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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "tempfile.h"
   6 #include "quote.h"
   7 #include "diff.h"
   8 #include "diffcore.h"
   9 #include "delta.h"
  10 #include "xdiff-interface.h"
  11 #include "color.h"
  12 #include "attr.h"
  13 #include "run-command.h"
  14 #include "utf8.h"
  15 #include "userdiff.h"
  16 #include "submodule-config.h"
  17 #include "submodule.h"
  18 #include "ll-merge.h"
  19 #include "string-list.h"
  20 #include "argv-array.h"
  21 #include "graph.h"
  22
  23 #ifdef NO_FAST_WORKING_DIRECTORY
  24 #define FAST_WORKING_DIRECTORY 0
  25 #else
  26 #define FAST_WORKING_DIRECTORY 1
  27 #endif
  28
  29 static int diff_detect_rename_default;
  30 static int diff_indent_heuristic; /* experimental */
  31 static int diff_rename_limit_default = 400;
  32 static int diff_suppress_blank_empty;
  33 static int diff_use_color_default = -1;
  34 static int diff_context_default = 3;
  35 static const char *diff_word_regex_cfg;
  36 static const char *external_diff_cmd_cfg;
  37 static const char *diff_order_file_cfg;
  38 int diff_auto_refresh_index = 1;
  39 static int diff_mnemonic_prefix;
  40 static int diff_no_prefix;
  41 static int diff_stat_graph_width;
  42 static int diff_dirstat_permille_default = 30;
  43 static struct diff_options default_diff_options;
  44 static long diff_algorithm;
  45 static unsigned ws_error_highlight_default = WSEH_NEW;
  46
  47 static char diff_colors[][COLOR_MAXLEN] = {
  48         GIT_COLOR_RESET,
  49         GIT_COLOR_NORMAL,       /* CONTEXT */
  50         GIT_COLOR_BOLD,         /* METAINFO */
  51         GIT_COLOR_CYAN,         /* FRAGINFO */
  52         GIT_COLOR_RED,          /* OLD */
  53         GIT_COLOR_GREEN,        /* NEW */
  54         GIT_COLOR_YELLOW,       /* COMMIT */
  55         GIT_COLOR_BG_RED,       /* WHITESPACE */
  56         GIT_COLOR_NORMAL,       /* FUNCINFO */
  57 };
  58
  59 static NORETURN void die_want_option(const char *option_name)
  60 {
  61         die(_("option '%s' requires a value"), option_name);
  62 }
  63
  64 static int parse_diff_color_slot(const char *var)
  65 {
  66         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  67                 return DIFF_CONTEXT;
  68         if (!strcasecmp(var, "meta"))
  69                 return DIFF_METAINFO;
  70         if (!strcasecmp(var, "frag"))
  71                 return DIFF_FRAGINFO;
  72         if (!strcasecmp(var, "old"))
  73                 return DIFF_FILE_OLD;
  74         if (!strcasecmp(var, "new"))
  75                 return DIFF_FILE_NEW;
  76         if (!strcasecmp(var, "commit"))
  77                 return DIFF_COMMIT;
  78         if (!strcasecmp(var, "whitespace"))
  79                 return DIFF_WHITESPACE;
  80         if (!strcasecmp(var, "func"))
  81                 return DIFF_FUNCINFO;
  82         return -1;
  83 }
  84
  85 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  86                                 struct strbuf *errmsg)
  87 {
  88         char *params_copy = xstrdup(params_string);
  89         struct string_list params = STRING_LIST_INIT_NODUP;
  90         int ret = 0;
  91         int i;
  92
  93         if (*params_copy)
  94                 string_list_split_in_place(&params, params_copy, ',', -1);
  95         for (i = 0; i < params.nr; i++) {
  96                 const char *p = params.items[i].string;
  97                 if (!strcmp(p, "changes")) {
  98                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  99                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 100                 } else if (!strcmp(p, "lines")) {
 101                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
 102                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 103                 } else if (!strcmp(p, "files")) {
 104                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 105                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 106                 } else if (!strcmp(p, "noncumulative")) {
 107                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 108                 } else if (!strcmp(p, "cumulative")) {
 109                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 110                 } else if (isdigit(*p)) {
 111                         char *end;
 112                         int permille = strtoul(p, &end, 10) * 10;
 113                         if (*end == '.' && isdigit(*++end)) {
 114                                 /* only use first digit */
 115                                 permille += *end - '0';
 116                                 /* .. and ignore any further digits */
 117                                 while (isdigit(*++end))
 118                                         ; /* nothing */
 119                         }
 120                         if (!*end)
 121                                 options->dirstat_permille = permille;
 122                         else {
 123                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 124                                             p);
 125                                 ret++;
 126                         }
 127                 } else {
 128                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 129                         ret++;
 130                 }
 131
 132         }
 133         string_list_clear(&params, 0);
 134         free(params_copy);
 135         return ret;
 136 }
 137
 138 static int parse_submodule_params(struct diff_options *options, const char *value)
 139 {
 140         if (!strcmp(value, "log"))
 141                 options->submodule_format = DIFF_SUBMODULE_LOG;
 142         else if (!strcmp(value, "short"))
 143                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 144         else if (!strcmp(value, "diff"))
 145                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 146         else
 147                 return -1;
 148         return 0;
 149 }
 150
 151 static int git_config_rename(const char *var, const char *value)
 152 {
 153         if (!value)
 154                 return DIFF_DETECT_RENAME;
 155         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 156                 return  DIFF_DETECT_COPY;
 157         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 158 }
 159
 160 long parse_algorithm_value(const char *value)
 161 {
 162         if (!value)
 163                 return -1;
 164         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 165                 return 0;
 166         else if (!strcasecmp(value, "minimal"))
 167                 return XDF_NEED_MINIMAL;
 168         else if (!strcasecmp(value, "patience"))
 169                 return XDF_PATIENCE_DIFF;
 170         else if (!strcasecmp(value, "histogram"))
 171                 return XDF_HISTOGRAM_DIFF;
 172         return -1;
 173 }
 174
 175 static int parse_one_token(const char **arg, const char *token)
 176 {
 177         const char *rest;
 178         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 179                 *arg = rest;
 180                 return 1;
 181         }
 182         return 0;
 183 }
 184
 185 static int parse_ws_error_highlight(const char *arg)
 186 {
 187         const char *orig_arg = arg;
 188         unsigned val = 0;
 189
 190         while (*arg) {
 191                 if (parse_one_token(&arg, "none"))
 192                         val = 0;
 193                 else if (parse_one_token(&arg, "default"))
 194                         val = WSEH_NEW;
 195                 else if (parse_one_token(&arg, "all"))
 196                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 197                 else if (parse_one_token(&arg, "new"))
 198                         val |= WSEH_NEW;
 199                 else if (parse_one_token(&arg, "old"))
 200                         val |= WSEH_OLD;
 201                 else if (parse_one_token(&arg, "context"))
 202                         val |= WSEH_CONTEXT;
 203                 else {
 204                         return -1 - (int)(arg - orig_arg);
 205                 }
 206                 if (*arg)
 207                         arg++;
 208         }
 209         return val;
 210 }
 211
 212 /*
 213  * These are to give UI layer defaults.
 214  * The core-level commands such as git-diff-files should
 215  * never be affected by the setting of diff.renames
 216  * the user happens to have in the configuration file.
 217  */
 218 void init_diff_ui_defaults(void)
 219 {
 220         diff_detect_rename_default = 1;
 221 }
 222
 223 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 224 {
 225         if (!strcmp(var, "diff.indentheuristic"))
 226                 diff_indent_heuristic = git_config_bool(var, value);
 227         return 0;
 228 }
 229
 230 int git_diff_ui_config(const char *var, const char *value, void *cb)
 231 {
 232         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 233                 diff_use_color_default = git_config_colorbool(var, value);
 234                 return 0;
 235         }
 236         if (!strcmp(var, "diff.context")) {
 237                 diff_context_default = git_config_int(var, value);
 238                 if (diff_context_default < 0)
 239                         return -1;
 240                 return 0;
 241         }
 242         if (!strcmp(var, "diff.renames")) {
 243                 diff_detect_rename_default = git_config_rename(var, value);
 244                 return 0;
 245         }
 246         if (!strcmp(var, "diff.autorefreshindex")) {
 247                 diff_auto_refresh_index = git_config_bool(var, value);
 248                 return 0;
 249         }
 250         if (!strcmp(var, "diff.mnemonicprefix")) {
 251                 diff_mnemonic_prefix = git_config_bool(var, value);
 252                 return 0;
 253         }
 254         if (!strcmp(var, "diff.noprefix")) {
 255                 diff_no_prefix = git_config_bool(var, value);
 256                 return 0;
 257         }
 258         if (!strcmp(var, "diff.statgraphwidth")) {
 259                 diff_stat_graph_width = git_config_int(var, value);
 260                 return 0;
 261         }
 262         if (!strcmp(var, "diff.external"))
 263                 return git_config_string(&external_diff_cmd_cfg, var, value);
 264         if (!strcmp(var, "diff.wordregex"))
 265                 return git_config_string(&diff_word_regex_cfg, var, value);
 266         if (!strcmp(var, "diff.orderfile"))
 267                 return git_config_pathname(&diff_order_file_cfg, var, value);
 268
 269         if (!strcmp(var, "diff.ignoresubmodules"))
 270                 handle_ignore_submodules_arg(&default_diff_options, value);
 271
 272         if (!strcmp(var, "diff.submodule")) {
 273                 if (parse_submodule_params(&default_diff_options, value))
 274                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 275                                 value);
 276                 return 0;
 277         }
 278
 279         if (!strcmp(var, "diff.algorithm")) {
 280                 diff_algorithm = parse_algorithm_value(value);
 281                 if (diff_algorithm < 0)
 282                         return -1;
 283                 return 0;
 284         }
 285
 286         if (git_diff_heuristic_config(var, value, cb) < 0)
 287                 return -1;
 288
 289         if (!strcmp(var, "diff.wserrorhighlight")) {
 290                 int val = parse_ws_error_highlight(value);
 291                 if (val < 0)
 292                         return -1;
 293                 ws_error_highlight_default = val;
 294                 return 0;
 295         }
 296
 297         if (git_color_config(var, value, cb) < 0)
 298                 return -1;
 299
 300         return git_diff_basic_config(var, value, cb);
 301 }
 302
 303 int git_diff_basic_config(const char *var, const char *value, void *cb)
 304 {
 305         const char *name;
 306
 307         if (!strcmp(var, "diff.renamelimit")) {
 308                 diff_rename_limit_default = git_config_int(var, value);
 309                 return 0;
 310         }
 311
 312         if (userdiff_config(var, value) < 0)
 313                 return -1;
 314
 315         if (skip_prefix(var, "diff.color.", &name) ||
 316             skip_prefix(var, "color.diff.", &name)) {
 317                 int slot = parse_diff_color_slot(name);
 318                 if (slot < 0)
 319                         return 0;
 320                 if (!value)
 321                         return config_error_nonbool(var);
 322                 return color_parse(value, diff_colors[slot]);
 323         }
 324
 325         /* like GNU diff's --suppress-blank-empty option  */
 326         if (!strcmp(var, "diff.suppressblankempty") ||
 327                         /* for backwards compatibility */
 328                         !strcmp(var, "diff.suppress-blank-empty")) {
 329                 diff_suppress_blank_empty = git_config_bool(var, value);
 330                 return 0;
 331         }
 332
 333         if (!strcmp(var, "diff.dirstat")) {
 334                 struct strbuf errmsg = STRBUF_INIT;
 335                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 336                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 337                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 338                                 errmsg.buf);
 339                 strbuf_release(&errmsg);
 340                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 341                 return 0;
 342         }
 343
 344         if (starts_with(var, "submodule."))
 345                 return parse_submodule_config_option(var, value);
 346
 347         return git_default_config(var, value, cb);
 348 }
 349
 350 static char *quote_two(const char *one, const char *two)
 351 {
 352         int need_one = quote_c_style(one, NULL, NULL, 1);
 353         int need_two = quote_c_style(two, NULL, NULL, 1);
 354         struct strbuf res = STRBUF_INIT;
 355
 356         if (need_one + need_two) {
 357                 strbuf_addch(&res, '"');
 358                 quote_c_style(one, &res, NULL, 1);
 359                 quote_c_style(two, &res, NULL, 1);
 360                 strbuf_addch(&res, '"');
 361         } else {
 362                 strbuf_addstr(&res, one);
 363                 strbuf_addstr(&res, two);
 364         }
 365         return strbuf_detach(&res, NULL);
 366 }
 367
 368 static const char *external_diff(void)
 369 {
 370         static const char *external_diff_cmd = NULL;
 371         static int done_preparing = 0;
 372
 373         if (done_preparing)
 374                 return external_diff_cmd;
 375         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 376         if (!external_diff_cmd)
 377                 external_diff_cmd = external_diff_cmd_cfg;
 378         done_preparing = 1;
 379         return external_diff_cmd;
 380 }
 381
 382 /*
 383  * Keep track of files used for diffing. Sometimes such an entry
 384  * refers to a temporary file, sometimes to an existing file, and
 385  * sometimes to "/dev/null".
 386  */
 387 static struct diff_tempfile {
 388         /*
 389          * filename external diff should read from, or NULL if this
 390          * entry is currently not in use:
 391          */
 392         const char *name;
 393
 394         char hex[GIT_SHA1_HEXSZ + 1];
 395         char mode[10];
 396
 397         /*
 398          * If this diff_tempfile instance refers to a temporary file,
 399          * this tempfile object is used to manage its lifetime.
 400          */
 401         struct tempfile tempfile;
 402 } diff_temp[2];
 403
 404 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 405
 406 struct emit_callback {
 407         int color_diff;
 408         unsigned ws_rule;
 409         int blank_at_eof_in_preimage;
 410         int blank_at_eof_in_postimage;
 411         int lno_in_preimage;
 412         int lno_in_postimage;
 413         sane_truncate_fn truncate;
 414         const char **label_path;
 415         struct diff_words_data *diff_words;
 416         struct diff_options *opt;
 417         struct strbuf *header;
 418 };
 419
 420 static int count_lines(const char *data, int size)
 421 {
 422         int count, ch, completely_empty = 1, nl_just_seen = 0;
 423         count = 0;
 424         while (0 < size--) {
 425                 ch = *data++;
 426                 if (ch == '\n') {
 427                         count++;
 428                         nl_just_seen = 1;
 429                         completely_empty = 0;
 430                 }
 431                 else {
 432                         nl_just_seen = 0;
 433                         completely_empty = 0;
 434                 }
 435         }
 436         if (completely_empty)
 437                 return 0;
 438         if (!nl_just_seen)
 439                 count++; /* no trailing newline */
 440         return count;
 441 }
 442
 443 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 444 {
 445         if (!DIFF_FILE_VALID(one)) {
 446                 mf->ptr = (char *)""; /* does not matter */
 447                 mf->size = 0;
 448                 return 0;
 449         }
 450         else if (diff_populate_filespec(one, 0))
 451                 return -1;
 452
 453         mf->ptr = one->data;
 454         mf->size = one->size;
 455         return 0;
 456 }
 457
 458 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 459 static unsigned long diff_filespec_size(struct diff_filespec *one)
 460 {
 461         if (!DIFF_FILE_VALID(one))
 462                 return 0;
 463         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 464         return one->size;
 465 }
 466
 467 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 468 {
 469         char *ptr = mf->ptr;
 470         long size = mf->size;
 471         int cnt = 0;
 472
 473         if (!size)
 474                 return cnt;
 475         ptr += size - 1; /* pointing at the very end */
 476         if (*ptr != '\n')
 477                 ; /* incomplete line */
 478         else
 479                 ptr--; /* skip the last LF */
 480         while (mf->ptr < ptr) {
 481                 char *prev_eol;
 482                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 483                         if (*prev_eol == '\n')
 484                                 break;
 485                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 486                         break;
 487                 cnt++;
 488                 ptr = prev_eol - 1;
 489         }
 490         return cnt;
 491 }
 492
 493 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 494                                struct emit_callback *ecbdata)
 495 {
 496         int l1, l2, at;
 497         unsigned ws_rule = ecbdata->ws_rule;
 498         l1 = count_trailing_blank(mf1, ws_rule);
 499         l2 = count_trailing_blank(mf2, ws_rule);
 500         if (l2 <= l1) {
 501                 ecbdata->blank_at_eof_in_preimage = 0;
 502                 ecbdata->blank_at_eof_in_postimage = 0;
 503                 return;
 504         }
 505         at = count_lines(mf1->ptr, mf1->size);
 506         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 507
 508         at = count_lines(mf2->ptr, mf2->size);
 509         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 510 }
 511
 512 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 513                         int first, const char *line, int len)
 514 {
 515         int has_trailing_newline, has_trailing_carriage_return;
 516         int nofirst;
 517         FILE *file = o->file;
 518
 519         fputs(diff_line_prefix(o), file);
 520
 521         if (len == 0) {
 522                 has_trailing_newline = (first == '\n');
 523                 has_trailing_carriage_return = (!has_trailing_newline &&
 524                                                 (first == '\r'));
 525                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 526         } else {
 527                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 528                 if (has_trailing_newline)
 529                         len--;
 530                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 531                 if (has_trailing_carriage_return)
 532                         len--;
 533                 nofirst = 0;
 534         }
 535
 536         if (len || !nofirst) {
 537                 fputs(set, file);
 538                 if (!nofirst)
 539                         fputc(first, file);
 540                 fwrite(line, len, 1, file);
 541                 fputs(reset, file);
 542         }
 543         if (has_trailing_carriage_return)
 544                 fputc('\r', file);
 545         if (has_trailing_newline)
 546                 fputc('\n', file);
 547 }
 548
 549 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 550                       const char *line, int len)
 551 {
 552         emit_line_0(o, set, reset, line[0], line+1, len-1);
 553 }
 554
 555 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 556 {
 557         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 558               ecbdata->blank_at_eof_in_preimage &&
 559               ecbdata->blank_at_eof_in_postimage &&
 560               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 561               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 562                 return 0;
 563         return ws_blank_line(line, len, ecbdata->ws_rule);
 564 }
 565
 566 static void emit_line_checked(const char *reset,
 567                               struct emit_callback *ecbdata,
 568                               const char *line, int len,
 569                               enum color_diff color,
 570                               unsigned ws_error_highlight,
 571                               char sign)
 572 {
 573         const char *set = diff_get_color(ecbdata->color_diff, color);
 574         const char *ws = NULL;
 575
 576         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
 577                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 578                 if (!*ws)
 579                         ws = NULL;
 580         }
 581
 582         if (!ws)
 583                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
 584         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
 585                 /* Blank line at EOF - paint '+' as well */
 586                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
 587         else {
 588                 /* Emit just the prefix, then the rest. */
 589                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
 590                 ws_check_emit(line, len, ecbdata->ws_rule,
 591                               ecbdata->opt->file, set, reset, ws);
 592         }
 593 }
 594
 595 static void emit_add_line(const char *reset,
 596                           struct emit_callback *ecbdata,
 597                           const char *line, int len)
 598 {
 599         emit_line_checked(reset, ecbdata, line, len,
 600                           DIFF_FILE_NEW, WSEH_NEW, '+');
 601 }
 602
 603 static void emit_del_line(const char *reset,
 604                           struct emit_callback *ecbdata,
 605                           const char *line, int len)
 606 {
 607         emit_line_checked(reset, ecbdata, line, len,
 608                           DIFF_FILE_OLD, WSEH_OLD, '-');
 609 }
 610
 611 static void emit_context_line(const char *reset,
 612                               struct emit_callback *ecbdata,
 613                               const char *line, int len)
 614 {
 615         emit_line_checked(reset, ecbdata, line, len,
 616                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
 617 }
 618
 619 static void emit_hunk_header(struct emit_callback *ecbdata,
 620                              const char *line, int len)
 621 {
 622         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 623         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 624         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 625         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 626         static const char atat[2] = { '@', '@' };
 627         const char *cp, *ep;
 628         struct strbuf msgbuf = STRBUF_INIT;
 629         int org_len = len;
 630         int i = 1;
 631
 632         /*
 633          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 634          * it always is at least 10 bytes long.
