git.oblomov.eu Git - git/blob - 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 "hashmap.h"
  20 #include "ll-merge.h"
  21 #include "string-list.h"
  22 #include "argv-array.h"
  23 #include "graph.h"
  24
  25 #ifdef NO_FAST_WORKING_DIRECTORY
  26 #define FAST_WORKING_DIRECTORY 0
  27 #else
  28 #define FAST_WORKING_DIRECTORY 1
  29 #endif
  30
  31 static int diff_detect_rename_default;
  32 static int diff_indent_heuristic = 1;
  33 static int diff_rename_limit_default = 400;
  34 static int diff_suppress_blank_empty;
  35 static int diff_use_color_default = -1;
  36 static int diff_color_moved_default;
  37 static int diff_context_default = 3;
  38 static int diff_interhunk_context_default;
  39 static const char *diff_word_regex_cfg;
  40 static const char *external_diff_cmd_cfg;
  41 static const char *diff_order_file_cfg;
  42 int diff_auto_refresh_index = 1;
  43 static int diff_mnemonic_prefix;
  44 static int diff_no_prefix;
  45 static int diff_stat_graph_width;
  46 static int diff_dirstat_permille_default = 30;
  47 static struct diff_options default_diff_options;
  48 static long diff_algorithm;
  49 static unsigned ws_error_highlight_default = WSEH_NEW;
  50
  51 static char diff_colors[][COLOR_MAXLEN] = {
  52         GIT_COLOR_RESET,
  53         GIT_COLOR_NORMAL,       /* CONTEXT */
  54         GIT_COLOR_BOLD,         /* METAINFO */
  55         GIT_COLOR_CYAN,         /* FRAGINFO */
  56         GIT_COLOR_RED,          /* OLD */
  57         GIT_COLOR_GREEN,        /* NEW */
  58         GIT_COLOR_YELLOW,       /* COMMIT */
  59         GIT_COLOR_BG_RED,       /* WHITESPACE */
  60         GIT_COLOR_NORMAL,       /* FUNCINFO */
  61         GIT_COLOR_MAGENTA,      /* OLD_MOVED */
  62         GIT_COLOR_BLUE,         /* OLD_MOVED ALTERNATIVE */
  63         GIT_COLOR_CYAN,         /* NEW_MOVED */
  64         GIT_COLOR_YELLOW,       /* NEW_MOVED ALTERNATIVE */
  65 };
  66
  67 static NORETURN void die_want_option(const char *option_name)
  68 {
  69         die(_("option '%s' requires a value"), option_name);
  70 }
  71
  72 static int parse_diff_color_slot(const char *var)
  73 {
  74         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  75                 return DIFF_CONTEXT;
  76         if (!strcasecmp(var, "meta"))
  77                 return DIFF_METAINFO;
  78         if (!strcasecmp(var, "frag"))
  79                 return DIFF_FRAGINFO;
  80         if (!strcasecmp(var, "old"))
  81                 return DIFF_FILE_OLD;
  82         if (!strcasecmp(var, "new"))
  83                 return DIFF_FILE_NEW;
  84         if (!strcasecmp(var, "commit"))
  85                 return DIFF_COMMIT;
  86         if (!strcasecmp(var, "whitespace"))
  87                 return DIFF_WHITESPACE;
  88         if (!strcasecmp(var, "func"))
  89                 return DIFF_FUNCINFO;
  90         if (!strcasecmp(var, "oldmoved"))
  91                 return DIFF_FILE_OLD_MOVED;
  92         if (!strcasecmp(var, "oldmovedalternative"))
  93                 return DIFF_FILE_OLD_MOVED_ALT;
  94         if (!strcasecmp(var, "newmoved"))
  95                 return DIFF_FILE_NEW_MOVED;
  96         if (!strcasecmp(var, "newmovedalternative"))
  97                 return DIFF_FILE_NEW_MOVED_ALT;
  98         return -1;
  99 }
 100
 101 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 102                                 struct strbuf *errmsg)
 103 {
 104         char *params_copy = xstrdup(params_string);
 105         struct string_list params = STRING_LIST_INIT_NODUP;
 106         int ret = 0;
 107         int i;
 108
 109         if (*params_copy)
 110                 string_list_split_in_place(&params, params_copy, ',', -1);
 111         for (i = 0; i < params.nr; i++) {
 112                 const char *p = params.items[i].string;
 113                 if (!strcmp(p, "changes")) {
 114                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 115                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 116                 } else if (!strcmp(p, "lines")) {
 117                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
 118                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 119                 } else if (!strcmp(p, "files")) {
 120                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 121                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 122                 } else if (!strcmp(p, "noncumulative")) {
 123                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 124                 } else if (!strcmp(p, "cumulative")) {
 125                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 126                 } else if (isdigit(*p)) {
 127                         char *end;
 128                         int permille = strtoul(p, &end, 10) * 10;
 129                         if (*end == '.' && isdigit(*++end)) {
 130                                 /* only use first digit */
 131                                 permille += *end - '0';
 132                                 /* .. and ignore any further digits */
 133                                 while (isdigit(*++end))
 134                                         ; /* nothing */
 135                         }
 136                         if (!*end)
 137                                 options->dirstat_permille = permille;
 138                         else {
 139                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 140                                             p);
 141                                 ret++;
 142                         }
 143                 } else {
 144                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 145                         ret++;
 146                 }
 147
 148         }
 149         string_list_clear(&params, 0);
 150         free(params_copy);
 151         return ret;
 152 }
 153
 154 static int parse_submodule_params(struct diff_options *options, const char *value)
 155 {
 156         if (!strcmp(value, "log"))
 157                 options->submodule_format = DIFF_SUBMODULE_LOG;
 158         else if (!strcmp(value, "short"))
 159                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 160         else if (!strcmp(value, "diff"))
 161                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 162         else
 163                 return -1;
 164         return 0;
 165 }
 166
 167 static int git_config_rename(const char *var, const char *value)
 168 {
 169         if (!value)
 170                 return DIFF_DETECT_RENAME;
 171         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 172                 return  DIFF_DETECT_COPY;
 173         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 174 }
 175
 176 long parse_algorithm_value(const char *value)
 177 {
 178         if (!value)
 179                 return -1;
 180         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 181                 return 0;
 182         else if (!strcasecmp(value, "minimal"))
 183                 return XDF_NEED_MINIMAL;
 184         else if (!strcasecmp(value, "patience"))
 185                 return XDF_PATIENCE_DIFF;
 186         else if (!strcasecmp(value, "histogram"))
 187                 return XDF_HISTOGRAM_DIFF;
 188         return -1;
 189 }
 190
 191 static int parse_one_token(const char **arg, const char *token)
 192 {
 193         const char *rest;
 194         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 195                 *arg = rest;
 196                 return 1;
 197         }
 198         return 0;
 199 }
 200
 201 static int parse_ws_error_highlight(const char *arg)
 202 {
 203         const char *orig_arg = arg;
 204         unsigned val = 0;
 205
 206         while (*arg) {
 207                 if (parse_one_token(&arg, "none"))
 208                         val = 0;
 209                 else if (parse_one_token(&arg, "default"))
 210                         val = WSEH_NEW;
 211                 else if (parse_one_token(&arg, "all"))
 212                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 213                 else if (parse_one_token(&arg, "new"))
 214                         val |= WSEH_NEW;
 215                 else if (parse_one_token(&arg, "old"))
 216                         val |= WSEH_OLD;
 217                 else if (parse_one_token(&arg, "context"))
 218                         val |= WSEH_CONTEXT;
 219                 else {
 220                         return -1 - (int)(arg - orig_arg);
 221                 }
 222                 if (*arg)
 223                         arg++;
 224         }
 225         return val;
 226 }
 227
 228 /*
 229  * These are to give UI layer defaults.
 230  * The core-level commands such as git-diff-files should
 231  * never be affected by the setting of diff.renames
 232  * the user happens to have in the configuration file.
 233  */
 234 void init_diff_ui_defaults(void)
 235 {
 236         diff_detect_rename_default = 1;
 237 }
 238
 239 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 240 {
 241         if (!strcmp(var, "diff.indentheuristic"))
 242                 diff_indent_heuristic = git_config_bool(var, value);
 243         return 0;
 244 }
 245
 246 static int parse_color_moved(const char *arg)
 247 {
 248         switch (git_parse_maybe_bool(arg)) {
 249         case 0:
 250                 return COLOR_MOVED_NO;
 251         case 1:
 252                 return COLOR_MOVED_DEFAULT;
 253         default:
 254                 break;
 255         }
 256
 257         if (!strcmp(arg, "no"))
 258                 return COLOR_MOVED_NO;
 259         else if (!strcmp(arg, "zebra"))
 260                 return COLOR_MOVED_ZEBRA;
 261         else if (!strcmp(arg, "default"))
 262                 return COLOR_MOVED_DEFAULT;
 263         else
 264                 return error(_("color moved setting must be one of 'no', 'default', 'zebra'"));
 265 }
 266
 267 int git_diff_ui_config(const char *var, const char *value, void *cb)
 268 {
 269         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 270                 diff_use_color_default = git_config_colorbool(var, value);
 271                 return 0;
 272         }
 273         if (!strcmp(var, "diff.colormoved")) {
 274                 int cm = parse_color_moved(value);
 275                 if (cm < 0)
 276                         return -1;
 277                 diff_color_moved_default = cm;
 278                 return 0;
 279         }
 280         if (!strcmp(var, "diff.context")) {
 281                 diff_context_default = git_config_int(var, value);
 282                 if (diff_context_default < 0)
 283                         return -1;
 284                 return 0;
 285         }
 286         if (!strcmp(var, "diff.interhunkcontext")) {
 287                 diff_interhunk_context_default = git_config_int(var, value);
 288                 if (diff_interhunk_context_default < 0)
 289                         return -1;
 290                 return 0;
 291         }
 292         if (!strcmp(var, "diff.renames")) {
 293                 diff_detect_rename_default = git_config_rename(var, value);
 294                 return 0;
 295         }
 296         if (!strcmp(var, "diff.autorefreshindex")) {
 297                 diff_auto_refresh_index = git_config_bool(var, value);
 298                 return 0;
 299         }
 300         if (!strcmp(var, "diff.mnemonicprefix")) {
 301                 diff_mnemonic_prefix = git_config_bool(var, value);
 302                 return 0;
 303         }
 304         if (!strcmp(var, "diff.noprefix")) {
 305                 diff_no_prefix = git_config_bool(var, value);
 306                 return 0;
 307         }
 308         if (!strcmp(var, "diff.statgraphwidth")) {
 309                 diff_stat_graph_width = git_config_int(var, value);
 310                 return 0;
 311         }
 312         if (!strcmp(var, "diff.external"))
 313                 return git_config_string(&external_diff_cmd_cfg, var, value);
 314         if (!strcmp(var, "diff.wordregex"))
 315                 return git_config_string(&diff_word_regex_cfg, var, value);
 316         if (!strcmp(var, "diff.orderfile"))
 317                 return git_config_pathname(&diff_order_file_cfg, var, value);
 318
 319         if (!strcmp(var, "diff.ignoresubmodules"))
 320                 handle_ignore_submodules_arg(&default_diff_options, value);
 321
 322         if (!strcmp(var, "diff.submodule")) {
 323                 if (parse_submodule_params(&default_diff_options, value))
 324                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 325                                 value);
 326                 return 0;
 327         }
 328
 329         if (!strcmp(var, "diff.algorithm")) {
 330                 diff_algorithm = parse_algorithm_value(value);
 331                 if (diff_algorithm < 0)
 332                         return -1;
 333                 return 0;
 334         }
 335
 336         if (!strcmp(var, "diff.wserrorhighlight")) {
 337                 int val = parse_ws_error_highlight(value);
 338                 if (val < 0)
 339                         return -1;
 340                 ws_error_highlight_default = val;
 341                 return 0;
 342         }
 343
 344         if (git_color_config(var, value, cb) < 0)
 345                 return -1;
 346
 347         return git_diff_basic_config(var, value, cb);
 348 }
 349
 350 int git_diff_basic_config(const char *var, const char *value, void *cb)
 351 {
 352         const char *name;
 353
 354         if (!strcmp(var, "diff.renamelimit")) {
 355                 diff_rename_limit_default = git_config_int(var, value);
 356                 return 0;
 357         }
 358
 359         if (userdiff_config(var, value) < 0)
 360                 return -1;
 361
 362         if (skip_prefix(var, "diff.color.", &name) ||
 363             skip_prefix(var, "color.diff.", &name)) {
 364                 int slot = parse_diff_color_slot(name);
 365                 if (slot < 0)
 366                         return 0;
 367                 if (!value)
 368                         return config_error_nonbool(var);
 369                 return color_parse(value, diff_colors[slot]);
 370         }
 371
 372         /* like GNU diff's --suppress-blank-empty option  */
 373         if (!strcmp(var, "diff.suppressblankempty") ||
 374                         /* for backwards compatibility */
 375                         !strcmp(var, "diff.suppress-blank-empty")) {
 376                 diff_suppress_blank_empty = git_config_bool(var, value);
 377                 return 0;
 378         }
 379
 380         if (!strcmp(var, "diff.dirstat")) {
 381                 struct strbuf errmsg = STRBUF_INIT;
 382                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 383                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 384                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 385                                 errmsg.buf);
 386                 strbuf_release(&errmsg);
 387                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 388                 return 0;
 389         }
 390
 391         if (starts_with(var, "submodule."))
 392                 return parse_submodule_config_option(var, value);
 393
 394         if (git_diff_heuristic_config(var, value, cb) < 0)
 395                 return -1;
 396
 397         return git_default_config(var, value, cb);
 398 }
 399
 400 static char *quote_two(const char *one, const char *two)
 401 {
 402         int need_one = quote_c_style(one, NULL, NULL, 1);
 403         int need_two = quote_c_style(two, NULL, NULL, 1);
 404         struct strbuf res = STRBUF_INIT;
 405
 406         if (need_one + need_two) {
 407                 strbuf_addch(&res, '"');
 408                 quote_c_style(one, &res, NULL, 1);
 409                 quote_c_style(two, &res, NULL, 1);
 410                 strbuf_addch(&res, '"');
 411         } else {
 412                 strbuf_addstr(&res, one);
 413                 strbuf_addstr(&res, two);
 414         }
 415         return strbuf_detach(&res, NULL);
 416 }
 417
 418 static const char *external_diff(void)
 419 {
 420         static const char *external_diff_cmd = NULL;
 421         static int done_preparing = 0;
 422
 423         if (done_preparing)
 424                 return external_diff_cmd;
 425         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 426         if (!external_diff_cmd)
 427                 external_diff_cmd = external_diff_cmd_cfg;
 428         done_preparing = 1;
 429         return external_diff_cmd;
 430 }
 431
 432 /*
 433  * Keep track of files used for diffing. Sometimes such an entry
 434  * refers to a temporary file, sometimes to an existing file, and
 435  * sometimes to "/dev/null".
 436  */
 437 static struct diff_tempfile {
 438         /*
 439          * filename external diff should read from, or NULL if this
 440          * entry is currently not in use:
 441          */
 442         const char *name;
 443
 444         char hex[GIT_MAX_HEXSZ + 1];
 445         char mode[10];
 446
 447         /*
 448          * If this diff_tempfile instance refers to a temporary file,
 449          * this tempfile object is used to manage its lifetime.
 450          */
 451         struct tempfile tempfile;
 452 } diff_temp[2];
 453
 454 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 455
 456 struct emit_callback {
 457         int color_diff;
 458         unsigned ws_rule;
 459         int blank_at_eof_in_preimage;
 460         int blank_at_eof_in_postimage;
 461         int lno_in_preimage;
 462         int lno_in_postimage;
 463         sane_truncate_fn truncate;
 464         const char **label_path;
 465         struct diff_words_data *diff_words;
 466         struct diff_options *opt;
 467         struct strbuf *header;
 468 };
 469
 470 static int count_lines(const char *data, int size)
 471 {
 472         int count, ch, completely_empty = 1, nl_just_seen = 0;
 473         count = 0;
 474         while (0 < size--) {
 475                 ch = *data++;
 476                 if (ch == '\n') {
 477                         count++;
 478                         nl_just_seen = 1;
 479                         completely_empty = 0;
 480                 }
 481                 else {
 482                         nl_just_seen = 0;
 483                         completely_empty = 0;
 484                 }
 485         }
 486         if (completely_empty)
 487                 return 0;
 488         if (!nl_just_seen)
 489                 count++; /* no trailing newline */
 490         return count;
 491 }
 492
 493 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 494 {
 495         if (!DIFF_FILE_VALID(one)) {
 496                 mf->ptr = (char *)""; /* does not matter */
 497                 mf->size = 0;
 498                 return 0;
 499         }
 500         else if (diff_populate_filespec(one, 0))
 501                 return -1;
 502
 503         mf->ptr = one->data;
 504         mf->size = one->size;
 505         return 0;
 506 }
 507
 508 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 509 static unsigned long diff_filespec_size(struct diff_filespec *one)
 510 {
 511         if (!DIFF_FILE_VALID(one))
 512                 return 0;
 513         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 514         return one->size;
 515 }
 516
 517 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 518 {
 519         char *ptr = mf->ptr;
 520         long size = mf->size;
 521         int cnt = 0;
 522
 523         if (!size)
 524                 return cnt;
 525         ptr += size - 1; /* pointing at the very end */
 526         if (*ptr != '\n')
 527                 ; /* incomplete line */
 528         else
 529                 ptr--; /* skip the last LF */
 530         while (mf->ptr < ptr) {
 531                 char *prev_eol;
 532                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 533                         if (*prev_eol == '\n')
 534                                 break;
 535                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 536                         break;
 537                 cnt++;
 538                 ptr = prev_eol - 1;
 539         }
 540         return cnt;
 541 }
 542
 543 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 544                                struct emit_callback *ecbdata)
 545 {
 546         int l1, l2, at;
 547         unsigned ws_rule = ecbdata->ws_rule;
 548         l1 = count_trailing_blank(mf1, ws_rule);
 549         l2 = count_trailing_blank(mf2, ws_rule);
 550         if (l2 <= l1) {
 551                 ecbdata->blank_at_eof_in_preimage = 0;
 552                 ecbdata->blank_at_eof_in_postimage = 0;
 553                 return;
 554         }
 555         at = count_lines(mf1->ptr, mf1->size);
 556         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 557
 558         at = count_lines(mf2->ptr, mf2->size);
 559         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 560 }
 561
 562 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 563                         int first, const char *line, int len)
 564 {
 565         int has_trailing_newline, has_trailing_carriage_return;
 566         int nofirst;
 567         FILE *file = o->file;
 568
 569         fputs(diff_line_prefix(o), file);
 570
 571         if (len == 0) {
 572                 has_trailing_newline = (first == '\n');
 573                 has_trailing_carriage_return = (!has_trailing_newline &&
 574                                                 (first == '\r'));
 575                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 576         } else {
 577                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 578                 if (has_trailing_newline)
 579                         len--;
 580                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 581                 if (has_trailing_carriage_return)
 582                         len--;
 583                 nofirst = 0;
 584         }
 585
 586         if (len || !nofirst) {
 587                 fputs(set, file);
 588                 if (!nofirst)
 589                         fputc(first, file);
 590                 fwrite(line, len, 1, file);
 591                 fputs(reset, file);
 592         }
 593         if (has_trailing_carriage_return)
 594                 fputc('\r', file);
 595         if (has_trailing_newline)
 596                 fputc('\n', file);
 597 }
 598
 599 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 600                       const char *line, int len)
 601 {
 602         emit_line_0(o, set, reset, line[0], line+1, len-1);
 603 }
 604
 605 enum diff_symbol {
 606         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 607         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 608         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 609         DIFF_SYMBOL_BINARY_DIFF_BODY,
 610         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 611         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 612         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 613         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 614         DIFF_SYMBOL_STATS_LINE,
 615         DIFF_SYMBOL_WORD_DIFF,
 616         DIFF_SYMBOL_STAT_SEP,
 617         DIFF_SYMBOL_SUMMARY,
 618         DIFF_SYMBOL_SUBMODULE_ADD,
 619         DIFF_SYMBOL_SUBMODULE_DEL,
 620         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 621         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 622         DIFF_SYMBOL_SUBMODULE_HEADER,
 623         DIFF_SYMBOL_SUBMODULE_ERROR,
 624         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 625         DIFF_SYMBOL_REWRITE_DIFF,
 626         DIFF_SYMBOL_BINARY_FILES,
 627         DIFF_SYMBOL_HEADER,
 628         DIFF_SYMBOL_FILEPAIR_PLUS,
 629         DIFF_SYMBOL_FILEPAIR_MINUS,
 630         DIFF_SYMBOL_WORDS_PORCELAIN,
 631         DIFF_SYMBOL_WORDS,
 632         DIFF_SYMBOL_CONTEXT,
 633         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 634         DIFF_SYMBOL_PLUS,
 635         DIFF_SYMBOL_MINUS,
 636         DIFF_SYMBOL_NO_LF_EOF,
 637         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 638         DIFF_SYMBOL_CONTEXT_MARKER,
 639         DIFF_SYMBOL_SEPARATOR
 640 };
 641 /*
 642  * Flags for content lines:
 643  * 0..12 are whitespace rules
 644  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 645  * 16 is marking if the line is blank at EOF
 646  */
 647 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 648 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 649 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 650 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 651
 652 /*
 653  * This struct is used when we need to buffer the output of the diff output.
 654  *
 655  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 656  * into the pre/post image file. This pointer could be a union with the
 657  * line pointer. By storing an offset into the file instead of the literal line,
 658  * we can decrease the memory footprint for the buffered output. At first we
 659  * may want to only have indirection for the content lines, but we could also
 660  * enhance the state for emitting prefabricated lines, e.g. the similarity
 661  * score line or hunk/file headers would only need to store a number or path
 662  * and then the output can be constructed later on depending on state.
 663  */
 664 struct emitted_diff_symbol {
 665         const char *line;
 666         int len;
 667         int flags;
 668         enum diff_symbol s;
 669 };
 670 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 671
 672 struct emitted_diff_symbols {
 673         struct emitted_diff_symbol *buf;
 674         int nr, alloc;
 675 };
 676 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 677
 678 static void append_emitted_diff_symbol(struct diff_options *o,
 679                                        struct emitted_diff_symbol *e)
 680 {
 681         struct emitted_diff_symbol *f;
 682
 683         ALLOC_GROW(o->emitted_symbols->buf,
 684                    o->emitted_symbols->nr + 1,
 685                    o->emitted_symbols->alloc);
 686         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 687
 688         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 689         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 690 }
 691
 692 struct moved_entry {
 693         struct hashmap_entry ent;
 694         const struct emitted_diff_symbol *es;
 695         struct moved_entry *next_line;
 696 };
 697
 698 static int next_byte(const char **cp, const char **endp,
 699                      const struct diff_options *diffopt)
 700 {
 701         int retval;
 702
 703         if (*cp > *endp)
 704                 return -1;
 705
 706         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_CHANGE)) {
 707                 while (*cp < *endp && isspace(**cp))
 708                         (*cp)++;
 709                 /*
 710                  * After skipping a couple of whitespaces, we still have to
 711                  * account for one space.
