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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "userdiff.h"
  17 #include "submodule-config.h"
  18 #include "submodule.h"
  19 #include "hashmap.h"
  20 #include "ll-merge.h"
  21 #include "string-list.h"
  22 #include "argv-array.h"
  23 #include "graph.h"
  24
  25 #ifdef NO_FAST_WORKING_DIRECTORY
  26 #define FAST_WORKING_DIRECTORY 0
  27 #else
  28 #define FAST_WORKING_DIRECTORY 1
  29 #endif
  30
  31 static int diff_detect_rename_default;
  32 static int diff_indent_heuristic = 1;
  33 static int diff_rename_limit_default = 400;
  34 static int diff_suppress_blank_empty;
  35 static int diff_use_color_default = -1;
  36 static int diff_color_moved_default;
  37 static int diff_context_default = 3;
  38 static int diff_interhunk_context_default;
  39 static const char *diff_word_regex_cfg;
  40 static const char *external_diff_cmd_cfg;
  41 static const char *diff_order_file_cfg;
  42 int diff_auto_refresh_index = 1;
  43 static int diff_mnemonic_prefix;
  44 static int diff_no_prefix;
  45 static int diff_stat_graph_width;
  46 static int diff_dirstat_permille_default = 30;
  47 static struct diff_options default_diff_options;
  48 static long diff_algorithm;
  49 static unsigned ws_error_highlight_default = WSEH_NEW;
  50
  51 static char diff_colors[][COLOR_MAXLEN] = {
  52         GIT_COLOR_RESET,
  53         GIT_COLOR_NORMAL,       /* CONTEXT */
  54         GIT_COLOR_BOLD,         /* METAINFO */
  55         GIT_COLOR_CYAN,         /* FRAGINFO */
  56         GIT_COLOR_RED,          /* OLD */
  57         GIT_COLOR_GREEN,        /* NEW */
  58         GIT_COLOR_YELLOW,       /* COMMIT */
  59         GIT_COLOR_BG_RED,       /* WHITESPACE */
  60         GIT_COLOR_NORMAL,       /* FUNCINFO */
  61         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  62         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  63         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  64         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  65         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  66         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  69 };
  70
  71 static NORETURN void die_want_option(const char *option_name)
  72 {
  73         die(_("option '%s' requires a value"), option_name);
  74 }
  75
  76 static int parse_diff_color_slot(const char *var)
  77 {
  78         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  79                 return DIFF_CONTEXT;
  80         if (!strcasecmp(var, "meta"))
  81                 return DIFF_METAINFO;
  82         if (!strcasecmp(var, "frag"))
  83                 return DIFF_FRAGINFO;
  84         if (!strcasecmp(var, "old"))
  85                 return DIFF_FILE_OLD;
  86         if (!strcasecmp(var, "new"))
  87                 return DIFF_FILE_NEW;
  88         if (!strcasecmp(var, "commit"))
  89                 return DIFF_COMMIT;
  90         if (!strcasecmp(var, "whitespace"))
  91                 return DIFF_WHITESPACE;
  92         if (!strcasecmp(var, "func"))
  93                 return DIFF_FUNCINFO;
  94         if (!strcasecmp(var, "oldmoved"))
  95                 return DIFF_FILE_OLD_MOVED;
  96         if (!strcasecmp(var, "oldmovedalternative"))
  97                 return DIFF_FILE_OLD_MOVED_ALT;
  98         if (!strcasecmp(var, "oldmoveddimmed"))
  99                 return DIFF_FILE_OLD_MOVED_DIM;
 100         if (!strcasecmp(var, "oldmovedalternativedimmed"))
 101                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
 102         if (!strcasecmp(var, "newmoved"))
 103                 return DIFF_FILE_NEW_MOVED;
 104         if (!strcasecmp(var, "newmovedalternative"))
 105                 return DIFF_FILE_NEW_MOVED_ALT;
 106         if (!strcasecmp(var, "newmoveddimmed"))
 107                 return DIFF_FILE_NEW_MOVED_DIM;
 108         if (!strcasecmp(var, "newmovedalternativedimmed"))
 109                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
 110         return -1;
 111 }
 112
 113 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 114                                 struct strbuf *errmsg)
 115 {
 116         char *params_copy = xstrdup(params_string);
 117         struct string_list params = STRING_LIST_INIT_NODUP;
 118         int ret = 0;
 119         int i;
 120
 121         if (*params_copy)
 122                 string_list_split_in_place(&params, params_copy, ',', -1);
 123         for (i = 0; i < params.nr; i++) {
 124                 const char *p = params.items[i].string;
 125                 if (!strcmp(p, "changes")) {
 126                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 127                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 128                 } else if (!strcmp(p, "lines")) {
 129                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
 130                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 131                 } else if (!strcmp(p, "files")) {
 132                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 133                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 134                 } else if (!strcmp(p, "noncumulative")) {
 135                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 136                 } else if (!strcmp(p, "cumulative")) {
 137                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 138                 } else if (isdigit(*p)) {
 139                         char *end;
 140                         int permille = strtoul(p, &end, 10) * 10;
 141                         if (*end == '.' && isdigit(*++end)) {
 142                                 /* only use first digit */
 143                                 permille += *end - '0';
 144                                 /* .. and ignore any further digits */
 145                                 while (isdigit(*++end))
 146                                         ; /* nothing */
 147                         }
 148                         if (!*end)
 149                                 options->dirstat_permille = permille;
 150                         else {
 151                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 152                                             p);
 153                                 ret++;
 154                         }
 155                 } else {
 156                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 157                         ret++;
 158                 }
 159
 160         }
 161         string_list_clear(&params, 0);
 162         free(params_copy);
 163         return ret;
 164 }
 165
 166 static int parse_submodule_params(struct diff_options *options, const char *value)
 167 {
 168         if (!strcmp(value, "log"))
 169                 options->submodule_format = DIFF_SUBMODULE_LOG;
 170         else if (!strcmp(value, "short"))
 171                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 172         else if (!strcmp(value, "diff"))
 173                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 174         else
 175                 return -1;
 176         return 0;
 177 }
 178
 179 static int git_config_rename(const char *var, const char *value)
 180 {
 181         if (!value)
 182                 return DIFF_DETECT_RENAME;
 183         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 184                 return  DIFF_DETECT_COPY;
 185         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 186 }
 187
 188 long parse_algorithm_value(const char *value)
 189 {
 190         if (!value)
 191                 return -1;
 192         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 193                 return 0;
 194         else if (!strcasecmp(value, "minimal"))
 195                 return XDF_NEED_MINIMAL;
 196         else if (!strcasecmp(value, "patience"))
 197                 return XDF_PATIENCE_DIFF;
 198         else if (!strcasecmp(value, "histogram"))
 199                 return XDF_HISTOGRAM_DIFF;
 200         return -1;
 201 }
 202
 203 static int parse_one_token(const char **arg, const char *token)
 204 {
 205         const char *rest;
 206         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 207                 *arg = rest;
 208                 return 1;
 209         }
 210         return 0;
 211 }
 212
 213 static int parse_ws_error_highlight(const char *arg)
 214 {
 215         const char *orig_arg = arg;
 216         unsigned val = 0;
 217
 218         while (*arg) {
 219                 if (parse_one_token(&arg, "none"))
 220                         val = 0;
 221                 else if (parse_one_token(&arg, "default"))
 222                         val = WSEH_NEW;
 223                 else if (parse_one_token(&arg, "all"))
 224                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 225                 else if (parse_one_token(&arg, "new"))
 226                         val |= WSEH_NEW;
 227                 else if (parse_one_token(&arg, "old"))
 228                         val |= WSEH_OLD;
 229                 else if (parse_one_token(&arg, "context"))
 230                         val |= WSEH_CONTEXT;
 231                 else {
 232                         return -1 - (int)(arg - orig_arg);
 233                 }
 234                 if (*arg)
 235                         arg++;
 236         }
 237         return val;
 238 }
 239
 240 /*
 241  * These are to give UI layer defaults.
 242  * The core-level commands such as git-diff-files should
 243  * never be affected by the setting of diff.renames
 244  * the user happens to have in the configuration file.
 245  */
 246 void init_diff_ui_defaults(void)
 247 {
 248         diff_detect_rename_default = 1;
 249 }
 250
 251 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 252 {
 253         if (!strcmp(var, "diff.indentheuristic"))
 254                 diff_indent_heuristic = git_config_bool(var, value);
 255         return 0;
 256 }
 257
 258 static int parse_color_moved(const char *arg)
 259 {
 260         switch (git_parse_maybe_bool(arg)) {
 261         case 0:
 262                 return COLOR_MOVED_NO;
 263         case 1:
 264                 return COLOR_MOVED_DEFAULT;
 265         default:
 266                 break;
 267         }
 268
 269         if (!strcmp(arg, "no"))
 270                 return COLOR_MOVED_NO;
 271         else if (!strcmp(arg, "plain"))
 272                 return COLOR_MOVED_PLAIN;
 273         else if (!strcmp(arg, "zebra"))
 274                 return COLOR_MOVED_ZEBRA;
 275         else if (!strcmp(arg, "default"))
 276                 return COLOR_MOVED_DEFAULT;
 277         else if (!strcmp(arg, "dimmed_zebra"))
 278                 return COLOR_MOVED_ZEBRA_DIM;
 279         else
 280                 return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'dimmed_zebra', 'plain'"));
 281 }
 282
 283 int git_diff_ui_config(const char *var, const char *value, void *cb)
 284 {
 285         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 286                 diff_use_color_default = git_config_colorbool(var, value);
 287                 return 0;
 288         }
 289         if (!strcmp(var, "diff.colormoved")) {
 290                 int cm = parse_color_moved(value);
 291                 if (cm < 0)
 292                         return -1;
 293                 diff_color_moved_default = cm;
 294                 return 0;
 295         }
 296         if (!strcmp(var, "diff.context")) {
 297                 diff_context_default = git_config_int(var, value);
 298                 if (diff_context_default < 0)
 299                         return -1;
 300                 return 0;
 301         }
 302         if (!strcmp(var, "diff.interhunkcontext")) {
 303                 diff_interhunk_context_default = git_config_int(var, value);
 304                 if (diff_interhunk_context_default < 0)
 305                         return -1;
 306                 return 0;
 307         }
 308         if (!strcmp(var, "diff.renames")) {
 309                 diff_detect_rename_default = git_config_rename(var, value);
 310                 return 0;
 311         }
 312         if (!strcmp(var, "diff.autorefreshindex")) {
 313                 diff_auto_refresh_index = git_config_bool(var, value);
 314                 return 0;
 315         }
 316         if (!strcmp(var, "diff.mnemonicprefix")) {
 317                 diff_mnemonic_prefix = git_config_bool(var, value);
 318                 return 0;
 319         }
 320         if (!strcmp(var, "diff.noprefix")) {
 321                 diff_no_prefix = git_config_bool(var, value);
 322                 return 0;
 323         }
 324         if (!strcmp(var, "diff.statgraphwidth")) {
 325                 diff_stat_graph_width = git_config_int(var, value);
 326                 return 0;
 327         }
 328         if (!strcmp(var, "diff.external"))
 329                 return git_config_string(&external_diff_cmd_cfg, var, value);
 330         if (!strcmp(var, "diff.wordregex"))
 331                 return git_config_string(&diff_word_regex_cfg, var, value);
 332         if (!strcmp(var, "diff.orderfile"))
 333                 return git_config_pathname(&diff_order_file_cfg, var, value);
 334
 335         if (!strcmp(var, "diff.ignoresubmodules"))
 336                 handle_ignore_submodules_arg(&default_diff_options, value);
 337
 338         if (!strcmp(var, "diff.submodule")) {
 339                 if (parse_submodule_params(&default_diff_options, value))
 340                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 341                                 value);
 342                 return 0;
 343         }
 344
 345         if (!strcmp(var, "diff.algorithm")) {
 346                 diff_algorithm = parse_algorithm_value(value);
 347                 if (diff_algorithm < 0)
 348                         return -1;
 349                 return 0;
 350         }
 351
 352         if (!strcmp(var, "diff.wserrorhighlight")) {
 353                 int val = parse_ws_error_highlight(value);
 354                 if (val < 0)
 355                         return -1;
 356                 ws_error_highlight_default = val;
 357                 return 0;
 358         }
 359
 360         if (git_color_config(var, value, cb) < 0)
 361                 return -1;
 362
 363         return git_diff_basic_config(var, value, cb);
 364 }
 365
 366 int git_diff_basic_config(const char *var, const char *value, void *cb)
 367 {
 368         const char *name;
 369
 370         if (!strcmp(var, "diff.renamelimit")) {
 371                 diff_rename_limit_default = git_config_int(var, value);
 372                 return 0;
 373         }
 374
 375         if (userdiff_config(var, value) < 0)
 376                 return -1;
 377
 378         if (skip_prefix(var, "diff.color.", &name) ||
 379             skip_prefix(var, "color.diff.", &name)) {
 380                 int slot = parse_diff_color_slot(name);
 381                 if (slot < 0)
 382                         return 0;
 383                 if (!value)
 384                         return config_error_nonbool(var);
 385                 return color_parse(value, diff_colors[slot]);
 386         }
 387
 388         /* like GNU diff's --suppress-blank-empty option  */
 389         if (!strcmp(var, "diff.suppressblankempty") ||
 390                         /* for backwards compatibility */
 391                         !strcmp(var, "diff.suppress-blank-empty")) {
 392                 diff_suppress_blank_empty = git_config_bool(var, value);
 393                 return 0;
 394         }
 395
 396         if (!strcmp(var, "diff.dirstat")) {
 397                 struct strbuf errmsg = STRBUF_INIT;
 398                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 399                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 400                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 401                                 errmsg.buf);
 402                 strbuf_release(&errmsg);
 403                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 404                 return 0;
 405         }
 406
 407         if (starts_with(var, "submodule."))
 408                 return parse_submodule_config_option(var, value);
 409
 410         if (git_diff_heuristic_config(var, value, cb) < 0)
 411                 return -1;
 412
 413         return git_default_config(var, value, cb);
 414 }
 415
 416 static char *quote_two(const char *one, const char *two)
 417 {
 418         int need_one = quote_c_style(one, NULL, NULL, 1);
 419         int need_two = quote_c_style(two, NULL, NULL, 1);
 420         struct strbuf res = STRBUF_INIT;
 421
 422         if (need_one + need_two) {
 423                 strbuf_addch(&res, '"');
 424                 quote_c_style(one, &res, NULL, 1);
 425                 quote_c_style(two, &res, NULL, 1);
 426                 strbuf_addch(&res, '"');
 427         } else {
 428                 strbuf_addstr(&res, one);
 429                 strbuf_addstr(&res, two);
 430         }
 431         return strbuf_detach(&res, NULL);
 432 }
 433
 434 static const char *external_diff(void)
 435 {
 436         static const char *external_diff_cmd = NULL;
 437         static int done_preparing = 0;
 438
 439         if (done_preparing)
 440                 return external_diff_cmd;
 441         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 442         if (!external_diff_cmd)
 443                 external_diff_cmd = external_diff_cmd_cfg;
 444         done_preparing = 1;
 445         return external_diff_cmd;
 446 }
 447
 448 /*
 449  * Keep track of files used for diffing. Sometimes such an entry
 450  * refers to a temporary file, sometimes to an existing file, and
 451  * sometimes to "/dev/null".
 452  */
 453 static struct diff_tempfile {
 454         /*
 455          * filename external diff should read from, or NULL if this
 456          * entry is currently not in use:
 457          */
 458         const char *name;
 459
 460         char hex[GIT_MAX_HEXSZ + 1];
 461         char mode[10];
 462
 463         /*
 464          * If this diff_tempfile instance refers to a temporary file,
 465          * this tempfile object is used to manage its lifetime.
 466          */
 467         struct tempfile tempfile;
 468 } diff_temp[2];
 469
 470 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 471
 472 struct emit_callback {
 473         int color_diff;
 474         unsigned ws_rule;
 475         int blank_at_eof_in_preimage;
 476         int blank_at_eof_in_postimage;
 477         int lno_in_preimage;
 478         int lno_in_postimage;
 479         sane_truncate_fn truncate;
 480         const char **label_path;
 481         struct diff_words_data *diff_words;
 482         struct diff_options *opt;
 483         struct strbuf *header;
 484 };
 485
 486 static int count_lines(const char *data, int size)
 487 {
 488         int count, ch, completely_empty = 1, nl_just_seen = 0;
 489         count = 0;
 490         while (0 < size--) {
 491                 ch = *data++;
 492                 if (ch == '\n') {
 493                         count++;
 494                         nl_just_seen = 1;
 495                         completely_empty = 0;
 496                 }
 497                 else {
 498                         nl_just_seen = 0;
 499                         completely_empty = 0;
 500                 }
 501         }
 502         if (completely_empty)
 503                 return 0;
 504         if (!nl_just_seen)
 505                 count++; /* no trailing newline */
 506         return count;
 507 }
 508
 509 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 510 {
 511         if (!DIFF_FILE_VALID(one)) {
 512                 mf->ptr = (char *)""; /* does not matter */
 513                 mf->size = 0;
 514                 return 0;
 515         }
 516         else if (diff_populate_filespec(one, 0))
 517                 return -1;
 518
 519         mf->ptr = one->data;
 520         mf->size = one->size;
 521         return 0;
 522 }
 523
 524 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 525 static unsigned long diff_filespec_size(struct diff_filespec *one)
 526 {
 527         if (!DIFF_FILE_VALID(one))
 528                 return 0;
 529         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 530         return one->size;
 531 }
 532
 533 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 534 {
 535         char *ptr = mf->ptr;
 536         long size = mf->size;
 537         int cnt = 0;
 538
 539         if (!size)
 540                 return cnt;
 541         ptr += size - 1; /* pointing at the very end */
 542         if (*ptr != '\n')
 543                 ; /* incomplete line */
 544         else
 545                 ptr--; /* skip the last LF */
 546         while (mf->ptr < ptr) {
 547                 char *prev_eol;
 548                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 549                         if (*prev_eol == '\n')
 550                                 break;
 551                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 552                         break;
 553                 cnt++;
 554                 ptr = prev_eol - 1;
 555         }
 556         return cnt;
 557 }
 558
 559 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 560                                struct emit_callback *ecbdata)
 561 {
 562         int l1, l2, at;
 563         unsigned ws_rule = ecbdata->ws_rule;
 564         l1 = count_trailing_blank(mf1, ws_rule);
 565         l2 = count_trailing_blank(mf2, ws_rule);
 566         if (l2 <= l1) {
 567                 ecbdata->blank_at_eof_in_preimage = 0;
 568                 ecbdata->blank_at_eof_in_postimage = 0;
 569                 return;
 570         }
 571         at = count_lines(mf1->ptr, mf1->size);
 572         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 573
 574         at = count_lines(mf2->ptr, mf2->size);
 575         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 576 }
 577
 578 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 579                         int first, const char *line, int len)
 580 {
 581         int has_trailing_newline, has_trailing_carriage_return;
 582         int nofirst;
 583         FILE *file = o->file;
 584
 585         fputs(diff_line_prefix(o), file);
 586
 587         if (len == 0) {
 588                 has_trailing_newline = (first == '\n');
 589                 has_trailing_carriage_return = (!has_trailing_newline &&
 590                                                 (first == '\r'));
 591                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 592         } else {
 593                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 594                 if (has_trailing_newline)
 595                         len--;
 596                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 597                 if (has_trailing_carriage_return)
 598                         len--;
 599                 nofirst = 0;
 600         }
 601
 602         if (len || !nofirst) {
 603                 fputs(set, file);
 604                 if (!nofirst)
 605                         fputc(first, file);
 606                 fwrite(line, len, 1, file);
 607                 fputs(reset, file);
 608         }
 609         if (has_trailing_carriage_return)
 610                 fputc('\r', file);
 611         if (has_trailing_newline)
 612                 fputc('\n', file);
 613 }
 614
 615 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 616                       const char *line, int len)
 617 {
 618         emit_line_0(o, set, reset, line[0], line+1, len-1);
 619 }
 620
 621 enum diff_symbol {
 622         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 623         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 624         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 625         DIFF_SYMBOL_BINARY_DIFF_BODY,
 626         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 627         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 628         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 629         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 630         DIFF_SYMBOL_STATS_LINE,
 631         DIFF_SYMBOL_WORD_DIFF,
 632         DIFF_SYMBOL_STAT_SEP,
 633         DIFF_SYMBOL_SUMMARY,
 634         DIFF_SYMBOL_SUBMODULE_ADD,
 635         DIFF_SYMBOL_SUBMODULE_DEL,
 636         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 637         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 638         DIFF_SYMBOL_SUBMODULE_HEADER,
 639         DIFF_SYMBOL_SUBMODULE_ERROR,
 640         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 641         DIFF_SYMBOL_REWRITE_DIFF,
 642         DIFF_SYMBOL_BINARY_FILES,
 643         DIFF_SYMBOL_HEADER,
 644         DIFF_SYMBOL_FILEPAIR_PLUS,
 645         DIFF_SYMBOL_FILEPAIR_MINUS,
 646         DIFF_SYMBOL_WORDS_PORCELAIN,
 647         DIFF_SYMBOL_WORDS,
 648         DIFF_SYMBOL_CONTEXT,
 649         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 650         DIFF_SYMBOL_PLUS,
 651         DIFF_SYMBOL_MINUS,
 652         DIFF_SYMBOL_NO_LF_EOF,
 653         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 654         DIFF_SYMBOL_CONTEXT_MARKER,
 655         DIFF_SYMBOL_SEPARATOR
 656 };
 657 /*
 658  * Flags for content lines:
 659  * 0..12 are whitespace rules
 660  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 661  * 16 is marking if the line is blank at EOF
 662  */
 663 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 664 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 665 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 666 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 667 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 668
 669 /*
 670  * This struct is used when we need to buffer the output of the diff output.
 671  *
 672  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 673  * into the pre/post image file. This pointer could be a union with the
 674  * line pointer. By storing an offset into the file instead of the literal line,
 675  * we can decrease the memory footprint for the buffered output. At first we
 676  * may want to only have indirection for the content lines, but we could also
 677  * enhance the state for emitting prefabricated lines, e.g. the similarity
 678  * score line or hunk/file headers would only need to store a number or path
 679  * and then the output can be constructed later on depending on state.
 680  */
 681 struct emitted_diff_symbol {
 682         const char *line;
 683         int len;
 684         int flags;
 685         enum diff_symbol s;
 686 };
 687 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 688
 689 struct emitted_diff_symbols {
 690         struct emitted_diff_symbol *buf;
 691         int nr, alloc;
 692 };
 693 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 694
 695 static void append_emitted_diff_symbol(struct diff_options *o,
 696                                        struct emitted_diff_symbol *e)
 697 {
 698         struct emitted_diff_symbol *f;
 699
 700         ALLOC_GROW(o->emitted_symbols->buf,
 701                    o->emitted_symbols->nr + 1,
 702                    o->emitted_symbols->alloc);
 703         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 704
 705         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 706         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 707 }
 708
 709 struct moved_entry {
 710         struct hashmap_entry ent;
 711         const struct emitted_diff_symbol *es;
 712         struct moved_entry *next_line;
 713 };
 714
 715 static int next_byte(const char **cp, const char **endp,
 716                      const struct diff_options *diffopt)
 717 {
 718         int retval;
 719
 720         if (*cp > *endp)
 721                 return -1;
 722
 723         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_CHANGE)) {
 724                 while (*cp < *endp && isspace(**cp))
 725                         (*cp)++;
 726                 /*
 727                  * After skipping a couple of whitespaces, we still have to
 728                  * account for one space.