 635          */
 636         if (len < 10 ||
 637             memcmp(line, atat, 2) ||
 638             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 639                 emit_line(ecbdata->opt, context, reset, line, len);
 640                 return;
 641         }
 642         ep += 2; /* skip over @@ */
 643
 644         /* The hunk header in fraginfo color */
 645         strbuf_addstr(&msgbuf, frag);
 646         strbuf_add(&msgbuf, line, ep - line);
 647         strbuf_addstr(&msgbuf, reset);
 648
 649         /*
 650          * trailing "\r\n"
 651          */
 652         for ( ; i < 3; i++)
 653                 if (line[len - i] == '\r' || line[len - i] == '\n')
 654                         len--;
 655
 656         /* blank before the func header */
 657         for (cp = ep; ep - line < len; ep++)
 658                 if (*ep != ' ' && *ep != '\t')
 659                         break;
 660         if (ep != cp) {
 661                 strbuf_addstr(&msgbuf, context);
 662                 strbuf_add(&msgbuf, cp, ep - cp);
 663                 strbuf_addstr(&msgbuf, reset);
 664         }
 665
 666         if (ep < line + len) {
 667                 strbuf_addstr(&msgbuf, func);
 668                 strbuf_add(&msgbuf, ep, line + len - ep);
 669                 strbuf_addstr(&msgbuf, reset);
 670         }
 671
 672         strbuf_add(&msgbuf, line + len, org_len - len);
 673         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 674         strbuf_release(&msgbuf);
 675 }
 676
 677 static struct diff_tempfile *claim_diff_tempfile(void) {
 678         int i;
 679         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 680                 if (!diff_temp[i].name)
 681                         return diff_temp + i;
 682         die("BUG: diff is failing to clean up its tempfiles");
 683 }
 684
 685 static void remove_tempfile(void)
 686 {
 687         int i;
 688         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 689                 if (is_tempfile_active(&diff_temp[i].tempfile))
 690                         delete_tempfile(&diff_temp[i].tempfile);
 691                 diff_temp[i].name = NULL;
 692         }
 693 }
 694
 695 static void print_line_count(FILE *file, int count)
 696 {
 697         switch (count) {
 698         case 0:
 699                 fprintf(file, "0,0");
 700                 break;
 701         case 1:
 702                 fprintf(file, "1");
 703                 break;
 704         default:
 705                 fprintf(file, "1,%d", count);
 706                 break;
 707         }
 708 }
 709
 710 static void emit_rewrite_lines(struct emit_callback *ecb,
 711                                int prefix, const char *data, int size)
 712 {
 713         const char *endp = NULL;
 714         static const char *nneof = " No newline at end of file\n";
 715         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 716
 717         while (0 < size) {
 718                 int len;
 719
 720                 endp = memchr(data, '\n', size);
 721                 len = endp ? (endp - data + 1) : size;
 722                 if (prefix != '+') {
 723                         ecb->lno_in_preimage++;
 724                         emit_del_line(reset, ecb, data, len);
 725                 } else {
 726                         ecb->lno_in_postimage++;
 727                         emit_add_line(reset, ecb, data, len);
 728                 }
 729                 size -= len;
 730                 data += len;
 731         }
 732         if (!endp) {
 733                 const char *context = diff_get_color(ecb->color_diff,
 734                                                      DIFF_CONTEXT);
 735                 putc('\n', ecb->opt->file);
 736                 emit_line_0(ecb->opt, context, reset, '\\',
 737                             nneof, strlen(nneof));
 738         }
 739 }
 740
 741 static void emit_rewrite_diff(const char *name_a,
 742                               const char *name_b,
 743                               struct diff_filespec *one,
 744                               struct diff_filespec *two,
 745                               struct userdiff_driver *textconv_one,
 746                               struct userdiff_driver *textconv_two,
 747                               struct diff_options *o)
 748 {
 749         int lc_a, lc_b;
 750         const char *name_a_tab, *name_b_tab;
 751         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 752         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 753         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 754         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 755         const char *a_prefix, *b_prefix;
 756         char *data_one, *data_two;
 757         size_t size_one, size_two;
 758         struct emit_callback ecbdata;
 759         const char *line_prefix = diff_line_prefix(o);
 760
 761         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 762                 a_prefix = o->b_prefix;
 763                 b_prefix = o->a_prefix;
 764         } else {
 765                 a_prefix = o->a_prefix;
 766                 b_prefix = o->b_prefix;
 767         }
 768
 769         name_a += (*name_a == '/');
 770         name_b += (*name_b == '/');
 771         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 772         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 773
 774         strbuf_reset(&a_name);
 775         strbuf_reset(&b_name);
 776         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 777         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 778
 779         size_one = fill_textconv(textconv_one, one, &data_one);
 780         size_two = fill_textconv(textconv_two, two, &data_two);
 781
 782         memset(&ecbdata, 0, sizeof(ecbdata));
 783         ecbdata.color_diff = want_color(o->use_color);
 784         ecbdata.ws_rule = whitespace_rule(name_b);
 785         ecbdata.opt = o;
 786         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 787                 mmfile_t mf1, mf2;
 788                 mf1.ptr = (char *)data_one;
 789                 mf2.ptr = (char *)data_two;
 790                 mf1.size = size_one;
 791                 mf2.size = size_two;
 792                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 793         }
 794         ecbdata.lno_in_preimage = 1;
 795         ecbdata.lno_in_postimage = 1;
 796
 797         lc_a = count_lines(data_one, size_one);
 798         lc_b = count_lines(data_two, size_two);
 799         fprintf(o->file,
 800                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 801                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 802                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 803                 line_prefix, fraginfo);
 804         if (!o->irreversible_delete)
 805                 print_line_count(o->file, lc_a);
 806         else
 807                 fprintf(o->file, "?,?");
 808         fprintf(o->file, " +");
 809         print_line_count(o->file, lc_b);
 810         fprintf(o->file, " @@%s\n", reset);
 811         if (lc_a && !o->irreversible_delete)
 812                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 813         if (lc_b)
 814                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 815         if (textconv_one)
 816                 free((char *)data_one);
 817         if (textconv_two)
 818                 free((char *)data_two);
 819 }
 820
 821 struct diff_words_buffer {
 822         mmfile_t text;
 823         long alloc;
 824         struct diff_words_orig {
 825                 const char *begin, *end;
 826         } *orig;
 827         int orig_nr, orig_alloc;
 828 };
 829
 830 static void diff_words_append(char *line, unsigned long len,
 831                 struct diff_words_buffer *buffer)
 832 {
 833         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 834         line++;
 835         len--;
 836         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 837         buffer->text.size += len;
 838         buffer->text.ptr[buffer->text.size] = '\0';
 839 }
 840
 841 struct diff_words_style_elem {
 842         const char *prefix;
 843         const char *suffix;
 844         const char *color; /* NULL; filled in by the setup code if
 845                             * color is enabled */
 846 };
 847
 848 struct diff_words_style {
 849         enum diff_words_type type;
 850         struct diff_words_style_elem new, old, ctx;
 851         const char *newline;
 852 };
 853
 854 static struct diff_words_style diff_words_styles[] = {
 855         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 856         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 857         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 858 };
 859
 860 struct diff_words_data {
 861         struct diff_words_buffer minus, plus;
 862         const char *current_plus;
 863         int last_minus;
 864         struct diff_options *opt;
 865         regex_t *word_regex;
 866         enum diff_words_type type;
 867         struct diff_words_style *style;
 868 };
 869
 870 static int fn_out_diff_words_write_helper(FILE *fp,
 871                                           struct diff_words_style_elem *st_el,
 872                                           const char *newline,
 873                                           size_t count, const char *buf,
 874                                           const char *line_prefix)
 875 {
 876         int print = 0;
 877
 878         while (count) {
 879                 char *p = memchr(buf, '\n', count);
 880                 if (print)
 881                         fputs(line_prefix, fp);
 882                 if (p != buf) {
 883                         if (st_el->color && fputs(st_el->color, fp) < 0)
 884                                 return -1;
 885                         if (fputs(st_el->prefix, fp) < 0 ||
 886                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 887                             fputs(st_el->suffix, fp) < 0)
 888                                 return -1;
 889                         if (st_el->color && *st_el->color
 890                             && fputs(GIT_COLOR_RESET, fp) < 0)
 891                                 return -1;
 892                 }
 893                 if (!p)
 894                         return 0;
 895                 if (fputs(newline, fp) < 0)
 896                         return -1;
 897                 count -= p + 1 - buf;
 898                 buf = p + 1;
 899                 print = 1;
 900         }
 901         return 0;
 902 }
 903
 904 /*
 905  * '--color-words' algorithm can be described as:
 906  *
 907  *   1. collect a the minus/plus lines of a diff hunk, divided into
 908  *      minus-lines and plus-lines;
 909  *
 910  *   2. break both minus-lines and plus-lines into words and
 911  *      place them into two mmfile_t with one word for each line;
 912  *
 913  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 914  *
 915  * And for the common parts of the both file, we output the plus side text.
 916  * diff_words->current_plus is used to trace the current position of the plus file
 917  * which printed. diff_words->last_minus is used to trace the last minus word
 918  * printed.
 919  *
 920  * For '--graph' to work with '--color-words', we need to output the graph prefix
 921  * on each line of color words output. Generally, there are two conditions on
 922  * which we should output the prefix.
 923  *
 924  *   1. diff_words->last_minus == 0 &&
 925  *      diff_words->current_plus == diff_words->plus.text.ptr
 926  *
 927  *      that is: the plus text must start as a new line, and if there is no minus
 928  *      word printed, a graph prefix must be printed.
 929  *
 930  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 931  *      *(diff_words->current_plus - 1) == '\n'
 932  *
 933  *      that is: a graph prefix must be printed following a '\n'
 934  */
 935 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 936 {
 937         if ((diff_words->last_minus == 0 &&
 938                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 939                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 940                 *(diff_words->current_plus - 1) == '\n')) {
 941                 return 1;
 942         } else {
 943                 return 0;
 944         }
 945 }
 946
 947 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 948 {
 949         struct diff_words_data *diff_words = priv;
 950         struct diff_words_style *style = diff_words->style;
 951         int minus_first, minus_len, plus_first, plus_len;
 952         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 953         struct diff_options *opt = diff_words->opt;
 954         const char *line_prefix;
 955
 956         if (line[0] != '@' || parse_hunk_header(line, len,
 957                         &minus_first, &minus_len, &plus_first, &plus_len))
 958                 return;
 959
 960         assert(opt);
 961         line_prefix = diff_line_prefix(opt);
 962
 963         /* POSIX requires that first be decremented by one if len == 0... */
 964         if (minus_len) {
 965                 minus_begin = diff_words->minus.orig[minus_first].begin;
 966                 minus_end =
 967                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 968         } else
 969                 minus_begin = minus_end =
 970                         diff_words->minus.orig[minus_first].end;
 971
 972         if (plus_len) {
 973                 plus_begin = diff_words->plus.orig[plus_first].begin;
 974                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 975         } else
 976                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 977
 978         if (color_words_output_graph_prefix(diff_words)) {
 979                 fputs(line_prefix, diff_words->opt->file);
 980         }
 981         if (diff_words->current_plus != plus_begin) {
 982                 fn_out_diff_words_write_helper(diff_words->opt->file,
 983                                 &style->ctx, style->newline,
 984                                 plus_begin - diff_words->current_plus,
 985                                 diff_words->current_plus, line_prefix);
 986                 if (*(plus_begin - 1) == '\n')
 987                         fputs(line_prefix, diff_words->opt->file);
 988         }
 989         if (minus_begin != minus_end) {
 990                 fn_out_diff_words_write_helper(diff_words->opt->file,
 991                                 &style->old, style->newline,
 992                                 minus_end - minus_begin, minus_begin,
 993                                 line_prefix);
 994         }
 995         if (plus_begin != plus_end) {
 996                 fn_out_diff_words_write_helper(diff_words->opt->file,
 997                                 &style->new, style->newline,
 998                                 plus_end - plus_begin, plus_begin,
 999                                 line_prefix);
1000         }
1001
1002         diff_words->current_plus = plus_end;
1003         diff_words->last_minus = minus_first;
1004 }
1005
1006 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1007 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1008                 int *begin, int *end)
1009 {
1010         if (word_regex && *begin < buffer->size) {
1011                 regmatch_t match[1];
1012                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1013                                  buffer->size - *begin, 1, match, 0)) {
1014                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1015                                         '\n', match[0].rm_eo - match[0].rm_so);
1016                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1017                         *begin += match[0].rm_so;
1018                         return *begin >= *end;
1019                 }
1020                 return -1;
1021         }
1022
1023         /* find the next word */
1024         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1025                 (*begin)++;
1026         if (*begin >= buffer->size)
1027                 return -1;
1028
1029         /* find the end of the word */
1030         *end = *begin + 1;
1031         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1032                 (*end)++;
1033
1034         return 0;
1035 }
1036
1037 /*
1038  * This function splits the words in buffer->text, stores the list with
1039  * newline separator into out, and saves the offsets of the original words
1040  * in buffer->orig.
1041  */
1042 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1043                 regex_t *word_regex)
1044 {
1045         int i, j;
1046         long alloc = 0;
1047
1048         out->size = 0;
1049         out->ptr = NULL;
1050
1051         /* fake an empty "0th" word */
1052         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1053         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1054         buffer->orig_nr = 1;
1055
1056         for (i = 0; i < buffer->text.size; i++) {
1057                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1058                         return;
1059
1060                 /* store original boundaries */
1061                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1062                                 buffer->orig_alloc);
1063                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1064                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1065                 buffer->orig_nr++;
1066
1067                 /* store one word */
1068                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1069                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1070                 out->ptr[out->size + j - i] = '\n';
1071                 out->size += j - i + 1;
1072
1073                 i = j - 1;
1074         }
1075 }
1076
1077 /* this executes the word diff on the accumulated buffers */
1078 static void diff_words_show(struct diff_words_data *diff_words)
1079 {
1080         xpparam_t xpp;
1081         xdemitconf_t xecfg;
1082         mmfile_t minus, plus;
1083         struct diff_words_style *style = diff_words->style;
1084
1085         struct diff_options *opt = diff_words->opt;
1086         const char *line_prefix;
1087
1088         assert(opt);
1089         line_prefix = diff_line_prefix(opt);
1090
1091         /* special case: only removal */
1092         if (!diff_words->plus.text.size) {
1093                 fputs(line_prefix, diff_words->opt->file);
1094                 fn_out_diff_words_write_helper(diff_words->opt->file,
1095                         &style->old, style->newline,
1096                         diff_words->minus.text.size,
1097                         diff_words->minus.text.ptr, line_prefix);
1098                 diff_words->minus.text.size = 0;
1099                 return;
1100         }
1101
1102         diff_words->current_plus = diff_words->plus.text.ptr;
1103         diff_words->last_minus = 0;
1104
1105         memset(&xpp, 0, sizeof(xpp));
1106         memset(&xecfg, 0, sizeof(xecfg));
1107         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1108         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1109         xpp.flags = 0;
1110         /* as only the hunk header will be parsed, we need a 0-context */
1111         xecfg.ctxlen = 0;
1112         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1113                           &xpp, &xecfg))
1114                 die("unable to generate word diff");
1115         free(minus.ptr);
1116         free(plus.ptr);
1117         if (diff_words->current_plus != diff_words->plus.text.ptr +
1118                         diff_words->plus.text.size) {
1119                 if (color_words_output_graph_prefix(diff_words))
1120                         fputs(line_prefix, diff_words->opt->file);
1121                 fn_out_diff_words_write_helper(diff_words->opt->file,
1122                         &style->ctx, style->newline,
1123                         diff_words->plus.text.ptr + diff_words->plus.text.size
1124                         - diff_words->current_plus, diff_words->current_plus,
1125                         line_prefix);
1126         }
1127         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1128 }
1129
1130 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1131 static void diff_words_flush(struct emit_callback *ecbdata)
1132 {
1133         if (ecbdata->diff_words->minus.text.size ||
1134             ecbdata->diff_words->plus.text.size)
1135                 diff_words_show(ecbdata->diff_words);
1136 }
1137
1138 static void diff_filespec_load_driver(struct diff_filespec *one)
1139 {
1140         /* Use already-loaded driver */
1141         if (one->driver)
1142                 return;
1143
1144         if (S_ISREG(one->mode))
1145                 one->driver = userdiff_find_by_path(one->path);
1146
1147         /* Fallback to default settings */
1148         if (!one->driver)
1149                 one->driver = userdiff_find_by_name("default");
1150 }
1151
1152 static const char *userdiff_word_regex(struct diff_filespec *one)
1153 {
1154         diff_filespec_load_driver(one);
1155         return one->driver->word_regex;
1156 }
1157
1158 static void init_diff_words_data(struct emit_callback *ecbdata,
1159                                  struct diff_options *orig_opts,
1160                                  struct diff_filespec *one,
1161                                  struct diff_filespec *two)
1162 {
1163         int i;
1164         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1165         memcpy(o, orig_opts, sizeof(struct diff_options));
1166
1167         ecbdata->diff_words =
1168                 xcalloc(1, sizeof(struct diff_words_data));
1169         ecbdata->diff_words->type = o->word_diff;
1170         ecbdata->diff_words->opt = o;
1171         if (!o->word_regex)
1172                 o->word_regex = userdiff_word_regex(one);
1173         if (!o->word_regex)
1174                 o->word_regex = userdiff_word_regex(two);
1175         if (!o->word_regex)
1176                 o->word_regex = diff_word_regex_cfg;
1177         if (o->word_regex) {
1178                 ecbdata->diff_words->word_regex = (regex_t *)
1179                         xmalloc(sizeof(regex_t));
1180                 if (regcomp(ecbdata->diff_words->word_regex,
1181                             o->word_regex,
1182                             REG_EXTENDED | REG_NEWLINE))
1183                         die ("Invalid regular expression: %s",
1184                              o->word_regex);
1185         }
1186         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1187                 if (o->word_diff == diff_words_styles[i].type) {
1188                         ecbdata->diff_words->style =
1189                                 &diff_words_styles[i];
1190                         break;
1191                 }
1192         }
1193         if (want_color(o->use_color)) {
1194                 struct diff_words_style *st = ecbdata->diff_words->style;
1195                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1196                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1197                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1198         }
1199 }
1200
1201 static void free_diff_words_data(struct emit_callback *ecbdata)
1202 {
1203         if (ecbdata->diff_words) {
1204                 diff_words_flush(ecbdata);
1205                 free (ecbdata->diff_words->opt);
1206                 free (ecbdata->diff_words->minus.text.ptr);
1207                 free (ecbdata->diff_words->minus.orig);
1208                 free (ecbdata->diff_words->plus.text.ptr);
1209                 free (ecbdata->diff_words->plus.orig);
1210                 if (ecbdata->diff_words->word_regex) {
1211                         regfree(ecbdata->diff_words->word_regex);
1212                         free(ecbdata->diff_words->word_regex);
1213                 }
1214                 free(ecbdata->diff_words);
1215                 ecbdata->diff_words = NULL;
1216         }
1217 }
1218
1219 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1220 {
1221         if (want_color(diff_use_color))
1222                 return diff_colors[ix];
1223         return "";
1224 }
1225
1226 const char *diff_line_prefix(struct diff_options *opt)
1227 {
1228         struct strbuf *msgbuf;
1229         if (!opt->output_prefix)
1230                 return "";
1231
1232         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1233         return msgbuf->buf;
1234 }
1235
1236 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1237 {
1238         const char *cp;
1239         unsigned long allot;
1240         size_t l = len;
1241
1242         if (ecb->truncate)
1243                 return ecb->truncate(line, len);
1244         cp = line;
1245         allot = l;
1246         while (0 < l) {
1247                 (void) utf8_width(&cp, &l);
1248                 if (!cp)
1249                         break; /* truncated in the middle? */
1250         }
1251         return allot - l;
1252 }
1253
1254 static void find_lno(const char *line, struct emit_callback *ecbdata)
1255 {
1256         const char *p;
1257         ecbdata->lno_in_preimage = 0;
1258         ecbdata->lno_in_postimage = 0;
1259         p = strchr(line, '-');
1260         if (!p)
1261                 return; /* cannot happen */
1262         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1263         p = strchr(p, '+');
1264         if (!p)
1265                 return; /* cannot happen */
1266         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1267 }
1268
1269 static void fn_out_consume(void *priv, char *line, unsigned long len)
1270 {
1271         struct emit_callback *ecbdata = priv;
1272         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1273         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1274         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1275         struct diff_options *o = ecbdata->opt;
1276         const char *line_prefix = diff_line_prefix(o);
1277
1278         o->found_changes = 1;
1279
1280         if (ecbdata->header) {
1281                 fprintf(o->file, "%s", ecbdata->header->buf);
1282                 strbuf_reset(ecbdata->header);
1283                 ecbdata->header = NULL;
1284         }
1285
1286         if (ecbdata->label_path[0]) {
1287                 const char *name_a_tab, *name_b_tab;
1288
1289                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1290                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1291
1292                 fprintf(o->file, "%s%s--- %s%s%s\n",
1293                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1294                 fprintf(o->file, "%s%s+++ %s%s%s\n",
1295                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1296                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1297         }
1298
1299         if (diff_suppress_blank_empty
1300             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1301                 line[0] = '\n';
1302                 len = 1;
1303         }
1304
1305         if (line[0] == '@') {
1306                 if (ecbdata->diff_words)
1307                         diff_words_flush(ecbdata);
1308                 len = sane_truncate_line(ecbdata, line, len);
1309                 find_lno(line, ecbdata);
1310                 emit_hunk_header(ecbdata, line, len);
1311                 if (line[len-1] != '\n')
1312                         putc('\n', o->file);
1313                 return;
1314         }
1315
1316         if (ecbdata->diff_words) {
1317                 if (line[0] == '-') {
1318                         diff_words_append(line, len,
1319                                           &ecbdata->diff_words->minus);
1320                         return;
1321                 } else if (line[0] == '+') {
1322                         diff_words_append(line, len,
1323                                           &ecbdata->diff_words->plus);
1324                         return;
1325                 } else if (starts_with(line, "\\ ")) {
1326                         /*
1327                          * Eat the "no newline at eof" marker as if we
1328                          * saw a "+" or "-" line with nothing on it,
1329                          * and return without diff_words_flush() to
1330                          * defer processing. If this is the end of
1331                          * preimage, more "+" lines may come after it.
