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 #include "ll-merge.h"
  18 #include "string-list.h"
  19 #include "argv-array.h"
  20
  21 #ifdef NO_FAST_WORKING_DIRECTORY
  22 #define FAST_WORKING_DIRECTORY 0
  23 #else
  24 #define FAST_WORKING_DIRECTORY 1
  25 #endif
  26
  27 static int diff_detect_rename_default;
  28 static int diff_rename_limit_default = 400;
  29 static int diff_suppress_blank_empty;
  30 static int diff_use_color_default = -1;
  31 static int diff_context_default = 3;
  32 static const char *diff_word_regex_cfg;
  33 static const char *external_diff_cmd_cfg;
  34 static const char *diff_order_file_cfg;
  35 int diff_auto_refresh_index = 1;
  36 static int diff_mnemonic_prefix;
  37 static int diff_no_prefix;
  38 static int diff_stat_graph_width;
  39 static int diff_dirstat_permille_default = 30;
  40 static struct diff_options default_diff_options;
  41 static long diff_algorithm;
  42
  43 static char diff_colors[][COLOR_MAXLEN] = {
  44         GIT_COLOR_RESET,
  45         GIT_COLOR_NORMAL,       /* PLAIN */
  46         GIT_COLOR_BOLD,         /* METAINFO */
  47         GIT_COLOR_CYAN,         /* FRAGINFO */
  48         GIT_COLOR_RED,          /* OLD */
  49         GIT_COLOR_GREEN,        /* NEW */
  50         GIT_COLOR_YELLOW,       /* COMMIT */
  51         GIT_COLOR_BG_RED,       /* WHITESPACE */
  52         GIT_COLOR_NORMAL,       /* FUNCINFO */
  53 };
  54
  55 static int parse_diff_color_slot(const char *var, int ofs)
  56 {
  57         if (!strcasecmp(var+ofs, "plain"))
  58                 return DIFF_PLAIN;
  59         if (!strcasecmp(var+ofs, "meta"))
  60                 return DIFF_METAINFO;
  61         if (!strcasecmp(var+ofs, "frag"))
  62                 return DIFF_FRAGINFO;
  63         if (!strcasecmp(var+ofs, "old"))
  64                 return DIFF_FILE_OLD;
  65         if (!strcasecmp(var+ofs, "new"))
  66                 return DIFF_FILE_NEW;
  67         if (!strcasecmp(var+ofs, "commit"))
  68                 return DIFF_COMMIT;
  69         if (!strcasecmp(var+ofs, "whitespace"))
  70                 return DIFF_WHITESPACE;
  71         if (!strcasecmp(var+ofs, "func"))
  72                 return DIFF_FUNCINFO;
  73         return -1;
  74 }
  75
  76 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  77                                 struct strbuf *errmsg)
  78 {
  79         char *params_copy = xstrdup(params_string);
  80         struct string_list params = STRING_LIST_INIT_NODUP;
  81         int ret = 0;
  82         int i;
  83
  84         if (*params_copy)
  85                 string_list_split_in_place(&params, params_copy, ',', -1);
  86         for (i = 0; i < params.nr; i++) {
  87                 const char *p = params.items[i].string;
  88                 if (!strcmp(p, "changes")) {
  89                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  90                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  91                 } else if (!strcmp(p, "lines")) {
  92                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  93                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  94                 } else if (!strcmp(p, "files")) {
  95                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  96                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
  97                 } else if (!strcmp(p, "noncumulative")) {
  98                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
  99                 } else if (!strcmp(p, "cumulative")) {
 100                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 101                 } else if (isdigit(*p)) {
 102                         char *end;
 103                         int permille = strtoul(p, &end, 10) * 10;
 104                         if (*end == '.' && isdigit(*++end)) {
 105                                 /* only use first digit */
 106                                 permille += *end - '0';
 107                                 /* .. and ignore any further digits */
 108                                 while (isdigit(*++end))
 109                                         ; /* nothing */
 110                         }
 111                         if (!*end)
 112                                 options->dirstat_permille = permille;
 113                         else {
 114                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 115                                             p);
 116                                 ret++;
 117                         }
 118                 } else {
 119                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 120                         ret++;
 121                 }
 122
 123         }
 124         string_list_clear(&params, 0);
 125         free(params_copy);
 126         return ret;
 127 }
 128
 129 static int parse_submodule_params(struct diff_options *options, const char *value)
 130 {
 131         if (!strcmp(value, "log"))
 132                 DIFF_OPT_SET(options, SUBMODULE_LOG);
 133         else if (!strcmp(value, "short"))
 134                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
 135         else
 136                 return -1;
 137         return 0;
 138 }
 139
 140 static int git_config_rename(const char *var, const char *value)
 141 {
 142         if (!value)
 143                 return DIFF_DETECT_RENAME;
 144         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 145                 return  DIFF_DETECT_COPY;
 146         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 147 }
 148
 149 long parse_algorithm_value(const char *value)
 150 {
 151         if (!value)
 152                 return -1;
 153         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 154                 return 0;
 155         else if (!strcasecmp(value, "minimal"))
 156                 return XDF_NEED_MINIMAL;
 157         else if (!strcasecmp(value, "patience"))
 158                 return XDF_PATIENCE_DIFF;
 159         else if (!strcasecmp(value, "histogram"))
 160                 return XDF_HISTOGRAM_DIFF;
 161         return -1;
 162 }
 163
 164 /*
 165  * These are to give UI layer defaults.
 166  * The core-level commands such as git-diff-files should
 167  * never be affected by the setting of diff.renames
 168  * the user happens to have in the configuration file.
 169  */
 170 int git_diff_ui_config(const char *var, const char *value, void *cb)
 171 {
 172         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 173                 diff_use_color_default = git_config_colorbool(var, value);
 174                 return 0;
 175         }
 176         if (!strcmp(var, "diff.context")) {
 177                 diff_context_default = git_config_int(var, value);
 178                 if (diff_context_default < 0)
 179                         return -1;
 180                 return 0;
 181         }
 182         if (!strcmp(var, "diff.renames")) {
 183                 diff_detect_rename_default = git_config_rename(var, value);
 184                 return 0;
 185         }
 186         if (!strcmp(var, "diff.autorefreshindex")) {
 187                 diff_auto_refresh_index = git_config_bool(var, value);
 188                 return 0;
 189         }
 190         if (!strcmp(var, "diff.mnemonicprefix")) {
 191                 diff_mnemonic_prefix = git_config_bool(var, value);
 192                 return 0;
 193         }
 194         if (!strcmp(var, "diff.noprefix")) {
 195                 diff_no_prefix = git_config_bool(var, value);
 196                 return 0;
 197         }
 198         if (!strcmp(var, "diff.statgraphwidth")) {
 199                 diff_stat_graph_width = git_config_int(var, value);
 200                 return 0;
 201         }
 202         if (!strcmp(var, "diff.external"))
 203                 return git_config_string(&external_diff_cmd_cfg, var, value);
 204         if (!strcmp(var, "diff.wordregex"))
 205                 return git_config_string(&diff_word_regex_cfg, var, value);
 206         if (!strcmp(var, "diff.orderfile"))
 207                 return git_config_pathname(&diff_order_file_cfg, var, value);
 208
 209         if (!strcmp(var, "diff.ignoresubmodules"))
 210                 handle_ignore_submodules_arg(&default_diff_options, value);
 211
 212         if (!strcmp(var, "diff.submodule")) {
 213                 if (parse_submodule_params(&default_diff_options, value))
 214                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 215                                 value);
 216                 return 0;
 217         }
 218
 219         if (!strcmp(var, "diff.algorithm")) {
 220                 diff_algorithm = parse_algorithm_value(value);
 221                 if (diff_algorithm < 0)
 222                         return -1;
 223                 return 0;
 224         }
 225
 226         if (git_color_config(var, value, cb) < 0)
 227                 return -1;
 228
 229         return git_diff_basic_config(var, value, cb);
 230 }
 231
 232 int git_diff_basic_config(const char *var, const char *value, void *cb)
 233 {
 234         if (!strcmp(var, "diff.renamelimit")) {
 235                 diff_rename_limit_default = git_config_int(var, value);
 236                 return 0;
 237         }
 238
 239         if (userdiff_config(var, value) < 0)
 240                 return -1;
 241
 242         if (starts_with(var, "diff.color.") || starts_with(var, "color.diff.")) {
 243                 int slot = parse_diff_color_slot(var, 11);
 244                 if (slot < 0)
 245                         return 0;
 246                 if (!value)
 247                         return config_error_nonbool(var);
 248                 color_parse(value, var, diff_colors[slot]);
 249                 return 0;
 250         }
 251
 252         /* like GNU diff's --suppress-blank-empty option  */
 253         if (!strcmp(var, "diff.suppressblankempty") ||
 254                         /* for backwards compatibility */
 255                         !strcmp(var, "diff.suppress-blank-empty")) {
 256                 diff_suppress_blank_empty = git_config_bool(var, value);
 257                 return 0;
 258         }
 259
 260         if (!strcmp(var, "diff.dirstat")) {
 261                 struct strbuf errmsg = STRBUF_INIT;
 262                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 263                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 264                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 265                                 errmsg.buf);
 266                 strbuf_release(&errmsg);
 267                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 268                 return 0;
 269         }
 270
 271         if (starts_with(var, "submodule."))
 272                 return parse_submodule_config_option(var, value);
 273
 274         return git_default_config(var, value, cb);
 275 }
 276
 277 static char *quote_two(const char *one, const char *two)
 278 {
 279         int need_one = quote_c_style(one, NULL, NULL, 1);
 280         int need_two = quote_c_style(two, NULL, NULL, 1);
 281         struct strbuf res = STRBUF_INIT;
 282
 283         if (need_one + need_two) {
 284                 strbuf_addch(&res, '"');
 285                 quote_c_style(one, &res, NULL, 1);
 286                 quote_c_style(two, &res, NULL, 1);
 287                 strbuf_addch(&res, '"');
 288         } else {
 289                 strbuf_addstr(&res, one);
 290                 strbuf_addstr(&res, two);
 291         }
 292         return strbuf_detach(&res, NULL);
 293 }
 294
 295 static const char *external_diff(void)
 296 {
 297         static const char *external_diff_cmd = NULL;
 298         static int done_preparing = 0;
 299
 300         if (done_preparing)
 301                 return external_diff_cmd;
 302         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 303         if (!external_diff_cmd)
 304                 external_diff_cmd = external_diff_cmd_cfg;
 305         done_preparing = 1;
 306         return external_diff_cmd;
 307 }
 308
 309 static struct diff_tempfile {
 310         const char *name; /* filename external diff should read from */
 311         char hex[41];
 312         char mode[10];
 313         char tmp_path[PATH_MAX];
 314 } diff_temp[2];
 315
 316 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 317
 318 struct emit_callback {
 319         int color_diff;
 320         unsigned ws_rule;
 321         int blank_at_eof_in_preimage;
 322         int blank_at_eof_in_postimage;
 323         int lno_in_preimage;
 324         int lno_in_postimage;
 325         sane_truncate_fn truncate;
 326         const char **label_path;
 327         struct diff_words_data *diff_words;
 328         struct diff_options *opt;
 329         int *found_changesp;
 330         struct strbuf *header;
 331 };
 332
 333 static int count_lines(const char *data, int size)
 334 {
 335         int count, ch, completely_empty = 1, nl_just_seen = 0;
 336         count = 0;
 337         while (0 < size--) {
 338                 ch = *data++;
 339                 if (ch == '\n') {
 340                         count++;
 341                         nl_just_seen = 1;
 342                         completely_empty = 0;
 343                 }
 344                 else {
 345                         nl_just_seen = 0;
 346                         completely_empty = 0;
 347                 }
 348         }
 349         if (completely_empty)
 350                 return 0;
 351         if (!nl_just_seen)
 352                 count++; /* no trailing newline */
 353         return count;
 354 }
 355
 356 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 357 {
 358         if (!DIFF_FILE_VALID(one)) {
 359                 mf->ptr = (char *)""; /* does not matter */
 360                 mf->size = 0;
 361                 return 0;
 362         }
 363         else if (diff_populate_filespec(one, 0))
 364                 return -1;
 365
 366         mf->ptr = one->data;
 367         mf->size = one->size;
 368         return 0;
 369 }
 370
 371 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 372 static unsigned long diff_filespec_size(struct diff_filespec *one)
 373 {
 374         if (!DIFF_FILE_VALID(one))
 375                 return 0;
 376         diff_populate_filespec(one, 1);
 377         return one->size;
 378 }
 379
 380 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 381 {
 382         char *ptr = mf->ptr;
 383         long size = mf->size;
 384         int cnt = 0;
 385
 386         if (!size)
 387                 return cnt;
 388         ptr += size - 1; /* pointing at the very end */
 389         if (*ptr != '\n')
 390                 ; /* incomplete line */
 391         else
 392                 ptr--; /* skip the last LF */
 393         while (mf->ptr < ptr) {
 394                 char *prev_eol;
 395                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 396                         if (*prev_eol == '\n')
 397                                 break;
 398                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 399                         break;
 400                 cnt++;
 401                 ptr = prev_eol - 1;
 402         }
 403         return cnt;
 404 }
 405
 406 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 407                                struct emit_callback *ecbdata)
 408 {
 409         int l1, l2, at;
 410         unsigned ws_rule = ecbdata->ws_rule;
 411         l1 = count_trailing_blank(mf1, ws_rule);
 412         l2 = count_trailing_blank(mf2, ws_rule);
 413         if (l2 <= l1) {
 414                 ecbdata->blank_at_eof_in_preimage = 0;
 415                 ecbdata->blank_at_eof_in_postimage = 0;
 416                 return;
 417         }
 418         at = count_lines(mf1->ptr, mf1->size);
 419         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 420
 421         at = count_lines(mf2->ptr, mf2->size);
 422         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 423 }
 424
 425 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 426                         int first, const char *line, int len)
 427 {
 428         int has_trailing_newline, has_trailing_carriage_return;
 429         int nofirst;
 430         FILE *file = o->file;
 431
 432         fputs(diff_line_prefix(o), file);
 433
 434         if (len == 0) {
 435                 has_trailing_newline = (first == '\n');
 436                 has_trailing_carriage_return = (!has_trailing_newline &&
 437                                                 (first == '\r'));
 438                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 439         } else {
 440                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 441                 if (has_trailing_newline)
 442                         len--;
 443                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 444                 if (has_trailing_carriage_return)
 445                         len--;
 446                 nofirst = 0;
 447         }
 448
 449         if (len || !nofirst) {
 450                 fputs(set, file);
 451                 if (!nofirst)
 452                         fputc(first, file);
 453                 fwrite(line, len, 1, file);
 454                 fputs(reset, file);
 455         }
 456         if (has_trailing_carriage_return)
 457                 fputc('\r', file);
 458         if (has_trailing_newline)
 459                 fputc('\n', file);
 460 }
 461
 462 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 463                       const char *line, int len)
 464 {
 465         emit_line_0(o, set, reset, line[0], line+1, len-1);
 466 }
 467
 468 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 469 {
 470         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 471               ecbdata->blank_at_eof_in_preimage &&
 472               ecbdata->blank_at_eof_in_postimage &&
 473               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 474               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 475                 return 0;
 476         return ws_blank_line(line, len, ecbdata->ws_rule);
 477 }
 478
 479 static void emit_add_line(const char *reset,
 480                           struct emit_callback *ecbdata,
 481                           const char *line, int len)
 482 {
 483         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 484         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 485
 486         if (!*ws)
 487                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
 488         else if (new_blank_line_at_eof(ecbdata, line, len))
 489                 /* Blank line at EOF - paint '+' as well */
 490                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
 491         else {
 492                 /* Emit just the prefix, then the rest. */
 493                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
 494                 ws_check_emit(line, len, ecbdata->ws_rule,
 495                               ecbdata->opt->file, set, reset, ws);
 496         }
 497 }
 498
 499 static void emit_hunk_header(struct emit_callback *ecbdata,
 500                              const char *line, int len)
 501 {
 502         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 503         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 504         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 505         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 506         static const char atat[2] = { '@', '@' };
 507         const char *cp, *ep;
 508         struct strbuf msgbuf = STRBUF_INIT;
 509         int org_len = len;
 510         int i = 1;
 511
 512         /*
 513          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 514          * it always is at least 10 bytes long.
