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 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed_zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else
 289                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 290 }
 291
 292 static int parse_color_moved_ws(const char *arg)
 293 {
 294         int ret = 0;
 295         struct string_list l = STRING_LIST_INIT_DUP;
 296         struct string_list_item *i;
 297
 298         string_list_split(&l, arg, ',', -1);
 299
 300         for_each_string_list_item(i, &l) {
 301                 struct strbuf sb = STRBUF_INIT;
 302                 strbuf_addstr(&sb, i->string);
 303                 strbuf_trim(&sb);
 304
 305                 if (!strcmp(sb.buf, "ignore-space-change"))
 306                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 307                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 308                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 309                 else if (!strcmp(sb.buf, "ignore-all-space"))
 310                         ret |= XDF_IGNORE_WHITESPACE;
 311                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 312                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 313                 else
 314                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 315
 316                 strbuf_release(&sb);
 317         }
 318
 319         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 320             (ret & XDF_WHITESPACE_FLAGS))
 321                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 322
 323         string_list_clear(&l, 0);
 324
 325         return ret;
 326 }
 327
 328 int git_diff_ui_config(const char *var, const char *value, void *cb)
 329 {
 330         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 331                 diff_use_color_default = git_config_colorbool(var, value);
 332                 return 0;
 333         }
 334         if (!strcmp(var, "diff.colormoved")) {
 335                 int cm = parse_color_moved(value);
 336                 if (cm < 0)
 337                         return -1;
 338                 diff_color_moved_default = cm;
 339                 return 0;
 340         }
 341         if (!strcmp(var, "diff.colormovedws")) {
 342                 int cm = parse_color_moved_ws(value);
 343                 if (cm < 0)
 344                         return -1;
 345                 diff_color_moved_ws_default = cm;
 346                 return 0;
 347         }
 348         if (!strcmp(var, "diff.context")) {
 349                 diff_context_default = git_config_int(var, value);
 350                 if (diff_context_default < 0)
 351                         return -1;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.interhunkcontext")) {
 355                 diff_interhunk_context_default = git_config_int(var, value);
 356                 if (diff_interhunk_context_default < 0)
 357                         return -1;
 358                 return 0;
 359         }
 360         if (!strcmp(var, "diff.renames")) {
 361                 diff_detect_rename_default = git_config_rename(var, value);
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.autorefreshindex")) {
 365                 diff_auto_refresh_index = git_config_bool(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.mnemonicprefix")) {
 369                 diff_mnemonic_prefix = git_config_bool(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.noprefix")) {
 373                 diff_no_prefix = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.statgraphwidth")) {
 377                 diff_stat_graph_width = git_config_int(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.external"))
 381                 return git_config_string(&external_diff_cmd_cfg, var, value);
 382         if (!strcmp(var, "diff.wordregex"))
 383                 return git_config_string(&diff_word_regex_cfg, var, value);
 384         if (!strcmp(var, "diff.orderfile"))
 385                 return git_config_pathname(&diff_order_file_cfg, var, value);
 386
 387         if (!strcmp(var, "diff.ignoresubmodules"))
 388                 handle_ignore_submodules_arg(&default_diff_options, value);
 389
 390         if (!strcmp(var, "diff.submodule")) {
 391                 if (parse_submodule_params(&default_diff_options, value))
 392                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 393                                 value);
 394                 return 0;
 395         }
 396
 397         if (!strcmp(var, "diff.algorithm")) {
 398                 diff_algorithm = parse_algorithm_value(value);
 399                 if (diff_algorithm < 0)
 400                         return -1;
 401                 return 0;
 402         }
 403
 404         if (!strcmp(var, "diff.wserrorhighlight")) {
 405                 int val = parse_ws_error_highlight(value);
 406                 if (val < 0)
 407                         return -1;
 408                 ws_error_highlight_default = val;
 409                 return 0;
 410         }
 411
 412         if (git_color_config(var, value, cb) < 0)
 413                 return -1;
 414
 415         return git_diff_basic_config(var, value, cb);
 416 }
 417
 418 int git_diff_basic_config(const char *var, const char *value, void *cb)
 419 {
 420         const char *name;
 421
 422         if (!strcmp(var, "diff.renamelimit")) {
 423                 diff_rename_limit_default = git_config_int(var, value);
 424                 return 0;
 425         }
 426
 427         if (userdiff_config(var, value) < 0)
 428                 return -1;
 429
 430         if (skip_prefix(var, "diff.color.", &name) ||
 431             skip_prefix(var, "color.diff.", &name)) {
 432                 int slot = parse_diff_color_slot(name);
 433                 if (slot < 0)
 434                         return 0;
 435                 if (!value)
 436                         return config_error_nonbool(var);
 437                 return color_parse(value, diff_colors[slot]);
 438         }
 439
 440         /* like GNU diff's --suppress-blank-empty option  */
 441         if (!strcmp(var, "diff.suppressblankempty") ||
 442                         /* for backwards compatibility */
 443                         !strcmp(var, "diff.suppress-blank-empty")) {
 444                 diff_suppress_blank_empty = git_config_bool(var, value);
 445                 return 0;
 446         }
 447
 448         if (!strcmp(var, "diff.dirstat")) {
 449                 struct strbuf errmsg = STRBUF_INIT;
 450                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 451                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 452                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 453                                 errmsg.buf);
 454                 strbuf_release(&errmsg);
 455                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 456                 return 0;
 457         }
 458
 459         if (git_diff_heuristic_config(var, value, cb) < 0)
 460                 return -1;
 461
 462         return git_default_config(var, value, cb);
 463 }
 464
 465 static char *quote_two(const char *one, const char *two)
 466 {
 467         int need_one = quote_c_style(one, NULL, NULL, 1);
 468         int need_two = quote_c_style(two, NULL, NULL, 1);
 469         struct strbuf res = STRBUF_INIT;
 470
 471         if (need_one + need_two) {
 472                 strbuf_addch(&res, '"');
 473                 quote_c_style(one, &res, NULL, 1);
 474                 quote_c_style(two, &res, NULL, 1);
 475                 strbuf_addch(&res, '"');
 476         } else {
 477                 strbuf_addstr(&res, one);
 478                 strbuf_addstr(&res, two);
 479         }
 480         return strbuf_detach(&res, NULL);
 481 }
 482
 483 static const char *external_diff(void)
 484 {
 485         static const char *external_diff_cmd = NULL;
 486         static int done_preparing = 0;
 487
 488         if (done_preparing)
 489                 return external_diff_cmd;
 490         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 491         if (!external_diff_cmd)
 492                 external_diff_cmd = external_diff_cmd_cfg;
 493         done_preparing = 1;
 494         return external_diff_cmd;
 495 }
 496
 497 /*
 498  * Keep track of files used for diffing. Sometimes such an entry
 499  * refers to a temporary file, sometimes to an existing file, and
 500  * sometimes to "/dev/null".
 501  */
 502 static struct diff_tempfile {
 503         /*
 504          * filename external diff should read from, or NULL if this
 505          * entry is currently not in use:
 506          */
 507         const char *name;
 508
 509         char hex[GIT_MAX_HEXSZ + 1];
 510         char mode[10];
 511
 512         /*
 513          * If this diff_tempfile instance refers to a temporary file,
 514          * this tempfile object is used to manage its lifetime.
