diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "userdiff.h"
  17 #include "submodule-config.h"
  18 #include "submodule.h"
  19 #include "ll-merge.h"
  20 #include "string-list.h"
  21 #include "argv-array.h"
  22 #include "graph.h"
  23
  24 #ifdef NO_FAST_WORKING_DIRECTORY
  25 #define FAST_WORKING_DIRECTORY 0
  26 #else
  27 #define FAST_WORKING_DIRECTORY 1
  28 #endif
  29
  30 static int diff_detect_rename_default;
  31 static int diff_indent_heuristic = 1;
  32 static int diff_rename_limit_default = 400;
  33 static int diff_suppress_blank_empty;
  34 static int diff_use_color_default = -1;
  35 static int diff_context_default = 3;
  36 static int diff_interhunk_context_default;
  37 static const char *diff_word_regex_cfg;
  38 static const char *external_diff_cmd_cfg;
  39 static const char *diff_order_file_cfg;
  40 int diff_auto_refresh_index = 1;
  41 static int diff_mnemonic_prefix;
  42 static int diff_no_prefix;
  43 static int diff_stat_graph_width;
  44 static int diff_dirstat_permille_default = 30;
  45 static struct diff_options default_diff_options;
  46 static long diff_algorithm;
  47 static unsigned ws_error_highlight_default = WSEH_NEW;
  48
  49 static char diff_colors[][COLOR_MAXLEN] = {
  50         GIT_COLOR_RESET,
  51         GIT_COLOR_NORMAL,       /* CONTEXT */
  52         GIT_COLOR_BOLD,         /* METAINFO */
  53         GIT_COLOR_CYAN,         /* FRAGINFO */
  54         GIT_COLOR_RED,          /* OLD */
  55         GIT_COLOR_GREEN,        /* NEW */
  56         GIT_COLOR_YELLOW,       /* COMMIT */
  57         GIT_COLOR_BG_RED,       /* WHITESPACE */
  58         GIT_COLOR_NORMAL,       /* FUNCINFO */
  59 };
  60
  61 static NORETURN void die_want_option(const char *option_name)
  62 {
  63         die(_("option '%s' requires a value"), option_name);
  64 }
  65
  66 static int parse_diff_color_slot(const char *var)
  67 {
  68         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  69                 return DIFF_CONTEXT;
  70         if (!strcasecmp(var, "meta"))
  71                 return DIFF_METAINFO;
  72         if (!strcasecmp(var, "frag"))
  73                 return DIFF_FRAGINFO;
  74         if (!strcasecmp(var, "old"))
  75                 return DIFF_FILE_OLD;
  76         if (!strcasecmp(var, "new"))
  77                 return DIFF_FILE_NEW;
  78         if (!strcasecmp(var, "commit"))
  79                 return DIFF_COMMIT;
  80         if (!strcasecmp(var, "whitespace"))
  81                 return DIFF_WHITESPACE;
  82         if (!strcasecmp(var, "func"))
  83                 return DIFF_FUNCINFO;
  84         return -1;
  85 }
  86
  87 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  88                                 struct strbuf *errmsg)
  89 {
  90         char *params_copy = xstrdup(params_string);
  91         struct string_list params = STRING_LIST_INIT_NODUP;
  92         int ret = 0;
  93         int i;
  94
  95         if (*params_copy)
  96                 string_list_split_in_place(&params, params_copy, ',', -1);
  97         for (i = 0; i < params.nr; i++) {
  98                 const char *p = params.items[i].string;
  99                 if (!strcmp(p, "changes")) {
 100                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 101                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 102                 } else if (!strcmp(p, "lines")) {
 103                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
 104                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 105                 } else if (!strcmp(p, "files")) {
 106                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 107                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 108                 } else if (!strcmp(p, "noncumulative")) {
 109                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 110                 } else if (!strcmp(p, "cumulative")) {
 111                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 112                 } else if (isdigit(*p)) {
 113                         char *end;
 114                         int permille = strtoul(p, &end, 10) * 10;
 115                         if (*end == '.' && isdigit(*++end)) {
 116                                 /* only use first digit */
 117                                 permille += *end - '0';
 118                                 /* .. and ignore any further digits */
 119                                 while (isdigit(*++end))
 120                                         ; /* nothing */
 121                         }
 122                         if (!*end)
 123                                 options->dirstat_permille = permille;
 124                         else {
 125                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 126                                             p);
 127                                 ret++;
 128                         }
 129                 } else {
 130                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 131                         ret++;
 132                 }
 133
 134         }
 135         string_list_clear(&params, 0);
 136         free(params_copy);
 137         return ret;
 138 }
 139
 140 static int parse_submodule_params(struct diff_options *options, const char *value)
 141 {
 142         if (!strcmp(value, "log"))
 143                 options->submodule_format = DIFF_SUBMODULE_LOG;
 144         else if (!strcmp(value, "short"))
 145                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 146         else if (!strcmp(value, "diff"))
 147                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 148         else
 149                 return -1;
 150         return 0;
 151 }
 152
 153 static int git_config_rename(const char *var, const char *value)
 154 {
 155         if (!value)
 156                 return DIFF_DETECT_RENAME;
 157         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 158                 return  DIFF_DETECT_COPY;
 159         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 160 }
 161
 162 long parse_algorithm_value(const char *value)
 163 {
 164         if (!value)
 165                 return -1;
 166         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 167                 return 0;
 168         else if (!strcasecmp(value, "minimal"))
 169                 return XDF_NEED_MINIMAL;
 170         else if (!strcasecmp(value, "patience"))
 171                 return XDF_PATIENCE_DIFF;
 172         else if (!strcasecmp(value, "histogram"))
 173                 return XDF_HISTOGRAM_DIFF;
 174         return -1;
 175 }
 176
 177 static int parse_one_token(const char **arg, const char *token)
 178 {
 179         const char *rest;
 180         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 181                 *arg = rest;
 182                 return 1;
 183         }
 184         return 0;
 185 }
 186
 187 static int parse_ws_error_highlight(const char *arg)
 188 {
 189         const char *orig_arg = arg;
 190         unsigned val = 0;
 191
 192         while (*arg) {
 193                 if (parse_one_token(&arg, "none"))
 194                         val = 0;
 195                 else if (parse_one_token(&arg, "default"))
 196                         val = WSEH_NEW;
 197                 else if (parse_one_token(&arg, "all"))
 198                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 199                 else if (parse_one_token(&arg, "new"))
 200                         val |= WSEH_NEW;
 201                 else if (parse_one_token(&arg, "old"))
 202                         val |= WSEH_OLD;
 203                 else if (parse_one_token(&arg, "context"))
 204                         val |= WSEH_CONTEXT;
 205                 else {
 206                         return -1 - (int)(arg - orig_arg);
 207                 }
 208                 if (*arg)
 209                         arg++;
 210         }
 211         return val;
 212 }
 213
 214 /*
 215  * These are to give UI layer defaults.
 216  * The core-level commands such as git-diff-files should
 217  * never be affected by the setting of diff.renames
 218  * the user happens to have in the configuration file.
 219  */
 220 void init_diff_ui_defaults(void)
 221 {
 222         diff_detect_rename_default = 1;
 223 }
 224
 225 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 226 {
 227         if (!strcmp(var, "diff.indentheuristic"))
 228                 diff_indent_heuristic = git_config_bool(var, value);
 229         return 0;
 230 }
 231
 232 int git_diff_ui_config(const char *var, const char *value, void *cb)
 233 {
 234         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 235                 diff_use_color_default = git_config_colorbool(var, value);
 236                 return 0;
 237         }
 238         if (!strcmp(var, "diff.context")) {
 239                 diff_context_default = git_config_int(var, value);
 240                 if (diff_context_default < 0)
 241                         return -1;
 242                 return 0;
 243         }
 244         if (!strcmp(var, "diff.interhunkcontext")) {
 245                 diff_interhunk_context_default = git_config_int(var, value);
 246                 if (diff_interhunk_context_default < 0)
 247                         return -1;
 248                 return 0;
 249         }
 250         if (!strcmp(var, "diff.renames")) {
 251                 diff_detect_rename_default = git_config_rename(var, value);
 252                 return 0;
 253         }
 254         if (!strcmp(var, "diff.autorefreshindex")) {
 255                 diff_auto_refresh_index = git_config_bool(var, value);
 256                 return 0;
 257         }
 258         if (!strcmp(var, "diff.mnemonicprefix")) {
 259                 diff_mnemonic_prefix = git_config_bool(var, value);
 260                 return 0;
 261         }
 262         if (!strcmp(var, "diff.noprefix")) {
 263                 diff_no_prefix = git_config_bool(var, value);
 264                 return 0;
 265         }
 266         if (!strcmp(var, "diff.statgraphwidth")) {
 267                 diff_stat_graph_width = git_config_int(var, value);
 268                 return 0;
 269         }
 270         if (!strcmp(var, "diff.external"))
 271                 return git_config_string(&external_diff_cmd_cfg, var, value);
 272         if (!strcmp(var, "diff.wordregex"))
 273                 return git_config_string(&diff_word_regex_cfg, var, value);
 274         if (!strcmp(var, "diff.orderfile"))
 275                 return git_config_pathname(&diff_order_file_cfg, var, value);
 276
 277         if (!strcmp(var, "diff.ignoresubmodules"))
 278                 handle_ignore_submodules_arg(&default_diff_options, value);
 279
 280         if (!strcmp(var, "diff.submodule")) {
 281                 if (parse_submodule_params(&default_diff_options, value))
 282                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 283                                 value);
 284                 return 0;
 285         }
 286
 287         if (!strcmp(var, "diff.algorithm")) {
 288                 diff_algorithm = parse_algorithm_value(value);
 289                 if (diff_algorithm < 0)
 290                         return -1;
 291                 return 0;
 292         }
 293
 294         if (!strcmp(var, "diff.wserrorhighlight")) {
 295                 int val = parse_ws_error_highlight(value);
 296                 if (val < 0)
 297                         return -1;
 298                 ws_error_highlight_default = val;
 299                 return 0;
 300         }
 301
 302         if (git_color_config(var, value, cb) < 0)
 303                 return -1;
 304
 305         return git_diff_basic_config(var, value, cb);
 306 }
 307
 308 int git_diff_basic_config(const char *var, const char *value, void *cb)
 309 {
 310         const char *name;
 311
 312         if (!strcmp(var, "diff.renamelimit")) {
 313                 diff_rename_limit_default = git_config_int(var, value);
 314                 return 0;
 315         }
 316
 317         if (userdiff_config(var, value) < 0)
 318                 return -1;
 319
 320         if (skip_prefix(var, "diff.color.", &name) ||
 321             skip_prefix(var, "color.diff.", &name)) {
 322                 int slot = parse_diff_color_slot(name);
 323                 if (slot < 0)
 324                         return 0;
 325                 if (!value)
 326                         return config_error_nonbool(var);
 327                 return color_parse(value, diff_colors[slot]);
 328         }
 329
 330         /* like GNU diff's --suppress-blank-empty option  */
 331         if (!strcmp(var, "diff.suppressblankempty") ||
 332                         /* for backwards compatibility */
 333                         !strcmp(var, "diff.suppress-blank-empty")) {
 334                 diff_suppress_blank_empty = git_config_bool(var, value);
 335                 return 0;
 336         }
 337
 338         if (!strcmp(var, "diff.dirstat")) {
 339                 struct strbuf errmsg = STRBUF_INIT;
 340                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 341                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 342                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 343                                 errmsg.buf);
 344                 strbuf_release(&errmsg);
 345                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 346                 return 0;
 347         }
 348
 349         if (starts_with(var, "submodule."))
 350                 return parse_submodule_config_option(var, value);
 351
 352         if (git_diff_heuristic_config(var, value, cb) < 0)
 353                 return -1;
 354
 355         return git_default_config(var, value, cb);
 356 }
 357
 358 static char *quote_two(const char *one, const char *two)
 359 {
 360         int need_one = quote_c_style(one, NULL, NULL, 1);
 361         int need_two = quote_c_style(two, NULL, NULL, 1);
 362         struct strbuf res = STRBUF_INIT;
 363
 364         if (need_one + need_two) {
 365                 strbuf_addch(&res, '"');
 366                 quote_c_style(one, &res, NULL, 1);
 367                 quote_c_style(two, &res, NULL, 1);
 368                 strbuf_addch(&res, '"');
 369         } else {
 370                 strbuf_addstr(&res, one);
 371                 strbuf_addstr(&res, two);
 372         }
 373         return strbuf_detach(&res, NULL);
 374 }
 375
 376 static const char *external_diff(void)
 377 {
 378         static const char *external_diff_cmd = NULL;
 379         static int done_preparing = 0;
 380
 381         if (done_preparing)
 382                 return external_diff_cmd;
 383         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 384         if (!external_diff_cmd)
 385                 external_diff_cmd = external_diff_cmd_cfg;
 386         done_preparing = 1;
 387         return external_diff_cmd;
 388 }
 389
 390 /*
 391  * Keep track of files used for diffing. Sometimes such an entry
 392  * refers to a temporary file, sometimes to an existing file, and
 393  * sometimes to "/dev/null".
 394  */
 395 static struct diff_tempfile {
 396         /*
 397          * filename external diff should read from, or NULL if this
 398          * entry is currently not in use:
 399          */
 400         const char *name;
 401
 402         char hex[GIT_MAX_HEXSZ + 1];
 403         char mode[10];
 404
 405         /*
 406          * If this diff_tempfile instance refers to a temporary file,
 407          * this tempfile object is used to manage its lifetime.
 408          */
 409         struct tempfile tempfile;
 410 } diff_temp[2];
 411
 412 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 413
 414 struct emit_callback {
 415         int color_diff;
 416         unsigned ws_rule;
 417         int blank_at_eof_in_preimage;
 418         int blank_at_eof_in_postimage;
 419         int lno_in_preimage;
 420         int lno_in_postimage;
 421         sane_truncate_fn truncate;
 422         const char **label_path;
 423         struct diff_words_data *diff_words;
 424         struct diff_options *opt;
 425         struct strbuf *header;
 426 };
 427
 428 static int count_lines(const char *data, int size)
 429 {
 430         int count, ch, completely_empty = 1, nl_just_seen = 0;
 431         count = 0;
 432         while (0 < size--) {
 433                 ch = *data++;
 434                 if (ch == '\n') {
 435                         count++;
 436                         nl_just_seen = 1;
 437                         completely_empty = 0;
 438                 }
 439                 else {
 440                         nl_just_seen = 0;
 441                         completely_empty = 0;
 442                 }
 443         }
 444         if (completely_empty)
 445                 return 0;
 446         if (!nl_just_seen)
 447                 count++; /* no trailing newline */
 448         return count;
 449 }
 450
 451 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 452 {
 453         if (!DIFF_FILE_VALID(one)) {
 454                 mf->ptr = (char *)""; /* does not matter */
 455                 mf->size = 0;
 456                 return 0;
 457         }
 458         else if (diff_populate_filespec(one, 0))
 459                 return -1;
 460
 461         mf->ptr = one->data;
 462         mf->size = one->size;
 463         return 0;
 464 }
 465
 466 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 467 static unsigned long diff_filespec_size(struct diff_filespec *one)
 468 {
 469         if (!DIFF_FILE_VALID(one))
 470                 return 0;
 471         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 472         return one->size;
 473 }
 474
 475 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 476 {
 477         char *ptr = mf->ptr;
 478         long size = mf->size;
 479         int cnt = 0;
 480
 481         if (!size)
 482                 return cnt;
 483         ptr += size - 1; /* pointing at the very end */
 484         if (*ptr != '\n')
 485                 ; /* incomplete line */
 486         else
 487                 ptr--; /* skip the last LF */
 488         while (mf->ptr < ptr) {
 489                 char *prev_eol;
 490                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 491                         if (*prev_eol == '\n')
 492                                 break;
 493                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 494                         break;
 495                 cnt++;
 496                 ptr = prev_eol - 1;
 497         }
 498         return cnt;
 499 }
 500
 501 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 502                                struct emit_callback *ecbdata)
 503 {
 504         int l1, l2, at;
 505         unsigned ws_rule = ecbdata->ws_rule;
 506         l1 = count_trailing_blank(mf1, ws_rule);
 507         l2 = count_trailing_blank(mf2, ws_rule);
 508         if (l2 <= l1) {
 509                 ecbdata->blank_at_eof_in_preimage = 0;
 510                 ecbdata->blank_at_eof_in_postimage = 0;
 511                 return;
 512         }
 513         at = count_lines(mf1->ptr, mf1->size);
 514         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 515
 516         at = count_lines(mf2->ptr, mf2->size);
 517         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 518 }
 519
 520 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 521                         int first, const char *line, int len)
 522 {
 523         int has_trailing_newline, has_trailing_carriage_return;
 524         int nofirst;
 525         FILE *file = o->file;
 526
 527         fputs(diff_line_prefix(o), file);
 528
 529         if (len == 0) {
 530                 has_trailing_newline = (first == '\n');
 531                 has_trailing_carriage_return = (!has_trailing_newline &&
 532                                                 (first == '\r'));
 533                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 534         } else {
 535                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 536                 if (has_trailing_newline)
 537                         len--;
 538                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 539                 if (has_trailing_carriage_return)
 540                         len--;
 541                 nofirst = 0;
 542         }
 543
 544         if (len || !nofirst) {
 545                 fputs(set, file);
 546                 if (!nofirst)
 547                         fputc(first, file);
 548                 fwrite(line, len, 1, file);
 549                 fputs(reset, file);
 550         }
 551         if (has_trailing_carriage_return)
 552                 fputc('\r', file);
 553         if (has_trailing_newline)
 554                 fputc('\n', file);
 555 }
 556
 557 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 558                       const char *line, int len)
 559 {
 560         emit_line_0(o, set, reset, line[0], line+1, len-1);
 561 }
 562
 563 enum diff_symbol {
 564         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 565         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 566         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 567         DIFF_SYMBOL_BINARY_DIFF_BODY,
 568         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 569         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 570         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 571         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 572         DIFF_SYMBOL_STATS_LINE,
 573         DIFF_SYMBOL_WORD_DIFF,
 574         DIFF_SYMBOL_SUBMODULE_ADD,
 575         DIFF_SYMBOL_SUBMODULE_DEL,
 576         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 577         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 578         DIFF_SYMBOL_SUBMODULE_HEADER,
 579         DIFF_SYMBOL_SUBMODULE_ERROR,
 580         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 581         DIFF_SYMBOL_REWRITE_DIFF,
 582         DIFF_SYMBOL_BINARY_FILES,
 583         DIFF_SYMBOL_HEADER,
 584         DIFF_SYMBOL_FILEPAIR_PLUS,
 585         DIFF_SYMBOL_FILEPAIR_MINUS,
 586         DIFF_SYMBOL_WORDS_PORCELAIN,
 587         DIFF_SYMBOL_WORDS,
 588         DIFF_SYMBOL_CONTEXT,
 589         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 590         DIFF_SYMBOL_PLUS,
 591         DIFF_SYMBOL_MINUS,
 592         DIFF_SYMBOL_NO_LF_EOF,
 593         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 594         DIFF_SYMBOL_CONTEXT_MARKER,
 595         DIFF_SYMBOL_SEPARATOR
 596 };
 597 /*
 598  * Flags for content lines:
 599  * 0..12 are whitespace rules
 600  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 601  * 16 is marking if the line is blank at EOF
 602  */
 603 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
 604 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 605
 606 static void emit_line_ws_markup(struct diff_options *o,
 607                                 const char *set, const char *reset,
 608                                 const char *line, int len, char sign,
 609                                 unsigned ws_rule, int blank_at_eof)
 610 {
 611         const char *ws = NULL;
 612
 613         if (o->ws_error_highlight & ws_rule) {
 614                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 615                 if (!*ws)
 616                         ws = NULL;
 617         }
 618
 619         if (!ws)
 620                 emit_line_0(o, set, reset, sign, line, len);
 621         else if (blank_at_eof)
 622                 /* Blank line at EOF - paint '+' as well */
 623                 emit_line_0(o, ws, reset, sign, line, len);
 624         else {
 625                 /* Emit just the prefix, then the rest. */
 626                 emit_line_0(o, set, reset, sign, "", 0);
 627                 ws_check_emit(line, len, ws_rule,
 628                               o->file, set, reset, ws);
 629         }
 630 }
 631
 632 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
 633                              const char *line, int len, unsigned flags)
 634 {
 635         static const char *nneof = " No newline at end of file\n";
 636         const char *context, *reset, *set, *meta, *fraginfo;
 637         struct strbuf sb = STRBUF_INIT;
 638         switch (s) {
 639         case DIFF_SYMBOL_NO_LF_EOF:
 640                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 641                 reset = diff_get_color_opt(o, DIFF_RESET);
 642                 putc('\n', o->file);
 643                 emit_line_0(o, context, reset, '\\',
 644                             nneof, strlen(nneof));
 645                 break;
 646         case DIFF_SYMBOL_SUBMODULE_HEADER:
 647         case DIFF_SYMBOL_SUBMODULE_ERROR:
 648         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
 649         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
 650         case DIFF_SYMBOL_STATS_LINE:
 651         case DIFF_SYMBOL_BINARY_DIFF_BODY:
 652         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
 653                 emit_line(o, "", "", line, len);
 654                 break;
 655         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
 656         case DIFF_SYMBOL_CONTEXT_MARKER:
 657                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 658                 reset = diff_get_color_opt(o, DIFF_RESET);
 659                 emit_line(o, context, reset, line, len);
 660                 break;
 661         case DIFF_SYMBOL_SEPARATOR:
 662                 fprintf(o->file, "%s%c",
 663                         diff_line_prefix(o),
 664                         o->line_termination);
 665                 break;
 666         case DIFF_SYMBOL_CONTEXT:
 667                 set = diff_get_color_opt(o, DIFF_CONTEXT);
 668                 reset = diff_get_color_opt(o, DIFF_RESET);
 669                 emit_line_ws_markup(o, set, reset, line, len, ' ',
 670                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
 671                 break;
 672         case DIFF_SYMBOL_PLUS:
 673                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
 674                 reset = diff_get_color_opt(o, DIFF_RESET);
 675                 emit_line_ws_markup(o, set, reset, line, len, '+',
 676                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
 677                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
 678                 break;
 679         case DIFF_SYMBOL_MINUS:
 680                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
 681                 reset = diff_get_color_opt(o, DIFF_RESET);
 682                 emit_line_ws_markup(o, set, reset, line, len, '-',
 683                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
 684                 break;
 685         case DIFF_SYMBOL_WORDS_PORCELAIN:
 686                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 687                 reset = diff_get_color_opt(o, DIFF_RESET);
 688                 emit_line(o, context, reset, line, len);
 689                 fputs("~\n", o->file);
 690                 break;
 691         case DIFF_SYMBOL_WORDS:
 692                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 693                 reset = diff_get_color_opt(o, DIFF_RESET);
 694                 /*
 695                  * Skip the prefix character, if any.  With
 696                  * diff_suppress_blank_empty, there may be
 697                  * none.
