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, "plain"))
 260                 return COLOR_MOVED_PLAIN;
 261         else if (!strcmp(arg, "zebra"))
 262                 return COLOR_MOVED_ZEBRA;
 263         else if (!strcmp(arg, "default"))
 264                 return COLOR_MOVED_DEFAULT;
 265         else
 266                 return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'plain'"));
 267 }
 268
 269 int git_diff_ui_config(const char *var, const char *value, void *cb)
 270 {
 271         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 272                 diff_use_color_default = git_config_colorbool(var, value);
 273                 return 0;
 274         }
 275         if (!strcmp(var, "diff.colormoved")) {
 276                 int cm = parse_color_moved(value);
 277                 if (cm < 0)
 278                         return -1;
 279                 diff_color_moved_default = cm;
 280                 return 0;
 281         }
 282         if (!strcmp(var, "diff.context")) {
 283                 diff_context_default = git_config_int(var, value);
 284                 if (diff_context_default < 0)
 285                         return -1;
 286                 return 0;
 287         }
 288         if (!strcmp(var, "diff.interhunkcontext")) {
 289                 diff_interhunk_context_default = git_config_int(var, value);
 290                 if (diff_interhunk_context_default < 0)
 291                         return -1;
 292                 return 0;
 293         }
 294         if (!strcmp(var, "diff.renames")) {
 295                 diff_detect_rename_default = git_config_rename(var, value);
 296                 return 0;
 297         }
 298         if (!strcmp(var, "diff.autorefreshindex")) {
 299                 diff_auto_refresh_index = git_config_bool(var, value);
 300                 return 0;
 301         }
 302         if (!strcmp(var, "diff.mnemonicprefix")) {
 303                 diff_mnemonic_prefix = git_config_bool(var, value);
 304                 return 0;
 305         }
 306         if (!strcmp(var, "diff.noprefix")) {
 307                 diff_no_prefix = git_config_bool(var, value);
 308                 return 0;
 309         }
 310         if (!strcmp(var, "diff.statgraphwidth")) {
 311                 diff_stat_graph_width = git_config_int(var, value);
 312                 return 0;
 313         }
 314         if (!strcmp(var, "diff.external"))
 315                 return git_config_string(&external_diff_cmd_cfg, var, value);
 316         if (!strcmp(var, "diff.wordregex"))
 317                 return git_config_string(&diff_word_regex_cfg, var, value);
 318         if (!strcmp(var, "diff.orderfile"))
 319                 return git_config_pathname(&diff_order_file_cfg, var, value);
 320
 321         if (!strcmp(var, "diff.ignoresubmodules"))
 322                 handle_ignore_submodules_arg(&default_diff_options, value);
 323
 324         if (!strcmp(var, "diff.submodule")) {
 325                 if (parse_submodule_params(&default_diff_options, value))
 326                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 327                                 value);
 328                 return 0;
 329         }
 330
 331         if (!strcmp(var, "diff.algorithm")) {
 332                 diff_algorithm = parse_algorithm_value(value);
 333                 if (diff_algorithm < 0)
 334                         return -1;
 335                 return 0;
 336         }
 337
 338         if (!strcmp(var, "diff.wserrorhighlight")) {
 339                 int val = parse_ws_error_highlight(value);
 340                 if (val < 0)
 341                         return -1;
 342                 ws_error_highlight_default = val;
 343                 return 0;
 344         }
 345
 346         if (git_color_config(var, value, cb) < 0)
 347                 return -1;
 348
 349         return git_diff_basic_config(var, value, cb);
 350 }
 351
 352 int git_diff_basic_config(const char *var, const char *value, void *cb)
 353 {
 354         const char *name;
 355
 356         if (!strcmp(var, "diff.renamelimit")) {
 357                 diff_rename_limit_default = git_config_int(var, value);
 358                 return 0;
 359         }
 360
 361         if (userdiff_config(var, value) < 0)
 362                 return -1;
 363
 364         if (skip_prefix(var, "diff.color.", &name) ||
 365             skip_prefix(var, "color.diff.", &name)) {
 366                 int slot = parse_diff_color_slot(name);
 367                 if (slot < 0)
 368                         return 0;
 369                 if (!value)
 370                         return config_error_nonbool(var);
 371                 return color_parse(value, diff_colors[slot]);
 372         }
 373
 374         /* like GNU diff's --suppress-blank-empty option  */
 375         if (!strcmp(var, "diff.suppressblankempty") ||
 376                         /* for backwards compatibility */
 377                         !strcmp(var, "diff.suppress-blank-empty")) {
 378                 diff_suppress_blank_empty = git_config_bool(var, value);
 379                 return 0;
 380         }
 381
 382         if (!strcmp(var, "diff.dirstat")) {
 383                 struct strbuf errmsg = STRBUF_INIT;
 384                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 385                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 386                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 387                                 errmsg.buf);
 388                 strbuf_release(&errmsg);
 389                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 390                 return 0;
 391         }
 392
 393         if (starts_with(var, "submodule."))
 394                 return parse_submodule_config_option(var, value);
 395
 396         if (git_diff_heuristic_config(var, value, cb) < 0)
 397                 return -1;
 398
 399         return git_default_config(var, value, cb);
 400 }
 401
 402 static char *quote_two(const char *one, const char *two)
 403 {
 404         int need_one = quote_c_style(one, NULL, NULL, 1);
 405         int need_two = quote_c_style(two, NULL, NULL, 1);
 406         struct strbuf res = STRBUF_INIT;
 407
 408         if (need_one + need_two) {
 409                 strbuf_addch(&res, '"');
 410                 quote_c_style(one, &res, NULL, 1);
 411                 quote_c_style(two, &res, NULL, 1);
 412                 strbuf_addch(&res, '"');
 413         } else {
 414                 strbuf_addstr(&res, one);
 415                 strbuf_addstr(&res, two);
 416         }
 417         return strbuf_detach(&res, NULL);
 418 }
 419
 420 static const char *external_diff(void)
 421 {
 422         static const char *external_diff_cmd = NULL;
 423         static int done_preparing = 0;
 424
 425         if (done_preparing)
 426                 return external_diff_cmd;
 427         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 428         if (!external_diff_cmd)
 429                 external_diff_cmd = external_diff_cmd_cfg;
 430         done_preparing = 1;
 431         return external_diff_cmd;
 432 }
 433
 434 /*
 435  * Keep track of files used for diffing. Sometimes such an entry
 436  * refers to a temporary file, sometimes to an existing file, and
 437  * sometimes to "/dev/null".
 438  */
 439 static struct diff_tempfile {
 440         /*
 441          * filename external diff should read from, or NULL if this
 442          * entry is currently not in use:
 443          */
 444         const char *name;
 445
 446         char hex[GIT_MAX_HEXSZ + 1];
 447         char mode[10];
 448
 449         /*
 450          * If this diff_tempfile instance refers to a temporary file,
 451          * this tempfile object is used to manage its lifetime.
 452          */
 453         struct tempfile tempfile;
 454 } diff_temp[2];
 455
 456 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 457
 458 struct emit_callback {
 459         int color_diff;
 460         unsigned ws_rule;
 461         int blank_at_eof_in_preimage;
 462         int blank_at_eof_in_postimage;
 463         int lno_in_preimage;
 464         int lno_in_postimage;
 465         sane_truncate_fn truncate;
 466         const char **label_path;
 467         struct diff_words_data *diff_words;
 468         struct diff_options *opt;
 469         struct strbuf *header;
 470 };
 471
 472 static int count_lines(const char *data, int size)
 473 {
 474         int count, ch, completely_empty = 1, nl_just_seen = 0;
 475         count = 0;
 476         while (0 < size--) {
 477                 ch = *data++;
 478                 if (ch == '\n') {
 479                         count++;
 480                         nl_just_seen = 1;
 481                         completely_empty = 0;
 482                 }
 483                 else {
 484                         nl_just_seen = 0;
 485                         completely_empty = 0;
 486                 }
 487         }
 488         if (completely_empty)
 489                 return 0;
 490         if (!nl_just_seen)
 491                 count++; /* no trailing newline */
 492         return count;
 493 }
 494
 495 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 496 {
 497         if (!DIFF_FILE_VALID(one)) {
 498                 mf->ptr = (char *)""; /* does not matter */
 499                 mf->size = 0;
 500                 return 0;
 501         }
 502         else if (diff_populate_filespec(one, 0))
 503                 return -1;
 504
 505         mf->ptr = one->data;
 506         mf->size = one->size;
 507         return 0;
 508 }
 509
 510 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 511 static unsigned long diff_filespec_size(struct diff_filespec *one)
 512 {
 513         if (!DIFF_FILE_VALID(one))
 514                 return 0;
 515         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 516         return one->size;
 517 }
 518
 519 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 520 {
 521         char *ptr = mf->ptr;
 522         long size = mf->size;
 523         int cnt = 0;
 524
 525         if (!size)
 526                 return cnt;
 527         ptr += size - 1; /* pointing at the very end */
 528         if (*ptr != '\n')
 529                 ; /* incomplete line */
 530         else
 531                 ptr--; /* skip the last LF */
 532         while (mf->ptr < ptr) {
 533                 char *prev_eol;
 534                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 535                         if (*prev_eol == '\n')
 536                                 break;
 537                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 538                         break;
 539                 cnt++;
 540                 ptr = prev_eol - 1;
 541         }
 542         return cnt;
 543 }
 544
 545 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 546                                struct emit_callback *ecbdata)
 547 {
 548         int l1, l2, at;
 549         unsigned ws_rule = ecbdata->ws_rule;
 550         l1 = count_trailing_blank(mf1, ws_rule);
 551         l2 = count_trailing_blank(mf2, ws_rule);
 552         if (l2 <= l1) {
 553                 ecbdata->blank_at_eof_in_preimage = 0;
 554                 ecbdata->blank_at_eof_in_postimage = 0;
 555                 return;
 556         }
 557         at = count_lines(mf1->ptr, mf1->size);
 558         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 559
 560         at = count_lines(mf2->ptr, mf2->size);
 561         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 562 }
 563
 564 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 565                         int first, const char *line, int len)
 566 {
 567         int has_trailing_newline, has_trailing_carriage_return;
 568         int nofirst;
 569         FILE *file = o->file;
 570
 571         fputs(diff_line_prefix(o), file);
 572
 573         if (len == 0) {
 574                 has_trailing_newline = (first == '\n');
 575                 has_trailing_carriage_return = (!has_trailing_newline &&
 576                                                 (first == '\r'));
 577                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 578         } else {
 579                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 580                 if (has_trailing_newline)
 581                         len--;
 582                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 583                 if (has_trailing_carriage_return)
 584                         len--;
 585                 nofirst = 0;
 586         }
 587
 588         if (len || !nofirst) {
 589                 fputs(set, file);
 590                 if (!nofirst)
 591                         fputc(first, file);
 592                 fwrite(line, len, 1, file);
 593                 fputs(reset, file);
 594         }
 595         if (has_trailing_carriage_return)
 596                 fputc('\r', file);
 597         if (has_trailing_newline)
 598                 fputc('\n', file);
 599 }
 600
 601 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 602                       const char *line, int len)
 603 {
 604         emit_line_0(o, set, reset, line[0], line+1, len-1);
 605 }
 606
 607 enum diff_symbol {
 608         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 609         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 610         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 611         DIFF_SYMBOL_BINARY_DIFF_BODY,
 612         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 613         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 614         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 615         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 616         DIFF_SYMBOL_STATS_LINE,
 617         DIFF_SYMBOL_WORD_DIFF,
 618         DIFF_SYMBOL_STAT_SEP,
 619         DIFF_SYMBOL_SUMMARY,
 620         DIFF_SYMBOL_SUBMODULE_ADD,
 621         DIFF_SYMBOL_SUBMODULE_DEL,
 622         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 623         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 624         DIFF_SYMBOL_SUBMODULE_HEADER,
 625         DIFF_SYMBOL_SUBMODULE_ERROR,
 626         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 627         DIFF_SYMBOL_REWRITE_DIFF,
 628         DIFF_SYMBOL_BINARY_FILES,
 629         DIFF_SYMBOL_HEADER,
 630         DIFF_SYMBOL_FILEPAIR_PLUS,
 631         DIFF_SYMBOL_FILEPAIR_MINUS,
 632         DIFF_SYMBOL_WORDS_PORCELAIN,
 633         DIFF_SYMBOL_WORDS,
 634         DIFF_SYMBOL_CONTEXT,
 635         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 636         DIFF_SYMBOL_PLUS,
 637         DIFF_SYMBOL_MINUS,
 638         DIFF_SYMBOL_NO_LF_EOF,
 639         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 640         DIFF_SYMBOL_CONTEXT_MARKER,
 641         DIFF_SYMBOL_SEPARATOR
 642 };
 643 /*
 644  * Flags for content lines:
 645  * 0..12 are whitespace rules
 646  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 647  * 16 is marking if the line is blank at EOF
 648  */
 649 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 650 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 651 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 652 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 653
 654 /*
 655  * This struct is used when we need to buffer the output of the diff output.
 656  *
 657  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 658  * into the pre/post image file. This pointer could be a union with the
 659  * line pointer. By storing an offset into the file instead of the literal line,
 660  * we can decrease the memory footprint for the buffered output. At first we
 661  * may want to only have indirection for the content lines, but we could also
 662  * enhance the state for emitting prefabricated lines, e.g. the similarity
 663  * score line or hunk/file headers would only need to store a number or path
 664  * and then the output can be constructed later on depending on state.
 665  */
 666 struct emitted_diff_symbol {
 667         const char *line;
 668         int len;
 669         int flags;
 670         enum diff_symbol s;
 671 };
 672 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 673
 674 struct emitted_diff_symbols {
 675         struct emitted_diff_symbol *buf;
 676         int nr, alloc;
 677 };
 678 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 679
 680 static void append_emitted_diff_symbol(struct diff_options *o,
 681                                        struct emitted_diff_symbol *e)
 682 {
 683         struct emitted_diff_symbol *f;
 684
 685         ALLOC_GROW(o->emitted_symbols->buf,
 686                    o->emitted_symbols->nr + 1,
 687                    o->emitted_symbols->alloc);
 688         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 689
 690         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 691         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 692 }
 693
 694 struct moved_entry {
 695         struct hashmap_entry ent;
 696         const struct emitted_diff_symbol *es;
 697         struct moved_entry *next_line;
 698 };
 699
 700 static int next_byte(const char **cp, const char **endp,
 701                      const struct diff_options *diffopt)
 702 {
 703         int retval;
 704
 705         if (*cp > *endp)
 706                 return -1;
 707
 708         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_CHANGE)) {
 709                 while (*cp < *endp && isspace(**cp))
 710                         (*cp)++;
 711                 /*
 712                  * After skipping a couple of whitespaces, we still have to
 713                  * account for one space.
 714                  */
 715                 return (int)' ';
 716         }
 717
 718         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE)) {
 719                 while (*cp < *endp && isspace(**cp))
 720                         (*cp)++;
 721                 /* return the first non-ws character via the usual below */
 722         }
 723
 724         retval = (unsigned char)(**cp);
 725         (*cp)++;
 726         return retval;
 727 }
 728
 729 static int moved_entry_cmp(const struct diff_options *diffopt,
 730                            const struct moved_entry *a,
 731                            const struct moved_entry *b,
 732                            const void *keydata)
 733 {
 734         const char *ap = a->es->line, *ae = a->es->line + a->es->len;
 735         const char *bp = b->es->line, *be = b->es->line + b->es->len;
 736
 737         if (!(diffopt->xdl_opts & XDF_WHITESPACE_FLAGS))
 738                 return a->es->len != b->es->len  || memcmp(ap, bp, a->es->len);
 739
 740         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_AT_EOL)) {
 741                 while (ae > ap && isspace(*ae))
 742                         ae--;
 743                 while (be > bp && isspace(*be))
 744                         be--;
 745         }
 746
 747         while (1) {
 748                 int ca, cb;
 749                 ca = next_byte(&ap, &ae, diffopt);
 750                 cb = next_byte(&bp, &be, diffopt);
 751                 if (ca != cb)
 752                         return 1;
 753                 if (ca < 0)
 754                         return 0;
 755         }
 756 }
 757
 758 static unsigned get_string_hash(struct emitted_diff_symbol *es, struct diff_options *o)
 759 {
 760         if (o->xdl_opts & XDF_WHITESPACE_FLAGS) {
 761                 static struct strbuf sb = STRBUF_INIT;
 762                 const char *ap = es->line, *ae = es->line + es->len;
 763                 int c;
 764
 765                 strbuf_reset(&sb);
 766                 while (ae > ap && isspace(*ae))
 767                         ae--;
 768                 while ((c = next_byte(&ap, &ae, o)) > 0)
 769                         strbuf_addch(&sb, c);
 770
 771                 return memhash(sb.buf, sb.len);
 772         } else {
 773                 return memhash(es->line, es->len);
 774         }
 775 }
 776
 777 static struct moved_entry *prepare_entry(struct diff_options *o,
 778                                          int line_no)
 779 {
 780         struct moved_entry *ret = xmalloc(sizeof(*ret));
 781         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 782
 783         ret->ent.hash = get_string_hash(l, o);
 784         ret->es = l;
 785         ret->next_line = NULL;
 786
 787         return ret;
 788 }
 789
 790 static void add_lines_to_move_detection(struct diff_options *o,
 791                                         struct hashmap *add_lines,
 792                                         struct hashmap *del_lines)
 793 {
 794         struct moved_entry *prev_line = NULL;
 795
 796         int n;
 797         for (n = 0; n < o->emitted_symbols->nr; n++) {
 798                 struct hashmap *hm;
 799                 struct moved_entry *key;
 800
 801                 switch (o->emitted_symbols->buf[n].s) {
 802                 case DIFF_SYMBOL_PLUS:
 803                         hm = add_lines;
 804                         break;
 805                 case DIFF_SYMBOL_MINUS:
 806                         hm = del_lines;
 807                         break;
 808                 default:
 809                         prev_line = NULL;
 810                         continue;
 811                 }
 812
 813                 key = prepare_entry(o, n);
 814                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 815                         prev_line->next_line = key;
 816
 817                 hashmap_add(hm, key);
 818                 prev_line = key;
 819         }
 820 }
 821
 822 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 823                                          int pmb_nr)
 824 {
 825         int lp, rp;
 826
 827         /* Shrink the set of potential block to the remaining running */
 828         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 829                 while (lp < pmb_nr && pmb[lp])
 830                         lp++;
 831                 /* lp points at the first NULL now */
 832
 833                 while (rp > -1 && !pmb[rp])
 834                         rp--;
 835                 /* rp points at the last non-NULL */
 836
 837                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 838                         pmb[lp] = pmb[rp];
 839                         pmb[rp] = NULL;
 840                         rp--;
 841                         lp++;
 842                 }
 843         }
 844
 845         /* Remember the number of running sets */
 846         return rp + 1;
 847 }
 848
 849 /* Find blocks of moved code, delegate actual coloring decision to helper */
 850 static void mark_color_as_moved(struct diff_options *o,
 851                                 struct hashmap *add_lines,
 852                                 struct hashmap *del_lines)
 853 {
 854         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 855         int pmb_nr = 0, pmb_alloc = 0;
 856         int n, flipped_block = 1, block_length = 0;
 857
 858
 859         for (n = 0; n < o->emitted_symbols->nr; n++) {
 860                 struct hashmap *hm = NULL;
 861                 struct moved_entry *key;
 862                 struct moved_entry *match = NULL;
 863                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 864                 int i;
 865
 866                 switch (l->s) {
 867                 case DIFF_SYMBOL_PLUS:
 868                         hm = del_lines;
 869                         key = prepare_entry(o, n);
 870                         match = hashmap_get(hm, key, o);
 871                         free(key);
 872                         break;
 873                 case DIFF_SYMBOL_MINUS:
 874                         hm = add_lines;
 875                         key = prepare_entry(o, n);
 876                         match = hashmap_get(hm, key, o);
 877                         free(key);
 878                         break;
 879                 default:
 880                         flipped_block = 1;
 881                 }
 882
 883                 if (!match) {
 884                         if (block_length < COLOR_MOVED_MIN_BLOCK_LENGTH &&
 885                             o->color_moved != COLOR_MOVED_PLAIN) {
 886                                 for (i = 0; i < block_length + 1; i++) {
 887                                         l = &o->emitted_symbols->buf[n - i];
 888                                         l->flags &= ~DIFF_SYMBOL_MOVED_LINE;
 889                                 }
 890                         }
 891                         pmb_nr = 0;
 892                         block_length = 0;
 893                         continue;
 894                 }
 895
 896                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 897                 block_length++;
 898
 899                 if (o->color_moved == COLOR_MOVED_PLAIN)
 900                         continue;
 901
 902                 /* Check any potential block runs, advance each or nullify */
 903                 for (i = 0; i < pmb_nr; i++) {
 904                         struct moved_entry *p = pmb[i];
 905                         struct moved_entry *pnext = (p && p->next_line) ?
 906                                         p->next_line : NULL;
 907                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 908                                 pmb[i] = p->next_line;
 909                         } else {
 910                                 pmb[i] = NULL;
 911                         }
 912                 }
 913
 914                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 915
 916                 if (pmb_nr == 0) {
 917                         /*
 918                          * The current line is the start of a new block.
 919                          * Setup the set of potential blocks.