 729                  */
 730                 return (int)' ';
 731         }
 732
 733         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE)) {
 734                 while (*cp < *endp && isspace(**cp))
 735                         (*cp)++;
 736                 /* return the first non-ws character via the usual below */
 737         }
 738
 739         retval = (unsigned char)(**cp);
 740         (*cp)++;
 741         return retval;
 742 }
 743
 744 static int moved_entry_cmp(const struct diff_options *diffopt,
 745                            const struct moved_entry *a,
 746                            const struct moved_entry *b,
 747                            const void *keydata)
 748 {
 749         const char *ap = a->es->line, *ae = a->es->line + a->es->len;
 750         const char *bp = b->es->line, *be = b->es->line + b->es->len;
 751
 752         if (!(diffopt->xdl_opts & XDF_WHITESPACE_FLAGS))
 753                 return a->es->len != b->es->len  || memcmp(ap, bp, a->es->len);
 754
 755         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_AT_EOL)) {
 756                 while (ae > ap && isspace(*ae))
 757                         ae--;
 758                 while (be > bp && isspace(*be))
 759                         be--;
 760         }
 761
 762         while (1) {
 763                 int ca, cb;
 764                 ca = next_byte(&ap, &ae, diffopt);
 765                 cb = next_byte(&bp, &be, diffopt);
 766                 if (ca != cb)
 767                         return 1;
 768                 if (ca < 0)
 769                         return 0;
 770         }
 771 }
 772
 773 static unsigned get_string_hash(struct emitted_diff_symbol *es, struct diff_options *o)
 774 {
 775         if (o->xdl_opts & XDF_WHITESPACE_FLAGS) {
 776                 static struct strbuf sb = STRBUF_INIT;
 777                 const char *ap = es->line, *ae = es->line + es->len;
 778                 int c;
 779
 780                 strbuf_reset(&sb);
 781                 while (ae > ap && isspace(*ae))
 782                         ae--;
 783                 while ((c = next_byte(&ap, &ae, o)) > 0)
 784                         strbuf_addch(&sb, c);
 785
 786                 return memhash(sb.buf, sb.len);
 787         } else {
 788                 return memhash(es->line, es->len);
 789         }
 790 }
 791
 792 static struct moved_entry *prepare_entry(struct diff_options *o,
 793                                          int line_no)
 794 {
 795         struct moved_entry *ret = xmalloc(sizeof(*ret));
 796         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 797
 798         ret->ent.hash = get_string_hash(l, o);
 799         ret->es = l;
 800         ret->next_line = NULL;
 801
 802         return ret;
 803 }
 804
 805 static void add_lines_to_move_detection(struct diff_options *o,
 806                                         struct hashmap *add_lines,
 807                                         struct hashmap *del_lines)
 808 {
 809         struct moved_entry *prev_line = NULL;
 810
 811         int n;
 812         for (n = 0; n < o->emitted_symbols->nr; n++) {
 813                 struct hashmap *hm;
 814                 struct moved_entry *key;
 815
 816                 switch (o->emitted_symbols->buf[n].s) {
 817                 case DIFF_SYMBOL_PLUS:
 818                         hm = add_lines;
 819                         break;
 820                 case DIFF_SYMBOL_MINUS:
 821                         hm = del_lines;
 822                         break;
 823                 default:
 824                         prev_line = NULL;
 825                         continue;
 826                 }
 827
 828                 key = prepare_entry(o, n);
 829                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 830                         prev_line->next_line = key;
 831
 832                 hashmap_add(hm, key);
 833                 prev_line = key;
 834         }
 835 }
 836
 837 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 838                                          int pmb_nr)
 839 {
 840         int lp, rp;
 841
 842         /* Shrink the set of potential block to the remaining running */
 843         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 844                 while (lp < pmb_nr && pmb[lp])
 845                         lp++;
 846                 /* lp points at the first NULL now */
 847
 848                 while (rp > -1 && !pmb[rp])
 849                         rp--;
 850                 /* rp points at the last non-NULL */
 851
 852                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 853                         pmb[lp] = pmb[rp];
 854                         pmb[rp] = NULL;
 855                         rp--;
 856                         lp++;
 857                 }
 858         }
 859
 860         /* Remember the number of running sets */
 861         return rp + 1;
 862 }
 863
 864 /*
 865  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 866  *
 867  * Otherwise, if the last block has fewer lines than
 868  * COLOR_MOVED_MIN_BLOCK_LENGTH, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 869  * that block.
 870  *
 871  * The last block consists of the (n - block_length)'th line up to but not
 872  * including the nth line.
 873  */
 874 static void adjust_last_block(struct diff_options *o, int n, int block_length)
 875 {
 876         int i;
 877         if (block_length >= COLOR_MOVED_MIN_BLOCK_LENGTH ||
 878             o->color_moved == COLOR_MOVED_PLAIN)
 879                 return;
 880         for (i = 1; i < block_length + 1; i++)
 881                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
 882 }
 883
 884 /* Find blocks of moved code, delegate actual coloring decision to helper */
 885 static void mark_color_as_moved(struct diff_options *o,
 886                                 struct hashmap *add_lines,
 887                                 struct hashmap *del_lines)
 888 {
 889         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 890         int pmb_nr = 0, pmb_alloc = 0;
 891         int n, flipped_block = 1, block_length = 0;
 892
 893
 894         for (n = 0; n < o->emitted_symbols->nr; n++) {
 895                 struct hashmap *hm = NULL;
 896                 struct moved_entry *key;
 897                 struct moved_entry *match = NULL;
 898                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 899                 int i;
 900
 901                 switch (l->s) {
 902                 case DIFF_SYMBOL_PLUS:
 903                         hm = del_lines;
 904                         key = prepare_entry(o, n);
 905                         match = hashmap_get(hm, key, o);
 906                         free(key);
 907                         break;
 908                 case DIFF_SYMBOL_MINUS:
 909                         hm = add_lines;
 910                         key = prepare_entry(o, n);
 911                         match = hashmap_get(hm, key, o);
 912                         free(key);
 913                         break;
 914                 default:
 915                         flipped_block = 1;
 916                 }
 917
 918                 if (!match) {
 919                         adjust_last_block(o, n, block_length);
 920                         pmb_nr = 0;
 921                         block_length = 0;
 922                         continue;
 923                 }
 924
 925                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 926                 block_length++;
 927
 928                 if (o->color_moved == COLOR_MOVED_PLAIN)
 929                         continue;
 930
 931                 /* Check any potential block runs, advance each or nullify */
 932                 for (i = 0; i < pmb_nr; i++) {
 933                         struct moved_entry *p = pmb[i];
 934                         struct moved_entry *pnext = (p && p->next_line) ?
 935                                         p->next_line : NULL;
 936                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 937                                 pmb[i] = p->next_line;
 938                         } else {
 939                                 pmb[i] = NULL;
 940                         }
 941                 }
 942
 943                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 944
 945                 if (pmb_nr == 0) {
 946                         /*
 947                          * The current line is the start of a new block.
 948                          * Setup the set of potential blocks.
 949                          */
 950                         for (; match; match = hashmap_get_next(hm, match)) {
 951                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 952                                 pmb[pmb_nr++] = match;
 953                         }
 954
 955                         flipped_block = (flipped_block + 1) % 2;
 956                 }
 957
 958                 if (flipped_block)
 959                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 960         }
 961         adjust_last_block(o, n, block_length);
 962
 963         free(pmb);
 964 }
 965
 966 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
 967   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
 968 static void dim_moved_lines(struct diff_options *o)
 969 {
 970         int n;
 971         for (n = 0; n < o->emitted_symbols->nr; n++) {
 972                 struct emitted_diff_symbol *prev = (n != 0) ?
 973                                 &o->emitted_symbols->buf[n - 1] : NULL;
 974                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 975                 struct emitted_diff_symbol *next =
 976                                 (n < o->emitted_symbols->nr - 1) ?
 977                                 &o->emitted_symbols->buf[n + 1] : NULL;
 978
 979                 /* Not a plus or minus line? */
 980                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
 981                         continue;
 982
 983                 /* Not a moved line? */
 984                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
 985                         continue;
 986
 987                 /*
 988                  * If prev or next are not a plus or minus line,
 989                  * pretend they don't exist
 990                  */
 991                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
 992                             prev->s != DIFF_SYMBOL_MINUS)
 993                         prev = NULL;
 994                 if (next && next->s != DIFF_SYMBOL_PLUS &&
 995                             next->s != DIFF_SYMBOL_MINUS)
 996                         next = NULL;
 997
 998                 /* Inside a block? */
 999                 if ((prev &&
1000                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1001                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1002                     (next &&
1003                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1004                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1005                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1006                         continue;
1007                 }
1008
1009                 /* Check if we are at an interesting bound: */
1010                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1011                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1012                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1013                         continue;
1014                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1015                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1016                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1017                         continue;
1018
1019                 /*
1020                  * The boundary to prev and next are not interesting,
1021                  * so this line is not interesting as a whole
1022                  */
1023                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1024         }
1025 }
1026
1027 static void emit_line_ws_markup(struct diff_options *o,
1028                                 const char *set, const char *reset,
1029                                 const char *line, int len, char sign,
1030                                 unsigned ws_rule, int blank_at_eof)
1031 {
1032         const char *ws = NULL;
1033
1034         if (o->ws_error_highlight & ws_rule) {
1035                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1036                 if (!*ws)
1037                         ws = NULL;
1038         }
1039
1040         if (!ws)
1041                 emit_line_0(o, set, reset, sign, line, len);
1042         else if (blank_at_eof)
1043                 /* Blank line at EOF - paint '+' as well */
1044                 emit_line_0(o, ws, reset, sign, line, len);
1045         else {
1046                 /* Emit just the prefix, then the rest. */
1047                 emit_line_0(o, set, reset, sign, "", 0);
1048                 ws_check_emit(line, len, ws_rule,
1049                               o->file, set, reset, ws);
1050         }
1051 }
1052
1053 static void emit_diff_symbol_from_struct(struct diff_options *o,
1054                                          struct emitted_diff_symbol *eds)
1055 {
1056         static const char *nneof = " No newline at end of file\n";
1057         const char *context, *reset, *set, *meta, *fraginfo;
1058         struct strbuf sb = STRBUF_INIT;
1059
1060         enum diff_symbol s = eds->s;
1061         const char *line = eds->line;
1062         int len = eds->len;
1063         unsigned flags = eds->flags;
1064
1065         switch (s) {
1066         case DIFF_SYMBOL_NO_LF_EOF:
1067                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1068                 reset = diff_get_color_opt(o, DIFF_RESET);
1069                 putc('\n', o->file);
1070                 emit_line_0(o, context, reset, '\\',
1071                             nneof, strlen(nneof));
1072                 break;
1073         case DIFF_SYMBOL_SUBMODULE_HEADER:
1074         case DIFF_SYMBOL_SUBMODULE_ERROR:
1075         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1076         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1077         case DIFF_SYMBOL_SUMMARY:
1078         case DIFF_SYMBOL_STATS_LINE:
1079         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1080         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1081                 emit_line(o, "", "", line, len);
1082                 break;
1083         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1084         case DIFF_SYMBOL_CONTEXT_MARKER:
1085                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1086                 reset = diff_get_color_opt(o, DIFF_RESET);
1087                 emit_line(o, context, reset, line, len);
1088                 break;
1089         case DIFF_SYMBOL_SEPARATOR:
1090                 fprintf(o->file, "%s%c",
1091                         diff_line_prefix(o),
1092                         o->line_termination);
1093                 break;
1094         case DIFF_SYMBOL_CONTEXT:
1095                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1096                 reset = diff_get_color_opt(o, DIFF_RESET);
1097                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1098                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1099                 break;
1100         case DIFF_SYMBOL_PLUS:
1101                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1102                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1103                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1104                 case DIFF_SYMBOL_MOVED_LINE |
1105                      DIFF_SYMBOL_MOVED_LINE_ALT |
1106                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1107                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1108                         break;
1109                 case DIFF_SYMBOL_MOVED_LINE |
1110                      DIFF_SYMBOL_MOVED_LINE_ALT:
1111                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1112                         break;
1113                 case DIFF_SYMBOL_MOVED_LINE |
1114                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1115                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1116                         break;
1117                 case DIFF_SYMBOL_MOVED_LINE:
1118                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1119                         break;
1120                 default:
1121                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1122                 }
1123                 reset = diff_get_color_opt(o, DIFF_RESET);
1124                 emit_line_ws_markup(o, set, reset, line, len, '+',
1125                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1126                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1127                 break;
1128         case DIFF_SYMBOL_MINUS:
1129                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1130                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1131                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1132                 case DIFF_SYMBOL_MOVED_LINE |
1133                      DIFF_SYMBOL_MOVED_LINE_ALT |
1134                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1135                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1136                         break;
1137                 case DIFF_SYMBOL_MOVED_LINE |
1138                      DIFF_SYMBOL_MOVED_LINE_ALT:
1139                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1140                         break;
1141                 case DIFF_SYMBOL_MOVED_LINE |
1142                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1143                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1144                         break;
1145                 case DIFF_SYMBOL_MOVED_LINE:
1146                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1147                         break;
1148                 default:
1149                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1150                 }
1151                 reset = diff_get_color_opt(o, DIFF_RESET);
1152                 emit_line_ws_markup(o, set, reset, line, len, '-',
1153                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1154                 break;
1155         case DIFF_SYMBOL_WORDS_PORCELAIN:
1156                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1157                 reset = diff_get_color_opt(o, DIFF_RESET);
1158                 emit_line(o, context, reset, line, len);
1159                 fputs("~\n", o->file);
1160                 break;
1161         case DIFF_SYMBOL_WORDS:
1162                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1163                 reset = diff_get_color_opt(o, DIFF_RESET);
1164                 /*
1165                  * Skip the prefix character, if any.  With
1166                  * diff_suppress_blank_empty, there may be
1167                  * none.
1168                  */
1169                 if (line[0] != '\n') {
1170                         line++;
1171                         len--;
1172                 }
1173                 emit_line(o, context, reset, line, len);
1174                 break;
1175         case DIFF_SYMBOL_FILEPAIR_PLUS:
1176                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1177                 reset = diff_get_color_opt(o, DIFF_RESET);
1178                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1179                         line, reset,
1180                         strchr(line, ' ') ? "\t" : "");
1181                 break;
1182         case DIFF_SYMBOL_FILEPAIR_MINUS:
1183                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1184                 reset = diff_get_color_opt(o, DIFF_RESET);
1185                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1186                         line, reset,
1187                         strchr(line, ' ') ? "\t" : "");
1188                 break;
1189         case DIFF_SYMBOL_BINARY_FILES:
1190         case DIFF_SYMBOL_HEADER:
1191                 fprintf(o->file, "%s", line);
1192                 break;
1193         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1194                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1195                 break;
1196         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1197                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1198                 break;
1199         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1200                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1201                 break;
1202         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1203                 fputs(diff_line_prefix(o), o->file);
1204                 fputc('\n', o->file);
1205                 break;
1206         case DIFF_SYMBOL_REWRITE_DIFF:
1207                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1208                 reset = diff_get_color_opt(o, DIFF_RESET);
1209                 emit_line(o, fraginfo, reset, line, len);
1210                 break;
1211         case DIFF_SYMBOL_SUBMODULE_ADD:
1212                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1213                 reset = diff_get_color_opt(o, DIFF_RESET);
1214                 emit_line(o, set, reset, line, len);
1215                 break;
1216         case DIFF_SYMBOL_SUBMODULE_DEL:
1217                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1218                 reset = diff_get_color_opt(o, DIFF_RESET);
1219                 emit_line(o, set, reset, line, len);
1220                 break;
1221         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1222                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1223                         diff_line_prefix(o), line);
1224                 break;
1225         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1226                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1227                         diff_line_prefix(o), line);
1228                 break;
1229         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1230                 emit_line(o, "", "", " 0 files changed\n",
1231                           strlen(" 0 files changed\n"));
1232                 break;
1233         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1234                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1235                 break;
1236         case DIFF_SYMBOL_WORD_DIFF:
1237                 fprintf(o->file, "%.*s", len, line);
1238                 break;
1239         case DIFF_SYMBOL_STAT_SEP:
1240                 fputs(o->stat_sep, o->file);
1241                 break;
1242         default:
1243                 die("BUG: unknown diff symbol");
1244         }
1245         strbuf_release(&sb);
1246 }
1247
1248 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1249                              const char *line, int len, unsigned flags)
1250 {
1251         struct emitted_diff_symbol e = {line, len, flags, s};
1252
1253         if (o->emitted_symbols)
1254                 append_emitted_diff_symbol(o, &e);
1255         else
1256                 emit_diff_symbol_from_struct(o, &e);
1257 }
1258
1259 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1260 {
1261         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1262 }
1263
1264 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1265 {
1266         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1267 }
1268
1269 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1270 {
1271         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1272                          path, strlen(path), 0);
1273 }
1274
1275 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1276 {
1277         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1278                          path, strlen(path), 0);
1279 }
1280
1281 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1282 {
1283         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1284                          header, strlen(header), 0);
1285 }
1286
1287 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1288 {
1289         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1290 }
1291
1292 void diff_emit_submodule_pipethrough(struct diff_options *o,
1293                                      const char *line, int len)
1294 {
1295         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1296 }
1297
1298 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1299 {
1300         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1301               ecbdata->blank_at_eof_in_preimage &&
1302               ecbdata->blank_at_eof_in_postimage &&
1303               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1304               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1305                 return 0;
1306         return ws_blank_line(line, len, ecbdata->ws_rule);
1307 }
1308
1309 static void emit_add_line(const char *reset,
1310                           struct emit_callback *ecbdata,
1311                           const char *line, int len)
1312 {
1313         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1314         if (new_blank_line_at_eof(ecbdata, line, len))
1315                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1316
1317         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1318 }
1319
1320 static void emit_del_line(const char *reset,
1321                           struct emit_callback *ecbdata,
1322                           const char *line, int len)
1323 {
1324         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1325         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1326 }
1327
1328 static void emit_context_line(const char *reset,
1329                               struct emit_callback *ecbdata,
1330                               const char *line, int len)
1331 {
1332         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1333         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1334 }
1335
1336 static void emit_hunk_header(struct emit_callback *ecbdata,
1337                              const char *line, int len)
1338 {
1339         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1340         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1341         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1342         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1343         static const char atat[2] = { '@', '@' };
1344         const char *cp, *ep;
1345         struct strbuf msgbuf = STRBUF_INIT;
1346         int org_len = len;
1347         int i = 1;
1348
1349         /*
1350          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1351          * it always is at least 10 bytes long.
1352          */
1353         if (len < 10 ||
1354             memcmp(line, atat, 2) ||
1355             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1356                 emit_diff_symbol(ecbdata->opt,
1357                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1358                 return;
1359         }
1360         ep += 2; /* skip over @@ */
1361
1362         /* The hunk header in fraginfo color */
1363         strbuf_addstr(&msgbuf, frag);
1364         strbuf_add(&msgbuf, line, ep - line);
1365         strbuf_addstr(&msgbuf, reset);
1366
1367         /*
1368          * trailing "\r\n"
1369          */
1370         for ( ; i < 3; i++)
1371                 if (line[len - i] == '\r' || line[len - i] == '\n')
1372                         len--;
1373
1374         /* blank before the func header */
1375         for (cp = ep; ep - line < len; ep++)
1376                 if (*ep != ' ' && *ep != '\t')
1377                         break;
1378         if (ep != cp) {
1379                 strbuf_addstr(&msgbuf, context);
1380                 strbuf_add(&msgbuf, cp, ep - cp);
1381                 strbuf_addstr(&msgbuf, reset);
1382         }
1383
1384         if (ep < line + len) {
1385                 strbuf_addstr(&msgbuf, func);
1386                 strbuf_add(&msgbuf, ep, line + len - ep);
1387                 strbuf_addstr(&msgbuf, reset);
1388         }
1389
1390         strbuf_add(&msgbuf, line + len, org_len - len);
1391         strbuf_complete_line(&msgbuf);
1392         emit_diff_symbol(ecbdata->opt,
1393                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1394         strbuf_release(&msgbuf);
1395 }
1396
1397 static struct diff_tempfile *claim_diff_tempfile(void) {
1398         int i;
1399         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1400                 if (!diff_temp[i].name)
1401                         return diff_temp + i;
1402         die("BUG: diff is failing to clean up its tempfiles");
1403 }
1404
1405 static void remove_tempfile(void)
1406 {
1407         int i;
1408         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1409                 if (is_tempfile_active(&diff_temp[i].tempfile))
1410                         delete_tempfile(&diff_temp[i].tempfile);
1411                 diff_temp[i].name = NULL;
1412         }
1413 }
1414
1415 static void add_line_count(struct strbuf *out, int count)
1416 {
1417         switch (count) {
1418         case 0:
1419                 strbuf_addstr(out, "0,0");
1420                 break;
1421         case 1:
1422                 strbuf_addstr(out, "1");
1423                 break;
1424         default:
1425                 strbuf_addf(out, "1,%d", count);
1426                 break;
1427         }
1428 }
1429
1430 static void emit_rewrite_lines(struct emit_callback *ecb,
1431                                int prefix, const char *data, int size)
1432 {
1433         const char *endp = NULL;
1434         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1435
1436         while (0 < size) {
1437                 int len;
1438
1439                 endp = memchr(data, '\n', size);
1440                 len = endp ? (endp - data + 1) : size;
1441                 if (prefix != '+') {
1442                         ecb->lno_in_preimage++;
1443                         emit_del_line(reset, ecb, data, len);
1444                 } else {
1445                         ecb->lno_in_postimage++;
1446                         emit_add_line(reset, ecb, data, len);
1447                 }
1448                 size -= len;
1449                 data += len;
1450         }
1451         if (!endp)
1452                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1453 }
1454
1455 static void emit_rewrite_diff(const char *name_a,
1456                               const char *name_b,
1457                               struct diff_filespec *one,
1458                               struct diff_filespec *two,
1459                               struct userdiff_driver *textconv_one,
1460                               struct userdiff_driver *textconv_two,
1461                               struct diff_options *o)
1462 {
1463         int lc_a, lc_b;
1464         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1465         const char *a_prefix, *b_prefix;
1466         char *data_one, *data_two;
1467         size_t size_one, size_two;
1468         struct emit_callback ecbdata;
1469         struct strbuf out = STRBUF_INIT;
1470
1471         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
1472                 a_prefix = o->b_prefix;
1473                 b_prefix = o->a_prefix;
1474         } else {
1475                 a_prefix = o->a_prefix;
1476                 b_prefix = o->b_prefix;
1477         }
1478
1479         name_a += (*name_a == '/');
1480         name_b += (*name_b == '/');
1481
1482         strbuf_reset(&a_name);
1483         strbuf_reset(&b_name);
1484         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1485         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1486
1487         size_one = fill_textconv(textconv_one, one, &data_one);
1488         size_two = fill_textconv(textconv_two, two, &data_two);
1489
1490         memset(&ecbdata, 0, sizeof(ecbdata));
1491         ecbdata.color_diff = want_color(o->use_color);
1492         ecbdata.ws_rule = whitespace_rule(name_b);
1493         ecbdata.opt = o;
1494         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1495                 mmfile_t mf1, mf2;
1496                 mf1.ptr = (char *)data_one;
1497                 mf2.ptr = (char *)data_two;
1498                 mf1.size = size_one;
1499                 mf2.size = size_two;
1500                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1501         }
1502         ecbdata.lno_in_preimage = 1;
1503         ecbdata.lno_in_postimage = 1;
1504
1505         lc_a = count_lines(data_one, size_one);
1506         lc_b = count_lines(data_two, size_two);
1507
1508         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1509                          a_name.buf, a_name.len, 0);
1510         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1511                          b_name.buf, b_name.len, 0);
1512
1513         strbuf_addstr(&out, "@@ -");
1514         if (!o->irreversible_delete)
1515                 add_line_count(&out, lc_a);
1516         else
1517                 strbuf_addstr(&out, "?,?");
1518         strbuf_addstr(&out, " +");
1519         add_line_count(&out, lc_b);
1520         strbuf_addstr(&out, " @@\n");
1521         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1522         strbuf_release(&out);
1523
1524         if (lc_a && !o->irreversible_delete)
1525                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1526         if (lc_b)
1527                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1528         if (textconv_one)
1529                 free((char *)data_one);
1530         if (textconv_two)
1531                 free((char *)data_two);
1532 }
1533
1534 struct diff_words_buffer {
1535         mmfile_t text;
1536         long alloc;
1537         struct diff_words_orig {
1538                 const char *begin, *end;
1539         } *orig;
1540         int orig_nr, orig_alloc;
1541 };
1542
1543 static void diff_words_append(char *line, unsigned long len,
1544                 struct diff_words_buffer *buffer)
1545 {
1546         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1547         line++;
1548         len--;
1549         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1550         buffer->text.size += len;
1551         buffer->text.ptr[buffer->text.size] = '\0';
1552 }
1553
1554 struct diff_words_style_elem {
1555         const char *prefix;
1556         const char *suffix;
1557         const char *color; /* NULL; filled in by the setup code if
1558                             * color is enabled */
1559 };
1560
1561 struct diff_words_style {
1562         enum diff_words_type type;
1563         struct diff_words_style_elem new, old, ctx;
1564         const char *newline;
1565 };
1566
1567 static struct diff_words_style diff_words_styles[] = {
1568         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1569         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1570         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1571 };
1572
1573 struct diff_words_data {
1574         struct diff_words_buffer minus, plus;
1575         const char *current_plus;
1576         int last_minus;
1577         struct diff_options *opt;
1578         regex_t *word_regex;
1579         enum diff_words_type type;
1580         struct diff_words_style *style;
1581 };
1582
1583 static int fn_out_diff_words_write_helper(struct diff_options *o,
1584                                           struct diff_words_style_elem *st_el,
1585                                           const char *newline,
1586                                           size_t count, const char *buf)
1587 {
1588         int print = 0;
1589         struct strbuf sb = STRBUF_INIT;
1590
1591         while (count) {
1592                 char *p = memchr(buf, '\n', count);
1593                 if (print)
1594                         strbuf_addstr(&sb, diff_line_prefix(o));
1595
1596                 if (p != buf) {
1597                         const char *reset = st_el->color && *st_el->color ?