 698                  */
 699                 if (line[0] != '\n') {
 700                         line++;
 701                         len--;
 702                 }
 703                 emit_line(o, context, reset, line, len);
 704                 break;
 705         case DIFF_SYMBOL_FILEPAIR_PLUS:
 706                 meta = diff_get_color_opt(o, DIFF_METAINFO);
 707                 reset = diff_get_color_opt(o, DIFF_RESET);
 708                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
 709                         line, reset,
 710                         strchr(line, ' ') ? "\t" : "");
 711                 break;
 712         case DIFF_SYMBOL_FILEPAIR_MINUS:
 713                 meta = diff_get_color_opt(o, DIFF_METAINFO);
 714                 reset = diff_get_color_opt(o, DIFF_RESET);
 715                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
 716                         line, reset,
 717                         strchr(line, ' ') ? "\t" : "");
 718                 break;
 719         case DIFF_SYMBOL_BINARY_FILES:
 720         case DIFF_SYMBOL_HEADER:
 721                 fprintf(o->file, "%s", line);
 722                 break;
 723         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
 724                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
 725                 break;
 726         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
 727                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
 728                 break;
 729         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
 730                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
 731                 break;
 732         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
 733                 fputs(diff_line_prefix(o), o->file);
 734                 fputc('\n', o->file);
 735                 break;
 736         case DIFF_SYMBOL_REWRITE_DIFF:
 737                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 738                 reset = diff_get_color_opt(o, DIFF_RESET);
 739                 emit_line(o, fraginfo, reset, line, len);
 740                 break;
 741         case DIFF_SYMBOL_SUBMODULE_ADD:
 742                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
 743                 reset = diff_get_color_opt(o, DIFF_RESET);
 744                 emit_line(o, set, reset, line, len);
 745                 break;
 746         case DIFF_SYMBOL_SUBMODULE_DEL:
 747                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
 748                 reset = diff_get_color_opt(o, DIFF_RESET);
 749                 emit_line(o, set, reset, line, len);
 750                 break;
 751         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
 752                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
 753                         diff_line_prefix(o), line);
 754                 break;
 755         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
 756                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
 757                         diff_line_prefix(o), line);
 758                 break;
 759         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
 760                 emit_line(o, "", "", " 0 files changed\n",
 761                           strlen(" 0 files changed\n"));
 762                 break;
 763         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
 764                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
 765                 break;
 766         case DIFF_SYMBOL_WORD_DIFF:
 767                 fprintf(o->file, "%.*s", len, line);
 768                 break;
 769         default:
 770                 die("BUG: unknown diff symbol");
 771         }
 772         strbuf_release(&sb);
 773 }
 774
 775 void diff_emit_submodule_del(struct diff_options *o, const char *line)
 776 {
 777         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
 778 }
 779
 780 void diff_emit_submodule_add(struct diff_options *o, const char *line)
 781 {
 782         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
 783 }
 784
 785 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
 786 {
 787         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 788                          path, strlen(path), 0);
 789 }
 790
 791 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
 792 {
 793         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
 794                          path, strlen(path), 0);
 795 }
 796
 797 void diff_emit_submodule_header(struct diff_options *o, const char *header)
 798 {
 799         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
 800                          header, strlen(header), 0);
 801 }
 802
 803 void diff_emit_submodule_error(struct diff_options *o, const char *err)
 804 {
 805         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
 806 }
 807
 808 void diff_emit_submodule_pipethrough(struct diff_options *o,
 809                                      const char *line, int len)
 810 {
 811         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
 812 }
 813
 814 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 815 {
 816         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 817               ecbdata->blank_at_eof_in_preimage &&
 818               ecbdata->blank_at_eof_in_postimage &&
 819               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 820               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 821                 return 0;
 822         return ws_blank_line(line, len, ecbdata->ws_rule);
 823 }
 824
 825 static void emit_add_line(const char *reset,
 826                           struct emit_callback *ecbdata,
 827                           const char *line, int len)
 828 {
 829         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
 830         if (new_blank_line_at_eof(ecbdata, line, len))
 831                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
 832
 833         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
 834 }
 835
 836 static void emit_del_line(const char *reset,
 837                           struct emit_callback *ecbdata,
 838                           const char *line, int len)
 839 {
 840         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
 841         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
 842 }
 843
 844 static void emit_context_line(const char *reset,
 845                               struct emit_callback *ecbdata,
 846                               const char *line, int len)
 847 {
 848         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
 849         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
 850 }
 851
 852 static void emit_hunk_header(struct emit_callback *ecbdata,
 853                              const char *line, int len)
 854 {
 855         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 856         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 857         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 858         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 859         static const char atat[2] = { '@', '@' };
 860         const char *cp, *ep;
 861         struct strbuf msgbuf = STRBUF_INIT;
 862         int org_len = len;
 863         int i = 1;
 864
 865         /*
 866          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 867          * it always is at least 10 bytes long.
 868          */
 869         if (len < 10 ||
 870             memcmp(line, atat, 2) ||
 871             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 872                 emit_diff_symbol(ecbdata->opt,
 873                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
 874                 return;
 875         }
 876         ep += 2; /* skip over @@ */
 877
 878         /* The hunk header in fraginfo color */
 879         strbuf_addstr(&msgbuf, frag);
 880         strbuf_add(&msgbuf, line, ep - line);
 881         strbuf_addstr(&msgbuf, reset);
 882
 883         /*
 884          * trailing "\r\n"
 885          */
 886         for ( ; i < 3; i++)
 887                 if (line[len - i] == '\r' || line[len - i] == '\n')
 888                         len--;
 889
 890         /* blank before the func header */
 891         for (cp = ep; ep - line < len; ep++)
 892                 if (*ep != ' ' && *ep != '\t')
 893                         break;
 894         if (ep != cp) {
 895                 strbuf_addstr(&msgbuf, context);
 896                 strbuf_add(&msgbuf, cp, ep - cp);
 897                 strbuf_addstr(&msgbuf, reset);
 898         }
 899
 900         if (ep < line + len) {
 901                 strbuf_addstr(&msgbuf, func);
 902                 strbuf_add(&msgbuf, ep, line + len - ep);
 903                 strbuf_addstr(&msgbuf, reset);
 904         }
 905
 906         strbuf_add(&msgbuf, line + len, org_len - len);
 907         strbuf_complete_line(&msgbuf);
 908         emit_diff_symbol(ecbdata->opt,
 909                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
 910         strbuf_release(&msgbuf);
 911 }
 912
 913 static struct diff_tempfile *claim_diff_tempfile(void) {
 914         int i;
 915         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 916                 if (!diff_temp[i].name)
 917                         return diff_temp + i;
 918         die("BUG: diff is failing to clean up its tempfiles");
 919 }
 920
 921 static void remove_tempfile(void)
 922 {
 923         int i;
 924         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 925                 if (is_tempfile_active(&diff_temp[i].tempfile))
 926                         delete_tempfile(&diff_temp[i].tempfile);
 927                 diff_temp[i].name = NULL;
 928         }
 929 }
 930
 931 static void add_line_count(struct strbuf *out, int count)
 932 {
 933         switch (count) {
 934         case 0:
 935                 strbuf_addstr(out, "0,0");
 936                 break;
 937         case 1:
 938                 strbuf_addstr(out, "1");
 939                 break;
 940         default:
 941                 strbuf_addf(out, "1,%d", count);
 942                 break;
 943         }
 944 }
 945
 946 static void emit_rewrite_lines(struct emit_callback *ecb,
 947                                int prefix, const char *data, int size)
 948 {
 949         const char *endp = NULL;
 950         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 951
 952         while (0 < size) {
 953                 int len;
 954
 955                 endp = memchr(data, '\n', size);
 956                 len = endp ? (endp - data + 1) : size;
 957                 if (prefix != '+') {
 958                         ecb->lno_in_preimage++;
 959                         emit_del_line(reset, ecb, data, len);
 960                 } else {
 961                         ecb->lno_in_postimage++;
 962                         emit_add_line(reset, ecb, data, len);
 963                 }
 964                 size -= len;
 965                 data += len;
 966         }
 967         if (!endp)
 968                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
 969 }
 970
 971 static void emit_rewrite_diff(const char *name_a,
 972                               const char *name_b,
 973                               struct diff_filespec *one,
 974                               struct diff_filespec *two,
 975                               struct userdiff_driver *textconv_one,
 976                               struct userdiff_driver *textconv_two,
 977                               struct diff_options *o)
 978 {
 979         int lc_a, lc_b;
 980         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 981         const char *a_prefix, *b_prefix;
 982         char *data_one, *data_two;
 983         size_t size_one, size_two;
 984         struct emit_callback ecbdata;
 985         struct strbuf out = STRBUF_INIT;
 986
 987         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 988                 a_prefix = o->b_prefix;
 989                 b_prefix = o->a_prefix;
 990         } else {
 991                 a_prefix = o->a_prefix;
 992                 b_prefix = o->b_prefix;
 993         }
 994
 995         name_a += (*name_a == '/');
 996         name_b += (*name_b == '/');
 997
 998         strbuf_reset(&a_name);
 999         strbuf_reset(&b_name);
1000         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1001         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1002
1003         size_one = fill_textconv(textconv_one, one, &data_one);
1004         size_two = fill_textconv(textconv_two, two, &data_two);
1005
1006         memset(&ecbdata, 0, sizeof(ecbdata));
1007         ecbdata.color_diff = want_color(o->use_color);
1008         ecbdata.ws_rule = whitespace_rule(name_b);
1009         ecbdata.opt = o;
1010         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1011                 mmfile_t mf1, mf2;
1012                 mf1.ptr = (char *)data_one;
1013                 mf2.ptr = (char *)data_two;
1014                 mf1.size = size_one;
1015                 mf2.size = size_two;
1016                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1017         }
1018         ecbdata.lno_in_preimage = 1;
1019         ecbdata.lno_in_postimage = 1;
1020
1021         lc_a = count_lines(data_one, size_one);
1022         lc_b = count_lines(data_two, size_two);
1023
1024         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1025                          a_name.buf, a_name.len, 0);
1026         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1027                          b_name.buf, b_name.len, 0);
1028
1029         strbuf_addstr(&out, "@@ -");
1030         if (!o->irreversible_delete)
1031                 add_line_count(&out, lc_a);
1032         else
1033                 strbuf_addstr(&out, "?,?");
1034         strbuf_addstr(&out, " +");
1035         add_line_count(&out, lc_b);
1036         strbuf_addstr(&out, " @@\n");
1037         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1038         strbuf_release(&out);
1039
1040         if (lc_a && !o->irreversible_delete)
1041                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1042         if (lc_b)
1043                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1044         if (textconv_one)
1045                 free((char *)data_one);
1046         if (textconv_two)
1047                 free((char *)data_two);
1048 }
1049
1050 struct diff_words_buffer {
1051         mmfile_t text;
1052         long alloc;
1053         struct diff_words_orig {
1054                 const char *begin, *end;
1055         } *orig;
1056         int orig_nr, orig_alloc;
1057 };
1058
1059 static void diff_words_append(char *line, unsigned long len,
1060                 struct diff_words_buffer *buffer)
1061 {
1062         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1063         line++;
1064         len--;
1065         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1066         buffer->text.size += len;
1067         buffer->text.ptr[buffer->text.size] = '\0';
1068 }
1069
1070 struct diff_words_style_elem {
1071         const char *prefix;
1072         const char *suffix;
1073         const char *color; /* NULL; filled in by the setup code if
1074                             * color is enabled */
1075 };
1076
1077 struct diff_words_style {
1078         enum diff_words_type type;
1079         struct diff_words_style_elem new, old, ctx;
1080         const char *newline;
1081 };
1082
1083 static struct diff_words_style diff_words_styles[] = {
1084         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1085         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1086         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1087 };
1088
1089 struct diff_words_data {
1090         struct diff_words_buffer minus, plus;
1091         const char *current_plus;
1092         int last_minus;
1093         struct diff_options *opt;
1094         regex_t *word_regex;
1095         enum diff_words_type type;
1096         struct diff_words_style *style;
1097 };
1098
1099 static int fn_out_diff_words_write_helper(struct diff_options *o,
1100                                           struct diff_words_style_elem *st_el,
1101                                           const char *newline,
1102                                           size_t count, const char *buf)
1103 {
1104         int print = 0;
1105         struct strbuf sb = STRBUF_INIT;
1106
1107         while (count) {
1108                 char *p = memchr(buf, '\n', count);
1109                 if (print)
1110                         strbuf_addstr(&sb, diff_line_prefix(o));
1111
1112                 if (p != buf) {
1113                         const char *reset = st_el->color && *st_el->color ?
1114                                             GIT_COLOR_RESET : NULL;
1115                         if (st_el->color && *st_el->color)
1116                                 strbuf_addstr(&sb, st_el->color);
1117                         strbuf_addstr(&sb, st_el->prefix);
1118                         strbuf_add(&sb, buf, p ? p - buf : count);
1119                         strbuf_addstr(&sb, st_el->suffix);
1120                         if (reset)
1121                                 strbuf_addstr(&sb, reset);
1122                 }
1123                 if (!p)
1124                         goto out;
1125
1126                 strbuf_addstr(&sb, newline);
1127                 count -= p + 1 - buf;
1128                 buf = p + 1;
1129                 print = 1;
1130                 if (count) {
1131                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1132                                          sb.buf, sb.len, 0);
1133                         strbuf_reset(&sb);
1134                 }
1135         }
1136
1137 out:
1138         if (sb.len)
1139                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1140                                  sb.buf, sb.len, 0);
1141         strbuf_release(&sb);
1142         return 0;
1143 }
1144
1145 /*
1146  * '--color-words' algorithm can be described as:
1147  *
1148  *   1. collect the minus/plus lines of a diff hunk, divided into
1149  *      minus-lines and plus-lines;
1150  *
1151  *   2. break both minus-lines and plus-lines into words and
1152  *      place them into two mmfile_t with one word for each line;
1153  *
1154  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1155  *
1156  * And for the common parts of the both file, we output the plus side text.
1157  * diff_words->current_plus is used to trace the current position of the plus file
1158  * which printed. diff_words->last_minus is used to trace the last minus word
1159  * printed.
1160  *
1161  * For '--graph' to work with '--color-words', we need to output the graph prefix
1162  * on each line of color words output. Generally, there are two conditions on
1163  * which we should output the prefix.
1164  *
1165  *   1. diff_words->last_minus == 0 &&
1166  *      diff_words->current_plus == diff_words->plus.text.ptr
1167  *
1168  *      that is: the plus text must start as a new line, and if there is no minus
1169  *      word printed, a graph prefix must be printed.
1170  *
1171  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1172  *      *(diff_words->current_plus - 1) == '\n'
1173  *
1174  *      that is: a graph prefix must be printed following a '\n'
1175  */
1176 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1177 {
1178         if ((diff_words->last_minus == 0 &&
1179                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1180                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1181                 *(diff_words->current_plus - 1) == '\n')) {
1182                 return 1;
1183         } else {
1184                 return 0;
1185         }
1186 }
1187
1188 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1189 {
1190         struct diff_words_data *diff_words = priv;
1191         struct diff_words_style *style = diff_words->style;
1192         int minus_first, minus_len, plus_first, plus_len;
1193         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1194         struct diff_options *opt = diff_words->opt;
1195         const char *line_prefix;
1196
1197         if (line[0] != '@' || parse_hunk_header(line, len,
1198                         &minus_first, &minus_len, &plus_first, &plus_len))
1199                 return;
1200
1201         assert(opt);
1202         line_prefix = diff_line_prefix(opt);
1203
1204         /* POSIX requires that first be decremented by one if len == 0... */
1205         if (minus_len) {
1206                 minus_begin = diff_words->minus.orig[minus_first].begin;
1207                 minus_end =
1208                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1209         } else
1210                 minus_begin = minus_end =
1211                         diff_words->minus.orig[minus_first].end;
1212
1213         if (plus_len) {
1214                 plus_begin = diff_words->plus.orig[plus_first].begin;
1215                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1216         } else
1217                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1218
1219         if (color_words_output_graph_prefix(diff_words)) {
1220                 fputs(line_prefix, diff_words->opt->file);
1221         }
1222         if (diff_words->current_plus != plus_begin) {
1223                 fn_out_diff_words_write_helper(diff_words->opt,
1224                                 &style->ctx, style->newline,
1225                                 plus_begin - diff_words->current_plus,
1226                                 diff_words->current_plus);
1227         }
1228         if (minus_begin != minus_end) {
1229                 fn_out_diff_words_write_helper(diff_words->opt,
1230                                 &style->old, style->newline,
1231                                 minus_end - minus_begin, minus_begin);
1232         }
1233         if (plus_begin != plus_end) {
1234                 fn_out_diff_words_write_helper(diff_words->opt,
1235                                 &style->new, style->newline,
1236                                 plus_end - plus_begin, plus_begin);
1237         }
1238
1239         diff_words->current_plus = plus_end;
1240         diff_words->last_minus = minus_first;
1241 }
1242
1243 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1244 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1245                 int *begin, int *end)
1246 {
1247         if (word_regex && *begin < buffer->size) {
1248                 regmatch_t match[1];
1249                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1250                                  buffer->size - *begin, 1, match, 0)) {
1251                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1252                                         '\n', match[0].rm_eo - match[0].rm_so);
1253                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1254                         *begin += match[0].rm_so;
1255                         return *begin >= *end;
1256                 }
1257                 return -1;
1258         }
1259
1260         /* find the next word */
1261         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1262                 (*begin)++;
1263         if (*begin >= buffer->size)
1264                 return -1;
1265
1266         /* find the end of the word */
1267         *end = *begin + 1;
1268         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1269                 (*end)++;
1270
1271         return 0;
1272 }
1273
1274 /*
1275  * This function splits the words in buffer->text, stores the list with
1276  * newline separator into out, and saves the offsets of the original words
1277  * in buffer->orig.