 712                  */
 713                 return (int)' ';
 714         }
 715
 716         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE)) {
 717                 while (*cp < *endp && isspace(**cp))
 718                         (*cp)++;
 719                 /* return the first non-ws character via the usual below */
 720         }
 721
 722         retval = (unsigned char)(**cp);
 723         (*cp)++;
 724         return retval;
 725 }
 726
 727 static int moved_entry_cmp(const struct diff_options *diffopt,
 728                            const struct moved_entry *a,
 729                            const struct moved_entry *b,
 730                            const void *keydata)
 731 {
 732         const char *ap = a->es->line, *ae = a->es->line + a->es->len;
 733         const char *bp = b->es->line, *be = b->es->line + b->es->len;
 734
 735         if (!(diffopt->xdl_opts & XDF_WHITESPACE_FLAGS))
 736                 return a->es->len != b->es->len  || memcmp(ap, bp, a->es->len);
 737
 738         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_AT_EOL)) {
 739                 while (ae > ap && isspace(*ae))
 740                         ae--;
 741                 while (be > bp && isspace(*be))
 742                         be--;
 743         }
 744
 745         while (1) {
 746                 int ca, cb;
 747                 ca = next_byte(&ap, &ae, diffopt);
 748                 cb = next_byte(&bp, &be, diffopt);
 749                 if (ca != cb)
 750                         return 1;
 751                 if (ca < 0)
 752                         return 0;
 753         }
 754 }
 755
 756 static unsigned get_string_hash(struct emitted_diff_symbol *es, struct diff_options *o)
 757 {
 758         if (o->xdl_opts & XDF_WHITESPACE_FLAGS) {
 759                 static struct strbuf sb = STRBUF_INIT;
 760                 const char *ap = es->line, *ae = es->line + es->len;
 761                 int c;
 762
 763                 strbuf_reset(&sb);
 764                 while (ae > ap && isspace(*ae))
 765                         ae--;
 766                 while ((c = next_byte(&ap, &ae, o)) > 0)
 767                         strbuf_addch(&sb, c);
 768
 769                 return memhash(sb.buf, sb.len);
 770         } else {
 771                 return memhash(es->line, es->len);
 772         }
 773 }
 774
 775 static struct moved_entry *prepare_entry(struct diff_options *o,
 776                                          int line_no)
 777 {
 778         struct moved_entry *ret = xmalloc(sizeof(*ret));
 779         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 780
 781         ret->ent.hash = get_string_hash(l, o);
 782         ret->es = l;
 783         ret->next_line = NULL;
 784
 785         return ret;
 786 }
 787
 788 static void add_lines_to_move_detection(struct diff_options *o,
 789                                         struct hashmap *add_lines,
 790                                         struct hashmap *del_lines)
 791 {
 792         struct moved_entry *prev_line = NULL;
 793
 794         int n;
 795         for (n = 0; n < o->emitted_symbols->nr; n++) {
 796                 struct hashmap *hm;
 797                 struct moved_entry *key;
 798
 799                 switch (o->emitted_symbols->buf[n].s) {
 800                 case DIFF_SYMBOL_PLUS:
 801                         hm = add_lines;
 802                         break;
 803                 case DIFF_SYMBOL_MINUS:
 804                         hm = del_lines;
 805                         break;
 806                 default:
 807                         prev_line = NULL;
 808                         continue;
 809                 }
 810
 811                 key = prepare_entry(o, n);
 812                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 813                         prev_line->next_line = key;
 814
 815                 hashmap_add(hm, key);
 816                 prev_line = key;
 817         }
 818 }
 819
 820 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 821                                          int pmb_nr)
 822 {
 823         int lp, rp;
 824
 825         /* Shrink the set of potential block to the remaining running */
 826         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 827                 while (lp < pmb_nr && pmb[lp])
 828                         lp++;
 829                 /* lp points at the first NULL now */
 830
 831                 while (rp > -1 && !pmb[rp])
 832                         rp--;
 833                 /* rp points at the last non-NULL */
 834
 835                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 836                         pmb[lp] = pmb[rp];
 837                         pmb[rp] = NULL;
 838                         rp--;
 839                         lp++;
 840                 }
 841         }
 842
 843         /* Remember the number of running sets */
 844         return rp + 1;
 845 }
 846
 847 /* Find blocks of moved code, delegate actual coloring decision to helper */
 848 static void mark_color_as_moved(struct diff_options *o,
 849                                 struct hashmap *add_lines,
 850                                 struct hashmap *del_lines)
 851 {
 852         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 853         int pmb_nr = 0, pmb_alloc = 0;
 854         int n, flipped_block = 1, block_length = 0;
 855
 856
 857         for (n = 0; n < o->emitted_symbols->nr; n++) {
 858                 struct hashmap *hm = NULL;
 859                 struct moved_entry *key;
 860                 struct moved_entry *match = NULL;
 861                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 862                 int i;
 863
 864                 switch (l->s) {
 865                 case DIFF_SYMBOL_PLUS:
 866                         hm = del_lines;
 867                         key = prepare_entry(o, n);
 868                         match = hashmap_get(hm, key, o);
 869                         free(key);
 870                         break;
 871                 case DIFF_SYMBOL_MINUS:
 872                         hm = add_lines;
 873                         key = prepare_entry(o, n);
 874                         match = hashmap_get(hm, key, o);
 875                         free(key);
 876                         break;
 877                 default:
 878                         flipped_block = 1;
 879                 }
 880
 881                 if (!match) {
 882                         if (block_length < COLOR_MOVED_MIN_BLOCK_LENGTH) {
 883                                 for (i = 0; i < block_length + 1; i++) {
 884                                         l = &o->emitted_symbols->buf[n - i];
 885                                         l->flags &= ~DIFF_SYMBOL_MOVED_LINE;
 886                                 }
 887                         }
 888                         pmb_nr = 0;
 889                         block_length = 0;
 890                         continue;
 891                 }
 892
 893                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 894                 block_length++;
 895
 896                 /* Check any potential block runs, advance each or nullify */
 897                 for (i = 0; i < pmb_nr; i++) {
 898                         struct moved_entry *p = pmb[i];
 899                         struct moved_entry *pnext = (p && p->next_line) ?
 900                                         p->next_line : NULL;
 901                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 902                                 pmb[i] = p->next_line;
 903                         } else {
 904                                 pmb[i] = NULL;
 905                         }
 906                 }
 907
 908                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 909
 910                 if (pmb_nr == 0) {
 911                         /*
 912                          * The current line is the start of a new block.
 913                          * Setup the set of potential blocks.
 914                          */
 915                         for (; match; match = hashmap_get_next(hm, match)) {
 916                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 917                                 pmb[pmb_nr++] = match;
 918                         }
 919
 920                         flipped_block = (flipped_block + 1) % 2;
 921                 }
 922
 923                 if (flipped_block)
 924                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 925         }
 926
 927         free(pmb);
 928 }
 929
 930 static void emit_line_ws_markup(struct diff_options *o,
 931                                 const char *set, const char *reset,
 932                                 const char *line, int len, char sign,
 933                                 unsigned ws_rule, int blank_at_eof)
 934 {
 935         const char *ws = NULL;
 936
 937         if (o->ws_error_highlight & ws_rule) {
 938                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 939                 if (!*ws)
 940                         ws = NULL;
 941         }
 942
 943         if (!ws)
 944                 emit_line_0(o, set, reset, sign, line, len);
 945         else if (blank_at_eof)
 946                 /* Blank line at EOF - paint '+' as well */
 947                 emit_line_0(o, ws, reset, sign, line, len);
 948         else {
 949                 /* Emit just the prefix, then the rest. */
 950                 emit_line_0(o, set, reset, sign, "", 0);
 951                 ws_check_emit(line, len, ws_rule,
 952                               o->file, set, reset, ws);
 953         }
 954 }
 955
 956 static void emit_diff_symbol_from_struct(struct diff_options *o,
 957                                          struct emitted_diff_symbol *eds)
 958 {
 959         static const char *nneof = " No newline at end of file\n";
 960         const char *context, *reset, *set, *meta, *fraginfo;
 961         struct strbuf sb = STRBUF_INIT;
 962
 963         enum diff_symbol s = eds->s;
 964         const char *line = eds->line;
 965         int len = eds->len;
 966         unsigned flags = eds->flags;
 967
 968         switch (s) {
 969         case DIFF_SYMBOL_NO_LF_EOF:
 970                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 971                 reset = diff_get_color_opt(o, DIFF_RESET);
 972                 putc('\n', o->file);
 973                 emit_line_0(o, context, reset, '\\',
 974                             nneof, strlen(nneof));
 975                 break;
 976         case DIFF_SYMBOL_SUBMODULE_HEADER:
 977         case DIFF_SYMBOL_SUBMODULE_ERROR:
 978         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
 979         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
 980         case DIFF_SYMBOL_SUMMARY:
 981         case DIFF_SYMBOL_STATS_LINE:
 982         case DIFF_SYMBOL_BINARY_DIFF_BODY:
 983         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
 984                 emit_line(o, "", "", line, len);
 985                 break;
 986         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
 987         case DIFF_SYMBOL_CONTEXT_MARKER:
 988                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 989                 reset = diff_get_color_opt(o, DIFF_RESET);
 990                 emit_line(o, context, reset, line, len);
 991                 break;
 992         case DIFF_SYMBOL_SEPARATOR:
 993                 fprintf(o->file, "%s%c",
 994                         diff_line_prefix(o),
 995                         o->line_termination);
 996                 break;
 997         case DIFF_SYMBOL_CONTEXT:
 998                 set = diff_get_color_opt(o, DIFF_CONTEXT);
 999                 reset = diff_get_color_opt(o, DIFF_RESET);
1000                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1001                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1002                 break;
1003         case DIFF_SYMBOL_PLUS:
1004                 if (flags & DIFF_SYMBOL_MOVED_LINE_ALT)
1005                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1006                 else if (flags & DIFF_SYMBOL_MOVED_LINE)
1007                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1008                 else
1009                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1010                 reset = diff_get_color_opt(o, DIFF_RESET);
1011                 emit_line_ws_markup(o, set, reset, line, len, '+',
1012                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1013                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1014                 break;
1015         case DIFF_SYMBOL_MINUS:
1016                 if (flags & DIFF_SYMBOL_MOVED_LINE_ALT)
1017                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1018                 else if (flags & DIFF_SYMBOL_MOVED_LINE)
1019                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1020                 else
1021                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1022                 reset = diff_get_color_opt(o, DIFF_RESET);
1023                 emit_line_ws_markup(o, set, reset, line, len, '-',
1024                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1025                 break;
1026         case DIFF_SYMBOL_WORDS_PORCELAIN:
1027                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1028                 reset = diff_get_color_opt(o, DIFF_RESET);
1029                 emit_line(o, context, reset, line, len);
1030                 fputs("~\n", o->file);
1031                 break;
1032         case DIFF_SYMBOL_WORDS:
1033                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1034                 reset = diff_get_color_opt(o, DIFF_RESET);
1035                 /*
1036                  * Skip the prefix character, if any.  With
1037                  * diff_suppress_blank_empty, there may be
1038                  * none.
1039                  */
1040                 if (line[0] != '\n') {
1041                         line++;
1042                         len--;
1043                 }
1044                 emit_line(o, context, reset, line, len);
1045                 break;
1046         case DIFF_SYMBOL_FILEPAIR_PLUS:
1047                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1048                 reset = diff_get_color_opt(o, DIFF_RESET);
1049                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1050                         line, reset,
1051                         strchr(line, ' ') ? "\t" : "");
1052                 break;
1053         case DIFF_SYMBOL_FILEPAIR_MINUS:
1054                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1055                 reset = diff_get_color_opt(o, DIFF_RESET);
1056                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1057                         line, reset,
1058                         strchr(line, ' ') ? "\t" : "");
1059                 break;
1060         case DIFF_SYMBOL_BINARY_FILES:
1061         case DIFF_SYMBOL_HEADER:
1062                 fprintf(o->file, "%s", line);
1063                 break;
1064         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1065                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1066                 break;
1067         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1068                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1069                 break;
1070         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1071                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1072                 break;
1073         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1074                 fputs(diff_line_prefix(o), o->file);
1075                 fputc('\n', o->file);
1076                 break;
1077         case DIFF_SYMBOL_REWRITE_DIFF:
1078                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1079                 reset = diff_get_color_opt(o, DIFF_RESET);
1080                 emit_line(o, fraginfo, reset, line, len);
1081                 break;
1082         case DIFF_SYMBOL_SUBMODULE_ADD:
1083                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1084                 reset = diff_get_color_opt(o, DIFF_RESET);
1085                 emit_line(o, set, reset, line, len);
1086                 break;
1087         case DIFF_SYMBOL_SUBMODULE_DEL:
1088                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1089                 reset = diff_get_color_opt(o, DIFF_RESET);
1090                 emit_line(o, set, reset, line, len);
1091                 break;
1092         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1093                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1094                         diff_line_prefix(o), line);
1095                 break;
1096         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1097                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1098                         diff_line_prefix(o), line);
1099                 break;
1100         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1101                 emit_line(o, "", "", " 0 files changed\n",
1102                           strlen(" 0 files changed\n"));
1103                 break;
1104         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1105                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1106                 break;
1107         case DIFF_SYMBOL_WORD_DIFF:
1108                 fprintf(o->file, "%.*s", len, line);
1109                 break;
1110         case DIFF_SYMBOL_STAT_SEP:
1111                 fputs(o->stat_sep, o->file);
1112                 break;
1113         default:
1114                 die("BUG: unknown diff symbol");
1115         }
1116         strbuf_release(&sb);
1117 }
1118
1119 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1120                              const char *line, int len, unsigned flags)
1121 {
1122         struct emitted_diff_symbol e = {line, len, flags, s};
1123
1124         if (o->emitted_symbols)
1125                 append_emitted_diff_symbol(o, &e);
1126         else
1127                 emit_diff_symbol_from_struct(o, &e);
1128 }
1129
1130 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1131 {
1132         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1133 }
1134
1135 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1136 {
1137         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1138 }
1139
1140 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1141 {
1142         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1143                          path, strlen(path), 0);
1144 }
1145
1146 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1147 {
1148         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1149                          path, strlen(path), 0);
1150 }
1151
1152 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1153 {
1154         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1155                          header, strlen(header), 0);
1156 }
1157
1158 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1159 {
1160         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1161 }
1162
1163 void diff_emit_submodule_pipethrough(struct diff_options *o,
1164                                      const char *line, int len)
1165 {
1166         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1167 }
1168
1169 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1170 {
1171         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1172               ecbdata->blank_at_eof_in_preimage &&
1173               ecbdata->blank_at_eof_in_postimage &&
1174               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1175               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1176                 return 0;
1177         return ws_blank_line(line, len, ecbdata->ws_rule);
1178 }
1179
1180 static void emit_add_line(const char *reset,
1181                           struct emit_callback *ecbdata,
1182                           const char *line, int len)
1183 {
1184         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1185         if (new_blank_line_at_eof(ecbdata, line, len))
1186                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1187
1188         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1189 }
1190
1191 static void emit_del_line(const char *reset,
1192                           struct emit_callback *ecbdata,
1193                           const char *line, int len)
1194 {
1195         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1196         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1197 }
1198
1199 static void emit_context_line(const char *reset,
1200                               struct emit_callback *ecbdata,
1201                               const char *line, int len)
1202 {
1203         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1204         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1205 }
1206
1207 static void emit_hunk_header(struct emit_callback *ecbdata,
1208                              const char *line, int len)
1209 {
1210         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1211         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1212         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1213         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1214         static const char atat[2] = { '@', '@' };
1215         const char *cp, *ep;
1216         struct strbuf msgbuf = STRBUF_INIT;
1217         int org_len = len;
1218         int i = 1;
1219
1220         /*
1221          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1222          * it always is at least 10 bytes long.
1223          */
1224         if (len < 10 ||
1225             memcmp(line, atat, 2) ||
1226             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1227                 emit_diff_symbol(ecbdata->opt,
1228                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1229                 return;
1230         }
1231         ep += 2; /* skip over @@ */
1232
1233         /* The hunk header in fraginfo color */
1234         strbuf_addstr(&msgbuf, frag);
1235         strbuf_add(&msgbuf, line, ep - line);
1236         strbuf_addstr(&msgbuf, reset);
1237
1238         /*
1239          * trailing "\r\n"
1240          */
1241         for ( ; i < 3; i++)
1242                 if (line[len - i] == '\r' || line[len - i] == '\n')
1243                         len--;
1244
1245         /* blank before the func header */
1246         for (cp = ep; ep - line < len; ep++)
1247                 if (*ep != ' ' && *ep != '\t')
1248                         break;
1249         if (ep != cp) {
1250                 strbuf_addstr(&msgbuf, context);
1251                 strbuf_add(&msgbuf, cp, ep - cp);
1252                 strbuf_addstr(&msgbuf, reset);
1253         }
1254
1255         if (ep < line + len) {
1256                 strbuf_addstr(&msgbuf, func);
1257                 strbuf_add(&msgbuf, ep, line + len - ep);
1258                 strbuf_addstr(&msgbuf, reset);
1259         }
1260
1261         strbuf_add(&msgbuf, line + len, org_len - len);
1262         strbuf_complete_line(&msgbuf);
1263         emit_diff_symbol(ecbdata->opt,
1264                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1265         strbuf_release(&msgbuf);
1266 }
1267
1268 static struct diff_tempfile *claim_diff_tempfile(void) {
1269         int i;
1270         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1271                 if (!diff_temp[i].name)
1272                         return diff_temp + i;
1273         die("BUG: diff is failing to clean up its tempfiles");
1274 }
1275
1276 static void remove_tempfile(void)
1277 {
1278         int i;
1279         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1280                 if (is_tempfile_active(&diff_temp[i].tempfile))
1281                         delete_tempfile(&diff_temp[i].tempfile);
1282                 diff_temp[i].name = NULL;
1283         }
1284 }
1285
1286 static void add_line_count(struct strbuf *out, int count)
1287 {
1288         switch (count) {
1289         case 0:
1290                 strbuf_addstr(out, "0,0");
1291                 break;
1292         case 1:
1293                 strbuf_addstr(out, "1");
1294                 break;
1295         default:
1296                 strbuf_addf(out, "1,%d", count);
1297                 break;
1298         }
1299 }
1300
1301 static void emit_rewrite_lines(struct emit_callback *ecb,
1302                                int prefix, const char *data, int size)
1303 {
1304         const char *endp = NULL;
1305         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1306
1307         while (0 < size) {
1308                 int len;
1309
1310                 endp = memchr(data, '\n', size);
1311                 len = endp ? (endp - data + 1) : size;
1312                 if (prefix != '+') {
1313                         ecb->lno_in_preimage++;
1314                         emit_del_line(reset, ecb, data, len);
1315                 } else {
1316                         ecb->lno_in_postimage++;
1317                         emit_add_line(reset, ecb, data, len);
1318                 }
1319                 size -= len;
1320                 data += len;
1321         }
1322         if (!endp)
1323                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1324 }
1325
1326 static void emit_rewrite_diff(const char *name_a,
1327                               const char *name_b,
1328                               struct diff_filespec *one,
1329                               struct diff_filespec *two,
1330                               struct userdiff_driver *textconv_one,
1331                               struct userdiff_driver *textconv_two,
1332                               struct diff_options *o)
1333 {
1334         int lc_a, lc_b;
1335         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1336         const char *a_prefix, *b_prefix;
1337         char *data_one, *data_two;
1338         size_t size_one, size_two;
1339         struct emit_callback ecbdata;
1340         struct strbuf out = STRBUF_INIT;
1341
1342         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
1343                 a_prefix = o->b_prefix;
1344                 b_prefix = o->a_prefix;
1345         } else {
1346                 a_prefix = o->a_prefix;
1347                 b_prefix = o->b_prefix;
1348         }
1349
1350         name_a += (*name_a == '/');
1351         name_b += (*name_b == '/');
1352
1353         strbuf_reset(&a_name);
1354         strbuf_reset(&b_name);
1355         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1356         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1357
1358         size_one = fill_textconv(textconv_one, one, &data_one);
1359         size_two = fill_textconv(textconv_two, two, &data_two);
1360
1361         memset(&ecbdata, 0, sizeof(ecbdata));
1362         ecbdata.color_diff = want_color(o->use_color);
1363         ecbdata.ws_rule = whitespace_rule(name_b);
1364         ecbdata.opt = o;
1365         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1366                 mmfile_t mf1, mf2;
1367                 mf1.ptr = (char *)data_one;
1368                 mf2.ptr = (char *)data_two;
1369                 mf1.size = size_one;
1370                 mf2.size = size_two;
1371                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1372         }
1373         ecbdata.lno_in_preimage = 1;
1374         ecbdata.lno_in_postimage = 1;
1375
1376         lc_a = count_lines(data_one, size_one);
1377         lc_b = count_lines(data_two, size_two);
1378
1379         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1380                          a_name.buf, a_name.len, 0);
1381         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1382                          b_name.buf, b_name.len, 0);
1383
1384         strbuf_addstr(&out, "@@ -");
1385         if (!o->irreversible_delete)
1386                 add_line_count(&out, lc_a);
1387         else
1388                 strbuf_addstr(&out, "?,?");
1389         strbuf_addstr(&out, " +");
1390         add_line_count(&out, lc_b);
1391         strbuf_addstr(&out, " @@\n");
1392         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1393         strbuf_release(&out);
1394
1395         if (lc_a && !o->irreversible_delete)
1396                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1397         if (lc_b)
1398                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1399         if (textconv_one)
1400                 free((char *)data_one);
1401         if (textconv_two)
1402                 free((char *)data_two);
1403 }
1404
1405 struct diff_words_buffer {
1406         mmfile_t text;
1407         long alloc;
1408         struct diff_words_orig {
1409                 const char *begin, *end;
1410         } *orig;
1411         int orig_nr, orig_alloc;
1412 };
1413
1414 static void diff_words_append(char *line, unsigned long len,
1415                 struct diff_words_buffer *buffer)
1416 {
1417         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1418         line++;
1419         len--;
1420         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1421         buffer->text.size += len;
1422         buffer->text.ptr[buffer->text.size] = '\0';
1423 }
1424
1425 struct diff_words_style_elem {
1426         const char *prefix;
1427         const char *suffix;
1428         const char *color; /* NULL; filled in by the setup code if
1429                             * color is enabled */
1430 };
1431
1432 struct diff_words_style {
1433         enum diff_words_type type;
1434         struct diff_words_style_elem new, old, ctx;
1435         const char *newline;
1436 };
1437
1438 static struct diff_words_style diff_words_styles[] = {
1439         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1440         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1441         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1442 };
1443
1444 struct diff_words_data {
1445         struct diff_words_buffer minus, plus;
1446         const char *current_plus;
1447         int last_minus;
1448         struct diff_options *opt;
1449         regex_t *word_regex;
1450         enum diff_words_type type;
1451         struct diff_words_style *style;
1452 };
1453
1454 static int fn_out_diff_words_write_helper(struct diff_options *o,
1455                                           struct diff_words_style_elem *st_el,
1456                                           const char *newline,
1457                                           size_t count, const char *buf)
1458 {
1459         int print = 0;
1460         struct strbuf sb = STRBUF_INIT;
1461
1462         while (count) {
1463                 char *p = memchr(buf, '\n', count);
1464                 if (print)
1465                         strbuf_addstr(&sb, diff_line_prefix(o));
1466
1467                 if (p != buf) {
1468                         const char *reset = st_el->color && *st_el->color ?
1469                                             GIT_COLOR_RESET : NULL;
1470                         if (st_el->color && *st_el->color)
1471                                 strbuf_addstr(&sb, st_el->color);
1472                         strbuf_addstr(&sb, st_el->prefix);
1473                         strbuf_add(&sb, buf, p ? p - buf : count);
1474                         strbuf_addstr(&sb, st_el->suffix);
1475                         if (reset)
1476                                 strbuf_addstr(&sb, reset);
1477                 }
1478                 if (!p)
1479                         goto out;
1480
1481                 strbuf_addstr(&sb, newline);
1482                 count -= p + 1 - buf;
1483                 buf = p + 1;
1484                 print = 1;
1485                 if (count) {
1486                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1487                                          sb.buf, sb.len, 0);
1488                         strbuf_reset(&sb);
1489                 }
1490         }
1491
1492 out:
1493         if (sb.len)
1494                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1495                                  sb.buf, sb.len, 0);
1496         strbuf_release(&sb);
1497         return 0;
1498 }
1499
1500 /*
1501  * '--color-words' algorithm can be described as:
1502  *
1503  *   1. collect the minus/plus lines of a diff hunk, divided into
1504  *      minus-lines and plus-lines;
1505  *
1506  *   2. break both minus-lines and plus-lines into words and
1507  *      place them into two mmfile_t with one word for each line;
1508  *
1509  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1510  *
1511  * And for the common parts of the both file, we output the plus side text.
1512  * diff_words->current_plus is used to trace the current position of the plus file
1513  * which printed. diff_words->last_minus is used to trace the last minus word
1514  * printed.