 515          */
 516         if (len < 10 ||
 517             memcmp(line, atat, 2) ||
 518             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 519                 emit_line(ecbdata->opt, plain, reset, line, len);
 520                 return;
 521         }
 522         ep += 2; /* skip over @@ */
 523
 524         /* The hunk header in fraginfo color */
 525         strbuf_add(&msgbuf, frag, strlen(frag));
 526         strbuf_add(&msgbuf, line, ep - line);
 527         strbuf_add(&msgbuf, reset, strlen(reset));
 528
 529         /*
 530          * trailing "\r\n"
 531          */
 532         for ( ; i < 3; i++)
 533                 if (line[len - i] == '\r' || line[len - i] == '\n')
 534                         len--;
 535
 536         /* blank before the func header */
 537         for (cp = ep; ep - line < len; ep++)
 538                 if (*ep != ' ' && *ep != '\t')
 539                         break;
 540         if (ep != cp) {
 541                 strbuf_add(&msgbuf, plain, strlen(plain));
 542                 strbuf_add(&msgbuf, cp, ep - cp);
 543                 strbuf_add(&msgbuf, reset, strlen(reset));
 544         }
 545
 546         if (ep < line + len) {
 547                 strbuf_add(&msgbuf, func, strlen(func));
 548                 strbuf_add(&msgbuf, ep, line + len - ep);
 549                 strbuf_add(&msgbuf, reset, strlen(reset));
 550         }
 551
 552         strbuf_add(&msgbuf, line + len, org_len - len);
 553         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 554         strbuf_release(&msgbuf);
 555 }
 556
 557 static struct diff_tempfile *claim_diff_tempfile(void) {
 558         int i;
 559         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 560                 if (!diff_temp[i].name)
 561                         return diff_temp + i;
 562         die("BUG: diff is failing to clean up its tempfiles");
 563 }
 564
 565 static int remove_tempfile_installed;
 566
 567 static void remove_tempfile(void)
 568 {
 569         int i;
 570         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 571                 if (diff_temp[i].name == diff_temp[i].tmp_path)
 572                         unlink_or_warn(diff_temp[i].name);
 573                 diff_temp[i].name = NULL;
 574         }
 575 }
 576
 577 static void remove_tempfile_on_signal(int signo)
 578 {
 579         remove_tempfile();
 580         sigchain_pop(signo);
 581         raise(signo);
 582 }
 583
 584 static void print_line_count(FILE *file, int count)
 585 {
 586         switch (count) {
 587         case 0:
 588                 fprintf(file, "0,0");
 589                 break;
 590         case 1:
 591                 fprintf(file, "1");
 592                 break;
 593         default:
 594                 fprintf(file, "1,%d", count);
 595                 break;
 596         }
 597 }
 598
 599 static void emit_rewrite_lines(struct emit_callback *ecb,
 600                                int prefix, const char *data, int size)
 601 {
 602         const char *endp = NULL;
 603         static const char *nneof = " No newline at end of file\n";
 604         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
 605         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 606
 607         while (0 < size) {
 608                 int len;
 609
 610                 endp = memchr(data, '\n', size);
 611                 len = endp ? (endp - data + 1) : size;
 612                 if (prefix != '+') {
 613                         ecb->lno_in_preimage++;
 614                         emit_line_0(ecb->opt, old, reset, '-',
 615                                     data, len);
 616                 } else {
 617                         ecb->lno_in_postimage++;
 618                         emit_add_line(reset, ecb, data, len);
 619                 }
 620                 size -= len;
 621                 data += len;
 622         }
 623         if (!endp) {
 624                 const char *plain = diff_get_color(ecb->color_diff,
 625                                                    DIFF_PLAIN);
 626                 putc('\n', ecb->opt->file);
 627                 emit_line_0(ecb->opt, plain, reset, '\\',
 628                             nneof, strlen(nneof));
 629         }
 630 }
 631
 632 static void emit_rewrite_diff(const char *name_a,
 633                               const char *name_b,
 634                               struct diff_filespec *one,
 635                               struct diff_filespec *two,
 636                               struct userdiff_driver *textconv_one,
 637                               struct userdiff_driver *textconv_two,
 638                               struct diff_options *o)
 639 {
 640         int lc_a, lc_b;
 641         const char *name_a_tab, *name_b_tab;
 642         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 643         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 644         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 645         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 646         const char *a_prefix, *b_prefix;
 647         char *data_one, *data_two;
 648         size_t size_one, size_two;
 649         struct emit_callback ecbdata;
 650         const char *line_prefix = diff_line_prefix(o);
 651
 652         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 653                 a_prefix = o->b_prefix;
 654                 b_prefix = o->a_prefix;
 655         } else {
 656                 a_prefix = o->a_prefix;
 657                 b_prefix = o->b_prefix;
 658         }
 659
 660         name_a += (*name_a == '/');
 661         name_b += (*name_b == '/');
 662         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 663         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 664
 665         strbuf_reset(&a_name);
 666         strbuf_reset(&b_name);
 667         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 668         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 669
 670         size_one = fill_textconv(textconv_one, one, &data_one);
 671         size_two = fill_textconv(textconv_two, two, &data_two);
 672
 673         memset(&ecbdata, 0, sizeof(ecbdata));
 674         ecbdata.color_diff = want_color(o->use_color);
 675         ecbdata.found_changesp = &o->found_changes;
 676         ecbdata.ws_rule = whitespace_rule(name_b);
 677         ecbdata.opt = o;
 678         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 679                 mmfile_t mf1, mf2;
 680                 mf1.ptr = (char *)data_one;
 681                 mf2.ptr = (char *)data_two;
 682                 mf1.size = size_one;
 683                 mf2.size = size_two;
 684                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 685         }
 686         ecbdata.lno_in_preimage = 1;
 687         ecbdata.lno_in_postimage = 1;
 688
 689         lc_a = count_lines(data_one, size_one);
 690         lc_b = count_lines(data_two, size_two);
 691         fprintf(o->file,
 692                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 693                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 694                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 695                 line_prefix, fraginfo);
 696         if (!o->irreversible_delete)
 697                 print_line_count(o->file, lc_a);
 698         else
 699                 fprintf(o->file, "?,?");
 700         fprintf(o->file, " +");
 701         print_line_count(o->file, lc_b);
 702         fprintf(o->file, " @@%s\n", reset);
 703         if (lc_a && !o->irreversible_delete)
 704                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 705         if (lc_b)
 706                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 707         if (textconv_one)
 708                 free((char *)data_one);
 709         if (textconv_two)
 710                 free((char *)data_two);
 711 }
 712
 713 struct diff_words_buffer {
 714         mmfile_t text;
 715         long alloc;
 716         struct diff_words_orig {
 717                 const char *begin, *end;
 718         } *orig;
 719         int orig_nr, orig_alloc;
 720 };
 721
 722 static void diff_words_append(char *line, unsigned long len,
 723                 struct diff_words_buffer *buffer)
 724 {
 725         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 726         line++;
 727         len--;
 728         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 729         buffer->text.size += len;
 730         buffer->text.ptr[buffer->text.size] = '\0';
 731 }
 732
 733 struct diff_words_style_elem {
 734         const char *prefix;
 735         const char *suffix;
 736         const char *color; /* NULL; filled in by the setup code if
 737                             * color is enabled */
 738 };
 739
 740 struct diff_words_style {
 741         enum diff_words_type type;
 742         struct diff_words_style_elem new, old, ctx;
 743         const char *newline;
 744 };
 745
 746 static struct diff_words_style diff_words_styles[] = {
 747         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 748         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 749         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 750 };
 751
 752 struct diff_words_data {
 753         struct diff_words_buffer minus, plus;
 754         const char *current_plus;
 755         int last_minus;
 756         struct diff_options *opt;
 757         regex_t *word_regex;
 758         enum diff_words_type type;
 759         struct diff_words_style *style;
 760 };
 761
 762 static int fn_out_diff_words_write_helper(FILE *fp,
 763                                           struct diff_words_style_elem *st_el,
 764                                           const char *newline,
 765                                           size_t count, const char *buf,
 766                                           const char *line_prefix)
 767 {
 768         int print = 0;
 769
 770         while (count) {
 771                 char *p = memchr(buf, '\n', count);
 772                 if (print)
 773                         fputs(line_prefix, fp);
 774                 if (p != buf) {
 775                         if (st_el->color && fputs(st_el->color, fp) < 0)
 776                                 return -1;
 777                         if (fputs(st_el->prefix, fp) < 0 ||
 778                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 779                             fputs(st_el->suffix, fp) < 0)
 780                                 return -1;
 781                         if (st_el->color && *st_el->color
 782                             && fputs(GIT_COLOR_RESET, fp) < 0)
 783                                 return -1;
 784                 }
 785                 if (!p)
 786                         return 0;
 787                 if (fputs(newline, fp) < 0)
 788                         return -1;
 789                 count -= p + 1 - buf;
 790                 buf = p + 1;
 791                 print = 1;
 792         }
 793         return 0;
 794 }
 795
 796 /*
 797  * '--color-words' algorithm can be described as:
 798  *
 799  *   1. collect a the minus/plus lines of a diff hunk, divided into
 800  *      minus-lines and plus-lines;
 801  *
 802  *   2. break both minus-lines and plus-lines into words and
 803  *      place them into two mmfile_t with one word for each line;
 804  *
 805  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 806  *
 807  * And for the common parts of the both file, we output the plus side text.
 808  * diff_words->current_plus is used to trace the current position of the plus file
 809  * which printed. diff_words->last_minus is used to trace the last minus word
 810  * printed.
 811  *
 812  * For '--graph' to work with '--color-words', we need to output the graph prefix
 813  * on each line of color words output. Generally, there are two conditions on
 814  * which we should output the prefix.
 815  *
 816  *   1. diff_words->last_minus == 0 &&
 817  *      diff_words->current_plus == diff_words->plus.text.ptr
 818  *
 819  *      that is: the plus text must start as a new line, and if there is no minus
 820  *      word printed, a graph prefix must be printed.
 821  *
 822  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 823  *      *(diff_words->current_plus - 1) == '\n'
 824  *
 825  *      that is: a graph prefix must be printed following a '\n'
 826  */
 827 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 828 {
 829         if ((diff_words->last_minus == 0 &&
 830                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 831                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 832                 *(diff_words->current_plus - 1) == '\n')) {
 833                 return 1;
 834         } else {
 835                 return 0;
 836         }
 837 }
 838
 839 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 840 {
 841         struct diff_words_data *diff_words = priv;
 842         struct diff_words_style *style = diff_words->style;
 843         int minus_first, minus_len, plus_first, plus_len;
 844         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 845         struct diff_options *opt = diff_words->opt;
 846         const char *line_prefix;
 847
 848         if (line[0] != '@' || parse_hunk_header(line, len,
 849                         &minus_first, &minus_len, &plus_first, &plus_len))
 850                 return;
 851
 852         assert(opt);
 853         line_prefix = diff_line_prefix(opt);
 854
 855         /* POSIX requires that first be decremented by one if len == 0... */
 856         if (minus_len) {
 857                 minus_begin = diff_words->minus.orig[minus_first].begin;
 858                 minus_end =
 859                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 860         } else
 861                 minus_begin = minus_end =
 862                         diff_words->minus.orig[minus_first].end;
 863
 864         if (plus_len) {
 865                 plus_begin = diff_words->plus.orig[plus_first].begin;
 866                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 867         } else
 868                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 869
 870         if (color_words_output_graph_prefix(diff_words)) {
 871                 fputs(line_prefix, diff_words->opt->file);
 872         }
 873         if (diff_words->current_plus != plus_begin) {
 874                 fn_out_diff_words_write_helper(diff_words->opt->file,
 875                                 &style->ctx, style->newline,
 876                                 plus_begin - diff_words->current_plus,
 877                                 diff_words->current_plus, line_prefix);
 878                 if (*(plus_begin - 1) == '\n')
 879                         fputs(line_prefix, diff_words->opt->file);
 880         }
 881         if (minus_begin != minus_end) {
 882                 fn_out_diff_words_write_helper(diff_words->opt->file,
 883                                 &style->old, style->newline,
 884                                 minus_end - minus_begin, minus_begin,
 885                                 line_prefix);
 886         }
 887         if (plus_begin != plus_end) {
 888                 fn_out_diff_words_write_helper(diff_words->opt->file,
 889                                 &style->new, style->newline,
 890                                 plus_end - plus_begin, plus_begin,
 891                                 line_prefix);
 892         }
 893
 894         diff_words->current_plus = plus_end;
 895         diff_words->last_minus = minus_first;
 896 }
 897
 898 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 899 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 900                 int *begin, int *end)
 901 {
 902         if (word_regex && *begin < buffer->size) {
 903                 regmatch_t match[1];
 904                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 905                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 906                                         '\n', match[0].rm_eo - match[0].rm_so);
 907                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 908                         *begin += match[0].rm_so;
 909                         return *begin >= *end;
 910                 }
 911                 return -1;
 912         }
 913
 914         /* find the next word */
 915         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
 916                 (*begin)++;
 917         if (*begin >= buffer->size)
 918                 return -1;
 919
 920         /* find the end of the word */
 921         *end = *begin + 1;
 922         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
 923                 (*end)++;
 924
 925         return 0;
 926 }
 927
 928 /*
 929  * This function splits the words in buffer->text, stores the list with
 930  * newline separator into out, and saves the offsets of the original words
 931  * in buffer->orig.
 932  */
 933 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
 934                 regex_t *word_regex)
 935 {
 936         int i, j;
 937         long alloc = 0;
 938
 939         out->size = 0;
 940         out->ptr = NULL;
 941
 942         /* fake an empty "0th" word */
 943         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
 944         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
 945         buffer->orig_nr = 1;
 946
 947         for (i = 0; i < buffer->text.size; i++) {
 948                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
 949                         return;
 950
 951                 /* store original boundaries */
 952                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
 953                                 buffer->orig_alloc);
 954                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
 955                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
 956                 buffer->orig_nr++;
 957
 958                 /* store one word */
 959                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
 960                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
 961                 out->ptr[out->size + j - i] = '\n';
 962                 out->size += j - i + 1;
 963
 964                 i = j - 1;
 965         }
 966 }
 967
 968 /* this executes the word diff on the accumulated buffers */
 969 static void diff_words_show(struct diff_words_data *diff_words)
 970 {
 971         xpparam_t xpp;
 972         xdemitconf_t xecfg;
 973         mmfile_t minus, plus;
 974         struct diff_words_style *style = diff_words->style;
 975
 976         struct diff_options *opt = diff_words->opt;
 977         const char *line_prefix;
 978
 979         assert(opt);
 980         line_prefix = diff_line_prefix(opt);
 981
 982         /* special case: only removal */
 983         if (!diff_words->plus.text.size) {
 984                 fputs(line_prefix, diff_words->opt->file);
 985                 fn_out_diff_words_write_helper(diff_words->opt->file,
 986                         &style->old, style->newline,
 987                         diff_words->minus.text.size,
 988                         diff_words->minus.text.ptr, line_prefix);
 989                 diff_words->minus.text.size = 0;
 990                 return;
 991         }
 992
 993         diff_words->current_plus = diff_words->plus.text.ptr;
 994         diff_words->last_minus = 0;
 995
 996         memset(&xpp, 0, sizeof(xpp));
 997         memset(&xecfg, 0, sizeof(xecfg));
 998         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
 999         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1000         xpp.flags = 0;
1001         /* as only the hunk header will be parsed, we need a 0-context */
1002         xecfg.ctxlen = 0;
1003         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1004                       &xpp, &xecfg);
1005         free(minus.ptr);
1006         free(plus.ptr);
1007         if (diff_words->current_plus != diff_words->plus.text.ptr +
1008                         diff_words->plus.text.size) {
1009                 if (color_words_output_graph_prefix(diff_words))
1010                         fputs(line_prefix, diff_words->opt->file);
1011                 fn_out_diff_words_write_helper(diff_words->opt->file,
1012                         &style->ctx, style->newline,
1013                         diff_words->plus.text.ptr + diff_words->plus.text.size
1014                         - diff_words->current_plus, diff_words->current_plus,
1015                         line_prefix);
1016         }
1017         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1018 }
1019
1020 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1021 static void diff_words_flush(struct emit_callback *ecbdata)
1022 {
1023         if (ecbdata->diff_words->minus.text.size ||
1024             ecbdata->diff_words->plus.text.size)
1025                 diff_words_show(ecbdata->diff_words);
1026 }
1027
1028 static void diff_filespec_load_driver(struct diff_filespec *one)
1029 {
1030         /* Use already-loaded driver */
1031         if (one->driver)
1032                 return;
1033
1034         if (S_ISREG(one->mode))
1035                 one->driver = userdiff_find_by_path(one->path);
1036
1037         /* Fallback to default settings */
1038         if (!one->driver)
1039                 one->driver = userdiff_find_by_name("default");
1040 }
1041
1042 static const char *userdiff_word_regex(struct diff_filespec *one)
1043 {
1044         diff_filespec_load_driver(one);
1045         return one->driver->word_regex;
1046 }
1047
1048 static void init_diff_words_data(struct emit_callback *ecbdata,
1049                                  struct diff_options *orig_opts,
1050                                  struct diff_filespec *one,
1051                                  struct diff_filespec *two)
1052 {
1053         int i;
1054         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1055         memcpy(o, orig_opts, sizeof(struct diff_options));
1056
1057         ecbdata->diff_words =
1058                 xcalloc(1, sizeof(struct diff_words_data));
1059         ecbdata->diff_words->type = o->word_diff;
1060         ecbdata->diff_words->opt = o;
1061         if (!o->word_regex)
1062                 o->word_regex = userdiff_word_regex(one);
1063         if (!o->word_regex)
1064                 o->word_regex = userdiff_word_regex(two);
1065         if (!o->word_regex)
1066                 o->word_regex = diff_word_regex_cfg;
1067         if (o->word_regex) {
1068                 ecbdata->diff_words->word_regex = (regex_t *)
1069                         xmalloc(sizeof(regex_t));
1070                 if (regcomp(ecbdata->diff_words->word_regex,
1071                             o->word_regex,
1072                             REG_EXTENDED | REG_NEWLINE))
1073                         die ("Invalid regular expression: %s",
1074                              o->word_regex);
1075         }
1076         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1077                 if (o->word_diff == diff_words_styles[i].type) {
1078                         ecbdata->diff_words->style =
1079                                 &diff_words_styles[i];
1080                         break;
1081                 }
1082         }
1083         if (want_color(o->use_color)) {
1084                 struct diff_words_style *st = ecbdata->diff_words->style;
1085                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1086                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1087                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1088         }
1089 }
1090
1091 static void free_diff_words_data(struct emit_callback *ecbdata)
1092 {
1093         if (ecbdata->diff_words) {
1094                 diff_words_flush(ecbdata);
1095                 free (ecbdata->diff_words->opt);
1096                 free (ecbdata->diff_words->minus.text.ptr);
1097                 free (ecbdata->diff_words->minus.orig);
1098                 free (ecbdata->diff_words->plus.text.ptr);
1099                 free (ecbdata->diff_words->plus.orig);
1100                 if (ecbdata->diff_words->word_regex) {
1101                         regfree(ecbdata->diff_words->word_regex);
1102                         free(ecbdata->diff_words->word_regex);
1103                 }
1104                 free(ecbdata->diff_words);
1105                 ecbdata->diff_words = NULL;
1106         }
1107 }
1108
1109 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1110 {
1111         if (want_color(diff_use_color))
1112                 return diff_colors[ix];
1113         return "";
1114 }
1115
1116 const char *diff_line_prefix(struct diff_options *opt)
1117 {
1118         struct strbuf *msgbuf;
1119         if (!opt->output_prefix)
1120                 return "";
1121
1122         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1123         return msgbuf->buf;
1124 }
1125
1126 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1127 {
1128         const char *cp;
1129         unsigned long allot;
1130         size_t l = len;
1131
1132         if (ecb->truncate)
1133                 return ecb->truncate(line, len);
1134         cp = line;
1135         allot = l;
1136         while (0 < l) {
1137                 (void) utf8_width(&cp, &l);
1138                 if (!cp)
1139                         break; /* truncated in the middle? */
1140         }
1141         return allot - l;
1142 }
1143
1144 static void find_lno(const char *line, struct emit_callback *ecbdata)
1145 {
1146         const char *p;
1147         ecbdata->lno_in_preimage = 0;
1148         ecbdata->lno_in_postimage = 0;
1149         p = strchr(line, '-');
1150         if (!p)
1151                 return; /* cannot happen */
1152         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1153         p = strchr(p, '+');
1154         if (!p)
1155                 return; /* cannot happen */
1156         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1157 }
1158
1159 static void fn_out_consume(void *priv, char *line, unsigned long len)
1160 {
1161         struct emit_callback *ecbdata = priv;
1162         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1163         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1164         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1165         struct diff_options *o = ecbdata->opt;
1166         const char *line_prefix = diff_line_prefix(o);
1167
1168         if (ecbdata->header) {
1169                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1170                 strbuf_reset(ecbdata->header);
1171                 ecbdata->header = NULL;
1172         }
1173         *(ecbdata->found_changesp) = 1;
1174
1175         if (ecbdata->label_path[0]) {
1176                 const char *name_a_tab, *name_b_tab;
1177
1178                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1179                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1180
1181                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1182                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1183                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1184                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1185                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1186         }
1187
1188         if (diff_suppress_blank_empty
1189             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1190                 line[0] = '\n';
1191                 len = 1;
1192         }
1193
1194         if (line[0] == '@') {
1195                 if (ecbdata->diff_words)
1196                         diff_words_flush(ecbdata);
1197                 len = sane_truncate_line(ecbdata, line, len);
1198                 find_lno(line, ecbdata);
1199                 emit_hunk_header(ecbdata, line, len);
1200                 if (line[len-1] != '\n')
1201                         putc('\n', ecbdata->opt->file);
1202                 return;
1203         }
1204
1205         if (len < 1) {
1206                 emit_line(ecbdata->opt, reset, reset, line, len);
1207                 if (ecbdata->diff_words
1208                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1209                         fputs("~\n", ecbdata->opt->file);
1210                 return;
1211         }
1212
1213         if (ecbdata->diff_words) {
1214                 if (line[0] == '-') {
1215                         diff_words_append(line, len,
1216                                           &ecbdata->diff_words->minus);
1217                         return;
1218                 } else if (line[0] == '+') {
1219                         diff_words_append(line, len,
1220                                           &ecbdata->diff_words->plus);
1221                         return;
1222                 } else if (starts_with(line, "\\ ")) {
1223                         /*
1224                          * Eat the "no newline at eof" marker as if we
1225                          * saw a "+" or "-" line with nothing on it,
1226                          * and return without diff_words_flush() to
1227                          * defer processing. If this is the end of
1228                          * preimage, more "+" lines may come after it.
1229                          */
1230                         return;
1231                 }
1232                 diff_words_flush(ecbdata);
1233                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1234                         emit_line(ecbdata->opt, plain, reset, line, len);
1235                         fputs("~\n", ecbdata->opt->file);
1236                 } else {
1237                         /*
1238                          * Skip the prefix character, if any.  With
1239                          * diff_suppress_blank_empty, there may be
1240                          * none.
1241                          */
1242                         if (line[0] != '\n') {
1243                               line++;
1244                               len--;
1245                         }
1246                         emit_line(ecbdata->opt, plain, reset, line, len);
1247                 }
1248                 return;
1249         }
1250
1251         if (line[0] != '+') {
1252                 const char *color =
1253                         diff_get_color(ecbdata->color_diff,
1254                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1255                 ecbdata->lno_in_preimage++;
1256                 if (line[0] == ' ')
1257                         ecbdata->lno_in_postimage++;
1258                 emit_line(ecbdata->opt, color, reset, line, len);
1259         } else {
1260                 ecbdata->lno_in_postimage++;
1261                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1262         }
1263 }
1264
1265 static char *pprint_rename(const char *a, const char *b)
1266 {
1267         const char *old = a;
1268         const char *new = b;
1269         struct strbuf name = STRBUF_INIT;
1270         int pfx_length, sfx_length;
1271         int pfx_adjust_for_slash;
1272         int len_a = strlen(a);
1273         int len_b = strlen(b);
1274         int a_midlen, b_midlen;
1275         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1276         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1277
1278         if (qlen_a || qlen_b) {
1279                 quote_c_style(a, &name, NULL, 0);
1280                 strbuf_addstr(&name, " => ");
1281                 quote_c_style(b, &name, NULL, 0);
1282                 return strbuf_detach(&name, NULL);
1283         }
1284
1285         /* Find common prefix */
1286         pfx_length = 0;
1287         while (*old && *new && *old == *new) {
1288                 if (*old == '/')
1289                         pfx_length = old - a + 1;
1290                 old++;
1291                 new++;
1292         }
1293
1294         /* Find common suffix */
1295         old = a + len_a;
1296         new = b + len_b;
1297         sfx_length = 0;
1298         /*
1299          * If there is a common prefix, it must end in a slash.  In
1300          * that case we let this loop run 1 into the prefix to see the
1301          * same slash.
