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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "tempfile.h"
   6 #include "quote.h"
   7 #include "diff.h"
   8 #include "diffcore.h"
   9 #include "delta.h"
  10 #include "xdiff-interface.h"
  11 #include "color.h"
  12 #include "attr.h"
  13 #include "run-command.h"
  14 #include "utf8.h"
  15 #include "userdiff.h"
  16 #include "submodule-config.h"
  17 #include "submodule.h"
  18 #include "ll-merge.h"
  19 #include "string-list.h"
  20 #include "argv-array.h"
  21 #include "graph.h"
  22
  23 #ifdef NO_FAST_WORKING_DIRECTORY
  24 #define FAST_WORKING_DIRECTORY 0
  25 #else
  26 #define FAST_WORKING_DIRECTORY 1
  27 #endif
  28
  29 static int diff_detect_rename_default;
  30 static int diff_indent_heuristic; /* experimental */
  31 static int diff_rename_limit_default = 400;
  32 static int diff_suppress_blank_empty;
  33 static int diff_use_color_default = -1;
  34 static int diff_context_default = 3;
  35 static int diff_interhunk_context_default;
  36 static const char *diff_word_regex_cfg;
  37 static const char *external_diff_cmd_cfg;
  38 static const char *diff_order_file_cfg;
  39 int diff_auto_refresh_index = 1;
  40 static int diff_mnemonic_prefix;
  41 static int diff_no_prefix;
  42 static int diff_stat_graph_width;
  43 static int diff_dirstat_permille_default = 30;
  44 static struct diff_options default_diff_options;
  45 static long diff_algorithm;
  46 static unsigned ws_error_highlight_default = WSEH_NEW;
  47
  48 static char diff_colors[][COLOR_MAXLEN] = {
  49         GIT_COLOR_RESET,
  50         GIT_COLOR_NORMAL,       /* CONTEXT */
  51         GIT_COLOR_BOLD,         /* METAINFO */
  52         GIT_COLOR_CYAN,         /* FRAGINFO */
  53         GIT_COLOR_RED,          /* OLD */
  54         GIT_COLOR_GREEN,        /* NEW */
  55         GIT_COLOR_YELLOW,       /* COMMIT */
  56         GIT_COLOR_BG_RED,       /* WHITESPACE */
  57         GIT_COLOR_NORMAL,       /* FUNCINFO */
  58 };
  59
  60 static NORETURN void die_want_option(const char *option_name)
  61 {
  62         die(_("option '%s' requires a value"), option_name);
  63 }
  64
  65 static int parse_diff_color_slot(const char *var)
  66 {
  67         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  68                 return DIFF_CONTEXT;
  69         if (!strcasecmp(var, "meta"))
  70                 return DIFF_METAINFO;
  71         if (!strcasecmp(var, "frag"))
  72                 return DIFF_FRAGINFO;
  73         if (!strcasecmp(var, "old"))
  74                 return DIFF_FILE_OLD;
  75         if (!strcasecmp(var, "new"))
  76                 return DIFF_FILE_NEW;
  77         if (!strcasecmp(var, "commit"))
  78                 return DIFF_COMMIT;
  79         if (!strcasecmp(var, "whitespace"))
  80                 return DIFF_WHITESPACE;
  81         if (!strcasecmp(var, "func"))
  82                 return DIFF_FUNCINFO;
  83         return -1;
  84 }
  85
  86 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  87                                 struct strbuf *errmsg)
  88 {
  89         char *params_copy = xstrdup(params_string);
  90         struct string_list params = STRING_LIST_INIT_NODUP;
  91         int ret = 0;
  92         int i;
  93
  94         if (*params_copy)
  95                 string_list_split_in_place(&params, params_copy, ',', -1);
  96         for (i = 0; i < params.nr; i++) {
  97                 const char *p = params.items[i].string;
  98                 if (!strcmp(p, "changes")) {
  99                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 100                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 101                 } else if (!strcmp(p, "lines")) {
 102                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
 103                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
 104                 } else if (!strcmp(p, "files")) {
 105                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
 106                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 107                 } else if (!strcmp(p, "noncumulative")) {
 108                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 109                 } else if (!strcmp(p, "cumulative")) {
 110                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 111                 } else if (isdigit(*p)) {
 112                         char *end;
 113                         int permille = strtoul(p, &end, 10) * 10;
 114                         if (*end == '.' && isdigit(*++end)) {
 115                                 /* only use first digit */
 116                                 permille += *end - '0';
 117                                 /* .. and ignore any further digits */
 118                                 while (isdigit(*++end))
 119                                         ; /* nothing */
 120                         }
 121                         if (!*end)
 122                                 options->dirstat_permille = permille;
 123                         else {
 124                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 125                                             p);
 126                                 ret++;
 127                         }
 128                 } else {
 129                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 130                         ret++;
 131                 }
 132
 133         }
 134         string_list_clear(&params, 0);
 135         free(params_copy);
 136         return ret;
 137 }
 138
 139 static int parse_submodule_params(struct diff_options *options, const char *value)
 140 {
 141         if (!strcmp(value, "log"))
 142                 options->submodule_format = DIFF_SUBMODULE_LOG;
 143         else if (!strcmp(value, "short"))
 144                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 145         else if (!strcmp(value, "diff"))
 146                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 147         else
 148                 return -1;
 149         return 0;
 150 }
 151
 152 static int git_config_rename(const char *var, const char *value)
 153 {
 154         if (!value)
 155                 return DIFF_DETECT_RENAME;
 156         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 157                 return  DIFF_DETECT_COPY;
 158         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 159 }
 160
 161 long parse_algorithm_value(const char *value)
 162 {
 163         if (!value)
 164                 return -1;
 165         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 166                 return 0;
 167         else if (!strcasecmp(value, "minimal"))
 168                 return XDF_NEED_MINIMAL;
 169         else if (!strcasecmp(value, "patience"))
 170                 return XDF_PATIENCE_DIFF;
 171         else if (!strcasecmp(value, "histogram"))
 172                 return XDF_HISTOGRAM_DIFF;
 173         return -1;
 174 }
 175
 176 static int parse_one_token(const char **arg, const char *token)
 177 {
 178         const char *rest;
 179         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 180                 *arg = rest;
 181                 return 1;
 182         }
 183         return 0;
 184 }
 185
 186 static int parse_ws_error_highlight(const char *arg)
 187 {
 188         const char *orig_arg = arg;
 189         unsigned val = 0;
 190
 191         while (*arg) {
 192                 if (parse_one_token(&arg, "none"))
 193                         val = 0;
 194                 else if (parse_one_token(&arg, "default"))
 195                         val = WSEH_NEW;
 196                 else if (parse_one_token(&arg, "all"))
 197                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 198                 else if (parse_one_token(&arg, "new"))
 199                         val |= WSEH_NEW;
 200                 else if (parse_one_token(&arg, "old"))
 201                         val |= WSEH_OLD;
 202                 else if (parse_one_token(&arg, "context"))
 203                         val |= WSEH_CONTEXT;
 204                 else {
 205                         return -1 - (int)(arg - orig_arg);
 206                 }
 207                 if (*arg)
 208                         arg++;
 209         }
 210         return val;
 211 }
 212
 213 /*
 214  * These are to give UI layer defaults.
 215  * The core-level commands such as git-diff-files should
 216  * never be affected by the setting of diff.renames
 217  * the user happens to have in the configuration file.
 218  */
 219 void init_diff_ui_defaults(void)
 220 {
 221         diff_detect_rename_default = 1;
 222 }
 223
 224 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 225 {
 226         if (!strcmp(var, "diff.indentheuristic"))
 227                 diff_indent_heuristic = git_config_bool(var, value);
 228         return 0;
 229 }
 230
 231 int git_diff_ui_config(const char *var, const char *value, void *cb)
 232 {
 233         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 234                 diff_use_color_default = git_config_colorbool(var, value);
 235                 return 0;
 236         }
 237         if (!strcmp(var, "diff.context")) {
 238                 diff_context_default = git_config_int(var, value);
 239                 if (diff_context_default < 0)
 240                         return -1;
 241                 return 0;
 242         }
 243         if (!strcmp(var, "diff.interhunkcontext")) {
 244                 diff_interhunk_context_default = git_config_int(var, value);
 245                 if (diff_interhunk_context_default < 0)
 246                         return -1;
 247                 return 0;
 248         }
 249         if (!strcmp(var, "diff.renames")) {
 250                 diff_detect_rename_default = git_config_rename(var, value);
 251                 return 0;
 252         }
 253         if (!strcmp(var, "diff.autorefreshindex")) {
 254                 diff_auto_refresh_index = git_config_bool(var, value);
 255                 return 0;
 256         }
 257         if (!strcmp(var, "diff.mnemonicprefix")) {
 258                 diff_mnemonic_prefix = git_config_bool(var, value);
 259                 return 0;
 260         }
 261         if (!strcmp(var, "diff.noprefix")) {
 262                 diff_no_prefix = git_config_bool(var, value);
 263                 return 0;
 264         }
 265         if (!strcmp(var, "diff.statgraphwidth")) {
 266                 diff_stat_graph_width = git_config_int(var, value);
 267                 return 0;
 268         }
 269         if (!strcmp(var, "diff.external"))
 270                 return git_config_string(&external_diff_cmd_cfg, var, value);
 271         if (!strcmp(var, "diff.wordregex"))
 272                 return git_config_string(&diff_word_regex_cfg, var, value);
 273         if (!strcmp(var, "diff.orderfile"))
 274                 return git_config_pathname(&diff_order_file_cfg, var, value);
 275
 276         if (!strcmp(var, "diff.ignoresubmodules"))
 277                 handle_ignore_submodules_arg(&default_diff_options, value);
 278
 279         if (!strcmp(var, "diff.submodule")) {
 280                 if (parse_submodule_params(&default_diff_options, value))
 281                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 282                                 value);
 283                 return 0;
 284         }
 285
 286         if (!strcmp(var, "diff.algorithm")) {
 287                 diff_algorithm = parse_algorithm_value(value);
 288                 if (diff_algorithm < 0)
 289                         return -1;
 290                 return 0;
 291         }
 292
 293         if (git_diff_heuristic_config(var, value, cb) < 0)
 294                 return -1;
 295
 296         if (!strcmp(var, "diff.wserrorhighlight")) {
 297                 int val = parse_ws_error_highlight(value);
 298                 if (val < 0)
 299                         return -1;
 300                 ws_error_highlight_default = val;
 301                 return 0;
 302         }
 303
 304         if (git_color_config(var, value, cb) < 0)
 305                 return -1;
 306
 307         return git_diff_basic_config(var, value, cb);
 308 }
 309
 310 int git_diff_basic_config(const char *var, const char *value, void *cb)
 311 {
 312         const char *name;
 313
 314         if (!strcmp(var, "diff.renamelimit")) {
 315                 diff_rename_limit_default = git_config_int(var, value);
 316                 return 0;
 317         }
 318
 319         if (userdiff_config(var, value) < 0)
 320                 return -1;
 321
 322         if (skip_prefix(var, "diff.color.", &name) ||
 323             skip_prefix(var, "color.diff.", &name)) {
 324                 int slot = parse_diff_color_slot(name);
 325                 if (slot < 0)
 326                         return 0;
 327                 if (!value)
 328                         return config_error_nonbool(var);
 329                 return color_parse(value, diff_colors[slot]);
 330         }
 331
 332         /* like GNU diff's --suppress-blank-empty option  */
 333         if (!strcmp(var, "diff.suppressblankempty") ||
 334                         /* for backwards compatibility */
 335                         !strcmp(var, "diff.suppress-blank-empty")) {
 336                 diff_suppress_blank_empty = git_config_bool(var, value);
 337                 return 0;
 338         }
 339
 340         if (!strcmp(var, "diff.dirstat")) {
 341                 struct strbuf errmsg = STRBUF_INIT;
 342                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 343                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 344                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 345                                 errmsg.buf);
 346                 strbuf_release(&errmsg);
 347                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 348                 return 0;
 349         }
 350
 351         if (starts_with(var, "submodule."))
 352                 return parse_submodule_config_option(var, value);
 353
 354         return git_default_config(var, value, cb);
 355 }
 356
 357 static char *quote_two(const char *one, const char *two)
 358 {
 359         int need_one = quote_c_style(one, NULL, NULL, 1);
 360         int need_two = quote_c_style(two, NULL, NULL, 1);
 361         struct strbuf res = STRBUF_INIT;
 362
 363         if (need_one + need_two) {
 364                 strbuf_addch(&res, '"');
 365                 quote_c_style(one, &res, NULL, 1);
 366                 quote_c_style(two, &res, NULL, 1);
 367                 strbuf_addch(&res, '"');
 368         } else {
 369                 strbuf_addstr(&res, one);
 370                 strbuf_addstr(&res, two);
 371         }
 372         return strbuf_detach(&res, NULL);
 373 }
 374
 375 static const char *external_diff(void)
 376 {
 377         static const char *external_diff_cmd = NULL;
 378         static int done_preparing = 0;
 379
 380         if (done_preparing)
 381                 return external_diff_cmd;
 382         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 383         if (!external_diff_cmd)
 384                 external_diff_cmd = external_diff_cmd_cfg;
 385         done_preparing = 1;
 386         return external_diff_cmd;
 387 }
 388
 389 /*
 390  * Keep track of files used for diffing. Sometimes such an entry
 391  * refers to a temporary file, sometimes to an existing file, and
 392  * sometimes to "/dev/null".
 393  */
 394 static struct diff_tempfile {
 395         /*
 396          * filename external diff should read from, or NULL if this
 397          * entry is currently not in use:
 398          */
 399         const char *name;
 400
 401         char hex[GIT_MAX_HEXSZ + 1];
 402         char mode[10];
 403
 404         /*
 405          * If this diff_tempfile instance refers to a temporary file,
 406          * this tempfile object is used to manage its lifetime.
 407          */
 408         struct tempfile tempfile;
 409 } diff_temp[2];
 410
 411 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 412
 413 struct emit_callback {
 414         int color_diff;
 415         unsigned ws_rule;
 416         int blank_at_eof_in_preimage;
 417         int blank_at_eof_in_postimage;
 418         int lno_in_preimage;
 419         int lno_in_postimage;
 420         sane_truncate_fn truncate;
 421         const char **label_path;
 422         struct diff_words_data *diff_words;
 423         struct diff_options *opt;
 424         struct strbuf *header;
 425 };
 426
 427 static int count_lines(const char *data, int size)
 428 {
 429         int count, ch, completely_empty = 1, nl_just_seen = 0;
 430         count = 0;
 431         while (0 < size--) {
 432                 ch = *data++;
 433                 if (ch == '\n') {
 434                         count++;
 435                         nl_just_seen = 1;
 436                         completely_empty = 0;
 437                 }
 438                 else {
 439                         nl_just_seen = 0;
 440                         completely_empty = 0;
 441                 }
 442         }
 443         if (completely_empty)
 444                 return 0;
 445         if (!nl_just_seen)
 446                 count++; /* no trailing newline */
 447         return count;
 448 }
 449
 450 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 451 {
 452         if (!DIFF_FILE_VALID(one)) {
 453                 mf->ptr = (char *)""; /* does not matter */
 454                 mf->size = 0;
 455                 return 0;
 456         }
 457         else if (diff_populate_filespec(one, 0))
 458                 return -1;
 459
 460         mf->ptr = one->data;
 461         mf->size = one->size;
 462         return 0;
 463 }
 464
 465 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 466 static unsigned long diff_filespec_size(struct diff_filespec *one)
 467 {
 468         if (!DIFF_FILE_VALID(one))
 469                 return 0;
 470         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 471         return one->size;
 472 }
 473
 474 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 475 {
 476         char *ptr = mf->ptr;
 477         long size = mf->size;
 478         int cnt = 0;
 479
 480         if (!size)
 481                 return cnt;
 482         ptr += size - 1; /* pointing at the very end */
 483         if (*ptr != '\n')
 484                 ; /* incomplete line */
 485         else
 486                 ptr--; /* skip the last LF */
 487         while (mf->ptr < ptr) {
 488                 char *prev_eol;
 489                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 490                         if (*prev_eol == '\n')
 491                                 break;
 492                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 493                         break;
 494                 cnt++;
 495                 ptr = prev_eol - 1;
 496         }
 497         return cnt;
 498 }
 499
 500 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 501                                struct emit_callback *ecbdata)
 502 {
 503         int l1, l2, at;
 504         unsigned ws_rule = ecbdata->ws_rule;
 505         l1 = count_trailing_blank(mf1, ws_rule);
 506         l2 = count_trailing_blank(mf2, ws_rule);
 507         if (l2 <= l1) {
 508                 ecbdata->blank_at_eof_in_preimage = 0;
 509                 ecbdata->blank_at_eof_in_postimage = 0;
 510                 return;
 511         }
 512         at = count_lines(mf1->ptr, mf1->size);
 513         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 514
 515         at = count_lines(mf2->ptr, mf2->size);
 516         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 517 }
 518
 519 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 520                         int first, const char *line, int len)
 521 {
 522         int has_trailing_newline, has_trailing_carriage_return;
 523         int nofirst;
 524         FILE *file = o->file;
 525
 526         fputs(diff_line_prefix(o), file);
 527
 528         if (len == 0) {
 529                 has_trailing_newline = (first == '\n');
 530                 has_trailing_carriage_return = (!has_trailing_newline &&
 531                                                 (first == '\r'));
 532                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 533         } else {
 534                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 535                 if (has_trailing_newline)
 536                         len--;
 537                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 538                 if (has_trailing_carriage_return)
 539                         len--;
 540                 nofirst = 0;
 541         }
 542
 543         if (len || !nofirst) {
 544                 fputs(set, file);
 545                 if (!nofirst)
 546                         fputc(first, file);
 547                 fwrite(line, len, 1, file);
 548                 fputs(reset, file);
 549         }
 550         if (has_trailing_carriage_return)
 551                 fputc('\r', file);
 552         if (has_trailing_newline)
 553                 fputc('\n', file);
 554 }
 555
 556 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 557                       const char *line, int len)
 558 {
 559         emit_line_0(o, set, reset, line[0], line+1, len-1);
 560 }
 561
 562 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 563 {
 564         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 565               ecbdata->blank_at_eof_in_preimage &&
 566               ecbdata->blank_at_eof_in_postimage &&
 567               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 568               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 569                 return 0;
 570         return ws_blank_line(line, len, ecbdata->ws_rule);
 571 }
 572
 573 static void emit_line_checked(const char *reset,
 574                               struct emit_callback *ecbdata,
 575                               const char *line, int len,
 576                               enum color_diff color,
 577                               unsigned ws_error_highlight,
 578                               char sign)
 579 {
 580         const char *set = diff_get_color(ecbdata->color_diff, color);
 581         const char *ws = NULL;
 582
 583         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
 584                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 585                 if (!*ws)
 586                         ws = NULL;
 587         }
 588
 589         if (!ws)
 590                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
 591         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
 592                 /* Blank line at EOF - paint '+' as well */
 593                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
 594         else {
 595                 /* Emit just the prefix, then the rest. */
 596                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
 597                 ws_check_emit(line, len, ecbdata->ws_rule,
 598                               ecbdata->opt->file, set, reset, ws);
 599         }
 600 }
 601
 602 static void emit_add_line(const char *reset,
 603                           struct emit_callback *ecbdata,
 604                           const char *line, int len)
 605 {
 606         emit_line_checked(reset, ecbdata, line, len,
 607                           DIFF_FILE_NEW, WSEH_NEW, '+');
 608 }
 609
 610 static void emit_del_line(const char *reset,
 611                           struct emit_callback *ecbdata,
 612                           const char *line, int len)
 613 {
 614         emit_line_checked(reset, ecbdata, line, len,
 615                           DIFF_FILE_OLD, WSEH_OLD, '-');
 616 }
 617
 618 static void emit_context_line(const char *reset,
 619                               struct emit_callback *ecbdata,
 620                               const char *line, int len)
 621 {
 622         emit_line_checked(reset, ecbdata, line, len,
 623                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
 624 }
 625
 626 static void emit_hunk_header(struct emit_callback *ecbdata,
 627                              const char *line, int len)
 628 {
 629         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 630         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 631         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 632         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 633         static const char atat[2] = { '@', '@' };
 634         const char *cp, *ep;
 635         struct strbuf msgbuf = STRBUF_INIT;
 636         int org_len = len;
 637         int i = 1;
 638
 639         /*
 640          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 641          * it always is at least 10 bytes long.