1515  *
1516  * For '--graph' to work with '--color-words', we need to output the graph prefix
1517  * on each line of color words output. Generally, there are two conditions on
1518  * which we should output the prefix.
1519  *
1520  *   1. diff_words->last_minus == 0 &&
1521  *      diff_words->current_plus == diff_words->plus.text.ptr
1522  *
1523  *      that is: the plus text must start as a new line, and if there is no minus
1524  *      word printed, a graph prefix must be printed.
1525  *
1526  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1527  *      *(diff_words->current_plus - 1) == '\n'
1528  *
1529  *      that is: a graph prefix must be printed following a '\n'
1530  */
1531 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1532 {
1533         if ((diff_words->last_minus == 0 &&
1534                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1535                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1536                 *(diff_words->current_plus - 1) == '\n')) {
1537                 return 1;
1538         } else {
1539                 return 0;
1540         }
1541 }
1542
1543 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1544 {
1545         struct diff_words_data *diff_words = priv;
1546         struct diff_words_style *style = diff_words->style;
1547         int minus_first, minus_len, plus_first, plus_len;
1548         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1549         struct diff_options *opt = diff_words->opt;
1550         const char *line_prefix;
1551
1552         if (line[0] != '@' || parse_hunk_header(line, len,
1553                         &minus_first, &minus_len, &plus_first, &plus_len))
1554                 return;
1555
1556         assert(opt);
1557         line_prefix = diff_line_prefix(opt);
1558
1559         /* POSIX requires that first be decremented by one if len == 0... */
1560         if (minus_len) {
1561                 minus_begin = diff_words->minus.orig[minus_first].begin;
1562                 minus_end =
1563                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1564         } else
1565                 minus_begin = minus_end =
1566                         diff_words->minus.orig[minus_first].end;
1567
1568         if (plus_len) {
1569                 plus_begin = diff_words->plus.orig[plus_first].begin;
1570                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1571         } else
1572                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1573
1574         if (color_words_output_graph_prefix(diff_words)) {
1575                 fputs(line_prefix, diff_words->opt->file);
1576         }
1577         if (diff_words->current_plus != plus_begin) {
1578                 fn_out_diff_words_write_helper(diff_words->opt,
1579                                 &style->ctx, style->newline,
1580                                 plus_begin - diff_words->current_plus,
1581                                 diff_words->current_plus);
1582         }
1583         if (minus_begin != minus_end) {
1584                 fn_out_diff_words_write_helper(diff_words->opt,
1585                                 &style->old, style->newline,
1586                                 minus_end - minus_begin, minus_begin);
1587         }
1588         if (plus_begin != plus_end) {
1589                 fn_out_diff_words_write_helper(diff_words->opt,
1590                                 &style->new, style->newline,
1591                                 plus_end - plus_begin, plus_begin);
1592         }
1593
1594         diff_words->current_plus = plus_end;
1595         diff_words->last_minus = minus_first;
1596 }
1597
1598 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1599 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1600                 int *begin, int *end)
1601 {
1602         if (word_regex && *begin < buffer->size) {
1603                 regmatch_t match[1];
1604                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1605                                  buffer->size - *begin, 1, match, 0)) {
1606                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1607                                         '\n', match[0].rm_eo - match[0].rm_so);
1608                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1609                         *begin += match[0].rm_so;
1610                         return *begin >= *end;
1611                 }
1612                 return -1;
1613         }
1614
1615         /* find the next word */
1616         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1617                 (*begin)++;
1618         if (*begin >= buffer->size)
1619                 return -1;
1620
1621         /* find the end of the word */
1622         *end = *begin + 1;
1623         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1624                 (*end)++;
1625
1626         return 0;
1627 }
1628
1629 /*
1630  * This function splits the words in buffer->text, stores the list with
1631  * newline separator into out, and saves the offsets of the original words
1632  * in buffer->orig.
1633  */
1634 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1635                 regex_t *word_regex)
1636 {
1637         int i, j;
1638         long alloc = 0;
1639
1640         out->size = 0;
1641         out->ptr = NULL;
1642
1643         /* fake an empty "0th" word */
1644         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1645         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1646         buffer->orig_nr = 1;
1647
1648         for (i = 0; i < buffer->text.size; i++) {
1649                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1650                         return;
1651
1652                 /* store original boundaries */
1653                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1654                                 buffer->orig_alloc);
1655                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1656                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1657                 buffer->orig_nr++;
1658
1659                 /* store one word */
1660                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1661                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1662                 out->ptr[out->size + j - i] = '\n';
1663                 out->size += j - i + 1;
1664
1665                 i = j - 1;
1666         }
1667 }
1668
1669 /* this executes the word diff on the accumulated buffers */
1670 static void diff_words_show(struct diff_words_data *diff_words)
1671 {
1672         xpparam_t xpp;
1673         xdemitconf_t xecfg;
1674         mmfile_t minus, plus;
1675         struct diff_words_style *style = diff_words->style;
1676
1677         struct diff_options *opt = diff_words->opt;
1678         const char *line_prefix;
1679
1680         assert(opt);
1681         line_prefix = diff_line_prefix(opt);
1682
1683         /* special case: only removal */
1684         if (!diff_words->plus.text.size) {
1685                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1686                                  line_prefix, strlen(line_prefix), 0);
1687                 fn_out_diff_words_write_helper(diff_words->opt,
1688                         &style->old, style->newline,
1689                         diff_words->minus.text.size,
1690                         diff_words->minus.text.ptr);
1691                 diff_words->minus.text.size = 0;
1692                 return;
1693         }
1694
1695         diff_words->current_plus = diff_words->plus.text.ptr;
1696         diff_words->last_minus = 0;
1697
1698         memset(&xpp, 0, sizeof(xpp));
1699         memset(&xecfg, 0, sizeof(xecfg));
1700         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1701         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1702         xpp.flags = 0;
1703         /* as only the hunk header will be parsed, we need a 0-context */
1704         xecfg.ctxlen = 0;
1705         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1706                           &xpp, &xecfg))
1707                 die("unable to generate word diff");
1708         free(minus.ptr);
1709         free(plus.ptr);
1710         if (diff_words->current_plus != diff_words->plus.text.ptr +
1711                         diff_words->plus.text.size) {
1712                 if (color_words_output_graph_prefix(diff_words))
1713                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1714                                          line_prefix, strlen(line_prefix), 0);
1715                 fn_out_diff_words_write_helper(diff_words->opt,
1716                         &style->ctx, style->newline,
1717                         diff_words->plus.text.ptr + diff_words->plus.text.size
1718                         - diff_words->current_plus, diff_words->current_plus);
1719         }
1720         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1721 }
1722
1723 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1724 static void diff_words_flush(struct emit_callback *ecbdata)
1725 {
1726         struct diff_options *wo = ecbdata->diff_words->opt;
1727
1728         if (ecbdata->diff_words->minus.text.size ||
1729             ecbdata->diff_words->plus.text.size)
1730                 diff_words_show(ecbdata->diff_words);
1731
1732         if (wo->emitted_symbols) {
1733                 struct diff_options *o = ecbdata->opt;
1734                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1735                 int i;
1736
1737                 /*
1738                  * NEEDSWORK:
1739                  * Instead of appending each, concat all words to a line?
1740                  */
1741                 for (i = 0; i < wol->nr; i++)
1742                         append_emitted_diff_symbol(o, &wol->buf[i]);
1743
1744                 for (i = 0; i < wol->nr; i++)
1745                         free((void *)wol->buf[i].line);
1746
1747                 wol->nr = 0;
1748         }
1749 }
1750
1751 static void diff_filespec_load_driver(struct diff_filespec *one)
1752 {
1753         /* Use already-loaded driver */
1754         if (one->driver)
1755                 return;
1756
1757         if (S_ISREG(one->mode))
1758                 one->driver = userdiff_find_by_path(one->path);
1759
1760         /* Fallback to default settings */
1761         if (!one->driver)
1762                 one->driver = userdiff_find_by_name("default");
1763 }
1764
1765 static const char *userdiff_word_regex(struct diff_filespec *one)
1766 {
1767         diff_filespec_load_driver(one);
1768         return one->driver->word_regex;
1769 }
1770
1771 static void init_diff_words_data(struct emit_callback *ecbdata,
1772                                  struct diff_options *orig_opts,
1773                                  struct diff_filespec *one,
1774                                  struct diff_filespec *two)
1775 {
1776         int i;
1777         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1778         memcpy(o, orig_opts, sizeof(struct diff_options));
1779
1780         ecbdata->diff_words =
1781                 xcalloc(1, sizeof(struct diff_words_data));
1782         ecbdata->diff_words->type = o->word_diff;
1783         ecbdata->diff_words->opt = o;
1784
1785         if (orig_opts->emitted_symbols)
1786                 o->emitted_symbols =
1787                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1788
1789         if (!o->word_regex)
1790                 o->word_regex = userdiff_word_regex(one);
1791         if (!o->word_regex)
1792                 o->word_regex = userdiff_word_regex(two);
1793         if (!o->word_regex)
1794                 o->word_regex = diff_word_regex_cfg;
1795         if (o->word_regex) {
1796                 ecbdata->diff_words->word_regex = (regex_t *)
1797                         xmalloc(sizeof(regex_t));
1798                 if (regcomp(ecbdata->diff_words->word_regex,
1799                             o->word_regex,
1800                             REG_EXTENDED | REG_NEWLINE))
1801                         die ("Invalid regular expression: %s",
1802                              o->word_regex);
1803         }
1804         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1805                 if (o->word_diff == diff_words_styles[i].type) {
1806                         ecbdata->diff_words->style =
1807                                 &diff_words_styles[i];
1808                         break;
1809                 }
1810         }
1811         if (want_color(o->use_color)) {
1812                 struct diff_words_style *st = ecbdata->diff_words->style;
1813                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1814                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1815                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1816         }
1817 }
1818
1819 static void free_diff_words_data(struct emit_callback *ecbdata)
1820 {
1821         if (ecbdata->diff_words) {
1822                 diff_words_flush(ecbdata);
1823                 free (ecbdata->diff_words->opt->emitted_symbols);
1824                 free (ecbdata->diff_words->opt);
1825                 free (ecbdata->diff_words->minus.text.ptr);
1826                 free (ecbdata->diff_words->minus.orig);
1827                 free (ecbdata->diff_words->plus.text.ptr);
1828                 free (ecbdata->diff_words->plus.orig);
1829                 if (ecbdata->diff_words->word_regex) {
1830                         regfree(ecbdata->diff_words->word_regex);
1831                         free(ecbdata->diff_words->word_regex);
1832                 }
1833                 FREE_AND_NULL(ecbdata->diff_words);
1834         }
1835 }
1836
1837 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1838 {
1839         if (want_color(diff_use_color))
1840                 return diff_colors[ix];
1841         return "";
1842 }
1843
1844 const char *diff_line_prefix(struct diff_options *opt)
1845 {
1846         struct strbuf *msgbuf;
1847         if (!opt->output_prefix)
1848                 return "";
1849
1850         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1851         return msgbuf->buf;
1852 }
1853
1854 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1855 {
1856         const char *cp;
1857         unsigned long allot;
1858         size_t l = len;
1859
1860         if (ecb->truncate)
1861                 return ecb->truncate(line, len);
1862         cp = line;
1863         allot = l;
1864         while (0 < l) {
1865                 (void) utf8_width(&cp, &l);
1866                 if (!cp)
1867                         break; /* truncated in the middle? */
1868         }
1869         return allot - l;
1870 }
1871
1872 static void find_lno(const char *line, struct emit_callback *ecbdata)
1873 {
1874         const char *p;
1875         ecbdata->lno_in_preimage = 0;
1876         ecbdata->lno_in_postimage = 0;
1877         p = strchr(line, '-');
1878         if (!p)
1879                 return; /* cannot happen */
1880         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1881         p = strchr(p, '+');
1882         if (!p)
1883                 return; /* cannot happen */
1884         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1885 }
1886
1887 static void fn_out_consume(void *priv, char *line, unsigned long len)
1888 {
1889         struct emit_callback *ecbdata = priv;
1890         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1891         struct diff_options *o = ecbdata->opt;
1892
1893         o->found_changes = 1;
1894
1895         if (ecbdata->header) {
1896                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1897                                  ecbdata->header->buf, ecbdata->header->len, 0);
1898                 strbuf_reset(ecbdata->header);
1899                 ecbdata->header = NULL;
1900         }
1901
1902         if (ecbdata->label_path[0]) {
1903                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1904                                  ecbdata->label_path[0],
1905                                  strlen(ecbdata->label_path[0]), 0);
1906                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1907                                  ecbdata->label_path[1],
1908                                  strlen(ecbdata->label_path[1]), 0);
1909                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1910         }
1911
1912         if (diff_suppress_blank_empty
1913             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1914                 line[0] = '\n';
1915                 len = 1;
1916         }
1917
1918         if (line[0] == '@') {
1919                 if (ecbdata->diff_words)
1920                         diff_words_flush(ecbdata);
1921                 len = sane_truncate_line(ecbdata, line, len);
1922                 find_lno(line, ecbdata);
1923                 emit_hunk_header(ecbdata, line, len);
1924                 return;
1925         }
1926
1927         if (ecbdata->diff_words) {
1928                 enum diff_symbol s =
1929                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
1930                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
1931                 if (line[0] == '-') {
1932                         diff_words_append(line, len,
1933                                           &ecbdata->diff_words->minus);
1934                         return;
1935                 } else if (line[0] == '+') {
1936                         diff_words_append(line, len,
1937                                           &ecbdata->diff_words->plus);
1938                         return;
1939                 } else if (starts_with(line, "\\ ")) {
1940                         /*
1941                          * Eat the "no newline at eof" marker as if we
1942                          * saw a "+" or "-" line with nothing on it,
1943                          * and return without diff_words_flush() to
1944                          * defer processing. If this is the end of
1945                          * preimage, more "+" lines may come after it.
1946                          */
1947                         return;
1948                 }
1949                 diff_words_flush(ecbdata);
1950                 emit_diff_symbol(o, s, line, len, 0);
1951                 return;
1952         }
1953
1954         switch (line[0]) {
1955         case '+':
1956                 ecbdata->lno_in_postimage++;
1957                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1958                 break;
1959         case '-':
1960                 ecbdata->lno_in_preimage++;
1961                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1962                 break;
1963         case ' ':
1964                 ecbdata->lno_in_postimage++;
1965                 ecbdata->lno_in_preimage++;
1966                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1967                 break;
1968         default:
1969                 /* incomplete line at the end */
1970                 ecbdata->lno_in_preimage++;
1971                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
1972                                  line, len, 0);
1973                 break;
1974         }
1975 }
1976
1977 static char *pprint_rename(const char *a, const char *b)
1978 {
1979         const char *old = a;
1980         const char *new = b;
1981         struct strbuf name = STRBUF_INIT;
1982         int pfx_length, sfx_length;
1983         int pfx_adjust_for_slash;
1984         int len_a = strlen(a);
1985         int len_b = strlen(b);
1986         int a_midlen, b_midlen;
1987         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1988         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1989
1990         if (qlen_a || qlen_b) {
1991                 quote_c_style(a, &name, NULL, 0);
1992                 strbuf_addstr(&name, " => ");
1993                 quote_c_style(b, &name, NULL, 0);
1994                 return strbuf_detach(&name, NULL);
1995         }
1996
1997         /* Find common prefix */
1998         pfx_length = 0;
1999         while (*old && *new && *old == *new) {
2000                 if (*old == '/')
2001                         pfx_length = old - a + 1;
2002                 old++;
2003                 new++;
2004         }
2005
2006         /* Find common suffix */
2007         old = a + len_a;
2008         new = b + len_b;
2009         sfx_length = 0;
2010         /*
2011          * If there is a common prefix, it must end in a slash.  In
2012          * that case we let this loop run 1 into the prefix to see the
2013          * same slash.
2014          *
2015          * If there is no common prefix, we cannot do this as it would
2016          * underrun the input strings.
2017          */
2018         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2019         while (a + pfx_length - pfx_adjust_for_slash <= old &&
2020                b + pfx_length - pfx_adjust_for_slash <= new &&
2021                *old == *new) {
2022                 if (*old == '/')
2023                         sfx_length = len_a - (old - a);
2024                 old--;
2025                 new--;
2026         }
2027
2028         /*
2029          * pfx{mid-a => mid-b}sfx
2030          * {pfx-a => pfx-b}sfx
2031          * pfx{sfx-a => sfx-b}
2032          * name-a => name-b
2033          */
2034         a_midlen = len_a - pfx_length - sfx_length;
2035         b_midlen = len_b - pfx_length - sfx_length;
2036         if (a_midlen < 0)
2037                 a_midlen = 0;
2038         if (b_midlen < 0)
2039                 b_midlen = 0;
2040
2041         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2042         if (pfx_length + sfx_length) {
2043                 strbuf_add(&name, a, pfx_length);
2044                 strbuf_addch(&name, '{');
2045         }
2046         strbuf_add(&name, a + pfx_length, a_midlen);
2047         strbuf_addstr(&name, " => ");
2048         strbuf_add(&name, b + pfx_length, b_midlen);
2049         if (pfx_length + sfx_length) {
2050                 strbuf_addch(&name, '}');
2051                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
2052         }
2053         return strbuf_detach(&name, NULL);
2054 }
2055
2056 struct diffstat_t {
2057         int nr;
2058         int alloc;
2059         struct diffstat_file {
2060                 char *from_name;
2061                 char *name;
2062                 char *print_name;
2063                 unsigned is_unmerged:1;
2064                 unsigned is_binary:1;
2065                 unsigned is_renamed:1;
2066                 unsigned is_interesting:1;
2067                 uintmax_t added, deleted;
2068         } **files;
2069 };
2070
2071 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2072                                           const char *name_a,
2073                                           const char *name_b)
2074 {
2075         struct diffstat_file *x;
2076         x = xcalloc(1, sizeof(*x));
2077         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2078         diffstat->files[diffstat->nr++] = x;
2079         if (name_b) {
2080                 x->from_name = xstrdup(name_a);
2081                 x->name = xstrdup(name_b);
2082                 x->is_renamed = 1;
2083         }
2084         else {
2085                 x->from_name = NULL;
2086                 x->name = xstrdup(name_a);
2087         }
2088         return x;
2089 }
2090
2091 static void diffstat_consume(void *priv, char *line, unsigned long len)
2092 {
2093         struct diffstat_t *diffstat = priv;
2094         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2095
2096         if (line[0] == '+')
2097                 x->added++;
2098         else if (line[0] == '-')
2099                 x->deleted++;
2100 }
2101
2102 const char mime_boundary_leader[] = "------------";
2103
2104 static int scale_linear(int it, int width, int max_change)
2105 {
2106         if (!it)
2107                 return 0;
2108         /*
2109          * make sure that at least one '-' or '+' is printed if
2110          * there is any change to this path. The easiest way is to
2111          * scale linearly as if the alloted width is one column shorter
2112          * than it is, and then add 1 to the result.
2113          */
2114         return 1 + (it * (width - 1) / max_change);
2115 }
2116
2117 static void show_graph(struct strbuf *out, char ch, int cnt,
2118                        const char *set, const char *reset)
2119 {
2120         if (cnt <= 0)
2121                 return;
2122         strbuf_addstr(out, set);
2123         strbuf_addchars(out, ch, cnt);
2124         strbuf_addstr(out, reset);
2125 }
2126
2127 static void fill_print_name(struct diffstat_file *file)
2128 {
2129         char *pname;
2130
2131         if (file->print_name)
2132                 return;
2133
2134         if (!file->is_renamed) {
2135                 struct strbuf buf = STRBUF_INIT;
2136                 if (quote_c_style(file->name, &buf, NULL, 0)) {
2137                         pname = strbuf_detach(&buf, NULL);
2138                 } else {
2139                         pname = file->name;
2140                         strbuf_release(&buf);
2141                 }
2142         } else {
2143                 pname = pprint_rename(file->from_name, file->name);
2144         }
2145         file->print_name = pname;
2146 }
2147
2148 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2149                 int files, int insertions, int deletions)
2150 {
2151         struct strbuf sb = STRBUF_INIT;
2152
2153         if (!files) {
2154                 assert(insertions == 0 && deletions == 0);
2155                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2156                                  NULL, 0, 0);
2157                 return;
2158         }
2159
2160         strbuf_addf(&sb,
2161                     (files == 1) ? " %d file changed" : " %d files changed",
2162                     files);
2163
2164         /*
2165          * For binary diff, the caller may want to print "x files
2166          * changed" with insertions == 0 && deletions == 0.
2167          *
2168          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2169          * is probably less confusing (i.e skip over "2 files changed
2170          * but nothing about added/removed lines? Is this a bug in Git?").
2171          */
2172         if (insertions || deletions == 0) {
2173                 strbuf_addf(&sb,
2174                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2175                             insertions);
2176         }
2177
2178         if (deletions || insertions == 0) {
2179                 strbuf_addf(&sb,
2180                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2181                             deletions);
2182         }
2183         strbuf_addch(&sb, '\n');
2184         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2185                          sb.buf, sb.len, 0);
2186         strbuf_release(&sb);
2187 }
2188
2189 void print_stat_summary(FILE *fp, int files,
2190                         int insertions, int deletions)
2191 {
2192         struct diff_options o;
2193         memset(&o, 0, sizeof(o));
2194         o.file = fp;
2195
2196         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2197 }
2198
2199 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2200 {
2201         int i, len, add, del, adds = 0, dels = 0;
2202         uintmax_t max_change = 0, max_len = 0;
2203         int total_files = data->nr, count;
2204         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2205         const char *reset, *add_c, *del_c;
2206         int extra_shown = 0;
2207         const char *line_prefix = diff_line_prefix(options);
2208         struct strbuf out = STRBUF_INIT;
2209
2210         if (data->nr == 0)
2211                 return;
2212
2213         count = options->stat_count ? options->stat_count : data->nr;
2214
2215         reset = diff_get_color_opt(options, DIFF_RESET);
2216         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2217         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2218
2219         /*
2220          * Find the longest filename and max number of changes
2221          */
2222         for (i = 0; (i < count) && (i < data->nr); i++) {
2223                 struct diffstat_file *file = data->files[i];
2224                 uintmax_t change = file->added + file->deleted;
2225
2226                 if (!file->is_interesting && (change == 0)) {
2227                         count++; /* not shown == room for one more */
2228                         continue;
2229                 }
2230                 fill_print_name(file);
2231                 len = strlen(file->print_name);
2232                 if (max_len < len)
2233                         max_len = len;
2234
2235                 if (file->is_unmerged) {
2236                         /* "Unmerged" is 8 characters */
2237                         bin_width = bin_width < 8 ? 8 : bin_width;
2238                         continue;
2239                 }
2240                 if (file->is_binary) {
2241                         /* "Bin XXX -> YYY bytes" */
2242                         int w = 14 + decimal_width(file->added)
2243                                 + decimal_width(file->deleted);
2244                         bin_width = bin_width < w ? w : bin_width;
2245                         /* Display change counts aligned with "Bin" */
2246                         number_width = 3;
2247                         continue;
2248                 }
2249
2250                 if (max_change < change)
2251                         max_change = change;
2252         }
2253         count = i; /* where we can stop scanning in data->files[] */
2254
2255         /*
2256          * We have width = stat_width or term_columns() columns total.
2257          * We want a maximum of min(max_len, stat_name_width) for the name part.
2258          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2259          * We also need 1 for " " and 4 + decimal_width(max_change)
2260          * for " | NNNN " and one the empty column at the end, altogether
2261          * 6 + decimal_width(max_change).
2262          *
2263          * If there's not enough space, we will use the smaller of
2264          * stat_name_width (if set) and 5/8*width for the filename,
2265          * and the rest for constant elements + graph part, but no more
2266          * than stat_graph_width for the graph part.
2267          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2268          * for the standard terminal size).
2269          *
2270          * In other words: stat_width limits the maximum width, and
2271          * stat_name_width fixes the maximum width of the filename,
2272          * and is also used to divide available columns if there
2273          * aren't enough.