1332                          */
1333                         return;
1334                 }
1335                 diff_words_flush(ecbdata);
1336                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1337                         emit_line(o, context, reset, line, len);
1338                         fputs("~\n", o->file);
1339                 } else {
1340                         /*
1341                          * Skip the prefix character, if any.  With
1342                          * diff_suppress_blank_empty, there may be
1343                          * none.
1344                          */
1345                         if (line[0] != '\n') {
1346                               line++;
1347                               len--;
1348                         }
1349                         emit_line(o, context, reset, line, len);
1350                 }
1351                 return;
1352         }
1353
1354         switch (line[0]) {
1355         case '+':
1356                 ecbdata->lno_in_postimage++;
1357                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1358                 break;
1359         case '-':
1360                 ecbdata->lno_in_preimage++;
1361                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1362                 break;
1363         case ' ':
1364                 ecbdata->lno_in_postimage++;
1365                 ecbdata->lno_in_preimage++;
1366                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1367                 break;
1368         default:
1369                 /* incomplete line at the end */
1370                 ecbdata->lno_in_preimage++;
1371                 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1372                           reset, line, len);
1373                 break;
1374         }
1375 }
1376
1377 static char *pprint_rename(const char *a, const char *b)
1378 {
1379         const char *old = a;
1380         const char *new = b;
1381         struct strbuf name = STRBUF_INIT;
1382         int pfx_length, sfx_length;
1383         int pfx_adjust_for_slash;
1384         int len_a = strlen(a);
1385         int len_b = strlen(b);
1386         int a_midlen, b_midlen;
1387         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1388         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1389
1390         if (qlen_a || qlen_b) {
1391                 quote_c_style(a, &name, NULL, 0);
1392                 strbuf_addstr(&name, " => ");
1393                 quote_c_style(b, &name, NULL, 0);
1394                 return strbuf_detach(&name, NULL);
1395         }
1396
1397         /* Find common prefix */
1398         pfx_length = 0;
1399         while (*old && *new && *old == *new) {
1400                 if (*old == '/')
1401                         pfx_length = old - a + 1;
1402                 old++;
1403                 new++;
1404         }
1405
1406         /* Find common suffix */
1407         old = a + len_a;
1408         new = b + len_b;
1409         sfx_length = 0;
1410         /*
1411          * If there is a common prefix, it must end in a slash.  In
1412          * that case we let this loop run 1 into the prefix to see the
1413          * same slash.
1414          *
1415          * If there is no common prefix, we cannot do this as it would
1416          * underrun the input strings.
1417          */
1418         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1419         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1420                b + pfx_length - pfx_adjust_for_slash <= new &&
1421                *old == *new) {
1422                 if (*old == '/')
1423                         sfx_length = len_a - (old - a);
1424                 old--;
1425                 new--;
1426         }
1427
1428         /*
1429          * pfx{mid-a => mid-b}sfx
1430          * {pfx-a => pfx-b}sfx
1431          * pfx{sfx-a => sfx-b}
1432          * name-a => name-b
1433          */
1434         a_midlen = len_a - pfx_length - sfx_length;
1435         b_midlen = len_b - pfx_length - sfx_length;
1436         if (a_midlen < 0)
1437                 a_midlen = 0;
1438         if (b_midlen < 0)
1439                 b_midlen = 0;
1440
1441         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1442         if (pfx_length + sfx_length) {
1443                 strbuf_add(&name, a, pfx_length);
1444                 strbuf_addch(&name, '{');
1445         }
1446         strbuf_add(&name, a + pfx_length, a_midlen);
1447         strbuf_addstr(&name, " => ");
1448         strbuf_add(&name, b + pfx_length, b_midlen);
1449         if (pfx_length + sfx_length) {
1450                 strbuf_addch(&name, '}');
1451                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1452         }
1453         return strbuf_detach(&name, NULL);
1454 }
1455
1456 struct diffstat_t {
1457         int nr;
1458         int alloc;
1459         struct diffstat_file {
1460                 char *from_name;
1461                 char *name;
1462                 char *print_name;
1463                 unsigned is_unmerged:1;
1464                 unsigned is_binary:1;
1465                 unsigned is_renamed:1;
1466                 unsigned is_interesting:1;
1467                 uintmax_t added, deleted;
1468         } **files;
1469 };
1470
1471 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1472                                           const char *name_a,
1473                                           const char *name_b)
1474 {
1475         struct diffstat_file *x;
1476         x = xcalloc(1, sizeof(*x));
1477         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1478         diffstat->files[diffstat->nr++] = x;
1479         if (name_b) {
1480                 x->from_name = xstrdup(name_a);
1481                 x->name = xstrdup(name_b);
1482                 x->is_renamed = 1;
1483         }
1484         else {
1485                 x->from_name = NULL;
1486                 x->name = xstrdup(name_a);
1487         }
1488         return x;
1489 }
1490
1491 static void diffstat_consume(void *priv, char *line, unsigned long len)
1492 {
1493         struct diffstat_t *diffstat = priv;
1494         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1495
1496         if (line[0] == '+')
1497                 x->added++;
1498         else if (line[0] == '-')
1499                 x->deleted++;
1500 }
1501
1502 const char mime_boundary_leader[] = "------------";
1503
1504 static int scale_linear(int it, int width, int max_change)
1505 {
1506         if (!it)
1507                 return 0;
1508         /*
1509          * make sure that at least one '-' or '+' is printed if
1510          * there is any change to this path. The easiest way is to
1511          * scale linearly as if the alloted width is one column shorter
1512          * than it is, and then add 1 to the result.
1513          */
1514         return 1 + (it * (width - 1) / max_change);
1515 }
1516
1517 static void show_name(FILE *file,
1518                       const char *prefix, const char *name, int len)
1519 {
1520         fprintf(file, " %s%-*s |", prefix, len, name);
1521 }
1522
1523 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1524 {
1525         if (cnt <= 0)
1526                 return;
1527         fprintf(file, "%s", set);
1528         while (cnt--)
1529                 putc(ch, file);
1530         fprintf(file, "%s", reset);
1531 }
1532
1533 static void fill_print_name(struct diffstat_file *file)
1534 {
1535         char *pname;
1536
1537         if (file->print_name)
1538                 return;
1539
1540         if (!file->is_renamed) {
1541                 struct strbuf buf = STRBUF_INIT;
1542                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1543                         pname = strbuf_detach(&buf, NULL);
1544                 } else {
1545                         pname = file->name;
1546                         strbuf_release(&buf);
1547                 }
1548         } else {
1549                 pname = pprint_rename(file->from_name, file->name);
1550         }
1551         file->print_name = pname;
1552 }
1553
1554 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1555 {
1556         struct strbuf sb = STRBUF_INIT;
1557         int ret;
1558
1559         if (!files) {
1560                 assert(insertions == 0 && deletions == 0);
1561                 return fprintf(fp, "%s\n", " 0 files changed");
1562         }
1563
1564         strbuf_addf(&sb,
1565                     (files == 1) ? " %d file changed" : " %d files changed",
1566                     files);
1567
1568         /*
1569          * For binary diff, the caller may want to print "x files
1570          * changed" with insertions == 0 && deletions == 0.
1571          *
1572          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1573          * is probably less confusing (i.e skip over "2 files changed
1574          * but nothing about added/removed lines? Is this a bug in Git?").
1575          */
1576         if (insertions || deletions == 0) {
1577                 strbuf_addf(&sb,
1578                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1579                             insertions);
1580         }
1581
1582         if (deletions || insertions == 0) {
1583                 strbuf_addf(&sb,
1584                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1585                             deletions);
1586         }
1587         strbuf_addch(&sb, '\n');
1588         ret = fputs(sb.buf, fp);
1589         strbuf_release(&sb);
1590         return ret;
1591 }
1592
1593 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1594 {
1595         int i, len, add, del, adds = 0, dels = 0;
1596         uintmax_t max_change = 0, max_len = 0;
1597         int total_files = data->nr, count;
1598         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1599         const char *reset, *add_c, *del_c;
1600         const char *line_prefix = "";
1601         int extra_shown = 0;
1602
1603         if (data->nr == 0)
1604                 return;
1605
1606         line_prefix = diff_line_prefix(options);
1607         count = options->stat_count ? options->stat_count : data->nr;
1608
1609         reset = diff_get_color_opt(options, DIFF_RESET);
1610         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1611         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1612
1613         /*
1614          * Find the longest filename and max number of changes
1615          */
1616         for (i = 0; (i < count) && (i < data->nr); i++) {
1617                 struct diffstat_file *file = data->files[i];
1618                 uintmax_t change = file->added + file->deleted;
1619
1620                 if (!file->is_interesting && (change == 0)) {
1621                         count++; /* not shown == room for one more */
1622                         continue;
1623                 }
1624                 fill_print_name(file);
1625                 len = strlen(file->print_name);
1626                 if (max_len < len)
1627                         max_len = len;
1628
1629                 if (file->is_unmerged) {
1630                         /* "Unmerged" is 8 characters */
1631                         bin_width = bin_width < 8 ? 8 : bin_width;
1632                         continue;
1633                 }
1634                 if (file->is_binary) {
1635                         /* "Bin XXX -> YYY bytes" */
1636                         int w = 14 + decimal_width(file->added)
1637                                 + decimal_width(file->deleted);
1638                         bin_width = bin_width < w ? w : bin_width;
1639                         /* Display change counts aligned with "Bin" */
1640                         number_width = 3;
1641                         continue;
1642                 }
1643
1644                 if (max_change < change)
1645                         max_change = change;
1646         }
1647         count = i; /* where we can stop scanning in data->files[] */
1648
1649         /*
1650          * We have width = stat_width or term_columns() columns total.
1651          * We want a maximum of min(max_len, stat_name_width) for the name part.
1652          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1653          * We also need 1 for " " and 4 + decimal_width(max_change)
1654          * for " | NNNN " and one the empty column at the end, altogether
1655          * 6 + decimal_width(max_change).
1656          *
1657          * If there's not enough space, we will use the smaller of
1658          * stat_name_width (if set) and 5/8*width for the filename,
1659          * and the rest for constant elements + graph part, but no more
1660          * than stat_graph_width for the graph part.
1661          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1662          * for the standard terminal size).
1663          *
1664          * In other words: stat_width limits the maximum width, and
1665          * stat_name_width fixes the maximum width of the filename,
1666          * and is also used to divide available columns if there
1667          * aren't enough.
1668          *
1669          * Binary files are displayed with "Bin XXX -> YYY bytes"
1670          * instead of the change count and graph. This part is treated
1671          * similarly to the graph part, except that it is not
1672          * "scaled". If total width is too small to accommodate the
1673          * guaranteed minimum width of the filename part and the
1674          * separators and this message, this message will "overflow"
1675          * making the line longer than the maximum width.
1676          */
1677
1678         if (options->stat_width == -1)
1679                 width = term_columns() - strlen(line_prefix);
1680         else
1681                 width = options->stat_width ? options->stat_width : 80;
1682         number_width = decimal_width(max_change) > number_width ?
1683                 decimal_width(max_change) : number_width;
1684
1685         if (options->stat_graph_width == -1)
1686                 options->stat_graph_width = diff_stat_graph_width;
1687
1688         /*
1689          * Guarantee 3/8*16==6 for the graph part
1690          * and 5/8*16==10 for the filename part
1691          */
1692         if (width < 16 + 6 + number_width)
1693                 width = 16 + 6 + number_width;
1694
1695         /*
1696          * First assign sizes that are wanted, ignoring available width.
1697          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1698          * starting from "XXX" should fit in graph_width.
1699          */
1700         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1701         if (options->stat_graph_width &&
1702             options->stat_graph_width < graph_width)
1703                 graph_width = options->stat_graph_width;
1704
1705         name_width = (options->stat_name_width > 0 &&
1706                       options->stat_name_width < max_len) ?
1707                 options->stat_name_width : max_len;
1708
1709         /*
1710          * Adjust adjustable widths not to exceed maximum width
1711          */
1712         if (name_width + number_width + 6 + graph_width > width) {
1713                 if (graph_width > width * 3/8 - number_width - 6) {
1714                         graph_width = width * 3/8 - number_width - 6;
1715                         if (graph_width < 6)
1716                                 graph_width = 6;
1717                 }
1718
1719                 if (options->stat_graph_width &&
1720                     graph_width > options->stat_graph_width)
1721                         graph_width = options->stat_graph_width;
1722                 if (name_width > width - number_width - 6 - graph_width)
1723                         name_width = width - number_width - 6 - graph_width;
1724                 else
1725                         graph_width = width - number_width - 6 - name_width;
1726         }
1727
1728         /*
1729          * From here name_width is the width of the name area,
1730          * and graph_width is the width of the graph area.
1731          * max_change is used to scale graph properly.
1732          */
1733         for (i = 0; i < count; i++) {
1734                 const char *prefix = "";
1735                 struct diffstat_file *file = data->files[i];
1736                 char *name = file->print_name;
1737                 uintmax_t added = file->added;
1738                 uintmax_t deleted = file->deleted;
1739                 int name_len;
1740
1741                 if (!file->is_interesting && (added + deleted == 0))
1742                         continue;
1743
1744                 /*
1745                  * "scale" the filename
1746                  */
1747                 len = name_width;
1748                 name_len = strlen(name);
1749                 if (name_width < name_len) {
1750                         char *slash;
1751                         prefix = "...";
1752                         len -= 3;
1753                         name += name_len - len;
1754                         slash = strchr(name, '/');
1755                         if (slash)
1756                                 name = slash;
1757                 }
1758
1759                 if (file->is_binary) {
1760                         fprintf(options->file, "%s", line_prefix);
1761                         show_name(options->file, prefix, name, len);
1762                         fprintf(options->file, " %*s", number_width, "Bin");
1763                         if (!added && !deleted) {
1764                                 putc('\n', options->file);
1765                                 continue;
1766                         }
1767                         fprintf(options->file, " %s%"PRIuMAX"%s",
1768                                 del_c, deleted, reset);
1769                         fprintf(options->file, " -> ");
1770                         fprintf(options->file, "%s%"PRIuMAX"%s",
1771                                 add_c, added, reset);
1772                         fprintf(options->file, " bytes");
1773                         fprintf(options->file, "\n");
1774                         continue;
1775                 }
1776                 else if (file->is_unmerged) {
1777                         fprintf(options->file, "%s", line_prefix);
1778                         show_name(options->file, prefix, name, len);
1779                         fprintf(options->file, " Unmerged\n");
1780                         continue;
1781                 }
1782
1783                 /*
1784                  * scale the add/delete
1785                  */
1786                 add = added;
1787                 del = deleted;
1788
1789                 if (graph_width <= max_change) {
1790                         int total = scale_linear(add + del, graph_width, max_change);
1791                         if (total < 2 && add && del)
1792                                 /* width >= 2 due to the sanity check */
1793                                 total = 2;
1794                         if (add < del) {
1795                                 add = scale_linear(add, graph_width, max_change);
1796                                 del = total - add;
1797                         } else {
1798                                 del = scale_linear(del, graph_width, max_change);
1799                                 add = total - del;
1800                         }
1801                 }
1802                 fprintf(options->file, "%s", line_prefix);
1803                 show_name(options->file, prefix, name, len);
1804                 fprintf(options->file, " %*"PRIuMAX"%s",
1805                         number_width, added + deleted,
1806                         added + deleted ? " " : "");
1807                 show_graph(options->file, '+', add, add_c, reset);
1808                 show_graph(options->file, '-', del, del_c, reset);
1809                 fprintf(options->file, "\n");
1810         }
1811
1812         for (i = 0; i < data->nr; i++) {
1813                 struct diffstat_file *file = data->files[i];
1814                 uintmax_t added = file->added;
1815                 uintmax_t deleted = file->deleted;
1816
1817                 if (file->is_unmerged ||
1818                     (!file->is_interesting && (added + deleted == 0))) {
1819                         total_files--;
1820                         continue;
1821                 }
1822
1823                 if (!file->is_binary) {
1824                         adds += added;
1825                         dels += deleted;
1826                 }
1827                 if (i < count)
1828                         continue;
1829                 if (!extra_shown)
1830                         fprintf(options->file, "%s ...\n", line_prefix);
1831                 extra_shown = 1;
1832         }
1833         fprintf(options->file, "%s", line_prefix);
1834         print_stat_summary(options->file, total_files, adds, dels);
1835 }
1836
1837 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1838 {
1839         int i, adds = 0, dels = 0, total_files = data->nr;
1840
1841         if (data->nr == 0)
1842                 return;
1843
1844         for (i = 0; i < data->nr; i++) {
1845                 int added = data->files[i]->added;
1846                 int deleted= data->files[i]->deleted;
1847
1848                 if (data->files[i]->is_unmerged ||
1849                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1850                         total_files--;
1851                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1852                         adds += added;
1853                         dels += deleted;
1854                 }
1855         }
1856         fprintf(options->file, "%s", diff_line_prefix(options));
1857         print_stat_summary(options->file, total_files, adds, dels);
1858 }
1859
1860 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1861 {
1862         int i;
1863
1864         if (data->nr == 0)
1865                 return;
1866
1867         for (i = 0; i < data->nr; i++) {
1868                 struct diffstat_file *file = data->files[i];
1869
1870                 fprintf(options->file, "%s", diff_line_prefix(options));
1871
1872                 if (file->is_binary)
1873                         fprintf(options->file, "-\t-\t");
1874                 else
1875                         fprintf(options->file,
1876                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1877                                 file->added, file->deleted);
1878                 if (options->line_termination) {
1879                         fill_print_name(file);
1880                         if (!file->is_renamed)
1881                                 write_name_quoted(file->name, options->file,
1882                                                   options->line_termination);
1883                         else {
1884                                 fputs(file->print_name, options->file);
1885                                 putc(options->line_termination, options->file);
1886                         }
1887                 } else {
1888                         if (file->is_renamed) {
1889                                 putc('\0', options->file);
1890                                 write_name_quoted(file->from_name, options->file, '\0');
1891                         }
1892                         write_name_quoted(file->name, options->file, '\0');
1893                 }
1894         }
1895 }
1896
1897 struct dirstat_file {
1898         const char *name;
1899         unsigned long changed;
1900 };
1901
1902 struct dirstat_dir {
1903         struct dirstat_file *files;
1904         int alloc, nr, permille, cumulative;
1905 };
1906
1907 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1908                 unsigned long changed, const char *base, int baselen)
1909 {
1910         unsigned long this_dir = 0;
1911         unsigned int sources = 0;
1912         const char *line_prefix = diff_line_prefix(opt);
1913
1914         while (dir->nr) {
1915                 struct dirstat_file *f = dir->files;
1916                 int namelen = strlen(f->name);
1917                 unsigned long this;
1918                 char *slash;
1919
1920                 if (namelen < baselen)
1921                         break;
1922                 if (memcmp(f->name, base, baselen))
1923                         break;
1924                 slash = strchr(f->name + baselen, '/');
1925                 if (slash) {
1926                         int newbaselen = slash + 1 - f->name;
1927                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1928                         sources++;
1929                 } else {
1930                         this = f->changed;
1931                         dir->files++;
1932                         dir->nr--;
1933                         sources += 2;
1934                 }
1935                 this_dir += this;
1936         }
1937
1938         /*
1939          * We don't report dirstat's for
1940          *  - the top level
1941          *  - or cases where everything came from a single directory
1942          *    under this directory (sources == 1).
1943          */
1944         if (baselen && sources != 1) {
1945                 if (this_dir) {
1946                         int permille = this_dir * 1000 / changed;
1947                         if (permille >= dir->permille) {
1948                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1949                                         permille / 10, permille % 10, baselen, base);
1950                                 if (!dir->cumulative)
1951                                         return 0;
1952                         }
1953                 }
1954         }
1955         return this_dir;
1956 }
1957
1958 static int dirstat_compare(const void *_a, const void *_b)
1959 {
1960         const struct dirstat_file *a = _a;
1961         const struct dirstat_file *b = _b;
1962         return strcmp(a->name, b->name);
1963 }
1964
1965 static void show_dirstat(struct diff_options *options)
1966 {
1967         int i;
1968         unsigned long changed;
1969         struct dirstat_dir dir;
1970         struct diff_queue_struct *q = &diff_queued_diff;
1971
1972         dir.files = NULL;
1973         dir.alloc = 0;
1974         dir.nr = 0;
1975         dir.permille = options->dirstat_permille;
1976         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1977
1978         changed = 0;
1979         for (i = 0; i < q->nr; i++) {
1980                 struct diff_filepair *p = q->queue[i];
1981                 const char *name;
1982                 unsigned long copied, added, damage;
1983                 int content_changed;
1984
1985                 name = p->two->path ? p->two->path : p->one->path;
1986
1987                 if (p->one->oid_valid && p->two->oid_valid)
1988                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
1989                 else
1990                         content_changed = 1;
1991
1992                 if (!content_changed) {
1993                         /*
1994                          * The SHA1 has not changed, so pre-/post-content is
1995                          * identical. We can therefore skip looking at the
1996                          * file contents altogether.
