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 = 1;
  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 (!strcmp(var, "diff.wserrorhighlight")) {
 294                 int val = parse_ws_error_highlight(value);
 295                 if (val < 0)
 296                         return -1;
 297                 ws_error_highlight_default = val;
 298                 return 0;
 299         }
 300
 301         if (git_color_config(var, value, cb) < 0)
 302                 return -1;
 303
 304         return git_diff_basic_config(var, value, cb);
 305 }
 306
 307 int git_diff_basic_config(const char *var, const char *value, void *cb)
 308 {
 309         const char *name;
 310
 311         if (!strcmp(var, "diff.renamelimit")) {
 312                 diff_rename_limit_default = git_config_int(var, value);
 313                 return 0;
 314         }
 315
 316         if (userdiff_config(var, value) < 0)
 317                 return -1;
 318
 319         if (skip_prefix(var, "diff.color.", &name) ||
 320             skip_prefix(var, "color.diff.", &name)) {
 321                 int slot = parse_diff_color_slot(name);
 322                 if (slot < 0)
 323                         return 0;
 324                 if (!value)
 325                         return config_error_nonbool(var);
 326                 return color_parse(value, diff_colors[slot]);
 327         }
 328
 329         /* like GNU diff's --suppress-blank-empty option  */
 330         if (!strcmp(var, "diff.suppressblankempty") ||
 331                         /* for backwards compatibility */
 332                         !strcmp(var, "diff.suppress-blank-empty")) {
 333                 diff_suppress_blank_empty = git_config_bool(var, value);
 334                 return 0;
 335         }
 336
 337         if (!strcmp(var, "diff.dirstat")) {
 338                 struct strbuf errmsg = STRBUF_INIT;
 339                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 340                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 341                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 342                                 errmsg.buf);
 343                 strbuf_release(&errmsg);
 344                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 345                 return 0;
 346         }
 347
 348         if (starts_with(var, "submodule."))
 349                 return parse_submodule_config_option(var, value);
 350
 351         if (git_diff_heuristic_config(var, value, cb) < 0)
 352                 return -1;
 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 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_AND_NULL(ecbdata->diff_words);
1222         }
1223 }
1224
1225 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1226 {
1227         if (want_color(diff_use_color))
1228                 return diff_colors[ix];
1229         return "";
1230 }
1231
1232 const char *diff_line_prefix(struct diff_options *opt)
1233 {
1234         struct strbuf *msgbuf;
1235         if (!opt->output_prefix)
1236                 return "";
1237
1238         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1239         return msgbuf->buf;
1240 }
1241
1242 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1243 {
1244         const char *cp;
1245         unsigned long allot;
1246         size_t l = len;
1247
1248         if (ecb->truncate)
1249                 return ecb->truncate(line, len);
1250         cp = line;
1251         allot = l;
1252         while (0 < l) {
1253                 (void) utf8_width(&cp, &l);
1254                 if (!cp)
1255                         break; /* truncated in the middle? */
1256         }
1257         return allot - l;
1258 }
1259
1260 static void find_lno(const char *line, struct emit_callback *ecbdata)
1261 {
1262         const char *p;
1263         ecbdata->lno_in_preimage = 0;
1264         ecbdata->lno_in_postimage = 0;
1265         p = strchr(line, '-');
1266         if (!p)
1267                 return; /* cannot happen */
1268         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1269         p = strchr(p, '+');
1270         if (!p)
1271                 return; /* cannot happen */
1272         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1273 }
1274
1275 static void fn_out_consume(void *priv, char *line, unsigned long len)
1276 {
1277         struct emit_callback *ecbdata = priv;
1278         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1279         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1280         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1281         struct diff_options *o = ecbdata->opt;
1282         const char *line_prefix = diff_line_prefix(o);
1283
1284         o->found_changes = 1;
1285
1286         if (ecbdata->header) {
1287                 fprintf(o->file, "%s", ecbdata->header->buf);
1288                 strbuf_reset(ecbdata->header);
1289                 ecbdata->header = NULL;
1290         }
1291
1292         if (ecbdata->label_path[0]) {
1293                 const char *name_a_tab, *name_b_tab;
1294
1295                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1296                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1297
1298                 fprintf(o->file, "%s%s--- %s%s%s\n",
1299                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1300                 fprintf(o->file, "%s%s+++ %s%s%s\n",
1301                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1302                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1303         }
1304
1305         if (diff_suppress_blank_empty
1306             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1307                 line[0] = '\n';
1308                 len = 1;
1309         }
1310
1311         if (line[0] == '@') {
1312                 if (ecbdata->diff_words)
1313                         diff_words_flush(ecbdata);
1314                 len = sane_truncate_line(ecbdata, line, len);
1315                 find_lno(line, ecbdata);
1316                 emit_hunk_header(ecbdata, line, len);
1317                 if (line[len-1] != '\n')
1318                         putc('\n', o->file);
1319                 return;
1320         }
1321
1322         if (ecbdata->diff_words) {
1323                 if (line[0] == '-') {
1324                         diff_words_append(line, len,
1325                                           &ecbdata->diff_words->minus);
1326                         return;
1327                 } else if (line[0] == '+') {
1328                         diff_words_append(line, len,
1329                                           &ecbdata->diff_words->plus);
1330                         return;
1331                 } else if (starts_with(line, "\\ ")) {
1332                         /*
1333                          * Eat the "no newline at eof" marker as if we
1334                          * saw a "+" or "-" line with nothing on it,
1335                          * and return without diff_words_flush() to
1336                          * defer processing. If this is the end of
1337                          * preimage, more "+" lines may come after it.
1338                          */
1339                         return;
1340                 }
1341                 diff_words_flush(ecbdata);
1342                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1343                         emit_line(o, context, reset, line, len);
1344                         fputs("~\n", o->file);
1345                 } else {
1346                         /*
1347                          * Skip the prefix character, if any.  With
1348                          * diff_suppress_blank_empty, there may be
1349                          * none.
1350                          */
1351                         if (line[0] != '\n') {
1352                               line++;
1353                               len--;
1354                         }
1355                         emit_line(o, context, reset, line, len);
1356                 }
1357                 return;
1358         }
1359
1360         switch (line[0]) {
1361         case '+':
1362                 ecbdata->lno_in_postimage++;
1363                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1364                 break;
1365         case '-':
1366                 ecbdata->lno_in_preimage++;
1367                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1368                 break;
1369         case ' ':
1370                 ecbdata->lno_in_postimage++;
1371                 ecbdata->lno_in_preimage++;
1372                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1373                 break;
1374         default:
1375                 /* incomplete line at the end */
1376                 ecbdata->lno_in_preimage++;
1377                 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1378                           reset, line, len);
1379                 break;
1380         }
1381 }
1382
1383 static char *pprint_rename(const char *a, const char *b)
1384 {
1385         const char *old = a;
1386         const char *new = b;
1387         struct strbuf name = STRBUF_INIT;
1388         int pfx_length, sfx_length;
1389         int pfx_adjust_for_slash;
1390         int len_a = strlen(a);
1391         int len_b = strlen(b);
1392         int a_midlen, b_midlen;
1393         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1394         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1395
1396         if (qlen_a || qlen_b) {
1397                 quote_c_style(a, &name, NULL, 0);
1398                 strbuf_addstr(&name, " => ");
1399                 quote_c_style(b, &name, NULL, 0);
1400                 return strbuf_detach(&name, NULL);
1401         }
1402
1403         /* Find common prefix */
1404         pfx_length = 0;
1405         while (*old && *new && *old == *new) {
1406                 if (*old == '/')
1407                         pfx_length = old - a + 1;
1408                 old++;
1409                 new++;
1410         }
1411
1412         /* Find common suffix */
1413         old = a + len_a;
1414         new = b + len_b;
1415         sfx_length = 0;
1416         /*
1417          * If there is a common prefix, it must end in a slash.  In
1418          * that case we let this loop run 1 into the prefix to see the
1419          * same slash.
1420          *
1421          * If there is no common prefix, we cannot do this as it would
1422          * underrun the input strings.
1423          */
1424         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1425         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1426                b + pfx_length - pfx_adjust_for_slash <= new &&
1427                *old == *new) {
1428                 if (*old == '/')
1429                         sfx_length = len_a - (old - a);
1430                 old--;
1431                 new--;
1432         }
1433
1434         /*
1435          * pfx{mid-a => mid-b}sfx
1436          * {pfx-a => pfx-b}sfx
1437          * pfx{sfx-a => sfx-b}
1438          * name-a => name-b
1439          */
1440         a_midlen = len_a - pfx_length - sfx_length;
1441         b_midlen = len_b - pfx_length - sfx_length;
1442         if (a_midlen < 0)
1443                 a_midlen = 0;
1444         if (b_midlen < 0)
1445                 b_midlen = 0;
1446
1447         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1448         if (pfx_length + sfx_length) {
1449                 strbuf_add(&name, a, pfx_length);
1450                 strbuf_addch(&name, '{');
1451         }
1452         strbuf_add(&name, a + pfx_length, a_midlen);
1453         strbuf_addstr(&name, " => ");
1454         strbuf_add(&name, b + pfx_length, b_midlen);
1455         if (pfx_length + sfx_length) {
1456                 strbuf_addch(&name, '}');
1457                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1458         }
1459         return strbuf_detach(&name, NULL);
1460 }
1461
1462 struct diffstat_t {
1463         int nr;
1464         int alloc;
1465         struct diffstat_file {
1466                 char *from_name;
1467                 char *name;
1468                 char *print_name;
1469                 unsigned is_unmerged:1;
1470                 unsigned is_binary:1;
1471                 unsigned is_renamed:1;
1472                 unsigned is_interesting:1;
1473                 uintmax_t added, deleted;
1474         } **files;
1475 };
1476
1477 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1478                                           const char *name_a,
1479                                           const char *name_b)
1480 {
1481         struct diffstat_file *x;
1482         x = xcalloc(1, sizeof(*x));
1483         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1484         diffstat->files[diffstat->nr++] = x;
1485         if (name_b) {
1486                 x->from_name = xstrdup(name_a);
1487                 x->name = xstrdup(name_b);
1488                 x->is_renamed = 1;
1489         }
1490         else {
1491                 x->from_name = NULL;
1492                 x->name = xstrdup(name_a);
1493         }
1494         return x;
1495 }
1496
1497 static void diffstat_consume(void *priv, char *line, unsigned long len)
1498 {
1499         struct diffstat_t *diffstat = priv;
1500         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1501
1502         if (line[0] == '+')
1503                 x->added++;
1504         else if (line[0] == '-')
1505                 x->deleted++;
1506 }
1507
1508 const char mime_boundary_leader[] = "------------";
1509
1510 static int scale_linear(int it, int width, int max_change)
1511 {
1512         if (!it)
1513                 return 0;
1514         /*
1515          * make sure that at least one '-' or '+' is printed if
1516          * there is any change to this path. The easiest way is to
1517          * scale linearly as if the alloted width is one column shorter
1518          * than it is, and then add 1 to the result.
1519          */
1520         return 1 + (it * (width - 1) / max_change);
1521 }
1522
1523 static void show_name(FILE *file,
1524                       const char *prefix, const char *name, int len)
1525 {
1526         fprintf(file, " %s%-*s |", prefix, len, name);
1527 }
1528
1529 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1530 {
1531         if (cnt <= 0)
1532                 return;
1533         fprintf(file, "%s", set);
1534         while (cnt--)
1535                 putc(ch, file);
1536         fprintf(file, "%s", reset);
1537 }
1538
1539 static void fill_print_name(struct diffstat_file *file)
1540 {
1541         char *pname;
1542
1543         if (file->print_name)
1544                 return;
1545
1546         if (!file->is_renamed) {
1547                 struct strbuf buf = STRBUF_INIT;
1548                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1549                         pname = strbuf_detach(&buf, NULL);
1550                 } else {
1551                         pname = file->name;
1552                         strbuf_release(&buf);
1553                 }
1554         } else {
1555                 pname = pprint_rename(file->from_name, file->name);
1556         }
1557         file->print_name = pname;
1558 }
1559
1560 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1561 {
1562         struct strbuf sb = STRBUF_INIT;
1563         int ret;
1564
1565         if (!files) {
1566                 assert(insertions == 0 && deletions == 0);
1567                 return fprintf(fp, "%s\n", " 0 files changed");
1568         }
1569
1570         strbuf_addf(&sb,
1571                     (files == 1) ? " %d file changed" : " %d files changed",
1572                     files);
1573
1574         /*
1575          * For binary diff, the caller may want to print "x files
1576          * changed" with insertions == 0 && deletions == 0.
1577          *
1578          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1579          * is probably less confusing (i.e skip over "2 files changed
1580          * but nothing about added/removed lines? Is this a bug in Git?").
1581          */
1582         if (insertions || deletions == 0) {
1583                 strbuf_addf(&sb,
1584                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1585                             insertions);
1586         }
1587
1588         if (deletions || insertions == 0) {
1589                 strbuf_addf(&sb,
1590                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1591                             deletions);
1592         }
1593         strbuf_addch(&sb, '\n');
1594         ret = fputs(sb.buf, fp);
1595         strbuf_release(&sb);
1596         return ret;
1597 }
1598
1599 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1600 {
1601         int i, len, add, del, adds = 0, dels = 0;
1602         uintmax_t max_change = 0, max_len = 0;
1603         int total_files = data->nr, count;
1604         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1605         const char *reset, *add_c, *del_c;
1606         const char *line_prefix = "";
1607         int extra_shown = 0;
1608
1609         if (data->nr == 0)
1610                 return;
1611
1612         line_prefix = diff_line_prefix(options);
1613         count = options->stat_count ? options->stat_count : data->nr;
1614
1615         reset = diff_get_color_opt(options, DIFF_RESET);
1616         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1617         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1618
1619         /*
1620          * Find the longest filename and max number of changes
1621          */
1622         for (i = 0; (i < count) && (i < data->nr); i++) {
1623                 struct diffstat_file *file = data->files[i];
1624                 uintmax_t change = file->added + file->deleted;
1625
1626                 if (!file->is_interesting && (change == 0)) {
1627                         count++; /* not shown == room for one more */
1628                         continue;
1629                 }
1630                 fill_print_name(file);
1631                 len = strlen(file->print_name);
1632                 if (max_len < len)
1633                         max_len = len;
1634
1635                 if (file->is_unmerged) {
1636                         /* "Unmerged" is 8 characters */
1637                         bin_width = bin_width < 8 ? 8 : bin_width;
1638                         continue;
1639                 }
1640                 if (file->is_binary) {
1641                         /* "Bin XXX -> YYY bytes" */
1642                         int w = 14 + decimal_width(file->added)
1643                                 + decimal_width(file->deleted);
1644                         bin_width = bin_width < w ? w : bin_width;
1645                         /* Display change counts aligned with "Bin" */
1646                         number_width = 3;
1647                         continue;
1648                 }
1649
1650                 if (max_change < change)
1651                         max_change = change;
1652         }
1653         count = i; /* where we can stop scanning in data->files[] */
1654
1655         /*
1656          * We have width = stat_width or term_columns() columns total.
1657          * We want a maximum of min(max_len, stat_name_width) for the name part.
1658          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1659          * We also need 1 for " " and 4 + decimal_width(max_change)
1660          * for " | NNNN " and one the empty column at the end, altogether
1661          * 6 + decimal_width(max_change).
1662          *
1663          * If there's not enough space, we will use the smaller of
1664          * stat_name_width (if set) and 5/8*width for the filename,
1665          * and the rest for constant elements + graph part, but no more
1666          * than stat_graph_width for the graph part.
1667          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1668          * for the standard terminal size).
1669          *
1670          * In other words: stat_width limits the maximum width, and
1671          * stat_name_width fixes the maximum width of the filename,
1672          * and is also used to divide available columns if there
1673          * aren't enough.
1674          *
1675          * Binary files are displayed with "Bin XXX -> YYY bytes"
1676          * instead of the change count and graph. This part is treated
1677          * similarly to the graph part, except that it is not
1678          * "scaled". If total width is too small to accommodate the
1679          * guaranteed minimum width of the filename part and the
1680          * separators and this message, this message will "overflow"
1681          * making the line longer than the maximum width.
1682          */
1683
1684         if (options->stat_width == -1)
1685                 width = term_columns() - strlen(line_prefix);
1686         else
1687                 width = options->stat_width ? options->stat_width : 80;
1688         number_width = decimal_width(max_change) > number_width ?
1689                 decimal_width(max_change) : number_width;
1690
1691         if (options->stat_graph_width == -1)
1692                 options->stat_graph_width = diff_stat_graph_width;
1693
1694         /*
1695          * Guarantee 3/8*16==6 for the graph part
1696          * and 5/8*16==10 for the filename part
1697          */
1698         if (width < 16 + 6 + number_width)
1699                 width = 16 + 6 + number_width;
1700
1701         /*
1702          * First assign sizes that are wanted, ignoring available width.