2274          *
2275          * Binary files are displayed with "Bin XXX -> YYY bytes"
2276          * instead of the change count and graph. This part is treated
2277          * similarly to the graph part, except that it is not
2278          * "scaled". If total width is too small to accommodate the
2279          * guaranteed minimum width of the filename part and the
2280          * separators and this message, this message will "overflow"
2281          * making the line longer than the maximum width.
2282          */
2283
2284         if (options->stat_width == -1)
2285                 width = term_columns() - strlen(line_prefix);
2286         else
2287                 width = options->stat_width ? options->stat_width : 80;
2288         number_width = decimal_width(max_change) > number_width ?
2289                 decimal_width(max_change) : number_width;
2290
2291         if (options->stat_graph_width == -1)
2292                 options->stat_graph_width = diff_stat_graph_width;
2293
2294         /*
2295          * Guarantee 3/8*16==6 for the graph part
2296          * and 5/8*16==10 for the filename part
2297          */
2298         if (width < 16 + 6 + number_width)
2299                 width = 16 + 6 + number_width;
2300
2301         /*
2302          * First assign sizes that are wanted, ignoring available width.
2303          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2304          * starting from "XXX" should fit in graph_width.
2305          */
2306         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2307         if (options->stat_graph_width &&
2308             options->stat_graph_width < graph_width)
2309                 graph_width = options->stat_graph_width;
2310
2311         name_width = (options->stat_name_width > 0 &&
2312                       options->stat_name_width < max_len) ?
2313                 options->stat_name_width : max_len;
2314
2315         /*
2316          * Adjust adjustable widths not to exceed maximum width
2317          */
2318         if (name_width + number_width + 6 + graph_width > width) {
2319                 if (graph_width > width * 3/8 - number_width - 6) {
2320                         graph_width = width * 3/8 - number_width - 6;
2321                         if (graph_width < 6)
2322                                 graph_width = 6;
2323                 }
2324
2325                 if (options->stat_graph_width &&
2326                     graph_width > options->stat_graph_width)
2327                         graph_width = options->stat_graph_width;
2328                 if (name_width > width - number_width - 6 - graph_width)
2329                         name_width = width - number_width - 6 - graph_width;
2330                 else
2331                         graph_width = width - number_width - 6 - name_width;
2332         }
2333
2334         /*
2335          * From here name_width is the width of the name area,
2336          * and graph_width is the width of the graph area.
2337          * max_change is used to scale graph properly.
2338          */
2339         for (i = 0; i < count; i++) {
2340                 const char *prefix = "";
2341                 struct diffstat_file *file = data->files[i];
2342                 char *name = file->print_name;
2343                 uintmax_t added = file->added;
2344                 uintmax_t deleted = file->deleted;
2345                 int name_len;
2346
2347                 if (!file->is_interesting && (added + deleted == 0))
2348                         continue;
2349
2350                 /*
2351                  * "scale" the filename
2352                  */
2353                 len = name_width;
2354                 name_len = strlen(name);
2355                 if (name_width < name_len) {
2356                         char *slash;
2357                         prefix = "...";
2358                         len -= 3;
2359                         name += name_len - len;
2360                         slash = strchr(name, '/');
2361                         if (slash)
2362                                 name = slash;
2363                 }
2364
2365                 if (file->is_binary) {
2366                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2367                         strbuf_addf(&out, " %*s", number_width, "Bin");
2368                         if (!added && !deleted) {
2369                                 strbuf_addch(&out, '\n');
2370                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2371                                                  out.buf, out.len, 0);
2372                                 strbuf_reset(&out);
2373                                 continue;
2374                         }
2375                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2376                                 del_c, deleted, reset);
2377                         strbuf_addstr(&out, " -> ");
2378                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2379                                 add_c, added, reset);
2380                         strbuf_addstr(&out, " bytes\n");
2381                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2382                                          out.buf, out.len, 0);
2383                         strbuf_reset(&out);
2384                         continue;
2385                 }
2386                 else if (file->is_unmerged) {
2387                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2388                         strbuf_addstr(&out, " Unmerged\n");
2389                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2390                                          out.buf, out.len, 0);
2391                         strbuf_reset(&out);
2392                         continue;
2393                 }
2394
2395                 /*
2396                  * scale the add/delete
2397                  */
2398                 add = added;
2399                 del = deleted;
2400
2401                 if (graph_width <= max_change) {
2402                         int total = scale_linear(add + del, graph_width, max_change);
2403                         if (total < 2 && add && del)
2404                                 /* width >= 2 due to the sanity check */
2405                                 total = 2;
2406                         if (add < del) {
2407                                 add = scale_linear(add, graph_width, max_change);
2408                                 del = total - add;
2409                         } else {
2410                                 del = scale_linear(del, graph_width, max_change);
2411                                 add = total - del;
2412                         }
2413                 }
2414                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2415                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2416                         number_width, added + deleted,
2417                         added + deleted ? " " : "");
2418                 show_graph(&out, '+', add, add_c, reset);
2419                 show_graph(&out, '-', del, del_c, reset);
2420                 strbuf_addch(&out, '\n');
2421                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2422                                  out.buf, out.len, 0);
2423                 strbuf_reset(&out);
2424         }
2425
2426         for (i = 0; i < data->nr; i++) {
2427                 struct diffstat_file *file = data->files[i];
2428                 uintmax_t added = file->added;
2429                 uintmax_t deleted = file->deleted;
2430
2431                 if (file->is_unmerged ||
2432                     (!file->is_interesting && (added + deleted == 0))) {
2433                         total_files--;
2434                         continue;
2435                 }
2436
2437                 if (!file->is_binary) {
2438                         adds += added;
2439                         dels += deleted;
2440                 }
2441                 if (i < count)
2442                         continue;
2443                 if (!extra_shown)
2444                         emit_diff_symbol(options,
2445                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2446                                          NULL, 0, 0);
2447                 extra_shown = 1;
2448         }
2449
2450         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2451 }
2452
2453 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2454 {
2455         int i, adds = 0, dels = 0, total_files = data->nr;
2456
2457         if (data->nr == 0)
2458                 return;
2459
2460         for (i = 0; i < data->nr; i++) {
2461                 int added = data->files[i]->added;
2462                 int deleted = data->files[i]->deleted;
2463
2464                 if (data->files[i]->is_unmerged ||
2465                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2466                         total_files--;
2467                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2468                         adds += added;
2469                         dels += deleted;
2470                 }
2471         }
2472         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2473 }
2474
2475 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2476 {
2477         int i;
2478
2479         if (data->nr == 0)
2480                 return;
2481
2482         for (i = 0; i < data->nr; i++) {
2483                 struct diffstat_file *file = data->files[i];
2484
2485                 fprintf(options->file, "%s", diff_line_prefix(options));
2486
2487                 if (file->is_binary)
2488                         fprintf(options->file, "-\t-\t");
2489                 else
2490                         fprintf(options->file,
2491                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2492                                 file->added, file->deleted);
2493                 if (options->line_termination) {
2494                         fill_print_name(file);
2495                         if (!file->is_renamed)
2496                                 write_name_quoted(file->name, options->file,
2497                                                   options->line_termination);
2498                         else {
2499                                 fputs(file->print_name, options->file);
2500                                 putc(options->line_termination, options->file);
2501                         }
2502                 } else {
2503                         if (file->is_renamed) {
2504                                 putc('\0', options->file);
2505                                 write_name_quoted(file->from_name, options->file, '\0');
2506                         }
2507                         write_name_quoted(file->name, options->file, '\0');
2508                 }
2509         }
2510 }
2511
2512 struct dirstat_file {
2513         const char *name;
2514         unsigned long changed;
2515 };
2516
2517 struct dirstat_dir {
2518         struct dirstat_file *files;
2519         int alloc, nr, permille, cumulative;
2520 };
2521
2522 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2523                 unsigned long changed, const char *base, int baselen)
2524 {
2525         unsigned long this_dir = 0;
2526         unsigned int sources = 0;
2527         const char *line_prefix = diff_line_prefix(opt);
2528
2529         while (dir->nr) {
2530                 struct dirstat_file *f = dir->files;
2531                 int namelen = strlen(f->name);
2532                 unsigned long this;
2533                 char *slash;
2534
2535                 if (namelen < baselen)
2536                         break;
2537                 if (memcmp(f->name, base, baselen))
2538                         break;
2539                 slash = strchr(f->name + baselen, '/');
2540                 if (slash) {
2541                         int newbaselen = slash + 1 - f->name;
2542                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2543                         sources++;
2544                 } else {
2545                         this = f->changed;
2546                         dir->files++;
2547                         dir->nr--;
2548                         sources += 2;
2549                 }
2550                 this_dir += this;
2551         }
2552
2553         /*
2554          * We don't report dirstat's for
2555          *  - the top level
2556          *  - or cases where everything came from a single directory
2557          *    under this directory (sources == 1).
2558          */
2559         if (baselen && sources != 1) {
2560                 if (this_dir) {
2561                         int permille = this_dir * 1000 / changed;
2562                         if (permille >= dir->permille) {
2563                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2564                                         permille / 10, permille % 10, baselen, base);
2565                                 if (!dir->cumulative)
2566                                         return 0;
2567                         }
2568                 }
2569         }
2570         return this_dir;
2571 }
2572
2573 static int dirstat_compare(const void *_a, const void *_b)
2574 {
2575         const struct dirstat_file *a = _a;
2576         const struct dirstat_file *b = _b;
2577         return strcmp(a->name, b->name);
2578 }
2579
2580 static void show_dirstat(struct diff_options *options)
2581 {
2582         int i;
2583         unsigned long changed;
2584         struct dirstat_dir dir;
2585         struct diff_queue_struct *q = &diff_queued_diff;
2586
2587         dir.files = NULL;
2588         dir.alloc = 0;
2589         dir.nr = 0;
2590         dir.permille = options->dirstat_permille;
2591         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2592
2593         changed = 0;
2594         for (i = 0; i < q->nr; i++) {
2595                 struct diff_filepair *p = q->queue[i];
2596                 const char *name;
2597                 unsigned long copied, added, damage;
2598                 int content_changed;
2599
2600                 name = p->two->path ? p->two->path : p->one->path;
2601
2602                 if (p->one->oid_valid && p->two->oid_valid)
2603                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2604                 else
2605                         content_changed = 1;
2606
2607                 if (!content_changed) {
2608                         /*
2609                          * The SHA1 has not changed, so pre-/post-content is
2610                          * identical. We can therefore skip looking at the
2611                          * file contents altogether.
2612                          */
2613                         damage = 0;
2614                         goto found_damage;
2615                 }
2616
2617                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2618                         /*
2619                          * In --dirstat-by-file mode, we don't really need to
2620                          * look at the actual file contents at all.
2621                          * The fact that the SHA1 changed is enough for us to
2622                          * add this file to the list of results
2623                          * (with each file contributing equal damage).
2624                          */
2625                         damage = 1;
2626                         goto found_damage;
2627                 }
2628
2629                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2630                         diff_populate_filespec(p->one, 0);
2631                         diff_populate_filespec(p->two, 0);
2632                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2633                                                &copied, &added);
2634                         diff_free_filespec_data(p->one);
2635                         diff_free_filespec_data(p->two);
2636                 } else if (DIFF_FILE_VALID(p->one)) {
2637                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2638                         copied = added = 0;
2639                         diff_free_filespec_data(p->one);
2640                 } else if (DIFF_FILE_VALID(p->two)) {
2641                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2642                         copied = 0;
2643                         added = p->two->size;
2644                         diff_free_filespec_data(p->two);
2645                 } else
2646                         continue;
2647
2648                 /*
2649                  * Original minus copied is the removed material,
2650                  * added is the new material.  They are both damages
2651                  * made to the preimage.
2652                  * If the resulting damage is zero, we know that
2653                  * diffcore_count_changes() considers the two entries to
2654                  * be identical, but since content_changed is true, we
2655                  * know that there must have been _some_ kind of change,
2656                  * so we force all entries to have damage > 0.
2657                  */
2658                 damage = (p->one->size - copied) + added;
2659                 if (!damage)
2660                         damage = 1;
2661
2662 found_damage:
2663                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2664                 dir.files[dir.nr].name = name;
2665                 dir.files[dir.nr].changed = damage;
2666                 changed += damage;
2667                 dir.nr++;
2668         }
2669
2670         /* This can happen even with many files, if everything was renames */
2671         if (!changed)
2672                 return;
2673
2674         /* Show all directories with more than x% of the changes */
2675         QSORT(dir.files, dir.nr, dirstat_compare);
2676         gather_dirstat(options, &dir, changed, "", 0);
2677 }
2678
2679 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2680 {
2681         int i;
2682         unsigned long changed;
2683         struct dirstat_dir dir;
2684
2685         if (data->nr == 0)
2686                 return;
2687
2688         dir.files = NULL;
2689         dir.alloc = 0;
2690         dir.nr = 0;
2691         dir.permille = options->dirstat_permille;
2692         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2693
2694         changed = 0;
2695         for (i = 0; i < data->nr; i++) {
2696                 struct diffstat_file *file = data->files[i];
2697                 unsigned long damage = file->added + file->deleted;
2698                 if (file->is_binary)
2699                         /*
2700                          * binary files counts bytes, not lines. Must find some
2701                          * way to normalize binary bytes vs. textual lines.
2702                          * The following heuristic assumes that there are 64
2703                          * bytes per "line".
2704                          * This is stupid and ugly, but very cheap...
2705                          */
2706                         damage = (damage + 63) / 64;
2707                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2708                 dir.files[dir.nr].name = file->name;
2709                 dir.files[dir.nr].changed = damage;
2710                 changed += damage;
2711                 dir.nr++;
2712         }
2713
2714         /* This can happen even with many files, if everything was renames */
2715         if (!changed)
2716                 return;
2717
2718         /* Show all directories with more than x% of the changes */
2719         QSORT(dir.files, dir.nr, dirstat_compare);
2720         gather_dirstat(options, &dir, changed, "", 0);
2721 }
2722
2723 static void free_diffstat_info(struct diffstat_t *diffstat)
2724 {
2725         int i;
2726         for (i = 0; i < diffstat->nr; i++) {
2727                 struct diffstat_file *f = diffstat->files[i];
2728                 if (f->name != f->print_name)
2729                         free(f->print_name);
2730                 free(f->name);
2731                 free(f->from_name);
2732                 free(f);
2733         }
2734         free(diffstat->files);
2735 }
2736
2737 struct checkdiff_t {
2738         const char *filename;
2739         int lineno;
2740         int conflict_marker_size;
2741         struct diff_options *o;
2742         unsigned ws_rule;
2743         unsigned status;
2744 };
2745
2746 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2747 {
2748         char firstchar;
2749         int cnt;
2750
2751         if (len < marker_size + 1)
2752                 return 0;
2753         firstchar = line[0];
2754         switch (firstchar) {
2755         case '=': case '>': case '<': case '|':
2756                 break;
2757         default:
2758                 return 0;
2759         }
2760         for (cnt = 1; cnt < marker_size; cnt++)
2761                 if (line[cnt] != firstchar)
2762                         return 0;
2763         /* line[1] thru line[marker_size-1] are same as firstchar */
2764         if (len < marker_size + 1 || !isspace(line[marker_size]))
2765                 return 0;
2766         return 1;
2767 }
2768
2769 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2770 {
2771         struct checkdiff_t *data = priv;
2772         int marker_size = data->conflict_marker_size;
2773         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2774         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2775         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2776         char *err;
2777         const char *line_prefix;
2778
2779         assert(data->o);
2780         line_prefix = diff_line_prefix(data->o);
2781
2782         if (line[0] == '+') {
2783                 unsigned bad;
2784                 data->lineno++;
2785                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2786                         data->status |= 1;
2787                         fprintf(data->o->file,
2788                                 "%s%s:%d: leftover conflict marker\n",
2789                                 line_prefix, data->filename, data->lineno);
2790                 }
2791                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2792                 if (!bad)
2793                         return;
2794                 data->status |= bad;
2795                 err = whitespace_error_string(bad);
2796                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2797                         line_prefix, data->filename, data->lineno, err);
2798                 free(err);
2799                 emit_line(data->o, set, reset, line, 1);
2800                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2801                               data->o->file, set, reset, ws);
2802         } else if (line[0] == ' ') {
2803                 data->lineno++;
2804         } else if (line[0] == '@') {
2805                 char *plus = strchr(line, '+');
2806                 if (plus)
2807                         data->lineno = strtol(plus, NULL, 10) - 1;
2808                 else
2809                         die("invalid diff");
2810         }
2811 }
2812
2813 static unsigned char *deflate_it(char *data,
2814                                  unsigned long size,
2815                                  unsigned long *result_size)
2816 {
2817         int bound;
2818         unsigned char *deflated;
2819         git_zstream stream;
2820
2821         git_deflate_init(&stream, zlib_compression_level);
2822         bound = git_deflate_bound(&stream, size);
2823         deflated = xmalloc(bound);
2824         stream.next_out = deflated;
2825         stream.avail_out = bound;
2826
2827         stream.next_in = (unsigned char *)data;
2828         stream.avail_in = size;
2829         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2830                 ; /* nothing */
2831         git_deflate_end(&stream);
2832         *result_size = stream.total_out;
2833         return deflated;
2834 }
2835
2836 static void emit_binary_diff_body(struct diff_options *o,
2837                                   mmfile_t *one, mmfile_t *two)
2838 {
2839         void *cp;
2840         void *delta;
2841         void *deflated;
2842         void *data;
2843         unsigned long orig_size;
2844         unsigned long delta_size;
2845         unsigned long deflate_size;
2846         unsigned long data_size;
2847
2848         /* We could do deflated delta, or we could do just deflated two,
2849          * whichever is smaller.
2850          */
2851         delta = NULL;
2852         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2853         if (one->size && two->size) {
2854                 delta = diff_delta(one->ptr, one->size,
2855                                    two->ptr, two->size,
2856                                    &delta_size, deflate_size);
2857                 if (delta) {
2858                         void *to_free = delta;
2859                         orig_size = delta_size;
2860                         delta = deflate_it(delta, delta_size, &delta_size);
2861                         free(to_free);
2862                 }
2863         }
2864
2865         if (delta && delta_size < deflate_size) {
2866                 char *s = xstrfmt("%lu", orig_size);
2867                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2868                                  s, strlen(s), 0);
2869                 free(s);
2870                 free(deflated);
2871                 data = delta;
2872                 data_size = delta_size;
2873         } else {
2874                 char *s = xstrfmt("%lu", two->size);
2875                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2876                                  s, strlen(s), 0);
2877                 free(s);
2878                 free(delta);
2879                 data = deflated;
2880                 data_size = deflate_size;
2881         }
2882
2883         /* emit data encoded in base85 */
2884         cp = data;
2885         while (data_size) {
2886                 int len;
2887                 int bytes = (52 < data_size) ? 52 : data_size;
2888                 char line[71];
2889                 data_size -= bytes;
2890                 if (bytes <= 26)
2891                         line[0] = bytes + 'A' - 1;
2892                 else
2893                         line[0] = bytes - 26 + 'a' - 1;
2894                 encode_85(line + 1, cp, bytes);
2895                 cp = (char *) cp + bytes;
2896
2897                 len = strlen(line);
2898                 line[len++] = '\n';
2899                 line[len] = '\0';
2900
2901                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2902                                  line, len, 0);
2903         }
2904         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2905         free(data);
2906 }
2907
2908 static void emit_binary_diff(struct diff_options *o,
2909                              mmfile_t *one, mmfile_t *two)
2910 {
2911         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2912         emit_binary_diff_body(o, one, two);
2913         emit_binary_diff_body(o, two, one);
2914 }
2915
2916 int diff_filespec_is_binary(struct diff_filespec *one)
2917 {
2918         if (one->is_binary == -1) {
2919                 diff_filespec_load_driver(one);
2920                 if (one->driver->binary != -1)
2921                         one->is_binary = one->driver->binary;
2922                 else {
2923                         if (!one->data && DIFF_FILE_VALID(one))
2924                                 diff_populate_filespec(one, CHECK_BINARY);
2925                         if (one->is_binary == -1 && one->data)
2926                                 one->is_binary = buffer_is_binary(one->data,
2927                                                 one->size);
2928                         if (one->is_binary == -1)
2929                                 one->is_binary = 0;
2930                 }
2931         }
2932         return one->is_binary;
2933 }
2934
2935 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2936 {
2937         diff_filespec_load_driver(one);
2938         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2939 }
2940
2941 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2942 {
2943         if (!options->a_prefix)
2944                 options->a_prefix = a;
2945         if (!options->b_prefix)
2946                 options->b_prefix = b;
2947 }
2948
2949 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2950 {
2951         if (!DIFF_FILE_VALID(one))
2952                 return NULL;
2953
2954         diff_filespec_load_driver(one);
2955         return userdiff_get_textconv(one->driver);
2956 }
2957
2958 static void builtin_diff(const char *name_a,
2959                          const char *name_b,
2960                          struct diff_filespec *one,
2961                          struct diff_filespec *two,
2962                          const char *xfrm_msg,
2963                          int must_show_header,
2964                          struct diff_options *o,
2965                          int complete_rewrite)
2966 {
2967         mmfile_t mf1, mf2;
2968         const char *lbl[2];
2969         char *a_one, *b_two;
2970         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2971         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2972         const char *a_prefix, *b_prefix;
2973         struct userdiff_driver *textconv_one = NULL;
2974         struct userdiff_driver *textconv_two = NULL;
2975         struct strbuf header = STRBUF_INIT;
2976         const char *line_prefix = diff_line_prefix(o);
2977
2978         diff_set_mnemonic_prefix(o, "a/", "b/");
2979         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2980                 a_prefix = o->b_prefix;
2981                 b_prefix = o->a_prefix;
2982         } else {
2983                 a_prefix = o->a_prefix;
2984                 b_prefix = o->b_prefix;
2985         }
2986
2987         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2988             (!one->mode || S_ISGITLINK(one->mode)) &&
2989             (!two->mode || S_ISGITLINK(two->mode))) {
2990                 show_submodule_summary(o, one->path ? one->path : two->path,
2991                                 &one->oid, &two->oid,
2992                                 two->dirty_submodule);
2993                 return;
2994         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2995                    (!one->mode || S_ISGITLINK(one->mode)) &&
2996                    (!two->mode || S_ISGITLINK(two->mode))) {
2997                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
2998                                 &one->oid, &two->oid,
2999                                 two->dirty_submodule);
3000                 return;
3001         }
3002
3003         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
3004                 textconv_one = get_textconv(one);
3005                 textconv_two = get_textconv(two);
3006         }
3007
3008         /* Never use a non-valid filename anywhere if at all possible */
3009         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3010         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3011
3012         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3013         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3014         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3015         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3016         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3017         if (lbl[0][0] == '/') {
3018                 /* /dev/null */
3019                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3020                 if (xfrm_msg)
3021                         strbuf_addstr(&header, xfrm_msg);
3022                 must_show_header = 1;
3023         }
3024         else if (lbl[1][0] == '/') {
3025                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3026                 if (xfrm_msg)
3027                         strbuf_addstr(&header, xfrm_msg);
3028                 must_show_header = 1;
3029         }
3030         else {
3031                 if (one->mode != two->mode) {
3032                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3033                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3034                         must_show_header = 1;
3035                 }
3036                 if (xfrm_msg)
3037                         strbuf_addstr(&header, xfrm_msg);
3038
3039                 /*
3040                  * we do not run diff between different kind
3041                  * of objects.
