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 #include "packfile.h"
  25
  26 #ifdef NO_FAST_WORKING_DIRECTORY
  27 #define FAST_WORKING_DIRECTORY 0
  28 #else
  29 #define FAST_WORKING_DIRECTORY 1
  30 #endif
  31
  32 static int diff_detect_rename_default;
  33 static int diff_indent_heuristic = 1;
  34 static int diff_rename_limit_default = 400;
  35 static int diff_suppress_blank_empty;
  36 static int diff_use_color_default = -1;
  37 static int diff_color_moved_default;
  38 static int diff_context_default = 3;
  39 static int diff_interhunk_context_default;
  40 static const char *diff_word_regex_cfg;
  41 static const char *external_diff_cmd_cfg;
  42 static const char *diff_order_file_cfg;
  43 int diff_auto_refresh_index = 1;
  44 static int diff_mnemonic_prefix;
  45 static int diff_no_prefix;
  46 static int diff_stat_graph_width;
  47 static int diff_dirstat_permille_default = 30;
  48 static struct diff_options default_diff_options;
  49 static long diff_algorithm;
  50 static unsigned ws_error_highlight_default = WSEH_NEW;
  51
  52 static char diff_colors[][COLOR_MAXLEN] = {
  53         GIT_COLOR_RESET,
  54         GIT_COLOR_NORMAL,       /* CONTEXT */
  55         GIT_COLOR_BOLD,         /* METAINFO */
  56         GIT_COLOR_CYAN,         /* FRAGINFO */
  57         GIT_COLOR_RED,          /* OLD */
  58         GIT_COLOR_GREEN,        /* NEW */
  59         GIT_COLOR_YELLOW,       /* COMMIT */
  60         GIT_COLOR_BG_RED,       /* WHITESPACE */
  61         GIT_COLOR_NORMAL,       /* FUNCINFO */
  62         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  63         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  64         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  65         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  66         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  67         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  70 };
  71
  72 static NORETURN void die_want_option(const char *option_name)
  73 {
  74         die(_("option '%s' requires a value"), option_name);
  75 }
  76
  77 static int parse_diff_color_slot(const char *var)
  78 {
  79         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  80                 return DIFF_CONTEXT;
  81         if (!strcasecmp(var, "meta"))
  82                 return DIFF_METAINFO;
  83         if (!strcasecmp(var, "frag"))
  84                 return DIFF_FRAGINFO;
  85         if (!strcasecmp(var, "old"))
  86                 return DIFF_FILE_OLD;
  87         if (!strcasecmp(var, "new"))
  88                 return DIFF_FILE_NEW;
  89         if (!strcasecmp(var, "commit"))
  90                 return DIFF_COMMIT;
  91         if (!strcasecmp(var, "whitespace"))
  92                 return DIFF_WHITESPACE;
  93         if (!strcasecmp(var, "func"))
  94                 return DIFF_FUNCINFO;
  95         if (!strcasecmp(var, "oldmoved"))
  96                 return DIFF_FILE_OLD_MOVED;
  97         if (!strcasecmp(var, "oldmovedalternative"))
  98                 return DIFF_FILE_OLD_MOVED_ALT;
  99         if (!strcasecmp(var, "oldmoveddimmed"))
 100                 return DIFF_FILE_OLD_MOVED_DIM;
 101         if (!strcasecmp(var, "oldmovedalternativedimmed"))
 102                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
 103         if (!strcasecmp(var, "newmoved"))
 104                 return DIFF_FILE_NEW_MOVED;
 105         if (!strcasecmp(var, "newmovedalternative"))
 106                 return DIFF_FILE_NEW_MOVED_ALT;
 107         if (!strcasecmp(var, "newmoveddimmed"))
 108                 return DIFF_FILE_NEW_MOVED_DIM;
 109         if (!strcasecmp(var, "newmovedalternativedimmed"))
 110                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
 111         return -1;
 112 }
 113
 114 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 115                                 struct strbuf *errmsg)
 116 {
 117         char *params_copy = xstrdup(params_string);
 118         struct string_list params = STRING_LIST_INIT_NODUP;
 119         int ret = 0;
 120         int i;
 121
 122         if (*params_copy)
 123                 string_list_split_in_place(&params, params_copy, ',', -1);
 124         for (i = 0; i < params.nr; i++) {
 125                 const char *p = params.items[i].string;
 126                 if (!strcmp(p, "changes")) {
 127                         options->flags.dirstat_by_line = 0;
 128                         options->flags.dirstat_by_file = 0;
 129                 } else if (!strcmp(p, "lines")) {
 130                         options->flags.dirstat_by_line = 1;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "files")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 1;
 135                 } else if (!strcmp(p, "noncumulative")) {
 136                         options->flags.dirstat_cumulative = 0;
 137                 } else if (!strcmp(p, "cumulative")) {
 138                         options->flags.dirstat_cumulative = 1;
 139                 } else if (isdigit(*p)) {
 140                         char *end;
 141                         int permille = strtoul(p, &end, 10) * 10;
 142                         if (*end == '.' && isdigit(*++end)) {
 143                                 /* only use first digit */
 144                                 permille += *end - '0';
 145                                 /* .. and ignore any further digits */
 146                                 while (isdigit(*++end))
 147                                         ; /* nothing */
 148                         }
 149                         if (!*end)
 150                                 options->dirstat_permille = permille;
 151                         else {
 152                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 153                                             p);
 154                                 ret++;
 155                         }
 156                 } else {
 157                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 158                         ret++;
 159                 }
 160
 161         }
 162         string_list_clear(&params, 0);
 163         free(params_copy);
 164         return ret;
 165 }
 166
 167 static int parse_submodule_params(struct diff_options *options, const char *value)
 168 {
 169         if (!strcmp(value, "log"))
 170                 options->submodule_format = DIFF_SUBMODULE_LOG;
 171         else if (!strcmp(value, "short"))
 172                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 173         else if (!strcmp(value, "diff"))
 174                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 175         else
 176                 return -1;
 177         return 0;
 178 }
 179
 180 static int git_config_rename(const char *var, const char *value)
 181 {
 182         if (!value)
 183                 return DIFF_DETECT_RENAME;
 184         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 185                 return  DIFF_DETECT_COPY;
 186         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 187 }
 188
 189 long parse_algorithm_value(const char *value)
 190 {
 191         if (!value)
 192                 return -1;
 193         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 194                 return 0;
 195         else if (!strcasecmp(value, "minimal"))
 196                 return XDF_NEED_MINIMAL;
 197         else if (!strcasecmp(value, "patience"))
 198                 return XDF_PATIENCE_DIFF;
 199         else if (!strcasecmp(value, "histogram"))
 200                 return XDF_HISTOGRAM_DIFF;
 201         return -1;
 202 }
 203
 204 static int parse_one_token(const char **arg, const char *token)
 205 {
 206         const char *rest;
 207         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 208                 *arg = rest;
 209                 return 1;
 210         }
 211         return 0;
 212 }
 213
 214 static int parse_ws_error_highlight(const char *arg)
 215 {
 216         const char *orig_arg = arg;
 217         unsigned val = 0;
 218
 219         while (*arg) {
 220                 if (parse_one_token(&arg, "none"))
 221                         val = 0;
 222                 else if (parse_one_token(&arg, "default"))
 223                         val = WSEH_NEW;
 224                 else if (parse_one_token(&arg, "all"))
 225                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 226                 else if (parse_one_token(&arg, "new"))
 227                         val |= WSEH_NEW;
 228                 else if (parse_one_token(&arg, "old"))
 229                         val |= WSEH_OLD;
 230                 else if (parse_one_token(&arg, "context"))
 231                         val |= WSEH_CONTEXT;
 232                 else {
 233                         return -1 - (int)(arg - orig_arg);
 234                 }
 235                 if (*arg)
 236                         arg++;
 237         }
 238         return val;
 239 }
 240
 241 /*
 242  * These are to give UI layer defaults.
 243  * The core-level commands such as git-diff-files should
 244  * never be affected by the setting of diff.renames
 245  * the user happens to have in the configuration file.
 246  */
 247 void init_diff_ui_defaults(void)
 248 {
 249         diff_detect_rename_default = DIFF_DETECT_RENAME;
 250 }
 251
 252 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 253 {
 254         if (!strcmp(var, "diff.indentheuristic"))
 255                 diff_indent_heuristic = git_config_bool(var, value);
 256         return 0;
 257 }
 258
 259 static int parse_color_moved(const char *arg)
 260 {
 261         switch (git_parse_maybe_bool(arg)) {
 262         case 0:
 263                 return COLOR_MOVED_NO;
 264         case 1:
 265                 return COLOR_MOVED_DEFAULT;
 266         default:
 267                 break;
 268         }
 269
 270         if (!strcmp(arg, "no"))
 271                 return COLOR_MOVED_NO;
 272         else if (!strcmp(arg, "plain"))
 273                 return COLOR_MOVED_PLAIN;
 274         else if (!strcmp(arg, "blocks"))
 275                 return COLOR_MOVED_BLOCKS;
 276         else if (!strcmp(arg, "zebra"))
 277                 return COLOR_MOVED_ZEBRA;
 278         else if (!strcmp(arg, "default"))
 279                 return COLOR_MOVED_DEFAULT;
 280         else if (!strcmp(arg, "dimmed_zebra"))
 281                 return COLOR_MOVED_ZEBRA_DIM;
 282         else
 283                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 284 }
 285
 286 static int parse_color_moved_ws(const char *arg)
 287 {
 288         int ret = 0;
 289         struct string_list l = STRING_LIST_INIT_DUP;
 290         struct string_list_item *i;
 291
 292         string_list_split(&l, arg, ',', -1);
 293
 294         for_each_string_list_item(i, &l) {
 295                 struct strbuf sb = STRBUF_INIT;
 296                 strbuf_addstr(&sb, i->string);
 297                 strbuf_trim(&sb);
 298
 299                 if (!strcmp(sb.buf, "ignore-space-change"))
 300                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 301                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 302                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 303                 else if (!strcmp(sb.buf, "ignore-all-space"))
 304                         ret |= XDF_IGNORE_WHITESPACE;
 305                 else
 306                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 307
 308                 strbuf_release(&sb);
 309         }
 310
 311         string_list_clear(&l, 0);
 312
 313         return ret;
 314 }
 315
 316 int git_diff_ui_config(const char *var, const char *value, void *cb)
 317 {
 318         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 319                 diff_use_color_default = git_config_colorbool(var, value);
 320                 return 0;
 321         }
 322         if (!strcmp(var, "diff.colormoved")) {
 323                 int cm = parse_color_moved(value);
 324                 if (cm < 0)
 325                         return -1;
 326                 diff_color_moved_default = cm;
 327                 return 0;
 328         }
 329         if (!strcmp(var, "diff.context")) {
 330                 diff_context_default = git_config_int(var, value);
 331                 if (diff_context_default < 0)
 332                         return -1;
 333                 return 0;
 334         }
 335         if (!strcmp(var, "diff.interhunkcontext")) {
 336                 diff_interhunk_context_default = git_config_int(var, value);
 337                 if (diff_interhunk_context_default < 0)
 338                         return -1;
 339                 return 0;
 340         }
 341         if (!strcmp(var, "diff.renames")) {
 342                 diff_detect_rename_default = git_config_rename(var, value);
 343                 return 0;
 344         }
 345         if (!strcmp(var, "diff.autorefreshindex")) {
 346                 diff_auto_refresh_index = git_config_bool(var, value);
 347                 return 0;
 348         }
 349         if (!strcmp(var, "diff.mnemonicprefix")) {
 350                 diff_mnemonic_prefix = git_config_bool(var, value);
 351                 return 0;
 352         }
 353         if (!strcmp(var, "diff.noprefix")) {
 354                 diff_no_prefix = git_config_bool(var, value);
 355                 return 0;
 356         }
 357         if (!strcmp(var, "diff.statgraphwidth")) {
 358                 diff_stat_graph_width = git_config_int(var, value);
 359                 return 0;
 360         }
 361         if (!strcmp(var, "diff.external"))
 362                 return git_config_string(&external_diff_cmd_cfg, var, value);
 363         if (!strcmp(var, "diff.wordregex"))
 364                 return git_config_string(&diff_word_regex_cfg, var, value);
 365         if (!strcmp(var, "diff.orderfile"))
 366                 return git_config_pathname(&diff_order_file_cfg, var, value);
 367
 368         if (!strcmp(var, "diff.ignoresubmodules"))
 369                 handle_ignore_submodules_arg(&default_diff_options, value);
 370
 371         if (!strcmp(var, "diff.submodule")) {
 372                 if (parse_submodule_params(&default_diff_options, value))
 373                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 374                                 value);
 375                 return 0;
 376         }
 377
 378         if (!strcmp(var, "diff.algorithm")) {
 379                 diff_algorithm = parse_algorithm_value(value);
 380                 if (diff_algorithm < 0)
 381                         return -1;
 382                 return 0;
 383         }
 384
 385         if (!strcmp(var, "diff.wserrorhighlight")) {
 386                 int val = parse_ws_error_highlight(value);
 387                 if (val < 0)
 388                         return -1;
 389                 ws_error_highlight_default = val;
 390                 return 0;
 391         }
 392
 393         if (git_color_config(var, value, cb) < 0)
 394                 return -1;
 395
 396         return git_diff_basic_config(var, value, cb);
 397 }
 398
 399 int git_diff_basic_config(const char *var, const char *value, void *cb)
 400 {
 401         const char *name;
 402
 403         if (!strcmp(var, "diff.renamelimit")) {
 404                 diff_rename_limit_default = git_config_int(var, value);
 405                 return 0;
 406         }
 407
 408         if (userdiff_config(var, value) < 0)
 409                 return -1;
 410
 411         if (skip_prefix(var, "diff.color.", &name) ||
 412             skip_prefix(var, "color.diff.", &name)) {
 413                 int slot = parse_diff_color_slot(name);
 414                 if (slot < 0)
 415                         return 0;
 416                 if (!value)
 417                         return config_error_nonbool(var);
 418                 return color_parse(value, diff_colors[slot]);
 419         }
 420
 421         /* like GNU diff's --suppress-blank-empty option  */
 422         if (!strcmp(var, "diff.suppressblankempty") ||
 423                         /* for backwards compatibility */
 424                         !strcmp(var, "diff.suppress-blank-empty")) {
 425                 diff_suppress_blank_empty = git_config_bool(var, value);
 426                 return 0;
 427         }
 428
 429         if (!strcmp(var, "diff.dirstat")) {
 430                 struct strbuf errmsg = STRBUF_INIT;
 431                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 432                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 433                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 434                                 errmsg.buf);
 435                 strbuf_release(&errmsg);
 436                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 437                 return 0;
 438         }
 439
 440         if (git_diff_heuristic_config(var, value, cb) < 0)
 441                 return -1;
 442
 443         return git_default_config(var, value, cb);
 444 }
 445
 446 static char *quote_two(const char *one, const char *two)
 447 {
 448         int need_one = quote_c_style(one, NULL, NULL, 1);
 449         int need_two = quote_c_style(two, NULL, NULL, 1);
 450         struct strbuf res = STRBUF_INIT;
 451
 452         if (need_one + need_two) {
 453                 strbuf_addch(&res, '"');
 454                 quote_c_style(one, &res, NULL, 1);
 455                 quote_c_style(two, &res, NULL, 1);
 456                 strbuf_addch(&res, '"');
 457         } else {
 458                 strbuf_addstr(&res, one);
 459                 strbuf_addstr(&res, two);
 460         }
 461         return strbuf_detach(&res, NULL);
 462 }
 463
 464 static const char *external_diff(void)
 465 {
 466         static const char *external_diff_cmd = NULL;
 467         static int done_preparing = 0;
 468
 469         if (done_preparing)
 470                 return external_diff_cmd;
 471         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 472         if (!external_diff_cmd)
 473                 external_diff_cmd = external_diff_cmd_cfg;
 474         done_preparing = 1;
 475         return external_diff_cmd;
 476 }
 477
 478 /*
 479  * Keep track of files used for diffing. Sometimes such an entry
 480  * refers to a temporary file, sometimes to an existing file, and
 481  * sometimes to "/dev/null".
 482  */
 483 static struct diff_tempfile {
 484         /*
 485          * filename external diff should read from, or NULL if this
 486          * entry is currently not in use:
 487          */
 488         const char *name;
 489
 490         char hex[GIT_MAX_HEXSZ + 1];
 491         char mode[10];
 492
 493         /*
 494          * If this diff_tempfile instance refers to a temporary file,
 495          * this tempfile object is used to manage its lifetime.
 496          */
 497         struct tempfile *tempfile;
 498 } diff_temp[2];
 499
 500 struct emit_callback {
 501         int color_diff;
 502         unsigned ws_rule;
 503         int blank_at_eof_in_preimage;
 504         int blank_at_eof_in_postimage;
 505         int lno_in_preimage;
 506         int lno_in_postimage;
 507         const char **label_path;
 508         struct diff_words_data *diff_words;
 509         struct diff_options *opt;
 510         struct strbuf *header;
 511 };
 512
 513 static int count_lines(const char *data, int size)
 514 {
 515         int count, ch, completely_empty = 1, nl_just_seen = 0;
 516         count = 0;
 517         while (0 < size--) {
 518                 ch = *data++;
 519                 if (ch == '\n') {
 520                         count++;
 521                         nl_just_seen = 1;
 522                         completely_empty = 0;
 523                 }
 524                 else {
 525                         nl_just_seen = 0;
 526                         completely_empty = 0;
 527                 }
 528         }
 529         if (completely_empty)
 530                 return 0;
 531         if (!nl_just_seen)
 532                 count++; /* no trailing newline */
 533         return count;
 534 }
 535
 536 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 537 {
 538         if (!DIFF_FILE_VALID(one)) {
 539                 mf->ptr = (char *)""; /* does not matter */
 540                 mf->size = 0;
 541                 return 0;
 542         }
 543         else if (diff_populate_filespec(one, 0))
 544                 return -1;
 545
 546         mf->ptr = one->data;
 547         mf->size = one->size;
 548         return 0;
 549 }
 550
 551 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 552 static unsigned long diff_filespec_size(struct diff_filespec *one)
 553 {
 554         if (!DIFF_FILE_VALID(one))
 555                 return 0;
 556         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 557         return one->size;
 558 }
 559
 560 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 561 {
 562         char *ptr = mf->ptr;
 563         long size = mf->size;
 564         int cnt = 0;
 565
 566         if (!size)
 567                 return cnt;
 568         ptr += size - 1; /* pointing at the very end */
 569         if (*ptr != '\n')
 570                 ; /* incomplete line */
 571         else
 572                 ptr--; /* skip the last LF */
 573         while (mf->ptr < ptr) {
 574                 char *prev_eol;
 575                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 576                         if (*prev_eol == '\n')
 577                                 break;
 578                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 579                         break;
 580                 cnt++;
 581                 ptr = prev_eol - 1;
 582         }
 583         return cnt;
 584 }
 585
 586 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 587                                struct emit_callback *ecbdata)
 588 {
 589         int l1, l2, at;
 590         unsigned ws_rule = ecbdata->ws_rule;
 591         l1 = count_trailing_blank(mf1, ws_rule);
 592         l2 = count_trailing_blank(mf2, ws_rule);
 593         if (l2 <= l1) {
 594                 ecbdata->blank_at_eof_in_preimage = 0;
 595                 ecbdata->blank_at_eof_in_postimage = 0;
 596                 return;
 597         }
 598         at = count_lines(mf1->ptr, mf1->size);
 599         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 600
 601         at = count_lines(mf2->ptr, mf2->size);
 602         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 603 }
 604
 605 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 606                         int first, const char *line, int len)
 607 {
 608         int has_trailing_newline, has_trailing_carriage_return;
 609         int nofirst;
 610         FILE *file = o->file;
 611
 612         fputs(diff_line_prefix(o), file);
 613
 614         if (len == 0) {
 615                 has_trailing_newline = (first == '\n');
 616                 has_trailing_carriage_return = (!has_trailing_newline &&
 617                                                 (first == '\r'));
 618                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 619         } else {
 620                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 621                 if (has_trailing_newline)
 622                         len--;
 623                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 624                 if (has_trailing_carriage_return)
 625                         len--;
 626                 nofirst = 0;
 627         }
 628
 629         if (len || !nofirst) {
 630                 fputs(set, file);
 631                 if (!nofirst)
 632                         fputc(first, file);
 633                 fwrite(line, len, 1, file);
 634                 fputs(reset, file);
 635         }
 636         if (has_trailing_carriage_return)
 637                 fputc('\r', file);
 638         if (has_trailing_newline)
 639                 fputc('\n', file);
 640 }
 641
 642 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 643                       const char *line, int len)
 644 {
 645         emit_line_0(o, set, reset, line[0], line+1, len-1);
 646 }
 647
 648 enum diff_symbol {
 649         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 650         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 651         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 652         DIFF_SYMBOL_BINARY_DIFF_BODY,
 653         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 654         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 655         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 656         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 657         DIFF_SYMBOL_STATS_LINE,
 658         DIFF_SYMBOL_WORD_DIFF,
 659         DIFF_SYMBOL_STAT_SEP,
 660         DIFF_SYMBOL_SUMMARY,
 661         DIFF_SYMBOL_SUBMODULE_ADD,
 662         DIFF_SYMBOL_SUBMODULE_DEL,
 663         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 664         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 665         DIFF_SYMBOL_SUBMODULE_HEADER,
 666         DIFF_SYMBOL_SUBMODULE_ERROR,
 667         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 668         DIFF_SYMBOL_REWRITE_DIFF,
 669         DIFF_SYMBOL_BINARY_FILES,
 670         DIFF_SYMBOL_HEADER,
 671         DIFF_SYMBOL_FILEPAIR_PLUS,
 672         DIFF_SYMBOL_FILEPAIR_MINUS,
 673         DIFF_SYMBOL_WORDS_PORCELAIN,
 674         DIFF_SYMBOL_WORDS,
 675         DIFF_SYMBOL_CONTEXT,
 676         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 677         DIFF_SYMBOL_PLUS,
 678         DIFF_SYMBOL_MINUS,
 679         DIFF_SYMBOL_NO_LF_EOF,
 680         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 681         DIFF_SYMBOL_CONTEXT_MARKER,
 682         DIFF_SYMBOL_SEPARATOR
 683 };
 684 /*
 685  * Flags for content lines:
 686  * 0..12 are whitespace rules
 687  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 688  * 16 is marking if the line is blank at EOF
 689  */
 690 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 691 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 692 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 693 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 694 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 695
 696 /*
 697  * This struct is used when we need to buffer the output of the diff output.
 698  *
 699  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 700  * into the pre/post image file. This pointer could be a union with the
 701  * line pointer. By storing an offset into the file instead of the literal line,
 702  * we can decrease the memory footprint for the buffered output. At first we
 703  * may want to only have indirection for the content lines, but we could also
 704  * enhance the state for emitting prefabricated lines, e.g. the similarity
 705  * score line or hunk/file headers would only need to store a number or path
 706  * and then the output can be constructed later on depending on state.
 707  */
 708 struct emitted_diff_symbol {
 709         const char *line;
 710         int len;
 711         int flags;
 712         enum diff_symbol s;
 713 };
 714 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 715
 716 struct emitted_diff_symbols {
 717         struct emitted_diff_symbol *buf;
 718         int nr, alloc;
 719 };
 720 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 721
 722 static void append_emitted_diff_symbol(struct diff_options *o,
 723                                        struct emitted_diff_symbol *e)
 724 {
 725         struct emitted_diff_symbol *f;
 726
 727         ALLOC_GROW(o->emitted_symbols->buf,
 728                    o->emitted_symbols->nr + 1,
 729                    o->emitted_symbols->alloc);
 730         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 731
 732         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 733         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 734 }
 735
 736 struct moved_entry {
 737         struct hashmap_entry ent;
 738         const struct emitted_diff_symbol *es;
 739         struct moved_entry *next_line;
 740 };
 741
 742 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 743                            const void *entry,
 744                            const void *entry_or_key,
 745                            const void *keydata)
 746 {
 747         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 748         const struct moved_entry *a = entry;
 749         const struct moved_entry *b = entry_or_key;
 750         unsigned flags = diffopt->color_moved_ws_handling
 751                          & XDF_WHITESPACE_FLAGS;
 752
 753         return !xdiff_compare_lines(a->es->line, a->es->len,
 754                                     b->es->line, b->es->len,
 755                                     flags);
 756 }
 757
 758 static struct moved_entry *prepare_entry(struct diff_options *o,
 759                                          int line_no)
 760 {
 761         struct moved_entry *ret = xmalloc(sizeof(*ret));
 762         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 763         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 764
 765         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 766         ret->es = l;
 767         ret->next_line = NULL;
 768
 769         return ret;
 770 }
 771
 772 static void add_lines_to_move_detection(struct diff_options *o,
 773                                         struct hashmap *add_lines,
 774                                         struct hashmap *del_lines)
 775 {
 776         struct moved_entry *prev_line = NULL;
 777
 778         int n;
 779         for (n = 0; n < o->emitted_symbols->nr; n++) {
 780                 struct hashmap *hm;
 781                 struct moved_entry *key;
 782
 783                 switch (o->emitted_symbols->buf[n].s) {
 784                 case DIFF_SYMBOL_PLUS:
 785                         hm = add_lines;
 786                         break;
 787                 case DIFF_SYMBOL_MINUS:
 788                         hm = del_lines;
 789                         break;
 790                 default:
 791                         prev_line = NULL;
 792                         continue;
 793                 }
 794
 795                 key = prepare_entry(o, n);
 796                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 797                         prev_line->next_line = key;
 798
 799                 hashmap_add(hm, key);
 800                 prev_line = key;
 801         }
 802 }
 803
 804 static void pmb_advance_or_null(struct diff_options *o,
 805                                 struct moved_entry *match,
 806                                 struct hashmap *hm,
 807                                 struct moved_entry **pmb,
 808                                 int pmb_nr)
 809 {
 810         int i;
 811         for (i = 0; i < pmb_nr; i++) {
 812                 struct moved_entry *prev = pmb[i];
 813                 struct moved_entry *cur = (prev && prev->next_line) ?
