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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26
  27 #ifdef NO_FAST_WORKING_DIRECTORY
  28 #define FAST_WORKING_DIRECTORY 0
  29 #else
  30 #define FAST_WORKING_DIRECTORY 1
  31 #endif
  32
  33 static int diff_detect_rename_default;
  34 static int diff_indent_heuristic = 1;
  35 static int diff_rename_limit_default = 400;
  36 static int diff_suppress_blank_empty;
  37 static int diff_use_color_default = -1;
  38 static int diff_color_moved_default;
  39 static int diff_context_default = 3;
  40 static int diff_interhunk_context_default;
  41 static const char *diff_word_regex_cfg;
  42 static const char *external_diff_cmd_cfg;
  43 static const char *diff_order_file_cfg;
  44 int diff_auto_refresh_index = 1;
  45 static int diff_mnemonic_prefix;
  46 static int diff_no_prefix;
  47 static int diff_stat_graph_width;
  48 static int diff_dirstat_permille_default = 30;
  49 static struct diff_options default_diff_options;
  50 static long diff_algorithm;
  51 static unsigned ws_error_highlight_default = WSEH_NEW;
  52
  53 static char diff_colors[][COLOR_MAXLEN] = {
  54         GIT_COLOR_RESET,
  55         GIT_COLOR_NORMAL,       /* CONTEXT */
  56         GIT_COLOR_BOLD,         /* METAINFO */
  57         GIT_COLOR_CYAN,         /* FRAGINFO */
  58         GIT_COLOR_RED,          /* OLD */
  59         GIT_COLOR_GREEN,        /* NEW */
  60         GIT_COLOR_YELLOW,       /* COMMIT */
  61         GIT_COLOR_BG_RED,       /* WHITESPACE */
  62         GIT_COLOR_NORMAL,       /* FUNCINFO */
  63         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  64         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  65         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  66         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  67         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  68         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  69         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  70         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  71 };
  72
  73 static NORETURN void die_want_option(const char *option_name)
  74 {
  75         die(_("option '%s' requires a value"), option_name);
  76 }
  77
  78 static int parse_diff_color_slot(const char *var)
  79 {
  80         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  81                 return DIFF_CONTEXT;
  82         if (!strcasecmp(var, "meta"))
  83                 return DIFF_METAINFO;
  84         if (!strcasecmp(var, "frag"))
  85                 return DIFF_FRAGINFO;
  86         if (!strcasecmp(var, "old"))
  87                 return DIFF_FILE_OLD;
  88         if (!strcasecmp(var, "new"))
  89                 return DIFF_FILE_NEW;
  90         if (!strcasecmp(var, "commit"))
  91                 return DIFF_COMMIT;
  92         if (!strcasecmp(var, "whitespace"))
  93                 return DIFF_WHITESPACE;
  94         if (!strcasecmp(var, "func"))
  95                 return DIFF_FUNCINFO;
  96         if (!strcasecmp(var, "oldmoved"))
  97                 return DIFF_FILE_OLD_MOVED;
  98         if (!strcasecmp(var, "oldmovedalternative"))
  99                 return DIFF_FILE_OLD_MOVED_ALT;
 100         if (!strcasecmp(var, "oldmoveddimmed"))
 101                 return DIFF_FILE_OLD_MOVED_DIM;
 102         if (!strcasecmp(var, "oldmovedalternativedimmed"))
 103                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
 104         if (!strcasecmp(var, "newmoved"))
 105                 return DIFF_FILE_NEW_MOVED;
 106         if (!strcasecmp(var, "newmovedalternative"))
 107                 return DIFF_FILE_NEW_MOVED_ALT;
 108         if (!strcasecmp(var, "newmoveddimmed"))
 109                 return DIFF_FILE_NEW_MOVED_DIM;
 110         if (!strcasecmp(var, "newmovedalternativedimmed"))
 111                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
 112         return -1;
 113 }
 114
 115 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 116                                 struct strbuf *errmsg)
 117 {
 118         char *params_copy = xstrdup(params_string);
 119         struct string_list params = STRING_LIST_INIT_NODUP;
 120         int ret = 0;
 121         int i;
 122
 123         if (*params_copy)
 124                 string_list_split_in_place(&params, params_copy, ',', -1);
 125         for (i = 0; i < params.nr; i++) {
 126                 const char *p = params.items[i].string;
 127                 if (!strcmp(p, "changes")) {
 128                         options->flags.dirstat_by_line = 0;
 129                         options->flags.dirstat_by_file = 0;
 130                 } else if (!strcmp(p, "lines")) {
 131                         options->flags.dirstat_by_line = 1;
 132                         options->flags.dirstat_by_file = 0;
 133                 } else if (!strcmp(p, "files")) {
 134                         options->flags.dirstat_by_line = 0;
 135                         options->flags.dirstat_by_file = 1;
 136                 } else if (!strcmp(p, "noncumulative")) {
 137                         options->flags.dirstat_cumulative = 0;
 138                 } else if (!strcmp(p, "cumulative")) {
 139                         options->flags.dirstat_cumulative = 1;
 140                 } else if (isdigit(*p)) {
 141                         char *end;
 142                         int permille = strtoul(p, &end, 10) * 10;
 143                         if (*end == '.' && isdigit(*++end)) {
 144                                 /* only use first digit */
 145                                 permille += *end - '0';
 146                                 /* .. and ignore any further digits */
 147                                 while (isdigit(*++end))
 148                                         ; /* nothing */
 149                         }
 150                         if (!*end)
 151                                 options->dirstat_permille = permille;
 152                         else {
 153                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 154                                             p);
 155                                 ret++;
 156                         }
 157                 } else {
 158                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 159                         ret++;
 160                 }
 161
 162         }
 163         string_list_clear(&params, 0);
 164         free(params_copy);
 165         return ret;
 166 }
 167
 168 static int parse_submodule_params(struct diff_options *options, const char *value)
 169 {
 170         if (!strcmp(value, "log"))
 171                 options->submodule_format = DIFF_SUBMODULE_LOG;
 172         else if (!strcmp(value, "short"))
 173                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 174         else if (!strcmp(value, "diff"))
 175                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 176         else
 177                 return -1;
 178         return 0;
 179 }
 180
 181 static int git_config_rename(const char *var, const char *value)
 182 {
 183         if (!value)
 184                 return DIFF_DETECT_RENAME;
 185         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 186                 return  DIFF_DETECT_COPY;
 187         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 188 }
 189
 190 long parse_algorithm_value(const char *value)
 191 {
 192         if (!value)
 193                 return -1;
 194         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 195                 return 0;
 196         else if (!strcasecmp(value, "minimal"))
 197                 return XDF_NEED_MINIMAL;
 198         else if (!strcasecmp(value, "patience"))
 199                 return XDF_PATIENCE_DIFF;
 200         else if (!strcasecmp(value, "histogram"))
 201                 return XDF_HISTOGRAM_DIFF;
 202         return -1;
 203 }
 204
 205 static int parse_one_token(const char **arg, const char *token)
 206 {
 207         const char *rest;
 208         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 209                 *arg = rest;
 210                 return 1;
 211         }
 212         return 0;
 213 }
 214
 215 static int parse_ws_error_highlight(const char *arg)
 216 {
 217         const char *orig_arg = arg;
 218         unsigned val = 0;
 219
 220         while (*arg) {
 221                 if (parse_one_token(&arg, "none"))
 222                         val = 0;
 223                 else if (parse_one_token(&arg, "default"))
 224                         val = WSEH_NEW;
 225                 else if (parse_one_token(&arg, "all"))
 226                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 227                 else if (parse_one_token(&arg, "new"))
 228                         val |= WSEH_NEW;
 229                 else if (parse_one_token(&arg, "old"))
 230                         val |= WSEH_OLD;
 231                 else if (parse_one_token(&arg, "context"))
 232                         val |= WSEH_CONTEXT;
 233                 else {
 234                         return -1 - (int)(arg - orig_arg);
 235                 }
 236                 if (*arg)
 237                         arg++;
 238         }
 239         return val;
 240 }
 241
 242 /*
 243  * These are to give UI layer defaults.
 244  * The core-level commands such as git-diff-files should
 245  * never be affected by the setting of diff.renames
 246  * the user happens to have in the configuration file.
 247  */
 248 void init_diff_ui_defaults(void)
 249 {
 250         diff_detect_rename_default = DIFF_DETECT_RENAME;
 251 }
 252
 253 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 254 {
 255         if (!strcmp(var, "diff.indentheuristic"))
 256                 diff_indent_heuristic = git_config_bool(var, value);
 257         return 0;
 258 }
 259
 260 static int parse_color_moved(const char *arg)
 261 {
 262         switch (git_parse_maybe_bool(arg)) {
 263         case 0:
 264                 return COLOR_MOVED_NO;
 265         case 1:
 266                 return COLOR_MOVED_DEFAULT;
 267         default:
 268                 break;
 269         }
 270
 271         if (!strcmp(arg, "no"))
 272                 return COLOR_MOVED_NO;
 273         else if (!strcmp(arg, "plain"))
 274                 return COLOR_MOVED_PLAIN;
 275         else if (!strcmp(arg, "zebra"))
 276                 return COLOR_MOVED_ZEBRA;
 277         else if (!strcmp(arg, "default"))
 278                 return COLOR_MOVED_DEFAULT;
 279         else if (!strcmp(arg, "dimmed_zebra"))
 280                 return COLOR_MOVED_ZEBRA_DIM;
 281         else
 282                 return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'dimmed_zebra', 'plain'"));
 283 }
 284
 285 int git_diff_ui_config(const char *var, const char *value, void *cb)
 286 {
 287         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 288                 diff_use_color_default = git_config_colorbool(var, value);
 289                 return 0;
 290         }
 291         if (!strcmp(var, "diff.colormoved")) {
 292                 int cm = parse_color_moved(value);
 293                 if (cm < 0)
 294                         return -1;
 295                 diff_color_moved_default = cm;
 296                 return 0;
 297         }
 298         if (!strcmp(var, "diff.context")) {
 299                 diff_context_default = git_config_int(var, value);
 300                 if (diff_context_default < 0)
 301                         return -1;
 302                 return 0;
 303         }
 304         if (!strcmp(var, "diff.interhunkcontext")) {
 305                 diff_interhunk_context_default = git_config_int(var, value);
 306                 if (diff_interhunk_context_default < 0)
 307                         return -1;
 308                 return 0;
 309         }
 310         if (!strcmp(var, "diff.renames")) {
 311                 diff_detect_rename_default = git_config_rename(var, value);
 312                 return 0;
 313         }
 314         if (!strcmp(var, "diff.autorefreshindex")) {
 315                 diff_auto_refresh_index = git_config_bool(var, value);
 316                 return 0;
 317         }
 318         if (!strcmp(var, "diff.mnemonicprefix")) {
 319                 diff_mnemonic_prefix = git_config_bool(var, value);
 320                 return 0;
 321         }
 322         if (!strcmp(var, "diff.noprefix")) {
 323                 diff_no_prefix = git_config_bool(var, value);
 324                 return 0;
 325         }
 326         if (!strcmp(var, "diff.statgraphwidth")) {
 327                 diff_stat_graph_width = git_config_int(var, value);
 328                 return 0;
 329         }
 330         if (!strcmp(var, "diff.external"))
 331                 return git_config_string(&external_diff_cmd_cfg, var, value);
 332         if (!strcmp(var, "diff.wordregex"))
 333                 return git_config_string(&diff_word_regex_cfg, var, value);
 334         if (!strcmp(var, "diff.orderfile"))
 335                 return git_config_pathname(&diff_order_file_cfg, var, value);
 336
 337         if (!strcmp(var, "diff.ignoresubmodules"))
 338                 handle_ignore_submodules_arg(&default_diff_options, value);
 339
 340         if (!strcmp(var, "diff.submodule")) {
 341                 if (parse_submodule_params(&default_diff_options, value))
 342                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 343                                 value);
 344                 return 0;
 345         }
 346
 347         if (!strcmp(var, "diff.algorithm")) {
 348                 diff_algorithm = parse_algorithm_value(value);
 349                 if (diff_algorithm < 0)
 350                         return -1;
 351                 return 0;
 352         }
 353
 354         if (!strcmp(var, "diff.wserrorhighlight")) {
 355                 int val = parse_ws_error_highlight(value);
 356                 if (val < 0)
 357                         return -1;
 358                 ws_error_highlight_default = val;
 359                 return 0;
 360         }
 361
 362         if (git_color_config(var, value, cb) < 0)
 363                 return -1;
 364
 365         return git_diff_basic_config(var, value, cb);
 366 }
 367
 368 int git_diff_basic_config(const char *var, const char *value, void *cb)
 369 {
 370         const char *name;
 371
 372         if (!strcmp(var, "diff.renamelimit")) {
 373                 diff_rename_limit_default = git_config_int(var, value);
 374                 return 0;
 375         }
 376
 377         if (userdiff_config(var, value) < 0)
 378                 return -1;
 379
 380         if (skip_prefix(var, "diff.color.", &name) ||
 381             skip_prefix(var, "color.diff.", &name)) {
 382                 int slot = parse_diff_color_slot(name);
 383                 if (slot < 0)
 384                         return 0;
 385                 if (!value)
 386                         return config_error_nonbool(var);
 387                 return color_parse(value, diff_colors[slot]);
 388         }
 389
 390         /* like GNU diff's --suppress-blank-empty option  */
 391         if (!strcmp(var, "diff.suppressblankempty") ||
 392                         /* for backwards compatibility */
 393                         !strcmp(var, "diff.suppress-blank-empty")) {
 394                 diff_suppress_blank_empty = git_config_bool(var, value);
 395                 return 0;
 396         }
 397
 398         if (!strcmp(var, "diff.dirstat")) {
 399                 struct strbuf errmsg = STRBUF_INIT;
 400                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 401                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 402                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 403                                 errmsg.buf);
 404                 strbuf_release(&errmsg);
 405                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 406                 return 0;
 407         }
 408
 409         if (git_diff_heuristic_config(var, value, cb) < 0)
 410                 return -1;
 411
 412         return git_default_config(var, value, cb);
 413 }
 414
 415 static char *quote_two(const char *one, const char *two)
 416 {
 417         int need_one = quote_c_style(one, NULL, NULL, 1);
 418         int need_two = quote_c_style(two, NULL, NULL, 1);
 419         struct strbuf res = STRBUF_INIT;
 420
 421         if (need_one + need_two) {
 422                 strbuf_addch(&res, '"');
 423                 quote_c_style(one, &res, NULL, 1);
 424                 quote_c_style(two, &res, NULL, 1);
 425                 strbuf_addch(&res, '"');
 426         } else {
 427                 strbuf_addstr(&res, one);
 428                 strbuf_addstr(&res, two);
 429         }
 430         return strbuf_detach(&res, NULL);
 431 }
 432
 433 static const char *external_diff(void)
 434 {
 435         static const char *external_diff_cmd = NULL;
 436         static int done_preparing = 0;
 437
 438         if (done_preparing)
 439                 return external_diff_cmd;
 440         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 441         if (!external_diff_cmd)
 442                 external_diff_cmd = external_diff_cmd_cfg;
 443         done_preparing = 1;
 444         return external_diff_cmd;
 445 }
 446
 447 /*
 448  * Keep track of files used for diffing. Sometimes such an entry
 449  * refers to a temporary file, sometimes to an existing file, and
 450  * sometimes to "/dev/null".
 451  */
 452 static struct diff_tempfile {
 453         /*
 454          * filename external diff should read from, or NULL if this
 455          * entry is currently not in use:
 456          */
 457         const char *name;
 458
 459         char hex[GIT_MAX_HEXSZ + 1];
 460         char mode[10];
 461
 462         /*
 463          * If this diff_tempfile instance refers to a temporary file,
 464          * this tempfile object is used to manage its lifetime.
 465          */
 466         struct tempfile *tempfile;
 467 } diff_temp[2];
 468
 469 struct emit_callback {
 470         int color_diff;
 471         unsigned ws_rule;
 472         int blank_at_eof_in_preimage;
 473         int blank_at_eof_in_postimage;
 474         int lno_in_preimage;
 475         int lno_in_postimage;
 476         const char **label_path;
 477         struct diff_words_data *diff_words;
 478         struct diff_options *opt;
 479         struct strbuf *header;
 480 };
 481
 482 static int count_lines(const char *data, int size)
 483 {
 484         int count, ch, completely_empty = 1, nl_just_seen = 0;
 485         count = 0;
 486         while (0 < size--) {
 487                 ch = *data++;
 488                 if (ch == '\n') {
 489                         count++;
 490                         nl_just_seen = 1;
 491                         completely_empty = 0;
 492                 }
 493                 else {
 494                         nl_just_seen = 0;
 495                         completely_empty = 0;
 496                 }
 497         }
 498         if (completely_empty)
 499                 return 0;
 500         if (!nl_just_seen)
 501                 count++; /* no trailing newline */
 502         return count;
 503 }
 504
 505 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 506 {
 507         if (!DIFF_FILE_VALID(one)) {
 508                 mf->ptr = (char *)""; /* does not matter */
 509                 mf->size = 0;
 510                 return 0;
 511         }
 512         else if (diff_populate_filespec(one, 0))
 513                 return -1;
 514
 515         mf->ptr = one->data;
 516         mf->size = one->size;
 517         return 0;
 518 }
 519
 520 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 521 static unsigned long diff_filespec_size(struct diff_filespec *one)
 522 {
 523         if (!DIFF_FILE_VALID(one))
 524                 return 0;
 525         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 526         return one->size;
 527 }
 528
 529 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 530 {
 531         char *ptr = mf->ptr;
 532         long size = mf->size;
 533         int cnt = 0;
 534
 535         if (!size)
 536                 return cnt;
 537         ptr += size - 1; /* pointing at the very end */
 538         if (*ptr != '\n')
 539                 ; /* incomplete line */
 540         else
 541                 ptr--; /* skip the last LF */
 542         while (mf->ptr < ptr) {
 543                 char *prev_eol;
 544                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 545                         if (*prev_eol == '\n')
 546                                 break;
 547                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 548                         break;
 549                 cnt++;
 550                 ptr = prev_eol - 1;
 551         }
 552         return cnt;
 553 }
 554
 555 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 556                                struct emit_callback *ecbdata)
 557 {
 558         int l1, l2, at;
 559         unsigned ws_rule = ecbdata->ws_rule;
 560         l1 = count_trailing_blank(mf1, ws_rule);
 561         l2 = count_trailing_blank(mf2, ws_rule);
 562         if (l2 <= l1) {
 563                 ecbdata->blank_at_eof_in_preimage = 0;
 564                 ecbdata->blank_at_eof_in_postimage = 0;
 565                 return;
 566         }
 567         at = count_lines(mf1->ptr, mf1->size);
 568         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 569
 570         at = count_lines(mf2->ptr, mf2->size);
 571         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 572 }
 573
 574 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 575                         int first, const char *line, int len)
 576 {
 577         int has_trailing_newline, has_trailing_carriage_return;
 578         int nofirst;
 579         FILE *file = o->file;
 580
 581         fputs(diff_line_prefix(o), file);
 582
 583         if (len == 0) {
 584                 has_trailing_newline = (first == '\n');
 585                 has_trailing_carriage_return = (!has_trailing_newline &&
 586                                                 (first == '\r'));
 587                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 588         } else {
 589                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 590                 if (has_trailing_newline)
 591                         len--;
 592                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 593                 if (has_trailing_carriage_return)
 594                         len--;
 595                 nofirst = 0;
 596         }
 597
 598         if (len || !nofirst) {
 599                 fputs(set, file);
 600                 if (!nofirst)
 601                         fputc(first, file);
 602                 fwrite(line, len, 1, file);
 603                 fputs(reset, file);
 604         }
 605         if (has_trailing_carriage_return)
 606                 fputc('\r', file);
 607         if (has_trailing_newline)
 608                 fputc('\n', file);
 609 }
 610
 611 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 612                       const char *line, int len)
 613 {
 614         emit_line_0(o, set, reset, line[0], line+1, len-1);
 615 }
 616
 617 enum diff_symbol {
 618         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 619         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 620         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 621         DIFF_SYMBOL_BINARY_DIFF_BODY,
 622         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 623         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 624         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 625         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 626         DIFF_SYMBOL_STATS_LINE,
 627         DIFF_SYMBOL_WORD_DIFF,
 628         DIFF_SYMBOL_STAT_SEP,
 629         DIFF_SYMBOL_SUMMARY,
 630         DIFF_SYMBOL_SUBMODULE_ADD,
 631         DIFF_SYMBOL_SUBMODULE_DEL,
 632         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 633         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 634         DIFF_SYMBOL_SUBMODULE_HEADER,
 635         DIFF_SYMBOL_SUBMODULE_ERROR,
 636         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 637         DIFF_SYMBOL_REWRITE_DIFF,
 638         DIFF_SYMBOL_BINARY_FILES,
 639         DIFF_SYMBOL_HEADER,
 640         DIFF_SYMBOL_FILEPAIR_PLUS,
 641         DIFF_SYMBOL_FILEPAIR_MINUS,
 642         DIFF_SYMBOL_WORDS_PORCELAIN,
 643         DIFF_SYMBOL_WORDS,
 644         DIFF_SYMBOL_CONTEXT,
 645         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 646         DIFF_SYMBOL_PLUS,
 647         DIFF_SYMBOL_MINUS,
 648         DIFF_SYMBOL_NO_LF_EOF,
 649         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 650         DIFF_SYMBOL_CONTEXT_MARKER,
 651         DIFF_SYMBOL_SEPARATOR
 652 };
 653 /*
 654  * Flags for content lines:
 655  * 0..12 are whitespace rules
 656  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 657  * 16 is marking if the line is blank at EOF
 658  */
 659 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 660 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 661 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 662 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 663 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 664
 665 /*
 666  * This struct is used when we need to buffer the output of the diff output.
 667  *
 668  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 669  * into the pre/post image file. This pointer could be a union with the
 670  * line pointer. By storing an offset into the file instead of the literal line,
 671  * we can decrease the memory footprint for the buffered output. At first we
 672  * may want to only have indirection for the content lines, but we could also
 673  * enhance the state for emitting prefabricated lines, e.g. the similarity
 674  * score line or hunk/file headers would only need to store a number or path
 675  * and then the output can be constructed later on depending on state.
 676  */
 677 struct emitted_diff_symbol {
 678         const char *line;
 679         int len;
 680         int flags;
 681         enum diff_symbol s;
 682 };
 683 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 684
 685 struct emitted_diff_symbols {
 686         struct emitted_diff_symbol *buf;
 687         int nr, alloc;
 688 };
 689 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 690
 691 static void append_emitted_diff_symbol(struct diff_options *o,
 692                                        struct emitted_diff_symbol *e)
 693 {
 694         struct emitted_diff_symbol *f;
 695
 696         ALLOC_GROW(o->emitted_symbols->buf,
 697                    o->emitted_symbols->nr + 1,
 698                    o->emitted_symbols->alloc);
 699         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 700
 701         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 702         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 703 }
 704
 705 struct moved_entry {
 706         struct hashmap_entry ent;
 707         const struct emitted_diff_symbol *es;
 708         struct moved_entry *next_line;
 709 };
 710
 711 static int moved_entry_cmp(const struct diff_options *diffopt,
 712                            const struct moved_entry *a,
 713                            const struct moved_entry *b,
 714                            const void *keydata)
 715 {
 716         return !xdiff_compare_lines(a->es->line, a->es->len,
 717                                     b->es->line, b->es->len,
 718                                     diffopt->xdl_opts);
 719 }
 720
 721 static struct moved_entry *prepare_entry(struct diff_options *o,
 722                                          int line_no)
 723 {
 724         struct moved_entry *ret = xmalloc(sizeof(*ret));
 725         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 726
 727         ret->ent.hash = xdiff_hash_string(l->line, l->len, o->xdl_opts);
 728         ret->es = l;
 729         ret->next_line = NULL;
 730
 731         return ret;
 732 }
 733
 734 static void add_lines_to_move_detection(struct diff_options *o,
 735                                         struct hashmap *add_lines,
 736                                         struct hashmap *del_lines)
 737 {
 738         struct moved_entry *prev_line = NULL;
 739
 740         int n;
 741         for (n = 0; n < o->emitted_symbols->nr; n++) {
 742                 struct hashmap *hm;
 743                 struct moved_entry *key;
 744
 745                 switch (o->emitted_symbols->buf[n].s) {
 746                 case DIFF_SYMBOL_PLUS:
 747                         hm = add_lines;
 748                         break;
 749                 case DIFF_SYMBOL_MINUS:
 750                         hm = del_lines;
 751                         break;
 752                 default:
 753                         prev_line = NULL;
 754                         continue;
 755                 }
 756
 757                 key = prepare_entry(o, n);
 758                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 759                         prev_line->next_line = key;
 760
 761                 hashmap_add(hm, key);
 762                 prev_line = key;
 763         }
 764 }
 765
 766 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 767                                          int pmb_nr)
 768 {
 769         int lp, rp;
 770
 771         /* Shrink the set of potential block to the remaining running */
 772         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 773                 while (lp < pmb_nr && pmb[lp])
 774                         lp++;
 775                 /* lp points at the first NULL now */
 776
 777                 while (rp > -1 && !pmb[rp])
 778                         rp--;
 779                 /* rp points at the last non-NULL */
 780
 781                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 782                         pmb[lp] = pmb[rp];
 783                         pmb[rp] = NULL;
 784                         rp--;
 785                         lp++;
 786                 }
 787         }
 788
 789         /* Remember the number of running sets */
 790         return rp + 1;
 791 }
 792
 793 /*
 794  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 795  *
 796  * Otherwise, if the last block has fewer alphanumeric characters than
 797  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 798  * that block.
