 0)) {
1140                         total_files--;
1141                         continue;
1142                 }
1143
1144                 /*
1145                  * scale the add/delete
1146                  */
1147                 add = added;
1148                 del = deleted;
1149                 adds += add;
1150                 dels += del;
1151
1152                 if (width <= max_change) {
1153                         add = scale_linear(add, width, max_change);
1154                         del = scale_linear(del, width, max_change);
1155                 }
1156                 show_name(options->file, prefix, name, len);
1157                 fprintf(options->file, "%5d%s", added + deleted,
1158                                 added + deleted ? " " : "");
1159                 show_graph(options->file, '+', add, add_c, reset);
1160                 show_graph(options->file, '-', del, del_c, reset);
1161                 fprintf(options->file, "\n");
1162         }
1163         fprintf(options->file,
1164                " %d files changed, %d insertions(+), %d deletions(-)\n",
1165                total_files, adds, dels);
1166 }
1167
1168 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1169 {
1170         int i, adds = 0, dels = 0, total_files = data->nr;
1171
1172         if (data->nr == 0)
1173                 return;
1174
1175         for (i = 0; i < data->nr; i++) {
1176                 if (!data->files[i]->is_binary &&
1177                     !data->files[i]->is_unmerged) {
1178                         int added = data->files[i]->added;
1179                         int deleted= data->files[i]->deleted;
1180                         if (!data->files[i]->is_renamed &&
1181                             (added + deleted == 0)) {
1182                                 total_files--;
1183                         } else {
1184                                 adds += added;
1185                                 dels += deleted;
1186                         }
1187                 }
1188         }
1189         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1190                total_files, adds, dels);
1191 }
1192
1193 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1194 {
1195         int i;
1196
1197         if (data->nr == 0)
1198                 return;
1199
1200         for (i = 0; i < data->nr; i++) {
1201                 struct diffstat_file *file = data->files[i];
1202
1203                 if (file->is_binary)
1204                         fprintf(options->file, "-\t-\t");
1205                 else
1206                         fprintf(options->file,
1207                                 "%d\t%d\t", file->added, file->deleted);
1208                 if (options->line_termination) {
1209                         fill_print_name(file);
1210                         if (!file->is_renamed)
1211                                 write_name_quoted(file->name, options->file,
1212                                                   options->line_termination);
1213                         else {
1214                                 fputs(file->print_name, options->file);
1215                                 putc(options->line_termination, options->file);
1216                         }
1217                 } else {
1218                         if (file->is_renamed) {
1219                                 putc('\0', options->file);
1220                                 write_name_quoted(file->from_name, options->file, '\0');
1221                         }
1222                         write_name_quoted(file->name, options->file, '\0');
1223                 }
1224         }
1225 }
1226
1227 struct dirstat_file {
1228         const char *name;
1229         unsigned long changed;
1230 };
1231
1232 struct dirstat_dir {
1233         struct dirstat_file *files;
1234         int alloc, nr, percent, cumulative;
1235 };
1236
1237 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1238 {
1239         unsigned long this_dir = 0;
1240         unsigned int sources = 0;
1241
1242         while (dir->nr) {
1243                 struct dirstat_file *f = dir->files;
1244                 int namelen = strlen(f->name);
1245                 unsigned long this;
1246                 char *slash;
1247
1248                 if (namelen < baselen)
1249                         break;
1250                 if (memcmp(f->name, base, baselen))
1251                         break;
1252                 slash = strchr(f->name + baselen, '/');
1253                 if (slash) {
1254                         int newbaselen = slash + 1 - f->name;
1255                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1256                         sources++;
1257                 } else {
1258                         this = f->changed;
1259                         dir->files++;
1260                         dir->nr--;
1261                         sources += 2;
1262                 }
1263                 this_dir += this;
1264         }
1265
1266         /*
1267          * We don't report dirstat's for
1268          *  - the top level
1269          *  - or cases where everything came from a single directory
1270          *    under this directory (sources == 1).
1271          */
1272         if (baselen && sources != 1) {
1273                 int permille = this_dir * 1000 / changed;
1274                 if (permille) {
1275                         int percent = permille / 10;
1276                         if (percent >= dir->percent) {
1277                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1278                                 if (!dir->cumulative)
1279                                         return 0;
1280                         }
1281                 }
1282         }
1283         return this_dir;
1284 }
1285
1286 static int dirstat_compare(const void *_a, const void *_b)
1287 {
1288         const struct dirstat_file *a = _a;
1289         const struct dirstat_file *b = _b;
1290         return strcmp(a->name, b->name);
1291 }
1292
1293 static void show_dirstat(struct diff_options *options)
1294 {
1295         int i;
1296         unsigned long changed;
1297         struct dirstat_dir dir;
1298         struct diff_queue_struct *q = &diff_queued_diff;
1299
1300         dir.files = NULL;
1301         dir.alloc = 0;
1302         dir.nr = 0;
1303         dir.percent = options->dirstat_percent;
1304         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1305
1306         changed = 0;
1307         for (i = 0; i < q->nr; i++) {
1308                 struct diff_filepair *p = q->queue[i];
1309                 const char *name;
1310                 unsigned long copied, added, damage;
1311
1312                 name = p->one->path ? p->one->path : p->two->path;
1313
1314                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1315                         diff_populate_filespec(p->one, 0);
1316                         diff_populate_filespec(p->two, 0);
1317                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1318                                                &copied, &added);
1319                         diff_free_filespec_data(p->one);
1320                         diff_free_filespec_data(p->two);
1321                 } else if (DIFF_FILE_VALID(p->one)) {
1322                         diff_populate_filespec(p->one, 1);
1323                         copied = added = 0;
1324                         diff_free_filespec_data(p->one);
1325                 } else if (DIFF_FILE_VALID(p->two)) {
1326                         diff_populate_filespec(p->two, 1);
1327                         copied = 0;
1328                         added = p->two->size;
1329                         diff_free_filespec_data(p->two);
1330                 } else
1331                         continue;
1332
1333                 /*
1334                  * Original minus copied is the removed material,
1335                  * added is the new material.  They are both damages
1336                  * made to the preimage. In --dirstat-by-file mode, count
1337                  * damaged files, not damaged lines. This is done by
1338                  * counting only a single damaged line per file.
1339                  */
1340                 damage = (p->one->size - copied) + added;
1341                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1342                         damage = 1;
1343
1344                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1345                 dir.files[dir.nr].name = name;
1346                 dir.files[dir.nr].changed = damage;
1347                 changed += damage;
1348                 dir.nr++;
1349         }
1350
1351         /* This can happen even with many files, if everything was renames */
1352         if (!changed)
1353                 return;
1354
1355         /* Show all directories with more than x% of the changes */
1356         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1357         gather_dirstat(options->file, &dir, changed, "", 0);
1358 }
1359
1360 static void free_diffstat_info(struct diffstat_t *diffstat)
1361 {
1362         int i;
1363         for (i = 0; i < diffstat->nr; i++) {
1364                 struct diffstat_file *f = diffstat->files[i];
1365                 if (f->name != f->print_name)
1366                         free(f->print_name);
1367                 free(f->name);
1368                 free(f->from_name);
1369                 free(f);
1370         }
1371         free(diffstat->files);
1372 }
1373
1374 struct checkdiff_t {
1375         const char *filename;
1376         int lineno;
1377         struct diff_options *o;
1378         unsigned ws_rule;
1379         unsigned status;
1380 };
1381
1382 static int is_conflict_marker(const char *line, unsigned long len)
1383 {
1384         char firstchar;
1385         int cnt;
1386
1387         if (len < 8)
1388                 return 0;
1389         firstchar = line[0];
1390         switch (firstchar) {
1391         case '=': case '>': case '<':
1392                 break;
1393         default:
1394                 return 0;
1395         }
1396         for (cnt = 1; cnt < 7; cnt++)
1397                 if (line[cnt] != firstchar)
1398                         return 0;
1399         /* line[0] thru line[6] are same as firstchar */
1400         if (firstchar == '=') {
1401                 /* divider between ours and theirs? */
1402                 if (len != 8 || line[7] != '\n')
1403                         return 0;
1404         } else if (len < 8 || !isspace(line[7])) {
1405                 /* not divider before ours nor after theirs */
1406                 return 0;
1407         }
1408         return 1;
1409 }
1410
1411 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1412 {
1413         struct checkdiff_t *data = priv;
1414         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1415         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1416         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1417         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1418         char *err;
1419
1420         if (line[0] == '+') {
1421                 unsigned bad;
1422                 data->lineno++;
1423                 if (is_conflict_marker(line + 1, len - 1)) {
1424                         data->status |= 1;
1425                         fprintf(data->o->file,
1426                                 "%s:%d: leftover conflict marker\n",
1427                                 data->filename, data->lineno);
1428                 }
1429                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1430                 if (!bad)
1431                         return;
1432                 data->status |= bad;
1433                 err = whitespace_error_string(bad);
1434                 fprintf(data->o->file, "%s:%d: %s.\n",
1435                         data->filename, data->lineno, err);
1436                 free(err);
1437                 emit_line(data->o->file, set, reset, line, 1);
1438                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1439                               data->o->file, set, reset, ws);
1440         } else if (line[0] == ' ') {
1441                 data->lineno++;
1442         } else if (line[0] == '@') {
1443                 char *plus = strchr(line, '+');
1444                 if (plus)
1445                         data->lineno = strtol(plus, NULL, 10) - 1;
1446                 else
1447                         die("invalid diff");
1448         }
1449 }
1450
1451 static unsigned char *deflate_it(char *data,
1452                                  unsigned long size,
1453                                  unsigned long *result_size)
1454 {
1455         int bound;
1456         unsigned char *deflated;
1457         z_stream stream;
1458
1459         memset(&stream, 0, sizeof(stream));
1460         deflateInit(&stream, zlib_compression_level);
1461         bound = deflateBound(&stream, size);
1462         deflated = xmalloc(bound);
1463         stream.next_out = deflated;
1464         stream.avail_out = bound;
1465
1466         stream.next_in = (unsigned char *)data;
1467         stream.avail_in = size;
1468         while (deflate(&stream, Z_FINISH) == Z_OK)
1469                 ; /* nothing */
1470         deflateEnd(&stream);
1471         *result_size = stream.total_out;
1472         return deflated;
1473 }
1474
1475 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1476 {
1477         void *cp;
1478         void *delta;
1479         void *deflated;
1480         void *data;
1481         unsigned long orig_size;
1482         unsigned long delta_size;
1483         unsigned long deflate_size;
1484         unsigned long data_size;
1485
1486         /* We could do deflated delta, or we could do just deflated two,
1487          * whichever is smaller.