 814                                 prev->next_line : NULL;
 815                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 816                         pmb[i] = cur;
 817                 } else {
 818                         pmb[i] = NULL;
 819                 }
 820         }
 821 }
 822
 823 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 824                                          int pmb_nr)
 825 {
 826         int lp, rp;
 827
 828         /* Shrink the set of potential block to the remaining running */
 829         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 830                 while (lp < pmb_nr && pmb[lp])
 831                         lp++;
 832                 /* lp points at the first NULL now */
 833
 834                 while (rp > -1 && !pmb[rp])
 835                         rp--;
 836                 /* rp points at the last non-NULL */
 837
 838                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 839                         pmb[lp] = pmb[rp];
 840                         pmb[rp] = NULL;
 841                         rp--;
 842                         lp++;
 843                 }
 844         }
 845
 846         /* Remember the number of running sets */
 847         return rp + 1;
 848 }
 849
 850 /*
 851  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 852  *
 853  * Otherwise, if the last block has fewer alphanumeric characters than
 854  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 855  * that block.
 856  *
 857  * The last block consists of the (n - block_length)'th line up to but not
 858  * including the nth line.
 859  *
 860  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
 861  * Think of a way to unify them.
 862  */
 863 static void adjust_last_block(struct diff_options *o, int n, int block_length)
 864 {
 865         int i, alnum_count = 0;
 866         if (o->color_moved == COLOR_MOVED_PLAIN)
 867                 return;
 868         for (i = 1; i < block_length + 1; i++) {
 869                 const char *c = o->emitted_symbols->buf[n - i].line;
 870                 for (; *c; c++) {
 871                         if (!isalnum(*c))
 872                                 continue;
 873                         alnum_count++;
 874                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
 875                                 return;
 876                 }
 877         }
 878         for (i = 1; i < block_length + 1; i++)
 879                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
 880 }
 881
 882 /* Find blocks of moved code, delegate actual coloring decision to helper */
 883 static void mark_color_as_moved(struct diff_options *o,
 884                                 struct hashmap *add_lines,
 885                                 struct hashmap *del_lines)
 886 {
 887         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 888         int pmb_nr = 0, pmb_alloc = 0;
 889         int n, flipped_block = 1, block_length = 0;
 890
 891
 892         for (n = 0; n < o->emitted_symbols->nr; n++) {
 893                 struct hashmap *hm = NULL;
 894                 struct moved_entry *key;
 895                 struct moved_entry *match = NULL;
 896                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 897
 898                 switch (l->s) {
 899                 case DIFF_SYMBOL_PLUS:
 900                         hm = del_lines;
 901                         key = prepare_entry(o, n);
 902                         match = hashmap_get(hm, key, NULL);
 903                         free(key);
 904                         break;
 905                 case DIFF_SYMBOL_MINUS:
 906                         hm = add_lines;
 907                         key = prepare_entry(o, n);
 908                         match = hashmap_get(hm, key, NULL);
 909                         free(key);
 910                         break;
 911                 default:
 912                         flipped_block = 1;
 913                 }
 914
 915                 if (!match) {
 916                         adjust_last_block(o, n, block_length);
 917                         pmb_nr = 0;
 918                         block_length = 0;
 919                         continue;
 920                 }
 921
 922                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 923
 924                 if (o->color_moved == COLOR_MOVED_PLAIN)
 925                         continue;
 926
 927                 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
 928
 929                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 930
 931                 if (pmb_nr == 0) {
 932                         /*
 933                          * The current line is the start of a new block.
 934                          * Setup the set of potential blocks.
 935                          */
 936                         for (; match; match = hashmap_get_next(hm, match)) {
 937                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 938                                 pmb[pmb_nr++] = match;
 939                         }
 940
 941                         flipped_block = (flipped_block + 1) % 2;
 942
 943                         adjust_last_block(o, n, block_length);
 944                         block_length = 0;
 945                 }
 946
 947                 block_length++;
 948
 949                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
 950                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 951         }
 952         adjust_last_block(o, n, block_length);
 953
 954         free(pmb);
 955 }
 956
 957 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
 958   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
 959 static void dim_moved_lines(struct diff_options *o)
 960 {
 961         int n;
 962         for (n = 0; n < o->emitted_symbols->nr; n++) {
 963                 struct emitted_diff_symbol *prev = (n != 0) ?
 964                                 &o->emitted_symbols->buf[n - 1] : NULL;
 965                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 966                 struct emitted_diff_symbol *next =
 967                                 (n < o->emitted_symbols->nr - 1) ?
 968                                 &o->emitted_symbols->buf[n + 1] : NULL;
 969
 970                 /* Not a plus or minus line? */
 971                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
 972                         continue;
 973
 974                 /* Not a moved line? */
 975                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
 976                         continue;
 977
 978                 /*
 979                  * If prev or next are not a plus or minus line,
 980                  * pretend they don't exist
 981                  */
 982                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
 983                             prev->s != DIFF_SYMBOL_MINUS)
 984                         prev = NULL;
 985                 if (next && next->s != DIFF_SYMBOL_PLUS &&
 986                             next->s != DIFF_SYMBOL_MINUS)
 987                         next = NULL;
 988
 989                 /* Inside a block? */
 990                 if ((prev &&
 991                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 992                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
 993                     (next &&
 994                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 995                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
 996                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 997                         continue;
 998                 }
 999
1000                 /* Check if we are at an interesting bound: */
1001                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1002                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1003                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1004                         continue;
1005                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1006                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1007                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1008                         continue;
1009
1010                 /*
1011                  * The boundary to prev and next are not interesting,
1012                  * so this line is not interesting as a whole
1013                  */
1014                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1015         }
1016 }
1017
1018 static void emit_line_ws_markup(struct diff_options *o,
1019                                 const char *set, const char *reset,
1020                                 const char *line, int len, char sign,
1021                                 unsigned ws_rule, int blank_at_eof)
1022 {
1023         const char *ws = NULL;
1024
1025         if (o->ws_error_highlight & ws_rule) {
1026                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1027                 if (!*ws)
1028                         ws = NULL;
1029         }
1030
1031         if (!ws)
1032                 emit_line_0(o, set, reset, sign, line, len);
1033         else if (blank_at_eof)
1034                 /* Blank line at EOF - paint '+' as well */
1035                 emit_line_0(o, ws, reset, sign, line, len);
1036         else {
1037                 /* Emit just the prefix, then the rest. */
1038                 emit_line_0(o, set, reset, sign, "", 0);
1039                 ws_check_emit(line, len, ws_rule,
1040                               o->file, set, reset, ws);
1041         }
1042 }
1043
1044 static void emit_diff_symbol_from_struct(struct diff_options *o,
1045                                          struct emitted_diff_symbol *eds)
1046 {
1047         static const char *nneof = " No newline at end of file\n";
1048         const char *context, *reset, *set, *meta, *fraginfo;
1049         struct strbuf sb = STRBUF_INIT;
1050
1051         enum diff_symbol s = eds->s;
1052         const char *line = eds->line;
1053         int len = eds->len;
1054         unsigned flags = eds->flags;
1055
1056         switch (s) {
1057         case DIFF_SYMBOL_NO_LF_EOF:
1058                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1059                 reset = diff_get_color_opt(o, DIFF_RESET);
1060                 putc('\n', o->file);
1061                 emit_line_0(o, context, reset, '\\',
1062                             nneof, strlen(nneof));
1063                 break;
1064         case DIFF_SYMBOL_SUBMODULE_HEADER:
1065         case DIFF_SYMBOL_SUBMODULE_ERROR:
1066         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1067         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1068         case DIFF_SYMBOL_SUMMARY:
1069         case DIFF_SYMBOL_STATS_LINE:
1070         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1071         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1072                 emit_line(o, "", "", line, len);
1073                 break;
1074         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1075         case DIFF_SYMBOL_CONTEXT_MARKER:
1076                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1077                 reset = diff_get_color_opt(o, DIFF_RESET);
1078                 emit_line(o, context, reset, line, len);
1079                 break;
1080         case DIFF_SYMBOL_SEPARATOR:
1081                 fprintf(o->file, "%s%c",
1082                         diff_line_prefix(o),
1083                         o->line_termination);
1084                 break;
1085         case DIFF_SYMBOL_CONTEXT:
1086                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1087                 reset = diff_get_color_opt(o, DIFF_RESET);
1088                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1089                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1090                 break;
1091         case DIFF_SYMBOL_PLUS:
1092                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1093                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1094                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1095                 case DIFF_SYMBOL_MOVED_LINE |
1096                      DIFF_SYMBOL_MOVED_LINE_ALT |
1097                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1098                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1099                         break;
1100                 case DIFF_SYMBOL_MOVED_LINE |
1101                      DIFF_SYMBOL_MOVED_LINE_ALT:
1102                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1103                         break;
1104                 case DIFF_SYMBOL_MOVED_LINE |
1105                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1106                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1107                         break;
1108                 case DIFF_SYMBOL_MOVED_LINE:
1109                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1110                         break;
1111                 default:
1112                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1113                 }
1114                 reset = diff_get_color_opt(o, DIFF_RESET);
1115                 emit_line_ws_markup(o, set, reset, line, len, '+',
1116                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1117                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1118                 break;
1119         case DIFF_SYMBOL_MINUS:
1120                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1121                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1122                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1123                 case DIFF_SYMBOL_MOVED_LINE |
1124                      DIFF_SYMBOL_MOVED_LINE_ALT |
1125                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1126                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1127                         break;
1128                 case DIFF_SYMBOL_MOVED_LINE |
1129                      DIFF_SYMBOL_MOVED_LINE_ALT:
1130                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1131                         break;
1132                 case DIFF_SYMBOL_MOVED_LINE |
1133                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1134                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1135                         break;
1136                 case DIFF_SYMBOL_MOVED_LINE:
1137                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1138                         break;
1139                 default:
1140                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1141                 }
1142                 reset = diff_get_color_opt(o, DIFF_RESET);
1143                 emit_line_ws_markup(o, set, reset, line, len, '-',
1144                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1145                 break;
1146         case DIFF_SYMBOL_WORDS_PORCELAIN:
1147                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1148                 reset = diff_get_color_opt(o, DIFF_RESET);
1149                 emit_line(o, context, reset, line, len);
1150                 fputs("~\n", o->file);
1151                 break;
1152         case DIFF_SYMBOL_WORDS:
1153                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1154                 reset = diff_get_color_opt(o, DIFF_RESET);
1155                 /*
1156                  * Skip the prefix character, if any.  With
1157                  * diff_suppress_blank_empty, there may be
1158                  * none.
1159                  */
1160                 if (line[0] != '\n') {
1161                         line++;
1162                         len--;
1163                 }
1164                 emit_line(o, context, reset, line, len);
1165                 break;
1166         case DIFF_SYMBOL_FILEPAIR_PLUS:
1167                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1168                 reset = diff_get_color_opt(o, DIFF_RESET);
1169                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1170                         line, reset,
1171                         strchr(line, ' ') ? "\t" : "");
1172                 break;
1173         case DIFF_SYMBOL_FILEPAIR_MINUS:
1174                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1175                 reset = diff_get_color_opt(o, DIFF_RESET);
1176                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1177                         line, reset,
1178                         strchr(line, ' ') ? "\t" : "");
1179                 break;
1180         case DIFF_SYMBOL_BINARY_FILES:
1181         case DIFF_SYMBOL_HEADER:
1182                 fprintf(o->file, "%s", line);
1183                 break;
1184         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1185                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1186                 break;
1187         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1188                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1189                 break;
1190         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1191                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1192                 break;
1193         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1194                 fputs(diff_line_prefix(o), o->file);
1195                 fputc('\n', o->file);
1196                 break;
1197         case DIFF_SYMBOL_REWRITE_DIFF:
1198                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1199                 reset = diff_get_color_opt(o, DIFF_RESET);
1200                 emit_line(o, fraginfo, reset, line, len);
1201                 break;
1202         case DIFF_SYMBOL_SUBMODULE_ADD:
1203                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1204                 reset = diff_get_color_opt(o, DIFF_RESET);
1205                 emit_line(o, set, reset, line, len);
1206                 break;
1207         case DIFF_SYMBOL_SUBMODULE_DEL:
1208                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1209                 reset = diff_get_color_opt(o, DIFF_RESET);
1210                 emit_line(o, set, reset, line, len);
1211                 break;
1212         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1213                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1214                         diff_line_prefix(o), line);
1215                 break;
1216         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1217                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1218                         diff_line_prefix(o), line);
1219                 break;
1220         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1221                 emit_line(o, "", "", " 0 files changed\n",
1222                           strlen(" 0 files changed\n"));
1223                 break;
1224         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1225                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1226                 break;
1227         case DIFF_SYMBOL_WORD_DIFF:
1228                 fprintf(o->file, "%.*s", len, line);
1229                 break;
1230         case DIFF_SYMBOL_STAT_SEP:
1231                 fputs(o->stat_sep, o->file);
1232                 break;
1233         default:
1234                 die("BUG: unknown diff symbol");
1235         }
1236         strbuf_release(&sb);
1237 }
1238
1239 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1240                              const char *line, int len, unsigned flags)
1241 {
1242         struct emitted_diff_symbol e = {line, len, flags, s};
1243
1244         if (o->emitted_symbols)
1245                 append_emitted_diff_symbol(o, &e);
1246         else
1247                 emit_diff_symbol_from_struct(o, &e);
1248 }
1249
1250 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1251 {
1252         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1253 }
1254
1255 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1256 {
1257         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1258 }
1259
1260 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1261 {
1262         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1263                          path, strlen(path), 0);
1264 }
1265
1266 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1267 {
1268         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1269                          path, strlen(path), 0);
1270 }
1271
1272 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1273 {
1274         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1275                          header, strlen(header), 0);
1276 }
1277
1278 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1279 {
1280         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1281 }
1282
1283 void diff_emit_submodule_pipethrough(struct diff_options *o,
1284                                      const char *line, int len)
1285 {
1286         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1287 }
1288
1289 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1290 {
1291         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1292               ecbdata->blank_at_eof_in_preimage &&
1293               ecbdata->blank_at_eof_in_postimage &&
1294               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1295               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1296                 return 0;
1297         return ws_blank_line(line, len, ecbdata->ws_rule);
1298 }
1299
1300 static void emit_add_line(const char *reset,
1301                           struct emit_callback *ecbdata,
1302                           const char *line, int len)
1303 {
1304         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1305         if (new_blank_line_at_eof(ecbdata, line, len))
1306                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1307
1308         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1309 }
1310
1311 static void emit_del_line(const char *reset,
1312                           struct emit_callback *ecbdata,
1313                           const char *line, int len)
1314 {
1315         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1316         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1317 }
1318
1319 static void emit_context_line(const char *reset,
1320                               struct emit_callback *ecbdata,
1321                               const char *line, int len)
1322 {
1323         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1324         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1325 }
1326
1327 static void emit_hunk_header(struct emit_callback *ecbdata,
1328                              const char *line, int len)
1329 {
1330         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1331         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1332         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1333         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1334         static const char atat[2] = { '@', '@' };
1335         const char *cp, *ep;
1336         struct strbuf msgbuf = STRBUF_INIT;
1337         int org_len = len;
1338         int i = 1;
1339
1340         /*
1341          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1342          * it always is at least 10 bytes long.
1343          */
1344         if (len < 10 ||
1345             memcmp(line, atat, 2) ||
1346             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1347                 emit_diff_symbol(ecbdata->opt,
1348                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1349                 return;
1350         }
1351         ep += 2; /* skip over @@ */
1352
1353         /* The hunk header in fraginfo color */
1354         strbuf_addstr(&msgbuf, frag);
1355         strbuf_add(&msgbuf, line, ep - line);
1356         strbuf_addstr(&msgbuf, reset);
1357
1358         /*
1359          * trailing "\r\n"
1360          */
1361         for ( ; i < 3; i++)
1362                 if (line[len - i] == '\r' || line[len - i] == '\n')
1363                         len--;
1364
1365         /* blank before the func header */
1366         for (cp = ep; ep - line < len; ep++)
1367                 if (*ep != ' ' && *ep != '\t')
1368                         break;
1369         if (ep != cp) {
1370                 strbuf_addstr(&msgbuf, context);
1371                 strbuf_add(&msgbuf, cp, ep - cp);
1372                 strbuf_addstr(&msgbuf, reset);
1373         }
1374
1375         if (ep < line + len) {
1376                 strbuf_addstr(&msgbuf, func);
1377                 strbuf_add(&msgbuf, ep, line + len - ep);
1378                 strbuf_addstr(&msgbuf, reset);
1379         }
1380
1381         strbuf_add(&msgbuf, line + len, org_len - len);
1382         strbuf_complete_line(&msgbuf);
1383         emit_diff_symbol(ecbdata->opt,
1384                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1385         strbuf_release(&msgbuf);
1386 }
1387
1388 static struct diff_tempfile *claim_diff_tempfile(void) {
1389         int i;
1390         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1391                 if (!diff_temp[i].name)
1392                         return diff_temp + i;
1393         die("BUG: diff is failing to clean up its tempfiles");
1394 }
1395
1396 static void remove_tempfile(void)
1397 {
1398         int i;
1399         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1400                 if (is_tempfile_active(diff_temp[i].tempfile))
1401                         delete_tempfile(&diff_temp[i].tempfile);
1402                 diff_temp[i].name = NULL;
1403         }
1404 }
1405
1406 static void add_line_count(struct strbuf *out, int count)
1407 {
1408         switch (count) {
1409         case 0:
1410                 strbuf_addstr(out, "0,0");
1411                 break;
1412         case 1:
1413                 strbuf_addstr(out, "1");
1414                 break;
1415         default:
1416                 strbuf_addf(out, "1,%d", count);
1417                 break;
1418         }
1419 }
1420
1421 static void emit_rewrite_lines(struct emit_callback *ecb,
1422                                int prefix, const char *data, int size)
1423 {
1424         const char *endp = NULL;
1425         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1426
1427         while (0 < size) {
1428                 int len;
1429
1430                 endp = memchr(data, '\n', size);
1431                 len = endp ? (endp - data + 1) : size;
1432                 if (prefix != '+') {
1433                         ecb->lno_in_preimage++;
1434                         emit_del_line(reset, ecb, data, len);
1435                 } else {
1436                         ecb->lno_in_postimage++;
1437                         emit_add_line(reset, ecb, data, len);
1438                 }
1439                 size -= len;
1440                 data += len;
1441         }
1442         if (!endp)
1443                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1444 }
1445
1446 static void emit_rewrite_diff(const char *name_a,
1447                               const char *name_b,
1448                               struct diff_filespec *one,
1449                               struct diff_filespec *two,
1450                               struct userdiff_driver *textconv_one,
1451                               struct userdiff_driver *textconv_two,
1452                               struct diff_options *o)
1453 {
1454         int lc_a, lc_b;
1455         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1456         const char *a_prefix, *b_prefix;
1457         char *data_one, *data_two;
1458         size_t size_one, size_two;
1459         struct emit_callback ecbdata;
1460         struct strbuf out = STRBUF_INIT;
1461
1462         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1463                 a_prefix = o->b_prefix;
1464                 b_prefix = o->a_prefix;
1465         } else {
1466                 a_prefix = o->a_prefix;
1467                 b_prefix = o->b_prefix;
1468         }
1469
1470         name_a += (*name_a == '/');
1471         name_b += (*name_b == '/');
1472
1473         strbuf_reset(&a_name);
1474         strbuf_reset(&b_name);
1475         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1476         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1477
1478         size_one = fill_textconv(textconv_one, one, &data_one);
1479         size_two = fill_textconv(textconv_two, two, &data_two);
1480
1481         memset(&ecbdata, 0, sizeof(ecbdata));
1482         ecbdata.color_diff = want_color(o->use_color);
1483         ecbdata.ws_rule = whitespace_rule(name_b);
1484         ecbdata.opt = o;
1485         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1486                 mmfile_t mf1, mf2;
1487                 mf1.ptr = (char *)data_one;
1488                 mf2.ptr = (char *)data_two;
1489                 mf1.size = size_one;
1490                 mf2.size = size_two;
1491                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1492         }
1493         ecbdata.lno_in_preimage = 1;
1494         ecbdata.lno_in_postimage = 1;
1495
1496         lc_a = count_lines(data_one, size_one);
1497         lc_b = count_lines(data_two, size_two);
1498
1499         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1500                          a_name.buf, a_name.len, 0);
1501         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1502                          b_name.buf, b_name.len, 0);
1503
1504         strbuf_addstr(&out, "@@ -");
1505         if (!o->irreversible_delete)
1506                 add_line_count(&out, lc_a);
1507         else
1508                 strbuf_addstr(&out, "?,?");
1509         strbuf_addstr(&out, " +");
1510         add_line_count(&out, lc_b);
1511         strbuf_addstr(&out, " @@\n");
1512         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1513         strbuf_release(&out);
1514
1515         if (lc_a && !o->irreversible_delete)
1516                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1517         if (lc_b)
1518                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1519         if (textconv_one)
1520                 free((char *)data_one);
1521         if (textconv_two)
1522                 free((char *)data_two);
1523 }
1524
1525 struct diff_words_buffer {
1526         mmfile_t text;
1527         unsigned long alloc;
1528         struct diff_words_orig {
1529                 const char *begin, *end;
1530         } *orig;
1531         int orig_nr, orig_alloc;
1532 };
1533
1534 static void diff_words_append(char *line, unsigned long len,
1535                 struct diff_words_buffer *buffer)
1536 {
1537         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1538         line++;
1539         len--;
1540         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1541         buffer->text.size += len;
1542         buffer->text.ptr[buffer->text.size] = '\0';
1543 }
1544
1545 struct diff_words_style_elem {
1546         const char *prefix;
1547         const char *suffix;
1548         const char *color; /* NULL; filled in by the setup code if
1549                             * color is enabled */
1550 };
1551
1552 struct diff_words_style {
1553         enum diff_words_type type;
1554         struct diff_words_style_elem new_word, old_word, ctx;
1555         const char *newline;
1556 };
1557
1558 static struct diff_words_style diff_words_styles[] = {
1559         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1560         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1561         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1562 };
1563
1564 struct diff_words_data {
1565         struct diff_words_buffer minus, plus;
1566         const char *current_plus;
1567         int last_minus;
1568         struct diff_options *opt;
1569         regex_t *word_regex;
1570         enum diff_words_type type;
1571         struct diff_words_style *style;
1572 };
1573
1574 static int fn_out_diff_words_write_helper(struct diff_options *o,
1575                                           struct diff_words_style_elem *st_el,
1576                                           const char *newline,
1577                                           size_t count, const char *buf)
1578 {
1579         int print = 0;
1580         struct strbuf sb = STRBUF_INIT;
1581
1582         while (count) {
1583                 char *p = memchr(buf, '\n', count);
1584                 if (print)
1585                         strbuf_addstr(&sb, diff_line_prefix(o));
1586
1587                 if (p != buf) {
1588                         const char *reset = st_el->color && *st_el->color ?
