git.oblomov.eu Git - git/blob - diff.c

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