 642          */
 643         if (len < 10 ||
 644             memcmp(line, atat, 2) ||
 645             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 646                 emit_line(ecbdata->opt, context, reset, line, len);
 647                 return;
 648         }
 649         ep += 2; /* skip over @@ */
 650
 651         /* The hunk header in fraginfo color */
 652         strbuf_addstr(&msgbuf, frag);
 653         strbuf_add(&msgbuf, line, ep - line);
 654         strbuf_addstr(&msgbuf, reset);
 655
 656         /*
 657          * trailing "\r\n"
 658          */
 659         for ( ; i < 3; i++)
 660                 if (line[len - i] == '\r' || line[len - i] == '\n')
 661                         len--;
 662
 663         /* blank before the func header */
 664         for (cp = ep; ep - line < len; ep++)
 665                 if (*ep != ' ' && *ep != '\t')
 666                         break;
 667         if (ep != cp) {
 668                 strbuf_addstr(&msgbuf, context);
 669                 strbuf_add(&msgbuf, cp, ep - cp);
 670                 strbuf_addstr(&msgbuf, reset);
 671         }
 672
 673         if (ep < line + len) {
 674                 strbuf_addstr(&msgbuf, func);
 675                 strbuf_add(&msgbuf, ep, line + len - ep);
 676                 strbuf_addstr(&msgbuf, reset);
 677         }
 678
 679         strbuf_add(&msgbuf, line + len, org_len - len);
 680         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 681         strbuf_release(&msgbuf);
 682 }
 683
 684 static struct diff_tempfile *claim_diff_tempfile(void) {
 685         int i;
 686         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 687                 if (!diff_temp[i].name)
 688                         return diff_temp + i;
 689         die("BUG: diff is failing to clean up its tempfiles");
 690 }
 691
 692 static void remove_tempfile(void)
 693 {
 694         int i;
 695         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 696                 if (is_tempfile_active(&diff_temp[i].tempfile))
 697                         delete_tempfile(&diff_temp[i].tempfile);
 698                 diff_temp[i].name = NULL;
 699         }
 700 }
 701
 702 static void print_line_count(FILE *file, int count)
 703 {
 704         switch (count) {
 705         case 0:
 706                 fprintf(file, "0,0");
 707                 break;
 708         case 1:
 709                 fprintf(file, "1");
 710                 break;
 711         default:
 712                 fprintf(file, "1,%d", count);
 713                 break;
 714         }
 715 }
 716
 717 static void emit_rewrite_lines(struct emit_callback *ecb,
 718                                int prefix, const char *data, int size)
 719 {
 720         const char *endp = NULL;
 721         static const char *nneof = " No newline at end of file\n";
 722         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 723
 724         while (0 < size) {
 725                 int len;
 726
 727                 endp = memchr(data, '\n', size);
 728                 len = endp ? (endp - data + 1) : size;
 729                 if (prefix != '+') {
 730                         ecb->lno_in_preimage++;
 731                         emit_del_line(reset, ecb, data, len);
 732                 } else {
 733                         ecb->lno_in_postimage++;
 734                         emit_add_line(reset, ecb, data, len);
 735                 }
 736                 size -= len;
 737                 data += len;
 738         }
 739         if (!endp) {
 740                 const char *context = diff_get_color(ecb->color_diff,
 741                                                      DIFF_CONTEXT);
 742                 putc('\n', ecb->opt->file);
 743                 emit_line_0(ecb->opt, context, reset, '\\',
 744                             nneof, strlen(nneof));
 745         }
 746 }
 747
 748 static void emit_rewrite_diff(const char *name_a,
 749                               const char *name_b,
 750                               struct diff_filespec *one,
 751                               struct diff_filespec *two,
 752                               struct userdiff_driver *textconv_one,
 753                               struct userdiff_driver *textconv_two,
 754                               struct diff_options *o)
 755 {
 756         int lc_a, lc_b;
 757         const char *name_a_tab, *name_b_tab;
 758         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 759         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 760         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 761         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 762         const char *a_prefix, *b_prefix;
 763         char *data_one, *data_two;
 764         size_t size_one, size_two;
 765         struct emit_callback ecbdata;
 766         const char *line_prefix = diff_line_prefix(o);
 767
 768         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 769                 a_prefix = o->b_prefix;
 770                 b_prefix = o->a_prefix;
 771         } else {
 772                 a_prefix = o->a_prefix;
 773                 b_prefix = o->b_prefix;
 774         }
 775
 776         name_a += (*name_a == '/');
 777         name_b += (*name_b == '/');
 778         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 779         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 780
 781         strbuf_reset(&a_name);
 782         strbuf_reset(&b_name);
 783         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 784         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 785
 786         size_one = fill_textconv(textconv_one, one, &data_one);
 787         size_two = fill_textconv(textconv_two, two, &data_two);
 788
 789         memset(&ecbdata, 0, sizeof(ecbdata));
 790         ecbdata.color_diff = want_color(o->use_color);
 791         ecbdata.ws_rule = whitespace_rule(name_b);
 792         ecbdata.opt = o;
 793         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 794                 mmfile_t mf1, mf2;
 795                 mf1.ptr = (char *)data_one;
 796                 mf2.ptr = (char *)data_two;
 797                 mf1.size = size_one;
 798                 mf2.size = size_two;
 799                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 800         }
 801         ecbdata.lno_in_preimage = 1;
 802         ecbdata.lno_in_postimage = 1;
 803
 804         lc_a = count_lines(data_one, size_one);
 805         lc_b = count_lines(data_two, size_two);
 806         fprintf(o->file,
 807                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 808                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 809                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 810                 line_prefix, fraginfo);
 811         if (!o->irreversible_delete)
 812                 print_line_count(o->file, lc_a);
 813         else
 814                 fprintf(o->file, "?,?");
 815         fprintf(o->file, " +");
 816         print_line_count(o->file, lc_b);
 817         fprintf(o->file, " @@%s\n", reset);
 818         if (lc_a && !o->irreversible_delete)
 819                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 820         if (lc_b)
 821                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 822         if (textconv_one)
 823                 free((char *)data_one);
 824         if (textconv_two)
 825                 free((char *)data_two);
 826 }
 827
 828 struct diff_words_buffer {
 829         mmfile_t text;
 830         long alloc;
 831         struct diff_words_orig {
 832                 const char *begin, *end;
 833         } *orig;
 834         int orig_nr, orig_alloc;
 835 };
 836
 837 static void diff_words_append(char *line, unsigned long len,
 838                 struct diff_words_buffer *buffer)
 839 {
 840         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 841         line++;
 842         len--;
 843         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 844         buffer->text.size += len;
 845         buffer->text.ptr[buffer->text.size] = '\0';
 846 }
 847
 848 struct diff_words_style_elem {
 849         const char *prefix;
 850         const char *suffix;
 851         const char *color; /* NULL; filled in by the setup code if
 852                             * color is enabled */
 853 };
 854
 855 struct diff_words_style {
 856         enum diff_words_type type;
 857         struct diff_words_style_elem new, old, ctx;
 858         const char *newline;
 859 };
 860
 861 static struct diff_words_style diff_words_styles[] = {
 862         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 863         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 864         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 865 };
 866
 867 struct diff_words_data {
 868         struct diff_words_buffer minus, plus;
 869         const char *current_plus;
 870         int last_minus;
 871         struct diff_options *opt;
 872         regex_t *word_regex;
 873         enum diff_words_type type;
 874         struct diff_words_style *style;
 875 };
 876
 877 static int fn_out_diff_words_write_helper(FILE *fp,
 878                                           struct diff_words_style_elem *st_el,
 879                                           const char *newline,
 880                                           size_t count, const char *buf,
 881                                           const char *line_prefix)
 882 {
 883         int print = 0;
 884
 885         while (count) {
 886                 char *p = memchr(buf, '\n', count);
 887                 if (print)
 888                         fputs(line_prefix, fp);
 889                 if (p != buf) {
 890                         if (st_el->color && fputs(st_el->color, fp) < 0)
 891                                 return -1;
 892                         if (fputs(st_el->prefix, fp) < 0 ||
 893                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 894                             fputs(st_el->suffix, fp) < 0)
 895                                 return -1;
 896                         if (st_el->color && *st_el->color
 897                             && fputs(GIT_COLOR_RESET, fp) < 0)
 898                                 return -1;
 899                 }
 900                 if (!p)
 901                         return 0;
 902                 if (fputs(newline, fp) < 0)
 903                         return -1;
 904                 count -= p + 1 - buf;
 905                 buf = p + 1;
 906                 print = 1;
 907         }
 908         return 0;
 909 }
 910
 911 /*
 912  * '--color-words' algorithm can be described as:
 913  *
 914  *   1. collect a the minus/plus lines of a diff hunk, divided into
 915  *      minus-lines and plus-lines;
 916  *
 917  *   2. break both minus-lines and plus-lines into words and
 918  *      place them into two mmfile_t with one word for each line;
 919  *
 920  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 921  *
 922  * And for the common parts of the both file, we output the plus side text.
 923  * diff_words->current_plus is used to trace the current position of the plus file
 924  * which printed. diff_words->last_minus is used to trace the last minus word
 925  * printed.
 926  *
 927  * For '--graph' to work with '--color-words', we need to output the graph prefix
 928  * on each line of color words output. Generally, there are two conditions on
 929  * which we should output the prefix.
 930  *
 931  *   1. diff_words->last_minus == 0 &&
 932  *      diff_words->current_plus == diff_words->plus.text.ptr
 933  *
 934  *      that is: the plus text must start as a new line, and if there is no minus
 935  *      word printed, a graph prefix must be printed.
 936  *
 937  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 938  *      *(diff_words->current_plus - 1) == '\n'
 939  *
 940  *      that is: a graph prefix must be printed following a '\n'
 941  */
 942 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 943 {
 944         if ((diff_words->last_minus == 0 &&
 945                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 946                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 947                 *(diff_words->current_plus - 1) == '\n')) {
 948                 return 1;
 949         } else {
 950                 return 0;
 951         }
 952 }
 953
 954 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 955 {
 956         struct diff_words_data *diff_words = priv;
 957         struct diff_words_style *style = diff_words->style;
 958         int minus_first, minus_len, plus_first, plus_len;
 959         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 960         struct diff_options *opt = diff_words->opt;
 961         const char *line_prefix;
 962
 963         if (line[0] != '@' || parse_hunk_header(line, len,
 964                         &minus_first, &minus_len, &plus_first, &plus_len))
 965                 return;
 966
 967         assert(opt);
 968         line_prefix = diff_line_prefix(opt);
 969
 970         /* POSIX requires that first be decremented by one if len == 0... */
 971         if (minus_len) {
 972                 minus_begin = diff_words->minus.orig[minus_first].begin;
 973                 minus_end =
 974                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 975         } else
 976                 minus_begin = minus_end =
 977                         diff_words->minus.orig[minus_first].end;
 978
 979         if (plus_len) {
 980                 plus_begin = diff_words->plus.orig[plus_first].begin;
 981                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 982         } else
 983                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 984
 985         if (color_words_output_graph_prefix(diff_words)) {
 986                 fputs(line_prefix, diff_words->opt->file);
 987         }
 988         if (diff_words->current_plus != plus_begin) {
 989                 fn_out_diff_words_write_helper(diff_words->opt->file,
 990                                 &style->ctx, style->newline,
 991                                 plus_begin - diff_words->current_plus,
 992                                 diff_words->current_plus, line_prefix);
 993                 if (*(plus_begin - 1) == '\n')
 994                         fputs(line_prefix, diff_words->opt->file);
 995         }
 996         if (minus_begin != minus_end) {
 997                 fn_out_diff_words_write_helper(diff_words->opt->file,
 998                                 &style->old, style->newline,
 999                                 minus_end - minus_begin, minus_begin,
1000                                 line_prefix);
1001         }
1002         if (plus_begin != plus_end) {
1003                 fn_out_diff_words_write_helper(diff_words->opt->file,
1004                                 &style->new, style->newline,
1005                                 plus_end - plus_begin, plus_begin,
1006                                 line_prefix);
1007         }
1008
1009         diff_words->current_plus = plus_end;
1010         diff_words->last_minus = minus_first;
1011 }
1012
1013 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1014 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1015                 int *begin, int *end)
1016 {
1017         if (word_regex && *begin < buffer->size) {
1018                 regmatch_t match[1];
1019                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1020                                  buffer->size - *begin, 1, match, 0)) {
1021                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1022                                         '\n', match[0].rm_eo - match[0].rm_so);
1023                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1024                         *begin += match[0].rm_so;
1025                         return *begin >= *end;
1026                 }
1027                 return -1;
1028         }
1029
1030         /* find the next word */
1031         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1032                 (*begin)++;
1033         if (*begin >= buffer->size)
1034                 return -1;
1035
1036         /* find the end of the word */
1037         *end = *begin + 1;
1038         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1039                 (*end)++;
1040
1041         return 0;
1042 }
1043
1044 /*
1045  * This function splits the words in buffer->text, stores the list with
1046  * newline separator into out, and saves the offsets of the original words
1047  * in buffer->orig.
1048  */
1049 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1050                 regex_t *word_regex)
1051 {
1052         int i, j;
1053         long alloc = 0;
1054
1055         out->size = 0;
1056         out->ptr = NULL;
1057
1058         /* fake an empty "0th" word */
1059         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1060         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1061         buffer->orig_nr = 1;
1062
1063         for (i = 0; i < buffer->text.size; i++) {
1064                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1065                         return;
1066
1067                 /* store original boundaries */
1068                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1069                                 buffer->orig_alloc);
1070                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1071                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1072                 buffer->orig_nr++;
1073
1074                 /* store one word */
1075                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1076                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1077                 out->ptr[out->size + j - i] = '\n';
1078                 out->size += j - i + 1;
1079
1080                 i = j - 1;
1081         }
1082 }
1083
1084 /* this executes the word diff on the accumulated buffers */
1085 static void diff_words_show(struct diff_words_data *diff_words)
1086 {
1087         xpparam_t xpp;
1088         xdemitconf_t xecfg;
1089         mmfile_t minus, plus;
1090         struct diff_words_style *style = diff_words->style;
1091
1092         struct diff_options *opt = diff_words->opt;
1093         const char *line_prefix;
1094
1095         assert(opt);
1096         line_prefix = diff_line_prefix(opt);
1097
1098         /* special case: only removal */
1099         if (!diff_words->plus.text.size) {
1100                 fputs(line_prefix, diff_words->opt->file);
1101                 fn_out_diff_words_write_helper(diff_words->opt->file,
1102                         &style->old, style->newline,
1103                         diff_words->minus.text.size,
1104                         diff_words->minus.text.ptr, line_prefix);
1105                 diff_words->minus.text.size = 0;
1106                 return;
1107         }
1108
1109         diff_words->current_plus = diff_words->plus.text.ptr;
1110         diff_words->last_minus = 0;
1111
1112         memset(&xpp, 0, sizeof(xpp));
1113         memset(&xecfg, 0, sizeof(xecfg));
1114         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1115         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1116         xpp.flags = 0;
1117         /* as only the hunk header will be parsed, we need a 0-context */
1118         xecfg.ctxlen = 0;
1119         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1120                           &xpp, &xecfg))
1121                 die("unable to generate word diff");
1122         free(minus.ptr);
1123         free(plus.ptr);
1124         if (diff_words->current_plus != diff_words->plus.text.ptr +
1125                         diff_words->plus.text.size) {
1126                 if (color_words_output_graph_prefix(diff_words))
1127                         fputs(line_prefix, diff_words->opt->file);
1128                 fn_out_diff_words_write_helper(diff_words->opt->file,
1129                         &style->ctx, style->newline,
1130                         diff_words->plus.text.ptr + diff_words->plus.text.size
1131                         - diff_words->current_plus, diff_words->current_plus,
1132                         line_prefix);
1133         }
1134         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1135 }
1136
1137 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1138 static void diff_words_flush(struct emit_callback *ecbdata)
1139 {
1140         if (ecbdata->diff_words->minus.text.size ||
1141             ecbdata->diff_words->plus.text.size)
1142                 diff_words_show(ecbdata->diff_words);
1143 }
1144
1145 static void diff_filespec_load_driver(struct diff_filespec *one)
1146 {
1147         /* Use already-loaded driver */
1148         if (one->driver)
1149                 return;
1150
1151         if (S_ISREG(one->mode))
1152                 one->driver = userdiff_find_by_path(one->path);
1153
1154         /* Fallback to default settings */
1155         if (!one->driver)
1156                 one->driver = userdiff_find_by_name("default");
1157 }
1158
1159 static const char *userdiff_word_regex(struct diff_filespec *one)
1160 {
1161         diff_filespec_load_driver(one);
1162         return one->driver->word_regex;
1163 }
1164
1165 static void init_diff_words_data(struct emit_callback *ecbdata,
1166                                  struct diff_options *orig_opts,
1167                                  struct diff_filespec *one,
1168                                  struct diff_filespec *two)
1169 {
1170         int i;
1171         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1172         memcpy(o, orig_opts, sizeof(struct diff_options));
1173
1174         ecbdata->diff_words =
1175                 xcalloc(1, sizeof(struct diff_words_data));
1176         ecbdata->diff_words->type = o->word_diff;
1177         ecbdata->diff_words->opt = o;
1178         if (!o->word_regex)
1179                 o->word_regex = userdiff_word_regex(one);
1180         if (!o->word_regex)
1181                 o->word_regex = userdiff_word_regex(two);
1182         if (!o->word_regex)
1183                 o->word_regex = diff_word_regex_cfg;
1184         if (o->word_regex) {
1185                 ecbdata->diff_words->word_regex = (regex_t *)
1186                         xmalloc(sizeof(regex_t));
1187                 if (regcomp(ecbdata->diff_words->word_regex,
1188                             o->word_regex,
1189                             REG_EXTENDED | REG_NEWLINE))
1190                         die ("Invalid regular expression: %s",
1191                              o->word_regex);
1192         }
1193         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1194                 if (o->word_diff == diff_words_styles[i].type) {
1195                         ecbdata->diff_words->style =
1196                                 &diff_words_styles[i];
1197                         break;
1198                 }
1199         }
1200         if (want_color(o->use_color)) {
1201                 struct diff_words_style *st = ecbdata->diff_words->style;
1202                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1203                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1204                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1205         }
1206 }
1207
1208 static void free_diff_words_data(struct emit_callback *ecbdata)
1209 {
1210         if (ecbdata->diff_words) {
1211                 diff_words_flush(ecbdata);
1212                 free (ecbdata->diff_words->opt);
1213                 free (ecbdata->diff_words->minus.text.ptr);
1214                 free (ecbdata->diff_words->minus.orig);
1215                 free (ecbdata->diff_words->plus.text.ptr);
1216                 free (ecbdata->diff_words->plus.orig);
1217                 if (ecbdata->diff_words->word_regex) {
1218                         regfree(ecbdata->diff_words->word_regex);
1219                         free(ecbdata->diff_words->word_regex);
1220                 }
1221                 free(ecbdata->diff_words);
1222                 ecbdata->diff_words = NULL;
1223         }
1224 }
1225
1226 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1227 {
1228         if (want_color(diff_use_color))
1229                 return diff_colors[ix];
1230         return "";
1231 }
1232
1233 const char *diff_line_prefix(struct diff_options *opt)
1234 {
1235         struct strbuf *msgbuf;
1236         if (!opt->output_prefix)
1237                 return "";
1238
1239         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1240         return msgbuf->buf;
1241 }
1242
1243 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1244 {
1245         const char *cp;
1246         unsigned long allot;
1247         size_t l = len;
1248
1249         if (ecb->truncate)
1250                 return ecb->truncate(line, len);
1251         cp = line;
1252         allot = l;
1253         while (0 < l) {
1254                 (void) utf8_width(&cp, &l);
1255                 if (!cp)
1256                         break; /* truncated in the middle? */
1257         }
1258         return allot - l;
1259 }
1260
1261 static void find_lno(const char *line, struct emit_callback *ecbdata)
1262 {
1263         const char *p;
1264         ecbdata->lno_in_preimage = 0;
1265         ecbdata->lno_in_postimage = 0;
1266         p = strchr(line, '-');
1267         if (!p)
1268                 return; /* cannot happen */
1269         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1270         p = strchr(p, '+');
1271         if (!p)
1272                 return; /* cannot happen */
1273         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1274 }
1275
1276 static void fn_out_consume(void *priv, char *line, unsigned long len)
1277 {
1278         struct emit_callback *ecbdata = priv;
1279         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1280         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1281         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1282         struct diff_options *o = ecbdata->opt;
1283         const char *line_prefix = diff_line_prefix(o);
1284
1285         o->found_changes = 1;
1286
1287         if (ecbdata->header) {
1288                 fprintf(o->file, "%s", ecbdata->header->buf);
1289                 strbuf_reset(ecbdata->header);
1290                 ecbdata->header = NULL;
1291         }
1292
1293         if (ecbdata->label_path[0]) {
1294                 const char *name_a_tab, *name_b_tab;
1295
1296                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1297                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1298
1299                 fprintf(o->file, "%s%s--- %s%s%s\n",
1300                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1301                 fprintf(o->file, "%s%s+++ %s%s%s\n",
1302                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1303                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1304         }
1305
1306         if (diff_suppress_blank_empty
1307             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1308                 line[0] = '\n';
1309                 len = 1;
1310         }
1311
1312         if (line[0] == '@') {
1313                 if (ecbdata->diff_words)
1314                         diff_words_flush(ecbdata);
1315                 len = sane_truncate_line(ecbdata, line, len);
1316                 find_lno(line, ecbdata);
1317                 emit_hunk_header(ecbdata, line, len);
1318                 if (line[len-1] != '\n')
1319                         putc('\n', o->file);
1320                 return;
1321         }
1322
1323         if (ecbdata->diff_words) {
1324                 if (line[0] == '-') {
1325                         diff_words_append(line, len,
1326                                           &ecbdata->diff_words->minus);
1327                         return;
1328                 } else if (line[0] == '+') {
1329                         diff_words_append(line, len,
1330                                           &ecbdata->diff_words->plus);
1331                         return;
1332                 } else if (starts_with(line, "\\ ")) {
1333                         /*
1334                          * Eat the "no newline at eof" marker as if we
1335                          * saw a "+" or "-" line with nothing on it,
1336                          * and return without diff_words_flush() to
1337                          * defer processing. If this is the end of
1338                          * preimage, more "+" lines may come after it.