1488          */
1489         delta = NULL;
1490         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1491         if (one->size && two->size) {
1492                 delta = diff_delta(one->ptr, one->size,
1493                                    two->ptr, two->size,
1494                                    &delta_size, deflate_size);
1495                 if (delta) {
1496                         void *to_free = delta;
1497                         orig_size = delta_size;
1498                         delta = deflate_it(delta, delta_size, &delta_size);
1499                         free(to_free);
1500                 }
1501         }
1502
1503         if (delta && delta_size < deflate_size) {
1504                 fprintf(file, "delta %lu\n", orig_size);
1505                 free(deflated);
1506                 data = delta;
1507                 data_size = delta_size;
1508         }
1509         else {
1510                 fprintf(file, "literal %lu\n", two->size);
1511                 free(delta);
1512                 data = deflated;
1513                 data_size = deflate_size;
1514         }
1515
1516         /* emit data encoded in base85 */
1517         cp = data;
1518         while (data_size) {
1519                 int bytes = (52 < data_size) ? 52 : data_size;
1520                 char line[70];
1521                 data_size -= bytes;
1522                 if (bytes <= 26)
1523                         line[0] = bytes + 'A' - 1;
1524                 else
1525                         line[0] = bytes - 26 + 'a' - 1;
1526                 encode_85(line + 1, cp, bytes);
1527                 cp = (char *) cp + bytes;
1528                 fputs(line, file);
1529                 fputc('\n', file);
1530         }
1531         fprintf(file, "\n");
1532         free(data);
1533 }
1534
1535 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1536 {
1537         fprintf(file, "GIT binary patch\n");
1538         emit_binary_diff_body(file, one, two);
1539         emit_binary_diff_body(file, two, one);
1540 }
1541
1542 static void diff_filespec_load_driver(struct diff_filespec *one)
1543 {
1544         if (!one->driver)
1545                 one->driver = userdiff_find_by_path(one->path);
1546         if (!one->driver)
1547                 one->driver = userdiff_find_by_name("default");
1548 }
1549
1550 int diff_filespec_is_binary(struct diff_filespec *one)
1551 {
1552         if (one->is_binary == -1) {
1553                 diff_filespec_load_driver(one);
1554                 if (one->driver->binary != -1)
1555                         one->is_binary = one->driver->binary;
1556                 else {
1557                         if (!one->data && DIFF_FILE_VALID(one))
1558                                 diff_populate_filespec(one, 0);
1559                         if (one->data)
1560                                 one->is_binary = buffer_is_binary(one->data,
1561                                                 one->size);
1562                         if (one->is_binary == -1)
1563                                 one->is_binary = 0;
1564                 }
1565         }
1566         return one->is_binary;
1567 }
1568
1569 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1570 {
1571         diff_filespec_load_driver(one);
1572         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1573 }
1574
1575 static const char *userdiff_word_regex(struct diff_filespec *one)
1576 {
1577         diff_filespec_load_driver(one);
1578         return one->driver->word_regex;
1579 }
1580
1581 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1582 {
1583         if (!options->a_prefix)
1584                 options->a_prefix = a;
1585         if (!options->b_prefix)
1586                 options->b_prefix = b;
1587 }
1588
1589 static const char *get_textconv(struct diff_filespec *one)
1590 {
1591         if (!DIFF_FILE_VALID(one))
1592                 return NULL;
1593         if (!S_ISREG(one->mode))
1594                 return NULL;
1595         diff_filespec_load_driver(one);
1596         return one->driver->textconv;
1597 }
1598
1599 static void builtin_diff(const char *name_a,
1600                          const char *name_b,
1601                          struct diff_filespec *one,
1602                          struct diff_filespec *two,
1603                          const char *xfrm_msg,
1604                          struct diff_options *o,
1605                          int complete_rewrite)
1606 {
1607         mmfile_t mf1, mf2;
1608         const char *lbl[2];
1609         char *a_one, *b_two;
1610         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1611         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1612         const char *a_prefix, *b_prefix;
1613         const char *textconv_one = NULL, *textconv_two = NULL;
1614         struct strbuf header = STRBUF_INIT;
1615
1616         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1617                         (!one->mode || S_ISGITLINK(one->mode)) &&
1618                         (!two->mode || S_ISGITLINK(two->mode))) {
1619                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1620                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1621                 show_submodule_summary(o->file, one ? one->path : two->path,
1622                                 one->sha1, two->sha1, two->dirty_submodule,
1623                                 del, add, reset);
1624                 return;
1625         }
1626
1627         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1628                 textconv_one = get_textconv(one);
1629                 textconv_two = get_textconv(two);
1630         }
1631
1632         diff_set_mnemonic_prefix(o, "a/", "b/");
1633         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1634                 a_prefix = o->b_prefix;
1635                 b_prefix = o->a_prefix;
1636         } else {
1637                 a_prefix = o->a_prefix;
1638                 b_prefix = o->b_prefix;
1639         }
1640
1641         /* Never use a non-valid filename anywhere if at all possible */
1642         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1643         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1644
1645         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1646         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1647         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1648         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1649         strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1650         if (lbl[0][0] == '/') {
1651                 /* /dev/null */
1652                 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1653                 if (xfrm_msg && xfrm_msg[0])
1654                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1655         }
1656         else if (lbl[1][0] == '/') {
1657                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1658                 if (xfrm_msg && xfrm_msg[0])
1659                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1660         }
1661         else {
1662                 if (one->mode != two->mode) {
1663                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1664                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1665                 }
1666                 if (xfrm_msg && xfrm_msg[0])
1667                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1668
1669                 /*
1670                  * we do not run diff between different kind
1671                  * of objects.
1672                  */
1673                 if ((one->mode ^ two->mode) & S_IFMT)
1674                         goto free_ab_and_return;
1675                 if (complete_rewrite &&
1676                     (textconv_one || !diff_filespec_is_binary(one)) &&
1677                     (textconv_two || !diff_filespec_is_binary(two))) {
1678                         fprintf(o->file, "%s", header.buf);
1679                         strbuf_reset(&header);
1680                         emit_rewrite_diff(name_a, name_b, one, two,
1681                                                 textconv_one, textconv_two, o);
1682                         o->found_changes = 1;
1683                         goto free_ab_and_return;
1684                 }
1685         }
1686
1687         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1688                 die("unable to read files to diff");
1689
1690         if (!DIFF_OPT_TST(o, TEXT) &&
1691             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1692               (diff_filespec_is_binary(two) && !textconv_two) )) {
1693                 /* Quite common confusing case */
1694                 if (mf1.size == mf2.size &&
1695                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1696                         goto free_ab_and_return;
1697                 fprintf(o->file, "%s", header.buf);
1698                 strbuf_reset(&header);
1699                 if (DIFF_OPT_TST(o, BINARY))
1700                         emit_binary_diff(o->file, &mf1, &mf2);
1701                 else
1702                         fprintf(o->file, "Binary files %s and %s differ\n",
1703                                 lbl[0], lbl[1]);
1704                 o->found_changes = 1;
1705         }
1706         else {
1707                 /* Crazy xdl interfaces.. */
1708                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1709                 xpparam_t xpp;
1710                 xdemitconf_t xecfg;
1711                 xdemitcb_t ecb;
1712                 struct emit_callback ecbdata;
1713                 const struct userdiff_funcname *pe;
1714
1715                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1716                         fprintf(o->file, "%s", header.buf);
1717                         strbuf_reset(&header);
1718                 }
1719
1720                 if (textconv_one) {
1721                         size_t size;
1722                         mf1.ptr = run_textconv(textconv_one, one, &size);
1723                         if (!mf1.ptr)
1724                                 die("unable to read files to diff");
1725                         mf1.size = size;
1726                 }
1727                 if (textconv_two) {
1728                         size_t size;
1729                         mf2.ptr = run_textconv(textconv_two, two, &size);
1730                         if (!mf2.ptr)
1731                                 die("unable to read files to diff");
1732                         mf2.size = size;
1733                 }
1734
1735                 pe = diff_funcname_pattern(one);
1736                 if (!pe)
1737                         pe = diff_funcname_pattern(two);
1738
1739                 memset(&xpp, 0, sizeof(xpp));
1740                 memset(&xecfg, 0, sizeof(xecfg));
1741                 memset(&ecbdata, 0, sizeof(ecbdata));
1742                 ecbdata.label_path = lbl;
1743                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1744                 ecbdata.found_changesp = &o->found_changes;
1745                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1746                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1747                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1748                 ecbdata.file = o->file;
1749                 ecbdata.header = header.len ? &header : NULL;
1750                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1751                 xecfg.ctxlen = o->context;
1752                 xecfg.interhunkctxlen = o->interhunkcontext;
1753                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1754                 if (pe)
1755                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1756                 if (!diffopts)
1757                         ;
1758                 else if (!prefixcmp(diffopts, "--unified="))
1759                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1760                 else if (!prefixcmp(diffopts, "-u"))
1761                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1762                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1763                         ecbdata.diff_words =
1764                                 xcalloc(1, sizeof(struct diff_words_data));
1765                         ecbdata.diff_words->file = o->file;
1766                         if (!o->word_regex)
1767                                 o->word_regex = userdiff_word_regex(one);
1768                         if (!o->word_regex)
1769                                 o->word_regex = userdiff_word_regex(two);
1770                         if (!o->word_regex)
1771                                 o->word_regex = diff_word_regex_cfg;
1772                         if (o->word_regex) {
1773                                 ecbdata.diff_words->word_regex = (regex_t *)
1774                                         xmalloc(sizeof(regex_t));
1775                                 if (regcomp(ecbdata.diff_words->word_regex,
1776                                                 o->word_regex,
1777                                                 REG_EXTENDED | REG_NEWLINE))
1778                                         die ("Invalid regular expression: %s",
1779                                                         o->word_regex);
1780                         }
1781                 }
1782                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1783                               &xpp, &xecfg, &ecb);
1784                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1785                         free_diff_words_data(&ecbdata);
1786                 if (textconv_one)
1787                         free(mf1.ptr);
1788                 if (textconv_two)
1789                         free(mf2.ptr);
1790                 xdiff_clear_find_func(&xecfg);
1791         }
1792
1793  free_ab_and_return:
1794         strbuf_release(&header);
1795         diff_free_filespec_data(one);
1796         diff_free_filespec_data(two);
1797         free(a_one);
1798         free(b_two);
1799         return;
1800 }
1801
1802 static void builtin_diffstat(const char *name_a, const char *name_b,
1803                              struct diff_filespec *one,
1804                              struct diff_filespec *two,
1805                              struct diffstat_t *diffstat,
1806                              struct diff_options *o,
1807                              int complete_rewrite)
1808 {
1809         mmfile_t mf1, mf2;
1810         struct diffstat_file *data;
1811
1812         data = diffstat_add(diffstat, name_a, name_b);
1813
1814         if (!one || !two) {
1815                 data->is_unmerged = 1;
1816                 return;
1817         }
1818         if (complete_rewrite) {
1819                 diff_populate_filespec(one, 0);
1820                 diff_populate_filespec(two, 0);
1821                 data->deleted = count_lines(one->data, one->size);
1822                 data->added = count_lines(two->data, two->size);
1823                 goto free_and_return;
1824         }
1825         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1826                 die("unable to read files to diff");
1827
1828         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1829                 data->is_binary = 1;
1830                 data->added = mf2.size;
1831                 data->deleted = mf1.size;
1832         } else {
1833                 /* Crazy xdl interfaces.. */
1834                 xpparam_t xpp;
1835                 xdemitconf_t xecfg;
1836                 xdemitcb_t ecb;
1837
1838                 memset(&xpp, 0, sizeof(xpp));
1839                 memset(&xecfg, 0, sizeof(xecfg));
1840                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1841                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1842                               &xpp, &xecfg, &ecb);
1843         }
1844
1845  free_and_return:
1846         diff_free_filespec_data(one);
1847         diff_free_filespec_data(two);
1848 }
1849
1850 static void builtin_checkdiff(const char *name_a, const char *name_b,
1851                               const char *attr_path,
1852                               struct diff_filespec *one,
1853                               struct diff_filespec *two,
1854                               struct diff_options *o)
1855 {
1856         mmfile_t mf1, mf2;
1857         struct checkdiff_t data;
1858
1859         if (!two)
1860                 return;
1861
1862         memset(&data, 0, sizeof(data));
1863         data.filename = name_b ? name_b : name_a;
1864         data.lineno = 0;
1865         data.o = o;
1866         data.ws_rule = whitespace_rule(attr_path);
1867
1868         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1869                 die("unable to read files to diff");
1870
1871         /*
1872          * All the other codepaths check both sides, but not checking
1873          * the "old" side here is deliberate.  We are checking the newly
1874          * introduced changes, and as long as the "new" side is text, we
1875          * can and should check what it introduces.