3042                  */
3043                 if ((one->mode ^ two->mode) & S_IFMT)
3044                         goto free_ab_and_return;
3045                 if (complete_rewrite &&
3046                     (textconv_one || !diff_filespec_is_binary(one)) &&
3047                     (textconv_two || !diff_filespec_is_binary(two))) {
3048                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3049                                          header.buf, header.len, 0);
3050                         strbuf_reset(&header);
3051                         emit_rewrite_diff(name_a, name_b, one, two,
3052                                                 textconv_one, textconv_two, o);
3053                         o->found_changes = 1;
3054                         goto free_ab_and_return;
3055                 }
3056         }
3057
3058         if (o->irreversible_delete && lbl[1][0] == '/') {
3059                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3060                                  header.len, 0);
3061                 strbuf_reset(&header);
3062                 goto free_ab_and_return;
3063         } else if (!DIFF_OPT_TST(o, TEXT) &&
3064             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3065               (!textconv_two && diff_filespec_is_binary(two)) )) {
3066                 struct strbuf sb = STRBUF_INIT;
3067                 if (!one->data && !two->data &&
3068                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3069                     !DIFF_OPT_TST(o, BINARY)) {
3070                         if (!oidcmp(&one->oid, &two->oid)) {
3071                                 if (must_show_header)
3072                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3073                                                          header.buf, header.len,
3074                                                          0);
3075                                 goto free_ab_and_return;
3076                         }
3077                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3078                                          header.buf, header.len, 0);
3079                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3080                                     diff_line_prefix(o), lbl[0], lbl[1]);
3081                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3082                                          sb.buf, sb.len, 0);
3083                         strbuf_release(&sb);
3084                         goto free_ab_and_return;
3085                 }
3086                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3087                         die("unable to read files to diff");
3088                 /* Quite common confusing case */
3089                 if (mf1.size == mf2.size &&
3090                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3091                         if (must_show_header)
3092                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3093                                                  header.buf, header.len, 0);
3094                         goto free_ab_and_return;
3095                 }
3096                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3097                 strbuf_reset(&header);
3098                 if (DIFF_OPT_TST(o, BINARY))
3099                         emit_binary_diff(o, &mf1, &mf2);
3100                 else {
3101                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3102                                     diff_line_prefix(o), lbl[0], lbl[1]);
3103                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3104                                          sb.buf, sb.len, 0);
3105                         strbuf_release(&sb);
3106                 }
3107                 o->found_changes = 1;
3108         } else {
3109                 /* Crazy xdl interfaces.. */
3110                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3111                 const char *v;
3112                 xpparam_t xpp;
3113                 xdemitconf_t xecfg;
3114                 struct emit_callback ecbdata;
3115                 const struct userdiff_funcname *pe;
3116
3117                 if (must_show_header) {
3118                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3119                                          header.buf, header.len, 0);
3120                         strbuf_reset(&header);
3121                 }
3122
3123                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3124                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3125
3126                 pe = diff_funcname_pattern(one);
3127                 if (!pe)
3128                         pe = diff_funcname_pattern(two);
3129
3130                 memset(&xpp, 0, sizeof(xpp));
3131                 memset(&xecfg, 0, sizeof(xecfg));
3132                 memset(&ecbdata, 0, sizeof(ecbdata));
3133                 ecbdata.label_path = lbl;
3134                 ecbdata.color_diff = want_color(o->use_color);
3135                 ecbdata.ws_rule = whitespace_rule(name_b);
3136                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3137                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3138                 ecbdata.opt = o;
3139                 ecbdata.header = header.len ? &header : NULL;
3140                 xpp.flags = o->xdl_opts;
3141                 xecfg.ctxlen = o->context;
3142                 xecfg.interhunkctxlen = o->interhunkcontext;
3143                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3144                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
3145                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3146                 if (pe)
3147                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3148                 if (!diffopts)
3149                         ;
3150                 else if (skip_prefix(diffopts, "--unified=", &v))
3151                         xecfg.ctxlen = strtoul(v, NULL, 10);
3152                 else if (skip_prefix(diffopts, "-u", &v))
3153                         xecfg.ctxlen = strtoul(v, NULL, 10);
3154                 if (o->word_diff)
3155                         init_diff_words_data(&ecbdata, o, one, two);
3156                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3157                                   &xpp, &xecfg))
3158                         die("unable to generate diff for %s", one->path);
3159                 if (o->word_diff)
3160                         free_diff_words_data(&ecbdata);
3161                 if (textconv_one)
3162                         free(mf1.ptr);
3163                 if (textconv_two)
3164                         free(mf2.ptr);
3165                 xdiff_clear_find_func(&xecfg);
3166         }
3167
3168  free_ab_and_return:
3169         strbuf_release(&header);
3170         diff_free_filespec_data(one);
3171         diff_free_filespec_data(two);
3172         free(a_one);
3173         free(b_two);
3174         return;
3175 }
3176
3177 static void builtin_diffstat(const char *name_a, const char *name_b,
3178                              struct diff_filespec *one,
3179                              struct diff_filespec *two,
3180                              struct diffstat_t *diffstat,
3181                              struct diff_options *o,
3182                              struct diff_filepair *p)
3183 {
3184         mmfile_t mf1, mf2;
3185         struct diffstat_file *data;
3186         int same_contents;
3187         int complete_rewrite = 0;
3188
3189         if (!DIFF_PAIR_UNMERGED(p)) {
3190                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3191                         complete_rewrite = 1;
3192         }
3193
3194         data = diffstat_add(diffstat, name_a, name_b);
3195         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3196
3197         if (!one || !two) {
3198                 data->is_unmerged = 1;
3199                 return;
3200         }
3201
3202         same_contents = !oidcmp(&one->oid, &two->oid);
3203
3204         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3205                 data->is_binary = 1;
3206                 if (same_contents) {
3207                         data->added = 0;
3208                         data->deleted = 0;
3209                 } else {
3210                         data->added = diff_filespec_size(two);
3211                         data->deleted = diff_filespec_size(one);
3212                 }
3213         }
3214
3215         else if (complete_rewrite) {
3216                 diff_populate_filespec(one, 0);
3217                 diff_populate_filespec(two, 0);
3218                 data->deleted = count_lines(one->data, one->size);
3219                 data->added = count_lines(two->data, two->size);
3220         }
3221
3222         else if (!same_contents) {
3223                 /* Crazy xdl interfaces.. */
3224                 xpparam_t xpp;
3225                 xdemitconf_t xecfg;
3226
3227                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3228                         die("unable to read files to diff");
3229
3230                 memset(&xpp, 0, sizeof(xpp));
3231                 memset(&xecfg, 0, sizeof(xecfg));
3232                 xpp.flags = o->xdl_opts;
3233                 xecfg.ctxlen = o->context;
3234                 xecfg.interhunkctxlen = o->interhunkcontext;
3235                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3236                                   &xpp, &xecfg))
3237                         die("unable to generate diffstat for %s", one->path);
3238         }
3239
3240         diff_free_filespec_data(one);
3241         diff_free_filespec_data(two);
3242 }
3243
3244 static void builtin_checkdiff(const char *name_a, const char *name_b,
3245                               const char *attr_path,
3246                               struct diff_filespec *one,
3247                               struct diff_filespec *two,
3248                               struct diff_options *o)
3249 {
3250         mmfile_t mf1, mf2;
3251         struct checkdiff_t data;
3252
3253         if (!two)
3254                 return;
3255
3256         memset(&data, 0, sizeof(data));
3257         data.filename = name_b ? name_b : name_a;
3258         data.lineno = 0;
3259         data.o = o;
3260         data.ws_rule = whitespace_rule(attr_path);
3261         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3262
3263         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3264                 die("unable to read files to diff");
3265
3266         /*
3267          * All the other codepaths check both sides, but not checking
3268          * the "old" side here is deliberate.  We are checking the newly
3269          * introduced changes, and as long as the "new" side is text, we
3270          * can and should check what it introduces.
3271          */
3272         if (diff_filespec_is_binary(two))
3273                 goto free_and_return;
3274         else {
3275                 /* Crazy xdl interfaces.. */
3276                 xpparam_t xpp;
3277                 xdemitconf_t xecfg;
3278
3279                 memset(&xpp, 0, sizeof(xpp));
3280                 memset(&xecfg, 0, sizeof(xecfg));
3281                 xecfg.ctxlen = 1; /* at least one context line */
3282                 xpp.flags = 0;
3283                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3284                                   &xpp, &xecfg))
3285                         die("unable to generate checkdiff for %s", one->path);
3286
3287                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3288                         struct emit_callback ecbdata;
3289                         int blank_at_eof;
3290
3291                         ecbdata.ws_rule = data.ws_rule;
3292                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3293                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3294
3295                         if (blank_at_eof) {
3296                                 static char *err;
3297                                 if (!err)
3298                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3299                                 fprintf(o->file, "%s:%d: %s.\n",
3300                                         data.filename, blank_at_eof, err);
3301                                 data.status = 1; /* report errors */
3302                         }
3303                 }
3304         }
3305  free_and_return:
3306         diff_free_filespec_data(one);
3307         diff_free_filespec_data(two);
3308         if (data.status)
3309                 DIFF_OPT_SET(o, CHECK_FAILED);
3310 }
3311
3312 struct diff_filespec *alloc_filespec(const char *path)
3313 {
3314         struct diff_filespec *spec;
3315
3316         FLEXPTR_ALLOC_STR(spec, path, path);
3317         spec->count = 1;
3318         spec->is_binary = -1;
3319         return spec;
3320 }
3321
3322 void free_filespec(struct diff_filespec *spec)
3323 {
3324         if (!--spec->count) {
3325                 diff_free_filespec_data(spec);
3326                 free(spec);
3327         }
3328 }
3329
3330 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3331                    int oid_valid, unsigned short mode)
3332 {
3333         if (mode) {
3334                 spec->mode = canon_mode(mode);
3335                 oidcpy(&spec->oid, oid);
3336                 spec->oid_valid = oid_valid;
3337         }
3338 }
3339
3340 /*
3341  * Given a name and sha1 pair, if the index tells us the file in
3342  * the work tree has that object contents, return true, so that
3343  * prepare_temp_file() does not have to inflate and extract.
3344  */
3345 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3346 {
3347         const struct cache_entry *ce;
3348         struct stat st;
3349         int pos, len;
3350
3351         /*
3352          * We do not read the cache ourselves here, because the
3353          * benchmark with my previous version that always reads cache
3354          * shows that it makes things worse for diff-tree comparing
3355          * two linux-2.6 kernel trees in an already checked out work
3356          * tree.  This is because most diff-tree comparisons deal with
3357          * only a small number of files, while reading the cache is
3358          * expensive for a large project, and its cost outweighs the
3359          * savings we get by not inflating the object to a temporary
3360          * file.  Practically, this code only helps when we are used
3361          * by diff-cache --cached, which does read the cache before
3362          * calling us.
3363          */
3364         if (!active_cache)
3365                 return 0;
3366
3367         /* We want to avoid the working directory if our caller
3368          * doesn't need the data in a normal file, this system
3369          * is rather slow with its stat/open/mmap/close syscalls,
3370          * and the object is contained in a pack file.  The pack
3371          * is probably already open and will be faster to obtain
3372          * the data through than the working directory.  Loose
3373          * objects however would tend to be slower as they need
3374          * to be individually opened and inflated.
3375          */
3376         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3377                 return 0;
3378
3379         /*
3380          * Similarly, if we'd have to convert the file contents anyway, that
3381          * makes the optimization not worthwhile.
3382          */
3383         if (!want_file && would_convert_to_git(&the_index, name))
3384                 return 0;
3385
3386         len = strlen(name);
3387         pos = cache_name_pos(name, len);
3388         if (pos < 0)
3389                 return 0;
3390         ce = active_cache[pos];
3391
3392         /*
3393          * This is not the sha1 we are looking for, or
3394          * unreusable because it is not a regular file.
3395          */
3396         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3397                 return 0;
3398
3399         /*
3400          * If ce is marked as "assume unchanged", there is no
3401          * guarantee that work tree matches what we are looking for.
3402          */
3403         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3404                 return 0;
3405
3406         /*
3407          * If ce matches the file in the work tree, we can reuse it.
3408          */
3409         if (ce_uptodate(ce) ||
3410             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3411                 return 1;
3412
3413         return 0;
3414 }
3415
3416 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3417 {
3418         struct strbuf buf = STRBUF_INIT;
3419         char *dirty = "";
3420
3421         /* Are we looking at the work tree? */
3422         if (s->dirty_submodule)
3423                 dirty = "-dirty";
3424
3425         strbuf_addf(&buf, "Subproject commit %s%s\n",
3426                     oid_to_hex(&s->oid), dirty);
3427         s->size = buf.len;
3428         if (size_only) {
3429                 s->data = NULL;
3430                 strbuf_release(&buf);
3431         } else {
3432                 s->data = strbuf_detach(&buf, NULL);
3433                 s->should_free = 1;
3434         }
3435         return 0;
3436 }
3437
3438 /*
3439  * While doing rename detection and pickaxe operation, we may need to
3440  * grab the data for the blob (or file) for our own in-core comparison.
3441  * diff_filespec has data and size fields for this purpose.
3442  */
3443 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3444 {
3445         int size_only = flags & CHECK_SIZE_ONLY;
3446         int err = 0;
3447         /*
3448          * demote FAIL to WARN to allow inspecting the situation
3449          * instead of refusing.
3450          */
3451         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3452                                     ? SAFE_CRLF_WARN
3453                                     : safe_crlf);
3454
3455         if (!DIFF_FILE_VALID(s))
3456                 die("internal error: asking to populate invalid file.");
3457         if (S_ISDIR(s->mode))
3458                 return -1;
3459
3460         if (s->data)
3461                 return 0;
3462
3463         if (size_only && 0 < s->size)
3464                 return 0;
3465
3466         if (S_ISGITLINK(s->mode))
3467                 return diff_populate_gitlink(s, size_only);
3468
3469         if (!s->oid_valid ||
3470             reuse_worktree_file(s->path, &s->oid, 0)) {
3471                 struct strbuf buf = STRBUF_INIT;
3472                 struct stat st;
3473                 int fd;
3474
3475                 if (lstat(s->path, &st) < 0) {
3476                         if (errno == ENOENT) {
3477                         err_empty:
3478                                 err = -1;
3479                         empty:
3480                                 s->data = (char *)"";
3481                                 s->size = 0;
3482                                 return err;
3483                         }
3484                 }
3485                 s->size = xsize_t(st.st_size);
3486                 if (!s->size)
3487                         goto empty;
3488                 if (S_ISLNK(st.st_mode)) {
3489                         struct strbuf sb = STRBUF_INIT;
3490
3491                         if (strbuf_readlink(&sb, s->path, s->size))
3492                                 goto err_empty;
3493                         s->size = sb.len;
3494                         s->data = strbuf_detach(&sb, NULL);
3495                         s->should_free = 1;
3496                         return 0;
3497                 }
3498
3499                 /*
3500                  * Even if the caller would be happy with getting
3501                  * only the size, we cannot return early at this
3502                  * point if the path requires us to run the content
3503                  * conversion.
3504                  */
3505                 if (size_only && !would_convert_to_git(&the_index, s->path))
3506                         return 0;
3507
3508                 /*
3509                  * Note: this check uses xsize_t(st.st_size) that may
3510                  * not be the true size of the blob after it goes
3511                  * through convert_to_git().  This may not strictly be
3512                  * correct, but the whole point of big_file_threshold
3513                  * and is_binary check being that we want to avoid
3514                  * opening the file and inspecting the contents, this
3515                  * is probably fine.
3516                  */
3517                 if ((flags & CHECK_BINARY) &&
3518                     s->size > big_file_threshold && s->is_binary == -1) {
3519                         s->is_binary = 1;
3520                         return 0;
3521                 }
3522                 fd = open(s->path, O_RDONLY);
3523                 if (fd < 0)
3524                         goto err_empty;
3525                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3526                 close(fd);
3527                 s->should_munmap = 1;
3528
3529                 /*
3530                  * Convert from working tree format to canonical git format
3531                  */
3532                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3533                         size_t size = 0;
3534                         munmap(s->data, s->size);
3535                         s->should_munmap = 0;
3536                         s->data = strbuf_detach(&buf, &size);
3537                         s->size = size;
3538                         s->should_free = 1;
3539                 }
3540         }
3541         else {
3542                 enum object_type type;
3543                 if (size_only || (flags & CHECK_BINARY)) {
3544                         type = sha1_object_info(s->oid.hash, &s->size);
3545                         if (type < 0)
3546                                 die("unable to read %s",
3547                                     oid_to_hex(&s->oid));
3548                         if (size_only)
3549                                 return 0;
3550                         if (s->size > big_file_threshold && s->is_binary == -1) {
3551                                 s->is_binary = 1;
3552                                 return 0;
3553                         }
3554                 }
3555                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3556                 if (!s->data)
3557                         die("unable to read %s", oid_to_hex(&s->oid));
3558                 s->should_free = 1;
3559         }
3560         return 0;
3561 }
3562
3563 void diff_free_filespec_blob(struct diff_filespec *s)
3564 {
3565         if (s->should_free)
3566                 free(s->data);
3567         else if (s->should_munmap)
3568                 munmap(s->data, s->size);
3569
3570         if (s->should_free || s->should_munmap) {
3571                 s->should_free = s->should_munmap = 0;
3572                 s->data = NULL;
3573         }
3574 }
3575
3576 void diff_free_filespec_data(struct diff_filespec *s)
3577 {
3578         diff_free_filespec_blob(s);
3579         FREE_AND_NULL(s->cnt_data);
3580 }
3581
3582 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3583                            void *blob,
3584                            unsigned long size,
3585                            const struct object_id *oid,
3586                            int mode)
3587 {
3588         int fd;
3589         struct strbuf buf = STRBUF_INIT;
3590         struct strbuf template = STRBUF_INIT;
3591         char *path_dup = xstrdup(path);
3592         const char *base = basename(path_dup);
3593
3594         /* Generate "XXXXXX_basename.ext" */
3595         strbuf_addstr(&template, "XXXXXX_");
3596         strbuf_addstr(&template, base);
3597
3598         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
3599         if (fd < 0)
3600                 die_errno("unable to create temp-file");
3601         if (convert_to_working_tree(path,
3602                         (const char *)blob, (size_t)size, &buf)) {
3603                 blob = buf.buf;
3604                 size = buf.len;
3605         }
3606         if (write_in_full(fd, blob, size) != size)
3607                 die_errno("unable to write temp-file");
3608         close_tempfile(&temp->tempfile);
3609         temp->name = get_tempfile_path(&temp->tempfile);
3610         oid_to_hex_r(temp->hex, oid);
3611         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3612         strbuf_release(&buf);
3613         strbuf_release(&template);
3614         free(path_dup);
3615 }
3616
3617 static struct diff_tempfile *prepare_temp_file(const char *name,
3618                 struct diff_filespec *one)
3619 {
3620         struct diff_tempfile *temp = claim_diff_tempfile();
3621
3622         if (!DIFF_FILE_VALID(one)) {
3623         not_a_valid_file:
3624                 /* A '-' entry produces this for file-2, and
3625                  * a '+' entry produces this for file-1.
3626                  */
3627                 temp->name = "/dev/null";
3628                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3629                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3630                 return temp;
3631         }
3632
3633         if (!S_ISGITLINK(one->mode) &&
3634             (!one->oid_valid ||
3635              reuse_worktree_file(name, &one->oid, 1))) {
3636                 struct stat st;
3637                 if (lstat(name, &st) < 0) {
3638                         if (errno == ENOENT)
3639                                 goto not_a_valid_file;
3640                         die_errno("stat(%s)", name);
3641                 }
3642                 if (S_ISLNK(st.st_mode)) {
3643                         struct strbuf sb = STRBUF_INIT;
3644                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3645                                 die_errno("readlink(%s)", name);
3646                         prep_temp_blob(name, temp, sb.buf, sb.len,
3647                                        (one->oid_valid ?
3648                                         &one->oid : &null_oid),
3649                                        (one->oid_valid ?
3650                                         one->mode : S_IFLNK));
3651                         strbuf_release(&sb);
3652                 }
3653                 else {
3654                         /* we can borrow from the file in the work tree */
3655                         temp->name = name;
3656                         if (!one->oid_valid)
3657                                 oid_to_hex_r(temp->hex, &null_oid);
3658                         else
3659                                 oid_to_hex_r(temp->hex, &one->oid);
3660                         /* Even though we may sometimes borrow the
3661                          * contents from the work tree, we always want
3662                          * one->mode.  mode is trustworthy even when
3663                          * !(one->oid_valid), as long as
3664                          * DIFF_FILE_VALID(one).