1703          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1704          * starting from "XXX" should fit in graph_width.
1705          */
1706         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1707         if (options->stat_graph_width &&
1708             options->stat_graph_width < graph_width)
1709                 graph_width = options->stat_graph_width;
1710
1711         name_width = (options->stat_name_width > 0 &&
1712                       options->stat_name_width < max_len) ?
1713                 options->stat_name_width : max_len;
1714
1715         /*
1716          * Adjust adjustable widths not to exceed maximum width
1717          */
1718         if (name_width + number_width + 6 + graph_width > width) {
1719                 if (graph_width > width * 3/8 - number_width - 6) {
1720                         graph_width = width * 3/8 - number_width - 6;
1721                         if (graph_width < 6)
1722                                 graph_width = 6;
1723                 }
1724
1725                 if (options->stat_graph_width &&
1726                     graph_width > options->stat_graph_width)
1727                         graph_width = options->stat_graph_width;
1728                 if (name_width > width - number_width - 6 - graph_width)
1729                         name_width = width - number_width - 6 - graph_width;
1730                 else
1731                         graph_width = width - number_width - 6 - name_width;
1732         }
1733
1734         /*
1735          * From here name_width is the width of the name area,
1736          * and graph_width is the width of the graph area.
1737          * max_change is used to scale graph properly.
1738          */
1739         for (i = 0; i < count; i++) {
1740                 const char *prefix = "";
1741                 struct diffstat_file *file = data->files[i];
1742                 char *name = file->print_name;
1743                 uintmax_t added = file->added;
1744                 uintmax_t deleted = file->deleted;
1745                 int name_len;
1746
1747                 if (!file->is_interesting && (added + deleted == 0))
1748                         continue;
1749
1750                 /*
1751                  * "scale" the filename
1752                  */
1753                 len = name_width;
1754                 name_len = strlen(name);
1755                 if (name_width < name_len) {
1756                         char *slash;
1757                         prefix = "...";
1758                         len -= 3;
1759                         name += name_len - len;
1760                         slash = strchr(name, '/');
1761                         if (slash)
1762                                 name = slash;
1763                 }
1764
1765                 if (file->is_binary) {
1766                         fprintf(options->file, "%s", line_prefix);
1767                         show_name(options->file, prefix, name, len);
1768                         fprintf(options->file, " %*s", number_width, "Bin");
1769                         if (!added && !deleted) {
1770                                 putc('\n', options->file);
1771                                 continue;
1772                         }
1773                         fprintf(options->file, " %s%"PRIuMAX"%s",
1774                                 del_c, deleted, reset);
1775                         fprintf(options->file, " -> ");
1776                         fprintf(options->file, "%s%"PRIuMAX"%s",
1777                                 add_c, added, reset);
1778                         fprintf(options->file, " bytes");
1779                         fprintf(options->file, "\n");
1780                         continue;
1781                 }
1782                 else if (file->is_unmerged) {
1783                         fprintf(options->file, "%s", line_prefix);
1784                         show_name(options->file, prefix, name, len);
1785                         fprintf(options->file, " Unmerged\n");
1786                         continue;
1787                 }
1788
1789                 /*
1790                  * scale the add/delete
1791                  */
1792                 add = added;
1793                 del = deleted;
1794
1795                 if (graph_width <= max_change) {
1796                         int total = scale_linear(add + del, graph_width, max_change);
1797                         if (total < 2 && add && del)
1798                                 /* width >= 2 due to the sanity check */
1799                                 total = 2;
1800                         if (add < del) {
1801                                 add = scale_linear(add, graph_width, max_change);
1802                                 del = total - add;
1803                         } else {
1804                                 del = scale_linear(del, graph_width, max_change);
1805                                 add = total - del;
1806                         }
1807                 }
1808                 fprintf(options->file, "%s", line_prefix);
1809                 show_name(options->file, prefix, name, len);
1810                 fprintf(options->file, " %*"PRIuMAX"%s",
1811                         number_width, added + deleted,
1812                         added + deleted ? " " : "");
1813                 show_graph(options->file, '+', add, add_c, reset);
1814                 show_graph(options->file, '-', del, del_c, reset);
1815                 fprintf(options->file, "\n");
1816         }
1817
1818         for (i = 0; i < data->nr; i++) {
1819                 struct diffstat_file *file = data->files[i];
1820                 uintmax_t added = file->added;
1821                 uintmax_t deleted = file->deleted;
1822
1823                 if (file->is_unmerged ||
1824                     (!file->is_interesting && (added + deleted == 0))) {
1825                         total_files--;
1826                         continue;
1827                 }
1828
1829                 if (!file->is_binary) {
1830                         adds += added;
1831                         dels += deleted;
1832                 }
1833                 if (i < count)
1834                         continue;
1835                 if (!extra_shown)
1836                         fprintf(options->file, "%s ...\n", line_prefix);
1837                 extra_shown = 1;
1838         }
1839         fprintf(options->file, "%s", line_prefix);
1840         print_stat_summary(options->file, total_files, adds, dels);
1841 }
1842
1843 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1844 {
1845         int i, adds = 0, dels = 0, total_files = data->nr;
1846
1847         if (data->nr == 0)
1848                 return;
1849
1850         for (i = 0; i < data->nr; i++) {
1851                 int added = data->files[i]->added;
1852                 int deleted= data->files[i]->deleted;
1853
1854                 if (data->files[i]->is_unmerged ||
1855                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1856                         total_files--;
1857                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1858                         adds += added;
1859                         dels += deleted;
1860                 }
1861         }
1862         fprintf(options->file, "%s", diff_line_prefix(options));
1863         print_stat_summary(options->file, total_files, adds, dels);
1864 }
1865
1866 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1867 {
1868         int i;
1869
1870         if (data->nr == 0)
1871                 return;
1872
1873         for (i = 0; i < data->nr; i++) {
1874                 struct diffstat_file *file = data->files[i];
1875
1876                 fprintf(options->file, "%s", diff_line_prefix(options));
1877
1878                 if (file->is_binary)
1879                         fprintf(options->file, "-\t-\t");
1880                 else
1881                         fprintf(options->file,
1882                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1883                                 file->added, file->deleted);
1884                 if (options->line_termination) {
1885                         fill_print_name(file);
1886                         if (!file->is_renamed)
1887                                 write_name_quoted(file->name, options->file,
1888                                                   options->line_termination);
1889                         else {
1890                                 fputs(file->print_name, options->file);
1891                                 putc(options->line_termination, options->file);
1892                         }
1893                 } else {
1894                         if (file->is_renamed) {
1895                                 putc('\0', options->file);
1896                                 write_name_quoted(file->from_name, options->file, '\0');
1897                         }
1898                         write_name_quoted(file->name, options->file, '\0');
1899                 }
1900         }
1901 }
1902
1903 struct dirstat_file {
1904         const char *name;
1905         unsigned long changed;
1906 };
1907
1908 struct dirstat_dir {
1909         struct dirstat_file *files;
1910         int alloc, nr, permille, cumulative;
1911 };
1912
1913 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1914                 unsigned long changed, const char *base, int baselen)
1915 {
1916         unsigned long this_dir = 0;
1917         unsigned int sources = 0;
1918         const char *line_prefix = diff_line_prefix(opt);
1919
1920         while (dir->nr) {
1921                 struct dirstat_file *f = dir->files;
1922                 int namelen = strlen(f->name);
1923                 unsigned long this;
1924                 char *slash;
1925
1926                 if (namelen < baselen)
1927                         break;
1928                 if (memcmp(f->name, base, baselen))
1929                         break;
1930                 slash = strchr(f->name + baselen, '/');
1931                 if (slash) {
1932                         int newbaselen = slash + 1 - f->name;
1933                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1934                         sources++;
1935                 } else {
1936                         this = f->changed;
1937                         dir->files++;
1938                         dir->nr--;
1939                         sources += 2;
1940                 }
1941                 this_dir += this;
1942         }
1943
1944         /*
1945          * We don't report dirstat's for
1946          *  - the top level
1947          *  - or cases where everything came from a single directory
1948          *    under this directory (sources == 1).
1949          */
1950         if (baselen && sources != 1) {
1951                 if (this_dir) {
1952                         int permille = this_dir * 1000 / changed;
1953                         if (permille >= dir->permille) {
1954                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1955                                         permille / 10, permille % 10, baselen, base);
1956                                 if (!dir->cumulative)
1957                                         return 0;
1958                         }
1959                 }
1960         }
1961         return this_dir;
1962 }
1963
1964 static int dirstat_compare(const void *_a, const void *_b)
1965 {
1966         const struct dirstat_file *a = _a;
1967         const struct dirstat_file *b = _b;
1968         return strcmp(a->name, b->name);
1969 }
1970
1971 static void show_dirstat(struct diff_options *options)
1972 {
1973         int i;
1974         unsigned long changed;
1975         struct dirstat_dir dir;
1976         struct diff_queue_struct *q = &diff_queued_diff;
1977
1978         dir.files = NULL;
1979         dir.alloc = 0;
1980         dir.nr = 0;
1981         dir.permille = options->dirstat_permille;
1982         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1983
1984         changed = 0;
1985         for (i = 0; i < q->nr; i++) {
1986                 struct diff_filepair *p = q->queue[i];
1987                 const char *name;
1988                 unsigned long copied, added, damage;
1989                 int content_changed;
1990
1991                 name = p->two->path ? p->two->path : p->one->path;
1992
1993                 if (p->one->oid_valid && p->two->oid_valid)
1994                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
1995                 else
1996                         content_changed = 1;
1997
1998                 if (!content_changed) {
1999                         /*
2000                          * The SHA1 has not changed, so pre-/post-content is
2001                          * identical. We can therefore skip looking at the
2002                          * file contents altogether.
2003                          */
2004                         damage = 0;
2005                         goto found_damage;
2006                 }
2007
2008                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2009                         /*
2010                          * In --dirstat-by-file mode, we don't really need to
2011                          * look at the actual file contents at all.
2012                          * The fact that the SHA1 changed is enough for us to
2013                          * add this file to the list of results
2014                          * (with each file contributing equal damage).
2015                          */
2016                         damage = 1;
2017                         goto found_damage;
2018                 }
2019
2020                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2021                         diff_populate_filespec(p->one, 0);
2022                         diff_populate_filespec(p->two, 0);
2023                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2024                                                &copied, &added);
2025                         diff_free_filespec_data(p->one);
2026                         diff_free_filespec_data(p->two);
2027                 } else if (DIFF_FILE_VALID(p->one)) {
2028                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2029                         copied = added = 0;
2030                         diff_free_filespec_data(p->one);
2031                 } else if (DIFF_FILE_VALID(p->two)) {
2032                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2033                         copied = 0;
2034                         added = p->two->size;
2035                         diff_free_filespec_data(p->two);
2036                 } else
2037                         continue;
2038
2039                 /*
2040                  * Original minus copied is the removed material,
2041                  * added is the new material.  They are both damages
2042                  * made to the preimage.
2043                  * If the resulting damage is zero, we know that
2044                  * diffcore_count_changes() considers the two entries to
2045                  * be identical, but since content_changed is true, we
2046                  * know that there must have been _some_ kind of change,
2047                  * so we force all entries to have damage > 0.
2048                  */
2049                 damage = (p->one->size - copied) + added;
2050                 if (!damage)
2051                         damage = 1;
2052
2053 found_damage:
2054                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2055                 dir.files[dir.nr].name = name;
2056                 dir.files[dir.nr].changed = damage;
2057                 changed += damage;
2058                 dir.nr++;
2059         }
2060
2061         /* This can happen even with many files, if everything was renames */
2062         if (!changed)
2063                 return;
2064
2065         /* Show all directories with more than x% of the changes */
2066         QSORT(dir.files, dir.nr, dirstat_compare);
2067         gather_dirstat(options, &dir, changed, "", 0);
2068 }
2069
2070 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2071 {
2072         int i;
2073         unsigned long changed;
2074         struct dirstat_dir dir;
2075
2076         if (data->nr == 0)
2077                 return;
2078
2079         dir.files = NULL;
2080         dir.alloc = 0;
2081         dir.nr = 0;
2082         dir.permille = options->dirstat_permille;
2083         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2084
2085         changed = 0;
2086         for (i = 0; i < data->nr; i++) {
2087                 struct diffstat_file *file = data->files[i];
2088                 unsigned long damage = file->added + file->deleted;
2089                 if (file->is_binary)
2090                         /*
2091                          * binary files counts bytes, not lines. Must find some
2092                          * way to normalize binary bytes vs. textual lines.
2093                          * The following heuristic assumes that there are 64
2094                          * bytes per "line".
2095                          * This is stupid and ugly, but very cheap...
2096                          */
2097                         damage = (damage + 63) / 64;
2098                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2099                 dir.files[dir.nr].name = file->name;
2100                 dir.files[dir.nr].changed = damage;
2101                 changed += damage;
2102                 dir.nr++;
2103         }
2104
2105         /* This can happen even with many files, if everything was renames */
2106         if (!changed)
2107                 return;
2108
2109         /* Show all directories with more than x% of the changes */
2110         QSORT(dir.files, dir.nr, dirstat_compare);
2111         gather_dirstat(options, &dir, changed, "", 0);
2112 }
2113
2114 static void free_diffstat_info(struct diffstat_t *diffstat)
2115 {
2116         int i;
2117         for (i = 0; i < diffstat->nr; i++) {
2118                 struct diffstat_file *f = diffstat->files[i];
2119                 if (f->name != f->print_name)
2120                         free(f->print_name);
2121                 free(f->name);
2122                 free(f->from_name);
2123                 free(f);
2124         }
2125         free(diffstat->files);
2126 }
2127
2128 struct checkdiff_t {
2129         const char *filename;
2130         int lineno;
2131         int conflict_marker_size;
2132         struct diff_options *o;
2133         unsigned ws_rule;
2134         unsigned status;
2135 };
2136
2137 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2138 {
2139         char firstchar;
2140         int cnt;
2141
2142         if (len < marker_size + 1)
2143                 return 0;
2144         firstchar = line[0];
2145         switch (firstchar) {
2146         case '=': case '>': case '<': case '|':
2147                 break;
2148         default:
2149                 return 0;
2150         }
2151         for (cnt = 1; cnt < marker_size; cnt++)
2152                 if (line[cnt] != firstchar)
2153                         return 0;
2154         /* line[1] thru line[marker_size-1] are same as firstchar */
2155         if (len < marker_size + 1 || !isspace(line[marker_size]))
2156                 return 0;
2157         return 1;
2158 }
2159
2160 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2161 {
2162         struct checkdiff_t *data = priv;
2163         int marker_size = data->conflict_marker_size;
2164         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2165         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2166         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2167         char *err;
2168         const char *line_prefix;
2169
2170         assert(data->o);
2171         line_prefix = diff_line_prefix(data->o);
2172
2173         if (line[0] == '+') {
2174                 unsigned bad;
2175                 data->lineno++;
2176                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2177                         data->status |= 1;
2178                         fprintf(data->o->file,
2179                                 "%s%s:%d: leftover conflict marker\n",
2180                                 line_prefix, data->filename, data->lineno);
2181                 }
2182                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2183                 if (!bad)
2184                         return;
2185                 data->status |= bad;
2186                 err = whitespace_error_string(bad);
2187                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2188                         line_prefix, data->filename, data->lineno, err);
2189                 free(err);
2190                 emit_line(data->o, set, reset, line, 1);
2191                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2192                               data->o->file, set, reset, ws);
2193         } else if (line[0] == ' ') {
2194                 data->lineno++;
2195         } else if (line[0] == '@') {
2196                 char *plus = strchr(line, '+');
2197                 if (plus)
2198                         data->lineno = strtol(plus, NULL, 10) - 1;
2199                 else
2200                         die("invalid diff");
2201         }
2202 }
2203
2204 static unsigned char *deflate_it(char *data,
2205                                  unsigned long size,
2206                                  unsigned long *result_size)
2207 {
2208         int bound;
2209         unsigned char *deflated;
2210         git_zstream stream;
2211
2212         git_deflate_init(&stream, zlib_compression_level);
2213         bound = git_deflate_bound(&stream, size);
2214         deflated = xmalloc(bound);
2215         stream.next_out = deflated;
2216         stream.avail_out = bound;
2217
2218         stream.next_in = (unsigned char *)data;
2219         stream.avail_in = size;
2220         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2221                 ; /* nothing */
2222         git_deflate_end(&stream);
2223         *result_size = stream.total_out;
2224         return deflated;
2225 }
2226
2227 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2228                                   const char *prefix)
2229 {
2230         void *cp;
2231         void *delta;
2232         void *deflated;
2233         void *data;
2234         unsigned long orig_size;
2235         unsigned long delta_size;
2236         unsigned long deflate_size;
2237         unsigned long data_size;
2238
2239         /* We could do deflated delta, or we could do just deflated two,
2240          * whichever is smaller.