1339                          */
1340                         return;
1341                 }
1342                 diff_words_flush(ecbdata);
1343                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1344                         emit_line(o, context, reset, line, len);
1345                         fputs("~\n", o->file);
1346                 } else {
1347                         /*
1348                          * Skip the prefix character, if any.  With
1349                          * diff_suppress_blank_empty, there may be
1350                          * none.
1351                          */
1352                         if (line[0] != '\n') {
1353                               line++;
1354                               len--;
1355                         }
1356                         emit_line(o, context, reset, line, len);
1357                 }
1358                 return;
1359         }
1360
1361         switch (line[0]) {
1362         case '+':
1363                 ecbdata->lno_in_postimage++;
1364                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1365                 break;
1366         case '-':
1367                 ecbdata->lno_in_preimage++;
1368                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1369                 break;
1370         case ' ':
1371                 ecbdata->lno_in_postimage++;
1372                 ecbdata->lno_in_preimage++;
1373                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1374                 break;
1375         default:
1376                 /* incomplete line at the end */
1377                 ecbdata->lno_in_preimage++;
1378                 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1379                           reset, line, len);
1380                 break;
1381         }
1382 }
1383
1384 static char *pprint_rename(const char *a, const char *b)
1385 {
1386         const char *old = a;
1387         const char *new = b;
1388         struct strbuf name = STRBUF_INIT;
1389         int pfx_length, sfx_length;
1390         int pfx_adjust_for_slash;
1391         int len_a = strlen(a);
1392         int len_b = strlen(b);
1393         int a_midlen, b_midlen;
1394         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1395         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1396
1397         if (qlen_a || qlen_b) {
1398                 quote_c_style(a, &name, NULL, 0);
1399                 strbuf_addstr(&name, " => ");
1400                 quote_c_style(b, &name, NULL, 0);
1401                 return strbuf_detach(&name, NULL);
1402         }
1403
1404         /* Find common prefix */
1405         pfx_length = 0;
1406         while (*old && *new && *old == *new) {
1407                 if (*old == '/')
1408                         pfx_length = old - a + 1;
1409                 old++;
1410                 new++;
1411         }
1412
1413         /* Find common suffix */
1414         old = a + len_a;
1415         new = b + len_b;
1416         sfx_length = 0;
1417         /*
1418          * If there is a common prefix, it must end in a slash.  In
1419          * that case we let this loop run 1 into the prefix to see the
1420          * same slash.
1421          *
1422          * If there is no common prefix, we cannot do this as it would
1423          * underrun the input strings.
1424          */
1425         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1426         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1427                b + pfx_length - pfx_adjust_for_slash <= new &&
1428                *old == *new) {
1429                 if (*old == '/')
1430                         sfx_length = len_a - (old - a);
1431                 old--;
1432                 new--;
1433         }
1434
1435         /*
1436          * pfx{mid-a => mid-b}sfx
1437          * {pfx-a => pfx-b}sfx
1438          * pfx{sfx-a => sfx-b}
1439          * name-a => name-b
1440          */
1441         a_midlen = len_a - pfx_length - sfx_length;
1442         b_midlen = len_b - pfx_length - sfx_length;
1443         if (a_midlen < 0)
1444                 a_midlen = 0;
1445         if (b_midlen < 0)
1446                 b_midlen = 0;
1447
1448         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1449         if (pfx_length + sfx_length) {
1450                 strbuf_add(&name, a, pfx_length);
1451                 strbuf_addch(&name, '{');
1452         }
1453         strbuf_add(&name, a + pfx_length, a_midlen);
1454         strbuf_addstr(&name, " => ");
1455         strbuf_add(&name, b + pfx_length, b_midlen);
1456         if (pfx_length + sfx_length) {
1457                 strbuf_addch(&name, '}');
1458                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1459         }
1460         return strbuf_detach(&name, NULL);
1461 }
1462
1463 struct diffstat_t {
1464         int nr;
1465         int alloc;
1466         struct diffstat_file {
1467                 char *from_name;
1468                 char *name;
1469                 char *print_name;
1470                 unsigned is_unmerged:1;
1471                 unsigned is_binary:1;
1472                 unsigned is_renamed:1;
1473                 unsigned is_interesting:1;
1474                 uintmax_t added, deleted;
1475         } **files;
1476 };
1477
1478 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1479                                           const char *name_a,
1480                                           const char *name_b)
1481 {
1482         struct diffstat_file *x;
1483         x = xcalloc(1, sizeof(*x));
1484         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1485         diffstat->files[diffstat->nr++] = x;
1486         if (name_b) {
1487                 x->from_name = xstrdup(name_a);
1488                 x->name = xstrdup(name_b);
1489                 x->is_renamed = 1;
1490         }
1491         else {
1492                 x->from_name = NULL;
1493                 x->name = xstrdup(name_a);
1494         }
1495         return x;
1496 }
1497
1498 static void diffstat_consume(void *priv, char *line, unsigned long len)
1499 {
1500         struct diffstat_t *diffstat = priv;
1501         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1502
1503         if (line[0] == '+')
1504                 x->added++;
1505         else if (line[0] == '-')
1506                 x->deleted++;
1507 }
1508
1509 const char mime_boundary_leader[] = "------------";
1510
1511 static int scale_linear(int it, int width, int max_change)
1512 {
1513         if (!it)
1514                 return 0;
1515         /*
1516          * make sure that at least one '-' or '+' is printed if
1517          * there is any change to this path. The easiest way is to
1518          * scale linearly as if the alloted width is one column shorter
1519          * than it is, and then add 1 to the result.
1520          */
1521         return 1 + (it * (width - 1) / max_change);
1522 }
1523
1524 static void show_name(FILE *file,
1525                       const char *prefix, const char *name, int len)
1526 {
1527         fprintf(file, " %s%-*s |", prefix, len, name);
1528 }
1529
1530 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1531 {
1532         if (cnt <= 0)
1533                 return;
1534         fprintf(file, "%s", set);
1535         while (cnt--)
1536                 putc(ch, file);
1537         fprintf(file, "%s", reset);
1538 }
1539
1540 static void fill_print_name(struct diffstat_file *file)
1541 {
1542         char *pname;
1543
1544         if (file->print_name)
1545                 return;
1546
1547         if (!file->is_renamed) {
1548                 struct strbuf buf = STRBUF_INIT;
1549                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1550                         pname = strbuf_detach(&buf, NULL);
1551                 } else {
1552                         pname = file->name;
1553                         strbuf_release(&buf);
1554                 }
1555         } else {
1556                 pname = pprint_rename(file->from_name, file->name);
1557         }
1558         file->print_name = pname;
1559 }
1560
1561 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1562 {
1563         struct strbuf sb = STRBUF_INIT;
1564         int ret;
1565
1566         if (!files) {
1567                 assert(insertions == 0 && deletions == 0);
1568                 return fprintf(fp, "%s\n", " 0 files changed");
1569         }
1570
1571         strbuf_addf(&sb,
1572                     (files == 1) ? " %d file changed" : " %d files changed",
1573                     files);
1574
1575         /*
1576          * For binary diff, the caller may want to print "x files
1577          * changed" with insertions == 0 && deletions == 0.
1578          *
1579          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1580          * is probably less confusing (i.e skip over "2 files changed
1581          * but nothing about added/removed lines? Is this a bug in Git?").
1582          */
1583         if (insertions || deletions == 0) {
1584                 strbuf_addf(&sb,
1585                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1586                             insertions);
1587         }
1588
1589         if (deletions || insertions == 0) {
1590                 strbuf_addf(&sb,
1591                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1592                             deletions);
1593         }
1594         strbuf_addch(&sb, '\n');
1595         ret = fputs(sb.buf, fp);
1596         strbuf_release(&sb);
1597         return ret;
1598 }
1599
1600 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1601 {
1602         int i, len, add, del, adds = 0, dels = 0;
1603         uintmax_t max_change = 0, max_len = 0;
1604         int total_files = data->nr, count;
1605         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1606         const char *reset, *add_c, *del_c;
1607         const char *line_prefix = "";
1608         int extra_shown = 0;
1609
1610         if (data->nr == 0)
1611                 return;
1612
1613         line_prefix = diff_line_prefix(options);
1614         count = options->stat_count ? options->stat_count : data->nr;
1615
1616         reset = diff_get_color_opt(options, DIFF_RESET);
1617         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1618         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1619
1620         /*
1621          * Find the longest filename and max number of changes
1622          */
1623         for (i = 0; (i < count) && (i < data->nr); i++) {
1624                 struct diffstat_file *file = data->files[i];
1625                 uintmax_t change = file->added + file->deleted;
1626
1627                 if (!file->is_interesting && (change == 0)) {
1628                         count++; /* not shown == room for one more */
1629                         continue;
1630                 }
1631                 fill_print_name(file);
1632                 len = strlen(file->print_name);
1633                 if (max_len < len)
1634                         max_len = len;
1635
1636                 if (file->is_unmerged) {
1637                         /* "Unmerged" is 8 characters */
1638                         bin_width = bin_width < 8 ? 8 : bin_width;
1639                         continue;
1640                 }
1641                 if (file->is_binary) {
1642                         /* "Bin XXX -> YYY bytes" */
1643                         int w = 14 + decimal_width(file->added)
1644                                 + decimal_width(file->deleted);
1645                         bin_width = bin_width < w ? w : bin_width;
1646                         /* Display change counts aligned with "Bin" */
1647                         number_width = 3;
1648                         continue;
1649                 }
1650
1651                 if (max_change < change)
1652                         max_change = change;
1653         }
1654         count = i; /* where we can stop scanning in data->files[] */
1655
1656         /*
1657          * We have width = stat_width or term_columns() columns total.
1658          * We want a maximum of min(max_len, stat_name_width) for the name part.
1659          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1660          * We also need 1 for " " and 4 + decimal_width(max_change)
1661          * for " | NNNN " and one the empty column at the end, altogether
1662          * 6 + decimal_width(max_change).
1663          *
1664          * If there's not enough space, we will use the smaller of
1665          * stat_name_width (if set) and 5/8*width for the filename,
1666          * and the rest for constant elements + graph part, but no more
1667          * than stat_graph_width for the graph part.
1668          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1669          * for the standard terminal size).
1670          *
1671          * In other words: stat_width limits the maximum width, and
1672          * stat_name_width fixes the maximum width of the filename,
1673          * and is also used to divide available columns if there
1674          * aren't enough.
1675          *
1676          * Binary files are displayed with "Bin XXX -> YYY bytes"
1677          * instead of the change count and graph. This part is treated
1678          * similarly to the graph part, except that it is not
1679          * "scaled". If total width is too small to accommodate the
1680          * guaranteed minimum width of the filename part and the
1681          * separators and this message, this message will "overflow"
1682          * making the line longer than the maximum width.
1683          */
1684
1685         if (options->stat_width == -1)
1686                 width = term_columns() - strlen(line_prefix);
1687         else
1688                 width = options->stat_width ? options->stat_width : 80;
1689         number_width = decimal_width(max_change) > number_width ?
1690                 decimal_width(max_change) : number_width;
1691
1692         if (options->stat_graph_width == -1)
1693                 options->stat_graph_width = diff_stat_graph_width;
1694
1695         /*
1696          * Guarantee 3/8*16==6 for the graph part
1697          * and 5/8*16==10 for the filename part
1698          */
1699         if (width < 16 + 6 + number_width)
1700                 width = 16 + 6 + number_width;
1701
1702         /*
1703          * First assign sizes that are wanted, ignoring available width.
1704          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1705          * starting from "XXX" should fit in graph_width.
1706          */
1707         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1708         if (options->stat_graph_width &&
1709             options->stat_graph_width < graph_width)
1710                 graph_width = options->stat_graph_width;
1711
1712         name_width = (options->stat_name_width > 0 &&
1713                       options->stat_name_width < max_len) ?
1714                 options->stat_name_width : max_len;
1715
1716         /*
1717          * Adjust adjustable widths not to exceed maximum width
1718          */
1719         if (name_width + number_width + 6 + graph_width > width) {
1720                 if (graph_width > width * 3/8 - number_width - 6) {
1721                         graph_width = width * 3/8 - number_width - 6;
1722                         if (graph_width < 6)
1723                                 graph_width = 6;
1724                 }
1725
1726                 if (options->stat_graph_width &&
1727                     graph_width > options->stat_graph_width)
1728                         graph_width = options->stat_graph_width;
1729                 if (name_width > width - number_width - 6 - graph_width)
1730                         name_width = width - number_width - 6 - graph_width;
1731                 else
1732                         graph_width = width - number_width - 6 - name_width;
1733         }
1734
1735         /*
1736          * From here name_width is the width of the name area,
1737          * and graph_width is the width of the graph area.
1738          * max_change is used to scale graph properly.
1739          */
1740         for (i = 0; i < count; i++) {
1741                 const char *prefix = "";
1742                 struct diffstat_file *file = data->files[i];
1743                 char *name = file->print_name;
1744                 uintmax_t added = file->added;
1745                 uintmax_t deleted = file->deleted;
1746                 int name_len;
1747
1748                 if (!file->is_interesting && (added + deleted == 0))
1749                         continue;
1750
1751                 /*
1752                  * "scale" the filename
1753                  */
1754                 len = name_width;
1755                 name_len = strlen(name);
1756                 if (name_width < name_len) {
1757                         char *slash;
1758                         prefix = "...";
1759                         len -= 3;
1760                         name += name_len - len;
1761                         slash = strchr(name, '/');
1762                         if (slash)
1763                                 name = slash;
1764                 }
1765
1766                 if (file->is_binary) {
1767                         fprintf(options->file, "%s", line_prefix);
1768                         show_name(options->file, prefix, name, len);
1769                         fprintf(options->file, " %*s", number_width, "Bin");
1770                         if (!added && !deleted) {
1771                                 putc('\n', options->file);
1772                                 continue;
1773                         }
1774                         fprintf(options->file, " %s%"PRIuMAX"%s",
1775                                 del_c, deleted, reset);
1776                         fprintf(options->file, " -> ");
1777                         fprintf(options->file, "%s%"PRIuMAX"%s",
1778                                 add_c, added, reset);
1779                         fprintf(options->file, " bytes");
1780                         fprintf(options->file, "\n");
1781                         continue;
1782                 }
1783                 else if (file->is_unmerged) {
1784                         fprintf(options->file, "%s", line_prefix);
1785                         show_name(options->file, prefix, name, len);
1786                         fprintf(options->file, " Unmerged\n");
1787                         continue;
1788                 }
1789
1790                 /*
1791                  * scale the add/delete
1792                  */
1793                 add = added;
1794                 del = deleted;
1795
1796                 if (graph_width <= max_change) {
1797                         int total = scale_linear(add + del, graph_width, max_change);
1798                         if (total < 2 && add && del)
1799                                 /* width >= 2 due to the sanity check */
1800                                 total = 2;
1801                         if (add < del) {
1802                                 add = scale_linear(add, graph_width, max_change);
1803                                 del = total - add;
1804                         } else {
1805                                 del = scale_linear(del, graph_width, max_change);
1806                                 add = total - del;
1807                         }
1808                 }
1809                 fprintf(options->file, "%s", line_prefix);
1810                 show_name(options->file, prefix, name, len);
1811                 fprintf(options->file, " %*"PRIuMAX"%s",
1812                         number_width, added + deleted,
1813                         added + deleted ? " " : "");
1814                 show_graph(options->file, '+', add, add_c, reset);
1815                 show_graph(options->file, '-', del, del_c, reset);
1816                 fprintf(options->file, "\n");
1817         }
1818
1819         for (i = 0; i < data->nr; i++) {
1820                 struct diffstat_file *file = data->files[i];
1821                 uintmax_t added = file->added;
1822                 uintmax_t deleted = file->deleted;
1823
1824                 if (file->is_unmerged ||
1825                     (!file->is_interesting && (added + deleted == 0))) {
1826                         total_files--;
1827                         continue;
1828                 }
1829
1830                 if (!file->is_binary) {
1831                         adds += added;
1832                         dels += deleted;
1833                 }
1834                 if (i < count)
1835                         continue;
1836                 if (!extra_shown)
1837                         fprintf(options->file, "%s ...\n", line_prefix);
1838                 extra_shown = 1;
1839         }
1840         fprintf(options->file, "%s", line_prefix);
1841         print_stat_summary(options->file, total_files, adds, dels);
1842 }
1843
1844 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1845 {
1846         int i, adds = 0, dels = 0, total_files = data->nr;
1847
1848         if (data->nr == 0)
1849                 return;
1850
1851         for (i = 0; i < data->nr; i++) {
1852                 int added = data->files[i]->added;
1853                 int deleted= data->files[i]->deleted;
1854
1855                 if (data->files[i]->is_unmerged ||
1856                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1857                         total_files--;
1858                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1859                         adds += added;
1860                         dels += deleted;
1861                 }
1862         }
1863         fprintf(options->file, "%s", diff_line_prefix(options));
1864         print_stat_summary(options->file, total_files, adds, dels);
1865 }
1866
1867 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1868 {
1869         int i;
1870
1871         if (data->nr == 0)
1872                 return;
1873
1874         for (i = 0; i < data->nr; i++) {
1875                 struct diffstat_file *file = data->files[i];
1876
1877                 fprintf(options->file, "%s", diff_line_prefix(options));
1878
1879                 if (file->is_binary)
1880                         fprintf(options->file, "-\t-\t");
1881                 else
1882                         fprintf(options->file,
1883                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1884                                 file->added, file->deleted);
1885                 if (options->line_termination) {
1886                         fill_print_name(file);
1887                         if (!file->is_renamed)
1888                                 write_name_quoted(file->name, options->file,
1889                                                   options->line_termination);
1890                         else {
1891                                 fputs(file->print_name, options->file);
1892                                 putc(options->line_termination, options->file);
1893                         }
1894                 } else {
1895                         if (file->is_renamed) {
1896                                 putc('\0', options->file);
1897                                 write_name_quoted(file->from_name, options->file, '\0');
1898                         }
1899                         write_name_quoted(file->name, options->file, '\0');
1900                 }
1901         }
1902 }
1903
1904 struct dirstat_file {
1905         const char *name;
1906         unsigned long changed;
1907 };
1908
1909 struct dirstat_dir {
1910         struct dirstat_file *files;
1911         int alloc, nr, permille, cumulative;
1912 };
1913
1914 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1915                 unsigned long changed, const char *base, int baselen)
1916 {
1917         unsigned long this_dir = 0;
1918         unsigned int sources = 0;
1919         const char *line_prefix = diff_line_prefix(opt);
1920
1921         while (dir->nr) {
1922                 struct dirstat_file *f = dir->files;
1923                 int namelen = strlen(f->name);
1924                 unsigned long this;
1925                 char *slash;
1926
1927                 if (namelen < baselen)
1928                         break;
1929                 if (memcmp(f->name, base, baselen))
1930                         break;
1931                 slash = strchr(f->name + baselen, '/');
1932                 if (slash) {
1933                         int newbaselen = slash + 1 - f->name;
1934                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1935                         sources++;
1936                 } else {
1937                         this = f->changed;
1938                         dir->files++;
1939                         dir->nr--;
1940                         sources += 2;
1941                 }
1942                 this_dir += this;
1943         }
1944
1945         /*
1946          * We don't report dirstat's for
1947          *  - the top level
1948          *  - or cases where everything came from a single directory
1949          *    under this directory (sources == 1).
1950          */
1951         if (baselen && sources != 1) {
1952                 if (this_dir) {
1953                         int permille = this_dir * 1000 / changed;
1954                         if (permille >= dir->permille) {
1955                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1956                                         permille / 10, permille % 10, baselen, base);
1957                                 if (!dir->cumulative)
1958                                         return 0;
1959                         }
1960                 }
1961         }
1962         return this_dir;
1963 }
1964
1965 static int dirstat_compare(const void *_a, const void *_b)
1966 {
1967         const struct dirstat_file *a = _a;
1968         const struct dirstat_file *b = _b;
1969         return strcmp(a->name, b->name);
1970 }
1971
1972 static void show_dirstat(struct diff_options *options)
1973 {
1974         int i;
1975         unsigned long changed;
1976         struct dirstat_dir dir;
1977         struct diff_queue_struct *q = &diff_queued_diff;
1978
1979         dir.files = NULL;
1980         dir.alloc = 0;
1981         dir.nr = 0;
1982         dir.permille = options->dirstat_permille;
1983         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1984
1985         changed = 0;
1986         for (i = 0; i < q->nr; i++) {
1987                 struct diff_filepair *p = q->queue[i];
1988                 const char *name;
1989                 unsigned long copied, added, damage;
1990                 int content_changed;
1991
1992                 name = p->two->path ? p->two->path : p->one->path;
1993
1994                 if (p->one->oid_valid && p->two->oid_valid)
1995                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
1996                 else
1997                         content_changed = 1;
1998
1999                 if (!content_changed) {
2000                         /*
2001                          * The SHA1 has not changed, so pre-/post-content is
2002                          * identical. We can therefore skip looking at the
2003                          * file contents altogether.
2004                          */
2005                         damage = 0;
2006                         goto found_damage;
2007                 }
2008
2009                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2010                         /*
2011                          * In --dirstat-by-file mode, we don't really need to
2012                          * look at the actual file contents at all.
2013                          * The fact that the SHA1 changed is enough for us to
2014                          * add this file to the list of results
2015                          * (with each file contributing equal damage).