1876          */
1877         if (diff_filespec_is_binary(two))
1878                 goto free_and_return;
1879         else {
1880                 /* Crazy xdl interfaces.. */
1881                 xpparam_t xpp;
1882                 xdemitconf_t xecfg;
1883                 xdemitcb_t ecb;
1884
1885                 memset(&xpp, 0, sizeof(xpp));
1886                 memset(&xecfg, 0, sizeof(xecfg));
1887                 xecfg.ctxlen = 1; /* at least one context line */
1888                 xpp.flags = XDF_NEED_MINIMAL;
1889                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1890                               &xpp, &xecfg, &ecb);
1891
1892                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1893                         struct emit_callback ecbdata;
1894                         int blank_at_eof;
1895
1896                         ecbdata.ws_rule = data.ws_rule;
1897                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1898                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1899
1900                         if (blank_at_eof) {
1901                                 static char *err;
1902                                 if (!err)
1903                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1904                                 fprintf(o->file, "%s:%d: %s.\n",
1905                                         data.filename, blank_at_eof, err);
1906                                 data.status = 1; /* report errors */
1907                         }
1908                 }
1909         }
1910  free_and_return:
1911         diff_free_filespec_data(one);
1912         diff_free_filespec_data(two);
1913         if (data.status)
1914                 DIFF_OPT_SET(o, CHECK_FAILED);
1915 }
1916
1917 struct diff_filespec *alloc_filespec(const char *path)
1918 {
1919         int namelen = strlen(path);
1920         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1921
1922         memset(spec, 0, sizeof(*spec));
1923         spec->path = (char *)(spec + 1);
1924         memcpy(spec->path, path, namelen+1);
1925         spec->count = 1;
1926         spec->is_binary = -1;
1927         return spec;
1928 }
1929
1930 void free_filespec(struct diff_filespec *spec)
1931 {
1932         if (!--spec->count) {
1933                 diff_free_filespec_data(spec);
1934                 free(spec);
1935         }
1936 }
1937
1938 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1939                    unsigned short mode)
1940 {
1941         if (mode) {
1942                 spec->mode = canon_mode(mode);
1943                 hashcpy(spec->sha1, sha1);
1944                 spec->sha1_valid = !is_null_sha1(sha1);
1945         }
1946 }
1947
1948 /*
1949  * Given a name and sha1 pair, if the index tells us the file in
1950  * the work tree has that object contents, return true, so that
1951  * prepare_temp_file() does not have to inflate and extract.
1952  */
1953 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1954 {
1955         struct cache_entry *ce;
1956         struct stat st;
1957         int pos, len;
1958
1959         /*
1960          * We do not read the cache ourselves here, because the
1961          * benchmark with my previous version that always reads cache
1962          * shows that it makes things worse for diff-tree comparing
1963          * two linux-2.6 kernel trees in an already checked out work
1964          * tree.  This is because most diff-tree comparisons deal with
1965          * only a small number of files, while reading the cache is
1966          * expensive for a large project, and its cost outweighs the
1967          * savings we get by not inflating the object to a temporary
1968          * file.  Practically, this code only helps when we are used
1969          * by diff-cache --cached, which does read the cache before
1970          * calling us.
1971          */
1972         if (!active_cache)
1973                 return 0;
1974
1975         /* We want to avoid the working directory if our caller
1976          * doesn't need the data in a normal file, this system
1977          * is rather slow with its stat/open/mmap/close syscalls,
1978          * and the object is contained in a pack file.  The pack
1979          * is probably already open and will be faster to obtain
1980          * the data through than the working directory.  Loose
1981          * objects however would tend to be slower as they need
1982          * to be individually opened and inflated.
1983          */
1984         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1985                 return 0;
1986
1987         len = strlen(name);
1988         pos = cache_name_pos(name, len);
1989         if (pos < 0)
1990                 return 0;
1991         ce = active_cache[pos];
1992
1993         /*
1994          * This is not the sha1 we are looking for, or
1995          * unreusable because it is not a regular file.
1996          */
1997         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1998                 return 0;
1999
2000         /*
2001          * If ce is marked as "assume unchanged", there is no
2002          * guarantee that work tree matches what we are looking for.
2003          */
2004         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2005                 return 0;
2006
2007         /*
2008          * If ce matches the file in the work tree, we can reuse it.
2009          */
2010         if (ce_uptodate(ce) ||
2011             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2012                 return 1;
2013
2014         return 0;
2015 }
2016
2017 static int populate_from_stdin(struct diff_filespec *s)
2018 {
2019         struct strbuf buf = STRBUF_INIT;
2020         size_t size = 0;
2021
2022         if (strbuf_read(&buf, 0, 0) < 0)
2023                 return error("error while reading from stdin %s",
2024                                      strerror(errno));
2025
2026         s->should_munmap = 0;
2027         s->data = strbuf_detach(&buf, &size);
2028         s->size = size;
2029         s->should_free = 1;
2030         return 0;
2031 }
2032
2033 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2034 {
2035         int len;
2036         char *data = xmalloc(100), *dirty = "";
2037
2038         /* Are we looking at the work tree? */
2039         if (s->dirty_submodule)
2040                 dirty = "-dirty";
2041
2042         len = snprintf(data, 100,
2043                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2044         s->data = data;
2045         s->size = len;
2046         s->should_free = 1;
2047         if (size_only) {
2048                 s->data = NULL;
2049                 free(data);
2050         }
2051         return 0;
2052 }
2053
2054 /*
2055  * While doing rename detection and pickaxe operation, we may need to
2056  * grab the data for the blob (or file) for our own in-core comparison.
2057  * diff_filespec has data and size fields for this purpose.
2058  */
2059 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2060 {
2061         int err = 0;
2062         if (!DIFF_FILE_VALID(s))
2063                 die("internal error: asking to populate invalid file.");
2064         if (S_ISDIR(s->mode))
2065                 return -1;
2066
2067         if (s->data)
2068                 return 0;
2069
2070         if (size_only && 0 < s->size)
2071                 return 0;
2072
2073         if (S_ISGITLINK(s->mode))
2074                 return diff_populate_gitlink(s, size_only);
2075
2076         if (!s->sha1_valid ||
2077             reuse_worktree_file(s->path, s->sha1, 0)) {
2078                 struct strbuf buf = STRBUF_INIT;
2079                 struct stat st;
2080                 int fd;
2081
2082                 if (!strcmp(s->path, "-"))
2083                         return populate_from_stdin(s);
2084
2085                 if (lstat(s->path, &st) < 0) {
2086                         if (errno == ENOENT) {
2087                         err_empty:
2088                                 err = -1;
2089                         empty:
2090                                 s->data = (char *)"";
2091                                 s->size = 0;
2092                                 return err;
2093                         }
2094                 }
2095                 s->size = xsize_t(st.st_size);
2096                 if (!s->size)
2097                         goto empty;
2098                 if (S_ISLNK(st.st_mode)) {
2099                         struct strbuf sb = STRBUF_INIT;
2100
2101                         if (strbuf_readlink(&sb, s->path, s->size))
2102                                 goto err_empty;
2103                         s->size = sb.len;
2104                         s->data = strbuf_detach(&sb, NULL);
2105                         s->should_free = 1;
2106                         return 0;
2107                 }
2108                 if (size_only)
2109                         return 0;
2110                 fd = open(s->path, O_RDONLY);
2111                 if (fd < 0)
2112                         goto err_empty;
2113                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2114                 close(fd);
2115                 s->should_munmap = 1;
2116
2117                 /*
2118                  * Convert from working tree format to canonical git format
2119                  */
2120                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2121                         size_t size = 0;
2122                         munmap(s->data, s->size);
2123                         s->should_munmap = 0;
2124                         s->data = strbuf_detach(&buf, &size);
2125                         s->size = size;
2126                         s->should_free = 1;
2127                 }
2128         }
2129         else {
2130                 enum object_type type;
2131                 if (size_only)
2132                         type = sha1_object_info(s->sha1, &s->size);
2133                 else {
2134                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2135                         s->should_free = 1;
2136                 }
2137         }
2138         return 0;
2139 }
2140
2141 void diff_free_filespec_blob(struct diff_filespec *s)
2142 {
2143         if (s->should_free)
2144                 free(s->data);
2145         else if (s->should_munmap)
2146                 munmap(s->data, s->size);
2147
2148         if (s->should_free || s->should_munmap) {
2149                 s->should_free = s->should_munmap = 0;
2150                 s->data = NULL;
2151         }
2152 }
2153
2154 void diff_free_filespec_data(struct diff_filespec *s)
2155 {
2156         diff_free_filespec_blob(s);
2157         free(s->cnt_data);
2158         s->cnt_data = NULL;
2159 }
2160
2161 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2162                            void *blob,
2163                            unsigned long size,
2164                            const unsigned char *sha1,
2165                            int mode)
2166 {
2167         int fd;
2168         struct strbuf buf = STRBUF_INIT;
2169         struct strbuf template = STRBUF_INIT;
2170         char *path_dup = xstrdup(path);
2171         const char *base = basename(path_dup);
2172
2173         /* Generate "XXXXXX_basename.ext" */
2174         strbuf_addstr(&template, "XXXXXX_");
2175         strbuf_addstr(&template, base);
2176
2177         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2178                         strlen(base) + 1);
2179         if (fd < 0)
2180                 die_errno("unable to create temp-file");
2181         if (convert_to_working_tree(path,
2182                         (const char *)blob, (size_t)size, &buf)) {
2183                 blob = buf.buf;
2184                 size = buf.len;
2185         }
2186         if (write_in_full(fd, blob, size) != size)
2187                 die_errno("unable to write temp-file");
2188         close(fd);
2189         temp->name = temp->tmp_path;
2190         strcpy(temp->hex, sha1_to_hex(sha1));
2191         temp->hex[40] = 0;
2192         sprintf(temp->mode, "%06o", mode);
2193         strbuf_release(&buf);
2194         strbuf_release(&template);
2195         free(path_dup);
2196 }
2197
2198 static struct diff_tempfile *prepare_temp_file(const char *name,
2199                 struct diff_filespec *one)
2200 {
2201         struct diff_tempfile *temp = claim_diff_tempfile();
2202
2203         if (!DIFF_FILE_VALID(one)) {
2204         not_a_valid_file:
2205                 /* A '-' entry produces this for file-2, and
2206                  * a '+' entry produces this for file-1.