2241          */
2242         delta = NULL;
2243         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2244         if (one->size && two->size) {
2245                 delta = diff_delta(one->ptr, one->size,
2246                                    two->ptr, two->size,
2247                                    &delta_size, deflate_size);
2248                 if (delta) {
2249                         void *to_free = delta;
2250                         orig_size = delta_size;
2251                         delta = deflate_it(delta, delta_size, &delta_size);
2252                         free(to_free);
2253                 }
2254         }
2255
2256         if (delta && delta_size < deflate_size) {
2257                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2258                 free(deflated);
2259                 data = delta;
2260                 data_size = delta_size;
2261         }
2262         else {
2263                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2264                 free(delta);
2265                 data = deflated;
2266                 data_size = deflate_size;
2267         }
2268
2269         /* emit data encoded in base85 */
2270         cp = data;
2271         while (data_size) {
2272                 int bytes = (52 < data_size) ? 52 : data_size;
2273                 char line[70];
2274                 data_size -= bytes;
2275                 if (bytes <= 26)
2276                         line[0] = bytes + 'A' - 1;
2277                 else
2278                         line[0] = bytes - 26 + 'a' - 1;
2279                 encode_85(line + 1, cp, bytes);
2280                 cp = (char *) cp + bytes;
2281                 fprintf(file, "%s", prefix);
2282                 fputs(line, file);
2283                 fputc('\n', file);
2284         }
2285         fprintf(file, "%s\n", prefix);
2286         free(data);
2287 }
2288
2289 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2290                              const char *prefix)
2291 {
2292         fprintf(file, "%sGIT binary patch\n", prefix);
2293         emit_binary_diff_body(file, one, two, prefix);
2294         emit_binary_diff_body(file, two, one, prefix);
2295 }
2296
2297 int diff_filespec_is_binary(struct diff_filespec *one)
2298 {
2299         if (one->is_binary == -1) {
2300                 diff_filespec_load_driver(one);
2301                 if (one->driver->binary != -1)
2302                         one->is_binary = one->driver->binary;
2303                 else {
2304                         if (!one->data && DIFF_FILE_VALID(one))
2305                                 diff_populate_filespec(one, CHECK_BINARY);
2306                         if (one->is_binary == -1 && one->data)
2307                                 one->is_binary = buffer_is_binary(one->data,
2308                                                 one->size);
2309                         if (one->is_binary == -1)
2310                                 one->is_binary = 0;
2311                 }
2312         }
2313         return one->is_binary;
2314 }
2315
2316 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2317 {
2318         diff_filespec_load_driver(one);
2319         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2320 }
2321
2322 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2323 {
2324         if (!options->a_prefix)
2325                 options->a_prefix = a;
2326         if (!options->b_prefix)
2327                 options->b_prefix = b;
2328 }
2329
2330 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2331 {
2332         if (!DIFF_FILE_VALID(one))
2333                 return NULL;
2334
2335         diff_filespec_load_driver(one);
2336         return userdiff_get_textconv(one->driver);
2337 }
2338
2339 static void builtin_diff(const char *name_a,
2340                          const char *name_b,
2341                          struct diff_filespec *one,
2342                          struct diff_filespec *two,
2343                          const char *xfrm_msg,
2344                          int must_show_header,
2345                          struct diff_options *o,
2346                          int complete_rewrite)
2347 {
2348         mmfile_t mf1, mf2;
2349         const char *lbl[2];
2350         char *a_one, *b_two;
2351         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2352         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2353         const char *a_prefix, *b_prefix;
2354         struct userdiff_driver *textconv_one = NULL;
2355         struct userdiff_driver *textconv_two = NULL;
2356         struct strbuf header = STRBUF_INIT;
2357         const char *line_prefix = diff_line_prefix(o);
2358
2359         diff_set_mnemonic_prefix(o, "a/", "b/");
2360         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2361                 a_prefix = o->b_prefix;
2362                 b_prefix = o->a_prefix;
2363         } else {
2364                 a_prefix = o->a_prefix;
2365                 b_prefix = o->b_prefix;
2366         }
2367
2368         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2369             (!one->mode || S_ISGITLINK(one->mode)) &&
2370             (!two->mode || S_ISGITLINK(two->mode))) {
2371                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2372                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2373                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2374                                 line_prefix,
2375                                 &one->oid, &two->oid,
2376                                 two->dirty_submodule,
2377                                 meta, del, add, reset);
2378                 return;
2379         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2380                    (!one->mode || S_ISGITLINK(one->mode)) &&
2381                    (!two->mode || S_ISGITLINK(two->mode))) {
2382                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2383                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2384                 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2385                                 line_prefix,
2386                                 &one->oid, &two->oid,
2387                                 two->dirty_submodule,
2388                                 meta, del, add, reset, o);
2389                 return;
2390         }
2391
2392         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2393                 textconv_one = get_textconv(one);
2394                 textconv_two = get_textconv(two);
2395         }
2396
2397         /* Never use a non-valid filename anywhere if at all possible */
2398         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2399         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2400
2401         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2402         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2403         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2404         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2405         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2406         if (lbl[0][0] == '/') {
2407                 /* /dev/null */
2408                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2409                 if (xfrm_msg)
2410                         strbuf_addstr(&header, xfrm_msg);
2411                 must_show_header = 1;
2412         }
2413         else if (lbl[1][0] == '/') {
2414                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2415                 if (xfrm_msg)
2416                         strbuf_addstr(&header, xfrm_msg);
2417                 must_show_header = 1;
2418         }
2419         else {
2420                 if (one->mode != two->mode) {
2421                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2422                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2423                         must_show_header = 1;
2424                 }
2425                 if (xfrm_msg)
2426                         strbuf_addstr(&header, xfrm_msg);
2427
2428                 /*
2429                  * we do not run diff between different kind
2430                  * of objects.
2431                  */
2432                 if ((one->mode ^ two->mode) & S_IFMT)
2433                         goto free_ab_and_return;
2434                 if (complete_rewrite &&
2435                     (textconv_one || !diff_filespec_is_binary(one)) &&
2436                     (textconv_two || !diff_filespec_is_binary(two))) {
2437                         fprintf(o->file, "%s", header.buf);
2438                         strbuf_reset(&header);
2439                         emit_rewrite_diff(name_a, name_b, one, two,
2440                                                 textconv_one, textconv_two, o);
2441                         o->found_changes = 1;
2442                         goto free_ab_and_return;
2443                 }
2444         }
2445
2446         if (o->irreversible_delete && lbl[1][0] == '/') {
2447                 fprintf(o->file, "%s", header.buf);
2448                 strbuf_reset(&header);
2449                 goto free_ab_and_return;
2450         } else if (!DIFF_OPT_TST(o, TEXT) &&
2451             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2452               (!textconv_two && diff_filespec_is_binary(two)) )) {
2453                 if (!one->data && !two->data &&
2454                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2455                     !DIFF_OPT_TST(o, BINARY)) {
2456                         if (!oidcmp(&one->oid, &two->oid)) {
2457                                 if (must_show_header)
2458                                         fprintf(o->file, "%s", header.buf);
2459                                 goto free_ab_and_return;
2460                         }
2461                         fprintf(o->file, "%s", header.buf);
2462                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2463                                 line_prefix, lbl[0], lbl[1]);
2464                         goto free_ab_and_return;
2465                 }
2466                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2467                         die("unable to read files to diff");
2468                 /* Quite common confusing case */
2469                 if (mf1.size == mf2.size &&
2470                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2471                         if (must_show_header)
2472                                 fprintf(o->file, "%s", header.buf);
2473                         goto free_ab_and_return;
2474                 }
2475                 fprintf(o->file, "%s", header.buf);
2476                 strbuf_reset(&header);
2477                 if (DIFF_OPT_TST(o, BINARY))
2478                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2479                 else
2480                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2481                                 line_prefix, lbl[0], lbl[1]);
2482                 o->found_changes = 1;
2483         } else {
2484                 /* Crazy xdl interfaces.. */
2485                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2486                 const char *v;
2487                 xpparam_t xpp;
2488                 xdemitconf_t xecfg;
2489                 struct emit_callback ecbdata;
2490                 const struct userdiff_funcname *pe;
2491
2492                 if (must_show_header) {
2493                         fprintf(o->file, "%s", header.buf);
2494                         strbuf_reset(&header);
2495                 }
2496
2497                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2498                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2499
2500                 pe = diff_funcname_pattern(one);
2501                 if (!pe)
2502                         pe = diff_funcname_pattern(two);
2503
2504                 memset(&xpp, 0, sizeof(xpp));
2505                 memset(&xecfg, 0, sizeof(xecfg));
2506                 memset(&ecbdata, 0, sizeof(ecbdata));
2507                 ecbdata.label_path = lbl;
2508                 ecbdata.color_diff = want_color(o->use_color);
2509                 ecbdata.ws_rule = whitespace_rule(name_b);
2510                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2511                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2512                 ecbdata.opt = o;
2513                 ecbdata.header = header.len ? &header : NULL;
2514                 xpp.flags = o->xdl_opts;
2515                 xecfg.ctxlen = o->context;
2516                 xecfg.interhunkctxlen = o->interhunkcontext;
2517                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2518                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2519                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2520                 if (pe)
2521                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2522                 if (!diffopts)
2523                         ;
2524                 else if (skip_prefix(diffopts, "--unified=", &v))
2525                         xecfg.ctxlen = strtoul(v, NULL, 10);
2526                 else if (skip_prefix(diffopts, "-u", &v))
2527                         xecfg.ctxlen = strtoul(v, NULL, 10);
2528                 if (o->word_diff)
2529                         init_diff_words_data(&ecbdata, o, one, two);
2530                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2531                                   &xpp, &xecfg))
2532                         die("unable to generate diff for %s", one->path);
2533                 if (o->word_diff)
2534                         free_diff_words_data(&ecbdata);
2535                 if (textconv_one)
2536                         free(mf1.ptr);
2537                 if (textconv_two)
2538                         free(mf2.ptr);
2539                 xdiff_clear_find_func(&xecfg);
2540         }
2541
2542  free_ab_and_return:
2543         strbuf_release(&header);
2544         diff_free_filespec_data(one);
2545         diff_free_filespec_data(two);
2546         free(a_one);
2547         free(b_two);
2548         return;
2549 }
2550
2551 static void builtin_diffstat(const char *name_a, const char *name_b,
2552                              struct diff_filespec *one,
2553                              struct diff_filespec *two,
2554                              struct diffstat_t *diffstat,
2555                              struct diff_options *o,
2556                              struct diff_filepair *p)
2557 {
2558         mmfile_t mf1, mf2;
2559         struct diffstat_file *data;
2560         int same_contents;
2561         int complete_rewrite = 0;
2562
2563         if (!DIFF_PAIR_UNMERGED(p)) {
2564                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2565                         complete_rewrite = 1;
2566         }
2567
2568         data = diffstat_add(diffstat, name_a, name_b);
2569         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2570
2571         if (!one || !two) {
2572                 data->is_unmerged = 1;
2573                 return;
2574         }
2575
2576         same_contents = !oidcmp(&one->oid, &two->oid);
2577
2578         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2579                 data->is_binary = 1;
2580                 if (same_contents) {
2581                         data->added = 0;
2582                         data->deleted = 0;
2583                 } else {
2584                         data->added = diff_filespec_size(two);
2585                         data->deleted = diff_filespec_size(one);
2586                 }
2587         }
2588
2589         else if (complete_rewrite) {
2590                 diff_populate_filespec(one, 0);
2591                 diff_populate_filespec(two, 0);
2592                 data->deleted = count_lines(one->data, one->size);
2593                 data->added = count_lines(two->data, two->size);
2594         }
2595
2596         else if (!same_contents) {
2597                 /* Crazy xdl interfaces.. */
2598                 xpparam_t xpp;
2599                 xdemitconf_t xecfg;
2600
2601                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2602                         die("unable to read files to diff");
2603
2604                 memset(&xpp, 0, sizeof(xpp));
2605                 memset(&xecfg, 0, sizeof(xecfg));
2606                 xpp.flags = o->xdl_opts;
2607                 xecfg.ctxlen = o->context;
2608                 xecfg.interhunkctxlen = o->interhunkcontext;
2609                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2610                                   &xpp, &xecfg))
2611                         die("unable to generate diffstat for %s", one->path);
2612         }
2613
2614         diff_free_filespec_data(one);
2615         diff_free_filespec_data(two);
2616 }
2617
2618 static void builtin_checkdiff(const char *name_a, const char *name_b,
2619                               const char *attr_path,
2620                               struct diff_filespec *one,
2621                               struct diff_filespec *two,
2622                               struct diff_options *o)
2623 {
2624         mmfile_t mf1, mf2;
2625         struct checkdiff_t data;
2626
2627         if (!two)
2628                 return;
2629
2630         memset(&data, 0, sizeof(data));
2631         data.filename = name_b ? name_b : name_a;
2632         data.lineno = 0;
2633         data.o = o;
2634         data.ws_rule = whitespace_rule(attr_path);
2635         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2636
2637         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2638                 die("unable to read files to diff");
2639
2640         /*
2641          * All the other codepaths check both sides, but not checking
2642          * the "old" side here is deliberate.  We are checking the newly
2643          * introduced changes, and as long as the "new" side is text, we
2644          * can and should check what it introduces.
2645          */
2646         if (diff_filespec_is_binary(two))
2647                 goto free_and_return;
2648         else {
2649                 /* Crazy xdl interfaces.. */
2650                 xpparam_t xpp;
2651                 xdemitconf_t xecfg;
2652
2653                 memset(&xpp, 0, sizeof(xpp));
2654                 memset(&xecfg, 0, sizeof(xecfg));
2655                 xecfg.ctxlen = 1; /* at least one context line */
2656                 xpp.flags = 0;
2657                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2658                                   &xpp, &xecfg))
2659                         die("unable to generate checkdiff for %s", one->path);
2660
2661                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2662                         struct emit_callback ecbdata;
2663                         int blank_at_eof;
2664
2665                         ecbdata.ws_rule = data.ws_rule;
2666                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2667                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2668
2669                         if (blank_at_eof) {
2670                                 static char *err;
2671                                 if (!err)
2672                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2673                                 fprintf(o->file, "%s:%d: %s.\n",
2674                                         data.filename, blank_at_eof, err);
2675                                 data.status = 1; /* report errors */
2676                         }
2677                 }
2678         }
2679  free_and_return:
2680         diff_free_filespec_data(one);
2681         diff_free_filespec_data(two);
2682         if (data.status)
2683                 DIFF_OPT_SET(o, CHECK_FAILED);
2684 }
2685
2686 struct diff_filespec *alloc_filespec(const char *path)
2687 {
2688         struct diff_filespec *spec;
2689
2690         FLEXPTR_ALLOC_STR(spec, path, path);
2691         spec->count = 1;
2692         spec->is_binary = -1;
2693         return spec;
2694 }
2695
2696 void free_filespec(struct diff_filespec *spec)
2697 {
2698         if (!--spec->count) {
2699                 diff_free_filespec_data(spec);
2700                 free(spec);
2701         }
2702 }
2703
2704 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2705                    int sha1_valid, unsigned short mode)
2706 {
2707         if (mode) {
2708                 spec->mode = canon_mode(mode);
2709                 hashcpy(spec->oid.hash, sha1);
2710                 spec->oid_valid = sha1_valid;
2711         }
2712 }
2713
2714 /*
2715  * Given a name and sha1 pair, if the index tells us the file in
2716  * the work tree has that object contents, return true, so that
2717  * prepare_temp_file() does not have to inflate and extract.
2718  */
2719 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2720 {
2721         const struct cache_entry *ce;
2722         struct stat st;
2723         int pos, len;
2724
2725         /*
2726          * We do not read the cache ourselves here, because the
2727          * benchmark with my previous version that always reads cache
2728          * shows that it makes things worse for diff-tree comparing
2729          * two linux-2.6 kernel trees in an already checked out work
2730          * tree.  This is because most diff-tree comparisons deal with
2731          * only a small number of files, while reading the cache is
2732          * expensive for a large project, and its cost outweighs the
2733          * savings we get by not inflating the object to a temporary
2734          * file.  Practically, this code only helps when we are used
2735          * by diff-cache --cached, which does read the cache before
2736          * calling us.
2737          */
2738         if (!active_cache)
2739                 return 0;
2740
2741         /* We want to avoid the working directory if our caller
2742          * doesn't need the data in a normal file, this system
2743          * is rather slow with its stat/open/mmap/close syscalls,
2744          * and the object is contained in a pack file.  The pack
2745          * is probably already open and will be faster to obtain
2746          * the data through than the working directory.  Loose
2747          * objects however would tend to be slower as they need
2748          * to be individually opened and inflated.