 799  *
 800  * The last block consists of the (n - block_length)'th line up to but not
 801  * including the nth line.
 802  *
 803  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
 804  * Think of a way to unify them.
 805  */
 806 static void adjust_last_block(struct diff_options *o, int n, int block_length)
 807 {
 808         int i, alnum_count = 0;
 809         if (o->color_moved == COLOR_MOVED_PLAIN)
 810                 return;
 811         for (i = 1; i < block_length + 1; i++) {
 812                 const char *c = o->emitted_symbols->buf[n - i].line;
 813                 for (; *c; c++) {
 814                         if (!isalnum(*c))
 815                                 continue;
 816                         alnum_count++;
 817                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
 818                                 return;
 819                 }
 820         }
 821         for (i = 1; i < block_length + 1; i++)
 822                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
 823 }
 824
 825 /* Find blocks of moved code, delegate actual coloring decision to helper */
 826 static void mark_color_as_moved(struct diff_options *o,
 827                                 struct hashmap *add_lines,
 828                                 struct hashmap *del_lines)
 829 {
 830         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 831         int pmb_nr = 0, pmb_alloc = 0;
 832         int n, flipped_block = 1, block_length = 0;
 833
 834
 835         for (n = 0; n < o->emitted_symbols->nr; n++) {
 836                 struct hashmap *hm = NULL;
 837                 struct moved_entry *key;
 838                 struct moved_entry *match = NULL;
 839                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 840                 int i;
 841
 842                 switch (l->s) {
 843                 case DIFF_SYMBOL_PLUS:
 844                         hm = del_lines;
 845                         key = prepare_entry(o, n);
 846                         match = hashmap_get(hm, key, o);
 847                         free(key);
 848                         break;
 849                 case DIFF_SYMBOL_MINUS:
 850                         hm = add_lines;
 851                         key = prepare_entry(o, n);
 852                         match = hashmap_get(hm, key, o);
 853                         free(key);
 854                         break;
 855                 default:
 856                         flipped_block = 1;
 857                 }
 858
 859                 if (!match) {
 860                         adjust_last_block(o, n, block_length);
 861                         pmb_nr = 0;
 862                         block_length = 0;
 863                         continue;
 864                 }
 865
 866                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 867
 868                 if (o->color_moved == COLOR_MOVED_PLAIN)
 869                         continue;
 870
 871                 /* Check any potential block runs, advance each or nullify */
 872                 for (i = 0; i < pmb_nr; i++) {
 873                         struct moved_entry *p = pmb[i];
 874                         struct moved_entry *pnext = (p && p->next_line) ?
 875                                         p->next_line : NULL;
 876                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 877                                 pmb[i] = p->next_line;
 878                         } else {
 879                                 pmb[i] = NULL;
 880                         }
 881                 }
 882
 883                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 884
 885                 if (pmb_nr == 0) {
 886                         /*
 887                          * The current line is the start of a new block.
 888                          * Setup the set of potential blocks.
 889                          */
 890                         for (; match; match = hashmap_get_next(hm, match)) {
 891                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 892                                 pmb[pmb_nr++] = match;
 893                         }
 894
 895                         flipped_block = (flipped_block + 1) % 2;
 896
 897                         adjust_last_block(o, n, block_length);
 898                         block_length = 0;
 899                 }
 900
 901                 block_length++;
 902
 903                 if (flipped_block)
 904                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 905         }
 906         adjust_last_block(o, n, block_length);
 907
 908         free(pmb);
 909 }
 910
 911 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
 912   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
 913 static void dim_moved_lines(struct diff_options *o)
 914 {
 915         int n;
 916         for (n = 0; n < o->emitted_symbols->nr; n++) {
 917                 struct emitted_diff_symbol *prev = (n != 0) ?
 918                                 &o->emitted_symbols->buf[n - 1] : NULL;
 919                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 920                 struct emitted_diff_symbol *next =
 921                                 (n < o->emitted_symbols->nr - 1) ?
 922                                 &o->emitted_symbols->buf[n + 1] : NULL;
 923
 924                 /* Not a plus or minus line? */
 925                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
 926                         continue;
 927
 928                 /* Not a moved line? */
 929                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
 930                         continue;
 931
 932                 /*
 933                  * If prev or next are not a plus or minus line,
 934                  * pretend they don't exist
 935                  */
 936                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
 937                             prev->s != DIFF_SYMBOL_MINUS)
 938                         prev = NULL;
 939                 if (next && next->s != DIFF_SYMBOL_PLUS &&
 940                             next->s != DIFF_SYMBOL_MINUS)
 941                         next = NULL;
 942
 943                 /* Inside a block? */
 944                 if ((prev &&
 945                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 946                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
 947                     (next &&
 948                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 949                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
 950                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 951                         continue;
 952                 }
 953
 954                 /* Check if we are at an interesting bound: */
 955                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
 956                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 957                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 958                         continue;
 959                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
 960                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 961                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 962                         continue;
 963
 964                 /*
 965                  * The boundary to prev and next are not interesting,
 966                  * so this line is not interesting as a whole
 967                  */
 968                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 969         }
 970 }
 971
 972 static void emit_line_ws_markup(struct diff_options *o,
 973                                 const char *set, const char *reset,
 974                                 const char *line, int len, char sign,
 975                                 unsigned ws_rule, int blank_at_eof)
 976 {
 977         const char *ws = NULL;
 978
 979         if (o->ws_error_highlight & ws_rule) {
 980                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 981                 if (!*ws)
 982                         ws = NULL;
 983         }
 984
 985         if (!ws)
 986                 emit_line_0(o, set, reset, sign, line, len);
 987         else if (blank_at_eof)
 988                 /* Blank line at EOF - paint '+' as well */
 989                 emit_line_0(o, ws, reset, sign, line, len);
 990         else {
 991                 /* Emit just the prefix, then the rest. */
 992                 emit_line_0(o, set, reset, sign, "", 0);
 993                 ws_check_emit(line, len, ws_rule,
 994                               o->file, set, reset, ws);
 995         }
 996 }
 997
 998 static void emit_diff_symbol_from_struct(struct diff_options *o,
 999                                          struct emitted_diff_symbol *eds)
1000 {
1001         static const char *nneof = " No newline at end of file\n";
1002         const char *context, *reset, *set, *meta, *fraginfo;
1003         struct strbuf sb = STRBUF_INIT;
1004
1005         enum diff_symbol s = eds->s;
1006         const char *line = eds->line;
1007         int len = eds->len;
1008         unsigned flags = eds->flags;
1009
1010         switch (s) {
1011         case DIFF_SYMBOL_NO_LF_EOF:
1012                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1013                 reset = diff_get_color_opt(o, DIFF_RESET);
1014                 putc('\n', o->file);
1015                 emit_line_0(o, context, reset, '\\',
1016                             nneof, strlen(nneof));
1017                 break;
1018         case DIFF_SYMBOL_SUBMODULE_HEADER:
1019         case DIFF_SYMBOL_SUBMODULE_ERROR:
1020         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1021         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1022         case DIFF_SYMBOL_SUMMARY:
1023         case DIFF_SYMBOL_STATS_LINE:
1024         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1025         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1026                 emit_line(o, "", "", line, len);
1027                 break;
1028         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1029         case DIFF_SYMBOL_CONTEXT_MARKER:
1030                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1031                 reset = diff_get_color_opt(o, DIFF_RESET);
1032                 emit_line(o, context, reset, line, len);
1033                 break;
1034         case DIFF_SYMBOL_SEPARATOR:
1035                 fprintf(o->file, "%s%c",
1036                         diff_line_prefix(o),
1037                         o->line_termination);
1038                 break;
1039         case DIFF_SYMBOL_CONTEXT:
1040                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1041                 reset = diff_get_color_opt(o, DIFF_RESET);
1042                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1043                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1044                 break;
1045         case DIFF_SYMBOL_PLUS:
1046                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1047                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1048                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1049                 case DIFF_SYMBOL_MOVED_LINE |
1050                      DIFF_SYMBOL_MOVED_LINE_ALT |
1051                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1052                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1053                         break;
1054                 case DIFF_SYMBOL_MOVED_LINE |
1055                      DIFF_SYMBOL_MOVED_LINE_ALT:
1056                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1057                         break;
1058                 case DIFF_SYMBOL_MOVED_LINE |
1059                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1060                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1061                         break;
1062                 case DIFF_SYMBOL_MOVED_LINE:
1063                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1064                         break;
1065                 default:
1066                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1067                 }
1068                 reset = diff_get_color_opt(o, DIFF_RESET);
1069                 emit_line_ws_markup(o, set, reset, line, len, '+',
1070                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1071                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1072                 break;
1073         case DIFF_SYMBOL_MINUS:
1074                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1075                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1076                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1077                 case DIFF_SYMBOL_MOVED_LINE |
1078                      DIFF_SYMBOL_MOVED_LINE_ALT |
1079                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1080                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1081                         break;
1082                 case DIFF_SYMBOL_MOVED_LINE |
1083                      DIFF_SYMBOL_MOVED_LINE_ALT:
1084                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1085                         break;
1086                 case DIFF_SYMBOL_MOVED_LINE |
1087                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1088                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1089                         break;
1090                 case DIFF_SYMBOL_MOVED_LINE:
1091                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1092                         break;
1093                 default:
1094                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1095                 }
1096                 reset = diff_get_color_opt(o, DIFF_RESET);
1097                 emit_line_ws_markup(o, set, reset, line, len, '-',
1098                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1099                 break;
1100         case DIFF_SYMBOL_WORDS_PORCELAIN:
1101                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1102                 reset = diff_get_color_opt(o, DIFF_RESET);
1103                 emit_line(o, context, reset, line, len);
1104                 fputs("~\n", o->file);
1105                 break;
1106         case DIFF_SYMBOL_WORDS:
1107                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1108                 reset = diff_get_color_opt(o, DIFF_RESET);
1109                 /*
1110                  * Skip the prefix character, if any.  With
1111                  * diff_suppress_blank_empty, there may be
1112                  * none.
1113                  */
1114                 if (line[0] != '\n') {
1115                         line++;
1116                         len--;
1117                 }
1118                 emit_line(o, context, reset, line, len);
1119                 break;
1120         case DIFF_SYMBOL_FILEPAIR_PLUS:
1121                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1122                 reset = diff_get_color_opt(o, DIFF_RESET);
1123                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1124                         line, reset,
1125                         strchr(line, ' ') ? "\t" : "");
1126                 break;
1127         case DIFF_SYMBOL_FILEPAIR_MINUS:
1128                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1129                 reset = diff_get_color_opt(o, DIFF_RESET);
1130                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1131                         line, reset,
1132                         strchr(line, ' ') ? "\t" : "");
1133                 break;
1134         case DIFF_SYMBOL_BINARY_FILES:
1135         case DIFF_SYMBOL_HEADER:
1136                 fprintf(o->file, "%s", line);
1137                 break;
1138         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1139                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1140                 break;
1141         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1142                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1143                 break;
1144         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1145                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1146                 break;
1147         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1148                 fputs(diff_line_prefix(o), o->file);
1149                 fputc('\n', o->file);
1150                 break;
1151         case DIFF_SYMBOL_REWRITE_DIFF:
1152                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1153                 reset = diff_get_color_opt(o, DIFF_RESET);
1154                 emit_line(o, fraginfo, reset, line, len);
1155                 break;
1156         case DIFF_SYMBOL_SUBMODULE_ADD:
1157                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1158                 reset = diff_get_color_opt(o, DIFF_RESET);
1159                 emit_line(o, set, reset, line, len);
1160                 break;
1161         case DIFF_SYMBOL_SUBMODULE_DEL:
1162                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1163                 reset = diff_get_color_opt(o, DIFF_RESET);
1164                 emit_line(o, set, reset, line, len);
1165                 break;
1166         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1167                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1168                         diff_line_prefix(o), line);
1169                 break;
1170         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1171                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1172                         diff_line_prefix(o), line);
1173                 break;
1174         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1175                 emit_line(o, "", "", " 0 files changed\n",
1176                           strlen(" 0 files changed\n"));
1177                 break;
1178         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1179                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1180                 break;
1181         case DIFF_SYMBOL_WORD_DIFF:
1182                 fprintf(o->file, "%.*s", len, line);
1183                 break;
1184         case DIFF_SYMBOL_STAT_SEP:
1185                 fputs(o->stat_sep, o->file);
1186                 break;
1187         default:
1188                 die("BUG: unknown diff symbol");
1189         }
1190         strbuf_release(&sb);
1191 }
1192
1193 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1194                              const char *line, int len, unsigned flags)
1195 {
1196         struct emitted_diff_symbol e = {line, len, flags, s};
1197
1198         if (o->emitted_symbols)
1199                 append_emitted_diff_symbol(o, &e);
1200         else
1201                 emit_diff_symbol_from_struct(o, &e);
1202 }
1203
1204 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1205 {
1206         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1207 }
1208
1209 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1210 {
1211         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1212 }
1213
1214 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1215 {
1216         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1217                          path, strlen(path), 0);
1218 }
1219
1220 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1221 {
1222         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1223                          path, strlen(path), 0);
1224 }
1225
1226 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1227 {
1228         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1229                          header, strlen(header), 0);
1230 }
1231
1232 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1233 {
1234         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1235 }
1236
1237 void diff_emit_submodule_pipethrough(struct diff_options *o,
1238                                      const char *line, int len)
1239 {
1240         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1241 }
1242
1243 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1244 {
1245         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1246               ecbdata->blank_at_eof_in_preimage &&
1247               ecbdata->blank_at_eof_in_postimage &&
1248               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1249               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1250                 return 0;
1251         return ws_blank_line(line, len, ecbdata->ws_rule);
1252 }
1253
1254 static void emit_add_line(const char *reset,
1255                           struct emit_callback *ecbdata,
1256                           const char *line, int len)
1257 {
1258         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1259         if (new_blank_line_at_eof(ecbdata, line, len))
1260                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1261
1262         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1263 }
1264
1265 static void emit_del_line(const char *reset,
1266                           struct emit_callback *ecbdata,
1267                           const char *line, int len)
1268 {
1269         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1270         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1271 }
1272
1273 static void emit_context_line(const char *reset,
1274                               struct emit_callback *ecbdata,
1275                               const char *line, int len)
1276 {
1277         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1278         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1279 }
1280
1281 static void emit_hunk_header(struct emit_callback *ecbdata,
1282                              const char *line, int len)
1283 {
1284         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1285         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1286         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1287         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1288         static const char atat[2] = { '@', '@' };
1289         const char *cp, *ep;
1290         struct strbuf msgbuf = STRBUF_INIT;
1291         int org_len = len;
1292         int i = 1;
1293
1294         /*
1295          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1296          * it always is at least 10 bytes long.
1297          */
1298         if (len < 10 ||
1299             memcmp(line, atat, 2) ||
1300             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1301                 emit_diff_symbol(ecbdata->opt,
1302                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1303                 return;
1304         }
1305         ep += 2; /* skip over @@ */
1306
1307         /* The hunk header in fraginfo color */
1308         strbuf_addstr(&msgbuf, frag);
1309         strbuf_add(&msgbuf, line, ep - line);
1310         strbuf_addstr(&msgbuf, reset);
1311
1312         /*
1313          * trailing "\r\n"
1314          */
1315         for ( ; i < 3; i++)
1316                 if (line[len - i] == '\r' || line[len - i] == '\n')
1317                         len--;
1318
1319         /* blank before the func header */
1320         for (cp = ep; ep - line < len; ep++)
1321                 if (*ep != ' ' && *ep != '\t')
1322                         break;
1323         if (ep != cp) {
1324                 strbuf_addstr(&msgbuf, context);
1325                 strbuf_add(&msgbuf, cp, ep - cp);
1326                 strbuf_addstr(&msgbuf, reset);
1327         }
1328
1329         if (ep < line + len) {
1330                 strbuf_addstr(&msgbuf, func);
1331                 strbuf_add(&msgbuf, ep, line + len - ep);
1332                 strbuf_addstr(&msgbuf, reset);
1333         }
1334
1335         strbuf_add(&msgbuf, line + len, org_len - len);
1336         strbuf_complete_line(&msgbuf);
1337         emit_diff_symbol(ecbdata->opt,
1338                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1339         strbuf_release(&msgbuf);
1340 }
1341
1342 static struct diff_tempfile *claim_diff_tempfile(void) {
1343         int i;
1344         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1345                 if (!diff_temp[i].name)
1346                         return diff_temp + i;
1347         die("BUG: diff is failing to clean up its tempfiles");
1348 }
1349
1350 static void remove_tempfile(void)
1351 {
1352         int i;
1353         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1354                 if (is_tempfile_active(diff_temp[i].tempfile))
1355                         delete_tempfile(&diff_temp[i].tempfile);
1356                 diff_temp[i].name = NULL;
1357         }
1358 }
1359
1360 static void add_line_count(struct strbuf *out, int count)
1361 {
1362         switch (count) {
1363         case 0:
1364                 strbuf_addstr(out, "0,0");
1365                 break;
1366         case 1:
1367                 strbuf_addstr(out, "1");
1368                 break;
1369         default:
1370                 strbuf_addf(out, "1,%d", count);
1371                 break;
1372         }
1373 }
1374
1375 static void emit_rewrite_lines(struct emit_callback *ecb,
1376                                int prefix, const char *data, int size)
1377 {
1378         const char *endp = NULL;
1379         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1380
1381         while (0 < size) {
1382                 int len;
1383
1384                 endp = memchr(data, '\n', size);
1385                 len = endp ? (endp - data + 1) : size;
1386                 if (prefix != '+') {
1387                         ecb->lno_in_preimage++;
1388                         emit_del_line(reset, ecb, data, len);
1389                 } else {
1390                         ecb->lno_in_postimage++;
1391                         emit_add_line(reset, ecb, data, len);
1392                 }
1393                 size -= len;
1394                 data += len;
1395         }
1396         if (!endp)
1397                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1398 }
1399
1400 static void emit_rewrite_diff(const char *name_a,
1401                               const char *name_b,
1402                               struct diff_filespec *one,
1403                               struct diff_filespec *two,
1404                               struct userdiff_driver *textconv_one,
1405                               struct userdiff_driver *textconv_two,
1406                               struct diff_options *o)
1407 {
1408         int lc_a, lc_b;
1409         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1410         const char *a_prefix, *b_prefix;
1411         char *data_one, *data_two;
1412         size_t size_one, size_two;
1413         struct emit_callback ecbdata;
1414         struct strbuf out = STRBUF_INIT;
1415
1416         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1417                 a_prefix = o->b_prefix;
1418                 b_prefix = o->a_prefix;
1419         } else {
1420                 a_prefix = o->a_prefix;
1421                 b_prefix = o->b_prefix;
1422         }
1423
1424         name_a += (*name_a == '/');
1425         name_b += (*name_b == '/');
1426
1427         strbuf_reset(&a_name);
1428         strbuf_reset(&b_name);
1429         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1430         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1431
1432         size_one = fill_textconv(textconv_one, one, &data_one);
1433         size_two = fill_textconv(textconv_two, two, &data_two);
1434
1435         memset(&ecbdata, 0, sizeof(ecbdata));
1436         ecbdata.color_diff = want_color(o->use_color);
1437         ecbdata.ws_rule = whitespace_rule(name_b);
1438         ecbdata.opt = o;
1439         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1440                 mmfile_t mf1, mf2;
1441                 mf1.ptr = (char *)data_one;
1442                 mf2.ptr = (char *)data_two;
1443                 mf1.size = size_one;
1444                 mf2.size = size_two;
1445                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1446         }
1447         ecbdata.lno_in_preimage = 1;
1448         ecbdata.lno_in_postimage = 1;
1449
1450         lc_a = count_lines(data_one, size_one);
1451         lc_b = count_lines(data_two, size_two);
1452
1453         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1454                          a_name.buf, a_name.len, 0);
1455         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1456                          b_name.buf, b_name.len, 0);
1457
1458         strbuf_addstr(&out, "@@ -");
1459         if (!o->irreversible_delete)
1460                 add_line_count(&out, lc_a);
1461         else
1462                 strbuf_addstr(&out, "?,?");
1463         strbuf_addstr(&out, " +");
1464         add_line_count(&out, lc_b);
1465         strbuf_addstr(&out, " @@\n");
1466         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1467         strbuf_release(&out);
1468
1469         if (lc_a && !o->irreversible_delete)
1470                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1471         if (lc_b)
1472                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1473         if (textconv_one)
1474                 free((char *)data_one);
1475         if (textconv_two)
1476                 free((char *)data_two);
1477 }
1478
1479 struct diff_words_buffer {
1480         mmfile_t text;
1481         unsigned long alloc;
1482         struct diff_words_orig {
1483                 const char *begin, *end;
1484         } *orig;
1485         int orig_nr, orig_alloc;
1486 };
1487
1488 static void diff_words_append(char *line, unsigned long len,
1489                 struct diff_words_buffer *buffer)
1490 {
1491         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1492         line++;
1493         len--;
1494         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1495         buffer->text.size += len;
1496         buffer->text.ptr[buffer->text.size] = '\0';
1497 }
1498
1499 struct diff_words_style_elem {
1500         const char *prefix;
1501         const char *suffix;
1502         const char *color; /* NULL; filled in by the setup code if
1503                             * color is enabled */
1504 };
1505
1506 struct diff_words_style {
1507         enum diff_words_type type;
1508         struct diff_words_style_elem new_word, old_word, ctx;
1509         const char *newline;
1510 };
1511
1512 static struct diff_words_style diff_words_styles[] = {
1513         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1514         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1515         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1516 };
1517
1518 struct diff_words_data {
1519         struct diff_words_buffer minus, plus;
1520         const char *current_plus;
1521         int last_minus;
1522         struct diff_options *opt;
1523         regex_t *word_regex;
1524         enum diff_words_type type;
1525         struct diff_words_style *style;
1526 };
1527
1528 static int fn_out_diff_words_write_helper(struct diff_options *o,
1529                                           struct diff_words_style_elem *st_el,
1530                                           const char *newline,
1531                                           size_t count, const char *buf)
1532 {
1533         int print = 0;
1534         struct strbuf sb = STRBUF_INIT;
1535
1536         while (count) {
1537                 char *p = memchr(buf, '\n', count);
1538                 if (print)
1539                         strbuf_addstr(&sb, diff_line_prefix(o));
1540
1541                 if (p != buf) {
1542                         const char *reset = st_el->color && *st_el->color ?
1543                                             GIT_COLOR_RESET : NULL;
1544                         if (st_el->color && *st_el->color)
1545                                 strbuf_addstr(&sb, st_el->color);
1546                         strbuf_addstr(&sb, st_el->prefix);
1547                         strbuf_add(&sb, buf, p ? p - buf : count);
1548                         strbuf_addstr(&sb, st_el->suffix);
1549                         if (reset)
1550                                 strbuf_addstr(&sb, reset);
1551                 }
1552                 if (!p)
1553                         goto out;
1554
1555                 strbuf_addstr(&sb, newline);
1556                 count -= p + 1 - buf;
1557                 buf = p + 1;
1558                 print = 1;
1559                 if (count) {
1560                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1561                                          sb.buf, sb.len, 0);
1562                         strbuf_reset(&sb);
1563                 }
1564         }
1565
1566 out:
1567         if (sb.len)
1568                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1569                                  sb.buf, sb.len, 0);
1570         strbuf_release(&sb);
1571         return 0;
1572 }
1573
1574 /*
1575  * '--color-words' algorithm can be described as:
1576  *
1577  *   1. collect the minus/plus lines of a diff hunk, divided into
1578  *      minus-lines and plus-lines;
1579  *
1580  *   2. break both minus-lines and plus-lines into words and
1581  *      place them into two mmfile_t with one word for each line;
1582  *
1583  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1584  *
1585  * And for the common parts of the both file, we output the plus side text.