 920                          */
 921                         for (; match; match = hashmap_get_next(hm, match)) {
 922                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 923                                 pmb[pmb_nr++] = match;
 924                         }
 925
 926                         flipped_block = (flipped_block + 1) % 2;
 927                 }
 928
 929                 if (flipped_block)
 930                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 931         }
 932
 933         free(pmb);
 934 }
 935
 936 static void emit_line_ws_markup(struct diff_options *o,
 937                                 const char *set, const char *reset,
 938                                 const char *line, int len, char sign,
 939                                 unsigned ws_rule, int blank_at_eof)
 940 {
 941         const char *ws = NULL;
 942
 943         if (o->ws_error_highlight & ws_rule) {
 944                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 945                 if (!*ws)
 946                         ws = NULL;
 947         }
 948
 949         if (!ws)
 950                 emit_line_0(o, set, reset, sign, line, len);
 951         else if (blank_at_eof)
 952                 /* Blank line at EOF - paint '+' as well */
 953                 emit_line_0(o, ws, reset, sign, line, len);
 954         else {
 955                 /* Emit just the prefix, then the rest. */
 956                 emit_line_0(o, set, reset, sign, "", 0);
 957                 ws_check_emit(line, len, ws_rule,
 958                               o->file, set, reset, ws);
 959         }
 960 }
 961
 962 static void emit_diff_symbol_from_struct(struct diff_options *o,
 963                                          struct emitted_diff_symbol *eds)
 964 {
 965         static const char *nneof = " No newline at end of file\n";
 966         const char *context, *reset, *set, *meta, *fraginfo;
 967         struct strbuf sb = STRBUF_INIT;
 968
 969         enum diff_symbol s = eds->s;
 970         const char *line = eds->line;
 971         int len = eds->len;
 972         unsigned flags = eds->flags;
 973
 974         switch (s) {
 975         case DIFF_SYMBOL_NO_LF_EOF:
 976                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 977                 reset = diff_get_color_opt(o, DIFF_RESET);
 978                 putc('\n', o->file);
 979                 emit_line_0(o, context, reset, '\\',
 980                             nneof, strlen(nneof));
 981                 break;
 982         case DIFF_SYMBOL_SUBMODULE_HEADER:
 983         case DIFF_SYMBOL_SUBMODULE_ERROR:
 984         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
 985         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
 986         case DIFF_SYMBOL_SUMMARY:
 987         case DIFF_SYMBOL_STATS_LINE:
 988         case DIFF_SYMBOL_BINARY_DIFF_BODY:
 989         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
 990                 emit_line(o, "", "", line, len);
 991                 break;
 992         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
 993         case DIFF_SYMBOL_CONTEXT_MARKER:
 994                 context = diff_get_color_opt(o, DIFF_CONTEXT);
 995                 reset = diff_get_color_opt(o, DIFF_RESET);
 996                 emit_line(o, context, reset, line, len);
 997                 break;
 998         case DIFF_SYMBOL_SEPARATOR:
 999                 fprintf(o->file, "%s%c",
1000                         diff_line_prefix(o),
1001                         o->line_termination);
1002                 break;
1003         case DIFF_SYMBOL_CONTEXT:
1004                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1005                 reset = diff_get_color_opt(o, DIFF_RESET);
1006                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1007                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1008                 break;
1009         case DIFF_SYMBOL_PLUS:
1010                 if (flags & DIFF_SYMBOL_MOVED_LINE_ALT)
1011                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1012                 else if (flags & DIFF_SYMBOL_MOVED_LINE)
1013                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1014                 else
1015                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1016                 reset = diff_get_color_opt(o, DIFF_RESET);
1017                 emit_line_ws_markup(o, set, reset, line, len, '+',
1018                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1019                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1020                 break;
1021         case DIFF_SYMBOL_MINUS:
1022                 if (flags & DIFF_SYMBOL_MOVED_LINE_ALT)
1023                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1024                 else if (flags & DIFF_SYMBOL_MOVED_LINE)
1025                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1026                 else
1027                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1028                 reset = diff_get_color_opt(o, DIFF_RESET);
1029                 emit_line_ws_markup(o, set, reset, line, len, '-',
1030                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1031                 break;
1032         case DIFF_SYMBOL_WORDS_PORCELAIN:
1033                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1034                 reset = diff_get_color_opt(o, DIFF_RESET);
1035                 emit_line(o, context, reset, line, len);
1036                 fputs("~\n", o->file);
1037                 break;
1038         case DIFF_SYMBOL_WORDS:
1039                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1040                 reset = diff_get_color_opt(o, DIFF_RESET);
1041                 /*
1042                  * Skip the prefix character, if any.  With
1043                  * diff_suppress_blank_empty, there may be
1044                  * none.
1045                  */
1046                 if (line[0] != '\n') {
1047                         line++;
1048                         len--;
1049                 }
1050                 emit_line(o, context, reset, line, len);
1051                 break;
1052         case DIFF_SYMBOL_FILEPAIR_PLUS:
1053                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1054                 reset = diff_get_color_opt(o, DIFF_RESET);
1055                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1056                         line, reset,
1057                         strchr(line, ' ') ? "\t" : "");
1058                 break;
1059         case DIFF_SYMBOL_FILEPAIR_MINUS:
1060                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1061                 reset = diff_get_color_opt(o, DIFF_RESET);
1062                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1063                         line, reset,
1064                         strchr(line, ' ') ? "\t" : "");
1065                 break;
1066         case DIFF_SYMBOL_BINARY_FILES:
1067         case DIFF_SYMBOL_HEADER:
1068                 fprintf(o->file, "%s", line);
1069                 break;
1070         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1071                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1072                 break;
1073         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1074                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1075                 break;
1076         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1077                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1078                 break;
1079         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1080                 fputs(diff_line_prefix(o), o->file);
1081                 fputc('\n', o->file);
1082                 break;
1083         case DIFF_SYMBOL_REWRITE_DIFF:
1084                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1085                 reset = diff_get_color_opt(o, DIFF_RESET);
1086                 emit_line(o, fraginfo, reset, line, len);
1087                 break;
1088         case DIFF_SYMBOL_SUBMODULE_ADD:
1089                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1090                 reset = diff_get_color_opt(o, DIFF_RESET);
1091                 emit_line(o, set, reset, line, len);
1092                 break;
1093         case DIFF_SYMBOL_SUBMODULE_DEL:
1094                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1095                 reset = diff_get_color_opt(o, DIFF_RESET);
1096                 emit_line(o, set, reset, line, len);
1097                 break;
1098         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1099                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1100                         diff_line_prefix(o), line);
1101                 break;
1102         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1103                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1104                         diff_line_prefix(o), line);
1105                 break;
1106         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1107                 emit_line(o, "", "", " 0 files changed\n",
1108                           strlen(" 0 files changed\n"));
1109                 break;
1110         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1111                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1112                 break;
1113         case DIFF_SYMBOL_WORD_DIFF:
1114                 fprintf(o->file, "%.*s", len, line);
1115                 break;
1116         case DIFF_SYMBOL_STAT_SEP:
1117                 fputs(o->stat_sep, o->file);
1118                 break;
1119         default:
1120                 die("BUG: unknown diff symbol");
1121         }
1122         strbuf_release(&sb);
1123 }
1124
1125 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1126                              const char *line, int len, unsigned flags)
1127 {
1128         struct emitted_diff_symbol e = {line, len, flags, s};
1129
1130         if (o->emitted_symbols)
1131                 append_emitted_diff_symbol(o, &e);
1132         else
1133                 emit_diff_symbol_from_struct(o, &e);
1134 }
1135
1136 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1137 {
1138         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1139 }
1140
1141 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1142 {
1143         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1144 }
1145
1146 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1147 {
1148         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1149                          path, strlen(path), 0);
1150 }
1151
1152 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1153 {
1154         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1155                          path, strlen(path), 0);
1156 }
1157
1158 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1159 {
1160         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1161                          header, strlen(header), 0);
1162 }
1163
1164 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1165 {
1166         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1167 }
1168
1169 void diff_emit_submodule_pipethrough(struct diff_options *o,
1170                                      const char *line, int len)
1171 {
1172         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1173 }
1174
1175 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1176 {
1177         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1178               ecbdata->blank_at_eof_in_preimage &&
1179               ecbdata->blank_at_eof_in_postimage &&
1180               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1181               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1182                 return 0;
1183         return ws_blank_line(line, len, ecbdata->ws_rule);
1184 }
1185
1186 static void emit_add_line(const char *reset,
1187                           struct emit_callback *ecbdata,
1188                           const char *line, int len)
1189 {
1190         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1191         if (new_blank_line_at_eof(ecbdata, line, len))
1192                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1193
1194         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1195 }
1196
1197 static void emit_del_line(const char *reset,
1198                           struct emit_callback *ecbdata,
1199                           const char *line, int len)
1200 {
1201         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1202         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1203 }
1204
1205 static void emit_context_line(const char *reset,
1206                               struct emit_callback *ecbdata,
1207                               const char *line, int len)
1208 {
1209         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1210         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1211 }
1212
1213 static void emit_hunk_header(struct emit_callback *ecbdata,
1214                              const char *line, int len)
1215 {
1216         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1217         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1218         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1219         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1220         static const char atat[2] = { '@', '@' };
1221         const char *cp, *ep;
1222         struct strbuf msgbuf = STRBUF_INIT;
1223         int org_len = len;
1224         int i = 1;
1225
1226         /*
1227          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1228          * it always is at least 10 bytes long.
1229          */
1230         if (len < 10 ||
1231             memcmp(line, atat, 2) ||
1232             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1233                 emit_diff_symbol(ecbdata->opt,
1234                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1235                 return;
1236         }
1237         ep += 2; /* skip over @@ */
1238
1239         /* The hunk header in fraginfo color */
1240         strbuf_addstr(&msgbuf, frag);
1241         strbuf_add(&msgbuf, line, ep - line);
1242         strbuf_addstr(&msgbuf, reset);
1243
1244         /*
1245          * trailing "\r\n"
1246          */
1247         for ( ; i < 3; i++)
1248                 if (line[len - i] == '\r' || line[len - i] == '\n')
1249                         len--;
1250
1251         /* blank before the func header */
1252         for (cp = ep; ep - line < len; ep++)
1253                 if (*ep != ' ' && *ep != '\t')
1254                         break;
1255         if (ep != cp) {
1256                 strbuf_addstr(&msgbuf, context);
1257                 strbuf_add(&msgbuf, cp, ep - cp);
1258                 strbuf_addstr(&msgbuf, reset);
1259         }
1260
1261         if (ep < line + len) {
1262                 strbuf_addstr(&msgbuf, func);
1263                 strbuf_add(&msgbuf, ep, line + len - ep);
1264                 strbuf_addstr(&msgbuf, reset);
1265         }
1266
1267         strbuf_add(&msgbuf, line + len, org_len - len);
1268         strbuf_complete_line(&msgbuf);
1269         emit_diff_symbol(ecbdata->opt,
1270                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1271         strbuf_release(&msgbuf);
1272 }
1273
1274 static struct diff_tempfile *claim_diff_tempfile(void) {
1275         int i;
1276         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1277                 if (!diff_temp[i].name)
1278                         return diff_temp + i;
1279         die("BUG: diff is failing to clean up its tempfiles");
1280 }
1281
1282 static void remove_tempfile(void)
1283 {
1284         int i;
1285         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1286                 if (is_tempfile_active(&diff_temp[i].tempfile))
1287                         delete_tempfile(&diff_temp[i].tempfile);
1288                 diff_temp[i].name = NULL;
1289         }
1290 }
1291
1292 static void add_line_count(struct strbuf *out, int count)
1293 {
1294         switch (count) {
1295         case 0:
1296                 strbuf_addstr(out, "0,0");
1297                 break;
1298         case 1:
1299                 strbuf_addstr(out, "1");
1300                 break;
1301         default:
1302                 strbuf_addf(out, "1,%d", count);
1303                 break;
1304         }
1305 }
1306
1307 static void emit_rewrite_lines(struct emit_callback *ecb,
1308                                int prefix, const char *data, int size)
1309 {
1310         const char *endp = NULL;
1311         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1312
1313         while (0 < size) {
1314                 int len;
1315
1316                 endp = memchr(data, '\n', size);
1317                 len = endp ? (endp - data + 1) : size;
1318                 if (prefix != '+') {
1319                         ecb->lno_in_preimage++;
1320                         emit_del_line(reset, ecb, data, len);
1321                 } else {
1322                         ecb->lno_in_postimage++;
1323                         emit_add_line(reset, ecb, data, len);
1324                 }
1325                 size -= len;
1326                 data += len;
1327         }
1328         if (!endp)
1329                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1330 }
1331
1332 static void emit_rewrite_diff(const char *name_a,
1333                               const char *name_b,
1334                               struct diff_filespec *one,
1335                               struct diff_filespec *two,
1336                               struct userdiff_driver *textconv_one,
1337                               struct userdiff_driver *textconv_two,
1338                               struct diff_options *o)
1339 {
1340         int lc_a, lc_b;
1341         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1342         const char *a_prefix, *b_prefix;
1343         char *data_one, *data_two;
1344         size_t size_one, size_two;
1345         struct emit_callback ecbdata;
1346         struct strbuf out = STRBUF_INIT;
1347
1348         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
1349                 a_prefix = o->b_prefix;
1350                 b_prefix = o->a_prefix;
1351         } else {
1352                 a_prefix = o->a_prefix;
1353                 b_prefix = o->b_prefix;
1354         }
1355
1356         name_a += (*name_a == '/');
1357         name_b += (*name_b == '/');
1358
1359         strbuf_reset(&a_name);
1360         strbuf_reset(&b_name);
1361         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1362         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1363
1364         size_one = fill_textconv(textconv_one, one, &data_one);
1365         size_two = fill_textconv(textconv_two, two, &data_two);
1366
1367         memset(&ecbdata, 0, sizeof(ecbdata));
1368         ecbdata.color_diff = want_color(o->use_color);
1369         ecbdata.ws_rule = whitespace_rule(name_b);
1370         ecbdata.opt = o;
1371         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1372                 mmfile_t mf1, mf2;
1373                 mf1.ptr = (char *)data_one;
1374                 mf2.ptr = (char *)data_two;
1375                 mf1.size = size_one;
1376                 mf2.size = size_two;
1377                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1378         }
1379         ecbdata.lno_in_preimage = 1;
1380         ecbdata.lno_in_postimage = 1;
1381
1382         lc_a = count_lines(data_one, size_one);
1383         lc_b = count_lines(data_two, size_two);
1384
1385         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1386                          a_name.buf, a_name.len, 0);
1387         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1388                          b_name.buf, b_name.len, 0);
1389
1390         strbuf_addstr(&out, "@@ -");
1391         if (!o->irreversible_delete)
1392                 add_line_count(&out, lc_a);
1393         else
1394                 strbuf_addstr(&out, "?,?");
1395         strbuf_addstr(&out, " +");
1396         add_line_count(&out, lc_b);
1397         strbuf_addstr(&out, " @@\n");
1398         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1399         strbuf_release(&out);
1400
1401         if (lc_a && !o->irreversible_delete)
1402                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1403         if (lc_b)
1404                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1405         if (textconv_one)
1406                 free((char *)data_one);
1407         if (textconv_two)
1408                 free((char *)data_two);
1409 }
1410
1411 struct diff_words_buffer {
1412         mmfile_t text;
1413         long alloc;
1414         struct diff_words_orig {
1415                 const char *begin, *end;
1416         } *orig;
1417         int orig_nr, orig_alloc;
1418 };
1419
1420 static void diff_words_append(char *line, unsigned long len,
1421                 struct diff_words_buffer *buffer)
1422 {
1423         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1424         line++;
1425         len--;
1426         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1427         buffer->text.size += len;
1428         buffer->text.ptr[buffer->text.size] = '\0';
1429 }
1430
1431 struct diff_words_style_elem {
1432         const char *prefix;
1433         const char *suffix;
1434         const char *color; /* NULL; filled in by the setup code if
1435                             * color is enabled */
1436 };
1437
1438 struct diff_words_style {
1439         enum diff_words_type type;
1440         struct diff_words_style_elem new, old, ctx;
1441         const char *newline;
1442 };
1443
1444 static struct diff_words_style diff_words_styles[] = {
1445         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1446         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1447         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1448 };
1449
1450 struct diff_words_data {
1451         struct diff_words_buffer minus, plus;
1452         const char *current_plus;
1453         int last_minus;
1454         struct diff_options *opt;
1455         regex_t *word_regex;
1456         enum diff_words_type type;
1457         struct diff_words_style *style;
1458 };
1459
1460 static int fn_out_diff_words_write_helper(struct diff_options *o,
1461                                           struct diff_words_style_elem *st_el,
1462                                           const char *newline,
1463                                           size_t count, const char *buf)
1464 {
1465         int print = 0;
1466         struct strbuf sb = STRBUF_INIT;
1467
1468         while (count) {
1469                 char *p = memchr(buf, '\n', count);
1470                 if (print)
1471                         strbuf_addstr(&sb, diff_line_prefix(o));
1472
1473                 if (p != buf) {
1474                         const char *reset = st_el->color && *st_el->color ?
1475                                             GIT_COLOR_RESET : NULL;
1476                         if (st_el->color && *st_el->color)
1477                                 strbuf_addstr(&sb, st_el->color);
1478                         strbuf_addstr(&sb, st_el->prefix);
1479                         strbuf_add(&sb, buf, p ? p - buf : count);
1480                         strbuf_addstr(&sb, st_el->suffix);
1481                         if (reset)
1482                                 strbuf_addstr(&sb, reset);
1483                 }
1484                 if (!p)
1485                         goto out;
1486
1487                 strbuf_addstr(&sb, newline);
1488                 count -= p + 1 - buf;
1489                 buf = p + 1;
1490                 print = 1;
1491                 if (count) {
1492                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1493                                          sb.buf, sb.len, 0);
1494                         strbuf_reset(&sb);
1495                 }
1496         }
1497
1498 out:
1499         if (sb.len)
1500                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1501                                  sb.buf, sb.len, 0);
1502         strbuf_release(&sb);
1503         return 0;
1504 }
1505
1506 /*
1507  * '--color-words' algorithm can be described as:
1508  *
1509  *   1. collect the minus/plus lines of a diff hunk, divided into
1510  *      minus-lines and plus-lines;
1511  *
1512  *   2. break both minus-lines and plus-lines into words and
1513  *      place them into two mmfile_t with one word for each line;
1514  *
1515  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1516  *
1517  * And for the common parts of the both file, we output the plus side text.
1518  * diff_words->current_plus is used to trace the current position of the plus file
1519  * which printed. diff_words->last_minus is used to trace the last minus word
1520  * printed.
1521  *
1522  * For '--graph' to work with '--color-words', we need to output the graph prefix
1523  * on each line of color words output. Generally, there are two conditions on
1524  * which we should output the prefix.
1525  *
1526  *   1. diff_words->last_minus == 0 &&
1527  *      diff_words->current_plus == diff_words->plus.text.ptr
1528  *
1529  *      that is: the plus text must start as a new line, and if there is no minus
1530  *      word printed, a graph prefix must be printed.
1531  *
1532  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1533  *      *(diff_words->current_plus - 1) == '\n'
1534  *
1535  *      that is: a graph prefix must be printed following a '\n'
1536  */
1537 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1538 {
1539         if ((diff_words->last_minus == 0 &&
1540                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1541                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1542                 *(diff_words->current_plus - 1) == '\n')) {
1543                 return 1;
1544         } else {
1545                 return 0;
1546         }
1547 }
1548
1549 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1550 {
1551         struct diff_words_data *diff_words = priv;
1552         struct diff_words_style *style = diff_words->style;
1553         int minus_first, minus_len, plus_first, plus_len;
1554         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1555         struct diff_options *opt = diff_words->opt;
1556         const char *line_prefix;
1557
1558         if (line[0] != '@' || parse_hunk_header(line, len,
1559                         &minus_first, &minus_len, &plus_first, &plus_len))
1560                 return;
1561
1562         assert(opt);
1563         line_prefix = diff_line_prefix(opt);
1564
1565         /* POSIX requires that first be decremented by one if len == 0... */
1566         if (minus_len) {
1567                 minus_begin = diff_words->minus.orig[minus_first].begin;
1568                 minus_end =
1569                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1570         } else
1571                 minus_begin = minus_end =
1572                         diff_words->minus.orig[minus_first].end;
1573
1574         if (plus_len) {
1575                 plus_begin = diff_words->plus.orig[plus_first].begin;
1576                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1577         } else
1578                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1579
1580         if (color_words_output_graph_prefix(diff_words)) {
1581                 fputs(line_prefix, diff_words->opt->file);
1582         }
1583         if (diff_words->current_plus != plus_begin) {
1584                 fn_out_diff_words_write_helper(diff_words->opt,
1585                                 &style->ctx, style->newline,
1586                                 plus_begin - diff_words->current_plus,
1587                                 diff_words->current_plus);
1588         }
1589         if (minus_begin != minus_end) {
1590                 fn_out_diff_words_write_helper(diff_words->opt,
1591                                 &style->old, style->newline,
1592                                 minus_end - minus_begin, minus_begin);
1593         }
1594         if (plus_begin != plus_end) {
1595                 fn_out_diff_words_write_helper(diff_words->opt,
1596                                 &style->new, style->newline,
1597                                 plus_end - plus_begin, plus_begin);
1598         }
1599
1600         diff_words->current_plus = plus_end;
1601         diff_words->last_minus = minus_first;
1602 }
1603
1604 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1605 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1606                 int *begin, int *end)
1607 {
1608         if (word_regex && *begin < buffer->size) {
1609                 regmatch_t match[1];
1610                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1611                                  buffer->size - *begin, 1, match, 0)) {
1612                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1613                                         '\n', match[0].rm_eo - match[0].rm_so);
1614                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1615                         *begin += match[0].rm_so;
1616                         return *begin >= *end;
1617                 }
1618                 return -1;
1619         }
1620
1621         /* find the next word */
1622         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1623                 (*begin)++;
1624         if (*begin >= buffer->size)
1625                 return -1;
1626
1627         /* find the end of the word */
1628         *end = *begin + 1;
1629         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1630                 (*end)++;
1631
1632         return 0;
1633 }
1634
1635 /*
1636  * This function splits the words in buffer->text, stores the list with
1637  * newline separator into out, and saves the offsets of the original words
1638  * in buffer->orig.
1639  */
1640 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1641                 regex_t *word_regex)
1642 {
1643         int i, j;
1644         long alloc = 0;
1645
1646         out->size = 0;
1647         out->ptr = NULL;
1648
1649         /* fake an empty "0th" word */
1650         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1651         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1652         buffer->orig_nr = 1;
1653
1654         for (i = 0; i < buffer->text.size; i++) {
1655                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1656                         return;
1657
1658                 /* store original boundaries */
1659                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1660                                 buffer->orig_alloc);
1661                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1662                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1663                 buffer->orig_nr++;
1664
1665                 /* store one word */
1666                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1667                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1668                 out->ptr[out->size + j - i] = '\n';
1669                 out->size += j - i + 1;
1670
1671                 i = j - 1;
1672         }
1673 }
1674
1675 /* this executes the word diff on the accumulated buffers */
1676 static void diff_words_show(struct diff_words_data *diff_words)
1677 {
1678         xpparam_t xpp;
1679         xdemitconf_t xecfg;
1680         mmfile_t minus, plus;
1681         struct diff_words_style *style = diff_words->style;
1682
1683         struct diff_options *opt = diff_words->opt;
1684         const char *line_prefix;
1685
1686         assert(opt);
1687         line_prefix = diff_line_prefix(opt);
1688
1689         /* special case: only removal */
1690         if (!diff_words->plus.text.size) {
1691                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1692                                  line_prefix, strlen(line_prefix), 0);
1693                 fn_out_diff_words_write_helper(diff_words->opt,
1694                         &style->old, style->newline,
1695                         diff_words->minus.text.size,
1696                         diff_words->minus.text.ptr);
1697                 diff_words->minus.text.size = 0;
1698                 return;
1699         }
1700
1701         diff_words->current_plus = diff_words->plus.text.ptr;
1702         diff_words->last_minus = 0;
1703
1704         memset(&xpp, 0, sizeof(xpp));
1705         memset(&xecfg, 0, sizeof(xecfg));
1706         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1707         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1708         xpp.flags = 0;
1709         /* as only the hunk header will be parsed, we need a 0-context */
1710         xecfg.ctxlen = 0;
1711         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1712                           &xpp, &xecfg))
1713                 die("unable to generate word diff");
1714         free(minus.ptr);
1715         free(plus.ptr);
1716         if (diff_words->current_plus != diff_words->plus.text.ptr +
1717                         diff_words->plus.text.size) {
1718                 if (color_words_output_graph_prefix(diff_words))
1719                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1720                                          line_prefix, strlen(line_prefix), 0);
1721                 fn_out_diff_words_write_helper(diff_words->opt,
1722                         &style->ctx, style->newline,
1723                         diff_words->plus.text.ptr + diff_words->plus.text.size
1724                         - diff_words->current_plus, diff_words->current_plus);
1725         }
1726         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1727 }
1728
1729 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1730 static void diff_words_flush(struct emit_callback *ecbdata)
1731 {
1732         struct diff_options *wo = ecbdata->diff_words->opt;
1733
1734         if (ecbdata->diff_words->minus.text.size ||
1735             ecbdata->diff_words->plus.text.size)
1736                 diff_words_show(ecbdata->diff_words);
1737
1738         if (wo->emitted_symbols) {
1739                 struct diff_options *o = ecbdata->opt;
1740                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1741                 int i;
1742
1743                 /*
1744                  * NEEDSWORK:
1745                  * Instead of appending each, concat all words to a line?