2207                  */
2208                 temp->name = "/dev/null";
2209                 strcpy(temp->hex, ".");
2210                 strcpy(temp->mode, ".");
2211                 return temp;
2212         }
2213
2214         if (!remove_tempfile_installed) {
2215                 atexit(remove_tempfile);
2216                 sigchain_push_common(remove_tempfile_on_signal);
2217                 remove_tempfile_installed = 1;
2218         }
2219
2220         if (!one->sha1_valid ||
2221             reuse_worktree_file(name, one->sha1, 1)) {
2222                 struct stat st;
2223                 if (lstat(name, &st) < 0) {
2224                         if (errno == ENOENT)
2225                                 goto not_a_valid_file;
2226                         die_errno("stat(%s)", name);
2227                 }
2228                 if (S_ISLNK(st.st_mode)) {
2229                         struct strbuf sb = STRBUF_INIT;
2230                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2231                                 die_errno("readlink(%s)", name);
2232                         prep_temp_blob(name, temp, sb.buf, sb.len,
2233                                        (one->sha1_valid ?
2234                                         one->sha1 : null_sha1),
2235                                        (one->sha1_valid ?
2236                                         one->mode : S_IFLNK));
2237                         strbuf_release(&sb);
2238                 }
2239                 else {
2240                         /* we can borrow from the file in the work tree */
2241                         temp->name = name;
2242                         if (!one->sha1_valid)
2243                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2244                         else
2245                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2246                         /* Even though we may sometimes borrow the
2247                          * contents from the work tree, we always want
2248                          * one->mode.  mode is trustworthy even when
2249                          * !(one->sha1_valid), as long as
2250                          * DIFF_FILE_VALID(one).
2251                          */
2252                         sprintf(temp->mode, "%06o", one->mode);
2253                 }
2254                 return temp;
2255         }
2256         else {
2257                 if (diff_populate_filespec(one, 0))
2258                         die("cannot read data blob for %s", one->path);
2259                 prep_temp_blob(name, temp, one->data, one->size,
2260                                one->sha1, one->mode);
2261         }
2262         return temp;
2263 }
2264
2265 /* An external diff command takes:
2266  *
2267  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2268  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2269  *
2270  */
2271 static void run_external_diff(const char *pgm,
2272                               const char *name,
2273                               const char *other,
2274                               struct diff_filespec *one,
2275                               struct diff_filespec *two,
2276                               const char *xfrm_msg,
2277                               int complete_rewrite)
2278 {
2279         const char *spawn_arg[10];
2280         int retval;
2281         const char **arg = &spawn_arg[0];
2282
2283         if (one && two) {
2284                 struct diff_tempfile *temp_one, *temp_two;
2285                 const char *othername = (other ? other : name);
2286                 temp_one = prepare_temp_file(name, one);
2287                 temp_two = prepare_temp_file(othername, two);
2288                 *arg++ = pgm;
2289                 *arg++ = name;
2290                 *arg++ = temp_one->name;
2291                 *arg++ = temp_one->hex;
2292                 *arg++ = temp_one->mode;
2293                 *arg++ = temp_two->name;
2294                 *arg++ = temp_two->hex;
2295                 *arg++ = temp_two->mode;
2296                 if (other) {
2297                         *arg++ = other;
2298                         *arg++ = xfrm_msg;
2299                 }
2300         } else {
2301                 *arg++ = pgm;
2302                 *arg++ = name;
2303         }
2304         *arg = NULL;
2305         fflush(NULL);
2306         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2307         remove_tempfile();
2308         if (retval) {
2309                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2310                 exit(1);
2311         }
2312 }
2313
2314 static int similarity_index(struct diff_filepair *p)
2315 {
2316         return p->score * 100 / MAX_SCORE;
2317 }
2318
2319 static void fill_metainfo(struct strbuf *msg,
2320                           const char *name,
2321                           const char *other,
2322                           struct diff_filespec *one,
2323                           struct diff_filespec *two,
2324                           struct diff_options *o,
2325                           struct diff_filepair *p)
2326 {
2327         strbuf_init(msg, PATH_MAX * 2 + 300);
2328         switch (p->status) {
2329         case DIFF_STATUS_COPIED:
2330                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2331                 strbuf_addstr(msg, "\ncopy from ");
2332                 quote_c_style(name, msg, NULL, 0);
2333                 strbuf_addstr(msg, "\ncopy to ");
2334                 quote_c_style(other, msg, NULL, 0);
2335                 strbuf_addch(msg, '\n');
2336                 break;
2337         case DIFF_STATUS_RENAMED:
2338                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2339                 strbuf_addstr(msg, "\nrename from ");
2340                 quote_c_style(name, msg, NULL, 0);
2341                 strbuf_addstr(msg, "\nrename to ");
2342                 quote_c_style(other, msg, NULL, 0);
2343                 strbuf_addch(msg, '\n');
2344                 break;
2345         case DIFF_STATUS_MODIFIED:
2346                 if (p->score) {
2347                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2348                                     similarity_index(p));
2349                         break;
2350                 }
2351                 /* fallthru */
2352         default:
2353                 /* nothing */
2354                 ;
2355         }
2356         if (one && two && hashcmp(one->sha1, two->sha1)) {
2357                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2358
2359                 if (DIFF_OPT_TST(o, BINARY)) {
2360                         mmfile_t mf;
2361                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2362                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2363                                 abbrev = 40;
2364                 }
2365                 strbuf_addf(msg, "index %.*s..%.*s",
2366                             abbrev, sha1_to_hex(one->sha1),
2367                             abbrev, sha1_to_hex(two->sha1));
2368                 if (one->mode == two->mode)
2369                         strbuf_addf(msg, " %06o", one->mode);
2370                 strbuf_addch(msg, '\n');
2371         }
2372         if (msg->len)
2373                 strbuf_setlen(msg, msg->len - 1);
2374 }
2375
2376 static void run_diff_cmd(const char *pgm,
2377                          const char *name,
2378                          const char *other,
2379                          const char *attr_path,
2380                          struct diff_filespec *one,
2381                          struct diff_filespec *two,
2382                          struct strbuf *msg,
2383                          struct diff_options *o,
2384                          struct diff_filepair *p)
2385 {
2386         const char *xfrm_msg = NULL;
2387         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2388
2389         if (msg) {
2390                 fill_metainfo(msg, name, other, one, two, o, p);
2391                 xfrm_msg = msg->len ? msg->buf : NULL;
2392         }
2393
2394         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2395                 pgm = NULL;
2396         else {
2397                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2398                 if (drv && drv->external)
2399                         pgm = drv->external;
2400         }
2401
2402         if (pgm) {
2403                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2404                                   complete_rewrite);
2405                 return;
2406         }
2407         if (one && two)
2408                 builtin_diff(name, other ? other : name,
2409                              one, two, xfrm_msg, o, complete_rewrite);
2410         else
2411                 fprintf(o->file, "* Unmerged path %s\n", name);
2412 }
2413
2414 static void diff_fill_sha1_info(struct diff_filespec *one)
2415 {
2416         if (DIFF_FILE_VALID(one)) {
2417                 if (!one->sha1_valid) {
2418                         struct stat st;
2419                         if (!strcmp(one->path, "-")) {
2420                                 hashcpy(one->sha1, null_sha1);
2421                                 return;
2422                         }
2423                         if (lstat(one->path, &st) < 0)
2424                                 die_errno("stat '%s'", one->path);
2425                         if (index_path(one->sha1, one->path, &st, 0))
2426                                 die("cannot hash %s", one->path);
2427                 }
2428         }
2429         else
2430                 hashclr(one->sha1);
2431 }
2432
2433 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2434 {
2435         /* Strip the prefix but do not molest /dev/null and absolute paths */
2436         if (*namep && **namep != '/')
2437                 *namep += prefix_length;
2438         if (*otherp && **otherp != '/')
2439                 *otherp += prefix_length;
2440 }
2441
2442 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2443 {
2444         const char *pgm = external_diff();
2445         struct strbuf msg;
2446         struct diff_filespec *one = p->one;
2447         struct diff_filespec *two = p->two;
2448         const char *name;
2449         const char *other;
2450         const char *attr_path;
2451
2452         name  = p->one->path;
2453         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2454         attr_path = name;
2455         if (o->prefix_length)
2456                 strip_prefix(o->prefix_length, &name, &other);
2457
2458         if (DIFF_PAIR_UNMERGED(p)) {
2459                 run_diff_cmd(pgm, name, NULL, attr_path,
2460                              NULL, NULL, NULL, o, p);
2461                 return;
2462         }
2463
2464         diff_fill_sha1_info(one);
2465         diff_fill_sha1_info(two);
2466
2467         if (!pgm &&
2468             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2469             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2470                 /*
2471                  * a filepair that changes between file and symlink
2472                  * needs to be split into deletion and creation.
2473                  */
2474                 struct diff_filespec *null = alloc_filespec(two->path);
2475                 run_diff_cmd(NULL, name, other, attr_path,
2476                              one, null, &msg, o, p);
2477                 free(null);
2478                 strbuf_release(&msg);
2479
2480                 null = alloc_filespec(one->path);
2481                 run_diff_cmd(NULL, name, other, attr_path,
2482                              null, two, &msg, o, p);
2483                 free(null);
2484         }
2485         else
2486                 run_diff_cmd(pgm, name, other, attr_path,
2487                              one, two, &msg, o, p);
2488
2489         strbuf_release(&msg);
2490 }
2491
2492 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2493                          struct diffstat_t *diffstat)
2494 {
2495         const char *name;
2496         const char *other;
2497         int complete_rewrite = 0;
2498
2499         if (DIFF_PAIR_UNMERGED(p)) {
2500                 /* unmerged */
2501                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2502                 return;
2503         }
2504
2505         name = p->one->path;
2506         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2507
2508         if (o->prefix_length)
2509                 strip_prefix(o->prefix_length, &name, &other);
2510
2511         diff_fill_sha1_info(p->one);
2512         diff_fill_sha1_info(p->two);
2513
2514         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2515                 complete_rewrite = 1;
2516         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2517 }
2518
2519 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2520 {
2521         const char *name;
2522         const char *other;
2523         const char *attr_path;
2524
2525         if (DIFF_PAIR_UNMERGED(p)) {
2526                 /* unmerged */
2527                 return;
2528         }
2529
2530         name = p->one->path;
2531         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2532         attr_path = other ? other : name;
2533
2534         if (o->prefix_length)
2535                 strip_prefix(o->prefix_length, &name, &other);
2536
2537         diff_fill_sha1_info(p->one);
2538         diff_fill_sha1_info(p->two);
2539
2540         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2541 }
2542
2543 void diff_setup(struct diff_options *options)
2544 {
2545         memset(options, 0, sizeof(*options));
2546
2547         options->file = stdout;
2548
2549         options->line_termination = '\n';
2550         options->break_opt = -1;
2551         options->rename_limit = -1;
2552         options->dirstat_percent = 3;
2553         options->context = 3;
2554
2555         options->change = diff_change;
2556         options->add_remove = diff_addremove;
2557         if (diff_use_color_default > 0)
2558                 DIFF_OPT_SET(options, COLOR_DIFF);
2559         options->detect_rename = diff_detect_rename_default;
2560
2561         if (!diff_mnemonic_prefix) {
2562                 options->a_prefix = "a/";
2563                 options->b_prefix = "b/";
2564         }
2565 }
2566
2567 int diff_setup_done(struct diff_options *options)
2568 {
2569         int count = 0;
2570
2571         if (options->output_format & DIFF_FORMAT_NAME)
2572                 count++;
2573         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2574                 count++;
2575         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2576                 count++;
2577         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2578                 count++;
2579         if (count > 1)
2580                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2581
2582         /*
2583          * Most of the time we can say "there are changes"
2584          * only by checking if there are changed paths, but
2585          * --ignore-whitespace* options force us to look
2586          * inside contents.