1586  * diff_words->current_plus is used to trace the current position of the plus file
1587  * which printed. diff_words->last_minus is used to trace the last minus word
1588  * printed.
1589  *
1590  * For '--graph' to work with '--color-words', we need to output the graph prefix
1591  * on each line of color words output. Generally, there are two conditions on
1592  * which we should output the prefix.
1593  *
1594  *   1. diff_words->last_minus == 0 &&
1595  *      diff_words->current_plus == diff_words->plus.text.ptr
1596  *
1597  *      that is: the plus text must start as a new line, and if there is no minus
1598  *      word printed, a graph prefix must be printed.
1599  *
1600  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1601  *      *(diff_words->current_plus - 1) == '\n'
1602  *
1603  *      that is: a graph prefix must be printed following a '\n'
1604  */
1605 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1606 {
1607         if ((diff_words->last_minus == 0 &&
1608                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1609                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1610                 *(diff_words->current_plus - 1) == '\n')) {
1611                 return 1;
1612         } else {
1613                 return 0;
1614         }
1615 }
1616
1617 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1618 {
1619         struct diff_words_data *diff_words = priv;
1620         struct diff_words_style *style = diff_words->style;
1621         int minus_first, minus_len, plus_first, plus_len;
1622         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1623         struct diff_options *opt = diff_words->opt;
1624         const char *line_prefix;
1625
1626         if (line[0] != '@' || parse_hunk_header(line, len,
1627                         &minus_first, &minus_len, &plus_first, &plus_len))
1628                 return;
1629
1630         assert(opt);
1631         line_prefix = diff_line_prefix(opt);
1632
1633         /* POSIX requires that first be decremented by one if len == 0... */
1634         if (minus_len) {
1635                 minus_begin = diff_words->minus.orig[minus_first].begin;
1636                 minus_end =
1637                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1638         } else
1639                 minus_begin = minus_end =
1640                         diff_words->minus.orig[minus_first].end;
1641
1642         if (plus_len) {
1643                 plus_begin = diff_words->plus.orig[plus_first].begin;
1644                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1645         } else
1646                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1647
1648         if (color_words_output_graph_prefix(diff_words)) {
1649                 fputs(line_prefix, diff_words->opt->file);
1650         }
1651         if (diff_words->current_plus != plus_begin) {
1652                 fn_out_diff_words_write_helper(diff_words->opt,
1653                                 &style->ctx, style->newline,
1654                                 plus_begin - diff_words->current_plus,
1655                                 diff_words->current_plus);
1656         }
1657         if (minus_begin != minus_end) {
1658                 fn_out_diff_words_write_helper(diff_words->opt,
1659                                 &style->old_word, style->newline,
1660                                 minus_end - minus_begin, minus_begin);
1661         }
1662         if (plus_begin != plus_end) {
1663                 fn_out_diff_words_write_helper(diff_words->opt,
1664                                 &style->new_word, style->newline,
1665                                 plus_end - plus_begin, plus_begin);
1666         }
1667
1668         diff_words->current_plus = plus_end;
1669         diff_words->last_minus = minus_first;
1670 }
1671
1672 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1673 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1674                 int *begin, int *end)
1675 {
1676         if (word_regex && *begin < buffer->size) {
1677                 regmatch_t match[1];
1678                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1679                                  buffer->size - *begin, 1, match, 0)) {
1680                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1681                                         '\n', match[0].rm_eo - match[0].rm_so);
1682                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1683                         *begin += match[0].rm_so;
1684                         return *begin >= *end;
1685                 }
1686                 return -1;
1687         }
1688
1689         /* find the next word */
1690         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1691                 (*begin)++;
1692         if (*begin >= buffer->size)
1693                 return -1;
1694
1695         /* find the end of the word */
1696         *end = *begin + 1;
1697         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1698                 (*end)++;
1699
1700         return 0;
1701 }
1702
1703 /*
1704  * This function splits the words in buffer->text, stores the list with
1705  * newline separator into out, and saves the offsets of the original words
1706  * in buffer->orig.
1707  */
1708 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1709                 regex_t *word_regex)
1710 {
1711         int i, j;
1712         long alloc = 0;
1713
1714         out->size = 0;
1715         out->ptr = NULL;
1716
1717         /* fake an empty "0th" word */
1718         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1719         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1720         buffer->orig_nr = 1;
1721
1722         for (i = 0; i < buffer->text.size; i++) {
1723                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1724                         return;
1725
1726                 /* store original boundaries */
1727                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1728                                 buffer->orig_alloc);
1729                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1730                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1731                 buffer->orig_nr++;
1732
1733                 /* store one word */
1734                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1735                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1736                 out->ptr[out->size + j - i] = '\n';
1737                 out->size += j - i + 1;
1738
1739                 i = j - 1;
1740         }
1741 }
1742
1743 /* this executes the word diff on the accumulated buffers */
1744 static void diff_words_show(struct diff_words_data *diff_words)
1745 {
1746         xpparam_t xpp;
1747         xdemitconf_t xecfg;
1748         mmfile_t minus, plus;
1749         struct diff_words_style *style = diff_words->style;
1750
1751         struct diff_options *opt = diff_words->opt;
1752         const char *line_prefix;
1753
1754         assert(opt);
1755         line_prefix = diff_line_prefix(opt);
1756
1757         /* special case: only removal */
1758         if (!diff_words->plus.text.size) {
1759                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1760                                  line_prefix, strlen(line_prefix), 0);
1761                 fn_out_diff_words_write_helper(diff_words->opt,
1762                         &style->old_word, style->newline,
1763                         diff_words->minus.text.size,
1764                         diff_words->minus.text.ptr);
1765                 diff_words->minus.text.size = 0;
1766                 return;
1767         }
1768
1769         diff_words->current_plus = diff_words->plus.text.ptr;
1770         diff_words->last_minus = 0;
1771
1772         memset(&xpp, 0, sizeof(xpp));
1773         memset(&xecfg, 0, sizeof(xecfg));
1774         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1775         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1776         xpp.flags = 0;
1777         /* as only the hunk header will be parsed, we need a 0-context */
1778         xecfg.ctxlen = 0;
1779         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1780                           &xpp, &xecfg))
1781                 die("unable to generate word diff");
1782         free(minus.ptr);
1783         free(plus.ptr);
1784         if (diff_words->current_plus != diff_words->plus.text.ptr +
1785                         diff_words->plus.text.size) {
1786                 if (color_words_output_graph_prefix(diff_words))
1787                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1788                                          line_prefix, strlen(line_prefix), 0);
1789                 fn_out_diff_words_write_helper(diff_words->opt,
1790                         &style->ctx, style->newline,
1791                         diff_words->plus.text.ptr + diff_words->plus.text.size
1792                         - diff_words->current_plus, diff_words->current_plus);
1793         }
1794         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1795 }
1796
1797 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1798 static void diff_words_flush(struct emit_callback *ecbdata)
1799 {
1800         struct diff_options *wo = ecbdata->diff_words->opt;
1801
1802         if (ecbdata->diff_words->minus.text.size ||
1803             ecbdata->diff_words->plus.text.size)
1804                 diff_words_show(ecbdata->diff_words);
1805
1806         if (wo->emitted_symbols) {
1807                 struct diff_options *o = ecbdata->opt;
1808                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1809                 int i;
1810
1811                 /*
1812                  * NEEDSWORK:
1813                  * Instead of appending each, concat all words to a line?
1814                  */
1815                 for (i = 0; i < wol->nr; i++)
1816                         append_emitted_diff_symbol(o, &wol->buf[i]);
1817
1818                 for (i = 0; i < wol->nr; i++)
1819                         free((void *)wol->buf[i].line);
1820
1821                 wol->nr = 0;
1822         }
1823 }
1824
1825 static void diff_filespec_load_driver(struct diff_filespec *one)
1826 {
1827         /* Use already-loaded driver */
1828         if (one->driver)
1829                 return;
1830
1831         if (S_ISREG(one->mode))
1832                 one->driver = userdiff_find_by_path(one->path);
1833
1834         /* Fallback to default settings */
1835         if (!one->driver)
1836                 one->driver = userdiff_find_by_name("default");
1837 }
1838
1839 static const char *userdiff_word_regex(struct diff_filespec *one)
1840 {
1841         diff_filespec_load_driver(one);
1842         return one->driver->word_regex;
1843 }
1844
1845 static void init_diff_words_data(struct emit_callback *ecbdata,
1846                                  struct diff_options *orig_opts,
1847                                  struct diff_filespec *one,
1848                                  struct diff_filespec *two)
1849 {
1850         int i;
1851         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1852         memcpy(o, orig_opts, sizeof(struct diff_options));
1853
1854         ecbdata->diff_words =
1855                 xcalloc(1, sizeof(struct diff_words_data));
1856         ecbdata->diff_words->type = o->word_diff;
1857         ecbdata->diff_words->opt = o;
1858
1859         if (orig_opts->emitted_symbols)
1860                 o->emitted_symbols =
1861                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1862
1863         if (!o->word_regex)
1864                 o->word_regex = userdiff_word_regex(one);
1865         if (!o->word_regex)
1866                 o->word_regex = userdiff_word_regex(two);
1867         if (!o->word_regex)
1868                 o->word_regex = diff_word_regex_cfg;
1869         if (o->word_regex) {
1870                 ecbdata->diff_words->word_regex = (regex_t *)
1871                         xmalloc(sizeof(regex_t));
1872                 if (regcomp(ecbdata->diff_words->word_regex,
1873                             o->word_regex,
1874                             REG_EXTENDED | REG_NEWLINE))
1875                         die ("Invalid regular expression: %s",
1876                              o->word_regex);
1877         }
1878         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1879                 if (o->word_diff == diff_words_styles[i].type) {
1880                         ecbdata->diff_words->style =
1881                                 &diff_words_styles[i];
1882                         break;
1883                 }
1884         }
1885         if (want_color(o->use_color)) {
1886                 struct diff_words_style *st = ecbdata->diff_words->style;
1887                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1888                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1889                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1890         }
1891 }
1892
1893 static void free_diff_words_data(struct emit_callback *ecbdata)
1894 {
1895         if (ecbdata->diff_words) {
1896                 diff_words_flush(ecbdata);
1897                 free (ecbdata->diff_words->opt->emitted_symbols);
1898                 free (ecbdata->diff_words->opt);
1899                 free (ecbdata->diff_words->minus.text.ptr);
1900                 free (ecbdata->diff_words->minus.orig);
1901                 free (ecbdata->diff_words->plus.text.ptr);
1902                 free (ecbdata->diff_words->plus.orig);
1903                 if (ecbdata->diff_words->word_regex) {
1904                         regfree(ecbdata->diff_words->word_regex);
1905                         free(ecbdata->diff_words->word_regex);
1906                 }
1907                 FREE_AND_NULL(ecbdata->diff_words);
1908         }
1909 }
1910
1911 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1912 {
1913         if (want_color(diff_use_color))
1914                 return diff_colors[ix];
1915         return "";
1916 }
1917
1918 const char *diff_line_prefix(struct diff_options *opt)
1919 {
1920         struct strbuf *msgbuf;
1921         if (!opt->output_prefix)
1922                 return "";
1923
1924         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1925         return msgbuf->buf;
1926 }
1927
1928 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1929 {
1930         const char *cp;
1931         unsigned long allot;
1932         size_t l = len;
1933
1934         cp = line;
1935         allot = l;
1936         while (0 < l) {
1937                 (void) utf8_width(&cp, &l);
1938                 if (!cp)
1939                         break; /* truncated in the middle? */
1940         }
1941         return allot - l;
1942 }
1943
1944 static void find_lno(const char *line, struct emit_callback *ecbdata)
1945 {
1946         const char *p;
1947         ecbdata->lno_in_preimage = 0;
1948         ecbdata->lno_in_postimage = 0;
1949         p = strchr(line, '-');
1950         if (!p)
1951                 return; /* cannot happen */
1952         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1953         p = strchr(p, '+');
1954         if (!p)
1955                 return; /* cannot happen */
1956         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1957 }
1958
1959 static void fn_out_consume(void *priv, char *line, unsigned long len)
1960 {
1961         struct emit_callback *ecbdata = priv;
1962         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1963         struct diff_options *o = ecbdata->opt;
1964
1965         o->found_changes = 1;
1966
1967         if (ecbdata->header) {
1968                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1969                                  ecbdata->header->buf, ecbdata->header->len, 0);
1970                 strbuf_reset(ecbdata->header);
1971                 ecbdata->header = NULL;
1972         }
1973
1974         if (ecbdata->label_path[0]) {
1975                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1976                                  ecbdata->label_path[0],
1977                                  strlen(ecbdata->label_path[0]), 0);
1978                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1979                                  ecbdata->label_path[1],
1980                                  strlen(ecbdata->label_path[1]), 0);
1981                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1982         }
1983
1984         if (diff_suppress_blank_empty
1985             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1986                 line[0] = '\n';
1987                 len = 1;
1988         }
1989
1990         if (line[0] == '@') {
1991                 if (ecbdata->diff_words)
1992                         diff_words_flush(ecbdata);
1993                 len = sane_truncate_line(ecbdata, line, len);
1994                 find_lno(line, ecbdata);
1995                 emit_hunk_header(ecbdata, line, len);
1996                 return;
1997         }
1998
1999         if (ecbdata->diff_words) {
2000                 enum diff_symbol s =
2001                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2002                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2003                 if (line[0] == '-') {
2004                         diff_words_append(line, len,
2005                                           &ecbdata->diff_words->minus);
2006                         return;
2007                 } else if (line[0] == '+') {
2008                         diff_words_append(line, len,
2009                                           &ecbdata->diff_words->plus);
2010                         return;
2011                 } else if (starts_with(line, "\\ ")) {
2012                         /*
2013                          * Eat the "no newline at eof" marker as if we
2014                          * saw a "+" or "-" line with nothing on it,
2015                          * and return without diff_words_flush() to
2016                          * defer processing. If this is the end of
2017                          * preimage, more "+" lines may come after it.
2018                          */
2019                         return;
2020                 }
2021                 diff_words_flush(ecbdata);
2022                 emit_diff_symbol(o, s, line, len, 0);
2023                 return;
2024         }
2025
2026         switch (line[0]) {
2027         case '+':
2028                 ecbdata->lno_in_postimage++;
2029                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2030                 break;
2031         case '-':
2032                 ecbdata->lno_in_preimage++;
2033                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2034                 break;
2035         case ' ':
2036                 ecbdata->lno_in_postimage++;
2037                 ecbdata->lno_in_preimage++;
2038                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2039                 break;
2040         default:
2041                 /* incomplete line at the end */
2042                 ecbdata->lno_in_preimage++;
2043                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2044                                  line, len, 0);
2045                 break;
2046         }
2047 }
2048
2049 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2050 {
2051         const char *old_name = a;
2052         const char *new_name = b;
2053         int pfx_length, sfx_length;
2054         int pfx_adjust_for_slash;
2055         int len_a = strlen(a);
2056         int len_b = strlen(b);
2057         int a_midlen, b_midlen;
2058         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2059         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2060
2061         if (qlen_a || qlen_b) {
2062                 quote_c_style(a, name, NULL, 0);
2063                 strbuf_addstr(name, " => ");
2064                 quote_c_style(b, name, NULL, 0);
2065                 return;
2066         }
2067
2068         /* Find common prefix */
2069         pfx_length = 0;
2070         while (*old_name && *new_name && *old_name == *new_name) {
2071                 if (*old_name == '/')
2072                         pfx_length = old_name - a + 1;
2073                 old_name++;
2074                 new_name++;
2075         }
2076
2077         /* Find common suffix */
2078         old_name = a + len_a;
2079         new_name = b + len_b;
2080         sfx_length = 0;
2081         /*
2082          * If there is a common prefix, it must end in a slash.  In
2083          * that case we let this loop run 1 into the prefix to see the
2084          * same slash.
2085          *
2086          * If there is no common prefix, we cannot do this as it would
2087          * underrun the input strings.
2088          */
2089         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2090         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2091                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2092                *old_name == *new_name) {
2093                 if (*old_name == '/')
2094                         sfx_length = len_a - (old_name - a);
2095                 old_name--;
2096                 new_name--;
2097         }
2098
2099         /*
2100          * pfx{mid-a => mid-b}sfx
2101          * {pfx-a => pfx-b}sfx
2102          * pfx{sfx-a => sfx-b}
2103          * name-a => name-b
2104          */
2105         a_midlen = len_a - pfx_length - sfx_length;
2106         b_midlen = len_b - pfx_length - sfx_length;
2107         if (a_midlen < 0)
2108                 a_midlen = 0;
2109         if (b_midlen < 0)
2110                 b_midlen = 0;
2111
2112         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2113         if (pfx_length + sfx_length) {
2114                 strbuf_add(name, a, pfx_length);
2115                 strbuf_addch(name, '{');
2116         }
2117         strbuf_add(name, a + pfx_length, a_midlen);
2118         strbuf_addstr(name, " => ");
2119         strbuf_add(name, b + pfx_length, b_midlen);
2120         if (pfx_length + sfx_length) {
2121                 strbuf_addch(name, '}');
2122                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2123         }
2124 }
2125
2126 struct diffstat_t {
2127         int nr;
2128         int alloc;
2129         struct diffstat_file {
2130                 char *from_name;
2131                 char *name;
2132                 char *print_name;
2133                 const char *comments;
2134                 unsigned is_unmerged:1;
2135                 unsigned is_binary:1;
2136                 unsigned is_renamed:1;
2137                 unsigned is_interesting:1;
2138                 uintmax_t added, deleted;
2139         } **files;
2140 };
2141
2142 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2143                                           const char *name_a,
2144                                           const char *name_b)
2145 {
2146         struct diffstat_file *x;
2147         x = xcalloc(1, sizeof(*x));
2148         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2149         diffstat->files[diffstat->nr++] = x;
2150         if (name_b) {
2151                 x->from_name = xstrdup(name_a);
2152                 x->name = xstrdup(name_b);
2153                 x->is_renamed = 1;
2154         }
2155         else {
2156                 x->from_name = NULL;
2157                 x->name = xstrdup(name_a);
2158         }
2159         return x;
2160 }
2161
2162 static void diffstat_consume(void *priv, char *line, unsigned long len)
2163 {
2164         struct diffstat_t *diffstat = priv;
2165         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2166
2167         if (line[0] == '+')
2168                 x->added++;
2169         else if (line[0] == '-')
2170                 x->deleted++;
2171 }
2172
2173 const char mime_boundary_leader[] = "------------";
2174
2175 static int scale_linear(int it, int width, int max_change)
2176 {
2177         if (!it)
2178                 return 0;
2179         /*
2180          * make sure that at least one '-' or '+' is printed if
2181          * there is any change to this path. The easiest way is to
2182          * scale linearly as if the alloted width is one column shorter
2183          * than it is, and then add 1 to the result.
2184          */
2185         return 1 + (it * (width - 1) / max_change);
2186 }
2187
2188 static void show_graph(struct strbuf *out, char ch, int cnt,
2189                        const char *set, const char *reset)
2190 {
2191         if (cnt <= 0)
2192                 return;
2193         strbuf_addstr(out, set);
2194         strbuf_addchars(out, ch, cnt);
2195         strbuf_addstr(out, reset);
2196 }
2197
2198 static void fill_print_name(struct diffstat_file *file)
2199 {
2200         struct strbuf pname = STRBUF_INIT;
2201
2202         if (file->print_name)
2203                 return;
2204
2205         if (file->is_renamed)
2206                 pprint_rename(&pname, file->from_name, file->name);
2207         else
2208                 quote_c_style(file->name, &pname, NULL, 0);
2209
2210         if (file->comments)
2211                 strbuf_addf(&pname, " (%s)", file->comments);
2212
2213         file->print_name = strbuf_detach(&pname, NULL);
2214 }
2215
2216 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2217                 int files, int insertions, int deletions)
2218 {
2219         struct strbuf sb = STRBUF_INIT;
2220
2221         if (!files) {
2222                 assert(insertions == 0 && deletions == 0);
2223                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2224                                  NULL, 0, 0);
2225                 return;
2226         }
2227
2228         strbuf_addf(&sb,
2229                     (files == 1) ? " %d file changed" : " %d files changed",
2230                     files);
2231
2232         /*
2233          * For binary diff, the caller may want to print "x files
2234          * changed" with insertions == 0 && deletions == 0.
2235          *
2236          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2237          * is probably less confusing (i.e skip over "2 files changed
2238          * but nothing about added/removed lines? Is this a bug in Git?").
2239          */
2240         if (insertions || deletions == 0) {
2241                 strbuf_addf(&sb,
2242                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2243                             insertions);
2244         }
2245
2246         if (deletions || insertions == 0) {
2247                 strbuf_addf(&sb,
2248                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2249                             deletions);
2250         }
2251         strbuf_addch(&sb, '\n');
2252         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2253                          sb.buf, sb.len, 0);
2254         strbuf_release(&sb);
2255 }
2256
2257 void print_stat_summary(FILE *fp, int files,
2258                         int insertions, int deletions)
2259 {
2260         struct diff_options o;
2261         memset(&o, 0, sizeof(o));
2262         o.file = fp;
2263
2264         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2265 }
2266
2267 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2268 {
2269         int i, len, add, del, adds = 0, dels = 0;
2270         uintmax_t max_change = 0, max_len = 0;
2271         int total_files = data->nr, count;
2272         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2273         const char *reset, *add_c, *del_c;
2274         int extra_shown = 0;
2275         const char *line_prefix = diff_line_prefix(options);
2276         struct strbuf out = STRBUF_INIT;
2277
2278         if (data->nr == 0)
2279                 return;
2280
2281         count = options->stat_count ? options->stat_count : data->nr;
2282
2283         reset = diff_get_color_opt(options, DIFF_RESET);
2284         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2285         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2286
2287         /*
2288          * Find the longest filename and max number of changes
2289          */
2290         for (i = 0; (i < count) && (i < data->nr); i++) {
2291                 struct diffstat_file *file = data->files[i];
2292                 uintmax_t change = file->added + file->deleted;
2293
2294                 if (!file->is_interesting && (change == 0)) {
2295                         count++; /* not shown == room for one more */
2296                         continue;
2297                 }
2298                 fill_print_name(file);
2299                 len = strlen(file->print_name);
2300                 if (max_len < len)
2301                         max_len = len;
2302
2303                 if (file->is_unmerged) {
2304                         /* "Unmerged" is 8 characters */
2305                         bin_width = bin_width < 8 ? 8 : bin_width;
2306                         continue;
2307                 }
2308                 if (file->is_binary) {
2309                         /* "Bin XXX -> YYY bytes" */
2310                         int w = 14 + decimal_width(file->added)
2311                                 + decimal_width(file->deleted);
2312                         bin_width = bin_width < w ? w : bin_width;
2313                         /* Display change counts aligned with "Bin" */
2314                         number_width = 3;
2315                         continue;
2316                 }
2317
2318                 if (max_change < change)
2319                         max_change = change;
2320         }
2321         count = i; /* where we can stop scanning in data->files[] */
2322
2323         /*
2324          * We have width = stat_width or term_columns() columns total.
2325          * We want a maximum of min(max_len, stat_name_width) for the name part.
2326          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2327          * We also need 1 for " " and 4 + decimal_width(max_change)
2328          * for " | NNNN " and one the empty column at the end, altogether
2329          * 6 + decimal_width(max_change).
2330          *
2331          * If there's not enough space, we will use the smaller of
2332          * stat_name_width (if set) and 5/8*width for the filename,
2333          * and the rest for constant elements + graph part, but no more
2334          * than stat_graph_width for the graph part.
2335          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2336          * for the standard terminal size).
2337          *
2338          * In other words: stat_width limits the maximum width, and
2339          * stat_name_width fixes the maximum width of the filename,
2340          * and is also used to divide available columns if there
2341          * aren't enough.
2342          *
2343          * Binary files are displayed with "Bin XXX -> YYY bytes"
2344          * instead of the change count and graph. This part is treated
2345          * similarly to the graph part, except that it is not
2346          * "scaled". If total width is too small to accommodate the
2347          * guaranteed minimum width of the filename part and the
2348          * separators and this message, this message will "overflow"
2349          * making the line longer than the maximum width.