3665                          */
3666                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3667                 }
3668                 return temp;
3669         }
3670         else {
3671                 if (diff_populate_filespec(one, 0))
3672                         die("cannot read data blob for %s", one->path);
3673                 prep_temp_blob(name, temp, one->data, one->size,
3674                                &one->oid, one->mode);
3675         }
3676         return temp;
3677 }
3678
3679 static void add_external_diff_name(struct argv_array *argv,
3680                                    const char *name,
3681                                    struct diff_filespec *df)
3682 {
3683         struct diff_tempfile *temp = prepare_temp_file(name, df);
3684         argv_array_push(argv, temp->name);
3685         argv_array_push(argv, temp->hex);
3686         argv_array_push(argv, temp->mode);
3687 }
3688
3689 /* An external diff command takes:
3690  *
3691  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3692  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3693  *
3694  */
3695 static void run_external_diff(const char *pgm,
3696                               const char *name,
3697                               const char *other,
3698                               struct diff_filespec *one,
3699                               struct diff_filespec *two,
3700                               const char *xfrm_msg,
3701                               int complete_rewrite,
3702                               struct diff_options *o)
3703 {
3704         struct argv_array argv = ARGV_ARRAY_INIT;
3705         struct argv_array env = ARGV_ARRAY_INIT;
3706         struct diff_queue_struct *q = &diff_queued_diff;
3707
3708         argv_array_push(&argv, pgm);
3709         argv_array_push(&argv, name);
3710
3711         if (one && two) {
3712                 add_external_diff_name(&argv, name, one);
3713                 if (!other)
3714                         add_external_diff_name(&argv, name, two);
3715                 else {
3716                         add_external_diff_name(&argv, other, two);
3717                         argv_array_push(&argv, other);
3718                         argv_array_push(&argv, xfrm_msg);
3719                 }
3720         }
3721
3722         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3723         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3724
3725         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3726                 die(_("external diff died, stopping at %s"), name);
3727
3728         remove_tempfile();
3729         argv_array_clear(&argv);
3730         argv_array_clear(&env);
3731 }
3732
3733 static int similarity_index(struct diff_filepair *p)
3734 {
3735         return p->score * 100 / MAX_SCORE;
3736 }
3737
3738 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3739 {
3740         if (startup_info->have_repository)
3741                 return find_unique_abbrev(oid->hash, abbrev);
3742         else {
3743                 char *hex = oid_to_hex(oid);
3744                 if (abbrev < 0)
3745                         abbrev = FALLBACK_DEFAULT_ABBREV;
3746                 if (abbrev > GIT_SHA1_HEXSZ)
3747                         die("BUG: oid abbreviation out of range: %d", abbrev);
3748                 if (abbrev)
3749                         hex[abbrev] = '\0';
3750                 return hex;
3751         }
3752 }
3753
3754 static void fill_metainfo(struct strbuf *msg,
3755                           const char *name,
3756                           const char *other,
3757                           struct diff_filespec *one,
3758                           struct diff_filespec *two,
3759                           struct diff_options *o,
3760                           struct diff_filepair *p,
3761                           int *must_show_header,
3762                           int use_color)
3763 {
3764         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3765         const char *reset = diff_get_color(use_color, DIFF_RESET);
3766         const char *line_prefix = diff_line_prefix(o);
3767
3768         *must_show_header = 1;
3769         strbuf_init(msg, PATH_MAX * 2 + 300);
3770         switch (p->status) {
3771         case DIFF_STATUS_COPIED:
3772                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3773                             line_prefix, set, similarity_index(p));
3774                 strbuf_addf(msg, "%s\n%s%scopy from ",
3775                             reset,  line_prefix, set);
3776                 quote_c_style(name, msg, NULL, 0);
3777                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3778                 quote_c_style(other, msg, NULL, 0);
3779                 strbuf_addf(msg, "%s\n", reset);
3780                 break;
3781         case DIFF_STATUS_RENAMED:
3782                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3783                             line_prefix, set, similarity_index(p));
3784                 strbuf_addf(msg, "%s\n%s%srename from ",
3785                             reset, line_prefix, set);
3786                 quote_c_style(name, msg, NULL, 0);
3787                 strbuf_addf(msg, "%s\n%s%srename to ",
3788                             reset, line_prefix, set);
3789                 quote_c_style(other, msg, NULL, 0);
3790                 strbuf_addf(msg, "%s\n", reset);
3791                 break;
3792         case DIFF_STATUS_MODIFIED:
3793                 if (p->score) {
3794                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3795                                     line_prefix,
3796                                     set, similarity_index(p), reset);
3797                         break;
3798                 }
3799                 /* fallthru */
3800         default:
3801                 *must_show_header = 0;
3802         }
3803         if (one && two && oidcmp(&one->oid, &two->oid)) {
3804                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3805
3806                 if (DIFF_OPT_TST(o, BINARY)) {
3807                         mmfile_t mf;
3808                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3809                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3810                                 abbrev = 40;
3811                 }
3812                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3813                             diff_abbrev_oid(&one->oid, abbrev),
3814                             diff_abbrev_oid(&two->oid, abbrev));
3815                 if (one->mode == two->mode)
3816                         strbuf_addf(msg, " %06o", one->mode);
3817                 strbuf_addf(msg, "%s\n", reset);
3818         }
3819 }
3820
3821 static void run_diff_cmd(const char *pgm,
3822                          const char *name,
3823                          const char *other,
3824                          const char *attr_path,
3825                          struct diff_filespec *one,
3826                          struct diff_filespec *two,
3827                          struct strbuf *msg,
3828                          struct diff_options *o,
3829                          struct diff_filepair *p)
3830 {
3831         const char *xfrm_msg = NULL;
3832         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3833         int must_show_header = 0;
3834
3835
3836         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3837                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3838                 if (drv && drv->external)
3839                         pgm = drv->external;
3840         }
3841
3842         if (msg) {
3843                 /*
3844                  * don't use colors when the header is intended for an
3845                  * external diff driver
3846                  */
3847                 fill_metainfo(msg, name, other, one, two, o, p,
3848                               &must_show_header,
3849                               want_color(o->use_color) && !pgm);
3850                 xfrm_msg = msg->len ? msg->buf : NULL;
3851         }
3852
3853         if (pgm) {
3854                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3855                                   complete_rewrite, o);
3856                 return;
3857         }
3858         if (one && two)
3859                 builtin_diff(name, other ? other : name,
3860                              one, two, xfrm_msg, must_show_header,
3861                              o, complete_rewrite);
3862         else
3863                 fprintf(o->file, "* Unmerged path %s\n", name);
3864 }
3865
3866 static void diff_fill_oid_info(struct diff_filespec *one)
3867 {
3868         if (DIFF_FILE_VALID(one)) {
3869                 if (!one->oid_valid) {
3870                         struct stat st;
3871                         if (one->is_stdin) {
3872                                 oidclr(&one->oid);
3873                                 return;
3874                         }
3875                         if (lstat(one->path, &st) < 0)
3876                                 die_errno("stat '%s'", one->path);
3877                         if (index_path(one->oid.hash, one->path, &st, 0))
3878                                 die("cannot hash %s", one->path);
3879                 }
3880         }
3881         else
3882                 oidclr(&one->oid);
3883 }
3884
3885 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3886 {
3887         /* Strip the prefix but do not molest /dev/null and absolute paths */
3888         if (*namep && **namep != '/') {
3889                 *namep += prefix_length;
3890                 if (**namep == '/')
3891                         ++*namep;
3892         }
3893         if (*otherp && **otherp != '/') {
3894                 *otherp += prefix_length;
3895                 if (**otherp == '/')
3896                         ++*otherp;
3897         }
3898 }
3899
3900 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3901 {
3902         const char *pgm = external_diff();
3903         struct strbuf msg;
3904         struct diff_filespec *one = p->one;
3905         struct diff_filespec *two = p->two;
3906         const char *name;
3907         const char *other;
3908         const char *attr_path;
3909
3910         name  = one->path;
3911         other = (strcmp(name, two->path) ? two->path : NULL);
3912         attr_path = name;
3913         if (o->prefix_length)
3914                 strip_prefix(o->prefix_length, &name, &other);
3915
3916         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3917                 pgm = NULL;
3918
3919         if (DIFF_PAIR_UNMERGED(p)) {
3920                 run_diff_cmd(pgm, name, NULL, attr_path,
3921                              NULL, NULL, NULL, o, p);
3922                 return;
3923         }
3924
3925         diff_fill_oid_info(one);
3926         diff_fill_oid_info(two);
3927
3928         if (!pgm &&
3929             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3930             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3931                 /*
3932                  * a filepair that changes between file and symlink
3933                  * needs to be split into deletion and creation.
3934                  */
3935                 struct diff_filespec *null = alloc_filespec(two->path);
3936                 run_diff_cmd(NULL, name, other, attr_path,
3937                              one, null, &msg, o, p);
3938                 free(null);
3939                 strbuf_release(&msg);
3940
3941                 null = alloc_filespec(one->path);
3942                 run_diff_cmd(NULL, name, other, attr_path,
3943                              null, two, &msg, o, p);
3944                 free(null);
3945         }
3946         else
3947                 run_diff_cmd(pgm, name, other, attr_path,
3948                              one, two, &msg, o, p);
3949
3950         strbuf_release(&msg);
3951 }
3952
3953 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3954                          struct diffstat_t *diffstat)
3955 {
3956         const char *name;
3957         const char *other;
3958
3959         if (DIFF_PAIR_UNMERGED(p)) {
3960                 /* unmerged */
3961                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3962                 return;
3963         }
3964
3965         name = p->one->path;
3966         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3967
3968         if (o->prefix_length)
3969                 strip_prefix(o->prefix_length, &name, &other);
3970
3971         diff_fill_oid_info(p->one);
3972         diff_fill_oid_info(p->two);
3973
3974         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3975 }
3976
3977 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3978 {
3979         const char *name;
3980         const char *other;
3981         const char *attr_path;
3982
3983         if (DIFF_PAIR_UNMERGED(p)) {
3984                 /* unmerged */
3985                 return;
3986         }
3987
3988         name = p->one->path;
3989         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3990         attr_path = other ? other : name;
3991
3992         if (o->prefix_length)
3993                 strip_prefix(o->prefix_length, &name, &other);
3994
3995         diff_fill_oid_info(p->one);
3996         diff_fill_oid_info(p->two);
3997
3998         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3999 }
4000
4001 void diff_setup(struct diff_options *options)
4002 {
4003         memcpy(options, &default_diff_options, sizeof(*options));
4004
4005         options->file = stdout;
4006
4007         options->abbrev = DEFAULT_ABBREV;
4008         options->line_termination = '\n';
4009         options->break_opt = -1;
4010         options->rename_limit = -1;
4011         options->dirstat_permille = diff_dirstat_permille_default;
4012         options->context = diff_context_default;
4013         options->interhunkcontext = diff_interhunk_context_default;
4014         options->ws_error_highlight = ws_error_highlight_default;
4015         DIFF_OPT_SET(options, RENAME_EMPTY);
4016
4017         /* pathchange left =NULL by default */
4018         options->change = diff_change;
4019         options->add_remove = diff_addremove;
4020         options->use_color = diff_use_color_default;
4021         options->detect_rename = diff_detect_rename_default;
4022         options->xdl_opts |= diff_algorithm;
4023         if (diff_indent_heuristic)
4024                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4025
4026         options->orderfile = diff_order_file_cfg;
4027
4028         if (diff_no_prefix) {
4029                 options->a_prefix = options->b_prefix = "";
4030         } else if (!diff_mnemonic_prefix) {
4031                 options->a_prefix = "a/";
4032                 options->b_prefix = "b/";
4033         }
4034
4035         options->color_moved = diff_color_moved_default;
4036 }
4037
4038 void diff_setup_done(struct diff_options *options)
4039 {
4040         int count = 0;
4041
4042         if (options->set_default)
4043                 options->set_default(options);
4044
4045         if (options->output_format & DIFF_FORMAT_NAME)
4046                 count++;
4047         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
4048                 count++;
4049         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
4050                 count++;
4051         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
4052                 count++;
4053         if (count > 1)
4054                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4055
4056         /*
4057          * Most of the time we can say "there are changes"
4058          * only by checking if there are changed paths, but
4059          * --ignore-whitespace* options force us to look
4060          * inside contents.
4061          */
4062
4063         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
4064             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
4065             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
4066                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
4067         else
4068                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
4069
4070         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
4071                 options->detect_rename = DIFF_DETECT_COPY;
4072
4073         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
4074                 options->prefix = NULL;
4075         if (options->prefix)
4076                 options->prefix_length = strlen(options->prefix);
4077         else
4078                 options->prefix_length = 0;
4079
4080         if (options->output_format & (DIFF_FORMAT_NAME |
4081                                       DIFF_FORMAT_NAME_STATUS |
4082                                       DIFF_FORMAT_CHECKDIFF |
4083                                       DIFF_FORMAT_NO_OUTPUT))
4084                 options->output_format &= ~(DIFF_FORMAT_RAW |
4085                                             DIFF_FORMAT_NUMSTAT |
4086                                             DIFF_FORMAT_DIFFSTAT |
4087                                             DIFF_FORMAT_SHORTSTAT |
4088                                             DIFF_FORMAT_DIRSTAT |
4089                                             DIFF_FORMAT_SUMMARY |
4090                                             DIFF_FORMAT_PATCH);
4091
4092         /*
4093          * These cases always need recursive; we do not drop caller-supplied
4094          * recursive bits for other formats here.
4095          */
4096         if (options->output_format & (DIFF_FORMAT_PATCH |
4097                                       DIFF_FORMAT_NUMSTAT |
4098                                       DIFF_FORMAT_DIFFSTAT |
4099                                       DIFF_FORMAT_SHORTSTAT |
4100                                       DIFF_FORMAT_DIRSTAT |
4101                                       DIFF_FORMAT_SUMMARY |
4102                                       DIFF_FORMAT_CHECKDIFF))
4103                 DIFF_OPT_SET(options, RECURSIVE);
4104         /*
4105          * Also pickaxe would not work very well if you do not say recursive
4106          */
4107         if (options->pickaxe)
4108                 DIFF_OPT_SET(options, RECURSIVE);
4109         /*
4110          * When patches are generated, submodules diffed against the work tree
4111          * must be checked for dirtiness too so it can be shown in the output
4112          */
4113         if (options->output_format & DIFF_FORMAT_PATCH)
4114                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
4115
4116         if (options->detect_rename && options->rename_limit < 0)
4117                 options->rename_limit = diff_rename_limit_default;
4118         if (options->setup & DIFF_SETUP_USE_CACHE) {
4119                 if (!active_cache)
4120                         /* read-cache does not die even when it fails
4121                          * so it is safe for us to do this here.  Also
4122                          * it does not smudge active_cache or active_nr
4123                          * when it fails, so we do not have to worry about
4124                          * cleaning it up ourselves either.
4125                          */
4126                         read_cache();
4127         }
4128         if (40 < options->abbrev)
4129                 options->abbrev = 40; /* full */
4130
4131         /*
4132          * It does not make sense to show the first hit we happened
4133          * to have found.  It does not make sense not to return with
4134          * exit code in such a case either.
4135          */
4136         if (DIFF_OPT_TST(options, QUICK)) {
4137                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4138                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4139         }
4140
4141         options->diff_path_counter = 0;
4142
4143         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
4144                 die(_("--follow requires exactly one pathspec"));
4145
4146         if (!options->use_color || external_diff())
4147                 options->color_moved = 0;
4148 }
4149
4150 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4151 {
4152         char c, *eq;
4153         int len;
4154
4155         if (*arg != '-')
4156                 return 0;
4157         c = *++arg;
4158         if (!c)
4159                 return 0;
4160         if (c == arg_short) {
4161                 c = *++arg;
4162                 if (!c)
4163                         return 1;
4164                 if (val && isdigit(c)) {
4165                         char *end;
4166                         int n = strtoul(arg, &end, 10);
4167                         if (*end)
4168                                 return 0;
4169                         *val = n;
4170                         return 1;
4171                 }
4172                 return 0;
4173         }
4174         if (c != '-')
4175                 return 0;
4176         arg++;
4177         eq = strchrnul(arg, '=');
4178         len = eq - arg;
4179         if (!len || strncmp(arg, arg_long, len))
4180                 return 0;
4181         if (*eq) {
4182                 int n;
4183                 char *end;
4184                 if (!isdigit(*++eq))
4185                         return 0;
4186                 n = strtoul(eq, &end, 10);
4187                 if (*end)
4188                         return 0;
4189                 *val = n;
4190         }
4191         return 1;
4192 }
4193
4194 static int diff_scoreopt_parse(const char *opt);
4195
4196 static inline int short_opt(char opt, const char **argv,
4197                             const char **optarg)
4198 {
4199         const char *arg = argv[0];
4200         if (arg[0] != '-' || arg[1] != opt)
4201                 return 0;
4202         if (arg[2] != '\0') {
4203                 *optarg = arg + 2;
4204                 return 1;
4205         }
4206         if (!argv[1])
4207                 die("Option '%c' requires a value", opt);
4208         *optarg = argv[1];
4209         return 2;
4210 }
4211
4212 int parse_long_opt(const char *opt, const char **argv,
4213                    const char **optarg)
4214 {
4215         const char *arg = argv[0];
4216         if (!skip_prefix(arg, "--", &arg))
4217                 return 0;
4218         if (!skip_prefix(arg, opt, &arg))
4219                 return 0;
4220         if (*arg == '=') { /* stuck form: --option=value */
4221                 *optarg = arg + 1;
4222                 return 1;
4223         }
4224         if (*arg != '\0')
4225                 return 0;
4226         /* separate form: --option value */
4227         if (!argv[1])
4228                 die("Option '--%s' requires a value", opt);
4229         *optarg = argv[1];
4230         return 2;
4231 }
4232
4233 static int stat_opt(struct diff_options *options, const char **av)
4234 {
4235         const char *arg = av[0];
4236         char *end;
4237         int width = options->stat_width;
4238         int name_width = options->stat_name_width;
4239         int graph_width = options->stat_graph_width;
4240         int count = options->stat_count;
4241         int argcount = 1;
4242
4243         if (!skip_prefix(arg, "--stat", &arg))
4244                 die("BUG: stat option does not begin with --stat: %s", arg);
4245         end = (char *)arg;
4246
4247         switch (*arg) {
4248         case '-':
4249                 if (skip_prefix(arg, "-width", &arg)) {
4250                         if (*arg == '=')
4251                                 width = strtoul(arg + 1, &end, 10);
4252                         else if (!*arg && !av[1])
4253                                 die_want_option("--stat-width");
4254                         else if (!*arg) {
4255                                 width = strtoul(av[1], &end, 10);
4256                                 argcount = 2;
4257                         }
4258                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4259                         if (*arg == '=')
4260                                 name_width = strtoul(arg + 1, &end, 10);
4261                         else if (!*arg && !av[1])
4262                                 die_want_option("--stat-name-width");
4263                         else if (!*arg) {
4264                                 name_width = strtoul(av[1], &end, 10);
4265                                 argcount = 2;
4266                         }
4267                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4268                         if (*arg == '=')
4269                                 graph_width = strtoul(arg + 1, &end, 10);
4270                         else if (!*arg && !av[1])
4271                                 die_want_option("--stat-graph-width");
4272                         else if (!*arg) {
4273                                 graph_width = strtoul(av[1], &end, 10);
4274                                 argcount = 2;
4275                         }
4276                 } else if (skip_prefix(arg, "-count", &arg)) {
4277                         if (*arg == '=')
4278                                 count = strtoul(arg + 1, &end, 10);
4279                         else if (!*arg && !av[1])
4280                                 die_want_option("--stat-count");
4281                         else if (!*arg) {
4282                                 count = strtoul(av[1], &end, 10);
4283                                 argcount = 2;
4284                         }
4285                 }
4286                 break;
4287         case '=':
4288                 width = strtoul(arg+1, &end, 10);
4289                 if (*end == ',')
4290                         name_width = strtoul(end+1, &end, 10);
4291                 if (*end == ',')
4292                         count = strtoul(end+1, &end, 10);
4293         }
4294
4295         /* Important! This checks all the error cases! */
4296         if (*end)
4297                 return 0;
4298         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4299         options->stat_name_width = name_width;
4300         options->stat_graph_width = graph_width;
4301         options->stat_width = width;
4302         options->stat_count = count;
4303         return argcount;
4304 }
4305
4306 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4307 {
4308         struct strbuf errmsg = STRBUF_INIT;
4309         if (parse_dirstat_params(options, params, &errmsg))
4310                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4311                     errmsg.buf);
4312         strbuf_release(&errmsg);
4313         /*
4314          * The caller knows a dirstat-related option is given from the command
4315          * line; allow it to say "return this_function();"
4316          */
4317         options->output_format |= DIFF_FORMAT_DIRSTAT;
4318         return 1;
4319 }
4320
4321 static int parse_submodule_opt(struct diff_options *options, const char *value)
4322 {
4323         if (parse_submodule_params(options, value))
4324                 die(_("Failed to parse --submodule option parameter: '%s'"),
4325                         value);
4326         return 1;
4327 }
4328
4329 static const char diff_status_letters[] = {
4330         DIFF_STATUS_ADDED,
4331         DIFF_STATUS_COPIED,
4332         DIFF_STATUS_DELETED,
4333         DIFF_STATUS_MODIFIED,
4334         DIFF_STATUS_RENAMED,
4335         DIFF_STATUS_TYPE_CHANGED,
4336         DIFF_STATUS_UNKNOWN,
4337         DIFF_STATUS_UNMERGED,
4338         DIFF_STATUS_FILTER_AON,
4339         DIFF_STATUS_FILTER_BROKEN,
4340         '\0',
4341 };
4342
4343 static unsigned int filter_bit['Z' + 1];
4344
4345 static void prepare_filter_bits(void)
4346 {
4347         int i;
4348
4349         if (!filter_bit[DIFF_STATUS_ADDED]) {
4350                 for (i = 0; diff_status_letters[i]; i++)
4351                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4352         }
4353 }
4354
4355 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4356 {
4357         return opt->filter & filter_bit[(int) status];
4358 }
4359
4360 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4361 {
4362         int i, optch;
4363
4364         prepare_filter_bits();
4365
4366         /*
4367          * If there is a negation e.g. 'd' in the input, and we haven't
4368          * initialized the filter field with another --diff-filter, start
4369          * from full set of bits, except for AON.