 515          */
 516         struct tempfile *tempfile;
 517 } diff_temp[2];
 518
 519 struct emit_callback {
 520         int color_diff;
 521         unsigned ws_rule;
 522         int blank_at_eof_in_preimage;
 523         int blank_at_eof_in_postimage;
 524         int lno_in_preimage;
 525         int lno_in_postimage;
 526         const char **label_path;
 527         struct diff_words_data *diff_words;
 528         struct diff_options *opt;
 529         struct strbuf *header;
 530 };
 531
 532 static int count_lines(const char *data, int size)
 533 {
 534         int count, ch, completely_empty = 1, nl_just_seen = 0;
 535         count = 0;
 536         while (0 < size--) {
 537                 ch = *data++;
 538                 if (ch == '\n') {
 539                         count++;
 540                         nl_just_seen = 1;
 541                         completely_empty = 0;
 542                 }
 543                 else {
 544                         nl_just_seen = 0;
 545                         completely_empty = 0;
 546                 }
 547         }
 548         if (completely_empty)
 549                 return 0;
 550         if (!nl_just_seen)
 551                 count++; /* no trailing newline */
 552         return count;
 553 }
 554
 555 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 556 {
 557         if (!DIFF_FILE_VALID(one)) {
 558                 mf->ptr = (char *)""; /* does not matter */
 559                 mf->size = 0;
 560                 return 0;
 561         }
 562         else if (diff_populate_filespec(one, 0))
 563                 return -1;
 564
 565         mf->ptr = one->data;
 566         mf->size = one->size;
 567         return 0;
 568 }
 569
 570 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 571 static unsigned long diff_filespec_size(struct diff_filespec *one)
 572 {
 573         if (!DIFF_FILE_VALID(one))
 574                 return 0;
 575         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 576         return one->size;
 577 }
 578
 579 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 580 {
 581         char *ptr = mf->ptr;
 582         long size = mf->size;
 583         int cnt = 0;
 584
 585         if (!size)
 586                 return cnt;
 587         ptr += size - 1; /* pointing at the very end */
 588         if (*ptr != '\n')
 589                 ; /* incomplete line */
 590         else
 591                 ptr--; /* skip the last LF */
 592         while (mf->ptr < ptr) {
 593                 char *prev_eol;
 594                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 595                         if (*prev_eol == '\n')
 596                                 break;
 597                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 598                         break;
 599                 cnt++;
 600                 ptr = prev_eol - 1;
 601         }
 602         return cnt;
 603 }
 604
 605 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 606                                struct emit_callback *ecbdata)
 607 {
 608         int l1, l2, at;
 609         unsigned ws_rule = ecbdata->ws_rule;
 610         l1 = count_trailing_blank(mf1, ws_rule);
 611         l2 = count_trailing_blank(mf2, ws_rule);
 612         if (l2 <= l1) {
 613                 ecbdata->blank_at_eof_in_preimage = 0;
 614                 ecbdata->blank_at_eof_in_postimage = 0;
 615                 return;
 616         }
 617         at = count_lines(mf1->ptr, mf1->size);
 618         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 619
 620         at = count_lines(mf2->ptr, mf2->size);
 621         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 622 }
 623
 624 static void emit_line_0(struct diff_options *o,
 625                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 626                         int first, const char *line, int len)
 627 {
 628         int has_trailing_newline, has_trailing_carriage_return;
 629         int nofirst;
 630         FILE *file = o->file;
 631
 632         if (first)
 633                 fputs(diff_line_prefix(o), file);
 634         else if (!len)
 635                 return;
 636
 637         if (len == 0) {
 638                 has_trailing_newline = (first == '\n');
 639                 has_trailing_carriage_return = (!has_trailing_newline &&
 640                                                 (first == '\r'));
 641                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 642         } else {
 643                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 644                 if (has_trailing_newline)
 645                         len--;
 646                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 647                 if (has_trailing_carriage_return)
 648                         len--;
 649                 nofirst = 0;
 650         }
 651
 652         if (len || !nofirst) {
 653                 if (reverse && want_color(o->use_color))
 654                         fputs(GIT_COLOR_REVERSE, file);
 655                 if (set_sign)
 656                         fputs(set_sign, file);
 657                 if (first && !nofirst)
 658                         fputc(first, file);
 659                 if (set && set != set_sign) {
 660                         if (set_sign)
 661                                 fputs(reset, file);
 662                         fputs(set, file);
 663                 }
 664                 fwrite(line, len, 1, file);
 665                 fputs(reset, file);
 666         }
 667         if (has_trailing_carriage_return)
 668                 fputc('\r', file);
 669         if (has_trailing_newline)
 670                 fputc('\n', file);
 671 }
 672
 673 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 674                       const char *line, int len)
 675 {
 676         emit_line_0(o, set, NULL, 0, reset, line[0], line+1, len-1);
 677 }
 678
 679 enum diff_symbol {
 680         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 681         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 682         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 683         DIFF_SYMBOL_BINARY_DIFF_BODY,
 684         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 685         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 686         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 687         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 688         DIFF_SYMBOL_STATS_LINE,
 689         DIFF_SYMBOL_WORD_DIFF,
 690         DIFF_SYMBOL_STAT_SEP,
 691         DIFF_SYMBOL_SUMMARY,
 692         DIFF_SYMBOL_SUBMODULE_ADD,
 693         DIFF_SYMBOL_SUBMODULE_DEL,
 694         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 695         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 696         DIFF_SYMBOL_SUBMODULE_HEADER,
 697         DIFF_SYMBOL_SUBMODULE_ERROR,
 698         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 699         DIFF_SYMBOL_REWRITE_DIFF,
 700         DIFF_SYMBOL_BINARY_FILES,
 701         DIFF_SYMBOL_HEADER,
 702         DIFF_SYMBOL_FILEPAIR_PLUS,
 703         DIFF_SYMBOL_FILEPAIR_MINUS,
 704         DIFF_SYMBOL_WORDS_PORCELAIN,
 705         DIFF_SYMBOL_WORDS,
 706         DIFF_SYMBOL_CONTEXT,
 707         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 708         DIFF_SYMBOL_PLUS,
 709         DIFF_SYMBOL_MINUS,
 710         DIFF_SYMBOL_NO_LF_EOF,
 711         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 712         DIFF_SYMBOL_CONTEXT_MARKER,
 713         DIFF_SYMBOL_SEPARATOR
 714 };
 715 /*
 716  * Flags for content lines:
 717  * 0..12 are whitespace rules
 718  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 719  * 16 is marking if the line is blank at EOF
 720  */
 721 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 722 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 723 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 724 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 725 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 726
 727 /*
 728  * This struct is used when we need to buffer the output of the diff output.
 729  *
 730  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 731  * into the pre/post image file. This pointer could be a union with the
 732  * line pointer. By storing an offset into the file instead of the literal line,
 733  * we can decrease the memory footprint for the buffered output. At first we
 734  * may want to only have indirection for the content lines, but we could also
 735  * enhance the state for emitting prefabricated lines, e.g. the similarity
 736  * score line or hunk/file headers would only need to store a number or path
 737  * and then the output can be constructed later on depending on state.
 738  */
 739 struct emitted_diff_symbol {
 740         const char *line;
 741         int len;
 742         int flags;
 743         enum diff_symbol s;
 744 };
 745 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 746
 747 struct emitted_diff_symbols {
 748         struct emitted_diff_symbol *buf;
 749         int nr, alloc;
 750 };
 751 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 752
 753 static void append_emitted_diff_symbol(struct diff_options *o,
 754                                        struct emitted_diff_symbol *e)
 755 {
 756         struct emitted_diff_symbol *f;
 757
 758         ALLOC_GROW(o->emitted_symbols->buf,
 759                    o->emitted_symbols->nr + 1,
 760                    o->emitted_symbols->alloc);
 761         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 762
 763         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 764         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 765 }
 766
 767 struct moved_entry {
 768         struct hashmap_entry ent;
 769         const struct emitted_diff_symbol *es;
 770         struct moved_entry *next_line;
 771         struct ws_delta *wsd;
 772 };
 773
 774 /**
 775  * The struct ws_delta holds white space differences between moved lines, i.e.
 776  * between '+' and '-' lines that have been detected to be a move.
 777  * The string contains the difference in leading white spaces, before the
 778  * rest of the line is compared using the white space config for move
 779  * coloring. The current_longer indicates if the first string in the
 780  * comparision is longer than the second.
 781  */
 782 struct ws_delta {
 783         char *string;
 784         unsigned int current_longer : 1;
 785 };
 786 #define WS_DELTA_INIT { NULL, 0 }
 787
 788 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 789                              const struct emitted_diff_symbol *b,
 790                              struct ws_delta *out)
 791 {
 792         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 793         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 794         int d = longer->len - shorter->len;
 795
 796         out->string = xmemdupz(longer->line, d);
 797         out->current_longer = (a == longer);
 798
 799         return !strncmp(longer->line + d, shorter->line, shorter->len);
 800 }
 801
 802 static int cmp_in_block_with_wsd(const struct diff_options *o,
 803                                  const struct moved_entry *cur,
 804                                  const struct moved_entry *match,
 805                                  struct moved_entry *pmb,
 806                                  int n)
 807 {
 808         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 809         int al = cur->es->len, cl = l->len;
 810         const char *a = cur->es->line,
 811                    *b = match->es->line,
 812                    *c = l->line;
 813
 814         int wslen;
 815
 816         /*
 817          * We need to check if 'cur' is equal to 'match'.
 818          * As those are from the same (+/-) side, we do not need to adjust for
 819          * indent changes. However these were found using fuzzy matching
 820          * so we do have to check if they are equal.
 821          */
 822         if (strcmp(a, b))
 823                 return 1;
 824
 825         if (!pmb->wsd)
 826                 /*
 827                  * No white space delta was carried forward? This can happen
 828                  * when we exit early in this function and do not carry
 829                  * forward ws.
 830                  */
 831                 return 1;
 832
 833         /*
 834          * The indent changes of the block are known and carried forward in
 835          * pmb->wsd; however we need to check if the indent changes of the
 836          * current line are still the same as before.
 837          *
 838          * To do so we need to compare 'l' to 'cur', adjusting the
 839          * one of them for the white spaces, depending which was longer.
 840          */
 841
 842         wslen = strlen(pmb->wsd->string);
 843         if (pmb->wsd->current_longer) {
 844                 c += wslen;
 845                 cl -= wslen;
 846         } else {
 847                 a += wslen;
 848                 al -= wslen;
 849         }
 850
 851         if (strcmp(a, c))
 852                 return 1;
 853
 854         return 0;
 855 }
 856
 857 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 858                            const void *entry,
 859                            const void *entry_or_key,
 860                            const void *keydata)
 861 {
 862         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 863         const struct moved_entry *a = entry;
 864         const struct moved_entry *b = entry_or_key;
 865         unsigned flags = diffopt->color_moved_ws_handling
 866                          & XDF_WHITESPACE_FLAGS;
 867
 868         if (diffopt->color_moved_ws_handling &
 869             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 870                 /*
 871                  * As there is not specific white space config given,
 872                  * we'd need to check for a new block, so ignore all
 873                  * white space. The setup of the white space
 874                  * configuration for the next block is done else where
 875                  */
 876                 flags |= XDF_IGNORE_WHITESPACE;
 877
 878         return !xdiff_compare_lines(a->es->line, a->es->len,
 879                                     b->es->line, b->es->len,
 880                                     flags);
 881 }
 882
 883 static struct moved_entry *prepare_entry(struct diff_options *o,
 884                                          int line_no)
 885 {
 886         struct moved_entry *ret = xmalloc(sizeof(*ret));
 887         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 888         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 889
 890         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 891         ret->es = l;
 892         ret->next_line = NULL;
 893         ret->wsd = NULL;
 894
 895         return ret;
 896 }
 897
 898 static void add_lines_to_move_detection(struct diff_options *o,
 899                                         struct hashmap *add_lines,
 900                                         struct hashmap *del_lines)
 901 {
 902         struct moved_entry *prev_line = NULL;
 903
 904         int n;
 905         for (n = 0; n < o->emitted_symbols->nr; n++) {
 906                 struct hashmap *hm;
 907                 struct moved_entry *key;
 908
 909                 switch (o->emitted_symbols->buf[n].s) {
 910                 case DIFF_SYMBOL_PLUS:
 911                         hm = add_lines;
 912                         break;
 913                 case DIFF_SYMBOL_MINUS:
 914                         hm = del_lines;
 915                         break;
 916                 default:
 917                         prev_line = NULL;
 918                         continue;
 919                 }
 920
 921                 key = prepare_entry(o, n);
 922                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 923                         prev_line->next_line = key;
 924
 925                 hashmap_add(hm, key);
 926                 prev_line = key;
 927         }
 928 }
 929
 930 static void pmb_advance_or_null(struct diff_options *o,
 931                                 struct moved_entry *match,
 932                                 struct hashmap *hm,
 933                                 struct moved_entry **pmb,
 934                                 int pmb_nr)
 935 {
 936         int i;
 937         for (i = 0; i < pmb_nr; i++) {
 938                 struct moved_entry *prev = pmb[i];
 939                 struct moved_entry *cur = (prev && prev->next_line) ?