1302          *
1303          * If there is no common prefix, we cannot do this as it would
1304          * underrun the input strings.
1305          */
1306         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1307         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1308                b + pfx_length - pfx_adjust_for_slash <= new &&
1309                *old == *new) {
1310                 if (*old == '/')
1311                         sfx_length = len_a - (old - a);
1312                 old--;
1313                 new--;
1314         }
1315
1316         /*
1317          * pfx{mid-a => mid-b}sfx
1318          * {pfx-a => pfx-b}sfx
1319          * pfx{sfx-a => sfx-b}
1320          * name-a => name-b
1321          */
1322         a_midlen = len_a - pfx_length - sfx_length;
1323         b_midlen = len_b - pfx_length - sfx_length;
1324         if (a_midlen < 0)
1325                 a_midlen = 0;
1326         if (b_midlen < 0)
1327                 b_midlen = 0;
1328
1329         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1330         if (pfx_length + sfx_length) {
1331                 strbuf_add(&name, a, pfx_length);
1332                 strbuf_addch(&name, '{');
1333         }
1334         strbuf_add(&name, a + pfx_length, a_midlen);
1335         strbuf_addstr(&name, " => ");
1336         strbuf_add(&name, b + pfx_length, b_midlen);
1337         if (pfx_length + sfx_length) {
1338                 strbuf_addch(&name, '}');
1339                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1340         }
1341         return strbuf_detach(&name, NULL);
1342 }
1343
1344 struct diffstat_t {
1345         int nr;
1346         int alloc;
1347         struct diffstat_file {
1348                 char *from_name;
1349                 char *name;
1350                 char *print_name;
1351                 unsigned is_unmerged:1;
1352                 unsigned is_binary:1;
1353                 unsigned is_renamed:1;
1354                 unsigned is_interesting:1;
1355                 uintmax_t added, deleted;
1356         } **files;
1357 };
1358
1359 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1360                                           const char *name_a,
1361                                           const char *name_b)
1362 {
1363         struct diffstat_file *x;
1364         x = xcalloc(sizeof (*x), 1);
1365         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1366         diffstat->files[diffstat->nr++] = x;
1367         if (name_b) {
1368                 x->from_name = xstrdup(name_a);
1369                 x->name = xstrdup(name_b);
1370                 x->is_renamed = 1;
1371         }
1372         else {
1373                 x->from_name = NULL;
1374                 x->name = xstrdup(name_a);
1375         }
1376         return x;
1377 }
1378
1379 static void diffstat_consume(void *priv, char *line, unsigned long len)
1380 {
1381         struct diffstat_t *diffstat = priv;
1382         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1383
1384         if (line[0] == '+')
1385                 x->added++;
1386         else if (line[0] == '-')
1387                 x->deleted++;
1388 }
1389
1390 const char mime_boundary_leader[] = "------------";
1391
1392 static int scale_linear(int it, int width, int max_change)
1393 {
1394         if (!it)
1395                 return 0;
1396         /*
1397          * make sure that at least one '-' or '+' is printed if
1398          * there is any change to this path. The easiest way is to
1399          * scale linearly as if the alloted width is one column shorter
1400          * than it is, and then add 1 to the result.
1401          */
1402         return 1 + (it * (width - 1) / max_change);
1403 }
1404
1405 static void show_name(FILE *file,
1406                       const char *prefix, const char *name, int len)
1407 {
1408         fprintf(file, " %s%-*s |", prefix, len, name);
1409 }
1410
1411 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1412 {
1413         if (cnt <= 0)
1414                 return;
1415         fprintf(file, "%s", set);
1416         while (cnt--)
1417                 putc(ch, file);
1418         fprintf(file, "%s", reset);
1419 }
1420
1421 static void fill_print_name(struct diffstat_file *file)
1422 {
1423         char *pname;
1424
1425         if (file->print_name)
1426                 return;
1427
1428         if (!file->is_renamed) {
1429                 struct strbuf buf = STRBUF_INIT;
1430                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1431                         pname = strbuf_detach(&buf, NULL);
1432                 } else {
1433                         pname = file->name;
1434                         strbuf_release(&buf);
1435                 }
1436         } else {
1437                 pname = pprint_rename(file->from_name, file->name);
1438         }
1439         file->print_name = pname;
1440 }
1441
1442 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1443 {
1444         struct strbuf sb = STRBUF_INIT;
1445         int ret;
1446
1447         if (!files) {
1448                 assert(insertions == 0 && deletions == 0);
1449                 return fprintf(fp, "%s\n", " 0 files changed");
1450         }
1451
1452         strbuf_addf(&sb,
1453                     (files == 1) ? " %d file changed" : " %d files changed",
1454                     files);
1455
1456         /*
1457          * For binary diff, the caller may want to print "x files
1458          * changed" with insertions == 0 && deletions == 0.
1459          *
1460          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1461          * is probably less confusing (i.e skip over "2 files changed
1462          * but nothing about added/removed lines? Is this a bug in Git?").
1463          */
1464         if (insertions || deletions == 0) {
1465                 strbuf_addf(&sb,
1466                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1467                             insertions);
1468         }
1469
1470         if (deletions || insertions == 0) {
1471                 strbuf_addf(&sb,
1472                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1473                             deletions);
1474         }
1475         strbuf_addch(&sb, '\n');
1476         ret = fputs(sb.buf, fp);
1477         strbuf_release(&sb);
1478         return ret;
1479 }
1480
1481 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1482 {
1483         int i, len, add, del, adds = 0, dels = 0;
1484         uintmax_t max_change = 0, max_len = 0;
1485         int total_files = data->nr, count;
1486         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1487         const char *reset, *add_c, *del_c;
1488         const char *line_prefix = "";
1489         int extra_shown = 0;
1490
1491         if (data->nr == 0)
1492                 return;
1493
1494         line_prefix = diff_line_prefix(options);
1495         count = options->stat_count ? options->stat_count : data->nr;
1496
1497         reset = diff_get_color_opt(options, DIFF_RESET);
1498         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1499         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1500
1501         /*
1502          * Find the longest filename and max number of changes
1503          */
1504         for (i = 0; (i < count) && (i < data->nr); i++) {
1505                 struct diffstat_file *file = data->files[i];
1506                 uintmax_t change = file->added + file->deleted;
1507
1508                 if (!file->is_interesting && (change == 0)) {
1509                         count++; /* not shown == room for one more */
1510                         continue;
1511                 }
1512                 fill_print_name(file);
1513                 len = strlen(file->print_name);
1514                 if (max_len < len)
1515                         max_len = len;
1516
1517                 if (file->is_unmerged) {
1518                         /* "Unmerged" is 8 characters */
1519                         bin_width = bin_width < 8 ? 8 : bin_width;
1520                         continue;
1521                 }
1522                 if (file->is_binary) {
1523                         /* "Bin XXX -> YYY bytes" */
1524                         int w = 14 + decimal_width(file->added)
1525                                 + decimal_width(file->deleted);
1526                         bin_width = bin_width < w ? w : bin_width;
1527                         /* Display change counts aligned with "Bin" */
1528                         number_width = 3;
1529                         continue;
1530                 }
1531
1532                 if (max_change < change)
1533                         max_change = change;
1534         }
1535         count = i; /* where we can stop scanning in data->files[] */
1536
1537         /*
1538          * We have width = stat_width or term_columns() columns total.
1539          * We want a maximum of min(max_len, stat_name_width) for the name part.
1540          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1541          * We also need 1 for " " and 4 + decimal_width(max_change)
1542          * for " | NNNN " and one the empty column at the end, altogether
1543          * 6 + decimal_width(max_change).
1544          *
1545          * If there's not enough space, we will use the smaller of
1546          * stat_name_width (if set) and 5/8*width for the filename,
1547          * and the rest for constant elements + graph part, but no more
1548          * than stat_graph_width for the graph part.
1549          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1550          * for the standard terminal size).
1551          *
1552          * In other words: stat_width limits the maximum width, and
1553          * stat_name_width fixes the maximum width of the filename,
1554          * and is also used to divide available columns if there
1555          * aren't enough.
1556          *
1557          * Binary files are displayed with "Bin XXX -> YYY bytes"
1558          * instead of the change count and graph. This part is treated
1559          * similarly to the graph part, except that it is not
1560          * "scaled". If total width is too small to accommodate the
1561          * guaranteed minimum width of the filename part and the
1562          * separators and this message, this message will "overflow"
1563          * making the line longer than the maximum width.
1564          */
1565
1566         if (options->stat_width == -1)
1567                 width = term_columns() - options->output_prefix_length;
1568         else
1569                 width = options->stat_width ? options->stat_width : 80;
1570         number_width = decimal_width(max_change) > number_width ?
1571                 decimal_width(max_change) : number_width;
1572
1573         if (options->stat_graph_width == -1)
1574                 options->stat_graph_width = diff_stat_graph_width;
1575
1576         /*
1577          * Guarantee 3/8*16==6 for the graph part
1578          * and 5/8*16==10 for the filename part
1579          */
1580         if (width < 16 + 6 + number_width)
1581                 width = 16 + 6 + number_width;
1582
1583         /*
1584          * First assign sizes that are wanted, ignoring available width.
1585          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1586          * starting from "XXX" should fit in graph_width.
1587          */
1588         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1589         if (options->stat_graph_width &&
1590             options->stat_graph_width < graph_width)
1591                 graph_width = options->stat_graph_width;
1592
1593         name_width = (options->stat_name_width > 0 &&
1594                       options->stat_name_width < max_len) ?
1595                 options->stat_name_width : max_len;
1596
1597         /*
1598          * Adjust adjustable widths not to exceed maximum width
1599          */
1600         if (name_width + number_width + 6 + graph_width > width) {
1601                 if (graph_width > width * 3/8 - number_width - 6) {
1602                         graph_width = width * 3/8 - number_width - 6;
1603                         if (graph_width < 6)
1604                                 graph_width = 6;
1605                 }
1606
1607                 if (options->stat_graph_width &&
1608                     graph_width > options->stat_graph_width)
1609                         graph_width = options->stat_graph_width;
1610                 if (name_width > width - number_width - 6 - graph_width)
1611                         name_width = width - number_width - 6 - graph_width;
1612                 else
1613                         graph_width = width - number_width - 6 - name_width;
1614         }
1615
1616         /*
1617          * From here name_width is the width of the name area,
1618          * and graph_width is the width of the graph area.
1619          * max_change is used to scale graph properly.
1620          */
1621         for (i = 0; i < count; i++) {
1622                 const char *prefix = "";
1623                 struct diffstat_file *file = data->files[i];
1624                 char *name = file->print_name;
1625                 uintmax_t added = file->added;
1626                 uintmax_t deleted = file->deleted;
1627                 int name_len;
1628
1629                 if (!file->is_interesting && (added + deleted == 0))
1630                         continue;
1631
1632                 /*
1633                  * "scale" the filename
1634                  */
1635                 len = name_width;
1636                 name_len = strlen(name);
1637                 if (name_width < name_len) {
1638                         char *slash;
1639                         prefix = "...";
1640                         len -= 3;
1641                         name += name_len - len;
1642                         slash = strchr(name, '/');
1643                         if (slash)
1644                                 name = slash;
1645                 }
1646
1647                 if (file->is_binary) {
1648                         fprintf(options->file, "%s", line_prefix);
1649                         show_name(options->file, prefix, name, len);
1650                         fprintf(options->file, " %*s", number_width, "Bin");
1651                         if (!added && !deleted) {
1652                                 putc('\n', options->file);
1653                                 continue;
1654                         }
1655                         fprintf(options->file, " %s%"PRIuMAX"%s",
1656                                 del_c, deleted, reset);
1657                         fprintf(options->file, " -> ");
1658                         fprintf(options->file, "%s%"PRIuMAX"%s",
1659                                 add_c, added, reset);
1660                         fprintf(options->file, " bytes");
1661                         fprintf(options->file, "\n");
1662                         continue;
1663                 }
1664                 else if (file->is_unmerged) {
1665                         fprintf(options->file, "%s", line_prefix);
1666                         show_name(options->file, prefix, name, len);
1667                         fprintf(options->file, " Unmerged\n");
1668                         continue;
1669                 }
1670
1671                 /*
1672                  * scale the add/delete
1673                  */
1674                 add = added;
1675                 del = deleted;
1676
1677                 if (graph_width <= max_change) {
1678                         int total = scale_linear(add + del, graph_width, max_change);
1679                         if (total < 2 && add && del)
1680                                 /* width >= 2 due to the sanity check */
1681                                 total = 2;
1682                         if (add < del) {
1683                                 add = scale_linear(add, graph_width, max_change);
1684                                 del = total - add;
1685                         } else {
1686                                 del = scale_linear(del, graph_width, max_change);
1687                                 add = total - del;
1688                         }
1689                 }
1690                 fprintf(options->file, "%s", line_prefix);
1691                 show_name(options->file, prefix, name, len);
1692                 fprintf(options->file, " %*"PRIuMAX"%s",
1693                         number_width, added + deleted,
1694                         added + deleted ? " " : "");
1695                 show_graph(options->file, '+', add, add_c, reset);
1696                 show_graph(options->file, '-', del, del_c, reset);
1697                 fprintf(options->file, "\n");
1698         }
1699
1700         for (i = 0; i < data->nr; i++) {
1701                 struct diffstat_file *file = data->files[i];
1702                 uintmax_t added = file->added;
1703                 uintmax_t deleted = file->deleted;
1704
1705                 if (file->is_unmerged ||
1706                     (!file->is_interesting && (added + deleted == 0))) {
1707                         total_files--;
1708                         continue;
1709                 }
1710
1711                 if (!file->is_binary) {
1712                         adds += added;
1713                         dels += deleted;
1714                 }
1715                 if (i < count)
1716                         continue;
1717                 if (!extra_shown)
1718                         fprintf(options->file, "%s ...\n", line_prefix);
1719                 extra_shown = 1;
1720         }
1721         fprintf(options->file, "%s", line_prefix);
1722         print_stat_summary(options->file, total_files, adds, dels);
1723 }
1724
1725 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1726 {
1727         int i, adds = 0, dels = 0, total_files = data->nr;
1728
1729         if (data->nr == 0)
1730                 return;
1731
1732         for (i = 0; i < data->nr; i++) {
1733                 int added = data->files[i]->added;
1734                 int deleted= data->files[i]->deleted;
1735
1736                 if (data->files[i]->is_unmerged ||
1737                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1738                         total_files--;
1739                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1740                         adds += added;
1741                         dels += deleted;
1742                 }
1743         }
1744         fprintf(options->file, "%s", diff_line_prefix(options));
1745         print_stat_summary(options->file, total_files, adds, dels);
1746 }
1747
1748 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1749 {
1750         int i;
1751
1752         if (data->nr == 0)
1753                 return;
1754
1755         for (i = 0; i < data->nr; i++) {
1756                 struct diffstat_file *file = data->files[i];
1757
1758                 fprintf(options->file, "%s", diff_line_prefix(options));
1759
1760                 if (file->is_binary)
1761                         fprintf(options->file, "-\t-\t");
1762                 else
1763                         fprintf(options->file,
1764                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1765                                 file->added, file->deleted);
1766                 if (options->line_termination) {
1767                         fill_print_name(file);
1768                         if (!file->is_renamed)
1769                                 write_name_quoted(file->name, options->file,
1770                                                   options->line_termination);
1771                         else {
1772                                 fputs(file->print_name, options->file);
1773                                 putc(options->line_termination, options->file);
1774                         }
1775                 } else {
1776                         if (file->is_renamed) {
1777                                 putc('\0', options->file);
1778                                 write_name_quoted(file->from_name, options->file, '\0');
1779                         }
1780                         write_name_quoted(file->name, options->file, '\0');
1781                 }
1782         }
1783 }
1784
1785 struct dirstat_file {
1786         const char *name;
1787         unsigned long changed;
1788 };
1789
1790 struct dirstat_dir {
1791         struct dirstat_file *files;
1792         int alloc, nr, permille, cumulative;
1793 };
1794
1795 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1796                 unsigned long changed, const char *base, int baselen)
1797 {
1798         unsigned long this_dir = 0;
1799         unsigned int sources = 0;
1800         const char *line_prefix = diff_line_prefix(opt);
1801
1802         while (dir->nr) {
1803                 struct dirstat_file *f = dir->files;
1804                 int namelen = strlen(f->name);
1805                 unsigned long this;
1806                 char *slash;
1807
1808                 if (namelen < baselen)
1809                         break;
1810                 if (memcmp(f->name, base, baselen))
1811                         break;
1812                 slash = strchr(f->name + baselen, '/');
1813                 if (slash) {
1814                         int newbaselen = slash + 1 - f->name;
1815                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1816                         sources++;
1817                 } else {
1818                         this = f->changed;
1819                         dir->files++;
1820                         dir->nr--;
1821                         sources += 2;
1822                 }
1823                 this_dir += this;
1824         }
1825
1826         /*
1827          * We don't report dirstat's for
1828          *  - the top level
1829          *  - or cases where everything came from a single directory
1830          *    under this directory (sources == 1).
1831          */
1832         if (baselen && sources != 1) {
1833                 if (this_dir) {
1834                         int permille = this_dir * 1000 / changed;
1835                         if (permille >= dir->permille) {
1836                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1837                                         permille / 10, permille % 10, baselen, base);
1838                                 if (!dir->cumulative)
1839                                         return 0;
1840                         }
1841                 }
1842         }
1843         return this_dir;
1844 }
1845
1846 static int dirstat_compare(const void *_a, const void *_b)
1847 {
1848         const struct dirstat_file *a = _a;
1849         const struct dirstat_file *b = _b;
1850         return strcmp(a->name, b->name);
1851 }
1852
1853 static void show_dirstat(struct diff_options *options)
1854 {
1855         int i;
1856         unsigned long changed;
1857         struct dirstat_dir dir;
1858         struct diff_queue_struct *q = &diff_queued_diff;
1859
1860         dir.files = NULL;
1861         dir.alloc = 0;
1862         dir.nr = 0;
1863         dir.permille = options->dirstat_permille;
1864         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1865
1866         changed = 0;
1867         for (i = 0; i < q->nr; i++) {
1868                 struct diff_filepair *p = q->queue[i];
1869                 const char *name;
1870                 unsigned long copied, added, damage;
1871                 int content_changed;
1872
1873                 name = p->two->path ? p->two->path : p->one->path;
1874
1875                 if (p->one->sha1_valid && p->two->sha1_valid)
1876                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1877                 else
1878                         content_changed = 1;
1879
1880                 if (!content_changed) {
1881                         /*
1882                          * The SHA1 has not changed, so pre-/post-content is
1883                          * identical. We can therefore skip looking at the
1884                          * file contents altogether.
1885                          */
1886                         damage = 0;
1887                         goto found_damage;
1888                 }
1889
1890                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1891                         /*
1892                          * In --dirstat-by-file mode, we don't really need to
1893                          * look at the actual file contents at all.
1894                          * The fact that the SHA1 changed is enough for us to
1895                          * add this file to the list of results
1896                          * (with each file contributing equal damage).