2350          */
2351
2352         if (options->stat_width == -1)
2353                 width = term_columns() - strlen(line_prefix);
2354         else
2355                 width = options->stat_width ? options->stat_width : 80;
2356         number_width = decimal_width(max_change) > number_width ?
2357                 decimal_width(max_change) : number_width;
2358
2359         if (options->stat_graph_width == -1)
2360                 options->stat_graph_width = diff_stat_graph_width;
2361
2362         /*
2363          * Guarantee 3/8*16==6 for the graph part
2364          * and 5/8*16==10 for the filename part
2365          */
2366         if (width < 16 + 6 + number_width)
2367                 width = 16 + 6 + number_width;
2368
2369         /*
2370          * First assign sizes that are wanted, ignoring available width.
2371          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2372          * starting from "XXX" should fit in graph_width.
2373          */
2374         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2375         if (options->stat_graph_width &&
2376             options->stat_graph_width < graph_width)
2377                 graph_width = options->stat_graph_width;
2378
2379         name_width = (options->stat_name_width > 0 &&
2380                       options->stat_name_width < max_len) ?
2381                 options->stat_name_width : max_len;
2382
2383         /*
2384          * Adjust adjustable widths not to exceed maximum width
2385          */
2386         if (name_width + number_width + 6 + graph_width > width) {
2387                 if (graph_width > width * 3/8 - number_width - 6) {
2388                         graph_width = width * 3/8 - number_width - 6;
2389                         if (graph_width < 6)
2390                                 graph_width = 6;
2391                 }
2392
2393                 if (options->stat_graph_width &&
2394                     graph_width > options->stat_graph_width)
2395                         graph_width = options->stat_graph_width;
2396                 if (name_width > width - number_width - 6 - graph_width)
2397                         name_width = width - number_width - 6 - graph_width;
2398                 else
2399                         graph_width = width - number_width - 6 - name_width;
2400         }
2401
2402         /*
2403          * From here name_width is the width of the name area,
2404          * and graph_width is the width of the graph area.
2405          * max_change is used to scale graph properly.
2406          */
2407         for (i = 0; i < count; i++) {
2408                 const char *prefix = "";
2409                 struct diffstat_file *file = data->files[i];
2410                 char *name = file->print_name;
2411                 uintmax_t added = file->added;
2412                 uintmax_t deleted = file->deleted;
2413                 int name_len;
2414
2415                 if (!file->is_interesting && (added + deleted == 0))
2416                         continue;
2417
2418                 /*
2419                  * "scale" the filename
2420                  */
2421                 len = name_width;
2422                 name_len = strlen(name);
2423                 if (name_width < name_len) {
2424                         char *slash;
2425                         prefix = "...";
2426                         len -= 3;
2427                         name += name_len - len;
2428                         slash = strchr(name, '/');
2429                         if (slash)
2430                                 name = slash;
2431                 }
2432
2433                 if (file->is_binary) {
2434                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2435                         strbuf_addf(&out, " %*s", number_width, "Bin");
2436                         if (!added && !deleted) {
2437                                 strbuf_addch(&out, '\n');
2438                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2439                                                  out.buf, out.len, 0);
2440                                 strbuf_reset(&out);
2441                                 continue;
2442                         }
2443                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2444                                 del_c, deleted, reset);
2445                         strbuf_addstr(&out, " -> ");
2446                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2447                                 add_c, added, reset);
2448                         strbuf_addstr(&out, " bytes\n");
2449                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2450                                          out.buf, out.len, 0);
2451                         strbuf_reset(&out);
2452                         continue;
2453                 }
2454                 else if (file->is_unmerged) {
2455                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2456                         strbuf_addstr(&out, " Unmerged\n");
2457                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2458                                          out.buf, out.len, 0);
2459                         strbuf_reset(&out);
2460                         continue;
2461                 }
2462
2463                 /*
2464                  * scale the add/delete
2465                  */
2466                 add = added;
2467                 del = deleted;
2468
2469                 if (graph_width <= max_change) {
2470                         int total = scale_linear(add + del, graph_width, max_change);
2471                         if (total < 2 && add && del)
2472                                 /* width >= 2 due to the sanity check */
2473                                 total = 2;
2474                         if (add < del) {
2475                                 add = scale_linear(add, graph_width, max_change);
2476                                 del = total - add;
2477                         } else {
2478                                 del = scale_linear(del, graph_width, max_change);
2479                                 add = total - del;
2480                         }
2481                 }
2482                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2483                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2484                         number_width, added + deleted,
2485                         added + deleted ? " " : "");
2486                 show_graph(&out, '+', add, add_c, reset);
2487                 show_graph(&out, '-', del, del_c, reset);
2488                 strbuf_addch(&out, '\n');
2489                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2490                                  out.buf, out.len, 0);
2491                 strbuf_reset(&out);
2492         }
2493
2494         for (i = 0; i < data->nr; i++) {
2495                 struct diffstat_file *file = data->files[i];
2496                 uintmax_t added = file->added;
2497                 uintmax_t deleted = file->deleted;
2498
2499                 if (file->is_unmerged ||
2500                     (!file->is_interesting && (added + deleted == 0))) {
2501                         total_files--;
2502                         continue;
2503                 }
2504
2505                 if (!file->is_binary) {
2506                         adds += added;
2507                         dels += deleted;
2508                 }
2509                 if (i < count)
2510                         continue;
2511                 if (!extra_shown)
2512                         emit_diff_symbol(options,
2513                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2514                                          NULL, 0, 0);
2515                 extra_shown = 1;
2516         }
2517
2518         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2519         strbuf_release(&out);
2520 }
2521
2522 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2523 {
2524         int i, adds = 0, dels = 0, total_files = data->nr;
2525
2526         if (data->nr == 0)
2527                 return;
2528
2529         for (i = 0; i < data->nr; i++) {
2530                 int added = data->files[i]->added;
2531                 int deleted = data->files[i]->deleted;
2532
2533                 if (data->files[i]->is_unmerged ||
2534                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2535                         total_files--;
2536                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2537                         adds += added;
2538                         dels += deleted;
2539                 }
2540         }
2541         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2542 }
2543
2544 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2545 {
2546         int i;
2547
2548         if (data->nr == 0)
2549                 return;
2550
2551         for (i = 0; i < data->nr; i++) {
2552                 struct diffstat_file *file = data->files[i];
2553
2554                 fprintf(options->file, "%s", diff_line_prefix(options));
2555
2556                 if (file->is_binary)
2557                         fprintf(options->file, "-\t-\t");
2558                 else
2559                         fprintf(options->file,
2560                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2561                                 file->added, file->deleted);
2562                 if (options->line_termination) {
2563                         fill_print_name(file);
2564                         if (!file->is_renamed)
2565                                 write_name_quoted(file->name, options->file,
2566                                                   options->line_termination);
2567                         else {
2568                                 fputs(file->print_name, options->file);
2569                                 putc(options->line_termination, options->file);
2570                         }
2571                 } else {
2572                         if (file->is_renamed) {
2573                                 putc('\0', options->file);
2574                                 write_name_quoted(file->from_name, options->file, '\0');
2575                         }
2576                         write_name_quoted(file->name, options->file, '\0');
2577                 }
2578         }
2579 }
2580
2581 struct dirstat_file {
2582         const char *name;
2583         unsigned long changed;
2584 };
2585
2586 struct dirstat_dir {
2587         struct dirstat_file *files;
2588         int alloc, nr, permille, cumulative;
2589 };
2590
2591 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2592                 unsigned long changed, const char *base, int baselen)
2593 {
2594         unsigned long sum_changes = 0;
2595         unsigned int sources = 0;
2596         const char *line_prefix = diff_line_prefix(opt);
2597
2598         while (dir->nr) {
2599                 struct dirstat_file *f = dir->files;
2600                 int namelen = strlen(f->name);
2601                 unsigned long changes;
2602                 char *slash;
2603
2604                 if (namelen < baselen)
2605                         break;
2606                 if (memcmp(f->name, base, baselen))
2607                         break;
2608                 slash = strchr(f->name + baselen, '/');
2609                 if (slash) {
2610                         int newbaselen = slash + 1 - f->name;
2611                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2612                         sources++;
2613                 } else {
2614                         changes = f->changed;
2615                         dir->files++;
2616                         dir->nr--;
2617                         sources += 2;
2618                 }
2619                 sum_changes += changes;
2620         }
2621
2622         /*
2623          * We don't report dirstat's for
2624          *  - the top level
2625          *  - or cases where everything came from a single directory
2626          *    under this directory (sources == 1).
2627          */
2628         if (baselen && sources != 1) {
2629                 if (sum_changes) {
2630                         int permille = sum_changes * 1000 / changed;
2631                         if (permille >= dir->permille) {
2632                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2633                                         permille / 10, permille % 10, baselen, base);
2634                                 if (!dir->cumulative)
2635                                         return 0;
2636                         }
2637                 }
2638         }
2639         return sum_changes;
2640 }
2641
2642 static int dirstat_compare(const void *_a, const void *_b)
2643 {
2644         const struct dirstat_file *a = _a;
2645         const struct dirstat_file *b = _b;
2646         return strcmp(a->name, b->name);
2647 }
2648
2649 static void show_dirstat(struct diff_options *options)
2650 {
2651         int i;
2652         unsigned long changed;
2653         struct dirstat_dir dir;
2654         struct diff_queue_struct *q = &diff_queued_diff;
2655
2656         dir.files = NULL;
2657         dir.alloc = 0;
2658         dir.nr = 0;
2659         dir.permille = options->dirstat_permille;
2660         dir.cumulative = options->flags.dirstat_cumulative;
2661
2662         changed = 0;
2663         for (i = 0; i < q->nr; i++) {
2664                 struct diff_filepair *p = q->queue[i];
2665                 const char *name;
2666                 unsigned long copied, added, damage;
2667                 int content_changed;
2668
2669                 name = p->two->path ? p->two->path : p->one->path;
2670
2671                 if (p->one->oid_valid && p->two->oid_valid)
2672                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2673                 else
2674                         content_changed = 1;
2675
2676                 if (!content_changed) {
2677                         /*
2678                          * The SHA1 has not changed, so pre-/post-content is
2679                          * identical. We can therefore skip looking at the
2680                          * file contents altogether.
2681                          */
2682                         damage = 0;
2683                         goto found_damage;
2684                 }
2685
2686                 if (options->flags.dirstat_by_file) {
2687                         /*
2688                          * In --dirstat-by-file mode, we don't really need to
2689                          * look at the actual file contents at all.
2690                          * The fact that the SHA1 changed is enough for us to
2691                          * add this file to the list of results
2692                          * (with each file contributing equal damage).
2693                          */
2694                         damage = 1;
2695                         goto found_damage;
2696                 }
2697
2698                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2699                         diff_populate_filespec(p->one, 0);
2700                         diff_populate_filespec(p->two, 0);
2701                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2702                                                &copied, &added);
2703                         diff_free_filespec_data(p->one);
2704                         diff_free_filespec_data(p->two);
2705                 } else if (DIFF_FILE_VALID(p->one)) {
2706                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2707                         copied = added = 0;
2708                         diff_free_filespec_data(p->one);
2709                 } else if (DIFF_FILE_VALID(p->two)) {
2710                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2711                         copied = 0;
2712                         added = p->two->size;
2713                         diff_free_filespec_data(p->two);
2714                 } else
2715                         continue;
2716
2717                 /*
2718                  * Original minus copied is the removed material,
2719                  * added is the new material.  They are both damages
2720                  * made to the preimage.
2721                  * If the resulting damage is zero, we know that
2722                  * diffcore_count_changes() considers the two entries to
2723                  * be identical, but since content_changed is true, we
2724                  * know that there must have been _some_ kind of change,
2725                  * so we force all entries to have damage > 0.
2726                  */
2727                 damage = (p->one->size - copied) + added;
2728                 if (!damage)
2729                         damage = 1;
2730
2731 found_damage:
2732                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2733                 dir.files[dir.nr].name = name;
2734                 dir.files[dir.nr].changed = damage;
2735                 changed += damage;
2736                 dir.nr++;
2737         }
2738
2739         /* This can happen even with many files, if everything was renames */
2740         if (!changed)
2741                 return;
2742
2743         /* Show all directories with more than x% of the changes */
2744         QSORT(dir.files, dir.nr, dirstat_compare);
2745         gather_dirstat(options, &dir, changed, "", 0);
2746 }
2747
2748 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2749 {
2750         int i;
2751         unsigned long changed;
2752         struct dirstat_dir dir;
2753
2754         if (data->nr == 0)
2755                 return;
2756
2757         dir.files = NULL;
2758         dir.alloc = 0;
2759         dir.nr = 0;
2760         dir.permille = options->dirstat_permille;
2761         dir.cumulative = options->flags.dirstat_cumulative;
2762
2763         changed = 0;
2764         for (i = 0; i < data->nr; i++) {
2765                 struct diffstat_file *file = data->files[i];
2766                 unsigned long damage = file->added + file->deleted;
2767                 if (file->is_binary)
2768                         /*
2769                          * binary files counts bytes, not lines. Must find some
2770                          * way to normalize binary bytes vs. textual lines.
2771                          * The following heuristic assumes that there are 64
2772                          * bytes per "line".
2773                          * This is stupid and ugly, but very cheap...
2774                          */
2775                         damage = DIV_ROUND_UP(damage, 64);
2776                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2777                 dir.files[dir.nr].name = file->name;
2778                 dir.files[dir.nr].changed = damage;
2779                 changed += damage;
2780                 dir.nr++;
2781         }
2782
2783         /* This can happen even with many files, if everything was renames */
2784         if (!changed)
2785                 return;
2786
2787         /* Show all directories with more than x% of the changes */
2788         QSORT(dir.files, dir.nr, dirstat_compare);
2789         gather_dirstat(options, &dir, changed, "", 0);
2790 }
2791
2792 static void free_diffstat_info(struct diffstat_t *diffstat)
2793 {
2794         int i;
2795         for (i = 0; i < diffstat->nr; i++) {
2796                 struct diffstat_file *f = diffstat->files[i];
2797                 free(f->print_name);
2798                 free(f->name);
2799                 free(f->from_name);
2800                 free(f);
2801         }
2802         free(diffstat->files);
2803 }
2804
2805 struct checkdiff_t {
2806         const char *filename;
2807         int lineno;
2808         int conflict_marker_size;
2809         struct diff_options *o;
2810         unsigned ws_rule;
2811         unsigned status;
2812 };
2813
2814 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2815 {
2816         char firstchar;
2817         int cnt;
2818
2819         if (len < marker_size + 1)
2820                 return 0;
2821         firstchar = line[0];
2822         switch (firstchar) {
2823         case '=': case '>': case '<': case '|':
2824                 break;
2825         default:
2826                 return 0;
2827         }
2828         for (cnt = 1; cnt < marker_size; cnt++)
2829                 if (line[cnt] != firstchar)
2830                         return 0;
2831         /* line[1] thru line[marker_size-1] are same as firstchar */
2832         if (len < marker_size + 1 || !isspace(line[marker_size]))
2833                 return 0;
2834         return 1;
2835 }
2836
2837 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2838 {
2839         struct checkdiff_t *data = priv;
2840         int marker_size = data->conflict_marker_size;
2841         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2842         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2843         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2844         char *err;
2845         const char *line_prefix;
2846
2847         assert(data->o);
2848         line_prefix = diff_line_prefix(data->o);
2849
2850         if (line[0] == '+') {
2851                 unsigned bad;
2852                 data->lineno++;
2853                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2854                         data->status |= 1;
2855                         fprintf(data->o->file,
2856                                 "%s%s:%d: leftover conflict marker\n",
2857                                 line_prefix, data->filename, data->lineno);
2858                 }
2859                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2860                 if (!bad)
2861                         return;
2862                 data->status |= bad;
2863                 err = whitespace_error_string(bad);
2864                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2865                         line_prefix, data->filename, data->lineno, err);
2866                 free(err);
2867                 emit_line(data->o, set, reset, line, 1);
2868                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2869                               data->o->file, set, reset, ws);
2870         } else if (line[0] == ' ') {
2871                 data->lineno++;
2872         } else if (line[0] == '@') {
2873                 char *plus = strchr(line, '+');
2874                 if (plus)
2875                         data->lineno = strtol(plus, NULL, 10) - 1;
2876                 else
2877                         die("invalid diff");
2878         }
2879 }
2880
2881 static unsigned char *deflate_it(char *data,
2882                                  unsigned long size,
2883                                  unsigned long *result_size)
2884 {
2885         int bound;
2886         unsigned char *deflated;
2887         git_zstream stream;
2888
2889         git_deflate_init(&stream, zlib_compression_level);
2890         bound = git_deflate_bound(&stream, size);
2891         deflated = xmalloc(bound);
2892         stream.next_out = deflated;
2893         stream.avail_out = bound;
2894
2895         stream.next_in = (unsigned char *)data;
2896         stream.avail_in = size;
2897         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2898                 ; /* nothing */
2899         git_deflate_end(&stream);
2900         *result_size = stream.total_out;
2901         return deflated;
2902 }
2903
2904 static void emit_binary_diff_body(struct diff_options *o,
2905                                   mmfile_t *one, mmfile_t *two)
2906 {
2907         void *cp;
2908         void *delta;
2909         void *deflated;
2910         void *data;
2911         unsigned long orig_size;
2912         unsigned long delta_size;
2913         unsigned long deflate_size;
2914         unsigned long data_size;
2915
2916         /* We could do deflated delta, or we could do just deflated two,
2917          * whichever is smaller.
2918          */
2919         delta = NULL;
2920         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2921         if (one->size && two->size) {
2922                 delta = diff_delta(one->ptr, one->size,
2923                                    two->ptr, two->size,
2924                                    &delta_size, deflate_size);
2925                 if (delta) {
2926                         void *to_free = delta;
2927                         orig_size = delta_size;
2928                         delta = deflate_it(delta, delta_size, &delta_size);
2929                         free(to_free);
2930                 }
2931         }
2932
2933         if (delta && delta_size < deflate_size) {
2934                 char *s = xstrfmt("%lu", orig_size);
2935                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2936                                  s, strlen(s), 0);
2937                 free(s);
2938                 free(deflated);
2939                 data = delta;
2940                 data_size = delta_size;
2941         } else {
2942                 char *s = xstrfmt("%lu", two->size);
2943                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2944                                  s, strlen(s), 0);
2945                 free(s);
2946                 free(delta);
2947                 data = deflated;
2948                 data_size = deflate_size;
2949         }
2950
2951         /* emit data encoded in base85 */
2952         cp = data;
2953         while (data_size) {
2954                 int len;
2955                 int bytes = (52 < data_size) ? 52 : data_size;
2956                 char line[71];
2957                 data_size -= bytes;
2958                 if (bytes <= 26)
2959                         line[0] = bytes + 'A' - 1;
2960                 else
2961                         line[0] = bytes - 26 + 'a' - 1;
2962                 encode_85(line + 1, cp, bytes);
2963                 cp = (char *) cp + bytes;
2964
2965                 len = strlen(line);
2966                 line[len++] = '\n';
2967                 line[len] = '\0';
2968
2969                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2970                                  line, len, 0);
2971         }
2972         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2973         free(data);
2974 }
2975
2976 static void emit_binary_diff(struct diff_options *o,
2977                              mmfile_t *one, mmfile_t *two)
2978 {
2979         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2980         emit_binary_diff_body(o, one, two);
2981         emit_binary_diff_body(o, two, one);
2982 }
2983
2984 int diff_filespec_is_binary(struct diff_filespec *one)
2985 {
2986         if (one->is_binary == -1) {
2987                 diff_filespec_load_driver(one);
2988                 if (one->driver->binary != -1)
2989                         one->is_binary = one->driver->binary;
2990                 else {
2991                         if (!one->data && DIFF_FILE_VALID(one))
2992                                 diff_populate_filespec(one, CHECK_BINARY);
2993                         if (one->is_binary == -1 && one->data)
2994                                 one->is_binary = buffer_is_binary(one->data,
2995                                                 one->size);
2996                         if (one->is_binary == -1)
2997                                 one->is_binary = 0;
2998                 }
2999         }
3000         return one->is_binary;
3001 }
3002
3003 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3004 {
3005         diff_filespec_load_driver(one);
3006         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3007 }
3008
3009 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3010 {
3011         if (!options->a_prefix)
3012                 options->a_prefix = a;
3013         if (!options->b_prefix)
3014                 options->b_prefix = b;
3015 }
3016
3017 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3018 {
3019         if (!DIFF_FILE_VALID(one))
3020                 return NULL;
3021
3022         diff_filespec_load_driver(one);
3023         return userdiff_get_textconv(one->driver);
3024 }
3025
3026 static void builtin_diff(const char *name_a,
3027                          const char *name_b,
3028                          struct diff_filespec *one,
3029                          struct diff_filespec *two,
3030                          const char *xfrm_msg,
3031                          int must_show_header,
3032                          struct diff_options *o,
3033                          int complete_rewrite)
3034 {
3035         mmfile_t mf1, mf2;
3036         const char *lbl[2];
3037         char *a_one, *b_two;
3038         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3039         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3040         const char *a_prefix, *b_prefix;
3041         struct userdiff_driver *textconv_one = NULL;
3042         struct userdiff_driver *textconv_two = NULL;
3043         struct strbuf header = STRBUF_INIT;
3044         const char *line_prefix = diff_line_prefix(o);
3045
3046         diff_set_mnemonic_prefix(o, "a/", "b/");
3047         if (o->flags.reverse_diff) {
3048                 a_prefix = o->b_prefix;
3049                 b_prefix = o->a_prefix;
3050         } else {
3051                 a_prefix = o->a_prefix;
3052                 b_prefix = o->b_prefix;
3053         }
3054
3055         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3056             (!one->mode || S_ISGITLINK(one->mode)) &&
3057             (!two->mode || S_ISGITLINK(two->mode))) {
3058                 show_submodule_summary(o, one->path ? one->path : two->path,
3059                                 &one->oid, &two->oid,
3060                                 two->dirty_submodule);
3061                 return;
3062         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3063                    (!one->mode || S_ISGITLINK(one->mode)) &&
3064                    (!two->mode || S_ISGITLINK(two->mode))) {
3065                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3066                                 &one->oid, &two->oid,
3067                                 two->dirty_submodule);
3068                 return;
3069         }
3070
3071         if (o->flags.allow_textconv) {
3072                 textconv_one = get_textconv(one);
3073                 textconv_two = get_textconv(two);
3074         }
3075
3076         /* Never use a non-valid filename anywhere if at all possible */
3077         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3078         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3079
3080         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3081         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3082         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3083         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3084         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3085         if (lbl[0][0] == '/') {
3086                 /* /dev/null */
3087                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3088                 if (xfrm_msg)
3089                         strbuf_addstr(&header, xfrm_msg);
3090                 must_show_header = 1;
3091         }
3092         else if (lbl[1][0] == '/') {
3093                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3094                 if (xfrm_msg)
3095                         strbuf_addstr(&header, xfrm_msg);
3096                 must_show_header = 1;
3097         }
3098         else {
3099                 if (one->mode != two->mode) {
3100                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3101                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3102                         must_show_header = 1;
3103                 }
3104                 if (xfrm_msg)
3105                         strbuf_addstr(&header, xfrm_msg);
3106
3107                 /*
3108                  * we do not run diff between different kind
3109                  * of objects.