4370          */
4371         if (!opt->filter) {
4372                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4373                         if (optch < 'a' || 'z' < optch)
4374                                 continue;
4375                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4376                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4377                         break;
4378                 }
4379         }
4380
4381         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4382                 unsigned int bit;
4383                 int negate;
4384
4385                 if ('a' <= optch && optch <= 'z') {
4386                         negate = 1;
4387                         optch = toupper(optch);
4388                 } else {
4389                         negate = 0;
4390                 }
4391
4392                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4393                 if (!bit)
4394                         return optarg[i];
4395                 if (negate)
4396                         opt->filter &= ~bit;
4397                 else
4398                         opt->filter |= bit;
4399         }
4400         return 0;
4401 }
4402
4403 static void enable_patch_output(int *fmt) {
4404         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4405         *fmt |= DIFF_FORMAT_PATCH;
4406 }
4407
4408 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4409 {
4410         int val = parse_ws_error_highlight(arg);
4411
4412         if (val < 0) {
4413                 error("unknown value after ws-error-highlight=%.*s",
4414                       -1 - val, arg);
4415                 return 0;
4416         }
4417         opt->ws_error_highlight = val;
4418         return 1;
4419 }
4420
4421 int diff_opt_parse(struct diff_options *options,
4422                    const char **av, int ac, const char *prefix)
4423 {
4424         const char *arg = av[0];
4425         const char *optarg;
4426         int argcount;
4427
4428         if (!prefix)
4429                 prefix = "";
4430
4431         /* Output format options */
4432         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4433             || opt_arg(arg, 'U', "unified", &options->context))
4434                 enable_patch_output(&options->output_format);
4435         else if (!strcmp(arg, "--raw"))
4436                 options->output_format |= DIFF_FORMAT_RAW;
4437         else if (!strcmp(arg, "--patch-with-raw")) {
4438                 enable_patch_output(&options->output_format);
4439                 options->output_format |= DIFF_FORMAT_RAW;
4440         } else if (!strcmp(arg, "--numstat"))
4441                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4442         else if (!strcmp(arg, "--shortstat"))
4443                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4444         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
4445                 return parse_dirstat_opt(options, "");
4446         else if (skip_prefix(arg, "-X", &arg))
4447                 return parse_dirstat_opt(options, arg);
4448         else if (skip_prefix(arg, "--dirstat=", &arg))
4449                 return parse_dirstat_opt(options, arg);
4450         else if (!strcmp(arg, "--cumulative"))
4451                 return parse_dirstat_opt(options, "cumulative");
4452         else if (!strcmp(arg, "--dirstat-by-file"))
4453                 return parse_dirstat_opt(options, "files");
4454         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
4455                 parse_dirstat_opt(options, "files");
4456                 return parse_dirstat_opt(options, arg);
4457         }
4458         else if (!strcmp(arg, "--check"))
4459                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4460         else if (!strcmp(arg, "--summary"))
4461                 options->output_format |= DIFF_FORMAT_SUMMARY;
4462         else if (!strcmp(arg, "--patch-with-stat")) {
4463                 enable_patch_output(&options->output_format);
4464                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4465         } else if (!strcmp(arg, "--name-only"))
4466                 options->output_format |= DIFF_FORMAT_NAME;
4467         else if (!strcmp(arg, "--name-status"))
4468                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4469         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4470                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4471         else if (starts_with(arg, "--stat"))
4472                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4473                 return stat_opt(options, av);
4474
4475         /* renames options */
4476         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
4477                  !strcmp(arg, "--break-rewrites")) {
4478                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4479                         return error("invalid argument to -B: %s", arg+2);
4480         }
4481         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
4482                  !strcmp(arg, "--find-renames")) {
4483                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4484                         return error("invalid argument to -M: %s", arg+2);
4485                 options->detect_rename = DIFF_DETECT_RENAME;
4486         }
4487         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4488                 options->irreversible_delete = 1;
4489         }
4490         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
4491                  !strcmp(arg, "--find-copies")) {
4492                 if (options->detect_rename == DIFF_DETECT_COPY)
4493                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4494                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4495                         return error("invalid argument to -C: %s", arg+2);
4496                 options->detect_rename = DIFF_DETECT_COPY;
4497         }
4498         else if (!strcmp(arg, "--no-renames"))
4499                 options->detect_rename = 0;
4500         else if (!strcmp(arg, "--rename-empty"))
4501                 DIFF_OPT_SET(options, RENAME_EMPTY);
4502         else if (!strcmp(arg, "--no-rename-empty"))
4503                 DIFF_OPT_CLR(options, RENAME_EMPTY);
4504         else if (!strcmp(arg, "--relative"))
4505                 DIFF_OPT_SET(options, RELATIVE_NAME);
4506         else if (skip_prefix(arg, "--relative=", &arg)) {
4507                 DIFF_OPT_SET(options, RELATIVE_NAME);
4508                 options->prefix = arg;
4509         }
4510
4511         /* xdiff options */
4512         else if (!strcmp(arg, "--minimal"))
4513                 DIFF_XDL_SET(options, NEED_MINIMAL);
4514         else if (!strcmp(arg, "--no-minimal"))
4515                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4516         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4517                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4518         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4519                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4520         else if (!strcmp(arg, "--ignore-space-at-eol"))
4521                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4522         else if (!strcmp(arg, "--ignore-blank-lines"))
4523                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4524         else if (!strcmp(arg, "--indent-heuristic"))
4525                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4526         else if (!strcmp(arg, "--no-indent-heuristic"))
4527                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4528         else if (!strcmp(arg, "--patience"))
4529                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4530         else if (!strcmp(arg, "--histogram"))
4531                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4532         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4533                 long value = parse_algorithm_value(optarg);
4534                 if (value < 0)
4535                         return error("option diff-algorithm accepts \"myers\", "
4536                                      "\"minimal\", \"patience\" and \"histogram\"");
4537                 /* clear out previous settings */
4538                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4539                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4540                 options->xdl_opts |= value;
4541                 return argcount;
4542         }
4543
4544         /* flags options */
4545         else if (!strcmp(arg, "--binary")) {
4546                 enable_patch_output(&options->output_format);
4547                 DIFF_OPT_SET(options, BINARY);
4548         }
4549         else if (!strcmp(arg, "--full-index"))
4550                 DIFF_OPT_SET(options, FULL_INDEX);
4551         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4552                 DIFF_OPT_SET(options, TEXT);
4553         else if (!strcmp(arg, "-R"))
4554                 DIFF_OPT_SET(options, REVERSE_DIFF);
4555         else if (!strcmp(arg, "--find-copies-harder"))
4556                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4557         else if (!strcmp(arg, "--follow"))
4558                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
4559         else if (!strcmp(arg, "--no-follow")) {
4560                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
4561                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
4562         } else if (!strcmp(arg, "--color"))
4563                 options->use_color = 1;
4564         else if (skip_prefix(arg, "--color=", &arg)) {
4565                 int value = git_config_colorbool(NULL, arg);
4566                 if (value < 0)
4567                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4568                 options->use_color = value;
4569         }
4570         else if (!strcmp(arg, "--no-color"))
4571                 options->use_color = 0;
4572         else if (!strcmp(arg, "--color-moved")) {
4573                 if (diff_color_moved_default)
4574                         options->color_moved = diff_color_moved_default;
4575                 if (options->color_moved == COLOR_MOVED_NO)
4576                         options->color_moved = COLOR_MOVED_DEFAULT;
4577         } else if (!strcmp(arg, "--no-color-moved"))
4578                 options->color_moved = COLOR_MOVED_NO;
4579         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4580                 int cm = parse_color_moved(arg);
4581                 if (cm < 0)
4582                         die("bad --color-moved argument: %s", arg);
4583                 options->color_moved = cm;
4584         } else if (!strcmp(arg, "--color-words")) {
4585                 options->use_color = 1;
4586                 options->word_diff = DIFF_WORDS_COLOR;
4587         }
4588         else if (skip_prefix(arg, "--color-words=", &arg)) {
4589                 options->use_color = 1;
4590                 options->word_diff = DIFF_WORDS_COLOR;
4591                 options->word_regex = arg;
4592         }
4593         else if (!strcmp(arg, "--word-diff")) {
4594                 if (options->word_diff == DIFF_WORDS_NONE)
4595                         options->word_diff = DIFF_WORDS_PLAIN;
4596         }
4597         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4598                 if (!strcmp(arg, "plain"))
4599                         options->word_diff = DIFF_WORDS_PLAIN;
4600                 else if (!strcmp(arg, "color")) {
4601                         options->use_color = 1;
4602                         options->word_diff = DIFF_WORDS_COLOR;
4603                 }
4604                 else if (!strcmp(arg, "porcelain"))
4605                         options->word_diff = DIFF_WORDS_PORCELAIN;
4606                 else if (!strcmp(arg, "none"))
4607                         options->word_diff = DIFF_WORDS_NONE;
4608                 else
4609                         die("bad --word-diff argument: %s", arg);
4610         }
4611         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4612                 if (options->word_diff == DIFF_WORDS_NONE)
4613                         options->word_diff = DIFF_WORDS_PLAIN;
4614                 options->word_regex = optarg;
4615                 return argcount;
4616         }
4617         else if (!strcmp(arg, "--exit-code"))
4618                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4619         else if (!strcmp(arg, "--quiet"))
4620                 DIFF_OPT_SET(options, QUICK);
4621         else if (!strcmp(arg, "--ext-diff"))
4622                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
4623         else if (!strcmp(arg, "--no-ext-diff"))
4624                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
4625         else if (!strcmp(arg, "--textconv"))
4626                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
4627         else if (!strcmp(arg, "--no-textconv"))
4628                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
4629         else if (!strcmp(arg, "--ignore-submodules")) {
4630                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4631                 handle_ignore_submodules_arg(options, "all");
4632         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
4633                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4634                 handle_ignore_submodules_arg(options, arg);
4635         } else if (!strcmp(arg, "--submodule"))
4636                 options->submodule_format = DIFF_SUBMODULE_LOG;
4637         else if (skip_prefix(arg, "--submodule=", &arg))
4638                 return parse_submodule_opt(options, arg);
4639         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4640                 return parse_ws_error_highlight_opt(options, arg);
4641         else if (!strcmp(arg, "--ita-invisible-in-index"))
4642                 options->ita_invisible_in_index = 1;
4643         else if (!strcmp(arg, "--ita-visible-in-index"))
4644                 options->ita_invisible_in_index = 0;
4645
4646         /* misc options */
4647         else if (!strcmp(arg, "-z"))
4648                 options->line_termination = 0;
4649         else if ((argcount = short_opt('l', av, &optarg))) {
4650                 options->rename_limit = strtoul(optarg, NULL, 10);
4651                 return argcount;
4652         }
4653         else if ((argcount = short_opt('S', av, &optarg))) {
4654                 options->pickaxe = optarg;
4655                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4656                 return argcount;
4657         } else if ((argcount = short_opt('G', av, &optarg))) {
4658                 options->pickaxe = optarg;
4659                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4660                 return argcount;
4661         }
4662         else if (!strcmp(arg, "--pickaxe-all"))
4663                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4664         else if (!strcmp(arg, "--pickaxe-regex"))
4665                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4666         else if ((argcount = short_opt('O', av, &optarg))) {
4667                 options->orderfile = prefix_filename(prefix, optarg);
4668                 return argcount;
4669         }
4670         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4671                 int offending = parse_diff_filter_opt(optarg, options);
4672                 if (offending)
4673                         die("unknown change class '%c' in --diff-filter=%s",
4674                             offending, optarg);
4675                 return argcount;
4676         }
4677         else if (!strcmp(arg, "--no-abbrev"))
4678                 options->abbrev = 0;
4679         else if (!strcmp(arg, "--abbrev"))
4680                 options->abbrev = DEFAULT_ABBREV;
4681         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4682                 options->abbrev = strtoul(arg, NULL, 10);
4683                 if (options->abbrev < MINIMUM_ABBREV)
4684                         options->abbrev = MINIMUM_ABBREV;
4685                 else if (40 < options->abbrev)
4686                         options->abbrev = 40;
4687         }
4688         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4689                 options->a_prefix = optarg;
4690                 return argcount;
4691         }
4692         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4693                 options->line_prefix = optarg;
4694                 options->line_prefix_length = strlen(options->line_prefix);
4695                 graph_setup_line_prefix(options);
4696                 return argcount;
4697         }
4698         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4699                 options->b_prefix = optarg;
4700                 return argcount;
4701         }
4702         else if (!strcmp(arg, "--no-prefix"))
4703                 options->a_prefix = options->b_prefix = "";
4704         else if (opt_arg(arg, '\0', "inter-hunk-context",
4705                          &options->interhunkcontext))
4706                 ;
4707         else if (!strcmp(arg, "-W"))
4708                 DIFF_OPT_SET(options, FUNCCONTEXT);
4709         else if (!strcmp(arg, "--function-context"))
4710                 DIFF_OPT_SET(options, FUNCCONTEXT);
4711         else if (!strcmp(arg, "--no-function-context"))
4712                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4713         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4714                 char *path = prefix_filename(prefix, optarg);
4715                 options->file = xfopen(path, "w");
4716                 options->close_file = 1;
4717                 if (options->use_color != GIT_COLOR_ALWAYS)
4718                         options->use_color = GIT_COLOR_NEVER;
4719                 free(path);
4720                 return argcount;
4721         } else
4722                 return 0;
4723         return 1;
4724 }
4725
4726 int parse_rename_score(const char **cp_p)
4727 {
4728         unsigned long num, scale;
4729         int ch, dot;
4730         const char *cp = *cp_p;
4731
4732         num = 0;
4733         scale = 1;
4734         dot = 0;
4735         for (;;) {
4736                 ch = *cp;
4737                 if ( !dot && ch == '.' ) {
4738                         scale = 1;
4739                         dot = 1;
4740                 } else if ( ch == '%' ) {
4741                         scale = dot ? scale*100 : 100;
4742                         cp++;   /* % is always at the end */
4743                         break;
4744                 } else if ( ch >= '0' && ch <= '9' ) {
4745                         if ( scale < 100000 ) {
4746                                 scale *= 10;
4747                                 num = (num*10) + (ch-'0');
4748                         }
4749                 } else {
4750                         break;
4751                 }
4752                 cp++;
4753         }
4754         *cp_p = cp;
4755
4756         /* user says num divided by scale and we say internally that
4757          * is MAX_SCORE * num / scale.
4758          */
4759         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4760 }
4761
4762 static int diff_scoreopt_parse(const char *opt)
4763 {
4764         int opt1, opt2, cmd;
4765
4766         if (*opt++ != '-')
4767                 return -1;
4768         cmd = *opt++;
4769         if (cmd == '-') {
4770                 /* convert the long-form arguments into short-form versions */
4771                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4772                         if (*opt == 0 || *opt++ == '=')
4773                                 cmd = 'B';
4774                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4775                         if (*opt == 0 || *opt++ == '=')
4776                                 cmd = 'C';
4777                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4778                         if (*opt == 0 || *opt++ == '=')
4779                                 cmd = 'M';
4780                 }
4781         }
4782         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4783                 return -1; /* that is not a -M, -C, or -B option */
4784
4785         opt1 = parse_rename_score(&opt);
4786         if (cmd != 'B')
4787                 opt2 = 0;
4788         else {
4789                 if (*opt == 0)
4790                         opt2 = 0;
4791                 else if (*opt != '/')
4792                         return -1; /* we expect -B80/99 or -B80 */
4793                 else {
4794                         opt++;
4795                         opt2 = parse_rename_score(&opt);
4796                 }
4797         }
4798         if (*opt != 0)
4799                 return -1;
4800         return opt1 | (opt2 << 16);
4801 }
4802
4803 struct diff_queue_struct diff_queued_diff;
4804
4805 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4806 {
4807         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4808         queue->queue[queue->nr++] = dp;
4809 }
4810
4811 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4812                                  struct diff_filespec *one,
4813                                  struct diff_filespec *two)
4814 {
4815         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4816         dp->one = one;
4817         dp->two = two;
4818         if (queue)
4819                 diff_q(queue, dp);
4820         return dp;
4821 }
4822
4823 void diff_free_filepair(struct diff_filepair *p)
4824 {
4825         free_filespec(p->one);
4826         free_filespec(p->two);
4827         free(p);
4828 }
4829
4830 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4831 {
4832         int abblen;
4833         const char *abbrev;
4834
4835         if (len == GIT_SHA1_HEXSZ)
4836                 return oid_to_hex(oid);
4837
4838         abbrev = diff_abbrev_oid(oid, len);
4839         abblen = strlen(abbrev);
4840
4841         /*
4842          * In well-behaved cases, where the abbbreviated result is the
4843          * same as the requested length, append three dots after the
4844          * abbreviation (hence the whole logic is limited to the case
4845          * where abblen < 37); when the actual abbreviated result is a
4846          * bit longer than the requested length, we reduce the number
4847          * of dots so that they match the well-behaved ones.  However,
4848          * if the actual abbreviation is longer than the requested
4849          * length by more than three, we give up on aligning, and add
4850          * three dots anyway, to indicate that the output is not the
4851          * full object name.  Yes, this may be suboptimal, but this
4852          * appears only in "diff --raw --abbrev" output and it is not
4853          * worth the effort to change it now.  Note that this would
4854          * likely to work fine when the automatic sizing of default
4855          * abbreviation length is used--we would be fed -1 in "len" in
4856          * that case, and will end up always appending three-dots, but
4857          * the automatic sizing is supposed to give abblen that ensures
4858          * uniqueness across all objects (statistically speaking).
4859          */
4860         if (abblen < GIT_SHA1_HEXSZ - 3) {
4861                 static char hex[GIT_MAX_HEXSZ + 1];
4862                 if (len < abblen && abblen <= len + 2)
4863                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4864                 else
4865                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4866                 return hex;
4867         }
4868
4869         return oid_to_hex(oid);
4870 }
4871
4872 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4873 {
4874         int line_termination = opt->line_termination;
4875         int inter_name_termination = line_termination ? '\t' : '\0';
4876
4877         fprintf(opt->file, "%s", diff_line_prefix(opt));
4878         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4879                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4880                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4881                 fprintf(opt->file, "%s ",
4882                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4883         }
4884         if (p->score) {
4885                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4886                         inter_name_termination);
4887         } else {
4888                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4889         }
4890
4891         if (p->status == DIFF_STATUS_COPIED ||
4892             p->status == DIFF_STATUS_RENAMED) {
4893                 const char *name_a, *name_b;
4894                 name_a = p->one->path;
4895                 name_b = p->two->path;
4896                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4897                 write_name_quoted(name_a, opt->file, inter_name_termination);
4898                 write_name_quoted(name_b, opt->file, line_termination);
4899         } else {
4900                 const char *name_a, *name_b;
4901                 name_a = p->one->mode ? p->one->path : p->two->path;
4902                 name_b = NULL;
4903                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4904                 write_name_quoted(name_a, opt->file, line_termination);
4905         }
4906 }
4907
4908 int diff_unmodified_pair(struct diff_filepair *p)
4909 {
4910         /* This function is written stricter than necessary to support
4911          * the currently implemented transformers, but the idea is to
4912          * let transformers to produce diff_filepairs any way they want,
4913          * and filter and clean them up here before producing the output.
4914          */
4915         struct diff_filespec *one = p->one, *two = p->two;
4916
4917         if (DIFF_PAIR_UNMERGED(p))
4918                 return 0; /* unmerged is interesting */
4919
4920         /* deletion, addition, mode or type change
4921          * and rename are all interesting.
4922          */
4923         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4924             DIFF_PAIR_MODE_CHANGED(p) ||
4925             strcmp(one->path, two->path))
4926                 return 0;
4927
4928         /* both are valid and point at the same path.  that is, we are
4929          * dealing with a change.
4930          */
4931         if (one->oid_valid && two->oid_valid &&
4932             !oidcmp(&one->oid, &two->oid) &&
4933             !one->dirty_submodule && !two->dirty_submodule)
4934                 return 1; /* no change */
4935         if (!one->oid_valid && !two->oid_valid)
4936                 return 1; /* both look at the same file on the filesystem. */
4937         return 0;
4938 }
4939
4940 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4941 {
4942         if (diff_unmodified_pair(p))
4943                 return;
4944
4945         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4946             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4947                 return; /* no tree diffs in patch format */
4948
4949         run_diff(p, o);
4950 }
4951
4952 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4953                             struct diffstat_t *diffstat)
4954 {
4955         if (diff_unmodified_pair(p))
4956                 return;
4957
4958         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4959             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4960                 return; /* no useful stat for tree diffs */
4961
4962         run_diffstat(p, o, diffstat);
4963 }
4964
4965 static void diff_flush_checkdiff(struct diff_filepair *p,
4966                 struct diff_options *o)
4967 {
4968         if (diff_unmodified_pair(p))
4969                 return;
4970
4971         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4972             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4973                 return; /* nothing to check in tree diffs */
4974
4975         run_checkdiff(p, o);
4976 }
4977
4978 int diff_queue_is_empty(void)
4979 {
4980         struct diff_queue_struct *q = &diff_queued_diff;
4981         int i;
4982         for (i = 0; i < q->nr; i++)
4983                 if (!diff_unmodified_pair(q->queue[i]))
4984                         return 0;
4985         return 1;
4986 }
4987
4988 #if DIFF_DEBUG
4989 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4990 {
4991         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4992                 x, one ? one : "",
4993                 s->path,
4994                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4995                 s->mode,
4996                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4997         fprintf(stderr, "queue[%d] %s size %lu\n",
4998                 x, one ? one : "",
4999                 s->size);
5000 }
5001
5002 void diff_debug_filepair(const struct diff_filepair *p, int i)
5003 {
5004         diff_debug_filespec(p->one, i, "one");
5005         diff_debug_filespec(p->two, i, "two");
5006         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5007                 p->score, p->status ? p->status : '?',
5008                 p->one->rename_used, p->broken_pair);
5009 }
5010
5011 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5012 {
5013         int i;
5014         if (msg)
5015                 fprintf(stderr, "%s\n", msg);
5016         fprintf(stderr, "q->nr = %d\n", q->nr);
5017         for (i = 0; i < q->nr; i++) {
5018                 struct diff_filepair *p = q->queue[i];
5019                 diff_debug_filepair(p, i);
5020         }
5021 }
5022 #endif
5023
5024 static void diff_resolve_rename_copy(void)
5025 {
5026         int i;
5027         struct diff_filepair *p;
5028         struct diff_queue_struct *q = &diff_queued_diff;
5029
5030         diff_debug_queue("resolve-rename-copy", q);
5031
5032         for (i = 0; i < q->nr; i++) {
5033                 p = q->queue[i];
5034                 p->status = 0; /* undecided */
5035                 if (DIFF_PAIR_UNMERGED(p))
5036                         p->status = DIFF_STATUS_UNMERGED;
5037                 else if (!DIFF_FILE_VALID(p->one))
5038                         p->status = DIFF_STATUS_ADDED;
5039                 else if (!DIFF_FILE_VALID(p->two))
5040                         p->status = DIFF_STATUS_DELETED;
5041                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5042                         p->status = DIFF_STATUS_TYPE_CHANGED;
5043
5044                 /* from this point on, we are dealing with a pair
5045                  * whose both sides are valid and of the same type, i.e.
5046                  * either in-place edit or rename/copy edit.
5047                  */
5048                 else if (DIFF_PAIR_RENAME(p)) {
5049                         /*
5050                          * A rename might have re-connected a broken
5051                          * pair up, causing the pathnames to be the
5052                          * same again. If so, that's not a rename at
5053                          * all, just a modification..
5054                          *
5055                          * Otherwise, see if this source was used for
5056                          * multiple renames, in which case we decrement
5057                          * the count, and call it a copy.
5058                          */
5059                         if (!strcmp(p->one->path, p->two->path))
5060                                 p->status = DIFF_STATUS_MODIFIED;
5061                         else if (--p->one->rename_used > 0)
5062                                 p->status = DIFF_STATUS_COPIED;
5063                         else
5064                                 p->status = DIFF_STATUS_RENAMED;
5065                 }
5066                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5067                          p->one->mode != p->two->mode ||
5068                          p->one->dirty_submodule ||
5069                          p->two->dirty_submodule ||
5070                          is_null_oid(&p->one->oid))
5071                         p->status = DIFF_STATUS_MODIFIED;
5072                 else {
5073                         /* This is a "no-change" entry and should not
5074                          * happen anymore, but prepare for broken callers.