1278  */
1279 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1280                 regex_t *word_regex)
1281 {
1282         int i, j;
1283         long alloc = 0;
1284
1285         out->size = 0;
1286         out->ptr = NULL;
1287
1288         /* fake an empty "0th" word */
1289         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1290         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1291         buffer->orig_nr = 1;
1292
1293         for (i = 0; i < buffer->text.size; i++) {
1294                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1295                         return;
1296
1297                 /* store original boundaries */
1298                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1299                                 buffer->orig_alloc);
1300                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1301                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1302                 buffer->orig_nr++;
1303
1304                 /* store one word */
1305                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1306                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1307                 out->ptr[out->size + j - i] = '\n';
1308                 out->size += j - i + 1;
1309
1310                 i = j - 1;
1311         }
1312 }
1313
1314 /* this executes the word diff on the accumulated buffers */
1315 static void diff_words_show(struct diff_words_data *diff_words)
1316 {
1317         xpparam_t xpp;
1318         xdemitconf_t xecfg;
1319         mmfile_t minus, plus;
1320         struct diff_words_style *style = diff_words->style;
1321
1322         struct diff_options *opt = diff_words->opt;
1323         const char *line_prefix;
1324
1325         assert(opt);
1326         line_prefix = diff_line_prefix(opt);
1327
1328         /* special case: only removal */
1329         if (!diff_words->plus.text.size) {
1330                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1331                                  line_prefix, strlen(line_prefix), 0);
1332                 fn_out_diff_words_write_helper(diff_words->opt,
1333                         &style->old, style->newline,
1334                         diff_words->minus.text.size,
1335                         diff_words->minus.text.ptr);
1336                 diff_words->minus.text.size = 0;
1337                 return;
1338         }
1339
1340         diff_words->current_plus = diff_words->plus.text.ptr;
1341         diff_words->last_minus = 0;
1342
1343         memset(&xpp, 0, sizeof(xpp));
1344         memset(&xecfg, 0, sizeof(xecfg));
1345         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1346         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1347         xpp.flags = 0;
1348         /* as only the hunk header will be parsed, we need a 0-context */
1349         xecfg.ctxlen = 0;
1350         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1351                           &xpp, &xecfg))
1352                 die("unable to generate word diff");
1353         free(minus.ptr);
1354         free(plus.ptr);
1355         if (diff_words->current_plus != diff_words->plus.text.ptr +
1356                         diff_words->plus.text.size) {
1357                 if (color_words_output_graph_prefix(diff_words))
1358                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1359                                          line_prefix, strlen(line_prefix), 0);
1360                 fn_out_diff_words_write_helper(diff_words->opt,
1361                         &style->ctx, style->newline,
1362                         diff_words->plus.text.ptr + diff_words->plus.text.size
1363                         - diff_words->current_plus, diff_words->current_plus);
1364         }
1365         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1366 }
1367
1368 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1369 static void diff_words_flush(struct emit_callback *ecbdata)
1370 {
1371         if (ecbdata->diff_words->minus.text.size ||
1372             ecbdata->diff_words->plus.text.size)
1373                 diff_words_show(ecbdata->diff_words);
1374 }
1375
1376 static void diff_filespec_load_driver(struct diff_filespec *one)
1377 {
1378         /* Use already-loaded driver */
1379         if (one->driver)
1380                 return;
1381
1382         if (S_ISREG(one->mode))
1383                 one->driver = userdiff_find_by_path(one->path);
1384
1385         /* Fallback to default settings */
1386         if (!one->driver)
1387                 one->driver = userdiff_find_by_name("default");
1388 }
1389
1390 static const char *userdiff_word_regex(struct diff_filespec *one)
1391 {
1392         diff_filespec_load_driver(one);
1393         return one->driver->word_regex;
1394 }
1395
1396 static void init_diff_words_data(struct emit_callback *ecbdata,
1397                                  struct diff_options *orig_opts,
1398                                  struct diff_filespec *one,
1399                                  struct diff_filespec *two)
1400 {
1401         int i;
1402         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1403         memcpy(o, orig_opts, sizeof(struct diff_options));
1404
1405         ecbdata->diff_words =
1406                 xcalloc(1, sizeof(struct diff_words_data));
1407         ecbdata->diff_words->type = o->word_diff;
1408         ecbdata->diff_words->opt = o;
1409         if (!o->word_regex)
1410                 o->word_regex = userdiff_word_regex(one);
1411         if (!o->word_regex)
1412                 o->word_regex = userdiff_word_regex(two);
1413         if (!o->word_regex)
1414                 o->word_regex = diff_word_regex_cfg;
1415         if (o->word_regex) {
1416                 ecbdata->diff_words->word_regex = (regex_t *)
1417                         xmalloc(sizeof(regex_t));
1418                 if (regcomp(ecbdata->diff_words->word_regex,
1419                             o->word_regex,
1420                             REG_EXTENDED | REG_NEWLINE))
1421                         die ("Invalid regular expression: %s",
1422                              o->word_regex);
1423         }
1424         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1425                 if (o->word_diff == diff_words_styles[i].type) {
1426                         ecbdata->diff_words->style =
1427                                 &diff_words_styles[i];
1428                         break;
1429                 }
1430         }
1431         if (want_color(o->use_color)) {
1432                 struct diff_words_style *st = ecbdata->diff_words->style;
1433                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1434                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1435                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1436         }
1437 }
1438
1439 static void free_diff_words_data(struct emit_callback *ecbdata)
1440 {
1441         if (ecbdata->diff_words) {
1442                 diff_words_flush(ecbdata);
1443                 free (ecbdata->diff_words->opt);
1444                 free (ecbdata->diff_words->minus.text.ptr);
1445                 free (ecbdata->diff_words->minus.orig);
1446                 free (ecbdata->diff_words->plus.text.ptr);
1447                 free (ecbdata->diff_words->plus.orig);
1448                 if (ecbdata->diff_words->word_regex) {
1449                         regfree(ecbdata->diff_words->word_regex);
1450                         free(ecbdata->diff_words->word_regex);
1451                 }
1452                 FREE_AND_NULL(ecbdata->diff_words);
1453         }
1454 }
1455
1456 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1457 {
1458         if (want_color(diff_use_color))
1459                 return diff_colors[ix];
1460         return "";
1461 }
1462
1463 const char *diff_line_prefix(struct diff_options *opt)
1464 {
1465         struct strbuf *msgbuf;
1466         if (!opt->output_prefix)
1467                 return "";
1468
1469         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1470         return msgbuf->buf;
1471 }
1472
1473 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1474 {
1475         const char *cp;
1476         unsigned long allot;
1477         size_t l = len;
1478
1479         if (ecb->truncate)
1480                 return ecb->truncate(line, len);
1481         cp = line;
1482         allot = l;
1483         while (0 < l) {
1484                 (void) utf8_width(&cp, &l);
1485                 if (!cp)
1486                         break; /* truncated in the middle? */
1487         }
1488         return allot - l;
1489 }
1490
1491 static void find_lno(const char *line, struct emit_callback *ecbdata)
1492 {
1493         const char *p;
1494         ecbdata->lno_in_preimage = 0;
1495         ecbdata->lno_in_postimage = 0;
1496         p = strchr(line, '-');
1497         if (!p)
1498                 return; /* cannot happen */
1499         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1500         p = strchr(p, '+');
1501         if (!p)
1502                 return; /* cannot happen */
1503         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1504 }
1505
1506 static void fn_out_consume(void *priv, char *line, unsigned long len)
1507 {
1508         struct emit_callback *ecbdata = priv;
1509         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1510         struct diff_options *o = ecbdata->opt;
1511
1512         o->found_changes = 1;
1513
1514         if (ecbdata->header) {
1515                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1516                                  ecbdata->header->buf, ecbdata->header->len, 0);
1517                 strbuf_reset(ecbdata->header);
1518                 ecbdata->header = NULL;
1519         }
1520
1521         if (ecbdata->label_path[0]) {
1522                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1523                                  ecbdata->label_path[0],
1524                                  strlen(ecbdata->label_path[0]), 0);
1525                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1526                                  ecbdata->label_path[1],
1527                                  strlen(ecbdata->label_path[1]), 0);
1528                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1529         }
1530
1531         if (diff_suppress_blank_empty
1532             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1533                 line[0] = '\n';
1534                 len = 1;
1535         }
1536
1537         if (line[0] == '@') {
1538                 if (ecbdata->diff_words)
1539                         diff_words_flush(ecbdata);
1540                 len = sane_truncate_line(ecbdata, line, len);
1541                 find_lno(line, ecbdata);
1542                 emit_hunk_header(ecbdata, line, len);
1543                 return;
1544         }
1545
1546         if (ecbdata->diff_words) {
1547                 enum diff_symbol s =
1548                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
1549                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
1550                 if (line[0] == '-') {
1551                         diff_words_append(line, len,
1552                                           &ecbdata->diff_words->minus);
1553                         return;
1554                 } else if (line[0] == '+') {
1555                         diff_words_append(line, len,
1556                                           &ecbdata->diff_words->plus);
1557                         return;
1558                 } else if (starts_with(line, "\\ ")) {
1559                         /*
1560                          * Eat the "no newline at eof" marker as if we
1561                          * saw a "+" or "-" line with nothing on it,
1562                          * and return without diff_words_flush() to
1563                          * defer processing. If this is the end of
1564                          * preimage, more "+" lines may come after it.
1565                          */
1566                         return;
1567                 }
1568                 diff_words_flush(ecbdata);
1569                 emit_diff_symbol(o, s, line, len, 0);
1570                 return;
1571         }
1572
1573         switch (line[0]) {
1574         case '+':
1575                 ecbdata->lno_in_postimage++;
1576                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1577                 break;
1578         case '-':
1579                 ecbdata->lno_in_preimage++;
1580                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1581                 break;
1582         case ' ':
1583                 ecbdata->lno_in_postimage++;
1584                 ecbdata->lno_in_preimage++;
1585                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1586                 break;
1587         default:
1588                 /* incomplete line at the end */
1589                 ecbdata->lno_in_preimage++;
1590                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
1591                                  line, len, 0);
1592                 break;
1593         }
1594 }
1595
1596 static char *pprint_rename(const char *a, const char *b)
1597 {
1598         const char *old = a;
1599         const char *new = b;
1600         struct strbuf name = STRBUF_INIT;
1601         int pfx_length, sfx_length;
1602         int pfx_adjust_for_slash;
1603         int len_a = strlen(a);
1604         int len_b = strlen(b);
1605         int a_midlen, b_midlen;
1606         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1607         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1608
1609         if (qlen_a || qlen_b) {
1610                 quote_c_style(a, &name, NULL, 0);
1611                 strbuf_addstr(&name, " => ");
1612                 quote_c_style(b, &name, NULL, 0);
1613                 return strbuf_detach(&name, NULL);
1614         }
1615
1616         /* Find common prefix */
1617         pfx_length = 0;
1618         while (*old && *new && *old == *new) {
1619                 if (*old == '/')
1620                         pfx_length = old - a + 1;
1621                 old++;
1622                 new++;
1623         }
1624
1625         /* Find common suffix */
1626         old = a + len_a;
1627         new = b + len_b;
1628         sfx_length = 0;
1629         /*
1630          * If there is a common prefix, it must end in a slash.  In
1631          * that case we let this loop run 1 into the prefix to see the
1632          * same slash.
1633          *
1634          * If there is no common prefix, we cannot do this as it would
1635          * underrun the input strings.
1636          */
1637         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1638         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1639                b + pfx_length - pfx_adjust_for_slash <= new &&
1640                *old == *new) {
1641                 if (*old == '/')
1642                         sfx_length = len_a - (old - a);
1643                 old--;
1644                 new--;
1645         }
1646
1647         /*
1648          * pfx{mid-a => mid-b}sfx
1649          * {pfx-a => pfx-b}sfx
1650          * pfx{sfx-a => sfx-b}
1651          * name-a => name-b
1652          */
1653         a_midlen = len_a - pfx_length - sfx_length;
1654         b_midlen = len_b - pfx_length - sfx_length;
1655         if (a_midlen < 0)
1656                 a_midlen = 0;
1657         if (b_midlen < 0)
1658                 b_midlen = 0;
1659
1660         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1661         if (pfx_length + sfx_length) {
1662                 strbuf_add(&name, a, pfx_length);
1663                 strbuf_addch(&name, '{');
1664         }
1665         strbuf_add(&name, a + pfx_length, a_midlen);
1666         strbuf_addstr(&name, " => ");
1667         strbuf_add(&name, b + pfx_length, b_midlen);
1668         if (pfx_length + sfx_length) {
1669                 strbuf_addch(&name, '}');
1670                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1671         }
1672         return strbuf_detach(&name, NULL);
1673 }
1674
1675 struct diffstat_t {
1676         int nr;
1677         int alloc;
1678         struct diffstat_file {
1679                 char *from_name;
1680                 char *name;
1681                 char *print_name;
1682                 unsigned is_unmerged:1;
1683                 unsigned is_binary:1;
1684                 unsigned is_renamed:1;
1685                 unsigned is_interesting:1;
1686                 uintmax_t added, deleted;
1687         } **files;
1688 };
1689
1690 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1691                                           const char *name_a,
1692                                           const char *name_b)
1693 {
1694         struct diffstat_file *x;
1695         x = xcalloc(1, sizeof(*x));
1696         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1697         diffstat->files[diffstat->nr++] = x;
1698         if (name_b) {
1699                 x->from_name = xstrdup(name_a);
1700                 x->name = xstrdup(name_b);
1701                 x->is_renamed = 1;
1702         }
1703         else {
1704                 x->from_name = NULL;
1705                 x->name = xstrdup(name_a);
1706         }
1707         return x;
1708 }
1709
1710 static void diffstat_consume(void *priv, char *line, unsigned long len)
1711 {
1712         struct diffstat_t *diffstat = priv;
1713         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1714
1715         if (line[0] == '+')
1716                 x->added++;
1717         else if (line[0] == '-')
1718                 x->deleted++;
1719 }
1720
1721 const char mime_boundary_leader[] = "------------";
1722
1723 static int scale_linear(int it, int width, int max_change)
1724 {
1725         if (!it)
1726                 return 0;
1727         /*
1728          * make sure that at least one '-' or '+' is printed if
1729          * there is any change to this path. The easiest way is to
1730          * scale linearly as if the alloted width is one column shorter
1731          * than it is, and then add 1 to the result.
1732          */
1733         return 1 + (it * (width - 1) / max_change);
1734 }
1735
1736 static void show_graph(struct strbuf *out, char ch, int cnt,
1737                        const char *set, const char *reset)
1738 {
1739         if (cnt <= 0)
1740                 return;
1741         strbuf_addstr(out, set);
1742         strbuf_addchars(out, ch, cnt);
1743         strbuf_addstr(out, reset);
1744 }
1745
1746 static void fill_print_name(struct diffstat_file *file)
1747 {
1748         char *pname;
1749
1750         if (file->print_name)
1751                 return;
1752
1753         if (!file->is_renamed) {
1754                 struct strbuf buf = STRBUF_INIT;
1755                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1756                         pname = strbuf_detach(&buf, NULL);
1757                 } else {
1758                         pname = file->name;
1759                         strbuf_release(&buf);
1760                 }
1761         } else {
1762                 pname = pprint_rename(file->from_name, file->name);
1763         }
1764         file->print_name = pname;
1765 }
1766
1767 static void print_stat_summary_inserts_deletes(struct diff_options *options,
1768                 int files, int insertions, int deletions)
1769 {
1770         struct strbuf sb = STRBUF_INIT;
1771
1772         if (!files) {
1773                 assert(insertions == 0 && deletions == 0);
1774                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
1775                                  NULL, 0, 0);
1776                 return;
1777         }
1778
1779         strbuf_addf(&sb,
1780                     (files == 1) ? " %d file changed" : " %d files changed",
1781                     files);
1782
1783         /*
1784          * For binary diff, the caller may want to print "x files
1785          * changed" with insertions == 0 && deletions == 0.
1786          *
1787          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1788          * is probably less confusing (i.e skip over "2 files changed
1789          * but nothing about added/removed lines? Is this a bug in Git?").
1790          */
1791         if (insertions || deletions == 0) {
1792                 strbuf_addf(&sb,
1793                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1794                             insertions);
1795         }
1796
1797         if (deletions || insertions == 0) {
1798                 strbuf_addf(&sb,
1799                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1800                             deletions);
1801         }
1802         strbuf_addch(&sb, '\n');
1803         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
1804                          sb.buf, sb.len, 0);
1805         strbuf_release(&sb);
1806 }
1807
1808 void print_stat_summary(FILE *fp, int files,
1809                         int insertions, int deletions)
1810 {
1811         struct diff_options o;
1812         memset(&o, 0, sizeof(o));
1813         o.file = fp;
1814
1815         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
1816 }
1817
1818 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1819 {
1820         int i, len, add, del, adds = 0, dels = 0;
1821         uintmax_t max_change = 0, max_len = 0;
1822         int total_files = data->nr, count;
1823         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1824         const char *reset, *add_c, *del_c;
1825         int extra_shown = 0;
1826         const char *line_prefix = diff_line_prefix(options);
1827         struct strbuf out = STRBUF_INIT;
1828
1829         if (data->nr == 0)
1830                 return;
1831
1832         count = options->stat_count ? options->stat_count : data->nr;
1833
1834         reset = diff_get_color_opt(options, DIFF_RESET);
1835         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1836         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1837
1838         /*
1839          * Find the longest filename and max number of changes
1840          */
1841         for (i = 0; (i < count) && (i < data->nr); i++) {
1842                 struct diffstat_file *file = data->files[i];
1843                 uintmax_t change = file->added + file->deleted;
1844
1845                 if (!file->is_interesting && (change == 0)) {
1846                         count++; /* not shown == room for one more */
1847                         continue;
1848                 }
1849                 fill_print_name(file);
1850                 len = strlen(file->print_name);
1851                 if (max_len < len)
1852                         max_len = len;
1853
1854                 if (file->is_unmerged) {
1855                         /* "Unmerged" is 8 characters */
1856                         bin_width = bin_width < 8 ? 8 : bin_width;
1857                         continue;
1858                 }
1859                 if (file->is_binary) {
1860                         /* "Bin XXX -> YYY bytes" */
1861                         int w = 14 + decimal_width(file->added)
1862                                 + decimal_width(file->deleted);
1863                         bin_width = bin_width < w ? w : bin_width;
1864                         /* Display change counts aligned with "Bin" */
1865                         number_width = 3;
1866                         continue;
1867                 }
1868
1869                 if (max_change < change)
1870                         max_change = change;
1871         }
1872         count = i; /* where we can stop scanning in data->files[] */
1873
1874         /*
1875          * We have width = stat_width or term_columns() columns total.
1876          * We want a maximum of min(max_len, stat_name_width) for the name part.
1877          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1878          * We also need 1 for " " and 4 + decimal_width(max_change)
1879          * for " | NNNN " and one the empty column at the end, altogether
1880          * 6 + decimal_width(max_change).
1881          *
1882          * If there's not enough space, we will use the smaller of
1883          * stat_name_width (if set) and 5/8*width for the filename,
1884          * and the rest for constant elements + graph part, but no more
1885          * than stat_graph_width for the graph part.
1886          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1887          * for the standard terminal size).
1888          *
1889          * In other words: stat_width limits the maximum width, and
1890          * stat_name_width fixes the maximum width of the filename,
1891          * and is also used to divide available columns if there
1892          * aren't enough.
1893          *
1894          * Binary files are displayed with "Bin XXX -> YYY bytes"
1895          * instead of the change count and graph. This part is treated
1896          * similarly to the graph part, except that it is not
1897          * "scaled". If total width is too small to accommodate the
1898          * guaranteed minimum width of the filename part and the
1899          * separators and this message, this message will "overflow"
1900          * making the line longer than the maximum width.
1901          */
1902
1903         if (options->stat_width == -1)
1904                 width = term_columns() - strlen(line_prefix);
1905         else
1906                 width = options->stat_width ? options->stat_width : 80;
1907         number_width = decimal_width(max_change) > number_width ?
1908                 decimal_width(max_change) : number_width;
1909
1910         if (options->stat_graph_width == -1)
1911                 options->stat_graph_width = diff_stat_graph_width;
1912
1913         /*
1914          * Guarantee 3/8*16==6 for the graph part
1915          * and 5/8*16==10 for the filename part
1916          */
1917         if (width < 16 + 6 + number_width)
1918                 width = 16 + 6 + number_width;
1919
1920         /*
1921          * First assign sizes that are wanted, ignoring available width.
1922          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1923          * starting from "XXX" should fit in graph_width.
1924          */
1925         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1926         if (options->stat_graph_width &&
1927             options->stat_graph_width < graph_width)
1928                 graph_width = options->stat_graph_width;
1929
1930         name_width = (options->stat_name_width > 0 &&
1931                       options->stat_name_width < max_len) ?
1932                 options->stat_name_width : max_len;
1933
1934         /*
1935          * Adjust adjustable widths not to exceed maximum width
1936          */
1937         if (name_width + number_width + 6 + graph_width > width) {
1938                 if (graph_width > width * 3/8 - number_width - 6) {
1939                         graph_width = width * 3/8 - number_width - 6;
1940                         if (graph_width < 6)
1941                                 graph_width = 6;
1942                 }
1943
1944                 if (options->stat_graph_width &&
1945                     graph_width > options->stat_graph_width)
1946                         graph_width = options->stat_graph_width;
1947                 if (name_width > width - number_width - 6 - graph_width)
1948                         name_width = width - number_width - 6 - graph_width;
1949                 else
1950                         graph_width = width - number_width - 6 - name_width;
1951         }
1952
1953         /*
1954          * From here name_width is the width of the name area,
1955          * and graph_width is the width of the graph area.
1956          * max_change is used to scale graph properly.
1957          */
1958         for (i = 0; i < count; i++) {
1959                 const char *prefix = "";
1960                 struct diffstat_file *file = data->files[i];
1961                 char *name = file->print_name;
1962                 uintmax_t added = file->added;
1963                 uintmax_t deleted = file->deleted;
1964                 int name_len;
1965
1966                 if (!file->is_interesting && (added + deleted == 0))
1967                         continue;
1968
1969                 /*
1970                  * "scale" the filename
1971                  */
1972                 len = name_width;
1973                 name_len = strlen(name);
1974                 if (name_width < name_len) {
1975                         char *slash;
1976                         prefix = "...";
1977                         len -= 3;
1978                         name += name_len - len;
1979                         slash = strchr(name, '/');
1980                         if (slash)
1981                                 name = slash;
1982                 }
1983
1984                 if (file->is_binary) {
1985                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
1986                         strbuf_addf(&out, " %*s", number_width, "Bin");
1987                         if (!added && !deleted) {
1988                                 strbuf_addch(&out, '\n');
1989                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
1990                                                  out.buf, out.len, 0);
1991                                 strbuf_reset(&out);
1992                                 continue;
1993                         }
1994                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
1995                                 del_c, deleted, reset);
1996                         strbuf_addstr(&out, " -> ");
1997                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
1998                                 add_c, added, reset);
1999                         strbuf_addstr(&out, " bytes\n");
2000                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2001                                          out.buf, out.len, 0);
2002                         strbuf_reset(&out);
2003                         continue;
2004                 }
2005                 else if (file->is_unmerged) {
2006                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2007                         strbuf_addstr(&out, " Unmerged\n");
2008                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2009                                          out.buf, out.len, 0);
2010                         strbuf_reset(&out);
2011                         continue;
2012                 }
2013
2014                 /*
2015                  * scale the add/delete
2016                  */
2017                 add = added;
2018                 del = deleted;
2019
2020                 if (graph_width <= max_change) {
2021                         int total = scale_linear(add + del, graph_width, max_change);
2022                         if (total < 2 && add && del)
2023                                 /* width >= 2 due to the sanity check */
2024                                 total = 2;
2025                         if (add < del) {
2026                                 add = scale_linear(add, graph_width, max_change);
2027                                 del = total - add;
2028                         } else {
2029                                 del = scale_linear(del, graph_width, max_change);
2030                                 add = total - del;
2031                         }
2032                 }
2033                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2034                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2035                         number_width, added + deleted,
2036                         added + deleted ? " " : "");
2037                 show_graph(&out, '+', add, add_c, reset);
2038                 show_graph(&out, '-', del, del_c, reset);
2039                 strbuf_addch(&out, '\n');
2040                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2041                                  out.buf, out.len, 0);
2042                 strbuf_reset(&out);
2043         }
2044
2045         for (i = 0; i < data->nr; i++) {
2046                 struct diffstat_file *file = data->files[i];
2047                 uintmax_t added = file->added;
2048                 uintmax_t deleted = file->deleted;
2049
2050                 if (file->is_unmerged ||
2051                     (!file->is_interesting && (added + deleted == 0))) {
2052                         total_files--;
2053                         continue;
2054                 }
2055
2056                 if (!file->is_binary) {
2057                         adds += added;
2058                         dels += deleted;
2059                 }
2060                 if (i < count)
2061                         continue;
2062                 if (!extra_shown)
2063                         emit_diff_symbol(options,
2064                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2065                                          NULL, 0, 0);
2066                 extra_shown = 1;
2067         }
2068
2069         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2070 }
2071
2072 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2073 {
2074         int i, adds = 0, dels = 0, total_files = data->nr;
2075
2076         if (data->nr == 0)
2077                 return;
2078
2079         for (i = 0; i < data->nr; i++) {
2080                 int added = data->files[i]->added;
2081                 int deleted = data->files[i]->deleted;
2082
2083                 if (data->files[i]->is_unmerged ||
2084                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2085                         total_files--;
2086                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2087                         adds += added;
2088                         dels += deleted;
2089                 }
2090         }
2091         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2092 }
2093
2094 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2095 {
2096         int i;
2097
2098         if (data->nr == 0)
2099                 return;
2100
2101         for (i = 0; i < data->nr; i++) {
2102                 struct diffstat_file *file = data->files[i];
2103
2104                 fprintf(options->file, "%s", diff_line_prefix(options));
2105
2106                 if (file->is_binary)
2107                         fprintf(options->file, "-\t-\t");
2108                 else
2109                         fprintf(options->file,
2110                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2111                                 file->added, file->deleted);
2112                 if (options->line_termination) {
2113                         fill_print_name(file);
2114                         if (!file->is_renamed)
2115                                 write_name_quoted(file->name, options->file,
2116                                                   options->line_termination);
2117                         else {
2118                                 fputs(file->print_name, options->file);
2119                                 putc(options->line_termination, options->file);
2120                         }
2121                 } else {
2122                         if (file->is_renamed) {
2123                                 putc('\0', options->file);
2124                                 write_name_quoted(file->from_name, options->file, '\0');
2125                         }
2126                         write_name_quoted(file->name, options->file, '\0');
2127                 }
2128         }
2129 }
2130
2131 struct dirstat_file {
2132         const char *name;
2133         unsigned long changed;
2134 };
2135
2136 struct dirstat_dir {
2137         struct dirstat_file *files;
2138         int alloc, nr, permille, cumulative;
2139 };
2140
2141 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2142                 unsigned long changed, const char *base, int baselen)
2143 {
2144         unsigned long this_dir = 0;
2145         unsigned int sources = 0;
2146         const char *line_prefix = diff_line_prefix(opt);
2147
2148         while (dir->nr) {
2149                 struct dirstat_file *f = dir->files;
2150                 int namelen = strlen(f->name);
2151                 unsigned long this;
2152                 char *slash;
2153
2154                 if (namelen < baselen)
2155                         break;
2156                 if (memcmp(f->name, base, baselen))
2157                         break;
2158                 slash = strchr(f->name + baselen, '/');
2159                 if (slash) {
2160                         int newbaselen = slash + 1 - f->name;
2161                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2162                         sources++;
2163                 } else {
2164                         this = f->changed;
2165                         dir->files++;
2166                         dir->nr--;
2167                         sources += 2;
2168                 }
2169                 this_dir += this;
2170         }
2171
2172         /*
2173          * We don't report dirstat's for
2174          *  - the top level
2175          *  - or cases where everything came from a single directory
2176          *    under this directory (sources == 1).