3110                  */
3111                 if ((one->mode ^ two->mode) & S_IFMT)
3112                         goto free_ab_and_return;
3113                 if (complete_rewrite &&
3114                     (textconv_one || !diff_filespec_is_binary(one)) &&
3115                     (textconv_two || !diff_filespec_is_binary(two))) {
3116                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3117                                          header.buf, header.len, 0);
3118                         strbuf_reset(&header);
3119                         emit_rewrite_diff(name_a, name_b, one, two,
3120                                                 textconv_one, textconv_two, o);
3121                         o->found_changes = 1;
3122                         goto free_ab_and_return;
3123                 }
3124         }
3125
3126         if (o->irreversible_delete && lbl[1][0] == '/') {
3127                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3128                                  header.len, 0);
3129                 strbuf_reset(&header);
3130                 goto free_ab_and_return;
3131         } else if (!o->flags.text &&
3132             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3133               (!textconv_two && diff_filespec_is_binary(two)) )) {
3134                 struct strbuf sb = STRBUF_INIT;
3135                 if (!one->data && !two->data &&
3136                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3137                     !o->flags.binary) {
3138                         if (!oidcmp(&one->oid, &two->oid)) {
3139                                 if (must_show_header)
3140                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3141                                                          header.buf, header.len,
3142                                                          0);
3143                                 goto free_ab_and_return;
3144                         }
3145                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3146                                          header.buf, header.len, 0);
3147                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3148                                     diff_line_prefix(o), lbl[0], lbl[1]);
3149                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3150                                          sb.buf, sb.len, 0);
3151                         strbuf_release(&sb);
3152                         goto free_ab_and_return;
3153                 }
3154                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3155                         die("unable to read files to diff");
3156                 /* Quite common confusing case */
3157                 if (mf1.size == mf2.size &&
3158                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3159                         if (must_show_header)
3160                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3161                                                  header.buf, header.len, 0);
3162                         goto free_ab_and_return;
3163                 }
3164                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3165                 strbuf_reset(&header);
3166                 if (o->flags.binary)
3167                         emit_binary_diff(o, &mf1, &mf2);
3168                 else {
3169                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3170                                     diff_line_prefix(o), lbl[0], lbl[1]);
3171                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3172                                          sb.buf, sb.len, 0);
3173                         strbuf_release(&sb);
3174                 }
3175                 o->found_changes = 1;
3176         } else {
3177                 /* Crazy xdl interfaces.. */
3178                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3179                 const char *v;
3180                 xpparam_t xpp;
3181                 xdemitconf_t xecfg;
3182                 struct emit_callback ecbdata;
3183                 const struct userdiff_funcname *pe;
3184
3185                 if (must_show_header) {
3186                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3187                                          header.buf, header.len, 0);
3188                         strbuf_reset(&header);
3189                 }
3190
3191                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3192                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3193
3194                 pe = diff_funcname_pattern(one);
3195                 if (!pe)
3196                         pe = diff_funcname_pattern(two);
3197
3198                 memset(&xpp, 0, sizeof(xpp));
3199                 memset(&xecfg, 0, sizeof(xecfg));
3200                 memset(&ecbdata, 0, sizeof(ecbdata));
3201                 ecbdata.label_path = lbl;
3202                 ecbdata.color_diff = want_color(o->use_color);
3203                 ecbdata.ws_rule = whitespace_rule(name_b);
3204                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3205                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3206                 ecbdata.opt = o;
3207                 ecbdata.header = header.len ? &header : NULL;
3208                 xpp.flags = o->xdl_opts;
3209                 xpp.anchors = o->anchors;
3210                 xpp.anchors_nr = o->anchors_nr;
3211                 xecfg.ctxlen = o->context;
3212                 xecfg.interhunkctxlen = o->interhunkcontext;
3213                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3214                 if (o->flags.funccontext)
3215                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3216                 if (pe)
3217                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3218                 if (!diffopts)
3219                         ;
3220                 else if (skip_prefix(diffopts, "--unified=", &v))
3221                         xecfg.ctxlen = strtoul(v, NULL, 10);
3222                 else if (skip_prefix(diffopts, "-u", &v))
3223                         xecfg.ctxlen = strtoul(v, NULL, 10);
3224                 if (o->word_diff)
3225                         init_diff_words_data(&ecbdata, o, one, two);
3226                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3227                                   &xpp, &xecfg))
3228                         die("unable to generate diff for %s", one->path);
3229                 if (o->word_diff)
3230                         free_diff_words_data(&ecbdata);
3231                 if (textconv_one)
3232                         free(mf1.ptr);
3233                 if (textconv_two)
3234                         free(mf2.ptr);
3235                 xdiff_clear_find_func(&xecfg);
3236         }
3237
3238  free_ab_and_return:
3239         strbuf_release(&header);
3240         diff_free_filespec_data(one);
3241         diff_free_filespec_data(two);
3242         free(a_one);
3243         free(b_two);
3244         return;
3245 }
3246
3247 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3248 {
3249         if (!is_renamed) {
3250                 if (p->status == DIFF_STATUS_ADDED) {
3251                         if (S_ISLNK(p->two->mode))
3252                                 return "new +l";
3253                         else if ((p->two->mode & 0777) == 0755)
3254                                 return "new +x";
3255                         else
3256                                 return "new";
3257                 } else if (p->status == DIFF_STATUS_DELETED)
3258                         return "gone";
3259         }
3260         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3261                 return "mode -l";
3262         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3263                 return "mode +l";
3264         else if ((p->one->mode & 0777) == 0644 &&
3265                  (p->two->mode & 0777) == 0755)
3266                 return "mode +x";
3267         else if ((p->one->mode & 0777) == 0755 &&
3268                  (p->two->mode & 0777) == 0644)
3269                 return "mode -x";
3270         return NULL;
3271 }
3272
3273 static void builtin_diffstat(const char *name_a, const char *name_b,
3274                              struct diff_filespec *one,
3275                              struct diff_filespec *two,
3276                              struct diffstat_t *diffstat,
3277                              struct diff_options *o,
3278                              struct diff_filepair *p)
3279 {
3280         mmfile_t mf1, mf2;
3281         struct diffstat_file *data;
3282         int same_contents;
3283         int complete_rewrite = 0;
3284
3285         if (!DIFF_PAIR_UNMERGED(p)) {
3286                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3287                         complete_rewrite = 1;
3288         }
3289
3290         data = diffstat_add(diffstat, name_a, name_b);
3291         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3292         if (o->flags.stat_with_summary)
3293                 data->comments = get_compact_summary(p, data->is_renamed);
3294
3295         if (!one || !two) {
3296                 data->is_unmerged = 1;
3297                 return;
3298         }
3299
3300         same_contents = !oidcmp(&one->oid, &two->oid);
3301
3302         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3303                 data->is_binary = 1;
3304                 if (same_contents) {
3305                         data->added = 0;
3306                         data->deleted = 0;
3307                 } else {
3308                         data->added = diff_filespec_size(two);
3309                         data->deleted = diff_filespec_size(one);
3310                 }
3311         }
3312
3313         else if (complete_rewrite) {
3314                 diff_populate_filespec(one, 0);
3315                 diff_populate_filespec(two, 0);
3316                 data->deleted = count_lines(one->data, one->size);
3317                 data->added = count_lines(two->data, two->size);
3318         }
3319
3320         else if (!same_contents) {
3321                 /* Crazy xdl interfaces.. */
3322                 xpparam_t xpp;
3323                 xdemitconf_t xecfg;
3324
3325                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3326                         die("unable to read files to diff");
3327
3328                 memset(&xpp, 0, sizeof(xpp));
3329                 memset(&xecfg, 0, sizeof(xecfg));
3330                 xpp.flags = o->xdl_opts;
3331                 xpp.anchors = o->anchors;
3332                 xpp.anchors_nr = o->anchors_nr;
3333                 xecfg.ctxlen = o->context;
3334                 xecfg.interhunkctxlen = o->interhunkcontext;
3335                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3336                                   &xpp, &xecfg))
3337                         die("unable to generate diffstat for %s", one->path);
3338         }
3339
3340         diff_free_filespec_data(one);
3341         diff_free_filespec_data(two);
3342 }
3343
3344 static void builtin_checkdiff(const char *name_a, const char *name_b,
3345                               const char *attr_path,
3346                               struct diff_filespec *one,
3347                               struct diff_filespec *two,
3348                               struct diff_options *o)
3349 {
3350         mmfile_t mf1, mf2;
3351         struct checkdiff_t data;
3352
3353         if (!two)
3354                 return;
3355
3356         memset(&data, 0, sizeof(data));
3357         data.filename = name_b ? name_b : name_a;
3358         data.lineno = 0;
3359         data.o = o;
3360         data.ws_rule = whitespace_rule(attr_path);
3361         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3362
3363         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3364                 die("unable to read files to diff");
3365
3366         /*
3367          * All the other codepaths check both sides, but not checking
3368          * the "old" side here is deliberate.  We are checking the newly
3369          * introduced changes, and as long as the "new" side is text, we
3370          * can and should check what it introduces.
3371          */
3372         if (diff_filespec_is_binary(two))
3373                 goto free_and_return;
3374         else {
3375                 /* Crazy xdl interfaces.. */
3376                 xpparam_t xpp;
3377                 xdemitconf_t xecfg;
3378
3379                 memset(&xpp, 0, sizeof(xpp));
3380                 memset(&xecfg, 0, sizeof(xecfg));
3381                 xecfg.ctxlen = 1; /* at least one context line */
3382                 xpp.flags = 0;
3383                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3384                                   &xpp, &xecfg))
3385                         die("unable to generate checkdiff for %s", one->path);
3386
3387                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3388                         struct emit_callback ecbdata;
3389                         int blank_at_eof;
3390
3391                         ecbdata.ws_rule = data.ws_rule;
3392                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3393                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3394
3395                         if (blank_at_eof) {
3396                                 static char *err;
3397                                 if (!err)
3398                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3399                                 fprintf(o->file, "%s:%d: %s.\n",
3400                                         data.filename, blank_at_eof, err);
3401                                 data.status = 1; /* report errors */
3402                         }
3403                 }
3404         }
3405  free_and_return:
3406         diff_free_filespec_data(one);
3407         diff_free_filespec_data(two);
3408         if (data.status)
3409                 o->flags.check_failed = 1;
3410 }
3411
3412 struct diff_filespec *alloc_filespec(const char *path)
3413 {
3414         struct diff_filespec *spec;
3415
3416         FLEXPTR_ALLOC_STR(spec, path, path);
3417         spec->count = 1;
3418         spec->is_binary = -1;
3419         return spec;
3420 }
3421
3422 void free_filespec(struct diff_filespec *spec)
3423 {
3424         if (!--spec->count) {
3425                 diff_free_filespec_data(spec);
3426                 free(spec);
3427         }
3428 }
3429
3430 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3431                    int oid_valid, unsigned short mode)
3432 {
3433         if (mode) {
3434                 spec->mode = canon_mode(mode);
3435                 oidcpy(&spec->oid, oid);
3436                 spec->oid_valid = oid_valid;
3437         }
3438 }
3439
3440 /*
3441  * Given a name and sha1 pair, if the index tells us the file in
3442  * the work tree has that object contents, return true, so that
3443  * prepare_temp_file() does not have to inflate and extract.
3444  */
3445 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3446 {
3447         const struct cache_entry *ce;
3448         struct stat st;
3449         int pos, len;
3450
3451         /*
3452          * We do not read the cache ourselves here, because the
3453          * benchmark with my previous version that always reads cache
3454          * shows that it makes things worse for diff-tree comparing
3455          * two linux-2.6 kernel trees in an already checked out work
3456          * tree.  This is because most diff-tree comparisons deal with
3457          * only a small number of files, while reading the cache is
3458          * expensive for a large project, and its cost outweighs the
3459          * savings we get by not inflating the object to a temporary
3460          * file.  Practically, this code only helps when we are used
3461          * by diff-cache --cached, which does read the cache before
3462          * calling us.
3463          */
3464         if (!active_cache)
3465                 return 0;
3466
3467         /* We want to avoid the working directory if our caller
3468          * doesn't need the data in a normal file, this system
3469          * is rather slow with its stat/open/mmap/close syscalls,
3470          * and the object is contained in a pack file.  The pack
3471          * is probably already open and will be faster to obtain
3472          * the data through than the working directory.  Loose
3473          * objects however would tend to be slower as they need
3474          * to be individually opened and inflated.
3475          */
3476         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3477                 return 0;
3478
3479         /*
3480          * Similarly, if we'd have to convert the file contents anyway, that
3481          * makes the optimization not worthwhile.
3482          */
3483         if (!want_file && would_convert_to_git(&the_index, name))
3484                 return 0;
3485
3486         len = strlen(name);
3487         pos = cache_name_pos(name, len);
3488         if (pos < 0)
3489                 return 0;
3490         ce = active_cache[pos];
3491
3492         /*
3493          * This is not the sha1 we are looking for, or
3494          * unreusable because it is not a regular file.
3495          */
3496         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3497                 return 0;
3498
3499         /*
3500          * If ce is marked as "assume unchanged", there is no
3501          * guarantee that work tree matches what we are looking for.
3502          */
3503         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3504                 return 0;
3505
3506         /*
3507          * If ce matches the file in the work tree, we can reuse it.
3508          */
3509         if (ce_uptodate(ce) ||
3510             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3511                 return 1;
3512
3513         return 0;
3514 }
3515
3516 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3517 {
3518         struct strbuf buf = STRBUF_INIT;
3519         char *dirty = "";
3520
3521         /* Are we looking at the work tree? */
3522         if (s->dirty_submodule)
3523                 dirty = "-dirty";
3524
3525         strbuf_addf(&buf, "Subproject commit %s%s\n",
3526                     oid_to_hex(&s->oid), dirty);
3527         s->size = buf.len;
3528         if (size_only) {
3529                 s->data = NULL;
3530                 strbuf_release(&buf);
3531         } else {
3532                 s->data = strbuf_detach(&buf, NULL);
3533                 s->should_free = 1;
3534         }
3535         return 0;
3536 }
3537
3538 /*
3539  * While doing rename detection and pickaxe operation, we may need to
3540  * grab the data for the blob (or file) for our own in-core comparison.
3541  * diff_filespec has data and size fields for this purpose.
3542  */
3543 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3544 {
3545         int size_only = flags & CHECK_SIZE_ONLY;
3546         int err = 0;
3547         int conv_flags = global_conv_flags_eol;
3548         /*
3549          * demote FAIL to WARN to allow inspecting the situation
3550          * instead of refusing.
3551          */
3552         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3553                 conv_flags = CONV_EOL_RNDTRP_WARN;
3554
3555         if (!DIFF_FILE_VALID(s))
3556                 die("internal error: asking to populate invalid file.");
3557         if (S_ISDIR(s->mode))
3558                 return -1;
3559
3560         if (s->data)
3561                 return 0;
3562
3563         if (size_only && 0 < s->size)
3564                 return 0;
3565
3566         if (S_ISGITLINK(s->mode))
3567                 return diff_populate_gitlink(s, size_only);
3568
3569         if (!s->oid_valid ||
3570             reuse_worktree_file(s->path, &s->oid, 0)) {
3571                 struct strbuf buf = STRBUF_INIT;
3572                 struct stat st;
3573                 int fd;
3574
3575                 if (lstat(s->path, &st) < 0) {
3576                 err_empty:
3577                         err = -1;
3578                 empty:
3579                         s->data = (char *)"";
3580                         s->size = 0;
3581                         return err;
3582                 }
3583                 s->size = xsize_t(st.st_size);
3584                 if (!s->size)
3585                         goto empty;
3586                 if (S_ISLNK(st.st_mode)) {
3587                         struct strbuf sb = STRBUF_INIT;
3588
3589                         if (strbuf_readlink(&sb, s->path, s->size))
3590                                 goto err_empty;
3591                         s->size = sb.len;
3592                         s->data = strbuf_detach(&sb, NULL);
3593                         s->should_free = 1;
3594                         return 0;
3595                 }
3596
3597                 /*
3598                  * Even if the caller would be happy with getting
3599                  * only the size, we cannot return early at this
3600                  * point if the path requires us to run the content
3601                  * conversion.
3602                  */
3603                 if (size_only && !would_convert_to_git(&the_index, s->path))
3604                         return 0;
3605
3606                 /*
3607                  * Note: this check uses xsize_t(st.st_size) that may
3608                  * not be the true size of the blob after it goes
3609                  * through convert_to_git().  This may not strictly be
3610                  * correct, but the whole point of big_file_threshold
3611                  * and is_binary check being that we want to avoid
3612                  * opening the file and inspecting the contents, this
3613                  * is probably fine.
3614                  */
3615                 if ((flags & CHECK_BINARY) &&
3616                     s->size > big_file_threshold && s->is_binary == -1) {
3617                         s->is_binary = 1;
3618                         return 0;
3619                 }
3620                 fd = open(s->path, O_RDONLY);
3621                 if (fd < 0)
3622                         goto err_empty;
3623                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3624                 close(fd);
3625                 s->should_munmap = 1;
3626
3627                 /*
3628                  * Convert from working tree format to canonical git format
3629                  */
3630                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3631                         size_t size = 0;
3632                         munmap(s->data, s->size);
3633                         s->should_munmap = 0;
3634                         s->data = strbuf_detach(&buf, &size);
3635                         s->size = size;
3636                         s->should_free = 1;
3637                 }
3638         }
3639         else {
3640                 enum object_type type;
3641                 if (size_only || (flags & CHECK_BINARY)) {
3642                         type = oid_object_info(the_repository, &s->oid,
3643                                                &s->size);
3644                         if (type < 0)
3645                                 die("unable to read %s",
3646                                     oid_to_hex(&s->oid));
3647                         if (size_only)
3648                                 return 0;
3649                         if (s->size > big_file_threshold && s->is_binary == -1) {
3650                                 s->is_binary = 1;
3651                                 return 0;
3652                         }
3653                 }
3654                 s->data = read_object_file(&s->oid, &type, &s->size);
3655                 if (!s->data)
3656                         die("unable to read %s", oid_to_hex(&s->oid));
3657                 s->should_free = 1;
3658         }
3659         return 0;
3660 }
3661
3662 void diff_free_filespec_blob(struct diff_filespec *s)
3663 {
3664         if (s->should_free)
3665                 free(s->data);
3666         else if (s->should_munmap)
3667                 munmap(s->data, s->size);
3668
3669         if (s->should_free || s->should_munmap) {
3670                 s->should_free = s->should_munmap = 0;
3671                 s->data = NULL;
3672         }
3673 }
3674
3675 void diff_free_filespec_data(struct diff_filespec *s)
3676 {
3677         diff_free_filespec_blob(s);
3678         FREE_AND_NULL(s->cnt_data);
3679 }
3680
3681 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3682                            void *blob,
3683                            unsigned long size,
3684                            const struct object_id *oid,
3685                            int mode)
3686 {
3687         struct strbuf buf = STRBUF_INIT;
3688         struct strbuf tempfile = STRBUF_INIT;
3689         char *path_dup = xstrdup(path);
3690         const char *base = basename(path_dup);
3691
3692         /* Generate "XXXXXX_basename.ext" */
3693         strbuf_addstr(&tempfile, "XXXXXX_");
3694         strbuf_addstr(&tempfile, base);
3695
3696         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3697         if (!temp->tempfile)
3698                 die_errno("unable to create temp-file");
3699         if (convert_to_working_tree(path,
3700                         (const char *)blob, (size_t)size, &buf)) {
3701                 blob = buf.buf;
3702                 size = buf.len;
3703         }
3704         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3705             close_tempfile_gently(temp->tempfile))
3706                 die_errno("unable to write temp-file");
3707         temp->name = get_tempfile_path(temp->tempfile);
3708         oid_to_hex_r(temp->hex, oid);
3709         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3710         strbuf_release(&buf);
3711         strbuf_release(&tempfile);
3712         free(path_dup);
3713 }
3714
3715 static struct diff_tempfile *prepare_temp_file(const char *name,
3716                 struct diff_filespec *one)
3717 {
3718         struct diff_tempfile *temp = claim_diff_tempfile();
3719
3720         if (!DIFF_FILE_VALID(one)) {
3721         not_a_valid_file:
3722                 /* A '-' entry produces this for file-2, and
3723                  * a '+' entry produces this for file-1.
3724                  */
3725                 temp->name = "/dev/null";
3726                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3727                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3728                 return temp;
3729         }
3730
3731         if (!S_ISGITLINK(one->mode) &&
3732             (!one->oid_valid ||
3733              reuse_worktree_file(name, &one->oid, 1))) {
3734                 struct stat st;
3735                 if (lstat(name, &st) < 0) {
3736                         if (errno == ENOENT)
3737                                 goto not_a_valid_file;
3738                         die_errno("stat(%s)", name);
3739                 }
3740                 if (S_ISLNK(st.st_mode)) {
3741                         struct strbuf sb = STRBUF_INIT;
3742                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3743                                 die_errno("readlink(%s)", name);
3744                         prep_temp_blob(name, temp, sb.buf, sb.len,
3745                                        (one->oid_valid ?
3746                                         &one->oid : &null_oid),
3747                                        (one->oid_valid ?
3748                                         one->mode : S_IFLNK));
3749                         strbuf_release(&sb);
3750                 }
3751                 else {
3752                         /* we can borrow from the file in the work tree */
3753                         temp->name = name;
3754                         if (!one->oid_valid)
3755                                 oid_to_hex_r(temp->hex, &null_oid);
3756                         else
3757                                 oid_to_hex_r(temp->hex, &one->oid);
3758                         /* Even though we may sometimes borrow the
3759                          * contents from the work tree, we always want
3760                          * one->mode.  mode is trustworthy even when
3761                          * !(one->oid_valid), as long as
3762                          * DIFF_FILE_VALID(one).