1598                                             GIT_COLOR_RESET : NULL;
1599                         if (st_el->color && *st_el->color)
1600                                 strbuf_addstr(&sb, st_el->color);
1601                         strbuf_addstr(&sb, st_el->prefix);
1602                         strbuf_add(&sb, buf, p ? p - buf : count);
1603                         strbuf_addstr(&sb, st_el->suffix);
1604                         if (reset)
1605                                 strbuf_addstr(&sb, reset);
1606                 }
1607                 if (!p)
1608                         goto out;
1609
1610                 strbuf_addstr(&sb, newline);
1611                 count -= p + 1 - buf;
1612                 buf = p + 1;
1613                 print = 1;
1614                 if (count) {
1615                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1616                                          sb.buf, sb.len, 0);
1617                         strbuf_reset(&sb);
1618                 }
1619         }
1620
1621 out:
1622         if (sb.len)
1623                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1624                                  sb.buf, sb.len, 0);
1625         strbuf_release(&sb);
1626         return 0;
1627 }
1628
1629 /*
1630  * '--color-words' algorithm can be described as:
1631  *
1632  *   1. collect the minus/plus lines of a diff hunk, divided into
1633  *      minus-lines and plus-lines;
1634  *
1635  *   2. break both minus-lines and plus-lines into words and
1636  *      place them into two mmfile_t with one word for each line;
1637  *
1638  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1639  *
1640  * And for the common parts of the both file, we output the plus side text.
1641  * diff_words->current_plus is used to trace the current position of the plus file
1642  * which printed. diff_words->last_minus is used to trace the last minus word
1643  * printed.
1644  *
1645  * For '--graph' to work with '--color-words', we need to output the graph prefix
1646  * on each line of color words output. Generally, there are two conditions on
1647  * which we should output the prefix.
1648  *
1649  *   1. diff_words->last_minus == 0 &&
1650  *      diff_words->current_plus == diff_words->plus.text.ptr
1651  *
1652  *      that is: the plus text must start as a new line, and if there is no minus
1653  *      word printed, a graph prefix must be printed.
1654  *
1655  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1656  *      *(diff_words->current_plus - 1) == '\n'
1657  *
1658  *      that is: a graph prefix must be printed following a '\n'
1659  */
1660 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1661 {
1662         if ((diff_words->last_minus == 0 &&
1663                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1664                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1665                 *(diff_words->current_plus - 1) == '\n')) {
1666                 return 1;
1667         } else {
1668                 return 0;
1669         }
1670 }
1671
1672 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1673 {
1674         struct diff_words_data *diff_words = priv;
1675         struct diff_words_style *style = diff_words->style;
1676         int minus_first, minus_len, plus_first, plus_len;
1677         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1678         struct diff_options *opt = diff_words->opt;
1679         const char *line_prefix;
1680
1681         if (line[0] != '@' || parse_hunk_header(line, len,
1682                         &minus_first, &minus_len, &plus_first, &plus_len))
1683                 return;
1684
1685         assert(opt);
1686         line_prefix = diff_line_prefix(opt);
1687
1688         /* POSIX requires that first be decremented by one if len == 0... */
1689         if (minus_len) {
1690                 minus_begin = diff_words->minus.orig[minus_first].begin;
1691                 minus_end =
1692                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1693         } else
1694                 minus_begin = minus_end =
1695                         diff_words->minus.orig[minus_first].end;
1696
1697         if (plus_len) {
1698                 plus_begin = diff_words->plus.orig[plus_first].begin;
1699                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1700         } else
1701                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1702
1703         if (color_words_output_graph_prefix(diff_words)) {
1704                 fputs(line_prefix, diff_words->opt->file);
1705         }
1706         if (diff_words->current_plus != plus_begin) {
1707                 fn_out_diff_words_write_helper(diff_words->opt,
1708                                 &style->ctx, style->newline,
1709                                 plus_begin - diff_words->current_plus,
1710                                 diff_words->current_plus);
1711         }
1712         if (minus_begin != minus_end) {
1713                 fn_out_diff_words_write_helper(diff_words->opt,
1714                                 &style->old, style->newline,
1715                                 minus_end - minus_begin, minus_begin);
1716         }
1717         if (plus_begin != plus_end) {
1718                 fn_out_diff_words_write_helper(diff_words->opt,
1719                                 &style->new, style->newline,
1720                                 plus_end - plus_begin, plus_begin);
1721         }
1722
1723         diff_words->current_plus = plus_end;
1724         diff_words->last_minus = minus_first;
1725 }
1726
1727 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1728 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1729                 int *begin, int *end)
1730 {
1731         if (word_regex && *begin < buffer->size) {
1732                 regmatch_t match[1];
1733                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1734                                  buffer->size - *begin, 1, match, 0)) {
1735                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1736                                         '\n', match[0].rm_eo - match[0].rm_so);
1737                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1738                         *begin += match[0].rm_so;
1739                         return *begin >= *end;
1740                 }
1741                 return -1;
1742         }
1743
1744         /* find the next word */
1745         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1746                 (*begin)++;
1747         if (*begin >= buffer->size)
1748                 return -1;
1749
1750         /* find the end of the word */
1751         *end = *begin + 1;
1752         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1753                 (*end)++;
1754
1755         return 0;
1756 }
1757
1758 /*
1759  * This function splits the words in buffer->text, stores the list with
1760  * newline separator into out, and saves the offsets of the original words
1761  * in buffer->orig.
1762  */
1763 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1764                 regex_t *word_regex)
1765 {
1766         int i, j;
1767         long alloc = 0;
1768
1769         out->size = 0;
1770         out->ptr = NULL;
1771
1772         /* fake an empty "0th" word */
1773         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1774         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1775         buffer->orig_nr = 1;
1776
1777         for (i = 0; i < buffer->text.size; i++) {
1778                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1779                         return;
1780
1781                 /* store original boundaries */
1782                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1783                                 buffer->orig_alloc);
1784                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1785                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1786                 buffer->orig_nr++;
1787
1788                 /* store one word */
1789                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1790                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1791                 out->ptr[out->size + j - i] = '\n';
1792                 out->size += j - i + 1;
1793
1794                 i = j - 1;
1795         }
1796 }
1797
1798 /* this executes the word diff on the accumulated buffers */
1799 static void diff_words_show(struct diff_words_data *diff_words)
1800 {
1801         xpparam_t xpp;
1802         xdemitconf_t xecfg;
1803         mmfile_t minus, plus;
1804         struct diff_words_style *style = diff_words->style;
1805
1806         struct diff_options *opt = diff_words->opt;
1807         const char *line_prefix;
1808
1809         assert(opt);
1810         line_prefix = diff_line_prefix(opt);
1811
1812         /* special case: only removal */
1813         if (!diff_words->plus.text.size) {
1814                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1815                                  line_prefix, strlen(line_prefix), 0);
1816                 fn_out_diff_words_write_helper(diff_words->opt,
1817                         &style->old, style->newline,
1818                         diff_words->minus.text.size,
1819                         diff_words->minus.text.ptr);
1820                 diff_words->minus.text.size = 0;
1821                 return;
1822         }
1823
1824         diff_words->current_plus = diff_words->plus.text.ptr;
1825         diff_words->last_minus = 0;
1826
1827         memset(&xpp, 0, sizeof(xpp));
1828         memset(&xecfg, 0, sizeof(xecfg));
1829         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1830         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1831         xpp.flags = 0;
1832         /* as only the hunk header will be parsed, we need a 0-context */
1833         xecfg.ctxlen = 0;
1834         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1835                           &xpp, &xecfg))
1836                 die("unable to generate word diff");
1837         free(minus.ptr);
1838         free(plus.ptr);
1839         if (diff_words->current_plus != diff_words->plus.text.ptr +
1840                         diff_words->plus.text.size) {
1841                 if (color_words_output_graph_prefix(diff_words))
1842                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1843                                          line_prefix, strlen(line_prefix), 0);
1844                 fn_out_diff_words_write_helper(diff_words->opt,
1845                         &style->ctx, style->newline,
1846                         diff_words->plus.text.ptr + diff_words->plus.text.size
1847                         - diff_words->current_plus, diff_words->current_plus);
1848         }
1849         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1850 }
1851
1852 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1853 static void diff_words_flush(struct emit_callback *ecbdata)
1854 {
1855         struct diff_options *wo = ecbdata->diff_words->opt;
1856
1857         if (ecbdata->diff_words->minus.text.size ||
1858             ecbdata->diff_words->plus.text.size)
1859                 diff_words_show(ecbdata->diff_words);
1860
1861         if (wo->emitted_symbols) {
1862                 struct diff_options *o = ecbdata->opt;
1863                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1864                 int i;
1865
1866                 /*
1867                  * NEEDSWORK:
1868                  * Instead of appending each, concat all words to a line?
1869                  */
1870                 for (i = 0; i < wol->nr; i++)
1871                         append_emitted_diff_symbol(o, &wol->buf[i]);
1872
1873                 for (i = 0; i < wol->nr; i++)
1874                         free((void *)wol->buf[i].line);
1875
1876                 wol->nr = 0;
1877         }
1878 }
1879
1880 static void diff_filespec_load_driver(struct diff_filespec *one)
1881 {
1882         /* Use already-loaded driver */
1883         if (one->driver)
1884                 return;
1885
1886         if (S_ISREG(one->mode))
1887                 one->driver = userdiff_find_by_path(one->path);
1888
1889         /* Fallback to default settings */
1890         if (!one->driver)
1891                 one->driver = userdiff_find_by_name("default");
1892 }
1893
1894 static const char *userdiff_word_regex(struct diff_filespec *one)
1895 {
1896         diff_filespec_load_driver(one);
1897         return one->driver->word_regex;
1898 }
1899
1900 static void init_diff_words_data(struct emit_callback *ecbdata,
1901                                  struct diff_options *orig_opts,
1902                                  struct diff_filespec *one,
1903                                  struct diff_filespec *two)
1904 {
1905         int i;
1906         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1907         memcpy(o, orig_opts, sizeof(struct diff_options));
1908
1909         ecbdata->diff_words =
1910                 xcalloc(1, sizeof(struct diff_words_data));
1911         ecbdata->diff_words->type = o->word_diff;
1912         ecbdata->diff_words->opt = o;
1913
1914         if (orig_opts->emitted_symbols)
1915                 o->emitted_symbols =
1916                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1917
1918         if (!o->word_regex)
1919                 o->word_regex = userdiff_word_regex(one);
1920         if (!o->word_regex)
1921                 o->word_regex = userdiff_word_regex(two);
1922         if (!o->word_regex)
1923                 o->word_regex = diff_word_regex_cfg;
1924         if (o->word_regex) {
1925                 ecbdata->diff_words->word_regex = (regex_t *)
1926                         xmalloc(sizeof(regex_t));
1927                 if (regcomp(ecbdata->diff_words->word_regex,
1928                             o->word_regex,
1929                             REG_EXTENDED | REG_NEWLINE))
1930                         die ("Invalid regular expression: %s",
1931                              o->word_regex);
1932         }
1933         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1934                 if (o->word_diff == diff_words_styles[i].type) {
1935                         ecbdata->diff_words->style =
1936                                 &diff_words_styles[i];
1937                         break;
1938                 }
1939         }
1940         if (want_color(o->use_color)) {
1941                 struct diff_words_style *st = ecbdata->diff_words->style;
1942                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1943                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1944                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1945         }
1946 }
1947
1948 static void free_diff_words_data(struct emit_callback *ecbdata)
1949 {
1950         if (ecbdata->diff_words) {
1951                 diff_words_flush(ecbdata);
1952                 free (ecbdata->diff_words->opt->emitted_symbols);
1953                 free (ecbdata->diff_words->opt);
1954                 free (ecbdata->diff_words->minus.text.ptr);
1955                 free (ecbdata->diff_words->minus.orig);
1956                 free (ecbdata->diff_words->plus.text.ptr);
1957                 free (ecbdata->diff_words->plus.orig);
1958                 if (ecbdata->diff_words->word_regex) {
1959                         regfree(ecbdata->diff_words->word_regex);
1960                         free(ecbdata->diff_words->word_regex);
1961                 }
1962                 FREE_AND_NULL(ecbdata->diff_words);
1963         }
1964 }
1965
1966 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1967 {
1968         if (want_color(diff_use_color))
1969                 return diff_colors[ix];
1970         return "";
1971 }
1972
1973 const char *diff_line_prefix(struct diff_options *opt)
1974 {
1975         struct strbuf *msgbuf;
1976         if (!opt->output_prefix)
1977                 return "";
1978
1979         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1980         return msgbuf->buf;
1981 }
1982
1983 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1984 {
1985         const char *cp;
1986         unsigned long allot;
1987         size_t l = len;
1988
1989         if (ecb->truncate)
1990                 return ecb->truncate(line, len);
1991         cp = line;
1992         allot = l;
1993         while (0 < l) {
1994                 (void) utf8_width(&cp, &l);
1995                 if (!cp)
1996                         break; /* truncated in the middle? */
1997         }
1998         return allot - l;
1999 }
2000
2001 static void find_lno(const char *line, struct emit_callback *ecbdata)
2002 {
2003         const char *p;
2004         ecbdata->lno_in_preimage = 0;
2005         ecbdata->lno_in_postimage = 0;
2006         p = strchr(line, '-');
2007         if (!p)
2008                 return; /* cannot happen */
2009         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2010         p = strchr(p, '+');
2011         if (!p)
2012                 return; /* cannot happen */
2013         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2014 }
2015
2016 static void fn_out_consume(void *priv, char *line, unsigned long len)
2017 {
2018         struct emit_callback *ecbdata = priv;
2019         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2020         struct diff_options *o = ecbdata->opt;
2021
2022         o->found_changes = 1;
2023
2024         if (ecbdata->header) {
2025                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2026                                  ecbdata->header->buf, ecbdata->header->len, 0);
2027                 strbuf_reset(ecbdata->header);
2028                 ecbdata->header = NULL;
2029         }
2030
2031         if (ecbdata->label_path[0]) {
2032                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2033                                  ecbdata->label_path[0],
2034                                  strlen(ecbdata->label_path[0]), 0);
2035                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2036                                  ecbdata->label_path[1],
2037                                  strlen(ecbdata->label_path[1]), 0);
2038                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2039         }
2040
2041         if (diff_suppress_blank_empty
2042             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2043                 line[0] = '\n';
2044                 len = 1;
2045         }
2046
2047         if (line[0] == '@') {
2048                 if (ecbdata->diff_words)
2049                         diff_words_flush(ecbdata);
2050                 len = sane_truncate_line(ecbdata, line, len);
2051                 find_lno(line, ecbdata);
2052                 emit_hunk_header(ecbdata, line, len);
2053                 return;
2054         }
2055
2056         if (ecbdata->diff_words) {
2057                 enum diff_symbol s =
2058                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2059                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2060                 if (line[0] == '-') {
2061                         diff_words_append(line, len,
2062                                           &ecbdata->diff_words->minus);
2063                         return;
2064                 } else if (line[0] == '+') {
2065                         diff_words_append(line, len,
2066                                           &ecbdata->diff_words->plus);
2067                         return;
2068                 } else if (starts_with(line, "\\ ")) {
2069                         /*
2070                          * Eat the "no newline at eof" marker as if we
2071                          * saw a "+" or "-" line with nothing on it,
2072                          * and return without diff_words_flush() to
2073                          * defer processing. If this is the end of
2074                          * preimage, more "+" lines may come after it.
2075                          */
2076                         return;
2077                 }
2078                 diff_words_flush(ecbdata);
2079                 emit_diff_symbol(o, s, line, len, 0);
2080                 return;
2081         }
2082
2083         switch (line[0]) {
2084         case '+':
2085                 ecbdata->lno_in_postimage++;
2086                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2087                 break;
2088         case '-':
2089                 ecbdata->lno_in_preimage++;
2090                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2091                 break;
2092         case ' ':
2093                 ecbdata->lno_in_postimage++;
2094                 ecbdata->lno_in_preimage++;
2095                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2096                 break;
2097         default:
2098                 /* incomplete line at the end */
2099                 ecbdata->lno_in_preimage++;
2100                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2101                                  line, len, 0);
2102                 break;
2103         }
2104 }
2105
2106 static char *pprint_rename(const char *a, const char *b)
2107 {
2108         const char *old = a;
2109         const char *new = b;
2110         struct strbuf name = STRBUF_INIT;
2111         int pfx_length, sfx_length;
2112         int pfx_adjust_for_slash;
2113         int len_a = strlen(a);
2114         int len_b = strlen(b);
2115         int a_midlen, b_midlen;
2116         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2117         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2118
2119         if (qlen_a || qlen_b) {
2120                 quote_c_style(a, &name, NULL, 0);
2121                 strbuf_addstr(&name, " => ");
2122                 quote_c_style(b, &name, NULL, 0);
2123                 return strbuf_detach(&name, NULL);
2124         }
2125
2126         /* Find common prefix */
2127         pfx_length = 0;
2128         while (*old && *new && *old == *new) {
2129                 if (*old == '/')
2130                         pfx_length = old - a + 1;
2131                 old++;
2132                 new++;
2133         }
2134
2135         /* Find common suffix */
2136         old = a + len_a;
2137         new = b + len_b;
2138         sfx_length = 0;
2139         /*
2140          * If there is a common prefix, it must end in a slash.  In
2141          * that case we let this loop run 1 into the prefix to see the
2142          * same slash.
2143          *
2144          * If there is no common prefix, we cannot do this as it would
2145          * underrun the input strings.
2146          */
2147         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2148         while (a + pfx_length - pfx_adjust_for_slash <= old &&
2149                b + pfx_length - pfx_adjust_for_slash <= new &&
2150                *old == *new) {
2151                 if (*old == '/')
2152                         sfx_length = len_a - (old - a);
2153                 old--;
2154                 new--;
2155         }
2156
2157         /*
2158          * pfx{mid-a => mid-b}sfx
2159          * {pfx-a => pfx-b}sfx
2160          * pfx{sfx-a => sfx-b}
2161          * name-a => name-b
2162          */
2163         a_midlen = len_a - pfx_length - sfx_length;
2164         b_midlen = len_b - pfx_length - sfx_length;
2165         if (a_midlen < 0)
2166                 a_midlen = 0;
2167         if (b_midlen < 0)
2168                 b_midlen = 0;
2169
2170         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2171         if (pfx_length + sfx_length) {
2172                 strbuf_add(&name, a, pfx_length);
2173                 strbuf_addch(&name, '{');
2174         }
2175         strbuf_add(&name, a + pfx_length, a_midlen);
2176         strbuf_addstr(&name, " => ");
2177         strbuf_add(&name, b + pfx_length, b_midlen);
2178         if (pfx_length + sfx_length) {
2179                 strbuf_addch(&name, '}');
2180                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
2181         }
2182         return strbuf_detach(&name, NULL);
2183 }
2184
2185 struct diffstat_t {
2186         int nr;
2187         int alloc;
2188         struct diffstat_file {
2189                 char *from_name;
2190                 char *name;
2191                 char *print_name;
2192                 unsigned is_unmerged:1;
2193                 unsigned is_binary:1;
2194                 unsigned is_renamed:1;
2195                 unsigned is_interesting:1;
2196                 uintmax_t added, deleted;
2197         } **files;
2198 };
2199
2200 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2201                                           const char *name_a,
2202                                           const char *name_b)
2203 {
2204         struct diffstat_file *x;
2205         x = xcalloc(1, sizeof(*x));
2206         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2207         diffstat->files[diffstat->nr++] = x;
2208         if (name_b) {
2209                 x->from_name = xstrdup(name_a);
2210                 x->name = xstrdup(name_b);
2211                 x->is_renamed = 1;
2212         }
2213         else {
2214                 x->from_name = NULL;
2215                 x->name = xstrdup(name_a);
2216         }
2217         return x;
2218 }
2219
2220 static void diffstat_consume(void *priv, char *line, unsigned long len)
2221 {
2222         struct diffstat_t *diffstat = priv;
2223         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2224
2225         if (line[0] == '+')
2226                 x->added++;
2227         else if (line[0] == '-')
2228                 x->deleted++;
2229 }
2230
2231 const char mime_boundary_leader[] = "------------";
2232
2233 static int scale_linear(int it, int width, int max_change)
2234 {
2235         if (!it)
2236                 return 0;
2237         /*
2238          * make sure that at least one '-' or '+' is printed if
2239          * there is any change to this path. The easiest way is to
2240          * scale linearly as if the alloted width is one column shorter
2241          * than it is, and then add 1 to the result.
2242          */
2243         return 1 + (it * (width - 1) / max_change);
2244 }
2245
2246 static void show_graph(struct strbuf *out, char ch, int cnt,
2247                        const char *set, const char *reset)
2248 {
2249         if (cnt <= 0)
2250                 return;
2251         strbuf_addstr(out, set);
2252         strbuf_addchars(out, ch, cnt);
2253         strbuf_addstr(out, reset);
2254 }
2255
2256 static void fill_print_name(struct diffstat_file *file)
2257 {
2258         char *pname;
2259
2260         if (file->print_name)
2261                 return;
2262
2263         if (!file->is_renamed) {
2264                 struct strbuf buf = STRBUF_INIT;
2265                 if (quote_c_style(file->name, &buf, NULL, 0)) {
2266                         pname = strbuf_detach(&buf, NULL);
2267                 } else {
2268                         pname = file->name;
2269                         strbuf_release(&buf);
2270                 }
2271         } else {
2272                 pname = pprint_rename(file->from_name, file->name);
2273         }
2274         file->print_name = pname;
2275 }
2276
2277 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2278                 int files, int insertions, int deletions)
2279 {
2280         struct strbuf sb = STRBUF_INIT;
2281
2282         if (!files) {
2283                 assert(insertions == 0 && deletions == 0);
2284                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2285                                  NULL, 0, 0);
2286                 return;
2287         }
2288
2289         strbuf_addf(&sb,
2290                     (files == 1) ? " %d file changed" : " %d files changed",
2291                     files);
2292
2293         /*
2294          * For binary diff, the caller may want to print "x files
2295          * changed" with insertions == 0 && deletions == 0.
2296          *
2297          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2298          * is probably less confusing (i.e skip over "2 files changed
2299          * but nothing about added/removed lines? Is this a bug in Git?").
2300          */
2301         if (insertions || deletions == 0) {
2302                 strbuf_addf(&sb,
2303                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2304                             insertions);
2305         }
2306
2307         if (deletions || insertions == 0) {
2308                 strbuf_addf(&sb,
2309                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2310                             deletions);
2311         }
2312         strbuf_addch(&sb, '\n');
2313         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2314                          sb.buf, sb.len, 0);
2315         strbuf_release(&sb);
2316 }
2317
2318 void print_stat_summary(FILE *fp, int files,
2319                         int insertions, int deletions)
2320 {
2321         struct diff_options o;
2322         memset(&o, 0, sizeof(o));
2323         o.file = fp;
2324
2325         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2326 }
2327
2328 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2329 {
2330         int i, len, add, del, adds = 0, dels = 0;
2331         uintmax_t max_change = 0, max_len = 0;
2332         int total_files = data->nr, count;
2333         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2334         const char *reset, *add_c, *del_c;
2335         int extra_shown = 0;
2336         const char *line_prefix = diff_line_prefix(options);
2337         struct strbuf out = STRBUF_INIT;
2338
2339         if (data->nr == 0)
2340                 return;
2341
2342         count = options->stat_count ? options->stat_count : data->nr;
2343
2344         reset = diff_get_color_opt(options, DIFF_RESET);
2345         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2346         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2347
2348         /*
2349          * Find the longest filename and max number of changes
2350          */
2351         for (i = 0; (i < count) && (i < data->nr); i++) {
2352                 struct diffstat_file *file = data->files[i];
2353                 uintmax_t change = file->added + file->deleted;
2354
2355                 if (!file->is_interesting && (change == 0)) {
2356                         count++; /* not shown == room for one more */
2357                         continue;
2358                 }
2359                 fill_print_name(file);
2360                 len = strlen(file->print_name);
2361                 if (max_len < len)
2362                         max_len = len;
2363
2364                 if (file->is_unmerged) {
2365                         /* "Unmerged" is 8 characters */
2366                         bin_width = bin_width < 8 ? 8 : bin_width;
2367                         continue;
2368                 }
2369                 if (file->is_binary) {
2370                         /* "Bin XXX -> YYY bytes" */
2371                         int w = 14 + decimal_width(file->added)
2372                                 + decimal_width(file->deleted);
2373                         bin_width = bin_width < w ? w : bin_width;
2374                         /* Display change counts aligned with "Bin" */
2375                         number_width = 3;
2376                         continue;
2377                 }
2378
2379                 if (max_change < change)
2380                         max_change = change;
2381         }
2382         count = i; /* where we can stop scanning in data->files[] */
2383
2384         /*
2385          * We have width = stat_width or term_columns() columns total.