2749          */
2750         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2751                 return 0;
2752
2753         /*
2754          * Similarly, if we'd have to convert the file contents anyway, that
2755          * makes the optimization not worthwhile.
2756          */
2757         if (!want_file && would_convert_to_git(name))
2758                 return 0;
2759
2760         len = strlen(name);
2761         pos = cache_name_pos(name, len);
2762         if (pos < 0)
2763                 return 0;
2764         ce = active_cache[pos];
2765
2766         /*
2767          * This is not the sha1 we are looking for, or
2768          * unreusable because it is not a regular file.
2769          */
2770         if (hashcmp(sha1, ce->oid.hash) || !S_ISREG(ce->ce_mode))
2771                 return 0;
2772
2773         /*
2774          * If ce is marked as "assume unchanged", there is no
2775          * guarantee that work tree matches what we are looking for.
2776          */
2777         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2778                 return 0;
2779
2780         /*
2781          * If ce matches the file in the work tree, we can reuse it.
2782          */
2783         if (ce_uptodate(ce) ||
2784             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2785                 return 1;
2786
2787         return 0;
2788 }
2789
2790 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2791 {
2792         struct strbuf buf = STRBUF_INIT;
2793         char *dirty = "";
2794
2795         /* Are we looking at the work tree? */
2796         if (s->dirty_submodule)
2797                 dirty = "-dirty";
2798
2799         strbuf_addf(&buf, "Subproject commit %s%s\n",
2800                     oid_to_hex(&s->oid), dirty);
2801         s->size = buf.len;
2802         if (size_only) {
2803                 s->data = NULL;
2804                 strbuf_release(&buf);
2805         } else {
2806                 s->data = strbuf_detach(&buf, NULL);
2807                 s->should_free = 1;
2808         }
2809         return 0;
2810 }
2811
2812 /*
2813  * While doing rename detection and pickaxe operation, we may need to
2814  * grab the data for the blob (or file) for our own in-core comparison.
2815  * diff_filespec has data and size fields for this purpose.
2816  */
2817 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2818 {
2819         int size_only = flags & CHECK_SIZE_ONLY;
2820         int err = 0;
2821         /*
2822          * demote FAIL to WARN to allow inspecting the situation
2823          * instead of refusing.
2824          */
2825         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2826                                     ? SAFE_CRLF_WARN
2827                                     : safe_crlf);
2828
2829         if (!DIFF_FILE_VALID(s))
2830                 die("internal error: asking to populate invalid file.");
2831         if (S_ISDIR(s->mode))
2832                 return -1;
2833
2834         if (s->data)
2835                 return 0;
2836
2837         if (size_only && 0 < s->size)
2838                 return 0;
2839
2840         if (S_ISGITLINK(s->mode))
2841                 return diff_populate_gitlink(s, size_only);
2842
2843         if (!s->oid_valid ||
2844             reuse_worktree_file(s->path, s->oid.hash, 0)) {
2845                 struct strbuf buf = STRBUF_INIT;
2846                 struct stat st;
2847                 int fd;
2848
2849                 if (lstat(s->path, &st) < 0) {
2850                         if (errno == ENOENT) {
2851                         err_empty:
2852                                 err = -1;
2853                         empty:
2854                                 s->data = (char *)"";
2855                                 s->size = 0;
2856                                 return err;
2857                         }
2858                 }
2859                 s->size = xsize_t(st.st_size);
2860                 if (!s->size)
2861                         goto empty;
2862                 if (S_ISLNK(st.st_mode)) {
2863                         struct strbuf sb = STRBUF_INIT;
2864
2865                         if (strbuf_readlink(&sb, s->path, s->size))
2866                                 goto err_empty;
2867                         s->size = sb.len;
2868                         s->data = strbuf_detach(&sb, NULL);
2869                         s->should_free = 1;
2870                         return 0;
2871                 }
2872
2873                 /*
2874                  * Even if the caller would be happy with getting
2875                  * only the size, we cannot return early at this
2876                  * point if the path requires us to run the content
2877                  * conversion.
2878                  */
2879                 if (size_only && !would_convert_to_git(s->path))
2880                         return 0;
2881
2882                 /*
2883                  * Note: this check uses xsize_t(st.st_size) that may
2884                  * not be the true size of the blob after it goes
2885                  * through convert_to_git().  This may not strictly be
2886                  * correct, but the whole point of big_file_threshold
2887                  * and is_binary check being that we want to avoid
2888                  * opening the file and inspecting the contents, this
2889                  * is probably fine.
2890                  */
2891                 if ((flags & CHECK_BINARY) &&
2892                     s->size > big_file_threshold && s->is_binary == -1) {
2893                         s->is_binary = 1;
2894                         return 0;
2895                 }
2896                 fd = open(s->path, O_RDONLY);
2897                 if (fd < 0)
2898                         goto err_empty;
2899                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2900                 close(fd);
2901                 s->should_munmap = 1;
2902
2903                 /*
2904                  * Convert from working tree format to canonical git format
2905                  */
2906                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2907                         size_t size = 0;
2908                         munmap(s->data, s->size);
2909                         s->should_munmap = 0;
2910                         s->data = strbuf_detach(&buf, &size);
2911                         s->size = size;
2912                         s->should_free = 1;
2913                 }
2914         }
2915         else {
2916                 enum object_type type;
2917                 if (size_only || (flags & CHECK_BINARY)) {
2918                         type = sha1_object_info(s->oid.hash, &s->size);
2919                         if (type < 0)
2920                                 die("unable to read %s",
2921                                     oid_to_hex(&s->oid));
2922                         if (size_only)
2923                                 return 0;
2924                         if (s->size > big_file_threshold && s->is_binary == -1) {
2925                                 s->is_binary = 1;
2926                                 return 0;
2927                         }
2928                 }
2929                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2930                 if (!s->data)
2931                         die("unable to read %s", oid_to_hex(&s->oid));
2932                 s->should_free = 1;
2933         }
2934         return 0;
2935 }
2936
2937 void diff_free_filespec_blob(struct diff_filespec *s)
2938 {
2939         if (s->should_free)
2940                 free(s->data);
2941         else if (s->should_munmap)
2942                 munmap(s->data, s->size);
2943
2944         if (s->should_free || s->should_munmap) {
2945                 s->should_free = s->should_munmap = 0;
2946                 s->data = NULL;
2947         }
2948 }
2949
2950 void diff_free_filespec_data(struct diff_filespec *s)
2951 {
2952         diff_free_filespec_blob(s);
2953         FREE_AND_NULL(s->cnt_data);
2954 }
2955
2956 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2957                            void *blob,
2958                            unsigned long size,
2959                            const struct object_id *oid,
2960                            int mode)
2961 {
2962         int fd;
2963         struct strbuf buf = STRBUF_INIT;
2964         struct strbuf template = STRBUF_INIT;
2965         char *path_dup = xstrdup(path);
2966         const char *base = basename(path_dup);
2967
2968         /* Generate "XXXXXX_basename.ext" */
2969         strbuf_addstr(&template, "XXXXXX_");
2970         strbuf_addstr(&template, base);
2971
2972         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2973         if (fd < 0)
2974                 die_errno("unable to create temp-file");
2975         if (convert_to_working_tree(path,
2976                         (const char *)blob, (size_t)size, &buf)) {
2977                 blob = buf.buf;
2978                 size = buf.len;
2979         }
2980         if (write_in_full(fd, blob, size) != size)
2981                 die_errno("unable to write temp-file");
2982         close_tempfile(&temp->tempfile);
2983         temp->name = get_tempfile_path(&temp->tempfile);
2984         oid_to_hex_r(temp->hex, oid);
2985         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2986         strbuf_release(&buf);
2987         strbuf_release(&template);
2988         free(path_dup);
2989 }
2990
2991 static struct diff_tempfile *prepare_temp_file(const char *name,
2992                 struct diff_filespec *one)
2993 {
2994         struct diff_tempfile *temp = claim_diff_tempfile();
2995
2996         if (!DIFF_FILE_VALID(one)) {
2997         not_a_valid_file:
2998                 /* A '-' entry produces this for file-2, and
2999                  * a '+' entry produces this for file-1.
3000                  */
3001                 temp->name = "/dev/null";
3002                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3003                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3004                 return temp;
3005         }
3006
3007         if (!S_ISGITLINK(one->mode) &&
3008             (!one->oid_valid ||
3009              reuse_worktree_file(name, one->oid.hash, 1))) {
3010                 struct stat st;
3011                 if (lstat(name, &st) < 0) {
3012                         if (errno == ENOENT)
3013                                 goto not_a_valid_file;
3014                         die_errno("stat(%s)", name);
3015                 }
3016                 if (S_ISLNK(st.st_mode)) {
3017                         struct strbuf sb = STRBUF_INIT;
3018                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3019                                 die_errno("readlink(%s)", name);
3020                         prep_temp_blob(name, temp, sb.buf, sb.len,
3021                                        (one->oid_valid ?
3022                                         &one->oid : &null_oid),
3023                                        (one->oid_valid ?
3024                                         one->mode : S_IFLNK));
3025                         strbuf_release(&sb);
3026                 }
3027                 else {
3028                         /* we can borrow from the file in the work tree */
3029                         temp->name = name;
3030                         if (!one->oid_valid)
3031                                 sha1_to_hex_r(temp->hex, null_sha1);
3032                         else
3033                                 oid_to_hex_r(temp->hex, &one->oid);
3034                         /* Even though we may sometimes borrow the
3035                          * contents from the work tree, we always want
3036                          * one->mode.  mode is trustworthy even when
3037                          * !(one->sha1_valid), as long as
3038                          * DIFF_FILE_VALID(one).
3039                          */
3040                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3041                 }
3042                 return temp;
3043         }
3044         else {
3045                 if (diff_populate_filespec(one, 0))
3046                         die("cannot read data blob for %s", one->path);
3047                 prep_temp_blob(name, temp, one->data, one->size,
3048                                &one->oid, one->mode);
3049         }
3050         return temp;
3051 }
3052
3053 static void add_external_diff_name(struct argv_array *argv,
3054                                    const char *name,
3055                                    struct diff_filespec *df)
3056 {
3057         struct diff_tempfile *temp = prepare_temp_file(name, df);
3058         argv_array_push(argv, temp->name);
3059         argv_array_push(argv, temp->hex);
3060         argv_array_push(argv, temp->mode);
3061 }
3062
3063 /* An external diff command takes:
3064  *
3065  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3066  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3067  *
3068  */
3069 static void run_external_diff(const char *pgm,
3070                               const char *name,
3071                               const char *other,
3072                               struct diff_filespec *one,
3073                               struct diff_filespec *two,
3074                               const char *xfrm_msg,
3075                               int complete_rewrite,
3076                               struct diff_options *o)
3077 {
3078         struct argv_array argv = ARGV_ARRAY_INIT;
3079         struct argv_array env = ARGV_ARRAY_INIT;
3080         struct diff_queue_struct *q = &diff_queued_diff;
3081
3082         argv_array_push(&argv, pgm);
3083         argv_array_push(&argv, name);
3084
3085         if (one && two) {
3086                 add_external_diff_name(&argv, name, one);
3087                 if (!other)
3088                         add_external_diff_name(&argv, name, two);
3089                 else {
3090                         add_external_diff_name(&argv, other, two);
3091                         argv_array_push(&argv, other);
3092                         argv_array_push(&argv, xfrm_msg);
3093                 }
3094         }
3095
3096         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3097         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3098
3099         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3100                 die(_("external diff died, stopping at %s"), name);
3101
3102         remove_tempfile();
3103         argv_array_clear(&argv);
3104         argv_array_clear(&env);
3105 }
3106
3107 static int similarity_index(struct diff_filepair *p)
3108 {
3109         return p->score * 100 / MAX_SCORE;
3110 }
3111
3112 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3113 {
3114         if (startup_info->have_repository)
3115                 return find_unique_abbrev(oid->hash, abbrev);
3116         else {
3117                 char *hex = oid_to_hex(oid);
3118                 if (abbrev < 0)
3119                         abbrev = FALLBACK_DEFAULT_ABBREV;
3120                 if (abbrev > GIT_SHA1_HEXSZ)
3121                         die("BUG: oid abbreviation out of range: %d", abbrev);
3122                 if (abbrev)
3123                         hex[abbrev] = '\0';
3124                 return hex;
3125         }
3126 }
3127
3128 static void fill_metainfo(struct strbuf *msg,
3129                           const char *name,
3130                           const char *other,
3131                           struct diff_filespec *one,
3132                           struct diff_filespec *two,
3133                           struct diff_options *o,
3134                           struct diff_filepair *p,
3135                           int *must_show_header,
3136                           int use_color)
3137 {
3138         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3139         const char *reset = diff_get_color(use_color, DIFF_RESET);
3140         const char *line_prefix = diff_line_prefix(o);
3141
3142         *must_show_header = 1;
3143         strbuf_init(msg, PATH_MAX * 2 + 300);
3144         switch (p->status) {
3145         case DIFF_STATUS_COPIED:
3146                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3147                             line_prefix, set, similarity_index(p));
3148                 strbuf_addf(msg, "%s\n%s%scopy from ",
3149                             reset,  line_prefix, set);
3150                 quote_c_style(name, msg, NULL, 0);
3151                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3152                 quote_c_style(other, msg, NULL, 0);
3153                 strbuf_addf(msg, "%s\n", reset);
3154                 break;
3155         case DIFF_STATUS_RENAMED:
3156                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3157                             line_prefix, set, similarity_index(p));
3158                 strbuf_addf(msg, "%s\n%s%srename from ",
3159                             reset, line_prefix, set);
3160                 quote_c_style(name, msg, NULL, 0);
3161                 strbuf_addf(msg, "%s\n%s%srename to ",
3162                             reset, line_prefix, set);
3163                 quote_c_style(other, msg, NULL, 0);
3164                 strbuf_addf(msg, "%s\n", reset);
3165                 break;
3166         case DIFF_STATUS_MODIFIED:
3167                 if (p->score) {
3168                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3169                                     line_prefix,
3170                                     set, similarity_index(p), reset);
3171                         break;
3172                 }
3173                 /* fallthru */
3174         default:
3175                 *must_show_header = 0;
3176         }
3177         if (one && two && oidcmp(&one->oid, &two->oid)) {
3178                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3179
3180                 if (DIFF_OPT_TST(o, BINARY)) {
3181                         mmfile_t mf;
3182                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3183                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3184                                 abbrev = 40;
3185                 }
3186                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3187                             diff_abbrev_oid(&one->oid, abbrev),
3188                             diff_abbrev_oid(&two->oid, abbrev));
3189                 if (one->mode == two->mode)
3190                         strbuf_addf(msg, " %06o", one->mode);
3191                 strbuf_addf(msg, "%s\n", reset);
3192         }
3193 }
3194
3195 static void run_diff_cmd(const char *pgm,
3196                          const char *name,
3197                          const char *other,
3198                          const char *attr_path,
3199                          struct diff_filespec *one,
3200                          struct diff_filespec *two,
3201                          struct strbuf *msg,
3202                          struct diff_options *o,
3203                          struct diff_filepair *p)
3204 {
3205         const char *xfrm_msg = NULL;
3206         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3207         int must_show_header = 0;
3208
3209
3210         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3211                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3212                 if (drv && drv->external)
3213                         pgm = drv->external;
3214         }
3215
3216         if (msg) {
3217                 /*
3218                  * don't use colors when the header is intended for an
3219                  * external diff driver
3220                  */
3221                 fill_metainfo(msg, name, other, one, two, o, p,
3222                               &must_show_header,
3223                               want_color(o->use_color) && !pgm);
3224                 xfrm_msg = msg->len ? msg->buf : NULL;
3225         }
3226
3227         if (pgm) {
3228                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3229                                   complete_rewrite, o);
3230                 return;
3231         }
3232         if (one && two)
3233                 builtin_diff(name, other ? other : name,
3234                              one, two, xfrm_msg, must_show_header,
3235                              o, complete_rewrite);
3236         else
3237                 fprintf(o->file, "* Unmerged path %s\n", name);
3238 }
3239
3240 static void diff_fill_sha1_info(struct diff_filespec *one)
3241 {
3242         if (DIFF_FILE_VALID(one)) {
3243                 if (!one->oid_valid) {
3244                         struct stat st;
3245                         if (one->is_stdin) {
3246                                 oidclr(&one->oid);
3247                                 return;
3248                         }
3249                         if (lstat(one->path, &st) < 0)
3250                                 die_errno("stat '%s'", one->path);
3251                         if (index_path(one->oid.hash, one->path, &st, 0))
3252                                 die("cannot hash %s", one->path);
3253                 }
3254         }
3255         else
3256                 oidclr(&one->oid);
3257 }
3258
3259 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3260 {
3261         /* Strip the prefix but do not molest /dev/null and absolute paths */
3262         if (*namep && **namep != '/') {
3263                 *namep += prefix_length;
3264                 if (**namep == '/')
3265                         ++*namep;
3266         }
3267         if (*otherp && **otherp != '/') {
3268                 *otherp += prefix_length;
3269                 if (**otherp == '/')
3270                         ++*otherp;
3271         }
3272 }
3273
3274 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3275 {
3276         const char *pgm = external_diff();
3277         struct strbuf msg;
3278         struct diff_filespec *one = p->one;
3279         struct diff_filespec *two = p->two;
3280         const char *name;
3281         const char *other;
3282         const char *attr_path;
3283
3284         name  = p->one->path;
3285         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3286         attr_path = name;
3287         if (o->prefix_length)
3288                 strip_prefix(o->prefix_length, &name, &other);
3289
3290         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3291                 pgm = NULL;
3292
3293         if (DIFF_PAIR_UNMERGED(p)) {
3294                 run_diff_cmd(pgm, name, NULL, attr_path,
3295                              NULL, NULL, NULL, o, p);
3296                 return;
3297         }
3298
3299         diff_fill_sha1_info(one);
3300         diff_fill_sha1_info(two);
3301
3302         if (!pgm &&
3303             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3304             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3305                 /*
3306                  * a filepair that changes between file and symlink
3307                  * needs to be split into deletion and creation.