2177          */
2178         if (baselen && sources != 1) {
2179                 if (this_dir) {
2180                         int permille = this_dir * 1000 / changed;
2181                         if (permille >= dir->permille) {
2182                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2183                                         permille / 10, permille % 10, baselen, base);
2184                                 if (!dir->cumulative)
2185                                         return 0;
2186                         }
2187                 }
2188         }
2189         return this_dir;
2190 }
2191
2192 static int dirstat_compare(const void *_a, const void *_b)
2193 {
2194         const struct dirstat_file *a = _a;
2195         const struct dirstat_file *b = _b;
2196         return strcmp(a->name, b->name);
2197 }
2198
2199 static void show_dirstat(struct diff_options *options)
2200 {
2201         int i;
2202         unsigned long changed;
2203         struct dirstat_dir dir;
2204         struct diff_queue_struct *q = &diff_queued_diff;
2205
2206         dir.files = NULL;
2207         dir.alloc = 0;
2208         dir.nr = 0;
2209         dir.permille = options->dirstat_permille;
2210         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2211
2212         changed = 0;
2213         for (i = 0; i < q->nr; i++) {
2214                 struct diff_filepair *p = q->queue[i];
2215                 const char *name;
2216                 unsigned long copied, added, damage;
2217                 int content_changed;
2218
2219                 name = p->two->path ? p->two->path : p->one->path;
2220
2221                 if (p->one->oid_valid && p->two->oid_valid)
2222                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2223                 else
2224                         content_changed = 1;
2225
2226                 if (!content_changed) {
2227                         /*
2228                          * The SHA1 has not changed, so pre-/post-content is
2229                          * identical. We can therefore skip looking at the
2230                          * file contents altogether.
2231                          */
2232                         damage = 0;
2233                         goto found_damage;
2234                 }
2235
2236                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2237                         /*
2238                          * In --dirstat-by-file mode, we don't really need to
2239                          * look at the actual file contents at all.
2240                          * The fact that the SHA1 changed is enough for us to
2241                          * add this file to the list of results
2242                          * (with each file contributing equal damage).
2243                          */
2244                         damage = 1;
2245                         goto found_damage;
2246                 }
2247
2248                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2249                         diff_populate_filespec(p->one, 0);
2250                         diff_populate_filespec(p->two, 0);
2251                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2252                                                &copied, &added);
2253                         diff_free_filespec_data(p->one);
2254                         diff_free_filespec_data(p->two);
2255                 } else if (DIFF_FILE_VALID(p->one)) {
2256                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2257                         copied = added = 0;
2258                         diff_free_filespec_data(p->one);
2259                 } else if (DIFF_FILE_VALID(p->two)) {
2260                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2261                         copied = 0;
2262                         added = p->two->size;
2263                         diff_free_filespec_data(p->two);
2264                 } else
2265                         continue;
2266
2267                 /*
2268                  * Original minus copied is the removed material,
2269                  * added is the new material.  They are both damages
2270                  * made to the preimage.
2271                  * If the resulting damage is zero, we know that
2272                  * diffcore_count_changes() considers the two entries to
2273                  * be identical, but since content_changed is true, we
2274                  * know that there must have been _some_ kind of change,
2275                  * so we force all entries to have damage > 0.
2276                  */
2277                 damage = (p->one->size - copied) + added;
2278                 if (!damage)
2279                         damage = 1;
2280
2281 found_damage:
2282                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2283                 dir.files[dir.nr].name = name;
2284                 dir.files[dir.nr].changed = damage;
2285                 changed += damage;
2286                 dir.nr++;
2287         }
2288
2289         /* This can happen even with many files, if everything was renames */
2290         if (!changed)
2291                 return;
2292
2293         /* Show all directories with more than x% of the changes */
2294         QSORT(dir.files, dir.nr, dirstat_compare);
2295         gather_dirstat(options, &dir, changed, "", 0);
2296 }
2297
2298 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2299 {
2300         int i;
2301         unsigned long changed;
2302         struct dirstat_dir dir;
2303
2304         if (data->nr == 0)
2305                 return;
2306
2307         dir.files = NULL;
2308         dir.alloc = 0;
2309         dir.nr = 0;
2310         dir.permille = options->dirstat_permille;
2311         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2312
2313         changed = 0;
2314         for (i = 0; i < data->nr; i++) {
2315                 struct diffstat_file *file = data->files[i];
2316                 unsigned long damage = file->added + file->deleted;
2317                 if (file->is_binary)
2318                         /*
2319                          * binary files counts bytes, not lines. Must find some
2320                          * way to normalize binary bytes vs. textual lines.
2321                          * The following heuristic assumes that there are 64
2322                          * bytes per "line".
2323                          * This is stupid and ugly, but very cheap...
2324                          */
2325                         damage = (damage + 63) / 64;
2326                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2327                 dir.files[dir.nr].name = file->name;
2328                 dir.files[dir.nr].changed = damage;
2329                 changed += damage;
2330                 dir.nr++;
2331         }
2332
2333         /* This can happen even with many files, if everything was renames */
2334         if (!changed)
2335                 return;
2336
2337         /* Show all directories with more than x% of the changes */
2338         QSORT(dir.files, dir.nr, dirstat_compare);
2339         gather_dirstat(options, &dir, changed, "", 0);
2340 }
2341
2342 static void free_diffstat_info(struct diffstat_t *diffstat)
2343 {
2344         int i;
2345         for (i = 0; i < diffstat->nr; i++) {
2346                 struct diffstat_file *f = diffstat->files[i];
2347                 if (f->name != f->print_name)
2348                         free(f->print_name);
2349                 free(f->name);
2350                 free(f->from_name);
2351                 free(f);
2352         }
2353         free(diffstat->files);
2354 }
2355
2356 struct checkdiff_t {
2357         const char *filename;
2358         int lineno;
2359         int conflict_marker_size;
2360         struct diff_options *o;
2361         unsigned ws_rule;
2362         unsigned status;
2363 };
2364
2365 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2366 {
2367         char firstchar;
2368         int cnt;
2369
2370         if (len < marker_size + 1)
2371                 return 0;
2372         firstchar = line[0];
2373         switch (firstchar) {
2374         case '=': case '>': case '<': case '|':
2375                 break;
2376         default:
2377                 return 0;
2378         }
2379         for (cnt = 1; cnt < marker_size; cnt++)
2380                 if (line[cnt] != firstchar)
2381                         return 0;
2382         /* line[1] thru line[marker_size-1] are same as firstchar */
2383         if (len < marker_size + 1 || !isspace(line[marker_size]))
2384                 return 0;
2385         return 1;
2386 }
2387
2388 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2389 {
2390         struct checkdiff_t *data = priv;
2391         int marker_size = data->conflict_marker_size;
2392         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2393         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2394         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2395         char *err;
2396         const char *line_prefix;
2397
2398         assert(data->o);
2399         line_prefix = diff_line_prefix(data->o);
2400
2401         if (line[0] == '+') {
2402                 unsigned bad;
2403                 data->lineno++;
2404                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2405                         data->status |= 1;
2406                         fprintf(data->o->file,
2407                                 "%s%s:%d: leftover conflict marker\n",
2408                                 line_prefix, data->filename, data->lineno);
2409                 }
2410                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2411                 if (!bad)
2412                         return;
2413                 data->status |= bad;
2414                 err = whitespace_error_string(bad);
2415                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2416                         line_prefix, data->filename, data->lineno, err);
2417                 free(err);
2418                 emit_line(data->o, set, reset, line, 1);
2419                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2420                               data->o->file, set, reset, ws);
2421         } else if (line[0] == ' ') {
2422                 data->lineno++;
2423         } else if (line[0] == '@') {
2424                 char *plus = strchr(line, '+');
2425                 if (plus)
2426                         data->lineno = strtol(plus, NULL, 10) - 1;
2427                 else
2428                         die("invalid diff");
2429         }
2430 }
2431
2432 static unsigned char *deflate_it(char *data,
2433                                  unsigned long size,
2434                                  unsigned long *result_size)
2435 {
2436         int bound;
2437         unsigned char *deflated;
2438         git_zstream stream;
2439
2440         git_deflate_init(&stream, zlib_compression_level);
2441         bound = git_deflate_bound(&stream, size);
2442         deflated = xmalloc(bound);
2443         stream.next_out = deflated;
2444         stream.avail_out = bound;
2445
2446         stream.next_in = (unsigned char *)data;
2447         stream.avail_in = size;
2448         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2449                 ; /* nothing */
2450         git_deflate_end(&stream);
2451         *result_size = stream.total_out;
2452         return deflated;
2453 }
2454
2455 static void emit_binary_diff_body(struct diff_options *o,
2456                                   mmfile_t *one, mmfile_t *two)
2457 {
2458         void *cp;
2459         void *delta;
2460         void *deflated;
2461         void *data;
2462         unsigned long orig_size;
2463         unsigned long delta_size;
2464         unsigned long deflate_size;
2465         unsigned long data_size;
2466
2467         /* We could do deflated delta, or we could do just deflated two,
2468          * whichever is smaller.
2469          */
2470         delta = NULL;
2471         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2472         if (one->size && two->size) {
2473                 delta = diff_delta(one->ptr, one->size,
2474                                    two->ptr, two->size,
2475                                    &delta_size, deflate_size);
2476                 if (delta) {
2477                         void *to_free = delta;
2478                         orig_size = delta_size;
2479                         delta = deflate_it(delta, delta_size, &delta_size);
2480                         free(to_free);
2481                 }
2482         }
2483
2484         if (delta && delta_size < deflate_size) {
2485                 char *s = xstrfmt("%lu", orig_size);
2486                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2487                                  s, strlen(s), 0);
2488                 free(s);
2489                 free(deflated);
2490                 data = delta;
2491                 data_size = delta_size;
2492         } else {
2493                 char *s = xstrfmt("%lu", two->size);
2494                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2495                                  s, strlen(s), 0);
2496                 free(s);
2497                 free(delta);
2498                 data = deflated;
2499                 data_size = deflate_size;
2500         }
2501
2502         /* emit data encoded in base85 */
2503         cp = data;
2504         while (data_size) {
2505                 int len;
2506                 int bytes = (52 < data_size) ? 52 : data_size;
2507                 char line[71];
2508                 data_size -= bytes;
2509                 if (bytes <= 26)
2510                         line[0] = bytes + 'A' - 1;
2511                 else
2512                         line[0] = bytes - 26 + 'a' - 1;
2513                 encode_85(line + 1, cp, bytes);
2514                 cp = (char *) cp + bytes;
2515
2516                 len = strlen(line);
2517                 line[len++] = '\n';
2518                 line[len] = '\0';
2519
2520                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2521                                  line, len, 0);
2522         }
2523         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2524         free(data);
2525 }
2526
2527 static void emit_binary_diff(struct diff_options *o,
2528                              mmfile_t *one, mmfile_t *two)
2529 {
2530         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2531         emit_binary_diff_body(o, one, two);
2532         emit_binary_diff_body(o, two, one);
2533 }
2534
2535 int diff_filespec_is_binary(struct diff_filespec *one)
2536 {
2537         if (one->is_binary == -1) {
2538                 diff_filespec_load_driver(one);
2539                 if (one->driver->binary != -1)
2540                         one->is_binary = one->driver->binary;
2541                 else {
2542                         if (!one->data && DIFF_FILE_VALID(one))
2543                                 diff_populate_filespec(one, CHECK_BINARY);
2544                         if (one->is_binary == -1 && one->data)
2545                                 one->is_binary = buffer_is_binary(one->data,
2546                                                 one->size);
2547                         if (one->is_binary == -1)
2548                                 one->is_binary = 0;
2549                 }
2550         }
2551         return one->is_binary;
2552 }
2553
2554 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2555 {
2556         diff_filespec_load_driver(one);
2557         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2558 }
2559
2560 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2561 {
2562         if (!options->a_prefix)
2563                 options->a_prefix = a;
2564         if (!options->b_prefix)
2565                 options->b_prefix = b;
2566 }
2567
2568 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2569 {
2570         if (!DIFF_FILE_VALID(one))
2571                 return NULL;
2572
2573         diff_filespec_load_driver(one);
2574         return userdiff_get_textconv(one->driver);
2575 }
2576
2577 static void builtin_diff(const char *name_a,
2578                          const char *name_b,
2579                          struct diff_filespec *one,
2580                          struct diff_filespec *two,
2581                          const char *xfrm_msg,
2582                          int must_show_header,
2583                          struct diff_options *o,
2584                          int complete_rewrite)
2585 {
2586         mmfile_t mf1, mf2;
2587         const char *lbl[2];
2588         char *a_one, *b_two;
2589         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2590         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2591         const char *a_prefix, *b_prefix;
2592         struct userdiff_driver *textconv_one = NULL;
2593         struct userdiff_driver *textconv_two = NULL;
2594         struct strbuf header = STRBUF_INIT;
2595         const char *line_prefix = diff_line_prefix(o);
2596
2597         diff_set_mnemonic_prefix(o, "a/", "b/");
2598         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2599                 a_prefix = o->b_prefix;
2600                 b_prefix = o->a_prefix;
2601         } else {
2602                 a_prefix = o->a_prefix;
2603                 b_prefix = o->b_prefix;
2604         }
2605
2606         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2607             (!one->mode || S_ISGITLINK(one->mode)) &&
2608             (!two->mode || S_ISGITLINK(two->mode))) {
2609                 show_submodule_summary(o, one->path ? one->path : two->path,
2610                                 &one->oid, &two->oid,
2611                                 two->dirty_submodule);
2612                 return;
2613         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2614                    (!one->mode || S_ISGITLINK(one->mode)) &&
2615                    (!two->mode || S_ISGITLINK(two->mode))) {
2616                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
2617                                 &one->oid, &two->oid,
2618                                 two->dirty_submodule);
2619                 return;
2620         }
2621
2622         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2623                 textconv_one = get_textconv(one);
2624                 textconv_two = get_textconv(two);
2625         }
2626
2627         /* Never use a non-valid filename anywhere if at all possible */
2628         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2629         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2630
2631         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2632         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2633         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2634         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2635         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2636         if (lbl[0][0] == '/') {
2637                 /* /dev/null */
2638                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2639                 if (xfrm_msg)
2640                         strbuf_addstr(&header, xfrm_msg);
2641                 must_show_header = 1;
2642         }
2643         else if (lbl[1][0] == '/') {
2644                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2645                 if (xfrm_msg)
2646                         strbuf_addstr(&header, xfrm_msg);
2647                 must_show_header = 1;
2648         }
2649         else {
2650                 if (one->mode != two->mode) {
2651                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2652                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2653                         must_show_header = 1;
2654                 }
2655                 if (xfrm_msg)
2656                         strbuf_addstr(&header, xfrm_msg);
2657
2658                 /*
2659                  * we do not run diff between different kind
2660                  * of objects.
2661                  */
2662                 if ((one->mode ^ two->mode) & S_IFMT)
2663                         goto free_ab_and_return;
2664                 if (complete_rewrite &&
2665                     (textconv_one || !diff_filespec_is_binary(one)) &&
2666                     (textconv_two || !diff_filespec_is_binary(two))) {
2667                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2668                                          header.buf, header.len, 0);
2669                         strbuf_reset(&header);
2670                         emit_rewrite_diff(name_a, name_b, one, two,
2671                                                 textconv_one, textconv_two, o);
2672                         o->found_changes = 1;
2673                         goto free_ab_and_return;
2674                 }
2675         }
2676
2677         if (o->irreversible_delete && lbl[1][0] == '/') {
2678                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
2679                                  header.len, 0);
2680                 strbuf_reset(&header);
2681                 goto free_ab_and_return;
2682         } else if (!DIFF_OPT_TST(o, TEXT) &&
2683             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2684               (!textconv_two && diff_filespec_is_binary(two)) )) {
2685                 struct strbuf sb = STRBUF_INIT;
2686                 if (!one->data && !two->data &&
2687                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2688                     !DIFF_OPT_TST(o, BINARY)) {
2689                         if (!oidcmp(&one->oid, &two->oid)) {
2690                                 if (must_show_header)
2691                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2692                                                          header.buf, header.len,
2693                                                          0);
2694                                 goto free_ab_and_return;
2695                         }
2696                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2697                                          header.buf, header.len, 0);
2698                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
2699                                     diff_line_prefix(o), lbl[0], lbl[1]);
2700                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
2701                                          sb.buf, sb.len, 0);
2702                         strbuf_release(&sb);
2703                         goto free_ab_and_return;
2704                 }
2705                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2706                         die("unable to read files to diff");
2707                 /* Quite common confusing case */
2708                 if (mf1.size == mf2.size &&
2709                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2710                         if (must_show_header)
2711                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2712                                                  header.buf, header.len, 0);
2713                         goto free_ab_and_return;
2714                 }
2715                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
2716                 strbuf_reset(&header);
2717                 if (DIFF_OPT_TST(o, BINARY))
2718                         emit_binary_diff(o, &mf1, &mf2);
2719                 else {
2720                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
2721                                     diff_line_prefix(o), lbl[0], lbl[1]);
2722                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
2723                                          sb.buf, sb.len, 0);
2724                         strbuf_release(&sb);
2725                 }
2726                 o->found_changes = 1;
2727         } else {
2728                 /* Crazy xdl interfaces.. */
2729                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2730                 const char *v;
2731                 xpparam_t xpp;
2732                 xdemitconf_t xecfg;
2733                 struct emit_callback ecbdata;
2734                 const struct userdiff_funcname *pe;
2735
2736                 if (must_show_header) {
2737                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2738                                          header.buf, header.len, 0);
2739                         strbuf_reset(&header);
2740                 }
2741
2742                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2743                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2744
2745                 pe = diff_funcname_pattern(one);
2746                 if (!pe)
2747                         pe = diff_funcname_pattern(two);
2748
2749                 memset(&xpp, 0, sizeof(xpp));
2750                 memset(&xecfg, 0, sizeof(xecfg));
2751                 memset(&ecbdata, 0, sizeof(ecbdata));
2752                 ecbdata.label_path = lbl;
2753                 ecbdata.color_diff = want_color(o->use_color);
2754                 ecbdata.ws_rule = whitespace_rule(name_b);
2755                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2756                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2757                 ecbdata.opt = o;
2758                 ecbdata.header = header.len ? &header : NULL;
2759                 xpp.flags = o->xdl_opts;
2760                 xecfg.ctxlen = o->context;
2761                 xecfg.interhunkctxlen = o->interhunkcontext;
2762                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2763                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2764                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2765                 if (pe)
2766                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2767                 if (!diffopts)
2768                         ;
2769                 else if (skip_prefix(diffopts, "--unified=", &v))
2770                         xecfg.ctxlen = strtoul(v, NULL, 10);
2771                 else if (skip_prefix(diffopts, "-u", &v))
2772                         xecfg.ctxlen = strtoul(v, NULL, 10);
2773                 if (o->word_diff)
2774                         init_diff_words_data(&ecbdata, o, one, two);
2775                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2776                                   &xpp, &xecfg))
2777                         die("unable to generate diff for %s", one->path);
2778                 if (o->word_diff)
2779                         free_diff_words_data(&ecbdata);
2780                 if (textconv_one)
2781                         free(mf1.ptr);
2782                 if (textconv_two)
2783                         free(mf2.ptr);
2784                 xdiff_clear_find_func(&xecfg);
2785         }
2786
2787  free_ab_and_return:
2788         strbuf_release(&header);
2789         diff_free_filespec_data(one);
2790         diff_free_filespec_data(two);
2791         free(a_one);
2792         free(b_two);
2793         return;
2794 }
2795
2796 static void builtin_diffstat(const char *name_a, const char *name_b,
2797                              struct diff_filespec *one,
2798                              struct diff_filespec *two,
2799                              struct diffstat_t *diffstat,
2800                              struct diff_options *o,
2801                              struct diff_filepair *p)
2802 {
2803         mmfile_t mf1, mf2;
2804         struct diffstat_file *data;
2805         int same_contents;
2806         int complete_rewrite = 0;
2807
2808         if (!DIFF_PAIR_UNMERGED(p)) {
2809                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2810                         complete_rewrite = 1;
2811         }
2812
2813         data = diffstat_add(diffstat, name_a, name_b);
2814         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2815
2816         if (!one || !two) {
2817                 data->is_unmerged = 1;
2818                 return;
2819         }
2820
2821         same_contents = !oidcmp(&one->oid, &two->oid);
2822
2823         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2824                 data->is_binary = 1;
2825                 if (same_contents) {
2826                         data->added = 0;
2827                         data->deleted = 0;
2828                 } else {
2829                         data->added = diff_filespec_size(two);
2830                         data->deleted = diff_filespec_size(one);
2831                 }
2832         }
2833
2834         else if (complete_rewrite) {
2835                 diff_populate_filespec(one, 0);
2836                 diff_populate_filespec(two, 0);
2837                 data->deleted = count_lines(one->data, one->size);
2838                 data->added = count_lines(two->data, two->size);
2839         }
2840
2841         else if (!same_contents) {
2842                 /* Crazy xdl interfaces.. */
2843                 xpparam_t xpp;
2844                 xdemitconf_t xecfg;
2845
2846                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2847                         die("unable to read files to diff");
2848
2849                 memset(&xpp, 0, sizeof(xpp));
2850                 memset(&xecfg, 0, sizeof(xecfg));
2851                 xpp.flags = o->xdl_opts;
2852                 xecfg.ctxlen = o->context;
2853                 xecfg.interhunkctxlen = o->interhunkcontext;
2854                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2855                                   &xpp, &xecfg))
2856                         die("unable to generate diffstat for %s", one->path);
2857         }
2858
2859         diff_free_filespec_data(one);
2860         diff_free_filespec_data(two);
2861 }
2862
2863 static void builtin_checkdiff(const char *name_a, const char *name_b,
2864                               const char *attr_path,
2865                               struct diff_filespec *one,
2866                               struct diff_filespec *two,
2867                               struct diff_options *o)
2868 {
2869         mmfile_t mf1, mf2;
2870         struct checkdiff_t data;
2871
2872         if (!two)
2873                 return;
2874
2875         memset(&data, 0, sizeof(data));
2876         data.filename = name_b ? name_b : name_a;
2877         data.lineno = 0;
2878         data.o = o;
2879         data.ws_rule = whitespace_rule(attr_path);
2880         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2881
2882         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2883                 die("unable to read files to diff");
2884
2885         /*
2886          * All the other codepaths check both sides, but not checking
2887          * the "old" side here is deliberate.  We are checking the newly
2888          * introduced changes, and as long as the "new" side is text, we
2889          * can and should check what it introduces.