1897                          */
1898                         damage = 1;
1899                         goto found_damage;
1900                 }
1901
1902                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1903                         diff_populate_filespec(p->one, 0);
1904                         diff_populate_filespec(p->two, 0);
1905                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1906                                                &copied, &added);
1907                         diff_free_filespec_data(p->one);
1908                         diff_free_filespec_data(p->two);
1909                 } else if (DIFF_FILE_VALID(p->one)) {
1910                         diff_populate_filespec(p->one, 1);
1911                         copied = added = 0;
1912                         diff_free_filespec_data(p->one);
1913                 } else if (DIFF_FILE_VALID(p->two)) {
1914                         diff_populate_filespec(p->two, 1);
1915                         copied = 0;
1916                         added = p->two->size;
1917                         diff_free_filespec_data(p->two);
1918                 } else
1919                         continue;
1920
1921                 /*
1922                  * Original minus copied is the removed material,
1923                  * added is the new material.  They are both damages
1924                  * made to the preimage.
1925                  * If the resulting damage is zero, we know that
1926                  * diffcore_count_changes() considers the two entries to
1927                  * be identical, but since content_changed is true, we
1928                  * know that there must have been _some_ kind of change,
1929                  * so we force all entries to have damage > 0.
1930                  */
1931                 damage = (p->one->size - copied) + added;
1932                 if (!damage)
1933                         damage = 1;
1934
1935 found_damage:
1936                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1937                 dir.files[dir.nr].name = name;
1938                 dir.files[dir.nr].changed = damage;
1939                 changed += damage;
1940                 dir.nr++;
1941         }
1942
1943         /* This can happen even with many files, if everything was renames */
1944         if (!changed)
1945                 return;
1946
1947         /* Show all directories with more than x% of the changes */
1948         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1949         gather_dirstat(options, &dir, changed, "", 0);
1950 }
1951
1952 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1953 {
1954         int i;
1955         unsigned long changed;
1956         struct dirstat_dir dir;
1957
1958         if (data->nr == 0)
1959                 return;
1960
1961         dir.files = NULL;
1962         dir.alloc = 0;
1963         dir.nr = 0;
1964         dir.permille = options->dirstat_permille;
1965         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1966
1967         changed = 0;
1968         for (i = 0; i < data->nr; i++) {
1969                 struct diffstat_file *file = data->files[i];
1970                 unsigned long damage = file->added + file->deleted;
1971                 if (file->is_binary)
1972                         /*
1973                          * binary files counts bytes, not lines. Must find some
1974                          * way to normalize binary bytes vs. textual lines.
1975                          * The following heuristic assumes that there are 64
1976                          * bytes per "line".
1977                          * This is stupid and ugly, but very cheap...
1978                          */
1979                         damage = (damage + 63) / 64;
1980                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1981                 dir.files[dir.nr].name = file->name;
1982                 dir.files[dir.nr].changed = damage;
1983                 changed += damage;
1984                 dir.nr++;
1985         }
1986
1987         /* This can happen even with many files, if everything was renames */
1988         if (!changed)
1989                 return;
1990
1991         /* Show all directories with more than x% of the changes */
1992         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1993         gather_dirstat(options, &dir, changed, "", 0);
1994 }
1995
1996 static void free_diffstat_info(struct diffstat_t *diffstat)
1997 {
1998         int i;
1999         for (i = 0; i < diffstat->nr; i++) {
2000                 struct diffstat_file *f = diffstat->files[i];
2001                 if (f->name != f->print_name)
2002                         free(f->print_name);
2003                 free(f->name);
2004                 free(f->from_name);
2005                 free(f);
2006         }
2007         free(diffstat->files);
2008 }
2009
2010 struct checkdiff_t {
2011         const char *filename;
2012         int lineno;
2013         int conflict_marker_size;
2014         struct diff_options *o;
2015         unsigned ws_rule;
2016         unsigned status;
2017 };
2018
2019 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2020 {
2021         char firstchar;
2022         int cnt;
2023
2024         if (len < marker_size + 1)
2025                 return 0;
2026         firstchar = line[0];
2027         switch (firstchar) {
2028         case '=': case '>': case '<': case '|':
2029                 break;
2030         default:
2031                 return 0;
2032         }
2033         for (cnt = 1; cnt < marker_size; cnt++)
2034                 if (line[cnt] != firstchar)
2035                         return 0;
2036         /* line[1] thru line[marker_size-1] are same as firstchar */
2037         if (len < marker_size + 1 || !isspace(line[marker_size]))
2038                 return 0;
2039         return 1;
2040 }
2041
2042 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2043 {
2044         struct checkdiff_t *data = priv;
2045         int marker_size = data->conflict_marker_size;
2046         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2047         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2048         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2049         char *err;
2050         const char *line_prefix;
2051
2052         assert(data->o);
2053         line_prefix = diff_line_prefix(data->o);
2054
2055         if (line[0] == '+') {
2056                 unsigned bad;
2057                 data->lineno++;
2058                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2059                         data->status |= 1;
2060                         fprintf(data->o->file,
2061                                 "%s%s:%d: leftover conflict marker\n",
2062                                 line_prefix, data->filename, data->lineno);
2063                 }
2064                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2065                 if (!bad)
2066                         return;
2067                 data->status |= bad;
2068                 err = whitespace_error_string(bad);
2069                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2070                         line_prefix, data->filename, data->lineno, err);
2071                 free(err);
2072                 emit_line(data->o, set, reset, line, 1);
2073                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2074                               data->o->file, set, reset, ws);
2075         } else if (line[0] == ' ') {
2076                 data->lineno++;
2077         } else if (line[0] == '@') {
2078                 char *plus = strchr(line, '+');
2079                 if (plus)
2080                         data->lineno = strtol(plus, NULL, 10) - 1;
2081                 else
2082                         die("invalid diff");
2083         }
2084 }
2085
2086 static unsigned char *deflate_it(char *data,
2087                                  unsigned long size,
2088                                  unsigned long *result_size)
2089 {
2090         int bound;
2091         unsigned char *deflated;
2092         git_zstream stream;
2093
2094         memset(&stream, 0, sizeof(stream));
2095         git_deflate_init(&stream, zlib_compression_level);
2096         bound = git_deflate_bound(&stream, size);
2097         deflated = xmalloc(bound);
2098         stream.next_out = deflated;
2099         stream.avail_out = bound;
2100
2101         stream.next_in = (unsigned char *)data;
2102         stream.avail_in = size;
2103         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2104                 ; /* nothing */
2105         git_deflate_end(&stream);
2106         *result_size = stream.total_out;
2107         return deflated;
2108 }
2109
2110 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2111                                   const char *prefix)
2112 {
2113         void *cp;
2114         void *delta;
2115         void *deflated;
2116         void *data;
2117         unsigned long orig_size;
2118         unsigned long delta_size;
2119         unsigned long deflate_size;
2120         unsigned long data_size;
2121
2122         /* We could do deflated delta, or we could do just deflated two,
2123          * whichever is smaller.
2124          */
2125         delta = NULL;
2126         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2127         if (one->size && two->size) {
2128                 delta = diff_delta(one->ptr, one->size,
2129                                    two->ptr, two->size,
2130                                    &delta_size, deflate_size);
2131                 if (delta) {
2132                         void *to_free = delta;
2133                         orig_size = delta_size;
2134                         delta = deflate_it(delta, delta_size, &delta_size);
2135                         free(to_free);
2136                 }
2137         }
2138
2139         if (delta && delta_size < deflate_size) {
2140                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2141                 free(deflated);
2142                 data = delta;
2143                 data_size = delta_size;
2144         }
2145         else {
2146                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2147                 free(delta);
2148                 data = deflated;
2149                 data_size = deflate_size;
2150         }
2151
2152         /* emit data encoded in base85 */
2153         cp = data;
2154         while (data_size) {
2155                 int bytes = (52 < data_size) ? 52 : data_size;
2156                 char line[70];
2157                 data_size -= bytes;
2158                 if (bytes <= 26)
2159                         line[0] = bytes + 'A' - 1;
2160                 else
2161                         line[0] = bytes - 26 + 'a' - 1;
2162                 encode_85(line + 1, cp, bytes);
2163                 cp = (char *) cp + bytes;
2164                 fprintf(file, "%s", prefix);
2165                 fputs(line, file);
2166                 fputc('\n', file);
2167         }
2168         fprintf(file, "%s\n", prefix);
2169         free(data);
2170 }
2171
2172 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2173                              const char *prefix)
2174 {
2175         fprintf(file, "%sGIT binary patch\n", prefix);
2176         emit_binary_diff_body(file, one, two, prefix);
2177         emit_binary_diff_body(file, two, one, prefix);
2178 }
2179
2180 int diff_filespec_is_binary(struct diff_filespec *one)
2181 {
2182         if (one->is_binary == -1) {
2183                 diff_filespec_load_driver(one);
2184                 if (one->driver->binary != -1)
2185                         one->is_binary = one->driver->binary;
2186                 else {
2187                         if (!one->data && DIFF_FILE_VALID(one))
2188                                 diff_populate_filespec(one, 0);
2189                         if (one->data)
2190                                 one->is_binary = buffer_is_binary(one->data,
2191                                                 one->size);
2192                         if (one->is_binary == -1)
2193                                 one->is_binary = 0;
2194                 }
2195         }
2196         return one->is_binary;
2197 }
2198
2199 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2200 {
2201         diff_filespec_load_driver(one);
2202         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2203 }
2204
2205 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2206 {
2207         if (!options->a_prefix)
2208                 options->a_prefix = a;
2209         if (!options->b_prefix)
2210                 options->b_prefix = b;
2211 }
2212
2213 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2214 {
2215         if (!DIFF_FILE_VALID(one))
2216                 return NULL;
2217
2218         diff_filespec_load_driver(one);
2219         return userdiff_get_textconv(one->driver);
2220 }
2221
2222 static void builtin_diff(const char *name_a,
2223                          const char *name_b,
2224                          struct diff_filespec *one,
2225                          struct diff_filespec *two,
2226                          const char *xfrm_msg,
2227                          int must_show_header,
2228                          struct diff_options *o,
2229                          int complete_rewrite)
2230 {
2231         mmfile_t mf1, mf2;
2232         const char *lbl[2];
2233         char *a_one, *b_two;
2234         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2235         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2236         const char *a_prefix, *b_prefix;
2237         struct userdiff_driver *textconv_one = NULL;
2238         struct userdiff_driver *textconv_two = NULL;
2239         struct strbuf header = STRBUF_INIT;
2240         const char *line_prefix = diff_line_prefix(o);
2241
2242         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2243                         (!one->mode || S_ISGITLINK(one->mode)) &&
2244                         (!two->mode || S_ISGITLINK(two->mode))) {
2245                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2246                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2247                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2248                                 line_prefix,
2249                                 one->sha1, two->sha1, two->dirty_submodule,
2250                                 meta, del, add, reset);
2251                 return;
2252         }
2253
2254         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2255                 textconv_one = get_textconv(one);
2256                 textconv_two = get_textconv(two);
2257         }
2258
2259         diff_set_mnemonic_prefix(o, "a/", "b/");
2260         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2261                 a_prefix = o->b_prefix;
2262                 b_prefix = o->a_prefix;
2263         } else {
2264                 a_prefix = o->a_prefix;
2265                 b_prefix = o->b_prefix;
2266         }
2267
2268         /* Never use a non-valid filename anywhere if at all possible */
2269         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2270         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2271
2272         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2273         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2274         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2275         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2276         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2277         if (lbl[0][0] == '/') {
2278                 /* /dev/null */
2279                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2280                 if (xfrm_msg)
2281                         strbuf_addstr(&header, xfrm_msg);
2282                 must_show_header = 1;
2283         }
2284         else if (lbl[1][0] == '/') {
2285                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2286                 if (xfrm_msg)
2287                         strbuf_addstr(&header, xfrm_msg);
2288                 must_show_header = 1;
2289         }
2290         else {
2291                 if (one->mode != two->mode) {
2292                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2293                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2294                         must_show_header = 1;
2295                 }
2296                 if (xfrm_msg)
2297                         strbuf_addstr(&header, xfrm_msg);
2298
2299                 /*
2300                  * we do not run diff between different kind
2301                  * of objects.
2302                  */
2303                 if ((one->mode ^ two->mode) & S_IFMT)
2304                         goto free_ab_and_return;
2305                 if (complete_rewrite &&
2306                     (textconv_one || !diff_filespec_is_binary(one)) &&
2307                     (textconv_two || !diff_filespec_is_binary(two))) {
2308                         fprintf(o->file, "%s", header.buf);
2309                         strbuf_reset(&header);
2310                         emit_rewrite_diff(name_a, name_b, one, two,
2311                                                 textconv_one, textconv_two, o);
2312                         o->found_changes = 1;
2313                         goto free_ab_and_return;
2314                 }
2315         }
2316
2317         if (o->irreversible_delete && lbl[1][0] == '/') {
2318                 fprintf(o->file, "%s", header.buf);
2319                 strbuf_reset(&header);
2320                 goto free_ab_and_return;
2321         } else if (!DIFF_OPT_TST(o, TEXT) &&
2322             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2323               (!textconv_two && diff_filespec_is_binary(two)) )) {
2324                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2325                         die("unable to read files to diff");
2326                 /* Quite common confusing case */
2327                 if (mf1.size == mf2.size &&
2328                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2329                         if (must_show_header)
2330                                 fprintf(o->file, "%s", header.buf);
2331                         goto free_ab_and_return;
2332                 }
2333                 fprintf(o->file, "%s", header.buf);
2334                 strbuf_reset(&header);
2335                 if (DIFF_OPT_TST(o, BINARY))
2336                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2337                 else
2338                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2339                                 line_prefix, lbl[0], lbl[1]);
2340                 o->found_changes = 1;
2341         } else {
2342                 /* Crazy xdl interfaces.. */
2343                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2344                 xpparam_t xpp;
2345                 xdemitconf_t xecfg;
2346                 struct emit_callback ecbdata;
2347                 const struct userdiff_funcname *pe;
2348
2349                 if (must_show_header) {
2350                         fprintf(o->file, "%s", header.buf);
2351                         strbuf_reset(&header);
2352                 }
2353
2354                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2355                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2356
2357                 pe = diff_funcname_pattern(one);
2358                 if (!pe)
2359                         pe = diff_funcname_pattern(two);
2360
2361                 memset(&xpp, 0, sizeof(xpp));
2362                 memset(&xecfg, 0, sizeof(xecfg));
2363                 memset(&ecbdata, 0, sizeof(ecbdata));
2364                 ecbdata.label_path = lbl;
2365                 ecbdata.color_diff = want_color(o->use_color);
2366                 ecbdata.found_changesp = &o->found_changes;
2367                 ecbdata.ws_rule = whitespace_rule(name_b);
2368                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2369                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2370                 ecbdata.opt = o;
2371                 ecbdata.header = header.len ? &header : NULL;
2372                 xpp.flags = o->xdl_opts;
2373                 xecfg.ctxlen = o->context;
2374                 xecfg.interhunkctxlen = o->interhunkcontext;
2375                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2376                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2377                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2378                 if (pe)
2379                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2380                 if (!diffopts)
2381                         ;
2382                 else if (starts_with(diffopts, "--unified="))
2383                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2384                 else if (starts_with(diffopts, "-u"))
2385                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2386                 if (o->word_diff)
2387                         init_diff_words_data(&ecbdata, o, one, two);
2388                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2389                               &xpp, &xecfg);
2390                 if (o->word_diff)
2391                         free_diff_words_data(&ecbdata);
2392                 if (textconv_one)
2393                         free(mf1.ptr);
2394                 if (textconv_two)
2395                         free(mf2.ptr);
2396                 xdiff_clear_find_func(&xecfg);
2397         }
2398
2399  free_ab_and_return:
2400         strbuf_release(&header);
2401         diff_free_filespec_data(one);
2402         diff_free_filespec_data(two);
2403         free(a_one);
2404         free(b_two);
2405         return;
2406 }
2407
2408 static void builtin_diffstat(const char *name_a, const char *name_b,
2409                              struct diff_filespec *one,
2410                              struct diff_filespec *two,
2411                              struct diffstat_t *diffstat,
2412                              struct diff_options *o,
2413                              struct diff_filepair *p)
2414 {
2415         mmfile_t mf1, mf2;
2416         struct diffstat_file *data;
2417         int same_contents;
2418         int complete_rewrite = 0;
2419
2420         if (!DIFF_PAIR_UNMERGED(p)) {
2421                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2422                         complete_rewrite = 1;
2423         }
2424
2425         data = diffstat_add(diffstat, name_a, name_b);
2426         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2427
2428         if (!one || !two) {
2429                 data->is_unmerged = 1;
2430                 return;
2431         }
2432
2433         same_contents = !hashcmp(one->sha1, two->sha1);
2434
2435         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2436                 data->is_binary = 1;
2437                 if (same_contents) {
2438                         data->added = 0;
2439                         data->deleted = 0;
2440                 } else {
2441                         data->added = diff_filespec_size(two);
2442                         data->deleted = diff_filespec_size(one);
2443                 }
2444         }
2445
2446         else if (complete_rewrite) {
2447                 diff_populate_filespec(one, 0);
2448                 diff_populate_filespec(two, 0);
2449                 data->deleted = count_lines(one->data, one->size);
2450                 data->added = count_lines(two->data, two->size);
2451         }
2452
2453         else if (!same_contents) {
2454                 /* Crazy xdl interfaces.. */
2455                 xpparam_t xpp;
2456                 xdemitconf_t xecfg;
2457
2458                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2459                         die("unable to read files to diff");
2460
2461                 memset(&xpp, 0, sizeof(xpp));
2462                 memset(&xecfg, 0, sizeof(xecfg));
2463                 xpp.flags = o->xdl_opts;
2464                 xecfg.ctxlen = o->context;
2465                 xecfg.interhunkctxlen = o->interhunkcontext;
2466                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2467                               &xpp, &xecfg);
2468         }
2469
2470         diff_free_filespec_data(one);
2471         diff_free_filespec_data(two);
2472 }
2473
2474 static void builtin_checkdiff(const char *name_a, const char *name_b,
2475                               const char *attr_path,
2476                               struct diff_filespec *one,
2477                               struct diff_filespec *two,
2478                               struct diff_options *o)
2479 {
2480         mmfile_t mf1, mf2;
2481         struct checkdiff_t data;
2482
2483         if (!two)
2484                 return;
2485
2486         memset(&data, 0, sizeof(data));
2487         data.filename = name_b ? name_b : name_a;
2488         data.lineno = 0;
2489         data.o = o;
2490         data.ws_rule = whitespace_rule(attr_path);
2491         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2492
2493         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2494                 die("unable to read files to diff");
2495
2496         /*
2497          * All the other codepaths check both sides, but not checking
2498          * the "old" side here is deliberate.  We are checking the newly
2499          * introduced changes, and as long as the "new" side is text, we
2500          * can and should check what it introduces.
2501          */
2502         if (diff_filespec_is_binary(two))
2503                 goto free_and_return;
2504         else {
2505                 /* Crazy xdl interfaces.. */
2506                 xpparam_t xpp;
2507                 xdemitconf_t xecfg;
2508
2509                 memset(&xpp, 0, sizeof(xpp));
2510                 memset(&xecfg, 0, sizeof(xecfg));
2511                 xecfg.ctxlen = 1; /* at least one context line */
2512                 xpp.flags = 0;
2513                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2514                               &xpp, &xecfg);
2515
2516                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2517                         struct emit_callback ecbdata;
2518                         int blank_at_eof;
2519
2520                         ecbdata.ws_rule = data.ws_rule;
2521                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2522                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2523
2524                         if (blank_at_eof) {
2525                                 static char *err;
2526                                 if (!err)
2527                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2528                                 fprintf(o->file, "%s:%d: %s.\n",
2529                                         data.filename, blank_at_eof, err);
2530                                 data.status = 1; /* report errors */
2531                         }
2532                 }
2533         }
2534  free_and_return:
2535         diff_free_filespec_data(one);
2536         diff_free_filespec_data(two);
2537         if (data.status)
2538                 DIFF_OPT_SET(o, CHECK_FAILED);
2539 }
2540
2541 struct diff_filespec *alloc_filespec(const char *path)
2542 {
2543         int namelen = strlen(path);
2544         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2545
2546         memset(spec, 0, sizeof(*spec));
2547         spec->path = (char *)(spec + 1);
2548         memcpy(spec->path, path, namelen+1);
2549         spec->count = 1;
2550         spec->is_binary = -1;
2551         return spec;
2552 }
2553
2554 void free_filespec(struct diff_filespec *spec)
2555 {
2556         if (!--spec->count) {
2557                 diff_free_filespec_data(spec);
2558                 free(spec);
2559         }
2560 }
2561
2562 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2563                    int sha1_valid, unsigned short mode)
2564 {
2565         if (mode) {
2566                 spec->mode = canon_mode(mode);
2567                 hashcpy(spec->sha1, sha1);
2568                 spec->sha1_valid = sha1_valid;
2569         }
2570 }
2571
2572 /*
2573  * Given a name and sha1 pair, if the index tells us the file in
2574  * the work tree has that object contents, return true, so that
2575  * prepare_temp_file() does not have to inflate and extract.