2587          */
2588
2589         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2590             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2591             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2592                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2593         else
2594                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2595
2596         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2597                 options->detect_rename = DIFF_DETECT_COPY;
2598
2599         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2600                 options->prefix = NULL;
2601         if (options->prefix)
2602                 options->prefix_length = strlen(options->prefix);
2603         else
2604                 options->prefix_length = 0;
2605
2606         if (options->output_format & (DIFF_FORMAT_NAME |
2607                                       DIFF_FORMAT_NAME_STATUS |
2608                                       DIFF_FORMAT_CHECKDIFF |
2609                                       DIFF_FORMAT_NO_OUTPUT))
2610                 options->output_format &= ~(DIFF_FORMAT_RAW |
2611                                             DIFF_FORMAT_NUMSTAT |
2612                                             DIFF_FORMAT_DIFFSTAT |
2613                                             DIFF_FORMAT_SHORTSTAT |
2614                                             DIFF_FORMAT_DIRSTAT |
2615                                             DIFF_FORMAT_SUMMARY |
2616                                             DIFF_FORMAT_PATCH);
2617
2618         /*
2619          * These cases always need recursive; we do not drop caller-supplied
2620          * recursive bits for other formats here.
2621          */
2622         if (options->output_format & (DIFF_FORMAT_PATCH |
2623                                       DIFF_FORMAT_NUMSTAT |
2624                                       DIFF_FORMAT_DIFFSTAT |
2625                                       DIFF_FORMAT_SHORTSTAT |
2626                                       DIFF_FORMAT_DIRSTAT |
2627                                       DIFF_FORMAT_SUMMARY |
2628                                       DIFF_FORMAT_CHECKDIFF))
2629                 DIFF_OPT_SET(options, RECURSIVE);
2630         /*
2631          * Also pickaxe would not work very well if you do not say recursive
2632          */
2633         if (options->pickaxe)
2634                 DIFF_OPT_SET(options, RECURSIVE);
2635         /*
2636          * When patches are generated, submodules diffed against the work tree
2637          * must be checked for dirtiness too so it can be shown in the output
2638          */
2639         if (options->output_format & DIFF_FORMAT_PATCH)
2640                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2641
2642         if (options->detect_rename && options->rename_limit < 0)
2643                 options->rename_limit = diff_rename_limit_default;
2644         if (options->setup & DIFF_SETUP_USE_CACHE) {
2645                 if (!active_cache)
2646                         /* read-cache does not die even when it fails
2647                          * so it is safe for us to do this here.  Also
2648                          * it does not smudge active_cache or active_nr
2649                          * when it fails, so we do not have to worry about
2650                          * cleaning it up ourselves either.
2651                          */
2652                         read_cache();
2653         }
2654         if (options->abbrev <= 0 || 40 < options->abbrev)
2655                 options->abbrev = 40; /* full */
2656
2657         /*
2658          * It does not make sense to show the first hit we happened
2659          * to have found.  It does not make sense not to return with
2660          * exit code in such a case either.
2661          */
2662         if (DIFF_OPT_TST(options, QUICK)) {
2663                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2664                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2665         }
2666
2667         return 0;
2668 }
2669
2670 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2671 {
2672         char c, *eq;
2673         int len;
2674
2675         if (*arg != '-')
2676                 return 0;
2677         c = *++arg;
2678         if (!c)
2679                 return 0;
2680         if (c == arg_short) {
2681                 c = *++arg;
2682                 if (!c)
2683                         return 1;
2684                 if (val && isdigit(c)) {
2685                         char *end;
2686                         int n = strtoul(arg, &end, 10);
2687                         if (*end)
2688                                 return 0;
2689                         *val = n;
2690                         return 1;
2691                 }
2692                 return 0;
2693         }
2694         if (c != '-')
2695                 return 0;
2696         arg++;
2697         eq = strchr(arg, '=');
2698         if (eq)
2699                 len = eq - arg;
2700         else
2701                 len = strlen(arg);
2702         if (!len || strncmp(arg, arg_long, len))
2703                 return 0;
2704         if (eq) {
2705                 int n;
2706                 char *end;
2707                 if (!isdigit(*++eq))
2708                         return 0;
2709                 n = strtoul(eq, &end, 10);
2710                 if (*end)
2711                         return 0;
2712                 *val = n;
2713         }
2714         return 1;
2715 }
2716
2717 static int diff_scoreopt_parse(const char *opt);
2718
2719 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2720 {
2721         const char *arg = av[0];
2722
2723         /* Output format options */
2724         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2725                 options->output_format |= DIFF_FORMAT_PATCH;
2726         else if (opt_arg(arg, 'U', "unified", &options->context))
2727                 options->output_format |= DIFF_FORMAT_PATCH;
2728         else if (!strcmp(arg, "--raw"))
2729                 options->output_format |= DIFF_FORMAT_RAW;
2730         else if (!strcmp(arg, "--patch-with-raw"))
2731                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2732         else if (!strcmp(arg, "--numstat"))
2733                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2734         else if (!strcmp(arg, "--shortstat"))
2735                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2736         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2737                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2738         else if (!strcmp(arg, "--cumulative")) {
2739                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2740                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2741         } else if (opt_arg(arg, 0, "dirstat-by-file",
2742                            &options->dirstat_percent)) {
2743                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2744                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2745         }
2746         else if (!strcmp(arg, "--check"))
2747                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2748         else if (!strcmp(arg, "--summary"))
2749                 options->output_format |= DIFF_FORMAT_SUMMARY;
2750         else if (!strcmp(arg, "--patch-with-stat"))
2751                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2752         else if (!strcmp(arg, "--name-only"))
2753                 options->output_format |= DIFF_FORMAT_NAME;
2754         else if (!strcmp(arg, "--name-status"))
2755                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2756         else if (!strcmp(arg, "-s"))
2757                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2758         else if (!prefixcmp(arg, "--stat")) {
2759                 char *end;
2760                 int width = options->stat_width;
2761                 int name_width = options->stat_name_width;
2762                 arg += 6;
2763                 end = (char *)arg;
2764
2765                 switch (*arg) {
2766                 case '-':
2767                         if (!prefixcmp(arg, "-width="))
2768                                 width = strtoul(arg + 7, &end, 10);
2769                         else if (!prefixcmp(arg, "-name-width="))
2770                                 name_width = strtoul(arg + 12, &end, 10);
2771                         break;
2772                 case '=':
2773                         width = strtoul(arg+1, &end, 10);
2774                         if (*end == ',')
2775                                 name_width = strtoul(end+1, &end, 10);
2776                 }
2777
2778                 /* Important! This checks all the error cases! */
2779                 if (*end)
2780                         return 0;
2781                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2782                 options->stat_name_width = name_width;
2783                 options->stat_width = width;
2784         }
2785
2786         /* renames options */
2787         else if (!prefixcmp(arg, "-B")) {
2788                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2789                         return -1;
2790         }
2791         else if (!prefixcmp(arg, "-M")) {
2792                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2793                         return -1;
2794                 options->detect_rename = DIFF_DETECT_RENAME;
2795         }
2796         else if (!prefixcmp(arg, "-C")) {
2797                 if (options->detect_rename == DIFF_DETECT_COPY)
2798                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2799                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2800                         return -1;
2801                 options->detect_rename = DIFF_DETECT_COPY;
2802         }
2803         else if (!strcmp(arg, "--no-renames"))
2804                 options->detect_rename = 0;
2805         else if (!strcmp(arg, "--relative"))
2806                 DIFF_OPT_SET(options, RELATIVE_NAME);
2807         else if (!prefixcmp(arg, "--relative=")) {
2808                 DIFF_OPT_SET(options, RELATIVE_NAME);
2809                 options->prefix = arg + 11;
2810         }
2811
2812         /* xdiff options */
2813         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2814                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2815         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2816                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2817         else if (!strcmp(arg, "--ignore-space-at-eol"))
2818                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2819         else if (!strcmp(arg, "--patience"))
2820                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2821
2822         /* flags options */
2823         else if (!strcmp(arg, "--binary")) {
2824                 options->output_format |= DIFF_FORMAT_PATCH;
2825                 DIFF_OPT_SET(options, BINARY);
2826         }
2827         else if (!strcmp(arg, "--full-index"))
2828                 DIFF_OPT_SET(options, FULL_INDEX);
2829         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2830                 DIFF_OPT_SET(options, TEXT);
2831         else if (!strcmp(arg, "-R"))
2832                 DIFF_OPT_SET(options, REVERSE_DIFF);
2833         else if (!strcmp(arg, "--find-copies-harder"))
2834                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2835         else if (!strcmp(arg, "--follow"))
2836                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2837         else if (!strcmp(arg, "--color"))
2838                 DIFF_OPT_SET(options, COLOR_DIFF);
2839         else if (!prefixcmp(arg, "--color=")) {
2840                 int value = git_config_colorbool(NULL, arg+8, -1);
2841                 if (value == 0)
2842                         DIFF_OPT_CLR(options, COLOR_DIFF);
2843                 else if (value > 0)
2844                         DIFF_OPT_SET(options, COLOR_DIFF);
2845                 else
2846                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
2847         }
2848         else if (!strcmp(arg, "--no-color"))
2849                 DIFF_OPT_CLR(options, COLOR_DIFF);
2850         else if (!strcmp(arg, "--color-words")) {
2851                 DIFF_OPT_SET(options, COLOR_DIFF);
2852                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2853         }
2854         else if (!prefixcmp(arg, "--color-words=")) {
2855                 DIFF_OPT_SET(options, COLOR_DIFF);
2856                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2857                 options->word_regex = arg + 14;
2858         }
2859         else if (!strcmp(arg, "--exit-code"))
2860                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2861         else if (!strcmp(arg, "--quiet"))
2862                 DIFF_OPT_SET(options, QUICK);
2863         else if (!strcmp(arg, "--ext-diff"))
2864                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2865         else if (!strcmp(arg, "--no-ext-diff"))
2866                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2867         else if (!strcmp(arg, "--textconv"))
2868                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2869         else if (!strcmp(arg, "--no-textconv"))
2870                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2871         else if (!strcmp(arg, "--ignore-submodules"))
2872                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2873         else if (!strcmp(arg, "--submodule"))
2874                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2875         else if (!prefixcmp(arg, "--submodule=")) {
2876                 if (!strcmp(arg + 12, "log"))
2877                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2878         }
2879
2880         /* misc options */
2881         else if (!strcmp(arg, "-z"))
2882                 options->line_termination = 0;
2883         else if (!prefixcmp(arg, "-l"))
2884                 options->rename_limit = strtoul(arg+2, NULL, 10);
2885         else if (!prefixcmp(arg, "-S"))
2886                 options->pickaxe = arg + 2;
2887         else if (!strcmp(arg, "--pickaxe-all"))
2888                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2889         else if (!strcmp(arg, "--pickaxe-regex"))
2890                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2891         else if (!prefixcmp(arg, "-O"))
2892                 options->orderfile = arg + 2;
2893         else if (!prefixcmp(arg, "--diff-filter="))
2894                 options->filter = arg + 14;
2895         else if (!strcmp(arg, "--abbrev"))
2896                 options->abbrev = DEFAULT_ABBREV;
2897         else if (!prefixcmp(arg, "--abbrev=")) {
2898                 options->abbrev = strtoul(arg + 9, NULL, 10);
2899                 if (options->abbrev < MINIMUM_ABBREV)
2900                         options->abbrev = MINIMUM_ABBREV;
2901                 else if (40 < options->abbrev)
2902                         options->abbrev = 40;
2903         }
2904         else if (!prefixcmp(arg, "--src-prefix="))
2905                 options->a_prefix = arg + 13;
2906         else if (!prefixcmp(arg, "--dst-prefix="))
2907                 options->b_prefix = arg + 13;
2908         else if (!strcmp(arg, "--no-prefix"))
2909                 options->a_prefix = options->b_prefix = "";
2910         else if (opt_arg(arg, '\0', "inter-hunk-context",
2911                          &options->interhunkcontext))
2912                 ;
2913         else if (!prefixcmp(arg, "--output=")) {
2914                 options->file = fopen(arg + strlen("--output="), "w");
2915                 if (!options->file)
2916                         die_errno("Could not open '%s'", arg + strlen("--output="));
2917                 options->close_file = 1;
2918         } else
2919                 return 0;
2920         return 1;
2921 }
2922
2923 static int parse_num(const char **cp_p)
2924 {
2925         unsigned long num, scale;
2926         int ch, dot;
2927         const char *cp = *cp_p;
2928
2929         num = 0;
2930         scale = 1;
2931         dot = 0;
2932         for (;;) {
2933                 ch = *cp;
2934                 if ( !dot && ch == '.' ) {
2935                         scale = 1;
2936                         dot = 1;
2937                 } else if ( ch == '%' ) {
2938                         scale = dot ? scale*100 : 100;
2939                         cp++;   /* % is always at the end */
2940                         break;
2941                 } else if ( ch >= '0' && ch <= '9' ) {
2942                         if ( scale < 100000 ) {
2943                                 scale *= 10;
2944                                 num = (num*10) + (ch-'0');
2945                         }
2946                 } else {
2947                         break;
2948                 }
2949                 cp++;
2950         }
2951         *cp_p = cp;
2952
2953         /* user says num divided by scale and we say internally that
2954          * is MAX_SCORE * num / scale.