1997                          */
1998                         damage = 0;
1999                         goto found_damage;
2000                 }
2001
2002                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2003                         /*
2004                          * In --dirstat-by-file mode, we don't really need to
2005                          * look at the actual file contents at all.
2006                          * The fact that the SHA1 changed is enough for us to
2007                          * add this file to the list of results
2008                          * (with each file contributing equal damage).
2009                          */
2010                         damage = 1;
2011                         goto found_damage;
2012                 }
2013
2014                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2015                         diff_populate_filespec(p->one, 0);
2016                         diff_populate_filespec(p->two, 0);
2017                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2018                                                &copied, &added);
2019                         diff_free_filespec_data(p->one);
2020                         diff_free_filespec_data(p->two);
2021                 } else if (DIFF_FILE_VALID(p->one)) {
2022                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2023                         copied = added = 0;
2024                         diff_free_filespec_data(p->one);
2025                 } else if (DIFF_FILE_VALID(p->two)) {
2026                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2027                         copied = 0;
2028                         added = p->two->size;
2029                         diff_free_filespec_data(p->two);
2030                 } else
2031                         continue;
2032
2033                 /*
2034                  * Original minus copied is the removed material,
2035                  * added is the new material.  They are both damages
2036                  * made to the preimage.
2037                  * If the resulting damage is zero, we know that
2038                  * diffcore_count_changes() considers the two entries to
2039                  * be identical, but since content_changed is true, we
2040                  * know that there must have been _some_ kind of change,
2041                  * so we force all entries to have damage > 0.
2042                  */
2043                 damage = (p->one->size - copied) + added;
2044                 if (!damage)
2045                         damage = 1;
2046
2047 found_damage:
2048                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2049                 dir.files[dir.nr].name = name;
2050                 dir.files[dir.nr].changed = damage;
2051                 changed += damage;
2052                 dir.nr++;
2053         }
2054
2055         /* This can happen even with many files, if everything was renames */
2056         if (!changed)
2057                 return;
2058
2059         /* Show all directories with more than x% of the changes */
2060         QSORT(dir.files, dir.nr, dirstat_compare);
2061         gather_dirstat(options, &dir, changed, "", 0);
2062 }
2063
2064 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2065 {
2066         int i;
2067         unsigned long changed;
2068         struct dirstat_dir dir;
2069
2070         if (data->nr == 0)
2071                 return;
2072
2073         dir.files = NULL;
2074         dir.alloc = 0;
2075         dir.nr = 0;
2076         dir.permille = options->dirstat_permille;
2077         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2078
2079         changed = 0;
2080         for (i = 0; i < data->nr; i++) {
2081                 struct diffstat_file *file = data->files[i];
2082                 unsigned long damage = file->added + file->deleted;
2083                 if (file->is_binary)
2084                         /*
2085                          * binary files counts bytes, not lines. Must find some
2086                          * way to normalize binary bytes vs. textual lines.
2087                          * The following heuristic assumes that there are 64
2088                          * bytes per "line".
2089                          * This is stupid and ugly, but very cheap...
2090                          */
2091                         damage = (damage + 63) / 64;
2092                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2093                 dir.files[dir.nr].name = file->name;
2094                 dir.files[dir.nr].changed = damage;
2095                 changed += damage;
2096                 dir.nr++;
2097         }
2098
2099         /* This can happen even with many files, if everything was renames */
2100         if (!changed)
2101                 return;
2102
2103         /* Show all directories with more than x% of the changes */
2104         QSORT(dir.files, dir.nr, dirstat_compare);
2105         gather_dirstat(options, &dir, changed, "", 0);
2106 }
2107
2108 static void free_diffstat_info(struct diffstat_t *diffstat)
2109 {
2110         int i;
2111         for (i = 0; i < diffstat->nr; i++) {
2112                 struct diffstat_file *f = diffstat->files[i];
2113                 if (f->name != f->print_name)
2114                         free(f->print_name);
2115                 free(f->name);
2116                 free(f->from_name);
2117                 free(f);
2118         }
2119         free(diffstat->files);
2120 }
2121
2122 struct checkdiff_t {
2123         const char *filename;
2124         int lineno;
2125         int conflict_marker_size;
2126         struct diff_options *o;
2127         unsigned ws_rule;
2128         unsigned status;
2129 };
2130
2131 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2132 {
2133         char firstchar;
2134         int cnt;
2135
2136         if (len < marker_size + 1)
2137                 return 0;
2138         firstchar = line[0];
2139         switch (firstchar) {
2140         case '=': case '>': case '<': case '|':
2141                 break;
2142         default:
2143                 return 0;
2144         }
2145         for (cnt = 1; cnt < marker_size; cnt++)
2146                 if (line[cnt] != firstchar)
2147                         return 0;
2148         /* line[1] thru line[marker_size-1] are same as firstchar */
2149         if (len < marker_size + 1 || !isspace(line[marker_size]))
2150                 return 0;
2151         return 1;
2152 }
2153
2154 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2155 {
2156         struct checkdiff_t *data = priv;
2157         int marker_size = data->conflict_marker_size;
2158         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2159         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2160         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2161         char *err;
2162         const char *line_prefix;
2163
2164         assert(data->o);
2165         line_prefix = diff_line_prefix(data->o);
2166
2167         if (line[0] == '+') {
2168                 unsigned bad;
2169                 data->lineno++;
2170                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2171                         data->status |= 1;
2172                         fprintf(data->o->file,
2173                                 "%s%s:%d: leftover conflict marker\n",
2174                                 line_prefix, data->filename, data->lineno);
2175                 }
2176                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2177                 if (!bad)
2178                         return;
2179                 data->status |= bad;
2180                 err = whitespace_error_string(bad);
2181                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2182                         line_prefix, data->filename, data->lineno, err);
2183                 free(err);
2184                 emit_line(data->o, set, reset, line, 1);
2185                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2186                               data->o->file, set, reset, ws);
2187         } else if (line[0] == ' ') {
2188                 data->lineno++;
2189         } else if (line[0] == '@') {
2190                 char *plus = strchr(line, '+');
2191                 if (plus)
2192                         data->lineno = strtol(plus, NULL, 10) - 1;
2193                 else
2194                         die("invalid diff");
2195         }
2196 }
2197
2198 static unsigned char *deflate_it(char *data,
2199                                  unsigned long size,
2200                                  unsigned long *result_size)
2201 {
2202         int bound;
2203         unsigned char *deflated;
2204         git_zstream stream;
2205
2206         git_deflate_init(&stream, zlib_compression_level);
2207         bound = git_deflate_bound(&stream, size);
2208         deflated = xmalloc(bound);
2209         stream.next_out = deflated;
2210         stream.avail_out = bound;
2211
2212         stream.next_in = (unsigned char *)data;
2213         stream.avail_in = size;
2214         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2215                 ; /* nothing */
2216         git_deflate_end(&stream);
2217         *result_size = stream.total_out;
2218         return deflated;
2219 }
2220
2221 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2222                                   const char *prefix)
2223 {
2224         void *cp;
2225         void *delta;
2226         void *deflated;
2227         void *data;
2228         unsigned long orig_size;
2229         unsigned long delta_size;
2230         unsigned long deflate_size;
2231         unsigned long data_size;
2232
2233         /* We could do deflated delta, or we could do just deflated two,
2234          * whichever is smaller.
2235          */
2236         delta = NULL;
2237         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2238         if (one->size && two->size) {
2239                 delta = diff_delta(one->ptr, one->size,
2240                                    two->ptr, two->size,
2241                                    &delta_size, deflate_size);
2242                 if (delta) {
2243                         void *to_free = delta;
2244                         orig_size = delta_size;
2245                         delta = deflate_it(delta, delta_size, &delta_size);
2246                         free(to_free);
2247                 }
2248         }
2249
2250         if (delta && delta_size < deflate_size) {
2251                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2252                 free(deflated);
2253                 data = delta;
2254                 data_size = delta_size;
2255         }
2256         else {
2257                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2258                 free(delta);
2259                 data = deflated;
2260                 data_size = deflate_size;
2261         }
2262
2263         /* emit data encoded in base85 */
2264         cp = data;
2265         while (data_size) {
2266                 int bytes = (52 < data_size) ? 52 : data_size;
2267                 char line[70];
2268                 data_size -= bytes;
2269                 if (bytes <= 26)
2270                         line[0] = bytes + 'A' - 1;
2271                 else
2272                         line[0] = bytes - 26 + 'a' - 1;
2273                 encode_85(line + 1, cp, bytes);
2274                 cp = (char *) cp + bytes;
2275                 fprintf(file, "%s", prefix);
2276                 fputs(line, file);
2277                 fputc('\n', file);
2278         }
2279         fprintf(file, "%s\n", prefix);
2280         free(data);
2281 }
2282
2283 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2284                              const char *prefix)
2285 {
2286         fprintf(file, "%sGIT binary patch\n", prefix);
2287         emit_binary_diff_body(file, one, two, prefix);
2288         emit_binary_diff_body(file, two, one, prefix);
2289 }
2290
2291 int diff_filespec_is_binary(struct diff_filespec *one)
2292 {
2293         if (one->is_binary == -1) {
2294                 diff_filespec_load_driver(one);
2295                 if (one->driver->binary != -1)
2296                         one->is_binary = one->driver->binary;
2297                 else {
2298                         if (!one->data && DIFF_FILE_VALID(one))
2299                                 diff_populate_filespec(one, CHECK_BINARY);
2300                         if (one->is_binary == -1 && one->data)
2301                                 one->is_binary = buffer_is_binary(one->data,
2302                                                 one->size);
2303                         if (one->is_binary == -1)
2304                                 one->is_binary = 0;
2305                 }
2306         }
2307         return one->is_binary;
2308 }
2309
2310 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2311 {
2312         diff_filespec_load_driver(one);
2313         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2314 }
2315
2316 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2317 {
2318         if (!options->a_prefix)
2319                 options->a_prefix = a;
2320         if (!options->b_prefix)
2321                 options->b_prefix = b;
2322 }
2323
2324 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2325 {
2326         if (!DIFF_FILE_VALID(one))
2327                 return NULL;
2328
2329         diff_filespec_load_driver(one);
2330         return userdiff_get_textconv(one->driver);
2331 }
2332
2333 static void builtin_diff(const char *name_a,
2334                          const char *name_b,
2335                          struct diff_filespec *one,
2336                          struct diff_filespec *two,
2337                          const char *xfrm_msg,
2338                          int must_show_header,
2339                          struct diff_options *o,
2340                          int complete_rewrite)
2341 {
2342         mmfile_t mf1, mf2;
2343         const char *lbl[2];
2344         char *a_one, *b_two;
2345         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2346         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2347         const char *a_prefix, *b_prefix;
2348         struct userdiff_driver *textconv_one = NULL;
2349         struct userdiff_driver *textconv_two = NULL;
2350         struct strbuf header = STRBUF_INIT;
2351         const char *line_prefix = diff_line_prefix(o);
2352
2353         diff_set_mnemonic_prefix(o, "a/", "b/");
2354         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2355                 a_prefix = o->b_prefix;
2356                 b_prefix = o->a_prefix;
2357         } else {
2358                 a_prefix = o->a_prefix;
2359                 b_prefix = o->b_prefix;
2360         }
2361
2362         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2363             (!one->mode || S_ISGITLINK(one->mode)) &&
2364             (!two->mode || S_ISGITLINK(two->mode))) {
2365                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2366                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2367                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2368                                 line_prefix,
2369                                 &one->oid, &two->oid,
2370                                 two->dirty_submodule,
2371                                 meta, del, add, reset);
2372                 return;
2373         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2374                    (!one->mode || S_ISGITLINK(one->mode)) &&
2375                    (!two->mode || S_ISGITLINK(two->mode))) {
2376                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2377                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2378                 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2379                                 line_prefix,
2380                                 &one->oid, &two->oid,
2381                                 two->dirty_submodule,
2382                                 meta, del, add, reset, o);
2383                 return;
2384         }
2385
2386         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2387                 textconv_one = get_textconv(one);
2388                 textconv_two = get_textconv(two);
2389         }
2390
2391         /* Never use a non-valid filename anywhere if at all possible */
2392         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2393         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2394
2395         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2396         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2397         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2398         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2399         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2400         if (lbl[0][0] == '/') {
2401                 /* /dev/null */
2402                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2403                 if (xfrm_msg)
2404                         strbuf_addstr(&header, xfrm_msg);
2405                 must_show_header = 1;
2406         }
2407         else if (lbl[1][0] == '/') {
2408                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2409                 if (xfrm_msg)
2410                         strbuf_addstr(&header, xfrm_msg);
2411                 must_show_header = 1;
2412         }
2413         else {
2414                 if (one->mode != two->mode) {
2415                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2416                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2417                         must_show_header = 1;
2418                 }
2419                 if (xfrm_msg)
2420                         strbuf_addstr(&header, xfrm_msg);
2421
2422                 /*
2423                  * we do not run diff between different kind
2424                  * of objects.
2425                  */
2426                 if ((one->mode ^ two->mode) & S_IFMT)
2427                         goto free_ab_and_return;
2428                 if (complete_rewrite &&
2429                     (textconv_one || !diff_filespec_is_binary(one)) &&
2430                     (textconv_two || !diff_filespec_is_binary(two))) {
2431                         fprintf(o->file, "%s", header.buf);
2432                         strbuf_reset(&header);
2433                         emit_rewrite_diff(name_a, name_b, one, two,
2434                                                 textconv_one, textconv_two, o);
2435                         o->found_changes = 1;
2436                         goto free_ab_and_return;
2437                 }
2438         }
2439
2440         if (o->irreversible_delete && lbl[1][0] == '/') {
2441                 fprintf(o->file, "%s", header.buf);
2442                 strbuf_reset(&header);
2443                 goto free_ab_and_return;
2444         } else if (!DIFF_OPT_TST(o, TEXT) &&
2445             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2446               (!textconv_two && diff_filespec_is_binary(two)) )) {
2447                 if (!one->data && !two->data &&
2448                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2449                     !DIFF_OPT_TST(o, BINARY)) {
2450                         if (!oidcmp(&one->oid, &two->oid)) {
2451                                 if (must_show_header)
2452                                         fprintf(o->file, "%s", header.buf);
2453                                 goto free_ab_and_return;
2454                         }
2455                         fprintf(o->file, "%s", header.buf);
2456                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2457                                 line_prefix, lbl[0], lbl[1]);
2458                         goto free_ab_and_return;
2459                 }
2460                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2461                         die("unable to read files to diff");
2462                 /* Quite common confusing case */
2463                 if (mf1.size == mf2.size &&
2464                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2465                         if (must_show_header)
2466                                 fprintf(o->file, "%s", header.buf);
2467                         goto free_ab_and_return;
2468                 }
2469                 fprintf(o->file, "%s", header.buf);
2470                 strbuf_reset(&header);
2471                 if (DIFF_OPT_TST(o, BINARY))
2472                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2473                 else
2474                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2475                                 line_prefix, lbl[0], lbl[1]);
2476                 o->found_changes = 1;
2477         } else {
2478                 /* Crazy xdl interfaces.. */
2479                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2480                 const char *v;
2481                 xpparam_t xpp;
2482                 xdemitconf_t xecfg;
2483                 struct emit_callback ecbdata;
2484                 const struct userdiff_funcname *pe;
2485
2486                 if (must_show_header) {
2487                         fprintf(o->file, "%s", header.buf);
2488                         strbuf_reset(&header);
2489                 }
2490
2491                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2492                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2493
2494                 pe = diff_funcname_pattern(one);
2495                 if (!pe)
2496                         pe = diff_funcname_pattern(two);
2497
2498                 memset(&xpp, 0, sizeof(xpp));
2499                 memset(&xecfg, 0, sizeof(xecfg));
2500                 memset(&ecbdata, 0, sizeof(ecbdata));
2501                 ecbdata.label_path = lbl;
2502                 ecbdata.color_diff = want_color(o->use_color);
2503                 ecbdata.ws_rule = whitespace_rule(name_b);
2504                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2505                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2506                 ecbdata.opt = o;
2507                 ecbdata.header = header.len ? &header : NULL;
2508                 xpp.flags = o->xdl_opts;
2509                 xecfg.ctxlen = o->context;
2510                 xecfg.interhunkctxlen = o->interhunkcontext;
2511                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2512                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2513                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2514                 if (pe)
2515                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2516                 if (!diffopts)
2517                         ;
2518                 else if (skip_prefix(diffopts, "--unified=", &v))
2519                         xecfg.ctxlen = strtoul(v, NULL, 10);
2520                 else if (skip_prefix(diffopts, "-u", &v))
2521                         xecfg.ctxlen = strtoul(v, NULL, 10);
2522                 if (o->word_diff)
2523                         init_diff_words_data(&ecbdata, o, one, two);
2524                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2525                                   &xpp, &xecfg))
2526                         die("unable to generate diff for %s", one->path);
2527                 if (o->word_diff)
2528                         free_diff_words_data(&ecbdata);
2529                 if (textconv_one)
2530                         free(mf1.ptr);
2531                 if (textconv_two)
2532                         free(mf2.ptr);
2533                 xdiff_clear_find_func(&xecfg);
2534         }
2535
2536  free_ab_and_return:
2537         strbuf_release(&header);
2538         diff_free_filespec_data(one);
2539         diff_free_filespec_data(two);
2540         free(a_one);
2541         free(b_two);
2542         return;
2543 }
2544
2545 static void builtin_diffstat(const char *name_a, const char *name_b,
2546                              struct diff_filespec *one,
2547                              struct diff_filespec *two,
2548                              struct diffstat_t *diffstat,
2549                              struct diff_options *o,
2550                              struct diff_filepair *p)
2551 {
2552         mmfile_t mf1, mf2;
2553         struct diffstat_file *data;
2554         int same_contents;
2555         int complete_rewrite = 0;
2556
2557         if (!DIFF_PAIR_UNMERGED(p)) {
2558                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2559                         complete_rewrite = 1;
2560         }
2561
2562         data = diffstat_add(diffstat, name_a, name_b);
2563         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2564
2565         if (!one || !two) {
2566                 data->is_unmerged = 1;
2567                 return;
2568         }
2569
2570         same_contents = !oidcmp(&one->oid, &two->oid);
2571
2572         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2573                 data->is_binary = 1;
2574                 if (same_contents) {
2575                         data->added = 0;
2576                         data->deleted = 0;
2577                 } else {
2578                         data->added = diff_filespec_size(two);
2579                         data->deleted = diff_filespec_size(one);
2580                 }
2581         }
2582
2583         else if (complete_rewrite) {
2584                 diff_populate_filespec(one, 0);
2585                 diff_populate_filespec(two, 0);
2586                 data->deleted = count_lines(one->data, one->size);
2587                 data->added = count_lines(two->data, two->size);
2588         }
2589
2590         else if (!same_contents) {
2591                 /* Crazy xdl interfaces.. */
2592                 xpparam_t xpp;
2593                 xdemitconf_t xecfg;
2594
2595                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2596                         die("unable to read files to diff");
2597
2598                 memset(&xpp, 0, sizeof(xpp));
2599                 memset(&xecfg, 0, sizeof(xecfg));
2600                 xpp.flags = o->xdl_opts;
2601                 xecfg.ctxlen = o->context;
2602                 xecfg.interhunkctxlen = o->interhunkcontext;
2603                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2604                                   &xpp, &xecfg))
2605                         die("unable to generate diffstat for %s", one->path);
2606         }
2607
2608         diff_free_filespec_data(one);
2609         diff_free_filespec_data(two);
2610 }
2611
2612 static void builtin_checkdiff(const char *name_a, const char *name_b,
2613                               const char *attr_path,
2614                               struct diff_filespec *one,
2615                               struct diff_filespec *two,
2616                               struct diff_options *o)
2617 {
2618         mmfile_t mf1, mf2;
2619         struct checkdiff_t data;
2620
2621         if (!two)
2622                 return;
2623
2624         memset(&data, 0, sizeof(data));
2625         data.filename = name_b ? name_b : name_a;
2626         data.lineno = 0;
2627         data.o = o;
2628         data.ws_rule = whitespace_rule(attr_path);
2629         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2630
2631         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2632                 die("unable to read files to diff");
2633
2634         /*
2635          * All the other codepaths check both sides, but not checking
2636          * the "old" side here is deliberate.  We are checking the newly
2637          * introduced changes, and as long as the "new" side is text, we
2638          * can and should check what it introduces.