2386          * We want a maximum of min(max_len, stat_name_width) for the name part.
2387          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2388          * We also need 1 for " " and 4 + decimal_width(max_change)
2389          * for " | NNNN " and one the empty column at the end, altogether
2390          * 6 + decimal_width(max_change).
2391          *
2392          * If there's not enough space, we will use the smaller of
2393          * stat_name_width (if set) and 5/8*width for the filename,
2394          * and the rest for constant elements + graph part, but no more
2395          * than stat_graph_width for the graph part.
2396          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2397          * for the standard terminal size).
2398          *
2399          * In other words: stat_width limits the maximum width, and
2400          * stat_name_width fixes the maximum width of the filename,
2401          * and is also used to divide available columns if there
2402          * aren't enough.
2403          *
2404          * Binary files are displayed with "Bin XXX -> YYY bytes"
2405          * instead of the change count and graph. This part is treated
2406          * similarly to the graph part, except that it is not
2407          * "scaled". If total width is too small to accommodate the
2408          * guaranteed minimum width of the filename part and the
2409          * separators and this message, this message will "overflow"
2410          * making the line longer than the maximum width.
2411          */
2412
2413         if (options->stat_width == -1)
2414                 width = term_columns() - strlen(line_prefix);
2415         else
2416                 width = options->stat_width ? options->stat_width : 80;
2417         number_width = decimal_width(max_change) > number_width ?
2418                 decimal_width(max_change) : number_width;
2419
2420         if (options->stat_graph_width == -1)
2421                 options->stat_graph_width = diff_stat_graph_width;
2422
2423         /*
2424          * Guarantee 3/8*16==6 for the graph part
2425          * and 5/8*16==10 for the filename part
2426          */
2427         if (width < 16 + 6 + number_width)
2428                 width = 16 + 6 + number_width;
2429
2430         /*
2431          * First assign sizes that are wanted, ignoring available width.
2432          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2433          * starting from "XXX" should fit in graph_width.
2434          */
2435         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2436         if (options->stat_graph_width &&
2437             options->stat_graph_width < graph_width)
2438                 graph_width = options->stat_graph_width;
2439
2440         name_width = (options->stat_name_width > 0 &&
2441                       options->stat_name_width < max_len) ?
2442                 options->stat_name_width : max_len;
2443
2444         /*
2445          * Adjust adjustable widths not to exceed maximum width
2446          */
2447         if (name_width + number_width + 6 + graph_width > width) {
2448                 if (graph_width > width * 3/8 - number_width - 6) {
2449                         graph_width = width * 3/8 - number_width - 6;
2450                         if (graph_width < 6)
2451                                 graph_width = 6;
2452                 }
2453
2454                 if (options->stat_graph_width &&
2455                     graph_width > options->stat_graph_width)
2456                         graph_width = options->stat_graph_width;
2457                 if (name_width > width - number_width - 6 - graph_width)
2458                         name_width = width - number_width - 6 - graph_width;
2459                 else
2460                         graph_width = width - number_width - 6 - name_width;
2461         }
2462
2463         /*
2464          * From here name_width is the width of the name area,
2465          * and graph_width is the width of the graph area.
2466          * max_change is used to scale graph properly.
2467          */
2468         for (i = 0; i < count; i++) {
2469                 const char *prefix = "";
2470                 struct diffstat_file *file = data->files[i];
2471                 char *name = file->print_name;
2472                 uintmax_t added = file->added;
2473                 uintmax_t deleted = file->deleted;
2474                 int name_len;
2475
2476                 if (!file->is_interesting && (added + deleted == 0))
2477                         continue;
2478
2479                 /*
2480                  * "scale" the filename
2481                  */
2482                 len = name_width;
2483                 name_len = strlen(name);
2484                 if (name_width < name_len) {
2485                         char *slash;
2486                         prefix = "...";
2487                         len -= 3;
2488                         name += name_len - len;
2489                         slash = strchr(name, '/');
2490                         if (slash)
2491                                 name = slash;
2492                 }
2493
2494                 if (file->is_binary) {
2495                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2496                         strbuf_addf(&out, " %*s", number_width, "Bin");
2497                         if (!added && !deleted) {
2498                                 strbuf_addch(&out, '\n');
2499                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2500                                                  out.buf, out.len, 0);
2501                                 strbuf_reset(&out);
2502                                 continue;
2503                         }
2504                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2505                                 del_c, deleted, reset);
2506                         strbuf_addstr(&out, " -> ");
2507                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2508                                 add_c, added, reset);
2509                         strbuf_addstr(&out, " bytes\n");
2510                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2511                                          out.buf, out.len, 0);
2512                         strbuf_reset(&out);
2513                         continue;
2514                 }
2515                 else if (file->is_unmerged) {
2516                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2517                         strbuf_addstr(&out, " Unmerged\n");
2518                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2519                                          out.buf, out.len, 0);
2520                         strbuf_reset(&out);
2521                         continue;
2522                 }
2523
2524                 /*
2525                  * scale the add/delete
2526                  */
2527                 add = added;
2528                 del = deleted;
2529
2530                 if (graph_width <= max_change) {
2531                         int total = scale_linear(add + del, graph_width, max_change);
2532                         if (total < 2 && add && del)
2533                                 /* width >= 2 due to the sanity check */
2534                                 total = 2;
2535                         if (add < del) {
2536                                 add = scale_linear(add, graph_width, max_change);
2537                                 del = total - add;
2538                         } else {
2539                                 del = scale_linear(del, graph_width, max_change);
2540                                 add = total - del;
2541                         }
2542                 }
2543                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2544                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2545                         number_width, added + deleted,
2546                         added + deleted ? " " : "");
2547                 show_graph(&out, '+', add, add_c, reset);
2548                 show_graph(&out, '-', del, del_c, reset);
2549                 strbuf_addch(&out, '\n');
2550                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2551                                  out.buf, out.len, 0);
2552                 strbuf_reset(&out);
2553         }
2554
2555         for (i = 0; i < data->nr; i++) {
2556                 struct diffstat_file *file = data->files[i];
2557                 uintmax_t added = file->added;
2558                 uintmax_t deleted = file->deleted;
2559
2560                 if (file->is_unmerged ||
2561                     (!file->is_interesting && (added + deleted == 0))) {
2562                         total_files--;
2563                         continue;
2564                 }
2565
2566                 if (!file->is_binary) {
2567                         adds += added;
2568                         dels += deleted;
2569                 }
2570                 if (i < count)
2571                         continue;
2572                 if (!extra_shown)
2573                         emit_diff_symbol(options,
2574                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2575                                          NULL, 0, 0);
2576                 extra_shown = 1;
2577         }
2578
2579         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2580 }
2581
2582 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2583 {
2584         int i, adds = 0, dels = 0, total_files = data->nr;
2585
2586         if (data->nr == 0)
2587                 return;
2588
2589         for (i = 0; i < data->nr; i++) {
2590                 int added = data->files[i]->added;
2591                 int deleted = data->files[i]->deleted;
2592
2593                 if (data->files[i]->is_unmerged ||
2594                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2595                         total_files--;
2596                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2597                         adds += added;
2598                         dels += deleted;
2599                 }
2600         }
2601         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2602 }
2603
2604 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2605 {
2606         int i;
2607
2608         if (data->nr == 0)
2609                 return;
2610
2611         for (i = 0; i < data->nr; i++) {
2612                 struct diffstat_file *file = data->files[i];
2613
2614                 fprintf(options->file, "%s", diff_line_prefix(options));
2615
2616                 if (file->is_binary)
2617                         fprintf(options->file, "-\t-\t");
2618                 else
2619                         fprintf(options->file,
2620                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2621                                 file->added, file->deleted);
2622                 if (options->line_termination) {
2623                         fill_print_name(file);
2624                         if (!file->is_renamed)
2625                                 write_name_quoted(file->name, options->file,
2626                                                   options->line_termination);
2627                         else {
2628                                 fputs(file->print_name, options->file);
2629                                 putc(options->line_termination, options->file);
2630                         }
2631                 } else {
2632                         if (file->is_renamed) {
2633                                 putc('\0', options->file);
2634                                 write_name_quoted(file->from_name, options->file, '\0');
2635                         }
2636                         write_name_quoted(file->name, options->file, '\0');
2637                 }
2638         }
2639 }
2640
2641 struct dirstat_file {
2642         const char *name;
2643         unsigned long changed;
2644 };
2645
2646 struct dirstat_dir {
2647         struct dirstat_file *files;
2648         int alloc, nr, permille, cumulative;
2649 };
2650
2651 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2652                 unsigned long changed, const char *base, int baselen)
2653 {
2654         unsigned long this_dir = 0;
2655         unsigned int sources = 0;
2656         const char *line_prefix = diff_line_prefix(opt);
2657
2658         while (dir->nr) {
2659                 struct dirstat_file *f = dir->files;
2660                 int namelen = strlen(f->name);
2661                 unsigned long this;
2662                 char *slash;
2663
2664                 if (namelen < baselen)
2665                         break;
2666                 if (memcmp(f->name, base, baselen))
2667                         break;
2668                 slash = strchr(f->name + baselen, '/');
2669                 if (slash) {
2670                         int newbaselen = slash + 1 - f->name;
2671                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2672                         sources++;
2673                 } else {
2674                         this = f->changed;
2675                         dir->files++;
2676                         dir->nr--;
2677                         sources += 2;
2678                 }
2679                 this_dir += this;
2680         }
2681
2682         /*
2683          * We don't report dirstat's for
2684          *  - the top level
2685          *  - or cases where everything came from a single directory
2686          *    under this directory (sources == 1).
2687          */
2688         if (baselen && sources != 1) {
2689                 if (this_dir) {
2690                         int permille = this_dir * 1000 / changed;
2691                         if (permille >= dir->permille) {
2692                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2693                                         permille / 10, permille % 10, baselen, base);
2694                                 if (!dir->cumulative)
2695                                         return 0;
2696                         }
2697                 }
2698         }
2699         return this_dir;
2700 }
2701
2702 static int dirstat_compare(const void *_a, const void *_b)
2703 {
2704         const struct dirstat_file *a = _a;
2705         const struct dirstat_file *b = _b;
2706         return strcmp(a->name, b->name);
2707 }
2708
2709 static void show_dirstat(struct diff_options *options)
2710 {
2711         int i;
2712         unsigned long changed;
2713         struct dirstat_dir dir;
2714         struct diff_queue_struct *q = &diff_queued_diff;
2715
2716         dir.files = NULL;
2717         dir.alloc = 0;
2718         dir.nr = 0;
2719         dir.permille = options->dirstat_permille;
2720         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2721
2722         changed = 0;
2723         for (i = 0; i < q->nr; i++) {
2724                 struct diff_filepair *p = q->queue[i];
2725                 const char *name;
2726                 unsigned long copied, added, damage;
2727                 int content_changed;
2728
2729                 name = p->two->path ? p->two->path : p->one->path;
2730
2731                 if (p->one->oid_valid && p->two->oid_valid)
2732                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2733                 else
2734                         content_changed = 1;
2735
2736                 if (!content_changed) {
2737                         /*
2738                          * The SHA1 has not changed, so pre-/post-content is
2739                          * identical. We can therefore skip looking at the
2740                          * file contents altogether.
2741                          */
2742                         damage = 0;
2743                         goto found_damage;
2744                 }
2745
2746                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2747                         /*
2748                          * In --dirstat-by-file mode, we don't really need to
2749                          * look at the actual file contents at all.
2750                          * The fact that the SHA1 changed is enough for us to
2751                          * add this file to the list of results
2752                          * (with each file contributing equal damage).
2753                          */
2754                         damage = 1;
2755                         goto found_damage;
2756                 }
2757
2758                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2759                         diff_populate_filespec(p->one, 0);
2760                         diff_populate_filespec(p->two, 0);
2761                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2762                                                &copied, &added);
2763                         diff_free_filespec_data(p->one);
2764                         diff_free_filespec_data(p->two);
2765                 } else if (DIFF_FILE_VALID(p->one)) {
2766                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2767                         copied = added = 0;
2768                         diff_free_filespec_data(p->one);
2769                 } else if (DIFF_FILE_VALID(p->two)) {
2770                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2771                         copied = 0;
2772                         added = p->two->size;
2773                         diff_free_filespec_data(p->two);
2774                 } else
2775                         continue;
2776
2777                 /*
2778                  * Original minus copied is the removed material,
2779                  * added is the new material.  They are both damages
2780                  * made to the preimage.
2781                  * If the resulting damage is zero, we know that
2782                  * diffcore_count_changes() considers the two entries to
2783                  * be identical, but since content_changed is true, we
2784                  * know that there must have been _some_ kind of change,
2785                  * so we force all entries to have damage > 0.
2786                  */
2787                 damage = (p->one->size - copied) + added;
2788                 if (!damage)
2789                         damage = 1;
2790
2791 found_damage:
2792                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2793                 dir.files[dir.nr].name = name;
2794                 dir.files[dir.nr].changed = damage;
2795                 changed += damage;
2796                 dir.nr++;
2797         }
2798
2799         /* This can happen even with many files, if everything was renames */
2800         if (!changed)
2801                 return;
2802
2803         /* Show all directories with more than x% of the changes */
2804         QSORT(dir.files, dir.nr, dirstat_compare);
2805         gather_dirstat(options, &dir, changed, "", 0);
2806 }
2807
2808 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2809 {
2810         int i;
2811         unsigned long changed;
2812         struct dirstat_dir dir;
2813
2814         if (data->nr == 0)
2815                 return;
2816
2817         dir.files = NULL;
2818         dir.alloc = 0;
2819         dir.nr = 0;
2820         dir.permille = options->dirstat_permille;
2821         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2822
2823         changed = 0;
2824         for (i = 0; i < data->nr; i++) {
2825                 struct diffstat_file *file = data->files[i];
2826                 unsigned long damage = file->added + file->deleted;
2827                 if (file->is_binary)
2828                         /*
2829                          * binary files counts bytes, not lines. Must find some
2830                          * way to normalize binary bytes vs. textual lines.
2831                          * The following heuristic assumes that there are 64
2832                          * bytes per "line".
2833                          * This is stupid and ugly, but very cheap...
2834                          */
2835                         damage = (damage + 63) / 64;
2836                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2837                 dir.files[dir.nr].name = file->name;
2838                 dir.files[dir.nr].changed = damage;
2839                 changed += damage;
2840                 dir.nr++;
2841         }
2842
2843         /* This can happen even with many files, if everything was renames */
2844         if (!changed)
2845                 return;
2846
2847         /* Show all directories with more than x% of the changes */
2848         QSORT(dir.files, dir.nr, dirstat_compare);
2849         gather_dirstat(options, &dir, changed, "", 0);
2850 }
2851
2852 static void free_diffstat_info(struct diffstat_t *diffstat)
2853 {
2854         int i;
2855         for (i = 0; i < diffstat->nr; i++) {
2856                 struct diffstat_file *f = diffstat->files[i];
2857                 if (f->name != f->print_name)
2858                         free(f->print_name);
2859                 free(f->name);
2860                 free(f->from_name);
2861                 free(f);
2862         }
2863         free(diffstat->files);
2864 }
2865
2866 struct checkdiff_t {
2867         const char *filename;
2868         int lineno;
2869         int conflict_marker_size;
2870         struct diff_options *o;
2871         unsigned ws_rule;
2872         unsigned status;
2873 };
2874
2875 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2876 {
2877         char firstchar;
2878         int cnt;
2879
2880         if (len < marker_size + 1)
2881                 return 0;
2882         firstchar = line[0];
2883         switch (firstchar) {
2884         case '=': case '>': case '<': case '|':
2885                 break;
2886         default:
2887                 return 0;
2888         }
2889         for (cnt = 1; cnt < marker_size; cnt++)
2890                 if (line[cnt] != firstchar)
2891                         return 0;
2892         /* line[1] thru line[marker_size-1] are same as firstchar */
2893         if (len < marker_size + 1 || !isspace(line[marker_size]))
2894                 return 0;
2895         return 1;
2896 }
2897
2898 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2899 {
2900         struct checkdiff_t *data = priv;
2901         int marker_size = data->conflict_marker_size;
2902         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2903         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2904         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2905         char *err;
2906         const char *line_prefix;
2907
2908         assert(data->o);
2909         line_prefix = diff_line_prefix(data->o);
2910
2911         if (line[0] == '+') {
2912                 unsigned bad;
2913                 data->lineno++;
2914                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2915                         data->status |= 1;
2916                         fprintf(data->o->file,
2917                                 "%s%s:%d: leftover conflict marker\n",
2918                                 line_prefix, data->filename, data->lineno);
2919                 }
2920                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2921                 if (!bad)
2922                         return;
2923                 data->status |= bad;
2924                 err = whitespace_error_string(bad);
2925                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2926                         line_prefix, data->filename, data->lineno, err);
2927                 free(err);
2928                 emit_line(data->o, set, reset, line, 1);
2929                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2930                               data->o->file, set, reset, ws);
2931         } else if (line[0] == ' ') {
2932                 data->lineno++;
2933         } else if (line[0] == '@') {
2934                 char *plus = strchr(line, '+');
2935                 if (plus)
2936                         data->lineno = strtol(plus, NULL, 10) - 1;
2937                 else
2938                         die("invalid diff");
2939         }
2940 }
2941
2942 static unsigned char *deflate_it(char *data,
2943                                  unsigned long size,
2944                                  unsigned long *result_size)
2945 {
2946         int bound;
2947         unsigned char *deflated;
2948         git_zstream stream;
2949
2950         git_deflate_init(&stream, zlib_compression_level);
2951         bound = git_deflate_bound(&stream, size);
2952         deflated = xmalloc(bound);
2953         stream.next_out = deflated;
2954         stream.avail_out = bound;
2955
2956         stream.next_in = (unsigned char *)data;
2957         stream.avail_in = size;
2958         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2959                 ; /* nothing */
2960         git_deflate_end(&stream);
2961         *result_size = stream.total_out;
2962         return deflated;
2963 }
2964
2965 static void emit_binary_diff_body(struct diff_options *o,
2966                                   mmfile_t *one, mmfile_t *two)
2967 {
2968         void *cp;
2969         void *delta;
2970         void *deflated;
2971         void *data;
2972         unsigned long orig_size;
2973         unsigned long delta_size;
2974         unsigned long deflate_size;
2975         unsigned long data_size;
2976
2977         /* We could do deflated delta, or we could do just deflated two,
2978          * whichever is smaller.
2979          */
2980         delta = NULL;
2981         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2982         if (one->size && two->size) {
2983                 delta = diff_delta(one->ptr, one->size,
2984                                    two->ptr, two->size,
2985                                    &delta_size, deflate_size);
2986                 if (delta) {
2987                         void *to_free = delta;
2988                         orig_size = delta_size;
2989                         delta = deflate_it(delta, delta_size, &delta_size);
2990                         free(to_free);
2991                 }
2992         }
2993
2994         if (delta && delta_size < deflate_size) {
2995                 char *s = xstrfmt("%lu", orig_size);
2996                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2997                                  s, strlen(s), 0);
2998                 free(s);
2999                 free(deflated);
3000                 data = delta;
3001                 data_size = delta_size;
3002         } else {
3003                 char *s = xstrfmt("%lu", two->size);
3004                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3005                                  s, strlen(s), 0);
3006                 free(s);
3007                 free(delta);
3008                 data = deflated;
3009                 data_size = deflate_size;
3010         }
3011
3012         /* emit data encoded in base85 */
3013         cp = data;
3014         while (data_size) {
3015                 int len;
3016                 int bytes = (52 < data_size) ? 52 : data_size;
3017                 char line[71];
3018                 data_size -= bytes;
3019                 if (bytes <= 26)
3020                         line[0] = bytes + 'A' - 1;
3021                 else
3022                         line[0] = bytes - 26 + 'a' - 1;
3023                 encode_85(line + 1, cp, bytes);
3024                 cp = (char *) cp + bytes;
3025
3026                 len = strlen(line);
3027                 line[len++] = '\n';
3028                 line[len] = '\0';
3029
3030                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3031                                  line, len, 0);
3032         }
3033         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3034         free(data);
3035 }
3036
3037 static void emit_binary_diff(struct diff_options *o,
3038                              mmfile_t *one, mmfile_t *two)
3039 {
3040         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3041         emit_binary_diff_body(o, one, two);
3042         emit_binary_diff_body(o, two, one);
3043 }
3044
3045 int diff_filespec_is_binary(struct diff_filespec *one)
3046 {
3047         if (one->is_binary == -1) {
3048                 diff_filespec_load_driver(one);
3049                 if (one->driver->binary != -1)
3050                         one->is_binary = one->driver->binary;
3051                 else {
3052                         if (!one->data && DIFF_FILE_VALID(one))
3053                                 diff_populate_filespec(one, CHECK_BINARY);
3054                         if (one->is_binary == -1 && one->data)
3055                                 one->is_binary = buffer_is_binary(one->data,
3056                                                 one->size);
3057                         if (one->is_binary == -1)
3058                                 one->is_binary = 0;
3059                 }
3060         }
3061         return one->is_binary;
3062 }
3063
3064 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3065 {
3066         diff_filespec_load_driver(one);
3067         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3068 }
3069
3070 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3071 {
3072         if (!options->a_prefix)
3073                 options->a_prefix = a;
3074         if (!options->b_prefix)
3075                 options->b_prefix = b;
3076 }
3077
3078 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3079 {
3080         if (!DIFF_FILE_VALID(one))
3081                 return NULL;
3082
3083         diff_filespec_load_driver(one);
3084         return userdiff_get_textconv(one->driver);
3085 }
3086
3087 static void builtin_diff(const char *name_a,
3088                          const char *name_b,
3089                          struct diff_filespec *one,
3090                          struct diff_filespec *two,
3091                          const char *xfrm_msg,
3092                          int must_show_header,
3093                          struct diff_options *o,
3094                          int complete_rewrite)
3095 {
3096         mmfile_t mf1, mf2;
3097         const char *lbl[2];
3098         char *a_one, *b_two;
3099         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3100         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3101         const char *a_prefix, *b_prefix;
3102         struct userdiff_driver *textconv_one = NULL;
3103         struct userdiff_driver *textconv_two = NULL;
3104         struct strbuf header = STRBUF_INIT;
3105         const char *line_prefix = diff_line_prefix(o);
3106
3107         diff_set_mnemonic_prefix(o, "a/", "b/");
3108         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
3109                 a_prefix = o->b_prefix;
3110                 b_prefix = o->a_prefix;
3111         } else {
3112                 a_prefix = o->a_prefix;
3113                 b_prefix = o->b_prefix;
3114         }
3115
3116         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3117             (!one->mode || S_ISGITLINK(one->mode)) &&
3118             (!two->mode || S_ISGITLINK(two->mode))) {
3119                 show_submodule_summary(o, one->path ? one->path : two->path,
3120                                 &one->oid, &two->oid,
3121                                 two->dirty_submodule);
3122                 return;
3123         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3124                    (!one->mode || S_ISGITLINK(one->mode)) &&
3125                    (!two->mode || S_ISGITLINK(two->mode))) {
3126                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3127                                 &one->oid, &two->oid,
3128                                 two->dirty_submodule);
3129                 return;
3130         }
3131
3132         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
3133                 textconv_one = get_textconv(one);
3134                 textconv_two = get_textconv(two);
3135         }
3136
3137         /* Never use a non-valid filename anywhere if at all possible */
3138         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3139         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3140
3141         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3142         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3143         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3144         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3145         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3146         if (lbl[0][0] == '/') {
3147                 /* /dev/null */
3148                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3149                 if (xfrm_msg)
3150                         strbuf_addstr(&header, xfrm_msg);
3151                 must_show_header = 1;
3152         }
3153         else if (lbl[1][0] == '/') {
3154                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3155                 if (xfrm_msg)
3156                         strbuf_addstr(&header, xfrm_msg);
3157                 must_show_header = 1;
3158         }
3159         else {
3160                 if (one->mode != two->mode) {
3161                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3162                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3163                         must_show_header = 1;
3164                 }
3165                 if (xfrm_msg)
3166                         strbuf_addstr(&header, xfrm_msg);
3167
3168                 /*
3169                  * we do not run diff between different kind
3170                  * of objects.