 940                                 prev->next_line : NULL;
 941                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 942                         pmb[i] = cur;
 943                 } else {
 944                         pmb[i] = NULL;
 945                 }
 946         }
 947 }
 948
 949 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 950                                             struct moved_entry *match,
 951                                             struct hashmap *hm,
 952                                             struct moved_entry **pmb,
 953                                             int pmb_nr, int n)
 954 {
 955         int i;
 956         char *got_match = xcalloc(1, pmb_nr);
 957
 958         for (; match; match = hashmap_get_next(hm, match)) {
 959                 for (i = 0; i < pmb_nr; i++) {
 960                         struct moved_entry *prev = pmb[i];
 961                         struct moved_entry *cur = (prev && prev->next_line) ?
 962                                         prev->next_line : NULL;
 963                         if (!cur)
 964                                 continue;
 965                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 966                                 got_match[i] |= 1;
 967                 }
 968         }
 969
 970         for (i = 0; i < pmb_nr; i++) {
 971                 if (got_match[i]) {
 972                         /* Carry the white space delta forward */
 973                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 974                         pmb[i] = pmb[i]->next_line;
 975                 } else
 976                         pmb[i] = NULL;
 977         }
 978 }
 979
 980 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 981                                          int pmb_nr)
 982 {
 983         int lp, rp;
 984
 985         /* Shrink the set of potential block to the remaining running */
 986         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 987                 while (lp < pmb_nr && pmb[lp])
 988                         lp++;
 989                 /* lp points at the first NULL now */
 990
 991                 while (rp > -1 && !pmb[rp])
 992                         rp--;
 993                 /* rp points at the last non-NULL */
 994
 995                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 996                         pmb[lp] = pmb[rp];
 997                         if (pmb[rp]->wsd) {
 998                                 free(pmb[rp]->wsd->string);
 999                                 FREE_AND_NULL(pmb[rp]->wsd);
1000                         }
1001                         pmb[rp] = NULL;
1002                         rp--;
1003                         lp++;
1004                 }
1005         }
1006
1007         /* Remember the number of running sets */
1008         return rp + 1;
1009 }
1010
1011 /*
1012  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1013  *
1014  * Otherwise, if the last block has fewer alphanumeric characters than
1015  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1016  * that block.
1017  *
1018  * The last block consists of the (n - block_length)'th line up to but not
1019  * including the nth line.
1020  *
1021  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1022  * Think of a way to unify them.
1023  */
1024 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1025 {
1026         int i, alnum_count = 0;
1027         if (o->color_moved == COLOR_MOVED_PLAIN)
1028                 return;
1029         for (i = 1; i < block_length + 1; i++) {
1030                 const char *c = o->emitted_symbols->buf[n - i].line;
1031                 for (; *c; c++) {
1032                         if (!isalnum(*c))
1033                                 continue;
1034                         alnum_count++;
1035                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1036                                 return;
1037                 }
1038         }
1039         for (i = 1; i < block_length + 1; i++)
1040                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1041 }
1042
1043 /* Find blocks of moved code, delegate actual coloring decision to helper */
1044 static void mark_color_as_moved(struct diff_options *o,
1045                                 struct hashmap *add_lines,
1046                                 struct hashmap *del_lines)
1047 {
1048         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1049         int pmb_nr = 0, pmb_alloc = 0;
1050         int n, flipped_block = 1, block_length = 0;
1051
1052
1053         for (n = 0; n < o->emitted_symbols->nr; n++) {
1054                 struct hashmap *hm = NULL;
1055                 struct moved_entry *key;
1056                 struct moved_entry *match = NULL;
1057                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1058
1059                 switch (l->s) {
1060                 case DIFF_SYMBOL_PLUS:
1061                         hm = del_lines;
1062                         key = prepare_entry(o, n);
1063                         match = hashmap_get(hm, key, NULL);
1064                         free(key);
1065                         break;
1066                 case DIFF_SYMBOL_MINUS:
1067                         hm = add_lines;
1068                         key = prepare_entry(o, n);
1069                         match = hashmap_get(hm, key, NULL);
1070                         free(key);
1071                         break;
1072                 default:
1073                         flipped_block = 1;
1074                 }
1075
1076                 if (!match) {
1077                         adjust_last_block(o, n, block_length);
1078                         pmb_nr = 0;
1079                         block_length = 0;
1080                         continue;
1081                 }
1082
1083                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1084
1085                 if (o->color_moved == COLOR_MOVED_PLAIN)
1086                         continue;
1087
1088                 if (o->color_moved_ws_handling &
1089                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1090                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1091                 else
1092                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1093
1094                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1095
1096                 if (pmb_nr == 0) {
1097                         /*
1098                          * The current line is the start of a new block.
1099                          * Setup the set of potential blocks.
1100                          */
1101                         for (; match; match = hashmap_get_next(hm, match)) {
1102                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1103                                 if (o->color_moved_ws_handling &
1104                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1105                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1106                                         if (compute_ws_delta(l, match->es, wsd)) {
1107                                                 match->wsd = wsd;
1108                                                 pmb[pmb_nr++] = match;
1109                                         } else
1110                                                 free(wsd);
1111                                 } else {
1112                                         pmb[pmb_nr++] = match;
1113                                 }
1114                         }
1115
1116                         flipped_block = (flipped_block + 1) % 2;
1117
1118                         adjust_last_block(o, n, block_length);
1119                         block_length = 0;
1120                 }
1121
1122                 block_length++;
1123
1124                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1125                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1126         }
1127         adjust_last_block(o, n, block_length);
1128
1129         free(pmb);
1130 }
1131
1132 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1133   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1134 static void dim_moved_lines(struct diff_options *o)
1135 {
1136         int n;
1137         for (n = 0; n < o->emitted_symbols->nr; n++) {
1138                 struct emitted_diff_symbol *prev = (n != 0) ?
1139                                 &o->emitted_symbols->buf[n - 1] : NULL;
1140                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1141                 struct emitted_diff_symbol *next =
1142                                 (n < o->emitted_symbols->nr - 1) ?
1143                                 &o->emitted_symbols->buf[n + 1] : NULL;
1144
1145                 /* Not a plus or minus line? */
1146                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1147                         continue;
1148
1149                 /* Not a moved line? */
1150                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1151                         continue;
1152
1153                 /*
1154                  * If prev or next are not a plus or minus line,
1155                  * pretend they don't exist
1156                  */
1157                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1158                             prev->s != DIFF_SYMBOL_MINUS)
1159                         prev = NULL;
1160                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1161                             next->s != DIFF_SYMBOL_MINUS)
1162                         next = NULL;
1163
1164                 /* Inside a block? */
1165                 if ((prev &&
1166                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1167                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1168                     (next &&
1169                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1170                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1171                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1172                         continue;
1173                 }
1174
1175                 /* Check if we are at an interesting bound: */
1176                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1177                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1178                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1179                         continue;
1180                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1181                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1182                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1183                         continue;
1184
1185                 /*
1186                  * The boundary to prev and next are not interesting,
1187                  * so this line is not interesting as a whole
1188                  */
1189                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1190         }
1191 }
1192
1193 static void emit_line_ws_markup(struct diff_options *o,
1194                                 const char *set_sign, const char *set,
1195                                 const char *reset,
1196                                 char sign, const char *line, int len,
1197                                 unsigned ws_rule, int blank_at_eof)
1198 {
1199         const char *ws = NULL;
1200
1201         if (o->ws_error_highlight & ws_rule) {
1202                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1203                 if (!*ws)
1204                         ws = NULL;
1205         }
1206
1207         if (!ws && !set_sign)
1208                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1209         else if (!ws) {
1210                 /* Emit just the prefix, then the rest. */
1211                 emit_line_0(o, set_sign, NULL, !!set_sign, reset, sign, "", 0);
1212                 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
1213         } else if (blank_at_eof)
1214                 /* Blank line at EOF - paint '+' as well */
1215                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1216         else {
1217                 /* Emit just the prefix, then the rest. */
1218                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1219                             sign, "", 0);
1220                 ws_check_emit(line, len, ws_rule,
1221                               o->file, set, reset, ws);
1222         }
1223 }
1224
1225 static void emit_diff_symbol_from_struct(struct diff_options *o,
1226                                          struct emitted_diff_symbol *eds)
1227 {
1228         static const char *nneof = " No newline at end of file\n";
1229         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1230         struct strbuf sb = STRBUF_INIT;
1231
1232         enum diff_symbol s = eds->s;
1233         const char *line = eds->line;
1234         int len = eds->len;
1235         unsigned flags = eds->flags;
1236
1237         switch (s) {
1238         case DIFF_SYMBOL_NO_LF_EOF:
1239                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1240                 reset = diff_get_color_opt(o, DIFF_RESET);
1241                 putc('\n', o->file);
1242                 emit_line_0(o, context, NULL, 0, reset, '\\',
1243                             nneof, strlen(nneof));
1244                 break;
1245         case DIFF_SYMBOL_SUBMODULE_HEADER:
1246         case DIFF_SYMBOL_SUBMODULE_ERROR:
1247         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1248         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1249         case DIFF_SYMBOL_SUMMARY:
1250         case DIFF_SYMBOL_STATS_LINE:
1251         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1252         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1253                 emit_line(o, "", "", line, len);
1254                 break;
1255         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1256         case DIFF_SYMBOL_CONTEXT_MARKER:
1257                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1258                 reset = diff_get_color_opt(o, DIFF_RESET);
1259                 emit_line(o, context, reset, line, len);
1260                 break;
1261         case DIFF_SYMBOL_SEPARATOR:
1262                 fprintf(o->file, "%s%c",
1263                         diff_line_prefix(o),
1264                         o->line_termination);
1265                 break;
1266         case DIFF_SYMBOL_CONTEXT:
1267                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1268                 reset = diff_get_color_opt(o, DIFF_RESET);
1269                 set_sign = NULL;
1270                 if (o->flags.dual_color_diffed_diffs) {
1271                         char c = !len ? 0 : line[0];
1272
1273                         if (c == '+')
1274                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1275                         else if (c == '@')
1276                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1277                         else if (c == '-')
1278                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1279                 }
1280                 emit_line_ws_markup(o, set_sign, set, reset, ' ', line, len,
1281                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1282                 break;
1283         case DIFF_SYMBOL_PLUS:
1284                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1285                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1286                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1287                 case DIFF_SYMBOL_MOVED_LINE |
1288                      DIFF_SYMBOL_MOVED_LINE_ALT |
1289                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1290                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1291                         break;
1292                 case DIFF_SYMBOL_MOVED_LINE |
1293                      DIFF_SYMBOL_MOVED_LINE_ALT:
1294                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1295                         break;
1296                 case DIFF_SYMBOL_MOVED_LINE |
1297                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1298                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1299                         break;
1300                 case DIFF_SYMBOL_MOVED_LINE:
1301                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1302                         break;
1303                 default:
1304                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1305                 }
1306                 reset = diff_get_color_opt(o, DIFF_RESET);
1307                 if (!o->flags.dual_color_diffed_diffs)
1308                         set_sign = NULL;
1309                 else {
1310                         char c = !len ? 0 : line[0];
1311
1312                         set_sign = set;
1313                         if (c == '-')
1314                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1315                         else if (c == '@')
1316                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1317                         else if (c == '+')
1318                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1319                         else
1320                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1321                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1322                 }
1323                 emit_line_ws_markup(o, set_sign, set, reset, '+', line, len,
1324                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1325                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1326                 break;
1327         case DIFF_SYMBOL_MINUS:
1328                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1329                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1330                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1331                 case DIFF_SYMBOL_MOVED_LINE |
1332                      DIFF_SYMBOL_MOVED_LINE_ALT |
1333                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1334                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1335                         break;
1336                 case DIFF_SYMBOL_MOVED_LINE |
1337                      DIFF_SYMBOL_MOVED_LINE_ALT:
1338                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1339                         break;
1340                 case DIFF_SYMBOL_MOVED_LINE |
1341                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1342                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1343                         break;
1344                 case DIFF_SYMBOL_MOVED_LINE:
1345                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1346                         break;
1347                 default:
1348                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1349                 }
1350                 reset = diff_get_color_opt(o, DIFF_RESET);
1351                 if (!o->flags.dual_color_diffed_diffs)
1352                         set_sign = NULL;
1353                 else {
1354                         char c = !len ? 0 : line[0];
1355
1356                         set_sign = set;
1357                         if (c == '+')
1358                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1359                         else if (c == '@')
1360                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1361                         else if (c == '-')
1362                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1363                         else
1364                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1365                 }
1366                 emit_line_ws_markup(o, set_sign, set, reset, '-', line, len,
1367                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1368                 break;
1369         case DIFF_SYMBOL_WORDS_PORCELAIN:
1370                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1371                 reset = diff_get_color_opt(o, DIFF_RESET);
1372                 emit_line(o, context, reset, line, len);
1373                 fputs("~\n", o->file);
1374                 break;
1375         case DIFF_SYMBOL_WORDS:
1376                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1377                 reset = diff_get_color_opt(o, DIFF_RESET);
1378                 /*
1379                  * Skip the prefix character, if any.  With
1380                  * diff_suppress_blank_empty, there may be
1381                  * none.
1382                  */
1383                 if (line[0] != '\n') {
1384                         line++;
1385                         len--;
1386                 }
1387                 emit_line(o, context, reset, line, len);
1388                 break;
1389         case DIFF_SYMBOL_FILEPAIR_PLUS:
1390                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1391                 reset = diff_get_color_opt(o, DIFF_RESET);
1392                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1393                         line, reset,
1394                         strchr(line, ' ') ? "\t" : "");
1395                 break;
1396         case DIFF_SYMBOL_FILEPAIR_MINUS:
1397                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1398                 reset = diff_get_color_opt(o, DIFF_RESET);
1399                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1400                         line, reset,
1401                         strchr(line, ' ') ? "\t" : "");
1402                 break;
1403         case DIFF_SYMBOL_BINARY_FILES:
1404         case DIFF_SYMBOL_HEADER:
1405                 fprintf(o->file, "%s", line);
1406                 break;
1407         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1408                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1409                 break;
1410         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1411                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1412                 break;
1413         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1414                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1415                 break;
1416         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1417                 fputs(diff_line_prefix(o), o->file);
1418                 fputc('\n', o->file);
1419                 break;
1420         case DIFF_SYMBOL_REWRITE_DIFF:
1421                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1422                 reset = diff_get_color_opt(o, DIFF_RESET);
1423                 emit_line(o, fraginfo, reset, line, len);
1424                 break;
1425         case DIFF_SYMBOL_SUBMODULE_ADD:
1426                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1427                 reset = diff_get_color_opt(o, DIFF_RESET);
1428                 emit_line(o, set, reset, line, len);
1429                 break;
1430         case DIFF_SYMBOL_SUBMODULE_DEL:
1431                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1432                 reset = diff_get_color_opt(o, DIFF_RESET);
1433                 emit_line(o, set, reset, line, len);
1434                 break;
1435         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1436                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1437                         diff_line_prefix(o), line);
1438                 break;
1439         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1440                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1441                         diff_line_prefix(o), line);
1442                 break;
1443         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1444                 emit_line(o, "", "", " 0 files changed\n",
1445                           strlen(" 0 files changed\n"));
1446                 break;
1447         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1448                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1449                 break;
1450         case DIFF_SYMBOL_WORD_DIFF:
1451                 fprintf(o->file, "%.*s", len, line);
1452                 break;
1453         case DIFF_SYMBOL_STAT_SEP:
1454                 fputs(o->stat_sep, o->file);
1455                 break;
1456         default:
1457                 BUG("unknown diff symbol");
1458         }
1459         strbuf_release(&sb);
1460 }
1461
1462 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1463                              const char *line, int len, unsigned flags)
1464 {
1465         struct emitted_diff_symbol e = {line, len, flags, s};
1466
1467         if (o->emitted_symbols)
1468                 append_emitted_diff_symbol(o, &e);
1469         else
1470                 emit_diff_symbol_from_struct(o, &e);
1471 }
1472
1473 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1474 {
1475         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1476 }
1477
1478 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1479 {
1480         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1481 }
1482
1483 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1484 {
1485         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1486                          path, strlen(path), 0);
1487 }
1488
1489 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1490 {
1491         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1492                          path, strlen(path), 0);
1493 }
1494
1495 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1496 {
1497         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1498                          header, strlen(header), 0);
1499 }
1500
1501 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1502 {
1503         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1504 }
1505
1506 void diff_emit_submodule_pipethrough(struct diff_options *o,
1507                                      const char *line, int len)
1508 {
1509         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1510 }
1511
1512 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1513 {
1514         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1515               ecbdata->blank_at_eof_in_preimage &&
1516               ecbdata->blank_at_eof_in_postimage &&
1517               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1518               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1519                 return 0;
1520         return ws_blank_line(line, len, ecbdata->ws_rule);
1521 }
1522
1523 static void emit_add_line(const char *reset,
1524                           struct emit_callback *ecbdata,
1525                           const char *line, int len)
1526 {
1527         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1528         if (new_blank_line_at_eof(ecbdata, line, len))
1529                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1530
1531         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1532 }
1533
1534 static void emit_del_line(const char *reset,
1535                           struct emit_callback *ecbdata,
1536                           const char *line, int len)
1537 {
1538         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1539         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1540 }
1541
1542 static void emit_context_line(const char *reset,
1543                               struct emit_callback *ecbdata,
1544                               const char *line, int len)
1545 {
1546         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1547         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1548 }
1549
1550 static void emit_hunk_header(struct emit_callback *ecbdata,
1551                              const char *line, int len)
1552 {
1553         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1554         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1555         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1556         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1557         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1558         static const char atat[2] = { '@', '@' };
1559         const char *cp, *ep;
1560         struct strbuf msgbuf = STRBUF_INIT;
1561         int org_len = len;
1562         int i = 1;
1563
1564         /*
1565          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1566          * it always is at least 10 bytes long.