3308                  */
3309                 struct diff_filespec *null = alloc_filespec(two->path);
3310                 run_diff_cmd(NULL, name, other, attr_path,
3311                              one, null, &msg, o, p);
3312                 free(null);
3313                 strbuf_release(&msg);
3314
3315                 null = alloc_filespec(one->path);
3316                 run_diff_cmd(NULL, name, other, attr_path,
3317                              null, two, &msg, o, p);
3318                 free(null);
3319         }
3320         else
3321                 run_diff_cmd(pgm, name, other, attr_path,
3322                              one, two, &msg, o, p);
3323
3324         strbuf_release(&msg);
3325 }
3326
3327 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3328                          struct diffstat_t *diffstat)
3329 {
3330         const char *name;
3331         const char *other;
3332
3333         if (DIFF_PAIR_UNMERGED(p)) {
3334                 /* unmerged */
3335                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3336                 return;
3337         }
3338
3339         name = p->one->path;
3340         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3341
3342         if (o->prefix_length)
3343                 strip_prefix(o->prefix_length, &name, &other);
3344
3345         diff_fill_sha1_info(p->one);
3346         diff_fill_sha1_info(p->two);
3347
3348         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3349 }
3350
3351 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3352 {
3353         const char *name;
3354         const char *other;
3355         const char *attr_path;
3356
3357         if (DIFF_PAIR_UNMERGED(p)) {
3358                 /* unmerged */
3359                 return;
3360         }
3361
3362         name = p->one->path;
3363         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3364         attr_path = other ? other : name;
3365
3366         if (o->prefix_length)
3367                 strip_prefix(o->prefix_length, &name, &other);
3368
3369         diff_fill_sha1_info(p->one);
3370         diff_fill_sha1_info(p->two);
3371
3372         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3373 }
3374
3375 void diff_setup(struct diff_options *options)
3376 {
3377         memcpy(options, &default_diff_options, sizeof(*options));
3378
3379         options->file = stdout;
3380
3381         options->abbrev = DEFAULT_ABBREV;
3382         options->line_termination = '\n';
3383         options->break_opt = -1;
3384         options->rename_limit = -1;
3385         options->dirstat_permille = diff_dirstat_permille_default;
3386         options->context = diff_context_default;
3387         options->interhunkcontext = diff_interhunk_context_default;
3388         options->ws_error_highlight = ws_error_highlight_default;
3389         DIFF_OPT_SET(options, RENAME_EMPTY);
3390
3391         /* pathchange left =NULL by default */
3392         options->change = diff_change;
3393         options->add_remove = diff_addremove;
3394         options->use_color = diff_use_color_default;
3395         options->detect_rename = diff_detect_rename_default;
3396         options->xdl_opts |= diff_algorithm;
3397         if (diff_indent_heuristic)
3398                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3399
3400         options->orderfile = diff_order_file_cfg;
3401
3402         if (diff_no_prefix) {
3403                 options->a_prefix = options->b_prefix = "";
3404         } else if (!diff_mnemonic_prefix) {
3405                 options->a_prefix = "a/";
3406                 options->b_prefix = "b/";
3407         }
3408 }
3409
3410 void diff_setup_done(struct diff_options *options)
3411 {
3412         int count = 0;
3413
3414         if (options->set_default)
3415                 options->set_default(options);
3416
3417         if (options->output_format & DIFF_FORMAT_NAME)
3418                 count++;
3419         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3420                 count++;
3421         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3422                 count++;
3423         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3424                 count++;
3425         if (count > 1)
3426                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
3427
3428         /*
3429          * Most of the time we can say "there are changes"
3430          * only by checking if there are changed paths, but
3431          * --ignore-whitespace* options force us to look
3432          * inside contents.
3433          */
3434
3435         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3436             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3437             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3438                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3439         else
3440                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3441
3442         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3443                 options->detect_rename = DIFF_DETECT_COPY;
3444
3445         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3446                 options->prefix = NULL;
3447         if (options->prefix)
3448                 options->prefix_length = strlen(options->prefix);
3449         else
3450                 options->prefix_length = 0;
3451
3452         if (options->output_format & (DIFF_FORMAT_NAME |
3453                                       DIFF_FORMAT_NAME_STATUS |
3454                                       DIFF_FORMAT_CHECKDIFF |
3455                                       DIFF_FORMAT_NO_OUTPUT))
3456                 options->output_format &= ~(DIFF_FORMAT_RAW |
3457                                             DIFF_FORMAT_NUMSTAT |
3458                                             DIFF_FORMAT_DIFFSTAT |
3459                                             DIFF_FORMAT_SHORTSTAT |
3460                                             DIFF_FORMAT_DIRSTAT |
3461                                             DIFF_FORMAT_SUMMARY |
3462                                             DIFF_FORMAT_PATCH);
3463
3464         /*
3465          * These cases always need recursive; we do not drop caller-supplied
3466          * recursive bits for other formats here.
3467          */
3468         if (options->output_format & (DIFF_FORMAT_PATCH |
3469                                       DIFF_FORMAT_NUMSTAT |
3470                                       DIFF_FORMAT_DIFFSTAT |
3471                                       DIFF_FORMAT_SHORTSTAT |
3472                                       DIFF_FORMAT_DIRSTAT |
3473                                       DIFF_FORMAT_SUMMARY |
3474                                       DIFF_FORMAT_CHECKDIFF))
3475                 DIFF_OPT_SET(options, RECURSIVE);
3476         /*
3477          * Also pickaxe would not work very well if you do not say recursive
3478          */
3479         if (options->pickaxe)
3480                 DIFF_OPT_SET(options, RECURSIVE);
3481         /*
3482          * When patches are generated, submodules diffed against the work tree
3483          * must be checked for dirtiness too so it can be shown in the output
3484          */
3485         if (options->output_format & DIFF_FORMAT_PATCH)
3486                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3487
3488         if (options->detect_rename && options->rename_limit < 0)
3489                 options->rename_limit = diff_rename_limit_default;
3490         if (options->setup & DIFF_SETUP_USE_CACHE) {
3491                 if (!active_cache)
3492                         /* read-cache does not die even when it fails
3493                          * so it is safe for us to do this here.  Also
3494                          * it does not smudge active_cache or active_nr
3495                          * when it fails, so we do not have to worry about
3496                          * cleaning it up ourselves either.
3497                          */
3498                         read_cache();
3499         }
3500         if (40 < options->abbrev)
3501                 options->abbrev = 40; /* full */
3502
3503         /*
3504          * It does not make sense to show the first hit we happened
3505          * to have found.  It does not make sense not to return with
3506          * exit code in such a case either.
3507          */
3508         if (DIFF_OPT_TST(options, QUICK)) {
3509                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3510                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3511         }
3512
3513         options->diff_path_counter = 0;
3514
3515         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3516                 die(_("--follow requires exactly one pathspec"));
3517 }
3518
3519 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3520 {
3521         char c, *eq;
3522         int len;
3523
3524         if (*arg != '-')
3525                 return 0;
3526         c = *++arg;
3527         if (!c)
3528                 return 0;
3529         if (c == arg_short) {
3530                 c = *++arg;
3531                 if (!c)
3532                         return 1;
3533                 if (val && isdigit(c)) {
3534                         char *end;
3535                         int n = strtoul(arg, &end, 10);
3536                         if (*end)
3537                                 return 0;
3538                         *val = n;
3539                         return 1;
3540                 }
3541                 return 0;
3542         }
3543         if (c != '-')
3544                 return 0;
3545         arg++;
3546         eq = strchrnul(arg, '=');
3547         len = eq - arg;
3548         if (!len || strncmp(arg, arg_long, len))
3549                 return 0;
3550         if (*eq) {
3551                 int n;
3552                 char *end;
3553                 if (!isdigit(*++eq))
3554                         return 0;
3555                 n = strtoul(eq, &end, 10);
3556                 if (*end)
3557                         return 0;
3558                 *val = n;
3559         }
3560         return 1;
3561 }
3562
3563 static int diff_scoreopt_parse(const char *opt);
3564
3565 static inline int short_opt(char opt, const char **argv,
3566                             const char **optarg)
3567 {
3568         const char *arg = argv[0];
3569         if (arg[0] != '-' || arg[1] != opt)
3570                 return 0;
3571         if (arg[2] != '\0') {
3572                 *optarg = arg + 2;
3573                 return 1;
3574         }
3575         if (!argv[1])
3576                 die("Option '%c' requires a value", opt);
3577         *optarg = argv[1];
3578         return 2;
3579 }
3580
3581 int parse_long_opt(const char *opt, const char **argv,
3582                    const char **optarg)
3583 {
3584         const char *arg = argv[0];
3585         if (!skip_prefix(arg, "--", &arg))
3586                 return 0;
3587         if (!skip_prefix(arg, opt, &arg))
3588                 return 0;
3589         if (*arg == '=') { /* stuck form: --option=value */
3590                 *optarg = arg + 1;
3591                 return 1;
3592         }
3593         if (*arg != '\0')
3594                 return 0;
3595         /* separate form: --option value */
3596         if (!argv[1])
3597                 die("Option '--%s' requires a value", opt);
3598         *optarg = argv[1];
3599         return 2;
3600 }
3601
3602 static int stat_opt(struct diff_options *options, const char **av)
3603 {
3604         const char *arg = av[0];
3605         char *end;
3606         int width = options->stat_width;
3607         int name_width = options->stat_name_width;
3608         int graph_width = options->stat_graph_width;
3609         int count = options->stat_count;
3610         int argcount = 1;
3611
3612         if (!skip_prefix(arg, "--stat", &arg))
3613                 die("BUG: stat option does not begin with --stat: %s", arg);
3614         end = (char *)arg;
3615
3616         switch (*arg) {
3617         case '-':
3618                 if (skip_prefix(arg, "-width", &arg)) {
3619                         if (*arg == '=')
3620                                 width = strtoul(arg + 1, &end, 10);
3621                         else if (!*arg && !av[1])
3622                                 die_want_option("--stat-width");
3623                         else if (!*arg) {
3624                                 width = strtoul(av[1], &end, 10);
3625                                 argcount = 2;
3626                         }
3627                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3628                         if (*arg == '=')
3629                                 name_width = strtoul(arg + 1, &end, 10);
3630                         else if (!*arg && !av[1])
3631                                 die_want_option("--stat-name-width");
3632                         else if (!*arg) {
3633                                 name_width = strtoul(av[1], &end, 10);
3634                                 argcount = 2;
3635                         }
3636                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3637                         if (*arg == '=')
3638                                 graph_width = strtoul(arg + 1, &end, 10);
3639                         else if (!*arg && !av[1])
3640                                 die_want_option("--stat-graph-width");
3641                         else if (!*arg) {
3642                                 graph_width = strtoul(av[1], &end, 10);
3643                                 argcount = 2;
3644                         }
3645                 } else if (skip_prefix(arg, "-count", &arg)) {
3646                         if (*arg == '=')
3647                                 count = strtoul(arg + 1, &end, 10);
3648                         else if (!*arg && !av[1])
3649                                 die_want_option("--stat-count");
3650                         else if (!*arg) {
3651                                 count = strtoul(av[1], &end, 10);
3652                                 argcount = 2;
3653                         }
3654                 }
3655                 break;
3656         case '=':
3657                 width = strtoul(arg+1, &end, 10);
3658                 if (*end == ',')
3659                         name_width = strtoul(end+1, &end, 10);
3660                 if (*end == ',')
3661                         count = strtoul(end+1, &end, 10);
3662         }
3663
3664         /* Important! This checks all the error cases! */
3665         if (*end)
3666                 return 0;
3667         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3668         options->stat_name_width = name_width;
3669         options->stat_graph_width = graph_width;
3670         options->stat_width = width;
3671         options->stat_count = count;
3672         return argcount;
3673 }
3674
3675 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3676 {
3677         struct strbuf errmsg = STRBUF_INIT;
3678         if (parse_dirstat_params(options, params, &errmsg))
3679                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3680                     errmsg.buf);
3681         strbuf_release(&errmsg);
3682         /*
3683          * The caller knows a dirstat-related option is given from the command
3684          * line; allow it to say "return this_function();"
3685          */
3686         options->output_format |= DIFF_FORMAT_DIRSTAT;
3687         return 1;
3688 }
3689
3690 static int parse_submodule_opt(struct diff_options *options, const char *value)
3691 {
3692         if (parse_submodule_params(options, value))
3693                 die(_("Failed to parse --submodule option parameter: '%s'"),
3694                         value);
3695         return 1;
3696 }
3697
3698 static const char diff_status_letters[] = {
3699         DIFF_STATUS_ADDED,
3700         DIFF_STATUS_COPIED,
3701         DIFF_STATUS_DELETED,
3702         DIFF_STATUS_MODIFIED,
3703         DIFF_STATUS_RENAMED,
3704         DIFF_STATUS_TYPE_CHANGED,
3705         DIFF_STATUS_UNKNOWN,
3706         DIFF_STATUS_UNMERGED,
3707         DIFF_STATUS_FILTER_AON,
3708         DIFF_STATUS_FILTER_BROKEN,
3709         '\0',
3710 };
3711
3712 static unsigned int filter_bit['Z' + 1];
3713
3714 static void prepare_filter_bits(void)
3715 {
3716         int i;
3717
3718         if (!filter_bit[DIFF_STATUS_ADDED]) {
3719                 for (i = 0; diff_status_letters[i]; i++)
3720                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3721         }
3722 }
3723
3724 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3725 {
3726         return opt->filter & filter_bit[(int) status];
3727 }
3728
3729 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3730 {
3731         int i, optch;
3732
3733         prepare_filter_bits();
3734
3735         /*
3736          * If there is a negation e.g. 'd' in the input, and we haven't
3737          * initialized the filter field with another --diff-filter, start
3738          * from full set of bits, except for AON.