1746                  */
1747                 for (i = 0; i < wol->nr; i++)
1748                         append_emitted_diff_symbol(o, &wol->buf[i]);
1749
1750                 for (i = 0; i < wol->nr; i++)
1751                         free((void *)wol->buf[i].line);
1752
1753                 wol->nr = 0;
1754         }
1755 }
1756
1757 static void diff_filespec_load_driver(struct diff_filespec *one)
1758 {
1759         /* Use already-loaded driver */
1760         if (one->driver)
1761                 return;
1762
1763         if (S_ISREG(one->mode))
1764                 one->driver = userdiff_find_by_path(one->path);
1765
1766         /* Fallback to default settings */
1767         if (!one->driver)
1768                 one->driver = userdiff_find_by_name("default");
1769 }
1770
1771 static const char *userdiff_word_regex(struct diff_filespec *one)
1772 {
1773         diff_filespec_load_driver(one);
1774         return one->driver->word_regex;
1775 }
1776
1777 static void init_diff_words_data(struct emit_callback *ecbdata,
1778                                  struct diff_options *orig_opts,
1779                                  struct diff_filespec *one,
1780                                  struct diff_filespec *two)
1781 {
1782         int i;
1783         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1784         memcpy(o, orig_opts, sizeof(struct diff_options));
1785
1786         ecbdata->diff_words =
1787                 xcalloc(1, sizeof(struct diff_words_data));
1788         ecbdata->diff_words->type = o->word_diff;
1789         ecbdata->diff_words->opt = o;
1790
1791         if (orig_opts->emitted_symbols)
1792                 o->emitted_symbols =
1793                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1794
1795         if (!o->word_regex)
1796                 o->word_regex = userdiff_word_regex(one);
1797         if (!o->word_regex)
1798                 o->word_regex = userdiff_word_regex(two);
1799         if (!o->word_regex)
1800                 o->word_regex = diff_word_regex_cfg;
1801         if (o->word_regex) {
1802                 ecbdata->diff_words->word_regex = (regex_t *)
1803                         xmalloc(sizeof(regex_t));
1804                 if (regcomp(ecbdata->diff_words->word_regex,
1805                             o->word_regex,
1806                             REG_EXTENDED | REG_NEWLINE))
1807                         die ("Invalid regular expression: %s",
1808                              o->word_regex);
1809         }
1810         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1811                 if (o->word_diff == diff_words_styles[i].type) {
1812                         ecbdata->diff_words->style =
1813                                 &diff_words_styles[i];
1814                         break;
1815                 }
1816         }
1817         if (want_color(o->use_color)) {
1818                 struct diff_words_style *st = ecbdata->diff_words->style;
1819                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1820                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1821                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1822         }
1823 }
1824
1825 static void free_diff_words_data(struct emit_callback *ecbdata)
1826 {
1827         if (ecbdata->diff_words) {
1828                 diff_words_flush(ecbdata);
1829                 free (ecbdata->diff_words->opt->emitted_symbols);
1830                 free (ecbdata->diff_words->opt);
1831                 free (ecbdata->diff_words->minus.text.ptr);
1832                 free (ecbdata->diff_words->minus.orig);
1833                 free (ecbdata->diff_words->plus.text.ptr);
1834                 free (ecbdata->diff_words->plus.orig);
1835                 if (ecbdata->diff_words->word_regex) {
1836                         regfree(ecbdata->diff_words->word_regex);
1837                         free(ecbdata->diff_words->word_regex);
1838                 }
1839                 FREE_AND_NULL(ecbdata->diff_words);
1840         }
1841 }
1842
1843 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1844 {
1845         if (want_color(diff_use_color))
1846                 return diff_colors[ix];
1847         return "";
1848 }
1849
1850 const char *diff_line_prefix(struct diff_options *opt)
1851 {
1852         struct strbuf *msgbuf;
1853         if (!opt->output_prefix)
1854                 return "";
1855
1856         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1857         return msgbuf->buf;
1858 }
1859
1860 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1861 {
1862         const char *cp;
1863         unsigned long allot;
1864         size_t l = len;
1865
1866         if (ecb->truncate)
1867                 return ecb->truncate(line, len);
1868         cp = line;
1869         allot = l;
1870         while (0 < l) {
1871                 (void) utf8_width(&cp, &l);
1872                 if (!cp)
1873                         break; /* truncated in the middle? */
1874         }
1875         return allot - l;
1876 }
1877
1878 static void find_lno(const char *line, struct emit_callback *ecbdata)
1879 {
1880         const char *p;
1881         ecbdata->lno_in_preimage = 0;
1882         ecbdata->lno_in_postimage = 0;
1883         p = strchr(line, '-');
1884         if (!p)
1885                 return; /* cannot happen */
1886         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1887         p = strchr(p, '+');
1888         if (!p)
1889                 return; /* cannot happen */
1890         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1891 }
1892
1893 static void fn_out_consume(void *priv, char *line, unsigned long len)
1894 {
1895         struct emit_callback *ecbdata = priv;
1896         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1897         struct diff_options *o = ecbdata->opt;
1898
1899         o->found_changes = 1;
1900
1901         if (ecbdata->header) {
1902                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1903                                  ecbdata->header->buf, ecbdata->header->len, 0);
1904                 strbuf_reset(ecbdata->header);
1905                 ecbdata->header = NULL;
1906         }
1907
1908         if (ecbdata->label_path[0]) {
1909                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1910                                  ecbdata->label_path[0],
1911                                  strlen(ecbdata->label_path[0]), 0);
1912                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1913                                  ecbdata->label_path[1],
1914                                  strlen(ecbdata->label_path[1]), 0);
1915                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1916         }
1917
1918         if (diff_suppress_blank_empty
1919             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1920                 line[0] = '\n';
1921                 len = 1;
1922         }
1923
1924         if (line[0] == '@') {
1925                 if (ecbdata->diff_words)
1926                         diff_words_flush(ecbdata);
1927                 len = sane_truncate_line(ecbdata, line, len);
1928                 find_lno(line, ecbdata);
1929                 emit_hunk_header(ecbdata, line, len);
1930                 return;
1931         }
1932
1933         if (ecbdata->diff_words) {
1934                 enum diff_symbol s =
1935                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
1936                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
1937                 if (line[0] == '-') {
1938                         diff_words_append(line, len,
1939                                           &ecbdata->diff_words->minus);
1940                         return;
1941                 } else if (line[0] == '+') {
1942                         diff_words_append(line, len,
1943                                           &ecbdata->diff_words->plus);
1944                         return;
1945                 } else if (starts_with(line, "\\ ")) {
1946                         /*
1947                          * Eat the "no newline at eof" marker as if we
1948                          * saw a "+" or "-" line with nothing on it,
1949                          * and return without diff_words_flush() to
1950                          * defer processing. If this is the end of
1951                          * preimage, more "+" lines may come after it.
1952                          */
1953                         return;
1954                 }
1955                 diff_words_flush(ecbdata);
1956                 emit_diff_symbol(o, s, line, len, 0);
1957                 return;
1958         }
1959
1960         switch (line[0]) {
1961         case '+':
1962                 ecbdata->lno_in_postimage++;
1963                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1964                 break;
1965         case '-':
1966                 ecbdata->lno_in_preimage++;
1967                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1968                 break;
1969         case ' ':
1970                 ecbdata->lno_in_postimage++;
1971                 ecbdata->lno_in_preimage++;
1972                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1973                 break;
1974         default:
1975                 /* incomplete line at the end */
1976                 ecbdata->lno_in_preimage++;
1977                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
1978                                  line, len, 0);
1979                 break;
1980         }
1981 }
1982
1983 static char *pprint_rename(const char *a, const char *b)
1984 {
1985         const char *old = a;
1986         const char *new = b;
1987         struct strbuf name = STRBUF_INIT;
1988         int pfx_length, sfx_length;
1989         int pfx_adjust_for_slash;
1990         int len_a = strlen(a);
1991         int len_b = strlen(b);
1992         int a_midlen, b_midlen;
1993         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1994         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1995
1996         if (qlen_a || qlen_b) {
1997                 quote_c_style(a, &name, NULL, 0);
1998                 strbuf_addstr(&name, " => ");
1999                 quote_c_style(b, &name, NULL, 0);
2000                 return strbuf_detach(&name, NULL);
2001         }
2002
2003         /* Find common prefix */
2004         pfx_length = 0;
2005         while (*old && *new && *old == *new) {
2006                 if (*old == '/')
2007                         pfx_length = old - a + 1;
2008                 old++;
2009                 new++;
2010         }
2011
2012         /* Find common suffix */
2013         old = a + len_a;
2014         new = b + len_b;
2015         sfx_length = 0;
2016         /*
2017          * If there is a common prefix, it must end in a slash.  In
2018          * that case we let this loop run 1 into the prefix to see the
2019          * same slash.
2020          *
2021          * If there is no common prefix, we cannot do this as it would
2022          * underrun the input strings.
2023          */
2024         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2025         while (a + pfx_length - pfx_adjust_for_slash <= old &&
2026                b + pfx_length - pfx_adjust_for_slash <= new &&
2027                *old == *new) {
2028                 if (*old == '/')
2029                         sfx_length = len_a - (old - a);
2030                 old--;
2031                 new--;
2032         }
2033
2034         /*
2035          * pfx{mid-a => mid-b}sfx
2036          * {pfx-a => pfx-b}sfx
2037          * pfx{sfx-a => sfx-b}
2038          * name-a => name-b
2039          */
2040         a_midlen = len_a - pfx_length - sfx_length;
2041         b_midlen = len_b - pfx_length - sfx_length;
2042         if (a_midlen < 0)
2043                 a_midlen = 0;
2044         if (b_midlen < 0)
2045                 b_midlen = 0;
2046
2047         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2048         if (pfx_length + sfx_length) {
2049                 strbuf_add(&name, a, pfx_length);
2050                 strbuf_addch(&name, '{');
2051         }
2052         strbuf_add(&name, a + pfx_length, a_midlen);
2053         strbuf_addstr(&name, " => ");
2054         strbuf_add(&name, b + pfx_length, b_midlen);
2055         if (pfx_length + sfx_length) {
2056                 strbuf_addch(&name, '}');
2057                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
2058         }
2059         return strbuf_detach(&name, NULL);
2060 }
2061
2062 struct diffstat_t {
2063         int nr;
2064         int alloc;
2065         struct diffstat_file {
2066                 char *from_name;
2067                 char *name;
2068                 char *print_name;
2069                 unsigned is_unmerged:1;
2070                 unsigned is_binary:1;
2071                 unsigned is_renamed:1;
2072                 unsigned is_interesting:1;
2073                 uintmax_t added, deleted;
2074         } **files;
2075 };
2076
2077 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2078                                           const char *name_a,
2079                                           const char *name_b)
2080 {
2081         struct diffstat_file *x;
2082         x = xcalloc(1, sizeof(*x));
2083         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2084         diffstat->files[diffstat->nr++] = x;
2085         if (name_b) {
2086                 x->from_name = xstrdup(name_a);
2087                 x->name = xstrdup(name_b);
2088                 x->is_renamed = 1;
2089         }
2090         else {
2091                 x->from_name = NULL;
2092                 x->name = xstrdup(name_a);
2093         }
2094         return x;
2095 }
2096
2097 static void diffstat_consume(void *priv, char *line, unsigned long len)
2098 {
2099         struct diffstat_t *diffstat = priv;
2100         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2101
2102         if (line[0] == '+')
2103                 x->added++;
2104         else if (line[0] == '-')
2105                 x->deleted++;
2106 }
2107
2108 const char mime_boundary_leader[] = "------------";
2109
2110 static int scale_linear(int it, int width, int max_change)
2111 {
2112         if (!it)
2113                 return 0;
2114         /*
2115          * make sure that at least one '-' or '+' is printed if
2116          * there is any change to this path. The easiest way is to
2117          * scale linearly as if the alloted width is one column shorter
2118          * than it is, and then add 1 to the result.
2119          */
2120         return 1 + (it * (width - 1) / max_change);
2121 }
2122
2123 static void show_graph(struct strbuf *out, char ch, int cnt,
2124                        const char *set, const char *reset)
2125 {
2126         if (cnt <= 0)
2127                 return;
2128         strbuf_addstr(out, set);
2129         strbuf_addchars(out, ch, cnt);
2130         strbuf_addstr(out, reset);
2131 }
2132
2133 static void fill_print_name(struct diffstat_file *file)
2134 {
2135         char *pname;
2136
2137         if (file->print_name)
2138                 return;
2139
2140         if (!file->is_renamed) {
2141                 struct strbuf buf = STRBUF_INIT;
2142                 if (quote_c_style(file->name, &buf, NULL, 0)) {
2143                         pname = strbuf_detach(&buf, NULL);
2144                 } else {
2145                         pname = file->name;
2146                         strbuf_release(&buf);
2147                 }
2148         } else {
2149                 pname = pprint_rename(file->from_name, file->name);
2150         }
2151         file->print_name = pname;
2152 }
2153
2154 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2155                 int files, int insertions, int deletions)
2156 {
2157         struct strbuf sb = STRBUF_INIT;
2158
2159         if (!files) {
2160                 assert(insertions == 0 && deletions == 0);
2161                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2162                                  NULL, 0, 0);
2163                 return;
2164         }
2165
2166         strbuf_addf(&sb,
2167                     (files == 1) ? " %d file changed" : " %d files changed",
2168                     files);
2169
2170         /*
2171          * For binary diff, the caller may want to print "x files
2172          * changed" with insertions == 0 && deletions == 0.
2173          *
2174          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2175          * is probably less confusing (i.e skip over "2 files changed
2176          * but nothing about added/removed lines? Is this a bug in Git?").
2177          */
2178         if (insertions || deletions == 0) {
2179                 strbuf_addf(&sb,
2180                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2181                             insertions);
2182         }
2183
2184         if (deletions || insertions == 0) {
2185                 strbuf_addf(&sb,
2186                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2187                             deletions);
2188         }
2189         strbuf_addch(&sb, '\n');
2190         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2191                          sb.buf, sb.len, 0);
2192         strbuf_release(&sb);
2193 }
2194
2195 void print_stat_summary(FILE *fp, int files,
2196                         int insertions, int deletions)
2197 {
2198         struct diff_options o;
2199         memset(&o, 0, sizeof(o));
2200         o.file = fp;
2201
2202         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2203 }
2204
2205 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2206 {
2207         int i, len, add, del, adds = 0, dels = 0;
2208         uintmax_t max_change = 0, max_len = 0;
2209         int total_files = data->nr, count;
2210         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2211         const char *reset, *add_c, *del_c;
2212         int extra_shown = 0;
2213         const char *line_prefix = diff_line_prefix(options);
2214         struct strbuf out = STRBUF_INIT;
2215
2216         if (data->nr == 0)
2217                 return;
2218
2219         count = options->stat_count ? options->stat_count : data->nr;
2220
2221         reset = diff_get_color_opt(options, DIFF_RESET);
2222         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2223         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2224
2225         /*
2226          * Find the longest filename and max number of changes
2227          */
2228         for (i = 0; (i < count) && (i < data->nr); i++) {
2229                 struct diffstat_file *file = data->files[i];
2230                 uintmax_t change = file->added + file->deleted;
2231
2232                 if (!file->is_interesting && (change == 0)) {
2233                         count++; /* not shown == room for one more */
2234                         continue;
2235                 }
2236                 fill_print_name(file);
2237                 len = strlen(file->print_name);
2238                 if (max_len < len)
2239                         max_len = len;
2240
2241                 if (file->is_unmerged) {
2242                         /* "Unmerged" is 8 characters */
2243                         bin_width = bin_width < 8 ? 8 : bin_width;
2244                         continue;
2245                 }
2246                 if (file->is_binary) {
2247                         /* "Bin XXX -> YYY bytes" */
2248                         int w = 14 + decimal_width(file->added)
2249                                 + decimal_width(file->deleted);
2250                         bin_width = bin_width < w ? w : bin_width;
2251                         /* Display change counts aligned with "Bin" */
2252                         number_width = 3;
2253                         continue;
2254                 }
2255
2256                 if (max_change < change)
2257                         max_change = change;
2258         }
2259         count = i; /* where we can stop scanning in data->files[] */
2260
2261         /*
2262          * We have width = stat_width or term_columns() columns total.
2263          * We want a maximum of min(max_len, stat_name_width) for the name part.
2264          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2265          * We also need 1 for " " and 4 + decimal_width(max_change)
2266          * for " | NNNN " and one the empty column at the end, altogether
2267          * 6 + decimal_width(max_change).
2268          *
2269          * If there's not enough space, we will use the smaller of
2270          * stat_name_width (if set) and 5/8*width for the filename,
2271          * and the rest for constant elements + graph part, but no more
2272          * than stat_graph_width for the graph part.
2273          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2274          * for the standard terminal size).
2275          *
2276          * In other words: stat_width limits the maximum width, and
2277          * stat_name_width fixes the maximum width of the filename,
2278          * and is also used to divide available columns if there
2279          * aren't enough.
2280          *
2281          * Binary files are displayed with "Bin XXX -> YYY bytes"
2282          * instead of the change count and graph. This part is treated
2283          * similarly to the graph part, except that it is not
2284          * "scaled". If total width is too small to accommodate the
2285          * guaranteed minimum width of the filename part and the
2286          * separators and this message, this message will "overflow"
2287          * making the line longer than the maximum width.
2288          */
2289
2290         if (options->stat_width == -1)
2291                 width = term_columns() - strlen(line_prefix);
2292         else
2293                 width = options->stat_width ? options->stat_width : 80;
2294         number_width = decimal_width(max_change) > number_width ?
2295                 decimal_width(max_change) : number_width;
2296
2297         if (options->stat_graph_width == -1)
2298                 options->stat_graph_width = diff_stat_graph_width;
2299
2300         /*
2301          * Guarantee 3/8*16==6 for the graph part
2302          * and 5/8*16==10 for the filename part
2303          */
2304         if (width < 16 + 6 + number_width)
2305                 width = 16 + 6 + number_width;
2306
2307         /*
2308          * First assign sizes that are wanted, ignoring available width.
2309          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2310          * starting from "XXX" should fit in graph_width.
2311          */
2312         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2313         if (options->stat_graph_width &&
2314             options->stat_graph_width < graph_width)
2315                 graph_width = options->stat_graph_width;
2316
2317         name_width = (options->stat_name_width > 0 &&
2318                       options->stat_name_width < max_len) ?
2319                 options->stat_name_width : max_len;
2320
2321         /*
2322          * Adjust adjustable widths not to exceed maximum width
2323          */
2324         if (name_width + number_width + 6 + graph_width > width) {
2325                 if (graph_width > width * 3/8 - number_width - 6) {
2326                         graph_width = width * 3/8 - number_width - 6;
2327                         if (graph_width < 6)
2328                                 graph_width = 6;
2329                 }
2330
2331                 if (options->stat_graph_width &&
2332                     graph_width > options->stat_graph_width)
2333                         graph_width = options->stat_graph_width;
2334                 if (name_width > width - number_width - 6 - graph_width)
2335                         name_width = width - number_width - 6 - graph_width;
2336                 else
2337                         graph_width = width - number_width - 6 - name_width;
2338         }
2339
2340         /*
2341          * From here name_width is the width of the name area,
2342          * and graph_width is the width of the graph area.
2343          * max_change is used to scale graph properly.