1567          */
1568         if (len < 10 ||
1569             memcmp(line, atat, 2) ||
1570             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1571                 emit_diff_symbol(ecbdata->opt,
1572                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1573                 return;
1574         }
1575         ep += 2; /* skip over @@ */
1576
1577         /* The hunk header in fraginfo color */
1578         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1579                 strbuf_addstr(&msgbuf, reverse);
1580         strbuf_addstr(&msgbuf, frag);
1581         strbuf_add(&msgbuf, line, ep - line);
1582         strbuf_addstr(&msgbuf, reset);
1583
1584         /*
1585          * trailing "\r\n"
1586          */
1587         for ( ; i < 3; i++)
1588                 if (line[len - i] == '\r' || line[len - i] == '\n')
1589                         len--;
1590
1591         /* blank before the func header */
1592         for (cp = ep; ep - line < len; ep++)
1593                 if (*ep != ' ' && *ep != '\t')
1594                         break;
1595         if (ep != cp) {
1596                 strbuf_addstr(&msgbuf, context);
1597                 strbuf_add(&msgbuf, cp, ep - cp);
1598                 strbuf_addstr(&msgbuf, reset);
1599         }
1600
1601         if (ep < line + len) {
1602                 strbuf_addstr(&msgbuf, func);
1603                 strbuf_add(&msgbuf, ep, line + len - ep);
1604                 strbuf_addstr(&msgbuf, reset);
1605         }
1606
1607         strbuf_add(&msgbuf, line + len, org_len - len);
1608         strbuf_complete_line(&msgbuf);
1609         emit_diff_symbol(ecbdata->opt,
1610                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1611         strbuf_release(&msgbuf);
1612 }
1613
1614 static struct diff_tempfile *claim_diff_tempfile(void) {
1615         int i;
1616         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1617                 if (!diff_temp[i].name)
1618                         return diff_temp + i;
1619         BUG("diff is failing to clean up its tempfiles");
1620 }
1621
1622 static void remove_tempfile(void)
1623 {
1624         int i;
1625         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1626                 if (is_tempfile_active(diff_temp[i].tempfile))
1627                         delete_tempfile(&diff_temp[i].tempfile);
1628                 diff_temp[i].name = NULL;
1629         }
1630 }
1631
1632 static void add_line_count(struct strbuf *out, int count)
1633 {
1634         switch (count) {
1635         case 0:
1636                 strbuf_addstr(out, "0,0");
1637                 break;
1638         case 1:
1639                 strbuf_addstr(out, "1");
1640                 break;
1641         default:
1642                 strbuf_addf(out, "1,%d", count);
1643                 break;
1644         }
1645 }
1646
1647 static void emit_rewrite_lines(struct emit_callback *ecb,
1648                                int prefix, const char *data, int size)
1649 {
1650         const char *endp = NULL;
1651         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1652
1653         while (0 < size) {
1654                 int len;
1655
1656                 endp = memchr(data, '\n', size);
1657                 len = endp ? (endp - data + 1) : size;
1658                 if (prefix != '+') {
1659                         ecb->lno_in_preimage++;
1660                         emit_del_line(reset, ecb, data, len);
1661                 } else {
1662                         ecb->lno_in_postimage++;
1663                         emit_add_line(reset, ecb, data, len);
1664                 }
1665                 size -= len;
1666                 data += len;
1667         }
1668         if (!endp)
1669                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1670 }
1671
1672 static void emit_rewrite_diff(const char *name_a,
1673                               const char *name_b,
1674                               struct diff_filespec *one,
1675                               struct diff_filespec *two,
1676                               struct userdiff_driver *textconv_one,
1677                               struct userdiff_driver *textconv_two,
1678                               struct diff_options *o)
1679 {
1680         int lc_a, lc_b;
1681         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1682         const char *a_prefix, *b_prefix;
1683         char *data_one, *data_two;
1684         size_t size_one, size_two;
1685         struct emit_callback ecbdata;
1686         struct strbuf out = STRBUF_INIT;
1687
1688         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1689                 a_prefix = o->b_prefix;
1690                 b_prefix = o->a_prefix;
1691         } else {
1692                 a_prefix = o->a_prefix;
1693                 b_prefix = o->b_prefix;
1694         }
1695
1696         name_a += (*name_a == '/');
1697         name_b += (*name_b == '/');
1698
1699         strbuf_reset(&a_name);
1700         strbuf_reset(&b_name);
1701         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1702         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1703
1704         size_one = fill_textconv(textconv_one, one, &data_one);
1705         size_two = fill_textconv(textconv_two, two, &data_two);
1706
1707         memset(&ecbdata, 0, sizeof(ecbdata));
1708         ecbdata.color_diff = want_color(o->use_color);
1709         ecbdata.ws_rule = whitespace_rule(name_b);
1710         ecbdata.opt = o;
1711         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1712                 mmfile_t mf1, mf2;
1713                 mf1.ptr = (char *)data_one;
1714                 mf2.ptr = (char *)data_two;
1715                 mf1.size = size_one;
1716                 mf2.size = size_two;
1717                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1718         }
1719         ecbdata.lno_in_preimage = 1;
1720         ecbdata.lno_in_postimage = 1;
1721
1722         lc_a = count_lines(data_one, size_one);
1723         lc_b = count_lines(data_two, size_two);
1724
1725         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1726                          a_name.buf, a_name.len, 0);
1727         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1728                          b_name.buf, b_name.len, 0);
1729
1730         strbuf_addstr(&out, "@@ -");
1731         if (!o->irreversible_delete)
1732                 add_line_count(&out, lc_a);
1733         else
1734                 strbuf_addstr(&out, "?,?");
1735         strbuf_addstr(&out, " +");
1736         add_line_count(&out, lc_b);
1737         strbuf_addstr(&out, " @@\n");
1738         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1739         strbuf_release(&out);
1740
1741         if (lc_a && !o->irreversible_delete)
1742                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1743         if (lc_b)
1744                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1745         if (textconv_one)
1746                 free((char *)data_one);
1747         if (textconv_two)
1748                 free((char *)data_two);
1749 }
1750
1751 struct diff_words_buffer {
1752         mmfile_t text;
1753         unsigned long alloc;
1754         struct diff_words_orig {
1755                 const char *begin, *end;
1756         } *orig;
1757         int orig_nr, orig_alloc;
1758 };
1759
1760 static void diff_words_append(char *line, unsigned long len,
1761                 struct diff_words_buffer *buffer)
1762 {
1763         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1764         line++;
1765         len--;
1766         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1767         buffer->text.size += len;
1768         buffer->text.ptr[buffer->text.size] = '\0';
1769 }
1770
1771 struct diff_words_style_elem {
1772         const char *prefix;
1773         const char *suffix;
1774         const char *color; /* NULL; filled in by the setup code if
1775                             * color is enabled */
1776 };
1777
1778 struct diff_words_style {
1779         enum diff_words_type type;
1780         struct diff_words_style_elem new_word, old_word, ctx;
1781         const char *newline;
1782 };
1783
1784 static struct diff_words_style diff_words_styles[] = {
1785         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1786         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1787         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1788 };
1789
1790 struct diff_words_data {
1791         struct diff_words_buffer minus, plus;
1792         const char *current_plus;
1793         int last_minus;
1794         struct diff_options *opt;
1795         regex_t *word_regex;
1796         enum diff_words_type type;
1797         struct diff_words_style *style;
1798 };
1799
1800 static int fn_out_diff_words_write_helper(struct diff_options *o,
1801                                           struct diff_words_style_elem *st_el,
1802                                           const char *newline,
1803                                           size_t count, const char *buf)
1804 {
1805         int print = 0;
1806         struct strbuf sb = STRBUF_INIT;
1807
1808         while (count) {
1809                 char *p = memchr(buf, '\n', count);
1810                 if (print)
1811                         strbuf_addstr(&sb, diff_line_prefix(o));
1812
1813                 if (p != buf) {
1814                         const char *reset = st_el->color && *st_el->color ?
1815                                             GIT_COLOR_RESET : NULL;
1816                         if (st_el->color && *st_el->color)
1817                                 strbuf_addstr(&sb, st_el->color);
1818                         strbuf_addstr(&sb, st_el->prefix);
1819                         strbuf_add(&sb, buf, p ? p - buf : count);
1820                         strbuf_addstr(&sb, st_el->suffix);
1821                         if (reset)
1822                                 strbuf_addstr(&sb, reset);
1823                 }
1824                 if (!p)
1825                         goto out;
1826
1827                 strbuf_addstr(&sb, newline);
1828                 count -= p + 1 - buf;
1829                 buf = p + 1;
1830                 print = 1;
1831                 if (count) {
1832                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1833                                          sb.buf, sb.len, 0);
1834                         strbuf_reset(&sb);
1835                 }
1836         }
1837
1838 out:
1839         if (sb.len)
1840                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1841                                  sb.buf, sb.len, 0);
1842         strbuf_release(&sb);
1843         return 0;
1844 }
1845
1846 /*
1847  * '--color-words' algorithm can be described as:
1848  *
1849  *   1. collect the minus/plus lines of a diff hunk, divided into
1850  *      minus-lines and plus-lines;
1851  *
1852  *   2. break both minus-lines and plus-lines into words and
1853  *      place them into two mmfile_t with one word for each line;
1854  *
1855  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1856  *
1857  * And for the common parts of the both file, we output the plus side text.
1858  * diff_words->current_plus is used to trace the current position of the plus file
1859  * which printed. diff_words->last_minus is used to trace the last minus word
1860  * printed.
1861  *
1862  * For '--graph' to work with '--color-words', we need to output the graph prefix
1863  * on each line of color words output. Generally, there are two conditions on
1864  * which we should output the prefix.
1865  *
1866  *   1. diff_words->last_minus == 0 &&
1867  *      diff_words->current_plus == diff_words->plus.text.ptr
1868  *
1869  *      that is: the plus text must start as a new line, and if there is no minus
1870  *      word printed, a graph prefix must be printed.
1871  *
1872  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1873  *      *(diff_words->current_plus - 1) == '\n'
1874  *
1875  *      that is: a graph prefix must be printed following a '\n'
1876  */
1877 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1878 {
1879         if ((diff_words->last_minus == 0 &&
1880                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1881                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1882                 *(diff_words->current_plus - 1) == '\n')) {
1883                 return 1;
1884         } else {
1885                 return 0;
1886         }
1887 }
1888
1889 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1890 {
1891         struct diff_words_data *diff_words = priv;
1892         struct diff_words_style *style = diff_words->style;
1893         int minus_first, minus_len, plus_first, plus_len;
1894         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1895         struct diff_options *opt = diff_words->opt;
1896         const char *line_prefix;
1897
1898         if (line[0] != '@' || parse_hunk_header(line, len,
1899                         &minus_first, &minus_len, &plus_first, &plus_len))
1900                 return;
1901
1902         assert(opt);
1903         line_prefix = diff_line_prefix(opt);
1904
1905         /* POSIX requires that first be decremented by one if len == 0... */
1906         if (minus_len) {
1907                 minus_begin = diff_words->minus.orig[minus_first].begin;
1908                 minus_end =
1909                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1910         } else
1911                 minus_begin = minus_end =
1912                         diff_words->minus.orig[minus_first].end;
1913
1914         if (plus_len) {
1915                 plus_begin = diff_words->plus.orig[plus_first].begin;
1916                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1917         } else
1918                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1919
1920         if (color_words_output_graph_prefix(diff_words)) {
1921                 fputs(line_prefix, diff_words->opt->file);
1922         }
1923         if (diff_words->current_plus != plus_begin) {
1924                 fn_out_diff_words_write_helper(diff_words->opt,
1925                                 &style->ctx, style->newline,
1926                                 plus_begin - diff_words->current_plus,
1927                                 diff_words->current_plus);
1928         }
1929         if (minus_begin != minus_end) {
1930                 fn_out_diff_words_write_helper(diff_words->opt,
1931                                 &style->old_word, style->newline,
1932                                 minus_end - minus_begin, minus_begin);
1933         }
1934         if (plus_begin != plus_end) {
1935                 fn_out_diff_words_write_helper(diff_words->opt,
1936                                 &style->new_word, style->newline,
1937                                 plus_end - plus_begin, plus_begin);
1938         }
1939
1940         diff_words->current_plus = plus_end;
1941         diff_words->last_minus = minus_first;
1942 }
1943
1944 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1945 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1946                 int *begin, int *end)
1947 {
1948         if (word_regex && *begin < buffer->size) {
1949                 regmatch_t match[1];
1950                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1951                                  buffer->size - *begin, 1, match, 0)) {
1952                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1953                                         '\n', match[0].rm_eo - match[0].rm_so);
1954                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1955                         *begin += match[0].rm_so;
1956                         return *begin >= *end;
1957                 }
1958                 return -1;
1959         }
1960
1961         /* find the next word */
1962         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1963                 (*begin)++;
1964         if (*begin >= buffer->size)
1965                 return -1;
1966
1967         /* find the end of the word */
1968         *end = *begin + 1;
1969         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1970                 (*end)++;
1971
1972         return 0;
1973 }
1974
1975 /*
1976  * This function splits the words in buffer->text, stores the list with
1977  * newline separator into out, and saves the offsets of the original words
1978  * in buffer->orig.