2576  */
2577 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2578 {
2579         const struct cache_entry *ce;
2580         struct stat st;
2581         int pos, len;
2582
2583         /*
2584          * We do not read the cache ourselves here, because the
2585          * benchmark with my previous version that always reads cache
2586          * shows that it makes things worse for diff-tree comparing
2587          * two linux-2.6 kernel trees in an already checked out work
2588          * tree.  This is because most diff-tree comparisons deal with
2589          * only a small number of files, while reading the cache is
2590          * expensive for a large project, and its cost outweighs the
2591          * savings we get by not inflating the object to a temporary
2592          * file.  Practically, this code only helps when we are used
2593          * by diff-cache --cached, which does read the cache before
2594          * calling us.
2595          */
2596         if (!active_cache)
2597                 return 0;
2598
2599         /* We want to avoid the working directory if our caller
2600          * doesn't need the data in a normal file, this system
2601          * is rather slow with its stat/open/mmap/close syscalls,
2602          * and the object is contained in a pack file.  The pack
2603          * is probably already open and will be faster to obtain
2604          * the data through than the working directory.  Loose
2605          * objects however would tend to be slower as they need
2606          * to be individually opened and inflated.
2607          */
2608         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2609                 return 0;
2610
2611         len = strlen(name);
2612         pos = cache_name_pos(name, len);
2613         if (pos < 0)
2614                 return 0;
2615         ce = active_cache[pos];
2616
2617         /*
2618          * This is not the sha1 we are looking for, or
2619          * unreusable because it is not a regular file.
2620          */
2621         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2622                 return 0;
2623
2624         /*
2625          * If ce is marked as "assume unchanged", there is no
2626          * guarantee that work tree matches what we are looking for.
2627          */
2628         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2629                 return 0;
2630
2631         /*
2632          * If ce matches the file in the work tree, we can reuse it.
2633          */
2634         if (ce_uptodate(ce) ||
2635             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2636                 return 1;
2637
2638         return 0;
2639 }
2640
2641 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2642 {
2643         int len;
2644         char *data = xmalloc(100), *dirty = "";
2645
2646         /* Are we looking at the work tree? */
2647         if (s->dirty_submodule)
2648                 dirty = "-dirty";
2649
2650         len = snprintf(data, 100,
2651                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2652         s->data = data;
2653         s->size = len;
2654         s->should_free = 1;
2655         if (size_only) {
2656                 s->data = NULL;
2657                 free(data);
2658         }
2659         return 0;
2660 }
2661
2662 /*
2663  * While doing rename detection and pickaxe operation, we may need to
2664  * grab the data for the blob (or file) for our own in-core comparison.
2665  * diff_filespec has data and size fields for this purpose.
2666  */
2667 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2668 {
2669         int err = 0;
2670         /*
2671          * demote FAIL to WARN to allow inspecting the situation
2672          * instead of refusing.
2673          */
2674         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2675                                     ? SAFE_CRLF_WARN
2676                                     : safe_crlf);
2677
2678         if (!DIFF_FILE_VALID(s))
2679                 die("internal error: asking to populate invalid file.");
2680         if (S_ISDIR(s->mode))
2681                 return -1;
2682
2683         if (s->data)
2684                 return 0;
2685
2686         if (size_only && 0 < s->size)
2687                 return 0;
2688
2689         if (S_ISGITLINK(s->mode))
2690                 return diff_populate_gitlink(s, size_only);
2691
2692         if (!s->sha1_valid ||
2693             reuse_worktree_file(s->path, s->sha1, 0)) {
2694                 struct strbuf buf = STRBUF_INIT;
2695                 struct stat st;
2696                 int fd;
2697
2698                 if (lstat(s->path, &st) < 0) {
2699                         if (errno == ENOENT) {
2700                         err_empty:
2701                                 err = -1;
2702                         empty:
2703                                 s->data = (char *)"";
2704                                 s->size = 0;
2705                                 return err;
2706                         }
2707                 }
2708                 s->size = xsize_t(st.st_size);
2709                 if (!s->size)
2710                         goto empty;
2711                 if (S_ISLNK(st.st_mode)) {
2712                         struct strbuf sb = STRBUF_INIT;
2713
2714                         if (strbuf_readlink(&sb, s->path, s->size))
2715                                 goto err_empty;
2716                         s->size = sb.len;
2717                         s->data = strbuf_detach(&sb, NULL);
2718                         s->should_free = 1;
2719                         return 0;
2720                 }
2721                 if (size_only)
2722                         return 0;
2723                 fd = open(s->path, O_RDONLY);
2724                 if (fd < 0)
2725                         goto err_empty;
2726                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2727                 close(fd);
2728                 s->should_munmap = 1;
2729
2730                 /*
2731                  * Convert from working tree format to canonical git format
2732                  */
2733                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2734                         size_t size = 0;
2735                         munmap(s->data, s->size);
2736                         s->should_munmap = 0;
2737                         s->data = strbuf_detach(&buf, &size);
2738                         s->size = size;
2739                         s->should_free = 1;
2740                 }
2741         }
2742         else {
2743                 enum object_type type;
2744                 if (size_only) {
2745                         type = sha1_object_info(s->sha1, &s->size);
2746                         if (type < 0)
2747                                 die("unable to read %s", sha1_to_hex(s->sha1));
2748                 } else {
2749                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2750                         if (!s->data)
2751                                 die("unable to read %s", sha1_to_hex(s->sha1));
2752                         s->should_free = 1;
2753                 }
2754         }
2755         return 0;
2756 }
2757
2758 void diff_free_filespec_blob(struct diff_filespec *s)
2759 {
2760         if (s->should_free)
2761                 free(s->data);
2762         else if (s->should_munmap)
2763                 munmap(s->data, s->size);
2764
2765         if (s->should_free || s->should_munmap) {
2766                 s->should_free = s->should_munmap = 0;
2767                 s->data = NULL;
2768         }
2769 }
2770
2771 void diff_free_filespec_data(struct diff_filespec *s)
2772 {
2773         diff_free_filespec_blob(s);
2774         free(s->cnt_data);
2775         s->cnt_data = NULL;
2776 }
2777
2778 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2779                            void *blob,
2780                            unsigned long size,
2781                            const unsigned char *sha1,
2782                            int mode)
2783 {
2784         int fd;
2785         struct strbuf buf = STRBUF_INIT;
2786         struct strbuf template = STRBUF_INIT;
2787         char *path_dup = xstrdup(path);
2788         const char *base = basename(path_dup);
2789
2790         /* Generate "XXXXXX_basename.ext" */
2791         strbuf_addstr(&template, "XXXXXX_");
2792         strbuf_addstr(&template, base);
2793
2794         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2795                         strlen(base) + 1);
2796         if (fd < 0)
2797                 die_errno("unable to create temp-file");
2798         if (convert_to_working_tree(path,
2799                         (const char *)blob, (size_t)size, &buf)) {
2800                 blob = buf.buf;
2801                 size = buf.len;
2802         }
2803         if (write_in_full(fd, blob, size) != size)
2804                 die_errno("unable to write temp-file");
2805         close(fd);
2806         temp->name = temp->tmp_path;
2807         strcpy(temp->hex, sha1_to_hex(sha1));
2808         temp->hex[40] = 0;
2809         sprintf(temp->mode, "%06o", mode);
2810         strbuf_release(&buf);
2811         strbuf_release(&template);
2812         free(path_dup);
2813 }
2814
2815 static struct diff_tempfile *prepare_temp_file(const char *name,
2816                 struct diff_filespec *one)
2817 {
2818         struct diff_tempfile *temp = claim_diff_tempfile();
2819
2820         if (!DIFF_FILE_VALID(one)) {
2821         not_a_valid_file:
2822                 /* A '-' entry produces this for file-2, and
2823                  * a '+' entry produces this for file-1.
2824                  */
2825                 temp->name = "/dev/null";
2826                 strcpy(temp->hex, ".");
2827                 strcpy(temp->mode, ".");
2828                 return temp;
2829         }
2830
2831         if (!remove_tempfile_installed) {
2832                 atexit(remove_tempfile);
2833                 sigchain_push_common(remove_tempfile_on_signal);
2834                 remove_tempfile_installed = 1;
2835         }
2836
2837         if (!S_ISGITLINK(one->mode) &&
2838             (!one->sha1_valid ||
2839              reuse_worktree_file(name, one->sha1, 1))) {
2840                 struct stat st;
2841                 if (lstat(name, &st) < 0) {
2842                         if (errno == ENOENT)
2843                                 goto not_a_valid_file;
2844                         die_errno("stat(%s)", name);
2845                 }
2846                 if (S_ISLNK(st.st_mode)) {
2847                         struct strbuf sb = STRBUF_INIT;
2848                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2849                                 die_errno("readlink(%s)", name);
2850                         prep_temp_blob(name, temp, sb.buf, sb.len,
2851                                        (one->sha1_valid ?
2852                                         one->sha1 : null_sha1),
2853                                        (one->sha1_valid ?
2854                                         one->mode : S_IFLNK));
2855                         strbuf_release(&sb);
2856                 }
2857                 else {
2858                         /* we can borrow from the file in the work tree */
2859                         temp->name = name;
2860                         if (!one->sha1_valid)
2861                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2862                         else
2863                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2864                         /* Even though we may sometimes borrow the
2865                          * contents from the work tree, we always want
2866                          * one->mode.  mode is trustworthy even when
2867                          * !(one->sha1_valid), as long as
2868                          * DIFF_FILE_VALID(one).
2869                          */
2870                         sprintf(temp->mode, "%06o", one->mode);
2871                 }
2872                 return temp;
2873         }
2874         else {
2875                 if (diff_populate_filespec(one, 0))
2876                         die("cannot read data blob for %s", one->path);
2877                 prep_temp_blob(name, temp, one->data, one->size,
2878                                one->sha1, one->mode);
2879         }
2880         return temp;
2881 }
2882
2883 /* An external diff command takes:
2884  *
2885  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2886  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2887  *
2888  */
2889 static void run_external_diff(const char *pgm,
2890                               const char *name,
2891                               const char *other,
2892                               struct diff_filespec *one,
2893                               struct diff_filespec *two,
2894                               const char *xfrm_msg,
2895                               int complete_rewrite,
2896                               struct diff_options *o)
2897 {
2898         struct argv_array argv = ARGV_ARRAY_INIT;
2899         int retval;
2900         struct diff_queue_struct *q = &diff_queued_diff;
2901         const char *env[3] = { NULL };
2902         char env_counter[50];
2903         char env_total[50];
2904
2905         if (one && two) {
2906                 struct diff_tempfile *temp_one, *temp_two;
2907                 const char *othername = (other ? other : name);
2908                 temp_one = prepare_temp_file(name, one);
2909                 temp_two = prepare_temp_file(othername, two);
2910                 argv_array_push(&argv, pgm);
2911                 argv_array_push(&argv, name);
2912                 argv_array_push(&argv, temp_one->name);
2913                 argv_array_push(&argv, temp_one->hex);
2914                 argv_array_push(&argv, temp_one->mode);
2915                 argv_array_push(&argv, temp_two->name);
2916                 argv_array_push(&argv, temp_two->hex);
2917                 argv_array_push(&argv, temp_two->mode);
2918                 if (other) {
2919                         argv_array_push(&argv, other);
2920                         argv_array_push(&argv, xfrm_msg);
2921                 }
2922         } else {
2923                 argv_array_push(&argv, pgm);
2924                 argv_array_push(&argv, name);
2925         }
2926         fflush(NULL);
2927
2928         env[0] = env_counter;
2929         snprintf(env_counter, sizeof(env_counter), "GIT_DIFF_PATH_COUNTER=%d",
2930                  ++o->diff_path_counter);
2931         env[1] = env_total;
2932         snprintf(env_total, sizeof(env_total), "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2933
2934         retval = run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env);
2935         remove_tempfile();
2936         argv_array_clear(&argv);
2937         if (retval) {
2938                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2939                 exit(1);
2940         }
2941 }
2942
2943 static int similarity_index(struct diff_filepair *p)
2944 {
2945         return p->score * 100 / MAX_SCORE;
2946 }
2947
2948 static void fill_metainfo(struct strbuf *msg,
2949                           const char *name,
2950                           const char *other,
2951                           struct diff_filespec *one,
2952                           struct diff_filespec *two,
2953                           struct diff_options *o,
2954                           struct diff_filepair *p,
2955                           int *must_show_header,
2956                           int use_color)
2957 {
2958         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2959         const char *reset = diff_get_color(use_color, DIFF_RESET);
2960         const char *line_prefix = diff_line_prefix(o);
2961
2962         *must_show_header = 1;
2963         strbuf_init(msg, PATH_MAX * 2 + 300);
2964         switch (p->status) {
2965         case DIFF_STATUS_COPIED:
2966                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2967                             line_prefix, set, similarity_index(p));
2968                 strbuf_addf(msg, "%s\n%s%scopy from ",
2969                             reset,  line_prefix, set);
2970                 quote_c_style(name, msg, NULL, 0);
2971                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2972                 quote_c_style(other, msg, NULL, 0);
2973                 strbuf_addf(msg, "%s\n", reset);
2974                 break;
2975         case DIFF_STATUS_RENAMED:
2976                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2977                             line_prefix, set, similarity_index(p));
2978                 strbuf_addf(msg, "%s\n%s%srename from ",
2979                             reset, line_prefix, set);
2980                 quote_c_style(name, msg, NULL, 0);
2981                 strbuf_addf(msg, "%s\n%s%srename to ",
2982                             reset, line_prefix, set);
2983                 quote_c_style(other, msg, NULL, 0);
2984                 strbuf_addf(msg, "%s\n", reset);
2985                 break;
2986         case DIFF_STATUS_MODIFIED:
2987                 if (p->score) {
2988                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2989                                     line_prefix,
2990                                     set, similarity_index(p), reset);
2991                         break;
2992                 }
2993                 /* fallthru */
2994         default:
2995                 *must_show_header = 0;
2996         }
2997         if (one && two && hashcmp(one->sha1, two->sha1)) {
2998                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2999
3000                 if (DIFF_OPT_TST(o, BINARY)) {
3001                         mmfile_t mf;
3002                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3003                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3004                                 abbrev = 40;
3005                 }
3006                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3007                             find_unique_abbrev(one->sha1, abbrev));
3008                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3009                 if (one->mode == two->mode)
3010                         strbuf_addf(msg, " %06o", one->mode);
3011                 strbuf_addf(msg, "%s\n", reset);
3012         }
3013 }
3014
3015 static void run_diff_cmd(const char *pgm,
3016                          const char *name,
3017                          const char *other,
3018                          const char *attr_path,
3019                          struct diff_filespec *one,
3020                          struct diff_filespec *two,
3021                          struct strbuf *msg,
3022                          struct diff_options *o,
3023                          struct diff_filepair *p)
3024 {
3025         const char *xfrm_msg = NULL;
3026         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3027         int must_show_header = 0;
3028
3029
3030         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3031                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3032                 if (drv && drv->external)
3033                         pgm = drv->external;
3034         }
3035
3036         if (msg) {
3037                 /*
3038                  * don't use colors when the header is intended for an
3039                  * external diff driver
3040                  */
3041                 fill_metainfo(msg, name, other, one, two, o, p,
3042                               &must_show_header,
3043                               want_color(o->use_color) && !pgm);
3044                 xfrm_msg = msg->len ? msg->buf : NULL;
3045         }
3046
3047         if (pgm) {
3048                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3049                                   complete_rewrite, o);
3050                 return;
3051         }
3052         if (one && two)
3053                 builtin_diff(name, other ? other : name,
3054                              one, two, xfrm_msg, must_show_header,
3055                              o, complete_rewrite);
3056         else
3057                 fprintf(o->file, "* Unmerged path %s\n", name);
3058 }
3059
3060 static void diff_fill_sha1_info(struct diff_filespec *one)
3061 {
3062         if (DIFF_FILE_VALID(one)) {
3063                 if (!one->sha1_valid) {
3064                         struct stat st;
3065                         if (one->is_stdin) {
3066                                 hashcpy(one->sha1, null_sha1);
3067                                 return;
3068                         }
3069                         if (lstat(one->path, &st) < 0)
3070                                 die_errno("stat '%s'", one->path);
3071                         if (index_path(one->sha1, one->path, &st, 0))
3072                                 die("cannot hash %s", one->path);
3073                 }
3074         }
3075         else
3076                 hashclr(one->sha1);
3077 }
3078
3079 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3080 {
3081         /* Strip the prefix but do not molest /dev/null and absolute paths */
3082         if (*namep && **namep != '/') {
3083                 *namep += prefix_length;
3084                 if (**namep == '/')
3085                         ++*namep;
3086         }
3087         if (*otherp && **otherp != '/') {
3088                 *otherp += prefix_length;
3089                 if (**otherp == '/')
3090                         ++*otherp;
3091         }
3092 }
3093
3094 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3095 {
3096         const char *pgm = external_diff();
3097         struct strbuf msg;
3098         struct diff_filespec *one = p->one;
3099         struct diff_filespec *two = p->two;
3100         const char *name;
3101         const char *other;
3102         const char *attr_path;
3103
3104         name  = p->one->path;
3105         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3106         attr_path = name;
3107         if (o->prefix_length)
3108                 strip_prefix(o->prefix_length, &name, &other);
3109
3110         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3111                 pgm = NULL;
3112
3113         if (DIFF_PAIR_UNMERGED(p)) {
3114                 run_diff_cmd(pgm, name, NULL, attr_path,
3115                              NULL, NULL, NULL, o, p);
3116                 return;
3117         }
3118
3119         diff_fill_sha1_info(one);
3120         diff_fill_sha1_info(two);
3121
3122         if (!pgm &&
3123             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3124             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3125                 /*
3126                  * a filepair that changes between file and symlink
3127                  * needs to be split into deletion and creation.
3128                  */
3129                 struct diff_filespec *null = alloc_filespec(two->path);
3130                 run_diff_cmd(NULL, name, other, attr_path,
3131                              one, null, &msg, o, p);
3132                 free(null);
3133                 strbuf_release(&msg);
3134
3135                 null = alloc_filespec(one->path);
3136                 run_diff_cmd(NULL, name, other, attr_path,
3137                              null, two, &msg, o, p);
3138                 free(null);
3139         }
3140         else
3141                 run_diff_cmd(pgm, name, other, attr_path,
3142                              one, two, &msg, o, p);
3143
3144         strbuf_release(&msg);
3145 }
3146
3147 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3148                          struct diffstat_t *diffstat)
3149 {
3150         const char *name;
3151         const char *other;
3152
3153         if (DIFF_PAIR_UNMERGED(p)) {
3154                 /* unmerged */
3155                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3156                 return;
3157         }
3158
3159         name = p->one->path;
3160         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3161
3162         if (o->prefix_length)
3163                 strip_prefix(o->prefix_length, &name, &other);
3164
3165         diff_fill_sha1_info(p->one);
3166         diff_fill_sha1_info(p->two);
3167
3168         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3169 }
3170
3171 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3172 {
3173         const char *name;
3174         const char *other;
3175         const char *attr_path;
3176
3177         if (DIFF_PAIR_UNMERGED(p)) {
3178                 /* unmerged */
3179                 return;
3180         }
3181
3182         name = p->one->path;
3183         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3184         attr_path = other ? other : name;
3185
3186         if (o->prefix_length)
3187                 strip_prefix(o->prefix_length, &name, &other);
3188
3189         diff_fill_sha1_info(p->one);
3190         diff_fill_sha1_info(p->two);
3191
3192         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3193 }
3194
3195 void diff_setup(struct diff_options *options)
3196 {
3197         memcpy(options, &default_diff_options, sizeof(*options));
3198
3199         options->file = stdout;
3200
3201         options->line_termination = '\n';
3202         options->break_opt = -1;
3203         options->rename_limit = -1;
3204         options->dirstat_permille = diff_dirstat_permille_default;
3205         options->context = diff_context_default;
3206         DIFF_OPT_SET(options, RENAME_EMPTY);
3207
3208         options->change = diff_change;
3209         options->add_remove = diff_addremove;
3210         options->use_color = diff_use_color_default;
3211         options->detect_rename = diff_detect_rename_default;
3212         options->xdl_opts |= diff_algorithm;
3213
3214         options->orderfile = diff_order_file_cfg;
3215
3216         if (diff_no_prefix) {
3217                 options->a_prefix = options->b_prefix = "";
3218         } else if (!diff_mnemonic_prefix) {
3219                 options->a_prefix = "a/";
3220                 options->b_prefix = "b/";
3221         }
3222 }
3223
3224 void diff_setup_done(struct diff_options *options)
3225 {
3226         int count = 0;
3227
3228         if (options->set_default)
3229                 options->set_default(options);
3230
3231         if (options->output_format & DIFF_FORMAT_NAME)
3232                 count++;
3233         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3234                 count++;
3235         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3236                 count++;
3237         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3238                 count++;
3239         if (count > 1)
3240                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3241
3242         /*
3243          * Most of the time we can say "there are changes"
3244          * only by checking if there are changed paths, but
3245          * --ignore-whitespace* options force us to look
3246          * inside contents.