2344          */
2345         for (i = 0; i < count; i++) {
2346                 const char *prefix = "";
2347                 struct diffstat_file *file = data->files[i];
2348                 char *name = file->print_name;
2349                 uintmax_t added = file->added;
2350                 uintmax_t deleted = file->deleted;
2351                 int name_len;
2352
2353                 if (!file->is_interesting && (added + deleted == 0))
2354                         continue;
2355
2356                 /*
2357                  * "scale" the filename
2358                  */
2359                 len = name_width;
2360                 name_len = strlen(name);
2361                 if (name_width < name_len) {
2362                         char *slash;
2363                         prefix = "...";
2364                         len -= 3;
2365                         name += name_len - len;
2366                         slash = strchr(name, '/');
2367                         if (slash)
2368                                 name = slash;
2369                 }
2370
2371                 if (file->is_binary) {
2372                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2373                         strbuf_addf(&out, " %*s", number_width, "Bin");
2374                         if (!added && !deleted) {
2375                                 strbuf_addch(&out, '\n');
2376                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2377                                                  out.buf, out.len, 0);
2378                                 strbuf_reset(&out);
2379                                 continue;
2380                         }
2381                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2382                                 del_c, deleted, reset);
2383                         strbuf_addstr(&out, " -> ");
2384                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2385                                 add_c, added, reset);
2386                         strbuf_addstr(&out, " bytes\n");
2387                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2388                                          out.buf, out.len, 0);
2389                         strbuf_reset(&out);
2390                         continue;
2391                 }
2392                 else if (file->is_unmerged) {
2393                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2394                         strbuf_addstr(&out, " Unmerged\n");
2395                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2396                                          out.buf, out.len, 0);
2397                         strbuf_reset(&out);
2398                         continue;
2399                 }
2400
2401                 /*
2402                  * scale the add/delete
2403                  */
2404                 add = added;
2405                 del = deleted;
2406
2407                 if (graph_width <= max_change) {
2408                         int total = scale_linear(add + del, graph_width, max_change);
2409                         if (total < 2 && add && del)
2410                                 /* width >= 2 due to the sanity check */
2411                                 total = 2;
2412                         if (add < del) {
2413                                 add = scale_linear(add, graph_width, max_change);
2414                                 del = total - add;
2415                         } else {
2416                                 del = scale_linear(del, graph_width, max_change);
2417                                 add = total - del;
2418                         }
2419                 }
2420                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2421                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2422                         number_width, added + deleted,
2423                         added + deleted ? " " : "");
2424                 show_graph(&out, '+', add, add_c, reset);
2425                 show_graph(&out, '-', del, del_c, reset);
2426                 strbuf_addch(&out, '\n');
2427                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2428                                  out.buf, out.len, 0);
2429                 strbuf_reset(&out);
2430         }
2431
2432         for (i = 0; i < data->nr; i++) {
2433                 struct diffstat_file *file = data->files[i];
2434                 uintmax_t added = file->added;
2435                 uintmax_t deleted = file->deleted;
2436
2437                 if (file->is_unmerged ||
2438                     (!file->is_interesting && (added + deleted == 0))) {
2439                         total_files--;
2440                         continue;
2441                 }
2442
2443                 if (!file->is_binary) {
2444                         adds += added;
2445                         dels += deleted;
2446                 }
2447                 if (i < count)
2448                         continue;
2449                 if (!extra_shown)
2450                         emit_diff_symbol(options,
2451                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2452                                          NULL, 0, 0);
2453                 extra_shown = 1;
2454         }
2455
2456         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2457 }
2458
2459 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2460 {
2461         int i, adds = 0, dels = 0, total_files = data->nr;
2462
2463         if (data->nr == 0)
2464                 return;
2465
2466         for (i = 0; i < data->nr; i++) {
2467                 int added = data->files[i]->added;
2468                 int deleted = data->files[i]->deleted;
2469
2470                 if (data->files[i]->is_unmerged ||
2471                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2472                         total_files--;
2473                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2474                         adds += added;
2475                         dels += deleted;
2476                 }
2477         }
2478         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2479 }
2480
2481 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2482 {
2483         int i;
2484
2485         if (data->nr == 0)
2486                 return;
2487
2488         for (i = 0; i < data->nr; i++) {
2489                 struct diffstat_file *file = data->files[i];
2490
2491                 fprintf(options->file, "%s", diff_line_prefix(options));
2492
2493                 if (file->is_binary)
2494                         fprintf(options->file, "-\t-\t");
2495                 else
2496                         fprintf(options->file,
2497                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2498                                 file->added, file->deleted);
2499                 if (options->line_termination) {
2500                         fill_print_name(file);
2501                         if (!file->is_renamed)
2502                                 write_name_quoted(file->name, options->file,
2503                                                   options->line_termination);
2504                         else {
2505                                 fputs(file->print_name, options->file);
2506                                 putc(options->line_termination, options->file);
2507                         }
2508                 } else {
2509                         if (file->is_renamed) {
2510                                 putc('\0', options->file);
2511                                 write_name_quoted(file->from_name, options->file, '\0');
2512                         }
2513                         write_name_quoted(file->name, options->file, '\0');
2514                 }
2515         }
2516 }
2517
2518 struct dirstat_file {
2519         const char *name;
2520         unsigned long changed;
2521 };
2522
2523 struct dirstat_dir {
2524         struct dirstat_file *files;
2525         int alloc, nr, permille, cumulative;
2526 };
2527
2528 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2529                 unsigned long changed, const char *base, int baselen)
2530 {
2531         unsigned long this_dir = 0;
2532         unsigned int sources = 0;
2533         const char *line_prefix = diff_line_prefix(opt);
2534
2535         while (dir->nr) {
2536                 struct dirstat_file *f = dir->files;
2537                 int namelen = strlen(f->name);
2538                 unsigned long this;
2539                 char *slash;
2540
2541                 if (namelen < baselen)
2542                         break;
2543                 if (memcmp(f->name, base, baselen))
2544                         break;
2545                 slash = strchr(f->name + baselen, '/');
2546                 if (slash) {
2547                         int newbaselen = slash + 1 - f->name;
2548                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2549                         sources++;
2550                 } else {
2551                         this = f->changed;
2552                         dir->files++;
2553                         dir->nr--;
2554                         sources += 2;
2555                 }
2556                 this_dir += this;
2557         }
2558
2559         /*
2560          * We don't report dirstat's for
2561          *  - the top level
2562          *  - or cases where everything came from a single directory
2563          *    under this directory (sources == 1).
2564          */
2565         if (baselen && sources != 1) {
2566                 if (this_dir) {
2567                         int permille = this_dir * 1000 / changed;
2568                         if (permille >= dir->permille) {
2569                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2570                                         permille / 10, permille % 10, baselen, base);
2571                                 if (!dir->cumulative)
2572                                         return 0;
2573                         }
2574                 }
2575         }
2576         return this_dir;
2577 }
2578
2579 static int dirstat_compare(const void *_a, const void *_b)
2580 {
2581         const struct dirstat_file *a = _a;
2582         const struct dirstat_file *b = _b;
2583         return strcmp(a->name, b->name);
2584 }
2585
2586 static void show_dirstat(struct diff_options *options)
2587 {
2588         int i;
2589         unsigned long changed;
2590         struct dirstat_dir dir;
2591         struct diff_queue_struct *q = &diff_queued_diff;
2592
2593         dir.files = NULL;
2594         dir.alloc = 0;
2595         dir.nr = 0;
2596         dir.permille = options->dirstat_permille;
2597         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2598
2599         changed = 0;
2600         for (i = 0; i < q->nr; i++) {
2601                 struct diff_filepair *p = q->queue[i];
2602                 const char *name;
2603                 unsigned long copied, added, damage;
2604                 int content_changed;
2605
2606                 name = p->two->path ? p->two->path : p->one->path;
2607
2608                 if (p->one->oid_valid && p->two->oid_valid)
2609                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2610                 else
2611                         content_changed = 1;
2612
2613                 if (!content_changed) {
2614                         /*
2615                          * The SHA1 has not changed, so pre-/post-content is
2616                          * identical. We can therefore skip looking at the
2617                          * file contents altogether.
2618                          */
2619                         damage = 0;
2620                         goto found_damage;
2621                 }
2622
2623                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2624                         /*
2625                          * In --dirstat-by-file mode, we don't really need to
2626                          * look at the actual file contents at all.
2627                          * The fact that the SHA1 changed is enough for us to
2628                          * add this file to the list of results
2629                          * (with each file contributing equal damage).
2630                          */
2631                         damage = 1;
2632                         goto found_damage;
2633                 }
2634
2635                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2636                         diff_populate_filespec(p->one, 0);
2637                         diff_populate_filespec(p->two, 0);
2638                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2639                                                &copied, &added);
2640                         diff_free_filespec_data(p->one);
2641                         diff_free_filespec_data(p->two);
2642                 } else if (DIFF_FILE_VALID(p->one)) {
2643                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2644                         copied = added = 0;
2645                         diff_free_filespec_data(p->one);
2646                 } else if (DIFF_FILE_VALID(p->two)) {
2647                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2648                         copied = 0;
2649                         added = p->two->size;
2650                         diff_free_filespec_data(p->two);
2651                 } else
2652                         continue;
2653
2654                 /*
2655                  * Original minus copied is the removed material,
2656                  * added is the new material.  They are both damages
2657                  * made to the preimage.
2658                  * If the resulting damage is zero, we know that
2659                  * diffcore_count_changes() considers the two entries to
2660                  * be identical, but since content_changed is true, we
2661                  * know that there must have been _some_ kind of change,
2662                  * so we force all entries to have damage > 0.
2663                  */
2664                 damage = (p->one->size - copied) + added;
2665                 if (!damage)
2666                         damage = 1;
2667
2668 found_damage:
2669                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2670                 dir.files[dir.nr].name = name;
2671                 dir.files[dir.nr].changed = damage;
2672                 changed += damage;
2673                 dir.nr++;
2674         }
2675
2676         /* This can happen even with many files, if everything was renames */
2677         if (!changed)
2678                 return;
2679
2680         /* Show all directories with more than x% of the changes */
2681         QSORT(dir.files, dir.nr, dirstat_compare);
2682         gather_dirstat(options, &dir, changed, "", 0);
2683 }
2684
2685 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2686 {
2687         int i;
2688         unsigned long changed;
2689         struct dirstat_dir dir;
2690
2691         if (data->nr == 0)
2692                 return;
2693
2694         dir.files = NULL;
2695         dir.alloc = 0;
2696         dir.nr = 0;
2697         dir.permille = options->dirstat_permille;
2698         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2699
2700         changed = 0;
2701         for (i = 0; i < data->nr; i++) {
2702                 struct diffstat_file *file = data->files[i];
2703                 unsigned long damage = file->added + file->deleted;
2704                 if (file->is_binary)
2705                         /*
2706                          * binary files counts bytes, not lines. Must find some
2707                          * way to normalize binary bytes vs. textual lines.
2708                          * The following heuristic assumes that there are 64
2709                          * bytes per "line".
2710                          * This is stupid and ugly, but very cheap...
2711                          */
2712                         damage = (damage + 63) / 64;
2713                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2714                 dir.files[dir.nr].name = file->name;
2715                 dir.files[dir.nr].changed = damage;
2716                 changed += damage;
2717                 dir.nr++;
2718         }
2719
2720         /* This can happen even with many files, if everything was renames */
2721         if (!changed)
2722                 return;
2723
2724         /* Show all directories with more than x% of the changes */
2725         QSORT(dir.files, dir.nr, dirstat_compare);
2726         gather_dirstat(options, &dir, changed, "", 0);
2727 }
2728
2729 static void free_diffstat_info(struct diffstat_t *diffstat)
2730 {
2731         int i;
2732         for (i = 0; i < diffstat->nr; i++) {
2733                 struct diffstat_file *f = diffstat->files[i];
2734                 if (f->name != f->print_name)
2735                         free(f->print_name);
2736                 free(f->name);
2737                 free(f->from_name);
2738                 free(f);
2739         }
2740         free(diffstat->files);
2741 }
2742
2743 struct checkdiff_t {
2744         const char *filename;
2745         int lineno;
2746         int conflict_marker_size;
2747         struct diff_options *o;
2748         unsigned ws_rule;
2749         unsigned status;
2750 };
2751
2752 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2753 {
2754         char firstchar;
2755         int cnt;
2756
2757         if (len < marker_size + 1)
2758                 return 0;
2759         firstchar = line[0];
2760         switch (firstchar) {
2761         case '=': case '>': case '<': case '|':
2762                 break;
2763         default:
2764                 return 0;
2765         }
2766         for (cnt = 1; cnt < marker_size; cnt++)
2767                 if (line[cnt] != firstchar)
2768                         return 0;
2769         /* line[1] thru line[marker_size-1] are same as firstchar */
2770         if (len < marker_size + 1 || !isspace(line[marker_size]))
2771                 return 0;
2772         return 1;
2773 }
2774
2775 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2776 {
2777         struct checkdiff_t *data = priv;
2778         int marker_size = data->conflict_marker_size;
2779         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2780         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2781         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2782         char *err;
2783         const char *line_prefix;
2784
2785         assert(data->o);
2786         line_prefix = diff_line_prefix(data->o);
2787
2788         if (line[0] == '+') {
2789                 unsigned bad;
2790                 data->lineno++;
2791                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2792                         data->status |= 1;
2793                         fprintf(data->o->file,
2794                                 "%s%s:%d: leftover conflict marker\n",
2795                                 line_prefix, data->filename, data->lineno);
2796                 }
2797                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2798                 if (!bad)
2799                         return;
2800                 data->status |= bad;
2801                 err = whitespace_error_string(bad);
2802                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2803                         line_prefix, data->filename, data->lineno, err);
2804                 free(err);
2805                 emit_line(data->o, set, reset, line, 1);
2806                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2807                               data->o->file, set, reset, ws);
2808         } else if (line[0] == ' ') {
2809                 data->lineno++;
2810         } else if (line[0] == '@') {
2811                 char *plus = strchr(line, '+');
2812                 if (plus)
2813                         data->lineno = strtol(plus, NULL, 10) - 1;
2814                 else
2815                         die("invalid diff");
2816         }
2817 }
2818
2819 static unsigned char *deflate_it(char *data,
2820                                  unsigned long size,
2821                                  unsigned long *result_size)
2822 {
2823         int bound;
2824         unsigned char *deflated;
2825         git_zstream stream;
2826
2827         git_deflate_init(&stream, zlib_compression_level);
2828         bound = git_deflate_bound(&stream, size);
2829         deflated = xmalloc(bound);
2830         stream.next_out = deflated;
2831         stream.avail_out = bound;
2832
2833         stream.next_in = (unsigned char *)data;
2834         stream.avail_in = size;
2835         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2836                 ; /* nothing */
2837         git_deflate_end(&stream);
2838         *result_size = stream.total_out;
2839         return deflated;
2840 }
2841
2842 static void emit_binary_diff_body(struct diff_options *o,
2843                                   mmfile_t *one, mmfile_t *two)
2844 {
2845         void *cp;
2846         void *delta;
2847         void *deflated;
2848         void *data;
2849         unsigned long orig_size;
2850         unsigned long delta_size;
2851         unsigned long deflate_size;
2852         unsigned long data_size;
2853
2854         /* We could do deflated delta, or we could do just deflated two,
2855          * whichever is smaller.
2856          */
2857         delta = NULL;
2858         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2859         if (one->size && two->size) {
2860                 delta = diff_delta(one->ptr, one->size,
2861                                    two->ptr, two->size,
2862                                    &delta_size, deflate_size);
2863                 if (delta) {
2864                         void *to_free = delta;
2865                         orig_size = delta_size;
2866                         delta = deflate_it(delta, delta_size, &delta_size);
2867                         free(to_free);
2868                 }
2869         }
2870
2871         if (delta && delta_size < deflate_size) {
2872                 char *s = xstrfmt("%lu", orig_size);
2873                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2874                                  s, strlen(s), 0);
2875                 free(s);
2876                 free(deflated);
2877                 data = delta;
2878                 data_size = delta_size;
2879         } else {
2880                 char *s = xstrfmt("%lu", two->size);
2881                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2882                                  s, strlen(s), 0);
2883                 free(s);
2884                 free(delta);
2885                 data = deflated;
2886                 data_size = deflate_size;
2887         }
2888
2889         /* emit data encoded in base85 */
2890         cp = data;
2891         while (data_size) {
2892                 int len;
2893                 int bytes = (52 < data_size) ? 52 : data_size;
2894                 char line[71];
2895                 data_size -= bytes;
2896                 if (bytes <= 26)
2897                         line[0] = bytes + 'A' - 1;
2898                 else
2899                         line[0] = bytes - 26 + 'a' - 1;
2900                 encode_85(line + 1, cp, bytes);
2901                 cp = (char *) cp + bytes;
2902
2903                 len = strlen(line);
2904                 line[len++] = '\n';
2905                 line[len] = '\0';
2906
2907                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2908                                  line, len, 0);
2909         }
2910         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2911         free(data);
2912 }
2913
2914 static void emit_binary_diff(struct diff_options *o,
2915                              mmfile_t *one, mmfile_t *two)
2916 {
2917         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2918         emit_binary_diff_body(o, one, two);
2919         emit_binary_diff_body(o, two, one);
2920 }
2921
2922 int diff_filespec_is_binary(struct diff_filespec *one)
2923 {
2924         if (one->is_binary == -1) {
2925                 diff_filespec_load_driver(one);
2926                 if (one->driver->binary != -1)
2927                         one->is_binary = one->driver->binary;
2928                 else {
2929                         if (!one->data && DIFF_FILE_VALID(one))
2930                                 diff_populate_filespec(one, CHECK_BINARY);
2931                         if (one->is_binary == -1 && one->data)
2932                                 one->is_binary = buffer_is_binary(one->data,
2933                                                 one->size);
2934                         if (one->is_binary == -1)
2935                                 one->is_binary = 0;
2936                 }
2937         }
2938         return one->is_binary;
2939 }
2940
2941 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2942 {
2943         diff_filespec_load_driver(one);
2944         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2945 }
2946
2947 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2948 {
2949         if (!options->a_prefix)
2950                 options->a_prefix = a;
2951         if (!options->b_prefix)
2952                 options->b_prefix = b;
2953 }
2954
2955 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2956 {
2957         if (!DIFF_FILE_VALID(one))
2958                 return NULL;
2959
2960         diff_filespec_load_driver(one);
2961         return userdiff_get_textconv(one->driver);
2962 }
2963
2964 static void builtin_diff(const char *name_a,
2965                          const char *name_b,
2966                          struct diff_filespec *one,
2967                          struct diff_filespec *two,
2968                          const char *xfrm_msg,
2969                          int must_show_header,
2970                          struct diff_options *o,
2971                          int complete_rewrite)
2972 {
2973         mmfile_t mf1, mf2;
2974         const char *lbl[2];
2975         char *a_one, *b_two;
2976         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2977         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2978         const char *a_prefix, *b_prefix;
2979         struct userdiff_driver *textconv_one = NULL;
2980         struct userdiff_driver *textconv_two = NULL;
2981         struct strbuf header = STRBUF_INIT;
2982         const char *line_prefix = diff_line_prefix(o);
2983
2984         diff_set_mnemonic_prefix(o, "a/", "b/");
2985         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2986                 a_prefix = o->b_prefix;
2987                 b_prefix = o->a_prefix;
2988         } else {
2989                 a_prefix = o->a_prefix;
2990                 b_prefix = o->b_prefix;
2991         }
2992
2993         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2994             (!one->mode || S_ISGITLINK(one->mode)) &&
2995             (!two->mode || S_ISGITLINK(two->mode))) {
2996                 show_submodule_summary(o, one->path ? one->path : two->path,
2997                                 &one->oid, &two->oid,
2998                                 two->dirty_submodule);
2999                 return;
3000         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3001                    (!one->mode || S_ISGITLINK(one->mode)) &&
3002                    (!two->mode || S_ISGITLINK(two->mode))) {
3003                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3004                                 &one->oid, &two->oid,
3005                                 two->dirty_submodule);
3006                 return;
3007         }
3008
3009         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
3010                 textconv_one = get_textconv(one);
3011                 textconv_two = get_textconv(two);
3012         }
3013
3014         /* Never use a non-valid filename anywhere if at all possible */
3015         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3016         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3017
3018         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3019         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3020         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3021         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3022         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3023         if (lbl[0][0] == '/') {
3024                 /* /dev/null */
3025                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3026                 if (xfrm_msg)
3027                         strbuf_addstr(&header, xfrm_msg);
3028                 must_show_header = 1;
3029         }
3030         else if (lbl[1][0] == '/') {
3031                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3032                 if (xfrm_msg)
3033                         strbuf_addstr(&header, xfrm_msg);
3034                 must_show_header = 1;
3035         }
3036         else {
3037                 if (one->mode != two->mode) {
3038                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3039                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3040                         must_show_header = 1;
3041                 }
3042                 if (xfrm_msg)
3043                         strbuf_addstr(&header, xfrm_msg);
3044
3045                 /*
3046                  * we do not run diff between different kind
3047                  * of objects.
3048                  */
3049                 if ((one->mode ^ two->mode) & S_IFMT)
3050                         goto free_ab_and_return;
3051                 if (complete_rewrite &&
3052                     (textconv_one || !diff_filespec_is_binary(one)) &&
3053                     (textconv_two || !diff_filespec_is_binary(two))) {
3054                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3055                                          header.buf, header.len, 0);
3056                         strbuf_reset(&header);
3057                         emit_rewrite_diff(name_a, name_b, one, two,
3058                                                 textconv_one, textconv_two, o);
3059                         o->found_changes = 1;
3060                         goto free_ab_and_return;
3061                 }
3062         }
3063
3064         if (o->irreversible_delete && lbl[1][0] == '/') {
3065                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3066                                  header.len, 0);
3067                 strbuf_reset(&header);
3068                 goto free_ab_and_return;
3069         } else if (!DIFF_OPT_TST(o, TEXT) &&
3070             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3071               (!textconv_two && diff_filespec_is_binary(two)) )) {
3072                 struct strbuf sb = STRBUF_INIT;
3073                 if (!one->data && !two->data &&
3074                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3075                     !DIFF_OPT_TST(o, BINARY)) {
3076                         if (!oidcmp(&one->oid, &two->oid)) {
3077                                 if (must_show_header)
3078                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3079                                                          header.buf, header.len,
3080                                                          0);
3081                                 goto free_ab_and_return;
3082                         }
3083                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3084                                          header.buf, header.len, 0);
3085                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3086                                     diff_line_prefix(o), lbl[0], lbl[1]);
3087                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3088                                          sb.buf, sb.len, 0);
3089                         strbuf_release(&sb);
3090                         goto free_ab_and_return;
3091                 }
3092                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3093                         die("unable to read files to diff");
3094                 /* Quite common confusing case */
3095                 if (mf1.size == mf2.size &&
3096                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3097                         if (must_show_header)
3098                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3099                                                  header.buf, header.len, 0);
3100                         goto free_ab_and_return;
3101                 }
3102                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3103                 strbuf_reset(&header);
3104                 if (DIFF_OPT_TST(o, BINARY))
3105                         emit_binary_diff(o, &mf1, &mf2);
3106                 else {
3107                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3108                                     diff_line_prefix(o), lbl[0], lbl[1]);
3109                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3110                                          sb.buf, sb.len, 0);
3111                         strbuf_release(&sb);
3112                 }
3113                 o->found_changes = 1;
3114         } else {
3115                 /* Crazy xdl interfaces.. */
3116                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3117                 const char *v;
3118                 xpparam_t xpp;
3119                 xdemitconf_t xecfg;
3120                 struct emit_callback ecbdata;
3121                 const struct userdiff_funcname *pe;
3122
3123                 if (must_show_header) {
3124                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3125                                          header.buf, header.len, 0);
3126                         strbuf_reset(&header);
3127                 }
3128
3129                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3130                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3131
3132                 pe = diff_funcname_pattern(one);
3133                 if (!pe)
3134                         pe = diff_funcname_pattern(two);
3135
3136                 memset(&xpp, 0, sizeof(xpp));
3137                 memset(&xecfg, 0, sizeof(xecfg));
3138                 memset(&ecbdata, 0, sizeof(ecbdata));
3139                 ecbdata.label_path = lbl;
3140                 ecbdata.color_diff = want_color(o->use_color);
3141                 ecbdata.ws_rule = whitespace_rule(name_b);
3142                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3143                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3144                 ecbdata.opt = o;
3145                 ecbdata.header = header.len ? &header : NULL;
3146                 xpp.flags = o->xdl_opts;
3147                 xecfg.ctxlen = o->context;
3148                 xecfg.interhunkctxlen = o->interhunkcontext;
3149                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3150                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
3151                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3152                 if (pe)
3153                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3154                 if (!diffopts)
3155                         ;
3156                 else if (skip_prefix(diffopts, "--unified=", &v))
3157                         xecfg.ctxlen = strtoul(v, NULL, 10);
3158                 else if (skip_prefix(diffopts, "-u", &v))
3159                         xecfg.ctxlen = strtoul(v, NULL, 10);
3160                 if (o->word_diff)
3161                         init_diff_words_data(&ecbdata, o, one, two);
3162                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3163                                   &xpp, &xecfg))
3164                         die("unable to generate diff for %s", one->path);
3165                 if (o->word_diff)
3166                         free_diff_words_data(&ecbdata);
3167                 if (textconv_one)
3168                         free(mf1.ptr);
3169                 if (textconv_two)
3170                         free(mf2.ptr);
3171                 xdiff_clear_find_func(&xecfg);
3172         }
3173
3174  free_ab_and_return:
3175         strbuf_release(&header);
3176         diff_free_filespec_data(one);
3177         diff_free_filespec_data(two);
3178         free(a_one);
3179         free(b_two);
3180         return;
3181 }
3182
3183 static void builtin_diffstat(const char *name_a, const char *name_b,
3184                              struct diff_filespec *one,
3185                              struct diff_filespec *two,
3186                              struct diffstat_t *diffstat,
3187                              struct diff_options *o,
3188                              struct diff_filepair *p)
3189 {
3190         mmfile_t mf1, mf2;
3191         struct diffstat_file *data;
3192         int same_contents;
3193         int complete_rewrite = 0;
3194
3195         if (!DIFF_PAIR_UNMERGED(p)) {
3196                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3197                         complete_rewrite = 1;
3198         }
3199
3200         data = diffstat_add(diffstat, name_a, name_b);
3201         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3202
3203         if (!one || !two) {
3204                 data->is_unmerged = 1;
3205                 return;
3206         }
3207
3208         same_contents = !oidcmp(&one->oid, &two->oid);
3209
3210         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3211                 data->is_binary = 1;
3212                 if (same_contents) {
3213                         data->added = 0;
3214                         data->deleted = 0;
3215                 } else {
3216                         data->added = diff_filespec_size(two);
3217                         data->deleted = diff_filespec_size(one);
3218                 }
3219         }
3220
3221         else if (complete_rewrite) {
3222                 diff_populate_filespec(one, 0);
3223                 diff_populate_filespec(two, 0);
3224                 data->deleted = count_lines(one->data, one->size);
3225                 data->added = count_lines(two->data, two->size);
3226         }
3227
3228         else if (!same_contents) {
3229                 /* Crazy xdl interfaces.. */
3230                 xpparam_t xpp;
3231                 xdemitconf_t xecfg;
3232
3233                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3234                         die("unable to read files to diff");
3235
3236                 memset(&xpp, 0, sizeof(xpp));
3237                 memset(&xecfg, 0, sizeof(xecfg));
3238                 xpp.flags = o->xdl_opts;
3239                 xecfg.ctxlen = o->context;
3240                 xecfg.interhunkctxlen = o->interhunkcontext;
3241                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3242                                   &xpp, &xecfg))
3243                         die("unable to generate diffstat for %s", one->path);
3244         }
3245
3246         diff_free_filespec_data(one);
3247         diff_free_filespec_data(two);
3248 }
3249
3250 static void builtin_checkdiff(const char *name_a, const char *name_b,
3251                               const char *attr_path,
3252                               struct diff_filespec *one,
3253                               struct diff_filespec *two,
3254                               struct diff_options *o)
3255 {
3256         mmfile_t mf1, mf2;
3257         struct checkdiff_t data;
3258
3259         if (!two)
3260                 return;
3261
3262         memset(&data, 0, sizeof(data));
3263         data.filename = name_b ? name_b : name_a;
3264         data.lineno = 0;
3265         data.o = o;
3266         data.ws_rule = whitespace_rule(attr_path);
3267         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3268
3269         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3270                 die("unable to read files to diff");
3271
3272         /*
3273          * All the other codepaths check both sides, but not checking
3274          * the "old" side here is deliberate.  We are checking the newly
3275          * introduced changes, and as long as the "new" side is text, we
3276          * can and should check what it introduces.