3739          */
3740         if (!opt->filter) {
3741                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3742                         if (optch < 'a' || 'z' < optch)
3743                                 continue;
3744                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3745                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3746                         break;
3747                 }
3748         }
3749
3750         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3751                 unsigned int bit;
3752                 int negate;
3753
3754                 if ('a' <= optch && optch <= 'z') {
3755                         negate = 1;
3756                         optch = toupper(optch);
3757                 } else {
3758                         negate = 0;
3759                 }
3760
3761                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3762                 if (!bit)
3763                         return optarg[i];
3764                 if (negate)
3765                         opt->filter &= ~bit;
3766                 else
3767                         opt->filter |= bit;
3768         }
3769         return 0;
3770 }
3771
3772 static void enable_patch_output(int *fmt) {
3773         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3774         *fmt |= DIFF_FORMAT_PATCH;
3775 }
3776
3777 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
3778 {
3779         int val = parse_ws_error_highlight(arg);
3780
3781         if (val < 0) {
3782                 error("unknown value after ws-error-highlight=%.*s",
3783                       -1 - val, arg);
3784                 return 0;
3785         }
3786         opt->ws_error_highlight = val;
3787         return 1;
3788 }
3789
3790 int diff_opt_parse(struct diff_options *options,
3791                    const char **av, int ac, const char *prefix)
3792 {
3793         const char *arg = av[0];
3794         const char *optarg;
3795         int argcount;
3796
3797         if (!prefix)
3798                 prefix = "";
3799
3800         /* Output format options */
3801         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3802             || opt_arg(arg, 'U', "unified", &options->context))
3803                 enable_patch_output(&options->output_format);
3804         else if (!strcmp(arg, "--raw"))
3805                 options->output_format |= DIFF_FORMAT_RAW;
3806         else if (!strcmp(arg, "--patch-with-raw")) {
3807                 enable_patch_output(&options->output_format);
3808                 options->output_format |= DIFF_FORMAT_RAW;
3809         } else if (!strcmp(arg, "--numstat"))
3810                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3811         else if (!strcmp(arg, "--shortstat"))
3812                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3813         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3814                 return parse_dirstat_opt(options, "");
3815         else if (skip_prefix(arg, "-X", &arg))
3816                 return parse_dirstat_opt(options, arg);
3817         else if (skip_prefix(arg, "--dirstat=", &arg))
3818                 return parse_dirstat_opt(options, arg);
3819         else if (!strcmp(arg, "--cumulative"))
3820                 return parse_dirstat_opt(options, "cumulative");
3821         else if (!strcmp(arg, "--dirstat-by-file"))
3822                 return parse_dirstat_opt(options, "files");
3823         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3824                 parse_dirstat_opt(options, "files");
3825                 return parse_dirstat_opt(options, arg);
3826         }
3827         else if (!strcmp(arg, "--check"))
3828                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3829         else if (!strcmp(arg, "--summary"))
3830                 options->output_format |= DIFF_FORMAT_SUMMARY;
3831         else if (!strcmp(arg, "--patch-with-stat")) {
3832                 enable_patch_output(&options->output_format);
3833                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3834         } else if (!strcmp(arg, "--name-only"))
3835                 options->output_format |= DIFF_FORMAT_NAME;
3836         else if (!strcmp(arg, "--name-status"))
3837                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3838         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3839                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3840         else if (starts_with(arg, "--stat"))
3841                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3842                 return stat_opt(options, av);
3843
3844         /* renames options */
3845         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3846                  !strcmp(arg, "--break-rewrites")) {
3847                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3848                         return error("invalid argument to -B: %s", arg+2);
3849         }
3850         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3851                  !strcmp(arg, "--find-renames")) {
3852                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3853                         return error("invalid argument to -M: %s", arg+2);
3854                 options->detect_rename = DIFF_DETECT_RENAME;
3855         }
3856         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3857                 options->irreversible_delete = 1;
3858         }
3859         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3860                  !strcmp(arg, "--find-copies")) {
3861                 if (options->detect_rename == DIFF_DETECT_COPY)
3862                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3863                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3864                         return error("invalid argument to -C: %s", arg+2);
3865                 options->detect_rename = DIFF_DETECT_COPY;
3866         }
3867         else if (!strcmp(arg, "--no-renames"))
3868                 options->detect_rename = 0;
3869         else if (!strcmp(arg, "--rename-empty"))
3870                 DIFF_OPT_SET(options, RENAME_EMPTY);
3871         else if (!strcmp(arg, "--no-rename-empty"))
3872                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3873         else if (!strcmp(arg, "--relative"))
3874                 DIFF_OPT_SET(options, RELATIVE_NAME);
3875         else if (skip_prefix(arg, "--relative=", &arg)) {
3876                 DIFF_OPT_SET(options, RELATIVE_NAME);
3877                 options->prefix = arg;
3878         }
3879
3880         /* xdiff options */
3881         else if (!strcmp(arg, "--minimal"))
3882                 DIFF_XDL_SET(options, NEED_MINIMAL);
3883         else if (!strcmp(arg, "--no-minimal"))
3884                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3885         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3886                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3887         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3888                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3889         else if (!strcmp(arg, "--ignore-space-at-eol"))
3890                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3891         else if (!strcmp(arg, "--ignore-blank-lines"))
3892                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3893         else if (!strcmp(arg, "--indent-heuristic"))
3894                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3895         else if (!strcmp(arg, "--no-indent-heuristic"))
3896                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3897         else if (!strcmp(arg, "--patience"))
3898                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3899         else if (!strcmp(arg, "--histogram"))
3900                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3901         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3902                 long value = parse_algorithm_value(optarg);
3903                 if (value < 0)
3904                         return error("option diff-algorithm accepts \"myers\", "
3905                                      "\"minimal\", \"patience\" and \"histogram\"");
3906                 /* clear out previous settings */
3907                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3908                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3909                 options->xdl_opts |= value;
3910                 return argcount;
3911         }
3912
3913         /* flags options */
3914         else if (!strcmp(arg, "--binary")) {
3915                 enable_patch_output(&options->output_format);
3916                 DIFF_OPT_SET(options, BINARY);
3917         }
3918         else if (!strcmp(arg, "--full-index"))
3919                 DIFF_OPT_SET(options, FULL_INDEX);
3920         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3921                 DIFF_OPT_SET(options, TEXT);
3922         else if (!strcmp(arg, "-R"))
3923                 DIFF_OPT_SET(options, REVERSE_DIFF);
3924         else if (!strcmp(arg, "--find-copies-harder"))
3925                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3926         else if (!strcmp(arg, "--follow"))
3927                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3928         else if (!strcmp(arg, "--no-follow")) {
3929                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3930                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3931         } else if (!strcmp(arg, "--color"))
3932                 options->use_color = 1;
3933         else if (skip_prefix(arg, "--color=", &arg)) {
3934                 int value = git_config_colorbool(NULL, arg);
3935                 if (value < 0)
3936                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3937                 options->use_color = value;
3938         }
3939         else if (!strcmp(arg, "--no-color"))
3940                 options->use_color = 0;
3941         else if (!strcmp(arg, "--color-words")) {
3942                 options->use_color = 1;
3943                 options->word_diff = DIFF_WORDS_COLOR;
3944         }
3945         else if (skip_prefix(arg, "--color-words=", &arg)) {
3946                 options->use_color = 1;
3947                 options->word_diff = DIFF_WORDS_COLOR;
3948                 options->word_regex = arg;
3949         }
3950         else if (!strcmp(arg, "--word-diff")) {
3951                 if (options->word_diff == DIFF_WORDS_NONE)
3952                         options->word_diff = DIFF_WORDS_PLAIN;
3953         }
3954         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3955                 if (!strcmp(arg, "plain"))
3956                         options->word_diff = DIFF_WORDS_PLAIN;
3957                 else if (!strcmp(arg, "color")) {
3958                         options->use_color = 1;
3959                         options->word_diff = DIFF_WORDS_COLOR;
3960                 }
3961                 else if (!strcmp(arg, "porcelain"))
3962                         options->word_diff = DIFF_WORDS_PORCELAIN;
3963                 else if (!strcmp(arg, "none"))
3964                         options->word_diff = DIFF_WORDS_NONE;
3965                 else
3966                         die("bad --word-diff argument: %s", arg);
3967         }
3968         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3969                 if (options->word_diff == DIFF_WORDS_NONE)
3970                         options->word_diff = DIFF_WORDS_PLAIN;
3971                 options->word_regex = optarg;
3972                 return argcount;
3973         }
3974         else if (!strcmp(arg, "--exit-code"))
3975                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3976         else if (!strcmp(arg, "--quiet"))
3977                 DIFF_OPT_SET(options, QUICK);
3978         else if (!strcmp(arg, "--ext-diff"))
3979                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3980         else if (!strcmp(arg, "--no-ext-diff"))
3981                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3982         else if (!strcmp(arg, "--textconv"))
3983                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3984         else if (!strcmp(arg, "--no-textconv"))
3985                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3986         else if (!strcmp(arg, "--ignore-submodules")) {
3987                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3988                 handle_ignore_submodules_arg(options, "all");
3989         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3990                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3991                 handle_ignore_submodules_arg(options, arg);
3992         } else if (!strcmp(arg, "--submodule"))
3993                 options->submodule_format = DIFF_SUBMODULE_LOG;
3994         else if (skip_prefix(arg, "--submodule=", &arg))
3995                 return parse_submodule_opt(options, arg);
3996         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3997                 return parse_ws_error_highlight_opt(options, arg);
3998         else if (!strcmp(arg, "--ita-invisible-in-index"))
3999                 options->ita_invisible_in_index = 1;
4000         else if (!strcmp(arg, "--ita-visible-in-index"))
4001                 options->ita_invisible_in_index = 0;
4002
4003         /* misc options */
4004         else if (!strcmp(arg, "-z"))
4005                 options->line_termination = 0;
4006         else if ((argcount = short_opt('l', av, &optarg))) {
4007                 options->rename_limit = strtoul(optarg, NULL, 10);
4008                 return argcount;
4009         }
4010         else if ((argcount = short_opt('S', av, &optarg))) {
4011                 options->pickaxe = optarg;
4012                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4013                 return argcount;
4014         } else if ((argcount = short_opt('G', av, &optarg))) {
4015                 options->pickaxe = optarg;
4016                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4017                 return argcount;
4018         }
4019         else if (!strcmp(arg, "--pickaxe-all"))
4020                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4021         else if (!strcmp(arg, "--pickaxe-regex"))
4022                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4023         else if ((argcount = short_opt('O', av, &optarg))) {
4024                 options->orderfile = prefix_filename(prefix, optarg);
4025                 return argcount;
4026         }
4027         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4028                 int offending = parse_diff_filter_opt(optarg, options);
4029                 if (offending)
4030                         die("unknown change class '%c' in --diff-filter=%s",
4031                             offending, optarg);
4032                 return argcount;
4033         }
4034         else if (!strcmp(arg, "--no-abbrev"))
4035                 options->abbrev = 0;
4036         else if (!strcmp(arg, "--abbrev"))
4037                 options->abbrev = DEFAULT_ABBREV;
4038         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4039                 options->abbrev = strtoul(arg, NULL, 10);
4040                 if (options->abbrev < MINIMUM_ABBREV)
4041                         options->abbrev = MINIMUM_ABBREV;
4042                 else if (40 < options->abbrev)
4043                         options->abbrev = 40;
4044         }
4045         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4046                 options->a_prefix = optarg;
4047                 return argcount;
4048         }
4049         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4050                 options->line_prefix = optarg;
4051                 options->line_prefix_length = strlen(options->line_prefix);
4052                 graph_setup_line_prefix(options);
4053                 return argcount;
4054         }
4055         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4056                 options->b_prefix = optarg;
4057                 return argcount;
4058         }
4059         else if (!strcmp(arg, "--no-prefix"))
4060                 options->a_prefix = options->b_prefix = "";
4061         else if (opt_arg(arg, '\0', "inter-hunk-context",
4062                          &options->interhunkcontext))
4063                 ;
4064         else if (!strcmp(arg, "-W"))
4065                 DIFF_OPT_SET(options, FUNCCONTEXT);
4066         else if (!strcmp(arg, "--function-context"))
4067                 DIFF_OPT_SET(options, FUNCCONTEXT);
4068         else if (!strcmp(arg, "--no-function-context"))
4069                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4070         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4071                 char *path = prefix_filename(prefix, optarg);
4072                 options->file = xfopen(path, "w");
4073                 options->close_file = 1;
4074                 if (options->use_color != GIT_COLOR_ALWAYS)
4075                         options->use_color = GIT_COLOR_NEVER;
4076                 free(path);
4077                 return argcount;
4078         } else
4079                 return 0;
4080         return 1;
4081 }
4082
4083 int parse_rename_score(const char **cp_p)
4084 {
4085         unsigned long num, scale;
4086         int ch, dot;
4087         const char *cp = *cp_p;
4088
4089         num = 0;
4090         scale = 1;
4091         dot = 0;
4092         for (;;) {
4093                 ch = *cp;
4094                 if ( !dot && ch == '.' ) {
4095                         scale = 1;
4096                         dot = 1;
4097                 } else if ( ch == '%' ) {
4098                         scale = dot ? scale*100 : 100;
4099                         cp++;   /* % is always at the end */
4100                         break;
4101                 } else if ( ch >= '0' && ch <= '9' ) {
4102                         if ( scale < 100000 ) {
4103                                 scale *= 10;
4104                                 num = (num*10) + (ch-'0');
4105                         }
4106                 } else {
4107                         break;
4108                 }
4109                 cp++;
4110         }
4111         *cp_p = cp;
4112
4113         /* user says num divided by scale and we say internally that
4114          * is MAX_SCORE * num / scale.
4115          */
4116         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4117 }
4118
4119 static int diff_scoreopt_parse(const char *opt)
4120 {
4121         int opt1, opt2, cmd;
4122
4123         if (*opt++ != '-')
4124                 return -1;
4125         cmd = *opt++;
4126         if (cmd == '-') {
4127                 /* convert the long-form arguments into short-form versions */
4128                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4129                         if (*opt == 0 || *opt++ == '=')
4130                                 cmd = 'B';
4131                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4132                         if (*opt == 0 || *opt++ == '=')
4133                                 cmd = 'C';
4134                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4135                         if (*opt == 0 || *opt++ == '=')
4136                                 cmd = 'M';
4137                 }
4138         }
4139         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4140                 return -1; /* that is not a -M, -C, or -B option */
4141
4142         opt1 = parse_rename_score(&opt);
4143         if (cmd != 'B')
4144                 opt2 = 0;
4145         else {
4146                 if (*opt == 0)
4147                         opt2 = 0;
4148                 else if (*opt != '/')
4149                         return -1; /* we expect -B80/99 or -B80 */
4150                 else {
4151                         opt++;
4152                         opt2 = parse_rename_score(&opt);
4153                 }
4154         }
4155         if (*opt != 0)
4156                 return -1;
4157         return opt1 | (opt2 << 16);
4158 }
4159
4160 struct diff_queue_struct diff_queued_diff;
4161
4162 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4163 {
4164         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4165         queue->queue[queue->nr++] = dp;
4166 }
4167
4168 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4169                                  struct diff_filespec *one,
4170                                  struct diff_filespec *two)
4171 {
4172         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4173         dp->one = one;
4174         dp->two = two;
4175         if (queue)
4176                 diff_q(queue, dp);
4177         return dp;
4178 }
4179
4180 void diff_free_filepair(struct diff_filepair *p)
4181 {
4182         free_filespec(p->one);
4183         free_filespec(p->two);
4184         free(p);
4185 }
4186
4187 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4188 {
4189         int abblen;
4190         const char *abbrev;
4191
4192         if (len == GIT_SHA1_HEXSZ)
4193                 return oid_to_hex(oid);
4194
4195         abbrev = diff_abbrev_oid(oid, len);
4196         abblen = strlen(abbrev);
4197
4198         /*
4199          * In well-behaved cases, where the abbbreviated result is the
4200          * same as the requested length, append three dots after the
4201          * abbreviation (hence the whole logic is limited to the case
4202          * where abblen < 37); when the actual abbreviated result is a
4203          * bit longer than the requested length, we reduce the number
4204          * of dots so that they match the well-behaved ones.  However,
4205          * if the actual abbreviation is longer than the requested
4206          * length by more than three, we give up on aligning, and add
4207          * three dots anyway, to indicate that the output is not the
4208          * full object name.  Yes, this may be suboptimal, but this
4209          * appears only in "diff --raw --abbrev" output and it is not
4210          * worth the effort to change it now.  Note that this would
4211          * likely to work fine when the automatic sizing of default
4212          * abbreviation length is used--we would be fed -1 in "len" in
4213          * that case, and will end up always appending three-dots, but
4214          * the automatic sizing is supposed to give abblen that ensures
4215          * uniqueness across all objects (statistically speaking).
4216          */
4217         if (abblen < GIT_SHA1_HEXSZ - 3) {
4218                 static char hex[GIT_MAX_HEXSZ + 1];
4219                 if (len < abblen && abblen <= len + 2)
4220                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4221                 else
4222                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4223                 return hex;
4224         }
4225
4226         return oid_to_hex(oid);
4227 }
4228
4229 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4230 {
4231         int line_termination = opt->line_termination;
4232         int inter_name_termination = line_termination ? '\t' : '\0';
4233
4234         fprintf(opt->file, "%s", diff_line_prefix(opt));
4235         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4236                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4237                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4238                 fprintf(opt->file, "%s ",
4239                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4240         }
4241         if (p->score) {
4242                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4243                         inter_name_termination);
4244         } else {
4245                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4246         }
4247
4248         if (p->status == DIFF_STATUS_COPIED ||
4249             p->status == DIFF_STATUS_RENAMED) {
4250                 const char *name_a, *name_b;
4251                 name_a = p->one->path;
4252                 name_b = p->two->path;
4253                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4254                 write_name_quoted(name_a, opt->file, inter_name_termination);
4255                 write_name_quoted(name_b, opt->file, line_termination);
4256         } else {
4257                 const char *name_a, *name_b;
4258                 name_a = p->one->mode ? p->one->path : p->two->path;
4259                 name_b = NULL;
4260                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4261                 write_name_quoted(name_a, opt->file, line_termination);
4262         }
4263 }
4264
4265 int diff_unmodified_pair(struct diff_filepair *p)
4266 {
4267         /* This function is written stricter than necessary to support
4268          * the currently implemented transformers, but the idea is to
4269          * let transformers to produce diff_filepairs any way they want,
4270          * and filter and clean them up here before producing the output.
