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