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