3247          */
3248
3249         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3250             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3251             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3252                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3253         else
3254                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3255
3256         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3257                 options->detect_rename = DIFF_DETECT_COPY;
3258
3259         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3260                 options->prefix = NULL;
3261         if (options->prefix)
3262                 options->prefix_length = strlen(options->prefix);
3263         else
3264                 options->prefix_length = 0;
3265
3266         if (options->output_format & (DIFF_FORMAT_NAME |
3267                                       DIFF_FORMAT_NAME_STATUS |
3268                                       DIFF_FORMAT_CHECKDIFF |
3269                                       DIFF_FORMAT_NO_OUTPUT))
3270                 options->output_format &= ~(DIFF_FORMAT_RAW |
3271                                             DIFF_FORMAT_NUMSTAT |
3272                                             DIFF_FORMAT_DIFFSTAT |
3273                                             DIFF_FORMAT_SHORTSTAT |
3274                                             DIFF_FORMAT_DIRSTAT |
3275                                             DIFF_FORMAT_SUMMARY |
3276                                             DIFF_FORMAT_PATCH);
3277
3278         /*
3279          * These cases always need recursive; we do not drop caller-supplied
3280          * recursive bits for other formats here.
3281          */
3282         if (options->output_format & (DIFF_FORMAT_PATCH |
3283                                       DIFF_FORMAT_NUMSTAT |
3284                                       DIFF_FORMAT_DIFFSTAT |
3285                                       DIFF_FORMAT_SHORTSTAT |
3286                                       DIFF_FORMAT_DIRSTAT |
3287                                       DIFF_FORMAT_SUMMARY |
3288                                       DIFF_FORMAT_CHECKDIFF))
3289                 DIFF_OPT_SET(options, RECURSIVE);
3290         /*
3291          * Also pickaxe would not work very well if you do not say recursive
3292          */
3293         if (options->pickaxe)
3294                 DIFF_OPT_SET(options, RECURSIVE);
3295         /*
3296          * When patches are generated, submodules diffed against the work tree
3297          * must be checked for dirtiness too so it can be shown in the output
3298          */
3299         if (options->output_format & DIFF_FORMAT_PATCH)
3300                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3301
3302         if (options->detect_rename && options->rename_limit < 0)
3303                 options->rename_limit = diff_rename_limit_default;
3304         if (options->setup & DIFF_SETUP_USE_CACHE) {
3305                 if (!active_cache)
3306                         /* read-cache does not die even when it fails
3307                          * so it is safe for us to do this here.  Also
3308                          * it does not smudge active_cache or active_nr
3309                          * when it fails, so we do not have to worry about
3310                          * cleaning it up ourselves either.
3311                          */
3312                         read_cache();
3313         }
3314         if (options->abbrev <= 0 || 40 < options->abbrev)
3315                 options->abbrev = 40; /* full */
3316
3317         /*
3318          * It does not make sense to show the first hit we happened
3319          * to have found.  It does not make sense not to return with
3320          * exit code in such a case either.
3321          */
3322         if (DIFF_OPT_TST(options, QUICK)) {
3323                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3324                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3325         }
3326
3327         options->diff_path_counter = 0;
3328
3329         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3330                 die(_("--follow requires exactly one pathspec"));
3331 }
3332
3333 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3334 {
3335         char c, *eq;
3336         int len;
3337
3338         if (*arg != '-')
3339                 return 0;
3340         c = *++arg;
3341         if (!c)
3342                 return 0;
3343         if (c == arg_short) {
3344                 c = *++arg;
3345                 if (!c)
3346                         return 1;
3347                 if (val && isdigit(c)) {
3348                         char *end;
3349                         int n = strtoul(arg, &end, 10);
3350                         if (*end)
3351                                 return 0;
3352                         *val = n;
3353                         return 1;
3354                 }
3355                 return 0;
3356         }
3357         if (c != '-')
3358                 return 0;
3359         arg++;
3360         eq = strchrnul(arg, '=');
3361         len = eq - arg;
3362         if (!len || strncmp(arg, arg_long, len))
3363                 return 0;
3364         if (*eq) {
3365                 int n;
3366                 char *end;
3367                 if (!isdigit(*++eq))
3368                         return 0;
3369                 n = strtoul(eq, &end, 10);
3370                 if (*end)
3371                         return 0;
3372                 *val = n;
3373         }
3374         return 1;
3375 }
3376
3377 static int diff_scoreopt_parse(const char *opt);
3378
3379 static inline int short_opt(char opt, const char **argv,
3380                             const char **optarg)
3381 {
3382         const char *arg = argv[0];
3383         if (arg[0] != '-' || arg[1] != opt)
3384                 return 0;
3385         if (arg[2] != '\0') {
3386                 *optarg = arg + 2;
3387                 return 1;
3388         }
3389         if (!argv[1])
3390                 die("Option '%c' requires a value", opt);
3391         *optarg = argv[1];
3392         return 2;
3393 }
3394
3395 int parse_long_opt(const char *opt, const char **argv,
3396                    const char **optarg)
3397 {
3398         const char *arg = argv[0];
3399         if (arg[0] != '-' || arg[1] != '-')
3400                 return 0;
3401         arg += strlen("--");
3402         if (!starts_with(arg, opt))
3403                 return 0;
3404         arg += strlen(opt);
3405         if (*arg == '=') { /* stuck form: --option=value */
3406                 *optarg = arg + 1;
3407                 return 1;
3408         }
3409         if (*arg != '\0')
3410                 return 0;
3411         /* separate form: --option value */
3412         if (!argv[1])
3413                 die("Option '--%s' requires a value", opt);
3414         *optarg = argv[1];
3415         return 2;
3416 }
3417
3418 static int stat_opt(struct diff_options *options, const char **av)
3419 {
3420         const char *arg = av[0];
3421         char *end;
3422         int width = options->stat_width;
3423         int name_width = options->stat_name_width;
3424         int graph_width = options->stat_graph_width;
3425         int count = options->stat_count;
3426         int argcount = 1;
3427
3428         arg += strlen("--stat");
3429         end = (char *)arg;
3430
3431         switch (*arg) {
3432         case '-':
3433                 if (starts_with(arg, "-width")) {
3434                         arg += strlen("-width");
3435                         if (*arg == '=')
3436                                 width = strtoul(arg + 1, &end, 10);
3437                         else if (!*arg && !av[1])
3438                                 die("Option '--stat-width' requires a value");
3439                         else if (!*arg) {
3440                                 width = strtoul(av[1], &end, 10);
3441                                 argcount = 2;
3442                         }
3443                 } else if (starts_with(arg, "-name-width")) {
3444                         arg += strlen("-name-width");
3445                         if (*arg == '=')
3446                                 name_width = strtoul(arg + 1, &end, 10);
3447                         else if (!*arg && !av[1])
3448                                 die("Option '--stat-name-width' requires a value");
3449                         else if (!*arg) {
3450                                 name_width = strtoul(av[1], &end, 10);
3451                                 argcount = 2;
3452                         }
3453                 } else if (starts_with(arg, "-graph-width")) {
3454                         arg += strlen("-graph-width");
3455                         if (*arg == '=')
3456                                 graph_width = strtoul(arg + 1, &end, 10);
3457                         else if (!*arg && !av[1])
3458                                 die("Option '--stat-graph-width' requires a value");
3459                         else if (!*arg) {
3460                                 graph_width = strtoul(av[1], &end, 10);
3461                                 argcount = 2;
3462                         }
3463                 } else if (starts_with(arg, "-count")) {
3464                         arg += strlen("-count");
3465                         if (*arg == '=')
3466                                 count = strtoul(arg + 1, &end, 10);
3467                         else if (!*arg && !av[1])
3468                                 die("Option '--stat-count' requires a value");
3469                         else if (!*arg) {
3470                                 count = strtoul(av[1], &end, 10);
3471                                 argcount = 2;
3472                         }
3473                 }
3474                 break;
3475         case '=':
3476                 width = strtoul(arg+1, &end, 10);
3477                 if (*end == ',')
3478                         name_width = strtoul(end+1, &end, 10);
3479                 if (*end == ',')
3480                         count = strtoul(end+1, &end, 10);
3481         }
3482
3483         /* Important! This checks all the error cases! */
3484         if (*end)
3485                 return 0;
3486         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3487         options->stat_name_width = name_width;
3488         options->stat_graph_width = graph_width;
3489         options->stat_width = width;
3490         options->stat_count = count;
3491         return argcount;
3492 }
3493
3494 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3495 {
3496         struct strbuf errmsg = STRBUF_INIT;
3497         if (parse_dirstat_params(options, params, &errmsg))
3498                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3499                     errmsg.buf);
3500         strbuf_release(&errmsg);
3501         /*
3502          * The caller knows a dirstat-related option is given from the command
3503          * line; allow it to say "return this_function();"
3504          */
3505         options->output_format |= DIFF_FORMAT_DIRSTAT;
3506         return 1;
3507 }
3508
3509 static int parse_submodule_opt(struct diff_options *options, const char *value)
3510 {
3511         if (parse_submodule_params(options, value))
3512                 die(_("Failed to parse --submodule option parameter: '%s'"),
3513                         value);
3514         return 1;
3515 }
3516
3517 static const char diff_status_letters[] = {
3518         DIFF_STATUS_ADDED,
3519         DIFF_STATUS_COPIED,
3520         DIFF_STATUS_DELETED,
3521         DIFF_STATUS_MODIFIED,
3522         DIFF_STATUS_RENAMED,
3523         DIFF_STATUS_TYPE_CHANGED,
3524         DIFF_STATUS_UNKNOWN,
3525         DIFF_STATUS_UNMERGED,
3526         DIFF_STATUS_FILTER_AON,
3527         DIFF_STATUS_FILTER_BROKEN,
3528         '\0',
3529 };
3530
3531 static unsigned int filter_bit['Z' + 1];
3532
3533 static void prepare_filter_bits(void)
3534 {
3535         int i;
3536
3537         if (!filter_bit[DIFF_STATUS_ADDED]) {
3538                 for (i = 0; diff_status_letters[i]; i++)
3539                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3540         }
3541 }
3542
3543 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3544 {
3545         return opt->filter & filter_bit[(int) status];
3546 }
3547
3548 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3549 {
3550         int i, optch;
3551
3552         prepare_filter_bits();
3553
3554         /*
3555          * If there is a negation e.g. 'd' in the input, and we haven't
3556          * initialized the filter field with another --diff-filter, start
3557          * from full set of bits, except for AON.
3558          */
3559         if (!opt->filter) {
3560                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3561                         if (optch < 'a' || 'z' < optch)
3562                                 continue;
3563                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3564                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3565                         break;
3566                 }
3567         }
3568
3569         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3570                 unsigned int bit;
3571                 int negate;
3572
3573                 if ('a' <= optch && optch <= 'z') {
3574                         negate = 1;
3575                         optch = toupper(optch);
3576                 } else {
3577                         negate = 0;
3578                 }
3579
3580                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3581                 if (!bit)
3582                         return optarg[i];
3583                 if (negate)
3584                         opt->filter &= ~bit;
3585                 else
3586                         opt->filter |= bit;
3587         }
3588         return 0;
3589 }
3590
3591 static void enable_patch_output(int *fmt) {
3592         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3593         *fmt |= DIFF_FORMAT_PATCH;
3594 }
3595
3596 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3597 {
3598         const char *arg = av[0];
3599         const char *optarg;
3600         int argcount;
3601
3602         /* Output format options */
3603         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3604             || opt_arg(arg, 'U', "unified", &options->context))
3605                 enable_patch_output(&options->output_format);
3606         else if (!strcmp(arg, "--raw"))
3607                 options->output_format |= DIFF_FORMAT_RAW;
3608         else if (!strcmp(arg, "--patch-with-raw")) {
3609                 enable_patch_output(&options->output_format);
3610                 options->output_format |= DIFF_FORMAT_RAW;
3611         } else if (!strcmp(arg, "--numstat"))
3612                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3613         else if (!strcmp(arg, "--shortstat"))
3614                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3615         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3616                 return parse_dirstat_opt(options, "");
3617         else if (starts_with(arg, "-X"))
3618                 return parse_dirstat_opt(options, arg + 2);
3619         else if (starts_with(arg, "--dirstat="))
3620                 return parse_dirstat_opt(options, arg + 10);
3621         else if (!strcmp(arg, "--cumulative"))
3622                 return parse_dirstat_opt(options, "cumulative");
3623         else if (!strcmp(arg, "--dirstat-by-file"))
3624                 return parse_dirstat_opt(options, "files");
3625         else if (starts_with(arg, "--dirstat-by-file=")) {
3626                 parse_dirstat_opt(options, "files");
3627                 return parse_dirstat_opt(options, arg + 18);
3628         }
3629         else if (!strcmp(arg, "--check"))
3630                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3631         else if (!strcmp(arg, "--summary"))
3632                 options->output_format |= DIFF_FORMAT_SUMMARY;
3633         else if (!strcmp(arg, "--patch-with-stat")) {
3634                 enable_patch_output(&options->output_format);
3635                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3636         } else if (!strcmp(arg, "--name-only"))
3637                 options->output_format |= DIFF_FORMAT_NAME;
3638         else if (!strcmp(arg, "--name-status"))
3639                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3640         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3641                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3642         else if (starts_with(arg, "--stat"))
3643                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3644                 return stat_opt(options, av);
3645
3646         /* renames options */
3647         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3648                  !strcmp(arg, "--break-rewrites")) {
3649                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3650                         return error("invalid argument to -B: %s", arg+2);
3651         }
3652         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3653                  !strcmp(arg, "--find-renames")) {
3654                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3655                         return error("invalid argument to -M: %s", arg+2);
3656                 options->detect_rename = DIFF_DETECT_RENAME;
3657         }
3658         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3659                 options->irreversible_delete = 1;
3660         }
3661         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3662                  !strcmp(arg, "--find-copies")) {
3663                 if (options->detect_rename == DIFF_DETECT_COPY)
3664                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3665                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3666                         return error("invalid argument to -C: %s", arg+2);
3667                 options->detect_rename = DIFF_DETECT_COPY;
3668         }
3669         else if (!strcmp(arg, "--no-renames"))
3670                 options->detect_rename = 0;
3671         else if (!strcmp(arg, "--rename-empty"))
3672                 DIFF_OPT_SET(options, RENAME_EMPTY);
3673         else if (!strcmp(arg, "--no-rename-empty"))
3674                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3675         else if (!strcmp(arg, "--relative"))
3676                 DIFF_OPT_SET(options, RELATIVE_NAME);
3677         else if (starts_with(arg, "--relative=")) {
3678                 DIFF_OPT_SET(options, RELATIVE_NAME);
3679                 options->prefix = arg + 11;
3680         }
3681
3682         /* xdiff options */
3683         else if (!strcmp(arg, "--minimal"))
3684                 DIFF_XDL_SET(options, NEED_MINIMAL);
3685         else if (!strcmp(arg, "--no-minimal"))
3686                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3687         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3688                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3689         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3690                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3691         else if (!strcmp(arg, "--ignore-space-at-eol"))
3692                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3693         else if (!strcmp(arg, "--ignore-blank-lines"))
3694                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3695         else if (!strcmp(arg, "--patience"))
3696                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3697         else if (!strcmp(arg, "--histogram"))
3698                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3699         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3700                 long value = parse_algorithm_value(optarg);
3701                 if (value < 0)
3702                         return error("option diff-algorithm accepts \"myers\", "
3703                                      "\"minimal\", \"patience\" and \"histogram\"");
3704                 /* clear out previous settings */
3705                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3706                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3707                 options->xdl_opts |= value;
3708                 return argcount;
3709         }
3710
3711         /* flags options */
3712         else if (!strcmp(arg, "--binary")) {
3713                 enable_patch_output(&options->output_format);
3714                 DIFF_OPT_SET(options, BINARY);
3715         }
3716         else if (!strcmp(arg, "--full-index"))
3717                 DIFF_OPT_SET(options, FULL_INDEX);
3718         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3719                 DIFF_OPT_SET(options, TEXT);
3720         else if (!strcmp(arg, "-R"))
3721                 DIFF_OPT_SET(options, REVERSE_DIFF);
3722         else if (!strcmp(arg, "--find-copies-harder"))
3723                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3724         else if (!strcmp(arg, "--follow"))
3725                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3726         else if (!strcmp(arg, "--no-follow"))
3727                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3728         else if (!strcmp(arg, "--color"))
3729                 options->use_color = 1;
3730         else if (starts_with(arg, "--color=")) {
3731                 int value = git_config_colorbool(NULL, arg+8);
3732                 if (value < 0)
3733                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3734                 options->use_color = value;
3735         }
3736         else if (!strcmp(arg, "--no-color"))
3737                 options->use_color = 0;
3738         else if (!strcmp(arg, "--color-words")) {
3739                 options->use_color = 1;
3740                 options->word_diff = DIFF_WORDS_COLOR;
3741         }
3742         else if (starts_with(arg, "--color-words=")) {
3743                 options->use_color = 1;
3744                 options->word_diff = DIFF_WORDS_COLOR;
3745                 options->word_regex = arg + 14;
3746         }
3747         else if (!strcmp(arg, "--word-diff")) {
3748                 if (options->word_diff == DIFF_WORDS_NONE)
3749                         options->word_diff = DIFF_WORDS_PLAIN;
3750         }
3751         else if (starts_with(arg, "--word-diff=")) {
3752                 const char *type = arg + 12;
3753                 if (!strcmp(type, "plain"))
3754                         options->word_diff = DIFF_WORDS_PLAIN;
3755                 else if (!strcmp(type, "color")) {
3756                         options->use_color = 1;
3757                         options->word_diff = DIFF_WORDS_COLOR;
3758                 }
3759                 else if (!strcmp(type, "porcelain"))
3760                         options->word_diff = DIFF_WORDS_PORCELAIN;
3761                 else if (!strcmp(type, "none"))
3762                         options->word_diff = DIFF_WORDS_NONE;
3763                 else
3764                         die("bad --word-diff argument: %s", type);
3765         }
3766         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3767                 if (options->word_diff == DIFF_WORDS_NONE)
3768                         options->word_diff = DIFF_WORDS_PLAIN;
3769                 options->word_regex = optarg;
3770                 return argcount;
3771         }
3772         else if (!strcmp(arg, "--exit-code"))
3773                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3774         else if (!strcmp(arg, "--quiet"))
3775                 DIFF_OPT_SET(options, QUICK);
3776         else if (!strcmp(arg, "--ext-diff"))
3777                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3778         else if (!strcmp(arg, "--no-ext-diff"))
3779                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3780         else if (!strcmp(arg, "--textconv"))
3781                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3782         else if (!strcmp(arg, "--no-textconv"))
3783                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3784         else if (!strcmp(arg, "--ignore-submodules")) {
3785                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3786                 handle_ignore_submodules_arg(options, "all");
3787         } else if (starts_with(arg, "--ignore-submodules=")) {
3788                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3789                 handle_ignore_submodules_arg(options, arg + 20);
3790         } else if (!strcmp(arg, "--submodule"))
3791                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3792         else if (starts_with(arg, "--submodule="))
3793                 return parse_submodule_opt(options, arg + 12);
3794
3795         /* misc options */
3796         else if (!strcmp(arg, "-z"))
3797                 options->line_termination = 0;
3798         else if ((argcount = short_opt('l', av, &optarg))) {
3799                 options->rename_limit = strtoul(optarg, NULL, 10);
3800                 return argcount;
3801         }
3802         else if ((argcount = short_opt('S', av, &optarg))) {
3803                 options->pickaxe = optarg;
3804                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3805                 return argcount;
3806         } else if ((argcount = short_opt('G', av, &optarg))) {
3807                 options->pickaxe = optarg;
3808                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3809                 return argcount;
3810         }
3811         else if (!strcmp(arg, "--pickaxe-all"))
3812                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3813         else if (!strcmp(arg, "--pickaxe-regex"))
3814                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3815         else if ((argcount = short_opt('O', av, &optarg))) {
3816                 options->orderfile = optarg;
3817                 return argcount;
3818         }
3819         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3820                 int offending = parse_diff_filter_opt(optarg, options);
3821                 if (offending)
3822                         die("unknown change class '%c' in --diff-filter=%s",
3823                             offending, optarg);
3824                 return argcount;
3825         }
3826         else if (!strcmp(arg, "--abbrev"))
3827                 options->abbrev = DEFAULT_ABBREV;
3828         else if (starts_with(arg, "--abbrev=")) {
3829                 options->abbrev = strtoul(arg + 9, NULL, 10);
3830                 if (options->abbrev < MINIMUM_ABBREV)
3831                         options->abbrev = MINIMUM_ABBREV;
3832                 else if (40 < options->abbrev)
3833                         options->abbrev = 40;
3834         }
3835         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3836                 options->a_prefix = optarg;
3837                 return argcount;
3838         }
3839         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3840                 options->b_prefix = optarg;
3841                 return argcount;
3842         }
3843         else if (!strcmp(arg, "--no-prefix"))
3844                 options->a_prefix = options->b_prefix = "";
3845         else if (opt_arg(arg, '\0', "inter-hunk-context",
3846                          &options->interhunkcontext))
3847                 ;
3848         else if (!strcmp(arg, "-W"))
3849                 DIFF_OPT_SET(options, FUNCCONTEXT);
3850         else if (!strcmp(arg, "--function-context"))
3851                 DIFF_OPT_SET(options, FUNCCONTEXT);
3852         else if (!strcmp(arg, "--no-function-context"))
3853                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3854         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3855                 options->file = fopen(optarg, "w");
3856                 if (!options->file)
3857                         die_errno("Could not open '%s'", optarg);
3858                 options->close_file = 1;
3859                 return argcount;
3860         } else
3861                 return 0;
3862         return 1;
3863 }
3864
3865 int parse_rename_score(const char **cp_p)
3866 {
3867         unsigned long num, scale;
3868         int ch, dot;
3869         const char *cp = *cp_p;
3870
3871         num = 0;
3872         scale = 1;
3873         dot = 0;
3874         for (;;) {
3875                 ch = *cp;
3876                 if ( !dot && ch == '.' ) {
3877                         scale = 1;
3878                         dot = 1;
3879                 } else if ( ch == '%' ) {
3880                         scale = dot ? scale*100 : 100;
3881                         cp++;   /* % is always at the end */
3882                         break;
3883                 } else if ( ch >= '0' && ch <= '9' ) {
3884                         if ( scale < 100000 ) {
3885                                 scale *= 10;
3886                                 num = (num*10) + (ch-'0');
3887                         }
3888                 } else {
3889                         break;
3890                 }
3891                 cp++;
3892         }
3893         *cp_p = cp;
3894
3895         /* user says num divided by scale and we say internally that
3896          * is MAX_SCORE * num / scale.