2890          */
2891         if (diff_filespec_is_binary(two))
2892                 goto free_and_return;
2893         else {
2894                 /* Crazy xdl interfaces.. */
2895                 xpparam_t xpp;
2896                 xdemitconf_t xecfg;
2897
2898                 memset(&xpp, 0, sizeof(xpp));
2899                 memset(&xecfg, 0, sizeof(xecfg));
2900                 xecfg.ctxlen = 1; /* at least one context line */
2901                 xpp.flags = 0;
2902                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2903                                   &xpp, &xecfg))
2904                         die("unable to generate checkdiff for %s", one->path);
2905
2906                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2907                         struct emit_callback ecbdata;
2908                         int blank_at_eof;
2909
2910                         ecbdata.ws_rule = data.ws_rule;
2911                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2912                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2913
2914                         if (blank_at_eof) {
2915                                 static char *err;
2916                                 if (!err)
2917                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2918                                 fprintf(o->file, "%s:%d: %s.\n",
2919                                         data.filename, blank_at_eof, err);
2920                                 data.status = 1; /* report errors */
2921                         }
2922                 }
2923         }
2924  free_and_return:
2925         diff_free_filespec_data(one);
2926         diff_free_filespec_data(two);
2927         if (data.status)
2928                 DIFF_OPT_SET(o, CHECK_FAILED);
2929 }
2930
2931 struct diff_filespec *alloc_filespec(const char *path)
2932 {
2933         struct diff_filespec *spec;
2934
2935         FLEXPTR_ALLOC_STR(spec, path, path);
2936         spec->count = 1;
2937         spec->is_binary = -1;
2938         return spec;
2939 }
2940
2941 void free_filespec(struct diff_filespec *spec)
2942 {
2943         if (!--spec->count) {
2944                 diff_free_filespec_data(spec);
2945                 free(spec);
2946         }
2947 }
2948
2949 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
2950                    int oid_valid, unsigned short mode)
2951 {
2952         if (mode) {
2953                 spec->mode = canon_mode(mode);
2954                 oidcpy(&spec->oid, oid);
2955                 spec->oid_valid = oid_valid;
2956         }
2957 }
2958
2959 /*
2960  * Given a name and sha1 pair, if the index tells us the file in
2961  * the work tree has that object contents, return true, so that
2962  * prepare_temp_file() does not have to inflate and extract.
2963  */
2964 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
2965 {
2966         const struct cache_entry *ce;
2967         struct stat st;
2968         int pos, len;
2969
2970         /*
2971          * We do not read the cache ourselves here, because the
2972          * benchmark with my previous version that always reads cache
2973          * shows that it makes things worse for diff-tree comparing
2974          * two linux-2.6 kernel trees in an already checked out work
2975          * tree.  This is because most diff-tree comparisons deal with
2976          * only a small number of files, while reading the cache is
2977          * expensive for a large project, and its cost outweighs the
2978          * savings we get by not inflating the object to a temporary
2979          * file.  Practically, this code only helps when we are used
2980          * by diff-cache --cached, which does read the cache before
2981          * calling us.
2982          */
2983         if (!active_cache)
2984                 return 0;
2985
2986         /* We want to avoid the working directory if our caller
2987          * doesn't need the data in a normal file, this system
2988          * is rather slow with its stat/open/mmap/close syscalls,
2989          * and the object is contained in a pack file.  The pack
2990          * is probably already open and will be faster to obtain
2991          * the data through than the working directory.  Loose
2992          * objects however would tend to be slower as they need
2993          * to be individually opened and inflated.
2994          */
2995         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
2996                 return 0;
2997
2998         /*
2999          * Similarly, if we'd have to convert the file contents anyway, that
3000          * makes the optimization not worthwhile.
3001          */
3002         if (!want_file && would_convert_to_git(&the_index, name))
3003                 return 0;
3004
3005         len = strlen(name);
3006         pos = cache_name_pos(name, len);
3007         if (pos < 0)
3008                 return 0;
3009         ce = active_cache[pos];
3010
3011         /*
3012          * This is not the sha1 we are looking for, or
3013          * unreusable because it is not a regular file.
3014          */
3015         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3016                 return 0;
3017
3018         /*
3019          * If ce is marked as "assume unchanged", there is no
3020          * guarantee that work tree matches what we are looking for.
3021          */
3022         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3023                 return 0;
3024
3025         /*
3026          * If ce matches the file in the work tree, we can reuse it.
3027          */
3028         if (ce_uptodate(ce) ||
3029             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3030                 return 1;
3031
3032         return 0;
3033 }
3034
3035 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3036 {
3037         struct strbuf buf = STRBUF_INIT;
3038         char *dirty = "";
3039
3040         /* Are we looking at the work tree? */
3041         if (s->dirty_submodule)
3042                 dirty = "-dirty";
3043
3044         strbuf_addf(&buf, "Subproject commit %s%s\n",
3045                     oid_to_hex(&s->oid), dirty);
3046         s->size = buf.len;
3047         if (size_only) {
3048                 s->data = NULL;
3049                 strbuf_release(&buf);
3050         } else {
3051                 s->data = strbuf_detach(&buf, NULL);
3052                 s->should_free = 1;
3053         }
3054         return 0;
3055 }
3056
3057 /*
3058  * While doing rename detection and pickaxe operation, we may need to
3059  * grab the data for the blob (or file) for our own in-core comparison.
3060  * diff_filespec has data and size fields for this purpose.
3061  */
3062 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3063 {
3064         int size_only = flags & CHECK_SIZE_ONLY;
3065         int err = 0;
3066         /*
3067          * demote FAIL to WARN to allow inspecting the situation
3068          * instead of refusing.
3069          */
3070         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3071                                     ? SAFE_CRLF_WARN
3072                                     : safe_crlf);
3073
3074         if (!DIFF_FILE_VALID(s))
3075                 die("internal error: asking to populate invalid file.");
3076         if (S_ISDIR(s->mode))
3077                 return -1;
3078
3079         if (s->data)
3080                 return 0;
3081
3082         if (size_only && 0 < s->size)
3083                 return 0;
3084
3085         if (S_ISGITLINK(s->mode))
3086                 return diff_populate_gitlink(s, size_only);
3087
3088         if (!s->oid_valid ||
3089             reuse_worktree_file(s->path, &s->oid, 0)) {
3090                 struct strbuf buf = STRBUF_INIT;
3091                 struct stat st;
3092                 int fd;
3093
3094                 if (lstat(s->path, &st) < 0) {
3095                         if (errno == ENOENT) {
3096                         err_empty:
3097                                 err = -1;
3098                         empty:
3099                                 s->data = (char *)"";
3100                                 s->size = 0;
3101                                 return err;
3102                         }
3103                 }
3104                 s->size = xsize_t(st.st_size);
3105                 if (!s->size)
3106                         goto empty;
3107                 if (S_ISLNK(st.st_mode)) {
3108                         struct strbuf sb = STRBUF_INIT;
3109
3110                         if (strbuf_readlink(&sb, s->path, s->size))
3111                                 goto err_empty;
3112                         s->size = sb.len;
3113                         s->data = strbuf_detach(&sb, NULL);
3114                         s->should_free = 1;
3115                         return 0;
3116                 }
3117
3118                 /*
3119                  * Even if the caller would be happy with getting
3120                  * only the size, we cannot return early at this
3121                  * point if the path requires us to run the content
3122                  * conversion.
3123                  */
3124                 if (size_only && !would_convert_to_git(&the_index, s->path))
3125                         return 0;
3126
3127                 /*
3128                  * Note: this check uses xsize_t(st.st_size) that may
3129                  * not be the true size of the blob after it goes
3130                  * through convert_to_git().  This may not strictly be
3131                  * correct, but the whole point of big_file_threshold
3132                  * and is_binary check being that we want to avoid
3133                  * opening the file and inspecting the contents, this
3134                  * is probably fine.
3135                  */
3136                 if ((flags & CHECK_BINARY) &&
3137                     s->size > big_file_threshold && s->is_binary == -1) {
3138                         s->is_binary = 1;
3139                         return 0;
3140                 }
3141                 fd = open(s->path, O_RDONLY);
3142                 if (fd < 0)
3143                         goto err_empty;
3144                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3145                 close(fd);
3146                 s->should_munmap = 1;
3147
3148                 /*
3149                  * Convert from working tree format to canonical git format
3150                  */
3151                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3152                         size_t size = 0;
3153                         munmap(s->data, s->size);
3154                         s->should_munmap = 0;
3155                         s->data = strbuf_detach(&buf, &size);
3156                         s->size = size;
3157                         s->should_free = 1;
3158                 }
3159         }
3160         else {
3161                 enum object_type type;
3162                 if (size_only || (flags & CHECK_BINARY)) {
3163                         type = sha1_object_info(s->oid.hash, &s->size);
3164                         if (type < 0)
3165                                 die("unable to read %s",
3166                                     oid_to_hex(&s->oid));
3167                         if (size_only)
3168                                 return 0;
3169                         if (s->size > big_file_threshold && s->is_binary == -1) {
3170                                 s->is_binary = 1;
3171                                 return 0;
3172                         }
3173                 }
3174                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3175                 if (!s->data)
3176                         die("unable to read %s", oid_to_hex(&s->oid));
3177                 s->should_free = 1;
3178         }
3179         return 0;
3180 }
3181
3182 void diff_free_filespec_blob(struct diff_filespec *s)
3183 {
3184         if (s->should_free)
3185                 free(s->data);
3186         else if (s->should_munmap)
3187                 munmap(s->data, s->size);
3188
3189         if (s->should_free || s->should_munmap) {
3190                 s->should_free = s->should_munmap = 0;
3191                 s->data = NULL;
3192         }
3193 }
3194
3195 void diff_free_filespec_data(struct diff_filespec *s)
3196 {
3197         diff_free_filespec_blob(s);
3198         FREE_AND_NULL(s->cnt_data);
3199 }
3200
3201 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3202                            void *blob,
3203                            unsigned long size,
3204                            const struct object_id *oid,
3205                            int mode)
3206 {
3207         int fd;
3208         struct strbuf buf = STRBUF_INIT;
3209         struct strbuf template = STRBUF_INIT;
3210         char *path_dup = xstrdup(path);
3211         const char *base = basename(path_dup);
3212
3213         /* Generate "XXXXXX_basename.ext" */
3214         strbuf_addstr(&template, "XXXXXX_");
3215         strbuf_addstr(&template, base);
3216
3217         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
3218         if (fd < 0)
3219                 die_errno("unable to create temp-file");
3220         if (convert_to_working_tree(path,
3221                         (const char *)blob, (size_t)size, &buf)) {
3222                 blob = buf.buf;
3223                 size = buf.len;
3224         }
3225         if (write_in_full(fd, blob, size) != size)
3226                 die_errno("unable to write temp-file");
3227         close_tempfile(&temp->tempfile);
3228         temp->name = get_tempfile_path(&temp->tempfile);
3229         oid_to_hex_r(temp->hex, oid);
3230         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3231         strbuf_release(&buf);
3232         strbuf_release(&template);
3233         free(path_dup);
3234 }
3235
3236 static struct diff_tempfile *prepare_temp_file(const char *name,
3237                 struct diff_filespec *one)
3238 {
3239         struct diff_tempfile *temp = claim_diff_tempfile();
3240
3241         if (!DIFF_FILE_VALID(one)) {
3242         not_a_valid_file:
3243                 /* A '-' entry produces this for file-2, and
3244                  * a '+' entry produces this for file-1.
3245                  */
3246                 temp->name = "/dev/null";
3247                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3248                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3249                 return temp;
3250         }
3251
3252         if (!S_ISGITLINK(one->mode) &&
3253             (!one->oid_valid ||
3254              reuse_worktree_file(name, &one->oid, 1))) {
3255                 struct stat st;
3256                 if (lstat(name, &st) < 0) {
3257                         if (errno == ENOENT)
3258                                 goto not_a_valid_file;
3259                         die_errno("stat(%s)", name);
3260                 }
3261                 if (S_ISLNK(st.st_mode)) {
3262                         struct strbuf sb = STRBUF_INIT;
3263                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3264                                 die_errno("readlink(%s)", name);
3265                         prep_temp_blob(name, temp, sb.buf, sb.len,
3266                                        (one->oid_valid ?
3267                                         &one->oid : &null_oid),
3268                                        (one->oid_valid ?
3269                                         one->mode : S_IFLNK));
3270                         strbuf_release(&sb);
3271                 }
3272                 else {
3273                         /* we can borrow from the file in the work tree */
3274                         temp->name = name;
3275                         if (!one->oid_valid)
3276                                 oid_to_hex_r(temp->hex, &null_oid);
3277                         else
3278                                 oid_to_hex_r(temp->hex, &one->oid);
3279                         /* Even though we may sometimes borrow the
3280                          * contents from the work tree, we always want
3281                          * one->mode.  mode is trustworthy even when
3282                          * !(one->oid_valid), as long as
3283                          * DIFF_FILE_VALID(one).
3284                          */
3285                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3286                 }
3287                 return temp;
3288         }
3289         else {
3290                 if (diff_populate_filespec(one, 0))
3291                         die("cannot read data blob for %s", one->path);
3292                 prep_temp_blob(name, temp, one->data, one->size,
3293                                &one->oid, one->mode);
3294         }
3295         return temp;
3296 }
3297
3298 static void add_external_diff_name(struct argv_array *argv,
3299                                    const char *name,
3300                                    struct diff_filespec *df)
3301 {
3302         struct diff_tempfile *temp = prepare_temp_file(name, df);
3303         argv_array_push(argv, temp->name);
3304         argv_array_push(argv, temp->hex);
3305         argv_array_push(argv, temp->mode);
3306 }
3307
3308 /* An external diff command takes:
3309  *
3310  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3311  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3312  *
3313  */
3314 static void run_external_diff(const char *pgm,
3315                               const char *name,
3316                               const char *other,
3317                               struct diff_filespec *one,
3318                               struct diff_filespec *two,
3319                               const char *xfrm_msg,
3320                               int complete_rewrite,
3321                               struct diff_options *o)
3322 {
3323         struct argv_array argv = ARGV_ARRAY_INIT;
3324         struct argv_array env = ARGV_ARRAY_INIT;
3325         struct diff_queue_struct *q = &diff_queued_diff;
3326
3327         argv_array_push(&argv, pgm);
3328         argv_array_push(&argv, name);
3329
3330         if (one && two) {
3331                 add_external_diff_name(&argv, name, one);
3332                 if (!other)
3333                         add_external_diff_name(&argv, name, two);
3334                 else {
3335                         add_external_diff_name(&argv, other, two);
3336                         argv_array_push(&argv, other);
3337                         argv_array_push(&argv, xfrm_msg);
3338                 }
3339         }
3340
3341         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3342         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3343
3344         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3345                 die(_("external diff died, stopping at %s"), name);
3346
3347         remove_tempfile();
3348         argv_array_clear(&argv);
3349         argv_array_clear(&env);
3350 }
3351
3352 static int similarity_index(struct diff_filepair *p)
3353 {
3354         return p->score * 100 / MAX_SCORE;
3355 }
3356
3357 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3358 {
3359         if (startup_info->have_repository)
3360                 return find_unique_abbrev(oid->hash, abbrev);
3361         else {
3362                 char *hex = oid_to_hex(oid);
3363                 if (abbrev < 0)
3364                         abbrev = FALLBACK_DEFAULT_ABBREV;
3365                 if (abbrev > GIT_SHA1_HEXSZ)
3366                         die("BUG: oid abbreviation out of range: %d", abbrev);
3367                 if (abbrev)
3368                         hex[abbrev] = '\0';
3369                 return hex;
3370         }
3371 }
3372
3373 static void fill_metainfo(struct strbuf *msg,
3374                           const char *name,
3375                           const char *other,
3376                           struct diff_filespec *one,
3377                           struct diff_filespec *two,
3378                           struct diff_options *o,
3379                           struct diff_filepair *p,
3380                           int *must_show_header,
3381                           int use_color)
3382 {
3383         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3384         const char *reset = diff_get_color(use_color, DIFF_RESET);
3385         const char *line_prefix = diff_line_prefix(o);
3386
3387         *must_show_header = 1;
3388         strbuf_init(msg, PATH_MAX * 2 + 300);
3389         switch (p->status) {
3390         case DIFF_STATUS_COPIED:
3391                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3392                             line_prefix, set, similarity_index(p));
3393                 strbuf_addf(msg, "%s\n%s%scopy from ",
3394                             reset,  line_prefix, set);
3395                 quote_c_style(name, msg, NULL, 0);
3396                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3397                 quote_c_style(other, msg, NULL, 0);
3398                 strbuf_addf(msg, "%s\n", reset);
3399                 break;
3400         case DIFF_STATUS_RENAMED:
3401                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3402                             line_prefix, set, similarity_index(p));
3403                 strbuf_addf(msg, "%s\n%s%srename from ",
3404                             reset, line_prefix, set);
3405                 quote_c_style(name, msg, NULL, 0);
3406                 strbuf_addf(msg, "%s\n%s%srename to ",
3407                             reset, line_prefix, set);
3408                 quote_c_style(other, msg, NULL, 0);
3409                 strbuf_addf(msg, "%s\n", reset);
3410                 break;
3411         case DIFF_STATUS_MODIFIED:
3412                 if (p->score) {
3413                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3414                                     line_prefix,
3415                                     set, similarity_index(p), reset);
3416                         break;
3417                 }
3418                 /* fallthru */
3419         default:
3420                 *must_show_header = 0;
3421         }
3422         if (one && two && oidcmp(&one->oid, &two->oid)) {
3423                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3424
3425                 if (DIFF_OPT_TST(o, BINARY)) {
3426                         mmfile_t mf;
3427                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3428                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3429                                 abbrev = 40;
3430                 }
3431                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3432                             diff_abbrev_oid(&one->oid, abbrev),
3433                             diff_abbrev_oid(&two->oid, abbrev));
3434                 if (one->mode == two->mode)
3435                         strbuf_addf(msg, " %06o", one->mode);
3436                 strbuf_addf(msg, "%s\n", reset);
3437         }
3438 }
3439
3440 static void run_diff_cmd(const char *pgm,
3441                          const char *name,
3442                          const char *other,
3443                          const char *attr_path,
3444                          struct diff_filespec *one,
3445                          struct diff_filespec *two,
3446                          struct strbuf *msg,
3447                          struct diff_options *o,
3448                          struct diff_filepair *p)
3449 {
3450         const char *xfrm_msg = NULL;
3451         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3452         int must_show_header = 0;
3453
3454
3455         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3456                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3457                 if (drv && drv->external)
3458                         pgm = drv->external;
3459         }
3460
3461         if (msg) {
3462                 /*
3463                  * don't use colors when the header is intended for an
3464                  * external diff driver
3465                  */
3466                 fill_metainfo(msg, name, other, one, two, o, p,
3467                               &must_show_header,
3468                               want_color(o->use_color) && !pgm);
3469                 xfrm_msg = msg->len ? msg->buf : NULL;
3470         }
3471
3472         if (pgm) {
3473                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3474                                   complete_rewrite, o);
3475                 return;
3476         }
3477         if (one && two)
3478                 builtin_diff(name, other ? other : name,
3479                              one, two, xfrm_msg, must_show_header,
3480                              o, complete_rewrite);
3481         else
3482                 fprintf(o->file, "* Unmerged path %s\n", name);
3483 }
3484
3485 static void diff_fill_oid_info(struct diff_filespec *one)
3486 {
3487         if (DIFF_FILE_VALID(one)) {
3488                 if (!one->oid_valid) {
3489                         struct stat st;
3490                         if (one->is_stdin) {
3491                                 oidclr(&one->oid);
3492                                 return;
3493                         }
3494                         if (lstat(one->path, &st) < 0)
3495                                 die_errno("stat '%s'", one->path);
3496                         if (index_path(one->oid.hash, one->path, &st, 0))
3497                                 die("cannot hash %s", one->path);
3498                 }
3499         }
3500         else
3501                 oidclr(&one->oid);
3502 }
3503
3504 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3505 {
3506         /* Strip the prefix but do not molest /dev/null and absolute paths */
3507         if (*namep && **namep != '/') {
3508                 *namep += prefix_length;
3509                 if (**namep == '/')
3510                         ++*namep;
3511         }
3512         if (*otherp && **otherp != '/') {
3513                 *otherp += prefix_length;
3514                 if (**otherp == '/')
3515                         ++*otherp;
3516         }
3517 }
3518
3519 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3520 {
3521         const char *pgm = external_diff();
3522         struct strbuf msg;
3523         struct diff_filespec *one = p->one;
3524         struct diff_filespec *two = p->two;
3525         const char *name;
3526         const char *other;
3527         const char *attr_path;
3528
3529         name  = one->path;
3530         other = (strcmp(name, two->path) ? two->path : NULL);
3531         attr_path = name;
3532         if (o->prefix_length)
3533                 strip_prefix(o->prefix_length, &name, &other);
3534
3535         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3536                 pgm = NULL;
3537
3538         if (DIFF_PAIR_UNMERGED(p)) {
3539                 run_diff_cmd(pgm, name, NULL, attr_path,
3540                              NULL, NULL, NULL, o, p);
3541                 return;
3542         }
3543
3544         diff_fill_oid_info(one);
3545         diff_fill_oid_info(two);
3546
3547         if (!pgm &&
3548             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3549             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3550                 /*
3551                  * a filepair that changes between file and symlink
3552                  * needs to be split into deletion and creation.