2639          */
2640         if (diff_filespec_is_binary(two))
2641                 goto free_and_return;
2642         else {
2643                 /* Crazy xdl interfaces.. */
2644                 xpparam_t xpp;
2645                 xdemitconf_t xecfg;
2646
2647                 memset(&xpp, 0, sizeof(xpp));
2648                 memset(&xecfg, 0, sizeof(xecfg));
2649                 xecfg.ctxlen = 1; /* at least one context line */
2650                 xpp.flags = 0;
2651                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2652                                   &xpp, &xecfg))
2653                         die("unable to generate checkdiff for %s", one->path);
2654
2655                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2656                         struct emit_callback ecbdata;
2657                         int blank_at_eof;
2658
2659                         ecbdata.ws_rule = data.ws_rule;
2660                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2661                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2662
2663                         if (blank_at_eof) {
2664                                 static char *err;
2665                                 if (!err)
2666                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2667                                 fprintf(o->file, "%s:%d: %s.\n",
2668                                         data.filename, blank_at_eof, err);
2669                                 data.status = 1; /* report errors */
2670                         }
2671                 }
2672         }
2673  free_and_return:
2674         diff_free_filespec_data(one);
2675         diff_free_filespec_data(two);
2676         if (data.status)
2677                 DIFF_OPT_SET(o, CHECK_FAILED);
2678 }
2679
2680 struct diff_filespec *alloc_filespec(const char *path)
2681 {
2682         struct diff_filespec *spec;
2683
2684         FLEXPTR_ALLOC_STR(spec, path, path);
2685         spec->count = 1;
2686         spec->is_binary = -1;
2687         return spec;
2688 }
2689
2690 void free_filespec(struct diff_filespec *spec)
2691 {
2692         if (!--spec->count) {
2693                 diff_free_filespec_data(spec);
2694                 free(spec);
2695         }
2696 }
2697
2698 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2699                    int sha1_valid, unsigned short mode)
2700 {
2701         if (mode) {
2702                 spec->mode = canon_mode(mode);
2703                 hashcpy(spec->oid.hash, sha1);
2704                 spec->oid_valid = sha1_valid;
2705         }
2706 }
2707
2708 /*
2709  * Given a name and sha1 pair, if the index tells us the file in
2710  * the work tree has that object contents, return true, so that
2711  * prepare_temp_file() does not have to inflate and extract.
2712  */
2713 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2714 {
2715         const struct cache_entry *ce;
2716         struct stat st;
2717         int pos, len;
2718
2719         /*
2720          * We do not read the cache ourselves here, because the
2721          * benchmark with my previous version that always reads cache
2722          * shows that it makes things worse for diff-tree comparing
2723          * two linux-2.6 kernel trees in an already checked out work
2724          * tree.  This is because most diff-tree comparisons deal with
2725          * only a small number of files, while reading the cache is
2726          * expensive for a large project, and its cost outweighs the
2727          * savings we get by not inflating the object to a temporary
2728          * file.  Practically, this code only helps when we are used
2729          * by diff-cache --cached, which does read the cache before
2730          * calling us.
2731          */
2732         if (!active_cache)
2733                 return 0;
2734
2735         /* We want to avoid the working directory if our caller
2736          * doesn't need the data in a normal file, this system
2737          * is rather slow with its stat/open/mmap/close syscalls,
2738          * and the object is contained in a pack file.  The pack
2739          * is probably already open and will be faster to obtain
2740          * the data through than the working directory.  Loose
2741          * objects however would tend to be slower as they need
2742          * to be individually opened and inflated.
2743          */
2744         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2745                 return 0;
2746
2747         /*
2748          * Similarly, if we'd have to convert the file contents anyway, that
2749          * makes the optimization not worthwhile.
2750          */
2751         if (!want_file && would_convert_to_git(name))
2752                 return 0;
2753
2754         len = strlen(name);
2755         pos = cache_name_pos(name, len);
2756         if (pos < 0)
2757                 return 0;
2758         ce = active_cache[pos];
2759
2760         /*
2761          * This is not the sha1 we are looking for, or
2762          * unreusable because it is not a regular file.
2763          */
2764         if (hashcmp(sha1, ce->oid.hash) || !S_ISREG(ce->ce_mode))
2765                 return 0;
2766
2767         /*
2768          * If ce is marked as "assume unchanged", there is no
2769          * guarantee that work tree matches what we are looking for.
2770          */
2771         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2772                 return 0;
2773
2774         /*
2775          * If ce matches the file in the work tree, we can reuse it.
2776          */
2777         if (ce_uptodate(ce) ||
2778             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2779                 return 1;
2780
2781         return 0;
2782 }
2783
2784 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2785 {
2786         struct strbuf buf = STRBUF_INIT;
2787         char *dirty = "";
2788
2789         /* Are we looking at the work tree? */
2790         if (s->dirty_submodule)
2791                 dirty = "-dirty";
2792
2793         strbuf_addf(&buf, "Subproject commit %s%s\n",
2794                     oid_to_hex(&s->oid), dirty);
2795         s->size = buf.len;
2796         if (size_only) {
2797                 s->data = NULL;
2798                 strbuf_release(&buf);
2799         } else {
2800                 s->data = strbuf_detach(&buf, NULL);
2801                 s->should_free = 1;
2802         }
2803         return 0;
2804 }
2805
2806 /*
2807  * While doing rename detection and pickaxe operation, we may need to
2808  * grab the data for the blob (or file) for our own in-core comparison.
2809  * diff_filespec has data and size fields for this purpose.
2810  */
2811 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2812 {
2813         int size_only = flags & CHECK_SIZE_ONLY;
2814         int err = 0;
2815         /*
2816          * demote FAIL to WARN to allow inspecting the situation
2817          * instead of refusing.
2818          */
2819         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2820                                     ? SAFE_CRLF_WARN
2821                                     : safe_crlf);
2822
2823         if (!DIFF_FILE_VALID(s))
2824                 die("internal error: asking to populate invalid file.");
2825         if (S_ISDIR(s->mode))
2826                 return -1;
2827
2828         if (s->data)
2829                 return 0;
2830
2831         if (size_only && 0 < s->size)
2832                 return 0;
2833
2834         if (S_ISGITLINK(s->mode))
2835                 return diff_populate_gitlink(s, size_only);
2836
2837         if (!s->oid_valid ||
2838             reuse_worktree_file(s->path, s->oid.hash, 0)) {
2839                 struct strbuf buf = STRBUF_INIT;
2840                 struct stat st;
2841                 int fd;
2842
2843                 if (lstat(s->path, &st) < 0) {
2844                         if (errno == ENOENT) {
2845                         err_empty:
2846                                 err = -1;
2847                         empty:
2848                                 s->data = (char *)"";
2849                                 s->size = 0;
2850                                 return err;
2851                         }
2852                 }
2853                 s->size = xsize_t(st.st_size);
2854                 if (!s->size)
2855                         goto empty;
2856                 if (S_ISLNK(st.st_mode)) {
2857                         struct strbuf sb = STRBUF_INIT;
2858
2859                         if (strbuf_readlink(&sb, s->path, s->size))
2860                                 goto err_empty;
2861                         s->size = sb.len;
2862                         s->data = strbuf_detach(&sb, NULL);
2863                         s->should_free = 1;
2864                         return 0;
2865                 }
2866                 if (size_only)
2867                         return 0;
2868                 if ((flags & CHECK_BINARY) &&
2869                     s->size > big_file_threshold && s->is_binary == -1) {
2870                         s->is_binary = 1;
2871                         return 0;
2872                 }
2873                 fd = open(s->path, O_RDONLY);
2874                 if (fd < 0)
2875                         goto err_empty;
2876                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2877                 close(fd);
2878                 s->should_munmap = 1;
2879
2880                 /*
2881                  * Convert from working tree format to canonical git format
2882                  */
2883                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2884                         size_t size = 0;
2885                         munmap(s->data, s->size);
2886                         s->should_munmap = 0;
2887                         s->data = strbuf_detach(&buf, &size);
2888                         s->size = size;
2889                         s->should_free = 1;
2890                 }
2891         }
2892         else {
2893                 enum object_type type;
2894                 if (size_only || (flags & CHECK_BINARY)) {
2895                         type = sha1_object_info(s->oid.hash, &s->size);
2896                         if (type < 0)
2897                                 die("unable to read %s",
2898                                     oid_to_hex(&s->oid));
2899                         if (size_only)
2900                                 return 0;
2901                         if (s->size > big_file_threshold && s->is_binary == -1) {
2902                                 s->is_binary = 1;
2903                                 return 0;
2904                         }
2905                 }
2906                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2907                 if (!s->data)
2908                         die("unable to read %s", oid_to_hex(&s->oid));
2909                 s->should_free = 1;
2910         }
2911         return 0;
2912 }
2913
2914 void diff_free_filespec_blob(struct diff_filespec *s)
2915 {
2916         if (s->should_free)
2917                 free(s->data);
2918         else if (s->should_munmap)
2919                 munmap(s->data, s->size);
2920
2921         if (s->should_free || s->should_munmap) {
2922                 s->should_free = s->should_munmap = 0;
2923                 s->data = NULL;
2924         }
2925 }
2926
2927 void diff_free_filespec_data(struct diff_filespec *s)
2928 {
2929         diff_free_filespec_blob(s);
2930         free(s->cnt_data);
2931         s->cnt_data = NULL;
2932 }
2933
2934 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2935                            void *blob,
2936                            unsigned long size,
2937                            const struct object_id *oid,
2938                            int mode)
2939 {
2940         int fd;
2941         struct strbuf buf = STRBUF_INIT;
2942         struct strbuf template = STRBUF_INIT;
2943         char *path_dup = xstrdup(path);
2944         const char *base = basename(path_dup);
2945
2946         /* Generate "XXXXXX_basename.ext" */
2947         strbuf_addstr(&template, "XXXXXX_");
2948         strbuf_addstr(&template, base);
2949
2950         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2951         if (fd < 0)
2952                 die_errno("unable to create temp-file");
2953         if (convert_to_working_tree(path,
2954                         (const char *)blob, (size_t)size, &buf)) {
2955                 blob = buf.buf;
2956                 size = buf.len;
2957         }
2958         if (write_in_full(fd, blob, size) != size)
2959                 die_errno("unable to write temp-file");
2960         close_tempfile(&temp->tempfile);
2961         temp->name = get_tempfile_path(&temp->tempfile);
2962         oid_to_hex_r(temp->hex, oid);
2963         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2964         strbuf_release(&buf);
2965         strbuf_release(&template);
2966         free(path_dup);
2967 }
2968
2969 static struct diff_tempfile *prepare_temp_file(const char *name,
2970                 struct diff_filespec *one)
2971 {
2972         struct diff_tempfile *temp = claim_diff_tempfile();
2973
2974         if (!DIFF_FILE_VALID(one)) {
2975         not_a_valid_file:
2976                 /* A '-' entry produces this for file-2, and
2977                  * a '+' entry produces this for file-1.
2978                  */
2979                 temp->name = "/dev/null";
2980                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2981                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2982                 return temp;
2983         }
2984
2985         if (!S_ISGITLINK(one->mode) &&
2986             (!one->oid_valid ||
2987              reuse_worktree_file(name, one->oid.hash, 1))) {
2988                 struct stat st;
2989                 if (lstat(name, &st) < 0) {
2990                         if (errno == ENOENT)
2991                                 goto not_a_valid_file;
2992                         die_errno("stat(%s)", name);
2993                 }
2994                 if (S_ISLNK(st.st_mode)) {
2995                         struct strbuf sb = STRBUF_INIT;
2996                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2997                                 die_errno("readlink(%s)", name);
2998                         prep_temp_blob(name, temp, sb.buf, sb.len,
2999                                        (one->oid_valid ?
3000                                         &one->oid : &null_oid),
3001                                        (one->oid_valid ?
3002                                         one->mode : S_IFLNK));
3003                         strbuf_release(&sb);
3004                 }
3005                 else {
3006                         /* we can borrow from the file in the work tree */
3007                         temp->name = name;
3008                         if (!one->oid_valid)
3009                                 sha1_to_hex_r(temp->hex, null_sha1);
3010                         else
3011                                 sha1_to_hex_r(temp->hex, one->oid.hash);
3012                         /* Even though we may sometimes borrow the
3013                          * contents from the work tree, we always want
3014                          * one->mode.  mode is trustworthy even when
3015                          * !(one->sha1_valid), as long as
3016                          * DIFF_FILE_VALID(one).
3017                          */
3018                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3019                 }
3020                 return temp;
3021         }
3022         else {
3023                 if (diff_populate_filespec(one, 0))
3024                         die("cannot read data blob for %s", one->path);
3025                 prep_temp_blob(name, temp, one->data, one->size,
3026                                &one->oid, one->mode);
3027         }
3028         return temp;
3029 }
3030
3031 static void add_external_diff_name(struct argv_array *argv,
3032                                    const char *name,
3033                                    struct diff_filespec *df)
3034 {
3035         struct diff_tempfile *temp = prepare_temp_file(name, df);
3036         argv_array_push(argv, temp->name);
3037         argv_array_push(argv, temp->hex);
3038         argv_array_push(argv, temp->mode);
3039 }
3040
3041 /* An external diff command takes:
3042  *
3043  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3044  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3045  *
3046  */
3047 static void run_external_diff(const char *pgm,
3048                               const char *name,
3049                               const char *other,
3050                               struct diff_filespec *one,
3051                               struct diff_filespec *two,
3052                               const char *xfrm_msg,
3053                               int complete_rewrite,
3054                               struct diff_options *o)
3055 {
3056         struct argv_array argv = ARGV_ARRAY_INIT;
3057         struct argv_array env = ARGV_ARRAY_INIT;
3058         struct diff_queue_struct *q = &diff_queued_diff;
3059
3060         argv_array_push(&argv, pgm);
3061         argv_array_push(&argv, name);
3062
3063         if (one && two) {
3064                 add_external_diff_name(&argv, name, one);
3065                 if (!other)
3066                         add_external_diff_name(&argv, name, two);
3067                 else {
3068                         add_external_diff_name(&argv, other, two);
3069                         argv_array_push(&argv, other);
3070                         argv_array_push(&argv, xfrm_msg);
3071                 }
3072         }
3073
3074         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3075         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3076
3077         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3078                 die(_("external diff died, stopping at %s"), name);
3079
3080         remove_tempfile();
3081         argv_array_clear(&argv);
3082         argv_array_clear(&env);
3083 }
3084
3085 static int similarity_index(struct diff_filepair *p)
3086 {
3087         return p->score * 100 / MAX_SCORE;
3088 }
3089
3090 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3091 {
3092         if (startup_info->have_repository)
3093                 return find_unique_abbrev(oid->hash, abbrev);
3094         else {
3095                 char *hex = oid_to_hex(oid);
3096                 if (abbrev < 0)
3097                         abbrev = FALLBACK_DEFAULT_ABBREV;
3098                 if (abbrev > GIT_SHA1_HEXSZ)
3099                         die("BUG: oid abbreviation out of range: %d", abbrev);
3100                 if (abbrev)
3101                         hex[abbrev] = '\0';
3102                 return hex;
3103         }
3104 }
3105
3106 static void fill_metainfo(struct strbuf *msg,
3107                           const char *name,
3108                           const char *other,
3109                           struct diff_filespec *one,
3110                           struct diff_filespec *two,
3111                           struct diff_options *o,
3112                           struct diff_filepair *p,
3113                           int *must_show_header,
3114                           int use_color)
3115 {
3116         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3117         const char *reset = diff_get_color(use_color, DIFF_RESET);
3118         const char *line_prefix = diff_line_prefix(o);
3119
3120         *must_show_header = 1;
3121         strbuf_init(msg, PATH_MAX * 2 + 300);
3122         switch (p->status) {
3123         case DIFF_STATUS_COPIED:
3124                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3125                             line_prefix, set, similarity_index(p));
3126                 strbuf_addf(msg, "%s\n%s%scopy from ",
3127                             reset,  line_prefix, set);
3128                 quote_c_style(name, msg, NULL, 0);
3129                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3130                 quote_c_style(other, msg, NULL, 0);
3131                 strbuf_addf(msg, "%s\n", reset);
3132                 break;
3133         case DIFF_STATUS_RENAMED:
3134                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3135                             line_prefix, set, similarity_index(p));
3136                 strbuf_addf(msg, "%s\n%s%srename from ",
3137                             reset, line_prefix, set);
3138                 quote_c_style(name, msg, NULL, 0);
3139                 strbuf_addf(msg, "%s\n%s%srename to ",
3140                             reset, line_prefix, set);
3141                 quote_c_style(other, msg, NULL, 0);
3142                 strbuf_addf(msg, "%s\n", reset);
3143                 break;
3144         case DIFF_STATUS_MODIFIED:
3145                 if (p->score) {
3146                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3147                                     line_prefix,
3148                                     set, similarity_index(p), reset);
3149                         break;
3150                 }
3151                 /* fallthru */
3152         default:
3153                 *must_show_header = 0;
3154         }
3155         if (one && two && oidcmp(&one->oid, &two->oid)) {
3156                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3157
3158                 if (DIFF_OPT_TST(o, BINARY)) {
3159                         mmfile_t mf;
3160                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3161                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3162                                 abbrev = 40;
3163                 }
3164                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3165                             diff_abbrev_oid(&one->oid, abbrev),
3166                             diff_abbrev_oid(&two->oid, abbrev));
3167                 if (one->mode == two->mode)
3168                         strbuf_addf(msg, " %06o", one->mode);
3169                 strbuf_addf(msg, "%s\n", reset);
3170         }
3171 }
3172
3173 static void run_diff_cmd(const char *pgm,
3174                          const char *name,
3175                          const char *other,
3176                          const char *attr_path,
3177                          struct diff_filespec *one,
3178                          struct diff_filespec *two,
3179                          struct strbuf *msg,
3180                          struct diff_options *o,
3181                          struct diff_filepair *p)
3182 {
3183         const char *xfrm_msg = NULL;
3184         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3185         int must_show_header = 0;
3186
3187
3188         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3189                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3190                 if (drv && drv->external)
3191                         pgm = drv->external;
3192         }
3193
3194         if (msg) {
3195                 /*
3196                  * don't use colors when the header is intended for an
3197                  * external diff driver
3198                  */
3199                 fill_metainfo(msg, name, other, one, two, o, p,
3200                               &must_show_header,
3201                               want_color(o->use_color) && !pgm);
3202                 xfrm_msg = msg->len ? msg->buf : NULL;
3203         }
3204
3205         if (pgm) {
3206                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3207                                   complete_rewrite, o);
3208                 return;
3209         }
3210         if (one && two)
3211                 builtin_diff(name, other ? other : name,
3212                              one, two, xfrm_msg, must_show_header,
3213                              o, complete_rewrite);
3214         else
3215                 fprintf(o->file, "* Unmerged path %s\n", name);
3216 }
3217
3218 static void diff_fill_sha1_info(struct diff_filespec *one)
3219 {
3220         if (DIFF_FILE_VALID(one)) {
3221                 if (!one->oid_valid) {
3222                         struct stat st;
3223                         if (one->is_stdin) {
3224                                 oidclr(&one->oid);
3225                                 return;
3226                         }
3227                         if (lstat(one->path, &st) < 0)
3228                                 die_errno("stat '%s'", one->path);
3229                         if (index_path(one->oid.hash, one->path, &st, 0))
3230                                 die("cannot hash %s", one->path);
3231                 }
3232         }
3233         else
3234                 oidclr(&one->oid);
3235 }
3236
3237 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3238 {
3239         /* Strip the prefix but do not molest /dev/null and absolute paths */
3240         if (*namep && **namep != '/') {
3241                 *namep += prefix_length;
3242                 if (**namep == '/')
3243                         ++*namep;
3244         }
3245         if (*otherp && **otherp != '/') {
3246                 *otherp += prefix_length;
3247                 if (**otherp == '/')
3248                         ++*otherp;
3249         }
3250 }
3251
3252 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3253 {
3254         const char *pgm = external_diff();
3255         struct strbuf msg;
3256         struct diff_filespec *one = p->one;
3257         struct diff_filespec *two = p->two;
3258         const char *name;
3259         const char *other;
3260         const char *attr_path;
3261
3262         name  = p->one->path;
3263         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3264         attr_path = name;
3265         if (o->prefix_length)
3266                 strip_prefix(o->prefix_length, &name, &other);
3267
3268         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3269                 pgm = NULL;
3270
3271         if (DIFF_PAIR_UNMERGED(p)) {
3272                 run_diff_cmd(pgm, name, NULL, attr_path,
3273                              NULL, NULL, NULL, o, p);
3274                 return;
3275         }
3276
3277         diff_fill_sha1_info(one);
3278         diff_fill_sha1_info(two);
3279
3280         if (!pgm &&
3281             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3282             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3283                 /*
3284                  * a filepair that changes between file and symlink
3285                  * needs to be split into deletion and creation.