3277          */
3278         if (diff_filespec_is_binary(two))
3279                 goto free_and_return;
3280         else {
3281                 /* Crazy xdl interfaces.. */
3282                 xpparam_t xpp;
3283                 xdemitconf_t xecfg;
3284
3285                 memset(&xpp, 0, sizeof(xpp));
3286                 memset(&xecfg, 0, sizeof(xecfg));
3287                 xecfg.ctxlen = 1; /* at least one context line */
3288                 xpp.flags = 0;
3289                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3290                                   &xpp, &xecfg))
3291                         die("unable to generate checkdiff for %s", one->path);
3292
3293                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3294                         struct emit_callback ecbdata;
3295                         int blank_at_eof;
3296
3297                         ecbdata.ws_rule = data.ws_rule;
3298                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3299                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3300
3301                         if (blank_at_eof) {
3302                                 static char *err;
3303                                 if (!err)
3304                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3305                                 fprintf(o->file, "%s:%d: %s.\n",
3306                                         data.filename, blank_at_eof, err);
3307                                 data.status = 1; /* report errors */
3308                         }
3309                 }
3310         }
3311  free_and_return:
3312         diff_free_filespec_data(one);
3313         diff_free_filespec_data(two);
3314         if (data.status)
3315                 DIFF_OPT_SET(o, CHECK_FAILED);
3316 }
3317
3318 struct diff_filespec *alloc_filespec(const char *path)
3319 {
3320         struct diff_filespec *spec;
3321
3322         FLEXPTR_ALLOC_STR(spec, path, path);
3323         spec->count = 1;
3324         spec->is_binary = -1;
3325         return spec;
3326 }
3327
3328 void free_filespec(struct diff_filespec *spec)
3329 {
3330         if (!--spec->count) {
3331                 diff_free_filespec_data(spec);
3332                 free(spec);
3333         }
3334 }
3335
3336 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3337                    int oid_valid, unsigned short mode)
3338 {
3339         if (mode) {
3340                 spec->mode = canon_mode(mode);
3341                 oidcpy(&spec->oid, oid);
3342                 spec->oid_valid = oid_valid;
3343         }
3344 }
3345
3346 /*
3347  * Given a name and sha1 pair, if the index tells us the file in
3348  * the work tree has that object contents, return true, so that
3349  * prepare_temp_file() does not have to inflate and extract.
3350  */
3351 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3352 {
3353         const struct cache_entry *ce;
3354         struct stat st;
3355         int pos, len;
3356
3357         /*
3358          * We do not read the cache ourselves here, because the
3359          * benchmark with my previous version that always reads cache
3360          * shows that it makes things worse for diff-tree comparing
3361          * two linux-2.6 kernel trees in an already checked out work
3362          * tree.  This is because most diff-tree comparisons deal with
3363          * only a small number of files, while reading the cache is
3364          * expensive for a large project, and its cost outweighs the
3365          * savings we get by not inflating the object to a temporary
3366          * file.  Practically, this code only helps when we are used
3367          * by diff-cache --cached, which does read the cache before
3368          * calling us.
3369          */
3370         if (!active_cache)
3371                 return 0;
3372
3373         /* We want to avoid the working directory if our caller
3374          * doesn't need the data in a normal file, this system
3375          * is rather slow with its stat/open/mmap/close syscalls,
3376          * and the object is contained in a pack file.  The pack
3377          * is probably already open and will be faster to obtain
3378          * the data through than the working directory.  Loose
3379          * objects however would tend to be slower as they need
3380          * to be individually opened and inflated.
3381          */
3382         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3383                 return 0;
3384
3385         /*
3386          * Similarly, if we'd have to convert the file contents anyway, that
3387          * makes the optimization not worthwhile.
3388          */
3389         if (!want_file && would_convert_to_git(&the_index, name))
3390                 return 0;
3391
3392         len = strlen(name);
3393         pos = cache_name_pos(name, len);
3394         if (pos < 0)
3395                 return 0;
3396         ce = active_cache[pos];
3397
3398         /*
3399          * This is not the sha1 we are looking for, or
3400          * unreusable because it is not a regular file.
3401          */
3402         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3403                 return 0;
3404
3405         /*
3406          * If ce is marked as "assume unchanged", there is no
3407          * guarantee that work tree matches what we are looking for.
3408          */
3409         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3410                 return 0;
3411
3412         /*
3413          * If ce matches the file in the work tree, we can reuse it.
3414          */
3415         if (ce_uptodate(ce) ||
3416             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3417                 return 1;
3418
3419         return 0;
3420 }
3421
3422 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3423 {
3424         struct strbuf buf = STRBUF_INIT;
3425         char *dirty = "";
3426
3427         /* Are we looking at the work tree? */
3428         if (s->dirty_submodule)
3429                 dirty = "-dirty";
3430
3431         strbuf_addf(&buf, "Subproject commit %s%s\n",
3432                     oid_to_hex(&s->oid), dirty);
3433         s->size = buf.len;
3434         if (size_only) {
3435                 s->data = NULL;
3436                 strbuf_release(&buf);
3437         } else {
3438                 s->data = strbuf_detach(&buf, NULL);
3439                 s->should_free = 1;
3440         }
3441         return 0;
3442 }
3443
3444 /*
3445  * While doing rename detection and pickaxe operation, we may need to
3446  * grab the data for the blob (or file) for our own in-core comparison.
3447  * diff_filespec has data and size fields for this purpose.
3448  */
3449 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3450 {
3451         int size_only = flags & CHECK_SIZE_ONLY;
3452         int err = 0;
3453         /*
3454          * demote FAIL to WARN to allow inspecting the situation
3455          * instead of refusing.
3456          */
3457         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3458                                     ? SAFE_CRLF_WARN
3459                                     : safe_crlf);
3460
3461         if (!DIFF_FILE_VALID(s))
3462                 die("internal error: asking to populate invalid file.");
3463         if (S_ISDIR(s->mode))
3464                 return -1;
3465
3466         if (s->data)
3467                 return 0;
3468
3469         if (size_only && 0 < s->size)
3470                 return 0;
3471
3472         if (S_ISGITLINK(s->mode))
3473                 return diff_populate_gitlink(s, size_only);
3474
3475         if (!s->oid_valid ||
3476             reuse_worktree_file(s->path, &s->oid, 0)) {
3477                 struct strbuf buf = STRBUF_INIT;
3478                 struct stat st;
3479                 int fd;
3480
3481                 if (lstat(s->path, &st) < 0) {
3482                         if (errno == ENOENT) {
3483                         err_empty:
3484                                 err = -1;
3485                         empty:
3486                                 s->data = (char *)"";
3487                                 s->size = 0;
3488                                 return err;
3489                         }
3490                 }
3491                 s->size = xsize_t(st.st_size);
3492                 if (!s->size)
3493                         goto empty;
3494                 if (S_ISLNK(st.st_mode)) {
3495                         struct strbuf sb = STRBUF_INIT;
3496
3497                         if (strbuf_readlink(&sb, s->path, s->size))
3498                                 goto err_empty;
3499                         s->size = sb.len;
3500                         s->data = strbuf_detach(&sb, NULL);
3501                         s->should_free = 1;
3502                         return 0;
3503                 }
3504
3505                 /*
3506                  * Even if the caller would be happy with getting
3507                  * only the size, we cannot return early at this
3508                  * point if the path requires us to run the content
3509                  * conversion.
3510                  */
3511                 if (size_only && !would_convert_to_git(&the_index, s->path))
3512                         return 0;
3513
3514                 /*
3515                  * Note: this check uses xsize_t(st.st_size) that may
3516                  * not be the true size of the blob after it goes
3517                  * through convert_to_git().  This may not strictly be
3518                  * correct, but the whole point of big_file_threshold
3519                  * and is_binary check being that we want to avoid
3520                  * opening the file and inspecting the contents, this
3521                  * is probably fine.
3522                  */
3523                 if ((flags & CHECK_BINARY) &&
3524                     s->size > big_file_threshold && s->is_binary == -1) {
3525                         s->is_binary = 1;
3526                         return 0;
3527                 }
3528                 fd = open(s->path, O_RDONLY);
3529                 if (fd < 0)
3530                         goto err_empty;
3531                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3532                 close(fd);
3533                 s->should_munmap = 1;
3534
3535                 /*
3536                  * Convert from working tree format to canonical git format
3537                  */
3538                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3539                         size_t size = 0;
3540                         munmap(s->data, s->size);
3541                         s->should_munmap = 0;
3542                         s->data = strbuf_detach(&buf, &size);
3543                         s->size = size;
3544                         s->should_free = 1;
3545                 }
3546         }
3547         else {
3548                 enum object_type type;
3549                 if (size_only || (flags & CHECK_BINARY)) {
3550                         type = sha1_object_info(s->oid.hash, &s->size);
3551                         if (type < 0)
3552                                 die("unable to read %s",
3553                                     oid_to_hex(&s->oid));
3554                         if (size_only)
3555                                 return 0;
3556                         if (s->size > big_file_threshold && s->is_binary == -1) {
3557                                 s->is_binary = 1;
3558                                 return 0;
3559                         }
3560                 }
3561                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3562                 if (!s->data)
3563                         die("unable to read %s", oid_to_hex(&s->oid));
3564                 s->should_free = 1;
3565         }
3566         return 0;
3567 }
3568
3569 void diff_free_filespec_blob(struct diff_filespec *s)
3570 {
3571         if (s->should_free)
3572                 free(s->data);
3573         else if (s->should_munmap)
3574                 munmap(s->data, s->size);
3575
3576         if (s->should_free || s->should_munmap) {
3577                 s->should_free = s->should_munmap = 0;
3578                 s->data = NULL;
3579         }
3580 }
3581
3582 void diff_free_filespec_data(struct diff_filespec *s)
3583 {
3584         diff_free_filespec_blob(s);
3585         FREE_AND_NULL(s->cnt_data);
3586 }
3587
3588 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3589                            void *blob,
3590                            unsigned long size,
3591                            const struct object_id *oid,
3592                            int mode)
3593 {
3594         int fd;
3595         struct strbuf buf = STRBUF_INIT;
3596         struct strbuf template = STRBUF_INIT;
3597         char *path_dup = xstrdup(path);
3598         const char *base = basename(path_dup);
3599
3600         /* Generate "XXXXXX_basename.ext" */
3601         strbuf_addstr(&template, "XXXXXX_");
3602         strbuf_addstr(&template, base);
3603
3604         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
3605         if (fd < 0)
3606                 die_errno("unable to create temp-file");
3607         if (convert_to_working_tree(path,
3608                         (const char *)blob, (size_t)size, &buf)) {
3609                 blob = buf.buf;
3610                 size = buf.len;
3611         }
3612         if (write_in_full(fd, blob, size) != size)
3613                 die_errno("unable to write temp-file");
3614         close_tempfile(&temp->tempfile);
3615         temp->name = get_tempfile_path(&temp->tempfile);
3616         oid_to_hex_r(temp->hex, oid);
3617         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3618         strbuf_release(&buf);
3619         strbuf_release(&template);
3620         free(path_dup);
3621 }
3622
3623 static struct diff_tempfile *prepare_temp_file(const char *name,
3624                 struct diff_filespec *one)
3625 {
3626         struct diff_tempfile *temp = claim_diff_tempfile();
3627
3628         if (!DIFF_FILE_VALID(one)) {
3629         not_a_valid_file:
3630                 /* A '-' entry produces this for file-2, and
3631                  * a '+' entry produces this for file-1.
3632                  */
3633                 temp->name = "/dev/null";
3634                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3635                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3636                 return temp;
3637         }
3638
3639         if (!S_ISGITLINK(one->mode) &&
3640             (!one->oid_valid ||
3641              reuse_worktree_file(name, &one->oid, 1))) {
3642                 struct stat st;
3643                 if (lstat(name, &st) < 0) {
3644                         if (errno == ENOENT)
3645                                 goto not_a_valid_file;
3646                         die_errno("stat(%s)", name);
3647                 }
3648                 if (S_ISLNK(st.st_mode)) {
3649                         struct strbuf sb = STRBUF_INIT;
3650                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3651                                 die_errno("readlink(%s)", name);
3652                         prep_temp_blob(name, temp, sb.buf, sb.len,
3653                                        (one->oid_valid ?
3654                                         &one->oid : &null_oid),
3655                                        (one->oid_valid ?
3656                                         one->mode : S_IFLNK));
3657                         strbuf_release(&sb);
3658                 }
3659                 else {
3660                         /* we can borrow from the file in the work tree */
3661                         temp->name = name;
3662                         if (!one->oid_valid)
3663                                 oid_to_hex_r(temp->hex, &null_oid);
3664                         else
3665                                 oid_to_hex_r(temp->hex, &one->oid);
3666                         /* Even though we may sometimes borrow the
3667                          * contents from the work tree, we always want
3668                          * one->mode.  mode is trustworthy even when
3669                          * !(one->oid_valid), as long as
3670                          * DIFF_FILE_VALID(one).
3671                          */
3672                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3673                 }
3674                 return temp;
3675         }
3676         else {
3677                 if (diff_populate_filespec(one, 0))
3678                         die("cannot read data blob for %s", one->path);
3679                 prep_temp_blob(name, temp, one->data, one->size,
3680                                &one->oid, one->mode);
3681         }
3682         return temp;
3683 }
3684
3685 static void add_external_diff_name(struct argv_array *argv,
3686                                    const char *name,
3687                                    struct diff_filespec *df)
3688 {
3689         struct diff_tempfile *temp = prepare_temp_file(name, df);
3690         argv_array_push(argv, temp->name);
3691         argv_array_push(argv, temp->hex);
3692         argv_array_push(argv, temp->mode);
3693 }
3694
3695 /* An external diff command takes:
3696  *
3697  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3698  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3699  *
3700  */
3701 static void run_external_diff(const char *pgm,
3702                               const char *name,
3703                               const char *other,
3704                               struct diff_filespec *one,
3705                               struct diff_filespec *two,
3706                               const char *xfrm_msg,
3707                               int complete_rewrite,
3708                               struct diff_options *o)
3709 {
3710         struct argv_array argv = ARGV_ARRAY_INIT;
3711         struct argv_array env = ARGV_ARRAY_INIT;
3712         struct diff_queue_struct *q = &diff_queued_diff;
3713
3714         argv_array_push(&argv, pgm);
3715         argv_array_push(&argv, name);
3716
3717         if (one && two) {
3718                 add_external_diff_name(&argv, name, one);
3719                 if (!other)
3720                         add_external_diff_name(&argv, name, two);
3721                 else {
3722                         add_external_diff_name(&argv, other, two);
3723                         argv_array_push(&argv, other);
3724                         argv_array_push(&argv, xfrm_msg);
3725                 }
3726         }
3727
3728         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3729         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3730
3731         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3732                 die(_("external diff died, stopping at %s"), name);
3733
3734         remove_tempfile();
3735         argv_array_clear(&argv);
3736         argv_array_clear(&env);
3737 }
3738
3739 static int similarity_index(struct diff_filepair *p)
3740 {
3741         return p->score * 100 / MAX_SCORE;
3742 }
3743
3744 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3745 {
3746         if (startup_info->have_repository)
3747                 return find_unique_abbrev(oid->hash, abbrev);
3748         else {
3749                 char *hex = oid_to_hex(oid);
3750                 if (abbrev < 0)
3751                         abbrev = FALLBACK_DEFAULT_ABBREV;
3752                 if (abbrev > GIT_SHA1_HEXSZ)
3753                         die("BUG: oid abbreviation out of range: %d", abbrev);
3754                 if (abbrev)
3755                         hex[abbrev] = '\0';
3756                 return hex;
3757         }
3758 }
3759
3760 static void fill_metainfo(struct strbuf *msg,
3761                           const char *name,
3762                           const char *other,
3763                           struct diff_filespec *one,
3764                           struct diff_filespec *two,
3765                           struct diff_options *o,
3766                           struct diff_filepair *p,
3767                           int *must_show_header,
3768                           int use_color)
3769 {
3770         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3771         const char *reset = diff_get_color(use_color, DIFF_RESET);
3772         const char *line_prefix = diff_line_prefix(o);
3773
3774         *must_show_header = 1;
3775         strbuf_init(msg, PATH_MAX * 2 + 300);
3776         switch (p->status) {
3777         case DIFF_STATUS_COPIED:
3778                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3779                             line_prefix, set, similarity_index(p));
3780                 strbuf_addf(msg, "%s\n%s%scopy from ",
3781                             reset,  line_prefix, set);
3782                 quote_c_style(name, msg, NULL, 0);
3783                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3784                 quote_c_style(other, msg, NULL, 0);
3785                 strbuf_addf(msg, "%s\n", reset);
3786                 break;
3787         case DIFF_STATUS_RENAMED:
3788                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3789                             line_prefix, set, similarity_index(p));
3790                 strbuf_addf(msg, "%s\n%s%srename from ",
3791                             reset, line_prefix, set);
3792                 quote_c_style(name, msg, NULL, 0);
3793                 strbuf_addf(msg, "%s\n%s%srename to ",
3794                             reset, line_prefix, set);
3795                 quote_c_style(other, msg, NULL, 0);
3796                 strbuf_addf(msg, "%s\n", reset);
3797                 break;
3798         case DIFF_STATUS_MODIFIED:
3799                 if (p->score) {
3800                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3801                                     line_prefix,
3802                                     set, similarity_index(p), reset);
3803                         break;
3804                 }
3805                 /* fallthru */
3806         default:
3807                 *must_show_header = 0;
3808         }
3809         if (one && two && oidcmp(&one->oid, &two->oid)) {
3810                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3811
3812                 if (DIFF_OPT_TST(o, BINARY)) {
3813                         mmfile_t mf;
3814                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3815                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3816                                 abbrev = 40;
3817                 }
3818                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3819                             diff_abbrev_oid(&one->oid, abbrev),
3820                             diff_abbrev_oid(&two->oid, abbrev));
3821                 if (one->mode == two->mode)
3822                         strbuf_addf(msg, " %06o", one->mode);
3823                 strbuf_addf(msg, "%s\n", reset);
3824         }
3825 }
3826
3827 static void run_diff_cmd(const char *pgm,
3828                          const char *name,
3829                          const char *other,
3830                          const char *attr_path,
3831                          struct diff_filespec *one,
3832                          struct diff_filespec *two,
3833                          struct strbuf *msg,
3834                          struct diff_options *o,
3835                          struct diff_filepair *p)
3836 {
3837         const char *xfrm_msg = NULL;
3838         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3839         int must_show_header = 0;
3840
3841
3842         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3843                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3844                 if (drv && drv->external)
3845                         pgm = drv->external;
3846         }
3847
3848         if (msg) {
3849                 /*
3850                  * don't use colors when the header is intended for an
3851                  * external diff driver
3852                  */
3853                 fill_metainfo(msg, name, other, one, two, o, p,
3854                               &must_show_header,
3855                               want_color(o->use_color) && !pgm);
3856                 xfrm_msg = msg->len ? msg->buf : NULL;
3857         }
3858
3859         if (pgm) {
3860                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3861                                   complete_rewrite, o);
3862                 return;
3863         }
3864         if (one && two)
3865                 builtin_diff(name, other ? other : name,
3866                              one, two, xfrm_msg, must_show_header,
3867                              o, complete_rewrite);
3868         else
3869                 fprintf(o->file, "* Unmerged path %s\n", name);
3870 }
3871
3872 static void diff_fill_oid_info(struct diff_filespec *one)
3873 {
3874         if (DIFF_FILE_VALID(one)) {
3875                 if (!one->oid_valid) {
3876                         struct stat st;
3877                         if (one->is_stdin) {
3878                                 oidclr(&one->oid);
3879                                 return;
3880                         }
3881                         if (lstat(one->path, &st) < 0)
3882                                 die_errno("stat '%s'", one->path);
3883                         if (index_path(one->oid.hash, one->path, &st, 0))
3884                                 die("cannot hash %s", one->path);
3885                 }
3886         }
3887         else
3888                 oidclr(&one->oid);
3889 }
3890
3891 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3892 {
3893         /* Strip the prefix but do not molest /dev/null and absolute paths */
3894         if (*namep && **namep != '/') {
3895                 *namep += prefix_length;
3896                 if (**namep == '/')
3897                         ++*namep;
3898         }
3899         if (*otherp && **otherp != '/') {
3900                 *otherp += prefix_length;
3901                 if (**otherp == '/')
3902                         ++*otherp;
3903         }
3904 }
3905
3906 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3907 {
3908         const char *pgm = external_diff();
3909         struct strbuf msg;
3910         struct diff_filespec *one = p->one;
3911         struct diff_filespec *two = p->two;
3912         const char *name;
3913         const char *other;
3914         const char *attr_path;
3915
3916         name  = one->path;
3917         other = (strcmp(name, two->path) ? two->path : NULL);
3918         attr_path = name;
3919         if (o->prefix_length)
3920                 strip_prefix(o->prefix_length, &name, &other);
3921
3922         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3923                 pgm = NULL;
3924
3925         if (DIFF_PAIR_UNMERGED(p)) {
3926                 run_diff_cmd(pgm, name, NULL, attr_path,
3927                              NULL, NULL, NULL, o, p);
3928                 return;
3929         }
3930
3931         diff_fill_oid_info(one);
3932         diff_fill_oid_info(two);
3933
3934         if (!pgm &&
3935             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3936             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3937                 /*
3938                  * a filepair that changes between file and symlink
3939                  * needs to be split into deletion and creation.