3553                  */
3554                 struct diff_filespec *null = alloc_filespec(two->path);
3555                 run_diff_cmd(NULL, name, other, attr_path,
3556                              one, null, &msg, o, p);
3557                 free(null);
3558                 strbuf_release(&msg);
3559
3560                 null = alloc_filespec(one->path);
3561                 run_diff_cmd(NULL, name, other, attr_path,
3562                              null, two, &msg, o, p);
3563                 free(null);
3564         }
3565         else
3566                 run_diff_cmd(pgm, name, other, attr_path,
3567                              one, two, &msg, o, p);
3568
3569         strbuf_release(&msg);
3570 }
3571
3572 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3573                          struct diffstat_t *diffstat)
3574 {
3575         const char *name;
3576         const char *other;
3577
3578         if (DIFF_PAIR_UNMERGED(p)) {
3579                 /* unmerged */
3580                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3581                 return;
3582         }
3583
3584         name = p->one->path;
3585         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3586
3587         if (o->prefix_length)
3588                 strip_prefix(o->prefix_length, &name, &other);
3589
3590         diff_fill_oid_info(p->one);
3591         diff_fill_oid_info(p->two);
3592
3593         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3594 }
3595
3596 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3597 {
3598         const char *name;
3599         const char *other;
3600         const char *attr_path;
3601
3602         if (DIFF_PAIR_UNMERGED(p)) {
3603                 /* unmerged */
3604                 return;
3605         }
3606
3607         name = p->one->path;
3608         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3609         attr_path = other ? other : name;
3610
3611         if (o->prefix_length)
3612                 strip_prefix(o->prefix_length, &name, &other);
3613
3614         diff_fill_oid_info(p->one);
3615         diff_fill_oid_info(p->two);
3616
3617         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3618 }
3619
3620 void diff_setup(struct diff_options *options)
3621 {
3622         memcpy(options, &default_diff_options, sizeof(*options));
3623
3624         options->file = stdout;
3625
3626         options->abbrev = DEFAULT_ABBREV;
3627         options->line_termination = '\n';
3628         options->break_opt = -1;
3629         options->rename_limit = -1;
3630         options->dirstat_permille = diff_dirstat_permille_default;
3631         options->context = diff_context_default;
3632         options->interhunkcontext = diff_interhunk_context_default;
3633         options->ws_error_highlight = ws_error_highlight_default;
3634         DIFF_OPT_SET(options, RENAME_EMPTY);
3635
3636         /* pathchange left =NULL by default */
3637         options->change = diff_change;
3638         options->add_remove = diff_addremove;
3639         options->use_color = diff_use_color_default;
3640         options->detect_rename = diff_detect_rename_default;
3641         options->xdl_opts |= diff_algorithm;
3642         if (diff_indent_heuristic)
3643                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3644
3645         options->orderfile = diff_order_file_cfg;
3646
3647         if (diff_no_prefix) {
3648                 options->a_prefix = options->b_prefix = "";
3649         } else if (!diff_mnemonic_prefix) {
3650                 options->a_prefix = "a/";
3651                 options->b_prefix = "b/";
3652         }
3653 }
3654
3655 void diff_setup_done(struct diff_options *options)
3656 {
3657         int count = 0;
3658
3659         if (options->set_default)
3660                 options->set_default(options);
3661
3662         if (options->output_format & DIFF_FORMAT_NAME)
3663                 count++;
3664         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3665                 count++;
3666         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3667                 count++;
3668         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3669                 count++;
3670         if (count > 1)
3671                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
3672
3673         /*
3674          * Most of the time we can say "there are changes"
3675          * only by checking if there are changed paths, but
3676          * --ignore-whitespace* options force us to look
3677          * inside contents.
3678          */
3679
3680         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3681             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3682             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3683                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3684         else
3685                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3686
3687         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3688                 options->detect_rename = DIFF_DETECT_COPY;
3689
3690         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3691                 options->prefix = NULL;
3692         if (options->prefix)
3693                 options->prefix_length = strlen(options->prefix);
3694         else
3695                 options->prefix_length = 0;
3696
3697         if (options->output_format & (DIFF_FORMAT_NAME |
3698                                       DIFF_FORMAT_NAME_STATUS |
3699                                       DIFF_FORMAT_CHECKDIFF |
3700                                       DIFF_FORMAT_NO_OUTPUT))
3701                 options->output_format &= ~(DIFF_FORMAT_RAW |
3702                                             DIFF_FORMAT_NUMSTAT |
3703                                             DIFF_FORMAT_DIFFSTAT |
3704                                             DIFF_FORMAT_SHORTSTAT |
3705                                             DIFF_FORMAT_DIRSTAT |
3706                                             DIFF_FORMAT_SUMMARY |
3707                                             DIFF_FORMAT_PATCH);
3708
3709         /*
3710          * These cases always need recursive; we do not drop caller-supplied
3711          * recursive bits for other formats here.
3712          */
3713         if (options->output_format & (DIFF_FORMAT_PATCH |
3714                                       DIFF_FORMAT_NUMSTAT |
3715                                       DIFF_FORMAT_DIFFSTAT |
3716                                       DIFF_FORMAT_SHORTSTAT |
3717                                       DIFF_FORMAT_DIRSTAT |
3718                                       DIFF_FORMAT_SUMMARY |
3719                                       DIFF_FORMAT_CHECKDIFF))
3720                 DIFF_OPT_SET(options, RECURSIVE);
3721         /*
3722          * Also pickaxe would not work very well if you do not say recursive
3723          */
3724         if (options->pickaxe)
3725                 DIFF_OPT_SET(options, RECURSIVE);
3726         /*
3727          * When patches are generated, submodules diffed against the work tree
3728          * must be checked for dirtiness too so it can be shown in the output
3729          */
3730         if (options->output_format & DIFF_FORMAT_PATCH)
3731                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3732
3733         if (options->detect_rename && options->rename_limit < 0)
3734                 options->rename_limit = diff_rename_limit_default;
3735         if (options->setup & DIFF_SETUP_USE_CACHE) {
3736                 if (!active_cache)
3737                         /* read-cache does not die even when it fails
3738                          * so it is safe for us to do this here.  Also
3739                          * it does not smudge active_cache or active_nr
3740                          * when it fails, so we do not have to worry about
3741                          * cleaning it up ourselves either.
3742                          */
3743                         read_cache();
3744         }
3745         if (40 < options->abbrev)
3746                 options->abbrev = 40; /* full */
3747
3748         /*
3749          * It does not make sense to show the first hit we happened
3750          * to have found.  It does not make sense not to return with
3751          * exit code in such a case either.
3752          */
3753         if (DIFF_OPT_TST(options, QUICK)) {
3754                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3755                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3756         }
3757
3758         options->diff_path_counter = 0;
3759
3760         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3761                 die(_("--follow requires exactly one pathspec"));
3762 }
3763
3764 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3765 {
3766         char c, *eq;
3767         int len;
3768
3769         if (*arg != '-')
3770                 return 0;
3771         c = *++arg;
3772         if (!c)
3773                 return 0;
3774         if (c == arg_short) {
3775                 c = *++arg;
3776                 if (!c)
3777                         return 1;
3778                 if (val && isdigit(c)) {
3779                         char *end;
3780                         int n = strtoul(arg, &end, 10);
3781                         if (*end)
3782                                 return 0;
3783                         *val = n;
3784                         return 1;
3785                 }
3786                 return 0;
3787         }
3788         if (c != '-')
3789                 return 0;
3790         arg++;
3791         eq = strchrnul(arg, '=');
3792         len = eq - arg;
3793         if (!len || strncmp(arg, arg_long, len))
3794                 return 0;
3795         if (*eq) {
3796                 int n;
3797                 char *end;
3798                 if (!isdigit(*++eq))
3799                         return 0;
3800                 n = strtoul(eq, &end, 10);
3801                 if (*end)
3802                         return 0;
3803                 *val = n;
3804         }
3805         return 1;
3806 }
3807
3808 static int diff_scoreopt_parse(const char *opt);
3809
3810 static inline int short_opt(char opt, const char **argv,
3811                             const char **optarg)
3812 {
3813         const char *arg = argv[0];
3814         if (arg[0] != '-' || arg[1] != opt)
3815                 return 0;
3816         if (arg[2] != '\0') {
3817                 *optarg = arg + 2;
3818                 return 1;
3819         }
3820         if (!argv[1])
3821                 die("Option '%c' requires a value", opt);
3822         *optarg = argv[1];
3823         return 2;
3824 }
3825
3826 int parse_long_opt(const char *opt, const char **argv,
3827                    const char **optarg)
3828 {
3829         const char *arg = argv[0];
3830         if (!skip_prefix(arg, "--", &arg))
3831                 return 0;
3832         if (!skip_prefix(arg, opt, &arg))
3833                 return 0;
3834         if (*arg == '=') { /* stuck form: --option=value */
3835                 *optarg = arg + 1;
3836                 return 1;
3837         }
3838         if (*arg != '\0')
3839                 return 0;
3840         /* separate form: --option value */
3841         if (!argv[1])
3842                 die("Option '--%s' requires a value", opt);
3843         *optarg = argv[1];
3844         return 2;
3845 }
3846
3847 static int stat_opt(struct diff_options *options, const char **av)
3848 {
3849         const char *arg = av[0];
3850         char *end;
3851         int width = options->stat_width;
3852         int name_width = options->stat_name_width;
3853         int graph_width = options->stat_graph_width;
3854         int count = options->stat_count;
3855         int argcount = 1;
3856
3857         if (!skip_prefix(arg, "--stat", &arg))
3858                 die("BUG: stat option does not begin with --stat: %s", arg);
3859         end = (char *)arg;
3860
3861         switch (*arg) {
3862         case '-':
3863                 if (skip_prefix(arg, "-width", &arg)) {
3864                         if (*arg == '=')
3865                                 width = strtoul(arg + 1, &end, 10);
3866                         else if (!*arg && !av[1])
3867                                 die_want_option("--stat-width");
3868                         else if (!*arg) {
3869                                 width = strtoul(av[1], &end, 10);
3870                                 argcount = 2;
3871                         }
3872                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3873                         if (*arg == '=')
3874                                 name_width = strtoul(arg + 1, &end, 10);
3875                         else if (!*arg && !av[1])
3876                                 die_want_option("--stat-name-width");
3877                         else if (!*arg) {
3878                                 name_width = strtoul(av[1], &end, 10);
3879                                 argcount = 2;
3880                         }
3881                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3882                         if (*arg == '=')
3883                                 graph_width = strtoul(arg + 1, &end, 10);
3884                         else if (!*arg && !av[1])
3885                                 die_want_option("--stat-graph-width");
3886                         else if (!*arg) {
3887                                 graph_width = strtoul(av[1], &end, 10);
3888                                 argcount = 2;
3889                         }
3890                 } else if (skip_prefix(arg, "-count", &arg)) {
3891                         if (*arg == '=')
3892                                 count = strtoul(arg + 1, &end, 10);
3893                         else if (!*arg && !av[1])
3894                                 die_want_option("--stat-count");
3895                         else if (!*arg) {
3896                                 count = strtoul(av[1], &end, 10);
3897                                 argcount = 2;
3898                         }
3899                 }
3900                 break;
3901         case '=':
3902                 width = strtoul(arg+1, &end, 10);
3903                 if (*end == ',')
3904                         name_width = strtoul(end+1, &end, 10);
3905                 if (*end == ',')
3906                         count = strtoul(end+1, &end, 10);
3907         }
3908
3909         /* Important! This checks all the error cases! */
3910         if (*end)
3911                 return 0;
3912         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3913         options->stat_name_width = name_width;
3914         options->stat_graph_width = graph_width;
3915         options->stat_width = width;
3916         options->stat_count = count;
3917         return argcount;
3918 }
3919
3920 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3921 {
3922         struct strbuf errmsg = STRBUF_INIT;
3923         if (parse_dirstat_params(options, params, &errmsg))
3924                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3925                     errmsg.buf);
3926         strbuf_release(&errmsg);
3927         /*
3928          * The caller knows a dirstat-related option is given from the command
3929          * line; allow it to say "return this_function();"
3930          */
3931         options->output_format |= DIFF_FORMAT_DIRSTAT;
3932         return 1;
3933 }
3934
3935 static int parse_submodule_opt(struct diff_options *options, const char *value)
3936 {
3937         if (parse_submodule_params(options, value))
3938                 die(_("Failed to parse --submodule option parameter: '%s'"),
3939                         value);
3940         return 1;
3941 }
3942
3943 static const char diff_status_letters[] = {
3944         DIFF_STATUS_ADDED,
3945         DIFF_STATUS_COPIED,
3946         DIFF_STATUS_DELETED,
3947         DIFF_STATUS_MODIFIED,
3948         DIFF_STATUS_RENAMED,
3949         DIFF_STATUS_TYPE_CHANGED,
3950         DIFF_STATUS_UNKNOWN,
3951         DIFF_STATUS_UNMERGED,
3952         DIFF_STATUS_FILTER_AON,
3953         DIFF_STATUS_FILTER_BROKEN,
3954         '\0',
3955 };
3956
3957 static unsigned int filter_bit['Z' + 1];
3958
3959 static void prepare_filter_bits(void)
3960 {
3961         int i;
3962
3963         if (!filter_bit[DIFF_STATUS_ADDED]) {
3964                 for (i = 0; diff_status_letters[i]; i++)
3965                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3966         }
3967 }
3968
3969 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3970 {
3971         return opt->filter & filter_bit[(int) status];
3972 }
3973
3974 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3975 {
3976         int i, optch;
3977
3978         prepare_filter_bits();
3979
3980         /*
3981          * If there is a negation e.g. 'd' in the input, and we haven't
3982          * initialized the filter field with another --diff-filter, start
3983          * from full set of bits, except for AON.
3984          */
3985         if (!opt->filter) {
3986                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3987                         if (optch < 'a' || 'z' < optch)
3988                                 continue;
3989                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3990                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3991                         break;
3992                 }
3993         }
3994
3995         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3996                 unsigned int bit;
3997                 int negate;
3998
3999                 if ('a' <= optch && optch <= 'z') {
4000                         negate = 1;
4001                         optch = toupper(optch);
4002                 } else {
4003                         negate = 0;
4004                 }
4005
4006                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4007                 if (!bit)
4008                         return optarg[i];
4009                 if (negate)
4010                         opt->filter &= ~bit;
4011                 else
4012                         opt->filter |= bit;
4013         }
4014         return 0;
4015 }
4016
4017 static void enable_patch_output(int *fmt) {
4018         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4019         *fmt |= DIFF_FORMAT_PATCH;
4020 }
4021
4022 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4023 {
4024         int val = parse_ws_error_highlight(arg);
4025
4026         if (val < 0) {
4027                 error("unknown value after ws-error-highlight=%.*s",
4028                       -1 - val, arg);
4029                 return 0;
4030         }
4031         opt->ws_error_highlight = val;
4032         return 1;
4033 }
4034
4035 int diff_opt_parse(struct diff_options *options,
4036                    const char **av, int ac, const char *prefix)
4037 {
4038         const char *arg = av[0];
4039         const char *optarg;
4040         int argcount;
4041
4042         if (!prefix)
4043                 prefix = "";
4044
4045         /* Output format options */
4046         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4047             || opt_arg(arg, 'U', "unified", &options->context))
4048                 enable_patch_output(&options->output_format);
4049         else if (!strcmp(arg, "--raw"))
4050                 options->output_format |= DIFF_FORMAT_RAW;
4051         else if (!strcmp(arg, "--patch-with-raw")) {
4052                 enable_patch_output(&options->output_format);
4053                 options->output_format |= DIFF_FORMAT_RAW;
4054         } else if (!strcmp(arg, "--numstat"))
4055                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4056         else if (!strcmp(arg, "--shortstat"))
4057                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4058         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
4059                 return parse_dirstat_opt(options, "");
4060         else if (skip_prefix(arg, "-X", &arg))
4061                 return parse_dirstat_opt(options, arg);
4062         else if (skip_prefix(arg, "--dirstat=", &arg))
4063                 return parse_dirstat_opt(options, arg);
4064         else if (!strcmp(arg, "--cumulative"))
4065                 return parse_dirstat_opt(options, "cumulative");
4066         else if (!strcmp(arg, "--dirstat-by-file"))
4067                 return parse_dirstat_opt(options, "files");
4068         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
4069                 parse_dirstat_opt(options, "files");
4070                 return parse_dirstat_opt(options, arg);
4071         }
4072         else if (!strcmp(arg, "--check"))
4073                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4074         else if (!strcmp(arg, "--summary"))
4075                 options->output_format |= DIFF_FORMAT_SUMMARY;
4076         else if (!strcmp(arg, "--patch-with-stat")) {
4077                 enable_patch_output(&options->output_format);
4078                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4079         } else if (!strcmp(arg, "--name-only"))
4080                 options->output_format |= DIFF_FORMAT_NAME;
4081         else if (!strcmp(arg, "--name-status"))
4082                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4083         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4084                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4085         else if (starts_with(arg, "--stat"))
4086                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4087                 return stat_opt(options, av);
4088
4089         /* renames options */
4090         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
4091                  !strcmp(arg, "--break-rewrites")) {
4092                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4093                         return error("invalid argument to -B: %s", arg+2);
4094         }
4095         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
4096                  !strcmp(arg, "--find-renames")) {
4097                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4098                         return error("invalid argument to -M: %s", arg+2);
4099                 options->detect_rename = DIFF_DETECT_RENAME;
4100         }
4101         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4102                 options->irreversible_delete = 1;
4103         }
4104         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
4105                  !strcmp(arg, "--find-copies")) {
4106                 if (options->detect_rename == DIFF_DETECT_COPY)
4107                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4108                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4109                         return error("invalid argument to -C: %s", arg+2);
4110                 options->detect_rename = DIFF_DETECT_COPY;
4111         }
4112         else if (!strcmp(arg, "--no-renames"))
4113                 options->detect_rename = 0;
4114         else if (!strcmp(arg, "--rename-empty"))
4115                 DIFF_OPT_SET(options, RENAME_EMPTY);
4116         else if (!strcmp(arg, "--no-rename-empty"))
4117                 DIFF_OPT_CLR(options, RENAME_EMPTY);
4118         else if (!strcmp(arg, "--relative"))
4119                 DIFF_OPT_SET(options, RELATIVE_NAME);
4120         else if (skip_prefix(arg, "--relative=", &arg)) {
4121                 DIFF_OPT_SET(options, RELATIVE_NAME);
4122                 options->prefix = arg;
4123         }
4124
4125         /* xdiff options */
4126         else if (!strcmp(arg, "--minimal"))
4127                 DIFF_XDL_SET(options, NEED_MINIMAL);
4128         else if (!strcmp(arg, "--no-minimal"))
4129                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4130         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4131                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4132         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4133                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4134         else if (!strcmp(arg, "--ignore-space-at-eol"))
4135                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4136         else if (!strcmp(arg, "--ignore-blank-lines"))
4137                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4138         else if (!strcmp(arg, "--indent-heuristic"))
4139                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4140         else if (!strcmp(arg, "--no-indent-heuristic"))
4141                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4142         else if (!strcmp(arg, "--patience"))
4143                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4144         else if (!strcmp(arg, "--histogram"))
4145                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4146         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4147                 long value = parse_algorithm_value(optarg);
4148                 if (value < 0)
4149                         return error("option diff-algorithm accepts \"myers\", "
4150                                      "\"minimal\", \"patience\" and \"histogram\"");
4151                 /* clear out previous settings */
4152                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4153                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4154                 options->xdl_opts |= value;
4155                 return argcount;
4156         }
4157
4158         /* flags options */
4159         else if (!strcmp(arg, "--binary")) {
4160                 enable_patch_output(&options->output_format);
4161                 DIFF_OPT_SET(options, BINARY);
4162         }
4163         else if (!strcmp(arg, "--full-index"))
4164                 DIFF_OPT_SET(options, FULL_INDEX);
4165         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4166                 DIFF_OPT_SET(options, TEXT);
4167         else if (!strcmp(arg, "-R"))
4168                 DIFF_OPT_SET(options, REVERSE_DIFF);
4169         else if (!strcmp(arg, "--find-copies-harder"))
4170                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4171         else if (!strcmp(arg, "--follow"))
4172                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
4173         else if (!strcmp(arg, "--no-follow")) {
4174                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
4175                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
4176         } else if (!strcmp(arg, "--color"))
4177                 options->use_color = 1;
4178         else if (skip_prefix(arg, "--color=", &arg)) {
4179                 int value = git_config_colorbool(NULL, arg);
4180                 if (value < 0)
4181                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4182                 options->use_color = value;
4183         }
4184         else if (!strcmp(arg, "--no-color"))
4185                 options->use_color = 0;
4186         else if (!strcmp(arg, "--color-words")) {
4187                 options->use_color = 1;
4188                 options->word_diff = DIFF_WORDS_COLOR;
4189         }
4190         else if (skip_prefix(arg, "--color-words=", &arg)) {
4191                 options->use_color = 1;
4192                 options->word_diff = DIFF_WORDS_COLOR;
4193                 options->word_regex = arg;
4194         }
4195         else if (!strcmp(arg, "--word-diff")) {
4196                 if (options->word_diff == DIFF_WORDS_NONE)
4197                         options->word_diff = DIFF_WORDS_PLAIN;
4198         }
4199         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4200                 if (!strcmp(arg, "plain"))
4201                         options->word_diff = DIFF_WORDS_PLAIN;
4202                 else if (!strcmp(arg, "color")) {
4203                         options->use_color = 1;
4204                         options->word_diff = DIFF_WORDS_COLOR;
4205                 }
4206                 else if (!strcmp(arg, "porcelain"))
4207                         options->word_diff = DIFF_WORDS_PORCELAIN;
4208                 else if (!strcmp(arg, "none"))
4209                         options->word_diff = DIFF_WORDS_NONE;
4210                 else
4211                         die("bad --word-diff argument: %s", arg);
4212         }
4213         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4214                 if (options->word_diff == DIFF_WORDS_NONE)
4215                         options->word_diff = DIFF_WORDS_PLAIN;
4216                 options->word_regex = optarg;
4217                 return argcount;
4218         }
4219         else if (!strcmp(arg, "--exit-code"))
4220                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4221         else if (!strcmp(arg, "--quiet"))
4222                 DIFF_OPT_SET(options, QUICK);
4223         else if (!strcmp(arg, "--ext-diff"))
4224                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
4225         else if (!strcmp(arg, "--no-ext-diff"))
4226                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
4227         else if (!strcmp(arg, "--textconv"))
4228                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
4229         else if (!strcmp(arg, "--no-textconv"))
4230                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
4231         else if (!strcmp(arg, "--ignore-submodules")) {
4232                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4233                 handle_ignore_submodules_arg(options, "all");
4234         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
4235                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4236                 handle_ignore_submodules_arg(options, arg);
4237         } else if (!strcmp(arg, "--submodule"))
4238                 options->submodule_format = DIFF_SUBMODULE_LOG;
4239         else if (skip_prefix(arg, "--submodule=", &arg))
4240                 return parse_submodule_opt(options, arg);
4241         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4242                 return parse_ws_error_highlight_opt(options, arg);
4243         else if (!strcmp(arg, "--ita-invisible-in-index"))
4244                 options->ita_invisible_in_index = 1;
4245         else if (!strcmp(arg, "--ita-visible-in-index"))
4246                 options->ita_invisible_in_index = 0;
4247
4248         /* misc options */
4249         else if (!strcmp(arg, "-z"))
4250                 options->line_termination = 0;
4251         else if ((argcount = short_opt('l', av, &optarg))) {
4252                 options->rename_limit = strtoul(optarg, NULL, 10);
4253                 return argcount;
4254         }
4255         else if ((argcount = short_opt('S', av, &optarg))) {
4256                 options->pickaxe = optarg;
4257                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4258                 return argcount;
4259         } else if ((argcount = short_opt('G', av, &optarg))) {
4260                 options->pickaxe = optarg;
4261                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4262                 return argcount;
4263         }
4264         else if (!strcmp(arg, "--pickaxe-all"))
4265                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4266         else if (!strcmp(arg, "--pickaxe-regex"))
4267                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4268         else if ((argcount = short_opt('O', av, &optarg))) {
4269                 options->orderfile = prefix_filename(prefix, optarg);
4270                 return argcount;
4271         }
4272         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4273                 int offending = parse_diff_filter_opt(optarg, options);
4274                 if (offending)
4275                         die("unknown change class '%c' in --diff-filter=%s",
4276                             offending, optarg);
4277                 return argcount;
4278         }
4279         else if (!strcmp(arg, "--no-abbrev"))
4280                 options->abbrev = 0;
4281         else if (!strcmp(arg, "--abbrev"))
4282                 options->abbrev = DEFAULT_ABBREV;
4283         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4284                 options->abbrev = strtoul(arg, NULL, 10);
4285                 if (options->abbrev < MINIMUM_ABBREV)
4286                         options->abbrev = MINIMUM_ABBREV;
4287                 else if (40 < options->abbrev)
4288                         options->abbrev = 40;
4289         }
4290         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4291                 options->a_prefix = optarg;
4292                 return argcount;
4293         }
4294         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4295                 options->line_prefix = optarg;
4296                 options->line_prefix_length = strlen(options->line_prefix);
4297                 graph_setup_line_prefix(options);
4298                 return argcount;
4299         }
4300         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4301                 options->b_prefix = optarg;
4302                 return argcount;
4303         }
4304         else if (!strcmp(arg, "--no-prefix"))
4305                 options->a_prefix = options->b_prefix = "";
4306         else if (opt_arg(arg, '\0', "inter-hunk-context",
4307                          &options->interhunkcontext))
4308                 ;
4309         else if (!strcmp(arg, "-W"))
4310                 DIFF_OPT_SET(options, FUNCCONTEXT);
4311         else if (!strcmp(arg, "--function-context"))
4312                 DIFF_OPT_SET(options, FUNCCONTEXT);
4313         else if (!strcmp(arg, "--no-function-context"))
4314                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4315         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4316                 char *path = prefix_filename(prefix, optarg);
4317                 options->file = xfopen(path, "w");
4318                 options->close_file = 1;
4319                 if (options->use_color != GIT_COLOR_ALWAYS)
4320                         options->use_color = GIT_COLOR_NEVER;
4321                 free(path);
4322                 return argcount;
4323         } else
4324                 return 0;
4325         return 1;
4326 }
4327
4328 int parse_rename_score(const char **cp_p)
4329 {
4330         unsigned long num, scale;
4331         int ch, dot;
4332         const char *cp = *cp_p;
4333
4334         num = 0;
4335         scale = 1;
4336         dot = 0;
4337         for (;;) {
4338                 ch = *cp;
4339                 if ( !dot && ch == '.' ) {
4340                         scale = 1;
4341                         dot = 1;
4342                 } else if ( ch == '%' ) {
4343                         scale = dot ? scale*100 : 100;
4344                         cp++;   /* % is always at the end */
4345                         break;
4346                 } else if ( ch >= '0' && ch <= '9' ) {
4347                         if ( scale < 100000 ) {
4348                                 scale *= 10;
4349                                 num = (num*10) + (ch-'0');
4350                         }
4351                 } else {
4352                         break;
4353                 }
4354                 cp++;
4355         }
4356         *cp_p = cp;
4357
4358         /* user says num divided by scale and we say internally that
4359          * is MAX_SCORE * num / scale.