3286                  */
3287                 struct diff_filespec *null = alloc_filespec(two->path);
3288                 run_diff_cmd(NULL, name, other, attr_path,
3289                              one, null, &msg, o, p);
3290                 free(null);
3291                 strbuf_release(&msg);
3292
3293                 null = alloc_filespec(one->path);
3294                 run_diff_cmd(NULL, name, other, attr_path,
3295                              null, two, &msg, o, p);
3296                 free(null);
3297         }
3298         else
3299                 run_diff_cmd(pgm, name, other, attr_path,
3300                              one, two, &msg, o, p);
3301
3302         strbuf_release(&msg);
3303 }
3304
3305 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3306                          struct diffstat_t *diffstat)
3307 {
3308         const char *name;
3309         const char *other;
3310
3311         if (DIFF_PAIR_UNMERGED(p)) {
3312                 /* unmerged */
3313                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3314                 return;
3315         }
3316
3317         name = p->one->path;
3318         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3319
3320         if (o->prefix_length)
3321                 strip_prefix(o->prefix_length, &name, &other);
3322
3323         diff_fill_sha1_info(p->one);
3324         diff_fill_sha1_info(p->two);
3325
3326         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3327 }
3328
3329 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3330 {
3331         const char *name;
3332         const char *other;
3333         const char *attr_path;
3334
3335         if (DIFF_PAIR_UNMERGED(p)) {
3336                 /* unmerged */
3337                 return;
3338         }
3339
3340         name = p->one->path;
3341         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3342         attr_path = other ? other : name;
3343
3344         if (o->prefix_length)
3345                 strip_prefix(o->prefix_length, &name, &other);
3346
3347         diff_fill_sha1_info(p->one);
3348         diff_fill_sha1_info(p->two);
3349
3350         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3351 }
3352
3353 void diff_setup(struct diff_options *options)
3354 {
3355         memcpy(options, &default_diff_options, sizeof(*options));
3356
3357         options->file = stdout;
3358
3359         options->abbrev = DEFAULT_ABBREV;
3360         options->line_termination = '\n';
3361         options->break_opt = -1;
3362         options->rename_limit = -1;
3363         options->dirstat_permille = diff_dirstat_permille_default;
3364         options->context = diff_context_default;
3365         options->ws_error_highlight = ws_error_highlight_default;
3366         DIFF_OPT_SET(options, RENAME_EMPTY);
3367
3368         /* pathchange left =NULL by default */
3369         options->change = diff_change;
3370         options->add_remove = diff_addremove;
3371         options->use_color = diff_use_color_default;
3372         options->detect_rename = diff_detect_rename_default;
3373         options->xdl_opts |= diff_algorithm;
3374         if (diff_indent_heuristic)
3375                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3376
3377         options->orderfile = diff_order_file_cfg;
3378
3379         if (diff_no_prefix) {
3380                 options->a_prefix = options->b_prefix = "";
3381         } else if (!diff_mnemonic_prefix) {
3382                 options->a_prefix = "a/";
3383                 options->b_prefix = "b/";
3384         }
3385 }
3386
3387 void diff_setup_done(struct diff_options *options)
3388 {
3389         int count = 0;
3390
3391         if (options->set_default)
3392                 options->set_default(options);
3393
3394         if (options->output_format & DIFF_FORMAT_NAME)
3395                 count++;
3396         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3397                 count++;
3398         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3399                 count++;
3400         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3401                 count++;
3402         if (count > 1)
3403                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
3404
3405         /*
3406          * Most of the time we can say "there are changes"
3407          * only by checking if there are changed paths, but
3408          * --ignore-whitespace* options force us to look
3409          * inside contents.
3410          */
3411
3412         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3413             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3414             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3415                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3416         else
3417                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3418
3419         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3420                 options->detect_rename = DIFF_DETECT_COPY;
3421
3422         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3423                 options->prefix = NULL;
3424         if (options->prefix)
3425                 options->prefix_length = strlen(options->prefix);
3426         else
3427                 options->prefix_length = 0;
3428
3429         if (options->output_format & (DIFF_FORMAT_NAME |
3430                                       DIFF_FORMAT_NAME_STATUS |
3431                                       DIFF_FORMAT_CHECKDIFF |
3432                                       DIFF_FORMAT_NO_OUTPUT))
3433                 options->output_format &= ~(DIFF_FORMAT_RAW |
3434                                             DIFF_FORMAT_NUMSTAT |
3435                                             DIFF_FORMAT_DIFFSTAT |
3436                                             DIFF_FORMAT_SHORTSTAT |
3437                                             DIFF_FORMAT_DIRSTAT |
3438                                             DIFF_FORMAT_SUMMARY |
3439                                             DIFF_FORMAT_PATCH);
3440
3441         /*
3442          * These cases always need recursive; we do not drop caller-supplied
3443          * recursive bits for other formats here.
3444          */
3445         if (options->output_format & (DIFF_FORMAT_PATCH |
3446                                       DIFF_FORMAT_NUMSTAT |
3447                                       DIFF_FORMAT_DIFFSTAT |
3448                                       DIFF_FORMAT_SHORTSTAT |
3449                                       DIFF_FORMAT_DIRSTAT |
3450                                       DIFF_FORMAT_SUMMARY |
3451                                       DIFF_FORMAT_CHECKDIFF))
3452                 DIFF_OPT_SET(options, RECURSIVE);
3453         /*
3454          * Also pickaxe would not work very well if you do not say recursive
3455          */
3456         if (options->pickaxe)
3457                 DIFF_OPT_SET(options, RECURSIVE);
3458         /*
3459          * When patches are generated, submodules diffed against the work tree
3460          * must be checked for dirtiness too so it can be shown in the output
3461          */
3462         if (options->output_format & DIFF_FORMAT_PATCH)
3463                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3464
3465         if (options->detect_rename && options->rename_limit < 0)
3466                 options->rename_limit = diff_rename_limit_default;
3467         if (options->setup & DIFF_SETUP_USE_CACHE) {
3468                 if (!active_cache)
3469                         /* read-cache does not die even when it fails
3470                          * so it is safe for us to do this here.  Also
3471                          * it does not smudge active_cache or active_nr
3472                          * when it fails, so we do not have to worry about
3473                          * cleaning it up ourselves either.
3474                          */
3475                         read_cache();
3476         }
3477         if (40 < options->abbrev)
3478                 options->abbrev = 40; /* full */
3479
3480         /*
3481          * It does not make sense to show the first hit we happened
3482          * to have found.  It does not make sense not to return with
3483          * exit code in such a case either.
3484          */
3485         if (DIFF_OPT_TST(options, QUICK)) {
3486                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3487                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3488         }
3489
3490         options->diff_path_counter = 0;
3491
3492         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3493                 die(_("--follow requires exactly one pathspec"));
3494 }
3495
3496 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3497 {
3498         char c, *eq;
3499         int len;
3500
3501         if (*arg != '-')
3502                 return 0;
3503         c = *++arg;
3504         if (!c)
3505                 return 0;
3506         if (c == arg_short) {
3507                 c = *++arg;
3508                 if (!c)
3509                         return 1;
3510                 if (val && isdigit(c)) {
3511                         char *end;
3512                         int n = strtoul(arg, &end, 10);
3513                         if (*end)
3514                                 return 0;
3515                         *val = n;
3516                         return 1;
3517                 }
3518                 return 0;
3519         }
3520         if (c != '-')
3521                 return 0;
3522         arg++;
3523         eq = strchrnul(arg, '=');
3524         len = eq - arg;
3525         if (!len || strncmp(arg, arg_long, len))
3526                 return 0;
3527         if (*eq) {
3528                 int n;
3529                 char *end;
3530                 if (!isdigit(*++eq))
3531                         return 0;
3532                 n = strtoul(eq, &end, 10);
3533                 if (*end)
3534                         return 0;
3535                 *val = n;
3536         }
3537         return 1;
3538 }
3539
3540 static int diff_scoreopt_parse(const char *opt);
3541
3542 static inline int short_opt(char opt, const char **argv,
3543                             const char **optarg)
3544 {
3545         const char *arg = argv[0];
3546         if (arg[0] != '-' || arg[1] != opt)
3547                 return 0;
3548         if (arg[2] != '\0') {
3549                 *optarg = arg + 2;
3550                 return 1;
3551         }
3552         if (!argv[1])
3553                 die("Option '%c' requires a value", opt);
3554         *optarg = argv[1];
3555         return 2;
3556 }
3557
3558 int parse_long_opt(const char *opt, const char **argv,
3559                    const char **optarg)
3560 {
3561         const char *arg = argv[0];
3562         if (!skip_prefix(arg, "--", &arg))
3563                 return 0;
3564         if (!skip_prefix(arg, opt, &arg))
3565                 return 0;
3566         if (*arg == '=') { /* stuck form: --option=value */
3567                 *optarg = arg + 1;
3568                 return 1;
3569         }
3570         if (*arg != '\0')
3571                 return 0;
3572         /* separate form: --option value */
3573         if (!argv[1])
3574                 die("Option '--%s' requires a value", opt);
3575         *optarg = argv[1];
3576         return 2;
3577 }
3578
3579 static int stat_opt(struct diff_options *options, const char **av)
3580 {
3581         const char *arg = av[0];
3582         char *end;
3583         int width = options->stat_width;
3584         int name_width = options->stat_name_width;
3585         int graph_width = options->stat_graph_width;
3586         int count = options->stat_count;
3587         int argcount = 1;
3588
3589         if (!skip_prefix(arg, "--stat", &arg))
3590                 die("BUG: stat option does not begin with --stat: %s", arg);
3591         end = (char *)arg;
3592
3593         switch (*arg) {
3594         case '-':
3595                 if (skip_prefix(arg, "-width", &arg)) {
3596                         if (*arg == '=')
3597                                 width = strtoul(arg + 1, &end, 10);
3598                         else if (!*arg && !av[1])
3599                                 die_want_option("--stat-width");
3600                         else if (!*arg) {
3601                                 width = strtoul(av[1], &end, 10);
3602                                 argcount = 2;
3603                         }
3604                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3605                         if (*arg == '=')
3606                                 name_width = strtoul(arg + 1, &end, 10);
3607                         else if (!*arg && !av[1])
3608                                 die_want_option("--stat-name-width");
3609                         else if (!*arg) {
3610                                 name_width = strtoul(av[1], &end, 10);
3611                                 argcount = 2;
3612                         }
3613                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3614                         if (*arg == '=')
3615                                 graph_width = strtoul(arg + 1, &end, 10);
3616                         else if (!*arg && !av[1])
3617                                 die_want_option("--stat-graph-width");
3618                         else if (!*arg) {
3619                                 graph_width = strtoul(av[1], &end, 10);
3620                                 argcount = 2;
3621                         }
3622                 } else if (skip_prefix(arg, "-count", &arg)) {
3623                         if (*arg == '=')
3624                                 count = strtoul(arg + 1, &end, 10);
3625                         else if (!*arg && !av[1])
3626                                 die_want_option("--stat-count");
3627                         else if (!*arg) {
3628                                 count = strtoul(av[1], &end, 10);
3629                                 argcount = 2;
3630                         }
3631                 }
3632                 break;
3633         case '=':
3634                 width = strtoul(arg+1, &end, 10);
3635                 if (*end == ',')
3636                         name_width = strtoul(end+1, &end, 10);
3637                 if (*end == ',')
3638                         count = strtoul(end+1, &end, 10);
3639         }
3640
3641         /* Important! This checks all the error cases! */
3642         if (*end)
3643                 return 0;
3644         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3645         options->stat_name_width = name_width;
3646         options->stat_graph_width = graph_width;
3647         options->stat_width = width;
3648         options->stat_count = count;
3649         return argcount;
3650 }
3651
3652 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3653 {
3654         struct strbuf errmsg = STRBUF_INIT;
3655         if (parse_dirstat_params(options, params, &errmsg))
3656                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3657                     errmsg.buf);
3658         strbuf_release(&errmsg);
3659         /*
3660          * The caller knows a dirstat-related option is given from the command
3661          * line; allow it to say "return this_function();"
3662          */
3663         options->output_format |= DIFF_FORMAT_DIRSTAT;
3664         return 1;
3665 }
3666
3667 static int parse_submodule_opt(struct diff_options *options, const char *value)
3668 {
3669         if (parse_submodule_params(options, value))
3670                 die(_("Failed to parse --submodule option parameter: '%s'"),
3671                         value);
3672         return 1;
3673 }
3674
3675 static const char diff_status_letters[] = {
3676         DIFF_STATUS_ADDED,
3677         DIFF_STATUS_COPIED,
3678         DIFF_STATUS_DELETED,
3679         DIFF_STATUS_MODIFIED,
3680         DIFF_STATUS_RENAMED,
3681         DIFF_STATUS_TYPE_CHANGED,
3682         DIFF_STATUS_UNKNOWN,
3683         DIFF_STATUS_UNMERGED,
3684         DIFF_STATUS_FILTER_AON,
3685         DIFF_STATUS_FILTER_BROKEN,
3686         '\0',
3687 };
3688
3689 static unsigned int filter_bit['Z' + 1];
3690
3691 static void prepare_filter_bits(void)
3692 {
3693         int i;
3694
3695         if (!filter_bit[DIFF_STATUS_ADDED]) {
3696                 for (i = 0; diff_status_letters[i]; i++)
3697                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3698         }
3699 }
3700
3701 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3702 {
3703         return opt->filter & filter_bit[(int) status];
3704 }
3705
3706 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3707 {
3708         int i, optch;
3709
3710         prepare_filter_bits();
3711
3712         /*
3713          * If there is a negation e.g. 'd' in the input, and we haven't
3714          * initialized the filter field with another --diff-filter, start
3715          * from full set of bits, except for AON.
3716          */
3717         if (!opt->filter) {
3718                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3719                         if (optch < 'a' || 'z' < optch)
3720                                 continue;
3721                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3722                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3723                         break;
3724                 }
3725         }
3726
3727         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3728                 unsigned int bit;
3729                 int negate;
3730
3731                 if ('a' <= optch && optch <= 'z') {
3732                         negate = 1;
3733                         optch = toupper(optch);
3734                 } else {
3735                         negate = 0;
3736                 }
3737
3738                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3739                 if (!bit)
3740                         return optarg[i];
3741                 if (negate)
3742                         opt->filter &= ~bit;
3743                 else
3744                         opt->filter |= bit;
3745         }
3746         return 0;
3747 }
3748
3749 static void enable_patch_output(int *fmt) {
3750         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3751         *fmt |= DIFF_FORMAT_PATCH;
3752 }
3753
3754 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
3755 {
3756         int val = parse_ws_error_highlight(arg);
3757
3758         if (val < 0) {
3759                 error("unknown value after ws-error-highlight=%.*s",
3760                       -1 - val, arg);
3761                 return 0;
3762         }
3763         opt->ws_error_highlight = val;
3764         return 1;
3765 }
3766
3767 int diff_opt_parse(struct diff_options *options,
3768                    const char **av, int ac, const char *prefix)
3769 {
3770         const char *arg = av[0];
3771         const char *optarg;
3772         int argcount;
3773
3774         if (!prefix)
3775                 prefix = "";
3776
3777         /* Output format options */
3778         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3779             || opt_arg(arg, 'U', "unified", &options->context))
3780                 enable_patch_output(&options->output_format);
3781         else if (!strcmp(arg, "--raw"))
3782                 options->output_format |= DIFF_FORMAT_RAW;
3783         else if (!strcmp(arg, "--patch-with-raw")) {
3784                 enable_patch_output(&options->output_format);
3785                 options->output_format |= DIFF_FORMAT_RAW;
3786         } else if (!strcmp(arg, "--numstat"))
3787                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3788         else if (!strcmp(arg, "--shortstat"))
3789                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3790         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3791                 return parse_dirstat_opt(options, "");
3792         else if (skip_prefix(arg, "-X", &arg))
3793                 return parse_dirstat_opt(options, arg);
3794         else if (skip_prefix(arg, "--dirstat=", &arg))
3795                 return parse_dirstat_opt(options, arg);
3796         else if (!strcmp(arg, "--cumulative"))
3797                 return parse_dirstat_opt(options, "cumulative");
3798         else if (!strcmp(arg, "--dirstat-by-file"))
3799                 return parse_dirstat_opt(options, "files");
3800         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3801                 parse_dirstat_opt(options, "files");
3802                 return parse_dirstat_opt(options, arg);
3803         }
3804         else if (!strcmp(arg, "--check"))
3805                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3806         else if (!strcmp(arg, "--summary"))
3807                 options->output_format |= DIFF_FORMAT_SUMMARY;
3808         else if (!strcmp(arg, "--patch-with-stat")) {
3809                 enable_patch_output(&options->output_format);
3810                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3811         } else if (!strcmp(arg, "--name-only"))
3812                 options->output_format |= DIFF_FORMAT_NAME;
3813         else if (!strcmp(arg, "--name-status"))
3814                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3815         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3816                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3817         else if (starts_with(arg, "--stat"))
3818                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3819                 return stat_opt(options, av);
3820
3821         /* renames options */
3822         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3823                  !strcmp(arg, "--break-rewrites")) {
3824                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3825                         return error("invalid argument to -B: %s", arg+2);
3826         }
3827         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3828                  !strcmp(arg, "--find-renames")) {
3829                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3830                         return error("invalid argument to -M: %s", arg+2);
3831                 options->detect_rename = DIFF_DETECT_RENAME;
3832         }
3833         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3834                 options->irreversible_delete = 1;
3835         }
3836         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3837                  !strcmp(arg, "--find-copies")) {
3838                 if (options->detect_rename == DIFF_DETECT_COPY)
3839                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3840                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3841                         return error("invalid argument to -C: %s", arg+2);
3842                 options->detect_rename = DIFF_DETECT_COPY;
3843         }
3844         else if (!strcmp(arg, "--no-renames"))
3845                 options->detect_rename = 0;
3846         else if (!strcmp(arg, "--rename-empty"))
3847                 DIFF_OPT_SET(options, RENAME_EMPTY);
3848         else if (!strcmp(arg, "--no-rename-empty"))
3849                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3850         else if (!strcmp(arg, "--relative"))
3851                 DIFF_OPT_SET(options, RELATIVE_NAME);
3852         else if (skip_prefix(arg, "--relative=", &arg)) {
3853                 DIFF_OPT_SET(options, RELATIVE_NAME);
3854                 options->prefix = arg;
3855         }
3856
3857         /* xdiff options */
3858         else if (!strcmp(arg, "--minimal"))
3859                 DIFF_XDL_SET(options, NEED_MINIMAL);
3860         else if (!strcmp(arg, "--no-minimal"))
3861                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3862         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3863                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3864         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3865                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3866         else if (!strcmp(arg, "--ignore-space-at-eol"))
3867                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3868         else if (!strcmp(arg, "--ignore-blank-lines"))
3869                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3870         else if (!strcmp(arg, "--indent-heuristic"))
3871                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3872         else if (!strcmp(arg, "--no-indent-heuristic"))
3873                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3874         else if (!strcmp(arg, "--patience"))
3875                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3876         else if (!strcmp(arg, "--histogram"))
3877                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3878         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3879                 long value = parse_algorithm_value(optarg);
3880                 if (value < 0)
3881                         return error("option diff-algorithm accepts \"myers\", "
3882                                      "\"minimal\", \"patience\" and \"histogram\"");
3883                 /* clear out previous settings */
3884                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3885                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3886                 options->xdl_opts |= value;
3887                 return argcount;
3888         }
3889
3890         /* flags options */
3891         else if (!strcmp(arg, "--binary")) {
3892                 enable_patch_output(&options->output_format);
3893                 DIFF_OPT_SET(options, BINARY);
3894         }
3895         else if (!strcmp(arg, "--full-index"))
3896                 DIFF_OPT_SET(options, FULL_INDEX);
3897         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3898                 DIFF_OPT_SET(options, TEXT);
3899         else if (!strcmp(arg, "-R"))
3900                 DIFF_OPT_SET(options, REVERSE_DIFF);
3901         else if (!strcmp(arg, "--find-copies-harder"))
3902                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3903         else if (!strcmp(arg, "--follow"))
3904                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3905         else if (!strcmp(arg, "--no-follow")) {
3906                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3907                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3908         } else if (!strcmp(arg, "--color"))
3909                 options->use_color = 1;
3910         else if (skip_prefix(arg, "--color=", &arg)) {
3911                 int value = git_config_colorbool(NULL, arg);
3912                 if (value < 0)
3913                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3914                 options->use_color = value;
3915         }
3916         else if (!strcmp(arg, "--no-color"))
3917                 options->use_color = 0;
3918         else if (!strcmp(arg, "--color-words")) {
3919                 options->use_color = 1;
3920                 options->word_diff = DIFF_WORDS_COLOR;
3921         }
3922         else if (skip_prefix(arg, "--color-words=", &arg)) {
3923                 options->use_color = 1;
3924                 options->word_diff = DIFF_WORDS_COLOR;
3925                 options->word_regex = arg;
3926         }
3927         else if (!strcmp(arg, "--word-diff")) {
3928                 if (options->word_diff == DIFF_WORDS_NONE)
3929                         options->word_diff = DIFF_WORDS_PLAIN;
3930         }
3931         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3932                 if (!strcmp(arg, "plain"))
3933                         options->word_diff = DIFF_WORDS_PLAIN;
3934                 else if (!strcmp(arg, "color")) {
3935                         options->use_color = 1;
3936                         options->word_diff = DIFF_WORDS_COLOR;
3937                 }
3938                 else if (!strcmp(arg, "porcelain"))
3939                         options->word_diff = DIFF_WORDS_PORCELAIN;
3940                 else if (!strcmp(arg, "none"))
3941                         options->word_diff = DIFF_WORDS_NONE;
3942                 else
3943                         die("bad --word-diff argument: %s", arg);
3944         }
3945         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3946                 if (options->word_diff == DIFF_WORDS_NONE)
3947                         options->word_diff = DIFF_WORDS_PLAIN;
3948                 options->word_regex = optarg;
3949                 return argcount;
3950         }
3951         else if (!strcmp(arg, "--exit-code"))
3952                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3953         else if (!strcmp(arg, "--quiet"))
3954                 DIFF_OPT_SET(options, QUICK);
3955         else if (!strcmp(arg, "--ext-diff"))
3956                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3957         else if (!strcmp(arg, "--no-ext-diff"))
3958                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3959         else if (!strcmp(arg, "--textconv"))
3960                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3961         else if (!strcmp(arg, "--no-textconv"))
3962                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3963         else if (!strcmp(arg, "--ignore-submodules")) {
3964                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3965                 handle_ignore_submodules_arg(options, "all");
3966         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3967                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3968                 handle_ignore_submodules_arg(options, arg);
3969         } else if (!strcmp(arg, "--submodule"))
3970                 options->submodule_format = DIFF_SUBMODULE_LOG;
3971         else if (skip_prefix(arg, "--submodule=", &arg))
3972                 return parse_submodule_opt(options, arg);
3973         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3974                 return parse_ws_error_highlight_opt(options, arg);
3975         else if (!strcmp(arg, "--ita-invisible-in-index"))
3976                 options->ita_invisible_in_index = 1;
3977         else if (!strcmp(arg, "--ita-visible-in-index"))
3978                 options->ita_invisible_in_index = 0;
3979
3980         /* misc options */
3981         else if (!strcmp(arg, "-z"))
3982                 options->line_termination = 0;
3983         else if ((argcount = short_opt('l', av, &optarg))) {
3984                 options->rename_limit = strtoul(optarg, NULL, 10);
3985                 return argcount;
3986         }
3987         else if ((argcount = short_opt('S', av, &optarg))) {
3988                 options->pickaxe = optarg;
3989                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3990                 return argcount;
3991         } else if ((argcount = short_opt('G', av, &optarg))) {
3992                 options->pickaxe = optarg;
3993                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3994                 return argcount;
3995         }
3996         else if (!strcmp(arg, "--pickaxe-all"))
3997                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3998         else if (!strcmp(arg, "--pickaxe-regex"))
3999                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4000         else if ((argcount = short_opt('O', av, &optarg))) {
4001                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
4002                 options->orderfile = xstrdup(path);
4003                 return argcount;
4004         }
4005         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4006                 int offending = parse_diff_filter_opt(optarg, options);
4007                 if (offending)
4008                         die("unknown change class '%c' in --diff-filter=%s",
4009                             offending, optarg);
4010                 return argcount;
4011         }
4012         else if (!strcmp(arg, "--no-abbrev"))
4013                 options->abbrev = 0;
4014         else if (!strcmp(arg, "--abbrev"))
4015                 options->abbrev = DEFAULT_ABBREV;
4016         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4017                 options->abbrev = strtoul(arg, NULL, 10);
4018                 if (options->abbrev < MINIMUM_ABBREV)
4019                         options->abbrev = MINIMUM_ABBREV;
4020                 else if (40 < options->abbrev)
4021                         options->abbrev = 40;
4022         }
4023         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4024                 options->a_prefix = optarg;
4025                 return argcount;
4026         }
4027         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4028                 options->line_prefix = optarg;
4029                 options->line_prefix_length = strlen(options->line_prefix);
4030                 graph_setup_line_prefix(options);
4031                 return argcount;
4032         }
4033         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4034                 options->b_prefix = optarg;
4035                 return argcount;
4036         }
4037         else if (!strcmp(arg, "--no-prefix"))
4038                 options->a_prefix = options->b_prefix = "";
4039         else if (opt_arg(arg, '\0', "inter-hunk-context",
4040                          &options->interhunkcontext))
4041                 ;
4042         else if (!strcmp(arg, "-W"))
4043                 DIFF_OPT_SET(options, FUNCCONTEXT);
4044         else if (!strcmp(arg, "--function-context"))
4045                 DIFF_OPT_SET(options, FUNCCONTEXT);
4046         else if (!strcmp(arg, "--no-function-context"))
4047                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4048         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4049                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
4050                 options->file = fopen(path, "w");
4051                 if (!options->file)
4052                         die_errno("Could not open '%s'", path);
4053                 options->close_file = 1;
4054                 if (options->use_color != GIT_COLOR_ALWAYS)
4055                         options->use_color = GIT_COLOR_NEVER;
4056                 return argcount;
4057         } else
4058                 return 0;
4059         return 1;
4060 }
4061
4062 int parse_rename_score(const char **cp_p)
4063 {
4064         unsigned long num, scale;
4065         int ch, dot;
4066         const char *cp = *cp_p;
4067
4068         num = 0;
4069         scale = 1;
4070         dot = 0;
4071         for (;;) {
4072                 ch = *cp;
4073                 if ( !dot && ch == '.' ) {
4074                         scale = 1;
4075                         dot = 1;
4076                 } else if ( ch == '%' ) {
4077                         scale = dot ? scale*100 : 100;
4078                         cp++;   /* % is always at the end */
4079                         break;
4080                 } else if ( ch >= '0' && ch <= '9' ) {
4081                         if ( scale < 100000 ) {
4082                                 scale *= 10;
4083                                 num = (num*10) + (ch-'0');
4084                         }
4085                 } else {
4086                         break;
4087                 }
4088                 cp++;
4089         }
4090         *cp_p = cp;
4091
4092         /* user says num divided by scale and we say internally that
4093          * is MAX_SCORE * num / scale.