3940                  */
3941                 struct diff_filespec *null = alloc_filespec(two->path);
3942                 run_diff_cmd(NULL, name, other, attr_path,
3943                              one, null, &msg, o, p);
3944                 free(null);
3945                 strbuf_release(&msg);
3946
3947                 null = alloc_filespec(one->path);
3948                 run_diff_cmd(NULL, name, other, attr_path,
3949                              null, two, &msg, o, p);
3950                 free(null);
3951         }
3952         else
3953                 run_diff_cmd(pgm, name, other, attr_path,
3954                              one, two, &msg, o, p);
3955
3956         strbuf_release(&msg);
3957 }
3958
3959 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3960                          struct diffstat_t *diffstat)
3961 {
3962         const char *name;
3963         const char *other;
3964
3965         if (DIFF_PAIR_UNMERGED(p)) {
3966                 /* unmerged */
3967                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3968                 return;
3969         }
3970
3971         name = p->one->path;
3972         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3973
3974         if (o->prefix_length)
3975                 strip_prefix(o->prefix_length, &name, &other);
3976
3977         diff_fill_oid_info(p->one);
3978         diff_fill_oid_info(p->two);
3979
3980         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3981 }
3982
3983 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3984 {
3985         const char *name;
3986         const char *other;
3987         const char *attr_path;
3988
3989         if (DIFF_PAIR_UNMERGED(p)) {
3990                 /* unmerged */
3991                 return;
3992         }
3993
3994         name = p->one->path;
3995         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3996         attr_path = other ? other : name;
3997
3998         if (o->prefix_length)
3999                 strip_prefix(o->prefix_length, &name, &other);
4000
4001         diff_fill_oid_info(p->one);
4002         diff_fill_oid_info(p->two);
4003
4004         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4005 }
4006
4007 void diff_setup(struct diff_options *options)
4008 {
4009         memcpy(options, &default_diff_options, sizeof(*options));
4010
4011         options->file = stdout;
4012
4013         options->abbrev = DEFAULT_ABBREV;
4014         options->line_termination = '\n';
4015         options->break_opt = -1;
4016         options->rename_limit = -1;
4017         options->dirstat_permille = diff_dirstat_permille_default;
4018         options->context = diff_context_default;
4019         options->interhunkcontext = diff_interhunk_context_default;
4020         options->ws_error_highlight = ws_error_highlight_default;
4021         DIFF_OPT_SET(options, RENAME_EMPTY);
4022
4023         /* pathchange left =NULL by default */
4024         options->change = diff_change;
4025         options->add_remove = diff_addremove;
4026         options->use_color = diff_use_color_default;
4027         options->detect_rename = diff_detect_rename_default;
4028         options->xdl_opts |= diff_algorithm;
4029         if (diff_indent_heuristic)
4030                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4031
4032         options->orderfile = diff_order_file_cfg;
4033
4034         if (diff_no_prefix) {
4035                 options->a_prefix = options->b_prefix = "";
4036         } else if (!diff_mnemonic_prefix) {
4037                 options->a_prefix = "a/";
4038                 options->b_prefix = "b/";
4039         }
4040
4041         options->color_moved = diff_color_moved_default;
4042 }
4043
4044 void diff_setup_done(struct diff_options *options)
4045 {
4046         int count = 0;
4047
4048         if (options->set_default)
4049                 options->set_default(options);
4050
4051         if (options->output_format & DIFF_FORMAT_NAME)
4052                 count++;
4053         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
4054                 count++;
4055         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
4056                 count++;
4057         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
4058                 count++;
4059         if (count > 1)
4060                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4061
4062         /*
4063          * Most of the time we can say "there are changes"
4064          * only by checking if there are changed paths, but
4065          * --ignore-whitespace* options force us to look
4066          * inside contents.
4067          */
4068
4069         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
4070             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
4071             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
4072                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
4073         else
4074                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
4075
4076         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
4077                 options->detect_rename = DIFF_DETECT_COPY;
4078
4079         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
4080                 options->prefix = NULL;
4081         if (options->prefix)
4082                 options->prefix_length = strlen(options->prefix);
4083         else
4084                 options->prefix_length = 0;
4085
4086         if (options->output_format & (DIFF_FORMAT_NAME |
4087                                       DIFF_FORMAT_NAME_STATUS |
4088                                       DIFF_FORMAT_CHECKDIFF |
4089                                       DIFF_FORMAT_NO_OUTPUT))
4090                 options->output_format &= ~(DIFF_FORMAT_RAW |
4091                                             DIFF_FORMAT_NUMSTAT |
4092                                             DIFF_FORMAT_DIFFSTAT |
4093                                             DIFF_FORMAT_SHORTSTAT |
4094                                             DIFF_FORMAT_DIRSTAT |
4095                                             DIFF_FORMAT_SUMMARY |
4096                                             DIFF_FORMAT_PATCH);
4097
4098         /*
4099          * These cases always need recursive; we do not drop caller-supplied
4100          * recursive bits for other formats here.
4101          */
4102         if (options->output_format & (DIFF_FORMAT_PATCH |
4103                                       DIFF_FORMAT_NUMSTAT |
4104                                       DIFF_FORMAT_DIFFSTAT |
4105                                       DIFF_FORMAT_SHORTSTAT |
4106                                       DIFF_FORMAT_DIRSTAT |
4107                                       DIFF_FORMAT_SUMMARY |
4108                                       DIFF_FORMAT_CHECKDIFF))
4109                 DIFF_OPT_SET(options, RECURSIVE);
4110         /*
4111          * Also pickaxe would not work very well if you do not say recursive
4112          */
4113         if (options->pickaxe)
4114                 DIFF_OPT_SET(options, RECURSIVE);
4115         /*
4116          * When patches are generated, submodules diffed against the work tree
4117          * must be checked for dirtiness too so it can be shown in the output
4118          */
4119         if (options->output_format & DIFF_FORMAT_PATCH)
4120                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
4121
4122         if (options->detect_rename && options->rename_limit < 0)
4123                 options->rename_limit = diff_rename_limit_default;
4124         if (options->setup & DIFF_SETUP_USE_CACHE) {
4125                 if (!active_cache)
4126                         /* read-cache does not die even when it fails
4127                          * so it is safe for us to do this here.  Also
4128                          * it does not smudge active_cache or active_nr
4129                          * when it fails, so we do not have to worry about
4130                          * cleaning it up ourselves either.
4131                          */
4132                         read_cache();
4133         }
4134         if (40 < options->abbrev)
4135                 options->abbrev = 40; /* full */
4136
4137         /*
4138          * It does not make sense to show the first hit we happened
4139          * to have found.  It does not make sense not to return with
4140          * exit code in such a case either.
4141          */
4142         if (DIFF_OPT_TST(options, QUICK)) {
4143                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4144                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4145         }
4146
4147         options->diff_path_counter = 0;
4148
4149         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
4150                 die(_("--follow requires exactly one pathspec"));
4151
4152         if (!options->use_color || external_diff())
4153                 options->color_moved = 0;
4154 }
4155
4156 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4157 {
4158         char c, *eq;
4159         int len;
4160
4161         if (*arg != '-')
4162                 return 0;
4163         c = *++arg;
4164         if (!c)
4165                 return 0;
4166         if (c == arg_short) {
4167                 c = *++arg;
4168                 if (!c)
4169                         return 1;
4170                 if (val && isdigit(c)) {
4171                         char *end;
4172                         int n = strtoul(arg, &end, 10);
4173                         if (*end)
4174                                 return 0;
4175                         *val = n;
4176                         return 1;
4177                 }
4178                 return 0;
4179         }
4180         if (c != '-')
4181                 return 0;
4182         arg++;
4183         eq = strchrnul(arg, '=');
4184         len = eq - arg;
4185         if (!len || strncmp(arg, arg_long, len))
4186                 return 0;
4187         if (*eq) {
4188                 int n;
4189                 char *end;
4190                 if (!isdigit(*++eq))
4191                         return 0;
4192                 n = strtoul(eq, &end, 10);
4193                 if (*end)
4194                         return 0;
4195                 *val = n;
4196         }
4197         return 1;
4198 }
4199
4200 static int diff_scoreopt_parse(const char *opt);
4201
4202 static inline int short_opt(char opt, const char **argv,
4203                             const char **optarg)
4204 {
4205         const char *arg = argv[0];
4206         if (arg[0] != '-' || arg[1] != opt)
4207                 return 0;
4208         if (arg[2] != '\0') {
4209                 *optarg = arg + 2;
4210                 return 1;
4211         }
4212         if (!argv[1])
4213                 die("Option '%c' requires a value", opt);
4214         *optarg = argv[1];
4215         return 2;
4216 }
4217
4218 int parse_long_opt(const char *opt, const char **argv,
4219                    const char **optarg)
4220 {
4221         const char *arg = argv[0];
4222         if (!skip_prefix(arg, "--", &arg))
4223                 return 0;
4224         if (!skip_prefix(arg, opt, &arg))
4225                 return 0;
4226         if (*arg == '=') { /* stuck form: --option=value */
4227                 *optarg = arg + 1;
4228                 return 1;
4229         }
4230         if (*arg != '\0')
4231                 return 0;
4232         /* separate form: --option value */
4233         if (!argv[1])
4234                 die("Option '--%s' requires a value", opt);
4235         *optarg = argv[1];
4236         return 2;
4237 }
4238
4239 static int stat_opt(struct diff_options *options, const char **av)
4240 {
4241         const char *arg = av[0];
4242         char *end;
4243         int width = options->stat_width;
4244         int name_width = options->stat_name_width;
4245         int graph_width = options->stat_graph_width;
4246         int count = options->stat_count;
4247         int argcount = 1;
4248
4249         if (!skip_prefix(arg, "--stat", &arg))
4250                 die("BUG: stat option does not begin with --stat: %s", arg);
4251         end = (char *)arg;
4252
4253         switch (*arg) {
4254         case '-':
4255                 if (skip_prefix(arg, "-width", &arg)) {
4256                         if (*arg == '=')
4257                                 width = strtoul(arg + 1, &end, 10);
4258                         else if (!*arg && !av[1])
4259                                 die_want_option("--stat-width");
4260                         else if (!*arg) {
4261                                 width = strtoul(av[1], &end, 10);
4262                                 argcount = 2;
4263                         }
4264                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4265                         if (*arg == '=')
4266                                 name_width = strtoul(arg + 1, &end, 10);
4267                         else if (!*arg && !av[1])
4268                                 die_want_option("--stat-name-width");
4269                         else if (!*arg) {
4270                                 name_width = strtoul(av[1], &end, 10);
4271                                 argcount = 2;
4272                         }
4273                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4274                         if (*arg == '=')
4275                                 graph_width = strtoul(arg + 1, &end, 10);
4276                         else if (!*arg && !av[1])
4277                                 die_want_option("--stat-graph-width");
4278                         else if (!*arg) {
4279                                 graph_width = strtoul(av[1], &end, 10);
4280                                 argcount = 2;
4281                         }
4282                 } else if (skip_prefix(arg, "-count", &arg)) {
4283                         if (*arg == '=')
4284                                 count = strtoul(arg + 1, &end, 10);
4285                         else if (!*arg && !av[1])
4286                                 die_want_option("--stat-count");
4287                         else if (!*arg) {
4288                                 count = strtoul(av[1], &end, 10);
4289                                 argcount = 2;
4290                         }
4291                 }
4292                 break;
4293         case '=':
4294                 width = strtoul(arg+1, &end, 10);
4295                 if (*end == ',')
4296                         name_width = strtoul(end+1, &end, 10);
4297                 if (*end == ',')
4298                         count = strtoul(end+1, &end, 10);
4299         }
4300
4301         /* Important! This checks all the error cases! */
4302         if (*end)
4303                 return 0;
4304         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4305         options->stat_name_width = name_width;
4306         options->stat_graph_width = graph_width;
4307         options->stat_width = width;
4308         options->stat_count = count;
4309         return argcount;
4310 }
4311
4312 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4313 {
4314         struct strbuf errmsg = STRBUF_INIT;
4315         if (parse_dirstat_params(options, params, &errmsg))
4316                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4317                     errmsg.buf);
4318         strbuf_release(&errmsg);
4319         /*
4320          * The caller knows a dirstat-related option is given from the command
4321          * line; allow it to say "return this_function();"
4322          */
4323         options->output_format |= DIFF_FORMAT_DIRSTAT;
4324         return 1;
4325 }
4326
4327 static int parse_submodule_opt(struct diff_options *options, const char *value)
4328 {
4329         if (parse_submodule_params(options, value))
4330                 die(_("Failed to parse --submodule option parameter: '%s'"),
4331                         value);
4332         return 1;
4333 }
4334
4335 static const char diff_status_letters[] = {
4336         DIFF_STATUS_ADDED,
4337         DIFF_STATUS_COPIED,
4338         DIFF_STATUS_DELETED,
4339         DIFF_STATUS_MODIFIED,
4340         DIFF_STATUS_RENAMED,
4341         DIFF_STATUS_TYPE_CHANGED,
4342         DIFF_STATUS_UNKNOWN,
4343         DIFF_STATUS_UNMERGED,
4344         DIFF_STATUS_FILTER_AON,
4345         DIFF_STATUS_FILTER_BROKEN,
4346         '\0',
4347 };
4348
4349 static unsigned int filter_bit['Z' + 1];
4350
4351 static void prepare_filter_bits(void)
4352 {
4353         int i;
4354
4355         if (!filter_bit[DIFF_STATUS_ADDED]) {
4356                 for (i = 0; diff_status_letters[i]; i++)
4357                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4358         }
4359 }
4360
4361 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4362 {
4363         return opt->filter & filter_bit[(int) status];
4364 }
4365
4366 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4367 {
4368         int i, optch;
4369
4370         prepare_filter_bits();
4371
4372         /*
4373          * If there is a negation e.g. 'd' in the input, and we haven't
4374          * initialized the filter field with another --diff-filter, start
4375          * from full set of bits, except for AON.
4376          */
4377         if (!opt->filter) {
4378                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4379                         if (optch < 'a' || 'z' < optch)
4380                                 continue;
4381                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4382                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4383                         break;
4384                 }
4385         }
4386
4387         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4388                 unsigned int bit;
4389                 int negate;
4390
4391                 if ('a' <= optch && optch <= 'z') {
4392                         negate = 1;
4393                         optch = toupper(optch);
4394                 } else {
4395                         negate = 0;
4396                 }
4397
4398                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4399                 if (!bit)
4400                         return optarg[i];
4401                 if (negate)
4402                         opt->filter &= ~bit;
4403                 else
4404                         opt->filter |= bit;
4405         }
4406         return 0;
4407 }
4408
4409 static void enable_patch_output(int *fmt) {
4410         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4411         *fmt |= DIFF_FORMAT_PATCH;
4412 }
4413
4414 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4415 {
4416         int val = parse_ws_error_highlight(arg);
4417
4418         if (val < 0) {
4419                 error("unknown value after ws-error-highlight=%.*s",
4420                       -1 - val, arg);
4421                 return 0;
4422         }
4423         opt->ws_error_highlight = val;
4424         return 1;
4425 }
4426
4427 int diff_opt_parse(struct diff_options *options,
4428                    const char **av, int ac, const char *prefix)
4429 {
4430         const char *arg = av[0];
4431         const char *optarg;
4432         int argcount;
4433
4434         if (!prefix)
4435                 prefix = "";
4436
4437         /* Output format options */
4438         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4439             || opt_arg(arg, 'U', "unified", &options->context))
4440                 enable_patch_output(&options->output_format);
4441         else if (!strcmp(arg, "--raw"))
4442                 options->output_format |= DIFF_FORMAT_RAW;
4443         else if (!strcmp(arg, "--patch-with-raw")) {
4444                 enable_patch_output(&options->output_format);
4445                 options->output_format |= DIFF_FORMAT_RAW;
4446         } else if (!strcmp(arg, "--numstat"))
4447                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4448         else if (!strcmp(arg, "--shortstat"))
4449                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4450         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
4451                 return parse_dirstat_opt(options, "");
4452         else if (skip_prefix(arg, "-X", &arg))
4453                 return parse_dirstat_opt(options, arg);
4454         else if (skip_prefix(arg, "--dirstat=", &arg))
4455                 return parse_dirstat_opt(options, arg);
4456         else if (!strcmp(arg, "--cumulative"))
4457                 return parse_dirstat_opt(options, "cumulative");
4458         else if (!strcmp(arg, "--dirstat-by-file"))
4459                 return parse_dirstat_opt(options, "files");
4460         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
4461                 parse_dirstat_opt(options, "files");
4462                 return parse_dirstat_opt(options, arg);
4463         }
4464         else if (!strcmp(arg, "--check"))
4465                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4466         else if (!strcmp(arg, "--summary"))
4467                 options->output_format |= DIFF_FORMAT_SUMMARY;
4468         else if (!strcmp(arg, "--patch-with-stat")) {
4469                 enable_patch_output(&options->output_format);
4470                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4471         } else if (!strcmp(arg, "--name-only"))
4472                 options->output_format |= DIFF_FORMAT_NAME;
4473         else if (!strcmp(arg, "--name-status"))
4474                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4475         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4476                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4477         else if (starts_with(arg, "--stat"))
4478                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4479                 return stat_opt(options, av);
4480
4481         /* renames options */
4482         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
4483                  !strcmp(arg, "--break-rewrites")) {
4484                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4485                         return error("invalid argument to -B: %s", arg+2);
4486         }
4487         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
4488                  !strcmp(arg, "--find-renames")) {
4489                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4490                         return error("invalid argument to -M: %s", arg+2);
4491                 options->detect_rename = DIFF_DETECT_RENAME;
4492         }
4493         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4494                 options->irreversible_delete = 1;
4495         }
4496         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
4497                  !strcmp(arg, "--find-copies")) {
4498                 if (options->detect_rename == DIFF_DETECT_COPY)
4499                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4500                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4501                         return error("invalid argument to -C: %s", arg+2);
4502                 options->detect_rename = DIFF_DETECT_COPY;
4503         }
4504         else if (!strcmp(arg, "--no-renames"))
4505                 options->detect_rename = 0;
4506         else if (!strcmp(arg, "--rename-empty"))
4507                 DIFF_OPT_SET(options, RENAME_EMPTY);
4508         else if (!strcmp(arg, "--no-rename-empty"))
4509                 DIFF_OPT_CLR(options, RENAME_EMPTY);
4510         else if (!strcmp(arg, "--relative"))
4511                 DIFF_OPT_SET(options, RELATIVE_NAME);
4512         else if (skip_prefix(arg, "--relative=", &arg)) {
4513                 DIFF_OPT_SET(options, RELATIVE_NAME);
4514                 options->prefix = arg;
4515         }
4516
4517         /* xdiff options */
4518         else if (!strcmp(arg, "--minimal"))
4519                 DIFF_XDL_SET(options, NEED_MINIMAL);
4520         else if (!strcmp(arg, "--no-minimal"))
4521                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4522         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4523                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4524         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4525                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4526         else if (!strcmp(arg, "--ignore-space-at-eol"))
4527                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4528         else if (!strcmp(arg, "--ignore-blank-lines"))
4529                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4530         else if (!strcmp(arg, "--indent-heuristic"))
4531                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4532         else if (!strcmp(arg, "--no-indent-heuristic"))
4533                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4534         else if (!strcmp(arg, "--patience"))
4535                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4536         else if (!strcmp(arg, "--histogram"))
4537                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4538         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4539                 long value = parse_algorithm_value(optarg);
4540                 if (value < 0)
4541                         return error("option diff-algorithm accepts \"myers\", "
4542                                      "\"minimal\", \"patience\" and \"histogram\"");
4543                 /* clear out previous settings */
4544                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4545                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4546                 options->xdl_opts |= value;
4547                 return argcount;
4548         }
4549
4550         /* flags options */
4551         else if (!strcmp(arg, "--binary")) {
4552                 enable_patch_output(&options->output_format);
4553                 DIFF_OPT_SET(options, BINARY);
4554         }
4555         else if (!strcmp(arg, "--full-index"))
4556                 DIFF_OPT_SET(options, FULL_INDEX);
4557         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4558                 DIFF_OPT_SET(options, TEXT);
4559         else if (!strcmp(arg, "-R"))
4560                 DIFF_OPT_SET(options, REVERSE_DIFF);
4561         else if (!strcmp(arg, "--find-copies-harder"))
4562                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4563         else if (!strcmp(arg, "--follow"))
4564                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
4565         else if (!strcmp(arg, "--no-follow")) {
4566                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
4567                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
4568         } else if (!strcmp(arg, "--color"))
4569                 options->use_color = 1;
4570         else if (skip_prefix(arg, "--color=", &arg)) {
4571                 int value = git_config_colorbool(NULL, arg);
4572                 if (value < 0)
4573                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4574                 options->use_color = value;
4575         }
4576         else if (!strcmp(arg, "--no-color"))
4577                 options->use_color = 0;
4578         else if (!strcmp(arg, "--color-moved")) {
4579                 if (diff_color_moved_default)
4580                         options->color_moved = diff_color_moved_default;
4581                 if (options->color_moved == COLOR_MOVED_NO)
4582                         options->color_moved = COLOR_MOVED_DEFAULT;
4583         } else if (!strcmp(arg, "--no-color-moved"))
4584                 options->color_moved = COLOR_MOVED_NO;
4585         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4586                 int cm = parse_color_moved(arg);
4587                 if (cm < 0)
4588                         die("bad --color-moved argument: %s", arg);
4589                 options->color_moved = cm;
4590         } else if (!strcmp(arg, "--color-words")) {
4591                 options->use_color = 1;
4592                 options->word_diff = DIFF_WORDS_COLOR;
4593         }
4594         else if (skip_prefix(arg, "--color-words=", &arg)) {
4595                 options->use_color = 1;
4596                 options->word_diff = DIFF_WORDS_COLOR;
4597                 options->word_regex = arg;
4598         }
4599         else if (!strcmp(arg, "--word-diff")) {
4600                 if (options->word_diff == DIFF_WORDS_NONE)
4601                         options->word_diff = DIFF_WORDS_PLAIN;
4602         }
4603         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4604                 if (!strcmp(arg, "plain"))
4605                         options->word_diff = DIFF_WORDS_PLAIN;
4606                 else if (!strcmp(arg, "color")) {
4607                         options->use_color = 1;
4608                         options->word_diff = DIFF_WORDS_COLOR;
4609                 }
4610                 else if (!strcmp(arg, "porcelain"))
4611                         options->word_diff = DIFF_WORDS_PORCELAIN;
4612                 else if (!strcmp(arg, "none"))
4613                         options->word_diff = DIFF_WORDS_NONE;
4614                 else
4615                         die("bad --word-diff argument: %s", arg);
4616         }
4617         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4618                 if (options->word_diff == DIFF_WORDS_NONE)
4619                         options->word_diff = DIFF_WORDS_PLAIN;
4620                 options->word_regex = optarg;
4621                 return argcount;
4622         }
4623         else if (!strcmp(arg, "--exit-code"))
4624                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4625         else if (!strcmp(arg, "--quiet"))
4626                 DIFF_OPT_SET(options, QUICK);
4627         else if (!strcmp(arg, "--ext-diff"))
4628                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
4629         else if (!strcmp(arg, "--no-ext-diff"))
4630                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
4631         else if (!strcmp(arg, "--textconv"))
4632                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
4633         else if (!strcmp(arg, "--no-textconv"))
4634                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
4635         else if (!strcmp(arg, "--ignore-submodules")) {
4636                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4637                 handle_ignore_submodules_arg(options, "all");
4638         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
4639                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4640                 handle_ignore_submodules_arg(options, arg);
4641         } else if (!strcmp(arg, "--submodule"))
4642                 options->submodule_format = DIFF_SUBMODULE_LOG;
4643         else if (skip_prefix(arg, "--submodule=", &arg))
4644                 return parse_submodule_opt(options, arg);
4645         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4646                 return parse_ws_error_highlight_opt(options, arg);
4647         else if (!strcmp(arg, "--ita-invisible-in-index"))
4648                 options->ita_invisible_in_index = 1;
4649         else if (!strcmp(arg, "--ita-visible-in-index"))
4650                 options->ita_invisible_in_index = 0;
4651
4652         /* misc options */
4653         else if (!strcmp(arg, "-z"))
4654                 options->line_termination = 0;
4655         else if ((argcount = short_opt('l', av, &optarg))) {
4656                 options->rename_limit = strtoul(optarg, NULL, 10);
4657                 return argcount;
4658         }
4659         else if ((argcount = short_opt('S', av, &optarg))) {
4660                 options->pickaxe = optarg;
4661                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4662                 return argcount;
4663         } else if ((argcount = short_opt('G', av, &optarg))) {
4664                 options->pickaxe = optarg;
4665                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4666                 return argcount;
4667         }
4668         else if (!strcmp(arg, "--pickaxe-all"))
4669                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4670         else if (!strcmp(arg, "--pickaxe-regex"))
4671                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4672         else if ((argcount = short_opt('O', av, &optarg))) {
4673                 options->orderfile = prefix_filename(prefix, optarg);
4674                 return argcount;
4675         }
4676         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4677                 int offending = parse_diff_filter_opt(optarg, options);
4678                 if (offending)
4679                         die("unknown change class '%c' in --diff-filter=%s",
4680                             offending, optarg);
4681                 return argcount;
4682         }
4683         else if (!strcmp(arg, "--no-abbrev"))
4684                 options->abbrev = 0;
4685         else if (!strcmp(arg, "--abbrev"))
4686                 options->abbrev = DEFAULT_ABBREV;
4687         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4688                 options->abbrev = strtoul(arg, NULL, 10);
4689                 if (options->abbrev < MINIMUM_ABBREV)
4690                         options->abbrev = MINIMUM_ABBREV;
4691                 else if (40 < options->abbrev)
4692                         options->abbrev = 40;
4693         }
4694         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4695                 options->a_prefix = optarg;
4696                 return argcount;
4697         }
4698         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4699                 options->line_prefix = optarg;
4700                 options->line_prefix_length = strlen(options->line_prefix);
4701                 graph_setup_line_prefix(options);
4702                 return argcount;
4703         }
4704         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4705                 options->b_prefix = optarg;
4706                 return argcount;
4707         }
4708         else if (!strcmp(arg, "--no-prefix"))
4709                 options->a_prefix = options->b_prefix = "";
4710         else if (opt_arg(arg, '\0', "inter-hunk-context",
4711                          &options->interhunkcontext))
4712                 ;
4713         else if (!strcmp(arg, "-W"))
4714                 DIFF_OPT_SET(options, FUNCCONTEXT);
4715         else if (!strcmp(arg, "--function-context"))
4716                 DIFF_OPT_SET(options, FUNCCONTEXT);
4717         else if (!strcmp(arg, "--no-function-context"))
4718                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4719         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4720                 char *path = prefix_filename(prefix, optarg);
4721                 options->file = xfopen(path, "w");
4722                 options->close_file = 1;
4723                 if (options->use_color != GIT_COLOR_ALWAYS)
4724                         options->use_color = GIT_COLOR_NEVER;
4725                 free(path);
4726                 return argcount;
4727         } else
4728                 return 0;
4729         return 1;
4730 }
4731
4732 int parse_rename_score(const char **cp_p)
4733 {
4734         unsigned long num, scale;
4735         int ch, dot;
4736         const char *cp = *cp_p;
4737
4738         num = 0;
4739         scale = 1;
4740         dot = 0;
4741         for (;;) {
4742                 ch = *cp;
4743                 if ( !dot && ch == '.' ) {
4744                         scale = 1;
4745                         dot = 1;
4746                 } else if ( ch == '%' ) {
4747                         scale = dot ? scale*100 : 100;
4748                         cp++;   /* % is always at the end */
4749                         break;
4750                 } else if ( ch >= '0' && ch <= '9' ) {
4751                         if ( scale < 100000 ) {
4752                                 scale *= 10;
4753                                 num = (num*10) + (ch-'0');
4754                         }
4755                 } else {
4756                         break;
4757                 }
4758                 cp++;
4759         }
4760         *cp_p = cp;
4761
4762         /* user says num divided by scale and we say internally that
4763          * is MAX_SCORE * num / scale.