1589                                             GIT_COLOR_RESET : NULL;
1590                         if (st_el->color && *st_el->color)
1591                                 strbuf_addstr(&sb, st_el->color);
1592                         strbuf_addstr(&sb, st_el->prefix);
1593                         strbuf_add(&sb, buf, p ? p - buf : count);
1594                         strbuf_addstr(&sb, st_el->suffix);
1595                         if (reset)
1596                                 strbuf_addstr(&sb, reset);
1597                 }
1598                 if (!p)
1599                         goto out;
1600
1601                 strbuf_addstr(&sb, newline);
1602                 count -= p + 1 - buf;
1603                 buf = p + 1;
1604                 print = 1;
1605                 if (count) {
1606                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1607                                          sb.buf, sb.len, 0);
1608                         strbuf_reset(&sb);
1609                 }
1610         }
1611
1612 out:
1613         if (sb.len)
1614                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1615                                  sb.buf, sb.len, 0);
1616         strbuf_release(&sb);
1617         return 0;
1618 }
1619
1620 /*
1621  * '--color-words' algorithm can be described as:
1622  *
1623  *   1. collect the minus/plus lines of a diff hunk, divided into
1624  *      minus-lines and plus-lines;
1625  *
1626  *   2. break both minus-lines and plus-lines into words and
1627  *      place them into two mmfile_t with one word for each line;
1628  *
1629  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1630  *
1631  * And for the common parts of the both file, we output the plus side text.
1632  * diff_words->current_plus is used to trace the current position of the plus file
1633  * which printed. diff_words->last_minus is used to trace the last minus word
1634  * printed.
1635  *
1636  * For '--graph' to work with '--color-words', we need to output the graph prefix
1637  * on each line of color words output. Generally, there are two conditions on
1638  * which we should output the prefix.
1639  *
1640  *   1. diff_words->last_minus == 0 &&
1641  *      diff_words->current_plus == diff_words->plus.text.ptr
1642  *
1643  *      that is: the plus text must start as a new line, and if there is no minus
1644  *      word printed, a graph prefix must be printed.
1645  *
1646  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1647  *      *(diff_words->current_plus - 1) == '\n'
1648  *
1649  *      that is: a graph prefix must be printed following a '\n'
1650  */
1651 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1652 {
1653         if ((diff_words->last_minus == 0 &&
1654                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1655                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1656                 *(diff_words->current_plus - 1) == '\n')) {
1657                 return 1;
1658         } else {
1659                 return 0;
1660         }
1661 }
1662
1663 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1664 {
1665         struct diff_words_data *diff_words = priv;
1666         struct diff_words_style *style = diff_words->style;
1667         int minus_first, minus_len, plus_first, plus_len;
1668         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1669         struct diff_options *opt = diff_words->opt;
1670         const char *line_prefix;
1671
1672         if (line[0] != '@' || parse_hunk_header(line, len,
1673                         &minus_first, &minus_len, &plus_first, &plus_len))
1674                 return;
1675
1676         assert(opt);
1677         line_prefix = diff_line_prefix(opt);
1678
1679         /* POSIX requires that first be decremented by one if len == 0... */
1680         if (minus_len) {
1681                 minus_begin = diff_words->minus.orig[minus_first].begin;
1682                 minus_end =
1683                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1684         } else
1685                 minus_begin = minus_end =
1686                         diff_words->minus.orig[minus_first].end;
1687
1688         if (plus_len) {
1689                 plus_begin = diff_words->plus.orig[plus_first].begin;
1690                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1691         } else
1692                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1693
1694         if (color_words_output_graph_prefix(diff_words)) {
1695                 fputs(line_prefix, diff_words->opt->file);
1696         }
1697         if (diff_words->current_plus != plus_begin) {
1698                 fn_out_diff_words_write_helper(diff_words->opt,
1699                                 &style->ctx, style->newline,
1700                                 plus_begin - diff_words->current_plus,
1701                                 diff_words->current_plus);
1702         }
1703         if (minus_begin != minus_end) {
1704                 fn_out_diff_words_write_helper(diff_words->opt,
1705                                 &style->old_word, style->newline,
1706                                 minus_end - minus_begin, minus_begin);
1707         }
1708         if (plus_begin != plus_end) {
1709                 fn_out_diff_words_write_helper(diff_words->opt,
1710                                 &style->new_word, style->newline,
1711                                 plus_end - plus_begin, plus_begin);
1712         }
1713
1714         diff_words->current_plus = plus_end;
1715         diff_words->last_minus = minus_first;
1716 }
1717
1718 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1719 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1720                 int *begin, int *end)
1721 {
1722         if (word_regex && *begin < buffer->size) {
1723                 regmatch_t match[1];
1724                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1725                                  buffer->size - *begin, 1, match, 0)) {
1726                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1727                                         '\n', match[0].rm_eo - match[0].rm_so);
1728                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1729                         *begin += match[0].rm_so;
1730                         return *begin >= *end;
1731                 }
1732                 return -1;
1733         }
1734
1735         /* find the next word */
1736         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1737                 (*begin)++;
1738         if (*begin >= buffer->size)
1739                 return -1;
1740
1741         /* find the end of the word */
1742         *end = *begin + 1;
1743         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1744                 (*end)++;
1745
1746         return 0;
1747 }
1748
1749 /*
1750  * This function splits the words in buffer->text, stores the list with
1751  * newline separator into out, and saves the offsets of the original words
1752  * in buffer->orig.
1753  */
1754 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1755                 regex_t *word_regex)
1756 {
1757         int i, j;
1758         long alloc = 0;
1759
1760         out->size = 0;
1761         out->ptr = NULL;
1762
1763         /* fake an empty "0th" word */
1764         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1765         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1766         buffer->orig_nr = 1;
1767
1768         for (i = 0; i < buffer->text.size; i++) {
1769                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1770                         return;
1771
1772                 /* store original boundaries */
1773                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1774                                 buffer->orig_alloc);
1775                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1776                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1777                 buffer->orig_nr++;
1778
1779                 /* store one word */
1780                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1781                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1782                 out->ptr[out->size + j - i] = '\n';
1783                 out->size += j - i + 1;
1784
1785                 i = j - 1;
1786         }
1787 }
1788
1789 /* this executes the word diff on the accumulated buffers */
1790 static void diff_words_show(struct diff_words_data *diff_words)
1791 {
1792         xpparam_t xpp;
1793         xdemitconf_t xecfg;
1794         mmfile_t minus, plus;
1795         struct diff_words_style *style = diff_words->style;
1796
1797         struct diff_options *opt = diff_words->opt;
1798         const char *line_prefix;
1799
1800         assert(opt);
1801         line_prefix = diff_line_prefix(opt);
1802
1803         /* special case: only removal */
1804         if (!diff_words->plus.text.size) {
1805                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1806                                  line_prefix, strlen(line_prefix), 0);
1807                 fn_out_diff_words_write_helper(diff_words->opt,
1808                         &style->old_word, style->newline,
1809                         diff_words->minus.text.size,
1810                         diff_words->minus.text.ptr);
1811                 diff_words->minus.text.size = 0;
1812                 return;
1813         }
1814
1815         diff_words->current_plus = diff_words->plus.text.ptr;
1816         diff_words->last_minus = 0;
1817
1818         memset(&xpp, 0, sizeof(xpp));
1819         memset(&xecfg, 0, sizeof(xecfg));
1820         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1821         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1822         xpp.flags = 0;
1823         /* as only the hunk header will be parsed, we need a 0-context */
1824         xecfg.ctxlen = 0;
1825         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1826                           &xpp, &xecfg))
1827                 die("unable to generate word diff");
1828         free(minus.ptr);
1829         free(plus.ptr);
1830         if (diff_words->current_plus != diff_words->plus.text.ptr +
1831                         diff_words->plus.text.size) {
1832                 if (color_words_output_graph_prefix(diff_words))
1833                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1834                                          line_prefix, strlen(line_prefix), 0);
1835                 fn_out_diff_words_write_helper(diff_words->opt,
1836                         &style->ctx, style->newline,
1837                         diff_words->plus.text.ptr + diff_words->plus.text.size
1838                         - diff_words->current_plus, diff_words->current_plus);
1839         }
1840         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1841 }
1842
1843 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1844 static void diff_words_flush(struct emit_callback *ecbdata)
1845 {
1846         struct diff_options *wo = ecbdata->diff_words->opt;
1847
1848         if (ecbdata->diff_words->minus.text.size ||
1849             ecbdata->diff_words->plus.text.size)
1850                 diff_words_show(ecbdata->diff_words);
1851
1852         if (wo->emitted_symbols) {
1853                 struct diff_options *o = ecbdata->opt;
1854                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1855                 int i;
1856
1857                 /*
1858                  * NEEDSWORK:
1859                  * Instead of appending each, concat all words to a line?
1860                  */
1861                 for (i = 0; i < wol->nr; i++)
1862                         append_emitted_diff_symbol(o, &wol->buf[i]);
1863
1864                 for (i = 0; i < wol->nr; i++)
1865                         free((void *)wol->buf[i].line);
1866
1867                 wol->nr = 0;
1868         }
1869 }
1870
1871 static void diff_filespec_load_driver(struct diff_filespec *one)
1872 {
1873         /* Use already-loaded driver */
1874         if (one->driver)
1875                 return;
1876
1877         if (S_ISREG(one->mode))
1878                 one->driver = userdiff_find_by_path(one->path);
1879
1880         /* Fallback to default settings */
1881         if (!one->driver)
1882                 one->driver = userdiff_find_by_name("default");
1883 }
1884
1885 static const char *userdiff_word_regex(struct diff_filespec *one)
1886 {
1887         diff_filespec_load_driver(one);
1888         return one->driver->word_regex;
1889 }
1890
1891 static void init_diff_words_data(struct emit_callback *ecbdata,
1892                                  struct diff_options *orig_opts,
1893                                  struct diff_filespec *one,
1894                                  struct diff_filespec *two)
1895 {
1896         int i;
1897         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1898         memcpy(o, orig_opts, sizeof(struct diff_options));
1899
1900         ecbdata->diff_words =
1901                 xcalloc(1, sizeof(struct diff_words_data));
1902         ecbdata->diff_words->type = o->word_diff;
1903         ecbdata->diff_words->opt = o;
1904
1905         if (orig_opts->emitted_symbols)
1906                 o->emitted_symbols =
1907                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1908
1909         if (!o->word_regex)
1910                 o->word_regex = userdiff_word_regex(one);
1911         if (!o->word_regex)
1912                 o->word_regex = userdiff_word_regex(two);
1913         if (!o->word_regex)
1914                 o->word_regex = diff_word_regex_cfg;
1915         if (o->word_regex) {
1916                 ecbdata->diff_words->word_regex = (regex_t *)
1917                         xmalloc(sizeof(regex_t));
1918                 if (regcomp(ecbdata->diff_words->word_regex,
1919                             o->word_regex,
1920                             REG_EXTENDED | REG_NEWLINE))
1921                         die ("Invalid regular expression: %s",
1922                              o->word_regex);
1923         }
1924         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1925                 if (o->word_diff == diff_words_styles[i].type) {
1926                         ecbdata->diff_words->style =
1927                                 &diff_words_styles[i];
1928                         break;
1929                 }
1930         }
1931         if (want_color(o->use_color)) {
1932                 struct diff_words_style *st = ecbdata->diff_words->style;
1933                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1934                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1935                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1936         }
1937 }
1938
1939 static void free_diff_words_data(struct emit_callback *ecbdata)
1940 {
1941         if (ecbdata->diff_words) {
1942                 diff_words_flush(ecbdata);
1943                 free (ecbdata->diff_words->opt->emitted_symbols);
1944                 free (ecbdata->diff_words->opt);
1945                 free (ecbdata->diff_words->minus.text.ptr);
1946                 free (ecbdata->diff_words->minus.orig);
1947                 free (ecbdata->diff_words->plus.text.ptr);
1948                 free (ecbdata->diff_words->plus.orig);
1949                 if (ecbdata->diff_words->word_regex) {
1950                         regfree(ecbdata->diff_words->word_regex);
1951                         free(ecbdata->diff_words->word_regex);
1952                 }
1953                 FREE_AND_NULL(ecbdata->diff_words);
1954         }
1955 }
1956
1957 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1958 {
1959         if (want_color(diff_use_color))
1960                 return diff_colors[ix];
1961         return "";
1962 }
1963
1964 const char *diff_line_prefix(struct diff_options *opt)
1965 {
1966         struct strbuf *msgbuf;
1967         if (!opt->output_prefix)
1968                 return "";
1969
1970         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1971         return msgbuf->buf;
1972 }
1973
1974 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1975 {
1976         const char *cp;
1977         unsigned long allot;
1978         size_t l = len;
1979
1980         cp = line;
1981         allot = l;
1982         while (0 < l) {
1983                 (void) utf8_width(&cp, &l);
1984                 if (!cp)
1985                         break; /* truncated in the middle? */
1986         }
1987         return allot - l;
1988 }
1989
1990 static void find_lno(const char *line, struct emit_callback *ecbdata)
1991 {
1992         const char *p;
1993         ecbdata->lno_in_preimage = 0;
1994         ecbdata->lno_in_postimage = 0;
1995         p = strchr(line, '-');
1996         if (!p)
1997                 return; /* cannot happen */
1998         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1999         p = strchr(p, '+');
2000         if (!p)
2001                 return; /* cannot happen */
2002         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2003 }
2004
2005 static void fn_out_consume(void *priv, char *line, unsigned long len)
2006 {
2007         struct emit_callback *ecbdata = priv;
2008         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2009         struct diff_options *o = ecbdata->opt;
2010
2011         o->found_changes = 1;
2012
2013         if (ecbdata->header) {
2014                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2015                                  ecbdata->header->buf, ecbdata->header->len, 0);
2016                 strbuf_reset(ecbdata->header);
2017                 ecbdata->header = NULL;
2018         }
2019
2020         if (ecbdata->label_path[0]) {
2021                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2022                                  ecbdata->label_path[0],
2023                                  strlen(ecbdata->label_path[0]), 0);
2024                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2025                                  ecbdata->label_path[1],
2026                                  strlen(ecbdata->label_path[1]), 0);
2027                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2028         }
2029
2030         if (diff_suppress_blank_empty
2031             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2032                 line[0] = '\n';
2033                 len = 1;
2034         }
2035
2036         if (line[0] == '@') {
2037                 if (ecbdata->diff_words)
2038                         diff_words_flush(ecbdata);
2039                 len = sane_truncate_line(ecbdata, line, len);
2040                 find_lno(line, ecbdata);
2041                 emit_hunk_header(ecbdata, line, len);
2042                 return;
2043         }
2044
2045         if (ecbdata->diff_words) {
2046                 enum diff_symbol s =
2047                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2048                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2049                 if (line[0] == '-') {
2050                         diff_words_append(line, len,
2051                                           &ecbdata->diff_words->minus);
2052                         return;
2053                 } else if (line[0] == '+') {
2054                         diff_words_append(line, len,
2055                                           &ecbdata->diff_words->plus);
2056                         return;
2057                 } else if (starts_with(line, "\\ ")) {
2058                         /*
2059                          * Eat the "no newline at eof" marker as if we
2060                          * saw a "+" or "-" line with nothing on it,
2061                          * and return without diff_words_flush() to
2062                          * defer processing. If this is the end of
2063                          * preimage, more "+" lines may come after it.
2064                          */
2065                         return;
2066                 }
2067                 diff_words_flush(ecbdata);
2068                 emit_diff_symbol(o, s, line, len, 0);
2069                 return;
2070         }
2071
2072         switch (line[0]) {
2073         case '+':
2074                 ecbdata->lno_in_postimage++;
2075                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2076                 break;
2077         case '-':
2078                 ecbdata->lno_in_preimage++;
2079                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2080                 break;
2081         case ' ':
2082                 ecbdata->lno_in_postimage++;
2083                 ecbdata->lno_in_preimage++;
2084                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2085                 break;
2086         default:
2087                 /* incomplete line at the end */
2088                 ecbdata->lno_in_preimage++;
2089                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2090                                  line, len, 0);
2091                 break;
2092         }
2093 }
2094
2095 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2096 {
2097         const char *old_name = a;
2098         const char *new_name = b;
2099         int pfx_length, sfx_length;
2100         int pfx_adjust_for_slash;
2101         int len_a = strlen(a);
2102         int len_b = strlen(b);
2103         int a_midlen, b_midlen;
2104         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2105         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2106
2107         if (qlen_a || qlen_b) {
2108                 quote_c_style(a, name, NULL, 0);
2109                 strbuf_addstr(name, " => ");
2110                 quote_c_style(b, name, NULL, 0);
2111                 return;
2112         }
2113
2114         /* Find common prefix */
2115         pfx_length = 0;
2116         while (*old_name && *new_name && *old_name == *new_name) {
2117                 if (*old_name == '/')
2118                         pfx_length = old_name - a + 1;
2119                 old_name++;
2120                 new_name++;
2121         }
2122
2123         /* Find common suffix */
2124         old_name = a + len_a;
2125         new_name = b + len_b;
2126         sfx_length = 0;
2127         /*
2128          * If there is a common prefix, it must end in a slash.  In
2129          * that case we let this loop run 1 into the prefix to see the
2130          * same slash.
2131          *
2132          * If there is no common prefix, we cannot do this as it would
2133          * underrun the input strings.
2134          */
2135         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2136         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2137                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2138                *old_name == *new_name) {
2139                 if (*old_name == '/')
2140                         sfx_length = len_a - (old_name - a);
2141                 old_name--;
2142                 new_name--;
2143         }
2144
2145         /*
2146          * pfx{mid-a => mid-b}sfx
2147          * {pfx-a => pfx-b}sfx
2148          * pfx{sfx-a => sfx-b}
2149          * name-a => name-b
2150          */
2151         a_midlen = len_a - pfx_length - sfx_length;
2152         b_midlen = len_b - pfx_length - sfx_length;
2153         if (a_midlen < 0)
2154                 a_midlen = 0;
2155         if (b_midlen < 0)
2156                 b_midlen = 0;
2157
2158         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2159         if (pfx_length + sfx_length) {
2160                 strbuf_add(name, a, pfx_length);
2161                 strbuf_addch(name, '{');
2162         }
2163         strbuf_add(name, a + pfx_length, a_midlen);
2164         strbuf_addstr(name, " => ");
2165         strbuf_add(name, b + pfx_length, b_midlen);
2166         if (pfx_length + sfx_length) {
2167                 strbuf_addch(name, '}');
2168                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2169         }
2170 }
2171
2172 struct diffstat_t {
2173         int nr;
2174         int alloc;
2175         struct diffstat_file {
2176                 char *from_name;
2177                 char *name;
2178                 char *print_name;
2179                 const char *comments;
2180                 unsigned is_unmerged:1;
2181                 unsigned is_binary:1;
2182                 unsigned is_renamed:1;
2183                 unsigned is_interesting:1;
2184                 uintmax_t added, deleted;
2185         } **files;
2186 };
2187
2188 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2189                                           const char *name_a,
2190                                           const char *name_b)
2191 {
2192         struct diffstat_file *x;
2193         x = xcalloc(1, sizeof(*x));
2194         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2195         diffstat->files[diffstat->nr++] = x;
2196         if (name_b) {
2197                 x->from_name = xstrdup(name_a);
2198                 x->name = xstrdup(name_b);
2199                 x->is_renamed = 1;
2200         }
2201         else {
2202                 x->from_name = NULL;
2203                 x->name = xstrdup(name_a);
2204         }
2205         return x;
2206 }
2207
2208 static void diffstat_consume(void *priv, char *line, unsigned long len)
2209 {
2210         struct diffstat_t *diffstat = priv;
2211         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2212
2213         if (line[0] == '+')
2214                 x->added++;
2215         else if (line[0] == '-')
2216                 x->deleted++;
2217 }
2218
2219 const char mime_boundary_leader[] = "------------";
2220
2221 static int scale_linear(int it, int width, int max_change)
2222 {
2223         if (!it)
2224                 return 0;
2225         /*
2226          * make sure that at least one '-' or '+' is printed if
2227          * there is any change to this path. The easiest way is to
2228          * scale linearly as if the alloted width is one column shorter
2229          * than it is, and then add 1 to the result.
2230          */
2231         return 1 + (it * (width - 1) / max_change);
2232 }
2233
2234 static void show_graph(struct strbuf *out, char ch, int cnt,
2235                        const char *set, const char *reset)
2236 {
2237         if (cnt <= 0)
2238                 return;
2239         strbuf_addstr(out, set);
2240         strbuf_addchars(out, ch, cnt);
2241         strbuf_addstr(out, reset);
2242 }
2243
2244 static void fill_print_name(struct diffstat_file *file)
2245 {
2246         struct strbuf pname = STRBUF_INIT;
2247
2248         if (file->print_name)
2249                 return;
2250
2251         if (file->is_renamed)
2252                 pprint_rename(&pname, file->from_name, file->name);
2253         else
2254                 quote_c_style(file->name, &pname, NULL, 0);
2255
2256         if (file->comments)
2257                 strbuf_addf(&pname, " (%s)", file->comments);
2258
2259         file->print_name = strbuf_detach(&pname, NULL);
2260 }
2261
2262 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2263                 int files, int insertions, int deletions)
2264 {
2265         struct strbuf sb = STRBUF_INIT;
2266
2267         if (!files) {
2268                 assert(insertions == 0 && deletions == 0);
2269                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2270                                  NULL, 0, 0);
2271                 return;
2272         }
2273
2274         strbuf_addf(&sb,
2275                     (files == 1) ? " %d file changed" : " %d files changed",
2276                     files);
2277
2278         /*
2279          * For binary diff, the caller may want to print "x files
2280          * changed" with insertions == 0 && deletions == 0.
2281          *
2282          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2283          * is probably less confusing (i.e skip over "2 files changed
2284          * but nothing about added/removed lines? Is this a bug in Git?").
2285          */
2286         if (insertions || deletions == 0) {
2287                 strbuf_addf(&sb,
2288                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2289                             insertions);
2290         }
2291
2292         if (deletions || insertions == 0) {
2293                 strbuf_addf(&sb,
2294                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2295                             deletions);
2296         }
2297         strbuf_addch(&sb, '\n');
2298         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2299                          sb.buf, sb.len, 0);
2300         strbuf_release(&sb);
2301 }
2302
2303 void print_stat_summary(FILE *fp, int files,
2304                         int insertions, int deletions)
2305 {
2306         struct diff_options o;
2307         memset(&o, 0, sizeof(o));
2308         o.file = fp;
2309
2310         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2311 }
2312
2313 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2314 {
2315         int i, len, add, del, adds = 0, dels = 0;
2316         uintmax_t max_change = 0, max_len = 0;
2317         int total_files = data->nr, count;
2318         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2319         const char *reset, *add_c, *del_c;
2320         int extra_shown = 0;
2321         const char *line_prefix = diff_line_prefix(options);
2322         struct strbuf out = STRBUF_INIT;
2323
2324         if (data->nr == 0)
2325                 return;
2326
2327         count = options->stat_count ? options->stat_count : data->nr;
2328
2329         reset = diff_get_color_opt(options, DIFF_RESET);
2330         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2331         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2332
2333         /*
2334          * Find the longest filename and max number of changes
2335          */
2336         for (i = 0; (i < count) && (i < data->nr); i++) {
2337                 struct diffstat_file *file = data->files[i];
2338                 uintmax_t change = file->added + file->deleted;
2339
2340                 if (!file->is_interesting && (change == 0)) {
2341                         count++; /* not shown == room for one more */
2342                         continue;
2343                 }
2344                 fill_print_name(file);
2345                 len = strlen(file->print_name);
2346                 if (max_len < len)
2347                         max_len = len;
2348
2349                 if (file->is_unmerged) {
2350                         /* "Unmerged" is 8 characters */
2351                         bin_width = bin_width < 8 ? 8 : bin_width;
2352                         continue;
2353                 }
2354                 if (file->is_binary) {
2355                         /* "Bin XXX -> YYY bytes" */
2356                         int w = 14 + decimal_width(file->added)
2357                                 + decimal_width(file->deleted);
2358                         bin_width = bin_width < w ? w : bin_width;
2359                         /* Display change counts aligned with "Bin" */
2360                         number_width = 3;
2361                         continue;
2362                 }
2363
2364                 if (max_change < change)
2365                         max_change = change;
2366         }
2367         count = i; /* where we can stop scanning in data->files[] */
2368
2369         /*
2370          * We have width = stat_width or term_columns() columns total.
2371          * We want a maximum of min(max_len, stat_name_width) for the name part.
2372          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2373          * We also need 1 for " " and 4 + decimal_width(max_change)
2374          * for " | NNNN " and one the empty column at the end, altogether
2375          * 6 + decimal_width(max_change).
2376          *
2377          * If there's not enough space, we will use the smaller of
2378          * stat_name_width (if set) and 5/8*width for the filename,
2379          * and the rest for constant elements + graph part, but no more
2380          * than stat_graph_width for the graph part.