2955          */
2956         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2957 }
2958
2959 static int diff_scoreopt_parse(const char *opt)
2960 {
2961         int opt1, opt2, cmd;
2962
2963         if (*opt++ != '-')
2964                 return -1;
2965         cmd = *opt++;
2966         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2967                 return -1; /* that is not a -M, -C nor -B option */
2968
2969         opt1 = parse_num(&opt);
2970         if (cmd != 'B')
2971                 opt2 = 0;
2972         else {
2973                 if (*opt == 0)
2974                         opt2 = 0;
2975                 else if (*opt != '/')
2976                         return -1; /* we expect -B80/99 or -B80 */
2977                 else {
2978                         opt++;
2979                         opt2 = parse_num(&opt);
2980                 }
2981         }
2982         if (*opt != 0)
2983                 return -1;
2984         return opt1 | (opt2 << 16);
2985 }
2986
2987 struct diff_queue_struct diff_queued_diff;
2988
2989 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2990 {
2991         if (queue->alloc <= queue->nr) {
2992                 queue->alloc = alloc_nr(queue->alloc);
2993                 queue->queue = xrealloc(queue->queue,
2994                                         sizeof(dp) * queue->alloc);
2995         }
2996         queue->queue[queue->nr++] = dp;
2997 }
2998
2999 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3000                                  struct diff_filespec *one,
3001                                  struct diff_filespec *two)
3002 {
3003         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3004         dp->one = one;
3005         dp->two = two;
3006         if (queue)
3007                 diff_q(queue, dp);
3008         return dp;
3009 }
3010
3011 void diff_free_filepair(struct diff_filepair *p)
3012 {
3013         free_filespec(p->one);
3014         free_filespec(p->two);
3015         free(p);
3016 }
3017
3018 /* This is different from find_unique_abbrev() in that
3019  * it stuffs the result with dots for alignment.
3020  */
3021 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3022 {
3023         int abblen;
3024         const char *abbrev;
3025         if (len == 40)
3026                 return sha1_to_hex(sha1);
3027
3028         abbrev = find_unique_abbrev(sha1, len);
3029         abblen = strlen(abbrev);
3030         if (abblen < 37) {
3031                 static char hex[41];
3032                 if (len < abblen && abblen <= len + 2)
3033                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3034                 else
3035                         sprintf(hex, "%s...", abbrev);
3036                 return hex;
3037         }
3038         return sha1_to_hex(sha1);
3039 }
3040
3041 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3042 {
3043         int line_termination = opt->line_termination;
3044         int inter_name_termination = line_termination ? '\t' : '\0';
3045
3046         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3047                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3048                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3049                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3050         }
3051         if (p->score) {
3052                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3053                         inter_name_termination);
3054         } else {
3055                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3056         }
3057
3058         if (p->status == DIFF_STATUS_COPIED ||
3059             p->status == DIFF_STATUS_RENAMED) {
3060                 const char *name_a, *name_b;
3061                 name_a = p->one->path;
3062                 name_b = p->two->path;
3063                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3064                 write_name_quoted(name_a, opt->file, inter_name_termination);
3065                 write_name_quoted(name_b, opt->file, line_termination);
3066         } else {
3067                 const char *name_a, *name_b;
3068                 name_a = p->one->mode ? p->one->path : p->two->path;
3069                 name_b = NULL;
3070                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3071                 write_name_quoted(name_a, opt->file, line_termination);
3072         }
3073 }
3074
3075 int diff_unmodified_pair(struct diff_filepair *p)
3076 {
3077         /* This function is written stricter than necessary to support
3078          * the currently implemented transformers, but the idea is to
3079          * let transformers to produce diff_filepairs any way they want,
3080          * and filter and clean them up here before producing the output.
3081          */
3082         struct diff_filespec *one = p->one, *two = p->two;
3083
3084         if (DIFF_PAIR_UNMERGED(p))
3085                 return 0; /* unmerged is interesting */
3086
3087         /* deletion, addition, mode or type change
3088          * and rename are all interesting.
3089          */
3090         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3091             DIFF_PAIR_MODE_CHANGED(p) ||
3092             strcmp(one->path, two->path))
3093                 return 0;
3094
3095         /* both are valid and point at the same path.  that is, we are
3096          * dealing with a change.
3097          */
3098         if (one->sha1_valid && two->sha1_valid &&
3099             !hashcmp(one->sha1, two->sha1) &&
3100             !one->dirty_submodule && !two->dirty_submodule)
3101                 return 1; /* no change */
3102         if (!one->sha1_valid && !two->sha1_valid)
3103                 return 1; /* both look at the same file on the filesystem. */
3104         return 0;
3105 }
3106
3107 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3108 {
3109         if (diff_unmodified_pair(p))
3110                 return;
3111
3112         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3113             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3114                 return; /* no tree diffs in patch format */
3115
3116         run_diff(p, o);
3117 }
3118
3119 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3120                             struct diffstat_t *diffstat)
3121 {
3122         if (diff_unmodified_pair(p))
3123                 return;
3124
3125         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3126             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3127                 return; /* no tree diffs in patch format */
3128
3129         run_diffstat(p, o, diffstat);
3130 }
3131
3132 static void diff_flush_checkdiff(struct diff_filepair *p,
3133                 struct diff_options *o)
3134 {
3135         if (diff_unmodified_pair(p))
3136                 return;
3137
3138         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3139             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3140                 return; /* no tree diffs in patch format */
3141
3142         run_checkdiff(p, o);
3143 }
3144
3145 int diff_queue_is_empty(void)
3146 {
3147         struct diff_queue_struct *q = &diff_queued_diff;
3148         int i;
3149         for (i = 0; i < q->nr; i++)
3150                 if (!diff_unmodified_pair(q->queue[i]))
3151                         return 0;
3152         return 1;
3153 }
3154
3155 #if DIFF_DEBUG
3156 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3157 {
3158         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3159                 x, one ? one : "",
3160                 s->path,
3161                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3162                 s->mode,
3163                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3164         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3165                 x, one ? one : "",
3166                 s->size, s->xfrm_flags);
3167 }
3168
3169 void diff_debug_filepair(const struct diff_filepair *p, int i)
3170 {
3171         diff_debug_filespec(p->one, i, "one");
3172         diff_debug_filespec(p->two, i, "two");
3173         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3174                 p->score, p->status ? p->status : '?',
3175                 p->one->rename_used, p->broken_pair);
3176 }
3177
3178 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3179 {
3180         int i;
3181         if (msg)
3182                 fprintf(stderr, "%s\n", msg);
3183         fprintf(stderr, "q->nr = %d\n", q->nr);
3184         for (i = 0; i < q->nr; i++) {
3185                 struct diff_filepair *p = q->queue[i];
3186                 diff_debug_filepair(p, i);
3187         }
3188 }
3189 #endif
3190
3191 static void diff_resolve_rename_copy(void)
3192 {
3193         int i;
3194         struct diff_filepair *p;
3195         struct diff_queue_struct *q = &diff_queued_diff;
3196
3197         diff_debug_queue("resolve-rename-copy", q);
3198
3199         for (i = 0; i < q->nr; i++) {
3200                 p = q->queue[i];
3201                 p->status = 0; /* undecided */
3202                 if (DIFF_PAIR_UNMERGED(p))
3203                         p->status = DIFF_STATUS_UNMERGED;
3204                 else if (!DIFF_FILE_VALID(p->one))
3205                         p->status = DIFF_STATUS_ADDED;
3206                 else if (!DIFF_FILE_VALID(p->two))
3207                         p->status = DIFF_STATUS_DELETED;
3208                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3209                         p->status = DIFF_STATUS_TYPE_CHANGED;
3210
3211                 /* from this point on, we are dealing with a pair
3212                  * whose both sides are valid and of the same type, i.e.
3213                  * either in-place edit or rename/copy edit.
3214                  */
3215                 else if (DIFF_PAIR_RENAME(p)) {
3216                         /*
3217                          * A rename might have re-connected a broken
3218                          * pair up, causing the pathnames to be the
3219                          * same again. If so, that's not a rename at
3220                          * all, just a modification..
3221                          *
3222                          * Otherwise, see if this source was used for
3223                          * multiple renames, in which case we decrement
3224                          * the count, and call it a copy.