4271          */
4272         struct diff_filespec *one = p->one, *two = p->two;
4273
4274         if (DIFF_PAIR_UNMERGED(p))
4275                 return 0; /* unmerged is interesting */
4276
4277         /* deletion, addition, mode or type change
4278          * and rename are all interesting.
4279          */
4280         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4281             DIFF_PAIR_MODE_CHANGED(p) ||
4282             strcmp(one->path, two->path))
4283                 return 0;
4284
4285         /* both are valid and point at the same path.  that is, we are
4286          * dealing with a change.
4287          */
4288         if (one->oid_valid && two->oid_valid &&
4289             !oidcmp(&one->oid, &two->oid) &&
4290             !one->dirty_submodule && !two->dirty_submodule)
4291                 return 1; /* no change */
4292         if (!one->oid_valid && !two->oid_valid)
4293                 return 1; /* both look at the same file on the filesystem. */
4294         return 0;
4295 }
4296
4297 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4298 {
4299         if (diff_unmodified_pair(p))
4300                 return;
4301
4302         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4303             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4304                 return; /* no tree diffs in patch format */
4305
4306         run_diff(p, o);
4307 }
4308
4309 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4310                             struct diffstat_t *diffstat)
4311 {
4312         if (diff_unmodified_pair(p))
4313                 return;
4314
4315         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4316             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4317                 return; /* no useful stat for tree diffs */
4318
4319         run_diffstat(p, o, diffstat);
4320 }
4321
4322 static void diff_flush_checkdiff(struct diff_filepair *p,
4323                 struct diff_options *o)
4324 {
4325         if (diff_unmodified_pair(p))
4326                 return;
4327
4328         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4329             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4330                 return; /* nothing to check in tree diffs */
4331
4332         run_checkdiff(p, o);
4333 }
4334
4335 int diff_queue_is_empty(void)
4336 {
4337         struct diff_queue_struct *q = &diff_queued_diff;
4338         int i;
4339         for (i = 0; i < q->nr; i++)
4340                 if (!diff_unmodified_pair(q->queue[i]))
4341                         return 0;
4342         return 1;
4343 }
4344
4345 #if DIFF_DEBUG
4346 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4347 {
4348         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4349                 x, one ? one : "",
4350                 s->path,
4351                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4352                 s->mode,
4353                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4354         fprintf(stderr, "queue[%d] %s size %lu\n",
4355                 x, one ? one : "",
4356                 s->size);
4357 }
4358
4359 void diff_debug_filepair(const struct diff_filepair *p, int i)
4360 {
4361         diff_debug_filespec(p->one, i, "one");
4362         diff_debug_filespec(p->two, i, "two");
4363         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4364                 p->score, p->status ? p->status : '?',
4365                 p->one->rename_used, p->broken_pair);
4366 }
4367
4368 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4369 {
4370         int i;
4371         if (msg)
4372                 fprintf(stderr, "%s\n", msg);
4373         fprintf(stderr, "q->nr = %d\n", q->nr);
4374         for (i = 0; i < q->nr; i++) {
4375                 struct diff_filepair *p = q->queue[i];
4376                 diff_debug_filepair(p, i);
4377         }
4378 }
4379 #endif
4380
4381 static void diff_resolve_rename_copy(void)
4382 {
4383         int i;
4384         struct diff_filepair *p;
4385         struct diff_queue_struct *q = &diff_queued_diff;
4386
4387         diff_debug_queue("resolve-rename-copy", q);
4388
4389         for (i = 0; i < q->nr; i++) {
4390                 p = q->queue[i];
4391                 p->status = 0; /* undecided */
4392                 if (DIFF_PAIR_UNMERGED(p))
4393                         p->status = DIFF_STATUS_UNMERGED;
4394                 else if (!DIFF_FILE_VALID(p->one))
4395                         p->status = DIFF_STATUS_ADDED;
4396                 else if (!DIFF_FILE_VALID(p->two))
4397                         p->status = DIFF_STATUS_DELETED;
4398                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4399                         p->status = DIFF_STATUS_TYPE_CHANGED;
4400
4401                 /* from this point on, we are dealing with a pair
4402                  * whose both sides are valid and of the same type, i.e.
4403                  * either in-place edit or rename/copy edit.
4404                  */
4405                 else if (DIFF_PAIR_RENAME(p)) {
4406                         /*
4407                          * A rename might have re-connected a broken
4408                          * pair up, causing the pathnames to be the
4409                          * same again. If so, that's not a rename at
4410                          * all, just a modification..
4411                          *
4412                          * Otherwise, see if this source was used for
4413                          * multiple renames, in which case we decrement
4414                          * the count, and call it a copy.
4415                          */
4416                         if (!strcmp(p->one->path, p->two->path))
4417                                 p->status = DIFF_STATUS_MODIFIED;
4418                         else if (--p->one->rename_used > 0)
4419                                 p->status = DIFF_STATUS_COPIED;
4420                         else
4421                                 p->status = DIFF_STATUS_RENAMED;
4422                 }
4423                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4424                          p->one->mode != p->two->mode ||
4425                          p->one->dirty_submodule ||
4426                          p->two->dirty_submodule ||
4427                          is_null_oid(&p->one->oid))
4428                         p->status = DIFF_STATUS_MODIFIED;
4429                 else {
4430                         /* This is a "no-change" entry and should not
4431                          * happen anymore, but prepare for broken callers.
4432                          */
4433                         error("feeding unmodified %s to diffcore",
4434                               p->one->path);
4435                         p->status = DIFF_STATUS_UNKNOWN;
4436                 }
4437         }
4438         diff_debug_queue("resolve-rename-copy done", q);
4439 }
4440
4441 static int check_pair_status(struct diff_filepair *p)
4442 {
4443         switch (p->status) {
4444         case DIFF_STATUS_UNKNOWN:
4445                 return 0;
4446         case 0:
4447                 die("internal error in diff-resolve-rename-copy");
4448         default:
4449                 return 1;
4450         }
4451 }
4452
4453 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4454 {
4455         int fmt = opt->output_format;
4456
4457         if (fmt & DIFF_FORMAT_CHECKDIFF)
4458                 diff_flush_checkdiff(p, opt);
4459         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4460                 diff_flush_raw(p, opt);
4461         else if (fmt & DIFF_FORMAT_NAME) {
4462                 const char *name_a, *name_b;
4463                 name_a = p->two->path;
4464                 name_b = NULL;
4465                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4466                 fprintf(opt->file, "%s", diff_line_prefix(opt));
4467                 write_name_quoted(name_a, opt->file, opt->line_termination);
4468         }
4469 }
4470
4471 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4472 {
4473         if (fs->mode)
4474                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4475         else
4476                 fprintf(file, " %s ", newdelete);
4477         write_name_quoted(fs->path, file, '\n');
4478 }
4479
4480
4481 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4482                 const char *line_prefix)
4483 {
4484         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4485                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4486                         p->two->mode, show_name ? ' ' : '\n');
4487                 if (show_name) {
4488                         write_name_quoted(p->two->path, file, '\n');
4489                 }
4490         }
4491 }
4492
4493 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4494                         const char *line_prefix)
4495 {
4496         char *names = pprint_rename(p->one->path, p->two->path);
4497
4498         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4499         free(names);
4500         show_mode_change(file, p, 0, line_prefix);
4501 }
4502
4503 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4504 {
4505         FILE *file = opt->file;
4506         const char *line_prefix = diff_line_prefix(opt);
4507
4508         switch(p->status) {
4509         case DIFF_STATUS_DELETED:
4510                 fputs(line_prefix, file);
4511                 show_file_mode_name(file, "delete", p->one);
4512                 break;
4513         case DIFF_STATUS_ADDED:
4514                 fputs(line_prefix, file);
4515                 show_file_mode_name(file, "create", p->two);
4516                 break;
4517         case DIFF_STATUS_COPIED:
4518                 fputs(line_prefix, file);
4519                 show_rename_copy(file, "copy", p, line_prefix);
4520                 break;
4521         case DIFF_STATUS_RENAMED:
4522                 fputs(line_prefix, file);
4523                 show_rename_copy(file, "rename", p, line_prefix);
4524                 break;
4525         default:
4526                 if (p->score) {
4527                         fprintf(file, "%s rewrite ", line_prefix);
4528                         write_name_quoted(p->two->path, file, ' ');
4529                         fprintf(file, "(%d%%)\n", similarity_index(p));
4530                 }
4531                 show_mode_change(file, p, !p->score, line_prefix);
4532                 break;
4533         }
4534 }
4535
4536 struct patch_id_t {
4537         git_SHA_CTX *ctx;
4538         int patchlen;
4539 };
4540
4541 static int remove_space(char *line, int len)
4542 {
4543         int i;
4544         char *dst = line;
4545         unsigned char c;
4546
4547         for (i = 0; i < len; i++)
4548                 if (!isspace((c = line[i])))
4549                         *dst++ = c;
4550
4551         return dst - line;
4552 }
4553
4554 static void patch_id_consume(void *priv, char *line, unsigned long len)
4555 {
4556         struct patch_id_t *data = priv;
4557         int new_len;
4558
4559         /* Ignore line numbers when computing the SHA1 of the patch */
4560         if (starts_with(line, "@@ -"))
4561                 return;
4562
4563         new_len = remove_space(line, len);
4564
4565         git_SHA1_Update(data->ctx, line, new_len);
4566         data->patchlen += new_len;
4567 }
4568
4569 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
4570 {
4571         git_SHA1_Update(ctx, str, strlen(str));
4572 }
4573
4574 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
4575 {
4576         /* large enough for 2^32 in octal */
4577         char buf[12];
4578         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
4579         git_SHA1_Update(ctx, buf, len);
4580 }
4581
4582 /* returns 0 upon success, and writes result into sha1 */
4583 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4584 {
4585         struct diff_queue_struct *q = &diff_queued_diff;
4586         int i;
4587         git_SHA_CTX ctx;
4588         struct patch_id_t data;
4589
4590         git_SHA1_Init(&ctx);
4591         memset(&data, 0, sizeof(struct patch_id_t));
4592         data.ctx = &ctx;
4593
4594         for (i = 0; i < q->nr; i++) {
4595                 xpparam_t xpp;
4596                 xdemitconf_t xecfg;
4597                 mmfile_t mf1, mf2;
4598                 struct diff_filepair *p = q->queue[i];
4599                 int len1, len2;
4600
4601                 memset(&xpp, 0, sizeof(xpp));
4602                 memset(&xecfg, 0, sizeof(xecfg));
4603                 if (p->status == 0)
4604                         return error("internal diff status error");
4605                 if (p->status == DIFF_STATUS_UNKNOWN)
4606                         continue;
4607                 if (diff_unmodified_pair(p))
4608                         continue;
4609                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4610                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4611                         continue;
4612                 if (DIFF_PAIR_UNMERGED(p))
4613                         continue;
4614
4615                 diff_fill_sha1_info(p->one);
4616                 diff_fill_sha1_info(p->two);
4617
4618                 len1 = remove_space(p->one->path, strlen(p->one->path));
4619                 len2 = remove_space(p->two->path, strlen(p->two->path));
4620                 patch_id_add_string(&ctx, "diff--git");
4621                 patch_id_add_string(&ctx, "a/");
4622                 git_SHA1_Update(&ctx, p->one->path, len1);
4623                 patch_id_add_string(&ctx, "b/");
4624                 git_SHA1_Update(&ctx, p->two->path, len2);
4625
4626                 if (p->one->mode == 0) {
4627                         patch_id_add_string(&ctx, "newfilemode");
4628                         patch_id_add_mode(&ctx, p->two->mode);
4629                         patch_id_add_string(&ctx, "---/dev/null");
4630                         patch_id_add_string(&ctx, "+++b/");
4631                         git_SHA1_Update(&ctx, p->two->path, len2);
4632                 } else if (p->two->mode == 0) {
4633                         patch_id_add_string(&ctx, "deletedfilemode");
4634                         patch_id_add_mode(&ctx, p->one->mode);
4635                         patch_id_add_string(&ctx, "---a/");
4636                         git_SHA1_Update(&ctx, p->one->path, len1);
4637                         patch_id_add_string(&ctx, "+++/dev/null");
4638                 } else {
4639                         patch_id_add_string(&ctx, "---a/");
4640                         git_SHA1_Update(&ctx, p->one->path, len1);
4641                         patch_id_add_string(&ctx, "+++b/");
4642                         git_SHA1_Update(&ctx, p->two->path, len2);
4643                 }
4644
4645                 if (diff_header_only)
4646                         continue;
4647
4648                 if (fill_mmfile(&mf1, p->one) < 0 ||
4649                     fill_mmfile(&mf2, p->two) < 0)
4650                         return error("unable to read files to diff");
4651
4652                 if (diff_filespec_is_binary(p->one) ||
4653                     diff_filespec_is_binary(p->two)) {
4654                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4655                                         40);
4656                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4657                                         40);
4658                         continue;
4659                 }
4660
4661                 xpp.flags = 0;
4662                 xecfg.ctxlen = 3;
4663                 xecfg.flags = 0;
4664                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4665                                   &xpp, &xecfg))
4666                         return error("unable to generate patch-id diff for %s",
4667                                      p->one->path);
4668         }
4669
4670         git_SHA1_Final(sha1, &ctx);
4671         return 0;
4672 }
4673
4674 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4675 {
4676         struct diff_queue_struct *q = &diff_queued_diff;
4677         int i;
4678         int result = diff_get_patch_id(options, sha1, diff_header_only);
4679
4680         for (i = 0; i < q->nr; i++)
4681                 diff_free_filepair(q->queue[i]);
4682
4683         free(q->queue);
4684         DIFF_QUEUE_CLEAR(q);
4685
4686         return result;
4687 }
4688
4689 static int is_summary_empty(const struct diff_queue_struct *q)
4690 {
4691         int i;
4692
4693         for (i = 0; i < q->nr; i++) {
4694                 const struct diff_filepair *p = q->queue[i];
4695
4696                 switch (p->status) {
4697                 case DIFF_STATUS_DELETED:
4698                 case DIFF_STATUS_ADDED:
4699                 case DIFF_STATUS_COPIED:
4700                 case DIFF_STATUS_RENAMED:
4701                         return 0;
4702                 default:
4703                         if (p->score)
4704                                 return 0;
4705                         if (p->one->mode && p->two->mode &&
4706                             p->one->mode != p->two->mode)
4707                                 return 0;
4708                         break;
4709                 }
4710         }
4711         return 1;
4712 }
4713
4714 static const char rename_limit_warning[] =
4715 N_("inexact rename detection was skipped due to too many files.");
4716
4717 static const char degrade_cc_to_c_warning[] =
4718 N_("only found copies from modified paths due to too many files.");
4719
4720 static const char rename_limit_advice[] =
4721 N_("you may want to set your %s variable to at least "
4722    "%d and retry the command.");
4723
4724 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4725 {
4726         if (degraded_cc)
4727                 warning(_(degrade_cc_to_c_warning));
4728         else if (needed)
4729                 warning(_(rename_limit_warning));
4730         else
4731                 return;
4732         if (0 < needed && needed < 32767)
4733                 warning(_(rename_limit_advice), varname, needed);
4734 }
4735
4736 void diff_flush(struct diff_options *options)
4737 {
4738         struct diff_queue_struct *q = &diff_queued_diff;
4739         int i, output_format = options->output_format;
4740         int separator = 0;
4741         int dirstat_by_line = 0;
4742
4743         /*
4744          * Order: raw, stat, summary, patch
4745          * or:    name/name-status/checkdiff (other bits clear)
4746          */
4747         if (!q->nr)
4748                 goto free_queue;
4749
4750         if (output_format & (DIFF_FORMAT_RAW |
4751                              DIFF_FORMAT_NAME |
4752                              DIFF_FORMAT_NAME_STATUS |
4753                              DIFF_FORMAT_CHECKDIFF)) {
4754                 for (i = 0; i < q->nr; i++) {
4755                         struct diff_filepair *p = q->queue[i];
4756                         if (check_pair_status(p))
4757                                 flush_one_pair(p, options);
4758                 }
4759                 separator++;
4760         }
4761
4762         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4763                 dirstat_by_line = 1;
4764
4765         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4766             dirstat_by_line) {
4767                 struct diffstat_t diffstat;
4768
4769                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4770                 for (i = 0; i < q->nr; i++) {
4771                         struct diff_filepair *p = q->queue[i];
4772                         if (check_pair_status(p))
4773                                 diff_flush_stat(p, options, &diffstat);
4774                 }
4775                 if (output_format & DIFF_FORMAT_NUMSTAT)
4776                         show_numstat(&diffstat, options);
4777                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4778                         show_stats(&diffstat, options);
4779                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4780                         show_shortstats(&diffstat, options);
4781                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4782                         show_dirstat_by_line(&diffstat, options);
4783                 free_diffstat_info(&diffstat);
4784                 separator++;
4785         }
4786         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4787                 show_dirstat(options);
4788
4789         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4790                 for (i = 0; i < q->nr; i++) {
4791                         diff_summary(options, q->queue[i]);
4792                 }
4793                 separator++;
4794         }
4795
4796         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4797             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4798             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4799                 /*
4800                  * run diff_flush_patch for the exit status. setting
4801                  * options->file to /dev/null should be safe, because we
4802                  * aren't supposed to produce any output anyway.