4360          */
4361         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4362 }
4363
4364 static int diff_scoreopt_parse(const char *opt)
4365 {
4366         int opt1, opt2, cmd;
4367
4368         if (*opt++ != '-')
4369                 return -1;
4370         cmd = *opt++;
4371         if (cmd == '-') {
4372                 /* convert the long-form arguments into short-form versions */
4373                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4374                         if (*opt == 0 || *opt++ == '=')
4375                                 cmd = 'B';
4376                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4377                         if (*opt == 0 || *opt++ == '=')
4378                                 cmd = 'C';
4379                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4380                         if (*opt == 0 || *opt++ == '=')
4381                                 cmd = 'M';
4382                 }
4383         }
4384         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4385                 return -1; /* that is not a -M, -C, or -B option */
4386
4387         opt1 = parse_rename_score(&opt);
4388         if (cmd != 'B')
4389                 opt2 = 0;
4390         else {
4391                 if (*opt == 0)
4392                         opt2 = 0;
4393                 else if (*opt != '/')
4394                         return -1; /* we expect -B80/99 or -B80 */
4395                 else {
4396                         opt++;
4397                         opt2 = parse_rename_score(&opt);
4398                 }
4399         }
4400         if (*opt != 0)
4401                 return -1;
4402         return opt1 | (opt2 << 16);
4403 }
4404
4405 struct diff_queue_struct diff_queued_diff;
4406
4407 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4408 {
4409         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4410         queue->queue[queue->nr++] = dp;
4411 }
4412
4413 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4414                                  struct diff_filespec *one,
4415                                  struct diff_filespec *two)
4416 {
4417         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4418         dp->one = one;
4419         dp->two = two;
4420         if (queue)
4421                 diff_q(queue, dp);
4422         return dp;
4423 }
4424
4425 void diff_free_filepair(struct diff_filepair *p)
4426 {
4427         free_filespec(p->one);
4428         free_filespec(p->two);
4429         free(p);
4430 }
4431
4432 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4433 {
4434         int abblen;
4435         const char *abbrev;
4436
4437         if (len == GIT_SHA1_HEXSZ)
4438                 return oid_to_hex(oid);
4439
4440         abbrev = diff_abbrev_oid(oid, len);
4441         abblen = strlen(abbrev);
4442
4443         /*
4444          * In well-behaved cases, where the abbbreviated result is the
4445          * same as the requested length, append three dots after the
4446          * abbreviation (hence the whole logic is limited to the case
4447          * where abblen < 37); when the actual abbreviated result is a
4448          * bit longer than the requested length, we reduce the number
4449          * of dots so that they match the well-behaved ones.  However,
4450          * if the actual abbreviation is longer than the requested
4451          * length by more than three, we give up on aligning, and add
4452          * three dots anyway, to indicate that the output is not the
4453          * full object name.  Yes, this may be suboptimal, but this
4454          * appears only in "diff --raw --abbrev" output and it is not
4455          * worth the effort to change it now.  Note that this would
4456          * likely to work fine when the automatic sizing of default
4457          * abbreviation length is used--we would be fed -1 in "len" in
4458          * that case, and will end up always appending three-dots, but
4459          * the automatic sizing is supposed to give abblen that ensures
4460          * uniqueness across all objects (statistically speaking).
4461          */
4462         if (abblen < GIT_SHA1_HEXSZ - 3) {
4463                 static char hex[GIT_MAX_HEXSZ + 1];
4464                 if (len < abblen && abblen <= len + 2)
4465                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4466                 else
4467                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4468                 return hex;
4469         }
4470
4471         return oid_to_hex(oid);
4472 }
4473
4474 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4475 {
4476         int line_termination = opt->line_termination;
4477         int inter_name_termination = line_termination ? '\t' : '\0';
4478
4479         fprintf(opt->file, "%s", diff_line_prefix(opt));
4480         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4481                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4482                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4483                 fprintf(opt->file, "%s ",
4484                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4485         }
4486         if (p->score) {
4487                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4488                         inter_name_termination);
4489         } else {
4490                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4491         }
4492
4493         if (p->status == DIFF_STATUS_COPIED ||
4494             p->status == DIFF_STATUS_RENAMED) {
4495                 const char *name_a, *name_b;
4496                 name_a = p->one->path;
4497                 name_b = p->two->path;
4498                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4499                 write_name_quoted(name_a, opt->file, inter_name_termination);
4500                 write_name_quoted(name_b, opt->file, line_termination);
4501         } else {
4502                 const char *name_a, *name_b;
4503                 name_a = p->one->mode ? p->one->path : p->two->path;
4504                 name_b = NULL;
4505                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4506                 write_name_quoted(name_a, opt->file, line_termination);
4507         }
4508 }
4509
4510 int diff_unmodified_pair(struct diff_filepair *p)
4511 {
4512         /* This function is written stricter than necessary to support
4513          * the currently implemented transformers, but the idea is to
4514          * let transformers to produce diff_filepairs any way they want,
4515          * and filter and clean them up here before producing the output.
4516          */
4517         struct diff_filespec *one = p->one, *two = p->two;
4518
4519         if (DIFF_PAIR_UNMERGED(p))
4520                 return 0; /* unmerged is interesting */
4521
4522         /* deletion, addition, mode or type change
4523          * and rename are all interesting.
4524          */
4525         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4526             DIFF_PAIR_MODE_CHANGED(p) ||
4527             strcmp(one->path, two->path))
4528                 return 0;
4529
4530         /* both are valid and point at the same path.  that is, we are
4531          * dealing with a change.
4532          */
4533         if (one->oid_valid && two->oid_valid &&
4534             !oidcmp(&one->oid, &two->oid) &&
4535             !one->dirty_submodule && !two->dirty_submodule)
4536                 return 1; /* no change */
4537         if (!one->oid_valid && !two->oid_valid)
4538                 return 1; /* both look at the same file on the filesystem. */
4539         return 0;
4540 }
4541
4542 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4543 {
4544         if (diff_unmodified_pair(p))
4545                 return;
4546
4547         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4548             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4549                 return; /* no tree diffs in patch format */
4550
4551         run_diff(p, o);
4552 }
4553
4554 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4555                             struct diffstat_t *diffstat)
4556 {
4557         if (diff_unmodified_pair(p))
4558                 return;
4559
4560         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4561             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4562                 return; /* no useful stat for tree diffs */
4563
4564         run_diffstat(p, o, diffstat);
4565 }
4566
4567 static void diff_flush_checkdiff(struct diff_filepair *p,
4568                 struct diff_options *o)
4569 {
4570         if (diff_unmodified_pair(p))
4571                 return;
4572
4573         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4574             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4575                 return; /* nothing to check in tree diffs */
4576
4577         run_checkdiff(p, o);
4578 }
4579
4580 int diff_queue_is_empty(void)
4581 {
4582         struct diff_queue_struct *q = &diff_queued_diff;
4583         int i;
4584         for (i = 0; i < q->nr; i++)
4585                 if (!diff_unmodified_pair(q->queue[i]))
4586                         return 0;
4587         return 1;
4588 }
4589
4590 #if DIFF_DEBUG
4591 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4592 {
4593         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4594                 x, one ? one : "",
4595                 s->path,
4596                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4597                 s->mode,
4598                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4599         fprintf(stderr, "queue[%d] %s size %lu\n",
4600                 x, one ? one : "",
4601                 s->size);
4602 }
4603
4604 void diff_debug_filepair(const struct diff_filepair *p, int i)
4605 {
4606         diff_debug_filespec(p->one, i, "one");
4607         diff_debug_filespec(p->two, i, "two");
4608         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4609                 p->score, p->status ? p->status : '?',
4610                 p->one->rename_used, p->broken_pair);
4611 }
4612
4613 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4614 {
4615         int i;
4616         if (msg)
4617                 fprintf(stderr, "%s\n", msg);
4618         fprintf(stderr, "q->nr = %d\n", q->nr);
4619         for (i = 0; i < q->nr; i++) {
4620                 struct diff_filepair *p = q->queue[i];
4621                 diff_debug_filepair(p, i);
4622         }
4623 }
4624 #endif
4625
4626 static void diff_resolve_rename_copy(void)
4627 {
4628         int i;
4629         struct diff_filepair *p;
4630         struct diff_queue_struct *q = &diff_queued_diff;
4631
4632         diff_debug_queue("resolve-rename-copy", q);
4633
4634         for (i = 0; i < q->nr; i++) {
4635                 p = q->queue[i];
4636                 p->status = 0; /* undecided */
4637                 if (DIFF_PAIR_UNMERGED(p))
4638                         p->status = DIFF_STATUS_UNMERGED;
4639                 else if (!DIFF_FILE_VALID(p->one))
4640                         p->status = DIFF_STATUS_ADDED;
4641                 else if (!DIFF_FILE_VALID(p->two))
4642                         p->status = DIFF_STATUS_DELETED;
4643                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4644                         p->status = DIFF_STATUS_TYPE_CHANGED;
4645
4646                 /* from this point on, we are dealing with a pair
4647                  * whose both sides are valid and of the same type, i.e.
4648                  * either in-place edit or rename/copy edit.
4649                  */
4650                 else if (DIFF_PAIR_RENAME(p)) {
4651                         /*
4652                          * A rename might have re-connected a broken
4653                          * pair up, causing the pathnames to be the
4654                          * same again. If so, that's not a rename at
4655                          * all, just a modification..
4656                          *
4657                          * Otherwise, see if this source was used for
4658                          * multiple renames, in which case we decrement
4659                          * the count, and call it a copy.
4660                          */
4661                         if (!strcmp(p->one->path, p->two->path))
4662                                 p->status = DIFF_STATUS_MODIFIED;
4663                         else if (--p->one->rename_used > 0)
4664                                 p->status = DIFF_STATUS_COPIED;
4665                         else
4666                                 p->status = DIFF_STATUS_RENAMED;
4667                 }
4668                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4669                          p->one->mode != p->two->mode ||
4670                          p->one->dirty_submodule ||
4671                          p->two->dirty_submodule ||
4672                          is_null_oid(&p->one->oid))
4673                         p->status = DIFF_STATUS_MODIFIED;
4674                 else {
4675                         /* This is a "no-change" entry and should not
4676                          * happen anymore, but prepare for broken callers.