2381          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2382          * for the standard terminal size).
2383          *
2384          * In other words: stat_width limits the maximum width, and
2385          * stat_name_width fixes the maximum width of the filename,
2386          * and is also used to divide available columns if there
2387          * aren't enough.
2388          *
2389          * Binary files are displayed with "Bin XXX -> YYY bytes"
2390          * instead of the change count and graph. This part is treated
2391          * similarly to the graph part, except that it is not
2392          * "scaled". If total width is too small to accommodate the
2393          * guaranteed minimum width of the filename part and the
2394          * separators and this message, this message will "overflow"
2395          * making the line longer than the maximum width.
2396          */
2397
2398         if (options->stat_width == -1)
2399                 width = term_columns() - strlen(line_prefix);
2400         else
2401                 width = options->stat_width ? options->stat_width : 80;
2402         number_width = decimal_width(max_change) > number_width ?
2403                 decimal_width(max_change) : number_width;
2404
2405         if (options->stat_graph_width == -1)
2406                 options->stat_graph_width = diff_stat_graph_width;
2407
2408         /*
2409          * Guarantee 3/8*16==6 for the graph part
2410          * and 5/8*16==10 for the filename part
2411          */
2412         if (width < 16 + 6 + number_width)
2413                 width = 16 + 6 + number_width;
2414
2415         /*
2416          * First assign sizes that are wanted, ignoring available width.
2417          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2418          * starting from "XXX" should fit in graph_width.
2419          */
2420         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2421         if (options->stat_graph_width &&
2422             options->stat_graph_width < graph_width)
2423                 graph_width = options->stat_graph_width;
2424
2425         name_width = (options->stat_name_width > 0 &&
2426                       options->stat_name_width < max_len) ?
2427                 options->stat_name_width : max_len;
2428
2429         /*
2430          * Adjust adjustable widths not to exceed maximum width
2431          */
2432         if (name_width + number_width + 6 + graph_width > width) {
2433                 if (graph_width > width * 3/8 - number_width - 6) {
2434                         graph_width = width * 3/8 - number_width - 6;
2435                         if (graph_width < 6)
2436                                 graph_width = 6;
2437                 }
2438
2439                 if (options->stat_graph_width &&
2440                     graph_width > options->stat_graph_width)
2441                         graph_width = options->stat_graph_width;
2442                 if (name_width > width - number_width - 6 - graph_width)
2443                         name_width = width - number_width - 6 - graph_width;
2444                 else
2445                         graph_width = width - number_width - 6 - name_width;
2446         }
2447
2448         /*
2449          * From here name_width is the width of the name area,
2450          * and graph_width is the width of the graph area.
2451          * max_change is used to scale graph properly.
2452          */
2453         for (i = 0; i < count; i++) {
2454                 const char *prefix = "";
2455                 struct diffstat_file *file = data->files[i];
2456                 char *name = file->print_name;
2457                 uintmax_t added = file->added;
2458                 uintmax_t deleted = file->deleted;
2459                 int name_len;
2460
2461                 if (!file->is_interesting && (added + deleted == 0))
2462                         continue;
2463
2464                 /*
2465                  * "scale" the filename
2466                  */
2467                 len = name_width;
2468                 name_len = strlen(name);
2469                 if (name_width < name_len) {
2470                         char *slash;
2471                         prefix = "...";
2472                         len -= 3;
2473                         name += name_len - len;
2474                         slash = strchr(name, '/');
2475                         if (slash)
2476                                 name = slash;
2477                 }
2478
2479                 if (file->is_binary) {
2480                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2481                         strbuf_addf(&out, " %*s", number_width, "Bin");
2482                         if (!added && !deleted) {
2483                                 strbuf_addch(&out, '\n');
2484                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2485                                                  out.buf, out.len, 0);
2486                                 strbuf_reset(&out);
2487                                 continue;
2488                         }
2489                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2490                                 del_c, deleted, reset);
2491                         strbuf_addstr(&out, " -> ");
2492                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2493                                 add_c, added, reset);
2494                         strbuf_addstr(&out, " bytes\n");
2495                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2496                                          out.buf, out.len, 0);
2497                         strbuf_reset(&out);
2498                         continue;
2499                 }
2500                 else if (file->is_unmerged) {
2501                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2502                         strbuf_addstr(&out, " Unmerged\n");
2503                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2504                                          out.buf, out.len, 0);
2505                         strbuf_reset(&out);
2506                         continue;
2507                 }
2508
2509                 /*
2510                  * scale the add/delete
2511                  */
2512                 add = added;
2513                 del = deleted;
2514
2515                 if (graph_width <= max_change) {
2516                         int total = scale_linear(add + del, graph_width, max_change);
2517                         if (total < 2 && add && del)
2518                                 /* width >= 2 due to the sanity check */
2519                                 total = 2;
2520                         if (add < del) {
2521                                 add = scale_linear(add, graph_width, max_change);
2522                                 del = total - add;
2523                         } else {
2524                                 del = scale_linear(del, graph_width, max_change);
2525                                 add = total - del;
2526                         }
2527                 }
2528                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2529                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2530                         number_width, added + deleted,
2531                         added + deleted ? " " : "");
2532                 show_graph(&out, '+', add, add_c, reset);
2533                 show_graph(&out, '-', del, del_c, reset);
2534                 strbuf_addch(&out, '\n');
2535                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2536                                  out.buf, out.len, 0);
2537                 strbuf_reset(&out);
2538         }
2539
2540         for (i = 0; i < data->nr; i++) {
2541                 struct diffstat_file *file = data->files[i];
2542                 uintmax_t added = file->added;
2543                 uintmax_t deleted = file->deleted;
2544
2545                 if (file->is_unmerged ||
2546                     (!file->is_interesting && (added + deleted == 0))) {
2547                         total_files--;
2548                         continue;
2549                 }
2550
2551                 if (!file->is_binary) {
2552                         adds += added;
2553                         dels += deleted;
2554                 }
2555                 if (i < count)
2556                         continue;
2557                 if (!extra_shown)
2558                         emit_diff_symbol(options,
2559                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2560                                          NULL, 0, 0);
2561                 extra_shown = 1;
2562         }
2563
2564         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2565         strbuf_release(&out);
2566 }
2567
2568 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2569 {
2570         int i, adds = 0, dels = 0, total_files = data->nr;
2571
2572         if (data->nr == 0)
2573                 return;
2574
2575         for (i = 0; i < data->nr; i++) {
2576                 int added = data->files[i]->added;
2577                 int deleted = data->files[i]->deleted;
2578
2579                 if (data->files[i]->is_unmerged ||
2580                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2581                         total_files--;
2582                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2583                         adds += added;
2584                         dels += deleted;
2585                 }
2586         }
2587         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2588 }
2589
2590 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2591 {
2592         int i;
2593
2594         if (data->nr == 0)
2595                 return;
2596
2597         for (i = 0; i < data->nr; i++) {
2598                 struct diffstat_file *file = data->files[i];
2599
2600                 fprintf(options->file, "%s", diff_line_prefix(options));
2601
2602                 if (file->is_binary)
2603                         fprintf(options->file, "-\t-\t");
2604                 else
2605                         fprintf(options->file,
2606                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2607                                 file->added, file->deleted);
2608                 if (options->line_termination) {
2609                         fill_print_name(file);
2610                         if (!file->is_renamed)
2611                                 write_name_quoted(file->name, options->file,
2612                                                   options->line_termination);
2613                         else {
2614                                 fputs(file->print_name, options->file);
2615                                 putc(options->line_termination, options->file);
2616                         }
2617                 } else {
2618                         if (file->is_renamed) {
2619                                 putc('\0', options->file);
2620                                 write_name_quoted(file->from_name, options->file, '\0');
2621                         }
2622                         write_name_quoted(file->name, options->file, '\0');
2623                 }
2624         }
2625 }
2626
2627 struct dirstat_file {
2628         const char *name;
2629         unsigned long changed;
2630 };
2631
2632 struct dirstat_dir {
2633         struct dirstat_file *files;
2634         int alloc, nr, permille, cumulative;
2635 };
2636
2637 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2638                 unsigned long changed, const char *base, int baselen)
2639 {
2640         unsigned long sum_changes = 0;
2641         unsigned int sources = 0;
2642         const char *line_prefix = diff_line_prefix(opt);
2643
2644         while (dir->nr) {
2645                 struct dirstat_file *f = dir->files;
2646                 int namelen = strlen(f->name);
2647                 unsigned long changes;
2648                 char *slash;
2649
2650                 if (namelen < baselen)
2651                         break;
2652                 if (memcmp(f->name, base, baselen))
2653                         break;
2654                 slash = strchr(f->name + baselen, '/');
2655                 if (slash) {
2656                         int newbaselen = slash + 1 - f->name;
2657                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2658                         sources++;
2659                 } else {
2660                         changes = f->changed;
2661                         dir->files++;
2662                         dir->nr--;
2663                         sources += 2;
2664                 }
2665                 sum_changes += changes;
2666         }
2667
2668         /*
2669          * We don't report dirstat's for
2670          *  - the top level
2671          *  - or cases where everything came from a single directory
2672          *    under this directory (sources == 1).
2673          */
2674         if (baselen && sources != 1) {
2675                 if (sum_changes) {
2676                         int permille = sum_changes * 1000 / changed;
2677                         if (permille >= dir->permille) {
2678                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2679                                         permille / 10, permille % 10, baselen, base);
2680                                 if (!dir->cumulative)
2681                                         return 0;
2682                         }
2683                 }
2684         }
2685         return sum_changes;
2686 }
2687
2688 static int dirstat_compare(const void *_a, const void *_b)
2689 {
2690         const struct dirstat_file *a = _a;
2691         const struct dirstat_file *b = _b;
2692         return strcmp(a->name, b->name);
2693 }
2694
2695 static void show_dirstat(struct diff_options *options)
2696 {
2697         int i;
2698         unsigned long changed;
2699         struct dirstat_dir dir;
2700         struct diff_queue_struct *q = &diff_queued_diff;
2701
2702         dir.files = NULL;
2703         dir.alloc = 0;
2704         dir.nr = 0;
2705         dir.permille = options->dirstat_permille;
2706         dir.cumulative = options->flags.dirstat_cumulative;
2707
2708         changed = 0;
2709         for (i = 0; i < q->nr; i++) {
2710                 struct diff_filepair *p = q->queue[i];
2711                 const char *name;
2712                 unsigned long copied, added, damage;
2713                 int content_changed;
2714
2715                 name = p->two->path ? p->two->path : p->one->path;
2716
2717                 if (p->one->oid_valid && p->two->oid_valid)
2718                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2719                 else
2720                         content_changed = 1;
2721
2722                 if (!content_changed) {
2723                         /*
2724                          * The SHA1 has not changed, so pre-/post-content is
2725                          * identical. We can therefore skip looking at the
2726                          * file contents altogether.
2727                          */
2728                         damage = 0;
2729                         goto found_damage;
2730                 }
2731
2732                 if (options->flags.dirstat_by_file) {
2733                         /*
2734                          * In --dirstat-by-file mode, we don't really need to
2735                          * look at the actual file contents at all.
2736                          * The fact that the SHA1 changed is enough for us to
2737                          * add this file to the list of results
2738                          * (with each file contributing equal damage).
2739                          */
2740                         damage = 1;
2741                         goto found_damage;
2742                 }
2743
2744                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2745                         diff_populate_filespec(p->one, 0);
2746                         diff_populate_filespec(p->two, 0);
2747                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2748                                                &copied, &added);
2749                         diff_free_filespec_data(p->one);
2750                         diff_free_filespec_data(p->two);
2751                 } else if (DIFF_FILE_VALID(p->one)) {
2752                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2753                         copied = added = 0;
2754                         diff_free_filespec_data(p->one);
2755                 } else if (DIFF_FILE_VALID(p->two)) {
2756                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2757                         copied = 0;
2758                         added = p->two->size;
2759                         diff_free_filespec_data(p->two);
2760                 } else
2761                         continue;
2762
2763                 /*
2764                  * Original minus copied is the removed material,
2765                  * added is the new material.  They are both damages
2766                  * made to the preimage.
2767                  * If the resulting damage is zero, we know that
2768                  * diffcore_count_changes() considers the two entries to
2769                  * be identical, but since content_changed is true, we
2770                  * know that there must have been _some_ kind of change,
2771                  * so we force all entries to have damage > 0.
2772                  */
2773                 damage = (p->one->size - copied) + added;
2774                 if (!damage)
2775                         damage = 1;
2776
2777 found_damage:
2778                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2779                 dir.files[dir.nr].name = name;
2780                 dir.files[dir.nr].changed = damage;
2781                 changed += damage;
2782                 dir.nr++;
2783         }
2784
2785         /* This can happen even with many files, if everything was renames */
2786         if (!changed)
2787                 return;
2788
2789         /* Show all directories with more than x% of the changes */
2790         QSORT(dir.files, dir.nr, dirstat_compare);
2791         gather_dirstat(options, &dir, changed, "", 0);
2792 }
2793
2794 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2795 {
2796         int i;
2797         unsigned long changed;
2798         struct dirstat_dir dir;
2799
2800         if (data->nr == 0)
2801                 return;
2802
2803         dir.files = NULL;
2804         dir.alloc = 0;
2805         dir.nr = 0;
2806         dir.permille = options->dirstat_permille;
2807         dir.cumulative = options->flags.dirstat_cumulative;
2808
2809         changed = 0;
2810         for (i = 0; i < data->nr; i++) {
2811                 struct diffstat_file *file = data->files[i];
2812                 unsigned long damage = file->added + file->deleted;
2813                 if (file->is_binary)
2814                         /*
2815                          * binary files counts bytes, not lines. Must find some
2816                          * way to normalize binary bytes vs. textual lines.
2817                          * The following heuristic assumes that there are 64
2818                          * bytes per "line".
2819                          * This is stupid and ugly, but very cheap...
2820                          */
2821                         damage = DIV_ROUND_UP(damage, 64);
2822                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2823                 dir.files[dir.nr].name = file->name;
2824                 dir.files[dir.nr].changed = damage;
2825                 changed += damage;
2826                 dir.nr++;
2827         }
2828
2829         /* This can happen even with many files, if everything was renames */
2830         if (!changed)
2831                 return;
2832
2833         /* Show all directories with more than x% of the changes */
2834         QSORT(dir.files, dir.nr, dirstat_compare);
2835         gather_dirstat(options, &dir, changed, "", 0);
2836 }
2837
2838 static void free_diffstat_info(struct diffstat_t *diffstat)
2839 {
2840         int i;
2841         for (i = 0; i < diffstat->nr; i++) {
2842                 struct diffstat_file *f = diffstat->files[i];
2843                 free(f->print_name);
2844                 free(f->name);
2845                 free(f->from_name);
2846                 free(f);
2847         }
2848         free(diffstat->files);
2849 }
2850
2851 struct checkdiff_t {
2852         const char *filename;
2853         int lineno;
2854         int conflict_marker_size;
2855         struct diff_options *o;
2856         unsigned ws_rule;
2857         unsigned status;
2858 };
2859
2860 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2861 {
2862         char firstchar;
2863         int cnt;
2864
2865         if (len < marker_size + 1)
2866                 return 0;
2867         firstchar = line[0];
2868         switch (firstchar) {
2869         case '=': case '>': case '<': case '|':
2870                 break;
2871         default:
2872                 return 0;
2873         }
2874         for (cnt = 1; cnt < marker_size; cnt++)
2875                 if (line[cnt] != firstchar)
2876                         return 0;
2877         /* line[1] thru line[marker_size-1] are same as firstchar */
2878         if (len < marker_size + 1 || !isspace(line[marker_size]))
2879                 return 0;
2880         return 1;
2881 }
2882
2883 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2884 {
2885         struct checkdiff_t *data = priv;
2886         int marker_size = data->conflict_marker_size;
2887         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2888         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2889         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2890         char *err;
2891         const char *line_prefix;
2892
2893         assert(data->o);
2894         line_prefix = diff_line_prefix(data->o);
2895
2896         if (line[0] == '+') {
2897                 unsigned bad;
2898                 data->lineno++;
2899                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2900                         data->status |= 1;
2901                         fprintf(data->o->file,
2902                                 "%s%s:%d: leftover conflict marker\n",
2903                                 line_prefix, data->filename, data->lineno);
2904                 }
2905                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2906                 if (!bad)
2907                         return;
2908                 data->status |= bad;
2909                 err = whitespace_error_string(bad);
2910                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2911                         line_prefix, data->filename, data->lineno, err);
2912                 free(err);
2913                 emit_line(data->o, set, reset, line, 1);
2914                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2915                               data->o->file, set, reset, ws);
2916         } else if (line[0] == ' ') {
2917                 data->lineno++;
2918         } else if (line[0] == '@') {
2919                 char *plus = strchr(line, '+');
2920                 if (plus)
2921                         data->lineno = strtol(plus, NULL, 10) - 1;
2922                 else
2923                         die("invalid diff");
2924         }
2925 }
2926
2927 static unsigned char *deflate_it(char *data,
2928                                  unsigned long size,
2929                                  unsigned long *result_size)
2930 {
2931         int bound;
2932         unsigned char *deflated;
2933         git_zstream stream;
2934
2935         git_deflate_init(&stream, zlib_compression_level);
2936         bound = git_deflate_bound(&stream, size);
2937         deflated = xmalloc(bound);
2938         stream.next_out = deflated;
2939         stream.avail_out = bound;
2940
2941         stream.next_in = (unsigned char *)data;
2942         stream.avail_in = size;
2943         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2944                 ; /* nothing */
2945         git_deflate_end(&stream);
2946         *result_size = stream.total_out;
2947         return deflated;
2948 }
2949
2950 static void emit_binary_diff_body(struct diff_options *o,
2951                                   mmfile_t *one, mmfile_t *two)
2952 {
2953         void *cp;
2954         void *delta;
2955         void *deflated;
2956         void *data;
2957         unsigned long orig_size;
2958         unsigned long delta_size;
2959         unsigned long deflate_size;
2960         unsigned long data_size;
2961
2962         /* We could do deflated delta, or we could do just deflated two,
2963          * whichever is smaller.
2964          */
2965         delta = NULL;
2966         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2967         if (one->size && two->size) {
2968                 delta = diff_delta(one->ptr, one->size,
2969                                    two->ptr, two->size,
2970                                    &delta_size, deflate_size);
2971                 if (delta) {
2972                         void *to_free = delta;
2973                         orig_size = delta_size;
2974                         delta = deflate_it(delta, delta_size, &delta_size);
2975                         free(to_free);
2976                 }
2977         }
2978
2979         if (delta && delta_size < deflate_size) {
2980                 char *s = xstrfmt("%lu", orig_size);
2981                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2982                                  s, strlen(s), 0);
2983                 free(s);
2984                 free(deflated);
2985                 data = delta;
2986                 data_size = delta_size;
2987         } else {
2988                 char *s = xstrfmt("%lu", two->size);
2989                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2990                                  s, strlen(s), 0);
2991                 free(s);
2992                 free(delta);
2993                 data = deflated;
2994                 data_size = deflate_size;
2995         }
2996
2997         /* emit data encoded in base85 */
2998         cp = data;
2999         while (data_size) {
3000                 int len;
3001                 int bytes = (52 < data_size) ? 52 : data_size;
3002                 char line[71];
3003                 data_size -= bytes;
3004                 if (bytes <= 26)
3005                         line[0] = bytes + 'A' - 1;
3006                 else
3007                         line[0] = bytes - 26 + 'a' - 1;
3008                 encode_85(line + 1, cp, bytes);
3009                 cp = (char *) cp + bytes;
3010
3011                 len = strlen(line);
3012                 line[len++] = '\n';
3013                 line[len] = '\0';
3014
3015                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3016                                  line, len, 0);
3017         }
3018         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3019         free(data);
3020 }
3021
3022 static void emit_binary_diff(struct diff_options *o,
3023                              mmfile_t *one, mmfile_t *two)
3024 {
3025         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3026         emit_binary_diff_body(o, one, two);
3027         emit_binary_diff_body(o, two, one);
3028 }
3029
3030 int diff_filespec_is_binary(struct diff_filespec *one)
3031 {
3032         if (one->is_binary == -1) {
3033                 diff_filespec_load_driver(one);
3034                 if (one->driver->binary != -1)
3035                         one->is_binary = one->driver->binary;
3036                 else {
3037                         if (!one->data && DIFF_FILE_VALID(one))
3038                                 diff_populate_filespec(one, CHECK_BINARY);
3039                         if (one->is_binary == -1 && one->data)
3040                                 one->is_binary = buffer_is_binary(one->data,
3041                                                 one->size);
3042                         if (one->is_binary == -1)
3043                                 one->is_binary = 0;
3044                 }
3045         }
3046         return one->is_binary;
3047 }
3048
3049 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3050 {
3051         diff_filespec_load_driver(one);
3052         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3053 }
3054
3055 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3056 {
3057         if (!options->a_prefix)
3058                 options->a_prefix = a;
3059         if (!options->b_prefix)
3060                 options->b_prefix = b;
3061 }
3062
3063 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3064 {
3065         if (!DIFF_FILE_VALID(one))
3066                 return NULL;
3067
3068         diff_filespec_load_driver(one);
3069         return userdiff_get_textconv(one->driver);
3070 }
3071
3072 static void builtin_diff(const char *name_a,
3073                          const char *name_b,
3074                          struct diff_filespec *one,
3075                          struct diff_filespec *two,
3076                          const char *xfrm_msg,
3077                          int must_show_header,
3078                          struct diff_options *o,
3079                          int complete_rewrite)
3080 {
3081         mmfile_t mf1, mf2;
3082         const char *lbl[2];
3083         char *a_one, *b_two;
3084         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3085         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3086         const char *a_prefix, *b_prefix;
3087         struct userdiff_driver *textconv_one = NULL;
3088         struct userdiff_driver *textconv_two = NULL;
3089         struct strbuf header = STRBUF_INIT;
3090         const char *line_prefix = diff_line_prefix(o);
3091
3092         diff_set_mnemonic_prefix(o, "a/", "b/");
3093         if (o->flags.reverse_diff) {
3094                 a_prefix = o->b_prefix;
3095                 b_prefix = o->a_prefix;
3096         } else {
3097                 a_prefix = o->a_prefix;
3098                 b_prefix = o->b_prefix;
3099         }
3100
3101         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3102             (!one->mode || S_ISGITLINK(one->mode)) &&
3103             (!two->mode || S_ISGITLINK(two->mode))) {
3104                 show_submodule_summary(o, one->path ? one->path : two->path,
3105                                 &one->oid, &two->oid,
3106                                 two->dirty_submodule);
3107                 return;
3108         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3109                    (!one->mode || S_ISGITLINK(one->mode)) &&
3110                    (!two->mode || S_ISGITLINK(two->mode))) {
3111                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3112                                 &one->oid, &two->oid,
3113                                 two->dirty_submodule);
3114                 return;
3115         }
3116
3117         if (o->flags.allow_textconv) {
3118                 textconv_one = get_textconv(one);
3119                 textconv_two = get_textconv(two);
3120         }
3121
3122         /* Never use a non-valid filename anywhere if at all possible */
3123         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3124         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3125
3126         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3127         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3128         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3129         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3130         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3131         if (lbl[0][0] == '/') {
3132                 /* /dev/null */
3133                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3134                 if (xfrm_msg)
3135                         strbuf_addstr(&header, xfrm_msg);
3136                 must_show_header = 1;
3137         }
3138         else if (lbl[1][0] == '/') {
3139                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3140                 if (xfrm_msg)
3141                         strbuf_addstr(&header, xfrm_msg);
3142                 must_show_header = 1;
3143         }
3144         else {
3145                 if (one->mode != two->mode) {
3146                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3147                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3148                         must_show_header = 1;
3149                 }
3150                 if (xfrm_msg)
3151                         strbuf_addstr(&header, xfrm_msg);
3152
3153                 /*
3154                  * we do not run diff between different kind
3155                  * of objects.