1979  */
1980 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1981                 regex_t *word_regex)
1982 {
1983         int i, j;
1984         long alloc = 0;
1985
1986         out->size = 0;
1987         out->ptr = NULL;
1988
1989         /* fake an empty "0th" word */
1990         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1991         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1992         buffer->orig_nr = 1;
1993
1994         for (i = 0; i < buffer->text.size; i++) {
1995                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1996                         return;
1997
1998                 /* store original boundaries */
1999                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2000                                 buffer->orig_alloc);
2001                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2002                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2003                 buffer->orig_nr++;
2004
2005                 /* store one word */
2006                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2007                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2008                 out->ptr[out->size + j - i] = '\n';
2009                 out->size += j - i + 1;
2010
2011                 i = j - 1;
2012         }
2013 }
2014
2015 /* this executes the word diff on the accumulated buffers */
2016 static void diff_words_show(struct diff_words_data *diff_words)
2017 {
2018         xpparam_t xpp;
2019         xdemitconf_t xecfg;
2020         mmfile_t minus, plus;
2021         struct diff_words_style *style = diff_words->style;
2022
2023         struct diff_options *opt = diff_words->opt;
2024         const char *line_prefix;
2025
2026         assert(opt);
2027         line_prefix = diff_line_prefix(opt);
2028
2029         /* special case: only removal */
2030         if (!diff_words->plus.text.size) {
2031                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2032                                  line_prefix, strlen(line_prefix), 0);
2033                 fn_out_diff_words_write_helper(diff_words->opt,
2034                         &style->old_word, style->newline,
2035                         diff_words->minus.text.size,
2036                         diff_words->minus.text.ptr);
2037                 diff_words->minus.text.size = 0;
2038                 return;
2039         }
2040
2041         diff_words->current_plus = diff_words->plus.text.ptr;
2042         diff_words->last_minus = 0;
2043
2044         memset(&xpp, 0, sizeof(xpp));
2045         memset(&xecfg, 0, sizeof(xecfg));
2046         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2047         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2048         xpp.flags = 0;
2049         /* as only the hunk header will be parsed, we need a 0-context */
2050         xecfg.ctxlen = 0;
2051         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2052                           &xpp, &xecfg))
2053                 die("unable to generate word diff");
2054         free(minus.ptr);
2055         free(plus.ptr);
2056         if (diff_words->current_plus != diff_words->plus.text.ptr +
2057                         diff_words->plus.text.size) {
2058                 if (color_words_output_graph_prefix(diff_words))
2059                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2060                                          line_prefix, strlen(line_prefix), 0);
2061                 fn_out_diff_words_write_helper(diff_words->opt,
2062                         &style->ctx, style->newline,
2063                         diff_words->plus.text.ptr + diff_words->plus.text.size
2064                         - diff_words->current_plus, diff_words->current_plus);
2065         }
2066         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2067 }
2068
2069 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2070 static void diff_words_flush(struct emit_callback *ecbdata)
2071 {
2072         struct diff_options *wo = ecbdata->diff_words->opt;
2073
2074         if (ecbdata->diff_words->minus.text.size ||
2075             ecbdata->diff_words->plus.text.size)
2076                 diff_words_show(ecbdata->diff_words);
2077
2078         if (wo->emitted_symbols) {
2079                 struct diff_options *o = ecbdata->opt;
2080                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2081                 int i;
2082
2083                 /*
2084                  * NEEDSWORK:
2085                  * Instead of appending each, concat all words to a line?
2086                  */
2087                 for (i = 0; i < wol->nr; i++)
2088                         append_emitted_diff_symbol(o, &wol->buf[i]);
2089
2090                 for (i = 0; i < wol->nr; i++)
2091                         free((void *)wol->buf[i].line);
2092
2093                 wol->nr = 0;
2094         }
2095 }
2096
2097 static void diff_filespec_load_driver(struct diff_filespec *one)
2098 {
2099         /* Use already-loaded driver */
2100         if (one->driver)
2101                 return;
2102
2103         if (S_ISREG(one->mode))
2104                 one->driver = userdiff_find_by_path(one->path);
2105
2106         /* Fallback to default settings */
2107         if (!one->driver)
2108                 one->driver = userdiff_find_by_name("default");
2109 }
2110
2111 static const char *userdiff_word_regex(struct diff_filespec *one)
2112 {
2113         diff_filespec_load_driver(one);
2114         return one->driver->word_regex;
2115 }
2116
2117 static void init_diff_words_data(struct emit_callback *ecbdata,
2118                                  struct diff_options *orig_opts,
2119                                  struct diff_filespec *one,
2120                                  struct diff_filespec *two)
2121 {
2122         int i;
2123         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2124         memcpy(o, orig_opts, sizeof(struct diff_options));
2125
2126         ecbdata->diff_words =
2127                 xcalloc(1, sizeof(struct diff_words_data));
2128         ecbdata->diff_words->type = o->word_diff;
2129         ecbdata->diff_words->opt = o;
2130
2131         if (orig_opts->emitted_symbols)
2132                 o->emitted_symbols =
2133                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2134
2135         if (!o->word_regex)
2136                 o->word_regex = userdiff_word_regex(one);
2137         if (!o->word_regex)
2138                 o->word_regex = userdiff_word_regex(two);
2139         if (!o->word_regex)
2140                 o->word_regex = diff_word_regex_cfg;
2141         if (o->word_regex) {
2142                 ecbdata->diff_words->word_regex = (regex_t *)
2143                         xmalloc(sizeof(regex_t));
2144                 if (regcomp(ecbdata->diff_words->word_regex,
2145                             o->word_regex,
2146                             REG_EXTENDED | REG_NEWLINE))
2147                         die ("Invalid regular expression: %s",
2148                              o->word_regex);
2149         }
2150         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2151                 if (o->word_diff == diff_words_styles[i].type) {
2152                         ecbdata->diff_words->style =
2153                                 &diff_words_styles[i];
2154                         break;
2155                 }
2156         }
2157         if (want_color(o->use_color)) {
2158                 struct diff_words_style *st = ecbdata->diff_words->style;
2159                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2160                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2161                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2162         }
2163 }
2164
2165 static void free_diff_words_data(struct emit_callback *ecbdata)
2166 {
2167         if (ecbdata->diff_words) {
2168                 diff_words_flush(ecbdata);
2169                 free (ecbdata->diff_words->opt->emitted_symbols);
2170                 free (ecbdata->diff_words->opt);
2171                 free (ecbdata->diff_words->minus.text.ptr);
2172                 free (ecbdata->diff_words->minus.orig);
2173                 free (ecbdata->diff_words->plus.text.ptr);
2174                 free (ecbdata->diff_words->plus.orig);
2175                 if (ecbdata->diff_words->word_regex) {
2176                         regfree(ecbdata->diff_words->word_regex);
2177                         free(ecbdata->diff_words->word_regex);
2178                 }
2179                 FREE_AND_NULL(ecbdata->diff_words);
2180         }
2181 }
2182
2183 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2184 {
2185         if (want_color(diff_use_color))
2186                 return diff_colors[ix];
2187         return "";
2188 }
2189
2190 const char *diff_line_prefix(struct diff_options *opt)
2191 {
2192         struct strbuf *msgbuf;
2193         if (!opt->output_prefix)
2194                 return "";
2195
2196         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2197         return msgbuf->buf;
2198 }
2199
2200 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2201 {
2202         const char *cp;
2203         unsigned long allot;
2204         size_t l = len;
2205
2206         cp = line;
2207         allot = l;
2208         while (0 < l) {
2209                 (void) utf8_width(&cp, &l);
2210                 if (!cp)
2211                         break; /* truncated in the middle? */
2212         }
2213         return allot - l;
2214 }
2215
2216 static void find_lno(const char *line, struct emit_callback *ecbdata)
2217 {
2218         const char *p;
2219         ecbdata->lno_in_preimage = 0;
2220         ecbdata->lno_in_postimage = 0;
2221         p = strchr(line, '-');
2222         if (!p)
2223                 return; /* cannot happen */
2224         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2225         p = strchr(p, '+');
2226         if (!p)
2227                 return; /* cannot happen */
2228         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2229 }
2230
2231 static void fn_out_consume(void *priv, char *line, unsigned long len)
2232 {
2233         struct emit_callback *ecbdata = priv;
2234         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2235         struct diff_options *o = ecbdata->opt;
2236
2237         o->found_changes = 1;
2238
2239         if (ecbdata->header) {
2240                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2241                                  ecbdata->header->buf, ecbdata->header->len, 0);
2242                 strbuf_reset(ecbdata->header);
2243                 ecbdata->header = NULL;
2244         }
2245
2246         if (ecbdata->label_path[0]) {
2247                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2248                                  ecbdata->label_path[0],
2249                                  strlen(ecbdata->label_path[0]), 0);
2250                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2251                                  ecbdata->label_path[1],
2252                                  strlen(ecbdata->label_path[1]), 0);
2253                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2254         }
2255
2256         if (diff_suppress_blank_empty
2257             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2258                 line[0] = '\n';
2259                 len = 1;
2260         }
2261
2262         if (line[0] == '@') {
2263                 if (ecbdata->diff_words)
2264                         diff_words_flush(ecbdata);
2265                 len = sane_truncate_line(ecbdata, line, len);
2266                 find_lno(line, ecbdata);
2267                 emit_hunk_header(ecbdata, line, len);
2268                 return;
2269         }
2270
2271         if (ecbdata->diff_words) {
2272                 enum diff_symbol s =
2273                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2274                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2275                 if (line[0] == '-') {
2276                         diff_words_append(line, len,
2277                                           &ecbdata->diff_words->minus);
2278                         return;
2279                 } else if (line[0] == '+') {
2280                         diff_words_append(line, len,
2281                                           &ecbdata->diff_words->plus);
2282                         return;
2283                 } else if (starts_with(line, "\\ ")) {
2284                         /*
2285                          * Eat the "no newline at eof" marker as if we
2286                          * saw a "+" or "-" line with nothing on it,
2287                          * and return without diff_words_flush() to
2288                          * defer processing. If this is the end of
2289                          * preimage, more "+" lines may come after it.