2016                          */
2017                         damage = 1;
2018                         goto found_damage;
2019                 }
2020
2021                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2022                         diff_populate_filespec(p->one, 0);
2023                         diff_populate_filespec(p->two, 0);
2024                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2025                                                &copied, &added);
2026                         diff_free_filespec_data(p->one);
2027                         diff_free_filespec_data(p->two);
2028                 } else if (DIFF_FILE_VALID(p->one)) {
2029                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2030                         copied = added = 0;
2031                         diff_free_filespec_data(p->one);
2032                 } else if (DIFF_FILE_VALID(p->two)) {
2033                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2034                         copied = 0;
2035                         added = p->two->size;
2036                         diff_free_filespec_data(p->two);
2037                 } else
2038                         continue;
2039
2040                 /*
2041                  * Original minus copied is the removed material,
2042                  * added is the new material.  They are both damages
2043                  * made to the preimage.
2044                  * If the resulting damage is zero, we know that
2045                  * diffcore_count_changes() considers the two entries to
2046                  * be identical, but since content_changed is true, we
2047                  * know that there must have been _some_ kind of change,
2048                  * so we force all entries to have damage > 0.
2049                  */
2050                 damage = (p->one->size - copied) + added;
2051                 if (!damage)
2052                         damage = 1;
2053
2054 found_damage:
2055                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2056                 dir.files[dir.nr].name = name;
2057                 dir.files[dir.nr].changed = damage;
2058                 changed += damage;
2059                 dir.nr++;
2060         }
2061
2062         /* This can happen even with many files, if everything was renames */
2063         if (!changed)
2064                 return;
2065
2066         /* Show all directories with more than x% of the changes */
2067         QSORT(dir.files, dir.nr, dirstat_compare);
2068         gather_dirstat(options, &dir, changed, "", 0);
2069 }
2070
2071 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2072 {
2073         int i;
2074         unsigned long changed;
2075         struct dirstat_dir dir;
2076
2077         if (data->nr == 0)
2078                 return;
2079
2080         dir.files = NULL;
2081         dir.alloc = 0;
2082         dir.nr = 0;
2083         dir.permille = options->dirstat_permille;
2084         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2085
2086         changed = 0;
2087         for (i = 0; i < data->nr; i++) {
2088                 struct diffstat_file *file = data->files[i];
2089                 unsigned long damage = file->added + file->deleted;
2090                 if (file->is_binary)
2091                         /*
2092                          * binary files counts bytes, not lines. Must find some
2093                          * way to normalize binary bytes vs. textual lines.
2094                          * The following heuristic assumes that there are 64
2095                          * bytes per "line".
2096                          * This is stupid and ugly, but very cheap...
2097                          */
2098                         damage = (damage + 63) / 64;
2099                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2100                 dir.files[dir.nr].name = file->name;
2101                 dir.files[dir.nr].changed = damage;
2102                 changed += damage;
2103                 dir.nr++;
2104         }
2105
2106         /* This can happen even with many files, if everything was renames */
2107         if (!changed)
2108                 return;
2109
2110         /* Show all directories with more than x% of the changes */
2111         QSORT(dir.files, dir.nr, dirstat_compare);
2112         gather_dirstat(options, &dir, changed, "", 0);
2113 }
2114
2115 static void free_diffstat_info(struct diffstat_t *diffstat)
2116 {
2117         int i;
2118         for (i = 0; i < diffstat->nr; i++) {
2119                 struct diffstat_file *f = diffstat->files[i];
2120                 if (f->name != f->print_name)
2121                         free(f->print_name);
2122                 free(f->name);
2123                 free(f->from_name);
2124                 free(f);
2125         }
2126         free(diffstat->files);
2127 }
2128
2129 struct checkdiff_t {
2130         const char *filename;
2131         int lineno;
2132         int conflict_marker_size;
2133         struct diff_options *o;
2134         unsigned ws_rule;
2135         unsigned status;
2136 };
2137
2138 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2139 {
2140         char firstchar;
2141         int cnt;
2142
2143         if (len < marker_size + 1)
2144                 return 0;
2145         firstchar = line[0];
2146         switch (firstchar) {
2147         case '=': case '>': case '<': case '|':
2148                 break;
2149         default:
2150                 return 0;
2151         }
2152         for (cnt = 1; cnt < marker_size; cnt++)
2153                 if (line[cnt] != firstchar)
2154                         return 0;
2155         /* line[1] thru line[marker_size-1] are same as firstchar */
2156         if (len < marker_size + 1 || !isspace(line[marker_size]))
2157                 return 0;
2158         return 1;
2159 }
2160
2161 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2162 {
2163         struct checkdiff_t *data = priv;
2164         int marker_size = data->conflict_marker_size;
2165         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2166         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2167         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2168         char *err;
2169         const char *line_prefix;
2170
2171         assert(data->o);
2172         line_prefix = diff_line_prefix(data->o);
2173
2174         if (line[0] == '+') {
2175                 unsigned bad;
2176                 data->lineno++;
2177                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2178                         data->status |= 1;
2179                         fprintf(data->o->file,
2180                                 "%s%s:%d: leftover conflict marker\n",
2181                                 line_prefix, data->filename, data->lineno);
2182                 }
2183                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2184                 if (!bad)
2185                         return;
2186                 data->status |= bad;
2187                 err = whitespace_error_string(bad);
2188                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2189                         line_prefix, data->filename, data->lineno, err);
2190                 free(err);
2191                 emit_line(data->o, set, reset, line, 1);
2192                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2193                               data->o->file, set, reset, ws);
2194         } else if (line[0] == ' ') {
2195                 data->lineno++;
2196         } else if (line[0] == '@') {
2197                 char *plus = strchr(line, '+');
2198                 if (plus)
2199                         data->lineno = strtol(plus, NULL, 10) - 1;
2200                 else
2201                         die("invalid diff");
2202         }
2203 }
2204
2205 static unsigned char *deflate_it(char *data,
2206                                  unsigned long size,
2207                                  unsigned long *result_size)
2208 {
2209         int bound;
2210         unsigned char *deflated;
2211         git_zstream stream;
2212
2213         git_deflate_init(&stream, zlib_compression_level);
2214         bound = git_deflate_bound(&stream, size);
2215         deflated = xmalloc(bound);
2216         stream.next_out = deflated;
2217         stream.avail_out = bound;
2218
2219         stream.next_in = (unsigned char *)data;
2220         stream.avail_in = size;
2221         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2222                 ; /* nothing */
2223         git_deflate_end(&stream);
2224         *result_size = stream.total_out;
2225         return deflated;
2226 }
2227
2228 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2229                                   const char *prefix)
2230 {
2231         void *cp;
2232         void *delta;
2233         void *deflated;
2234         void *data;
2235         unsigned long orig_size;
2236         unsigned long delta_size;
2237         unsigned long deflate_size;
2238         unsigned long data_size;
2239
2240         /* We could do deflated delta, or we could do just deflated two,
2241          * whichever is smaller.
2242          */
2243         delta = NULL;
2244         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2245         if (one->size && two->size) {
2246                 delta = diff_delta(one->ptr, one->size,
2247                                    two->ptr, two->size,
2248                                    &delta_size, deflate_size);
2249                 if (delta) {
2250                         void *to_free = delta;
2251                         orig_size = delta_size;
2252                         delta = deflate_it(delta, delta_size, &delta_size);
2253                         free(to_free);
2254                 }
2255         }
2256
2257         if (delta && delta_size < deflate_size) {
2258                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2259                 free(deflated);
2260                 data = delta;
2261                 data_size = delta_size;
2262         }
2263         else {
2264                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2265                 free(delta);
2266                 data = deflated;
2267                 data_size = deflate_size;
2268         }
2269
2270         /* emit data encoded in base85 */
2271         cp = data;
2272         while (data_size) {
2273                 int bytes = (52 < data_size) ? 52 : data_size;
2274                 char line[70];
2275                 data_size -= bytes;
2276                 if (bytes <= 26)
2277                         line[0] = bytes + 'A' - 1;
2278                 else
2279                         line[0] = bytes - 26 + 'a' - 1;
2280                 encode_85(line + 1, cp, bytes);
2281                 cp = (char *) cp + bytes;
2282                 fprintf(file, "%s", prefix);
2283                 fputs(line, file);
2284                 fputc('\n', file);
2285         }
2286         fprintf(file, "%s\n", prefix);
2287         free(data);
2288 }
2289
2290 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2291                              const char *prefix)
2292 {
2293         fprintf(file, "%sGIT binary patch\n", prefix);
2294         emit_binary_diff_body(file, one, two, prefix);
2295         emit_binary_diff_body(file, two, one, prefix);
2296 }
2297
2298 int diff_filespec_is_binary(struct diff_filespec *one)
2299 {
2300         if (one->is_binary == -1) {
2301                 diff_filespec_load_driver(one);
2302                 if (one->driver->binary != -1)
2303                         one->is_binary = one->driver->binary;
2304                 else {
2305                         if (!one->data && DIFF_FILE_VALID(one))
2306                                 diff_populate_filespec(one, CHECK_BINARY);
2307                         if (one->is_binary == -1 && one->data)
2308                                 one->is_binary = buffer_is_binary(one->data,
2309                                                 one->size);
2310                         if (one->is_binary == -1)
2311                                 one->is_binary = 0;
2312                 }
2313         }
2314         return one->is_binary;
2315 }
2316
2317 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2318 {
2319         diff_filespec_load_driver(one);
2320         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2321 }
2322
2323 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2324 {
2325         if (!options->a_prefix)
2326                 options->a_prefix = a;
2327         if (!options->b_prefix)
2328                 options->b_prefix = b;
2329 }
2330
2331 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2332 {
2333         if (!DIFF_FILE_VALID(one))
2334                 return NULL;
2335
2336         diff_filespec_load_driver(one);
2337         return userdiff_get_textconv(one->driver);
2338 }
2339
2340 static void builtin_diff(const char *name_a,
2341                          const char *name_b,
2342                          struct diff_filespec *one,
2343                          struct diff_filespec *two,
2344                          const char *xfrm_msg,
2345                          int must_show_header,
2346                          struct diff_options *o,
2347                          int complete_rewrite)
2348 {
2349         mmfile_t mf1, mf2;
2350         const char *lbl[2];
2351         char *a_one, *b_two;
2352         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2353         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2354         const char *a_prefix, *b_prefix;
2355         struct userdiff_driver *textconv_one = NULL;
2356         struct userdiff_driver *textconv_two = NULL;
2357         struct strbuf header = STRBUF_INIT;
2358         const char *line_prefix = diff_line_prefix(o);
2359
2360         diff_set_mnemonic_prefix(o, "a/", "b/");
2361         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2362                 a_prefix = o->b_prefix;
2363                 b_prefix = o->a_prefix;
2364         } else {
2365                 a_prefix = o->a_prefix;
2366                 b_prefix = o->b_prefix;
2367         }
2368
2369         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2370             (!one->mode || S_ISGITLINK(one->mode)) &&
2371             (!two->mode || S_ISGITLINK(two->mode))) {
2372                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2373                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2374                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2375                                 line_prefix,
2376                                 &one->oid, &two->oid,
2377                                 two->dirty_submodule,
2378                                 meta, del, add, reset);
2379                 return;
2380         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2381                    (!one->mode || S_ISGITLINK(one->mode)) &&
2382                    (!two->mode || S_ISGITLINK(two->mode))) {
2383                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2384                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2385                 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2386                                 line_prefix,
2387                                 &one->oid, &two->oid,
2388                                 two->dirty_submodule,
2389                                 meta, del, add, reset, o);
2390                 return;
2391         }
2392
2393         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2394                 textconv_one = get_textconv(one);
2395                 textconv_two = get_textconv(two);
2396         }
2397
2398         /* Never use a non-valid filename anywhere if at all possible */
2399         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2400         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2401
2402         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2403         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2404         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2405         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2406         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2407         if (lbl[0][0] == '/') {
2408                 /* /dev/null */
2409                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2410                 if (xfrm_msg)
2411                         strbuf_addstr(&header, xfrm_msg);
2412                 must_show_header = 1;
2413         }
2414         else if (lbl[1][0] == '/') {
2415                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2416                 if (xfrm_msg)
2417                         strbuf_addstr(&header, xfrm_msg);
2418                 must_show_header = 1;
2419         }
2420         else {
2421                 if (one->mode != two->mode) {
2422                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2423                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2424                         must_show_header = 1;
2425                 }
2426                 if (xfrm_msg)
2427                         strbuf_addstr(&header, xfrm_msg);
2428
2429                 /*
2430                  * we do not run diff between different kind
2431                  * of objects.
2432                  */
2433                 if ((one->mode ^ two->mode) & S_IFMT)
2434                         goto free_ab_and_return;
2435                 if (complete_rewrite &&
2436                     (textconv_one || !diff_filespec_is_binary(one)) &&
2437                     (textconv_two || !diff_filespec_is_binary(two))) {
2438                         fprintf(o->file, "%s", header.buf);
2439                         strbuf_reset(&header);
2440                         emit_rewrite_diff(name_a, name_b, one, two,
2441                                                 textconv_one, textconv_two, o);
2442                         o->found_changes = 1;
2443                         goto free_ab_and_return;
2444                 }
2445         }
2446
2447         if (o->irreversible_delete && lbl[1][0] == '/') {
2448                 fprintf(o->file, "%s", header.buf);
2449                 strbuf_reset(&header);
2450                 goto free_ab_and_return;
2451         } else if (!DIFF_OPT_TST(o, TEXT) &&
2452             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2453               (!textconv_two && diff_filespec_is_binary(two)) )) {
2454                 if (!one->data && !two->data &&
2455                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2456                     !DIFF_OPT_TST(o, BINARY)) {
2457                         if (!oidcmp(&one->oid, &two->oid)) {
2458                                 if (must_show_header)
2459                                         fprintf(o->file, "%s", header.buf);
2460                                 goto free_ab_and_return;
2461                         }
2462                         fprintf(o->file, "%s", header.buf);
2463                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2464                                 line_prefix, lbl[0], lbl[1]);
2465                         goto free_ab_and_return;
2466                 }
2467                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2468                         die("unable to read files to diff");
2469                 /* Quite common confusing case */
2470                 if (mf1.size == mf2.size &&
2471                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2472                         if (must_show_header)
2473                                 fprintf(o->file, "%s", header.buf);
2474                         goto free_ab_and_return;
2475                 }
2476                 fprintf(o->file, "%s", header.buf);
2477                 strbuf_reset(&header);
2478                 if (DIFF_OPT_TST(o, BINARY))
2479                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2480                 else
2481                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2482                                 line_prefix, lbl[0], lbl[1]);
2483                 o->found_changes = 1;
2484         } else {
2485                 /* Crazy xdl interfaces.. */
2486                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2487                 const char *v;
2488                 xpparam_t xpp;
2489                 xdemitconf_t xecfg;
2490                 struct emit_callback ecbdata;
2491                 const struct userdiff_funcname *pe;
2492
2493                 if (must_show_header) {
2494                         fprintf(o->file, "%s", header.buf);
2495                         strbuf_reset(&header);
2496                 }
2497
2498                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2499                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2500
2501                 pe = diff_funcname_pattern(one);
2502                 if (!pe)
2503                         pe = diff_funcname_pattern(two);
2504
2505                 memset(&xpp, 0, sizeof(xpp));
2506                 memset(&xecfg, 0, sizeof(xecfg));
2507                 memset(&ecbdata, 0, sizeof(ecbdata));
2508                 ecbdata.label_path = lbl;
2509                 ecbdata.color_diff = want_color(o->use_color);
2510                 ecbdata.ws_rule = whitespace_rule(name_b);
2511                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2512                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2513                 ecbdata.opt = o;
2514                 ecbdata.header = header.len ? &header : NULL;
2515                 xpp.flags = o->xdl_opts;
2516                 xecfg.ctxlen = o->context;
2517                 xecfg.interhunkctxlen = o->interhunkcontext;
2518                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2519                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2520                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2521                 if (pe)
2522                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2523                 if (!diffopts)
2524                         ;
2525                 else if (skip_prefix(diffopts, "--unified=", &v))
2526                         xecfg.ctxlen = strtoul(v, NULL, 10);
2527                 else if (skip_prefix(diffopts, "-u", &v))
2528                         xecfg.ctxlen = strtoul(v, NULL, 10);
2529                 if (o->word_diff)
2530                         init_diff_words_data(&ecbdata, o, one, two);
2531                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2532                                   &xpp, &xecfg))
2533                         die("unable to generate diff for %s", one->path);
2534                 if (o->word_diff)
2535                         free_diff_words_data(&ecbdata);
2536                 if (textconv_one)
2537                         free(mf1.ptr);
2538                 if (textconv_two)
2539                         free(mf2.ptr);
2540                 xdiff_clear_find_func(&xecfg);
2541         }
2542
2543  free_ab_and_return:
2544         strbuf_release(&header);
2545         diff_free_filespec_data(one);
2546         diff_free_filespec_data(two);
2547         free(a_one);
2548         free(b_two);
2549         return;
2550 }
2551
2552 static void builtin_diffstat(const char *name_a, const char *name_b,
2553                              struct diff_filespec *one,
2554                              struct diff_filespec *two,
2555                              struct diffstat_t *diffstat,
2556                              struct diff_options *o,
2557                              struct diff_filepair *p)
2558 {
2559         mmfile_t mf1, mf2;
2560         struct diffstat_file *data;
2561         int same_contents;
2562         int complete_rewrite = 0;
2563
2564         if (!DIFF_PAIR_UNMERGED(p)) {
2565                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2566                         complete_rewrite = 1;
2567         }
2568
2569         data = diffstat_add(diffstat, name_a, name_b);
2570         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2571
2572         if (!one || !two) {
2573                 data->is_unmerged = 1;
2574                 return;
2575         }
2576
2577         same_contents = !oidcmp(&one->oid, &two->oid);
2578
2579         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2580                 data->is_binary = 1;
2581                 if (same_contents) {
2582                         data->added = 0;
2583                         data->deleted = 0;
2584                 } else {
2585                         data->added = diff_filespec_size(two);
2586                         data->deleted = diff_filespec_size(one);
2587                 }
2588         }
2589
2590         else if (complete_rewrite) {
2591                 diff_populate_filespec(one, 0);
2592                 diff_populate_filespec(two, 0);
2593                 data->deleted = count_lines(one->data, one->size);
2594                 data->added = count_lines(two->data, two->size);
2595         }
2596
2597         else if (!same_contents) {
2598                 /* Crazy xdl interfaces.. */
2599                 xpparam_t xpp;
2600                 xdemitconf_t xecfg;
2601
2602                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2603                         die("unable to read files to diff");
2604
2605                 memset(&xpp, 0, sizeof(xpp));
2606                 memset(&xecfg, 0, sizeof(xecfg));
2607                 xpp.flags = o->xdl_opts;
2608                 xecfg.ctxlen = o->context;
2609                 xecfg.interhunkctxlen = o->interhunkcontext;
2610                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2611                                   &xpp, &xecfg))
2612                         die("unable to generate diffstat for %s", one->path);
2613         }
2614
2615         diff_free_filespec_data(one);
2616         diff_free_filespec_data(two);
2617 }
2618
2619 static void builtin_checkdiff(const char *name_a, const char *name_b,
2620                               const char *attr_path,
2621                               struct diff_filespec *one,
2622                               struct diff_filespec *two,
2623                               struct diff_options *o)
2624 {
2625         mmfile_t mf1, mf2;
2626         struct checkdiff_t data;
2627
2628         if (!two)
2629                 return;
2630
2631         memset(&data, 0, sizeof(data));
2632         data.filename = name_b ? name_b : name_a;
2633         data.lineno = 0;
2634         data.o = o;
2635         data.ws_rule = whitespace_rule(attr_path);
2636         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2637
2638         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2639                 die("unable to read files to diff");
2640
2641         /*
2642          * All the other codepaths check both sides, but not checking
2643          * the "old" side here is deliberate.  We are checking the newly
2644          * introduced changes, and as long as the "new" side is text, we
2645          * can and should check what it introduces.
2646          */
2647         if (diff_filespec_is_binary(two))
2648                 goto free_and_return;
2649         else {
2650                 /* Crazy xdl interfaces.. */
2651                 xpparam_t xpp;
2652                 xdemitconf_t xecfg;
2653
2654                 memset(&xpp, 0, sizeof(xpp));
2655                 memset(&xecfg, 0, sizeof(xecfg));
2656                 xecfg.ctxlen = 1; /* at least one context line */
2657                 xpp.flags = 0;
2658                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2659                                   &xpp, &xecfg))
2660                         die("unable to generate checkdiff for %s", one->path);
2661
2662                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2663                         struct emit_callback ecbdata;
2664                         int blank_at_eof;
2665
2666                         ecbdata.ws_rule = data.ws_rule;
2667                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2668                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2669
2670                         if (blank_at_eof) {
2671                                 static char *err;
2672                                 if (!err)
2673                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2674                                 fprintf(o->file, "%s:%d: %s.\n",
2675                                         data.filename, blank_at_eof, err);
2676                                 data.status = 1; /* report errors */
2677                         }
2678                 }
2679         }
2680  free_and_return:
2681         diff_free_filespec_data(one);
2682         diff_free_filespec_data(two);
2683         if (data.status)
2684                 DIFF_OPT_SET(o, CHECK_FAILED);
2685 }
2686
2687 struct diff_filespec *alloc_filespec(const char *path)
2688 {
2689         struct diff_filespec *spec;
2690
2691         FLEXPTR_ALLOC_STR(spec, path, path);
2692         spec->count = 1;
2693         spec->is_binary = -1;
2694         return spec;
2695 }
2696
2697 void free_filespec(struct diff_filespec *spec)
2698 {
2699         if (!--spec->count) {
2700                 diff_free_filespec_data(spec);
2701                 free(spec);
2702         }
2703 }
2704
2705 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2706                    int sha1_valid, unsigned short mode)
2707 {
2708         if (mode) {
2709                 spec->mode = canon_mode(mode);
2710                 hashcpy(spec->oid.hash, sha1);
2711                 spec->oid_valid = sha1_valid;
2712         }
2713 }
2714
2715 /*
2716  * Given a name and sha1 pair, if the index tells us the file in
2717  * the work tree has that object contents, return true, so that
2718  * prepare_temp_file() does not have to inflate and extract.