3156                  */
3157                 if ((one->mode ^ two->mode) & S_IFMT)
3158                         goto free_ab_and_return;
3159                 if (complete_rewrite &&
3160                     (textconv_one || !diff_filespec_is_binary(one)) &&
3161                     (textconv_two || !diff_filespec_is_binary(two))) {
3162                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3163                                          header.buf, header.len, 0);
3164                         strbuf_reset(&header);
3165                         emit_rewrite_diff(name_a, name_b, one, two,
3166                                                 textconv_one, textconv_two, o);
3167                         o->found_changes = 1;
3168                         goto free_ab_and_return;
3169                 }
3170         }
3171
3172         if (o->irreversible_delete && lbl[1][0] == '/') {
3173                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3174                                  header.len, 0);
3175                 strbuf_reset(&header);
3176                 goto free_ab_and_return;
3177         } else if (!o->flags.text &&
3178             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3179               (!textconv_two && diff_filespec_is_binary(two)) )) {
3180                 struct strbuf sb = STRBUF_INIT;
3181                 if (!one->data && !two->data &&
3182                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3183                     !o->flags.binary) {
3184                         if (!oidcmp(&one->oid, &two->oid)) {
3185                                 if (must_show_header)
3186                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3187                                                          header.buf, header.len,
3188                                                          0);
3189                                 goto free_ab_and_return;
3190                         }
3191                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3192                                          header.buf, header.len, 0);
3193                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3194                                     diff_line_prefix(o), lbl[0], lbl[1]);
3195                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3196                                          sb.buf, sb.len, 0);
3197                         strbuf_release(&sb);
3198                         goto free_ab_and_return;
3199                 }
3200                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3201                         die("unable to read files to diff");
3202                 /* Quite common confusing case */
3203                 if (mf1.size == mf2.size &&
3204                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3205                         if (must_show_header)
3206                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3207                                                  header.buf, header.len, 0);
3208                         goto free_ab_and_return;
3209                 }
3210                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3211                 strbuf_reset(&header);
3212                 if (o->flags.binary)
3213                         emit_binary_diff(o, &mf1, &mf2);
3214                 else {
3215                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3216                                     diff_line_prefix(o), lbl[0], lbl[1]);
3217                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3218                                          sb.buf, sb.len, 0);
3219                         strbuf_release(&sb);
3220                 }
3221                 o->found_changes = 1;
3222         } else {
3223                 /* Crazy xdl interfaces.. */
3224                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3225                 const char *v;
3226                 xpparam_t xpp;
3227                 xdemitconf_t xecfg;
3228                 struct emit_callback ecbdata;
3229                 const struct userdiff_funcname *pe;
3230
3231                 if (must_show_header) {
3232                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3233                                          header.buf, header.len, 0);
3234                         strbuf_reset(&header);
3235                 }
3236
3237                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3238                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3239
3240                 pe = diff_funcname_pattern(one);
3241                 if (!pe)
3242                         pe = diff_funcname_pattern(two);
3243
3244                 memset(&xpp, 0, sizeof(xpp));
3245                 memset(&xecfg, 0, sizeof(xecfg));
3246                 memset(&ecbdata, 0, sizeof(ecbdata));
3247                 ecbdata.label_path = lbl;
3248                 ecbdata.color_diff = want_color(o->use_color);
3249                 ecbdata.ws_rule = whitespace_rule(name_b);
3250                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3251                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3252                 ecbdata.opt = o;
3253                 ecbdata.header = header.len ? &header : NULL;
3254                 xpp.flags = o->xdl_opts;
3255                 xpp.anchors = o->anchors;
3256                 xpp.anchors_nr = o->anchors_nr;
3257                 xecfg.ctxlen = o->context;
3258                 xecfg.interhunkctxlen = o->interhunkcontext;
3259                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3260                 if (o->flags.funccontext)
3261                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3262                 if (pe)
3263                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3264                 if (!diffopts)
3265                         ;
3266                 else if (skip_prefix(diffopts, "--unified=", &v))
3267                         xecfg.ctxlen = strtoul(v, NULL, 10);
3268                 else if (skip_prefix(diffopts, "-u", &v))
3269                         xecfg.ctxlen = strtoul(v, NULL, 10);
3270                 if (o->word_diff)
3271                         init_diff_words_data(&ecbdata, o, one, two);
3272                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3273                                   &xpp, &xecfg))
3274                         die("unable to generate diff for %s", one->path);
3275                 if (o->word_diff)
3276                         free_diff_words_data(&ecbdata);
3277                 if (textconv_one)
3278                         free(mf1.ptr);
3279                 if (textconv_two)
3280                         free(mf2.ptr);
3281                 xdiff_clear_find_func(&xecfg);
3282         }
3283
3284  free_ab_and_return:
3285         strbuf_release(&header);
3286         diff_free_filespec_data(one);
3287         diff_free_filespec_data(two);
3288         free(a_one);
3289         free(b_two);
3290         return;
3291 }
3292
3293 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3294 {
3295         if (!is_renamed) {
3296                 if (p->status == DIFF_STATUS_ADDED) {
3297                         if (S_ISLNK(p->two->mode))
3298                                 return "new +l";
3299                         else if ((p->two->mode & 0777) == 0755)
3300                                 return "new +x";
3301                         else
3302                                 return "new";
3303                 } else if (p->status == DIFF_STATUS_DELETED)
3304                         return "gone";
3305         }
3306         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3307                 return "mode -l";
3308         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3309                 return "mode +l";
3310         else if ((p->one->mode & 0777) == 0644 &&
3311                  (p->two->mode & 0777) == 0755)
3312                 return "mode +x";
3313         else if ((p->one->mode & 0777) == 0755 &&
3314                  (p->two->mode & 0777) == 0644)
3315                 return "mode -x";
3316         return NULL;
3317 }
3318
3319 static void builtin_diffstat(const char *name_a, const char *name_b,
3320                              struct diff_filespec *one,
3321                              struct diff_filespec *two,
3322                              struct diffstat_t *diffstat,
3323                              struct diff_options *o,
3324                              struct diff_filepair *p)
3325 {
3326         mmfile_t mf1, mf2;
3327         struct diffstat_file *data;
3328         int same_contents;
3329         int complete_rewrite = 0;
3330
3331         if (!DIFF_PAIR_UNMERGED(p)) {
3332                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3333                         complete_rewrite = 1;
3334         }
3335
3336         data = diffstat_add(diffstat, name_a, name_b);
3337         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3338         if (o->flags.stat_with_summary)
3339                 data->comments = get_compact_summary(p, data->is_renamed);
3340
3341         if (!one || !two) {
3342                 data->is_unmerged = 1;
3343                 return;
3344         }
3345
3346         same_contents = !oidcmp(&one->oid, &two->oid);
3347
3348         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3349                 data->is_binary = 1;
3350                 if (same_contents) {
3351                         data->added = 0;
3352                         data->deleted = 0;
3353                 } else {
3354                         data->added = diff_filespec_size(two);
3355                         data->deleted = diff_filespec_size(one);
3356                 }
3357         }
3358
3359         else if (complete_rewrite) {
3360                 diff_populate_filespec(one, 0);
3361                 diff_populate_filespec(two, 0);
3362                 data->deleted = count_lines(one->data, one->size);
3363                 data->added = count_lines(two->data, two->size);
3364         }
3365
3366         else if (!same_contents) {
3367                 /* Crazy xdl interfaces.. */
3368                 xpparam_t xpp;
3369                 xdemitconf_t xecfg;
3370
3371                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3372                         die("unable to read files to diff");
3373
3374                 memset(&xpp, 0, sizeof(xpp));
3375                 memset(&xecfg, 0, sizeof(xecfg));
3376                 xpp.flags = o->xdl_opts;
3377                 xpp.anchors = o->anchors;
3378                 xpp.anchors_nr = o->anchors_nr;
3379                 xecfg.ctxlen = o->context;
3380                 xecfg.interhunkctxlen = o->interhunkcontext;
3381                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3382                                   &xpp, &xecfg))
3383                         die("unable to generate diffstat for %s", one->path);
3384         }
3385
3386         diff_free_filespec_data(one);
3387         diff_free_filespec_data(two);
3388 }
3389
3390 static void builtin_checkdiff(const char *name_a, const char *name_b,
3391                               const char *attr_path,
3392                               struct diff_filespec *one,
3393                               struct diff_filespec *two,
3394                               struct diff_options *o)
3395 {
3396         mmfile_t mf1, mf2;
3397         struct checkdiff_t data;
3398
3399         if (!two)
3400                 return;
3401
3402         memset(&data, 0, sizeof(data));
3403         data.filename = name_b ? name_b : name_a;
3404         data.lineno = 0;
3405         data.o = o;
3406         data.ws_rule = whitespace_rule(attr_path);
3407         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3408
3409         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3410                 die("unable to read files to diff");
3411
3412         /*
3413          * All the other codepaths check both sides, but not checking
3414          * the "old" side here is deliberate.  We are checking the newly
3415          * introduced changes, and as long as the "new" side is text, we
3416          * can and should check what it introduces.
3417          */
3418         if (diff_filespec_is_binary(two))
3419                 goto free_and_return;
3420         else {
3421                 /* Crazy xdl interfaces.. */
3422                 xpparam_t xpp;
3423                 xdemitconf_t xecfg;
3424
3425                 memset(&xpp, 0, sizeof(xpp));
3426                 memset(&xecfg, 0, sizeof(xecfg));
3427                 xecfg.ctxlen = 1; /* at least one context line */
3428                 xpp.flags = 0;
3429                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3430                                   &xpp, &xecfg))
3431                         die("unable to generate checkdiff for %s", one->path);
3432
3433                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3434                         struct emit_callback ecbdata;
3435                         int blank_at_eof;
3436
3437                         ecbdata.ws_rule = data.ws_rule;
3438                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3439                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3440
3441                         if (blank_at_eof) {
3442                                 static char *err;
3443                                 if (!err)
3444                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3445                                 fprintf(o->file, "%s:%d: %s.\n",
3446                                         data.filename, blank_at_eof, err);
3447                                 data.status = 1; /* report errors */
3448                         }
3449                 }
3450         }
3451  free_and_return:
3452         diff_free_filespec_data(one);
3453         diff_free_filespec_data(two);
3454         if (data.status)
3455                 o->flags.check_failed = 1;
3456 }
3457
3458 struct diff_filespec *alloc_filespec(const char *path)
3459 {
3460         struct diff_filespec *spec;
3461
3462         FLEXPTR_ALLOC_STR(spec, path, path);
3463         spec->count = 1;
3464         spec->is_binary = -1;
3465         return spec;
3466 }
3467
3468 void free_filespec(struct diff_filespec *spec)
3469 {
3470         if (!--spec->count) {
3471                 diff_free_filespec_data(spec);
3472                 free(spec);
3473         }
3474 }
3475
3476 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3477                    int oid_valid, unsigned short mode)
3478 {
3479         if (mode) {
3480                 spec->mode = canon_mode(mode);
3481                 oidcpy(&spec->oid, oid);
3482                 spec->oid_valid = oid_valid;
3483         }
3484 }
3485
3486 /*
3487  * Given a name and sha1 pair, if the index tells us the file in
3488  * the work tree has that object contents, return true, so that
3489  * prepare_temp_file() does not have to inflate and extract.
3490  */
3491 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3492 {
3493         const struct cache_entry *ce;
3494         struct stat st;
3495         int pos, len;
3496
3497         /*
3498          * We do not read the cache ourselves here, because the
3499          * benchmark with my previous version that always reads cache
3500          * shows that it makes things worse for diff-tree comparing
3501          * two linux-2.6 kernel trees in an already checked out work
3502          * tree.  This is because most diff-tree comparisons deal with
3503          * only a small number of files, while reading the cache is
3504          * expensive for a large project, and its cost outweighs the
3505          * savings we get by not inflating the object to a temporary
3506          * file.  Practically, this code only helps when we are used
3507          * by diff-cache --cached, which does read the cache before
3508          * calling us.
3509          */
3510         if (!active_cache)
3511                 return 0;
3512
3513         /* We want to avoid the working directory if our caller
3514          * doesn't need the data in a normal file, this system
3515          * is rather slow with its stat/open/mmap/close syscalls,
3516          * and the object is contained in a pack file.  The pack
3517          * is probably already open and will be faster to obtain
3518          * the data through than the working directory.  Loose
3519          * objects however would tend to be slower as they need
3520          * to be individually opened and inflated.
3521          */
3522         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3523                 return 0;
3524
3525         /*
3526          * Similarly, if we'd have to convert the file contents anyway, that
3527          * makes the optimization not worthwhile.
3528          */
3529         if (!want_file && would_convert_to_git(&the_index, name))
3530                 return 0;
3531
3532         len = strlen(name);
3533         pos = cache_name_pos(name, len);
3534         if (pos < 0)
3535                 return 0;
3536         ce = active_cache[pos];
3537
3538         /*
3539          * This is not the sha1 we are looking for, or
3540          * unreusable because it is not a regular file.
3541          */
3542         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3543                 return 0;
3544
3545         /*
3546          * If ce is marked as "assume unchanged", there is no
3547          * guarantee that work tree matches what we are looking for.
3548          */
3549         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3550                 return 0;
3551
3552         /*
3553          * If ce matches the file in the work tree, we can reuse it.
3554          */
3555         if (ce_uptodate(ce) ||
3556             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3557                 return 1;
3558
3559         return 0;
3560 }
3561
3562 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3563 {
3564         struct strbuf buf = STRBUF_INIT;
3565         char *dirty = "";
3566
3567         /* Are we looking at the work tree? */
3568         if (s->dirty_submodule)
3569                 dirty = "-dirty";
3570
3571         strbuf_addf(&buf, "Subproject commit %s%s\n",
3572                     oid_to_hex(&s->oid), dirty);
3573         s->size = buf.len;
3574         if (size_only) {
3575                 s->data = NULL;
3576                 strbuf_release(&buf);
3577         } else {
3578                 s->data = strbuf_detach(&buf, NULL);
3579                 s->should_free = 1;
3580         }
3581         return 0;
3582 }
3583
3584 /*
3585  * While doing rename detection and pickaxe operation, we may need to
3586  * grab the data for the blob (or file) for our own in-core comparison.
3587  * diff_filespec has data and size fields for this purpose.
3588  */
3589 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3590 {
3591         int size_only = flags & CHECK_SIZE_ONLY;
3592         int err = 0;
3593         int conv_flags = global_conv_flags_eol;
3594         /*
3595          * demote FAIL to WARN to allow inspecting the situation
3596          * instead of refusing.
3597          */
3598         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3599                 conv_flags = CONV_EOL_RNDTRP_WARN;
3600
3601         if (!DIFF_FILE_VALID(s))
3602                 die("internal error: asking to populate invalid file.");
3603         if (S_ISDIR(s->mode))
3604                 return -1;
3605
3606         if (s->data)
3607                 return 0;
3608
3609         if (size_only && 0 < s->size)
3610                 return 0;
3611
3612         if (S_ISGITLINK(s->mode))
3613                 return diff_populate_gitlink(s, size_only);
3614
3615         if (!s->oid_valid ||
3616             reuse_worktree_file(s->path, &s->oid, 0)) {
3617                 struct strbuf buf = STRBUF_INIT;
3618                 struct stat st;
3619                 int fd;
3620
3621                 if (lstat(s->path, &st) < 0) {
3622                 err_empty:
3623                         err = -1;
3624                 empty:
3625                         s->data = (char *)"";
3626                         s->size = 0;
3627                         return err;
3628                 }
3629                 s->size = xsize_t(st.st_size);
3630                 if (!s->size)
3631                         goto empty;
3632                 if (S_ISLNK(st.st_mode)) {
3633                         struct strbuf sb = STRBUF_INIT;
3634
3635                         if (strbuf_readlink(&sb, s->path, s->size))
3636                                 goto err_empty;
3637                         s->size = sb.len;
3638                         s->data = strbuf_detach(&sb, NULL);
3639                         s->should_free = 1;
3640                         return 0;
3641                 }
3642
3643                 /*
3644                  * Even if the caller would be happy with getting
3645                  * only the size, we cannot return early at this
3646                  * point if the path requires us to run the content
3647                  * conversion.
3648                  */
3649                 if (size_only && !would_convert_to_git(&the_index, s->path))
3650                         return 0;
3651
3652                 /*
3653                  * Note: this check uses xsize_t(st.st_size) that may
3654                  * not be the true size of the blob after it goes
3655                  * through convert_to_git().  This may not strictly be
3656                  * correct, but the whole point of big_file_threshold
3657                  * and is_binary check being that we want to avoid
3658                  * opening the file and inspecting the contents, this
3659                  * is probably fine.
3660                  */
3661                 if ((flags & CHECK_BINARY) &&
3662                     s->size > big_file_threshold && s->is_binary == -1) {
3663                         s->is_binary = 1;
3664                         return 0;
3665                 }
3666                 fd = open(s->path, O_RDONLY);
3667                 if (fd < 0)
3668                         goto err_empty;
3669                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3670                 close(fd);
3671                 s->should_munmap = 1;
3672
3673                 /*
3674                  * Convert from working tree format to canonical git format
3675                  */
3676                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3677                         size_t size = 0;
3678                         munmap(s->data, s->size);
3679                         s->should_munmap = 0;
3680                         s->data = strbuf_detach(&buf, &size);
3681                         s->size = size;
3682                         s->should_free = 1;
3683                 }
3684         }
3685         else {
3686                 enum object_type type;
3687                 if (size_only || (flags & CHECK_BINARY)) {
3688                         type = oid_object_info(&s->oid, &s->size);
3689                         if (type < 0)
3690                                 die("unable to read %s",
3691                                     oid_to_hex(&s->oid));
3692                         if (size_only)
3693                                 return 0;
3694                         if (s->size > big_file_threshold && s->is_binary == -1) {
3695                                 s->is_binary = 1;
3696                                 return 0;
3697                         }
3698                 }
3699                 s->data = read_object_file(&s->oid, &type, &s->size);
3700                 if (!s->data)
3701                         die("unable to read %s", oid_to_hex(&s->oid));
3702                 s->should_free = 1;
3703         }
3704         return 0;
3705 }
3706
3707 void diff_free_filespec_blob(struct diff_filespec *s)
3708 {
3709         if (s->should_free)
3710                 free(s->data);
3711         else if (s->should_munmap)
3712                 munmap(s->data, s->size);
3713
3714         if (s->should_free || s->should_munmap) {
3715                 s->should_free = s->should_munmap = 0;
3716                 s->data = NULL;
3717         }
3718 }
3719
3720 void diff_free_filespec_data(struct diff_filespec *s)
3721 {
3722         diff_free_filespec_blob(s);
3723         FREE_AND_NULL(s->cnt_data);
3724 }
3725
3726 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3727                            void *blob,
3728                            unsigned long size,
3729                            const struct object_id *oid,
3730                            int mode)
3731 {
3732         struct strbuf buf = STRBUF_INIT;
3733         struct strbuf tempfile = STRBUF_INIT;
3734         char *path_dup = xstrdup(path);
3735         const char *base = basename(path_dup);
3736
3737         /* Generate "XXXXXX_basename.ext" */
3738         strbuf_addstr(&tempfile, "XXXXXX_");
3739         strbuf_addstr(&tempfile, base);
3740
3741         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3742         if (!temp->tempfile)
3743                 die_errno("unable to create temp-file");
3744         if (convert_to_working_tree(path,
3745                         (const char *)blob, (size_t)size, &buf)) {
3746                 blob = buf.buf;
3747                 size = buf.len;
3748         }
3749         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3750             close_tempfile_gently(temp->tempfile))
3751                 die_errno("unable to write temp-file");
3752         temp->name = get_tempfile_path(temp->tempfile);
3753         oid_to_hex_r(temp->hex, oid);
3754         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3755         strbuf_release(&buf);
3756         strbuf_release(&tempfile);
3757         free(path_dup);
3758 }
3759
3760 static struct diff_tempfile *prepare_temp_file(const char *name,
3761                 struct diff_filespec *one)
3762 {
3763         struct diff_tempfile *temp = claim_diff_tempfile();
3764
3765         if (!DIFF_FILE_VALID(one)) {
3766         not_a_valid_file:
3767                 /* A '-' entry produces this for file-2, and
3768                  * a '+' entry produces this for file-1.
3769                  */
3770                 temp->name = "/dev/null";
3771                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3772                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3773                 return temp;
3774         }
3775
3776         if (!S_ISGITLINK(one->mode) &&
3777             (!one->oid_valid ||
3778              reuse_worktree_file(name, &one->oid, 1))) {
3779                 struct stat st;
3780                 if (lstat(name, &st) < 0) {
3781                         if (errno == ENOENT)
3782                                 goto not_a_valid_file;
3783                         die_errno("stat(%s)", name);
3784                 }
3785                 if (S_ISLNK(st.st_mode)) {
3786                         struct strbuf sb = STRBUF_INIT;
3787                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3788                                 die_errno("readlink(%s)", name);
3789                         prep_temp_blob(name, temp, sb.buf, sb.len,
3790                                        (one->oid_valid ?
3791                                         &one->oid : &null_oid),
3792                                        (one->oid_valid ?
3793                                         one->mode : S_IFLNK));
3794                         strbuf_release(&sb);
3795                 }
3796                 else {
3797                         /* we can borrow from the file in the work tree */
3798                         temp->name = name;
3799                         if (!one->oid_valid)
3800                                 oid_to_hex_r(temp->hex, &null_oid);
3801                         else
3802                                 oid_to_hex_r(temp->hex, &one->oid);
3803                         /* Even though we may sometimes borrow the
3804                          * contents from the work tree, we always want
3805                          * one->mode.  mode is trustworthy even when
3806                          * !(one->oid_valid), as long as
3807                          * DIFF_FILE_VALID(one).