3225                          */
3226                         if (!strcmp(p->one->path, p->two->path))
3227                                 p->status = DIFF_STATUS_MODIFIED;
3228                         else if (--p->one->rename_used > 0)
3229                                 p->status = DIFF_STATUS_COPIED;
3230                         else
3231                                 p->status = DIFF_STATUS_RENAMED;
3232                 }
3233                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3234                          p->one->mode != p->two->mode ||
3235                          p->one->dirty_submodule ||
3236                          p->two->dirty_submodule ||
3237                          is_null_sha1(p->one->sha1))
3238                         p->status = DIFF_STATUS_MODIFIED;
3239                 else {
3240                         /* This is a "no-change" entry and should not
3241                          * happen anymore, but prepare for broken callers.
3242                          */
3243                         error("feeding unmodified %s to diffcore",
3244                               p->one->path);
3245                         p->status = DIFF_STATUS_UNKNOWN;
3246                 }
3247         }
3248         diff_debug_queue("resolve-rename-copy done", q);
3249 }
3250
3251 static int check_pair_status(struct diff_filepair *p)
3252 {
3253         switch (p->status) {
3254         case DIFF_STATUS_UNKNOWN:
3255                 return 0;
3256         case 0:
3257                 die("internal error in diff-resolve-rename-copy");
3258         default:
3259                 return 1;
3260         }
3261 }
3262
3263 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3264 {
3265         int fmt = opt->output_format;
3266
3267         if (fmt & DIFF_FORMAT_CHECKDIFF)
3268                 diff_flush_checkdiff(p, opt);
3269         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3270                 diff_flush_raw(p, opt);
3271         else if (fmt & DIFF_FORMAT_NAME) {
3272                 const char *name_a, *name_b;
3273                 name_a = p->two->path;
3274                 name_b = NULL;
3275                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3276                 write_name_quoted(name_a, opt->file, opt->line_termination);
3277         }
3278 }
3279
3280 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3281 {
3282         if (fs->mode)
3283                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3284         else
3285                 fprintf(file, " %s ", newdelete);
3286         write_name_quoted(fs->path, file, '\n');
3287 }
3288
3289
3290 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3291 {
3292         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3293                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3294                         show_name ? ' ' : '\n');
3295                 if (show_name) {
3296                         write_name_quoted(p->two->path, file, '\n');
3297                 }
3298         }
3299 }
3300
3301 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3302 {
3303         char *names = pprint_rename(p->one->path, p->two->path);
3304
3305         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3306         free(names);
3307         show_mode_change(file, p, 0);
3308 }
3309
3310 static void diff_summary(FILE *file, struct diff_filepair *p)
3311 {
3312         switch(p->status) {
3313         case DIFF_STATUS_DELETED:
3314                 show_file_mode_name(file, "delete", p->one);
3315                 break;
3316         case DIFF_STATUS_ADDED:
3317                 show_file_mode_name(file, "create", p->two);
3318                 break;
3319         case DIFF_STATUS_COPIED:
3320                 show_rename_copy(file, "copy", p);
3321                 break;
3322         case DIFF_STATUS_RENAMED:
3323                 show_rename_copy(file, "rename", p);
3324                 break;
3325         default:
3326                 if (p->score) {
3327                         fputs(" rewrite ", file);
3328                         write_name_quoted(p->two->path, file, ' ');
3329                         fprintf(file, "(%d%%)\n", similarity_index(p));
3330                 }
3331                 show_mode_change(file, p, !p->score);
3332                 break;
3333         }
3334 }
3335
3336 struct patch_id_t {
3337         git_SHA_CTX *ctx;
3338         int patchlen;
3339 };
3340
3341 static int remove_space(char *line, int len)
3342 {
3343         int i;
3344         char *dst = line;
3345         unsigned char c;
3346
3347         for (i = 0; i < len; i++)
3348                 if (!isspace((c = line[i])))
3349                         *dst++ = c;
3350
3351         return dst - line;
3352 }
3353
3354 static void patch_id_consume(void *priv, char *line, unsigned long len)
3355 {
3356         struct patch_id_t *data = priv;
3357         int new_len;
3358
3359         /* Ignore line numbers when computing the SHA1 of the patch */
3360         if (!prefixcmp(line, "@@ -"))
3361                 return;
3362
3363         new_len = remove_space(line, len);
3364
3365         git_SHA1_Update(data->ctx, line, new_len);
3366         data->patchlen += new_len;
3367 }
3368
3369 /* returns 0 upon success, and writes result into sha1 */
3370 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3371 {
3372         struct diff_queue_struct *q = &diff_queued_diff;
3373         int i;
3374         git_SHA_CTX ctx;
3375         struct patch_id_t data;
3376         char buffer[PATH_MAX * 4 + 20];
3377
3378         git_SHA1_Init(&ctx);
3379         memset(&data, 0, sizeof(struct patch_id_t));
3380         data.ctx = &ctx;
3381
3382         for (i = 0; i < q->nr; i++) {
3383                 xpparam_t xpp;
3384                 xdemitconf_t xecfg;
3385                 xdemitcb_t ecb;
3386                 mmfile_t mf1, mf2;
3387                 struct diff_filepair *p = q->queue[i];
3388                 int len1, len2;
3389
3390                 memset(&xpp, 0, sizeof(xpp));
3391                 memset(&xecfg, 0, sizeof(xecfg));
3392                 if (p->status == 0)
3393                         return error("internal diff status error");
3394                 if (p->status == DIFF_STATUS_UNKNOWN)
3395                         continue;
3396                 if (diff_unmodified_pair(p))
3397                         continue;
3398                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3399                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3400                         continue;
3401                 if (DIFF_PAIR_UNMERGED(p))
3402                         continue;
3403
3404                 diff_fill_sha1_info(p->one);
3405                 diff_fill_sha1_info(p->two);
3406                 if (fill_mmfile(&mf1, p->one) < 0 ||
3407                                 fill_mmfile(&mf2, p->two) < 0)
3408                         return error("unable to read files to diff");
3409
3410                 len1 = remove_space(p->one->path, strlen(p->one->path));
3411                 len2 = remove_space(p->two->path, strlen(p->two->path));
3412                 if (p->one->mode == 0)
3413                         len1 = snprintf(buffer, sizeof(buffer),
3414                                         "diff--gita/%.*sb/%.*s"
3415                                         "newfilemode%06o"
3416                                         "---/dev/null"
3417                                         "+++b/%.*s",
3418                                         len1, p->one->path,
3419                                         len2, p->two->path,
3420                                         p->two->mode,
3421                                         len2, p->two->path);
3422                 else if (p->two->mode == 0)
3423                         len1 = snprintf(buffer, sizeof(buffer),
3424                                         "diff--gita/%.*sb/%.*s"
3425                                         "deletedfilemode%06o"
3426                                         "---a/%.*s"
3427                                         "+++/dev/null",
3428                                         len1, p->one->path,
3429                                         len2, p->two->path,
3430                                         p->one->mode,
3431                                         len1, p->one->path);
3432                 else
3433                         len1 = snprintf(buffer, sizeof(buffer),
3434                                         "diff--gita/%.*sb/%.*s"
3435                                         "---a/%.*s"
3436                                         "+++b/%.*s",
3437                                         len1, p->one->path,
3438                                         len2, p->two->path,
3439                                         len1, p->one->path,
3440                                         len2, p->two->path);
3441                 git_SHA1_Update(&ctx, buffer, len1);
3442
3443                 xpp.flags = XDF_NEED_MINIMAL;
3444                 xecfg.ctxlen = 3;
3445                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3446                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3447                               &xpp, &xecfg, &ecb);
3448         }
3449
3450         git_SHA1_Final(sha1, &ctx);
3451         return 0;
3452 }
3453
3454 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3455 {
3456         struct diff_queue_struct *q = &diff_queued_diff;
3457         int i;
3458         int result = diff_get_patch_id(options, sha1);
3459
3460         for (i = 0; i < q->nr; i++)
3461                 diff_free_filepair(q->queue[i]);
3462
3463         free(q->queue);
3464         q->queue = NULL;
3465         q->nr = q->alloc = 0;
3466
3467         return result;
3468 }
3469
3470 static int is_summary_empty(const struct diff_queue_struct *q)
3471 {
3472         int i;
3473
3474         for (i = 0; i < q->nr; i++) {
3475                 const struct diff_filepair *p = q->queue[i];
3476
3477                 switch (p->status) {
3478                 case DIFF_STATUS_DELETED:
3479                 case DIFF_STATUS_ADDED:
3480                 case DIFF_STATUS_COPIED:
3481                 case DIFF_STATUS_RENAMED:
3482                         return 0;
3483                 default:
3484                         if (p->score)
3485                                 return 0;
3486                         if (p->one->mode && p->two->mode &&
3487                             p->one->mode != p->two->mode)
3488                                 return 0;
3489                         break;
3490                 }
3491         }
3492         return 1;
3493 }
3494
3495 void diff_flush(struct diff_options *options)
3496 {
3497         struct diff_queue_struct *q = &diff_queued_diff;
3498         int i, output_format = options->output_format;
3499         int separator = 0;
3500
3501         /*
3502          * Order: raw, stat, summary, patch
3503          * or:    name/name-status/checkdiff (other bits clear)
3504          */
3505         if (!q->nr)
3506                 goto free_queue;
3507
3508         if (output_format & (DIFF_FORMAT_RAW |
3509                              DIFF_FORMAT_NAME |
3510                              DIFF_FORMAT_NAME_STATUS |
3511                              DIFF_FORMAT_CHECKDIFF)) {
3512                 for (i = 0; i < q->nr; i++) {
3513                         struct diff_filepair *p = q->queue[i];
3514                         if (check_pair_status(p))
3515                                 flush_one_pair(p, options);
3516                 }
3517                 separator++;
3518         }
3519
3520         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3521                 struct diffstat_t diffstat;
3522
3523                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3524                 for (i = 0; i < q->nr; i++) {
3525                         struct diff_filepair *p = q->queue[i];
3526                         if (check_pair_status(p))
3527                                 diff_flush_stat(p, options, &diffstat);
3528                 }
3529                 if (output_format & DIFF_FORMAT_NUMSTAT)
3530                         show_numstat(&diffstat, options);
3531                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3532                         show_stats(&diffstat, options);
3533                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3534                         show_shortstats(&diffstat, options);
3535                 free_diffstat_info(&diffstat);
3536                 separator++;
3537         }
3538         if (output_format & DIFF_FORMAT_DIRSTAT)
3539                 show_dirstat(options);
3540
3541         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3542                 for (i = 0; i < q->nr; i++)
3543                         diff_summary(options->file, q->queue[i]);
3544                 separator++;
3545         }
3546
3547         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3548             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3549             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3550                 /*
3551                  * run diff_flush_patch for the exit status. setting
3552                  * options->file to /dev/null should be safe, becaue we
3553                  * aren't supposed to produce any output anyway.