3171                  */
3172                 if ((one->mode ^ two->mode) & S_IFMT)
3173                         goto free_ab_and_return;
3174                 if (complete_rewrite &&
3175                     (textconv_one || !diff_filespec_is_binary(one)) &&
3176                     (textconv_two || !diff_filespec_is_binary(two))) {
3177                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3178                                          header.buf, header.len, 0);
3179                         strbuf_reset(&header);
3180                         emit_rewrite_diff(name_a, name_b, one, two,
3181                                                 textconv_one, textconv_two, o);
3182                         o->found_changes = 1;
3183                         goto free_ab_and_return;
3184                 }
3185         }
3186
3187         if (o->irreversible_delete && lbl[1][0] == '/') {
3188                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3189                                  header.len, 0);
3190                 strbuf_reset(&header);
3191                 goto free_ab_and_return;
3192         } else if (!DIFF_OPT_TST(o, TEXT) &&
3193             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3194               (!textconv_two && diff_filespec_is_binary(two)) )) {
3195                 struct strbuf sb = STRBUF_INIT;
3196                 if (!one->data && !two->data &&
3197                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3198                     !DIFF_OPT_TST(o, BINARY)) {
3199                         if (!oidcmp(&one->oid, &two->oid)) {
3200                                 if (must_show_header)
3201                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3202                                                          header.buf, header.len,
3203                                                          0);
3204                                 goto free_ab_and_return;
3205                         }
3206                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3207                                          header.buf, header.len, 0);
3208                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3209                                     diff_line_prefix(o), lbl[0], lbl[1]);
3210                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3211                                          sb.buf, sb.len, 0);
3212                         strbuf_release(&sb);
3213                         goto free_ab_and_return;
3214                 }
3215                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3216                         die("unable to read files to diff");
3217                 /* Quite common confusing case */
3218                 if (mf1.size == mf2.size &&
3219                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3220                         if (must_show_header)
3221                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3222                                                  header.buf, header.len, 0);
3223                         goto free_ab_and_return;
3224                 }
3225                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3226                 strbuf_reset(&header);
3227                 if (DIFF_OPT_TST(o, BINARY))
3228                         emit_binary_diff(o, &mf1, &mf2);
3229                 else {
3230                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3231                                     diff_line_prefix(o), lbl[0], lbl[1]);
3232                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3233                                          sb.buf, sb.len, 0);
3234                         strbuf_release(&sb);
3235                 }
3236                 o->found_changes = 1;
3237         } else {
3238                 /* Crazy xdl interfaces.. */
3239                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3240                 const char *v;
3241                 xpparam_t xpp;
3242                 xdemitconf_t xecfg;
3243                 struct emit_callback ecbdata;
3244                 const struct userdiff_funcname *pe;
3245
3246                 if (must_show_header) {
3247                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3248                                          header.buf, header.len, 0);
3249                         strbuf_reset(&header);
3250                 }
3251
3252                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3253                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3254
3255                 pe = diff_funcname_pattern(one);
3256                 if (!pe)
3257                         pe = diff_funcname_pattern(two);
3258
3259                 memset(&xpp, 0, sizeof(xpp));
3260                 memset(&xecfg, 0, sizeof(xecfg));
3261                 memset(&ecbdata, 0, sizeof(ecbdata));
3262                 ecbdata.label_path = lbl;
3263                 ecbdata.color_diff = want_color(o->use_color);
3264                 ecbdata.ws_rule = whitespace_rule(name_b);
3265                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3266                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3267                 ecbdata.opt = o;
3268                 ecbdata.header = header.len ? &header : NULL;
3269                 xpp.flags = o->xdl_opts;
3270                 xecfg.ctxlen = o->context;
3271                 xecfg.interhunkctxlen = o->interhunkcontext;
3272                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3273                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
3274                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3275                 if (pe)
3276                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3277                 if (!diffopts)
3278                         ;
3279                 else if (skip_prefix(diffopts, "--unified=", &v))
3280                         xecfg.ctxlen = strtoul(v, NULL, 10);
3281                 else if (skip_prefix(diffopts, "-u", &v))
3282                         xecfg.ctxlen = strtoul(v, NULL, 10);
3283                 if (o->word_diff)
3284                         init_diff_words_data(&ecbdata, o, one, two);
3285                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3286                                   &xpp, &xecfg))
3287                         die("unable to generate diff for %s", one->path);
3288                 if (o->word_diff)
3289                         free_diff_words_data(&ecbdata);
3290                 if (textconv_one)
3291                         free(mf1.ptr);
3292                 if (textconv_two)
3293                         free(mf2.ptr);
3294                 xdiff_clear_find_func(&xecfg);
3295         }
3296
3297  free_ab_and_return:
3298         strbuf_release(&header);
3299         diff_free_filespec_data(one);
3300         diff_free_filespec_data(two);
3301         free(a_one);
3302         free(b_two);
3303         return;
3304 }
3305
3306 static void builtin_diffstat(const char *name_a, const char *name_b,
3307                              struct diff_filespec *one,
3308                              struct diff_filespec *two,
3309                              struct diffstat_t *diffstat,
3310                              struct diff_options *o,
3311                              struct diff_filepair *p)
3312 {
3313         mmfile_t mf1, mf2;
3314         struct diffstat_file *data;
3315         int same_contents;
3316         int complete_rewrite = 0;
3317
3318         if (!DIFF_PAIR_UNMERGED(p)) {
3319                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3320                         complete_rewrite = 1;
3321         }
3322
3323         data = diffstat_add(diffstat, name_a, name_b);
3324         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3325
3326         if (!one || !two) {
3327                 data->is_unmerged = 1;
3328                 return;
3329         }
3330
3331         same_contents = !oidcmp(&one->oid, &two->oid);
3332
3333         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3334                 data->is_binary = 1;
3335                 if (same_contents) {
3336                         data->added = 0;
3337                         data->deleted = 0;
3338                 } else {
3339                         data->added = diff_filespec_size(two);
3340                         data->deleted = diff_filespec_size(one);
3341                 }
3342         }
3343
3344         else if (complete_rewrite) {
3345                 diff_populate_filespec(one, 0);
3346                 diff_populate_filespec(two, 0);
3347                 data->deleted = count_lines(one->data, one->size);
3348                 data->added = count_lines(two->data, two->size);
3349         }
3350
3351         else if (!same_contents) {
3352                 /* Crazy xdl interfaces.. */
3353                 xpparam_t xpp;
3354                 xdemitconf_t xecfg;
3355
3356                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3357                         die("unable to read files to diff");
3358
3359                 memset(&xpp, 0, sizeof(xpp));
3360                 memset(&xecfg, 0, sizeof(xecfg));
3361                 xpp.flags = o->xdl_opts;
3362                 xecfg.ctxlen = o->context;
3363                 xecfg.interhunkctxlen = o->interhunkcontext;
3364                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3365                                   &xpp, &xecfg))
3366                         die("unable to generate diffstat for %s", one->path);
3367         }
3368
3369         diff_free_filespec_data(one);
3370         diff_free_filespec_data(two);
3371 }
3372
3373 static void builtin_checkdiff(const char *name_a, const char *name_b,
3374                               const char *attr_path,
3375                               struct diff_filespec *one,
3376                               struct diff_filespec *two,
3377                               struct diff_options *o)
3378 {
3379         mmfile_t mf1, mf2;
3380         struct checkdiff_t data;
3381
3382         if (!two)
3383                 return;
3384
3385         memset(&data, 0, sizeof(data));
3386         data.filename = name_b ? name_b : name_a;
3387         data.lineno = 0;
3388         data.o = o;
3389         data.ws_rule = whitespace_rule(attr_path);
3390         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3391
3392         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3393                 die("unable to read files to diff");
3394
3395         /*
3396          * All the other codepaths check both sides, but not checking
3397          * the "old" side here is deliberate.  We are checking the newly
3398          * introduced changes, and as long as the "new" side is text, we
3399          * can and should check what it introduces.
3400          */
3401         if (diff_filespec_is_binary(two))
3402                 goto free_and_return;
3403         else {
3404                 /* Crazy xdl interfaces.. */
3405                 xpparam_t xpp;
3406                 xdemitconf_t xecfg;
3407
3408                 memset(&xpp, 0, sizeof(xpp));
3409                 memset(&xecfg, 0, sizeof(xecfg));
3410                 xecfg.ctxlen = 1; /* at least one context line */
3411                 xpp.flags = 0;
3412                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3413                                   &xpp, &xecfg))
3414                         die("unable to generate checkdiff for %s", one->path);
3415
3416                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3417                         struct emit_callback ecbdata;
3418                         int blank_at_eof;
3419
3420                         ecbdata.ws_rule = data.ws_rule;
3421                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3422                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3423
3424                         if (blank_at_eof) {
3425                                 static char *err;
3426                                 if (!err)
3427                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3428                                 fprintf(o->file, "%s:%d: %s.\n",
3429                                         data.filename, blank_at_eof, err);
3430                                 data.status = 1; /* report errors */
3431                         }
3432                 }
3433         }
3434  free_and_return:
3435         diff_free_filespec_data(one);
3436         diff_free_filespec_data(two);
3437         if (data.status)
3438                 DIFF_OPT_SET(o, CHECK_FAILED);
3439 }
3440
3441 struct diff_filespec *alloc_filespec(const char *path)
3442 {
3443         struct diff_filespec *spec;
3444
3445         FLEXPTR_ALLOC_STR(spec, path, path);
3446         spec->count = 1;
3447         spec->is_binary = -1;
3448         return spec;
3449 }
3450
3451 void free_filespec(struct diff_filespec *spec)
3452 {
3453         if (!--spec->count) {
3454                 diff_free_filespec_data(spec);
3455                 free(spec);
3456         }
3457 }
3458
3459 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3460                    int oid_valid, unsigned short mode)
3461 {
3462         if (mode) {
3463                 spec->mode = canon_mode(mode);
3464                 oidcpy(&spec->oid, oid);
3465                 spec->oid_valid = oid_valid;
3466         }
3467 }
3468
3469 /*
3470  * Given a name and sha1 pair, if the index tells us the file in
3471  * the work tree has that object contents, return true, so that
3472  * prepare_temp_file() does not have to inflate and extract.
3473  */
3474 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3475 {
3476         const struct cache_entry *ce;
3477         struct stat st;
3478         int pos, len;
3479
3480         /*
3481          * We do not read the cache ourselves here, because the
3482          * benchmark with my previous version that always reads cache
3483          * shows that it makes things worse for diff-tree comparing
3484          * two linux-2.6 kernel trees in an already checked out work
3485          * tree.  This is because most diff-tree comparisons deal with
3486          * only a small number of files, while reading the cache is
3487          * expensive for a large project, and its cost outweighs the
3488          * savings we get by not inflating the object to a temporary
3489          * file.  Practically, this code only helps when we are used
3490          * by diff-cache --cached, which does read the cache before
3491          * calling us.
3492          */
3493         if (!active_cache)
3494                 return 0;
3495
3496         /* We want to avoid the working directory if our caller
3497          * doesn't need the data in a normal file, this system
3498          * is rather slow with its stat/open/mmap/close syscalls,
3499          * and the object is contained in a pack file.  The pack
3500          * is probably already open and will be faster to obtain
3501          * the data through than the working directory.  Loose
3502          * objects however would tend to be slower as they need
3503          * to be individually opened and inflated.
3504          */
3505         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3506                 return 0;
3507
3508         /*
3509          * Similarly, if we'd have to convert the file contents anyway, that
3510          * makes the optimization not worthwhile.
3511          */
3512         if (!want_file && would_convert_to_git(&the_index, name))
3513                 return 0;
3514
3515         len = strlen(name);
3516         pos = cache_name_pos(name, len);
3517         if (pos < 0)
3518                 return 0;
3519         ce = active_cache[pos];
3520
3521         /*
3522          * This is not the sha1 we are looking for, or
3523          * unreusable because it is not a regular file.
3524          */
3525         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3526                 return 0;
3527
3528         /*
3529          * If ce is marked as "assume unchanged", there is no
3530          * guarantee that work tree matches what we are looking for.
3531          */
3532         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3533                 return 0;
3534
3535         /*
3536          * If ce matches the file in the work tree, we can reuse it.
3537          */
3538         if (ce_uptodate(ce) ||
3539             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3540                 return 1;
3541
3542         return 0;
3543 }
3544
3545 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3546 {
3547         struct strbuf buf = STRBUF_INIT;
3548         char *dirty = "";
3549
3550         /* Are we looking at the work tree? */
3551         if (s->dirty_submodule)
3552                 dirty = "-dirty";
3553
3554         strbuf_addf(&buf, "Subproject commit %s%s\n",
3555                     oid_to_hex(&s->oid), dirty);
3556         s->size = buf.len;
3557         if (size_only) {
3558                 s->data = NULL;
3559                 strbuf_release(&buf);
3560         } else {
3561                 s->data = strbuf_detach(&buf, NULL);
3562                 s->should_free = 1;
3563         }
3564         return 0;
3565 }
3566
3567 /*
3568  * While doing rename detection and pickaxe operation, we may need to
3569  * grab the data for the blob (or file) for our own in-core comparison.
3570  * diff_filespec has data and size fields for this purpose.
3571  */
3572 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3573 {
3574         int size_only = flags & CHECK_SIZE_ONLY;
3575         int err = 0;
3576         /*
3577          * demote FAIL to WARN to allow inspecting the situation
3578          * instead of refusing.
3579          */
3580         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3581                                     ? SAFE_CRLF_WARN
3582                                     : safe_crlf);
3583
3584         if (!DIFF_FILE_VALID(s))
3585                 die("internal error: asking to populate invalid file.");
3586         if (S_ISDIR(s->mode))
3587                 return -1;
3588
3589         if (s->data)
3590                 return 0;
3591
3592         if (size_only && 0 < s->size)
3593                 return 0;
3594
3595         if (S_ISGITLINK(s->mode))
3596                 return diff_populate_gitlink(s, size_only);
3597
3598         if (!s->oid_valid ||
3599             reuse_worktree_file(s->path, &s->oid, 0)) {
3600                 struct strbuf buf = STRBUF_INIT;
3601                 struct stat st;
3602                 int fd;
3603
3604                 if (lstat(s->path, &st) < 0) {
3605                         if (errno == ENOENT) {
3606                         err_empty:
3607                                 err = -1;
3608                         empty:
3609                                 s->data = (char *)"";
3610                                 s->size = 0;
3611                                 return err;
3612                         }
3613                 }
3614                 s->size = xsize_t(st.st_size);
3615                 if (!s->size)
3616                         goto empty;
3617                 if (S_ISLNK(st.st_mode)) {
3618                         struct strbuf sb = STRBUF_INIT;
3619
3620                         if (strbuf_readlink(&sb, s->path, s->size))
3621                                 goto err_empty;
3622                         s->size = sb.len;
3623                         s->data = strbuf_detach(&sb, NULL);
3624                         s->should_free = 1;
3625                         return 0;
3626                 }
3627
3628                 /*
3629                  * Even if the caller would be happy with getting
3630                  * only the size, we cannot return early at this
3631                  * point if the path requires us to run the content
3632                  * conversion.
3633                  */
3634                 if (size_only && !would_convert_to_git(&the_index, s->path))
3635                         return 0;
3636
3637                 /*
3638                  * Note: this check uses xsize_t(st.st_size) that may
3639                  * not be the true size of the blob after it goes
3640                  * through convert_to_git().  This may not strictly be
3641                  * correct, but the whole point of big_file_threshold
3642                  * and is_binary check being that we want to avoid
3643                  * opening the file and inspecting the contents, this
3644                  * is probably fine.
3645                  */
3646                 if ((flags & CHECK_BINARY) &&
3647                     s->size > big_file_threshold && s->is_binary == -1) {
3648                         s->is_binary = 1;
3649                         return 0;
3650                 }
3651                 fd = open(s->path, O_RDONLY);
3652                 if (fd < 0)
3653                         goto err_empty;
3654                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3655                 close(fd);
3656                 s->should_munmap = 1;
3657
3658                 /*
3659                  * Convert from working tree format to canonical git format
3660                  */
3661                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3662                         size_t size = 0;
3663                         munmap(s->data, s->size);
3664                         s->should_munmap = 0;
3665                         s->data = strbuf_detach(&buf, &size);
3666                         s->size = size;
3667                         s->should_free = 1;
3668                 }
3669         }
3670         else {
3671                 enum object_type type;
3672                 if (size_only || (flags & CHECK_BINARY)) {
3673                         type = sha1_object_info(s->oid.hash, &s->size);
3674                         if (type < 0)
3675                                 die("unable to read %s",
3676                                     oid_to_hex(&s->oid));
3677                         if (size_only)
3678                                 return 0;
3679                         if (s->size > big_file_threshold && s->is_binary == -1) {
3680                                 s->is_binary = 1;
3681                                 return 0;
3682                         }
3683                 }
3684                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3685                 if (!s->data)
3686                         die("unable to read %s", oid_to_hex(&s->oid));
3687                 s->should_free = 1;
3688         }
3689         return 0;
3690 }
3691
3692 void diff_free_filespec_blob(struct diff_filespec *s)
3693 {
3694         if (s->should_free)
3695                 free(s->data);
3696         else if (s->should_munmap)
3697                 munmap(s->data, s->size);
3698
3699         if (s->should_free || s->should_munmap) {
3700                 s->should_free = s->should_munmap = 0;
3701                 s->data = NULL;
3702         }
3703 }
3704
3705 void diff_free_filespec_data(struct diff_filespec *s)
3706 {
3707         diff_free_filespec_blob(s);
3708         FREE_AND_NULL(s->cnt_data);
3709 }
3710
3711 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3712                            void *blob,
3713                            unsigned long size,
3714                            const struct object_id *oid,
3715                            int mode)
3716 {
3717         int fd;
3718         struct strbuf buf = STRBUF_INIT;
3719         struct strbuf template = STRBUF_INIT;
3720         char *path_dup = xstrdup(path);
3721         const char *base = basename(path_dup);
3722
3723         /* Generate "XXXXXX_basename.ext" */
3724         strbuf_addstr(&template, "XXXXXX_");
3725         strbuf_addstr(&template, base);
3726
3727         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
3728         if (fd < 0)
3729                 die_errno("unable to create temp-file");
3730         if (convert_to_working_tree(path,
3731                         (const char *)blob, (size_t)size, &buf)) {
3732                 blob = buf.buf;
3733                 size = buf.len;
3734         }
3735         if (write_in_full(fd, blob, size) != size)
3736                 die_errno("unable to write temp-file");
3737         close_tempfile(&temp->tempfile);
3738         temp->name = get_tempfile_path(&temp->tempfile);
3739         oid_to_hex_r(temp->hex, oid);
3740         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3741         strbuf_release(&buf);
3742         strbuf_release(&template);
3743         free(path_dup);
3744 }
3745
3746 static struct diff_tempfile *prepare_temp_file(const char *name,
3747                 struct diff_filespec *one)
3748 {
3749         struct diff_tempfile *temp = claim_diff_tempfile();
3750
3751         if (!DIFF_FILE_VALID(one)) {
3752         not_a_valid_file:
3753                 /* A '-' entry produces this for file-2, and
3754                  * a '+' entry produces this for file-1.
3755                  */
3756                 temp->name = "/dev/null";
3757                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3758                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3759                 return temp;
3760         }
3761
3762         if (!S_ISGITLINK(one->mode) &&
3763             (!one->oid_valid ||
3764              reuse_worktree_file(name, &one->oid, 1))) {
3765                 struct stat st;
3766                 if (lstat(name, &st) < 0) {
3767                         if (errno == ENOENT)
3768                                 goto not_a_valid_file;
3769                         die_errno("stat(%s)", name);
3770                 }
3771                 if (S_ISLNK(st.st_mode)) {
3772                         struct strbuf sb = STRBUF_INIT;
3773                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3774                                 die_errno("readlink(%s)", name);
3775                         prep_temp_blob(name, temp, sb.buf, sb.len,
3776                                        (one->oid_valid ?
3777                                         &one->oid : &null_oid),
3778                                        (one->oid_valid ?
3779                                         one->mode : S_IFLNK));
3780                         strbuf_release(&sb);
3781                 }
3782                 else {
3783                         /* we can borrow from the file in the work tree */
3784                         temp->name = name;
3785                         if (!one->oid_valid)
3786                                 oid_to_hex_r(temp->hex, &null_oid);
3787                         else
3788                                 oid_to_hex_r(temp->hex, &one->oid);
3789                         /* Even though we may sometimes borrow the
3790                          * contents from the work tree, we always want
3791                          * one->mode.  mode is trustworthy even when
3792                          * !(one->oid_valid), as long as
3793                          * DIFF_FILE_VALID(one).