2719  */
2720 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2721 {
2722         const struct cache_entry *ce;
2723         struct stat st;
2724         int pos, len;
2725
2726         /*
2727          * We do not read the cache ourselves here, because the
2728          * benchmark with my previous version that always reads cache
2729          * shows that it makes things worse for diff-tree comparing
2730          * two linux-2.6 kernel trees in an already checked out work
2731          * tree.  This is because most diff-tree comparisons deal with
2732          * only a small number of files, while reading the cache is
2733          * expensive for a large project, and its cost outweighs the
2734          * savings we get by not inflating the object to a temporary
2735          * file.  Practically, this code only helps when we are used
2736          * by diff-cache --cached, which does read the cache before
2737          * calling us.
2738          */
2739         if (!active_cache)
2740                 return 0;
2741
2742         /* We want to avoid the working directory if our caller
2743          * doesn't need the data in a normal file, this system
2744          * is rather slow with its stat/open/mmap/close syscalls,
2745          * and the object is contained in a pack file.  The pack
2746          * is probably already open and will be faster to obtain
2747          * the data through than the working directory.  Loose
2748          * objects however would tend to be slower as they need
2749          * to be individually opened and inflated.
2750          */
2751         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2752                 return 0;
2753
2754         /*
2755          * Similarly, if we'd have to convert the file contents anyway, that
2756          * makes the optimization not worthwhile.
2757          */
2758         if (!want_file && would_convert_to_git(name))
2759                 return 0;
2760
2761         len = strlen(name);
2762         pos = cache_name_pos(name, len);
2763         if (pos < 0)
2764                 return 0;
2765         ce = active_cache[pos];
2766
2767         /*
2768          * This is not the sha1 we are looking for, or
2769          * unreusable because it is not a regular file.
2770          */
2771         if (hashcmp(sha1, ce->oid.hash) || !S_ISREG(ce->ce_mode))
2772                 return 0;
2773
2774         /*
2775          * If ce is marked as "assume unchanged", there is no
2776          * guarantee that work tree matches what we are looking for.
2777          */
2778         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2779                 return 0;
2780
2781         /*
2782          * If ce matches the file in the work tree, we can reuse it.
2783          */
2784         if (ce_uptodate(ce) ||
2785             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2786                 return 1;
2787
2788         return 0;
2789 }
2790
2791 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2792 {
2793         struct strbuf buf = STRBUF_INIT;
2794         char *dirty = "";
2795
2796         /* Are we looking at the work tree? */
2797         if (s->dirty_submodule)
2798                 dirty = "-dirty";
2799
2800         strbuf_addf(&buf, "Subproject commit %s%s\n",
2801                     oid_to_hex(&s->oid), dirty);
2802         s->size = buf.len;
2803         if (size_only) {
2804                 s->data = NULL;
2805                 strbuf_release(&buf);
2806         } else {
2807                 s->data = strbuf_detach(&buf, NULL);
2808                 s->should_free = 1;
2809         }
2810         return 0;
2811 }
2812
2813 /*
2814  * While doing rename detection and pickaxe operation, we may need to
2815  * grab the data for the blob (or file) for our own in-core comparison.
2816  * diff_filespec has data and size fields for this purpose.
2817  */
2818 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2819 {
2820         int size_only = flags & CHECK_SIZE_ONLY;
2821         int err = 0;
2822         /*
2823          * demote FAIL to WARN to allow inspecting the situation
2824          * instead of refusing.
2825          */
2826         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2827                                     ? SAFE_CRLF_WARN
2828                                     : safe_crlf);
2829
2830         if (!DIFF_FILE_VALID(s))
2831                 die("internal error: asking to populate invalid file.");
2832         if (S_ISDIR(s->mode))
2833                 return -1;
2834
2835         if (s->data)
2836                 return 0;
2837
2838         if (size_only && 0 < s->size)
2839                 return 0;
2840
2841         if (S_ISGITLINK(s->mode))
2842                 return diff_populate_gitlink(s, size_only);
2843
2844         if (!s->oid_valid ||
2845             reuse_worktree_file(s->path, s->oid.hash, 0)) {
2846                 struct strbuf buf = STRBUF_INIT;
2847                 struct stat st;
2848                 int fd;
2849
2850                 if (lstat(s->path, &st) < 0) {
2851                         if (errno == ENOENT) {
2852                         err_empty:
2853                                 err = -1;
2854                         empty:
2855                                 s->data = (char *)"";
2856                                 s->size = 0;
2857                                 return err;
2858                         }
2859                 }
2860                 s->size = xsize_t(st.st_size);
2861                 if (!s->size)
2862                         goto empty;
2863                 if (S_ISLNK(st.st_mode)) {
2864                         struct strbuf sb = STRBUF_INIT;
2865
2866                         if (strbuf_readlink(&sb, s->path, s->size))
2867                                 goto err_empty;
2868                         s->size = sb.len;
2869                         s->data = strbuf_detach(&sb, NULL);
2870                         s->should_free = 1;
2871                         return 0;
2872                 }
2873
2874                 /*
2875                  * Even if the caller would be happy with getting
2876                  * only the size, we cannot return early at this
2877                  * point if the path requires us to run the content
2878                  * conversion.
2879                  */
2880                 if (size_only && !would_convert_to_git(s->path))
2881                         return 0;
2882
2883                 /*
2884                  * Note: this check uses xsize_t(st.st_size) that may
2885                  * not be the true size of the blob after it goes
2886                  * through convert_to_git().  This may not strictly be
2887                  * correct, but the whole point of big_file_threshold
2888                  * and is_binary check being that we want to avoid
2889                  * opening the file and inspecting the contents, this
2890                  * is probably fine.
2891                  */
2892                 if ((flags & CHECK_BINARY) &&
2893                     s->size > big_file_threshold && s->is_binary == -1) {
2894                         s->is_binary = 1;
2895                         return 0;
2896                 }
2897                 fd = open(s->path, O_RDONLY);
2898                 if (fd < 0)
2899                         goto err_empty;
2900                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2901                 close(fd);
2902                 s->should_munmap = 1;
2903
2904                 /*
2905                  * Convert from working tree format to canonical git format
2906                  */
2907                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2908                         size_t size = 0;
2909                         munmap(s->data, s->size);
2910                         s->should_munmap = 0;
2911                         s->data = strbuf_detach(&buf, &size);
2912                         s->size = size;
2913                         s->should_free = 1;
2914                 }
2915         }
2916         else {
2917                 enum object_type type;
2918                 if (size_only || (flags & CHECK_BINARY)) {
2919                         type = sha1_object_info(s->oid.hash, &s->size);
2920                         if (type < 0)
2921                                 die("unable to read %s",
2922                                     oid_to_hex(&s->oid));
2923                         if (size_only)
2924                                 return 0;
2925                         if (s->size > big_file_threshold && s->is_binary == -1) {
2926                                 s->is_binary = 1;
2927                                 return 0;
2928                         }
2929                 }
2930                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2931                 if (!s->data)
2932                         die("unable to read %s", oid_to_hex(&s->oid));
2933                 s->should_free = 1;
2934         }
2935         return 0;
2936 }
2937
2938 void diff_free_filespec_blob(struct diff_filespec *s)
2939 {
2940         if (s->should_free)
2941                 free(s->data);
2942         else if (s->should_munmap)
2943                 munmap(s->data, s->size);
2944
2945         if (s->should_free || s->should_munmap) {
2946                 s->should_free = s->should_munmap = 0;
2947                 s->data = NULL;
2948         }
2949 }
2950
2951 void diff_free_filespec_data(struct diff_filespec *s)
2952 {
2953         diff_free_filespec_blob(s);
2954         free(s->cnt_data);
2955         s->cnt_data = NULL;
2956 }
2957
2958 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2959                            void *blob,
2960                            unsigned long size,
2961                            const struct object_id *oid,
2962                            int mode)
2963 {
2964         int fd;
2965         struct strbuf buf = STRBUF_INIT;
2966         struct strbuf template = STRBUF_INIT;
2967         char *path_dup = xstrdup(path);
2968         const char *base = basename(path_dup);
2969
2970         /* Generate "XXXXXX_basename.ext" */
2971         strbuf_addstr(&template, "XXXXXX_");
2972         strbuf_addstr(&template, base);
2973
2974         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2975         if (fd < 0)
2976                 die_errno("unable to create temp-file");
2977         if (convert_to_working_tree(path,
2978                         (const char *)blob, (size_t)size, &buf)) {
2979                 blob = buf.buf;
2980                 size = buf.len;
2981         }
2982         if (write_in_full(fd, blob, size) != size)
2983                 die_errno("unable to write temp-file");
2984         close_tempfile(&temp->tempfile);
2985         temp->name = get_tempfile_path(&temp->tempfile);
2986         oid_to_hex_r(temp->hex, oid);
2987         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2988         strbuf_release(&buf);
2989         strbuf_release(&template);
2990         free(path_dup);
2991 }
2992
2993 static struct diff_tempfile *prepare_temp_file(const char *name,
2994                 struct diff_filespec *one)
2995 {
2996         struct diff_tempfile *temp = claim_diff_tempfile();
2997
2998         if (!DIFF_FILE_VALID(one)) {
2999         not_a_valid_file:
3000                 /* A '-' entry produces this for file-2, and
3001                  * a '+' entry produces this for file-1.
3002                  */
3003                 temp->name = "/dev/null";
3004                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3005                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3006                 return temp;
3007         }
3008
3009         if (!S_ISGITLINK(one->mode) &&
3010             (!one->oid_valid ||
3011              reuse_worktree_file(name, one->oid.hash, 1))) {
3012                 struct stat st;
3013                 if (lstat(name, &st) < 0) {
3014                         if (errno == ENOENT)
3015                                 goto not_a_valid_file;
3016                         die_errno("stat(%s)", name);
3017                 }
3018                 if (S_ISLNK(st.st_mode)) {
3019                         struct strbuf sb = STRBUF_INIT;
3020                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3021                                 die_errno("readlink(%s)", name);
3022                         prep_temp_blob(name, temp, sb.buf, sb.len,
3023                                        (one->oid_valid ?
3024                                         &one->oid : &null_oid),
3025                                        (one->oid_valid ?
3026                                         one->mode : S_IFLNK));
3027                         strbuf_release(&sb);
3028                 }
3029                 else {
3030                         /* we can borrow from the file in the work tree */
3031                         temp->name = name;
3032                         if (!one->oid_valid)
3033                                 sha1_to_hex_r(temp->hex, null_sha1);
3034                         else
3035                                 oid_to_hex_r(temp->hex, &one->oid);
3036                         /* Even though we may sometimes borrow the
3037                          * contents from the work tree, we always want
3038                          * one->mode.  mode is trustworthy even when
3039                          * !(one->sha1_valid), as long as
3040                          * DIFF_FILE_VALID(one).
3041                          */
3042                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3043                 }
3044                 return temp;
3045         }
3046         else {
3047                 if (diff_populate_filespec(one, 0))
3048                         die("cannot read data blob for %s", one->path);
3049                 prep_temp_blob(name, temp, one->data, one->size,
3050                                &one->oid, one->mode);
3051         }
3052         return temp;
3053 }
3054
3055 static void add_external_diff_name(struct argv_array *argv,
3056                                    const char *name,
3057                                    struct diff_filespec *df)
3058 {
3059         struct diff_tempfile *temp = prepare_temp_file(name, df);
3060         argv_array_push(argv, temp->name);
3061         argv_array_push(argv, temp->hex);
3062         argv_array_push(argv, temp->mode);
3063 }
3064
3065 /* An external diff command takes:
3066  *
3067  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3068  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3069  *
3070  */
3071 static void run_external_diff(const char *pgm,
3072                               const char *name,
3073                               const char *other,
3074                               struct diff_filespec *one,
3075                               struct diff_filespec *two,
3076                               const char *xfrm_msg,
3077                               int complete_rewrite,
3078                               struct diff_options *o)
3079 {
3080         struct argv_array argv = ARGV_ARRAY_INIT;
3081         struct argv_array env = ARGV_ARRAY_INIT;
3082         struct diff_queue_struct *q = &diff_queued_diff;
3083
3084         argv_array_push(&argv, pgm);
3085         argv_array_push(&argv, name);
3086
3087         if (one && two) {
3088                 add_external_diff_name(&argv, name, one);
3089                 if (!other)
3090                         add_external_diff_name(&argv, name, two);
3091                 else {
3092                         add_external_diff_name(&argv, other, two);
3093                         argv_array_push(&argv, other);
3094                         argv_array_push(&argv, xfrm_msg);
3095                 }
3096         }
3097
3098         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3099         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3100
3101         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3102                 die(_("external diff died, stopping at %s"), name);
3103
3104         remove_tempfile();
3105         argv_array_clear(&argv);
3106         argv_array_clear(&env);
3107 }
3108
3109 static int similarity_index(struct diff_filepair *p)
3110 {
3111         return p->score * 100 / MAX_SCORE;
3112 }
3113
3114 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3115 {
3116         if (startup_info->have_repository)
3117                 return find_unique_abbrev(oid->hash, abbrev);
3118         else {
3119                 char *hex = oid_to_hex(oid);
3120                 if (abbrev < 0)
3121                         abbrev = FALLBACK_DEFAULT_ABBREV;
3122                 if (abbrev > GIT_SHA1_HEXSZ)
3123                         die("BUG: oid abbreviation out of range: %d", abbrev);
3124                 if (abbrev)
3125                         hex[abbrev] = '\0';
3126                 return hex;
3127         }
3128 }
3129
3130 static void fill_metainfo(struct strbuf *msg,
3131                           const char *name,
3132                           const char *other,
3133                           struct diff_filespec *one,
3134                           struct diff_filespec *two,
3135                           struct diff_options *o,
3136                           struct diff_filepair *p,
3137                           int *must_show_header,
3138                           int use_color)
3139 {
3140         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3141         const char *reset = diff_get_color(use_color, DIFF_RESET);
3142         const char *line_prefix = diff_line_prefix(o);
3143
3144         *must_show_header = 1;
3145         strbuf_init(msg, PATH_MAX * 2 + 300);
3146         switch (p->status) {
3147         case DIFF_STATUS_COPIED:
3148                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3149                             line_prefix, set, similarity_index(p));
3150                 strbuf_addf(msg, "%s\n%s%scopy from ",
3151                             reset,  line_prefix, set);
3152                 quote_c_style(name, msg, NULL, 0);
3153                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3154                 quote_c_style(other, msg, NULL, 0);
3155                 strbuf_addf(msg, "%s\n", reset);
3156                 break;
3157         case DIFF_STATUS_RENAMED:
3158                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3159                             line_prefix, set, similarity_index(p));
3160                 strbuf_addf(msg, "%s\n%s%srename from ",
3161                             reset, line_prefix, set);
3162                 quote_c_style(name, msg, NULL, 0);
3163                 strbuf_addf(msg, "%s\n%s%srename to ",
3164                             reset, line_prefix, set);
3165                 quote_c_style(other, msg, NULL, 0);
3166                 strbuf_addf(msg, "%s\n", reset);
3167                 break;
3168         case DIFF_STATUS_MODIFIED:
3169                 if (p->score) {
3170                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3171                                     line_prefix,
3172                                     set, similarity_index(p), reset);
3173                         break;
3174                 }
3175                 /* fallthru */
3176         default:
3177                 *must_show_header = 0;
3178         }
3179         if (one && two && oidcmp(&one->oid, &two->oid)) {
3180                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3181
3182                 if (DIFF_OPT_TST(o, BINARY)) {
3183                         mmfile_t mf;
3184                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3185                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3186                                 abbrev = 40;
3187                 }
3188                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3189                             diff_abbrev_oid(&one->oid, abbrev),
3190                             diff_abbrev_oid(&two->oid, abbrev));
3191                 if (one->mode == two->mode)
3192                         strbuf_addf(msg, " %06o", one->mode);
3193                 strbuf_addf(msg, "%s\n", reset);
3194         }
3195 }
3196
3197 static void run_diff_cmd(const char *pgm,
3198                          const char *name,
3199                          const char *other,
3200                          const char *attr_path,
3201                          struct diff_filespec *one,
3202                          struct diff_filespec *two,
3203                          struct strbuf *msg,
3204                          struct diff_options *o,
3205                          struct diff_filepair *p)
3206 {
3207         const char *xfrm_msg = NULL;
3208         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3209         int must_show_header = 0;
3210
3211
3212         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3213                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3214                 if (drv && drv->external)
3215                         pgm = drv->external;
3216         }
3217
3218         if (msg) {
3219                 /*
3220                  * don't use colors when the header is intended for an
3221                  * external diff driver
3222                  */
3223                 fill_metainfo(msg, name, other, one, two, o, p,
3224                               &must_show_header,
3225                               want_color(o->use_color) && !pgm);
3226                 xfrm_msg = msg->len ? msg->buf : NULL;
3227         }
3228
3229         if (pgm) {
3230                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3231                                   complete_rewrite, o);
3232                 return;
3233         }
3234         if (one && two)
3235                 builtin_diff(name, other ? other : name,
3236                              one, two, xfrm_msg, must_show_header,
3237                              o, complete_rewrite);
3238         else
3239                 fprintf(o->file, "* Unmerged path %s\n", name);
3240 }
3241
3242 static void diff_fill_sha1_info(struct diff_filespec *one)
3243 {
3244         if (DIFF_FILE_VALID(one)) {
3245                 if (!one->oid_valid) {
3246                         struct stat st;
3247                         if (one->is_stdin) {
3248                                 oidclr(&one->oid);
3249                                 return;
3250                         }
3251                         if (lstat(one->path, &st) < 0)
3252                                 die_errno("stat '%s'", one->path);
3253                         if (index_path(one->oid.hash, one->path, &st, 0))
3254                                 die("cannot hash %s", one->path);
3255                 }
3256         }
3257         else
3258                 oidclr(&one->oid);
3259 }
3260
3261 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3262 {
3263         /* Strip the prefix but do not molest /dev/null and absolute paths */
3264         if (*namep && **namep != '/') {
3265                 *namep += prefix_length;
3266                 if (**namep == '/')
3267                         ++*namep;
3268         }
3269         if (*otherp && **otherp != '/') {
3270                 *otherp += prefix_length;
3271                 if (**otherp == '/')
3272                         ++*otherp;
3273         }
3274 }
3275
3276 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3277 {
3278         const char *pgm = external_diff();
3279         struct strbuf msg;
3280         struct diff_filespec *one = p->one;
3281         struct diff_filespec *two = p->two;
3282         const char *name;
3283         const char *other;
3284         const char *attr_path;
3285
3286         name  = p->one->path;
3287         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3288         attr_path = name;
3289         if (o->prefix_length)
3290                 strip_prefix(o->prefix_length, &name, &other);
3291
3292         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3293                 pgm = NULL;
3294
3295         if (DIFF_PAIR_UNMERGED(p)) {
3296                 run_diff_cmd(pgm, name, NULL, attr_path,
3297                              NULL, NULL, NULL, o, p);
3298                 return;
3299         }
3300
3301         diff_fill_sha1_info(one);
3302         diff_fill_sha1_info(two);
3303
3304         if (!pgm &&
3305             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3306             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3307                 /*
3308                  * a filepair that changes between file and symlink
3309                  * needs to be split into deletion and creation.