2290                          */
2291                         return;
2292                 }
2293                 diff_words_flush(ecbdata);
2294                 emit_diff_symbol(o, s, line, len, 0);
2295                 return;
2296         }
2297
2298         switch (line[0]) {
2299         case '+':
2300                 ecbdata->lno_in_postimage++;
2301                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2302                 break;
2303         case '-':
2304                 ecbdata->lno_in_preimage++;
2305                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2306                 break;
2307         case ' ':
2308                 ecbdata->lno_in_postimage++;
2309                 ecbdata->lno_in_preimage++;
2310                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2311                 break;
2312         default:
2313                 /* incomplete line at the end */
2314                 ecbdata->lno_in_preimage++;
2315                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2316                                  line, len, 0);
2317                 break;
2318         }
2319 }
2320
2321 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2322 {
2323         const char *old_name = a;
2324         const char *new_name = b;
2325         int pfx_length, sfx_length;
2326         int pfx_adjust_for_slash;
2327         int len_a = strlen(a);
2328         int len_b = strlen(b);
2329         int a_midlen, b_midlen;
2330         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2331         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2332
2333         if (qlen_a || qlen_b) {
2334                 quote_c_style(a, name, NULL, 0);
2335                 strbuf_addstr(name, " => ");
2336                 quote_c_style(b, name, NULL, 0);
2337                 return;
2338         }
2339
2340         /* Find common prefix */
2341         pfx_length = 0;
2342         while (*old_name && *new_name && *old_name == *new_name) {
2343                 if (*old_name == '/')
2344                         pfx_length = old_name - a + 1;
2345                 old_name++;
2346                 new_name++;
2347         }
2348
2349         /* Find common suffix */
2350         old_name = a + len_a;
2351         new_name = b + len_b;
2352         sfx_length = 0;
2353         /*
2354          * If there is a common prefix, it must end in a slash.  In
2355          * that case we let this loop run 1 into the prefix to see the
2356          * same slash.
2357          *
2358          * If there is no common prefix, we cannot do this as it would
2359          * underrun the input strings.
2360          */
2361         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2362         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2363                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2364                *old_name == *new_name) {
2365                 if (*old_name == '/')
2366                         sfx_length = len_a - (old_name - a);
2367                 old_name--;
2368                 new_name--;
2369         }
2370
2371         /*
2372          * pfx{mid-a => mid-b}sfx
2373          * {pfx-a => pfx-b}sfx
2374          * pfx{sfx-a => sfx-b}
2375          * name-a => name-b
2376          */
2377         a_midlen = len_a - pfx_length - sfx_length;
2378         b_midlen = len_b - pfx_length - sfx_length;
2379         if (a_midlen < 0)
2380                 a_midlen = 0;
2381         if (b_midlen < 0)
2382                 b_midlen = 0;
2383
2384         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2385         if (pfx_length + sfx_length) {
2386                 strbuf_add(name, a, pfx_length);
2387                 strbuf_addch(name, '{');
2388         }
2389         strbuf_add(name, a + pfx_length, a_midlen);
2390         strbuf_addstr(name, " => ");
2391         strbuf_add(name, b + pfx_length, b_midlen);
2392         if (pfx_length + sfx_length) {
2393                 strbuf_addch(name, '}');
2394                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2395         }
2396 }
2397
2398 struct diffstat_t {
2399         int nr;
2400         int alloc;
2401         struct diffstat_file {
2402                 char *from_name;
2403                 char *name;
2404                 char *print_name;
2405                 const char *comments;
2406                 unsigned is_unmerged:1;
2407                 unsigned is_binary:1;
2408                 unsigned is_renamed:1;
2409                 unsigned is_interesting:1;
2410                 uintmax_t added, deleted;
2411         } **files;
2412 };
2413
2414 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2415                                           const char *name_a,
2416                                           const char *name_b)
2417 {
2418         struct diffstat_file *x;
2419         x = xcalloc(1, sizeof(*x));
2420         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2421         diffstat->files[diffstat->nr++] = x;
2422         if (name_b) {
2423                 x->from_name = xstrdup(name_a);
2424                 x->name = xstrdup(name_b);
2425                 x->is_renamed = 1;
2426         }
2427         else {
2428                 x->from_name = NULL;
2429                 x->name = xstrdup(name_a);
2430         }
2431         return x;
2432 }
2433
2434 static void diffstat_consume(void *priv, char *line, unsigned long len)
2435 {
2436         struct diffstat_t *diffstat = priv;
2437         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2438
2439         if (line[0] == '+')
2440                 x->added++;
2441         else if (line[0] == '-')
2442                 x->deleted++;
2443 }
2444
2445 const char mime_boundary_leader[] = "------------";
2446
2447 static int scale_linear(int it, int width, int max_change)
2448 {
2449         if (!it)
2450                 return 0;
2451         /*
2452          * make sure that at least one '-' or '+' is printed if
2453          * there is any change to this path. The easiest way is to
2454          * scale linearly as if the alloted width is one column shorter
2455          * than it is, and then add 1 to the result.
2456          */
2457         return 1 + (it * (width - 1) / max_change);
2458 }
2459
2460 static void show_graph(struct strbuf *out, char ch, int cnt,
2461                        const char *set, const char *reset)
2462 {
2463         if (cnt <= 0)
2464                 return;
2465         strbuf_addstr(out, set);
2466         strbuf_addchars(out, ch, cnt);
2467         strbuf_addstr(out, reset);
2468 }
2469
2470 static void fill_print_name(struct diffstat_file *file)
2471 {
2472         struct strbuf pname = STRBUF_INIT;
2473
2474         if (file->print_name)
2475                 return;
2476
2477         if (file->is_renamed)
2478                 pprint_rename(&pname, file->from_name, file->name);
2479         else
2480                 quote_c_style(file->name, &pname, NULL, 0);
2481
2482         if (file->comments)
2483                 strbuf_addf(&pname, " (%s)", file->comments);
2484
2485         file->print_name = strbuf_detach(&pname, NULL);
2486 }
2487
2488 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2489                 int files, int insertions, int deletions)
2490 {
2491         struct strbuf sb = STRBUF_INIT;
2492
2493         if (!files) {
2494                 assert(insertions == 0 && deletions == 0);
2495                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2496                                  NULL, 0, 0);
2497                 return;
2498         }
2499
2500         strbuf_addf(&sb,
2501                     (files == 1) ? " %d file changed" : " %d files changed",
2502                     files);
2503
2504         /*
2505          * For binary diff, the caller may want to print "x files
2506          * changed" with insertions == 0 && deletions == 0.
2507          *
2508          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2509          * is probably less confusing (i.e skip over "2 files changed
2510          * but nothing about added/removed lines? Is this a bug in Git?").
2511          */
2512         if (insertions || deletions == 0) {
2513                 strbuf_addf(&sb,
2514                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2515                             insertions);
2516         }
2517
2518         if (deletions || insertions == 0) {
2519                 strbuf_addf(&sb,
2520                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2521                             deletions);
2522         }
2523         strbuf_addch(&sb, '\n');
2524         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2525                          sb.buf, sb.len, 0);
2526         strbuf_release(&sb);
2527 }
2528
2529 void print_stat_summary(FILE *fp, int files,
2530                         int insertions, int deletions)
2531 {
2532         struct diff_options o;
2533         memset(&o, 0, sizeof(o));
2534         o.file = fp;
2535
2536         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2537 }
2538
2539 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2540 {
2541         int i, len, add, del, adds = 0, dels = 0;
2542         uintmax_t max_change = 0, max_len = 0;
2543         int total_files = data->nr, count;
2544         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2545         const char *reset, *add_c, *del_c;
2546         int extra_shown = 0;
2547         const char *line_prefix = diff_line_prefix(options);
2548         struct strbuf out = STRBUF_INIT;
2549
2550         if (data->nr == 0)
2551                 return;
2552
2553         count = options->stat_count ? options->stat_count : data->nr;
2554
2555         reset = diff_get_color_opt(options, DIFF_RESET);
2556         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2557         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2558
2559         /*
2560          * Find the longest filename and max number of changes
2561          */
2562         for (i = 0; (i < count) && (i < data->nr); i++) {
2563                 struct diffstat_file *file = data->files[i];
2564                 uintmax_t change = file->added + file->deleted;
2565
2566                 if (!file->is_interesting && (change == 0)) {
2567                         count++; /* not shown == room for one more */
2568                         continue;
2569                 }
2570                 fill_print_name(file);
2571                 len = strlen(file->print_name);
2572                 if (max_len < len)
2573                         max_len = len;
2574
2575                 if (file->is_unmerged) {
2576                         /* "Unmerged" is 8 characters */
2577                         bin_width = bin_width < 8 ? 8 : bin_width;
2578                         continue;
2579                 }
2580                 if (file->is_binary) {
2581                         /* "Bin XXX -> YYY bytes" */
2582                         int w = 14 + decimal_width(file->added)
2583                                 + decimal_width(file->deleted);
2584                         bin_width = bin_width < w ? w : bin_width;
2585                         /* Display change counts aligned with "Bin" */
2586                         number_width = 3;
2587                         continue;
2588                 }
2589
2590                 if (max_change < change)
2591                         max_change = change;
2592         }
2593         count = i; /* where we can stop scanning in data->files[] */
2594
2595         /*
2596          * We have width = stat_width or term_columns() columns total.