4764          */
4765         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4766 }
4767
4768 static int diff_scoreopt_parse(const char *opt)
4769 {
4770         int opt1, opt2, cmd;
4771
4772         if (*opt++ != '-')
4773                 return -1;
4774         cmd = *opt++;
4775         if (cmd == '-') {
4776                 /* convert the long-form arguments into short-form versions */
4777                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4778                         if (*opt == 0 || *opt++ == '=')
4779                                 cmd = 'B';
4780                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4781                         if (*opt == 0 || *opt++ == '=')
4782                                 cmd = 'C';
4783                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4784                         if (*opt == 0 || *opt++ == '=')
4785                                 cmd = 'M';
4786                 }
4787         }
4788         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4789                 return -1; /* that is not a -M, -C, or -B option */
4790
4791         opt1 = parse_rename_score(&opt);
4792         if (cmd != 'B')
4793                 opt2 = 0;
4794         else {
4795                 if (*opt == 0)
4796                         opt2 = 0;
4797                 else if (*opt != '/')
4798                         return -1; /* we expect -B80/99 or -B80 */
4799                 else {
4800                         opt++;
4801                         opt2 = parse_rename_score(&opt);
4802                 }
4803         }
4804         if (*opt != 0)
4805                 return -1;
4806         return opt1 | (opt2 << 16);
4807 }
4808
4809 struct diff_queue_struct diff_queued_diff;
4810
4811 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4812 {
4813         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4814         queue->queue[queue->nr++] = dp;
4815 }
4816
4817 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4818                                  struct diff_filespec *one,
4819                                  struct diff_filespec *two)
4820 {
4821         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4822         dp->one = one;
4823         dp->two = two;
4824         if (queue)
4825                 diff_q(queue, dp);
4826         return dp;
4827 }
4828
4829 void diff_free_filepair(struct diff_filepair *p)
4830 {
4831         free_filespec(p->one);
4832         free_filespec(p->two);
4833         free(p);
4834 }
4835
4836 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4837 {
4838         int abblen;
4839         const char *abbrev;
4840
4841         if (len == GIT_SHA1_HEXSZ)
4842                 return oid_to_hex(oid);
4843
4844         abbrev = diff_abbrev_oid(oid, len);
4845         abblen = strlen(abbrev);
4846
4847         /*
4848          * In well-behaved cases, where the abbbreviated result is the
4849          * same as the requested length, append three dots after the
4850          * abbreviation (hence the whole logic is limited to the case
4851          * where abblen < 37); when the actual abbreviated result is a
4852          * bit longer than the requested length, we reduce the number
4853          * of dots so that they match the well-behaved ones.  However,
4854          * if the actual abbreviation is longer than the requested
4855          * length by more than three, we give up on aligning, and add
4856          * three dots anyway, to indicate that the output is not the
4857          * full object name.  Yes, this may be suboptimal, but this
4858          * appears only in "diff --raw --abbrev" output and it is not
4859          * worth the effort to change it now.  Note that this would
4860          * likely to work fine when the automatic sizing of default
4861          * abbreviation length is used--we would be fed -1 in "len" in
4862          * that case, and will end up always appending three-dots, but
4863          * the automatic sizing is supposed to give abblen that ensures
4864          * uniqueness across all objects (statistically speaking).
4865          */
4866         if (abblen < GIT_SHA1_HEXSZ - 3) {
4867                 static char hex[GIT_MAX_HEXSZ + 1];
4868                 if (len < abblen && abblen <= len + 2)
4869                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4870                 else
4871                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4872                 return hex;
4873         }
4874
4875         return oid_to_hex(oid);
4876 }
4877
4878 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4879 {
4880         int line_termination = opt->line_termination;
4881         int inter_name_termination = line_termination ? '\t' : '\0';
4882
4883         fprintf(opt->file, "%s", diff_line_prefix(opt));
4884         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4885                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4886                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4887                 fprintf(opt->file, "%s ",
4888                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4889         }
4890         if (p->score) {
4891                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4892                         inter_name_termination);
4893         } else {
4894                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4895         }
4896
4897         if (p->status == DIFF_STATUS_COPIED ||
4898             p->status == DIFF_STATUS_RENAMED) {
4899                 const char *name_a, *name_b;
4900                 name_a = p->one->path;
4901                 name_b = p->two->path;
4902                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4903                 write_name_quoted(name_a, opt->file, inter_name_termination);
4904                 write_name_quoted(name_b, opt->file, line_termination);
4905         } else {
4906                 const char *name_a, *name_b;
4907                 name_a = p->one->mode ? p->one->path : p->two->path;
4908                 name_b = NULL;
4909                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4910                 write_name_quoted(name_a, opt->file, line_termination);
4911         }
4912 }
4913
4914 int diff_unmodified_pair(struct diff_filepair *p)
4915 {
4916         /* This function is written stricter than necessary to support
4917          * the currently implemented transformers, but the idea is to
4918          * let transformers to produce diff_filepairs any way they want,
4919          * and filter and clean them up here before producing the output.
4920          */
4921         struct diff_filespec *one = p->one, *two = p->two;
4922
4923         if (DIFF_PAIR_UNMERGED(p))
4924                 return 0; /* unmerged is interesting */
4925
4926         /* deletion, addition, mode or type change
4927          * and rename are all interesting.
4928          */
4929         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4930             DIFF_PAIR_MODE_CHANGED(p) ||
4931             strcmp(one->path, two->path))
4932                 return 0;
4933
4934         /* both are valid and point at the same path.  that is, we are
4935          * dealing with a change.
4936          */
4937         if (one->oid_valid && two->oid_valid &&
4938             !oidcmp(&one->oid, &two->oid) &&
4939             !one->dirty_submodule && !two->dirty_submodule)
4940                 return 1; /* no change */
4941         if (!one->oid_valid && !two->oid_valid)
4942                 return 1; /* both look at the same file on the filesystem. */
4943         return 0;
4944 }
4945
4946 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4947 {
4948         if (diff_unmodified_pair(p))
4949                 return;
4950
4951         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4952             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4953                 return; /* no tree diffs in patch format */
4954
4955         run_diff(p, o);
4956 }
4957
4958 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4959                             struct diffstat_t *diffstat)
4960 {
4961         if (diff_unmodified_pair(p))
4962                 return;
4963
4964         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4965             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4966                 return; /* no useful stat for tree diffs */
4967
4968         run_diffstat(p, o, diffstat);
4969 }
4970
4971 static void diff_flush_checkdiff(struct diff_filepair *p,
4972                 struct diff_options *o)
4973 {
4974         if (diff_unmodified_pair(p))
4975                 return;
4976
4977         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4978             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4979                 return; /* nothing to check in tree diffs */
4980
4981         run_checkdiff(p, o);
4982 }
4983
4984 int diff_queue_is_empty(void)
4985 {
4986         struct diff_queue_struct *q = &diff_queued_diff;
4987         int i;
4988         for (i = 0; i < q->nr; i++)
4989                 if (!diff_unmodified_pair(q->queue[i]))
4990                         return 0;
4991         return 1;
4992 }
4993
4994 #if DIFF_DEBUG
4995 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4996 {
4997         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4998                 x, one ? one : "",
4999                 s->path,
5000                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5001                 s->mode,
5002                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5003         fprintf(stderr, "queue[%d] %s size %lu\n",
5004                 x, one ? one : "",
5005                 s->size);
5006 }
5007
5008 void diff_debug_filepair(const struct diff_filepair *p, int i)
5009 {
5010         diff_debug_filespec(p->one, i, "one");
5011         diff_debug_filespec(p->two, i, "two");
5012         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5013                 p->score, p->status ? p->status : '?',
5014                 p->one->rename_used, p->broken_pair);
5015 }
5016
5017 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5018 {
5019         int i;
5020         if (msg)
5021                 fprintf(stderr, "%s\n", msg);
5022         fprintf(stderr, "q->nr = %d\n", q->nr);
5023         for (i = 0; i < q->nr; i++) {
5024                 struct diff_filepair *p = q->queue[i];
5025                 diff_debug_filepair(p, i);
5026         }
5027 }
5028 #endif
5029
5030 static void diff_resolve_rename_copy(void)
5031 {
5032         int i;
5033         struct diff_filepair *p;
5034         struct diff_queue_struct *q = &diff_queued_diff;
5035
5036         diff_debug_queue("resolve-rename-copy", q);
5037
5038         for (i = 0; i < q->nr; i++) {
5039                 p = q->queue[i];
5040                 p->status = 0; /* undecided */
5041                 if (DIFF_PAIR_UNMERGED(p))
5042                         p->status = DIFF_STATUS_UNMERGED;
5043                 else if (!DIFF_FILE_VALID(p->one))
5044                         p->status = DIFF_STATUS_ADDED;
5045                 else if (!DIFF_FILE_VALID(p->two))
5046                         p->status = DIFF_STATUS_DELETED;
5047                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5048                         p->status = DIFF_STATUS_TYPE_CHANGED;
5049
5050                 /* from this point on, we are dealing with a pair
5051                  * whose both sides are valid and of the same type, i.e.
5052                  * either in-place edit or rename/copy edit.
5053                  */
5054                 else if (DIFF_PAIR_RENAME(p)) {
5055                         /*
5056                          * A rename might have re-connected a broken
5057                          * pair up, causing the pathnames to be the
5058                          * same again. If so, that's not a rename at
5059                          * all, just a modification..
5060                          *
5061                          * Otherwise, see if this source was used for
5062                          * multiple renames, in which case we decrement
5063                          * the count, and call it a copy.
5064                          */
5065                         if (!strcmp(p->one->path, p->two->path))
5066                                 p->status = DIFF_STATUS_MODIFIED;
5067                         else if (--p->one->rename_used > 0)
5068                                 p->status = DIFF_STATUS_COPIED;
5069                         else
5070                                 p->status = DIFF_STATUS_RENAMED;
5071                 }
5072                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5073                          p->one->mode != p->two->mode ||
5074                          p->one->dirty_submodule ||
5075                          p->two->dirty_submodule ||
5076                          is_null_oid(&p->one->oid))
5077                         p->status = DIFF_STATUS_MODIFIED;
5078                 else {
5079                         /* This is a "no-change" entry and should not
5080                          * happen anymore, but prepare for broken callers.