4803                  */
4804                 if (options->close_file)
4805                         fclose(options->file);
4806                 options->file = xfopen("/dev/null", "w");
4807                 options->close_file = 1;
4808                 for (i = 0; i < q->nr; i++) {
4809                         struct diff_filepair *p = q->queue[i];
4810                         if (check_pair_status(p))
4811                                 diff_flush_patch(p, options);
4812                         if (options->found_changes)
4813                                 break;
4814                 }
4815         }
4816
4817         if (output_format & DIFF_FORMAT_PATCH) {
4818                 if (separator) {
4819                         fprintf(options->file, "%s%c",
4820                                 diff_line_prefix(options),
4821                                 options->line_termination);
4822                         if (options->stat_sep) {
4823                                 /* attach patch instead of inline */
4824                                 fputs(options->stat_sep, options->file);
4825                         }
4826                 }
4827
4828                 for (i = 0; i < q->nr; i++) {
4829                         struct diff_filepair *p = q->queue[i];
4830                         if (check_pair_status(p))
4831                                 diff_flush_patch(p, options);
4832                 }
4833         }
4834
4835         if (output_format & DIFF_FORMAT_CALLBACK)
4836                 options->format_callback(q, options, options->format_callback_data);
4837
4838         for (i = 0; i < q->nr; i++)
4839                 diff_free_filepair(q->queue[i]);
4840 free_queue:
4841         free(q->queue);
4842         DIFF_QUEUE_CLEAR(q);
4843         if (options->close_file)
4844                 fclose(options->file);
4845
4846         /*
4847          * Report the content-level differences with HAS_CHANGES;
4848          * diff_addremove/diff_change does not set the bit when
4849          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4850          */
4851         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4852                 if (options->found_changes)
4853                         DIFF_OPT_SET(options, HAS_CHANGES);
4854                 else
4855                         DIFF_OPT_CLR(options, HAS_CHANGES);
4856         }
4857 }
4858
4859 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4860 {
4861         return (((p->status == DIFF_STATUS_MODIFIED) &&
4862                  ((p->score &&
4863                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4864                   (!p->score &&
4865                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4866                 ((p->status != DIFF_STATUS_MODIFIED) &&
4867                  filter_bit_tst(p->status, options)));
4868 }
4869
4870 static void diffcore_apply_filter(struct diff_options *options)
4871 {
4872         int i;
4873         struct diff_queue_struct *q = &diff_queued_diff;
4874         struct diff_queue_struct outq;
4875
4876         DIFF_QUEUE_CLEAR(&outq);
4877
4878         if (!options->filter)
4879                 return;
4880
4881         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4882                 int found;
4883                 for (i = found = 0; !found && i < q->nr; i++) {
4884                         if (match_filter(options, q->queue[i]))
4885                                 found++;
4886                 }
4887                 if (found)
4888                         return;
4889
4890                 /* otherwise we will clear the whole queue
4891                  * by copying the empty outq at the end of this
4892                  * function, but first clear the current entries
4893                  * in the queue.
4894                  */
4895                 for (i = 0; i < q->nr; i++)
4896                         diff_free_filepair(q->queue[i]);
4897         }
4898         else {
4899                 /* Only the matching ones */
4900                 for (i = 0; i < q->nr; i++) {
4901                         struct diff_filepair *p = q->queue[i];
4902                         if (match_filter(options, p))
4903                                 diff_q(&outq, p);
4904                         else
4905                                 diff_free_filepair(p);
4906                 }
4907         }
4908         free(q->queue);
4909         *q = outq;
4910 }
4911
4912 /* Check whether two filespecs with the same mode and size are identical */
4913 static int diff_filespec_is_identical(struct diff_filespec *one,
4914                                       struct diff_filespec *two)
4915 {
4916         if (S_ISGITLINK(one->mode))
4917                 return 0;
4918         if (diff_populate_filespec(one, 0))
4919                 return 0;
4920         if (diff_populate_filespec(two, 0))
4921                 return 0;
4922         return !memcmp(one->data, two->data, one->size);
4923 }
4924
4925 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4926 {
4927         if (p->done_skip_stat_unmatch)
4928                 return p->skip_stat_unmatch_result;
4929
4930         p->done_skip_stat_unmatch = 1;
4931         p->skip_stat_unmatch_result = 0;
4932         /*
4933          * 1. Entries that come from stat info dirtiness
4934          *    always have both sides (iow, not create/delete),
4935          *    one side of the object name is unknown, with
4936          *    the same mode and size.  Keep the ones that
4937          *    do not match these criteria.  They have real
4938          *    differences.
4939          *
4940          * 2. At this point, the file is known to be modified,
4941          *    with the same mode and size, and the object
4942          *    name of one side is unknown.  Need to inspect
4943          *    the identical contents.
4944          */
4945         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4946             !DIFF_FILE_VALID(p->two) ||
4947             (p->one->oid_valid && p->two->oid_valid) ||
4948             (p->one->mode != p->two->mode) ||
4949             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4950             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4951             (p->one->size != p->two->size) ||
4952             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4953                 p->skip_stat_unmatch_result = 1;
4954         return p->skip_stat_unmatch_result;
4955 }
4956
4957 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4958 {
4959         int i;
4960         struct diff_queue_struct *q = &diff_queued_diff;
4961         struct diff_queue_struct outq;
4962         DIFF_QUEUE_CLEAR(&outq);
4963
4964         for (i = 0; i < q->nr; i++) {
4965                 struct diff_filepair *p = q->queue[i];
4966
4967                 if (diff_filespec_check_stat_unmatch(p))
4968                         diff_q(&outq, p);
4969                 else {
4970                         /*
4971                          * The caller can subtract 1 from skip_stat_unmatch
4972                          * to determine how many paths were dirty only
4973                          * due to stat info mismatch.
4974                          */
4975                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4976                                 diffopt->skip_stat_unmatch++;
4977                         diff_free_filepair(p);
4978                 }
4979         }
4980         free(q->queue);
4981         *q = outq;
4982 }
4983
4984 static int diffnamecmp(const void *a_, const void *b_)
4985 {
4986         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4987         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4988         const char *name_a, *name_b;
4989
4990         name_a = a->one ? a->one->path : a->two->path;
4991         name_b = b->one ? b->one->path : b->two->path;
4992         return strcmp(name_a, name_b);
4993 }
4994
4995 void diffcore_fix_diff_index(struct diff_options *options)
4996 {
4997         struct diff_queue_struct *q = &diff_queued_diff;
4998         QSORT(q->queue, q->nr, diffnamecmp);
4999 }
5000
5001 void diffcore_std(struct diff_options *options)
5002 {
5003         /* NOTE please keep the following in sync with diff_tree_combined() */
5004         if (options->skip_stat_unmatch)
5005                 diffcore_skip_stat_unmatch(options);
5006         if (!options->found_follow) {
5007                 /* See try_to_follow_renames() in tree-diff.c */
5008                 if (options->break_opt != -1)
5009                         diffcore_break(options->break_opt);
5010                 if (options->detect_rename)
5011                         diffcore_rename(options);
5012                 if (options->break_opt != -1)
5013                         diffcore_merge_broken();
5014         }
5015         if (options->pickaxe)
5016                 diffcore_pickaxe(options);
5017         if (options->orderfile)
5018                 diffcore_order(options->orderfile);
5019         if (!options->found_follow)
5020                 /* See try_to_follow_renames() in tree-diff.c */
5021                 diff_resolve_rename_copy();
5022         diffcore_apply_filter(options);
5023
5024         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5025                 DIFF_OPT_SET(options, HAS_CHANGES);
5026         else
5027                 DIFF_OPT_CLR(options, HAS_CHANGES);
5028
5029         options->found_follow = 0;
5030 }
5031
5032 int diff_result_code(struct diff_options *opt, int status)
5033 {
5034         int result = 0;
5035
5036         diff_warn_rename_limit("diff.renameLimit",
5037                                opt->needed_rename_limit,
5038                                opt->degraded_cc_to_c);
5039         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5040             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5041                 return status;
5042         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5043             DIFF_OPT_TST(opt, HAS_CHANGES))
5044                 result |= 01;
5045         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5046             DIFF_OPT_TST(opt, CHECK_FAILED))
5047                 result |= 02;
5048         return result;
5049 }
5050
5051 int diff_can_quit_early(struct diff_options *opt)
5052 {
5053         return (DIFF_OPT_TST(opt, QUICK) &&
5054                 !opt->filter &&
5055                 DIFF_OPT_TST(opt, HAS_CHANGES));
5056 }
5057
5058 /*
5059  * Shall changes to this submodule be ignored?
5060  *
5061  * Submodule changes can be configured to be ignored separately for each path,
5062  * but that configuration can be overridden from the command line.
5063  */
5064 static int is_submodule_ignored(const char *path, struct diff_options *options)
5065 {
5066         int ignored = 0;
5067         unsigned orig_flags = options->flags;
5068         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5069                 set_diffopt_flags_from_submodule_config(options, path);
5070         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5071                 ignored = 1;
5072         options->flags = orig_flags;
5073         return ignored;
5074 }
5075
5076 void diff_addremove(struct diff_options *options,
5077                     int addremove, unsigned mode,
5078                     const unsigned char *sha1,
5079                     int sha1_valid,
5080                     const char *concatpath, unsigned dirty_submodule)
5081 {
5082         struct diff_filespec *one, *two;
5083
5084         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5085                 return;
5086
5087         /* This may look odd, but it is a preparation for
5088          * feeding "there are unchanged files which should
5089          * not produce diffs, but when you are doing copy
5090          * detection you would need them, so here they are"
5091          * entries to the diff-core.  They will be prefixed
5092          * with something like '=' or '*' (I haven't decided
5093          * which but should not make any difference).
5094          * Feeding the same new and old to diff_change()
5095          * also has the same effect.
5096          * Before the final output happens, they are pruned after
5097          * merged into rename/copy pairs as appropriate.
5098          */
5099         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5100                 addremove = (addremove == '+' ? '-' :
5101                              addremove == '-' ? '+' : addremove);
5102
5103         if (options->prefix &&
5104             strncmp(concatpath, options->prefix, options->prefix_length))
5105                 return;
5106
5107         one = alloc_filespec(concatpath);
5108         two = alloc_filespec(concatpath);
5109
5110         if (addremove != '+')
5111                 fill_filespec(one, sha1, sha1_valid, mode);
5112         if (addremove != '-') {
5113                 fill_filespec(two, sha1, sha1_valid, mode);
5114                 two->dirty_submodule = dirty_submodule;
5115         }
5116
5117         diff_queue(&diff_queued_diff, one, two);
5118         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5119                 DIFF_OPT_SET(options, HAS_CHANGES);
5120 }
5121
5122 void diff_change(struct diff_options *options,
5123                  unsigned old_mode, unsigned new_mode,
5124                  const unsigned char *old_sha1,
5125                  const unsigned char *new_sha1,
5126                  int old_sha1_valid, int new_sha1_valid,
5127                  const char *concatpath,
5128                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5129 {
5130         struct diff_filespec *one, *two;
5131         struct diff_filepair *p;
5132
5133         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5134             is_submodule_ignored(concatpath, options))
5135                 return;
5136
5137         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5138                 SWAP(old_mode, new_mode);
5139                 SWAP(old_sha1, new_sha1);
5140                 SWAP(old_sha1_valid, new_sha1_valid);
5141                 SWAP(old_dirty_submodule, new_dirty_submodule);
5142         }
5143
5144         if (options->prefix &&
5145             strncmp(concatpath, options->prefix, options->prefix_length))
5146                 return;
5147
5148         one = alloc_filespec(concatpath);
5149         two = alloc_filespec(concatpath);
5150         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5151         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5152         one->dirty_submodule = old_dirty_submodule;
5153         two->dirty_submodule = new_dirty_submodule;
5154         p = diff_queue(&diff_queued_diff, one, two);
5155
5156         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5157                 return;
5158
5159         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5160             !diff_filespec_check_stat_unmatch(p))
5161                 return;
5162
5163         DIFF_OPT_SET(options, HAS_CHANGES);
5164 }
5165
5166 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5167 {
5168         struct diff_filepair *pair;
5169         struct diff_filespec *one, *two;
5170
5171         if (options->prefix &&
5172             strncmp(path, options->prefix, options->prefix_length))
5173                 return NULL;
5174
5175         one = alloc_filespec(path);
5176         two = alloc_filespec(path);
5177         pair = diff_queue(&diff_queued_diff, one, two);
5178         pair->is_unmerged = 1;
5179         return pair;
5180 }
5181
5182 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5183                 size_t *outsize)
5184 {
5185         struct diff_tempfile *temp;
5186         const char *argv[3];
5187         const char **arg = argv;
5188         struct child_process child = CHILD_PROCESS_INIT;
5189         struct strbuf buf = STRBUF_INIT;
5190         int err = 0;
5191
5192         temp = prepare_temp_file(spec->path, spec);
5193         *arg++ = pgm;
5194         *arg++ = temp->name;
5195         *arg = NULL;
5196
5197         child.use_shell = 1;
5198         child.argv = argv;
5199         child.out = -1;
5200         if (start_command(&child)) {
5201                 remove_tempfile();
5202                 return NULL;
5203         }
5204
5205         if (strbuf_read(&buf, child.out, 0) < 0)
5206                 err = error("error reading from textconv command '%s'", pgm);
5207         close(child.out);
5208
5209         if (finish_command(&child) || err) {
5210                 strbuf_release(&buf);
5211                 remove_tempfile();
5212                 return NULL;
5213         }
5214         remove_tempfile();
5215
5216         return strbuf_detach(&buf, outsize);
5217 }
5218
5219 size_t fill_textconv(struct userdiff_driver *driver,
5220                      struct diff_filespec *df,
5221                      char **outbuf)
5222 {
5223         size_t size;
5224
5225         if (!driver) {
5226                 if (!DIFF_FILE_VALID(df)) {
5227                         *outbuf = "";
5228                         return 0;
5229                 }
5230                 if (diff_populate_filespec(df, 0))
5231                         die("unable to read files to diff");
5232                 *outbuf = df->data;
5233                 return df->size;
5234         }
5235
5236         if (!driver->textconv)
5237                 die("BUG: fill_textconv called with non-textconv driver");
5238
5239         if (driver->textconv_cache && df->oid_valid) {
5240                 *outbuf = notes_cache_get(driver->textconv_cache,
5241                                           &df->oid,
5242                                           &size);
5243                 if (*outbuf)
5244                         return size;
5245         }
5246
5247         *outbuf = run_textconv(driver->textconv, df, &size);
5248         if (!*outbuf)
5249                 die("unable to read files to diff");
5250
5251         if (driver->textconv_cache && df->oid_valid) {
5252                 /* ignore errors, as we might be in a readonly repository */
5253                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
5254                                 size);
5255                 /*
5256                  * we could save up changes and flush them all at the end,
5257                  * but we would need an extra call after all diffing is done.
5258                  * Since generating a cache entry is the slow path anyway,
5259                  * this extra overhead probably isn't a big deal.
5260                  */
5261                 notes_cache_write(driver->textconv_cache);
5262         }
5263
5264         return size;
5265 }
5266
5267 int textconv_object(const char *path,
5268                     unsigned mode,
5269                     const struct object_id *oid,
5270                     int oid_valid,
5271                     char **buf,
5272                     unsigned long *buf_size)
5273 {
5274         struct diff_filespec *df;
5275         struct userdiff_driver *textconv;
5276
5277         df = alloc_filespec(path);
5278         fill_filespec(df, oid->hash, oid_valid, mode);
5279         textconv = get_textconv(df);
5280         if (!textconv) {
5281                 free_filespec(df);
5282                 return 0;
5283         }
5284
5285         *buf_size = fill_textconv(textconv, df, buf);
5286         free_filespec(df);
5287         return 1;
5288 }
5289
5290 void setup_diff_pager(struct diff_options *opt)
5291 {
5292         /*
5293          * If the user asked for our exit code, then either they want --quiet
5294          * or --exit-code. We should definitely not bother with a pager in the
5295          * former case, as we will generate no output. Since we still properly
5296          * report our exit code even when a pager is run, we _could_ run a
5297          * pager with --exit-code. But since we have not done so historically,
5298          * and because it is easy to find people oneline advising "git diff
5299          * --exit-code" in hooks and other scripts, we do not do so.
5300          */
5301         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5302             check_pager_config("diff") != 0)
5303                 setup_pager();
5304 }