3794                          */
3795                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3796                 }
3797                 return temp;
3798         }
3799         else {
3800                 if (diff_populate_filespec(one, 0))
3801                         die("cannot read data blob for %s", one->path);
3802                 prep_temp_blob(name, temp, one->data, one->size,
3803                                &one->oid, one->mode);
3804         }
3805         return temp;
3806 }
3807
3808 static void add_external_diff_name(struct argv_array *argv,
3809                                    const char *name,
3810                                    struct diff_filespec *df)
3811 {
3812         struct diff_tempfile *temp = prepare_temp_file(name, df);
3813         argv_array_push(argv, temp->name);
3814         argv_array_push(argv, temp->hex);
3815         argv_array_push(argv, temp->mode);
3816 }
3817
3818 /* An external diff command takes:
3819  *
3820  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3821  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3822  *
3823  */
3824 static void run_external_diff(const char *pgm,
3825                               const char *name,
3826                               const char *other,
3827                               struct diff_filespec *one,
3828                               struct diff_filespec *two,
3829                               const char *xfrm_msg,
3830                               int complete_rewrite,
3831                               struct diff_options *o)
3832 {
3833         struct argv_array argv = ARGV_ARRAY_INIT;
3834         struct argv_array env = ARGV_ARRAY_INIT;
3835         struct diff_queue_struct *q = &diff_queued_diff;
3836
3837         argv_array_push(&argv, pgm);
3838         argv_array_push(&argv, name);
3839
3840         if (one && two) {
3841                 add_external_diff_name(&argv, name, one);
3842                 if (!other)
3843                         add_external_diff_name(&argv, name, two);
3844                 else {
3845                         add_external_diff_name(&argv, other, two);
3846                         argv_array_push(&argv, other);
3847                         argv_array_push(&argv, xfrm_msg);
3848                 }
3849         }
3850
3851         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3852         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3853
3854         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3855                 die(_("external diff died, stopping at %s"), name);
3856
3857         remove_tempfile();
3858         argv_array_clear(&argv);
3859         argv_array_clear(&env);
3860 }
3861
3862 static int similarity_index(struct diff_filepair *p)
3863 {
3864         return p->score * 100 / MAX_SCORE;
3865 }
3866
3867 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3868 {
3869         if (startup_info->have_repository)
3870                 return find_unique_abbrev(oid->hash, abbrev);
3871         else {
3872                 char *hex = oid_to_hex(oid);
3873                 if (abbrev < 0)
3874                         abbrev = FALLBACK_DEFAULT_ABBREV;
3875                 if (abbrev > GIT_SHA1_HEXSZ)
3876                         die("BUG: oid abbreviation out of range: %d", abbrev);
3877                 if (abbrev)
3878                         hex[abbrev] = '\0';
3879                 return hex;
3880         }
3881 }
3882
3883 static void fill_metainfo(struct strbuf *msg,
3884                           const char *name,
3885                           const char *other,
3886                           struct diff_filespec *one,
3887                           struct diff_filespec *two,
3888                           struct diff_options *o,
3889                           struct diff_filepair *p,
3890                           int *must_show_header,
3891                           int use_color)
3892 {
3893         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3894         const char *reset = diff_get_color(use_color, DIFF_RESET);
3895         const char *line_prefix = diff_line_prefix(o);
3896
3897         *must_show_header = 1;
3898         strbuf_init(msg, PATH_MAX * 2 + 300);
3899         switch (p->status) {
3900         case DIFF_STATUS_COPIED:
3901                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3902                             line_prefix, set, similarity_index(p));
3903                 strbuf_addf(msg, "%s\n%s%scopy from ",
3904                             reset,  line_prefix, set);
3905                 quote_c_style(name, msg, NULL, 0);
3906                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3907                 quote_c_style(other, msg, NULL, 0);
3908                 strbuf_addf(msg, "%s\n", reset);
3909                 break;
3910         case DIFF_STATUS_RENAMED:
3911                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3912                             line_prefix, set, similarity_index(p));
3913                 strbuf_addf(msg, "%s\n%s%srename from ",
3914                             reset, line_prefix, set);
3915                 quote_c_style(name, msg, NULL, 0);
3916                 strbuf_addf(msg, "%s\n%s%srename to ",
3917                             reset, line_prefix, set);
3918                 quote_c_style(other, msg, NULL, 0);
3919                 strbuf_addf(msg, "%s\n", reset);
3920                 break;
3921         case DIFF_STATUS_MODIFIED:
3922                 if (p->score) {
3923                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3924                                     line_prefix,
3925                                     set, similarity_index(p), reset);
3926                         break;
3927                 }
3928                 /* fallthru */
3929         default:
3930                 *must_show_header = 0;
3931         }
3932         if (one && two && oidcmp(&one->oid, &two->oid)) {
3933                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3934
3935                 if (DIFF_OPT_TST(o, BINARY)) {
3936                         mmfile_t mf;
3937                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3938                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3939                                 abbrev = 40;
3940                 }
3941                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3942                             diff_abbrev_oid(&one->oid, abbrev),
3943                             diff_abbrev_oid(&two->oid, abbrev));
3944                 if (one->mode == two->mode)
3945                         strbuf_addf(msg, " %06o", one->mode);
3946                 strbuf_addf(msg, "%s\n", reset);
3947         }
3948 }
3949
3950 static void run_diff_cmd(const char *pgm,
3951                          const char *name,
3952                          const char *other,
3953                          const char *attr_path,
3954                          struct diff_filespec *one,
3955                          struct diff_filespec *two,
3956                          struct strbuf *msg,
3957                          struct diff_options *o,
3958                          struct diff_filepair *p)
3959 {
3960         const char *xfrm_msg = NULL;
3961         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3962         int must_show_header = 0;
3963
3964
3965         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3966                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3967                 if (drv && drv->external)
3968                         pgm = drv->external;
3969         }
3970
3971         if (msg) {
3972                 /*
3973                  * don't use colors when the header is intended for an
3974                  * external diff driver
3975                  */
3976                 fill_metainfo(msg, name, other, one, two, o, p,
3977                               &must_show_header,
3978                               want_color(o->use_color) && !pgm);
3979                 xfrm_msg = msg->len ? msg->buf : NULL;
3980         }
3981
3982         if (pgm) {
3983                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3984                                   complete_rewrite, o);
3985                 return;
3986         }
3987         if (one && two)
3988                 builtin_diff(name, other ? other : name,
3989                              one, two, xfrm_msg, must_show_header,
3990                              o, complete_rewrite);
3991         else
3992                 fprintf(o->file, "* Unmerged path %s\n", name);
3993 }
3994
3995 static void diff_fill_oid_info(struct diff_filespec *one)
3996 {
3997         if (DIFF_FILE_VALID(one)) {
3998                 if (!one->oid_valid) {
3999                         struct stat st;
4000                         if (one->is_stdin) {
4001                                 oidclr(&one->oid);
4002                                 return;
4003                         }
4004                         if (lstat(one->path, &st) < 0)
4005                                 die_errno("stat '%s'", one->path);
4006                         if (index_path(one->oid.hash, one->path, &st, 0))
4007                                 die("cannot hash %s", one->path);
4008                 }
4009         }
4010         else
4011                 oidclr(&one->oid);
4012 }
4013
4014 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4015 {
4016         /* Strip the prefix but do not molest /dev/null and absolute paths */
4017         if (*namep && **namep != '/') {
4018                 *namep += prefix_length;
4019                 if (**namep == '/')
4020                         ++*namep;
4021         }
4022         if (*otherp && **otherp != '/') {
4023                 *otherp += prefix_length;
4024                 if (**otherp == '/')
4025                         ++*otherp;
4026         }
4027 }
4028
4029 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4030 {
4031         const char *pgm = external_diff();
4032         struct strbuf msg;
4033         struct diff_filespec *one = p->one;
4034         struct diff_filespec *two = p->two;
4035         const char *name;
4036         const char *other;
4037         const char *attr_path;
4038
4039         name  = one->path;
4040         other = (strcmp(name, two->path) ? two->path : NULL);
4041         attr_path = name;
4042         if (o->prefix_length)
4043                 strip_prefix(o->prefix_length, &name, &other);
4044
4045         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
4046                 pgm = NULL;
4047
4048         if (DIFF_PAIR_UNMERGED(p)) {
4049                 run_diff_cmd(pgm, name, NULL, attr_path,
4050                              NULL, NULL, NULL, o, p);
4051                 return;
4052         }
4053
4054         diff_fill_oid_info(one);
4055         diff_fill_oid_info(two);
4056
4057         if (!pgm &&
4058             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4059             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4060                 /*
4061                  * a filepair that changes between file and symlink
4062                  * needs to be split into deletion and creation.
4063                  */
4064                 struct diff_filespec *null = alloc_filespec(two->path);
4065                 run_diff_cmd(NULL, name, other, attr_path,
4066                              one, null, &msg, o, p);
4067                 free(null);
4068                 strbuf_release(&msg);
4069
4070                 null = alloc_filespec(one->path);
4071                 run_diff_cmd(NULL, name, other, attr_path,
4072                              null, two, &msg, o, p);
4073                 free(null);
4074         }
4075         else
4076                 run_diff_cmd(pgm, name, other, attr_path,
4077                              one, two, &msg, o, p);
4078
4079         strbuf_release(&msg);
4080 }
4081
4082 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4083                          struct diffstat_t *diffstat)
4084 {
4085         const char *name;
4086         const char *other;
4087
4088         if (DIFF_PAIR_UNMERGED(p)) {
4089                 /* unmerged */
4090                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4091                 return;
4092         }
4093
4094         name = p->one->path;
4095         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4096
4097         if (o->prefix_length)
4098                 strip_prefix(o->prefix_length, &name, &other);
4099
4100         diff_fill_oid_info(p->one);
4101         diff_fill_oid_info(p->two);
4102
4103         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4104 }
4105
4106 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4107 {
4108         const char *name;
4109         const char *other;
4110         const char *attr_path;
4111
4112         if (DIFF_PAIR_UNMERGED(p)) {
4113                 /* unmerged */
4114                 return;
4115         }
4116
4117         name = p->one->path;
4118         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4119         attr_path = other ? other : name;
4120
4121         if (o->prefix_length)
4122                 strip_prefix(o->prefix_length, &name, &other);
4123
4124         diff_fill_oid_info(p->one);
4125         diff_fill_oid_info(p->two);
4126
4127         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4128 }
4129
4130 void diff_setup(struct diff_options *options)
4131 {
4132         memcpy(options, &default_diff_options, sizeof(*options));
4133
4134         options->file = stdout;
4135
4136         options->abbrev = DEFAULT_ABBREV;
4137         options->line_termination = '\n';
4138         options->break_opt = -1;
4139         options->rename_limit = -1;
4140         options->dirstat_permille = diff_dirstat_permille_default;
4141         options->context = diff_context_default;
4142         options->interhunkcontext = diff_interhunk_context_default;
4143         options->ws_error_highlight = ws_error_highlight_default;
4144         DIFF_OPT_SET(options, RENAME_EMPTY);
4145
4146         /* pathchange left =NULL by default */
4147         options->change = diff_change;
4148         options->add_remove = diff_addremove;
4149         options->use_color = diff_use_color_default;
4150         options->detect_rename = diff_detect_rename_default;
4151         options->xdl_opts |= diff_algorithm;
4152         if (diff_indent_heuristic)
4153                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4154
4155         options->orderfile = diff_order_file_cfg;
4156
4157         if (diff_no_prefix) {
4158                 options->a_prefix = options->b_prefix = "";
4159         } else if (!diff_mnemonic_prefix) {
4160                 options->a_prefix = "a/";
4161                 options->b_prefix = "b/";
4162         }
4163
4164         options->color_moved = diff_color_moved_default;
4165 }
4166
4167 void diff_setup_done(struct diff_options *options)
4168 {
4169         int count = 0;
4170
4171         if (options->set_default)
4172                 options->set_default(options);
4173
4174         if (options->output_format & DIFF_FORMAT_NAME)
4175                 count++;
4176         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
4177                 count++;
4178         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
4179                 count++;
4180         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
4181                 count++;
4182         if (count > 1)
4183                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4184
4185         /*
4186          * Most of the time we can say "there are changes"
4187          * only by checking if there are changed paths, but
4188          * --ignore-whitespace* options force us to look
4189          * inside contents.
4190          */
4191
4192         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
4193             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
4194             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
4195                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
4196         else
4197                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
4198
4199         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
4200                 options->detect_rename = DIFF_DETECT_COPY;
4201
4202         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
4203                 options->prefix = NULL;
4204         if (options->prefix)
4205                 options->prefix_length = strlen(options->prefix);
4206         else
4207                 options->prefix_length = 0;
4208
4209         if (options->output_format & (DIFF_FORMAT_NAME |
4210                                       DIFF_FORMAT_NAME_STATUS |
4211                                       DIFF_FORMAT_CHECKDIFF |
4212                                       DIFF_FORMAT_NO_OUTPUT))
4213                 options->output_format &= ~(DIFF_FORMAT_RAW |
4214                                             DIFF_FORMAT_NUMSTAT |
4215                                             DIFF_FORMAT_DIFFSTAT |
4216                                             DIFF_FORMAT_SHORTSTAT |
4217                                             DIFF_FORMAT_DIRSTAT |
4218                                             DIFF_FORMAT_SUMMARY |
4219                                             DIFF_FORMAT_PATCH);
4220
4221         /*
4222          * These cases always need recursive; we do not drop caller-supplied
4223          * recursive bits for other formats here.
4224          */
4225         if (options->output_format & (DIFF_FORMAT_PATCH |
4226                                       DIFF_FORMAT_NUMSTAT |
4227                                       DIFF_FORMAT_DIFFSTAT |
4228                                       DIFF_FORMAT_SHORTSTAT |
4229                                       DIFF_FORMAT_DIRSTAT |
4230                                       DIFF_FORMAT_SUMMARY |
4231                                       DIFF_FORMAT_CHECKDIFF))
4232                 DIFF_OPT_SET(options, RECURSIVE);
4233         /*
4234          * Also pickaxe would not work very well if you do not say recursive
4235          */
4236         if (options->pickaxe)
4237                 DIFF_OPT_SET(options, RECURSIVE);
4238         /*
4239          * When patches are generated, submodules diffed against the work tree
4240          * must be checked for dirtiness too so it can be shown in the output
4241          */
4242         if (options->output_format & DIFF_FORMAT_PATCH)
4243                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
4244
4245         if (options->detect_rename && options->rename_limit < 0)
4246                 options->rename_limit = diff_rename_limit_default;
4247         if (options->setup & DIFF_SETUP_USE_CACHE) {
4248                 if (!active_cache)
4249                         /* read-cache does not die even when it fails
4250                          * so it is safe for us to do this here.  Also
4251                          * it does not smudge active_cache or active_nr
4252                          * when it fails, so we do not have to worry about
4253                          * cleaning it up ourselves either.
4254                          */
4255                         read_cache();
4256         }
4257         if (40 < options->abbrev)
4258                 options->abbrev = 40; /* full */
4259
4260         /*
4261          * It does not make sense to show the first hit we happened
4262          * to have found.  It does not make sense not to return with
4263          * exit code in such a case either.
4264          */
4265         if (DIFF_OPT_TST(options, QUICK)) {
4266                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4267                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4268         }
4269
4270         options->diff_path_counter = 0;
4271
4272         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
4273                 die(_("--follow requires exactly one pathspec"));
4274
4275         if (!options->use_color || external_diff())
4276                 options->color_moved = 0;
4277 }
4278
4279 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4280 {
4281         char c, *eq;
4282         int len;
4283
4284         if (*arg != '-')
4285                 return 0;
4286         c = *++arg;
4287         if (!c)
4288                 return 0;
4289         if (c == arg_short) {
4290                 c = *++arg;
4291                 if (!c)
4292                         return 1;
4293                 if (val && isdigit(c)) {
4294                         char *end;
4295                         int n = strtoul(arg, &end, 10);
4296                         if (*end)
4297                                 return 0;
4298                         *val = n;
4299                         return 1;
4300                 }
4301                 return 0;
4302         }
4303         if (c != '-')
4304                 return 0;
4305         arg++;
4306         eq = strchrnul(arg, '=');
4307         len = eq - arg;
4308         if (!len || strncmp(arg, arg_long, len))
4309                 return 0;
4310         if (*eq) {
4311                 int n;
4312                 char *end;
4313                 if (!isdigit(*++eq))
4314                         return 0;
4315                 n = strtoul(eq, &end, 10);
4316                 if (*end)
4317                         return 0;
4318                 *val = n;
4319         }
4320         return 1;
4321 }
4322
4323 static int diff_scoreopt_parse(const char *opt);
4324
4325 static inline int short_opt(char opt, const char **argv,
4326                             const char **optarg)
4327 {
4328         const char *arg = argv[0];
4329         if (arg[0] != '-' || arg[1] != opt)
4330                 return 0;
4331         if (arg[2] != '\0') {
4332                 *optarg = arg + 2;
4333                 return 1;
4334         }
4335         if (!argv[1])
4336                 die("Option '%c' requires a value", opt);
4337         *optarg = argv[1];
4338         return 2;
4339 }
4340
4341 int parse_long_opt(const char *opt, const char **argv,
4342                    const char **optarg)
4343 {
4344         const char *arg = argv[0];
4345         if (!skip_prefix(arg, "--", &arg))
4346                 return 0;
4347         if (!skip_prefix(arg, opt, &arg))
4348                 return 0;
4349         if (*arg == '=') { /* stuck form: --option=value */
4350                 *optarg = arg + 1;
4351                 return 1;
4352         }
4353         if (*arg != '\0')
4354                 return 0;
4355         /* separate form: --option value */
4356         if (!argv[1])
4357                 die("Option '--%s' requires a value", opt);
4358         *optarg = argv[1];
4359         return 2;
4360 }
4361
4362 static int stat_opt(struct diff_options *options, const char **av)
4363 {
4364         const char *arg = av[0];
4365         char *end;
4366         int width = options->stat_width;
4367         int name_width = options->stat_name_width;
4368         int graph_width = options->stat_graph_width;
4369         int count = options->stat_count;
4370         int argcount = 1;
4371
4372         if (!skip_prefix(arg, "--stat", &arg))
4373                 die("BUG: stat option does not begin with --stat: %s", arg);
4374         end = (char *)arg;
4375
4376         switch (*arg) {
4377         case '-':
4378                 if (skip_prefix(arg, "-width", &arg)) {
4379                         if (*arg == '=')
4380                                 width = strtoul(arg + 1, &end, 10);
4381                         else if (!*arg && !av[1])
4382                                 die_want_option("--stat-width");
4383                         else if (!*arg) {
4384                                 width = strtoul(av[1], &end, 10);
4385                                 argcount = 2;
4386                         }
4387                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4388                         if (*arg == '=')
4389                                 name_width = strtoul(arg + 1, &end, 10);
4390                         else if (!*arg && !av[1])
4391                                 die_want_option("--stat-name-width");
4392                         else if (!*arg) {
4393                                 name_width = strtoul(av[1], &end, 10);
4394                                 argcount = 2;
4395                         }
4396                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4397                         if (*arg == '=')
4398                                 graph_width = strtoul(arg + 1, &end, 10);
4399                         else if (!*arg && !av[1])
4400                                 die_want_option("--stat-graph-width");
4401                         else if (!*arg) {
4402                                 graph_width = strtoul(av[1], &end, 10);
4403                                 argcount = 2;
4404                         }
4405                 } else if (skip_prefix(arg, "-count", &arg)) {
4406                         if (*arg == '=')
4407                                 count = strtoul(arg + 1, &end, 10);
4408                         else if (!*arg && !av[1])
4409                                 die_want_option("--stat-count");
4410                         else if (!*arg) {
4411                                 count = strtoul(av[1], &end, 10);
4412                                 argcount = 2;
4413                         }
4414                 }
4415                 break;
4416         case '=':
4417                 width = strtoul(arg+1, &end, 10);
4418                 if (*end == ',')
4419                         name_width = strtoul(end+1, &end, 10);
4420                 if (*end == ',')
4421                         count = strtoul(end+1, &end, 10);
4422         }
4423
4424         /* Important! This checks all the error cases! */
4425         if (*end)
4426                 return 0;
4427         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4428         options->stat_name_width = name_width;
4429         options->stat_graph_width = graph_width;
4430         options->stat_width = width;
4431         options->stat_count = count;
4432         return argcount;
4433 }
4434
4435 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4436 {
4437         struct strbuf errmsg = STRBUF_INIT;
4438         if (parse_dirstat_params(options, params, &errmsg))
4439                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4440                     errmsg.buf);
4441         strbuf_release(&errmsg);
4442         /*
4443          * The caller knows a dirstat-related option is given from the command
4444          * line; allow it to say "return this_function();"
4445          */
4446         options->output_format |= DIFF_FORMAT_DIRSTAT;
4447         return 1;
4448 }
4449
4450 static int parse_submodule_opt(struct diff_options *options, const char *value)
4451 {
4452         if (parse_submodule_params(options, value))
4453                 die(_("Failed to parse --submodule option parameter: '%s'"),
4454                         value);
4455         return 1;
4456 }
4457
4458 static const char diff_status_letters[] = {
4459         DIFF_STATUS_ADDED,
4460         DIFF_STATUS_COPIED,
4461         DIFF_STATUS_DELETED,
4462         DIFF_STATUS_MODIFIED,
4463         DIFF_STATUS_RENAMED,
4464         DIFF_STATUS_TYPE_CHANGED,
4465         DIFF_STATUS_UNKNOWN,
4466         DIFF_STATUS_UNMERGED,
4467         DIFF_STATUS_FILTER_AON,
4468         DIFF_STATUS_FILTER_BROKEN,
4469         '\0',
4470 };
4471
4472 static unsigned int filter_bit['Z' + 1];
4473
4474 static void prepare_filter_bits(void)
4475 {
4476         int i;
4477
4478         if (!filter_bit[DIFF_STATUS_ADDED]) {
4479                 for (i = 0; diff_status_letters[i]; i++)
4480                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4481         }
4482 }
4483
4484 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4485 {
4486         return opt->filter & filter_bit[(int) status];
4487 }
4488
4489 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4490 {
4491         int i, optch;
4492
4493         prepare_filter_bits();
4494
4495         /*
4496          * If there is a negation e.g. 'd' in the input, and we haven't
4497          * initialized the filter field with another --diff-filter, start
4498          * from full set of bits, except for AON.