3897          */
3898         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3899 }
3900
3901 static int diff_scoreopt_parse(const char *opt)
3902 {
3903         int opt1, opt2, cmd;
3904
3905         if (*opt++ != '-')
3906                 return -1;
3907         cmd = *opt++;
3908         if (cmd == '-') {
3909                 /* convert the long-form arguments into short-form versions */
3910                 if (starts_with(opt, "break-rewrites")) {
3911                         opt += strlen("break-rewrites");
3912                         if (*opt == 0 || *opt++ == '=')
3913                                 cmd = 'B';
3914                 } else if (starts_with(opt, "find-copies")) {
3915                         opt += strlen("find-copies");
3916                         if (*opt == 0 || *opt++ == '=')
3917                                 cmd = 'C';
3918                 } else if (starts_with(opt, "find-renames")) {
3919                         opt += strlen("find-renames");
3920                         if (*opt == 0 || *opt++ == '=')
3921                                 cmd = 'M';
3922                 }
3923         }
3924         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3925                 return -1; /* that is not a -M, -C, or -B option */
3926
3927         opt1 = parse_rename_score(&opt);
3928         if (cmd != 'B')
3929                 opt2 = 0;
3930         else {
3931                 if (*opt == 0)
3932                         opt2 = 0;
3933                 else if (*opt != '/')
3934                         return -1; /* we expect -B80/99 or -B80 */
3935                 else {
3936                         opt++;
3937                         opt2 = parse_rename_score(&opt);
3938                 }
3939         }
3940         if (*opt != 0)
3941                 return -1;
3942         return opt1 | (opt2 << 16);
3943 }
3944
3945 struct diff_queue_struct diff_queued_diff;
3946
3947 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3948 {
3949         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
3950         queue->queue[queue->nr++] = dp;
3951 }
3952
3953 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3954                                  struct diff_filespec *one,
3955                                  struct diff_filespec *two)
3956 {
3957         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3958         dp->one = one;
3959         dp->two = two;
3960         if (queue)
3961                 diff_q(queue, dp);
3962         return dp;
3963 }
3964
3965 void diff_free_filepair(struct diff_filepair *p)
3966 {
3967         free_filespec(p->one);
3968         free_filespec(p->two);
3969         free(p);
3970 }
3971
3972 /* This is different from find_unique_abbrev() in that
3973  * it stuffs the result with dots for alignment.
3974  */
3975 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3976 {
3977         int abblen;
3978         const char *abbrev;
3979         if (len == 40)
3980                 return sha1_to_hex(sha1);
3981
3982         abbrev = find_unique_abbrev(sha1, len);
3983         abblen = strlen(abbrev);
3984         if (abblen < 37) {
3985                 static char hex[41];
3986                 if (len < abblen && abblen <= len + 2)
3987                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3988                 else
3989                         sprintf(hex, "%s...", abbrev);
3990                 return hex;
3991         }
3992         return sha1_to_hex(sha1);
3993 }
3994
3995 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3996 {
3997         int line_termination = opt->line_termination;
3998         int inter_name_termination = line_termination ? '\t' : '\0';
3999
4000         fprintf(opt->file, "%s", diff_line_prefix(opt));
4001         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4002                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4003                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4004                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4005         }
4006         if (p->score) {
4007                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4008                         inter_name_termination);
4009         } else {
4010                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4011         }
4012
4013         if (p->status == DIFF_STATUS_COPIED ||
4014             p->status == DIFF_STATUS_RENAMED) {
4015                 const char *name_a, *name_b;
4016                 name_a = p->one->path;
4017                 name_b = p->two->path;
4018                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4019                 write_name_quoted(name_a, opt->file, inter_name_termination);
4020                 write_name_quoted(name_b, opt->file, line_termination);
4021         } else {
4022                 const char *name_a, *name_b;
4023                 name_a = p->one->mode ? p->one->path : p->two->path;
4024                 name_b = NULL;
4025                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4026                 write_name_quoted(name_a, opt->file, line_termination);
4027         }
4028 }
4029
4030 int diff_unmodified_pair(struct diff_filepair *p)
4031 {
4032         /* This function is written stricter than necessary to support
4033          * the currently implemented transformers, but the idea is to
4034          * let transformers to produce diff_filepairs any way they want,
4035          * and filter and clean them up here before producing the output.
4036          */
4037         struct diff_filespec *one = p->one, *two = p->two;
4038
4039         if (DIFF_PAIR_UNMERGED(p))
4040                 return 0; /* unmerged is interesting */
4041
4042         /* deletion, addition, mode or type change
4043          * and rename are all interesting.
4044          */
4045         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4046             DIFF_PAIR_MODE_CHANGED(p) ||
4047             strcmp(one->path, two->path))
4048                 return 0;
4049
4050         /* both are valid and point at the same path.  that is, we are
4051          * dealing with a change.
4052          */
4053         if (one->sha1_valid && two->sha1_valid &&
4054             !hashcmp(one->sha1, two->sha1) &&
4055             !one->dirty_submodule && !two->dirty_submodule)
4056                 return 1; /* no change */
4057         if (!one->sha1_valid && !two->sha1_valid)
4058                 return 1; /* both look at the same file on the filesystem. */
4059         return 0;
4060 }
4061
4062 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4063 {
4064         if (diff_unmodified_pair(p))
4065                 return;
4066
4067         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4068             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4069                 return; /* no tree diffs in patch format */
4070
4071         run_diff(p, o);
4072 }
4073
4074 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4075                             struct diffstat_t *diffstat)
4076 {
4077         if (diff_unmodified_pair(p))
4078                 return;
4079
4080         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4081             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4082                 return; /* no useful stat for tree diffs */
4083
4084         run_diffstat(p, o, diffstat);
4085 }
4086
4087 static void diff_flush_checkdiff(struct diff_filepair *p,
4088                 struct diff_options *o)
4089 {
4090         if (diff_unmodified_pair(p))
4091                 return;
4092
4093         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4094             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4095                 return; /* nothing to check in tree diffs */
4096
4097         run_checkdiff(p, o);
4098 }
4099
4100 int diff_queue_is_empty(void)
4101 {
4102         struct diff_queue_struct *q = &diff_queued_diff;
4103         int i;
4104         for (i = 0; i < q->nr; i++)
4105                 if (!diff_unmodified_pair(q->queue[i]))
4106                         return 0;
4107         return 1;
4108 }
4109
4110 #if DIFF_DEBUG
4111 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4112 {
4113         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4114                 x, one ? one : "",
4115                 s->path,
4116                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4117                 s->mode,
4118                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4119         fprintf(stderr, "queue[%d] %s size %lu\n",
4120                 x, one ? one : "",
4121                 s->size);
4122 }
4123
4124 void diff_debug_filepair(const struct diff_filepair *p, int i)
4125 {
4126         diff_debug_filespec(p->one, i, "one");
4127         diff_debug_filespec(p->two, i, "two");
4128         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4129                 p->score, p->status ? p->status : '?',
4130                 p->one->rename_used, p->broken_pair);
4131 }
4132
4133 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4134 {
4135         int i;
4136         if (msg)
4137                 fprintf(stderr, "%s\n", msg);
4138         fprintf(stderr, "q->nr = %d\n", q->nr);
4139         for (i = 0; i < q->nr; i++) {
4140                 struct diff_filepair *p = q->queue[i];
4141                 diff_debug_filepair(p, i);
4142         }
4143 }
4144 #endif
4145
4146 static void diff_resolve_rename_copy(void)
4147 {
4148         int i;
4149         struct diff_filepair *p;
4150         struct diff_queue_struct *q = &diff_queued_diff;
4151
4152         diff_debug_queue("resolve-rename-copy", q);
4153
4154         for (i = 0; i < q->nr; i++) {
4155                 p = q->queue[i];
4156                 p->status = 0; /* undecided */
4157                 if (DIFF_PAIR_UNMERGED(p))
4158                         p->status = DIFF_STATUS_UNMERGED;
4159                 else if (!DIFF_FILE_VALID(p->one))
4160                         p->status = DIFF_STATUS_ADDED;
4161                 else if (!DIFF_FILE_VALID(p->two))
4162                         p->status = DIFF_STATUS_DELETED;
4163                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4164                         p->status = DIFF_STATUS_TYPE_CHANGED;
4165
4166                 /* from this point on, we are dealing with a pair
4167                  * whose both sides are valid and of the same type, i.e.
4168                  * either in-place edit or rename/copy edit.
4169                  */
4170                 else if (DIFF_PAIR_RENAME(p)) {
4171                         /*
4172                          * A rename might have re-connected a broken
4173                          * pair up, causing the pathnames to be the
4174                          * same again. If so, that's not a rename at
4175                          * all, just a modification..
4176                          *
4177                          * Otherwise, see if this source was used for
4178                          * multiple renames, in which case we decrement
4179                          * the count, and call it a copy.
4180                          */
4181                         if (!strcmp(p->one->path, p->two->path))
4182                                 p->status = DIFF_STATUS_MODIFIED;
4183                         else if (--p->one->rename_used > 0)
4184                                 p->status = DIFF_STATUS_COPIED;
4185                         else
4186                                 p->status = DIFF_STATUS_RENAMED;
4187                 }
4188                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4189                          p->one->mode != p->two->mode ||
4190                          p->one->dirty_submodule ||
4191                          p->two->dirty_submodule ||
4192                          is_null_sha1(p->one->sha1))
4193                         p->status = DIFF_STATUS_MODIFIED;
4194                 else {
4195                         /* This is a "no-change" entry and should not
4196                          * happen anymore, but prepare for broken callers.
4197                          */
4198                         error("feeding unmodified %s to diffcore",
4199                               p->one->path);
4200                         p->status = DIFF_STATUS_UNKNOWN;
4201                 }
4202         }
4203         diff_debug_queue("resolve-rename-copy done", q);
4204 }
4205
4206 static int check_pair_status(struct diff_filepair *p)
4207 {
4208         switch (p->status) {
4209         case DIFF_STATUS_UNKNOWN:
4210                 return 0;
4211         case 0:
4212                 die("internal error in diff-resolve-rename-copy");
4213         default:
4214                 return 1;
4215         }
4216 }
4217
4218 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4219 {
4220         int fmt = opt->output_format;
4221
4222         if (fmt & DIFF_FORMAT_CHECKDIFF)
4223                 diff_flush_checkdiff(p, opt);
4224         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4225                 diff_flush_raw(p, opt);
4226         else if (fmt & DIFF_FORMAT_NAME) {
4227                 const char *name_a, *name_b;
4228                 name_a = p->two->path;
4229                 name_b = NULL;
4230                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4231                 write_name_quoted(name_a, opt->file, opt->line_termination);
4232         }
4233 }
4234
4235 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4236 {
4237         if (fs->mode)
4238                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4239         else
4240                 fprintf(file, " %s ", newdelete);
4241         write_name_quoted(fs->path, file, '\n');
4242 }
4243
4244
4245 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4246                 const char *line_prefix)
4247 {
4248         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4249                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4250                         p->two->mode, show_name ? ' ' : '\n');
4251                 if (show_name) {
4252                         write_name_quoted(p->two->path, file, '\n');
4253                 }
4254         }
4255 }
4256
4257 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4258                         const char *line_prefix)
4259 {
4260         char *names = pprint_rename(p->one->path, p->two->path);
4261
4262         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4263         free(names);
4264         show_mode_change(file, p, 0, line_prefix);
4265 }
4266
4267 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4268 {
4269         FILE *file = opt->file;
4270         const char *line_prefix = diff_line_prefix(opt);
4271
4272         switch(p->status) {
4273         case DIFF_STATUS_DELETED:
4274                 fputs(line_prefix, file);
4275                 show_file_mode_name(file, "delete", p->one);
4276                 break;
4277         case DIFF_STATUS_ADDED:
4278                 fputs(line_prefix, file);
4279                 show_file_mode_name(file, "create", p->two);
4280                 break;
4281         case DIFF_STATUS_COPIED:
4282                 fputs(line_prefix, file);
4283                 show_rename_copy(file, "copy", p, line_prefix);
4284                 break;
4285         case DIFF_STATUS_RENAMED:
4286                 fputs(line_prefix, file);
4287                 show_rename_copy(file, "rename", p, line_prefix);
4288                 break;
4289         default:
4290                 if (p->score) {
4291                         fprintf(file, "%s rewrite ", line_prefix);
4292                         write_name_quoted(p->two->path, file, ' ');
4293                         fprintf(file, "(%d%%)\n", similarity_index(p));
4294                 }
4295                 show_mode_change(file, p, !p->score, line_prefix);
4296                 break;
4297         }
4298 }
4299
4300 struct patch_id_t {
4301         git_SHA_CTX *ctx;
4302         int patchlen;
4303 };
4304
4305 static int remove_space(char *line, int len)
4306 {
4307         int i;
4308         char *dst = line;
4309         unsigned char c;
4310
4311         for (i = 0; i < len; i++)
4312                 if (!isspace((c = line[i])))
4313                         *dst++ = c;
4314
4315         return dst - line;
4316 }
4317
4318 static void patch_id_consume(void *priv, char *line, unsigned long len)
4319 {
4320         struct patch_id_t *data = priv;
4321         int new_len;
4322
4323         /* Ignore line numbers when computing the SHA1 of the patch */
4324         if (starts_with(line, "@@ -"))
4325                 return;
4326
4327         new_len = remove_space(line, len);
4328
4329         git_SHA1_Update(data->ctx, line, new_len);
4330         data->patchlen += new_len;
4331 }
4332
4333 /* returns 0 upon success, and writes result into sha1 */
4334 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4335 {
4336         struct diff_queue_struct *q = &diff_queued_diff;
4337         int i;
4338         git_SHA_CTX ctx;
4339         struct patch_id_t data;
4340         char buffer[PATH_MAX * 4 + 20];
4341
4342         git_SHA1_Init(&ctx);
4343         memset(&data, 0, sizeof(struct patch_id_t));
4344         data.ctx = &ctx;
4345
4346         for (i = 0; i < q->nr; i++) {
4347                 xpparam_t xpp;
4348                 xdemitconf_t xecfg;
4349                 mmfile_t mf1, mf2;
4350                 struct diff_filepair *p = q->queue[i];
4351                 int len1, len2;
4352
4353                 memset(&xpp, 0, sizeof(xpp));
4354                 memset(&xecfg, 0, sizeof(xecfg));
4355                 if (p->status == 0)
4356                         return error("internal diff status error");
4357                 if (p->status == DIFF_STATUS_UNKNOWN)
4358                         continue;
4359                 if (diff_unmodified_pair(p))
4360                         continue;
4361                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4362                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4363                         continue;
4364                 if (DIFF_PAIR_UNMERGED(p))
4365                         continue;
4366
4367                 diff_fill_sha1_info(p->one);
4368                 diff_fill_sha1_info(p->two);
4369                 if (fill_mmfile(&mf1, p->one) < 0 ||
4370                                 fill_mmfile(&mf2, p->two) < 0)
4371                         return error("unable to read files to diff");
4372
4373                 len1 = remove_space(p->one->path, strlen(p->one->path));
4374                 len2 = remove_space(p->two->path, strlen(p->two->path));
4375                 if (p->one->mode == 0)
4376                         len1 = snprintf(buffer, sizeof(buffer),
4377                                         "diff--gita/%.*sb/%.*s"
4378                                         "newfilemode%06o"
4379                                         "---/dev/null"
4380                                         "+++b/%.*s",
4381                                         len1, p->one->path,
4382                                         len2, p->two->path,
4383                                         p->two->mode,
4384                                         len2, p->two->path);
4385                 else if (p->two->mode == 0)
4386                         len1 = snprintf(buffer, sizeof(buffer),
4387                                         "diff--gita/%.*sb/%.*s"
4388                                         "deletedfilemode%06o"
4389                                         "---a/%.*s"
4390                                         "+++/dev/null",
4391                                         len1, p->one->path,
4392                                         len2, p->two->path,
4393                                         p->one->mode,
4394                                         len1, p->one->path);
4395                 else
4396                         len1 = snprintf(buffer, sizeof(buffer),
4397                                         "diff--gita/%.*sb/%.*s"
4398                                         "---a/%.*s"
4399                                         "+++b/%.*s",
4400                                         len1, p->one->path,
4401                                         len2, p->two->path,
4402                                         len1, p->one->path,
4403                                         len2, p->two->path);
4404                 git_SHA1_Update(&ctx, buffer, len1);
4405
4406                 if (diff_filespec_is_binary(p->one) ||
4407                     diff_filespec_is_binary(p->two)) {
4408                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4409                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4410                         continue;
4411                 }
4412
4413                 xpp.flags = 0;
4414                 xecfg.ctxlen = 3;
4415                 xecfg.flags = 0;
4416                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4417                               &xpp, &xecfg);
4418         }
4419
4420         git_SHA1_Final(sha1, &ctx);
4421         return 0;
4422 }
4423
4424 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4425 {
4426         struct diff_queue_struct *q = &diff_queued_diff;
4427         int i;
4428         int result = diff_get_patch_id(options, sha1);
4429
4430         for (i = 0; i < q->nr; i++)
4431                 diff_free_filepair(q->queue[i]);
4432
4433         free(q->queue);
4434         DIFF_QUEUE_CLEAR(q);
4435
4436         return result;
4437 }
4438
4439 static int is_summary_empty(const struct diff_queue_struct *q)
4440 {
4441         int i;
4442
4443         for (i = 0; i < q->nr; i++) {
4444                 const struct diff_filepair *p = q->queue[i];
4445
4446                 switch (p->status) {
4447                 case DIFF_STATUS_DELETED:
4448                 case DIFF_STATUS_ADDED:
4449                 case DIFF_STATUS_COPIED:
4450                 case DIFF_STATUS_RENAMED:
4451                         return 0;
4452                 default:
4453                         if (p->score)
4454                                 return 0;
4455                         if (p->one->mode && p->two->mode &&
4456                             p->one->mode != p->two->mode)
4457                                 return 0;
4458                         break;
4459                 }
4460         }
4461         return 1;
4462 }
4463
4464 static const char rename_limit_warning[] =
4465 "inexact rename detection was skipped due to too many files.";
4466
4467 static const char degrade_cc_to_c_warning[] =
4468 "only found copies from modified paths due to too many files.";
4469
4470 static const char rename_limit_advice[] =
4471 "you may want to set your %s variable to at least "
4472 "%d and retry the command.";
4473
4474 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4475 {
4476         if (degraded_cc)
4477                 warning(degrade_cc_to_c_warning);
4478         else if (needed)
4479                 warning(rename_limit_warning);
4480         else
4481                 return;
4482         if (0 < needed && needed < 32767)
4483                 warning(rename_limit_advice, varname, needed);
4484 }
4485
4486 void diff_flush(struct diff_options *options)
4487 {
4488         struct diff_queue_struct *q = &diff_queued_diff;
4489         int i, output_format = options->output_format;
4490         int separator = 0;
4491         int dirstat_by_line = 0;
4492
4493         /*
4494          * Order: raw, stat, summary, patch
4495          * or:    name/name-status/checkdiff (other bits clear)
4496          */
4497         if (!q->nr)
4498                 goto free_queue;
4499
4500         if (output_format & (DIFF_FORMAT_RAW |
4501                              DIFF_FORMAT_NAME |
4502                              DIFF_FORMAT_NAME_STATUS |
4503                              DIFF_FORMAT_CHECKDIFF)) {
4504                 for (i = 0; i < q->nr; i++) {
4505                         struct diff_filepair *p = q->queue[i];
4506                         if (check_pair_status(p))
4507                                 flush_one_pair(p, options);
4508                 }
4509                 separator++;
4510         }
4511
4512         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4513                 dirstat_by_line = 1;
4514
4515         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4516             dirstat_by_line) {
4517                 struct diffstat_t diffstat;
4518
4519                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4520                 for (i = 0; i < q->nr; i++) {
4521                         struct diff_filepair *p = q->queue[i];
4522                         if (check_pair_status(p))
4523                                 diff_flush_stat(p, options, &diffstat);
4524                 }
4525                 if (output_format & DIFF_FORMAT_NUMSTAT)
4526                         show_numstat(&diffstat, options);
4527                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4528                         show_stats(&diffstat, options);
4529                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4530                         show_shortstats(&diffstat, options);
4531                 if (output_format & DIFF_FORMAT_DIRSTAT)
4532                         show_dirstat_by_line(&diffstat, options);
4533                 free_diffstat_info(&diffstat);
4534                 separator++;
4535         }
4536         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4537                 show_dirstat(options);
4538
4539         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4540                 for (i = 0; i < q->nr; i++) {
4541                         diff_summary(options, q->queue[i]);
4542                 }
4543                 separator++;
4544         }
4545
4546         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4547             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4548             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4549                 /*
4550                  * run diff_flush_patch for the exit status. setting
4551                  * options->file to /dev/null should be safe, because we
4552                  * aren't supposed to produce any output anyway.