3808                          */
3809                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3810                 }
3811                 return temp;
3812         }
3813         else {
3814                 if (diff_populate_filespec(one, 0))
3815                         die("cannot read data blob for %s", one->path);
3816                 prep_temp_blob(name, temp, one->data, one->size,
3817                                &one->oid, one->mode);
3818         }
3819         return temp;
3820 }
3821
3822 static void add_external_diff_name(struct argv_array *argv,
3823                                    const char *name,
3824                                    struct diff_filespec *df)
3825 {
3826         struct diff_tempfile *temp = prepare_temp_file(name, df);
3827         argv_array_push(argv, temp->name);
3828         argv_array_push(argv, temp->hex);
3829         argv_array_push(argv, temp->mode);
3830 }
3831
3832 /* An external diff command takes:
3833  *
3834  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3835  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3836  *
3837  */
3838 static void run_external_diff(const char *pgm,
3839                               const char *name,
3840                               const char *other,
3841                               struct diff_filespec *one,
3842                               struct diff_filespec *two,
3843                               const char *xfrm_msg,
3844                               int complete_rewrite,
3845                               struct diff_options *o)
3846 {
3847         struct argv_array argv = ARGV_ARRAY_INIT;
3848         struct argv_array env = ARGV_ARRAY_INIT;
3849         struct diff_queue_struct *q = &diff_queued_diff;
3850
3851         argv_array_push(&argv, pgm);
3852         argv_array_push(&argv, name);
3853
3854         if (one && two) {
3855                 add_external_diff_name(&argv, name, one);
3856                 if (!other)
3857                         add_external_diff_name(&argv, name, two);
3858                 else {
3859                         add_external_diff_name(&argv, other, two);
3860                         argv_array_push(&argv, other);
3861                         argv_array_push(&argv, xfrm_msg);
3862                 }
3863         }
3864
3865         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3866         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3867
3868         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3869                 die(_("external diff died, stopping at %s"), name);
3870
3871         remove_tempfile();
3872         argv_array_clear(&argv);
3873         argv_array_clear(&env);
3874 }
3875
3876 static int similarity_index(struct diff_filepair *p)
3877 {
3878         return p->score * 100 / MAX_SCORE;
3879 }
3880
3881 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3882 {
3883         if (startup_info->have_repository)
3884                 return find_unique_abbrev(oid, abbrev);
3885         else {
3886                 char *hex = oid_to_hex(oid);
3887                 if (abbrev < 0)
3888                         abbrev = FALLBACK_DEFAULT_ABBREV;
3889                 if (abbrev > GIT_SHA1_HEXSZ)
3890                         die("BUG: oid abbreviation out of range: %d", abbrev);
3891                 if (abbrev)
3892                         hex[abbrev] = '\0';
3893                 return hex;
3894         }
3895 }
3896
3897 static void fill_metainfo(struct strbuf *msg,
3898                           const char *name,
3899                           const char *other,
3900                           struct diff_filespec *one,
3901                           struct diff_filespec *two,
3902                           struct diff_options *o,
3903                           struct diff_filepair *p,
3904                           int *must_show_header,
3905                           int use_color)
3906 {
3907         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3908         const char *reset = diff_get_color(use_color, DIFF_RESET);
3909         const char *line_prefix = diff_line_prefix(o);
3910
3911         *must_show_header = 1;
3912         strbuf_init(msg, PATH_MAX * 2 + 300);
3913         switch (p->status) {
3914         case DIFF_STATUS_COPIED:
3915                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3916                             line_prefix, set, similarity_index(p));
3917                 strbuf_addf(msg, "%s\n%s%scopy from ",
3918                             reset,  line_prefix, set);
3919                 quote_c_style(name, msg, NULL, 0);
3920                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3921                 quote_c_style(other, msg, NULL, 0);
3922                 strbuf_addf(msg, "%s\n", reset);
3923                 break;
3924         case DIFF_STATUS_RENAMED:
3925                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3926                             line_prefix, set, similarity_index(p));
3927                 strbuf_addf(msg, "%s\n%s%srename from ",
3928                             reset, line_prefix, set);
3929                 quote_c_style(name, msg, NULL, 0);
3930                 strbuf_addf(msg, "%s\n%s%srename to ",
3931                             reset, line_prefix, set);
3932                 quote_c_style(other, msg, NULL, 0);
3933                 strbuf_addf(msg, "%s\n", reset);
3934                 break;
3935         case DIFF_STATUS_MODIFIED:
3936                 if (p->score) {
3937                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3938                                     line_prefix,
3939                                     set, similarity_index(p), reset);
3940                         break;
3941                 }
3942                 /* fallthru */
3943         default:
3944                 *must_show_header = 0;
3945         }
3946         if (one && two && oidcmp(&one->oid, &two->oid)) {
3947                 int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
3948
3949                 if (o->flags.binary) {
3950                         mmfile_t mf;
3951                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3952                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3953                                 abbrev = 40;
3954                 }
3955                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3956                             diff_abbrev_oid(&one->oid, abbrev),
3957                             diff_abbrev_oid(&two->oid, abbrev));
3958                 if (one->mode == two->mode)
3959                         strbuf_addf(msg, " %06o", one->mode);
3960                 strbuf_addf(msg, "%s\n", reset);
3961         }
3962 }
3963
3964 static void run_diff_cmd(const char *pgm,
3965                          const char *name,
3966                          const char *other,
3967                          const char *attr_path,
3968                          struct diff_filespec *one,
3969                          struct diff_filespec *two,
3970                          struct strbuf *msg,
3971                          struct diff_options *o,
3972                          struct diff_filepair *p)
3973 {
3974         const char *xfrm_msg = NULL;
3975         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3976         int must_show_header = 0;
3977
3978
3979         if (o->flags.allow_external) {
3980                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3981                 if (drv && drv->external)
3982                         pgm = drv->external;
3983         }
3984
3985         if (msg) {
3986                 /*
3987                  * don't use colors when the header is intended for an
3988                  * external diff driver
3989                  */
3990                 fill_metainfo(msg, name, other, one, two, o, p,
3991                               &must_show_header,
3992                               want_color(o->use_color) && !pgm);
3993                 xfrm_msg = msg->len ? msg->buf : NULL;
3994         }
3995
3996         if (pgm) {
3997                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3998                                   complete_rewrite, o);
3999                 return;
4000         }
4001         if (one && two)
4002                 builtin_diff(name, other ? other : name,
4003                              one, two, xfrm_msg, must_show_header,
4004                              o, complete_rewrite);
4005         else
4006                 fprintf(o->file, "* Unmerged path %s\n", name);
4007 }
4008
4009 static void diff_fill_oid_info(struct diff_filespec *one)
4010 {
4011         if (DIFF_FILE_VALID(one)) {
4012                 if (!one->oid_valid) {
4013                         struct stat st;
4014                         if (one->is_stdin) {
4015                                 oidclr(&one->oid);
4016                                 return;
4017                         }
4018                         if (lstat(one->path, &st) < 0)
4019                                 die_errno("stat '%s'", one->path);
4020                         if (index_path(&one->oid, one->path, &st, 0))
4021                                 die("cannot hash %s", one->path);
4022                 }
4023         }
4024         else
4025                 oidclr(&one->oid);
4026 }
4027
4028 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4029 {
4030         /* Strip the prefix but do not molest /dev/null and absolute paths */
4031         if (*namep && **namep != '/') {
4032                 *namep += prefix_length;
4033                 if (**namep == '/')
4034                         ++*namep;
4035         }
4036         if (*otherp && **otherp != '/') {
4037                 *otherp += prefix_length;
4038                 if (**otherp == '/')
4039                         ++*otherp;
4040         }
4041 }
4042
4043 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4044 {
4045         const char *pgm = external_diff();
4046         struct strbuf msg;
4047         struct diff_filespec *one = p->one;
4048         struct diff_filespec *two = p->two;
4049         const char *name;
4050         const char *other;
4051         const char *attr_path;
4052
4053         name  = one->path;
4054         other = (strcmp(name, two->path) ? two->path : NULL);
4055         attr_path = name;
4056         if (o->prefix_length)
4057                 strip_prefix(o->prefix_length, &name, &other);
4058
4059         if (!o->flags.allow_external)
4060                 pgm = NULL;
4061
4062         if (DIFF_PAIR_UNMERGED(p)) {
4063                 run_diff_cmd(pgm, name, NULL, attr_path,
4064                              NULL, NULL, NULL, o, p);
4065                 return;
4066         }
4067
4068         diff_fill_oid_info(one);
4069         diff_fill_oid_info(two);
4070
4071         if (!pgm &&
4072             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4073             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4074                 /*
4075                  * a filepair that changes between file and symlink
4076                  * needs to be split into deletion and creation.
4077                  */
4078                 struct diff_filespec *null = alloc_filespec(two->path);
4079                 run_diff_cmd(NULL, name, other, attr_path,
4080                              one, null, &msg, o, p);
4081                 free(null);
4082                 strbuf_release(&msg);
4083
4084                 null = alloc_filespec(one->path);
4085                 run_diff_cmd(NULL, name, other, attr_path,
4086                              null, two, &msg, o, p);
4087                 free(null);
4088         }
4089         else
4090                 run_diff_cmd(pgm, name, other, attr_path,
4091                              one, two, &msg, o, p);
4092
4093         strbuf_release(&msg);
4094 }
4095
4096 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4097                          struct diffstat_t *diffstat)
4098 {
4099         const char *name;
4100         const char *other;
4101
4102         if (DIFF_PAIR_UNMERGED(p)) {
4103                 /* unmerged */
4104                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4105                 return;
4106         }
4107
4108         name = p->one->path;
4109         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4110
4111         if (o->prefix_length)
4112                 strip_prefix(o->prefix_length, &name, &other);
4113
4114         diff_fill_oid_info(p->one);
4115         diff_fill_oid_info(p->two);
4116
4117         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4118 }
4119
4120 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4121 {
4122         const char *name;
4123         const char *other;
4124         const char *attr_path;
4125
4126         if (DIFF_PAIR_UNMERGED(p)) {
4127                 /* unmerged */
4128                 return;
4129         }
4130
4131         name = p->one->path;
4132         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4133         attr_path = other ? other : name;
4134
4135         if (o->prefix_length)
4136                 strip_prefix(o->prefix_length, &name, &other);
4137
4138         diff_fill_oid_info(p->one);
4139         diff_fill_oid_info(p->two);
4140
4141         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4142 }
4143
4144 void diff_setup(struct diff_options *options)
4145 {
4146         memcpy(options, &default_diff_options, sizeof(*options));
4147
4148         options->file = stdout;
4149
4150         options->abbrev = DEFAULT_ABBREV;
4151         options->line_termination = '\n';
4152         options->break_opt = -1;
4153         options->rename_limit = -1;
4154         options->dirstat_permille = diff_dirstat_permille_default;
4155         options->context = diff_context_default;
4156         options->interhunkcontext = diff_interhunk_context_default;
4157         options->ws_error_highlight = ws_error_highlight_default;
4158         options->flags.rename_empty = 1;
4159         options->objfind = NULL;
4160
4161         /* pathchange left =NULL by default */
4162         options->change = diff_change;
4163         options->add_remove = diff_addremove;
4164         options->use_color = diff_use_color_default;
4165         options->detect_rename = diff_detect_rename_default;
4166         options->xdl_opts |= diff_algorithm;
4167         if (diff_indent_heuristic)
4168                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4169
4170         options->orderfile = diff_order_file_cfg;
4171
4172         if (diff_no_prefix) {
4173                 options->a_prefix = options->b_prefix = "";
4174         } else if (!diff_mnemonic_prefix) {
4175                 options->a_prefix = "a/";
4176                 options->b_prefix = "b/";
4177         }
4178
4179         options->color_moved = diff_color_moved_default;
4180 }
4181
4182 void diff_setup_done(struct diff_options *options)
4183 {
4184         unsigned check_mask = DIFF_FORMAT_NAME |
4185                               DIFF_FORMAT_NAME_STATUS |
4186                               DIFF_FORMAT_CHECKDIFF |
4187                               DIFF_FORMAT_NO_OUTPUT;
4188
4189         if (options->set_default)
4190                 options->set_default(options);
4191
4192         if (HAS_MULTI_BITS(options->output_format & check_mask))
4193                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4194
4195         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4196                 die(_("-G, -S and --find-object are mutually exclusive"));
4197
4198         /*
4199          * Most of the time we can say "there are changes"
4200          * only by checking if there are changed paths, but
4201          * --ignore-whitespace* options force us to look
4202          * inside contents.
4203          */
4204
4205         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4206                 options->flags.diff_from_contents = 1;
4207         else
4208                 options->flags.diff_from_contents = 0;
4209
4210         if (options->flags.find_copies_harder)
4211                 options->detect_rename = DIFF_DETECT_COPY;
4212
4213         if (!options->flags.relative_name)
4214                 options->prefix = NULL;
4215         if (options->prefix)
4216                 options->prefix_length = strlen(options->prefix);
4217         else
4218                 options->prefix_length = 0;
4219
4220         if (options->output_format & (DIFF_FORMAT_NAME |
4221                                       DIFF_FORMAT_NAME_STATUS |
4222                                       DIFF_FORMAT_CHECKDIFF |
4223                                       DIFF_FORMAT_NO_OUTPUT))
4224                 options->output_format &= ~(DIFF_FORMAT_RAW |
4225                                             DIFF_FORMAT_NUMSTAT |
4226                                             DIFF_FORMAT_DIFFSTAT |
4227                                             DIFF_FORMAT_SHORTSTAT |
4228                                             DIFF_FORMAT_DIRSTAT |
4229                                             DIFF_FORMAT_SUMMARY |
4230                                             DIFF_FORMAT_PATCH);
4231
4232         /*
4233          * These cases always need recursive; we do not drop caller-supplied
4234          * recursive bits for other formats here.
4235          */
4236         if (options->output_format & (DIFF_FORMAT_PATCH |
4237                                       DIFF_FORMAT_NUMSTAT |
4238                                       DIFF_FORMAT_DIFFSTAT |
4239                                       DIFF_FORMAT_SHORTSTAT |
4240                                       DIFF_FORMAT_DIRSTAT |
4241                                       DIFF_FORMAT_SUMMARY |
4242                                       DIFF_FORMAT_CHECKDIFF))
4243                 options->flags.recursive = 1;
4244         /*
4245          * Also pickaxe would not work very well if you do not say recursive
4246          */
4247         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4248                 options->flags.recursive = 1;
4249         /*
4250          * When patches are generated, submodules diffed against the work tree
4251          * must be checked for dirtiness too so it can be shown in the output
4252          */
4253         if (options->output_format & DIFF_FORMAT_PATCH)
4254                 options->flags.dirty_submodules = 1;
4255
4256         if (options->detect_rename && options->rename_limit < 0)
4257                 options->rename_limit = diff_rename_limit_default;
4258         if (options->setup & DIFF_SETUP_USE_CACHE) {
4259                 if (!active_cache)
4260                         /* read-cache does not die even when it fails
4261                          * so it is safe for us to do this here.  Also
4262                          * it does not smudge active_cache or active_nr
4263                          * when it fails, so we do not have to worry about
4264                          * cleaning it up ourselves either.
4265                          */
4266                         read_cache();
4267         }
4268         if (40 < options->abbrev)
4269                 options->abbrev = 40; /* full */
4270
4271         /*
4272          * It does not make sense to show the first hit we happened
4273          * to have found.  It does not make sense not to return with
4274          * exit code in such a case either.
4275          */
4276         if (options->flags.quick) {
4277                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4278                 options->flags.exit_with_status = 1;
4279         }
4280
4281         options->diff_path_counter = 0;
4282
4283         if (options->flags.follow_renames && options->pathspec.nr != 1)
4284                 die(_("--follow requires exactly one pathspec"));
4285
4286         if (!options->use_color || external_diff())
4287                 options->color_moved = 0;
4288 }
4289
4290 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4291 {
4292         char c, *eq;
4293         int len;
4294
4295         if (*arg != '-')
4296                 return 0;
4297         c = *++arg;
4298         if (!c)
4299                 return 0;
4300         if (c == arg_short) {
4301                 c = *++arg;
4302                 if (!c)
4303                         return 1;
4304                 if (val && isdigit(c)) {
4305                         char *end;
4306                         int n = strtoul(arg, &end, 10);
4307                         if (*end)
4308                                 return 0;
4309                         *val = n;
4310                         return 1;
4311                 }
4312                 return 0;
4313         }
4314         if (c != '-')
4315                 return 0;
4316         arg++;
4317         eq = strchrnul(arg, '=');
4318         len = eq - arg;
4319         if (!len || strncmp(arg, arg_long, len))
4320                 return 0;
4321         if (*eq) {
4322                 int n;
4323                 char *end;
4324                 if (!isdigit(*++eq))
4325                         return 0;
4326                 n = strtoul(eq, &end, 10);
4327                 if (*end)
4328                         return 0;
4329                 *val = n;
4330         }
4331         return 1;
4332 }
4333
4334 static int diff_scoreopt_parse(const char *opt);
4335
4336 static inline int short_opt(char opt, const char **argv,
4337                             const char **optarg)
4338 {
4339         const char *arg = argv[0];
4340         if (arg[0] != '-' || arg[1] != opt)
4341                 return 0;
4342         if (arg[2] != '\0') {
4343                 *optarg = arg + 2;
4344                 return 1;
4345         }
4346         if (!argv[1])
4347                 die("Option '%c' requires a value", opt);
4348         *optarg = argv[1];
4349         return 2;
4350 }
4351
4352 int parse_long_opt(const char *opt, const char **argv,
4353                    const char **optarg)
4354 {
4355         const char *arg = argv[0];
4356         if (!skip_prefix(arg, "--", &arg))
4357                 return 0;
4358         if (!skip_prefix(arg, opt, &arg))
4359                 return 0;
4360         if (*arg == '=') { /* stuck form: --option=value */
4361                 *optarg = arg + 1;
4362                 return 1;
4363         }
4364         if (*arg != '\0')
4365                 return 0;
4366         /* separate form: --option value */
4367         if (!argv[1])
4368                 die("Option '--%s' requires a value", opt);
4369         *optarg = argv[1];
4370         return 2;
4371 }
4372
4373 static int stat_opt(struct diff_options *options, const char **av)
4374 {
4375         const char *arg = av[0];
4376         char *end;
4377         int width = options->stat_width;
4378         int name_width = options->stat_name_width;
4379         int graph_width = options->stat_graph_width;
4380         int count = options->stat_count;
4381         int argcount = 1;
4382
4383         if (!skip_prefix(arg, "--stat", &arg))
4384                 die("BUG: stat option does not begin with --stat: %s", arg);
4385         end = (char *)arg;
4386
4387         switch (*arg) {
4388         case '-':
4389                 if (skip_prefix(arg, "-width", &arg)) {
4390                         if (*arg == '=')
4391                                 width = strtoul(arg + 1, &end, 10);
4392                         else if (!*arg && !av[1])
4393                                 die_want_option("--stat-width");
4394                         else if (!*arg) {
4395                                 width = strtoul(av[1], &end, 10);
4396                                 argcount = 2;
4397                         }
4398                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4399                         if (*arg == '=')
4400                                 name_width = strtoul(arg + 1, &end, 10);
4401                         else if (!*arg && !av[1])
4402                                 die_want_option("--stat-name-width");
4403                         else if (!*arg) {
4404                                 name_width = strtoul(av[1], &end, 10);
4405                                 argcount = 2;
4406                         }
4407                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4408                         if (*arg == '=')
4409                                 graph_width = strtoul(arg + 1, &end, 10);
4410                         else if (!*arg && !av[1])
4411                                 die_want_option("--stat-graph-width");
4412                         else if (!*arg) {
4413                                 graph_width = strtoul(av[1], &end, 10);
4414                                 argcount = 2;
4415                         }
4416                 } else if (skip_prefix(arg, "-count", &arg)) {
4417                         if (*arg == '=')
4418                                 count = strtoul(arg + 1, &end, 10);
4419                         else if (!*arg && !av[1])
4420                                 die_want_option("--stat-count");
4421                         else if (!*arg) {
4422                                 count = strtoul(av[1], &end, 10);
4423                                 argcount = 2;
4424                         }
4425                 }
4426                 break;
4427         case '=':
4428                 width = strtoul(arg+1, &end, 10);
4429                 if (*end == ',')
4430                         name_width = strtoul(end+1, &end, 10);
4431                 if (*end == ',')
4432                         count = strtoul(end+1, &end, 10);
4433         }
4434
4435         /* Important! This checks all the error cases! */
4436         if (*end)
4437                 return 0;
4438         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4439         options->stat_name_width = name_width;
4440         options->stat_graph_width = graph_width;
4441         options->stat_width = width;
4442         options->stat_count = count;
4443         return argcount;
4444 }
4445
4446 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4447 {
4448         struct strbuf errmsg = STRBUF_INIT;
4449         if (parse_dirstat_params(options, params, &errmsg))
4450                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4451                     errmsg.buf);
4452         strbuf_release(&errmsg);
4453         /*
4454          * The caller knows a dirstat-related option is given from the command
4455          * line; allow it to say "return this_function();"
4456          */
4457         options->output_format |= DIFF_FORMAT_DIRSTAT;
4458         return 1;
4459 }
4460
4461 static int parse_submodule_opt(struct diff_options *options, const char *value)
4462 {
4463         if (parse_submodule_params(options, value))
4464                 die(_("Failed to parse --submodule option parameter: '%s'"),
4465                         value);
4466         return 1;
4467 }
4468
4469 static const char diff_status_letters[] = {
4470         DIFF_STATUS_ADDED,
4471         DIFF_STATUS_COPIED,
4472         DIFF_STATUS_DELETED,
4473         DIFF_STATUS_MODIFIED,
4474         DIFF_STATUS_RENAMED,
4475         DIFF_STATUS_TYPE_CHANGED,
4476         DIFF_STATUS_UNKNOWN,
4477         DIFF_STATUS_UNMERGED,
4478         DIFF_STATUS_FILTER_AON,
4479         DIFF_STATUS_FILTER_BROKEN,
4480         '\0',
4481 };
4482
4483 static unsigned int filter_bit['Z' + 1];
4484
4485 static void prepare_filter_bits(void)
4486 {
4487         int i;
4488
4489         if (!filter_bit[DIFF_STATUS_ADDED]) {
4490                 for (i = 0; diff_status_letters[i]; i++)
4491                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4492         }
4493 }
4494
4495 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4496 {
4497         return opt->filter & filter_bit[(int) status];
4498 }
4499
4500 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4501 {
4502         int i, optch;
4503
4504         prepare_filter_bits();
4505
4506         /*
4507          * If there is a negation e.g. 'd' in the input, and we haven't
4508          * initialized the filter field with another --diff-filter, start
4509          * from full set of bits, except for AON.
4510          */
4511         if (!opt->filter) {
4512                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4513                         if (optch < 'a' || 'z' < optch)
4514                                 continue;
4515                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4516                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4517                         break;
4518                 }
4519         }
4520
4521         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4522                 unsigned int bit;
4523                 int negate;
4524
4525                 if ('a' <= optch && optch <= 'z') {
4526                         negate = 1;
4527                         optch = toupper(optch);
4528                 } else {
4529                         negate = 0;
4530                 }
4531
4532                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4533                 if (!bit)
4534                         return optarg[i];
4535                 if (negate)
4536                         opt->filter &= ~bit;
4537                 else
4538                         opt->filter |= bit;
4539         }
4540         return 0;
4541 }
4542
4543 static void enable_patch_output(int *fmt) {
4544         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4545         *fmt |= DIFF_FORMAT_PATCH;
4546 }
4547
4548 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4549 {
4550         int val = parse_ws_error_highlight(arg);
4551
4552         if (val < 0) {
4553                 error("unknown value after ws-error-highlight=%.*s",
4554                       -1 - val, arg);
4555                 return 0;
4556         }
4557         opt->ws_error_highlight = val;
4558         return 1;
4559 }
4560
4561 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4562 {
4563         struct object_id oid;
4564
4565         if (get_oid(arg, &oid))
4566                 return error("unable to resolve '%s'", arg);
4567
4568         if (!opt->objfind)
4569                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4570
4571         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4572         opt->flags.recursive = 1;
4573         opt->flags.tree_in_recursive = 1;
4574         oidset_insert(opt->objfind, &oid);
4575         return 1;
4576 }
4577
4578 int diff_opt_parse(struct diff_options *options,
4579                    const char **av, int ac, const char *prefix)
4580 {
4581         const char *arg = av[0];
4582         const char *optarg;
4583         int argcount;
4584
4585         if (!prefix)
4586                 prefix = "";
4587
4588         /* Output format options */
4589         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4590             || opt_arg(arg, 'U', "unified", &options->context))
4591                 enable_patch_output(&options->output_format);
4592         else if (!strcmp(arg, "--raw"))
4593                 options->output_format |= DIFF_FORMAT_RAW;
4594         else if (!strcmp(arg, "--patch-with-raw")) {
4595                 enable_patch_output(&options->output_format);
4596                 options->output_format |= DIFF_FORMAT_RAW;
4597         } else if (!strcmp(arg, "--numstat"))
4598                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4599         else if (!strcmp(arg, "--shortstat"))
4600                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4601         else if (skip_prefix(arg, "-X", &arg) ||
4602                  skip_to_optional_arg(arg, "--dirstat", &arg))
4603                 return parse_dirstat_opt(options, arg);
4604         else if (!strcmp(arg, "--cumulative"))
4605                 return parse_dirstat_opt(options, "cumulative");
4606         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4607                 parse_dirstat_opt(options, "files");
4608                 return parse_dirstat_opt(options, arg);
4609         }
4610         else if (!strcmp(arg, "--check"))
4611                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4612         else if (!strcmp(arg, "--summary"))
4613                 options->output_format |= DIFF_FORMAT_SUMMARY;
4614         else if (!strcmp(arg, "--patch-with-stat")) {
4615                 enable_patch_output(&options->output_format);
4616                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4617         } else if (!strcmp(arg, "--name-only"))
4618                 options->output_format |= DIFF_FORMAT_NAME;
4619         else if (!strcmp(arg, "--name-status"))
4620                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4621         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4622                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4623         else if (starts_with(arg, "--stat"))
4624                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4625                 return stat_opt(options, av);
4626         else if (!strcmp(arg, "--compact-summary")) {
4627                  options->flags.stat_with_summary = 1;
4628                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4629         } else if (!strcmp(arg, "--no-compact-summary"))
4630                  options->flags.stat_with_summary = 0;
4631
4632         /* renames options */
4633         else if (starts_with(arg, "-B") ||
4634                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4635                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4636                         return error("invalid argument to -B: %s", arg+2);
4637         }
4638         else if (starts_with(arg, "-M") ||
4639                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4640                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4641                         return error("invalid argument to -M: %s", arg+2);
4642                 options->detect_rename = DIFF_DETECT_RENAME;
4643         }
4644         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4645                 options->irreversible_delete = 1;
4646         }
4647         else if (starts_with(arg, "-C") ||
4648                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4649                 if (options->detect_rename == DIFF_DETECT_COPY)
4650                         options->flags.find_copies_harder = 1;
4651                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4652                         return error("invalid argument to -C: %s", arg+2);
4653                 options->detect_rename = DIFF_DETECT_COPY;
4654         }
4655         else if (!strcmp(arg, "--no-renames"))
4656                 options->detect_rename = 0;
4657         else if (!strcmp(arg, "--rename-empty"))
4658                 options->flags.rename_empty = 1;
4659         else if (!strcmp(arg, "--no-rename-empty"))
4660                 options->flags.rename_empty = 0;
4661         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4662                 options->flags.relative_name = 1;
4663                 if (arg)
4664                         options->prefix = arg;
4665         }
4666
4667         /* xdiff options */
4668         else if (!strcmp(arg, "--minimal"))
4669                 DIFF_XDL_SET(options, NEED_MINIMAL);
4670         else if (!strcmp(arg, "--no-minimal"))
4671                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4672         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4673                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4674         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4675                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4676         else if (!strcmp(arg, "--ignore-space-at-eol"))
4677                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4678         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4679                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4680         else if (!strcmp(arg, "--ignore-blank-lines"))
4681                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4682         else if (!strcmp(arg, "--indent-heuristic"))
4683                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4684         else if (!strcmp(arg, "--no-indent-heuristic"))
4685                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4686         else if (!strcmp(arg, "--patience")) {
4687                 int i;
4688                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4689                 /*
4690                  * Both --patience and --anchored use PATIENCE_DIFF
4691                  * internally, so remove any anchors previously
4692                  * specified.