4677                          */
4678                         error("feeding unmodified %s to diffcore",
4679                               p->one->path);
4680                         p->status = DIFF_STATUS_UNKNOWN;
4681                 }
4682         }
4683         diff_debug_queue("resolve-rename-copy done", q);
4684 }
4685
4686 static int check_pair_status(struct diff_filepair *p)
4687 {
4688         switch (p->status) {
4689         case DIFF_STATUS_UNKNOWN:
4690                 return 0;
4691         case 0:
4692                 die("internal error in diff-resolve-rename-copy");
4693         default:
4694                 return 1;
4695         }
4696 }
4697
4698 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4699 {
4700         int fmt = opt->output_format;
4701
4702         if (fmt & DIFF_FORMAT_CHECKDIFF)
4703                 diff_flush_checkdiff(p, opt);
4704         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4705                 diff_flush_raw(p, opt);
4706         else if (fmt & DIFF_FORMAT_NAME) {
4707                 const char *name_a, *name_b;
4708                 name_a = p->two->path;
4709                 name_b = NULL;
4710                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4711                 fprintf(opt->file, "%s", diff_line_prefix(opt));
4712                 write_name_quoted(name_a, opt->file, opt->line_termination);
4713         }
4714 }
4715
4716 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4717 {
4718         if (fs->mode)
4719                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4720         else
4721                 fprintf(file, " %s ", newdelete);
4722         write_name_quoted(fs->path, file, '\n');
4723 }
4724
4725
4726 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4727                 const char *line_prefix)
4728 {
4729         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4730                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4731                         p->two->mode, show_name ? ' ' : '\n');
4732                 if (show_name) {
4733                         write_name_quoted(p->two->path, file, '\n');
4734                 }
4735         }
4736 }
4737
4738 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4739                         const char *line_prefix)
4740 {
4741         char *names = pprint_rename(p->one->path, p->two->path);
4742
4743         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4744         free(names);
4745         show_mode_change(file, p, 0, line_prefix);
4746 }
4747
4748 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4749 {
4750         FILE *file = opt->file;
4751         const char *line_prefix = diff_line_prefix(opt);
4752
4753         switch(p->status) {
4754         case DIFF_STATUS_DELETED:
4755                 fputs(line_prefix, file);
4756                 show_file_mode_name(file, "delete", p->one);
4757                 break;
4758         case DIFF_STATUS_ADDED:
4759                 fputs(line_prefix, file);
4760                 show_file_mode_name(file, "create", p->two);
4761                 break;
4762         case DIFF_STATUS_COPIED:
4763                 fputs(line_prefix, file);
4764                 show_rename_copy(file, "copy", p, line_prefix);
4765                 break;
4766         case DIFF_STATUS_RENAMED:
4767                 fputs(line_prefix, file);
4768                 show_rename_copy(file, "rename", p, line_prefix);
4769                 break;
4770         default:
4771                 if (p->score) {
4772                         fprintf(file, "%s rewrite ", line_prefix);
4773                         write_name_quoted(p->two->path, file, ' ');
4774                         fprintf(file, "(%d%%)\n", similarity_index(p));
4775                 }
4776                 show_mode_change(file, p, !p->score, line_prefix);
4777                 break;
4778         }
4779 }
4780
4781 struct patch_id_t {
4782         git_SHA_CTX *ctx;
4783         int patchlen;
4784 };
4785
4786 static int remove_space(char *line, int len)
4787 {
4788         int i;
4789         char *dst = line;
4790         unsigned char c;
4791
4792         for (i = 0; i < len; i++)
4793                 if (!isspace((c = line[i])))
4794                         *dst++ = c;
4795
4796         return dst - line;
4797 }
4798
4799 static void patch_id_consume(void *priv, char *line, unsigned long len)
4800 {
4801         struct patch_id_t *data = priv;
4802         int new_len;
4803
4804         /* Ignore line numbers when computing the SHA1 of the patch */
4805         if (starts_with(line, "@@ -"))
4806                 return;
4807
4808         new_len = remove_space(line, len);
4809
4810         git_SHA1_Update(data->ctx, line, new_len);
4811         data->patchlen += new_len;
4812 }
4813
4814 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
4815 {
4816         git_SHA1_Update(ctx, str, strlen(str));
4817 }
4818
4819 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
4820 {
4821         /* large enough for 2^32 in octal */
4822         char buf[12];
4823         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
4824         git_SHA1_Update(ctx, buf, len);
4825 }
4826
4827 /* returns 0 upon success, and writes result into sha1 */
4828 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
4829 {
4830         struct diff_queue_struct *q = &diff_queued_diff;
4831         int i;
4832         git_SHA_CTX ctx;
4833         struct patch_id_t data;
4834
4835         git_SHA1_Init(&ctx);
4836         memset(&data, 0, sizeof(struct patch_id_t));
4837         data.ctx = &ctx;
4838
4839         for (i = 0; i < q->nr; i++) {
4840                 xpparam_t xpp;
4841                 xdemitconf_t xecfg;
4842                 mmfile_t mf1, mf2;
4843                 struct diff_filepair *p = q->queue[i];
4844                 int len1, len2;
4845
4846                 memset(&xpp, 0, sizeof(xpp));
4847                 memset(&xecfg, 0, sizeof(xecfg));
4848                 if (p->status == 0)
4849                         return error("internal diff status error");
4850                 if (p->status == DIFF_STATUS_UNKNOWN)
4851                         continue;
4852                 if (diff_unmodified_pair(p))
4853                         continue;
4854                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4855                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4856                         continue;
4857                 if (DIFF_PAIR_UNMERGED(p))
4858                         continue;
4859
4860                 diff_fill_oid_info(p->one);
4861                 diff_fill_oid_info(p->two);
4862
4863                 len1 = remove_space(p->one->path, strlen(p->one->path));
4864                 len2 = remove_space(p->two->path, strlen(p->two->path));
4865                 patch_id_add_string(&ctx, "diff--git");
4866                 patch_id_add_string(&ctx, "a/");
4867                 git_SHA1_Update(&ctx, p->one->path, len1);
4868                 patch_id_add_string(&ctx, "b/");
4869                 git_SHA1_Update(&ctx, p->two->path, len2);
4870
4871                 if (p->one->mode == 0) {
4872                         patch_id_add_string(&ctx, "newfilemode");
4873                         patch_id_add_mode(&ctx, p->two->mode);
4874                         patch_id_add_string(&ctx, "---/dev/null");
4875                         patch_id_add_string(&ctx, "+++b/");
4876                         git_SHA1_Update(&ctx, p->two->path, len2);
4877                 } else if (p->two->mode == 0) {
4878                         patch_id_add_string(&ctx, "deletedfilemode");
4879                         patch_id_add_mode(&ctx, p->one->mode);
4880                         patch_id_add_string(&ctx, "---a/");
4881                         git_SHA1_Update(&ctx, p->one->path, len1);
4882                         patch_id_add_string(&ctx, "+++/dev/null");
4883                 } else {
4884                         patch_id_add_string(&ctx, "---a/");
4885                         git_SHA1_Update(&ctx, p->one->path, len1);
4886                         patch_id_add_string(&ctx, "+++b/");
4887                         git_SHA1_Update(&ctx, p->two->path, len2);
4888                 }
4889
4890                 if (diff_header_only)
4891                         continue;
4892
4893                 if (fill_mmfile(&mf1, p->one) < 0 ||
4894                     fill_mmfile(&mf2, p->two) < 0)
4895                         return error("unable to read files to diff");
4896
4897                 if (diff_filespec_is_binary(p->one) ||
4898                     diff_filespec_is_binary(p->two)) {
4899                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4900                                         GIT_SHA1_HEXSZ);
4901                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4902                                         GIT_SHA1_HEXSZ);
4903                         continue;
4904                 }
4905
4906                 xpp.flags = 0;
4907                 xecfg.ctxlen = 3;
4908                 xecfg.flags = 0;
4909                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4910                                   &xpp, &xecfg))
4911                         return error("unable to generate patch-id diff for %s",
4912                                      p->one->path);
4913         }
4914
4915         git_SHA1_Final(oid->hash, &ctx);
4916         return 0;
4917 }
4918
4919 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
4920 {
4921         struct diff_queue_struct *q = &diff_queued_diff;
4922         int i;
4923         int result = diff_get_patch_id(options, oid, diff_header_only);
4924
4925         for (i = 0; i < q->nr; i++)
4926                 diff_free_filepair(q->queue[i]);
4927
4928         free(q->queue);
4929         DIFF_QUEUE_CLEAR(q);
4930
4931         return result;
4932 }
4933
4934 static int is_summary_empty(const struct diff_queue_struct *q)
4935 {
4936         int i;
4937
4938         for (i = 0; i < q->nr; i++) {
4939                 const struct diff_filepair *p = q->queue[i];
4940
4941                 switch (p->status) {
4942                 case DIFF_STATUS_DELETED:
4943                 case DIFF_STATUS_ADDED:
4944                 case DIFF_STATUS_COPIED:
4945                 case DIFF_STATUS_RENAMED:
4946                         return 0;
4947                 default:
4948                         if (p->score)
4949                                 return 0;
4950                         if (p->one->mode && p->two->mode &&
4951                             p->one->mode != p->two->mode)
4952                                 return 0;
4953                         break;
4954                 }
4955         }
4956         return 1;
4957 }
4958
4959 static const char rename_limit_warning[] =
4960 N_("inexact rename detection was skipped due to too many files.");
4961
4962 static const char degrade_cc_to_c_warning[] =
4963 N_("only found copies from modified paths due to too many files.");
4964
4965 static const char rename_limit_advice[] =
4966 N_("you may want to set your %s variable to at least "
4967    "%d and retry the command.");
4968
4969 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4970 {
4971         if (degraded_cc)
4972                 warning(_(degrade_cc_to_c_warning));
4973         else if (needed)
4974                 warning(_(rename_limit_warning));
4975         else
4976                 return;
4977         if (0 < needed && needed < 32767)
4978                 warning(_(rename_limit_advice), varname, needed);
4979 }
4980
4981 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
4982 {
4983         int i;
4984         struct diff_queue_struct *q = &diff_queued_diff;
4985
4986         if (WSEH_NEW & WS_RULE_MASK)
4987                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
4988
4989         for (i = 0; i < q->nr; i++) {
4990                 struct diff_filepair *p = q->queue[i];
4991                 if (check_pair_status(p))
4992                         diff_flush_patch(p, o);
4993         }
4994 }
4995
4996 void diff_flush(struct diff_options *options)
4997 {
4998         struct diff_queue_struct *q = &diff_queued_diff;
4999         int i, output_format = options->output_format;
5000         int separator = 0;
5001         int dirstat_by_line = 0;
5002
5003         /*
5004          * Order: raw, stat, summary, patch
5005          * or:    name/name-status/checkdiff (other bits clear)
5006          */
5007         if (!q->nr)
5008                 goto free_queue;
5009
5010         if (output_format & (DIFF_FORMAT_RAW |
5011                              DIFF_FORMAT_NAME |
5012                              DIFF_FORMAT_NAME_STATUS |
5013                              DIFF_FORMAT_CHECKDIFF)) {
5014                 for (i = 0; i < q->nr; i++) {
5015                         struct diff_filepair *p = q->queue[i];
5016                         if (check_pair_status(p))
5017                                 flush_one_pair(p, options);
5018                 }
5019                 separator++;
5020         }
5021
5022         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
5023                 dirstat_by_line = 1;
5024
5025         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5026             dirstat_by_line) {
5027                 struct diffstat_t diffstat;
5028
5029                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5030                 for (i = 0; i < q->nr; i++) {
5031                         struct diff_filepair *p = q->queue[i];
5032                         if (check_pair_status(p))
5033                                 diff_flush_stat(p, options, &diffstat);
5034                 }
5035                 if (output_format & DIFF_FORMAT_NUMSTAT)
5036                         show_numstat(&diffstat, options);
5037                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5038                         show_stats(&diffstat, options);
5039                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5040                         show_shortstats(&diffstat, options);
5041                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5042                         show_dirstat_by_line(&diffstat, options);
5043                 free_diffstat_info(&diffstat);
5044                 separator++;
5045         }
5046         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5047                 show_dirstat(options);
5048
5049         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5050                 for (i = 0; i < q->nr; i++) {
5051                         diff_summary(options, q->queue[i]);
5052                 }
5053                 separator++;
5054         }
5055
5056         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5057             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
5058             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5059                 /*
5060                  * run diff_flush_patch for the exit status. setting
5061                  * options->file to /dev/null should be safe, because we
5062                  * aren't supposed to produce any output anyway.
5063                  */
5064                 if (options->close_file)
5065                         fclose(options->file);
5066                 options->file = xfopen("/dev/null", "w");
5067                 options->close_file = 1;
5068                 for (i = 0; i < q->nr; i++) {
5069                         struct diff_filepair *p = q->queue[i];
5070                         if (check_pair_status(p))
5071                                 diff_flush_patch(p, options);
5072                         if (options->found_changes)
5073                                 break;
5074                 }
5075         }
5076
5077         if (output_format & DIFF_FORMAT_PATCH) {
5078                 if (separator) {
5079                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5080                         if (options->stat_sep) {
5081                                 /* attach patch instead of inline */
5082                                 fputs(options->stat_sep, options->file);
5083                         }
5084                 }
5085
5086                 diff_flush_patch_all_file_pairs(options);
5087         }
5088
5089         if (output_format & DIFF_FORMAT_CALLBACK)
5090                 options->format_callback(q, options, options->format_callback_data);
5091
5092         for (i = 0; i < q->nr; i++)
5093                 diff_free_filepair(q->queue[i]);
5094 free_queue:
5095         free(q->queue);
5096         DIFF_QUEUE_CLEAR(q);
5097         if (options->close_file)
5098                 fclose(options->file);
5099
5100         /*
5101          * Report the content-level differences with HAS_CHANGES;
5102          * diff_addremove/diff_change does not set the bit when
5103          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5104          */
5105         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5106                 if (options->found_changes)
5107                         DIFF_OPT_SET(options, HAS_CHANGES);
5108                 else
5109                         DIFF_OPT_CLR(options, HAS_CHANGES);
5110         }
5111 }
5112
5113 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5114 {
5115         return (((p->status == DIFF_STATUS_MODIFIED) &&
5116                  ((p->score &&
5117                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5118                   (!p->score &&
5119                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5120                 ((p->status != DIFF_STATUS_MODIFIED) &&
5121                  filter_bit_tst(p->status, options)));
5122 }
5123
5124 static void diffcore_apply_filter(struct diff_options *options)
5125 {
5126         int i;
5127         struct diff_queue_struct *q = &diff_queued_diff;
5128         struct diff_queue_struct outq;
5129
5130         DIFF_QUEUE_CLEAR(&outq);
5131
5132         if (!options->filter)
5133                 return;
5134
5135         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5136                 int found;
5137                 for (i = found = 0; !found && i < q->nr; i++) {
5138                         if (match_filter(options, q->queue[i]))
5139                                 found++;
5140                 }
5141                 if (found)
5142                         return;
5143
5144                 /* otherwise we will clear the whole queue
5145                  * by copying the empty outq at the end of this
5146                  * function, but first clear the current entries
5147                  * in the queue.
5148                  */
5149                 for (i = 0; i < q->nr; i++)
5150                         diff_free_filepair(q->queue[i]);
5151         }
5152         else {
5153                 /* Only the matching ones */
5154                 for (i = 0; i < q->nr; i++) {
5155                         struct diff_filepair *p = q->queue[i];
5156                         if (match_filter(options, p))
5157                                 diff_q(&outq, p);
5158                         else
5159                                 diff_free_filepair(p);
5160                 }
5161         }
5162         free(q->queue);
5163         *q = outq;
5164 }
5165
5166 /* Check whether two filespecs with the same mode and size are identical */
5167 static int diff_filespec_is_identical(struct diff_filespec *one,
5168                                       struct diff_filespec *two)
5169 {
5170         if (S_ISGITLINK(one->mode))
5171                 return 0;
5172         if (diff_populate_filespec(one, 0))
5173                 return 0;
5174         if (diff_populate_filespec(two, 0))
5175                 return 0;
5176         return !memcmp(one->data, two->data, one->size);
5177 }
5178
5179 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5180 {
5181         if (p->done_skip_stat_unmatch)
5182                 return p->skip_stat_unmatch_result;
5183
5184         p->done_skip_stat_unmatch = 1;
5185         p->skip_stat_unmatch_result = 0;
5186         /*
5187          * 1. Entries that come from stat info dirtiness
5188          *    always have both sides (iow, not create/delete),
5189          *    one side of the object name is unknown, with
5190          *    the same mode and size.  Keep the ones that
5191          *    do not match these criteria.  They have real
5192          *    differences.
5193          *
5194          * 2. At this point, the file is known to be modified,
5195          *    with the same mode and size, and the object
5196          *    name of one side is unknown.  Need to inspect
5197          *    the identical contents.
5198          */
5199         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5200             !DIFF_FILE_VALID(p->two) ||
5201             (p->one->oid_valid && p->two->oid_valid) ||
5202             (p->one->mode != p->two->mode) ||
5203             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5204             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5205             (p->one->size != p->two->size) ||
5206             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5207                 p->skip_stat_unmatch_result = 1;
5208         return p->skip_stat_unmatch_result;
5209 }
5210
5211 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5212 {
5213         int i;
5214         struct diff_queue_struct *q = &diff_queued_diff;
5215         struct diff_queue_struct outq;
5216         DIFF_QUEUE_CLEAR(&outq);
5217
5218         for (i = 0; i < q->nr; i++) {
5219                 struct diff_filepair *p = q->queue[i];
5220
5221                 if (diff_filespec_check_stat_unmatch(p))
5222                         diff_q(&outq, p);
5223                 else {
5224                         /*
5225                          * The caller can subtract 1 from skip_stat_unmatch
5226                          * to determine how many paths were dirty only
5227                          * due to stat info mismatch.
5228                          */
5229                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
5230                                 diffopt->skip_stat_unmatch++;
5231                         diff_free_filepair(p);
5232                 }
5233         }
5234         free(q->queue);
5235         *q = outq;
5236 }
5237
5238 static int diffnamecmp(const void *a_, const void *b_)
5239 {
5240         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5241         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5242         const char *name_a, *name_b;
5243
5244         name_a = a->one ? a->one->path : a->two->path;
5245         name_b = b->one ? b->one->path : b->two->path;
5246         return strcmp(name_a, name_b);
5247 }
5248
5249 void diffcore_fix_diff_index(struct diff_options *options)
5250 {
5251         struct diff_queue_struct *q = &diff_queued_diff;
5252         QSORT(q->queue, q->nr, diffnamecmp);
5253 }
5254
5255 void diffcore_std(struct diff_options *options)
5256 {
5257         /* NOTE please keep the following in sync with diff_tree_combined() */
5258         if (options->skip_stat_unmatch)
5259                 diffcore_skip_stat_unmatch(options);
5260         if (!options->found_follow) {
5261                 /* See try_to_follow_renames() in tree-diff.c */
5262                 if (options->break_opt != -1)
5263                         diffcore_break(options->break_opt);
5264                 if (options->detect_rename)
5265                         diffcore_rename(options);
5266                 if (options->break_opt != -1)
5267                         diffcore_merge_broken();
5268         }
5269         if (options->pickaxe)
5270                 diffcore_pickaxe(options);
5271         if (options->orderfile)
5272                 diffcore_order(options->orderfile);
5273         if (!options->found_follow)
5274                 /* See try_to_follow_renames() in tree-diff.c */
5275                 diff_resolve_rename_copy();
5276         diffcore_apply_filter(options);
5277
5278         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5279                 DIFF_OPT_SET(options, HAS_CHANGES);
5280         else
5281                 DIFF_OPT_CLR(options, HAS_CHANGES);
5282
5283         options->found_follow = 0;
5284 }
5285
5286 int diff_result_code(struct diff_options *opt, int status)
5287 {
5288         int result = 0;
5289
5290         diff_warn_rename_limit("diff.renameLimit",
5291                                opt->needed_rename_limit,
5292                                opt->degraded_cc_to_c);
5293         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5294             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5295                 return status;
5296         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5297             DIFF_OPT_TST(opt, HAS_CHANGES))
5298                 result |= 01;
5299         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5300             DIFF_OPT_TST(opt, CHECK_FAILED))
5301                 result |= 02;
5302         return result;
5303 }
5304
5305 int diff_can_quit_early(struct diff_options *opt)
5306 {
5307         return (DIFF_OPT_TST(opt, QUICK) &&
5308                 !opt->filter &&
5309                 DIFF_OPT_TST(opt, HAS_CHANGES));
5310 }
5311
5312 /*
5313  * Shall changes to this submodule be ignored?
5314  *
5315  * Submodule changes can be configured to be ignored separately for each path,
5316  * but that configuration can be overridden from the command line.
5317  */
5318 static int is_submodule_ignored(const char *path, struct diff_options *options)
5319 {
5320         int ignored = 0;
5321         unsigned orig_flags = options->flags;
5322         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5323                 set_diffopt_flags_from_submodule_config(options, path);
5324         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5325                 ignored = 1;
5326         options->flags = orig_flags;
5327         return ignored;
5328 }
5329
5330 void diff_addremove(struct diff_options *options,
5331                     int addremove, unsigned mode,
5332                     const struct object_id *oid,
5333                     int oid_valid,
5334                     const char *concatpath, unsigned dirty_submodule)
5335 {
5336         struct diff_filespec *one, *two;
5337
5338         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5339                 return;
5340
5341         /* This may look odd, but it is a preparation for
5342          * feeding "there are unchanged files which should
5343          * not produce diffs, but when you are doing copy
5344          * detection you would need them, so here they are"
5345          * entries to the diff-core.  They will be prefixed
5346          * with something like '=' or '*' (I haven't decided
5347          * which but should not make any difference).
5348          * Feeding the same new and old to diff_change()
5349          * also has the same effect.
5350          * Before the final output happens, they are pruned after
5351          * merged into rename/copy pairs as appropriate.
5352          */
5353         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5354                 addremove = (addremove == '+' ? '-' :
5355                              addremove == '-' ? '+' : addremove);
5356
5357         if (options->prefix &&
5358             strncmp(concatpath, options->prefix, options->prefix_length))
5359                 return;
5360
5361         one = alloc_filespec(concatpath);
5362         two = alloc_filespec(concatpath);
5363
5364         if (addremove != '+')
5365                 fill_filespec(one, oid, oid_valid, mode);
5366         if (addremove != '-') {
5367                 fill_filespec(two, oid, oid_valid, mode);
5368                 two->dirty_submodule = dirty_submodule;
5369         }
5370
5371         diff_queue(&diff_queued_diff, one, two);
5372         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5373                 DIFF_OPT_SET(options, HAS_CHANGES);
5374 }
5375
5376 void diff_change(struct diff_options *options,
5377                  unsigned old_mode, unsigned new_mode,
5378                  const struct object_id *old_oid,
5379                  const struct object_id *new_oid,
5380                  int old_oid_valid, int new_oid_valid,
5381                  const char *concatpath,
5382                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5383 {
5384         struct diff_filespec *one, *two;
5385         struct diff_filepair *p;
5386
5387         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5388             is_submodule_ignored(concatpath, options))
5389                 return;
5390
5391         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5392                 SWAP(old_mode, new_mode);
5393                 SWAP(old_oid, new_oid);
5394                 SWAP(old_oid_valid, new_oid_valid);
5395                 SWAP(old_dirty_submodule, new_dirty_submodule);
5396         }
5397
5398         if (options->prefix &&
5399             strncmp(concatpath, options->prefix, options->prefix_length))
5400                 return;
5401
5402         one = alloc_filespec(concatpath);
5403         two = alloc_filespec(concatpath);
5404         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5405         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5406         one->dirty_submodule = old_dirty_submodule;
5407         two->dirty_submodule = new_dirty_submodule;
5408         p = diff_queue(&diff_queued_diff, one, two);
5409
5410         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5411                 return;
5412
5413         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5414             !diff_filespec_check_stat_unmatch(p))
5415                 return;
5416
5417         DIFF_OPT_SET(options, HAS_CHANGES);
5418 }
5419
5420 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5421 {
5422         struct diff_filepair *pair;
5423         struct diff_filespec *one, *two;
5424
5425         if (options->prefix &&
5426             strncmp(path, options->prefix, options->prefix_length))
5427                 return NULL;
5428
5429         one = alloc_filespec(path);
5430         two = alloc_filespec(path);
5431         pair = diff_queue(&diff_queued_diff, one, two);
5432         pair->is_unmerged = 1;
5433         return pair;
5434 }
5435
5436 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5437                 size_t *outsize)
5438 {
5439         struct diff_tempfile *temp;
5440         const char *argv[3];
5441         const char **arg = argv;
5442         struct child_process child = CHILD_PROCESS_INIT;
5443         struct strbuf buf = STRBUF_INIT;
5444         int err = 0;
5445
5446         temp = prepare_temp_file(spec->path, spec);
5447         *arg++ = pgm;
5448         *arg++ = temp->name;
5449         *arg = NULL;
5450
5451         child.use_shell = 1;
5452         child.argv = argv;
5453         child.out = -1;
5454         if (start_command(&child)) {
5455                 remove_tempfile();
5456                 return NULL;
5457         }
5458
5459         if (strbuf_read(&buf, child.out, 0) < 0)
5460                 err = error("error reading from textconv command '%s'", pgm);
5461         close(child.out);
5462
5463         if (finish_command(&child) || err) {
5464                 strbuf_release(&buf);
5465                 remove_tempfile();
5466                 return NULL;
5467         }
5468         remove_tempfile();
5469
5470         return strbuf_detach(&buf, outsize);
5471 }
5472
5473 size_t fill_textconv(struct userdiff_driver *driver,
5474                      struct diff_filespec *df,
5475                      char **outbuf)
5476 {
5477         size_t size;
5478
5479         if (!driver) {
5480                 if (!DIFF_FILE_VALID(df)) {
5481                         *outbuf = "";
5482                         return 0;
5483                 }
5484                 if (diff_populate_filespec(df, 0))
5485                         die("unable to read files to diff");
5486                 *outbuf = df->data;
5487                 return df->size;
5488         }
5489
5490         if (!driver->textconv)
5491                 die("BUG: fill_textconv called with non-textconv driver");
5492
5493         if (driver->textconv_cache && df->oid_valid) {
5494                 *outbuf = notes_cache_get(driver->textconv_cache,
5495                                           &df->oid,
5496                                           &size);
5497                 if (*outbuf)
5498                         return size;
5499         }
5500
5501         *outbuf = run_textconv(driver->textconv, df, &size);
5502         if (!*outbuf)
5503                 die("unable to read files to diff");
5504
5505         if (driver->textconv_cache && df->oid_valid) {
5506                 /* ignore errors, as we might be in a readonly repository */
5507                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
5508                                 size);
5509                 /*
5510                  * we could save up changes and flush them all at the end,
5511                  * but we would need an extra call after all diffing is done.
5512                  * Since generating a cache entry is the slow path anyway,
5513                  * this extra overhead probably isn't a big deal.
5514                  */
5515                 notes_cache_write(driver->textconv_cache);
5516         }
5517
5518         return size;
5519 }
5520
5521 int textconv_object(const char *path,
5522                     unsigned mode,
5523                     const struct object_id *oid,
5524                     int oid_valid,
5525                     char **buf,
5526                     unsigned long *buf_size)
5527 {
5528         struct diff_filespec *df;
5529         struct userdiff_driver *textconv;
5530
5531         df = alloc_filespec(path);
5532         fill_filespec(df, oid, oid_valid, mode);
5533         textconv = get_textconv(df);
5534         if (!textconv) {
5535                 free_filespec(df);
5536                 return 0;
5537         }
5538
5539         *buf_size = fill_textconv(textconv, df, buf);
5540         free_filespec(df);
5541         return 1;
5542 }
5543
5544 void setup_diff_pager(struct diff_options *opt)
5545 {
5546         /*
5547          * If the user asked for our exit code, then either they want --quiet
5548          * or --exit-code. We should definitely not bother with a pager in the
5549          * former case, as we will generate no output. Since we still properly
5550          * report our exit code even when a pager is run, we _could_ run a
5551          * pager with --exit-code. But since we have not done so historically,
5552          * and because it is easy to find people oneline advising "git diff
5553          * --exit-code" in hooks and other scripts, we do not do so.
5554          */
5555         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5556             check_pager_config("diff") != 0)
5557                 setup_pager();
5558 }