3763                          */
3764                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3765                 }
3766                 return temp;
3767         }
3768         else {
3769                 if (diff_populate_filespec(one, 0))
3770                         die("cannot read data blob for %s", one->path);
3771                 prep_temp_blob(name, temp, one->data, one->size,
3772                                &one->oid, one->mode);
3773         }
3774         return temp;
3775 }
3776
3777 static void add_external_diff_name(struct argv_array *argv,
3778                                    const char *name,
3779                                    struct diff_filespec *df)
3780 {
3781         struct diff_tempfile *temp = prepare_temp_file(name, df);
3782         argv_array_push(argv, temp->name);
3783         argv_array_push(argv, temp->hex);
3784         argv_array_push(argv, temp->mode);
3785 }
3786
3787 /* An external diff command takes:
3788  *
3789  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3790  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3791  *
3792  */
3793 static void run_external_diff(const char *pgm,
3794                               const char *name,
3795                               const char *other,
3796                               struct diff_filespec *one,
3797                               struct diff_filespec *two,
3798                               const char *xfrm_msg,
3799                               int complete_rewrite,
3800                               struct diff_options *o)
3801 {
3802         struct argv_array argv = ARGV_ARRAY_INIT;
3803         struct argv_array env = ARGV_ARRAY_INIT;
3804         struct diff_queue_struct *q = &diff_queued_diff;
3805
3806         argv_array_push(&argv, pgm);
3807         argv_array_push(&argv, name);
3808
3809         if (one && two) {
3810                 add_external_diff_name(&argv, name, one);
3811                 if (!other)
3812                         add_external_diff_name(&argv, name, two);
3813                 else {
3814                         add_external_diff_name(&argv, other, two);
3815                         argv_array_push(&argv, other);
3816                         argv_array_push(&argv, xfrm_msg);
3817                 }
3818         }
3819
3820         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3821         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3822
3823         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3824                 die(_("external diff died, stopping at %s"), name);
3825
3826         remove_tempfile();
3827         argv_array_clear(&argv);
3828         argv_array_clear(&env);
3829 }
3830
3831 static int similarity_index(struct diff_filepair *p)
3832 {
3833         return p->score * 100 / MAX_SCORE;
3834 }
3835
3836 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3837 {
3838         if (startup_info->have_repository)
3839                 return find_unique_abbrev(oid, abbrev);
3840         else {
3841                 char *hex = oid_to_hex(oid);
3842                 if (abbrev < 0)
3843                         abbrev = FALLBACK_DEFAULT_ABBREV;
3844                 if (abbrev > GIT_SHA1_HEXSZ)
3845                         die("BUG: oid abbreviation out of range: %d", abbrev);
3846                 if (abbrev)
3847                         hex[abbrev] = '\0';
3848                 return hex;
3849         }
3850 }
3851
3852 static void fill_metainfo(struct strbuf *msg,
3853                           const char *name,
3854                           const char *other,
3855                           struct diff_filespec *one,
3856                           struct diff_filespec *two,
3857                           struct diff_options *o,
3858                           struct diff_filepair *p,
3859                           int *must_show_header,
3860                           int use_color)
3861 {
3862         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3863         const char *reset = diff_get_color(use_color, DIFF_RESET);
3864         const char *line_prefix = diff_line_prefix(o);
3865
3866         *must_show_header = 1;
3867         strbuf_init(msg, PATH_MAX * 2 + 300);
3868         switch (p->status) {
3869         case DIFF_STATUS_COPIED:
3870                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3871                             line_prefix, set, similarity_index(p));
3872                 strbuf_addf(msg, "%s\n%s%scopy from ",
3873                             reset,  line_prefix, set);
3874                 quote_c_style(name, msg, NULL, 0);
3875                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3876                 quote_c_style(other, msg, NULL, 0);
3877                 strbuf_addf(msg, "%s\n", reset);
3878                 break;
3879         case DIFF_STATUS_RENAMED:
3880                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3881                             line_prefix, set, similarity_index(p));
3882                 strbuf_addf(msg, "%s\n%s%srename from ",
3883                             reset, line_prefix, set);
3884                 quote_c_style(name, msg, NULL, 0);
3885                 strbuf_addf(msg, "%s\n%s%srename to ",
3886                             reset, line_prefix, set);
3887                 quote_c_style(other, msg, NULL, 0);
3888                 strbuf_addf(msg, "%s\n", reset);
3889                 break;
3890         case DIFF_STATUS_MODIFIED:
3891                 if (p->score) {
3892                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3893                                     line_prefix,
3894                                     set, similarity_index(p), reset);
3895                         break;
3896                 }
3897                 /* fallthru */
3898         default:
3899                 *must_show_header = 0;
3900         }
3901         if (one && two && oidcmp(&one->oid, &two->oid)) {
3902                 int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
3903
3904                 if (o->flags.binary) {
3905                         mmfile_t mf;
3906                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3907                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3908                                 abbrev = 40;
3909                 }
3910                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3911                             diff_abbrev_oid(&one->oid, abbrev),
3912                             diff_abbrev_oid(&two->oid, abbrev));
3913                 if (one->mode == two->mode)
3914                         strbuf_addf(msg, " %06o", one->mode);
3915                 strbuf_addf(msg, "%s\n", reset);
3916         }
3917 }
3918
3919 static void run_diff_cmd(const char *pgm,
3920                          const char *name,
3921                          const char *other,
3922                          const char *attr_path,
3923                          struct diff_filespec *one,
3924                          struct diff_filespec *two,
3925                          struct strbuf *msg,
3926                          struct diff_options *o,
3927                          struct diff_filepair *p)
3928 {
3929         const char *xfrm_msg = NULL;
3930         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3931         int must_show_header = 0;
3932
3933
3934         if (o->flags.allow_external) {
3935                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3936                 if (drv && drv->external)
3937                         pgm = drv->external;
3938         }
3939
3940         if (msg) {
3941                 /*
3942                  * don't use colors when the header is intended for an
3943                  * external diff driver
3944                  */
3945                 fill_metainfo(msg, name, other, one, two, o, p,
3946                               &must_show_header,
3947                               want_color(o->use_color) && !pgm);
3948                 xfrm_msg = msg->len ? msg->buf : NULL;
3949         }
3950
3951         if (pgm) {
3952                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3953                                   complete_rewrite, o);
3954                 return;
3955         }
3956         if (one && two)
3957                 builtin_diff(name, other ? other : name,
3958                              one, two, xfrm_msg, must_show_header,
3959                              o, complete_rewrite);
3960         else
3961                 fprintf(o->file, "* Unmerged path %s\n", name);
3962 }
3963
3964 static void diff_fill_oid_info(struct diff_filespec *one)
3965 {
3966         if (DIFF_FILE_VALID(one)) {
3967                 if (!one->oid_valid) {
3968                         struct stat st;
3969                         if (one->is_stdin) {
3970                                 oidclr(&one->oid);
3971                                 return;
3972                         }
3973                         if (lstat(one->path, &st) < 0)
3974                                 die_errno("stat '%s'", one->path);
3975                         if (index_path(&one->oid, one->path, &st, 0))
3976                                 die("cannot hash %s", one->path);
3977                 }
3978         }
3979         else
3980                 oidclr(&one->oid);
3981 }
3982
3983 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3984 {
3985         /* Strip the prefix but do not molest /dev/null and absolute paths */
3986         if (*namep && **namep != '/') {
3987                 *namep += prefix_length;
3988                 if (**namep == '/')
3989                         ++*namep;
3990         }
3991         if (*otherp && **otherp != '/') {
3992                 *otherp += prefix_length;
3993                 if (**otherp == '/')
3994                         ++*otherp;
3995         }
3996 }
3997
3998 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3999 {
4000         const char *pgm = external_diff();
4001         struct strbuf msg;
4002         struct diff_filespec *one = p->one;
4003         struct diff_filespec *two = p->two;
4004         const char *name;
4005         const char *other;
4006         const char *attr_path;
4007
4008         name  = one->path;
4009         other = (strcmp(name, two->path) ? two->path : NULL);
4010         attr_path = name;
4011         if (o->prefix_length)
4012                 strip_prefix(o->prefix_length, &name, &other);
4013
4014         if (!o->flags.allow_external)
4015                 pgm = NULL;
4016
4017         if (DIFF_PAIR_UNMERGED(p)) {
4018                 run_diff_cmd(pgm, name, NULL, attr_path,
4019                              NULL, NULL, NULL, o, p);
4020                 return;
4021         }
4022
4023         diff_fill_oid_info(one);
4024         diff_fill_oid_info(two);
4025
4026         if (!pgm &&
4027             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4028             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4029                 /*
4030                  * a filepair that changes between file and symlink
4031                  * needs to be split into deletion and creation.
4032                  */
4033                 struct diff_filespec *null = alloc_filespec(two->path);
4034                 run_diff_cmd(NULL, name, other, attr_path,
4035                              one, null, &msg, o, p);
4036                 free(null);
4037                 strbuf_release(&msg);
4038
4039                 null = alloc_filespec(one->path);
4040                 run_diff_cmd(NULL, name, other, attr_path,
4041                              null, two, &msg, o, p);
4042                 free(null);
4043         }
4044         else
4045                 run_diff_cmd(pgm, name, other, attr_path,
4046                              one, two, &msg, o, p);
4047
4048         strbuf_release(&msg);
4049 }
4050
4051 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4052                          struct diffstat_t *diffstat)
4053 {
4054         const char *name;
4055         const char *other;
4056
4057         if (DIFF_PAIR_UNMERGED(p)) {
4058                 /* unmerged */
4059                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4060                 return;
4061         }
4062
4063         name = p->one->path;
4064         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4065
4066         if (o->prefix_length)
4067                 strip_prefix(o->prefix_length, &name, &other);
4068
4069         diff_fill_oid_info(p->one);
4070         diff_fill_oid_info(p->two);
4071
4072         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4073 }
4074
4075 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4076 {
4077         const char *name;
4078         const char *other;
4079         const char *attr_path;
4080
4081         if (DIFF_PAIR_UNMERGED(p)) {
4082                 /* unmerged */
4083                 return;
4084         }
4085
4086         name = p->one->path;
4087         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4088         attr_path = other ? other : name;
4089
4090         if (o->prefix_length)
4091                 strip_prefix(o->prefix_length, &name, &other);
4092
4093         diff_fill_oid_info(p->one);
4094         diff_fill_oid_info(p->two);
4095
4096         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4097 }
4098
4099 void diff_setup(struct diff_options *options)
4100 {
4101         memcpy(options, &default_diff_options, sizeof(*options));
4102
4103         options->file = stdout;
4104
4105         options->abbrev = DEFAULT_ABBREV;
4106         options->line_termination = '\n';
4107         options->break_opt = -1;
4108         options->rename_limit = -1;
4109         options->dirstat_permille = diff_dirstat_permille_default;
4110         options->context = diff_context_default;
4111         options->interhunkcontext = diff_interhunk_context_default;
4112         options->ws_error_highlight = ws_error_highlight_default;
4113         options->flags.rename_empty = 1;
4114         options->objfind = NULL;
4115
4116         /* pathchange left =NULL by default */
4117         options->change = diff_change;
4118         options->add_remove = diff_addremove;
4119         options->use_color = diff_use_color_default;
4120         options->detect_rename = diff_detect_rename_default;
4121         options->xdl_opts |= diff_algorithm;
4122         if (diff_indent_heuristic)
4123                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4124
4125         options->orderfile = diff_order_file_cfg;
4126
4127         if (diff_no_prefix) {
4128                 options->a_prefix = options->b_prefix = "";
4129         } else if (!diff_mnemonic_prefix) {
4130                 options->a_prefix = "a/";
4131                 options->b_prefix = "b/";
4132         }
4133
4134         options->color_moved = diff_color_moved_default;
4135 }
4136
4137 void diff_setup_done(struct diff_options *options)
4138 {
4139         unsigned check_mask = DIFF_FORMAT_NAME |
4140                               DIFF_FORMAT_NAME_STATUS |
4141                               DIFF_FORMAT_CHECKDIFF |
4142                               DIFF_FORMAT_NO_OUTPUT;
4143
4144         if (options->set_default)
4145                 options->set_default(options);
4146
4147         if (HAS_MULTI_BITS(options->output_format & check_mask))
4148                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4149
4150         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4151                 die(_("-G, -S and --find-object are mutually exclusive"));
4152
4153         /*
4154          * Most of the time we can say "there are changes"
4155          * only by checking if there are changed paths, but
4156          * --ignore-whitespace* options force us to look
4157          * inside contents.
4158          */
4159
4160         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4161                 options->flags.diff_from_contents = 1;
4162         else
4163                 options->flags.diff_from_contents = 0;
4164
4165         if (options->flags.find_copies_harder)
4166                 options->detect_rename = DIFF_DETECT_COPY;
4167
4168         if (!options->flags.relative_name)
4169                 options->prefix = NULL;
4170         if (options->prefix)
4171                 options->prefix_length = strlen(options->prefix);
4172         else
4173                 options->prefix_length = 0;
4174
4175         if (options->output_format & (DIFF_FORMAT_NAME |
4176                                       DIFF_FORMAT_NAME_STATUS |
4177                                       DIFF_FORMAT_CHECKDIFF |
4178                                       DIFF_FORMAT_NO_OUTPUT))
4179                 options->output_format &= ~(DIFF_FORMAT_RAW |
4180                                             DIFF_FORMAT_NUMSTAT |
4181                                             DIFF_FORMAT_DIFFSTAT |
4182                                             DIFF_FORMAT_SHORTSTAT |
4183                                             DIFF_FORMAT_DIRSTAT |
4184                                             DIFF_FORMAT_SUMMARY |
4185                                             DIFF_FORMAT_PATCH);
4186
4187         /*
4188          * These cases always need recursive; we do not drop caller-supplied
4189          * recursive bits for other formats here.
4190          */
4191         if (options->output_format & (DIFF_FORMAT_PATCH |
4192                                       DIFF_FORMAT_NUMSTAT |
4193                                       DIFF_FORMAT_DIFFSTAT |
4194                                       DIFF_FORMAT_SHORTSTAT |
4195                                       DIFF_FORMAT_DIRSTAT |
4196                                       DIFF_FORMAT_SUMMARY |
4197                                       DIFF_FORMAT_CHECKDIFF))
4198                 options->flags.recursive = 1;
4199         /*
4200          * Also pickaxe would not work very well if you do not say recursive
4201          */
4202         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4203                 options->flags.recursive = 1;
4204         /*
4205          * When patches are generated, submodules diffed against the work tree
4206          * must be checked for dirtiness too so it can be shown in the output
4207          */
4208         if (options->output_format & DIFF_FORMAT_PATCH)
4209                 options->flags.dirty_submodules = 1;
4210
4211         if (options->detect_rename && options->rename_limit < 0)
4212                 options->rename_limit = diff_rename_limit_default;
4213         if (options->setup & DIFF_SETUP_USE_CACHE) {
4214                 if (!active_cache)
4215                         /* read-cache does not die even when it fails
4216                          * so it is safe for us to do this here.  Also
4217                          * it does not smudge active_cache or active_nr
4218                          * when it fails, so we do not have to worry about
4219                          * cleaning it up ourselves either.
4220                          */
4221                         read_cache();
4222         }
4223         if (40 < options->abbrev)
4224                 options->abbrev = 40; /* full */
4225
4226         /*
4227          * It does not make sense to show the first hit we happened
4228          * to have found.  It does not make sense not to return with
4229          * exit code in such a case either.
4230          */
4231         if (options->flags.quick) {
4232                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4233                 options->flags.exit_with_status = 1;
4234         }
4235
4236         options->diff_path_counter = 0;
4237
4238         if (options->flags.follow_renames && options->pathspec.nr != 1)
4239                 die(_("--follow requires exactly one pathspec"));
4240
4241         if (!options->use_color || external_diff())
4242                 options->color_moved = 0;
4243 }
4244
4245 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4246 {
4247         char c, *eq;
4248         int len;
4249
4250         if (*arg != '-')
4251                 return 0;
4252         c = *++arg;
4253         if (!c)
4254                 return 0;
4255         if (c == arg_short) {
4256                 c = *++arg;
4257                 if (!c)
4258                         return 1;
4259                 if (val && isdigit(c)) {
4260                         char *end;
4261                         int n = strtoul(arg, &end, 10);
4262                         if (*end)
4263                                 return 0;
4264                         *val = n;
4265                         return 1;
4266                 }
4267                 return 0;
4268         }
4269         if (c != '-')
4270                 return 0;
4271         arg++;
4272         eq = strchrnul(arg, '=');
4273         len = eq - arg;
4274         if (!len || strncmp(arg, arg_long, len))
4275                 return 0;
4276         if (*eq) {
4277                 int n;
4278                 char *end;
4279                 if (!isdigit(*++eq))
4280                         return 0;
4281                 n = strtoul(eq, &end, 10);
4282                 if (*end)
4283                         return 0;
4284                 *val = n;
4285         }
4286         return 1;
4287 }
4288
4289 static int diff_scoreopt_parse(const char *opt);
4290
4291 static inline int short_opt(char opt, const char **argv,
4292                             const char **optarg)
4293 {
4294         const char *arg = argv[0];
4295         if (arg[0] != '-' || arg[1] != opt)
4296                 return 0;
4297         if (arg[2] != '\0') {
4298                 *optarg = arg + 2;
4299                 return 1;
4300         }
4301         if (!argv[1])
4302                 die("Option '%c' requires a value", opt);
4303         *optarg = argv[1];
4304         return 2;
4305 }
4306
4307 int parse_long_opt(const char *opt, const char **argv,
4308                    const char **optarg)
4309 {
4310         const char *arg = argv[0];
4311         if (!skip_prefix(arg, "--", &arg))
4312                 return 0;
4313         if (!skip_prefix(arg, opt, &arg))
4314                 return 0;
4315         if (*arg == '=') { /* stuck form: --option=value */
4316                 *optarg = arg + 1;
4317                 return 1;
4318         }
4319         if (*arg != '\0')
4320                 return 0;
4321         /* separate form: --option value */
4322         if (!argv[1])
4323                 die("Option '--%s' requires a value", opt);
4324         *optarg = argv[1];
4325         return 2;
4326 }
4327
4328 static int stat_opt(struct diff_options *options, const char **av)
4329 {
4330         const char *arg = av[0];
4331         char *end;
4332         int width = options->stat_width;
4333         int name_width = options->stat_name_width;
4334         int graph_width = options->stat_graph_width;
4335         int count = options->stat_count;
4336         int argcount = 1;
4337
4338         if (!skip_prefix(arg, "--stat", &arg))
4339                 die("BUG: stat option does not begin with --stat: %s", arg);
4340         end = (char *)arg;
4341
4342         switch (*arg) {
4343         case '-':
4344                 if (skip_prefix(arg, "-width", &arg)) {
4345                         if (*arg == '=')
4346                                 width = strtoul(arg + 1, &end, 10);
4347                         else if (!*arg && !av[1])
4348                                 die_want_option("--stat-width");
4349                         else if (!*arg) {
4350                                 width = strtoul(av[1], &end, 10);
4351                                 argcount = 2;
4352                         }
4353                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4354                         if (*arg == '=')
4355                                 name_width = strtoul(arg + 1, &end, 10);
4356                         else if (!*arg && !av[1])
4357                                 die_want_option("--stat-name-width");
4358                         else if (!*arg) {
4359                                 name_width = strtoul(av[1], &end, 10);
4360                                 argcount = 2;
4361                         }
4362                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4363                         if (*arg == '=')
4364                                 graph_width = strtoul(arg + 1, &end, 10);
4365                         else if (!*arg && !av[1])
4366                                 die_want_option("--stat-graph-width");
4367                         else if (!*arg) {
4368                                 graph_width = strtoul(av[1], &end, 10);
4369                                 argcount = 2;
4370                         }
4371                 } else if (skip_prefix(arg, "-count", &arg)) {
4372                         if (*arg == '=')
4373                                 count = strtoul(arg + 1, &end, 10);
4374                         else if (!*arg && !av[1])
4375                                 die_want_option("--stat-count");
4376                         else if (!*arg) {
4377                                 count = strtoul(av[1], &end, 10);
4378                                 argcount = 2;
4379                         }
4380                 }
4381                 break;
4382         case '=':
4383                 width = strtoul(arg+1, &end, 10);
4384                 if (*end == ',')
4385                         name_width = strtoul(end+1, &end, 10);
4386                 if (*end == ',')
4387                         count = strtoul(end+1, &end, 10);
4388         }
4389
4390         /* Important! This checks all the error cases! */
4391         if (*end)
4392                 return 0;
4393         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4394         options->stat_name_width = name_width;
4395         options->stat_graph_width = graph_width;
4396         options->stat_width = width;
4397         options->stat_count = count;
4398         return argcount;
4399 }
4400
4401 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4402 {
4403         struct strbuf errmsg = STRBUF_INIT;
4404         if (parse_dirstat_params(options, params, &errmsg))
4405                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4406                     errmsg.buf);
4407         strbuf_release(&errmsg);
4408         /*
4409          * The caller knows a dirstat-related option is given from the command
4410          * line; allow it to say "return this_function();"
4411          */
4412         options->output_format |= DIFF_FORMAT_DIRSTAT;
4413         return 1;
4414 }
4415
4416 static int parse_submodule_opt(struct diff_options *options, const char *value)
4417 {
4418         if (parse_submodule_params(options, value))
4419                 die(_("Failed to parse --submodule option parameter: '%s'"),
4420                         value);
4421         return 1;
4422 }
4423
4424 static const char diff_status_letters[] = {
4425         DIFF_STATUS_ADDED,
4426         DIFF_STATUS_COPIED,
4427         DIFF_STATUS_DELETED,
4428         DIFF_STATUS_MODIFIED,
4429         DIFF_STATUS_RENAMED,
4430         DIFF_STATUS_TYPE_CHANGED,
4431         DIFF_STATUS_UNKNOWN,
4432         DIFF_STATUS_UNMERGED,
4433         DIFF_STATUS_FILTER_AON,
4434         DIFF_STATUS_FILTER_BROKEN,
4435         '\0',
4436 };
4437
4438 static unsigned int filter_bit['Z' + 1];
4439
4440 static void prepare_filter_bits(void)
4441 {
4442         int i;
4443
4444         if (!filter_bit[DIFF_STATUS_ADDED]) {
4445                 for (i = 0; diff_status_letters[i]; i++)
4446                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4447         }
4448 }
4449
4450 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4451 {
4452         return opt->filter & filter_bit[(int) status];
4453 }
4454
4455 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4456 {
4457         int i, optch;
4458
4459         prepare_filter_bits();
4460
4461         /*
4462          * If there is a negation e.g. 'd' in the input, and we haven't
4463          * initialized the filter field with another --diff-filter, start
4464          * from full set of bits, except for AON.
4465          */
4466         if (!opt->filter) {
4467                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4468                         if (optch < 'a' || 'z' < optch)
4469                                 continue;
4470                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4471                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4472                         break;
4473                 }
4474         }
4475
4476         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4477                 unsigned int bit;
4478                 int negate;
4479
4480                 if ('a' <= optch && optch <= 'z') {
4481                         negate = 1;
4482                         optch = toupper(optch);
4483                 } else {
4484                         negate = 0;
4485                 }
4486
4487                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4488                 if (!bit)
4489                         return optarg[i];
4490                 if (negate)
4491                         opt->filter &= ~bit;
4492                 else
4493                         opt->filter |= bit;
4494         }
4495         return 0;
4496 }
4497
4498 static void enable_patch_output(int *fmt) {
4499         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4500         *fmt |= DIFF_FORMAT_PATCH;
4501 }
4502
4503 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4504 {
4505         int val = parse_ws_error_highlight(arg);
4506
4507         if (val < 0) {
4508                 error("unknown value after ws-error-highlight=%.*s",
4509                       -1 - val, arg);
4510                 return 0;
4511         }
4512         opt->ws_error_highlight = val;
4513         return 1;
4514 }
4515
4516 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4517 {
4518         struct object_id oid;
4519
4520         if (get_oid(arg, &oid))
4521                 return error("unable to resolve '%s'", arg);
4522
4523         if (!opt->objfind)
4524                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4525
4526         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4527         opt->flags.recursive = 1;
4528         opt->flags.tree_in_recursive = 1;
4529         oidset_insert(opt->objfind, &oid);
4530         return 1;
4531 }
4532
4533 int diff_opt_parse(struct diff_options *options,
4534                    const char **av, int ac, const char *prefix)
4535 {
4536         const char *arg = av[0];
4537         const char *optarg;
4538         int argcount;
4539
4540         if (!prefix)
4541                 prefix = "";
4542
4543         /* Output format options */
4544         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4545             || opt_arg(arg, 'U', "unified", &options->context))
4546                 enable_patch_output(&options->output_format);
4547         else if (!strcmp(arg, "--raw"))
4548                 options->output_format |= DIFF_FORMAT_RAW;
4549         else if (!strcmp(arg, "--patch-with-raw")) {
4550                 enable_patch_output(&options->output_format);
4551                 options->output_format |= DIFF_FORMAT_RAW;
4552         } else if (!strcmp(arg, "--numstat"))
4553                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4554         else if (!strcmp(arg, "--shortstat"))
4555                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4556         else if (skip_prefix(arg, "-X", &arg) ||
4557                  skip_to_optional_arg(arg, "--dirstat", &arg))
4558                 return parse_dirstat_opt(options, arg);
4559         else if (!strcmp(arg, "--cumulative"))
4560                 return parse_dirstat_opt(options, "cumulative");
4561         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4562                 parse_dirstat_opt(options, "files");
4563                 return parse_dirstat_opt(options, arg);
4564         }
4565         else if (!strcmp(arg, "--check"))
4566                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4567         else if (!strcmp(arg, "--summary"))
4568                 options->output_format |= DIFF_FORMAT_SUMMARY;
4569         else if (!strcmp(arg, "--patch-with-stat")) {
4570                 enable_patch_output(&options->output_format);
4571                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4572         } else if (!strcmp(arg, "--name-only"))
4573                 options->output_format |= DIFF_FORMAT_NAME;
4574         else if (!strcmp(arg, "--name-status"))
4575                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4576         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4577                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4578         else if (starts_with(arg, "--stat"))
4579                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4580                 return stat_opt(options, av);
4581         else if (!strcmp(arg, "--compact-summary")) {
4582                  options->flags.stat_with_summary = 1;
4583                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4584         } else if (!strcmp(arg, "--no-compact-summary"))
4585                  options->flags.stat_with_summary = 0;
4586
4587         /* renames options */
4588         else if (starts_with(arg, "-B") ||
4589                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4590                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4591                         return error("invalid argument to -B: %s", arg+2);
4592         }
4593         else if (starts_with(arg, "-M") ||
4594                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4595                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4596                         return error("invalid argument to -M: %s", arg+2);
4597                 options->detect_rename = DIFF_DETECT_RENAME;
4598         }
4599         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4600                 options->irreversible_delete = 1;
4601         }
4602         else if (starts_with(arg, "-C") ||
4603                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4604                 if (options->detect_rename == DIFF_DETECT_COPY)
4605                         options->flags.find_copies_harder = 1;
4606                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4607                         return error("invalid argument to -C: %s", arg+2);
4608                 options->detect_rename = DIFF_DETECT_COPY;
4609         }
4610         else if (!strcmp(arg, "--no-renames"))
4611                 options->detect_rename = 0;
4612         else if (!strcmp(arg, "--rename-empty"))
4613                 options->flags.rename_empty = 1;
4614         else if (!strcmp(arg, "--no-rename-empty"))
4615                 options->flags.rename_empty = 0;
4616         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4617                 options->flags.relative_name = 1;
4618                 if (arg)
4619                         options->prefix = arg;
4620         }
4621
4622         /* xdiff options */
4623         else if (!strcmp(arg, "--minimal"))
4624                 DIFF_XDL_SET(options, NEED_MINIMAL);
4625         else if (!strcmp(arg, "--no-minimal"))
4626                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4627         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4628                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4629         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4630                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4631         else if (!strcmp(arg, "--ignore-space-at-eol"))
4632                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4633         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4634                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4635         else if (!strcmp(arg, "--ignore-blank-lines"))
4636                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4637         else if (!strcmp(arg, "--indent-heuristic"))
4638                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4639         else if (!strcmp(arg, "--no-indent-heuristic"))
4640                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4641         else if (!strcmp(arg, "--patience")) {
4642                 int i;
4643                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4644                 /*
4645                  * Both --patience and --anchored use PATIENCE_DIFF
4646                  * internally, so remove any anchors previously
4647                  * specified.