5081                          */
5082                         error("feeding unmodified %s to diffcore",
5083                               p->one->path);
5084                         p->status = DIFF_STATUS_UNKNOWN;
5085                 }
5086         }
5087         diff_debug_queue("resolve-rename-copy done", q);
5088 }
5089
5090 static int check_pair_status(struct diff_filepair *p)
5091 {
5092         switch (p->status) {
5093         case DIFF_STATUS_UNKNOWN:
5094                 return 0;
5095         case 0:
5096                 die("internal error in diff-resolve-rename-copy");
5097         default:
5098                 return 1;
5099         }
5100 }
5101
5102 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5103 {
5104         int fmt = opt->output_format;
5105
5106         if (fmt & DIFF_FORMAT_CHECKDIFF)
5107                 diff_flush_checkdiff(p, opt);
5108         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5109                 diff_flush_raw(p, opt);
5110         else if (fmt & DIFF_FORMAT_NAME) {
5111                 const char *name_a, *name_b;
5112                 name_a = p->two->path;
5113                 name_b = NULL;
5114                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5115                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5116                 write_name_quoted(name_a, opt->file, opt->line_termination);
5117         }
5118 }
5119
5120 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5121 {
5122         struct strbuf sb = STRBUF_INIT;
5123         if (fs->mode)
5124                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5125         else
5126                 strbuf_addf(&sb, " %s ", newdelete);
5127
5128         quote_c_style(fs->path, &sb, NULL, 0);
5129         strbuf_addch(&sb, '\n');
5130         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5131                          sb.buf, sb.len, 0);
5132         strbuf_release(&sb);
5133 }
5134
5135 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5136                 int show_name)
5137 {
5138         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5139                 struct strbuf sb = STRBUF_INIT;
5140                 strbuf_addf(&sb, " mode change %06o => %06o",
5141                             p->one->mode, p->two->mode);
5142                 if (show_name) {
5143                         strbuf_addch(&sb, ' ');
5144                         quote_c_style(p->two->path, &sb, NULL, 0);
5145                 }
5146                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5147                                  sb.buf, sb.len, 0);
5148                 strbuf_release(&sb);
5149         }
5150 }
5151
5152 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5153                 struct diff_filepair *p)
5154 {
5155         struct strbuf sb = STRBUF_INIT;
5156         char *names = pprint_rename(p->one->path, p->two->path);
5157         strbuf_addf(&sb, " %s %s (%d%%)\n",
5158                         renamecopy, names, similarity_index(p));
5159         free(names);
5160         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5161                                  sb.buf, sb.len, 0);
5162         show_mode_change(opt, p, 0);
5163 }
5164
5165 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5166 {
5167         switch(p->status) {
5168         case DIFF_STATUS_DELETED:
5169                 show_file_mode_name(opt, "delete", p->one);
5170                 break;
5171         case DIFF_STATUS_ADDED:
5172                 show_file_mode_name(opt, "create", p->two);
5173                 break;
5174         case DIFF_STATUS_COPIED:
5175                 show_rename_copy(opt, "copy", p);
5176                 break;
5177         case DIFF_STATUS_RENAMED:
5178                 show_rename_copy(opt, "rename", p);
5179                 break;
5180         default:
5181                 if (p->score) {
5182                         struct strbuf sb = STRBUF_INIT;
5183                         strbuf_addstr(&sb, " rewrite ");
5184                         quote_c_style(p->two->path, &sb, NULL, 0);
5185                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5186                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5187                                          sb.buf, sb.len, 0);
5188                 }
5189                 show_mode_change(opt, p, !p->score);
5190                 break;
5191         }
5192 }
5193
5194 struct patch_id_t {
5195         git_SHA_CTX *ctx;
5196         int patchlen;
5197 };
5198
5199 static int remove_space(char *line, int len)
5200 {
5201         int i;
5202         char *dst = line;
5203         unsigned char c;
5204
5205         for (i = 0; i < len; i++)
5206                 if (!isspace((c = line[i])))
5207                         *dst++ = c;
5208
5209         return dst - line;
5210 }
5211
5212 static void patch_id_consume(void *priv, char *line, unsigned long len)
5213 {
5214         struct patch_id_t *data = priv;
5215         int new_len;
5216
5217         /* Ignore line numbers when computing the SHA1 of the patch */
5218         if (starts_with(line, "@@ -"))
5219                 return;
5220
5221         new_len = remove_space(line, len);
5222
5223         git_SHA1_Update(data->ctx, line, new_len);
5224         data->patchlen += new_len;
5225 }
5226
5227 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5228 {
5229         git_SHA1_Update(ctx, str, strlen(str));
5230 }
5231
5232 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5233 {
5234         /* large enough for 2^32 in octal */
5235         char buf[12];
5236         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5237         git_SHA1_Update(ctx, buf, len);
5238 }
5239
5240 /* returns 0 upon success, and writes result into sha1 */
5241 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5242 {
5243         struct diff_queue_struct *q = &diff_queued_diff;
5244         int i;
5245         git_SHA_CTX ctx;
5246         struct patch_id_t data;
5247
5248         git_SHA1_Init(&ctx);
5249         memset(&data, 0, sizeof(struct patch_id_t));
5250         data.ctx = &ctx;
5251
5252         for (i = 0; i < q->nr; i++) {
5253                 xpparam_t xpp;
5254                 xdemitconf_t xecfg;
5255                 mmfile_t mf1, mf2;
5256                 struct diff_filepair *p = q->queue[i];
5257                 int len1, len2;
5258
5259                 memset(&xpp, 0, sizeof(xpp));
5260                 memset(&xecfg, 0, sizeof(xecfg));
5261                 if (p->status == 0)
5262                         return error("internal diff status error");
5263                 if (p->status == DIFF_STATUS_UNKNOWN)
5264                         continue;
5265                 if (diff_unmodified_pair(p))
5266                         continue;
5267                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5268                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5269                         continue;
5270                 if (DIFF_PAIR_UNMERGED(p))
5271                         continue;
5272
5273                 diff_fill_oid_info(p->one);
5274                 diff_fill_oid_info(p->two);
5275
5276                 len1 = remove_space(p->one->path, strlen(p->one->path));
5277                 len2 = remove_space(p->two->path, strlen(p->two->path));
5278                 patch_id_add_string(&ctx, "diff--git");
5279                 patch_id_add_string(&ctx, "a/");
5280                 git_SHA1_Update(&ctx, p->one->path, len1);
5281                 patch_id_add_string(&ctx, "b/");
5282                 git_SHA1_Update(&ctx, p->two->path, len2);
5283
5284                 if (p->one->mode == 0) {
5285                         patch_id_add_string(&ctx, "newfilemode");
5286                         patch_id_add_mode(&ctx, p->two->mode);
5287                         patch_id_add_string(&ctx, "---/dev/null");
5288                         patch_id_add_string(&ctx, "+++b/");
5289                         git_SHA1_Update(&ctx, p->two->path, len2);
5290                 } else if (p->two->mode == 0) {
5291                         patch_id_add_string(&ctx, "deletedfilemode");
5292                         patch_id_add_mode(&ctx, p->one->mode);
5293                         patch_id_add_string(&ctx, "---a/");
5294                         git_SHA1_Update(&ctx, p->one->path, len1);
5295                         patch_id_add_string(&ctx, "+++/dev/null");
5296                 } else {
5297                         patch_id_add_string(&ctx, "---a/");
5298                         git_SHA1_Update(&ctx, p->one->path, len1);
5299                         patch_id_add_string(&ctx, "+++b/");
5300                         git_SHA1_Update(&ctx, p->two->path, len2);
5301                 }
5302
5303                 if (diff_header_only)
5304                         continue;
5305
5306                 if (fill_mmfile(&mf1, p->one) < 0 ||
5307                     fill_mmfile(&mf2, p->two) < 0)
5308                         return error("unable to read files to diff");
5309
5310                 if (diff_filespec_is_binary(p->one) ||
5311                     diff_filespec_is_binary(p->two)) {
5312                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5313                                         GIT_SHA1_HEXSZ);
5314                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5315                                         GIT_SHA1_HEXSZ);
5316                         continue;
5317                 }
5318
5319                 xpp.flags = 0;
5320                 xecfg.ctxlen = 3;
5321                 xecfg.flags = 0;
5322                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5323                                   &xpp, &xecfg))
5324                         return error("unable to generate patch-id diff for %s",
5325                                      p->one->path);
5326         }
5327
5328         git_SHA1_Final(oid->hash, &ctx);
5329         return 0;
5330 }
5331
5332 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5333 {
5334         struct diff_queue_struct *q = &diff_queued_diff;
5335         int i;
5336         int result = diff_get_patch_id(options, oid, diff_header_only);
5337
5338         for (i = 0; i < q->nr; i++)
5339                 diff_free_filepair(q->queue[i]);
5340
5341         free(q->queue);
5342         DIFF_QUEUE_CLEAR(q);
5343
5344         return result;
5345 }
5346
5347 static int is_summary_empty(const struct diff_queue_struct *q)
5348 {
5349         int i;
5350
5351         for (i = 0; i < q->nr; i++) {
5352                 const struct diff_filepair *p = q->queue[i];
5353
5354                 switch (p->status) {
5355                 case DIFF_STATUS_DELETED:
5356                 case DIFF_STATUS_ADDED:
5357                 case DIFF_STATUS_COPIED:
5358                 case DIFF_STATUS_RENAMED:
5359                         return 0;
5360                 default:
5361                         if (p->score)
5362                                 return 0;
5363                         if (p->one->mode && p->two->mode &&
5364                             p->one->mode != p->two->mode)
5365                                 return 0;
5366                         break;
5367                 }
5368         }
5369         return 1;
5370 }
5371
5372 static const char rename_limit_warning[] =
5373 N_("inexact rename detection was skipped due to too many files.");
5374
5375 static const char degrade_cc_to_c_warning[] =
5376 N_("only found copies from modified paths due to too many files.");
5377
5378 static const char rename_limit_advice[] =
5379 N_("you may want to set your %s variable to at least "
5380    "%d and retry the command.");
5381
5382 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5383 {
5384         if (degraded_cc)
5385                 warning(_(degrade_cc_to_c_warning));
5386         else if (needed)
5387                 warning(_(rename_limit_warning));
5388         else
5389                 return;
5390         if (0 < needed && needed < 32767)
5391                 warning(_(rename_limit_advice), varname, needed);
5392 }
5393
5394 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5395 {
5396         int i;
5397         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5398         struct diff_queue_struct *q = &diff_queued_diff;
5399
5400         if (WSEH_NEW & WS_RULE_MASK)
5401                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5402
5403         if (o->color_moved)
5404                 o->emitted_symbols = &esm;
5405
5406         for (i = 0; i < q->nr; i++) {
5407                 struct diff_filepair *p = q->queue[i];
5408                 if (check_pair_status(p))
5409                         diff_flush_patch(p, o);
5410         }
5411
5412         if (o->emitted_symbols) {
5413                 if (o->color_moved) {
5414                         struct hashmap add_lines, del_lines;
5415
5416                         hashmap_init(&del_lines,
5417                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5418                         hashmap_init(&add_lines,
5419                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5420
5421                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5422                         mark_color_as_moved(o, &add_lines, &del_lines);
5423
5424                         hashmap_free(&add_lines, 0);
5425                         hashmap_free(&del_lines, 0);
5426                 }
5427
5428                 for (i = 0; i < esm.nr; i++)
5429                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5430
5431                 for (i = 0; i < esm.nr; i++)
5432                         free((void *)esm.buf[i].line);
5433         }
5434         esm.nr = 0;
5435 }
5436
5437 void diff_flush(struct diff_options *options)
5438 {
5439         struct diff_queue_struct *q = &diff_queued_diff;
5440         int i, output_format = options->output_format;
5441         int separator = 0;
5442         int dirstat_by_line = 0;
5443
5444         /*
5445          * Order: raw, stat, summary, patch
5446          * or:    name/name-status/checkdiff (other bits clear)
5447          */
5448         if (!q->nr)
5449                 goto free_queue;
5450
5451         if (output_format & (DIFF_FORMAT_RAW |
5452                              DIFF_FORMAT_NAME |
5453                              DIFF_FORMAT_NAME_STATUS |
5454                              DIFF_FORMAT_CHECKDIFF)) {
5455                 for (i = 0; i < q->nr; i++) {
5456                         struct diff_filepair *p = q->queue[i];
5457                         if (check_pair_status(p))
5458                                 flush_one_pair(p, options);
5459                 }
5460                 separator++;
5461         }
5462
5463         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
5464                 dirstat_by_line = 1;
5465
5466         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5467             dirstat_by_line) {
5468                 struct diffstat_t diffstat;
5469
5470                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5471                 for (i = 0; i < q->nr; i++) {
5472                         struct diff_filepair *p = q->queue[i];
5473                         if (check_pair_status(p))
5474                                 diff_flush_stat(p, options, &diffstat);
5475                 }
5476                 if (output_format & DIFF_FORMAT_NUMSTAT)
5477                         show_numstat(&diffstat, options);
5478                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5479                         show_stats(&diffstat, options);
5480                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5481                         show_shortstats(&diffstat, options);
5482                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5483                         show_dirstat_by_line(&diffstat, options);
5484                 free_diffstat_info(&diffstat);
5485                 separator++;
5486         }
5487         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5488                 show_dirstat(options);
5489
5490         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5491                 for (i = 0; i < q->nr; i++) {
5492                         diff_summary(options, q->queue[i]);
5493                 }
5494                 separator++;
5495         }
5496
5497         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5498             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
5499             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5500                 /*
5501                  * run diff_flush_patch for the exit status. setting
5502                  * options->file to /dev/null should be safe, because we
5503                  * aren't supposed to produce any output anyway.
5504                  */
5505                 if (options->close_file)
5506                         fclose(options->file);
5507                 options->file = xfopen("/dev/null", "w");
5508                 options->close_file = 1;
5509                 options->color_moved = 0;
5510                 for (i = 0; i < q->nr; i++) {
5511                         struct diff_filepair *p = q->queue[i];
5512                         if (check_pair_status(p))
5513                                 diff_flush_patch(p, options);
5514                         if (options->found_changes)
5515                                 break;
5516                 }
5517         }
5518
5519         if (output_format & DIFF_FORMAT_PATCH) {
5520                 if (separator) {
5521                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5522                         if (options->stat_sep)
5523                                 /* attach patch instead of inline */
5524                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5525                                                  NULL, 0, 0);
5526                 }
5527
5528                 diff_flush_patch_all_file_pairs(options);
5529         }
5530
5531         if (output_format & DIFF_FORMAT_CALLBACK)
5532                 options->format_callback(q, options, options->format_callback_data);
5533
5534         for (i = 0; i < q->nr; i++)
5535                 diff_free_filepair(q->queue[i]);
5536 free_queue:
5537         free(q->queue);
5538         DIFF_QUEUE_CLEAR(q);
5539         if (options->close_file)
5540                 fclose(options->file);
5541
5542         /*
5543          * Report the content-level differences with HAS_CHANGES;
5544          * diff_addremove/diff_change does not set the bit when
5545          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5546          */
5547         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5548                 if (options->found_changes)
5549                         DIFF_OPT_SET(options, HAS_CHANGES);
5550                 else
5551                         DIFF_OPT_CLR(options, HAS_CHANGES);
5552         }
5553 }
5554
5555 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5556 {
5557         return (((p->status == DIFF_STATUS_MODIFIED) &&
5558                  ((p->score &&
5559                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5560                   (!p->score &&
5561                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5562                 ((p->status != DIFF_STATUS_MODIFIED) &&
5563                  filter_bit_tst(p->status, options)));
5564 }
5565
5566 static void diffcore_apply_filter(struct diff_options *options)
5567 {
5568         int i;
5569         struct diff_queue_struct *q = &diff_queued_diff;
5570         struct diff_queue_struct outq;
5571
5572         DIFF_QUEUE_CLEAR(&outq);
5573
5574         if (!options->filter)
5575                 return;
5576
5577         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5578                 int found;
5579                 for (i = found = 0; !found && i < q->nr; i++) {
5580                         if (match_filter(options, q->queue[i]))
5581                                 found++;
5582                 }
5583                 if (found)
5584                         return;
5585
5586                 /* otherwise we will clear the whole queue
5587                  * by copying the empty outq at the end of this
5588                  * function, but first clear the current entries
5589                  * in the queue.
5590                  */
5591                 for (i = 0; i < q->nr; i++)
5592                         diff_free_filepair(q->queue[i]);
5593         }
5594         else {
5595                 /* Only the matching ones */
5596                 for (i = 0; i < q->nr; i++) {
5597                         struct diff_filepair *p = q->queue[i];
5598                         if (match_filter(options, p))
5599                                 diff_q(&outq, p);
5600                         else
5601                                 diff_free_filepair(p);
5602                 }
5603         }
5604         free(q->queue);
5605         *q = outq;
5606 }
5607
5608 /* Check whether two filespecs with the same mode and size are identical */
5609 static int diff_filespec_is_identical(struct diff_filespec *one,
5610                                       struct diff_filespec *two)
5611 {
5612         if (S_ISGITLINK(one->mode))
5613                 return 0;
5614         if (diff_populate_filespec(one, 0))
5615                 return 0;
5616         if (diff_populate_filespec(two, 0))
5617                 return 0;
5618         return !memcmp(one->data, two->data, one->size);
5619 }
5620
5621 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5622 {
5623         if (p->done_skip_stat_unmatch)
5624                 return p->skip_stat_unmatch_result;
5625
5626         p->done_skip_stat_unmatch = 1;
5627         p->skip_stat_unmatch_result = 0;
5628         /*
5629          * 1. Entries that come from stat info dirtiness
5630          *    always have both sides (iow, not create/delete),
5631          *    one side of the object name is unknown, with
5632          *    the same mode and size.  Keep the ones that
5633          *    do not match these criteria.  They have real
5634          *    differences.
5635          *
5636          * 2. At this point, the file is known to be modified,
5637          *    with the same mode and size, and the object
5638          *    name of one side is unknown.  Need to inspect
5639          *    the identical contents.
5640          */
5641         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5642             !DIFF_FILE_VALID(p->two) ||
5643             (p->one->oid_valid && p->two->oid_valid) ||
5644             (p->one->mode != p->two->mode) ||
5645             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5646             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5647             (p->one->size != p->two->size) ||
5648             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5649                 p->skip_stat_unmatch_result = 1;
5650         return p->skip_stat_unmatch_result;
5651 }
5652
5653 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5654 {
5655         int i;
5656         struct diff_queue_struct *q = &diff_queued_diff;
5657         struct diff_queue_struct outq;
5658         DIFF_QUEUE_CLEAR(&outq);
5659
5660         for (i = 0; i < q->nr; i++) {
5661                 struct diff_filepair *p = q->queue[i];
5662
5663                 if (diff_filespec_check_stat_unmatch(p))
5664                         diff_q(&outq, p);
5665                 else {
5666                         /*
5667                          * The caller can subtract 1 from skip_stat_unmatch
5668                          * to determine how many paths were dirty only
5669                          * due to stat info mismatch.
5670                          */
5671                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
5672                                 diffopt->skip_stat_unmatch++;
5673                         diff_free_filepair(p);
5674                 }
5675         }
5676         free(q->queue);
5677         *q = outq;
5678 }
5679
5680 static int diffnamecmp(const void *a_, const void *b_)
5681 {
5682         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5683         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5684         const char *name_a, *name_b;
5685
5686         name_a = a->one ? a->one->path : a->two->path;
5687         name_b = b->one ? b->one->path : b->two->path;
5688         return strcmp(name_a, name_b);
5689 }
5690
5691 void diffcore_fix_diff_index(struct diff_options *options)
5692 {
5693         struct diff_queue_struct *q = &diff_queued_diff;
5694         QSORT(q->queue, q->nr, diffnamecmp);
5695 }
5696
5697 void diffcore_std(struct diff_options *options)
5698 {
5699         /* NOTE please keep the following in sync with diff_tree_combined() */
5700         if (options->skip_stat_unmatch)
5701                 diffcore_skip_stat_unmatch(options);
5702         if (!options->found_follow) {
5703                 /* See try_to_follow_renames() in tree-diff.c */
5704                 if (options->break_opt != -1)
5705                         diffcore_break(options->break_opt);
5706                 if (options->detect_rename)
5707                         diffcore_rename(options);
5708                 if (options->break_opt != -1)
5709                         diffcore_merge_broken();
5710         }
5711         if (options->pickaxe)
5712                 diffcore_pickaxe(options);
5713         if (options->orderfile)
5714                 diffcore_order(options->orderfile);
5715         if (!options->found_follow)
5716                 /* See try_to_follow_renames() in tree-diff.c */
5717                 diff_resolve_rename_copy();
5718         diffcore_apply_filter(options);
5719
5720         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5721                 DIFF_OPT_SET(options, HAS_CHANGES);
5722         else
5723                 DIFF_OPT_CLR(options, HAS_CHANGES);
5724
5725         options->found_follow = 0;
5726 }
5727
5728 int diff_result_code(struct diff_options *opt, int status)
5729 {
5730         int result = 0;
5731
5732         diff_warn_rename_limit("diff.renameLimit",
5733                                opt->needed_rename_limit,
5734                                opt->degraded_cc_to_c);
5735         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5736             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5737                 return status;
5738         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5739             DIFF_OPT_TST(opt, HAS_CHANGES))
5740                 result |= 01;
5741         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5742             DIFF_OPT_TST(opt, CHECK_FAILED))
5743                 result |= 02;
5744         return result;
5745 }
5746
5747 int diff_can_quit_early(struct diff_options *opt)
5748 {
5749         return (DIFF_OPT_TST(opt, QUICK) &&
5750                 !opt->filter &&
5751                 DIFF_OPT_TST(opt, HAS_CHANGES));
5752 }
5753
5754 /*
5755  * Shall changes to this submodule be ignored?
5756  *
5757  * Submodule changes can be configured to be ignored separately for each path,
5758  * but that configuration can be overridden from the command line.
5759  */
5760 static int is_submodule_ignored(const char *path, struct diff_options *options)
5761 {
5762         int ignored = 0;
5763         unsigned orig_flags = options->flags;
5764         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5765                 set_diffopt_flags_from_submodule_config(options, path);
5766         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5767                 ignored = 1;
5768         options->flags = orig_flags;
5769         return ignored;
5770 }
5771
5772 void diff_addremove(struct diff_options *options,
5773                     int addremove, unsigned mode,
5774                     const struct object_id *oid,
5775                     int oid_valid,
5776                     const char *concatpath, unsigned dirty_submodule)
5777 {
5778         struct diff_filespec *one, *two;
5779
5780         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5781                 return;
5782
5783         /* This may look odd, but it is a preparation for
5784          * feeding "there are unchanged files which should
5785          * not produce diffs, but when you are doing copy
5786          * detection you would need them, so here they are"
5787          * entries to the diff-core.  They will be prefixed
5788          * with something like '=' or '*' (I haven't decided
5789          * which but should not make any difference).
5790          * Feeding the same new and old to diff_change()
5791          * also has the same effect.
5792          * Before the final output happens, they are pruned after
5793          * merged into rename/copy pairs as appropriate.
5794          */
5795         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5796                 addremove = (addremove == '+' ? '-' :
5797                              addremove == '-' ? '+' : addremove);
5798
5799         if (options->prefix &&
5800             strncmp(concatpath, options->prefix, options->prefix_length))
5801                 return;
5802
5803         one = alloc_filespec(concatpath);
5804         two = alloc_filespec(concatpath);
5805
5806         if (addremove != '+')
5807                 fill_filespec(one, oid, oid_valid, mode);
5808         if (addremove != '-') {
5809                 fill_filespec(two, oid, oid_valid, mode);
5810                 two->dirty_submodule = dirty_submodule;
5811         }
5812
5813         diff_queue(&diff_queued_diff, one, two);
5814         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5815                 DIFF_OPT_SET(options, HAS_CHANGES);
5816 }
5817
5818 void diff_change(struct diff_options *options,
5819                  unsigned old_mode, unsigned new_mode,
5820                  const struct object_id *old_oid,
5821                  const struct object_id *new_oid,
5822                  int old_oid_valid, int new_oid_valid,
5823                  const char *concatpath,
5824                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5825 {
5826         struct diff_filespec *one, *two;
5827         struct diff_filepair *p;
5828
5829         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5830             is_submodule_ignored(concatpath, options))
5831                 return;
5832
5833         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5834                 SWAP(old_mode, new_mode);
5835                 SWAP(old_oid, new_oid);
5836                 SWAP(old_oid_valid, new_oid_valid);
5837                 SWAP(old_dirty_submodule, new_dirty_submodule);
5838         }
5839
5840         if (options->prefix &&
5841             strncmp(concatpath, options->prefix, options->prefix_length))
5842                 return;
5843
5844         one = alloc_filespec(concatpath);
5845         two = alloc_filespec(concatpath);
5846         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5847         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5848         one->dirty_submodule = old_dirty_submodule;
5849         two->dirty_submodule = new_dirty_submodule;
5850         p = diff_queue(&diff_queued_diff, one, two);
5851
5852         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5853                 return;
5854
5855         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5856             !diff_filespec_check_stat_unmatch(p))
5857                 return;
5858
5859         DIFF_OPT_SET(options, HAS_CHANGES);
5860 }
5861
5862 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5863 {
5864         struct diff_filepair *pair;
5865         struct diff_filespec *one, *two;
5866
5867         if (options->prefix &&
5868             strncmp(path, options->prefix, options->prefix_length))
5869                 return NULL;
5870
5871         one = alloc_filespec(path);
5872         two = alloc_filespec(path);
5873         pair = diff_queue(&diff_queued_diff, one, two);
5874         pair->is_unmerged = 1;
5875         return pair;
5876 }
5877
5878 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5879                 size_t *outsize)
5880 {
5881         struct diff_tempfile *temp;
5882         const char *argv[3];
5883         const char **arg = argv;
5884         struct child_process child = CHILD_PROCESS_INIT;
5885         struct strbuf buf = STRBUF_INIT;
5886         int err = 0;
5887
5888         temp = prepare_temp_file(spec->path, spec);
5889         *arg++ = pgm;
5890         *arg++ = temp->name;
5891         *arg = NULL;
5892
5893         child.use_shell = 1;
5894         child.argv = argv;
5895         child.out = -1;
5896         if (start_command(&child)) {
5897                 remove_tempfile();
5898                 return NULL;
5899         }
5900
5901         if (strbuf_read(&buf, child.out, 0) < 0)
5902                 err = error("error reading from textconv command '%s'", pgm);
5903         close(child.out);
5904
5905         if (finish_command(&child) || err) {
5906                 strbuf_release(&buf);
5907                 remove_tempfile();
5908                 return NULL;
5909         }
5910         remove_tempfile();
5911
5912         return strbuf_detach(&buf, outsize);
5913 }
5914
5915 size_t fill_textconv(struct userdiff_driver *driver,
5916                      struct diff_filespec *df,
5917                      char **outbuf)
5918 {
5919         size_t size;
5920
5921         if (!driver) {
5922                 if (!DIFF_FILE_VALID(df)) {
5923                         *outbuf = "";
5924                         return 0;
5925                 }
5926                 if (diff_populate_filespec(df, 0))
5927                         die("unable to read files to diff");
5928                 *outbuf = df->data;
5929                 return df->size;
5930         }
5931
5932         if (!driver->textconv)
5933                 die("BUG: fill_textconv called with non-textconv driver");
5934
5935         if (driver->textconv_cache && df->oid_valid) {
5936                 *outbuf = notes_cache_get(driver->textconv_cache,
5937                                           &df->oid,
5938                                           &size);
5939                 if (*outbuf)
5940                         return size;
5941         }
5942
5943         *outbuf = run_textconv(driver->textconv, df, &size);
5944         if (!*outbuf)
5945                 die("unable to read files to diff");
5946
5947         if (driver->textconv_cache && df->oid_valid) {
5948                 /* ignore errors, as we might be in a readonly repository */
5949                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
5950                                 size);
5951                 /*
5952                  * we could save up changes and flush them all at the end,
5953                  * but we would need an extra call after all diffing is done.
5954                  * Since generating a cache entry is the slow path anyway,
5955                  * this extra overhead probably isn't a big deal.
5956                  */
5957                 notes_cache_write(driver->textconv_cache);
5958         }
5959
5960         return size;
5961 }
5962
5963 int textconv_object(const char *path,
5964                     unsigned mode,
5965                     const struct object_id *oid,
5966                     int oid_valid,
5967                     char **buf,
5968                     unsigned long *buf_size)
5969 {
5970         struct diff_filespec *df;
5971         struct userdiff_driver *textconv;
5972
5973         df = alloc_filespec(path);
5974         fill_filespec(df, oid, oid_valid, mode);
5975         textconv = get_textconv(df);
5976         if (!textconv) {
5977                 free_filespec(df);
5978                 return 0;
5979         }
5980
5981         *buf_size = fill_textconv(textconv, df, buf);
5982         free_filespec(df);
5983         return 1;
5984 }
5985
5986 void setup_diff_pager(struct diff_options *opt)
5987 {
5988         /*
5989          * If the user asked for our exit code, then either they want --quiet
5990          * or --exit-code. We should definitely not bother with a pager in the
5991          * former case, as we will generate no output. Since we still properly
5992          * report our exit code even when a pager is run, we _could_ run a
5993          * pager with --exit-code. But since we have not done so historically,
5994          * and because it is easy to find people oneline advising "git diff
5995          * --exit-code" in hooks and other scripts, we do not do so.
5996          */
5997         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5998             check_pager_config("diff") != 0)
5999                 setup_pager();
6000 }