3310                  */
3311                 struct diff_filespec *null = alloc_filespec(two->path);
3312                 run_diff_cmd(NULL, name, other, attr_path,
3313                              one, null, &msg, o, p);
3314                 free(null);
3315                 strbuf_release(&msg);
3316
3317                 null = alloc_filespec(one->path);
3318                 run_diff_cmd(NULL, name, other, attr_path,
3319                              null, two, &msg, o, p);
3320                 free(null);
3321         }
3322         else
3323                 run_diff_cmd(pgm, name, other, attr_path,
3324                              one, two, &msg, o, p);
3325
3326         strbuf_release(&msg);
3327 }
3328
3329 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3330                          struct diffstat_t *diffstat)
3331 {
3332         const char *name;
3333         const char *other;
3334
3335         if (DIFF_PAIR_UNMERGED(p)) {
3336                 /* unmerged */
3337                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3338                 return;
3339         }
3340
3341         name = p->one->path;
3342         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3343
3344         if (o->prefix_length)
3345                 strip_prefix(o->prefix_length, &name, &other);
3346
3347         diff_fill_sha1_info(p->one);
3348         diff_fill_sha1_info(p->two);
3349
3350         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3351 }
3352
3353 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3354 {
3355         const char *name;
3356         const char *other;
3357         const char *attr_path;
3358
3359         if (DIFF_PAIR_UNMERGED(p)) {
3360                 /* unmerged */
3361                 return;
3362         }
3363
3364         name = p->one->path;
3365         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3366         attr_path = other ? other : name;
3367
3368         if (o->prefix_length)
3369                 strip_prefix(o->prefix_length, &name, &other);
3370
3371         diff_fill_sha1_info(p->one);
3372         diff_fill_sha1_info(p->two);
3373
3374         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3375 }
3376
3377 void diff_setup(struct diff_options *options)
3378 {
3379         memcpy(options, &default_diff_options, sizeof(*options));
3380
3381         options->file = stdout;
3382
3383         options->abbrev = DEFAULT_ABBREV;
3384         options->line_termination = '\n';
3385         options->break_opt = -1;
3386         options->rename_limit = -1;
3387         options->dirstat_permille = diff_dirstat_permille_default;
3388         options->context = diff_context_default;
3389         options->interhunkcontext = diff_interhunk_context_default;
3390         options->ws_error_highlight = ws_error_highlight_default;
3391         DIFF_OPT_SET(options, RENAME_EMPTY);
3392
3393         /* pathchange left =NULL by default */
3394         options->change = diff_change;
3395         options->add_remove = diff_addremove;
3396         options->use_color = diff_use_color_default;
3397         options->detect_rename = diff_detect_rename_default;
3398         options->xdl_opts |= diff_algorithm;
3399         if (diff_indent_heuristic)
3400                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3401
3402         options->orderfile = diff_order_file_cfg;
3403
3404         if (diff_no_prefix) {
3405                 options->a_prefix = options->b_prefix = "";
3406         } else if (!diff_mnemonic_prefix) {
3407                 options->a_prefix = "a/";
3408                 options->b_prefix = "b/";
3409         }
3410 }
3411
3412 void diff_setup_done(struct diff_options *options)
3413 {
3414         int count = 0;
3415
3416         if (options->set_default)
3417                 options->set_default(options);
3418
3419         if (options->output_format & DIFF_FORMAT_NAME)
3420                 count++;
3421         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3422                 count++;
3423         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3424                 count++;
3425         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3426                 count++;
3427         if (count > 1)
3428                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
3429
3430         /*
3431          * Most of the time we can say "there are changes"
3432          * only by checking if there are changed paths, but
3433          * --ignore-whitespace* options force us to look
3434          * inside contents.
3435          */
3436
3437         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3438             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3439             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3440                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3441         else
3442                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3443
3444         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3445                 options->detect_rename = DIFF_DETECT_COPY;
3446
3447         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3448                 options->prefix = NULL;
3449         if (options->prefix)
3450                 options->prefix_length = strlen(options->prefix);
3451         else
3452                 options->prefix_length = 0;
3453
3454         if (options->output_format & (DIFF_FORMAT_NAME |
3455                                       DIFF_FORMAT_NAME_STATUS |
3456                                       DIFF_FORMAT_CHECKDIFF |
3457                                       DIFF_FORMAT_NO_OUTPUT))
3458                 options->output_format &= ~(DIFF_FORMAT_RAW |
3459                                             DIFF_FORMAT_NUMSTAT |
3460                                             DIFF_FORMAT_DIFFSTAT |
3461                                             DIFF_FORMAT_SHORTSTAT |
3462                                             DIFF_FORMAT_DIRSTAT |
3463                                             DIFF_FORMAT_SUMMARY |
3464                                             DIFF_FORMAT_PATCH);
3465
3466         /*
3467          * These cases always need recursive; we do not drop caller-supplied
3468          * recursive bits for other formats here.
3469          */
3470         if (options->output_format & (DIFF_FORMAT_PATCH |
3471                                       DIFF_FORMAT_NUMSTAT |
3472                                       DIFF_FORMAT_DIFFSTAT |
3473                                       DIFF_FORMAT_SHORTSTAT |
3474                                       DIFF_FORMAT_DIRSTAT |
3475                                       DIFF_FORMAT_SUMMARY |
3476                                       DIFF_FORMAT_CHECKDIFF))
3477                 DIFF_OPT_SET(options, RECURSIVE);
3478         /*
3479          * Also pickaxe would not work very well if you do not say recursive
3480          */
3481         if (options->pickaxe)
3482                 DIFF_OPT_SET(options, RECURSIVE);
3483         /*
3484          * When patches are generated, submodules diffed against the work tree
3485          * must be checked for dirtiness too so it can be shown in the output
3486          */
3487         if (options->output_format & DIFF_FORMAT_PATCH)
3488                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3489
3490         if (options->detect_rename && options->rename_limit < 0)
3491                 options->rename_limit = diff_rename_limit_default;
3492         if (options->setup & DIFF_SETUP_USE_CACHE) {
3493                 if (!active_cache)
3494                         /* read-cache does not die even when it fails
3495                          * so it is safe for us to do this here.  Also
3496                          * it does not smudge active_cache or active_nr
3497                          * when it fails, so we do not have to worry about
3498                          * cleaning it up ourselves either.
3499                          */
3500                         read_cache();
3501         }
3502         if (40 < options->abbrev)
3503                 options->abbrev = 40; /* full */
3504
3505         /*
3506          * It does not make sense to show the first hit we happened
3507          * to have found.  It does not make sense not to return with
3508          * exit code in such a case either.
3509          */
3510         if (DIFF_OPT_TST(options, QUICK)) {
3511                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3512                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3513         }
3514
3515         options->diff_path_counter = 0;
3516
3517         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3518                 die(_("--follow requires exactly one pathspec"));
3519 }
3520
3521 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3522 {
3523         char c, *eq;
3524         int len;
3525
3526         if (*arg != '-')
3527                 return 0;
3528         c = *++arg;
3529         if (!c)
3530                 return 0;
3531         if (c == arg_short) {
3532                 c = *++arg;
3533                 if (!c)
3534                         return 1;
3535                 if (val && isdigit(c)) {
3536                         char *end;
3537                         int n = strtoul(arg, &end, 10);
3538                         if (*end)
3539                                 return 0;
3540                         *val = n;
3541                         return 1;
3542                 }
3543                 return 0;
3544         }
3545         if (c != '-')
3546                 return 0;
3547         arg++;
3548         eq = strchrnul(arg, '=');
3549         len = eq - arg;
3550         if (!len || strncmp(arg, arg_long, len))
3551                 return 0;
3552         if (*eq) {
3553                 int n;
3554                 char *end;
3555                 if (!isdigit(*++eq))
3556                         return 0;
3557                 n = strtoul(eq, &end, 10);
3558                 if (*end)
3559                         return 0;
3560                 *val = n;
3561         }
3562         return 1;
3563 }
3564
3565 static int diff_scoreopt_parse(const char *opt);
3566
3567 static inline int short_opt(char opt, const char **argv,
3568                             const char **optarg)
3569 {
3570         const char *arg = argv[0];
3571         if (arg[0] != '-' || arg[1] != opt)
3572                 return 0;
3573         if (arg[2] != '\0') {
3574                 *optarg = arg + 2;
3575                 return 1;
3576         }
3577         if (!argv[1])
3578                 die("Option '%c' requires a value", opt);
3579         *optarg = argv[1];
3580         return 2;
3581 }
3582
3583 int parse_long_opt(const char *opt, const char **argv,
3584                    const char **optarg)
3585 {
3586         const char *arg = argv[0];
3587         if (!skip_prefix(arg, "--", &arg))
3588                 return 0;
3589         if (!skip_prefix(arg, opt, &arg))
3590                 return 0;
3591         if (*arg == '=') { /* stuck form: --option=value */
3592                 *optarg = arg + 1;
3593                 return 1;
3594         }
3595         if (*arg != '\0')
3596                 return 0;
3597         /* separate form: --option value */
3598         if (!argv[1])
3599                 die("Option '--%s' requires a value", opt);
3600         *optarg = argv[1];
3601         return 2;
3602 }
3603
3604 static int stat_opt(struct diff_options *options, const char **av)
3605 {
3606         const char *arg = av[0];
3607         char *end;
3608         int width = options->stat_width;
3609         int name_width = options->stat_name_width;
3610         int graph_width = options->stat_graph_width;
3611         int count = options->stat_count;
3612         int argcount = 1;
3613
3614         if (!skip_prefix(arg, "--stat", &arg))
3615                 die("BUG: stat option does not begin with --stat: %s", arg);
3616         end = (char *)arg;
3617
3618         switch (*arg) {
3619         case '-':
3620                 if (skip_prefix(arg, "-width", &arg)) {
3621                         if (*arg == '=')
3622                                 width = strtoul(arg + 1, &end, 10);
3623                         else if (!*arg && !av[1])
3624                                 die_want_option("--stat-width");
3625                         else if (!*arg) {
3626                                 width = strtoul(av[1], &end, 10);
3627                                 argcount = 2;
3628                         }
3629                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3630                         if (*arg == '=')
3631                                 name_width = strtoul(arg + 1, &end, 10);
3632                         else if (!*arg && !av[1])
3633                                 die_want_option("--stat-name-width");
3634                         else if (!*arg) {
3635                                 name_width = strtoul(av[1], &end, 10);
3636                                 argcount = 2;
3637                         }
3638                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3639                         if (*arg == '=')
3640                                 graph_width = strtoul(arg + 1, &end, 10);
3641                         else if (!*arg && !av[1])
3642                                 die_want_option("--stat-graph-width");
3643                         else if (!*arg) {
3644                                 graph_width = strtoul(av[1], &end, 10);
3645                                 argcount = 2;
3646                         }
3647                 } else if (skip_prefix(arg, "-count", &arg)) {
3648                         if (*arg == '=')
3649                                 count = strtoul(arg + 1, &end, 10);
3650                         else if (!*arg && !av[1])
3651                                 die_want_option("--stat-count");
3652                         else if (!*arg) {
3653                                 count = strtoul(av[1], &end, 10);
3654                                 argcount = 2;
3655                         }
3656                 }
3657                 break;
3658         case '=':
3659                 width = strtoul(arg+1, &end, 10);
3660                 if (*end == ',')
3661                         name_width = strtoul(end+1, &end, 10);
3662                 if (*end == ',')
3663                         count = strtoul(end+1, &end, 10);
3664         }
3665
3666         /* Important! This checks all the error cases! */
3667         if (*end)
3668                 return 0;
3669         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3670         options->stat_name_width = name_width;
3671         options->stat_graph_width = graph_width;
3672         options->stat_width = width;
3673         options->stat_count = count;
3674         return argcount;
3675 }
3676
3677 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3678 {
3679         struct strbuf errmsg = STRBUF_INIT;
3680         if (parse_dirstat_params(options, params, &errmsg))
3681                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3682                     errmsg.buf);
3683         strbuf_release(&errmsg);
3684         /*
3685          * The caller knows a dirstat-related option is given from the command
3686          * line; allow it to say "return this_function();"
3687          */
3688         options->output_format |= DIFF_FORMAT_DIRSTAT;
3689         return 1;
3690 }
3691
3692 static int parse_submodule_opt(struct diff_options *options, const char *value)
3693 {
3694         if (parse_submodule_params(options, value))
3695                 die(_("Failed to parse --submodule option parameter: '%s'"),
3696                         value);
3697         return 1;
3698 }
3699
3700 static const char diff_status_letters[] = {
3701         DIFF_STATUS_ADDED,
3702         DIFF_STATUS_COPIED,
3703         DIFF_STATUS_DELETED,
3704         DIFF_STATUS_MODIFIED,
3705         DIFF_STATUS_RENAMED,
3706         DIFF_STATUS_TYPE_CHANGED,
3707         DIFF_STATUS_UNKNOWN,
3708         DIFF_STATUS_UNMERGED,
3709         DIFF_STATUS_FILTER_AON,
3710         DIFF_STATUS_FILTER_BROKEN,
3711         '\0',
3712 };
3713
3714 static unsigned int filter_bit['Z' + 1];
3715
3716 static void prepare_filter_bits(void)
3717 {
3718         int i;
3719
3720         if (!filter_bit[DIFF_STATUS_ADDED]) {
3721                 for (i = 0; diff_status_letters[i]; i++)
3722                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3723         }
3724 }
3725
3726 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3727 {
3728         return opt->filter & filter_bit[(int) status];
3729 }
3730
3731 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3732 {
3733         int i, optch;
3734
3735         prepare_filter_bits();
3736
3737         /*
3738          * If there is a negation e.g. 'd' in the input, and we haven't
3739          * initialized the filter field with another --diff-filter, start
3740          * from full set of bits, except for AON.
3741          */
3742         if (!opt->filter) {
3743                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3744                         if (optch < 'a' || 'z' < optch)
3745                                 continue;
3746                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3747                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3748                         break;
3749                 }
3750         }
3751
3752         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3753                 unsigned int bit;
3754                 int negate;
3755
3756                 if ('a' <= optch && optch <= 'z') {
3757                         negate = 1;
3758                         optch = toupper(optch);
3759                 } else {
3760                         negate = 0;
3761                 }
3762
3763                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3764                 if (!bit)
3765                         return optarg[i];
3766                 if (negate)
3767                         opt->filter &= ~bit;
3768                 else
3769                         opt->filter |= bit;
3770         }
3771         return 0;
3772 }
3773
3774 static void enable_patch_output(int *fmt) {
3775         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3776         *fmt |= DIFF_FORMAT_PATCH;
3777 }
3778
3779 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
3780 {
3781         int val = parse_ws_error_highlight(arg);
3782
3783         if (val < 0) {
3784                 error("unknown value after ws-error-highlight=%.*s",
3785                       -1 - val, arg);
3786                 return 0;
3787         }
3788         opt->ws_error_highlight = val;
3789         return 1;
3790 }
3791
3792 int diff_opt_parse(struct diff_options *options,
3793                    const char **av, int ac, const char *prefix)
3794 {
3795         const char *arg = av[0];
3796         const char *optarg;
3797         int argcount;
3798
3799         if (!prefix)
3800                 prefix = "";
3801
3802         /* Output format options */
3803         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3804             || opt_arg(arg, 'U', "unified", &options->context))
3805                 enable_patch_output(&options->output_format);
3806         else if (!strcmp(arg, "--raw"))
3807                 options->output_format |= DIFF_FORMAT_RAW;
3808         else if (!strcmp(arg, "--patch-with-raw")) {
3809                 enable_patch_output(&options->output_format);
3810                 options->output_format |= DIFF_FORMAT_RAW;
3811         } else if (!strcmp(arg, "--numstat"))
3812                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3813         else if (!strcmp(arg, "--shortstat"))
3814                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3815         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3816                 return parse_dirstat_opt(options, "");
3817         else if (skip_prefix(arg, "-X", &arg))
3818                 return parse_dirstat_opt(options, arg);
3819         else if (skip_prefix(arg, "--dirstat=", &arg))
3820                 return parse_dirstat_opt(options, arg);
3821         else if (!strcmp(arg, "--cumulative"))
3822                 return parse_dirstat_opt(options, "cumulative");
3823         else if (!strcmp(arg, "--dirstat-by-file"))
3824                 return parse_dirstat_opt(options, "files");
3825         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3826                 parse_dirstat_opt(options, "files");
3827                 return parse_dirstat_opt(options, arg);
3828         }
3829         else if (!strcmp(arg, "--check"))
3830                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3831         else if (!strcmp(arg, "--summary"))
3832                 options->output_format |= DIFF_FORMAT_SUMMARY;
3833         else if (!strcmp(arg, "--patch-with-stat")) {
3834                 enable_patch_output(&options->output_format);
3835                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3836         } else if (!strcmp(arg, "--name-only"))
3837                 options->output_format |= DIFF_FORMAT_NAME;
3838         else if (!strcmp(arg, "--name-status"))
3839                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3840         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3841                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3842         else if (starts_with(arg, "--stat"))
3843                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3844                 return stat_opt(options, av);
3845
3846         /* renames options */
3847         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3848                  !strcmp(arg, "--break-rewrites")) {
3849                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3850                         return error("invalid argument to -B: %s", arg+2);
3851         }
3852         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3853                  !strcmp(arg, "--find-renames")) {
3854                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3855                         return error("invalid argument to -M: %s", arg+2);
3856                 options->detect_rename = DIFF_DETECT_RENAME;
3857         }
3858         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3859                 options->irreversible_delete = 1;
3860         }
3861         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3862                  !strcmp(arg, "--find-copies")) {
3863                 if (options->detect_rename == DIFF_DETECT_COPY)
3864                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3865                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3866                         return error("invalid argument to -C: %s", arg+2);
3867                 options->detect_rename = DIFF_DETECT_COPY;
3868         }
3869         else if (!strcmp(arg, "--no-renames"))
3870                 options->detect_rename = 0;
3871         else if (!strcmp(arg, "--rename-empty"))
3872                 DIFF_OPT_SET(options, RENAME_EMPTY);
3873         else if (!strcmp(arg, "--no-rename-empty"))
3874                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3875         else if (!strcmp(arg, "--relative"))
3876                 DIFF_OPT_SET(options, RELATIVE_NAME);
3877         else if (skip_prefix(arg, "--relative=", &arg)) {
3878                 DIFF_OPT_SET(options, RELATIVE_NAME);
3879                 options->prefix = arg;
3880         }
3881
3882         /* xdiff options */
3883         else if (!strcmp(arg, "--minimal"))
3884                 DIFF_XDL_SET(options, NEED_MINIMAL);
3885         else if (!strcmp(arg, "--no-minimal"))
3886                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3887         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3888                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3889         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3890                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3891         else if (!strcmp(arg, "--ignore-space-at-eol"))
3892                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3893         else if (!strcmp(arg, "--ignore-blank-lines"))
3894                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3895         else if (!strcmp(arg, "--indent-heuristic"))
3896                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3897         else if (!strcmp(arg, "--no-indent-heuristic"))
3898                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3899         else if (!strcmp(arg, "--patience"))
3900                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3901         else if (!strcmp(arg, "--histogram"))
3902                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3903         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3904                 long value = parse_algorithm_value(optarg);
3905                 if (value < 0)
3906                         return error("option diff-algorithm accepts \"myers\", "
3907                                      "\"minimal\", \"patience\" and \"histogram\"");
3908                 /* clear out previous settings */
3909                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3910                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3911                 options->xdl_opts |= value;
3912                 return argcount;
3913         }
3914
3915         /* flags options */
3916         else if (!strcmp(arg, "--binary")) {
3917                 enable_patch_output(&options->output_format);
3918                 DIFF_OPT_SET(options, BINARY);
3919         }
3920         else if (!strcmp(arg, "--full-index"))
3921                 DIFF_OPT_SET(options, FULL_INDEX);
3922         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3923                 DIFF_OPT_SET(options, TEXT);
3924         else if (!strcmp(arg, "-R"))
3925                 DIFF_OPT_SET(options, REVERSE_DIFF);
3926         else if (!strcmp(arg, "--find-copies-harder"))
3927                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3928         else if (!strcmp(arg, "--follow"))
3929                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3930         else if (!strcmp(arg, "--no-follow")) {
3931                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3932                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3933         } else if (!strcmp(arg, "--color"))
3934                 options->use_color = 1;
3935         else if (skip_prefix(arg, "--color=", &arg)) {
3936                 int value = git_config_colorbool(NULL, arg);
3937                 if (value < 0)
3938                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3939                 options->use_color = value;
3940         }
3941         else if (!strcmp(arg, "--no-color"))
3942                 options->use_color = 0;
3943         else if (!strcmp(arg, "--color-words")) {
3944                 options->use_color = 1;
3945                 options->word_diff = DIFF_WORDS_COLOR;
3946         }
3947         else if (skip_prefix(arg, "--color-words=", &arg)) {
3948                 options->use_color = 1;
3949                 options->word_diff = DIFF_WORDS_COLOR;
3950                 options->word_regex = arg;
3951         }
3952         else if (!strcmp(arg, "--word-diff")) {
3953                 if (options->word_diff == DIFF_WORDS_NONE)
3954                         options->word_diff = DIFF_WORDS_PLAIN;
3955         }
3956         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3957                 if (!strcmp(arg, "plain"))
3958                         options->word_diff = DIFF_WORDS_PLAIN;
3959                 else if (!strcmp(arg, "color")) {
3960                         options->use_color = 1;
3961                         options->word_diff = DIFF_WORDS_COLOR;
3962                 }
3963                 else if (!strcmp(arg, "porcelain"))
3964                         options->word_diff = DIFF_WORDS_PORCELAIN;
3965                 else if (!strcmp(arg, "none"))
3966                         options->word_diff = DIFF_WORDS_NONE;
3967                 else
3968                         die("bad --word-diff argument: %s", arg);
3969         }
3970         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3971                 if (options->word_diff == DIFF_WORDS_NONE)
3972                         options->word_diff = DIFF_WORDS_PLAIN;
3973                 options->word_regex = optarg;
3974                 return argcount;
3975         }
3976         else if (!strcmp(arg, "--exit-code"))
3977                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3978         else if (!strcmp(arg, "--quiet"))
3979                 DIFF_OPT_SET(options, QUICK);
3980         else if (!strcmp(arg, "--ext-diff"))
3981                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3982         else if (!strcmp(arg, "--no-ext-diff"))
3983                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3984         else if (!strcmp(arg, "--textconv"))
3985                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3986         else if (!strcmp(arg, "--no-textconv"))
3987                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3988         else if (!strcmp(arg, "--ignore-submodules")) {
3989                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3990                 handle_ignore_submodules_arg(options, "all");
3991         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3992                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3993                 handle_ignore_submodules_arg(options, arg);
3994         } else if (!strcmp(arg, "--submodule"))
3995                 options->submodule_format = DIFF_SUBMODULE_LOG;
3996         else if (skip_prefix(arg, "--submodule=", &arg))
3997                 return parse_submodule_opt(options, arg);
3998         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3999                 return parse_ws_error_highlight_opt(options, arg);
4000         else if (!strcmp(arg, "--ita-invisible-in-index"))
4001                 options->ita_invisible_in_index = 1;
4002         else if (!strcmp(arg, "--ita-visible-in-index"))
4003                 options->ita_invisible_in_index = 0;
4004
4005         /* misc options */
4006         else if (!strcmp(arg, "-z"))
4007                 options->line_termination = 0;
4008         else if ((argcount = short_opt('l', av, &optarg))) {
4009                 options->rename_limit = strtoul(optarg, NULL, 10);
4010                 return argcount;
4011         }
4012         else if ((argcount = short_opt('S', av, &optarg))) {
4013                 options->pickaxe = optarg;
4014                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4015                 return argcount;
4016         } else if ((argcount = short_opt('G', av, &optarg))) {
4017                 options->pickaxe = optarg;
4018                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4019                 return argcount;
4020         }
4021         else if (!strcmp(arg, "--pickaxe-all"))
4022                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4023         else if (!strcmp(arg, "--pickaxe-regex"))
4024                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4025         else if ((argcount = short_opt('O', av, &optarg))) {
4026                 options->orderfile = prefix_filename(prefix, optarg);
4027                 return argcount;
4028         }
4029         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4030                 int offending = parse_diff_filter_opt(optarg, options);
4031                 if (offending)
4032                         die("unknown change class '%c' in --diff-filter=%s",
4033                             offending, optarg);
4034                 return argcount;
4035         }
4036         else if (!strcmp(arg, "--no-abbrev"))
4037                 options->abbrev = 0;
4038         else if (!strcmp(arg, "--abbrev"))
4039                 options->abbrev = DEFAULT_ABBREV;
4040         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4041                 options->abbrev = strtoul(arg, NULL, 10);
4042                 if (options->abbrev < MINIMUM_ABBREV)
4043                         options->abbrev = MINIMUM_ABBREV;
4044                 else if (40 < options->abbrev)
4045                         options->abbrev = 40;
4046         }
4047         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4048                 options->a_prefix = optarg;
4049                 return argcount;
4050         }
4051         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4052                 options->line_prefix = optarg;
4053                 options->line_prefix_length = strlen(options->line_prefix);
4054                 graph_setup_line_prefix(options);
4055                 return argcount;
4056         }
4057         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4058                 options->b_prefix = optarg;
4059                 return argcount;
4060         }
4061         else if (!strcmp(arg, "--no-prefix"))
4062                 options->a_prefix = options->b_prefix = "";
4063         else if (opt_arg(arg, '\0', "inter-hunk-context",
4064                          &options->interhunkcontext))
4065                 ;
4066         else if (!strcmp(arg, "-W"))
4067                 DIFF_OPT_SET(options, FUNCCONTEXT);
4068         else if (!strcmp(arg, "--function-context"))
4069                 DIFF_OPT_SET(options, FUNCCONTEXT);
4070         else if (!strcmp(arg, "--no-function-context"))
4071                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4072         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4073                 char *path = prefix_filename(prefix, optarg);
4074                 options->file = fopen(path, "w");
4075                 if (!options->file)
4076                         die_errno("Could not open '%s'", path);
4077                 options->close_file = 1;
4078                 if (options->use_color != GIT_COLOR_ALWAYS)
4079                         options->use_color = GIT_COLOR_NEVER;
4080                 free(path);
4081                 return argcount;
4082         } else
4083                 return 0;
4084         return 1;
4085 }
4086
4087 int parse_rename_score(const char **cp_p)
4088 {
4089         unsigned long num, scale;
4090         int ch, dot;
4091         const char *cp = *cp_p;
4092
4093         num = 0;
4094         scale = 1;
4095         dot = 0;
4096         for (;;) {
4097                 ch = *cp;
4098                 if ( !dot && ch == '.' ) {
4099                         scale = 1;
4100                         dot = 1;
4101                 } else if ( ch == '%' ) {
4102                         scale = dot ? scale*100 : 100;
4103                         cp++;   /* % is always at the end */
4104                         break;
4105                 } else if ( ch >= '0' && ch <= '9' ) {
4106                         if ( scale < 100000 ) {
4107                                 scale *= 10;
4108                                 num = (num*10) + (ch-'0');
4109                         }
4110                 } else {
4111                         break;
4112                 }
4113                 cp++;
4114         }
4115         *cp_p = cp;
4116
4117         /* user says num divided by scale and we say internally that
4118          * is MAX_SCORE * num / scale.