4648                  */
4649                 for (i = 0; i < options->anchors_nr; i++)
4650                         free(options->anchors[i]);
4651                 options->anchors_nr = 0;
4652         } else if (!strcmp(arg, "--histogram"))
4653                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4654         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4655                 long value = parse_algorithm_value(optarg);
4656                 if (value < 0)
4657                         return error("option diff-algorithm accepts \"myers\", "
4658                                      "\"minimal\", \"patience\" and \"histogram\"");
4659                 /* clear out previous settings */
4660                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4661                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4662                 options->xdl_opts |= value;
4663                 return argcount;
4664         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4665                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4666                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4667                            options->anchors_alloc);
4668                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4669         }
4670
4671         /* flags options */
4672         else if (!strcmp(arg, "--binary")) {
4673                 enable_patch_output(&options->output_format);
4674                 options->flags.binary = 1;
4675         }
4676         else if (!strcmp(arg, "--full-index"))
4677                 options->flags.full_index = 1;
4678         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4679                 options->flags.text = 1;
4680         else if (!strcmp(arg, "-R"))
4681                 options->flags.reverse_diff = 1;
4682         else if (!strcmp(arg, "--find-copies-harder"))
4683                 options->flags.find_copies_harder = 1;
4684         else if (!strcmp(arg, "--follow"))
4685                 options->flags.follow_renames = 1;
4686         else if (!strcmp(arg, "--no-follow")) {
4687                 options->flags.follow_renames = 0;
4688                 options->flags.default_follow_renames = 0;
4689         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4690                 int value = git_config_colorbool(NULL, arg);
4691                 if (value < 0)
4692                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4693                 options->use_color = value;
4694         }
4695         else if (!strcmp(arg, "--no-color"))
4696                 options->use_color = 0;
4697         else if (!strcmp(arg, "--color-moved")) {
4698                 if (diff_color_moved_default)
4699                         options->color_moved = diff_color_moved_default;
4700                 if (options->color_moved == COLOR_MOVED_NO)
4701                         options->color_moved = COLOR_MOVED_DEFAULT;
4702         } else if (!strcmp(arg, "--no-color-moved"))
4703                 options->color_moved = COLOR_MOVED_NO;
4704         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4705                 int cm = parse_color_moved(arg);
4706                 if (cm < 0)
4707                         die("bad --color-moved argument: %s", arg);
4708                 options->color_moved = cm;
4709         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4710                 options->use_color = 1;
4711                 options->word_diff = DIFF_WORDS_COLOR;
4712         }
4713         else if (!strcmp(arg, "--word-diff")) {
4714                 if (options->word_diff == DIFF_WORDS_NONE)
4715                         options->word_diff = DIFF_WORDS_PLAIN;
4716         }
4717         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4718                 if (!strcmp(arg, "plain"))
4719                         options->word_diff = DIFF_WORDS_PLAIN;
4720                 else if (!strcmp(arg, "color")) {
4721                         options->use_color = 1;
4722                         options->word_diff = DIFF_WORDS_COLOR;
4723                 }
4724                 else if (!strcmp(arg, "porcelain"))
4725                         options->word_diff = DIFF_WORDS_PORCELAIN;
4726                 else if (!strcmp(arg, "none"))
4727                         options->word_diff = DIFF_WORDS_NONE;
4728                 else
4729                         die("bad --word-diff argument: %s", arg);
4730         }
4731         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4732                 if (options->word_diff == DIFF_WORDS_NONE)
4733                         options->word_diff = DIFF_WORDS_PLAIN;
4734                 options->word_regex = optarg;
4735                 return argcount;
4736         }
4737         else if (!strcmp(arg, "--exit-code"))
4738                 options->flags.exit_with_status = 1;
4739         else if (!strcmp(arg, "--quiet"))
4740                 options->flags.quick = 1;
4741         else if (!strcmp(arg, "--ext-diff"))
4742                 options->flags.allow_external = 1;
4743         else if (!strcmp(arg, "--no-ext-diff"))
4744                 options->flags.allow_external = 0;
4745         else if (!strcmp(arg, "--textconv")) {
4746                 options->flags.allow_textconv = 1;
4747                 options->flags.textconv_set_via_cmdline = 1;
4748         } else if (!strcmp(arg, "--no-textconv"))
4749                 options->flags.allow_textconv = 0;
4750         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
4751                 options->flags.override_submodule_config = 1;
4752                 handle_ignore_submodules_arg(options, arg);
4753         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
4754                 return parse_submodule_opt(options, arg);
4755         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4756                 return parse_ws_error_highlight_opt(options, arg);
4757         else if (!strcmp(arg, "--ita-invisible-in-index"))
4758                 options->ita_invisible_in_index = 1;
4759         else if (!strcmp(arg, "--ita-visible-in-index"))
4760                 options->ita_invisible_in_index = 0;
4761
4762         /* misc options */
4763         else if (!strcmp(arg, "-z"))
4764                 options->line_termination = 0;
4765         else if ((argcount = short_opt('l', av, &optarg))) {
4766                 options->rename_limit = strtoul(optarg, NULL, 10);
4767                 return argcount;
4768         }
4769         else if ((argcount = short_opt('S', av, &optarg))) {
4770                 options->pickaxe = optarg;
4771                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4772                 return argcount;
4773         } else if ((argcount = short_opt('G', av, &optarg))) {
4774                 options->pickaxe = optarg;
4775                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4776                 return argcount;
4777         }
4778         else if (!strcmp(arg, "--pickaxe-all"))
4779                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4780         else if (!strcmp(arg, "--pickaxe-regex"))
4781                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4782         else if ((argcount = short_opt('O', av, &optarg))) {
4783                 options->orderfile = prefix_filename(prefix, optarg);
4784                 return argcount;
4785         } else if (skip_prefix(arg, "--find-object=", &arg))
4786                 return parse_objfind_opt(options, arg);
4787         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4788                 int offending = parse_diff_filter_opt(optarg, options);
4789                 if (offending)
4790                         die("unknown change class '%c' in --diff-filter=%s",
4791                             offending, optarg);
4792                 return argcount;
4793         }
4794         else if (!strcmp(arg, "--no-abbrev"))
4795                 options->abbrev = 0;
4796         else if (!strcmp(arg, "--abbrev"))
4797                 options->abbrev = DEFAULT_ABBREV;
4798         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4799                 options->abbrev = strtoul(arg, NULL, 10);
4800                 if (options->abbrev < MINIMUM_ABBREV)
4801                         options->abbrev = MINIMUM_ABBREV;
4802                 else if (40 < options->abbrev)
4803                         options->abbrev = 40;
4804         }
4805         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4806                 options->a_prefix = optarg;
4807                 return argcount;
4808         }
4809         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4810                 options->line_prefix = optarg;
4811                 options->line_prefix_length = strlen(options->line_prefix);
4812                 graph_setup_line_prefix(options);
4813                 return argcount;
4814         }
4815         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4816                 options->b_prefix = optarg;
4817                 return argcount;
4818         }
4819         else if (!strcmp(arg, "--no-prefix"))
4820                 options->a_prefix = options->b_prefix = "";
4821         else if (opt_arg(arg, '\0', "inter-hunk-context",
4822                          &options->interhunkcontext))
4823                 ;
4824         else if (!strcmp(arg, "-W"))
4825                 options->flags.funccontext = 1;
4826         else if (!strcmp(arg, "--function-context"))
4827                 options->flags.funccontext = 1;
4828         else if (!strcmp(arg, "--no-function-context"))
4829                 options->flags.funccontext = 0;
4830         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4831                 char *path = prefix_filename(prefix, optarg);
4832                 options->file = xfopen(path, "w");
4833                 options->close_file = 1;
4834                 if (options->use_color != GIT_COLOR_ALWAYS)
4835                         options->use_color = GIT_COLOR_NEVER;
4836                 free(path);
4837                 return argcount;
4838         } else
4839                 return 0;
4840         return 1;
4841 }
4842
4843 int parse_rename_score(const char **cp_p)
4844 {
4845         unsigned long num, scale;
4846         int ch, dot;
4847         const char *cp = *cp_p;
4848
4849         num = 0;
4850         scale = 1;
4851         dot = 0;
4852         for (;;) {
4853                 ch = *cp;
4854                 if ( !dot && ch == '.' ) {
4855                         scale = 1;
4856                         dot = 1;
4857                 } else if ( ch == '%' ) {
4858                         scale = dot ? scale*100 : 100;
4859                         cp++;   /* % is always at the end */
4860                         break;
4861                 } else if ( ch >= '0' && ch <= '9' ) {
4862                         if ( scale < 100000 ) {
4863                                 scale *= 10;
4864                                 num = (num*10) + (ch-'0');
4865                         }
4866                 } else {
4867                         break;
4868                 }
4869                 cp++;
4870         }
4871         *cp_p = cp;
4872
4873         /* user says num divided by scale and we say internally that
4874          * is MAX_SCORE * num / scale.
4875          */
4876         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4877 }
4878
4879 static int diff_scoreopt_parse(const char *opt)
4880 {
4881         int opt1, opt2, cmd;
4882
4883         if (*opt++ != '-')
4884                 return -1;
4885         cmd = *opt++;
4886         if (cmd == '-') {
4887                 /* convert the long-form arguments into short-form versions */
4888                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4889                         if (*opt == 0 || *opt++ == '=')
4890                                 cmd = 'B';
4891                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4892                         if (*opt == 0 || *opt++ == '=')
4893                                 cmd = 'C';
4894                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4895                         if (*opt == 0 || *opt++ == '=')
4896                                 cmd = 'M';
4897                 }
4898         }
4899         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4900                 return -1; /* that is not a -M, -C, or -B option */
4901
4902         opt1 = parse_rename_score(&opt);
4903         if (cmd != 'B')
4904                 opt2 = 0;
4905         else {
4906                 if (*opt == 0)
4907                         opt2 = 0;
4908                 else if (*opt != '/')
4909                         return -1; /* we expect -B80/99 or -B80 */
4910                 else {
4911                         opt++;
4912                         opt2 = parse_rename_score(&opt);
4913                 }
4914         }
4915         if (*opt != 0)
4916                 return -1;
4917         return opt1 | (opt2 << 16);
4918 }
4919
4920 struct diff_queue_struct diff_queued_diff;
4921
4922 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4923 {
4924         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4925         queue->queue[queue->nr++] = dp;
4926 }
4927
4928 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4929                                  struct diff_filespec *one,
4930                                  struct diff_filespec *two)
4931 {
4932         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4933         dp->one = one;
4934         dp->two = two;
4935         if (queue)
4936                 diff_q(queue, dp);
4937         return dp;
4938 }
4939
4940 void diff_free_filepair(struct diff_filepair *p)
4941 {
4942         free_filespec(p->one);
4943         free_filespec(p->two);
4944         free(p);
4945 }
4946
4947 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4948 {
4949         int abblen;
4950         const char *abbrev;
4951
4952         /* Do we want all 40 hex characters? */
4953         if (len == GIT_SHA1_HEXSZ)
4954                 return oid_to_hex(oid);
4955
4956         /* An abbreviated value is fine, possibly followed by an ellipsis. */
4957         abbrev = diff_abbrev_oid(oid, len);
4958
4959         if (!print_sha1_ellipsis())
4960                 return abbrev;
4961
4962         abblen = strlen(abbrev);
4963
4964         /*
4965          * In well-behaved cases, where the abbreviated result is the
4966          * same as the requested length, append three dots after the
4967          * abbreviation (hence the whole logic is limited to the case
4968          * where abblen < 37); when the actual abbreviated result is a
4969          * bit longer than the requested length, we reduce the number
4970          * of dots so that they match the well-behaved ones.  However,
4971          * if the actual abbreviation is longer than the requested
4972          * length by more than three, we give up on aligning, and add
4973          * three dots anyway, to indicate that the output is not the
4974          * full object name.  Yes, this may be suboptimal, but this
4975          * appears only in "diff --raw --abbrev" output and it is not
4976          * worth the effort to change it now.  Note that this would
4977          * likely to work fine when the automatic sizing of default
4978          * abbreviation length is used--we would be fed -1 in "len" in
4979          * that case, and will end up always appending three-dots, but
4980          * the automatic sizing is supposed to give abblen that ensures
4981          * uniqueness across all objects (statistically speaking).
4982          */
4983         if (abblen < GIT_SHA1_HEXSZ - 3) {
4984                 static char hex[GIT_MAX_HEXSZ + 1];
4985                 if (len < abblen && abblen <= len + 2)
4986                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4987                 else
4988                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4989                 return hex;
4990         }
4991
4992         return oid_to_hex(oid);
4993 }
4994
4995 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4996 {
4997         int line_termination = opt->line_termination;
4998         int inter_name_termination = line_termination ? '\t' : '\0';
4999
5000         fprintf(opt->file, "%s", diff_line_prefix(opt));
5001         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5002                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5003                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5004                 fprintf(opt->file, "%s ",
5005                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5006         }
5007         if (p->score) {
5008                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5009                         inter_name_termination);
5010         } else {
5011                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5012         }
5013
5014         if (p->status == DIFF_STATUS_COPIED ||
5015             p->status == DIFF_STATUS_RENAMED) {
5016                 const char *name_a, *name_b;
5017                 name_a = p->one->path;
5018                 name_b = p->two->path;
5019                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5020                 write_name_quoted(name_a, opt->file, inter_name_termination);
5021                 write_name_quoted(name_b, opt->file, line_termination);
5022         } else {
5023                 const char *name_a, *name_b;
5024                 name_a = p->one->mode ? p->one->path : p->two->path;
5025                 name_b = NULL;
5026                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5027                 write_name_quoted(name_a, opt->file, line_termination);
5028         }
5029 }
5030
5031 int diff_unmodified_pair(struct diff_filepair *p)
5032 {
5033         /* This function is written stricter than necessary to support
5034          * the currently implemented transformers, but the idea is to
5035          * let transformers to produce diff_filepairs any way they want,
5036          * and filter and clean them up here before producing the output.
5037          */
5038         struct diff_filespec *one = p->one, *two = p->two;
5039
5040         if (DIFF_PAIR_UNMERGED(p))
5041                 return 0; /* unmerged is interesting */
5042
5043         /* deletion, addition, mode or type change
5044          * and rename are all interesting.
5045          */
5046         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5047             DIFF_PAIR_MODE_CHANGED(p) ||
5048             strcmp(one->path, two->path))
5049                 return 0;
5050
5051         /* both are valid and point at the same path.  that is, we are
5052          * dealing with a change.
5053          */
5054         if (one->oid_valid && two->oid_valid &&
5055             !oidcmp(&one->oid, &two->oid) &&
5056             !one->dirty_submodule && !two->dirty_submodule)
5057                 return 1; /* no change */
5058         if (!one->oid_valid && !two->oid_valid)
5059                 return 1; /* both look at the same file on the filesystem. */
5060         return 0;
5061 }
5062
5063 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5064 {
5065         if (diff_unmodified_pair(p))
5066                 return;
5067
5068         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5069             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5070                 return; /* no tree diffs in patch format */
5071
5072         run_diff(p, o);
5073 }
5074
5075 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5076                             struct diffstat_t *diffstat)
5077 {
5078         if (diff_unmodified_pair(p))
5079                 return;
5080
5081         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5082             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5083                 return; /* no useful stat for tree diffs */
5084
5085         run_diffstat(p, o, diffstat);
5086 }
5087
5088 static void diff_flush_checkdiff(struct diff_filepair *p,
5089                 struct diff_options *o)
5090 {
5091         if (diff_unmodified_pair(p))
5092                 return;
5093
5094         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5095             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5096                 return; /* nothing to check in tree diffs */
5097
5098         run_checkdiff(p, o);
5099 }
5100
5101 int diff_queue_is_empty(void)
5102 {
5103         struct diff_queue_struct *q = &diff_queued_diff;
5104         int i;
5105         for (i = 0; i < q->nr; i++)
5106                 if (!diff_unmodified_pair(q->queue[i]))
5107                         return 0;
5108         return 1;
5109 }
5110
5111 #if DIFF_DEBUG
5112 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5113 {
5114         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5115                 x, one ? one : "",
5116                 s->path,
5117                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5118                 s->mode,
5119                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5120         fprintf(stderr, "queue[%d] %s size %lu\n",
5121                 x, one ? one : "",
5122                 s->size);
5123 }
5124
5125 void diff_debug_filepair(const struct diff_filepair *p, int i)
5126 {
5127         diff_debug_filespec(p->one, i, "one");
5128         diff_debug_filespec(p->two, i, "two");
5129         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5130                 p->score, p->status ? p->status : '?',
5131                 p->one->rename_used, p->broken_pair);
5132 }
5133
5134 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5135 {
5136         int i;
5137         if (msg)
5138                 fprintf(stderr, "%s\n", msg);
5139         fprintf(stderr, "q->nr = %d\n", q->nr);
5140         for (i = 0; i < q->nr; i++) {
5141                 struct diff_filepair *p = q->queue[i];
5142                 diff_debug_filepair(p, i);
5143         }
5144 }
5145 #endif
5146
5147 static void diff_resolve_rename_copy(void)
5148 {
5149         int i;
5150         struct diff_filepair *p;
5151         struct diff_queue_struct *q = &diff_queued_diff;
5152
5153         diff_debug_queue("resolve-rename-copy", q);
5154
5155         for (i = 0; i < q->nr; i++) {
5156                 p = q->queue[i];
5157                 p->status = 0; /* undecided */
5158                 if (DIFF_PAIR_UNMERGED(p))
5159                         p->status = DIFF_STATUS_UNMERGED;
5160                 else if (!DIFF_FILE_VALID(p->one))
5161                         p->status = DIFF_STATUS_ADDED;
5162                 else if (!DIFF_FILE_VALID(p->two))
5163                         p->status = DIFF_STATUS_DELETED;
5164                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5165                         p->status = DIFF_STATUS_TYPE_CHANGED;
5166
5167                 /* from this point on, we are dealing with a pair
5168                  * whose both sides are valid and of the same type, i.e.
5169                  * either in-place edit or rename/copy edit.
5170                  */
5171                 else if (DIFF_PAIR_RENAME(p)) {
5172                         /*
5173                          * A rename might have re-connected a broken
5174                          * pair up, causing the pathnames to be the
5175                          * same again. If so, that's not a rename at
5176                          * all, just a modification..
5177                          *
5178                          * Otherwise, see if this source was used for
5179                          * multiple renames, in which case we decrement
5180                          * the count, and call it a copy.
5181                          */
5182                         if (!strcmp(p->one->path, p->two->path))
5183                                 p->status = DIFF_STATUS_MODIFIED;
5184                         else if (--p->one->rename_used > 0)
5185                                 p->status = DIFF_STATUS_COPIED;
5186                         else
5187                                 p->status = DIFF_STATUS_RENAMED;
5188                 }
5189                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5190                          p->one->mode != p->two->mode ||
5191                          p->one->dirty_submodule ||
5192                          p->two->dirty_submodule ||
5193                          is_null_oid(&p->one->oid))
5194                         p->status = DIFF_STATUS_MODIFIED;
5195                 else {
5196                         /* This is a "no-change" entry and should not
5197                          * happen anymore, but prepare for broken callers.