4693                  */
4694                 for (i = 0; i < options->anchors_nr; i++)
4695                         free(options->anchors[i]);
4696                 options->anchors_nr = 0;
4697         } else if (!strcmp(arg, "--histogram"))
4698                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4699         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4700                 long value = parse_algorithm_value(optarg);
4701                 if (value < 0)
4702                         return error("option diff-algorithm accepts \"myers\", "
4703                                      "\"minimal\", \"patience\" and \"histogram\"");
4704                 /* clear out previous settings */
4705                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4706                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4707                 options->xdl_opts |= value;
4708                 return argcount;
4709         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4710                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4711                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4712                            options->anchors_alloc);
4713                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4714         }
4715
4716         /* flags options */
4717         else if (!strcmp(arg, "--binary")) {
4718                 enable_patch_output(&options->output_format);
4719                 options->flags.binary = 1;
4720         }
4721         else if (!strcmp(arg, "--full-index"))
4722                 options->flags.full_index = 1;
4723         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4724                 options->flags.text = 1;
4725         else if (!strcmp(arg, "-R"))
4726                 options->flags.reverse_diff = 1;
4727         else if (!strcmp(arg, "--find-copies-harder"))
4728                 options->flags.find_copies_harder = 1;
4729         else if (!strcmp(arg, "--follow"))
4730                 options->flags.follow_renames = 1;
4731         else if (!strcmp(arg, "--no-follow")) {
4732                 options->flags.follow_renames = 0;
4733                 options->flags.default_follow_renames = 0;
4734         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4735                 int value = git_config_colorbool(NULL, arg);
4736                 if (value < 0)
4737                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4738                 options->use_color = value;
4739         }
4740         else if (!strcmp(arg, "--no-color"))
4741                 options->use_color = 0;
4742         else if (!strcmp(arg, "--color-moved")) {
4743                 if (diff_color_moved_default)
4744                         options->color_moved = diff_color_moved_default;
4745                 if (options->color_moved == COLOR_MOVED_NO)
4746                         options->color_moved = COLOR_MOVED_DEFAULT;
4747         } else if (!strcmp(arg, "--no-color-moved"))
4748                 options->color_moved = COLOR_MOVED_NO;
4749         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4750                 int cm = parse_color_moved(arg);
4751                 if (cm < 0)
4752                         die("bad --color-moved argument: %s", arg);
4753                 options->color_moved = cm;
4754         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
4755                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
4756         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4757                 options->use_color = 1;
4758                 options->word_diff = DIFF_WORDS_COLOR;
4759         }
4760         else if (!strcmp(arg, "--word-diff")) {
4761                 if (options->word_diff == DIFF_WORDS_NONE)
4762                         options->word_diff = DIFF_WORDS_PLAIN;
4763         }
4764         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4765                 if (!strcmp(arg, "plain"))
4766                         options->word_diff = DIFF_WORDS_PLAIN;
4767                 else if (!strcmp(arg, "color")) {
4768                         options->use_color = 1;
4769                         options->word_diff = DIFF_WORDS_COLOR;
4770                 }
4771                 else if (!strcmp(arg, "porcelain"))
4772                         options->word_diff = DIFF_WORDS_PORCELAIN;
4773                 else if (!strcmp(arg, "none"))
4774                         options->word_diff = DIFF_WORDS_NONE;
4775                 else
4776                         die("bad --word-diff argument: %s", arg);
4777         }
4778         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4779                 if (options->word_diff == DIFF_WORDS_NONE)
4780                         options->word_diff = DIFF_WORDS_PLAIN;
4781                 options->word_regex = optarg;
4782                 return argcount;
4783         }
4784         else if (!strcmp(arg, "--exit-code"))
4785                 options->flags.exit_with_status = 1;
4786         else if (!strcmp(arg, "--quiet"))
4787                 options->flags.quick = 1;
4788         else if (!strcmp(arg, "--ext-diff"))
4789                 options->flags.allow_external = 1;
4790         else if (!strcmp(arg, "--no-ext-diff"))
4791                 options->flags.allow_external = 0;
4792         else if (!strcmp(arg, "--textconv")) {
4793                 options->flags.allow_textconv = 1;
4794                 options->flags.textconv_set_via_cmdline = 1;
4795         } else if (!strcmp(arg, "--no-textconv"))
4796                 options->flags.allow_textconv = 0;
4797         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
4798                 options->flags.override_submodule_config = 1;
4799                 handle_ignore_submodules_arg(options, arg);
4800         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
4801                 return parse_submodule_opt(options, arg);
4802         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4803                 return parse_ws_error_highlight_opt(options, arg);
4804         else if (!strcmp(arg, "--ita-invisible-in-index"))
4805                 options->ita_invisible_in_index = 1;
4806         else if (!strcmp(arg, "--ita-visible-in-index"))
4807                 options->ita_invisible_in_index = 0;
4808
4809         /* misc options */
4810         else if (!strcmp(arg, "-z"))
4811                 options->line_termination = 0;
4812         else if ((argcount = short_opt('l', av, &optarg))) {
4813                 options->rename_limit = strtoul(optarg, NULL, 10);
4814                 return argcount;
4815         }
4816         else if ((argcount = short_opt('S', av, &optarg))) {
4817                 options->pickaxe = optarg;
4818                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4819                 return argcount;
4820         } else if ((argcount = short_opt('G', av, &optarg))) {
4821                 options->pickaxe = optarg;
4822                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4823                 return argcount;
4824         }
4825         else if (!strcmp(arg, "--pickaxe-all"))
4826                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4827         else if (!strcmp(arg, "--pickaxe-regex"))
4828                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4829         else if ((argcount = short_opt('O', av, &optarg))) {
4830                 options->orderfile = prefix_filename(prefix, optarg);
4831                 return argcount;
4832         } else if (skip_prefix(arg, "--find-object=", &arg))
4833                 return parse_objfind_opt(options, arg);
4834         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4835                 int offending = parse_diff_filter_opt(optarg, options);
4836                 if (offending)
4837                         die("unknown change class '%c' in --diff-filter=%s",
4838                             offending, optarg);
4839                 return argcount;
4840         }
4841         else if (!strcmp(arg, "--no-abbrev"))
4842                 options->abbrev = 0;
4843         else if (!strcmp(arg, "--abbrev"))
4844                 options->abbrev = DEFAULT_ABBREV;
4845         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4846                 options->abbrev = strtoul(arg, NULL, 10);
4847                 if (options->abbrev < MINIMUM_ABBREV)
4848                         options->abbrev = MINIMUM_ABBREV;
4849                 else if (40 < options->abbrev)
4850                         options->abbrev = 40;
4851         }
4852         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4853                 options->a_prefix = optarg;
4854                 return argcount;
4855         }
4856         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4857                 options->line_prefix = optarg;
4858                 options->line_prefix_length = strlen(options->line_prefix);
4859                 graph_setup_line_prefix(options);
4860                 return argcount;
4861         }
4862         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4863                 options->b_prefix = optarg;
4864                 return argcount;
4865         }
4866         else if (!strcmp(arg, "--no-prefix"))
4867                 options->a_prefix = options->b_prefix = "";
4868         else if (opt_arg(arg, '\0', "inter-hunk-context",
4869                          &options->interhunkcontext))
4870                 ;
4871         else if (!strcmp(arg, "-W"))
4872                 options->flags.funccontext = 1;
4873         else if (!strcmp(arg, "--function-context"))
4874                 options->flags.funccontext = 1;
4875         else if (!strcmp(arg, "--no-function-context"))
4876                 options->flags.funccontext = 0;
4877         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4878                 char *path = prefix_filename(prefix, optarg);
4879                 options->file = xfopen(path, "w");
4880                 options->close_file = 1;
4881                 if (options->use_color != GIT_COLOR_ALWAYS)
4882                         options->use_color = GIT_COLOR_NEVER;
4883                 free(path);
4884                 return argcount;
4885         } else
4886                 return 0;
4887         return 1;
4888 }
4889
4890 int parse_rename_score(const char **cp_p)
4891 {
4892         unsigned long num, scale;
4893         int ch, dot;
4894         const char *cp = *cp_p;
4895
4896         num = 0;
4897         scale = 1;
4898         dot = 0;
4899         for (;;) {
4900                 ch = *cp;
4901                 if ( !dot && ch == '.' ) {
4902                         scale = 1;
4903                         dot = 1;
4904                 } else if ( ch == '%' ) {
4905                         scale = dot ? scale*100 : 100;
4906                         cp++;   /* % is always at the end */
4907                         break;
4908                 } else if ( ch >= '0' && ch <= '9' ) {
4909                         if ( scale < 100000 ) {
4910                                 scale *= 10;
4911                                 num = (num*10) + (ch-'0');
4912                         }
4913                 } else {
4914                         break;
4915                 }
4916                 cp++;
4917         }
4918         *cp_p = cp;
4919
4920         /* user says num divided by scale and we say internally that
4921          * is MAX_SCORE * num / scale.
4922          */
4923         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4924 }
4925
4926 static int diff_scoreopt_parse(const char *opt)
4927 {
4928         int opt1, opt2, cmd;
4929
4930         if (*opt++ != '-')
4931                 return -1;
4932         cmd = *opt++;
4933         if (cmd == '-') {
4934                 /* convert the long-form arguments into short-form versions */
4935                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4936                         if (*opt == 0 || *opt++ == '=')
4937                                 cmd = 'B';
4938                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4939                         if (*opt == 0 || *opt++ == '=')
4940                                 cmd = 'C';
4941                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4942                         if (*opt == 0 || *opt++ == '=')
4943                                 cmd = 'M';
4944                 }
4945         }
4946         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4947                 return -1; /* that is not a -M, -C, or -B option */
4948
4949         opt1 = parse_rename_score(&opt);
4950         if (cmd != 'B')
4951                 opt2 = 0;
4952         else {
4953                 if (*opt == 0)
4954                         opt2 = 0;
4955                 else if (*opt != '/')
4956                         return -1; /* we expect -B80/99 or -B80 */
4957                 else {
4958                         opt++;
4959                         opt2 = parse_rename_score(&opt);
4960                 }
4961         }
4962         if (*opt != 0)
4963                 return -1;
4964         return opt1 | (opt2 << 16);
4965 }
4966
4967 struct diff_queue_struct diff_queued_diff;
4968
4969 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4970 {
4971         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4972         queue->queue[queue->nr++] = dp;
4973 }
4974
4975 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4976                                  struct diff_filespec *one,
4977                                  struct diff_filespec *two)
4978 {
4979         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4980         dp->one = one;
4981         dp->two = two;
4982         if (queue)
4983                 diff_q(queue, dp);
4984         return dp;
4985 }
4986
4987 void diff_free_filepair(struct diff_filepair *p)
4988 {
4989         free_filespec(p->one);
4990         free_filespec(p->two);
4991         free(p);
4992 }
4993
4994 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4995 {
4996         int abblen;
4997         const char *abbrev;
4998
4999         /* Do we want all 40 hex characters? */
5000         if (len == GIT_SHA1_HEXSZ)
5001                 return oid_to_hex(oid);
5002
5003         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5004         abbrev = diff_abbrev_oid(oid, len);
5005
5006         if (!print_sha1_ellipsis())
5007                 return abbrev;
5008
5009         abblen = strlen(abbrev);
5010
5011         /*
5012          * In well-behaved cases, where the abbreviated result is the
5013          * same as the requested length, append three dots after the
5014          * abbreviation (hence the whole logic is limited to the case
5015          * where abblen < 37); when the actual abbreviated result is a
5016          * bit longer than the requested length, we reduce the number
5017          * of dots so that they match the well-behaved ones.  However,
5018          * if the actual abbreviation is longer than the requested
5019          * length by more than three, we give up on aligning, and add
5020          * three dots anyway, to indicate that the output is not the
5021          * full object name.  Yes, this may be suboptimal, but this
5022          * appears only in "diff --raw --abbrev" output and it is not
5023          * worth the effort to change it now.  Note that this would
5024          * likely to work fine when the automatic sizing of default
5025          * abbreviation length is used--we would be fed -1 in "len" in
5026          * that case, and will end up always appending three-dots, but
5027          * the automatic sizing is supposed to give abblen that ensures
5028          * uniqueness across all objects (statistically speaking).
5029          */
5030         if (abblen < GIT_SHA1_HEXSZ - 3) {
5031                 static char hex[GIT_MAX_HEXSZ + 1];
5032                 if (len < abblen && abblen <= len + 2)
5033                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5034                 else
5035                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5036                 return hex;
5037         }
5038
5039         return oid_to_hex(oid);
5040 }
5041
5042 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5043 {
5044         int line_termination = opt->line_termination;
5045         int inter_name_termination = line_termination ? '\t' : '\0';
5046
5047         fprintf(opt->file, "%s", diff_line_prefix(opt));
5048         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5049                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5050                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5051                 fprintf(opt->file, "%s ",
5052                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5053         }
5054         if (p->score) {
5055                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5056                         inter_name_termination);
5057         } else {
5058                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5059         }
5060
5061         if (p->status == DIFF_STATUS_COPIED ||
5062             p->status == DIFF_STATUS_RENAMED) {
5063                 const char *name_a, *name_b;
5064                 name_a = p->one->path;
5065                 name_b = p->two->path;
5066                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5067                 write_name_quoted(name_a, opt->file, inter_name_termination);
5068                 write_name_quoted(name_b, opt->file, line_termination);
5069         } else {
5070                 const char *name_a, *name_b;
5071                 name_a = p->one->mode ? p->one->path : p->two->path;
5072                 name_b = NULL;
5073                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5074                 write_name_quoted(name_a, opt->file, line_termination);
5075         }
5076 }
5077
5078 int diff_unmodified_pair(struct diff_filepair *p)
5079 {
5080         /* This function is written stricter than necessary to support
5081          * the currently implemented transformers, but the idea is to
5082          * let transformers to produce diff_filepairs any way they want,
5083          * and filter and clean them up here before producing the output.
5084          */
5085         struct diff_filespec *one = p->one, *two = p->two;
5086
5087         if (DIFF_PAIR_UNMERGED(p))
5088                 return 0; /* unmerged is interesting */
5089
5090         /* deletion, addition, mode or type change
5091          * and rename are all interesting.
5092          */
5093         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5094             DIFF_PAIR_MODE_CHANGED(p) ||
5095             strcmp(one->path, two->path))
5096                 return 0;
5097
5098         /* both are valid and point at the same path.  that is, we are
5099          * dealing with a change.
5100          */
5101         if (one->oid_valid && two->oid_valid &&
5102             !oidcmp(&one->oid, &two->oid) &&
5103             !one->dirty_submodule && !two->dirty_submodule)
5104                 return 1; /* no change */
5105         if (!one->oid_valid && !two->oid_valid)
5106                 return 1; /* both look at the same file on the filesystem. */
5107         return 0;
5108 }
5109
5110 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5111 {
5112         if (diff_unmodified_pair(p))
5113                 return;
5114
5115         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5116             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5117                 return; /* no tree diffs in patch format */
5118
5119         run_diff(p, o);
5120 }
5121
5122 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5123                             struct diffstat_t *diffstat)
5124 {
5125         if (diff_unmodified_pair(p))
5126                 return;
5127
5128         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5129             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5130                 return; /* no useful stat for tree diffs */
5131
5132         run_diffstat(p, o, diffstat);
5133 }
5134
5135 static void diff_flush_checkdiff(struct diff_filepair *p,
5136                 struct diff_options *o)
5137 {
5138         if (diff_unmodified_pair(p))
5139                 return;
5140
5141         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5142             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5143                 return; /* nothing to check in tree diffs */
5144
5145         run_checkdiff(p, o);
5146 }
5147
5148 int diff_queue_is_empty(void)
5149 {
5150         struct diff_queue_struct *q = &diff_queued_diff;
5151         int i;
5152         for (i = 0; i < q->nr; i++)
5153                 if (!diff_unmodified_pair(q->queue[i]))
5154                         return 0;
5155         return 1;
5156 }
5157
5158 #if DIFF_DEBUG
5159 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5160 {
5161         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5162                 x, one ? one : "",
5163                 s->path,
5164                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5165                 s->mode,
5166                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5167         fprintf(stderr, "queue[%d] %s size %lu\n",
5168                 x, one ? one : "",
5169                 s->size);
5170 }
5171
5172 void diff_debug_filepair(const struct diff_filepair *p, int i)
5173 {
5174         diff_debug_filespec(p->one, i, "one");
5175         diff_debug_filespec(p->two, i, "two");
5176         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5177                 p->score, p->status ? p->status : '?',
5178                 p->one->rename_used, p->broken_pair);
5179 }
5180
5181 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5182 {
5183         int i;
5184         if (msg)
5185                 fprintf(stderr, "%s\n", msg);
5186         fprintf(stderr, "q->nr = %d\n", q->nr);
5187         for (i = 0; i < q->nr; i++) {
5188                 struct diff_filepair *p = q->queue[i];
5189                 diff_debug_filepair(p, i);
5190         }
5191 }
5192 #endif
5193
5194 static void diff_resolve_rename_copy(void)
5195 {
5196         int i;
5197         struct diff_filepair *p;
5198         struct diff_queue_struct *q = &diff_queued_diff;
5199
5200         diff_debug_queue("resolve-rename-copy", q);
5201
5202         for (i = 0; i < q->nr; i++) {
5203                 p = q->queue[i];
5204                 p->status = 0; /* undecided */
5205                 if (DIFF_PAIR_UNMERGED(p))
5206                         p->status = DIFF_STATUS_UNMERGED;
5207                 else if (!DIFF_FILE_VALID(p->one))
5208                         p->status = DIFF_STATUS_ADDED;
5209                 else if (!DIFF_FILE_VALID(p->two))
5210                         p->status = DIFF_STATUS_DELETED;
5211                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5212                         p->status = DIFF_STATUS_TYPE_CHANGED;
5213
5214                 /* from this point on, we are dealing with a pair
5215                  * whose both sides are valid and of the same type, i.e.
5216                  * either in-place edit or rename/copy edit.
5217                  */
5218                 else if (DIFF_PAIR_RENAME(p)) {
5219                         /*
5220                          * A rename might have re-connected a broken
5221                          * pair up, causing the pathnames to be the
5222                          * same again. If so, that's not a rename at
5223                          * all, just a modification..
5224                          *
5225                          * Otherwise, see if this source was used for
5226                          * multiple renames, in which case we decrement
5227                          * the count, and call it a copy.
5228                          */
5229                         if (!strcmp(p->one->path, p->two->path))
5230                                 p->status = DIFF_STATUS_MODIFIED;
5231                         else if (--p->one->rename_used > 0)
5232                                 p->status = DIFF_STATUS_COPIED;
5233                         else
5234                                 p->status = DIFF_STATUS_RENAMED;
5235                 }
5236                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5237                          p->one->mode != p->two->mode ||
5238                          p->one->dirty_submodule ||
5239                          p->two->dirty_submodule ||
5240                          is_null_oid(&p->one->oid))
5241                         p->status = DIFF_STATUS_MODIFIED;
5242                 else {
5243                         /* This is a "no-change" entry and should not
5244                          * happen anymore, but prepare for broken callers.