4553                  */
4554                 if (options->close_file)
4555                         fclose(options->file);
4556                 options->file = fopen("/dev/null", "w");
4557                 if (!options->file)
4558                         die_errno("Could not open /dev/null");
4559                 options->close_file = 1;
4560                 for (i = 0; i < q->nr; i++) {
4561                         struct diff_filepair *p = q->queue[i];
4562                         if (check_pair_status(p))
4563                                 diff_flush_patch(p, options);
4564                         if (options->found_changes)
4565                                 break;
4566                 }
4567         }
4568
4569         if (output_format & DIFF_FORMAT_PATCH) {
4570                 if (separator) {
4571                         fprintf(options->file, "%s%c",
4572                                 diff_line_prefix(options),
4573                                 options->line_termination);
4574                         if (options->stat_sep) {
4575                                 /* attach patch instead of inline */
4576                                 fputs(options->stat_sep, options->file);
4577                         }
4578                 }
4579
4580                 for (i = 0; i < q->nr; i++) {
4581                         struct diff_filepair *p = q->queue[i];
4582                         if (check_pair_status(p))
4583                                 diff_flush_patch(p, options);
4584                 }
4585         }
4586
4587         if (output_format & DIFF_FORMAT_CALLBACK)
4588                 options->format_callback(q, options, options->format_callback_data);
4589
4590         for (i = 0; i < q->nr; i++)
4591                 diff_free_filepair(q->queue[i]);
4592 free_queue:
4593         free(q->queue);
4594         DIFF_QUEUE_CLEAR(q);
4595         if (options->close_file)
4596                 fclose(options->file);
4597
4598         /*
4599          * Report the content-level differences with HAS_CHANGES;
4600          * diff_addremove/diff_change does not set the bit when
4601          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4602          */
4603         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4604                 if (options->found_changes)
4605                         DIFF_OPT_SET(options, HAS_CHANGES);
4606                 else
4607                         DIFF_OPT_CLR(options, HAS_CHANGES);
4608         }
4609 }
4610
4611 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4612 {
4613         return (((p->status == DIFF_STATUS_MODIFIED) &&
4614                  ((p->score &&
4615                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4616                   (!p->score &&
4617                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4618                 ((p->status != DIFF_STATUS_MODIFIED) &&
4619                  filter_bit_tst(p->status, options)));
4620 }
4621
4622 static void diffcore_apply_filter(struct diff_options *options)
4623 {
4624         int i;
4625         struct diff_queue_struct *q = &diff_queued_diff;
4626         struct diff_queue_struct outq;
4627
4628         DIFF_QUEUE_CLEAR(&outq);
4629
4630         if (!options->filter)
4631                 return;
4632
4633         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4634                 int found;
4635                 for (i = found = 0; !found && i < q->nr; i++) {
4636                         if (match_filter(options, q->queue[i]))
4637                                 found++;
4638                 }
4639                 if (found)
4640                         return;
4641
4642                 /* otherwise we will clear the whole queue
4643                  * by copying the empty outq at the end of this
4644                  * function, but first clear the current entries
4645                  * in the queue.
4646                  */
4647                 for (i = 0; i < q->nr; i++)
4648                         diff_free_filepair(q->queue[i]);
4649         }
4650         else {
4651                 /* Only the matching ones */
4652                 for (i = 0; i < q->nr; i++) {
4653                         struct diff_filepair *p = q->queue[i];
4654                         if (match_filter(options, p))
4655                                 diff_q(&outq, p);
4656                         else
4657                                 diff_free_filepair(p);
4658                 }
4659         }
4660         free(q->queue);
4661         *q = outq;
4662 }
4663
4664 /* Check whether two filespecs with the same mode and size are identical */
4665 static int diff_filespec_is_identical(struct diff_filespec *one,
4666                                       struct diff_filespec *two)
4667 {
4668         if (S_ISGITLINK(one->mode))
4669                 return 0;
4670         if (diff_populate_filespec(one, 0))
4671                 return 0;
4672         if (diff_populate_filespec(two, 0))
4673                 return 0;
4674         return !memcmp(one->data, two->data, one->size);
4675 }
4676
4677 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4678 {
4679         if (p->done_skip_stat_unmatch)
4680                 return p->skip_stat_unmatch_result;
4681
4682         p->done_skip_stat_unmatch = 1;
4683         p->skip_stat_unmatch_result = 0;
4684         /*
4685          * 1. Entries that come from stat info dirtiness
4686          *    always have both sides (iow, not create/delete),
4687          *    one side of the object name is unknown, with
4688          *    the same mode and size.  Keep the ones that
4689          *    do not match these criteria.  They have real
4690          *    differences.
4691          *
4692          * 2. At this point, the file is known to be modified,
4693          *    with the same mode and size, and the object
4694          *    name of one side is unknown.  Need to inspect
4695          *    the identical contents.
4696          */
4697         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4698             !DIFF_FILE_VALID(p->two) ||
4699             (p->one->sha1_valid && p->two->sha1_valid) ||
4700             (p->one->mode != p->two->mode) ||
4701             diff_populate_filespec(p->one, 1) ||
4702             diff_populate_filespec(p->two, 1) ||
4703             (p->one->size != p->two->size) ||
4704             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4705                 p->skip_stat_unmatch_result = 1;
4706         return p->skip_stat_unmatch_result;
4707 }
4708
4709 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4710 {
4711         int i;
4712         struct diff_queue_struct *q = &diff_queued_diff;
4713         struct diff_queue_struct outq;
4714         DIFF_QUEUE_CLEAR(&outq);
4715
4716         for (i = 0; i < q->nr; i++) {
4717                 struct diff_filepair *p = q->queue[i];
4718
4719                 if (diff_filespec_check_stat_unmatch(p))
4720                         diff_q(&outq, p);
4721                 else {
4722                         /*
4723                          * The caller can subtract 1 from skip_stat_unmatch
4724                          * to determine how many paths were dirty only
4725                          * due to stat info mismatch.
4726                          */
4727                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4728                                 diffopt->skip_stat_unmatch++;
4729                         diff_free_filepair(p);
4730                 }
4731         }
4732         free(q->queue);
4733         *q = outq;
4734 }
4735
4736 static int diffnamecmp(const void *a_, const void *b_)
4737 {
4738         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4739         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4740         const char *name_a, *name_b;
4741
4742         name_a = a->one ? a->one->path : a->two->path;
4743         name_b = b->one ? b->one->path : b->two->path;
4744         return strcmp(name_a, name_b);
4745 }
4746
4747 void diffcore_fix_diff_index(struct diff_options *options)
4748 {
4749         struct diff_queue_struct *q = &diff_queued_diff;
4750         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4751 }
4752
4753 void diffcore_std(struct diff_options *options)
4754 {
4755         if (options->skip_stat_unmatch)
4756                 diffcore_skip_stat_unmatch(options);
4757         if (!options->found_follow) {
4758                 /* See try_to_follow_renames() in tree-diff.c */
4759                 if (options->break_opt != -1)
4760                         diffcore_break(options->break_opt);
4761                 if (options->detect_rename)
4762                         diffcore_rename(options);
4763                 if (options->break_opt != -1)
4764                         diffcore_merge_broken();
4765         }
4766         if (options->pickaxe)
4767                 diffcore_pickaxe(options);
4768         if (options->orderfile)
4769                 diffcore_order(options->orderfile);
4770         if (!options->found_follow)
4771                 /* See try_to_follow_renames() in tree-diff.c */
4772                 diff_resolve_rename_copy();
4773         diffcore_apply_filter(options);
4774
4775         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4776                 DIFF_OPT_SET(options, HAS_CHANGES);
4777         else
4778                 DIFF_OPT_CLR(options, HAS_CHANGES);
4779
4780         options->found_follow = 0;
4781 }
4782
4783 int diff_result_code(struct diff_options *opt, int status)
4784 {
4785         int result = 0;
4786
4787         diff_warn_rename_limit("diff.renameLimit",
4788                                opt->needed_rename_limit,
4789                                opt->degraded_cc_to_c);
4790         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4791             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4792                 return status;
4793         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4794             DIFF_OPT_TST(opt, HAS_CHANGES))
4795                 result |= 01;
4796         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4797             DIFF_OPT_TST(opt, CHECK_FAILED))
4798                 result |= 02;
4799         return result;
4800 }
4801
4802 int diff_can_quit_early(struct diff_options *opt)
4803 {
4804         return (DIFF_OPT_TST(opt, QUICK) &&
4805                 !opt->filter &&
4806                 DIFF_OPT_TST(opt, HAS_CHANGES));
4807 }
4808
4809 /*
4810  * Shall changes to this submodule be ignored?
4811  *
4812  * Submodule changes can be configured to be ignored separately for each path,
4813  * but that configuration can be overridden from the command line.
4814  */
4815 static int is_submodule_ignored(const char *path, struct diff_options *options)
4816 {
4817         int ignored = 0;
4818         unsigned orig_flags = options->flags;
4819         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4820                 set_diffopt_flags_from_submodule_config(options, path);
4821         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4822                 ignored = 1;
4823         options->flags = orig_flags;
4824         return ignored;
4825 }
4826
4827 void diff_addremove(struct diff_options *options,
4828                     int addremove, unsigned mode,
4829                     const unsigned char *sha1,
4830                     int sha1_valid,
4831                     const char *concatpath, unsigned dirty_submodule)
4832 {
4833         struct diff_filespec *one, *two;
4834
4835         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4836                 return;
4837
4838         /* This may look odd, but it is a preparation for
4839          * feeding "there are unchanged files which should
4840          * not produce diffs, but when you are doing copy
4841          * detection you would need them, so here they are"
4842          * entries to the diff-core.  They will be prefixed
4843          * with something like '=' or '*' (I haven't decided
4844          * which but should not make any difference).
4845          * Feeding the same new and old to diff_change()
4846          * also has the same effect.
4847          * Before the final output happens, they are pruned after
4848          * merged into rename/copy pairs as appropriate.
4849          */
4850         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4851                 addremove = (addremove == '+' ? '-' :
4852                              addremove == '-' ? '+' : addremove);
4853
4854         if (options->prefix &&
4855             strncmp(concatpath, options->prefix, options->prefix_length))
4856                 return;
4857
4858         one = alloc_filespec(concatpath);
4859         two = alloc_filespec(concatpath);
4860
4861         if (addremove != '+')
4862                 fill_filespec(one, sha1, sha1_valid, mode);
4863         if (addremove != '-') {
4864                 fill_filespec(two, sha1, sha1_valid, mode);
4865                 two->dirty_submodule = dirty_submodule;
4866         }
4867
4868         diff_queue(&diff_queued_diff, one, two);
4869         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4870                 DIFF_OPT_SET(options, HAS_CHANGES);
4871 }
4872
4873 void diff_change(struct diff_options *options,
4874                  unsigned old_mode, unsigned new_mode,
4875                  const unsigned char *old_sha1,
4876                  const unsigned char *new_sha1,
4877                  int old_sha1_valid, int new_sha1_valid,
4878                  const char *concatpath,
4879                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4880 {
4881         struct diff_filespec *one, *two;
4882         struct diff_filepair *p;
4883
4884         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4885             is_submodule_ignored(concatpath, options))
4886                 return;
4887
4888         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4889                 unsigned tmp;
4890                 const unsigned char *tmp_c;
4891                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4892                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4893                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4894                         new_sha1_valid = tmp;
4895                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4896                         new_dirty_submodule = tmp;
4897         }
4898
4899         if (options->prefix &&
4900             strncmp(concatpath, options->prefix, options->prefix_length))
4901                 return;
4902
4903         one = alloc_filespec(concatpath);
4904         two = alloc_filespec(concatpath);
4905         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4906         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4907         one->dirty_submodule = old_dirty_submodule;
4908         two->dirty_submodule = new_dirty_submodule;
4909         p = diff_queue(&diff_queued_diff, one, two);
4910
4911         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4912                 return;
4913
4914         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4915             !diff_filespec_check_stat_unmatch(p))
4916                 return;
4917
4918         DIFF_OPT_SET(options, HAS_CHANGES);
4919 }
4920
4921 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4922 {
4923         struct diff_filepair *pair;
4924         struct diff_filespec *one, *two;
4925
4926         if (options->prefix &&
4927             strncmp(path, options->prefix, options->prefix_length))
4928                 return NULL;
4929
4930         one = alloc_filespec(path);
4931         two = alloc_filespec(path);
4932         pair = diff_queue(&diff_queued_diff, one, two);
4933         pair->is_unmerged = 1;
4934         return pair;
4935 }
4936
4937 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4938                 size_t *outsize)
4939 {
4940         struct diff_tempfile *temp;
4941         const char *argv[3];
4942         const char **arg = argv;
4943         struct child_process child;
4944         struct strbuf buf = STRBUF_INIT;
4945         int err = 0;
4946
4947         temp = prepare_temp_file(spec->path, spec);
4948         *arg++ = pgm;
4949         *arg++ = temp->name;
4950         *arg = NULL;
4951
4952         memset(&child, 0, sizeof(child));
4953         child.use_shell = 1;
4954         child.argv = argv;
4955         child.out = -1;
4956         if (start_command(&child)) {
4957                 remove_tempfile();
4958                 return NULL;
4959         }
4960
4961         if (strbuf_read(&buf, child.out, 0) < 0)
4962                 err = error("error reading from textconv command '%s'", pgm);
4963         close(child.out);
4964
4965         if (finish_command(&child) || err) {
4966                 strbuf_release(&buf);
4967                 remove_tempfile();
4968                 return NULL;
4969         }
4970         remove_tempfile();
4971
4972         return strbuf_detach(&buf, outsize);
4973 }
4974
4975 size_t fill_textconv(struct userdiff_driver *driver,
4976                      struct diff_filespec *df,
4977                      char **outbuf)
4978 {
4979         size_t size;
4980
4981         if (!driver || !driver->textconv) {
4982                 if (!DIFF_FILE_VALID(df)) {
4983                         *outbuf = "";
4984                         return 0;
4985                 }
4986                 if (diff_populate_filespec(df, 0))
4987                         die("unable to read files to diff");
4988                 *outbuf = df->data;
4989                 return df->size;
4990         }
4991
4992         if (driver->textconv_cache && df->sha1_valid) {
4993                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4994                                           &size);
4995                 if (*outbuf)
4996                         return size;
4997         }
4998
4999         *outbuf = run_textconv(driver->textconv, df, &size);
5000         if (!*outbuf)
5001                 die("unable to read files to diff");
5002
5003         if (driver->textconv_cache && df->sha1_valid) {
5004                 /* ignore errors, as we might be in a readonly repository */
5005                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5006                                 size);
5007                 /*
5008                  * we could save up changes and flush them all at the end,
5009                  * but we would need an extra call after all diffing is done.
5010                  * Since generating a cache entry is the slow path anyway,
5011                  * this extra overhead probably isn't a big deal.
5012                  */
5013                 notes_cache_write(driver->textconv_cache);
5014         }
5015
5016         return size;
5017 }
5018
5019 void setup_diff_pager(struct diff_options *opt)
5020 {
5021         /*
5022          * If the user asked for our exit code, then either they want --quiet
5023          * or --exit-code. We should definitely not bother with a pager in the
5024          * former case, as we will generate no output. Since we still properly
5025          * report our exit code even when a pager is run, we _could_ run a
5026          * pager with --exit-code. But since we have not done so historically,
5027          * and because it is easy to find people oneline advising "git diff
5028          * --exit-code" in hooks and other scripts, we do not do so.
5029          */
5030         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5031             check_pager_config("diff") != 0)
5032                 setup_pager();
5033 }