5075                          */
5076                         error("feeding unmodified %s to diffcore",
5077                               p->one->path);
5078                         p->status = DIFF_STATUS_UNKNOWN;
5079                 }
5080         }
5081         diff_debug_queue("resolve-rename-copy done", q);
5082 }
5083
5084 static int check_pair_status(struct diff_filepair *p)
5085 {
5086         switch (p->status) {
5087         case DIFF_STATUS_UNKNOWN:
5088                 return 0;
5089         case 0:
5090                 die("internal error in diff-resolve-rename-copy");
5091         default:
5092                 return 1;
5093         }
5094 }
5095
5096 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5097 {
5098         int fmt = opt->output_format;
5099
5100         if (fmt & DIFF_FORMAT_CHECKDIFF)
5101                 diff_flush_checkdiff(p, opt);
5102         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5103                 diff_flush_raw(p, opt);
5104         else if (fmt & DIFF_FORMAT_NAME) {
5105                 const char *name_a, *name_b;
5106                 name_a = p->two->path;
5107                 name_b = NULL;
5108                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5109                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5110                 write_name_quoted(name_a, opt->file, opt->line_termination);
5111         }
5112 }
5113
5114 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5115 {
5116         struct strbuf sb = STRBUF_INIT;
5117         if (fs->mode)
5118                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5119         else
5120                 strbuf_addf(&sb, " %s ", newdelete);
5121
5122         quote_c_style(fs->path, &sb, NULL, 0);
5123         strbuf_addch(&sb, '\n');
5124         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5125                          sb.buf, sb.len, 0);
5126         strbuf_release(&sb);
5127 }
5128
5129 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5130                 int show_name)
5131 {
5132         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5133                 struct strbuf sb = STRBUF_INIT;
5134                 strbuf_addf(&sb, " mode change %06o => %06o",
5135                             p->one->mode, p->two->mode);
5136                 if (show_name) {
5137                         strbuf_addch(&sb, ' ');
5138                         quote_c_style(p->two->path, &sb, NULL, 0);
5139                 }
5140                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5141                                  sb.buf, sb.len, 0);
5142                 strbuf_release(&sb);
5143         }
5144 }
5145
5146 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5147                 struct diff_filepair *p)
5148 {
5149         struct strbuf sb = STRBUF_INIT;
5150         char *names = pprint_rename(p->one->path, p->two->path);
5151         strbuf_addf(&sb, " %s %s (%d%%)\n",
5152                         renamecopy, names, similarity_index(p));
5153         free(names);
5154         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5155                                  sb.buf, sb.len, 0);
5156         show_mode_change(opt, p, 0);
5157 }
5158
5159 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5160 {
5161         switch(p->status) {
5162         case DIFF_STATUS_DELETED:
5163                 show_file_mode_name(opt, "delete", p->one);
5164                 break;
5165         case DIFF_STATUS_ADDED:
5166                 show_file_mode_name(opt, "create", p->two);
5167                 break;
5168         case DIFF_STATUS_COPIED:
5169                 show_rename_copy(opt, "copy", p);
5170                 break;
5171         case DIFF_STATUS_RENAMED:
5172                 show_rename_copy(opt, "rename", p);
5173                 break;
5174         default:
5175                 if (p->score) {
5176                         struct strbuf sb = STRBUF_INIT;
5177                         strbuf_addstr(&sb, " rewrite ");
5178                         quote_c_style(p->two->path, &sb, NULL, 0);
5179                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5180                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5181                                          sb.buf, sb.len, 0);
5182                 }
5183                 show_mode_change(opt, p, !p->score);
5184                 break;
5185         }
5186 }
5187
5188 struct patch_id_t {
5189         git_SHA_CTX *ctx;
5190         int patchlen;
5191 };
5192
5193 static int remove_space(char *line, int len)
5194 {
5195         int i;
5196         char *dst = line;
5197         unsigned char c;
5198
5199         for (i = 0; i < len; i++)
5200                 if (!isspace((c = line[i])))
5201                         *dst++ = c;
5202
5203         return dst - line;
5204 }
5205
5206 static void patch_id_consume(void *priv, char *line, unsigned long len)
5207 {
5208         struct patch_id_t *data = priv;
5209         int new_len;
5210
5211         /* Ignore line numbers when computing the SHA1 of the patch */
5212         if (starts_with(line, "@@ -"))
5213                 return;
5214
5215         new_len = remove_space(line, len);
5216
5217         git_SHA1_Update(data->ctx, line, new_len);
5218         data->patchlen += new_len;
5219 }
5220
5221 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5222 {
5223         git_SHA1_Update(ctx, str, strlen(str));
5224 }
5225
5226 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5227 {
5228         /* large enough for 2^32 in octal */
5229         char buf[12];
5230         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5231         git_SHA1_Update(ctx, buf, len);
5232 }
5233
5234 /* returns 0 upon success, and writes result into sha1 */
5235 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5236 {
5237         struct diff_queue_struct *q = &diff_queued_diff;
5238         int i;
5239         git_SHA_CTX ctx;
5240         struct patch_id_t data;
5241
5242         git_SHA1_Init(&ctx);
5243         memset(&data, 0, sizeof(struct patch_id_t));
5244         data.ctx = &ctx;
5245
5246         for (i = 0; i < q->nr; i++) {
5247                 xpparam_t xpp;
5248                 xdemitconf_t xecfg;
5249                 mmfile_t mf1, mf2;
5250                 struct diff_filepair *p = q->queue[i];
5251                 int len1, len2;
5252
5253                 memset(&xpp, 0, sizeof(xpp));
5254                 memset(&xecfg, 0, sizeof(xecfg));
5255                 if (p->status == 0)
5256                         return error("internal diff status error");
5257                 if (p->status == DIFF_STATUS_UNKNOWN)
5258                         continue;
5259                 if (diff_unmodified_pair(p))
5260                         continue;
5261                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5262                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5263                         continue;
5264                 if (DIFF_PAIR_UNMERGED(p))
5265                         continue;
5266
5267                 diff_fill_oid_info(p->one);
5268                 diff_fill_oid_info(p->two);
5269
5270                 len1 = remove_space(p->one->path, strlen(p->one->path));
5271                 len2 = remove_space(p->two->path, strlen(p->two->path));
5272                 patch_id_add_string(&ctx, "diff--git");
5273                 patch_id_add_string(&ctx, "a/");
5274                 git_SHA1_Update(&ctx, p->one->path, len1);
5275                 patch_id_add_string(&ctx, "b/");
5276                 git_SHA1_Update(&ctx, p->two->path, len2);
5277
5278                 if (p->one->mode == 0) {
5279                         patch_id_add_string(&ctx, "newfilemode");
5280                         patch_id_add_mode(&ctx, p->two->mode);
5281                         patch_id_add_string(&ctx, "---/dev/null");
5282                         patch_id_add_string(&ctx, "+++b/");
5283                         git_SHA1_Update(&ctx, p->two->path, len2);
5284                 } else if (p->two->mode == 0) {
5285                         patch_id_add_string(&ctx, "deletedfilemode");
5286                         patch_id_add_mode(&ctx, p->one->mode);
5287                         patch_id_add_string(&ctx, "---a/");
5288                         git_SHA1_Update(&ctx, p->one->path, len1);
5289                         patch_id_add_string(&ctx, "+++/dev/null");
5290                 } else {
5291                         patch_id_add_string(&ctx, "---a/");
5292                         git_SHA1_Update(&ctx, p->one->path, len1);
5293                         patch_id_add_string(&ctx, "+++b/");
5294                         git_SHA1_Update(&ctx, p->two->path, len2);
5295                 }
5296
5297                 if (diff_header_only)
5298                         continue;
5299
5300                 if (fill_mmfile(&mf1, p->one) < 0 ||
5301                     fill_mmfile(&mf2, p->two) < 0)
5302                         return error("unable to read files to diff");
5303
5304                 if (diff_filespec_is_binary(p->one) ||
5305                     diff_filespec_is_binary(p->two)) {
5306                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5307                                         GIT_SHA1_HEXSZ);
5308                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5309                                         GIT_SHA1_HEXSZ);
5310                         continue;
5311                 }
5312
5313                 xpp.flags = 0;
5314                 xecfg.ctxlen = 3;
5315                 xecfg.flags = 0;
5316                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5317                                   &xpp, &xecfg))
5318                         return error("unable to generate patch-id diff for %s",
5319                                      p->one->path);
5320         }
5321
5322         git_SHA1_Final(oid->hash, &ctx);
5323         return 0;
5324 }
5325
5326 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5327 {
5328         struct diff_queue_struct *q = &diff_queued_diff;
5329         int i;
5330         int result = diff_get_patch_id(options, oid, diff_header_only);
5331
5332         for (i = 0; i < q->nr; i++)
5333                 diff_free_filepair(q->queue[i]);
5334
5335         free(q->queue);
5336         DIFF_QUEUE_CLEAR(q);
5337
5338         return result;
5339 }
5340
5341 static int is_summary_empty(const struct diff_queue_struct *q)
5342 {
5343         int i;
5344
5345         for (i = 0; i < q->nr; i++) {
5346                 const struct diff_filepair *p = q->queue[i];
5347
5348                 switch (p->status) {
5349                 case DIFF_STATUS_DELETED:
5350                 case DIFF_STATUS_ADDED:
5351                 case DIFF_STATUS_COPIED:
5352                 case DIFF_STATUS_RENAMED:
5353                         return 0;
5354                 default:
5355                         if (p->score)
5356                                 return 0;
5357                         if (p->one->mode && p->two->mode &&
5358                             p->one->mode != p->two->mode)
5359                                 return 0;
5360                         break;
5361                 }
5362         }
5363         return 1;
5364 }
5365
5366 static const char rename_limit_warning[] =
5367 N_("inexact rename detection was skipped due to too many files.");
5368
5369 static const char degrade_cc_to_c_warning[] =
5370 N_("only found copies from modified paths due to too many files.");
5371
5372 static const char rename_limit_advice[] =
5373 N_("you may want to set your %s variable to at least "
5374    "%d and retry the command.");
5375
5376 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5377 {
5378         if (degraded_cc)
5379                 warning(_(degrade_cc_to_c_warning));
5380         else if (needed)
5381                 warning(_(rename_limit_warning));
5382         else
5383                 return;
5384         if (0 < needed && needed < 32767)
5385                 warning(_(rename_limit_advice), varname, needed);
5386 }
5387
5388 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5389 {
5390         int i;
5391         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5392         struct diff_queue_struct *q = &diff_queued_diff;
5393
5394         if (WSEH_NEW & WS_RULE_MASK)
5395                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5396
5397         if (o->color_moved)
5398                 o->emitted_symbols = &esm;
5399
5400         for (i = 0; i < q->nr; i++) {
5401                 struct diff_filepair *p = q->queue[i];
5402                 if (check_pair_status(p))
5403                         diff_flush_patch(p, o);
5404         }
5405
5406         if (o->emitted_symbols) {
5407                 if (o->color_moved) {
5408                         struct hashmap add_lines, del_lines;
5409
5410                         hashmap_init(&del_lines,
5411                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5412                         hashmap_init(&add_lines,
5413                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5414
5415                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5416                         mark_color_as_moved(o, &add_lines, &del_lines);
5417
5418                         hashmap_free(&add_lines, 0);
5419                         hashmap_free(&del_lines, 0);
5420                 }
5421
5422                 for (i = 0; i < esm.nr; i++)
5423                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5424
5425                 for (i = 0; i < esm.nr; i++)
5426                         free((void *)esm.buf[i].line);
5427         }
5428         esm.nr = 0;
5429 }
5430
5431 void diff_flush(struct diff_options *options)
5432 {
5433         struct diff_queue_struct *q = &diff_queued_diff;
5434         int i, output_format = options->output_format;
5435         int separator = 0;
5436         int dirstat_by_line = 0;
5437
5438         /*
5439          * Order: raw, stat, summary, patch
5440          * or:    name/name-status/checkdiff (other bits clear)
5441          */
5442         if (!q->nr)
5443                 goto free_queue;
5444
5445         if (output_format & (DIFF_FORMAT_RAW |
5446                              DIFF_FORMAT_NAME |
5447                              DIFF_FORMAT_NAME_STATUS |
5448                              DIFF_FORMAT_CHECKDIFF)) {
5449                 for (i = 0; i < q->nr; i++) {
5450                         struct diff_filepair *p = q->queue[i];
5451                         if (check_pair_status(p))
5452                                 flush_one_pair(p, options);
5453                 }
5454                 separator++;
5455         }
5456
5457         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
5458                 dirstat_by_line = 1;
5459
5460         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5461             dirstat_by_line) {
5462                 struct diffstat_t diffstat;
5463
5464                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5465                 for (i = 0; i < q->nr; i++) {
5466                         struct diff_filepair *p = q->queue[i];
5467                         if (check_pair_status(p))
5468                                 diff_flush_stat(p, options, &diffstat);
5469                 }
5470                 if (output_format & DIFF_FORMAT_NUMSTAT)
5471                         show_numstat(&diffstat, options);
5472                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5473                         show_stats(&diffstat, options);
5474                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5475                         show_shortstats(&diffstat, options);
5476                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5477                         show_dirstat_by_line(&diffstat, options);
5478                 free_diffstat_info(&diffstat);
5479                 separator++;
5480         }
5481         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5482                 show_dirstat(options);
5483
5484         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5485                 for (i = 0; i < q->nr; i++) {
5486                         diff_summary(options, q->queue[i]);
5487                 }
5488                 separator++;
5489         }
5490
5491         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5492             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
5493             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5494                 /*
5495                  * run diff_flush_patch for the exit status. setting
5496                  * options->file to /dev/null should be safe, because we
5497                  * aren't supposed to produce any output anyway.
5498                  */
5499                 if (options->close_file)
5500                         fclose(options->file);
5501                 options->file = xfopen("/dev/null", "w");
5502                 options->close_file = 1;
5503                 options->color_moved = 0;
5504                 for (i = 0; i < q->nr; i++) {
5505                         struct diff_filepair *p = q->queue[i];
5506                         if (check_pair_status(p))
5507                                 diff_flush_patch(p, options);
5508                         if (options->found_changes)
5509                                 break;
5510                 }
5511         }
5512
5513         if (output_format & DIFF_FORMAT_PATCH) {
5514                 if (separator) {
5515                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5516                         if (options->stat_sep)
5517                                 /* attach patch instead of inline */
5518                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5519                                                  NULL, 0, 0);
5520                 }
5521
5522                 diff_flush_patch_all_file_pairs(options);
5523         }
5524
5525         if (output_format & DIFF_FORMAT_CALLBACK)
5526                 options->format_callback(q, options, options->format_callback_data);
5527
5528         for (i = 0; i < q->nr; i++)
5529                 diff_free_filepair(q->queue[i]);
5530 free_queue:
5531         free(q->queue);
5532         DIFF_QUEUE_CLEAR(q);
5533         if (options->close_file)
5534                 fclose(options->file);
5535
5536         /*
5537          * Report the content-level differences with HAS_CHANGES;
5538          * diff_addremove/diff_change does not set the bit when
5539          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5540          */
5541         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5542                 if (options->found_changes)
5543                         DIFF_OPT_SET(options, HAS_CHANGES);
5544                 else
5545                         DIFF_OPT_CLR(options, HAS_CHANGES);
5546         }
5547 }
5548
5549 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5550 {
5551         return (((p->status == DIFF_STATUS_MODIFIED) &&
5552                  ((p->score &&
5553                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5554                   (!p->score &&
5555                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5556                 ((p->status != DIFF_STATUS_MODIFIED) &&
5557                  filter_bit_tst(p->status, options)));
5558 }
5559
5560 static void diffcore_apply_filter(struct diff_options *options)
5561 {
5562         int i;
5563         struct diff_queue_struct *q = &diff_queued_diff;
5564         struct diff_queue_struct outq;
5565
5566         DIFF_QUEUE_CLEAR(&outq);
5567
5568         if (!options->filter)
5569                 return;
5570
5571         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5572                 int found;
5573                 for (i = found = 0; !found && i < q->nr; i++) {
5574                         if (match_filter(options, q->queue[i]))
5575                                 found++;
5576                 }
5577                 if (found)
5578                         return;
5579
5580                 /* otherwise we will clear the whole queue
5581                  * by copying the empty outq at the end of this
5582                  * function, but first clear the current entries
5583                  * in the queue.
5584                  */
5585                 for (i = 0; i < q->nr; i++)
5586                         diff_free_filepair(q->queue[i]);
5587         }
5588         else {
5589                 /* Only the matching ones */
5590                 for (i = 0; i < q->nr; i++) {
5591                         struct diff_filepair *p = q->queue[i];
5592                         if (match_filter(options, p))
5593                                 diff_q(&outq, p);
5594                         else
5595                                 diff_free_filepair(p);
5596                 }
5597         }
5598         free(q->queue);
5599         *q = outq;
5600 }
5601
5602 /* Check whether two filespecs with the same mode and size are identical */
5603 static int diff_filespec_is_identical(struct diff_filespec *one,
5604                                       struct diff_filespec *two)
5605 {
5606         if (S_ISGITLINK(one->mode))
5607                 return 0;
5608         if (diff_populate_filespec(one, 0))
5609                 return 0;
5610         if (diff_populate_filespec(two, 0))
5611                 return 0;
5612         return !memcmp(one->data, two->data, one->size);
5613 }
5614
5615 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5616 {
5617         if (p->done_skip_stat_unmatch)
5618                 return p->skip_stat_unmatch_result;
5619
5620         p->done_skip_stat_unmatch = 1;
5621         p->skip_stat_unmatch_result = 0;
5622         /*
5623          * 1. Entries that come from stat info dirtiness
5624          *    always have both sides (iow, not create/delete),
5625          *    one side of the object name is unknown, with
5626          *    the same mode and size.  Keep the ones that
5627          *    do not match these criteria.  They have real
5628          *    differences.
5629          *
5630          * 2. At this point, the file is known to be modified,
5631          *    with the same mode and size, and the object
5632          *    name of one side is unknown.  Need to inspect
5633          *    the identical contents.
5634          */
5635         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5636             !DIFF_FILE_VALID(p->two) ||
5637             (p->one->oid_valid && p->two->oid_valid) ||
5638             (p->one->mode != p->two->mode) ||
5639             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5640             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5641             (p->one->size != p->two->size) ||
5642             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5643                 p->skip_stat_unmatch_result = 1;
5644         return p->skip_stat_unmatch_result;
5645 }
5646
5647 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5648 {
5649         int i;
5650         struct diff_queue_struct *q = &diff_queued_diff;
5651         struct diff_queue_struct outq;
5652         DIFF_QUEUE_CLEAR(&outq);
5653
5654         for (i = 0; i < q->nr; i++) {
5655                 struct diff_filepair *p = q->queue[i];
5656
5657                 if (diff_filespec_check_stat_unmatch(p))
5658                         diff_q(&outq, p);
5659                 else {
5660                         /*
5661                          * The caller can subtract 1 from skip_stat_unmatch
5662                          * to determine how many paths were dirty only
5663                          * due to stat info mismatch.
5664                          */
5665                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
5666                                 diffopt->skip_stat_unmatch++;
5667                         diff_free_filepair(p);
5668                 }
5669         }
5670         free(q->queue);
5671         *q = outq;
5672 }
5673
5674 static int diffnamecmp(const void *a_, const void *b_)
5675 {
5676         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5677         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5678         const char *name_a, *name_b;
5679
5680         name_a = a->one ? a->one->path : a->two->path;
5681         name_b = b->one ? b->one->path : b->two->path;
5682         return strcmp(name_a, name_b);
5683 }
5684
5685 void diffcore_fix_diff_index(struct diff_options *options)
5686 {
5687         struct diff_queue_struct *q = &diff_queued_diff;
5688         QSORT(q->queue, q->nr, diffnamecmp);
5689 }
5690
5691 void diffcore_std(struct diff_options *options)
5692 {
5693         /* NOTE please keep the following in sync with diff_tree_combined() */
5694         if (options->skip_stat_unmatch)
5695                 diffcore_skip_stat_unmatch(options);
5696         if (!options->found_follow) {
5697                 /* See try_to_follow_renames() in tree-diff.c */
5698                 if (options->break_opt != -1)
5699                         diffcore_break(options->break_opt);
5700                 if (options->detect_rename)
5701                         diffcore_rename(options);
5702                 if (options->break_opt != -1)
5703                         diffcore_merge_broken();
5704         }
5705         if (options->pickaxe)
5706                 diffcore_pickaxe(options);
5707         if (options->orderfile)
5708                 diffcore_order(options->orderfile);
5709         if (!options->found_follow)
5710                 /* See try_to_follow_renames() in tree-diff.c */
5711                 diff_resolve_rename_copy();
5712         diffcore_apply_filter(options);
5713
5714         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5715                 DIFF_OPT_SET(options, HAS_CHANGES);
5716         else
5717                 DIFF_OPT_CLR(options, HAS_CHANGES);
5718
5719         options->found_follow = 0;
5720 }
5721
5722 int diff_result_code(struct diff_options *opt, int status)
5723 {
5724         int result = 0;
5725
5726         diff_warn_rename_limit("diff.renameLimit",
5727                                opt->needed_rename_limit,
5728                                opt->degraded_cc_to_c);
5729         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5730             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5731                 return status;
5732         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5733             DIFF_OPT_TST(opt, HAS_CHANGES))
5734                 result |= 01;
5735         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5736             DIFF_OPT_TST(opt, CHECK_FAILED))
5737                 result |= 02;
5738         return result;
5739 }
5740
5741 int diff_can_quit_early(struct diff_options *opt)
5742 {
5743         return (DIFF_OPT_TST(opt, QUICK) &&
5744                 !opt->filter &&
5745                 DIFF_OPT_TST(opt, HAS_CHANGES));
5746 }
5747
5748 /*
5749  * Shall changes to this submodule be ignored?
5750  *
5751  * Submodule changes can be configured to be ignored separately for each path,
5752  * but that configuration can be overridden from the command line.
5753  */
5754 static int is_submodule_ignored(const char *path, struct diff_options *options)
5755 {
5756         int ignored = 0;
5757         unsigned orig_flags = options->flags;
5758         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5759                 set_diffopt_flags_from_submodule_config(options, path);
5760         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5761                 ignored = 1;
5762         options->flags = orig_flags;
5763         return ignored;
5764 }
5765
5766 void diff_addremove(struct diff_options *options,
5767                     int addremove, unsigned mode,
5768                     const struct object_id *oid,
5769                     int oid_valid,
5770                     const char *concatpath, unsigned dirty_submodule)
5771 {
5772         struct diff_filespec *one, *two;
5773
5774         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5775                 return;
5776
5777         /* This may look odd, but it is a preparation for
5778          * feeding "there are unchanged files which should
5779          * not produce diffs, but when you are doing copy
5780          * detection you would need them, so here they are"
5781          * entries to the diff-core.  They will be prefixed
5782          * with something like '=' or '*' (I haven't decided
5783          * which but should not make any difference).
5784          * Feeding the same new and old to diff_change()
5785          * also has the same effect.
5786          * Before the final output happens, they are pruned after
5787          * merged into rename/copy pairs as appropriate.
5788          */
5789         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5790                 addremove = (addremove == '+' ? '-' :
5791                              addremove == '-' ? '+' : addremove);
5792
5793         if (options->prefix &&
5794             strncmp(concatpath, options->prefix, options->prefix_length))
5795                 return;
5796
5797         one = alloc_filespec(concatpath);
5798         two = alloc_filespec(concatpath);
5799
5800         if (addremove != '+')
5801                 fill_filespec(one, oid, oid_valid, mode);
5802         if (addremove != '-') {
5803                 fill_filespec(two, oid, oid_valid, mode);
5804                 two->dirty_submodule = dirty_submodule;
5805         }
5806
5807         diff_queue(&diff_queued_diff, one, two);
5808         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5809                 DIFF_OPT_SET(options, HAS_CHANGES);
5810 }
5811
5812 void diff_change(struct diff_options *options,
5813                  unsigned old_mode, unsigned new_mode,
5814                  const struct object_id *old_oid,
5815                  const struct object_id *new_oid,
5816                  int old_oid_valid, int new_oid_valid,
5817                  const char *concatpath,
5818                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5819 {
5820         struct diff_filespec *one, *two;
5821         struct diff_filepair *p;
5822
5823         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5824             is_submodule_ignored(concatpath, options))
5825                 return;
5826
5827         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5828                 SWAP(old_mode, new_mode);
5829                 SWAP(old_oid, new_oid);
5830                 SWAP(old_oid_valid, new_oid_valid);
5831                 SWAP(old_dirty_submodule, new_dirty_submodule);
5832         }
5833
5834         if (options->prefix &&
5835             strncmp(concatpath, options->prefix, options->prefix_length))
5836                 return;
5837
5838         one = alloc_filespec(concatpath);
5839         two = alloc_filespec(concatpath);
5840         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5841         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5842         one->dirty_submodule = old_dirty_submodule;
5843         two->dirty_submodule = new_dirty_submodule;
5844         p = diff_queue(&diff_queued_diff, one, two);
5845
5846         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5847                 return;
5848
5849         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5850             !diff_filespec_check_stat_unmatch(p))
5851                 return;
5852
5853         DIFF_OPT_SET(options, HAS_CHANGES);
5854 }
5855
5856 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5857 {
5858         struct diff_filepair *pair;
5859         struct diff_filespec *one, *two;
5860
5861         if (options->prefix &&
5862             strncmp(path, options->prefix, options->prefix_length))
5863                 return NULL;
5864
5865         one = alloc_filespec(path);
5866         two = alloc_filespec(path);
5867         pair = diff_queue(&diff_queued_diff, one, two);
5868         pair->is_unmerged = 1;
5869         return pair;
5870 }
5871
5872 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5873                 size_t *outsize)
5874 {
5875         struct diff_tempfile *temp;
5876         const char *argv[3];
5877         const char **arg = argv;
5878         struct child_process child = CHILD_PROCESS_INIT;
5879         struct strbuf buf = STRBUF_INIT;
5880         int err = 0;
5881
5882         temp = prepare_temp_file(spec->path, spec);
5883         *arg++ = pgm;
5884         *arg++ = temp->name;
5885         *arg = NULL;
5886
5887         child.use_shell = 1;
5888         child.argv = argv;
5889         child.out = -1;
5890         if (start_command(&child)) {
5891                 remove_tempfile();
5892                 return NULL;
5893         }
5894
5895         if (strbuf_read(&buf, child.out, 0) < 0)
5896                 err = error("error reading from textconv command '%s'", pgm);
5897         close(child.out);
5898
5899         if (finish_command(&child) || err) {
5900                 strbuf_release(&buf);
5901                 remove_tempfile();
5902                 return NULL;
5903         }
5904         remove_tempfile();
5905
5906         return strbuf_detach(&buf, outsize);
5907 }
5908
5909 size_t fill_textconv(struct userdiff_driver *driver,
5910                      struct diff_filespec *df,
5911                      char **outbuf)
5912 {
5913         size_t size;
5914
5915         if (!driver) {
5916                 if (!DIFF_FILE_VALID(df)) {
5917                         *outbuf = "";
5918                         return 0;
5919                 }
5920                 if (diff_populate_filespec(df, 0))
5921                         die("unable to read files to diff");
5922                 *outbuf = df->data;
5923                 return df->size;
5924         }
5925
5926         if (!driver->textconv)
5927                 die("BUG: fill_textconv called with non-textconv driver");
5928
5929         if (driver->textconv_cache && df->oid_valid) {
5930                 *outbuf = notes_cache_get(driver->textconv_cache,
5931                                           &df->oid,
5932                                           &size);
5933                 if (*outbuf)
5934                         return size;
5935         }
5936
5937         *outbuf = run_textconv(driver->textconv, df, &size);
5938         if (!*outbuf)
5939                 die("unable to read files to diff");
5940
5941         if (driver->textconv_cache && df->oid_valid) {
5942                 /* ignore errors, as we might be in a readonly repository */
5943                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
5944                                 size);
5945                 /*
5946                  * we could save up changes and flush them all at the end,
5947                  * but we would need an extra call after all diffing is done.
5948                  * Since generating a cache entry is the slow path anyway,
5949                  * this extra overhead probably isn't a big deal.
5950                  */
5951                 notes_cache_write(driver->textconv_cache);
5952         }
5953
5954         return size;
5955 }
5956
5957 int textconv_object(const char *path,
5958                     unsigned mode,
5959                     const struct object_id *oid,
5960                     int oid_valid,
5961                     char **buf,
5962                     unsigned long *buf_size)
5963 {
5964         struct diff_filespec *df;
5965         struct userdiff_driver *textconv;
5966
5967         df = alloc_filespec(path);
5968         fill_filespec(df, oid, oid_valid, mode);
5969         textconv = get_textconv(df);
5970         if (!textconv) {
5971                 free_filespec(df);
5972                 return 0;
5973         }
5974
5975         *buf_size = fill_textconv(textconv, df, buf);
5976         free_filespec(df);
5977         return 1;
5978 }
5979
5980 void setup_diff_pager(struct diff_options *opt)
5981 {
5982         /*
5983          * If the user asked for our exit code, then either they want --quiet
5984          * or --exit-code. We should definitely not bother with a pager in the
5985          * former case, as we will generate no output. Since we still properly
5986          * report our exit code even when a pager is run, we _could_ run a
5987          * pager with --exit-code. But since we have not done so historically,
5988          * and because it is easy to find people oneline advising "git diff
5989          * --exit-code" in hooks and other scripts, we do not do so.
5990          */
5991         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5992             check_pager_config("diff") != 0)
5993                 setup_pager();
5994 }