4094          */
4095         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4096 }
4097
4098 static int diff_scoreopt_parse(const char *opt)
4099 {
4100         int opt1, opt2, cmd;
4101
4102         if (*opt++ != '-')
4103                 return -1;
4104         cmd = *opt++;
4105         if (cmd == '-') {
4106                 /* convert the long-form arguments into short-form versions */
4107                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4108                         if (*opt == 0 || *opt++ == '=')
4109                                 cmd = 'B';
4110                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4111                         if (*opt == 0 || *opt++ == '=')
4112                                 cmd = 'C';
4113                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4114                         if (*opt == 0 || *opt++ == '=')
4115                                 cmd = 'M';
4116                 }
4117         }
4118         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4119                 return -1; /* that is not a -M, -C, or -B option */
4120
4121         opt1 = parse_rename_score(&opt);
4122         if (cmd != 'B')
4123                 opt2 = 0;
4124         else {
4125                 if (*opt == 0)
4126                         opt2 = 0;
4127                 else if (*opt != '/')
4128                         return -1; /* we expect -B80/99 or -B80 */
4129                 else {
4130                         opt++;
4131                         opt2 = parse_rename_score(&opt);
4132                 }
4133         }
4134         if (*opt != 0)
4135                 return -1;
4136         return opt1 | (opt2 << 16);
4137 }
4138
4139 struct diff_queue_struct diff_queued_diff;
4140
4141 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4142 {
4143         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4144         queue->queue[queue->nr++] = dp;
4145 }
4146
4147 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4148                                  struct diff_filespec *one,
4149                                  struct diff_filespec *two)
4150 {
4151         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4152         dp->one = one;
4153         dp->two = two;
4154         if (queue)
4155                 diff_q(queue, dp);
4156         return dp;
4157 }
4158
4159 void diff_free_filepair(struct diff_filepair *p)
4160 {
4161         free_filespec(p->one);
4162         free_filespec(p->two);
4163         free(p);
4164 }
4165
4166 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4167 {
4168         int abblen;
4169         const char *abbrev;
4170
4171         if (len == GIT_SHA1_HEXSZ)
4172                 return oid_to_hex(oid);
4173
4174         abbrev = diff_abbrev_oid(oid, len);
4175         abblen = strlen(abbrev);
4176
4177         /*
4178          * In well-behaved cases, where the abbbreviated result is the
4179          * same as the requested length, append three dots after the
4180          * abbreviation (hence the whole logic is limited to the case
4181          * where abblen < 37); when the actual abbreviated result is a
4182          * bit longer than the requested length, we reduce the number
4183          * of dots so that they match the well-behaved ones.  However,
4184          * if the actual abbreviation is longer than the requested
4185          * length by more than three, we give up on aligning, and add
4186          * three dots anyway, to indicate that the output is not the
4187          * full object name.  Yes, this may be suboptimal, but this
4188          * appears only in "diff --raw --abbrev" output and it is not
4189          * worth the effort to change it now.  Note that this would
4190          * likely to work fine when the automatic sizing of default
4191          * abbreviation length is used--we would be fed -1 in "len" in
4192          * that case, and will end up always appending three-dots, but
4193          * the automatic sizing is supposed to give abblen that ensures
4194          * uniqueness across all objects (statistically speaking).
4195          */
4196         if (abblen < GIT_SHA1_HEXSZ - 3) {
4197                 static char hex[GIT_SHA1_HEXSZ + 1];
4198                 if (len < abblen && abblen <= len + 2)
4199                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4200                 else
4201                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4202                 return hex;
4203         }
4204
4205         return oid_to_hex(oid);
4206 }
4207
4208 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4209 {
4210         int line_termination = opt->line_termination;
4211         int inter_name_termination = line_termination ? '\t' : '\0';
4212
4213         fprintf(opt->file, "%s", diff_line_prefix(opt));
4214         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4215                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4216                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4217                 fprintf(opt->file, "%s ",
4218                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4219         }
4220         if (p->score) {
4221                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4222                         inter_name_termination);
4223         } else {
4224                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4225         }
4226
4227         if (p->status == DIFF_STATUS_COPIED ||
4228             p->status == DIFF_STATUS_RENAMED) {
4229                 const char *name_a, *name_b;
4230                 name_a = p->one->path;
4231                 name_b = p->two->path;
4232                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4233                 write_name_quoted(name_a, opt->file, inter_name_termination);
4234                 write_name_quoted(name_b, opt->file, line_termination);
4235         } else {
4236                 const char *name_a, *name_b;
4237                 name_a = p->one->mode ? p->one->path : p->two->path;
4238                 name_b = NULL;
4239                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4240                 write_name_quoted(name_a, opt->file, line_termination);
4241         }
4242 }
4243
4244 int diff_unmodified_pair(struct diff_filepair *p)
4245 {
4246         /* This function is written stricter than necessary to support
4247          * the currently implemented transformers, but the idea is to
4248          * let transformers to produce diff_filepairs any way they want,
4249          * and filter and clean them up here before producing the output.
4250          */
4251         struct diff_filespec *one = p->one, *two = p->two;
4252
4253         if (DIFF_PAIR_UNMERGED(p))
4254                 return 0; /* unmerged is interesting */
4255
4256         /* deletion, addition, mode or type change
4257          * and rename are all interesting.
4258          */
4259         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4260             DIFF_PAIR_MODE_CHANGED(p) ||
4261             strcmp(one->path, two->path))
4262                 return 0;
4263
4264         /* both are valid and point at the same path.  that is, we are
4265          * dealing with a change.
4266          */
4267         if (one->oid_valid && two->oid_valid &&
4268             !oidcmp(&one->oid, &two->oid) &&
4269             !one->dirty_submodule && !two->dirty_submodule)
4270                 return 1; /* no change */
4271         if (!one->oid_valid && !two->oid_valid)
4272                 return 1; /* both look at the same file on the filesystem. */
4273         return 0;
4274 }
4275
4276 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4277 {
4278         if (diff_unmodified_pair(p))
4279                 return;
4280
4281         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4282             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4283                 return; /* no tree diffs in patch format */
4284
4285         run_diff(p, o);
4286 }
4287
4288 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4289                             struct diffstat_t *diffstat)
4290 {
4291         if (diff_unmodified_pair(p))
4292                 return;
4293
4294         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4295             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4296                 return; /* no useful stat for tree diffs */
4297
4298         run_diffstat(p, o, diffstat);
4299 }
4300
4301 static void diff_flush_checkdiff(struct diff_filepair *p,
4302                 struct diff_options *o)
4303 {
4304         if (diff_unmodified_pair(p))
4305                 return;
4306
4307         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4308             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4309                 return; /* nothing to check in tree diffs */
4310
4311         run_checkdiff(p, o);
4312 }
4313
4314 int diff_queue_is_empty(void)
4315 {
4316         struct diff_queue_struct *q = &diff_queued_diff;
4317         int i;
4318         for (i = 0; i < q->nr; i++)
4319                 if (!diff_unmodified_pair(q->queue[i]))
4320                         return 0;
4321         return 1;
4322 }
4323
4324 #if DIFF_DEBUG
4325 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4326 {
4327         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4328                 x, one ? one : "",
4329                 s->path,
4330                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4331                 s->mode,
4332                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4333         fprintf(stderr, "queue[%d] %s size %lu\n",
4334                 x, one ? one : "",
4335                 s->size);
4336 }
4337
4338 void diff_debug_filepair(const struct diff_filepair *p, int i)
4339 {
4340         diff_debug_filespec(p->one, i, "one");
4341         diff_debug_filespec(p->two, i, "two");
4342         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4343                 p->score, p->status ? p->status : '?',
4344                 p->one->rename_used, p->broken_pair);
4345 }
4346
4347 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4348 {
4349         int i;
4350         if (msg)
4351                 fprintf(stderr, "%s\n", msg);
4352         fprintf(stderr, "q->nr = %d\n", q->nr);
4353         for (i = 0; i < q->nr; i++) {
4354                 struct diff_filepair *p = q->queue[i];
4355                 diff_debug_filepair(p, i);
4356         }
4357 }
4358 #endif
4359
4360 static void diff_resolve_rename_copy(void)
4361 {
4362         int i;
4363         struct diff_filepair *p;
4364         struct diff_queue_struct *q = &diff_queued_diff;
4365
4366         diff_debug_queue("resolve-rename-copy", q);
4367
4368         for (i = 0; i < q->nr; i++) {
4369                 p = q->queue[i];
4370                 p->status = 0; /* undecided */
4371                 if (DIFF_PAIR_UNMERGED(p))
4372                         p->status = DIFF_STATUS_UNMERGED;
4373                 else if (!DIFF_FILE_VALID(p->one))
4374                         p->status = DIFF_STATUS_ADDED;
4375                 else if (!DIFF_FILE_VALID(p->two))
4376                         p->status = DIFF_STATUS_DELETED;
4377                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4378                         p->status = DIFF_STATUS_TYPE_CHANGED;
4379
4380                 /* from this point on, we are dealing with a pair
4381                  * whose both sides are valid and of the same type, i.e.
4382                  * either in-place edit or rename/copy edit.
4383                  */
4384                 else if (DIFF_PAIR_RENAME(p)) {
4385                         /*
4386                          * A rename might have re-connected a broken
4387                          * pair up, causing the pathnames to be the
4388                          * same again. If so, that's not a rename at
4389                          * all, just a modification..
4390                          *
4391                          * Otherwise, see if this source was used for
4392                          * multiple renames, in which case we decrement
4393                          * the count, and call it a copy.
4394                          */
4395                         if (!strcmp(p->one->path, p->two->path))
4396                                 p->status = DIFF_STATUS_MODIFIED;
4397                         else if (--p->one->rename_used > 0)
4398                                 p->status = DIFF_STATUS_COPIED;
4399                         else
4400                                 p->status = DIFF_STATUS_RENAMED;
4401                 }
4402                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4403                          p->one->mode != p->two->mode ||
4404                          p->one->dirty_submodule ||
4405                          p->two->dirty_submodule ||
4406                          is_null_oid(&p->one->oid))
4407                         p->status = DIFF_STATUS_MODIFIED;
4408                 else {
4409                         /* This is a "no-change" entry and should not
4410                          * happen anymore, but prepare for broken callers.
4411                          */
4412                         error("feeding unmodified %s to diffcore",
4413                               p->one->path);
4414                         p->status = DIFF_STATUS_UNKNOWN;
4415                 }
4416         }
4417         diff_debug_queue("resolve-rename-copy done", q);
4418 }
4419
4420 static int check_pair_status(struct diff_filepair *p)
4421 {
4422         switch (p->status) {
4423         case DIFF_STATUS_UNKNOWN:
4424                 return 0;
4425         case 0:
4426                 die("internal error in diff-resolve-rename-copy");
4427         default:
4428                 return 1;
4429         }
4430 }
4431
4432 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4433 {
4434         int fmt = opt->output_format;
4435
4436         if (fmt & DIFF_FORMAT_CHECKDIFF)
4437                 diff_flush_checkdiff(p, opt);
4438         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4439                 diff_flush_raw(p, opt);
4440         else if (fmt & DIFF_FORMAT_NAME) {
4441                 const char *name_a, *name_b;
4442                 name_a = p->two->path;
4443                 name_b = NULL;
4444                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4445                 write_name_quoted(name_a, opt->file, opt->line_termination);
4446         }
4447 }
4448
4449 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4450 {
4451         if (fs->mode)
4452                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4453         else
4454                 fprintf(file, " %s ", newdelete);
4455         write_name_quoted(fs->path, file, '\n');
4456 }
4457
4458
4459 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4460                 const char *line_prefix)
4461 {
4462         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4463                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4464                         p->two->mode, show_name ? ' ' : '\n');
4465                 if (show_name) {
4466                         write_name_quoted(p->two->path, file, '\n');
4467                 }
4468         }
4469 }
4470
4471 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4472                         const char *line_prefix)
4473 {
4474         char *names = pprint_rename(p->one->path, p->two->path);
4475
4476         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4477         free(names);
4478         show_mode_change(file, p, 0, line_prefix);
4479 }
4480
4481 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4482 {
4483         FILE *file = opt->file;
4484         const char *line_prefix = diff_line_prefix(opt);
4485
4486         switch(p->status) {
4487         case DIFF_STATUS_DELETED:
4488                 fputs(line_prefix, file);
4489                 show_file_mode_name(file, "delete", p->one);
4490                 break;
4491         case DIFF_STATUS_ADDED:
4492                 fputs(line_prefix, file);
4493                 show_file_mode_name(file, "create", p->two);
4494                 break;
4495         case DIFF_STATUS_COPIED:
4496                 fputs(line_prefix, file);
4497                 show_rename_copy(file, "copy", p, line_prefix);
4498                 break;
4499         case DIFF_STATUS_RENAMED:
4500                 fputs(line_prefix, file);
4501                 show_rename_copy(file, "rename", p, line_prefix);
4502                 break;
4503         default:
4504                 if (p->score) {
4505                         fprintf(file, "%s rewrite ", line_prefix);
4506                         write_name_quoted(p->two->path, file, ' ');
4507                         fprintf(file, "(%d%%)\n", similarity_index(p));
4508                 }
4509                 show_mode_change(file, p, !p->score, line_prefix);
4510                 break;
4511         }
4512 }
4513
4514 struct patch_id_t {
4515         git_SHA_CTX *ctx;
4516         int patchlen;
4517 };
4518
4519 static int remove_space(char *line, int len)
4520 {
4521         int i;
4522         char *dst = line;
4523         unsigned char c;
4524
4525         for (i = 0; i < len; i++)
4526                 if (!isspace((c = line[i])))
4527                         *dst++ = c;
4528
4529         return dst - line;
4530 }
4531
4532 static void patch_id_consume(void *priv, char *line, unsigned long len)
4533 {
4534         struct patch_id_t *data = priv;
4535         int new_len;
4536
4537         /* Ignore line numbers when computing the SHA1 of the patch */
4538         if (starts_with(line, "@@ -"))
4539                 return;
4540
4541         new_len = remove_space(line, len);
4542
4543         git_SHA1_Update(data->ctx, line, new_len);
4544         data->patchlen += new_len;
4545 }
4546
4547 /* returns 0 upon success, and writes result into sha1 */
4548 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4549 {
4550         struct diff_queue_struct *q = &diff_queued_diff;
4551         int i;
4552         git_SHA_CTX ctx;
4553         struct patch_id_t data;
4554         char buffer[PATH_MAX * 4 + 20];
4555
4556         git_SHA1_Init(&ctx);
4557         memset(&data, 0, sizeof(struct patch_id_t));
4558         data.ctx = &ctx;
4559
4560         for (i = 0; i < q->nr; i++) {
4561                 xpparam_t xpp;
4562                 xdemitconf_t xecfg;
4563                 mmfile_t mf1, mf2;
4564                 struct diff_filepair *p = q->queue[i];
4565                 int len1, len2;
4566
4567                 memset(&xpp, 0, sizeof(xpp));
4568                 memset(&xecfg, 0, sizeof(xecfg));
4569                 if (p->status == 0)
4570                         return error("internal diff status error");
4571                 if (p->status == DIFF_STATUS_UNKNOWN)
4572                         continue;
4573                 if (diff_unmodified_pair(p))
4574                         continue;
4575                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4576                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4577                         continue;
4578                 if (DIFF_PAIR_UNMERGED(p))
4579                         continue;
4580
4581                 diff_fill_sha1_info(p->one);
4582                 diff_fill_sha1_info(p->two);
4583
4584                 len1 = remove_space(p->one->path, strlen(p->one->path));
4585                 len2 = remove_space(p->two->path, strlen(p->two->path));
4586                 if (p->one->mode == 0)
4587                         len1 = snprintf(buffer, sizeof(buffer),
4588                                         "diff--gita/%.*sb/%.*s"
4589                                         "newfilemode%06o"
4590                                         "---/dev/null"
4591                                         "+++b/%.*s",
4592                                         len1, p->one->path,
4593                                         len2, p->two->path,
4594                                         p->two->mode,
4595                                         len2, p->two->path);
4596                 else if (p->two->mode == 0)
4597                         len1 = snprintf(buffer, sizeof(buffer),
4598                                         "diff--gita/%.*sb/%.*s"
4599                                         "deletedfilemode%06o"
4600                                         "---a/%.*s"
4601                                         "+++/dev/null",
4602                                         len1, p->one->path,
4603                                         len2, p->two->path,
4604                                         p->one->mode,
4605                                         len1, p->one->path);
4606                 else
4607                         len1 = snprintf(buffer, sizeof(buffer),
4608                                         "diff--gita/%.*sb/%.*s"
4609                                         "---a/%.*s"
4610                                         "+++b/%.*s",
4611                                         len1, p->one->path,
4612                                         len2, p->two->path,
4613                                         len1, p->one->path,
4614                                         len2, p->two->path);
4615                 git_SHA1_Update(&ctx, buffer, len1);
4616
4617                 if (diff_header_only)
4618                         continue;
4619
4620                 if (fill_mmfile(&mf1, p->one) < 0 ||
4621                     fill_mmfile(&mf2, p->two) < 0)
4622                         return error("unable to read files to diff");
4623
4624                 if (diff_filespec_is_binary(p->one) ||
4625                     diff_filespec_is_binary(p->two)) {
4626                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4627                                         40);
4628                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4629                                         40);
4630                         continue;
4631                 }
4632
4633                 xpp.flags = 0;
4634                 xecfg.ctxlen = 3;
4635                 xecfg.flags = 0;
4636                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4637                                   &xpp, &xecfg))
4638                         return error("unable to generate patch-id diff for %s",
4639                                      p->one->path);
4640         }
4641
4642         git_SHA1_Final(sha1, &ctx);
4643         return 0;
4644 }
4645
4646 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4647 {
4648         struct diff_queue_struct *q = &diff_queued_diff;
4649         int i;
4650         int result = diff_get_patch_id(options, sha1, diff_header_only);
4651
4652         for (i = 0; i < q->nr; i++)
4653                 diff_free_filepair(q->queue[i]);
4654
4655         free(q->queue);
4656         DIFF_QUEUE_CLEAR(q);
4657
4658         return result;
4659 }
4660
4661 static int is_summary_empty(const struct diff_queue_struct *q)
4662 {
4663         int i;
4664
4665         for (i = 0; i < q->nr; i++) {
4666                 const struct diff_filepair *p = q->queue[i];
4667
4668                 switch (p->status) {
4669                 case DIFF_STATUS_DELETED:
4670                 case DIFF_STATUS_ADDED:
4671                 case DIFF_STATUS_COPIED:
4672                 case DIFF_STATUS_RENAMED:
4673                         return 0;
4674                 default:
4675                         if (p->score)
4676                                 return 0;
4677                         if (p->one->mode && p->two->mode &&
4678                             p->one->mode != p->two->mode)
4679                                 return 0;
4680                         break;
4681                 }
4682         }
4683         return 1;
4684 }
4685
4686 static const char rename_limit_warning[] =
4687 N_("inexact rename detection was skipped due to too many files.");
4688
4689 static const char degrade_cc_to_c_warning[] =
4690 N_("only found copies from modified paths due to too many files.");
4691
4692 static const char rename_limit_advice[] =
4693 N_("you may want to set your %s variable to at least "
4694    "%d and retry the command.");
4695
4696 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4697 {
4698         if (degraded_cc)
4699                 warning(_(degrade_cc_to_c_warning));
4700         else if (needed)
4701                 warning(_(rename_limit_warning));
4702         else
4703                 return;
4704         if (0 < needed && needed < 32767)
4705                 warning(_(rename_limit_advice), varname, needed);
4706 }
4707
4708 void diff_flush(struct diff_options *options)
4709 {
4710         struct diff_queue_struct *q = &diff_queued_diff;
4711         int i, output_format = options->output_format;
4712         int separator = 0;
4713         int dirstat_by_line = 0;
4714
4715         /*
4716          * Order: raw, stat, summary, patch
4717          * or:    name/name-status/checkdiff (other bits clear)
4718          */
4719         if (!q->nr)
4720                 goto free_queue;
4721
4722         if (output_format & (DIFF_FORMAT_RAW |
4723                              DIFF_FORMAT_NAME |
4724                              DIFF_FORMAT_NAME_STATUS |
4725                              DIFF_FORMAT_CHECKDIFF)) {
4726                 for (i = 0; i < q->nr; i++) {
4727                         struct diff_filepair *p = q->queue[i];
4728                         if (check_pair_status(p))
4729                                 flush_one_pair(p, options);
4730                 }
4731                 separator++;
4732         }
4733
4734         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4735                 dirstat_by_line = 1;
4736
4737         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4738             dirstat_by_line) {
4739                 struct diffstat_t diffstat;
4740
4741                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4742                 for (i = 0; i < q->nr; i++) {
4743                         struct diff_filepair *p = q->queue[i];
4744                         if (check_pair_status(p))
4745                                 diff_flush_stat(p, options, &diffstat);
4746                 }
4747                 if (output_format & DIFF_FORMAT_NUMSTAT)
4748                         show_numstat(&diffstat, options);
4749                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4750                         show_stats(&diffstat, options);
4751                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4752                         show_shortstats(&diffstat, options);
4753                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4754                         show_dirstat_by_line(&diffstat, options);
4755                 free_diffstat_info(&diffstat);
4756                 separator++;
4757         }
4758         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4759                 show_dirstat(options);
4760
4761         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4762                 for (i = 0; i < q->nr; i++) {
4763                         diff_summary(options, q->queue[i]);
4764                 }
4765                 separator++;
4766         }
4767
4768         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4769             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4770             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4771                 /*
4772                  * run diff_flush_patch for the exit status. setting
4773                  * options->file to /dev/null should be safe, because we
4774                  * aren't supposed to produce any output anyway.