4499          */
4500         if (!opt->filter) {
4501                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4502                         if (optch < 'a' || 'z' < optch)
4503                                 continue;
4504                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4505                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4506                         break;
4507                 }
4508         }
4509
4510         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4511                 unsigned int bit;
4512                 int negate;
4513
4514                 if ('a' <= optch && optch <= 'z') {
4515                         negate = 1;
4516                         optch = toupper(optch);
4517                 } else {
4518                         negate = 0;
4519                 }
4520
4521                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4522                 if (!bit)
4523                         return optarg[i];
4524                 if (negate)
4525                         opt->filter &= ~bit;
4526                 else
4527                         opt->filter |= bit;
4528         }
4529         return 0;
4530 }
4531
4532 static void enable_patch_output(int *fmt) {
4533         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4534         *fmt |= DIFF_FORMAT_PATCH;
4535 }
4536
4537 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4538 {
4539         int val = parse_ws_error_highlight(arg);
4540
4541         if (val < 0) {
4542                 error("unknown value after ws-error-highlight=%.*s",
4543                       -1 - val, arg);
4544                 return 0;
4545         }
4546         opt->ws_error_highlight = val;
4547         return 1;
4548 }
4549
4550 int diff_opt_parse(struct diff_options *options,
4551                    const char **av, int ac, const char *prefix)
4552 {
4553         const char *arg = av[0];
4554         const char *optarg;
4555         int argcount;
4556
4557         if (!prefix)
4558                 prefix = "";
4559
4560         /* Output format options */
4561         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4562             || opt_arg(arg, 'U', "unified", &options->context))
4563                 enable_patch_output(&options->output_format);
4564         else if (!strcmp(arg, "--raw"))
4565                 options->output_format |= DIFF_FORMAT_RAW;
4566         else if (!strcmp(arg, "--patch-with-raw")) {
4567                 enable_patch_output(&options->output_format);
4568                 options->output_format |= DIFF_FORMAT_RAW;
4569         } else if (!strcmp(arg, "--numstat"))
4570                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4571         else if (!strcmp(arg, "--shortstat"))
4572                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4573         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
4574                 return parse_dirstat_opt(options, "");
4575         else if (skip_prefix(arg, "-X", &arg))
4576                 return parse_dirstat_opt(options, arg);
4577         else if (skip_prefix(arg, "--dirstat=", &arg))
4578                 return parse_dirstat_opt(options, arg);
4579         else if (!strcmp(arg, "--cumulative"))
4580                 return parse_dirstat_opt(options, "cumulative");
4581         else if (!strcmp(arg, "--dirstat-by-file"))
4582                 return parse_dirstat_opt(options, "files");
4583         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
4584                 parse_dirstat_opt(options, "files");
4585                 return parse_dirstat_opt(options, arg);
4586         }
4587         else if (!strcmp(arg, "--check"))
4588                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4589         else if (!strcmp(arg, "--summary"))
4590                 options->output_format |= DIFF_FORMAT_SUMMARY;
4591         else if (!strcmp(arg, "--patch-with-stat")) {
4592                 enable_patch_output(&options->output_format);
4593                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4594         } else if (!strcmp(arg, "--name-only"))
4595                 options->output_format |= DIFF_FORMAT_NAME;
4596         else if (!strcmp(arg, "--name-status"))
4597                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4598         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4599                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4600         else if (starts_with(arg, "--stat"))
4601                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4602                 return stat_opt(options, av);
4603
4604         /* renames options */
4605         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
4606                  !strcmp(arg, "--break-rewrites")) {
4607                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4608                         return error("invalid argument to -B: %s", arg+2);
4609         }
4610         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
4611                  !strcmp(arg, "--find-renames")) {
4612                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4613                         return error("invalid argument to -M: %s", arg+2);
4614                 options->detect_rename = DIFF_DETECT_RENAME;
4615         }
4616         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4617                 options->irreversible_delete = 1;
4618         }
4619         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
4620                  !strcmp(arg, "--find-copies")) {
4621                 if (options->detect_rename == DIFF_DETECT_COPY)
4622                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4623                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4624                         return error("invalid argument to -C: %s", arg+2);
4625                 options->detect_rename = DIFF_DETECT_COPY;
4626         }
4627         else if (!strcmp(arg, "--no-renames"))
4628                 options->detect_rename = 0;
4629         else if (!strcmp(arg, "--rename-empty"))
4630                 DIFF_OPT_SET(options, RENAME_EMPTY);
4631         else if (!strcmp(arg, "--no-rename-empty"))
4632                 DIFF_OPT_CLR(options, RENAME_EMPTY);
4633         else if (!strcmp(arg, "--relative"))
4634                 DIFF_OPT_SET(options, RELATIVE_NAME);
4635         else if (skip_prefix(arg, "--relative=", &arg)) {
4636                 DIFF_OPT_SET(options, RELATIVE_NAME);
4637                 options->prefix = arg;
4638         }
4639
4640         /* xdiff options */
4641         else if (!strcmp(arg, "--minimal"))
4642                 DIFF_XDL_SET(options, NEED_MINIMAL);
4643         else if (!strcmp(arg, "--no-minimal"))
4644                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4645         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4646                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4647         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4648                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4649         else if (!strcmp(arg, "--ignore-space-at-eol"))
4650                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4651         else if (!strcmp(arg, "--ignore-blank-lines"))
4652                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4653         else if (!strcmp(arg, "--indent-heuristic"))
4654                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4655         else if (!strcmp(arg, "--no-indent-heuristic"))
4656                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4657         else if (!strcmp(arg, "--patience"))
4658                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4659         else if (!strcmp(arg, "--histogram"))
4660                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4661         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4662                 long value = parse_algorithm_value(optarg);
4663                 if (value < 0)
4664                         return error("option diff-algorithm accepts \"myers\", "
4665                                      "\"minimal\", \"patience\" and \"histogram\"");
4666                 /* clear out previous settings */
4667                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4668                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4669                 options->xdl_opts |= value;
4670                 return argcount;
4671         }
4672
4673         /* flags options */
4674         else if (!strcmp(arg, "--binary")) {
4675                 enable_patch_output(&options->output_format);
4676                 DIFF_OPT_SET(options, BINARY);
4677         }
4678         else if (!strcmp(arg, "--full-index"))
4679                 DIFF_OPT_SET(options, FULL_INDEX);
4680         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4681                 DIFF_OPT_SET(options, TEXT);
4682         else if (!strcmp(arg, "-R"))
4683                 DIFF_OPT_SET(options, REVERSE_DIFF);
4684         else if (!strcmp(arg, "--find-copies-harder"))
4685                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4686         else if (!strcmp(arg, "--follow"))
4687                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
4688         else if (!strcmp(arg, "--no-follow")) {
4689                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
4690                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
4691         } else if (!strcmp(arg, "--color"))
4692                 options->use_color = 1;
4693         else if (skip_prefix(arg, "--color=", &arg)) {
4694                 int value = git_config_colorbool(NULL, arg);
4695                 if (value < 0)
4696                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4697                 options->use_color = value;
4698         }
4699         else if (!strcmp(arg, "--no-color"))
4700                 options->use_color = 0;
4701         else if (!strcmp(arg, "--color-moved")) {
4702                 if (diff_color_moved_default)
4703                         options->color_moved = diff_color_moved_default;
4704                 if (options->color_moved == COLOR_MOVED_NO)
4705                         options->color_moved = COLOR_MOVED_DEFAULT;
4706         } else if (!strcmp(arg, "--no-color-moved"))
4707                 options->color_moved = COLOR_MOVED_NO;
4708         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4709                 int cm = parse_color_moved(arg);
4710                 if (cm < 0)
4711                         die("bad --color-moved argument: %s", arg);
4712                 options->color_moved = cm;
4713         } else if (!strcmp(arg, "--color-words")) {
4714                 options->use_color = 1;
4715                 options->word_diff = DIFF_WORDS_COLOR;
4716         }
4717         else if (skip_prefix(arg, "--color-words=", &arg)) {
4718                 options->use_color = 1;
4719                 options->word_diff = DIFF_WORDS_COLOR;
4720                 options->word_regex = arg;
4721         }
4722         else if (!strcmp(arg, "--word-diff")) {
4723                 if (options->word_diff == DIFF_WORDS_NONE)
4724                         options->word_diff = DIFF_WORDS_PLAIN;
4725         }
4726         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4727                 if (!strcmp(arg, "plain"))
4728                         options->word_diff = DIFF_WORDS_PLAIN;
4729                 else if (!strcmp(arg, "color")) {
4730                         options->use_color = 1;
4731                         options->word_diff = DIFF_WORDS_COLOR;
4732                 }
4733                 else if (!strcmp(arg, "porcelain"))
4734                         options->word_diff = DIFF_WORDS_PORCELAIN;
4735                 else if (!strcmp(arg, "none"))
4736                         options->word_diff = DIFF_WORDS_NONE;
4737                 else
4738                         die("bad --word-diff argument: %s", arg);
4739         }
4740         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4741                 if (options->word_diff == DIFF_WORDS_NONE)
4742                         options->word_diff = DIFF_WORDS_PLAIN;
4743                 options->word_regex = optarg;
4744                 return argcount;
4745         }
4746         else if (!strcmp(arg, "--exit-code"))
4747                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4748         else if (!strcmp(arg, "--quiet"))
4749                 DIFF_OPT_SET(options, QUICK);
4750         else if (!strcmp(arg, "--ext-diff"))
4751                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
4752         else if (!strcmp(arg, "--no-ext-diff"))
4753                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
4754         else if (!strcmp(arg, "--textconv"))
4755                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
4756         else if (!strcmp(arg, "--no-textconv"))
4757                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
4758         else if (!strcmp(arg, "--ignore-submodules")) {
4759                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4760                 handle_ignore_submodules_arg(options, "all");
4761         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
4762                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4763                 handle_ignore_submodules_arg(options, arg);
4764         } else if (!strcmp(arg, "--submodule"))
4765                 options->submodule_format = DIFF_SUBMODULE_LOG;
4766         else if (skip_prefix(arg, "--submodule=", &arg))
4767                 return parse_submodule_opt(options, arg);
4768         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4769                 return parse_ws_error_highlight_opt(options, arg);
4770         else if (!strcmp(arg, "--ita-invisible-in-index"))
4771                 options->ita_invisible_in_index = 1;
4772         else if (!strcmp(arg, "--ita-visible-in-index"))
4773                 options->ita_invisible_in_index = 0;
4774
4775         /* misc options */
4776         else if (!strcmp(arg, "-z"))
4777                 options->line_termination = 0;
4778         else if ((argcount = short_opt('l', av, &optarg))) {
4779                 options->rename_limit = strtoul(optarg, NULL, 10);
4780                 return argcount;
4781         }
4782         else if ((argcount = short_opt('S', av, &optarg))) {
4783                 options->pickaxe = optarg;
4784                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4785                 return argcount;
4786         } else if ((argcount = short_opt('G', av, &optarg))) {
4787                 options->pickaxe = optarg;
4788                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4789                 return argcount;
4790         }
4791         else if (!strcmp(arg, "--pickaxe-all"))
4792                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4793         else if (!strcmp(arg, "--pickaxe-regex"))
4794                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4795         else if ((argcount = short_opt('O', av, &optarg))) {
4796                 options->orderfile = prefix_filename(prefix, optarg);
4797                 return argcount;
4798         }
4799         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4800                 int offending = parse_diff_filter_opt(optarg, options);
4801                 if (offending)
4802                         die("unknown change class '%c' in --diff-filter=%s",
4803                             offending, optarg);
4804                 return argcount;
4805         }
4806         else if (!strcmp(arg, "--no-abbrev"))
4807                 options->abbrev = 0;
4808         else if (!strcmp(arg, "--abbrev"))
4809                 options->abbrev = DEFAULT_ABBREV;
4810         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4811                 options->abbrev = strtoul(arg, NULL, 10);
4812                 if (options->abbrev < MINIMUM_ABBREV)
4813                         options->abbrev = MINIMUM_ABBREV;
4814                 else if (40 < options->abbrev)
4815                         options->abbrev = 40;
4816         }
4817         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4818                 options->a_prefix = optarg;
4819                 return argcount;
4820         }
4821         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4822                 options->line_prefix = optarg;
4823                 options->line_prefix_length = strlen(options->line_prefix);
4824                 graph_setup_line_prefix(options);
4825                 return argcount;
4826         }
4827         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4828                 options->b_prefix = optarg;
4829                 return argcount;
4830         }
4831         else if (!strcmp(arg, "--no-prefix"))
4832                 options->a_prefix = options->b_prefix = "";
4833         else if (opt_arg(arg, '\0', "inter-hunk-context",
4834                          &options->interhunkcontext))
4835                 ;
4836         else if (!strcmp(arg, "-W"))
4837                 DIFF_OPT_SET(options, FUNCCONTEXT);
4838         else if (!strcmp(arg, "--function-context"))
4839                 DIFF_OPT_SET(options, FUNCCONTEXT);
4840         else if (!strcmp(arg, "--no-function-context"))
4841                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4842         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4843                 char *path = prefix_filename(prefix, optarg);
4844                 options->file = xfopen(path, "w");
4845                 options->close_file = 1;
4846                 if (options->use_color != GIT_COLOR_ALWAYS)
4847                         options->use_color = GIT_COLOR_NEVER;
4848                 free(path);
4849                 return argcount;
4850         } else
4851                 return 0;
4852         return 1;
4853 }
4854
4855 int parse_rename_score(const char **cp_p)
4856 {
4857         unsigned long num, scale;
4858         int ch, dot;
4859         const char *cp = *cp_p;
4860
4861         num = 0;
4862         scale = 1;
4863         dot = 0;
4864         for (;;) {
4865                 ch = *cp;
4866                 if ( !dot && ch == '.' ) {
4867                         scale = 1;
4868                         dot = 1;
4869                 } else if ( ch == '%' ) {
4870                         scale = dot ? scale*100 : 100;
4871                         cp++;   /* % is always at the end */
4872                         break;
4873                 } else if ( ch >= '0' && ch <= '9' ) {
4874                         if ( scale < 100000 ) {
4875                                 scale *= 10;
4876                                 num = (num*10) + (ch-'0');
4877                         }
4878                 } else {
4879                         break;
4880                 }
4881                 cp++;
4882         }
4883         *cp_p = cp;
4884
4885         /* user says num divided by scale and we say internally that
4886          * is MAX_SCORE * num / scale.
4887          */
4888         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4889 }
4890
4891 static int diff_scoreopt_parse(const char *opt)
4892 {
4893         int opt1, opt2, cmd;
4894
4895         if (*opt++ != '-')
4896                 return -1;
4897         cmd = *opt++;
4898         if (cmd == '-') {
4899                 /* convert the long-form arguments into short-form versions */
4900                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4901                         if (*opt == 0 || *opt++ == '=')
4902                                 cmd = 'B';
4903                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4904                         if (*opt == 0 || *opt++ == '=')
4905                                 cmd = 'C';
4906                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4907                         if (*opt == 0 || *opt++ == '=')
4908                                 cmd = 'M';
4909                 }
4910         }
4911         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4912                 return -1; /* that is not a -M, -C, or -B option */
4913
4914         opt1 = parse_rename_score(&opt);
4915         if (cmd != 'B')
4916                 opt2 = 0;
4917         else {
4918                 if (*opt == 0)
4919                         opt2 = 0;
4920                 else if (*opt != '/')
4921                         return -1; /* we expect -B80/99 or -B80 */
4922                 else {
4923                         opt++;
4924                         opt2 = parse_rename_score(&opt);
4925                 }
4926         }
4927         if (*opt != 0)
4928                 return -1;
4929         return opt1 | (opt2 << 16);
4930 }
4931
4932 struct diff_queue_struct diff_queued_diff;
4933
4934 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4935 {
4936         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4937         queue->queue[queue->nr++] = dp;
4938 }
4939
4940 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4941                                  struct diff_filespec *one,
4942                                  struct diff_filespec *two)
4943 {
4944         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4945         dp->one = one;
4946         dp->two = two;
4947         if (queue)
4948                 diff_q(queue, dp);
4949         return dp;
4950 }
4951
4952 void diff_free_filepair(struct diff_filepair *p)
4953 {
4954         free_filespec(p->one);
4955         free_filespec(p->two);
4956         free(p);
4957 }
4958
4959 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4960 {
4961         int abblen;
4962         const char *abbrev;
4963
4964         if (len == GIT_SHA1_HEXSZ)
4965                 return oid_to_hex(oid);
4966
4967         abbrev = diff_abbrev_oid(oid, len);
4968         abblen = strlen(abbrev);
4969
4970         /*
4971          * In well-behaved cases, where the abbbreviated result is the
4972          * same as the requested length, append three dots after the
4973          * abbreviation (hence the whole logic is limited to the case
4974          * where abblen < 37); when the actual abbreviated result is a
4975          * bit longer than the requested length, we reduce the number
4976          * of dots so that they match the well-behaved ones.  However,
4977          * if the actual abbreviation is longer than the requested
4978          * length by more than three, we give up on aligning, and add
4979          * three dots anyway, to indicate that the output is not the
4980          * full object name.  Yes, this may be suboptimal, but this
4981          * appears only in "diff --raw --abbrev" output and it is not
4982          * worth the effort to change it now.  Note that this would
4983          * likely to work fine when the automatic sizing of default
4984          * abbreviation length is used--we would be fed -1 in "len" in
4985          * that case, and will end up always appending three-dots, but
4986          * the automatic sizing is supposed to give abblen that ensures
4987          * uniqueness across all objects (statistically speaking).
4988          */
4989         if (abblen < GIT_SHA1_HEXSZ - 3) {
4990                 static char hex[GIT_MAX_HEXSZ + 1];
4991                 if (len < abblen && abblen <= len + 2)
4992                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4993                 else
4994                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4995                 return hex;
4996         }
4997
4998         return oid_to_hex(oid);
4999 }
5000
5001 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5002 {
5003         int line_termination = opt->line_termination;
5004         int inter_name_termination = line_termination ? '\t' : '\0';
5005
5006         fprintf(opt->file, "%s", diff_line_prefix(opt));
5007         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5008                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5009                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5010                 fprintf(opt->file, "%s ",
5011                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5012         }
5013         if (p->score) {
5014                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5015                         inter_name_termination);
5016         } else {
5017                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5018         }
5019
5020         if (p->status == DIFF_STATUS_COPIED ||
5021             p->status == DIFF_STATUS_RENAMED) {
5022                 const char *name_a, *name_b;
5023                 name_a = p->one->path;
5024                 name_b = p->two->path;
5025                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5026                 write_name_quoted(name_a, opt->file, inter_name_termination);
5027                 write_name_quoted(name_b, opt->file, line_termination);
5028         } else {
5029                 const char *name_a, *name_b;
5030                 name_a = p->one->mode ? p->one->path : p->two->path;
5031                 name_b = NULL;
5032                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5033                 write_name_quoted(name_a, opt->file, line_termination);
5034         }
5035 }
5036
5037 int diff_unmodified_pair(struct diff_filepair *p)
5038 {
5039         /* This function is written stricter than necessary to support
5040          * the currently implemented transformers, but the idea is to
5041          * let transformers to produce diff_filepairs any way they want,
5042          * and filter and clean them up here before producing the output.
5043          */
5044         struct diff_filespec *one = p->one, *two = p->two;
5045
5046         if (DIFF_PAIR_UNMERGED(p))
5047                 return 0; /* unmerged is interesting */
5048
5049         /* deletion, addition, mode or type change
5050          * and rename are all interesting.
5051          */
5052         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5053             DIFF_PAIR_MODE_CHANGED(p) ||
5054             strcmp(one->path, two->path))
5055                 return 0;
5056
5057         /* both are valid and point at the same path.  that is, we are
5058          * dealing with a change.
5059          */
5060         if (one->oid_valid && two->oid_valid &&
5061             !oidcmp(&one->oid, &two->oid) &&
5062             !one->dirty_submodule && !two->dirty_submodule)
5063                 return 1; /* no change */
5064         if (!one->oid_valid && !two->oid_valid)
5065                 return 1; /* both look at the same file on the filesystem. */
5066         return 0;
5067 }
5068
5069 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5070 {
5071         if (diff_unmodified_pair(p))
5072                 return;
5073
5074         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5075             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5076                 return; /* no tree diffs in patch format */
5077
5078         run_diff(p, o);
5079 }
5080
5081 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5082                             struct diffstat_t *diffstat)
5083 {
5084         if (diff_unmodified_pair(p))
5085                 return;
5086
5087         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5088             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5089                 return; /* no useful stat for tree diffs */
5090
5091         run_diffstat(p, o, diffstat);
5092 }
5093
5094 static void diff_flush_checkdiff(struct diff_filepair *p,
5095                 struct diff_options *o)
5096 {
5097         if (diff_unmodified_pair(p))
5098                 return;
5099
5100         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5101             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5102                 return; /* nothing to check in tree diffs */
5103
5104         run_checkdiff(p, o);
5105 }
5106
5107 int diff_queue_is_empty(void)
5108 {
5109         struct diff_queue_struct *q = &diff_queued_diff;
5110         int i;
5111         for (i = 0; i < q->nr; i++)
5112                 if (!diff_unmodified_pair(q->queue[i]))
5113                         return 0;
5114         return 1;
5115 }
5116
5117 #if DIFF_DEBUG
5118 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5119 {
5120         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5121                 x, one ? one : "",
5122                 s->path,
5123                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5124                 s->mode,
5125                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5126         fprintf(stderr, "queue[%d] %s size %lu\n",
5127                 x, one ? one : "",
5128                 s->size);
5129 }
5130
5131 void diff_debug_filepair(const struct diff_filepair *p, int i)
5132 {
5133         diff_debug_filespec(p->one, i, "one");
5134         diff_debug_filespec(p->two, i, "two");
5135         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5136                 p->score, p->status ? p->status : '?',
5137                 p->one->rename_used, p->broken_pair);
5138 }
5139
5140 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5141 {
5142         int i;
5143         if (msg)
5144                 fprintf(stderr, "%s\n", msg);
5145         fprintf(stderr, "q->nr = %d\n", q->nr);
5146         for (i = 0; i < q->nr; i++) {
5147                 struct diff_filepair *p = q->queue[i];
5148                 diff_debug_filepair(p, i);
5149         }
5150 }
5151 #endif
5152
5153 static void diff_resolve_rename_copy(void)
5154 {
5155         int i;
5156         struct diff_filepair *p;
5157         struct diff_queue_struct *q = &diff_queued_diff;
5158
5159         diff_debug_queue("resolve-rename-copy", q);
5160
5161         for (i = 0; i < q->nr; i++) {
5162                 p = q->queue[i];
5163                 p->status = 0; /* undecided */
5164                 if (DIFF_PAIR_UNMERGED(p))
5165                         p->status = DIFF_STATUS_UNMERGED;
5166                 else if (!DIFF_FILE_VALID(p->one))
5167                         p->status = DIFF_STATUS_ADDED;
5168                 else if (!DIFF_FILE_VALID(p->two))
5169                         p->status = DIFF_STATUS_DELETED;
5170                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5171                         p->status = DIFF_STATUS_TYPE_CHANGED;
5172
5173                 /* from this point on, we are dealing with a pair
5174                  * whose both sides are valid and of the same type, i.e.
5175                  * either in-place edit or rename/copy edit.
5176                  */
5177                 else if (DIFF_PAIR_RENAME(p)) {
5178                         /*
5179                          * A rename might have re-connected a broken
5180                          * pair up, causing the pathnames to be the
5181                          * same again. If so, that's not a rename at
5182                          * all, just a modification..
5183                          *
5184                          * Otherwise, see if this source was used for
5185                          * multiple renames, in which case we decrement
5186                          * the count, and call it a copy.
5187                          */
5188                         if (!strcmp(p->one->path, p->two->path))
5189                                 p->status = DIFF_STATUS_MODIFIED;
5190                         else if (--p->one->rename_used > 0)
5191                                 p->status = DIFF_STATUS_COPIED;
5192                         else
5193                                 p->status = DIFF_STATUS_RENAMED;
5194                 }
5195                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5196                          p->one->mode != p->two->mode ||
5197                          p->one->dirty_submodule ||
5198                          p->two->dirty_submodule ||
5199                          is_null_oid(&p->one->oid))
5200                         p->status = DIFF_STATUS_MODIFIED;
5201                 else {
5202                         /* This is a "no-change" entry and should not
5203                          * happen anymore, but prepare for broken callers.