5198                          */
5199                         error("feeding unmodified %s to diffcore",
5200                               p->one->path);
5201                         p->status = DIFF_STATUS_UNKNOWN;
5202                 }
5203         }
5204         diff_debug_queue("resolve-rename-copy done", q);
5205 }
5206
5207 static int check_pair_status(struct diff_filepair *p)
5208 {
5209         switch (p->status) {
5210         case DIFF_STATUS_UNKNOWN:
5211                 return 0;
5212         case 0:
5213                 die("internal error in diff-resolve-rename-copy");
5214         default:
5215                 return 1;
5216         }
5217 }
5218
5219 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5220 {
5221         int fmt = opt->output_format;
5222
5223         if (fmt & DIFF_FORMAT_CHECKDIFF)
5224                 diff_flush_checkdiff(p, opt);
5225         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5226                 diff_flush_raw(p, opt);
5227         else if (fmt & DIFF_FORMAT_NAME) {
5228                 const char *name_a, *name_b;
5229                 name_a = p->two->path;
5230                 name_b = NULL;
5231                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5232                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5233                 write_name_quoted(name_a, opt->file, opt->line_termination);
5234         }
5235 }
5236
5237 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5238 {
5239         struct strbuf sb = STRBUF_INIT;
5240         if (fs->mode)
5241                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5242         else
5243                 strbuf_addf(&sb, " %s ", newdelete);
5244
5245         quote_c_style(fs->path, &sb, NULL, 0);
5246         strbuf_addch(&sb, '\n');
5247         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5248                          sb.buf, sb.len, 0);
5249         strbuf_release(&sb);
5250 }
5251
5252 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5253                 int show_name)
5254 {
5255         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5256                 struct strbuf sb = STRBUF_INIT;
5257                 strbuf_addf(&sb, " mode change %06o => %06o",
5258                             p->one->mode, p->two->mode);
5259                 if (show_name) {
5260                         strbuf_addch(&sb, ' ');
5261                         quote_c_style(p->two->path, &sb, NULL, 0);
5262                 }
5263                 strbuf_addch(&sb, '\n');
5264                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5265                                  sb.buf, sb.len, 0);
5266                 strbuf_release(&sb);
5267         }
5268 }
5269
5270 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5271                 struct diff_filepair *p)
5272 {
5273         struct strbuf sb = STRBUF_INIT;
5274         struct strbuf names = STRBUF_INIT;
5275
5276         pprint_rename(&names, p->one->path, p->two->path);
5277         strbuf_addf(&sb, " %s %s (%d%%)\n",
5278                     renamecopy, names.buf, similarity_index(p));
5279         strbuf_release(&names);
5280         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5281                                  sb.buf, sb.len, 0);
5282         show_mode_change(opt, p, 0);
5283         strbuf_release(&sb);
5284 }
5285
5286 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5287 {
5288         switch(p->status) {
5289         case DIFF_STATUS_DELETED:
5290                 show_file_mode_name(opt, "delete", p->one);
5291                 break;
5292         case DIFF_STATUS_ADDED:
5293                 show_file_mode_name(opt, "create", p->two);
5294                 break;
5295         case DIFF_STATUS_COPIED:
5296                 show_rename_copy(opt, "copy", p);
5297                 break;
5298         case DIFF_STATUS_RENAMED:
5299                 show_rename_copy(opt, "rename", p);
5300                 break;
5301         default:
5302                 if (p->score) {
5303                         struct strbuf sb = STRBUF_INIT;
5304                         strbuf_addstr(&sb, " rewrite ");
5305                         quote_c_style(p->two->path, &sb, NULL, 0);
5306                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5307                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5308                                          sb.buf, sb.len, 0);
5309                         strbuf_release(&sb);
5310                 }
5311                 show_mode_change(opt, p, !p->score);
5312                 break;
5313         }
5314 }
5315
5316 struct patch_id_t {
5317         git_SHA_CTX *ctx;
5318         int patchlen;
5319 };
5320
5321 static int remove_space(char *line, int len)
5322 {
5323         int i;
5324         char *dst = line;
5325         unsigned char c;
5326
5327         for (i = 0; i < len; i++)
5328                 if (!isspace((c = line[i])))
5329                         *dst++ = c;
5330
5331         return dst - line;
5332 }
5333
5334 static void patch_id_consume(void *priv, char *line, unsigned long len)
5335 {
5336         struct patch_id_t *data = priv;
5337         int new_len;
5338
5339         /* Ignore line numbers when computing the SHA1 of the patch */
5340         if (starts_with(line, "@@ -"))
5341                 return;
5342
5343         new_len = remove_space(line, len);
5344
5345         git_SHA1_Update(data->ctx, line, new_len);
5346         data->patchlen += new_len;
5347 }
5348
5349 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5350 {
5351         git_SHA1_Update(ctx, str, strlen(str));
5352 }
5353
5354 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5355 {
5356         /* large enough for 2^32 in octal */
5357         char buf[12];
5358         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5359         git_SHA1_Update(ctx, buf, len);
5360 }
5361
5362 /* returns 0 upon success, and writes result into sha1 */
5363 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5364 {
5365         struct diff_queue_struct *q = &diff_queued_diff;
5366         int i;
5367         git_SHA_CTX ctx;
5368         struct patch_id_t data;
5369
5370         git_SHA1_Init(&ctx);
5371         memset(&data, 0, sizeof(struct patch_id_t));
5372         data.ctx = &ctx;
5373
5374         for (i = 0; i < q->nr; i++) {
5375                 xpparam_t xpp;
5376                 xdemitconf_t xecfg;
5377                 mmfile_t mf1, mf2;
5378                 struct diff_filepair *p = q->queue[i];
5379                 int len1, len2;
5380
5381                 memset(&xpp, 0, sizeof(xpp));
5382                 memset(&xecfg, 0, sizeof(xecfg));
5383                 if (p->status == 0)
5384                         return error("internal diff status error");
5385                 if (p->status == DIFF_STATUS_UNKNOWN)
5386                         continue;
5387                 if (diff_unmodified_pair(p))
5388                         continue;
5389                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5390                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5391                         continue;
5392                 if (DIFF_PAIR_UNMERGED(p))
5393                         continue;
5394
5395                 diff_fill_oid_info(p->one);
5396                 diff_fill_oid_info(p->two);
5397
5398                 len1 = remove_space(p->one->path, strlen(p->one->path));
5399                 len2 = remove_space(p->two->path, strlen(p->two->path));
5400                 patch_id_add_string(&ctx, "diff--git");
5401                 patch_id_add_string(&ctx, "a/");
5402                 git_SHA1_Update(&ctx, p->one->path, len1);
5403                 patch_id_add_string(&ctx, "b/");
5404                 git_SHA1_Update(&ctx, p->two->path, len2);
5405
5406                 if (p->one->mode == 0) {
5407                         patch_id_add_string(&ctx, "newfilemode");
5408                         patch_id_add_mode(&ctx, p->two->mode);
5409                         patch_id_add_string(&ctx, "---/dev/null");
5410                         patch_id_add_string(&ctx, "+++b/");
5411                         git_SHA1_Update(&ctx, p->two->path, len2);
5412                 } else if (p->two->mode == 0) {
5413                         patch_id_add_string(&ctx, "deletedfilemode");
5414                         patch_id_add_mode(&ctx, p->one->mode);
5415                         patch_id_add_string(&ctx, "---a/");
5416                         git_SHA1_Update(&ctx, p->one->path, len1);
5417                         patch_id_add_string(&ctx, "+++/dev/null");
5418                 } else {
5419                         patch_id_add_string(&ctx, "---a/");
5420                         git_SHA1_Update(&ctx, p->one->path, len1);
5421                         patch_id_add_string(&ctx, "+++b/");
5422                         git_SHA1_Update(&ctx, p->two->path, len2);
5423                 }
5424
5425                 if (diff_header_only)
5426                         continue;
5427
5428                 if (fill_mmfile(&mf1, p->one) < 0 ||
5429                     fill_mmfile(&mf2, p->two) < 0)
5430                         return error("unable to read files to diff");
5431
5432                 if (diff_filespec_is_binary(p->one) ||
5433                     diff_filespec_is_binary(p->two)) {
5434                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5435                                         GIT_SHA1_HEXSZ);
5436                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5437                                         GIT_SHA1_HEXSZ);
5438                         continue;
5439                 }
5440
5441                 xpp.flags = 0;
5442                 xecfg.ctxlen = 3;
5443                 xecfg.flags = 0;
5444                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5445                                   &xpp, &xecfg))
5446                         return error("unable to generate patch-id diff for %s",
5447                                      p->one->path);
5448         }
5449
5450         git_SHA1_Final(oid->hash, &ctx);
5451         return 0;
5452 }
5453
5454 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5455 {
5456         struct diff_queue_struct *q = &diff_queued_diff;
5457         int i;
5458         int result = diff_get_patch_id(options, oid, diff_header_only);
5459
5460         for (i = 0; i < q->nr; i++)
5461                 diff_free_filepair(q->queue[i]);
5462
5463         free(q->queue);
5464         DIFF_QUEUE_CLEAR(q);
5465
5466         return result;
5467 }
5468
5469 static int is_summary_empty(const struct diff_queue_struct *q)
5470 {
5471         int i;
5472
5473         for (i = 0; i < q->nr; i++) {
5474                 const struct diff_filepair *p = q->queue[i];
5475
5476                 switch (p->status) {
5477                 case DIFF_STATUS_DELETED:
5478                 case DIFF_STATUS_ADDED:
5479                 case DIFF_STATUS_COPIED:
5480                 case DIFF_STATUS_RENAMED:
5481                         return 0;
5482                 default:
5483                         if (p->score)
5484                                 return 0;
5485                         if (p->one->mode && p->two->mode &&
5486                             p->one->mode != p->two->mode)
5487                                 return 0;
5488                         break;
5489                 }
5490         }
5491         return 1;
5492 }
5493
5494 static const char rename_limit_warning[] =
5495 N_("inexact rename detection was skipped due to too many files.");
5496
5497 static const char degrade_cc_to_c_warning[] =
5498 N_("only found copies from modified paths due to too many files.");
5499
5500 static const char rename_limit_advice[] =
5501 N_("you may want to set your %s variable to at least "
5502    "%d and retry the command.");
5503
5504 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5505 {
5506         fflush(stdout);
5507         if (degraded_cc)
5508                 warning(_(degrade_cc_to_c_warning));
5509         else if (needed)
5510                 warning(_(rename_limit_warning));
5511         else
5512                 return;
5513         if (0 < needed)
5514                 warning(_(rename_limit_advice), varname, needed);
5515 }
5516
5517 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5518 {
5519         int i;
5520         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5521         struct diff_queue_struct *q = &diff_queued_diff;
5522
5523         if (WSEH_NEW & WS_RULE_MASK)
5524                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5525
5526         if (o->color_moved)
5527                 o->emitted_symbols = &esm;
5528
5529         for (i = 0; i < q->nr; i++) {
5530                 struct diff_filepair *p = q->queue[i];
5531                 if (check_pair_status(p))
5532                         diff_flush_patch(p, o);
5533         }
5534
5535         if (o->emitted_symbols) {
5536                 if (o->color_moved) {
5537                         struct hashmap add_lines, del_lines;
5538
5539                         hashmap_init(&del_lines,
5540                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5541                         hashmap_init(&add_lines,
5542                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5543
5544                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5545                         mark_color_as_moved(o, &add_lines, &del_lines);
5546                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5547                                 dim_moved_lines(o);
5548
5549                         hashmap_free(&add_lines, 0);
5550                         hashmap_free(&del_lines, 0);
5551                 }
5552
5553                 for (i = 0; i < esm.nr; i++)
5554                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5555
5556                 for (i = 0; i < esm.nr; i++)
5557                         free((void *)esm.buf[i].line);
5558         }
5559         esm.nr = 0;
5560 }
5561
5562 void diff_flush(struct diff_options *options)
5563 {
5564         struct diff_queue_struct *q = &diff_queued_diff;
5565         int i, output_format = options->output_format;
5566         int separator = 0;
5567         int dirstat_by_line = 0;
5568
5569         /*
5570          * Order: raw, stat, summary, patch
5571          * or:    name/name-status/checkdiff (other bits clear)
5572          */
5573         if (!q->nr)
5574                 goto free_queue;
5575
5576         if (output_format & (DIFF_FORMAT_RAW |
5577                              DIFF_FORMAT_NAME |
5578                              DIFF_FORMAT_NAME_STATUS |
5579                              DIFF_FORMAT_CHECKDIFF)) {
5580                 for (i = 0; i < q->nr; i++) {
5581                         struct diff_filepair *p = q->queue[i];
5582                         if (check_pair_status(p))
5583                                 flush_one_pair(p, options);
5584                 }
5585                 separator++;
5586         }
5587
5588         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5589                 dirstat_by_line = 1;
5590
5591         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5592             dirstat_by_line) {
5593                 struct diffstat_t diffstat;
5594
5595                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5596                 for (i = 0; i < q->nr; i++) {
5597                         struct diff_filepair *p = q->queue[i];
5598                         if (check_pair_status(p))
5599                                 diff_flush_stat(p, options, &diffstat);
5600                 }
5601                 if (output_format & DIFF_FORMAT_NUMSTAT)
5602                         show_numstat(&diffstat, options);
5603                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5604                         show_stats(&diffstat, options);
5605                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5606                         show_shortstats(&diffstat, options);
5607                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5608                         show_dirstat_by_line(&diffstat, options);
5609                 free_diffstat_info(&diffstat);
5610                 separator++;
5611         }
5612         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5613                 show_dirstat(options);
5614
5615         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5616                 for (i = 0; i < q->nr; i++) {
5617                         diff_summary(options, q->queue[i]);
5618                 }
5619                 separator++;
5620         }
5621
5622         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5623             options->flags.exit_with_status &&
5624             options->flags.diff_from_contents) {
5625                 /*
5626                  * run diff_flush_patch for the exit status. setting
5627                  * options->file to /dev/null should be safe, because we
5628                  * aren't supposed to produce any output anyway.
5629                  */
5630                 if (options->close_file)
5631                         fclose(options->file);
5632                 options->file = xfopen("/dev/null", "w");
5633                 options->close_file = 1;
5634                 options->color_moved = 0;
5635                 for (i = 0; i < q->nr; i++) {
5636                         struct diff_filepair *p = q->queue[i];
5637                         if (check_pair_status(p))
5638                                 diff_flush_patch(p, options);
5639                         if (options->found_changes)
5640                                 break;
5641                 }
5642         }
5643
5644         if (output_format & DIFF_FORMAT_PATCH) {
5645                 if (separator) {
5646                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5647                         if (options->stat_sep)
5648                                 /* attach patch instead of inline */
5649                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5650                                                  NULL, 0, 0);
5651                 }
5652
5653                 diff_flush_patch_all_file_pairs(options);
5654         }
5655
5656         if (output_format & DIFF_FORMAT_CALLBACK)
5657                 options->format_callback(q, options, options->format_callback_data);
5658
5659         for (i = 0; i < q->nr; i++)
5660                 diff_free_filepair(q->queue[i]);
5661 free_queue:
5662         free(q->queue);
5663         DIFF_QUEUE_CLEAR(q);
5664         if (options->close_file)
5665                 fclose(options->file);
5666
5667         /*
5668          * Report the content-level differences with HAS_CHANGES;
5669          * diff_addremove/diff_change does not set the bit when
5670          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5671          */
5672         if (options->flags.diff_from_contents) {
5673                 if (options->found_changes)
5674                         options->flags.has_changes = 1;
5675                 else
5676                         options->flags.has_changes = 0;
5677         }
5678 }
5679
5680 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5681 {
5682         return (((p->status == DIFF_STATUS_MODIFIED) &&
5683                  ((p->score &&
5684                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5685                   (!p->score &&
5686                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5687                 ((p->status != DIFF_STATUS_MODIFIED) &&
5688                  filter_bit_tst(p->status, options)));
5689 }
5690
5691 static void diffcore_apply_filter(struct diff_options *options)
5692 {
5693         int i;
5694         struct diff_queue_struct *q = &diff_queued_diff;
5695         struct diff_queue_struct outq;
5696
5697         DIFF_QUEUE_CLEAR(&outq);
5698
5699         if (!options->filter)
5700                 return;
5701
5702         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5703                 int found;
5704                 for (i = found = 0; !found && i < q->nr; i++) {
5705                         if (match_filter(options, q->queue[i]))
5706                                 found++;
5707                 }
5708                 if (found)
5709                         return;
5710
5711                 /* otherwise we will clear the whole queue
5712                  * by copying the empty outq at the end of this
5713                  * function, but first clear the current entries
5714                  * in the queue.
5715                  */
5716                 for (i = 0; i < q->nr; i++)
5717                         diff_free_filepair(q->queue[i]);
5718         }
5719         else {
5720                 /* Only the matching ones */
5721                 for (i = 0; i < q->nr; i++) {
5722                         struct diff_filepair *p = q->queue[i];
5723                         if (match_filter(options, p))
5724                                 diff_q(&outq, p);
5725                         else
5726                                 diff_free_filepair(p);
5727                 }
5728         }
5729         free(q->queue);
5730         *q = outq;
5731 }
5732
5733 /* Check whether two filespecs with the same mode and size are identical */
5734 static int diff_filespec_is_identical(struct diff_filespec *one,
5735                                       struct diff_filespec *two)
5736 {
5737         if (S_ISGITLINK(one->mode))
5738                 return 0;
5739         if (diff_populate_filespec(one, 0))
5740                 return 0;
5741         if (diff_populate_filespec(two, 0))
5742                 return 0;
5743         return !memcmp(one->data, two->data, one->size);
5744 }
5745
5746 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5747 {
5748         if (p->done_skip_stat_unmatch)
5749                 return p->skip_stat_unmatch_result;
5750
5751         p->done_skip_stat_unmatch = 1;
5752         p->skip_stat_unmatch_result = 0;
5753         /*
5754          * 1. Entries that come from stat info dirtiness
5755          *    always have both sides (iow, not create/delete),
5756          *    one side of the object name is unknown, with
5757          *    the same mode and size.  Keep the ones that
5758          *    do not match these criteria.  They have real
5759          *    differences.
5760          *
5761          * 2. At this point, the file is known to be modified,
5762          *    with the same mode and size, and the object
5763          *    name of one side is unknown.  Need to inspect
5764          *    the identical contents.
5765          */
5766         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5767             !DIFF_FILE_VALID(p->two) ||
5768             (p->one->oid_valid && p->two->oid_valid) ||
5769             (p->one->mode != p->two->mode) ||
5770             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5771             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5772             (p->one->size != p->two->size) ||
5773             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5774                 p->skip_stat_unmatch_result = 1;
5775         return p->skip_stat_unmatch_result;
5776 }
5777
5778 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5779 {
5780         int i;
5781         struct diff_queue_struct *q = &diff_queued_diff;
5782         struct diff_queue_struct outq;
5783         DIFF_QUEUE_CLEAR(&outq);
5784
5785         for (i = 0; i < q->nr; i++) {
5786                 struct diff_filepair *p = q->queue[i];
5787
5788                 if (diff_filespec_check_stat_unmatch(p))
5789                         diff_q(&outq, p);
5790                 else {
5791                         /*
5792                          * The caller can subtract 1 from skip_stat_unmatch
5793                          * to determine how many paths were dirty only
5794                          * due to stat info mismatch.
5795                          */
5796                         if (!diffopt->flags.no_index)
5797                                 diffopt->skip_stat_unmatch++;
5798                         diff_free_filepair(p);
5799                 }
5800         }
5801         free(q->queue);
5802         *q = outq;
5803 }
5804
5805 static int diffnamecmp(const void *a_, const void *b_)
5806 {
5807         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5808         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5809         const char *name_a, *name_b;
5810
5811         name_a = a->one ? a->one->path : a->two->path;
5812         name_b = b->one ? b->one->path : b->two->path;
5813         return strcmp(name_a, name_b);
5814 }
5815
5816 void diffcore_fix_diff_index(struct diff_options *options)
5817 {
5818         struct diff_queue_struct *q = &diff_queued_diff;
5819         QSORT(q->queue, q->nr, diffnamecmp);
5820 }
5821
5822 void diffcore_std(struct diff_options *options)
5823 {
5824         /* NOTE please keep the following in sync with diff_tree_combined() */
5825         if (options->skip_stat_unmatch)
5826                 diffcore_skip_stat_unmatch(options);
5827         if (!options->found_follow) {
5828                 /* See try_to_follow_renames() in tree-diff.c */
5829                 if (options->break_opt != -1)
5830                         diffcore_break(options->break_opt);
5831                 if (options->detect_rename)
5832                         diffcore_rename(options);
5833                 if (options->break_opt != -1)
5834                         diffcore_merge_broken();
5835         }
5836         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
5837                 diffcore_pickaxe(options);
5838         if (options->orderfile)
5839                 diffcore_order(options->orderfile);
5840         if (!options->found_follow)
5841                 /* See try_to_follow_renames() in tree-diff.c */
5842                 diff_resolve_rename_copy();
5843         diffcore_apply_filter(options);
5844
5845         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
5846                 options->flags.has_changes = 1;
5847         else
5848                 options->flags.has_changes = 0;
5849
5850         options->found_follow = 0;
5851 }
5852
5853 int diff_result_code(struct diff_options *opt, int status)
5854 {
5855         int result = 0;
5856
5857         diff_warn_rename_limit("diff.renameLimit",
5858                                opt->needed_rename_limit,
5859                                opt->degraded_cc_to_c);
5860         if (!opt->flags.exit_with_status &&
5861             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5862                 return status;
5863         if (opt->flags.exit_with_status &&
5864             opt->flags.has_changes)
5865                 result |= 01;
5866         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5867             opt->flags.check_failed)
5868                 result |= 02;
5869         return result;
5870 }
5871
5872 int diff_can_quit_early(struct diff_options *opt)
5873 {
5874         return (opt->flags.quick &&
5875                 !opt->filter &&
5876                 opt->flags.has_changes);
5877 }
5878
5879 /*
5880  * Shall changes to this submodule be ignored?
5881  *
5882  * Submodule changes can be configured to be ignored separately for each path,
5883  * but that configuration can be overridden from the command line.
5884  */
5885 static int is_submodule_ignored(const char *path, struct diff_options *options)
5886 {
5887         int ignored = 0;
5888         struct diff_flags orig_flags = options->flags;
5889         if (!options->flags.override_submodule_config)
5890                 set_diffopt_flags_from_submodule_config(options, path);
5891         if (options->flags.ignore_submodules)
5892                 ignored = 1;
5893         options->flags = orig_flags;
5894         return ignored;
5895 }
5896
5897 void diff_addremove(struct diff_options *options,
5898                     int addremove, unsigned mode,
5899                     const struct object_id *oid,
5900                     int oid_valid,
5901                     const char *concatpath, unsigned dirty_submodule)
5902 {
5903         struct diff_filespec *one, *two;
5904
5905         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5906                 return;
5907
5908         /* This may look odd, but it is a preparation for
5909          * feeding "there are unchanged files which should
5910          * not produce diffs, but when you are doing copy
5911          * detection you would need them, so here they are"
5912          * entries to the diff-core.  They will be prefixed
5913          * with something like '=' or '*' (I haven't decided
5914          * which but should not make any difference).
5915          * Feeding the same new and old to diff_change()
5916          * also has the same effect.
5917          * Before the final output happens, they are pruned after
5918          * merged into rename/copy pairs as appropriate.
5919          */
5920         if (options->flags.reverse_diff)
5921                 addremove = (addremove == '+' ? '-' :
5922                              addremove == '-' ? '+' : addremove);
5923
5924         if (options->prefix &&
5925             strncmp(concatpath, options->prefix, options->prefix_length))
5926                 return;
5927
5928         one = alloc_filespec(concatpath);
5929         two = alloc_filespec(concatpath);
5930
5931         if (addremove != '+')
5932                 fill_filespec(one, oid, oid_valid, mode);
5933         if (addremove != '-') {
5934                 fill_filespec(two, oid, oid_valid, mode);
5935                 two->dirty_submodule = dirty_submodule;
5936         }
5937
5938         diff_queue(&diff_queued_diff, one, two);
5939         if (!options->flags.diff_from_contents)
5940                 options->flags.has_changes = 1;
5941 }
5942
5943 void diff_change(struct diff_options *options,
5944                  unsigned old_mode, unsigned new_mode,
5945                  const struct object_id *old_oid,
5946                  const struct object_id *new_oid,
5947                  int old_oid_valid, int new_oid_valid,
5948                  const char *concatpath,
5949                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5950 {
5951         struct diff_filespec *one, *two;
5952         struct diff_filepair *p;
5953
5954         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5955             is_submodule_ignored(concatpath, options))
5956                 return;
5957
5958         if (options->flags.reverse_diff) {
5959                 SWAP(old_mode, new_mode);
5960                 SWAP(old_oid, new_oid);
5961                 SWAP(old_oid_valid, new_oid_valid);
5962                 SWAP(old_dirty_submodule, new_dirty_submodule);
5963         }
5964
5965         if (options->prefix &&
5966             strncmp(concatpath, options->prefix, options->prefix_length))
5967                 return;
5968
5969         one = alloc_filespec(concatpath);
5970         two = alloc_filespec(concatpath);
5971         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5972         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5973         one->dirty_submodule = old_dirty_submodule;
5974         two->dirty_submodule = new_dirty_submodule;
5975         p = diff_queue(&diff_queued_diff, one, two);
5976
5977         if (options->flags.diff_from_contents)
5978                 return;
5979
5980         if (options->flags.quick && options->skip_stat_unmatch &&
5981             !diff_filespec_check_stat_unmatch(p))
5982                 return;
5983
5984         options->flags.has_changes = 1;
5985 }
5986
5987 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5988 {
5989         struct diff_filepair *pair;
5990         struct diff_filespec *one, *two;
5991
5992         if (options->prefix &&
5993             strncmp(path, options->prefix, options->prefix_length))
5994                 return NULL;
5995
5996         one = alloc_filespec(path);
5997         two = alloc_filespec(path);
5998         pair = diff_queue(&diff_queued_diff, one, two);
5999         pair->is_unmerged = 1;
6000         return pair;
6001 }
6002
6003 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6004                 size_t *outsize)
6005 {
6006         struct diff_tempfile *temp;
6007         const char *argv[3];
6008         const char **arg = argv;
6009         struct child_process child = CHILD_PROCESS_INIT;
6010         struct strbuf buf = STRBUF_INIT;
6011         int err = 0;
6012
6013         temp = prepare_temp_file(spec->path, spec);
6014         *arg++ = pgm;
6015         *arg++ = temp->name;
6016         *arg = NULL;
6017
6018         child.use_shell = 1;
6019         child.argv = argv;
6020         child.out = -1;
6021         if (start_command(&child)) {
6022                 remove_tempfile();
6023                 return NULL;
6024         }
6025
6026         if (strbuf_read(&buf, child.out, 0) < 0)
6027                 err = error("error reading from textconv command '%s'", pgm);
6028         close(child.out);
6029
6030         if (finish_command(&child) || err) {
6031                 strbuf_release(&buf);
6032                 remove_tempfile();
6033                 return NULL;
6034         }
6035         remove_tempfile();
6036
6037         return strbuf_detach(&buf, outsize);
6038 }
6039
6040 size_t fill_textconv(struct userdiff_driver *driver,
6041                      struct diff_filespec *df,
6042                      char **outbuf)
6043 {
6044         size_t size;
6045
6046         if (!driver) {
6047                 if (!DIFF_FILE_VALID(df)) {
6048                         *outbuf = "";
6049                         return 0;
6050                 }
6051                 if (diff_populate_filespec(df, 0))
6052                         die("unable to read files to diff");
6053                 *outbuf = df->data;
6054                 return df->size;
6055         }
6056
6057         if (!driver->textconv)
6058                 die("BUG: fill_textconv called with non-textconv driver");
6059
6060         if (driver->textconv_cache && df->oid_valid) {
6061                 *outbuf = notes_cache_get(driver->textconv_cache,
6062                                           &df->oid,
6063                                           &size);
6064                 if (*outbuf)
6065                         return size;
6066         }
6067
6068         *outbuf = run_textconv(driver->textconv, df, &size);
6069         if (!*outbuf)
6070                 die("unable to read files to diff");
6071
6072         if (driver->textconv_cache && df->oid_valid) {
6073                 /* ignore errors, as we might be in a readonly repository */
6074                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6075                                 size);
6076                 /*
6077                  * we could save up changes and flush them all at the end,
6078                  * but we would need an extra call after all diffing is done.
6079                  * Since generating a cache entry is the slow path anyway,
6080                  * this extra overhead probably isn't a big deal.
6081                  */
6082                 notes_cache_write(driver->textconv_cache);
6083         }
6084
6085         return size;
6086 }
6087
6088 int textconv_object(const char *path,
6089                     unsigned mode,
6090                     const struct object_id *oid,
6091                     int oid_valid,
6092                     char **buf,
6093                     unsigned long *buf_size)
6094 {
6095         struct diff_filespec *df;
6096         struct userdiff_driver *textconv;
6097
6098         df = alloc_filespec(path);
6099         fill_filespec(df, oid, oid_valid, mode);
6100         textconv = get_textconv(df);
6101         if (!textconv) {
6102                 free_filespec(df);
6103                 return 0;
6104         }
6105
6106         *buf_size = fill_textconv(textconv, df, buf);
6107         free_filespec(df);
6108         return 1;
6109 }
6110
6111 void setup_diff_pager(struct diff_options *opt)
6112 {
6113         /*
6114          * If the user asked for our exit code, then either they want --quiet
6115          * or --exit-code. We should definitely not bother with a pager in the
6116          * former case, as we will generate no output. Since we still properly
6117          * report our exit code even when a pager is run, we _could_ run a
6118          * pager with --exit-code. But since we have not done so historically,
6119          * and because it is easy to find people oneline advising "git diff
6120          * --exit-code" in hooks and other scripts, we do not do so.
6121          */
6122         if (!opt->flags.exit_with_status &&
6123             check_pager_config("diff") != 0)
6124                 setup_pager();
6125 }