4775                  */
4776                 if (options->close_file)
4777                         fclose(options->file);
4778                 options->file = fopen("/dev/null", "w");
4779                 if (!options->file)
4780                         die_errno("Could not open /dev/null");
4781                 options->close_file = 1;
4782                 for (i = 0; i < q->nr; i++) {
4783                         struct diff_filepair *p = q->queue[i];
4784                         if (check_pair_status(p))
4785                                 diff_flush_patch(p, options);
4786                         if (options->found_changes)
4787                                 break;
4788                 }
4789         }
4790
4791         if (output_format & DIFF_FORMAT_PATCH) {
4792                 if (separator) {
4793                         fprintf(options->file, "%s%c",
4794                                 diff_line_prefix(options),
4795                                 options->line_termination);
4796                         if (options->stat_sep) {
4797                                 /* attach patch instead of inline */
4798                                 fputs(options->stat_sep, options->file);
4799                         }
4800                 }
4801
4802                 for (i = 0; i < q->nr; i++) {
4803                         struct diff_filepair *p = q->queue[i];
4804                         if (check_pair_status(p))
4805                                 diff_flush_patch(p, options);
4806                 }
4807         }
4808
4809         if (output_format & DIFF_FORMAT_CALLBACK)
4810                 options->format_callback(q, options, options->format_callback_data);
4811
4812         for (i = 0; i < q->nr; i++)
4813                 diff_free_filepair(q->queue[i]);
4814 free_queue:
4815         free(q->queue);
4816         DIFF_QUEUE_CLEAR(q);
4817         if (options->close_file)
4818                 fclose(options->file);
4819
4820         /*
4821          * Report the content-level differences with HAS_CHANGES;
4822          * diff_addremove/diff_change does not set the bit when
4823          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4824          */
4825         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4826                 if (options->found_changes)
4827                         DIFF_OPT_SET(options, HAS_CHANGES);
4828                 else
4829                         DIFF_OPT_CLR(options, HAS_CHANGES);
4830         }
4831 }
4832
4833 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4834 {
4835         return (((p->status == DIFF_STATUS_MODIFIED) &&
4836                  ((p->score &&
4837                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4838                   (!p->score &&
4839                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4840                 ((p->status != DIFF_STATUS_MODIFIED) &&
4841                  filter_bit_tst(p->status, options)));
4842 }
4843
4844 static void diffcore_apply_filter(struct diff_options *options)
4845 {
4846         int i;
4847         struct diff_queue_struct *q = &diff_queued_diff;
4848         struct diff_queue_struct outq;
4849
4850         DIFF_QUEUE_CLEAR(&outq);
4851
4852         if (!options->filter)
4853                 return;
4854
4855         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4856                 int found;
4857                 for (i = found = 0; !found && i < q->nr; i++) {
4858                         if (match_filter(options, q->queue[i]))
4859                                 found++;
4860                 }
4861                 if (found)
4862                         return;
4863
4864                 /* otherwise we will clear the whole queue
4865                  * by copying the empty outq at the end of this
4866                  * function, but first clear the current entries
4867                  * in the queue.
4868                  */
4869                 for (i = 0; i < q->nr; i++)
4870                         diff_free_filepair(q->queue[i]);
4871         }
4872         else {
4873                 /* Only the matching ones */
4874                 for (i = 0; i < q->nr; i++) {
4875                         struct diff_filepair *p = q->queue[i];
4876                         if (match_filter(options, p))
4877                                 diff_q(&outq, p);
4878                         else
4879                                 diff_free_filepair(p);
4880                 }
4881         }
4882         free(q->queue);
4883         *q = outq;
4884 }
4885
4886 /* Check whether two filespecs with the same mode and size are identical */
4887 static int diff_filespec_is_identical(struct diff_filespec *one,
4888                                       struct diff_filespec *two)
4889 {
4890         if (S_ISGITLINK(one->mode))
4891                 return 0;
4892         if (diff_populate_filespec(one, 0))
4893                 return 0;
4894         if (diff_populate_filespec(two, 0))
4895                 return 0;
4896         return !memcmp(one->data, two->data, one->size);
4897 }
4898
4899 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4900 {
4901         if (p->done_skip_stat_unmatch)
4902                 return p->skip_stat_unmatch_result;
4903
4904         p->done_skip_stat_unmatch = 1;
4905         p->skip_stat_unmatch_result = 0;
4906         /*
4907          * 1. Entries that come from stat info dirtiness
4908          *    always have both sides (iow, not create/delete),
4909          *    one side of the object name is unknown, with
4910          *    the same mode and size.  Keep the ones that
4911          *    do not match these criteria.  They have real
4912          *    differences.
4913          *
4914          * 2. At this point, the file is known to be modified,
4915          *    with the same mode and size, and the object
4916          *    name of one side is unknown.  Need to inspect
4917          *    the identical contents.
4918          */
4919         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4920             !DIFF_FILE_VALID(p->two) ||
4921             (p->one->oid_valid && p->two->oid_valid) ||
4922             (p->one->mode != p->two->mode) ||
4923             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4924             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4925             (p->one->size != p->two->size) ||
4926             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4927                 p->skip_stat_unmatch_result = 1;
4928         return p->skip_stat_unmatch_result;
4929 }
4930
4931 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4932 {
4933         int i;
4934         struct diff_queue_struct *q = &diff_queued_diff;
4935         struct diff_queue_struct outq;
4936         DIFF_QUEUE_CLEAR(&outq);
4937
4938         for (i = 0; i < q->nr; i++) {
4939                 struct diff_filepair *p = q->queue[i];
4940
4941                 if (diff_filespec_check_stat_unmatch(p))
4942                         diff_q(&outq, p);
4943                 else {
4944                         /*
4945                          * The caller can subtract 1 from skip_stat_unmatch
4946                          * to determine how many paths were dirty only
4947                          * due to stat info mismatch.
4948                          */
4949                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4950                                 diffopt->skip_stat_unmatch++;
4951                         diff_free_filepair(p);
4952                 }
4953         }
4954         free(q->queue);
4955         *q = outq;
4956 }
4957
4958 static int diffnamecmp(const void *a_, const void *b_)
4959 {
4960         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4961         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4962         const char *name_a, *name_b;
4963
4964         name_a = a->one ? a->one->path : a->two->path;
4965         name_b = b->one ? b->one->path : b->two->path;
4966         return strcmp(name_a, name_b);
4967 }
4968
4969 void diffcore_fix_diff_index(struct diff_options *options)
4970 {
4971         struct diff_queue_struct *q = &diff_queued_diff;
4972         QSORT(q->queue, q->nr, diffnamecmp);
4973 }
4974
4975 void diffcore_std(struct diff_options *options)
4976 {
4977         /* NOTE please keep the following in sync with diff_tree_combined() */
4978         if (options->skip_stat_unmatch)
4979                 diffcore_skip_stat_unmatch(options);
4980         if (!options->found_follow) {
4981                 /* See try_to_follow_renames() in tree-diff.c */
4982                 if (options->break_opt != -1)
4983                         diffcore_break(options->break_opt);
4984                 if (options->detect_rename)
4985                         diffcore_rename(options);
4986                 if (options->break_opt != -1)
4987                         diffcore_merge_broken();
4988         }
4989         if (options->pickaxe)
4990                 diffcore_pickaxe(options);
4991         if (options->orderfile)
4992                 diffcore_order(options->orderfile);
4993         if (!options->found_follow)
4994                 /* See try_to_follow_renames() in tree-diff.c */
4995                 diff_resolve_rename_copy();
4996         diffcore_apply_filter(options);
4997
4998         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4999                 DIFF_OPT_SET(options, HAS_CHANGES);
5000         else
5001                 DIFF_OPT_CLR(options, HAS_CHANGES);
5002
5003         options->found_follow = 0;
5004 }
5005
5006 int diff_result_code(struct diff_options *opt, int status)
5007 {
5008         int result = 0;
5009
5010         diff_warn_rename_limit("diff.renameLimit",
5011                                opt->needed_rename_limit,
5012                                opt->degraded_cc_to_c);
5013         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5014             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5015                 return status;
5016         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5017             DIFF_OPT_TST(opt, HAS_CHANGES))
5018                 result |= 01;
5019         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5020             DIFF_OPT_TST(opt, CHECK_FAILED))
5021                 result |= 02;
5022         return result;
5023 }
5024
5025 int diff_can_quit_early(struct diff_options *opt)
5026 {
5027         return (DIFF_OPT_TST(opt, QUICK) &&
5028                 !opt->filter &&
5029                 DIFF_OPT_TST(opt, HAS_CHANGES));
5030 }
5031
5032 /*
5033  * Shall changes to this submodule be ignored?
5034  *
5035  * Submodule changes can be configured to be ignored separately for each path,
5036  * but that configuration can be overridden from the command line.
5037  */
5038 static int is_submodule_ignored(const char *path, struct diff_options *options)
5039 {
5040         int ignored = 0;
5041         unsigned orig_flags = options->flags;
5042         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5043                 set_diffopt_flags_from_submodule_config(options, path);
5044         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5045                 ignored = 1;
5046         options->flags = orig_flags;
5047         return ignored;
5048 }
5049
5050 void diff_addremove(struct diff_options *options,
5051                     int addremove, unsigned mode,
5052                     const unsigned char *sha1,
5053                     int sha1_valid,
5054                     const char *concatpath, unsigned dirty_submodule)
5055 {
5056         struct diff_filespec *one, *two;
5057
5058         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5059                 return;
5060
5061         /* This may look odd, but it is a preparation for
5062          * feeding "there are unchanged files which should
5063          * not produce diffs, but when you are doing copy
5064          * detection you would need them, so here they are"
5065          * entries to the diff-core.  They will be prefixed
5066          * with something like '=' or '*' (I haven't decided
5067          * which but should not make any difference).
5068          * Feeding the same new and old to diff_change()
5069          * also has the same effect.
5070          * Before the final output happens, they are pruned after
5071          * merged into rename/copy pairs as appropriate.
5072          */
5073         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5074                 addremove = (addremove == '+' ? '-' :
5075                              addremove == '-' ? '+' : addremove);
5076
5077         if (options->prefix &&
5078             strncmp(concatpath, options->prefix, options->prefix_length))
5079                 return;
5080
5081         one = alloc_filespec(concatpath);
5082         two = alloc_filespec(concatpath);
5083
5084         if (addremove != '+')
5085                 fill_filespec(one, sha1, sha1_valid, mode);
5086         if (addremove != '-') {
5087                 fill_filespec(two, sha1, sha1_valid, mode);
5088                 two->dirty_submodule = dirty_submodule;
5089         }
5090
5091         diff_queue(&diff_queued_diff, one, two);
5092         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5093                 DIFF_OPT_SET(options, HAS_CHANGES);
5094 }
5095
5096 void diff_change(struct diff_options *options,
5097                  unsigned old_mode, unsigned new_mode,
5098                  const unsigned char *old_sha1,
5099                  const unsigned char *new_sha1,
5100                  int old_sha1_valid, int new_sha1_valid,
5101                  const char *concatpath,
5102                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5103 {
5104         struct diff_filespec *one, *two;
5105         struct diff_filepair *p;
5106
5107         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5108             is_submodule_ignored(concatpath, options))
5109                 return;
5110
5111         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5112                 unsigned tmp;
5113                 const unsigned char *tmp_c;
5114                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5115                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5116                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5117                         new_sha1_valid = tmp;
5118                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5119                         new_dirty_submodule = tmp;
5120         }
5121
5122         if (options->prefix &&
5123             strncmp(concatpath, options->prefix, options->prefix_length))
5124                 return;
5125
5126         one = alloc_filespec(concatpath);
5127         two = alloc_filespec(concatpath);
5128         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5129         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5130         one->dirty_submodule = old_dirty_submodule;
5131         two->dirty_submodule = new_dirty_submodule;
5132         p = diff_queue(&diff_queued_diff, one, two);
5133
5134         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5135                 return;
5136
5137         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5138             !diff_filespec_check_stat_unmatch(p))
5139                 return;
5140
5141         DIFF_OPT_SET(options, HAS_CHANGES);
5142 }
5143
5144 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5145 {
5146         struct diff_filepair *pair;
5147         struct diff_filespec *one, *two;
5148
5149         if (options->prefix &&
5150             strncmp(path, options->prefix, options->prefix_length))
5151                 return NULL;
5152
5153         one = alloc_filespec(path);
5154         two = alloc_filespec(path);
5155         pair = diff_queue(&diff_queued_diff, one, two);
5156         pair->is_unmerged = 1;
5157         return pair;
5158 }
5159
5160 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5161                 size_t *outsize)
5162 {
5163         struct diff_tempfile *temp;
5164         const char *argv[3];
5165         const char **arg = argv;
5166         struct child_process child = CHILD_PROCESS_INIT;
5167         struct strbuf buf = STRBUF_INIT;
5168         int err = 0;
5169
5170         temp = prepare_temp_file(spec->path, spec);
5171         *arg++ = pgm;
5172         *arg++ = temp->name;
5173         *arg = NULL;
5174
5175         child.use_shell = 1;
5176         child.argv = argv;
5177         child.out = -1;
5178         if (start_command(&child)) {
5179                 remove_tempfile();
5180                 return NULL;
5181         }
5182
5183         if (strbuf_read(&buf, child.out, 0) < 0)
5184                 err = error("error reading from textconv command '%s'", pgm);
5185         close(child.out);
5186
5187         if (finish_command(&child) || err) {
5188                 strbuf_release(&buf);
5189                 remove_tempfile();
5190                 return NULL;
5191         }
5192         remove_tempfile();
5193
5194         return strbuf_detach(&buf, outsize);
5195 }
5196
5197 size_t fill_textconv(struct userdiff_driver *driver,
5198                      struct diff_filespec *df,
5199                      char **outbuf)
5200 {
5201         size_t size;
5202
5203         if (!driver) {
5204                 if (!DIFF_FILE_VALID(df)) {
5205                         *outbuf = "";
5206                         return 0;
5207                 }
5208                 if (diff_populate_filespec(df, 0))
5209                         die("unable to read files to diff");
5210                 *outbuf = df->data;
5211                 return df->size;
5212         }
5213
5214         if (!driver->textconv)
5215                 die("BUG: fill_textconv called with non-textconv driver");
5216
5217         if (driver->textconv_cache && df->oid_valid) {
5218                 *outbuf = notes_cache_get(driver->textconv_cache,
5219                                           df->oid.hash,
5220                                           &size);
5221                 if (*outbuf)
5222                         return size;
5223         }
5224
5225         *outbuf = run_textconv(driver->textconv, df, &size);
5226         if (!*outbuf)
5227                 die("unable to read files to diff");
5228
5229         if (driver->textconv_cache && df->oid_valid) {
5230                 /* ignore errors, as we might be in a readonly repository */
5231                 notes_cache_put(driver->textconv_cache, df->oid.hash, *outbuf,
5232                                 size);
5233                 /*
5234                  * we could save up changes and flush them all at the end,
5235                  * but we would need an extra call after all diffing is done.
5236                  * Since generating a cache entry is the slow path anyway,
5237                  * this extra overhead probably isn't a big deal.
5238                  */
5239                 notes_cache_write(driver->textconv_cache);
5240         }
5241
5242         return size;
5243 }
5244
5245 void setup_diff_pager(struct diff_options *opt)
5246 {
5247         /*
5248          * If the user asked for our exit code, then either they want --quiet
5249          * or --exit-code. We should definitely not bother with a pager in the
5250          * former case, as we will generate no output. Since we still properly
5251          * report our exit code even when a pager is run, we _could_ run a
5252          * pager with --exit-code. But since we have not done so historically,
5253          * and because it is easy to find people oneline advising "git diff
5254          * --exit-code" in hooks and other scripts, we do not do so.
5255          */
5256         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5257             check_pager_config("diff") != 0)
5258                 setup_pager();
5259 }