5245                          */
5246                         error("feeding unmodified %s to diffcore",
5247                               p->one->path);
5248                         p->status = DIFF_STATUS_UNKNOWN;
5249                 }
5250         }
5251         diff_debug_queue("resolve-rename-copy done", q);
5252 }
5253
5254 static int check_pair_status(struct diff_filepair *p)
5255 {
5256         switch (p->status) {
5257         case DIFF_STATUS_UNKNOWN:
5258                 return 0;
5259         case 0:
5260                 die("internal error in diff-resolve-rename-copy");
5261         default:
5262                 return 1;
5263         }
5264 }
5265
5266 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5267 {
5268         int fmt = opt->output_format;
5269
5270         if (fmt & DIFF_FORMAT_CHECKDIFF)
5271                 diff_flush_checkdiff(p, opt);
5272         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5273                 diff_flush_raw(p, opt);
5274         else if (fmt & DIFF_FORMAT_NAME) {
5275                 const char *name_a, *name_b;
5276                 name_a = p->two->path;
5277                 name_b = NULL;
5278                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5279                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5280                 write_name_quoted(name_a, opt->file, opt->line_termination);
5281         }
5282 }
5283
5284 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5285 {
5286         struct strbuf sb = STRBUF_INIT;
5287         if (fs->mode)
5288                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5289         else
5290                 strbuf_addf(&sb, " %s ", newdelete);
5291
5292         quote_c_style(fs->path, &sb, NULL, 0);
5293         strbuf_addch(&sb, '\n');
5294         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5295                          sb.buf, sb.len, 0);
5296         strbuf_release(&sb);
5297 }
5298
5299 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5300                 int show_name)
5301 {
5302         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5303                 struct strbuf sb = STRBUF_INIT;
5304                 strbuf_addf(&sb, " mode change %06o => %06o",
5305                             p->one->mode, p->two->mode);
5306                 if (show_name) {
5307                         strbuf_addch(&sb, ' ');
5308                         quote_c_style(p->two->path, &sb, NULL, 0);
5309                 }
5310                 strbuf_addch(&sb, '\n');
5311                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5312                                  sb.buf, sb.len, 0);
5313                 strbuf_release(&sb);
5314         }
5315 }
5316
5317 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5318                 struct diff_filepair *p)
5319 {
5320         struct strbuf sb = STRBUF_INIT;
5321         struct strbuf names = STRBUF_INIT;
5322
5323         pprint_rename(&names, p->one->path, p->two->path);
5324         strbuf_addf(&sb, " %s %s (%d%%)\n",
5325                     renamecopy, names.buf, similarity_index(p));
5326         strbuf_release(&names);
5327         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5328                                  sb.buf, sb.len, 0);
5329         show_mode_change(opt, p, 0);
5330         strbuf_release(&sb);
5331 }
5332
5333 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5334 {
5335         switch(p->status) {
5336         case DIFF_STATUS_DELETED:
5337                 show_file_mode_name(opt, "delete", p->one);
5338                 break;
5339         case DIFF_STATUS_ADDED:
5340                 show_file_mode_name(opt, "create", p->two);
5341                 break;
5342         case DIFF_STATUS_COPIED:
5343                 show_rename_copy(opt, "copy", p);
5344                 break;
5345         case DIFF_STATUS_RENAMED:
5346                 show_rename_copy(opt, "rename", p);
5347                 break;
5348         default:
5349                 if (p->score) {
5350                         struct strbuf sb = STRBUF_INIT;
5351                         strbuf_addstr(&sb, " rewrite ");
5352                         quote_c_style(p->two->path, &sb, NULL, 0);
5353                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5354                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5355                                          sb.buf, sb.len, 0);
5356                         strbuf_release(&sb);
5357                 }
5358                 show_mode_change(opt, p, !p->score);
5359                 break;
5360         }
5361 }
5362
5363 struct patch_id_t {
5364         git_SHA_CTX *ctx;
5365         int patchlen;
5366 };
5367
5368 static int remove_space(char *line, int len)
5369 {
5370         int i;
5371         char *dst = line;
5372         unsigned char c;
5373
5374         for (i = 0; i < len; i++)
5375                 if (!isspace((c = line[i])))
5376                         *dst++ = c;
5377
5378         return dst - line;
5379 }
5380
5381 static void patch_id_consume(void *priv, char *line, unsigned long len)
5382 {
5383         struct patch_id_t *data = priv;
5384         int new_len;
5385
5386         /* Ignore line numbers when computing the SHA1 of the patch */
5387         if (starts_with(line, "@@ -"))
5388                 return;
5389
5390         new_len = remove_space(line, len);
5391
5392         git_SHA1_Update(data->ctx, line, new_len);
5393         data->patchlen += new_len;
5394 }
5395
5396 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5397 {
5398         git_SHA1_Update(ctx, str, strlen(str));
5399 }
5400
5401 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5402 {
5403         /* large enough for 2^32 in octal */
5404         char buf[12];
5405         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5406         git_SHA1_Update(ctx, buf, len);
5407 }
5408
5409 /* returns 0 upon success, and writes result into sha1 */
5410 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5411 {
5412         struct diff_queue_struct *q = &diff_queued_diff;
5413         int i;
5414         git_SHA_CTX ctx;
5415         struct patch_id_t data;
5416
5417         git_SHA1_Init(&ctx);
5418         memset(&data, 0, sizeof(struct patch_id_t));
5419         data.ctx = &ctx;
5420
5421         for (i = 0; i < q->nr; i++) {
5422                 xpparam_t xpp;
5423                 xdemitconf_t xecfg;
5424                 mmfile_t mf1, mf2;
5425                 struct diff_filepair *p = q->queue[i];
5426                 int len1, len2;
5427
5428                 memset(&xpp, 0, sizeof(xpp));
5429                 memset(&xecfg, 0, sizeof(xecfg));
5430                 if (p->status == 0)
5431                         return error("internal diff status error");
5432                 if (p->status == DIFF_STATUS_UNKNOWN)
5433                         continue;
5434                 if (diff_unmodified_pair(p))
5435                         continue;
5436                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5437                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5438                         continue;
5439                 if (DIFF_PAIR_UNMERGED(p))
5440                         continue;
5441
5442                 diff_fill_oid_info(p->one);
5443                 diff_fill_oid_info(p->two);
5444
5445                 len1 = remove_space(p->one->path, strlen(p->one->path));
5446                 len2 = remove_space(p->two->path, strlen(p->two->path));
5447                 patch_id_add_string(&ctx, "diff--git");
5448                 patch_id_add_string(&ctx, "a/");
5449                 git_SHA1_Update(&ctx, p->one->path, len1);
5450                 patch_id_add_string(&ctx, "b/");
5451                 git_SHA1_Update(&ctx, p->two->path, len2);
5452
5453                 if (p->one->mode == 0) {
5454                         patch_id_add_string(&ctx, "newfilemode");
5455                         patch_id_add_mode(&ctx, p->two->mode);
5456                         patch_id_add_string(&ctx, "---/dev/null");
5457                         patch_id_add_string(&ctx, "+++b/");
5458                         git_SHA1_Update(&ctx, p->two->path, len2);
5459                 } else if (p->two->mode == 0) {
5460                         patch_id_add_string(&ctx, "deletedfilemode");
5461                         patch_id_add_mode(&ctx, p->one->mode);
5462                         patch_id_add_string(&ctx, "---a/");
5463                         git_SHA1_Update(&ctx, p->one->path, len1);
5464                         patch_id_add_string(&ctx, "+++/dev/null");
5465                 } else {
5466                         patch_id_add_string(&ctx, "---a/");
5467                         git_SHA1_Update(&ctx, p->one->path, len1);
5468                         patch_id_add_string(&ctx, "+++b/");
5469                         git_SHA1_Update(&ctx, p->two->path, len2);
5470                 }
5471
5472                 if (diff_header_only)
5473                         continue;
5474
5475                 if (fill_mmfile(&mf1, p->one) < 0 ||
5476                     fill_mmfile(&mf2, p->two) < 0)
5477                         return error("unable to read files to diff");
5478
5479                 if (diff_filespec_is_binary(p->one) ||
5480                     diff_filespec_is_binary(p->two)) {
5481                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5482                                         GIT_SHA1_HEXSZ);
5483                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5484                                         GIT_SHA1_HEXSZ);
5485                         continue;
5486                 }
5487
5488                 xpp.flags = 0;
5489                 xecfg.ctxlen = 3;
5490                 xecfg.flags = 0;
5491                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5492                                   &xpp, &xecfg))
5493                         return error("unable to generate patch-id diff for %s",
5494                                      p->one->path);
5495         }
5496
5497         git_SHA1_Final(oid->hash, &ctx);
5498         return 0;
5499 }
5500
5501 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5502 {
5503         struct diff_queue_struct *q = &diff_queued_diff;
5504         int i;
5505         int result = diff_get_patch_id(options, oid, diff_header_only);
5506
5507         for (i = 0; i < q->nr; i++)
5508                 diff_free_filepair(q->queue[i]);
5509
5510         free(q->queue);
5511         DIFF_QUEUE_CLEAR(q);
5512
5513         return result;
5514 }
5515
5516 static int is_summary_empty(const struct diff_queue_struct *q)
5517 {
5518         int i;
5519
5520         for (i = 0; i < q->nr; i++) {
5521                 const struct diff_filepair *p = q->queue[i];
5522
5523                 switch (p->status) {
5524                 case DIFF_STATUS_DELETED:
5525                 case DIFF_STATUS_ADDED:
5526                 case DIFF_STATUS_COPIED:
5527                 case DIFF_STATUS_RENAMED:
5528                         return 0;
5529                 default:
5530                         if (p->score)
5531                                 return 0;
5532                         if (p->one->mode && p->two->mode &&
5533                             p->one->mode != p->two->mode)
5534                                 return 0;
5535                         break;
5536                 }
5537         }
5538         return 1;
5539 }
5540
5541 static const char rename_limit_warning[] =
5542 N_("inexact rename detection was skipped due to too many files.");
5543
5544 static const char degrade_cc_to_c_warning[] =
5545 N_("only found copies from modified paths due to too many files.");
5546
5547 static const char rename_limit_advice[] =
5548 N_("you may want to set your %s variable to at least "
5549    "%d and retry the command.");
5550
5551 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5552 {
5553         fflush(stdout);
5554         if (degraded_cc)
5555                 warning(_(degrade_cc_to_c_warning));
5556         else if (needed)
5557                 warning(_(rename_limit_warning));
5558         else
5559                 return;
5560         if (0 < needed)
5561                 warning(_(rename_limit_advice), varname, needed);
5562 }
5563
5564 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5565 {
5566         int i;
5567         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5568         struct diff_queue_struct *q = &diff_queued_diff;
5569
5570         if (WSEH_NEW & WS_RULE_MASK)
5571                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5572
5573         if (o->color_moved)
5574                 o->emitted_symbols = &esm;
5575
5576         for (i = 0; i < q->nr; i++) {
5577                 struct diff_filepair *p = q->queue[i];
5578                 if (check_pair_status(p))
5579                         diff_flush_patch(p, o);
5580         }
5581
5582         if (o->emitted_symbols) {
5583                 if (o->color_moved) {
5584                         struct hashmap add_lines, del_lines;
5585
5586                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5587                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5588
5589                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5590                         mark_color_as_moved(o, &add_lines, &del_lines);
5591                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5592                                 dim_moved_lines(o);
5593
5594                         hashmap_free(&add_lines, 0);
5595                         hashmap_free(&del_lines, 0);
5596                 }
5597
5598                 for (i = 0; i < esm.nr; i++)
5599                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5600
5601                 for (i = 0; i < esm.nr; i++)
5602                         free((void *)esm.buf[i].line);
5603         }
5604         esm.nr = 0;
5605 }
5606
5607 void diff_flush(struct diff_options *options)
5608 {
5609         struct diff_queue_struct *q = &diff_queued_diff;
5610         int i, output_format = options->output_format;
5611         int separator = 0;
5612         int dirstat_by_line = 0;
5613
5614         /*
5615          * Order: raw, stat, summary, patch
5616          * or:    name/name-status/checkdiff (other bits clear)
5617          */
5618         if (!q->nr)
5619                 goto free_queue;
5620
5621         if (output_format & (DIFF_FORMAT_RAW |
5622                              DIFF_FORMAT_NAME |
5623                              DIFF_FORMAT_NAME_STATUS |
5624                              DIFF_FORMAT_CHECKDIFF)) {
5625                 for (i = 0; i < q->nr; i++) {
5626                         struct diff_filepair *p = q->queue[i];
5627                         if (check_pair_status(p))
5628                                 flush_one_pair(p, options);
5629                 }
5630                 separator++;
5631         }
5632
5633         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5634                 dirstat_by_line = 1;
5635
5636         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5637             dirstat_by_line) {
5638                 struct diffstat_t diffstat;
5639
5640                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5641                 for (i = 0; i < q->nr; i++) {
5642                         struct diff_filepair *p = q->queue[i];
5643                         if (check_pair_status(p))
5644                                 diff_flush_stat(p, options, &diffstat);
5645                 }
5646                 if (output_format & DIFF_FORMAT_NUMSTAT)
5647                         show_numstat(&diffstat, options);
5648                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5649                         show_stats(&diffstat, options);
5650                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5651                         show_shortstats(&diffstat, options);
5652                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5653                         show_dirstat_by_line(&diffstat, options);
5654                 free_diffstat_info(&diffstat);
5655                 separator++;
5656         }
5657         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5658                 show_dirstat(options);
5659
5660         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5661                 for (i = 0; i < q->nr; i++) {
5662                         diff_summary(options, q->queue[i]);
5663                 }
5664                 separator++;
5665         }
5666
5667         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5668             options->flags.exit_with_status &&
5669             options->flags.diff_from_contents) {
5670                 /*
5671                  * run diff_flush_patch for the exit status. setting
5672                  * options->file to /dev/null should be safe, because we
5673                  * aren't supposed to produce any output anyway.
5674                  */
5675                 if (options->close_file)
5676                         fclose(options->file);
5677                 options->file = xfopen("/dev/null", "w");
5678                 options->close_file = 1;
5679                 options->color_moved = 0;
5680                 for (i = 0; i < q->nr; i++) {
5681                         struct diff_filepair *p = q->queue[i];
5682                         if (check_pair_status(p))
5683                                 diff_flush_patch(p, options);
5684                         if (options->found_changes)
5685                                 break;
5686                 }
5687         }
5688
5689         if (output_format & DIFF_FORMAT_PATCH) {
5690                 if (separator) {
5691                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5692                         if (options->stat_sep)
5693                                 /* attach patch instead of inline */
5694                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5695                                                  NULL, 0, 0);
5696                 }
5697
5698                 diff_flush_patch_all_file_pairs(options);
5699         }
5700
5701         if (output_format & DIFF_FORMAT_CALLBACK)
5702                 options->format_callback(q, options, options->format_callback_data);
5703
5704         for (i = 0; i < q->nr; i++)
5705                 diff_free_filepair(q->queue[i]);
5706 free_queue:
5707         free(q->queue);
5708         DIFF_QUEUE_CLEAR(q);
5709         if (options->close_file)
5710                 fclose(options->file);
5711
5712         /*
5713          * Report the content-level differences with HAS_CHANGES;
5714          * diff_addremove/diff_change does not set the bit when
5715          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5716          */
5717         if (options->flags.diff_from_contents) {
5718                 if (options->found_changes)
5719                         options->flags.has_changes = 1;
5720                 else
5721                         options->flags.has_changes = 0;
5722         }
5723 }
5724
5725 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5726 {
5727         return (((p->status == DIFF_STATUS_MODIFIED) &&
5728                  ((p->score &&
5729                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5730                   (!p->score &&
5731                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5732                 ((p->status != DIFF_STATUS_MODIFIED) &&
5733                  filter_bit_tst(p->status, options)));
5734 }
5735
5736 static void diffcore_apply_filter(struct diff_options *options)
5737 {
5738         int i;
5739         struct diff_queue_struct *q = &diff_queued_diff;
5740         struct diff_queue_struct outq;
5741
5742         DIFF_QUEUE_CLEAR(&outq);
5743
5744         if (!options->filter)
5745                 return;
5746
5747         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5748                 int found;
5749                 for (i = found = 0; !found && i < q->nr; i++) {
5750                         if (match_filter(options, q->queue[i]))
5751                                 found++;
5752                 }
5753                 if (found)
5754                         return;
5755
5756                 /* otherwise we will clear the whole queue
5757                  * by copying the empty outq at the end of this
5758                  * function, but first clear the current entries
5759                  * in the queue.
5760                  */
5761                 for (i = 0; i < q->nr; i++)
5762                         diff_free_filepair(q->queue[i]);
5763         }
5764         else {
5765                 /* Only the matching ones */
5766                 for (i = 0; i < q->nr; i++) {
5767                         struct diff_filepair *p = q->queue[i];
5768                         if (match_filter(options, p))
5769                                 diff_q(&outq, p);
5770                         else
5771                                 diff_free_filepair(p);
5772                 }
5773         }
5774         free(q->queue);
5775         *q = outq;
5776 }
5777
5778 /* Check whether two filespecs with the same mode and size are identical */
5779 static int diff_filespec_is_identical(struct diff_filespec *one,
5780                                       struct diff_filespec *two)
5781 {
5782         if (S_ISGITLINK(one->mode))
5783                 return 0;
5784         if (diff_populate_filespec(one, 0))
5785                 return 0;
5786         if (diff_populate_filespec(two, 0))
5787                 return 0;
5788         return !memcmp(one->data, two->data, one->size);
5789 }
5790
5791 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5792 {
5793         if (p->done_skip_stat_unmatch)
5794                 return p->skip_stat_unmatch_result;
5795
5796         p->done_skip_stat_unmatch = 1;
5797         p->skip_stat_unmatch_result = 0;
5798         /*
5799          * 1. Entries that come from stat info dirtiness
5800          *    always have both sides (iow, not create/delete),
5801          *    one side of the object name is unknown, with
5802          *    the same mode and size.  Keep the ones that
5803          *    do not match these criteria.  They have real
5804          *    differences.
5805          *
5806          * 2. At this point, the file is known to be modified,
5807          *    with the same mode and size, and the object
5808          *    name of one side is unknown.  Need to inspect
5809          *    the identical contents.
5810          */
5811         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5812             !DIFF_FILE_VALID(p->two) ||
5813             (p->one->oid_valid && p->two->oid_valid) ||
5814             (p->one->mode != p->two->mode) ||
5815             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5816             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5817             (p->one->size != p->two->size) ||
5818             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5819                 p->skip_stat_unmatch_result = 1;
5820         return p->skip_stat_unmatch_result;
5821 }
5822
5823 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5824 {
5825         int i;
5826         struct diff_queue_struct *q = &diff_queued_diff;
5827         struct diff_queue_struct outq;
5828         DIFF_QUEUE_CLEAR(&outq);
5829
5830         for (i = 0; i < q->nr; i++) {
5831                 struct diff_filepair *p = q->queue[i];
5832
5833                 if (diff_filespec_check_stat_unmatch(p))
5834                         diff_q(&outq, p);
5835                 else {
5836                         /*
5837                          * The caller can subtract 1 from skip_stat_unmatch
5838                          * to determine how many paths were dirty only
5839                          * due to stat info mismatch.
5840                          */
5841                         if (!diffopt->flags.no_index)
5842                                 diffopt->skip_stat_unmatch++;
5843                         diff_free_filepair(p);
5844                 }
5845         }
5846         free(q->queue);
5847         *q = outq;
5848 }
5849
5850 static int diffnamecmp(const void *a_, const void *b_)
5851 {
5852         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5853         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5854         const char *name_a, *name_b;
5855
5856         name_a = a->one ? a->one->path : a->two->path;
5857         name_b = b->one ? b->one->path : b->two->path;
5858         return strcmp(name_a, name_b);
5859 }
5860
5861 void diffcore_fix_diff_index(struct diff_options *options)
5862 {
5863         struct diff_queue_struct *q = &diff_queued_diff;
5864         QSORT(q->queue, q->nr, diffnamecmp);
5865 }
5866
5867 void diffcore_std(struct diff_options *options)
5868 {
5869         /* NOTE please keep the following in sync with diff_tree_combined() */
5870         if (options->skip_stat_unmatch)
5871                 diffcore_skip_stat_unmatch(options);
5872         if (!options->found_follow) {
5873                 /* See try_to_follow_renames() in tree-diff.c */
5874                 if (options->break_opt != -1)
5875                         diffcore_break(options->break_opt);
5876                 if (options->detect_rename)
5877                         diffcore_rename(options);
5878                 if (options->break_opt != -1)
5879                         diffcore_merge_broken();
5880         }
5881         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
5882                 diffcore_pickaxe(options);
5883         if (options->orderfile)
5884                 diffcore_order(options->orderfile);
5885         if (!options->found_follow)
5886                 /* See try_to_follow_renames() in tree-diff.c */
5887                 diff_resolve_rename_copy();
5888         diffcore_apply_filter(options);
5889
5890         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
5891                 options->flags.has_changes = 1;
5892         else
5893                 options->flags.has_changes = 0;
5894
5895         options->found_follow = 0;
5896 }
5897
5898 int diff_result_code(struct diff_options *opt, int status)
5899 {
5900         int result = 0;
5901
5902         diff_warn_rename_limit("diff.renameLimit",
5903                                opt->needed_rename_limit,
5904                                opt->degraded_cc_to_c);
5905         if (!opt->flags.exit_with_status &&
5906             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5907                 return status;
5908         if (opt->flags.exit_with_status &&
5909             opt->flags.has_changes)
5910                 result |= 01;
5911         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5912             opt->flags.check_failed)
5913                 result |= 02;
5914         return result;
5915 }
5916
5917 int diff_can_quit_early(struct diff_options *opt)
5918 {
5919         return (opt->flags.quick &&
5920                 !opt->filter &&
5921                 opt->flags.has_changes);
5922 }
5923
5924 /*
5925  * Shall changes to this submodule be ignored?
5926  *
5927  * Submodule changes can be configured to be ignored separately for each path,
5928  * but that configuration can be overridden from the command line.
5929  */
5930 static int is_submodule_ignored(const char *path, struct diff_options *options)
5931 {
5932         int ignored = 0;
5933         struct diff_flags orig_flags = options->flags;
5934         if (!options->flags.override_submodule_config)
5935                 set_diffopt_flags_from_submodule_config(options, path);
5936         if (options->flags.ignore_submodules)
5937                 ignored = 1;
5938         options->flags = orig_flags;
5939         return ignored;
5940 }
5941
5942 void diff_addremove(struct diff_options *options,
5943                     int addremove, unsigned mode,
5944                     const struct object_id *oid,
5945                     int oid_valid,
5946                     const char *concatpath, unsigned dirty_submodule)
5947 {
5948         struct diff_filespec *one, *two;
5949
5950         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5951                 return;
5952
5953         /* This may look odd, but it is a preparation for
5954          * feeding "there are unchanged files which should
5955          * not produce diffs, but when you are doing copy
5956          * detection you would need them, so here they are"
5957          * entries to the diff-core.  They will be prefixed
5958          * with something like '=' or '*' (I haven't decided
5959          * which but should not make any difference).
5960          * Feeding the same new and old to diff_change()
5961          * also has the same effect.
5962          * Before the final output happens, they are pruned after
5963          * merged into rename/copy pairs as appropriate.
5964          */
5965         if (options->flags.reverse_diff)
5966                 addremove = (addremove == '+' ? '-' :
5967                              addremove == '-' ? '+' : addremove);
5968
5969         if (options->prefix &&
5970             strncmp(concatpath, options->prefix, options->prefix_length))
5971                 return;
5972
5973         one = alloc_filespec(concatpath);
5974         two = alloc_filespec(concatpath);
5975
5976         if (addremove != '+')
5977                 fill_filespec(one, oid, oid_valid, mode);
5978         if (addremove != '-') {
5979                 fill_filespec(two, oid, oid_valid, mode);
5980                 two->dirty_submodule = dirty_submodule;
5981         }
5982
5983         diff_queue(&diff_queued_diff, one, two);
5984         if (!options->flags.diff_from_contents)
5985                 options->flags.has_changes = 1;
5986 }
5987
5988 void diff_change(struct diff_options *options,
5989                  unsigned old_mode, unsigned new_mode,
5990                  const struct object_id *old_oid,
5991                  const struct object_id *new_oid,
5992                  int old_oid_valid, int new_oid_valid,
5993                  const char *concatpath,
5994                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5995 {
5996         struct diff_filespec *one, *two;
5997         struct diff_filepair *p;
5998
5999         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6000             is_submodule_ignored(concatpath, options))
6001                 return;
6002
6003         if (options->flags.reverse_diff) {
6004                 SWAP(old_mode, new_mode);
6005                 SWAP(old_oid, new_oid);
6006                 SWAP(old_oid_valid, new_oid_valid);
6007                 SWAP(old_dirty_submodule, new_dirty_submodule);
6008         }
6009
6010         if (options->prefix &&
6011             strncmp(concatpath, options->prefix, options->prefix_length))
6012                 return;
6013
6014         one = alloc_filespec(concatpath);
6015         two = alloc_filespec(concatpath);
6016         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6017         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6018         one->dirty_submodule = old_dirty_submodule;
6019         two->dirty_submodule = new_dirty_submodule;
6020         p = diff_queue(&diff_queued_diff, one, two);
6021
6022         if (options->flags.diff_from_contents)
6023                 return;
6024
6025         if (options->flags.quick && options->skip_stat_unmatch &&
6026             !diff_filespec_check_stat_unmatch(p))
6027                 return;
6028
6029         options->flags.has_changes = 1;
6030 }
6031
6032 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6033 {
6034         struct diff_filepair *pair;
6035         struct diff_filespec *one, *two;
6036
6037         if (options->prefix &&
6038             strncmp(path, options->prefix, options->prefix_length))
6039                 return NULL;
6040
6041         one = alloc_filespec(path);
6042         two = alloc_filespec(path);
6043         pair = diff_queue(&diff_queued_diff, one, two);
6044         pair->is_unmerged = 1;
6045         return pair;
6046 }
6047
6048 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6049                 size_t *outsize)
6050 {
6051         struct diff_tempfile *temp;
6052         const char *argv[3];
6053         const char **arg = argv;
6054         struct child_process child = CHILD_PROCESS_INIT;
6055         struct strbuf buf = STRBUF_INIT;
6056         int err = 0;
6057
6058         temp = prepare_temp_file(spec->path, spec);
6059         *arg++ = pgm;
6060         *arg++ = temp->name;
6061         *arg = NULL;
6062
6063         child.use_shell = 1;
6064         child.argv = argv;
6065         child.out = -1;
6066         if (start_command(&child)) {
6067                 remove_tempfile();
6068                 return NULL;
6069         }
6070
6071         if (strbuf_read(&buf, child.out, 0) < 0)
6072                 err = error("error reading from textconv command '%s'", pgm);
6073         close(child.out);
6074
6075         if (finish_command(&child) || err) {
6076                 strbuf_release(&buf);
6077                 remove_tempfile();
6078                 return NULL;
6079         }
6080         remove_tempfile();
6081
6082         return strbuf_detach(&buf, outsize);
6083 }
6084
6085 size_t fill_textconv(struct userdiff_driver *driver,
6086                      struct diff_filespec *df,
6087                      char **outbuf)
6088 {
6089         size_t size;
6090
6091         if (!driver) {
6092                 if (!DIFF_FILE_VALID(df)) {
6093                         *outbuf = "";
6094                         return 0;
6095                 }
6096                 if (diff_populate_filespec(df, 0))
6097                         die("unable to read files to diff");
6098                 *outbuf = df->data;
6099                 return df->size;
6100         }
6101
6102         if (!driver->textconv)
6103                 die("BUG: fill_textconv called with non-textconv driver");
6104
6105         if (driver->textconv_cache && df->oid_valid) {
6106                 *outbuf = notes_cache_get(driver->textconv_cache,
6107                                           &df->oid,
6108                                           &size);
6109                 if (*outbuf)
6110                         return size;
6111         }
6112
6113         *outbuf = run_textconv(driver->textconv, df, &size);
6114         if (!*outbuf)
6115                 die("unable to read files to diff");
6116
6117         if (driver->textconv_cache && df->oid_valid) {
6118                 /* ignore errors, as we might be in a readonly repository */
6119                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6120                                 size);
6121                 /*
6122                  * we could save up changes and flush them all at the end,
6123                  * but we would need an extra call after all diffing is done.
6124                  * Since generating a cache entry is the slow path anyway,
6125                  * this extra overhead probably isn't a big deal.
6126                  */
6127                 notes_cache_write(driver->textconv_cache);
6128         }
6129
6130         return size;
6131 }
6132
6133 int textconv_object(const char *path,
6134                     unsigned mode,
6135                     const struct object_id *oid,
6136                     int oid_valid,
6137                     char **buf,
6138                     unsigned long *buf_size)
6139 {
6140         struct diff_filespec *df;
6141         struct userdiff_driver *textconv;
6142
6143         df = alloc_filespec(path);
6144         fill_filespec(df, oid, oid_valid, mode);
6145         textconv = get_textconv(df);
6146         if (!textconv) {
6147                 free_filespec(df);
6148                 return 0;
6149         }
6150
6151         *buf_size = fill_textconv(textconv, df, buf);
6152         free_filespec(df);
6153         return 1;
6154 }
6155
6156 void setup_diff_pager(struct diff_options *opt)
6157 {
6158         /*
6159          * If the user asked for our exit code, then either they want --quiet
6160          * or --exit-code. We should definitely not bother with a pager in the
6161          * former case, as we will generate no output. Since we still properly
6162          * report our exit code even when a pager is run, we _could_ run a
6163          * pager with --exit-code. But since we have not done so historically,
6164          * and because it is easy to find people oneline advising "git diff
6165          * --exit-code" in hooks and other scripts, we do not do so.
6166          */
6167         if (!opt->flags.exit_with_status &&
6168             check_pager_config("diff") != 0)
6169                 setup_pager();
6170 }