5204                          */
5205                         error("feeding unmodified %s to diffcore",
5206                               p->one->path);
5207                         p->status = DIFF_STATUS_UNKNOWN;
5208                 }
5209         }
5210         diff_debug_queue("resolve-rename-copy done", q);
5211 }
5212
5213 static int check_pair_status(struct diff_filepair *p)
5214 {
5215         switch (p->status) {
5216         case DIFF_STATUS_UNKNOWN:
5217                 return 0;
5218         case 0:
5219                 die("internal error in diff-resolve-rename-copy");
5220         default:
5221                 return 1;
5222         }
5223 }
5224
5225 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5226 {
5227         int fmt = opt->output_format;
5228
5229         if (fmt & DIFF_FORMAT_CHECKDIFF)
5230                 diff_flush_checkdiff(p, opt);
5231         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5232                 diff_flush_raw(p, opt);
5233         else if (fmt & DIFF_FORMAT_NAME) {
5234                 const char *name_a, *name_b;
5235                 name_a = p->two->path;
5236                 name_b = NULL;
5237                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5238                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5239                 write_name_quoted(name_a, opt->file, opt->line_termination);
5240         }
5241 }
5242
5243 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5244 {
5245         struct strbuf sb = STRBUF_INIT;
5246         if (fs->mode)
5247                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5248         else
5249                 strbuf_addf(&sb, " %s ", newdelete);
5250
5251         quote_c_style(fs->path, &sb, NULL, 0);
5252         strbuf_addch(&sb, '\n');
5253         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5254                          sb.buf, sb.len, 0);
5255         strbuf_release(&sb);
5256 }
5257
5258 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5259                 int show_name)
5260 {
5261         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5262                 struct strbuf sb = STRBUF_INIT;
5263                 strbuf_addf(&sb, " mode change %06o => %06o",
5264                             p->one->mode, p->two->mode);
5265                 if (show_name) {
5266                         strbuf_addch(&sb, ' ');
5267                         quote_c_style(p->two->path, &sb, NULL, 0);
5268                 }
5269                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5270                                  sb.buf, sb.len, 0);
5271                 strbuf_release(&sb);
5272         }
5273 }
5274
5275 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5276                 struct diff_filepair *p)
5277 {
5278         struct strbuf sb = STRBUF_INIT;
5279         char *names = pprint_rename(p->one->path, p->two->path);
5280         strbuf_addf(&sb, " %s %s (%d%%)\n",
5281                         renamecopy, names, similarity_index(p));
5282         free(names);
5283         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5284                                  sb.buf, sb.len, 0);
5285         show_mode_change(opt, p, 0);
5286 }
5287
5288 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5289 {
5290         switch(p->status) {
5291         case DIFF_STATUS_DELETED:
5292                 show_file_mode_name(opt, "delete", p->one);
5293                 break;
5294         case DIFF_STATUS_ADDED:
5295                 show_file_mode_name(opt, "create", p->two);
5296                 break;
5297         case DIFF_STATUS_COPIED:
5298                 show_rename_copy(opt, "copy", p);
5299                 break;
5300         case DIFF_STATUS_RENAMED:
5301                 show_rename_copy(opt, "rename", p);
5302                 break;
5303         default:
5304                 if (p->score) {
5305                         struct strbuf sb = STRBUF_INIT;
5306                         strbuf_addstr(&sb, " rewrite ");
5307                         quote_c_style(p->two->path, &sb, NULL, 0);
5308                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5309                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5310                                          sb.buf, sb.len, 0);
5311                 }
5312                 show_mode_change(opt, p, !p->score);
5313                 break;
5314         }
5315 }
5316
5317 struct patch_id_t {
5318         git_SHA_CTX *ctx;
5319         int patchlen;
5320 };
5321
5322 static int remove_space(char *line, int len)
5323 {
5324         int i;
5325         char *dst = line;
5326         unsigned char c;
5327
5328         for (i = 0; i < len; i++)
5329                 if (!isspace((c = line[i])))
5330                         *dst++ = c;
5331
5332         return dst - line;
5333 }
5334
5335 static void patch_id_consume(void *priv, char *line, unsigned long len)
5336 {
5337         struct patch_id_t *data = priv;
5338         int new_len;
5339
5340         /* Ignore line numbers when computing the SHA1 of the patch */
5341         if (starts_with(line, "@@ -"))
5342                 return;
5343
5344         new_len = remove_space(line, len);
5345
5346         git_SHA1_Update(data->ctx, line, new_len);
5347         data->patchlen += new_len;
5348 }
5349
5350 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5351 {
5352         git_SHA1_Update(ctx, str, strlen(str));
5353 }
5354
5355 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5356 {
5357         /* large enough for 2^32 in octal */
5358         char buf[12];
5359         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5360         git_SHA1_Update(ctx, buf, len);
5361 }
5362
5363 /* returns 0 upon success, and writes result into sha1 */
5364 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5365 {
5366         struct diff_queue_struct *q = &diff_queued_diff;
5367         int i;
5368         git_SHA_CTX ctx;
5369         struct patch_id_t data;
5370
5371         git_SHA1_Init(&ctx);
5372         memset(&data, 0, sizeof(struct patch_id_t));
5373         data.ctx = &ctx;
5374
5375         for (i = 0; i < q->nr; i++) {
5376                 xpparam_t xpp;
5377                 xdemitconf_t xecfg;
5378                 mmfile_t mf1, mf2;
5379                 struct diff_filepair *p = q->queue[i];
5380                 int len1, len2;
5381
5382                 memset(&xpp, 0, sizeof(xpp));
5383                 memset(&xecfg, 0, sizeof(xecfg));
5384                 if (p->status == 0)
5385                         return error("internal diff status error");
5386                 if (p->status == DIFF_STATUS_UNKNOWN)
5387                         continue;
5388                 if (diff_unmodified_pair(p))
5389                         continue;
5390                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5391                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5392                         continue;
5393                 if (DIFF_PAIR_UNMERGED(p))
5394                         continue;
5395
5396                 diff_fill_oid_info(p->one);
5397                 diff_fill_oid_info(p->two);
5398
5399                 len1 = remove_space(p->one->path, strlen(p->one->path));
5400                 len2 = remove_space(p->two->path, strlen(p->two->path));
5401                 patch_id_add_string(&ctx, "diff--git");
5402                 patch_id_add_string(&ctx, "a/");
5403                 git_SHA1_Update(&ctx, p->one->path, len1);
5404                 patch_id_add_string(&ctx, "b/");
5405                 git_SHA1_Update(&ctx, p->two->path, len2);
5406
5407                 if (p->one->mode == 0) {
5408                         patch_id_add_string(&ctx, "newfilemode");
5409                         patch_id_add_mode(&ctx, p->two->mode);
5410                         patch_id_add_string(&ctx, "---/dev/null");
5411                         patch_id_add_string(&ctx, "+++b/");
5412                         git_SHA1_Update(&ctx, p->two->path, len2);
5413                 } else if (p->two->mode == 0) {
5414                         patch_id_add_string(&ctx, "deletedfilemode");
5415                         patch_id_add_mode(&ctx, p->one->mode);
5416                         patch_id_add_string(&ctx, "---a/");
5417                         git_SHA1_Update(&ctx, p->one->path, len1);
5418                         patch_id_add_string(&ctx, "+++/dev/null");
5419                 } else {
5420                         patch_id_add_string(&ctx, "---a/");
5421                         git_SHA1_Update(&ctx, p->one->path, len1);
5422                         patch_id_add_string(&ctx, "+++b/");
5423                         git_SHA1_Update(&ctx, p->two->path, len2);
5424                 }
5425
5426                 if (diff_header_only)
5427                         continue;
5428
5429                 if (fill_mmfile(&mf1, p->one) < 0 ||
5430                     fill_mmfile(&mf2, p->two) < 0)
5431                         return error("unable to read files to diff");
5432
5433                 if (diff_filespec_is_binary(p->one) ||
5434                     diff_filespec_is_binary(p->two)) {
5435                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5436                                         GIT_SHA1_HEXSZ);
5437                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5438                                         GIT_SHA1_HEXSZ);
5439                         continue;
5440                 }
5441
5442                 xpp.flags = 0;
5443                 xecfg.ctxlen = 3;
5444                 xecfg.flags = 0;
5445                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5446                                   &xpp, &xecfg))
5447                         return error("unable to generate patch-id diff for %s",
5448                                      p->one->path);
5449         }
5450
5451         git_SHA1_Final(oid->hash, &ctx);
5452         return 0;
5453 }
5454
5455 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5456 {
5457         struct diff_queue_struct *q = &diff_queued_diff;
5458         int i;
5459         int result = diff_get_patch_id(options, oid, diff_header_only);
5460
5461         for (i = 0; i < q->nr; i++)
5462                 diff_free_filepair(q->queue[i]);
5463
5464         free(q->queue);
5465         DIFF_QUEUE_CLEAR(q);
5466
5467         return result;
5468 }
5469
5470 static int is_summary_empty(const struct diff_queue_struct *q)
5471 {
5472         int i;
5473
5474         for (i = 0; i < q->nr; i++) {
5475                 const struct diff_filepair *p = q->queue[i];
5476
5477                 switch (p->status) {
5478                 case DIFF_STATUS_DELETED:
5479                 case DIFF_STATUS_ADDED:
5480                 case DIFF_STATUS_COPIED:
5481                 case DIFF_STATUS_RENAMED:
5482                         return 0;
5483                 default:
5484                         if (p->score)
5485                                 return 0;
5486                         if (p->one->mode && p->two->mode &&
5487                             p->one->mode != p->two->mode)
5488                                 return 0;
5489                         break;
5490                 }
5491         }
5492         return 1;
5493 }
5494
5495 static const char rename_limit_warning[] =
5496 N_("inexact rename detection was skipped due to too many files.");
5497
5498 static const char degrade_cc_to_c_warning[] =
5499 N_("only found copies from modified paths due to too many files.");
5500
5501 static const char rename_limit_advice[] =
5502 N_("you may want to set your %s variable to at least "
5503    "%d and retry the command.");
5504
5505 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5506 {
5507         if (degraded_cc)
5508                 warning(_(degrade_cc_to_c_warning));
5509         else if (needed)
5510                 warning(_(rename_limit_warning));
5511         else
5512                 return;
5513         if (0 < needed && needed < 32767)
5514                 warning(_(rename_limit_advice), varname, needed);
5515 }
5516
5517 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5518 {
5519         int i;
5520         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5521         struct diff_queue_struct *q = &diff_queued_diff;
5522
5523         if (WSEH_NEW & WS_RULE_MASK)
5524                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5525
5526         if (o->color_moved)
5527                 o->emitted_symbols = &esm;
5528
5529         for (i = 0; i < q->nr; i++) {
5530                 struct diff_filepair *p = q->queue[i];
5531                 if (check_pair_status(p))
5532                         diff_flush_patch(p, o);
5533         }
5534
5535         if (o->emitted_symbols) {
5536                 if (o->color_moved) {
5537                         struct hashmap add_lines, del_lines;
5538
5539                         hashmap_init(&del_lines,
5540                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5541                         hashmap_init(&add_lines,
5542                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5543
5544                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5545                         mark_color_as_moved(o, &add_lines, &del_lines);
5546                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5547                                 dim_moved_lines(o);
5548
5549                         hashmap_free(&add_lines, 0);
5550                         hashmap_free(&del_lines, 0);
5551                 }
5552
5553                 for (i = 0; i < esm.nr; i++)
5554                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5555
5556                 for (i = 0; i < esm.nr; i++)
5557                         free((void *)esm.buf[i].line);
5558         }
5559         esm.nr = 0;
5560 }
5561
5562 void diff_flush(struct diff_options *options)
5563 {
5564         struct diff_queue_struct *q = &diff_queued_diff;
5565         int i, output_format = options->output_format;
5566         int separator = 0;
5567         int dirstat_by_line = 0;
5568
5569         /*
5570          * Order: raw, stat, summary, patch
5571          * or:    name/name-status/checkdiff (other bits clear)
5572          */
5573         if (!q->nr)
5574                 goto free_queue;
5575
5576         if (output_format & (DIFF_FORMAT_RAW |
5577                              DIFF_FORMAT_NAME |
5578                              DIFF_FORMAT_NAME_STATUS |
5579                              DIFF_FORMAT_CHECKDIFF)) {
5580                 for (i = 0; i < q->nr; i++) {
5581                         struct diff_filepair *p = q->queue[i];
5582                         if (check_pair_status(p))
5583                                 flush_one_pair(p, options);
5584                 }
5585                 separator++;
5586         }
5587
5588         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
5589                 dirstat_by_line = 1;
5590
5591         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5592             dirstat_by_line) {
5593                 struct diffstat_t diffstat;
5594
5595                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5596                 for (i = 0; i < q->nr; i++) {
5597                         struct diff_filepair *p = q->queue[i];
5598                         if (check_pair_status(p))
5599                                 diff_flush_stat(p, options, &diffstat);
5600                 }
5601                 if (output_format & DIFF_FORMAT_NUMSTAT)
5602                         show_numstat(&diffstat, options);
5603                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5604                         show_stats(&diffstat, options);
5605                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5606                         show_shortstats(&diffstat, options);
5607                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5608                         show_dirstat_by_line(&diffstat, options);
5609                 free_diffstat_info(&diffstat);
5610                 separator++;
5611         }
5612         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5613                 show_dirstat(options);
5614
5615         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5616                 for (i = 0; i < q->nr; i++) {
5617                         diff_summary(options, q->queue[i]);
5618                 }
5619                 separator++;
5620         }
5621
5622         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5623             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
5624             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5625                 /*
5626                  * run diff_flush_patch for the exit status. setting
5627                  * options->file to /dev/null should be safe, because we
5628                  * aren't supposed to produce any output anyway.
5629                  */
5630                 if (options->close_file)
5631                         fclose(options->file);
5632                 options->file = xfopen("/dev/null", "w");
5633                 options->close_file = 1;
5634                 options->color_moved = 0;
5635                 for (i = 0; i < q->nr; i++) {
5636                         struct diff_filepair *p = q->queue[i];
5637                         if (check_pair_status(p))
5638                                 diff_flush_patch(p, options);
5639                         if (options->found_changes)
5640                                 break;
5641                 }
5642         }
5643
5644         if (output_format & DIFF_FORMAT_PATCH) {
5645                 if (separator) {
5646                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5647                         if (options->stat_sep)
5648                                 /* attach patch instead of inline */
5649                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5650                                                  NULL, 0, 0);
5651                 }
5652
5653                 diff_flush_patch_all_file_pairs(options);
5654         }
5655
5656         if (output_format & DIFF_FORMAT_CALLBACK)
5657                 options->format_callback(q, options, options->format_callback_data);
5658
5659         for (i = 0; i < q->nr; i++)
5660                 diff_free_filepair(q->queue[i]);
5661 free_queue:
5662         free(q->queue);
5663         DIFF_QUEUE_CLEAR(q);
5664         if (options->close_file)
5665                 fclose(options->file);
5666
5667         /*
5668          * Report the content-level differences with HAS_CHANGES;
5669          * diff_addremove/diff_change does not set the bit when
5670          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5671          */
5672         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5673                 if (options->found_changes)
5674                         DIFF_OPT_SET(options, HAS_CHANGES);
5675                 else
5676                         DIFF_OPT_CLR(options, HAS_CHANGES);
5677         }
5678 }
5679
5680 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5681 {
5682         return (((p->status == DIFF_STATUS_MODIFIED) &&
5683                  ((p->score &&
5684                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5685                   (!p->score &&
5686                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5687                 ((p->status != DIFF_STATUS_MODIFIED) &&
5688                  filter_bit_tst(p->status, options)));
5689 }
5690
5691 static void diffcore_apply_filter(struct diff_options *options)
5692 {
5693         int i;
5694         struct diff_queue_struct *q = &diff_queued_diff;
5695         struct diff_queue_struct outq;
5696
5697         DIFF_QUEUE_CLEAR(&outq);
5698
5699         if (!options->filter)
5700                 return;
5701
5702         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5703                 int found;
5704                 for (i = found = 0; !found && i < q->nr; i++) {
5705                         if (match_filter(options, q->queue[i]))
5706                                 found++;
5707                 }
5708                 if (found)
5709                         return;
5710
5711                 /* otherwise we will clear the whole queue
5712                  * by copying the empty outq at the end of this
5713                  * function, but first clear the current entries
5714                  * in the queue.
5715                  */
5716                 for (i = 0; i < q->nr; i++)
5717                         diff_free_filepair(q->queue[i]);
5718         }
5719         else {
5720                 /* Only the matching ones */
5721                 for (i = 0; i < q->nr; i++) {
5722                         struct diff_filepair *p = q->queue[i];
5723                         if (match_filter(options, p))
5724                                 diff_q(&outq, p);
5725                         else
5726                                 diff_free_filepair(p);
5727                 }
5728         }
5729         free(q->queue);
5730         *q = outq;
5731 }
5732
5733 /* Check whether two filespecs with the same mode and size are identical */
5734 static int diff_filespec_is_identical(struct diff_filespec *one,
5735                                       struct diff_filespec *two)
5736 {
5737         if (S_ISGITLINK(one->mode))
5738                 return 0;
5739         if (diff_populate_filespec(one, 0))
5740                 return 0;
5741         if (diff_populate_filespec(two, 0))
5742                 return 0;
5743         return !memcmp(one->data, two->data, one->size);
5744 }
5745
5746 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5747 {
5748         if (p->done_skip_stat_unmatch)
5749                 return p->skip_stat_unmatch_result;
5750
5751         p->done_skip_stat_unmatch = 1;
5752         p->skip_stat_unmatch_result = 0;
5753         /*
5754          * 1. Entries that come from stat info dirtiness
5755          *    always have both sides (iow, not create/delete),
5756          *    one side of the object name is unknown, with
5757          *    the same mode and size.  Keep the ones that
5758          *    do not match these criteria.  They have real
5759          *    differences.
5760          *
5761          * 2. At this point, the file is known to be modified,
5762          *    with the same mode and size, and the object
5763          *    name of one side is unknown.  Need to inspect
5764          *    the identical contents.
5765          */
5766         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5767             !DIFF_FILE_VALID(p->two) ||
5768             (p->one->oid_valid && p->two->oid_valid) ||
5769             (p->one->mode != p->two->mode) ||
5770             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5771             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5772             (p->one->size != p->two->size) ||
5773             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5774                 p->skip_stat_unmatch_result = 1;
5775         return p->skip_stat_unmatch_result;
5776 }
5777
5778 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5779 {
5780         int i;
5781         struct diff_queue_struct *q = &diff_queued_diff;
5782         struct diff_queue_struct outq;
5783         DIFF_QUEUE_CLEAR(&outq);
5784
5785         for (i = 0; i < q->nr; i++) {
5786                 struct diff_filepair *p = q->queue[i];
5787
5788                 if (diff_filespec_check_stat_unmatch(p))
5789                         diff_q(&outq, p);
5790                 else {
5791                         /*
5792                          * The caller can subtract 1 from skip_stat_unmatch
5793                          * to determine how many paths were dirty only
5794                          * due to stat info mismatch.
5795                          */
5796                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
5797                                 diffopt->skip_stat_unmatch++;
5798                         diff_free_filepair(p);
5799                 }
5800         }
5801         free(q->queue);
5802         *q = outq;
5803 }
5804
5805 static int diffnamecmp(const void *a_, const void *b_)
5806 {
5807         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5808         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5809         const char *name_a, *name_b;
5810
5811         name_a = a->one ? a->one->path : a->two->path;
5812         name_b = b->one ? b->one->path : b->two->path;
5813         return strcmp(name_a, name_b);
5814 }
5815
5816 void diffcore_fix_diff_index(struct diff_options *options)
5817 {
5818         struct diff_queue_struct *q = &diff_queued_diff;
5819         QSORT(q->queue, q->nr, diffnamecmp);
5820 }
5821
5822 void diffcore_std(struct diff_options *options)
5823 {
5824         /* NOTE please keep the following in sync with diff_tree_combined() */
5825         if (options->skip_stat_unmatch)
5826                 diffcore_skip_stat_unmatch(options);
5827         if (!options->found_follow) {
5828                 /* See try_to_follow_renames() in tree-diff.c */
5829                 if (options->break_opt != -1)
5830                         diffcore_break(options->break_opt);
5831                 if (options->detect_rename)
5832                         diffcore_rename(options);
5833                 if (options->break_opt != -1)
5834                         diffcore_merge_broken();
5835         }
5836         if (options->pickaxe)
5837                 diffcore_pickaxe(options);
5838         if (options->orderfile)
5839                 diffcore_order(options->orderfile);
5840         if (!options->found_follow)
5841                 /* See try_to_follow_renames() in tree-diff.c */
5842                 diff_resolve_rename_copy();
5843         diffcore_apply_filter(options);
5844
5845         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5846                 DIFF_OPT_SET(options, HAS_CHANGES);
5847         else
5848                 DIFF_OPT_CLR(options, HAS_CHANGES);
5849
5850         options->found_follow = 0;
5851 }
5852
5853 int diff_result_code(struct diff_options *opt, int status)
5854 {
5855         int result = 0;
5856
5857         diff_warn_rename_limit("diff.renameLimit",
5858                                opt->needed_rename_limit,
5859                                opt->degraded_cc_to_c);
5860         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5861             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5862                 return status;
5863         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5864             DIFF_OPT_TST(opt, HAS_CHANGES))
5865                 result |= 01;
5866         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5867             DIFF_OPT_TST(opt, CHECK_FAILED))
5868                 result |= 02;
5869         return result;
5870 }
5871
5872 int diff_can_quit_early(struct diff_options *opt)
5873 {
5874         return (DIFF_OPT_TST(opt, QUICK) &&
5875                 !opt->filter &&
5876                 DIFF_OPT_TST(opt, HAS_CHANGES));
5877 }
5878
5879 /*
5880  * Shall changes to this submodule be ignored?
5881  *
5882  * Submodule changes can be configured to be ignored separately for each path,
5883  * but that configuration can be overridden from the command line.
5884  */
5885 static int is_submodule_ignored(const char *path, struct diff_options *options)
5886 {
5887         int ignored = 0;
5888         unsigned orig_flags = options->flags;
5889         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5890                 set_diffopt_flags_from_submodule_config(options, path);
5891         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5892                 ignored = 1;
5893         options->flags = orig_flags;
5894         return ignored;
5895 }
5896
5897 void diff_addremove(struct diff_options *options,
5898                     int addremove, unsigned mode,
5899                     const struct object_id *oid,
5900                     int oid_valid,
5901                     const char *concatpath, unsigned dirty_submodule)
5902 {
5903         struct diff_filespec *one, *two;
5904
5905         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5906                 return;
5907
5908         /* This may look odd, but it is a preparation for
5909          * feeding "there are unchanged files which should
5910          * not produce diffs, but when you are doing copy
5911          * detection you would need them, so here they are"
5912          * entries to the diff-core.  They will be prefixed
5913          * with something like '=' or '*' (I haven't decided
5914          * which but should not make any difference).
5915          * Feeding the same new and old to diff_change()
5916          * also has the same effect.
5917          * Before the final output happens, they are pruned after
5918          * merged into rename/copy pairs as appropriate.
5919          */
5920         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5921                 addremove = (addremove == '+' ? '-' :
5922                              addremove == '-' ? '+' : addremove);
5923
5924         if (options->prefix &&
5925             strncmp(concatpath, options->prefix, options->prefix_length))
5926                 return;
5927
5928         one = alloc_filespec(concatpath);
5929         two = alloc_filespec(concatpath);
5930
5931         if (addremove != '+')
5932                 fill_filespec(one, oid, oid_valid, mode);
5933         if (addremove != '-') {
5934                 fill_filespec(two, oid, oid_valid, mode);
5935                 two->dirty_submodule = dirty_submodule;
5936         }
5937
5938         diff_queue(&diff_queued_diff, one, two);
5939         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5940                 DIFF_OPT_SET(options, HAS_CHANGES);
5941 }
5942
5943 void diff_change(struct diff_options *options,
5944                  unsigned old_mode, unsigned new_mode,
5945                  const struct object_id *old_oid,
5946                  const struct object_id *new_oid,
5947                  int old_oid_valid, int new_oid_valid,
5948                  const char *concatpath,
5949                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5950 {
5951         struct diff_filespec *one, *two;
5952         struct diff_filepair *p;
5953
5954         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5955             is_submodule_ignored(concatpath, options))
5956                 return;
5957
5958         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5959                 SWAP(old_mode, new_mode);
5960                 SWAP(old_oid, new_oid);
5961                 SWAP(old_oid_valid, new_oid_valid);
5962                 SWAP(old_dirty_submodule, new_dirty_submodule);
5963         }
5964
5965         if (options->prefix &&
5966             strncmp(concatpath, options->prefix, options->prefix_length))
5967                 return;
5968
5969         one = alloc_filespec(concatpath);
5970         two = alloc_filespec(concatpath);
5971         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5972         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5973         one->dirty_submodule = old_dirty_submodule;
5974         two->dirty_submodule = new_dirty_submodule;
5975         p = diff_queue(&diff_queued_diff, one, two);
5976
5977         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5978                 return;
5979
5980         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5981             !diff_filespec_check_stat_unmatch(p))
5982                 return;
5983
5984         DIFF_OPT_SET(options, HAS_CHANGES);
5985 }
5986
5987 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5988 {
5989         struct diff_filepair *pair;
5990         struct diff_filespec *one, *two;
5991
5992         if (options->prefix &&
5993             strncmp(path, options->prefix, options->prefix_length))
5994                 return NULL;
5995
5996         one = alloc_filespec(path);
5997         two = alloc_filespec(path);
5998         pair = diff_queue(&diff_queued_diff, one, two);
5999         pair->is_unmerged = 1;
6000         return pair;
6001 }
6002
6003 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6004                 size_t *outsize)
6005 {
6006         struct diff_tempfile *temp;
6007         const char *argv[3];
6008         const char **arg = argv;
6009         struct child_process child = CHILD_PROCESS_INIT;
6010         struct strbuf buf = STRBUF_INIT;
6011         int err = 0;
6012
6013         temp = prepare_temp_file(spec->path, spec);
6014         *arg++ = pgm;
6015         *arg++ = temp->name;
6016         *arg = NULL;
6017
6018         child.use_shell = 1;
6019         child.argv = argv;
6020         child.out = -1;
6021         if (start_command(&child)) {
6022                 remove_tempfile();
6023                 return NULL;
6024         }
6025
6026         if (strbuf_read(&buf, child.out, 0) < 0)
6027                 err = error("error reading from textconv command '%s'", pgm);
6028         close(child.out);
6029
6030         if (finish_command(&child) || err) {
6031                 strbuf_release(&buf);
6032                 remove_tempfile();
6033                 return NULL;
6034         }
6035         remove_tempfile();
6036
6037         return strbuf_detach(&buf, outsize);
6038 }
6039
6040 size_t fill_textconv(struct userdiff_driver *driver,
6041                      struct diff_filespec *df,
6042                      char **outbuf)
6043 {
6044         size_t size;
6045
6046         if (!driver) {
6047                 if (!DIFF_FILE_VALID(df)) {
6048                         *outbuf = "";
6049                         return 0;
6050                 }
6051                 if (diff_populate_filespec(df, 0))
6052                         die("unable to read files to diff");
6053                 *outbuf = df->data;
6054                 return df->size;
6055         }
6056
6057         if (!driver->textconv)
6058                 die("BUG: fill_textconv called with non-textconv driver");
6059
6060         if (driver->textconv_cache && df->oid_valid) {
6061                 *outbuf = notes_cache_get(driver->textconv_cache,
6062                                           &df->oid,
6063                                           &size);
6064                 if (*outbuf)
6065                         return size;
6066         }
6067
6068         *outbuf = run_textconv(driver->textconv, df, &size);
6069         if (!*outbuf)
6070                 die("unable to read files to diff");
6071
6072         if (driver->textconv_cache && df->oid_valid) {
6073                 /* ignore errors, as we might be in a readonly repository */
6074                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6075                                 size);
6076                 /*
6077                  * we could save up changes and flush them all at the end,
6078                  * but we would need an extra call after all diffing is done.
6079                  * Since generating a cache entry is the slow path anyway,
6080                  * this extra overhead probably isn't a big deal.
6081                  */
6082                 notes_cache_write(driver->textconv_cache);
6083         }
6084
6085         return size;
6086 }
6087
6088 int textconv_object(const char *path,
6089                     unsigned mode,
6090                     const struct object_id *oid,
6091                     int oid_valid,
6092                     char **buf,
6093                     unsigned long *buf_size)
6094 {
6095         struct diff_filespec *df;
6096         struct userdiff_driver *textconv;
6097
6098         df = alloc_filespec(path);
6099         fill_filespec(df, oid, oid_valid, mode);
6100         textconv = get_textconv(df);
6101         if (!textconv) {
6102                 free_filespec(df);
6103                 return 0;
6104         }
6105
6106         *buf_size = fill_textconv(textconv, df, buf);
6107         free_filespec(df);
6108         return 1;
6109 }
6110
6111 void setup_diff_pager(struct diff_options *opt)
6112 {
6113         /*
6114          * If the user asked for our exit code, then either they want --quiet
6115          * or --exit-code. We should definitely not bother with a pager in the
6116          * former case, as we will generate no output. Since we still properly
6117          * report our exit code even when a pager is run, we _could_ run a
6118          * pager with --exit-code. But since we have not done so historically,
6119          * and because it is easy to find people oneline advising "git diff
6120          * --exit-code" in hooks and other scripts, we do not do so.
6121          */
6122         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
6123             check_pager_config("diff") != 0)
6124                 setup_pager();
6125 }