3554                  */
3555                 if (options->close_file)
3556                         fclose(options->file);
3557                 options->file = fopen("/dev/null", "w");
3558                 if (!options->file)
3559                         die_errno("Could not open /dev/null");
3560                 options->close_file = 1;
3561                 for (i = 0; i < q->nr; i++) {
3562                         struct diff_filepair *p = q->queue[i];
3563                         if (check_pair_status(p))
3564                                 diff_flush_patch(p, options);
3565                         if (options->found_changes)
3566                                 break;
3567                 }
3568         }
3569
3570         if (output_format & DIFF_FORMAT_PATCH) {
3571                 if (separator) {
3572                         putc(options->line_termination, options->file);
3573                         if (options->stat_sep) {
3574                                 /* attach patch instead of inline */
3575                                 fputs(options->stat_sep, options->file);
3576                         }
3577                 }
3578
3579                 for (i = 0; i < q->nr; i++) {
3580                         struct diff_filepair *p = q->queue[i];
3581                         if (check_pair_status(p))
3582                                 diff_flush_patch(p, options);
3583                 }
3584         }
3585
3586         if (output_format & DIFF_FORMAT_CALLBACK)
3587                 options->format_callback(q, options, options->format_callback_data);
3588
3589         for (i = 0; i < q->nr; i++)
3590                 diff_free_filepair(q->queue[i]);
3591 free_queue:
3592         free(q->queue);
3593         q->queue = NULL;
3594         q->nr = q->alloc = 0;
3595         if (options->close_file)
3596                 fclose(options->file);
3597
3598         /*
3599          * Report the content-level differences with HAS_CHANGES;
3600          * diff_addremove/diff_change does not set the bit when
3601          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3602          */
3603         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3604                 if (options->found_changes)
3605                         DIFF_OPT_SET(options, HAS_CHANGES);
3606                 else
3607                         DIFF_OPT_CLR(options, HAS_CHANGES);
3608         }
3609 }
3610
3611 static void diffcore_apply_filter(const char *filter)
3612 {
3613         int i;
3614         struct diff_queue_struct *q = &diff_queued_diff;
3615         struct diff_queue_struct outq;
3616         outq.queue = NULL;
3617         outq.nr = outq.alloc = 0;
3618
3619         if (!filter)
3620                 return;
3621
3622         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3623                 int found;
3624                 for (i = found = 0; !found && i < q->nr; i++) {
3625                         struct diff_filepair *p = q->queue[i];
3626                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3627                              ((p->score &&
3628                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3629                               (!p->score &&
3630                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3631                             ((p->status != DIFF_STATUS_MODIFIED) &&
3632                              strchr(filter, p->status)))
3633                                 found++;
3634                 }
3635                 if (found)
3636                         return;
3637
3638                 /* otherwise we will clear the whole queue
3639                  * by copying the empty outq at the end of this
3640                  * function, but first clear the current entries
3641                  * in the queue.
3642                  */
3643                 for (i = 0; i < q->nr; i++)
3644                         diff_free_filepair(q->queue[i]);
3645         }
3646         else {
3647                 /* Only the matching ones */
3648                 for (i = 0; i < q->nr; i++) {
3649                         struct diff_filepair *p = q->queue[i];
3650
3651                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3652                              ((p->score &&
3653                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3654                               (!p->score &&
3655                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3656                             ((p->status != DIFF_STATUS_MODIFIED) &&
3657                              strchr(filter, p->status)))
3658                                 diff_q(&outq, p);
3659                         else
3660                                 diff_free_filepair(p);
3661                 }
3662         }
3663         free(q->queue);
3664         *q = outq;
3665 }
3666
3667 /* Check whether two filespecs with the same mode and size are identical */
3668 static int diff_filespec_is_identical(struct diff_filespec *one,
3669                                       struct diff_filespec *two)
3670 {
3671         if (S_ISGITLINK(one->mode))
3672                 return 0;
3673         if (diff_populate_filespec(one, 0))
3674                 return 0;
3675         if (diff_populate_filespec(two, 0))
3676                 return 0;
3677         return !memcmp(one->data, two->data, one->size);
3678 }
3679
3680 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3681 {
3682         int i;
3683         struct diff_queue_struct *q = &diff_queued_diff;
3684         struct diff_queue_struct outq;
3685         outq.queue = NULL;
3686         outq.nr = outq.alloc = 0;
3687
3688         for (i = 0; i < q->nr; i++) {
3689                 struct diff_filepair *p = q->queue[i];
3690
3691                 /*
3692                  * 1. Entries that come from stat info dirtiness
3693                  *    always have both sides (iow, not create/delete),
3694                  *    one side of the object name is unknown, with
3695                  *    the same mode and size.  Keep the ones that
3696                  *    do not match these criteria.  They have real
3697                  *    differences.
3698                  *
3699                  * 2. At this point, the file is known to be modified,
3700                  *    with the same mode and size, and the object
3701                  *    name of one side is unknown.  Need to inspect
3702                  *    the identical contents.
3703                  */
3704                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3705                     !DIFF_FILE_VALID(p->two) ||
3706                     (p->one->sha1_valid && p->two->sha1_valid) ||
3707                     (p->one->mode != p->two->mode) ||
3708                     diff_populate_filespec(p->one, 1) ||
3709                     diff_populate_filespec(p->two, 1) ||
3710                     (p->one->size != p->two->size) ||
3711                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3712                         diff_q(&outq, p);
3713                 else {
3714                         /*
3715                          * The caller can subtract 1 from skip_stat_unmatch
3716                          * to determine how many paths were dirty only
3717                          * due to stat info mismatch.
3718                          */
3719                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3720                                 diffopt->skip_stat_unmatch++;
3721                         diff_free_filepair(p);
3722                 }
3723         }
3724         free(q->queue);
3725         *q = outq;
3726 }
3727
3728 static int diffnamecmp(const void *a_, const void *b_)
3729 {
3730         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3731         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3732         const char *name_a, *name_b;
3733
3734         name_a = a->one ? a->one->path : a->two->path;
3735         name_b = b->one ? b->one->path : b->two->path;
3736         return strcmp(name_a, name_b);
3737 }
3738
3739 void diffcore_fix_diff_index(struct diff_options *options)
3740 {
3741         struct diff_queue_struct *q = &diff_queued_diff;
3742         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3743 }
3744
3745 void diffcore_std(struct diff_options *options)
3746 {
3747         if (options->skip_stat_unmatch)
3748                 diffcore_skip_stat_unmatch(options);
3749         if (options->break_opt != -1)
3750                 diffcore_break(options->break_opt);
3751         if (options->detect_rename)
3752                 diffcore_rename(options);
3753         if (options->break_opt != -1)
3754                 diffcore_merge_broken();
3755         if (options->pickaxe)
3756                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3757         if (options->orderfile)
3758                 diffcore_order(options->orderfile);
3759         diff_resolve_rename_copy();
3760         diffcore_apply_filter(options->filter);
3761
3762         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3763                 DIFF_OPT_SET(options, HAS_CHANGES);
3764         else
3765                 DIFF_OPT_CLR(options, HAS_CHANGES);
3766 }
3767
3768 int diff_result_code(struct diff_options *opt, int status)
3769 {
3770         int result = 0;
3771         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3772             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3773                 return status;
3774         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3775             DIFF_OPT_TST(opt, HAS_CHANGES))
3776                 result |= 01;
3777         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3778             DIFF_OPT_TST(opt, CHECK_FAILED))
3779                 result |= 02;
3780         return result;
3781 }
3782
3783 void diff_addremove(struct diff_options *options,
3784                     int addremove, unsigned mode,
3785                     const unsigned char *sha1,
3786                     const char *concatpath, unsigned dirty_submodule)
3787 {
3788         struct diff_filespec *one, *two;
3789
3790         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3791                 return;
3792
3793         /* This may look odd, but it is a preparation for
3794          * feeding "there are unchanged files which should
3795          * not produce diffs, but when you are doing copy
3796          * detection you would need them, so here they are"
3797          * entries to the diff-core.  They will be prefixed
3798          * with something like '=' or '*' (I haven't decided
3799          * which but should not make any difference).
3800          * Feeding the same new and old to diff_change()
3801          * also has the same effect.
3802          * Before the final output happens, they are pruned after
3803          * merged into rename/copy pairs as appropriate.
3804          */
3805         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3806                 addremove = (addremove == '+' ? '-' :
3807                              addremove == '-' ? '+' : addremove);
3808
3809         if (options->prefix &&
3810             strncmp(concatpath, options->prefix, options->prefix_length))
3811                 return;
3812
3813         one = alloc_filespec(concatpath);
3814         two = alloc_filespec(concatpath);
3815
3816         if (addremove != '+')
3817                 fill_filespec(one, sha1, mode);
3818         if (addremove != '-') {
3819                 fill_filespec(two, sha1, mode);
3820                 two->dirty_submodule = dirty_submodule;
3821         }
3822
3823         diff_queue(&diff_queued_diff, one, two);
3824         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3825                 DIFF_OPT_SET(options, HAS_CHANGES);
3826 }
3827
3828 void diff_change(struct diff_options *options,
3829                  unsigned old_mode, unsigned new_mode,
3830                  const unsigned char *old_sha1,
3831                  const unsigned char *new_sha1,
3832                  const char *concatpath,
3833                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3834 {
3835         struct diff_filespec *one, *two;
3836
3837         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3838                         && S_ISGITLINK(new_mode))
3839                 return;
3840
3841         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3842                 unsigned tmp;
3843                 const unsigned char *tmp_c;
3844                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3845                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3846                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3847                         new_dirty_submodule = tmp;
3848         }
3849
3850         if (options->prefix &&
3851             strncmp(concatpath, options->prefix, options->prefix_length))
3852                 return;
3853
3854         one = alloc_filespec(concatpath);
3855         two = alloc_filespec(concatpath);
3856         fill_filespec(one, old_sha1, old_mode);
3857         fill_filespec(two, new_sha1, new_mode);
3858         one->dirty_submodule = old_dirty_submodule;
3859         two->dirty_submodule = new_dirty_submodule;
3860
3861         diff_queue(&diff_queued_diff, one, two);
3862         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3863                 DIFF_OPT_SET(options, HAS_CHANGES);
3864 }
3865
3866 void diff_unmerge(struct diff_options *options,
3867                   const char *path,
3868                   unsigned mode, const unsigned char *sha1)
3869 {
3870         struct diff_filespec *one, *two;
3871
3872         if (options->prefix &&
3873             strncmp(path, options->prefix, options->prefix_length))
3874                 return;
3875
3876         one = alloc_filespec(path);
3877         two = alloc_filespec(path);
3878         fill_filespec(one, sha1, mode);
3879         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3880 }
3881
3882 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3883                 size_t *outsize)
3884 {
3885         struct diff_tempfile *temp;
3886         const char *argv[3];
3887         const char **arg = argv;
3888         struct child_process child;
3889         struct strbuf buf = STRBUF_INIT;
3890         int err = 0;
3891
3892         temp = prepare_temp_file(spec->path, spec);
3893         *arg++ = pgm;
3894         *arg++ = temp->name;
3895         *arg = NULL;
3896
3897         memset(&child, 0, sizeof(child));
3898         child.use_shell = 1;
3899         child.argv = argv;
3900         child.out = -1;
3901         if (start_command(&child)) {
3902                 remove_tempfile();
3903                 return NULL;
3904         }
3905
3906         if (strbuf_read(&buf, child.out, 0) < 0)
3907                 err = error("error reading from textconv command '%s'", pgm);
3908         close(child.out);
3909
3910         if (finish_command(&child) || err) {
3911                 strbuf_release(&buf);
3912                 remove_tempfile();
3913                 return NULL;
3914         }
3915         remove_tempfile();
3916
3917         return strbuf_detach(&buf, outsize);
3918 }