4119          */
4120         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4121 }
4122
4123 static int diff_scoreopt_parse(const char *opt)
4124 {
4125         int opt1, opt2, cmd;
4126
4127         if (*opt++ != '-')
4128                 return -1;
4129         cmd = *opt++;
4130         if (cmd == '-') {
4131                 /* convert the long-form arguments into short-form versions */
4132                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4133                         if (*opt == 0 || *opt++ == '=')
4134                                 cmd = 'B';
4135                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4136                         if (*opt == 0 || *opt++ == '=')
4137                                 cmd = 'C';
4138                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4139                         if (*opt == 0 || *opt++ == '=')
4140                                 cmd = 'M';
4141                 }
4142         }
4143         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4144                 return -1; /* that is not a -M, -C, or -B option */
4145
4146         opt1 = parse_rename_score(&opt);
4147         if (cmd != 'B')
4148                 opt2 = 0;
4149         else {
4150                 if (*opt == 0)
4151                         opt2 = 0;
4152                 else if (*opt != '/')
4153                         return -1; /* we expect -B80/99 or -B80 */
4154                 else {
4155                         opt++;
4156                         opt2 = parse_rename_score(&opt);
4157                 }
4158         }
4159         if (*opt != 0)
4160                 return -1;
4161         return opt1 | (opt2 << 16);
4162 }
4163
4164 struct diff_queue_struct diff_queued_diff;
4165
4166 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4167 {
4168         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4169         queue->queue[queue->nr++] = dp;
4170 }
4171
4172 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4173                                  struct diff_filespec *one,
4174                                  struct diff_filespec *two)
4175 {
4176         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4177         dp->one = one;
4178         dp->two = two;
4179         if (queue)
4180                 diff_q(queue, dp);
4181         return dp;
4182 }
4183
4184 void diff_free_filepair(struct diff_filepair *p)
4185 {
4186         free_filespec(p->one);
4187         free_filespec(p->two);
4188         free(p);
4189 }
4190
4191 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4192 {
4193         int abblen;
4194         const char *abbrev;
4195
4196         if (len == GIT_SHA1_HEXSZ)
4197                 return oid_to_hex(oid);
4198
4199         abbrev = diff_abbrev_oid(oid, len);
4200         abblen = strlen(abbrev);
4201
4202         /*
4203          * In well-behaved cases, where the abbbreviated result is the
4204          * same as the requested length, append three dots after the
4205          * abbreviation (hence the whole logic is limited to the case
4206          * where abblen < 37); when the actual abbreviated result is a
4207          * bit longer than the requested length, we reduce the number
4208          * of dots so that they match the well-behaved ones.  However,
4209          * if the actual abbreviation is longer than the requested
4210          * length by more than three, we give up on aligning, and add
4211          * three dots anyway, to indicate that the output is not the
4212          * full object name.  Yes, this may be suboptimal, but this
4213          * appears only in "diff --raw --abbrev" output and it is not
4214          * worth the effort to change it now.  Note that this would
4215          * likely to work fine when the automatic sizing of default
4216          * abbreviation length is used--we would be fed -1 in "len" in
4217          * that case, and will end up always appending three-dots, but
4218          * the automatic sizing is supposed to give abblen that ensures
4219          * uniqueness across all objects (statistically speaking).
4220          */
4221         if (abblen < GIT_SHA1_HEXSZ - 3) {
4222                 static char hex[GIT_MAX_HEXSZ + 1];
4223                 if (len < abblen && abblen <= len + 2)
4224                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4225                 else
4226                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4227                 return hex;
4228         }
4229
4230         return oid_to_hex(oid);
4231 }
4232
4233 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4234 {
4235         int line_termination = opt->line_termination;
4236         int inter_name_termination = line_termination ? '\t' : '\0';
4237
4238         fprintf(opt->file, "%s", diff_line_prefix(opt));
4239         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4240                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4241                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4242                 fprintf(opt->file, "%s ",
4243                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4244         }
4245         if (p->score) {
4246                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4247                         inter_name_termination);
4248         } else {
4249                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4250         }
4251
4252         if (p->status == DIFF_STATUS_COPIED ||
4253             p->status == DIFF_STATUS_RENAMED) {
4254                 const char *name_a, *name_b;
4255                 name_a = p->one->path;
4256                 name_b = p->two->path;
4257                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4258                 write_name_quoted(name_a, opt->file, inter_name_termination);
4259                 write_name_quoted(name_b, opt->file, line_termination);
4260         } else {
4261                 const char *name_a, *name_b;
4262                 name_a = p->one->mode ? p->one->path : p->two->path;
4263                 name_b = NULL;
4264                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4265                 write_name_quoted(name_a, opt->file, line_termination);
4266         }
4267 }
4268
4269 int diff_unmodified_pair(struct diff_filepair *p)
4270 {
4271         /* This function is written stricter than necessary to support
4272          * the currently implemented transformers, but the idea is to
4273          * let transformers to produce diff_filepairs any way they want,
4274          * and filter and clean them up here before producing the output.
4275          */
4276         struct diff_filespec *one = p->one, *two = p->two;
4277
4278         if (DIFF_PAIR_UNMERGED(p))
4279                 return 0; /* unmerged is interesting */
4280
4281         /* deletion, addition, mode or type change
4282          * and rename are all interesting.
4283          */
4284         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4285             DIFF_PAIR_MODE_CHANGED(p) ||
4286             strcmp(one->path, two->path))
4287                 return 0;
4288
4289         /* both are valid and point at the same path.  that is, we are
4290          * dealing with a change.
4291          */
4292         if (one->oid_valid && two->oid_valid &&
4293             !oidcmp(&one->oid, &two->oid) &&
4294             !one->dirty_submodule && !two->dirty_submodule)
4295                 return 1; /* no change */
4296         if (!one->oid_valid && !two->oid_valid)
4297                 return 1; /* both look at the same file on the filesystem. */
4298         return 0;
4299 }
4300
4301 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4302 {
4303         if (diff_unmodified_pair(p))
4304                 return;
4305
4306         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4307             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4308                 return; /* no tree diffs in patch format */
4309
4310         run_diff(p, o);
4311 }
4312
4313 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4314                             struct diffstat_t *diffstat)
4315 {
4316         if (diff_unmodified_pair(p))
4317                 return;
4318
4319         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4320             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4321                 return; /* no useful stat for tree diffs */
4322
4323         run_diffstat(p, o, diffstat);
4324 }
4325
4326 static void diff_flush_checkdiff(struct diff_filepair *p,
4327                 struct diff_options *o)
4328 {
4329         if (diff_unmodified_pair(p))
4330                 return;
4331
4332         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4333             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4334                 return; /* nothing to check in tree diffs */
4335
4336         run_checkdiff(p, o);
4337 }
4338
4339 int diff_queue_is_empty(void)
4340 {
4341         struct diff_queue_struct *q = &diff_queued_diff;
4342         int i;
4343         for (i = 0; i < q->nr; i++)
4344                 if (!diff_unmodified_pair(q->queue[i]))
4345                         return 0;
4346         return 1;
4347 }
4348
4349 #if DIFF_DEBUG
4350 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4351 {
4352         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4353                 x, one ? one : "",
4354                 s->path,
4355                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4356                 s->mode,
4357                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4358         fprintf(stderr, "queue[%d] %s size %lu\n",
4359                 x, one ? one : "",
4360                 s->size);
4361 }
4362
4363 void diff_debug_filepair(const struct diff_filepair *p, int i)
4364 {
4365         diff_debug_filespec(p->one, i, "one");
4366         diff_debug_filespec(p->two, i, "two");
4367         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4368                 p->score, p->status ? p->status : '?',
4369                 p->one->rename_used, p->broken_pair);
4370 }
4371
4372 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4373 {
4374         int i;
4375         if (msg)
4376                 fprintf(stderr, "%s\n", msg);
4377         fprintf(stderr, "q->nr = %d\n", q->nr);
4378         for (i = 0; i < q->nr; i++) {
4379                 struct diff_filepair *p = q->queue[i];
4380                 diff_debug_filepair(p, i);
4381         }
4382 }
4383 #endif
4384
4385 static void diff_resolve_rename_copy(void)
4386 {
4387         int i;
4388         struct diff_filepair *p;
4389         struct diff_queue_struct *q = &diff_queued_diff;
4390
4391         diff_debug_queue("resolve-rename-copy", q);
4392
4393         for (i = 0; i < q->nr; i++) {
4394                 p = q->queue[i];
4395                 p->status = 0; /* undecided */
4396                 if (DIFF_PAIR_UNMERGED(p))
4397                         p->status = DIFF_STATUS_UNMERGED;
4398                 else if (!DIFF_FILE_VALID(p->one))
4399                         p->status = DIFF_STATUS_ADDED;
4400                 else if (!DIFF_FILE_VALID(p->two))
4401                         p->status = DIFF_STATUS_DELETED;
4402                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4403                         p->status = DIFF_STATUS_TYPE_CHANGED;
4404
4405                 /* from this point on, we are dealing with a pair
4406                  * whose both sides are valid and of the same type, i.e.
4407                  * either in-place edit or rename/copy edit.
4408                  */
4409                 else if (DIFF_PAIR_RENAME(p)) {
4410                         /*
4411                          * A rename might have re-connected a broken
4412                          * pair up, causing the pathnames to be the
4413                          * same again. If so, that's not a rename at
4414                          * all, just a modification..
4415                          *
4416                          * Otherwise, see if this source was used for
4417                          * multiple renames, in which case we decrement
4418                          * the count, and call it a copy.
4419                          */
4420                         if (!strcmp(p->one->path, p->two->path))
4421                                 p->status = DIFF_STATUS_MODIFIED;
4422                         else if (--p->one->rename_used > 0)
4423                                 p->status = DIFF_STATUS_COPIED;
4424                         else
4425                                 p->status = DIFF_STATUS_RENAMED;
4426                 }
4427                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4428                          p->one->mode != p->two->mode ||
4429                          p->one->dirty_submodule ||
4430                          p->two->dirty_submodule ||
4431                          is_null_oid(&p->one->oid))
4432                         p->status = DIFF_STATUS_MODIFIED;
4433                 else {
4434                         /* This is a "no-change" entry and should not
4435                          * happen anymore, but prepare for broken callers.
4436                          */
4437                         error("feeding unmodified %s to diffcore",
4438                               p->one->path);
4439                         p->status = DIFF_STATUS_UNKNOWN;
4440                 }
4441         }
4442         diff_debug_queue("resolve-rename-copy done", q);
4443 }
4444
4445 static int check_pair_status(struct diff_filepair *p)
4446 {
4447         switch (p->status) {
4448         case DIFF_STATUS_UNKNOWN:
4449                 return 0;
4450         case 0:
4451                 die("internal error in diff-resolve-rename-copy");
4452         default:
4453                 return 1;
4454         }
4455 }
4456
4457 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4458 {
4459         int fmt = opt->output_format;
4460
4461         if (fmt & DIFF_FORMAT_CHECKDIFF)
4462                 diff_flush_checkdiff(p, opt);
4463         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4464                 diff_flush_raw(p, opt);
4465         else if (fmt & DIFF_FORMAT_NAME) {
4466                 const char *name_a, *name_b;
4467                 name_a = p->two->path;
4468                 name_b = NULL;
4469                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4470                 fprintf(opt->file, "%s", diff_line_prefix(opt));
4471                 write_name_quoted(name_a, opt->file, opt->line_termination);
4472         }
4473 }
4474
4475 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4476 {
4477         if (fs->mode)
4478                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4479         else
4480                 fprintf(file, " %s ", newdelete);
4481         write_name_quoted(fs->path, file, '\n');
4482 }
4483
4484
4485 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4486                 const char *line_prefix)
4487 {
4488         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4489                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4490                         p->two->mode, show_name ? ' ' : '\n');
4491                 if (show_name) {
4492                         write_name_quoted(p->two->path, file, '\n');
4493                 }
4494         }
4495 }
4496
4497 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4498                         const char *line_prefix)
4499 {
4500         char *names = pprint_rename(p->one->path, p->two->path);
4501
4502         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4503         free(names);
4504         show_mode_change(file, p, 0, line_prefix);
4505 }
4506
4507 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4508 {
4509         FILE *file = opt->file;
4510         const char *line_prefix = diff_line_prefix(opt);
4511
4512         switch(p->status) {
4513         case DIFF_STATUS_DELETED:
4514                 fputs(line_prefix, file);
4515                 show_file_mode_name(file, "delete", p->one);
4516                 break;
4517         case DIFF_STATUS_ADDED:
4518                 fputs(line_prefix, file);
4519                 show_file_mode_name(file, "create", p->two);
4520                 break;
4521         case DIFF_STATUS_COPIED:
4522                 fputs(line_prefix, file);
4523                 show_rename_copy(file, "copy", p, line_prefix);
4524                 break;
4525         case DIFF_STATUS_RENAMED:
4526                 fputs(line_prefix, file);
4527                 show_rename_copy(file, "rename", p, line_prefix);
4528                 break;
4529         default:
4530                 if (p->score) {
4531                         fprintf(file, "%s rewrite ", line_prefix);
4532                         write_name_quoted(p->two->path, file, ' ');
4533                         fprintf(file, "(%d%%)\n", similarity_index(p));
4534                 }
4535                 show_mode_change(file, p, !p->score, line_prefix);
4536                 break;
4537         }
4538 }
4539
4540 struct patch_id_t {
4541         git_SHA_CTX *ctx;
4542         int patchlen;
4543 };
4544
4545 static int remove_space(char *line, int len)
4546 {
4547         int i;
4548         char *dst = line;
4549         unsigned char c;
4550
4551         for (i = 0; i < len; i++)
4552                 if (!isspace((c = line[i])))
4553                         *dst++ = c;
4554
4555         return dst - line;
4556 }
4557
4558 static void patch_id_consume(void *priv, char *line, unsigned long len)
4559 {
4560         struct patch_id_t *data = priv;
4561         int new_len;
4562
4563         /* Ignore line numbers when computing the SHA1 of the patch */
4564         if (starts_with(line, "@@ -"))
4565                 return;
4566
4567         new_len = remove_space(line, len);
4568
4569         git_SHA1_Update(data->ctx, line, new_len);
4570         data->patchlen += new_len;
4571 }
4572
4573 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
4574 {
4575         git_SHA1_Update(ctx, str, strlen(str));
4576 }
4577
4578 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
4579 {
4580         /* large enough for 2^32 in octal */
4581         char buf[12];
4582         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
4583         git_SHA1_Update(ctx, buf, len);
4584 }
4585
4586 /* returns 0 upon success, and writes result into sha1 */
4587 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4588 {
4589         struct diff_queue_struct *q = &diff_queued_diff;
4590         int i;
4591         git_SHA_CTX ctx;
4592         struct patch_id_t data;
4593
4594         git_SHA1_Init(&ctx);
4595         memset(&data, 0, sizeof(struct patch_id_t));
4596         data.ctx = &ctx;
4597
4598         for (i = 0; i < q->nr; i++) {
4599                 xpparam_t xpp;
4600                 xdemitconf_t xecfg;
4601                 mmfile_t mf1, mf2;
4602                 struct diff_filepair *p = q->queue[i];
4603                 int len1, len2;
4604
4605                 memset(&xpp, 0, sizeof(xpp));
4606                 memset(&xecfg, 0, sizeof(xecfg));
4607                 if (p->status == 0)
4608                         return error("internal diff status error");
4609                 if (p->status == DIFF_STATUS_UNKNOWN)
4610                         continue;
4611                 if (diff_unmodified_pair(p))
4612                         continue;
4613                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4614                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4615                         continue;
4616                 if (DIFF_PAIR_UNMERGED(p))
4617                         continue;
4618
4619                 diff_fill_sha1_info(p->one);
4620                 diff_fill_sha1_info(p->two);
4621
4622                 len1 = remove_space(p->one->path, strlen(p->one->path));
4623                 len2 = remove_space(p->two->path, strlen(p->two->path));
4624                 patch_id_add_string(&ctx, "diff--git");
4625                 patch_id_add_string(&ctx, "a/");
4626                 git_SHA1_Update(&ctx, p->one->path, len1);
4627                 patch_id_add_string(&ctx, "b/");
4628                 git_SHA1_Update(&ctx, p->two->path, len2);
4629
4630                 if (p->one->mode == 0) {
4631                         patch_id_add_string(&ctx, "newfilemode");
4632                         patch_id_add_mode(&ctx, p->two->mode);
4633                         patch_id_add_string(&ctx, "---/dev/null");
4634                         patch_id_add_string(&ctx, "+++b/");
4635                         git_SHA1_Update(&ctx, p->two->path, len2);
4636                 } else if (p->two->mode == 0) {
4637                         patch_id_add_string(&ctx, "deletedfilemode");
4638                         patch_id_add_mode(&ctx, p->one->mode);
4639                         patch_id_add_string(&ctx, "---a/");
4640                         git_SHA1_Update(&ctx, p->one->path, len1);
4641                         patch_id_add_string(&ctx, "+++/dev/null");
4642                 } else {
4643                         patch_id_add_string(&ctx, "---a/");
4644                         git_SHA1_Update(&ctx, p->one->path, len1);
4645                         patch_id_add_string(&ctx, "+++b/");
4646                         git_SHA1_Update(&ctx, p->two->path, len2);
4647                 }
4648
4649                 if (diff_header_only)
4650                         continue;
4651
4652                 if (fill_mmfile(&mf1, p->one) < 0 ||
4653                     fill_mmfile(&mf2, p->two) < 0)
4654                         return error("unable to read files to diff");
4655
4656                 if (diff_filespec_is_binary(p->one) ||
4657                     diff_filespec_is_binary(p->two)) {
4658                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4659                                         40);
4660                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4661                                         40);
4662                         continue;
4663                 }
4664
4665                 xpp.flags = 0;
4666                 xecfg.ctxlen = 3;
4667                 xecfg.flags = 0;
4668                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4669                                   &xpp, &xecfg))
4670                         return error("unable to generate patch-id diff for %s",
4671                                      p->one->path);
4672         }
4673
4674         git_SHA1_Final(sha1, &ctx);
4675         return 0;
4676 }
4677
4678 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4679 {
4680         struct diff_queue_struct *q = &diff_queued_diff;
4681         int i;
4682         int result = diff_get_patch_id(options, sha1, diff_header_only);
4683
4684         for (i = 0; i < q->nr; i++)
4685                 diff_free_filepair(q->queue[i]);
4686
4687         free(q->queue);
4688         DIFF_QUEUE_CLEAR(q);
4689
4690         return result;
4691 }
4692
4693 static int is_summary_empty(const struct diff_queue_struct *q)
4694 {
4695         int i;
4696
4697         for (i = 0; i < q->nr; i++) {
4698                 const struct diff_filepair *p = q->queue[i];
4699
4700                 switch (p->status) {
4701                 case DIFF_STATUS_DELETED:
4702                 case DIFF_STATUS_ADDED:
4703                 case DIFF_STATUS_COPIED:
4704                 case DIFF_STATUS_RENAMED:
4705                         return 0;
4706                 default:
4707                         if (p->score)
4708                                 return 0;
4709                         if (p->one->mode && p->two->mode &&
4710                             p->one->mode != p->two->mode)
4711                                 return 0;
4712                         break;
4713                 }
4714         }
4715         return 1;
4716 }
4717
4718 static const char rename_limit_warning[] =
4719 N_("inexact rename detection was skipped due to too many files.");
4720
4721 static const char degrade_cc_to_c_warning[] =
4722 N_("only found copies from modified paths due to too many files.");
4723
4724 static const char rename_limit_advice[] =
4725 N_("you may want to set your %s variable to at least "
4726    "%d and retry the command.");
4727
4728 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4729 {
4730         if (degraded_cc)
4731                 warning(_(degrade_cc_to_c_warning));
4732         else if (needed)
4733                 warning(_(rename_limit_warning));
4734         else
4735                 return;
4736         if (0 < needed && needed < 32767)
4737                 warning(_(rename_limit_advice), varname, needed);
4738 }
4739
4740 void diff_flush(struct diff_options *options)
4741 {
4742         struct diff_queue_struct *q = &diff_queued_diff;
4743         int i, output_format = options->output_format;
4744         int separator = 0;
4745         int dirstat_by_line = 0;
4746
4747         /*
4748          * Order: raw, stat, summary, patch
4749          * or:    name/name-status/checkdiff (other bits clear)
4750          */
4751         if (!q->nr)
4752                 goto free_queue;
4753
4754         if (output_format & (DIFF_FORMAT_RAW |
4755                              DIFF_FORMAT_NAME |
4756                              DIFF_FORMAT_NAME_STATUS |
4757                              DIFF_FORMAT_CHECKDIFF)) {
4758                 for (i = 0; i < q->nr; i++) {
4759                         struct diff_filepair *p = q->queue[i];
4760                         if (check_pair_status(p))
4761                                 flush_one_pair(p, options);
4762                 }
4763                 separator++;
4764         }
4765
4766         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4767                 dirstat_by_line = 1;
4768
4769         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4770             dirstat_by_line) {
4771                 struct diffstat_t diffstat;
4772
4773                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4774                 for (i = 0; i < q->nr; i++) {
4775                         struct diff_filepair *p = q->queue[i];
4776                         if (check_pair_status(p))
4777                                 diff_flush_stat(p, options, &diffstat);
4778                 }
4779                 if (output_format & DIFF_FORMAT_NUMSTAT)
4780                         show_numstat(&diffstat, options);
4781                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4782                         show_stats(&diffstat, options);
4783                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4784                         show_shortstats(&diffstat, options);
4785                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4786                         show_dirstat_by_line(&diffstat, options);
4787                 free_diffstat_info(&diffstat);
4788                 separator++;
4789         }
4790         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4791                 show_dirstat(options);
4792
4793         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4794                 for (i = 0; i < q->nr; i++) {
4795                         diff_summary(options, q->queue[i]);
4796                 }
4797                 separator++;
4798         }
4799
4800         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4801             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4802             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4803                 /*
4804                  * run diff_flush_patch for the exit status. setting
4805                  * options->file to /dev/null should be safe, because we
4806                  * aren't supposed to produce any output anyway.
4807                  */
4808                 if (options->close_file)
4809                         fclose(options->file);
4810                 options->file = fopen("/dev/null", "w");
4811                 if (!options->file)
4812                         die_errno("Could not open /dev/null");
4813                 options->close_file = 1;
4814                 for (i = 0; i < q->nr; i++) {
4815                         struct diff_filepair *p = q->queue[i];
4816                         if (check_pair_status(p))
4817                                 diff_flush_patch(p, options);
4818                         if (options->found_changes)
4819                                 break;
4820                 }
4821         }
4822
4823         if (output_format & DIFF_FORMAT_PATCH) {
4824                 if (separator) {
4825                         fprintf(options->file, "%s%c",
4826                                 diff_line_prefix(options),
4827                                 options->line_termination);
4828                         if (options->stat_sep) {
4829                                 /* attach patch instead of inline */
4830                                 fputs(options->stat_sep, options->file);
4831                         }
4832                 }
4833
4834                 for (i = 0; i < q->nr; i++) {
4835                         struct diff_filepair *p = q->queue[i];
4836                         if (check_pair_status(p))
4837                                 diff_flush_patch(p, options);
4838                 }
4839         }
4840
4841         if (output_format & DIFF_FORMAT_CALLBACK)
4842                 options->format_callback(q, options, options->format_callback_data);
4843
4844         for (i = 0; i < q->nr; i++)
4845                 diff_free_filepair(q->queue[i]);
4846 free_queue:
4847         free(q->queue);
4848         DIFF_QUEUE_CLEAR(q);
4849         if (options->close_file)
4850                 fclose(options->file);
4851
4852         /*
4853          * Report the content-level differences with HAS_CHANGES;
4854          * diff_addremove/diff_change does not set the bit when
4855          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4856          */
4857         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4858                 if (options->found_changes)
4859                         DIFF_OPT_SET(options, HAS_CHANGES);
4860                 else
4861                         DIFF_OPT_CLR(options, HAS_CHANGES);
4862         }
4863 }
4864
4865 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4866 {
4867         return (((p->status == DIFF_STATUS_MODIFIED) &&
4868                  ((p->score &&
4869                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4870                   (!p->score &&
4871                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4872                 ((p->status != DIFF_STATUS_MODIFIED) &&
4873                  filter_bit_tst(p->status, options)));
4874 }
4875
4876 static void diffcore_apply_filter(struct diff_options *options)
4877 {
4878         int i;
4879         struct diff_queue_struct *q = &diff_queued_diff;
4880         struct diff_queue_struct outq;
4881
4882         DIFF_QUEUE_CLEAR(&outq);
4883
4884         if (!options->filter)
4885                 return;
4886
4887         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4888                 int found;
4889                 for (i = found = 0; !found && i < q->nr; i++) {
4890                         if (match_filter(options, q->queue[i]))
4891                                 found++;
4892                 }
4893                 if (found)
4894                         return;
4895
4896                 /* otherwise we will clear the whole queue
4897                  * by copying the empty outq at the end of this
4898                  * function, but first clear the current entries
4899                  * in the queue.
4900                  */
4901                 for (i = 0; i < q->nr; i++)
4902                         diff_free_filepair(q->queue[i]);
4903         }
4904         else {
4905                 /* Only the matching ones */
4906                 for (i = 0; i < q->nr; i++) {
4907                         struct diff_filepair *p = q->queue[i];
4908                         if (match_filter(options, p))
4909                                 diff_q(&outq, p);
4910                         else
4911                                 diff_free_filepair(p);
4912                 }
4913         }
4914         free(q->queue);
4915         *q = outq;
4916 }
4917
4918 /* Check whether two filespecs with the same mode and size are identical */
4919 static int diff_filespec_is_identical(struct diff_filespec *one,
4920                                       struct diff_filespec *two)
4921 {
4922         if (S_ISGITLINK(one->mode))
4923                 return 0;
4924         if (diff_populate_filespec(one, 0))
4925                 return 0;
4926         if (diff_populate_filespec(two, 0))
4927                 return 0;
4928         return !memcmp(one->data, two->data, one->size);
4929 }
4930
4931 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4932 {
4933         if (p->done_skip_stat_unmatch)
4934                 return p->skip_stat_unmatch_result;
4935
4936         p->done_skip_stat_unmatch = 1;
4937         p->skip_stat_unmatch_result = 0;
4938         /*
4939          * 1. Entries that come from stat info dirtiness
4940          *    always have both sides (iow, not create/delete),
4941          *    one side of the object name is unknown, with
4942          *    the same mode and size.  Keep the ones that
4943          *    do not match these criteria.  They have real
4944          *    differences.
4945          *
4946          * 2. At this point, the file is known to be modified,
4947          *    with the same mode and size, and the object
4948          *    name of one side is unknown.  Need to inspect
4949          *    the identical contents.
4950          */
4951         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4952             !DIFF_FILE_VALID(p->two) ||
4953             (p->one->oid_valid && p->two->oid_valid) ||
4954             (p->one->mode != p->two->mode) ||
4955             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4956             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4957             (p->one->size != p->two->size) ||
4958             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4959                 p->skip_stat_unmatch_result = 1;
4960         return p->skip_stat_unmatch_result;
4961 }
4962
4963 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4964 {
4965         int i;
4966         struct diff_queue_struct *q = &diff_queued_diff;
4967         struct diff_queue_struct outq;
4968         DIFF_QUEUE_CLEAR(&outq);
4969
4970         for (i = 0; i < q->nr; i++) {
4971                 struct diff_filepair *p = q->queue[i];
4972
4973                 if (diff_filespec_check_stat_unmatch(p))
4974                         diff_q(&outq, p);
4975                 else {
4976                         /*
4977                          * The caller can subtract 1 from skip_stat_unmatch
4978                          * to determine how many paths were dirty only
4979                          * due to stat info mismatch.
4980                          */
4981                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4982                                 diffopt->skip_stat_unmatch++;
4983                         diff_free_filepair(p);
4984                 }
4985         }
4986         free(q->queue);
4987         *q = outq;
4988 }
4989
4990 static int diffnamecmp(const void *a_, const void *b_)
4991 {
4992         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4993         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4994         const char *name_a, *name_b;
4995
4996         name_a = a->one ? a->one->path : a->two->path;
4997         name_b = b->one ? b->one->path : b->two->path;
4998         return strcmp(name_a, name_b);
4999 }
5000
5001 void diffcore_fix_diff_index(struct diff_options *options)
5002 {
5003         struct diff_queue_struct *q = &diff_queued_diff;
5004         QSORT(q->queue, q->nr, diffnamecmp);
5005 }
5006
5007 void diffcore_std(struct diff_options *options)
5008 {
5009         /* NOTE please keep the following in sync with diff_tree_combined() */
5010         if (options->skip_stat_unmatch)
5011                 diffcore_skip_stat_unmatch(options);
5012         if (!options->found_follow) {
5013                 /* See try_to_follow_renames() in tree-diff.c */
5014                 if (options->break_opt != -1)
5015                         diffcore_break(options->break_opt);
5016                 if (options->detect_rename)
5017                         diffcore_rename(options);
5018                 if (options->break_opt != -1)
5019                         diffcore_merge_broken();
5020         }
5021         if (options->pickaxe)
5022                 diffcore_pickaxe(options);
5023         if (options->orderfile)
5024                 diffcore_order(options->orderfile);
5025         if (!options->found_follow)
5026                 /* See try_to_follow_renames() in tree-diff.c */
5027                 diff_resolve_rename_copy();
5028         diffcore_apply_filter(options);
5029
5030         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5031                 DIFF_OPT_SET(options, HAS_CHANGES);
5032         else
5033                 DIFF_OPT_CLR(options, HAS_CHANGES);
5034
5035         options->found_follow = 0;
5036 }
5037
5038 int diff_result_code(struct diff_options *opt, int status)
5039 {
5040         int result = 0;
5041
5042         diff_warn_rename_limit("diff.renameLimit",
5043                                opt->needed_rename_limit,
5044                                opt->degraded_cc_to_c);
5045         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5046             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5047                 return status;
5048         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5049             DIFF_OPT_TST(opt, HAS_CHANGES))
5050                 result |= 01;
5051         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5052             DIFF_OPT_TST(opt, CHECK_FAILED))
5053                 result |= 02;
5054         return result;
5055 }
5056
5057 int diff_can_quit_early(struct diff_options *opt)
5058 {
5059         return (DIFF_OPT_TST(opt, QUICK) &&
5060                 !opt->filter &&
5061                 DIFF_OPT_TST(opt, HAS_CHANGES));
5062 }
5063
5064 /*
5065  * Shall changes to this submodule be ignored?
5066  *
5067  * Submodule changes can be configured to be ignored separately for each path,
5068  * but that configuration can be overridden from the command line.
5069  */
5070 static int is_submodule_ignored(const char *path, struct diff_options *options)
5071 {
5072         int ignored = 0;
5073         unsigned orig_flags = options->flags;
5074         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5075                 set_diffopt_flags_from_submodule_config(options, path);
5076         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5077                 ignored = 1;
5078         options->flags = orig_flags;
5079         return ignored;
5080 }
5081
5082 void diff_addremove(struct diff_options *options,
5083                     int addremove, unsigned mode,
5084                     const unsigned char *sha1,
5085                     int sha1_valid,
5086                     const char *concatpath, unsigned dirty_submodule)
5087 {
5088         struct diff_filespec *one, *two;
5089
5090         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5091                 return;
5092
5093         /* This may look odd, but it is a preparation for
5094          * feeding "there are unchanged files which should
5095          * not produce diffs, but when you are doing copy
5096          * detection you would need them, so here they are"
5097          * entries to the diff-core.  They will be prefixed
5098          * with something like '=' or '*' (I haven't decided
5099          * which but should not make any difference).
5100          * Feeding the same new and old to diff_change()
5101          * also has the same effect.
5102          * Before the final output happens, they are pruned after
5103          * merged into rename/copy pairs as appropriate.
5104          */
5105         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5106                 addremove = (addremove == '+' ? '-' :
5107                              addremove == '-' ? '+' : addremove);
5108
5109         if (options->prefix &&
5110             strncmp(concatpath, options->prefix, options->prefix_length))
5111                 return;
5112
5113         one = alloc_filespec(concatpath);
5114         two = alloc_filespec(concatpath);
5115
5116         if (addremove != '+')
5117                 fill_filespec(one, sha1, sha1_valid, mode);
5118         if (addremove != '-') {
5119                 fill_filespec(two, sha1, sha1_valid, mode);
5120                 two->dirty_submodule = dirty_submodule;
5121         }
5122
5123         diff_queue(&diff_queued_diff, one, two);
5124         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5125                 DIFF_OPT_SET(options, HAS_CHANGES);
5126 }
5127
5128 void diff_change(struct diff_options *options,
5129                  unsigned old_mode, unsigned new_mode,
5130                  const unsigned char *old_sha1,
5131                  const unsigned char *new_sha1,
5132                  int old_sha1_valid, int new_sha1_valid,
5133                  const char *concatpath,
5134                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5135 {
5136         struct diff_filespec *one, *two;
5137         struct diff_filepair *p;
5138
5139         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5140             is_submodule_ignored(concatpath, options))
5141                 return;
5142
5143         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5144                 SWAP(old_mode, new_mode);
5145                 SWAP(old_sha1, new_sha1);
5146                 SWAP(old_sha1_valid, new_sha1_valid);
5147                 SWAP(old_dirty_submodule, new_dirty_submodule);
5148         }
5149
5150         if (options->prefix &&
5151             strncmp(concatpath, options->prefix, options->prefix_length))
5152                 return;
5153
5154         one = alloc_filespec(concatpath);
5155         two = alloc_filespec(concatpath);
5156         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5157         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5158         one->dirty_submodule = old_dirty_submodule;
5159         two->dirty_submodule = new_dirty_submodule;
5160         p = diff_queue(&diff_queued_diff, one, two);
5161
5162         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5163                 return;
5164
5165         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5166             !diff_filespec_check_stat_unmatch(p))
5167                 return;
5168
5169         DIFF_OPT_SET(options, HAS_CHANGES);
5170 }
5171
5172 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5173 {
5174         struct diff_filepair *pair;
5175         struct diff_filespec *one, *two;
5176
5177         if (options->prefix &&
5178             strncmp(path, options->prefix, options->prefix_length))
5179                 return NULL;
5180
5181         one = alloc_filespec(path);
5182         two = alloc_filespec(path);
5183         pair = diff_queue(&diff_queued_diff, one, two);
5184         pair->is_unmerged = 1;
5185         return pair;
5186 }
5187
5188 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5189                 size_t *outsize)
5190 {
5191         struct diff_tempfile *temp;
5192         const char *argv[3];
5193         const char **arg = argv;
5194         struct child_process child = CHILD_PROCESS_INIT;
5195         struct strbuf buf = STRBUF_INIT;
5196         int err = 0;
5197
5198         temp = prepare_temp_file(spec->path, spec);
5199         *arg++ = pgm;
5200         *arg++ = temp->name;
5201         *arg = NULL;
5202
5203         child.use_shell = 1;
5204         child.argv = argv;
5205         child.out = -1;
5206         if (start_command(&child)) {
5207                 remove_tempfile();
5208                 return NULL;
5209         }
5210
5211         if (strbuf_read(&buf, child.out, 0) < 0)
5212                 err = error("error reading from textconv command '%s'", pgm);
5213         close(child.out);
5214
5215         if (finish_command(&child) || err) {
5216                 strbuf_release(&buf);
5217                 remove_tempfile();
5218                 return NULL;
5219         }
5220         remove_tempfile();
5221
5222         return strbuf_detach(&buf, outsize);
5223 }
5224
5225 size_t fill_textconv(struct userdiff_driver *driver,
5226                      struct diff_filespec *df,
5227                      char **outbuf)
5228 {
5229         size_t size;
5230
5231         if (!driver) {
5232                 if (!DIFF_FILE_VALID(df)) {
5233                         *outbuf = "";
5234                         return 0;
5235                 }
5236                 if (diff_populate_filespec(df, 0))
5237                         die("unable to read files to diff");
5238                 *outbuf = df->data;
5239                 return df->size;
5240         }
5241
5242         if (!driver->textconv)
5243                 die("BUG: fill_textconv called with non-textconv driver");
5244
5245         if (driver->textconv_cache && df->oid_valid) {
5246                 *outbuf = notes_cache_get(driver->textconv_cache,
5247                                           df->oid.hash,
5248                                           &size);
5249                 if (*outbuf)
5250                         return size;
5251         }
5252
5253         *outbuf = run_textconv(driver->textconv, df, &size);
5254         if (!*outbuf)
5255                 die("unable to read files to diff");
5256
5257         if (driver->textconv_cache && df->oid_valid) {
5258                 /* ignore errors, as we might be in a readonly repository */
5259                 notes_cache_put(driver->textconv_cache, df->oid.hash, *outbuf,
5260                                 size);
5261                 /*
5262                  * we could save up changes and flush them all at the end,
5263                  * but we would need an extra call after all diffing is done.
5264                  * Since generating a cache entry is the slow path anyway,
5265                  * this extra overhead probably isn't a big deal.
5266                  */
5267                 notes_cache_write(driver->textconv_cache);
5268         }
5269
5270         return size;
5271 }
5272
5273 void setup_diff_pager(struct diff_options *opt)
5274 {
5275         /*
5276          * If the user asked for our exit code, then either they want --quiet
5277          * or --exit-code. We should definitely not bother with a pager in the
5278          * former case, as we will generate no output. Since we still properly
5279          * report our exit code even when a pager is run, we _could_ run a
5280          * pager with --exit-code. But since we have not done so historically,
5281          * and because it is easy to find people oneline advising "git diff
5282          * --exit-code" in hooks and other scripts, we do not do so.
5283          */
5284         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5285             check_pager_config("diff") != 0)
5286                 setup_pager();
5287 }