2597          * We want a maximum of min(max_len, stat_name_width) for the name part.
2598          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2599          * We also need 1 for " " and 4 + decimal_width(max_change)
2600          * for " | NNNN " and one the empty column at the end, altogether
2601          * 6 + decimal_width(max_change).
2602          *
2603          * If there's not enough space, we will use the smaller of
2604          * stat_name_width (if set) and 5/8*width for the filename,
2605          * and the rest for constant elements + graph part, but no more
2606          * than stat_graph_width for the graph part.
2607          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2608          * for the standard terminal size).
2609          *
2610          * In other words: stat_width limits the maximum width, and
2611          * stat_name_width fixes the maximum width of the filename,
2612          * and is also used to divide available columns if there
2613          * aren't enough.
2614          *
2615          * Binary files are displayed with "Bin XXX -> YYY bytes"
2616          * instead of the change count and graph. This part is treated
2617          * similarly to the graph part, except that it is not
2618          * "scaled". If total width is too small to accommodate the
2619          * guaranteed minimum width of the filename part and the
2620          * separators and this message, this message will "overflow"
2621          * making the line longer than the maximum width.
2622          */
2623
2624         if (options->stat_width == -1)
2625                 width = term_columns() - strlen(line_prefix);
2626         else
2627                 width = options->stat_width ? options->stat_width : 80;
2628         number_width = decimal_width(max_change) > number_width ?
2629                 decimal_width(max_change) : number_width;
2630
2631         if (options->stat_graph_width == -1)
2632                 options->stat_graph_width = diff_stat_graph_width;
2633
2634         /*
2635          * Guarantee 3/8*16==6 for the graph part
2636          * and 5/8*16==10 for the filename part
2637          */
2638         if (width < 16 + 6 + number_width)
2639                 width = 16 + 6 + number_width;
2640
2641         /*
2642          * First assign sizes that are wanted, ignoring available width.
2643          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2644          * starting from "XXX" should fit in graph_width.
2645          */
2646         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2647         if (options->stat_graph_width &&
2648             options->stat_graph_width < graph_width)
2649                 graph_width = options->stat_graph_width;
2650
2651         name_width = (options->stat_name_width > 0 &&
2652                       options->stat_name_width < max_len) ?
2653                 options->stat_name_width : max_len;
2654
2655         /*
2656          * Adjust adjustable widths not to exceed maximum width
2657          */
2658         if (name_width + number_width + 6 + graph_width > width) {
2659                 if (graph_width > width * 3/8 - number_width - 6) {
2660                         graph_width = width * 3/8 - number_width - 6;
2661                         if (graph_width < 6)
2662                                 graph_width = 6;
2663                 }
2664
2665                 if (options->stat_graph_width &&
2666                     graph_width > options->stat_graph_width)
2667                         graph_width = options->stat_graph_width;
2668                 if (name_width > width - number_width - 6 - graph_width)
2669                         name_width = width - number_width - 6 - graph_width;
2670                 else
2671                         graph_width = width - number_width - 6 - name_width;
2672         }
2673
2674         /*
2675          * From here name_width is the width of the name area,
2676          * and graph_width is the width of the graph area.
2677          * max_change is used to scale graph properly.
2678          */
2679         for (i = 0; i < count; i++) {
2680                 const char *prefix = "";
2681                 struct diffstat_file *file = data->files[i];
2682                 char *name = file->print_name;
2683                 uintmax_t added = file->added;
2684                 uintmax_t deleted = file->deleted;
2685                 int name_len;
2686
2687                 if (!file->is_interesting && (added + deleted == 0))
2688                         continue;
2689
2690                 /*
2691                  * "scale" the filename
2692                  */
2693                 len = name_width;
2694                 name_len = strlen(name);
2695                 if (name_width < name_len) {
2696                         char *slash;
2697                         prefix = "...";
2698                         len -= 3;
2699                         name += name_len - len;
2700                         slash = strchr(name, '/');
2701                         if (slash)
2702                                 name = slash;
2703                 }
2704
2705                 if (file->is_binary) {
2706                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2707                         strbuf_addf(&out, " %*s", number_width, "Bin");
2708                         if (!added && !deleted) {
2709                                 strbuf_addch(&out, '\n');
2710                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2711                                                  out.buf, out.len, 0);
2712                                 strbuf_reset(&out);
2713                                 continue;
2714                         }
2715                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2716                                 del_c, deleted, reset);
2717                         strbuf_addstr(&out, " -> ");
2718                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2719                                 add_c, added, reset);
2720                         strbuf_addstr(&out, " bytes\n");
2721                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2722                                          out.buf, out.len, 0);
2723                         strbuf_reset(&out);
2724                         continue;
2725                 }
2726                 else if (file->is_unmerged) {
2727                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2728                         strbuf_addstr(&out, " Unmerged\n");
2729                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2730                                          out.buf, out.len, 0);
2731                         strbuf_reset(&out);
2732                         continue;
2733                 }
2734
2735                 /*
2736                  * scale the add/delete
2737                  */
2738                 add = added;
2739                 del = deleted;
2740
2741                 if (graph_width <= max_change) {
2742                         int total = scale_linear(add + del, graph_width, max_change);
2743                         if (total < 2 && add && del)
2744                                 /* width >= 2 due to the sanity check */
2745                                 total = 2;
2746                         if (add < del) {
2747                                 add = scale_linear(add, graph_width, max_change);
2748                                 del = total - add;
2749                         } else {
2750                                 del = scale_linear(del, graph_width, max_change);
2751                                 add = total - del;
2752                         }
2753                 }
2754                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2755                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2756                         number_width, added + deleted,
2757                         added + deleted ? " " : "");
2758                 show_graph(&out, '+', add, add_c, reset);
2759                 show_graph(&out, '-', del, del_c, reset);
2760                 strbuf_addch(&out, '\n');
2761                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2762                                  out.buf, out.len, 0);
2763                 strbuf_reset(&out);
2764         }
2765
2766         for (i = 0; i < data->nr; i++) {
2767                 struct diffstat_file *file = data->files[i];
2768                 uintmax_t added = file->added;
2769                 uintmax_t deleted = file->deleted;
2770
2771                 if (file->is_unmerged ||
2772                     (!file->is_interesting && (added + deleted == 0))) {
2773                         total_files--;
2774                         continue;
2775                 }
2776
2777                 if (!file->is_binary) {
2778                         adds += added;
2779                         dels += deleted;
2780                 }
2781                 if (i < count)
2782                         continue;
2783                 if (!extra_shown)
2784                         emit_diff_symbol(options,
2785                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2786                                          NULL, 0, 0);
2787                 extra_shown = 1;
2788         }
2789
2790         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2791         strbuf_release(&out);
2792 }
2793
2794 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2795 {
2796         int i, adds = 0, dels = 0, total_files = data->nr;
2797
2798         if (data->nr == 0)
2799                 return;
2800
2801         for (i = 0; i < data->nr; i++) {
2802                 int added = data->files[i]->added;
2803                 int deleted = data->files[i]->deleted;
2804
2805                 if (data->files[i]->is_unmerged ||
2806                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2807                         total_files--;
2808                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2809                         adds += added;
2810                         dels += deleted;
2811                 }
2812         }
2813         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2814 }
2815
2816 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2817 {
2818         int i;
2819
2820         if (data->nr == 0)
2821                 return;
2822
2823         for (i = 0; i < data->nr; i++) {
2824                 struct diffstat_file *file = data->files[i];
2825
2826                 fprintf(options->file, "%s", diff_line_prefix(options));
2827
2828                 if (file->is_binary)
2829                         fprintf(options->file, "-\t-\t");
2830                 else
2831                         fprintf(options->file,
2832                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2833                                 file->added, file->deleted);
2834                 if (options->line_termination) {
2835                         fill_print_name(file);
2836                         if (!file->is_renamed)
2837                                 write_name_quoted(file->name, options->file,
2838                                                   options->line_termination);
2839                         else {
2840                                 fputs(file->print_name, options->file);
2841                                 putc(options->line_termination, options->file);
2842                         }
2843                 } else {
2844                         if (file->is_renamed) {
2845                                 putc('\0', options->file);
2846                                 write_name_quoted(file->from_name, options->file, '\0');
2847                         }
2848                         write_name_quoted(file->name, options->file, '\0');
2849                 }
2850         }
2851 }
2852
2853 struct dirstat_file {
2854         const char *name;
2855         unsigned long changed;
2856 };
2857
2858 struct dirstat_dir {
2859         struct dirstat_file *files;
2860         int alloc, nr, permille, cumulative;
2861 };
2862
2863 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2864                 unsigned long changed, const char *base, int baselen)
2865 {
2866         unsigned long sum_changes = 0;
2867         unsigned int sources = 0;
2868         const char *line_prefix = diff_line_prefix(opt);
2869
2870         while (dir->nr) {
2871                 struct dirstat_file *f = dir->files;
2872                 int namelen = strlen(f->name);
2873                 unsigned long changes;
2874                 char *slash;
2875
2876                 if (namelen < baselen)
2877                         break;
2878                 if (memcmp(f->name, base, baselen))
2879                         break;
2880                 slash = strchr(f->name + baselen, '/');
2881                 if (slash) {
2882                         int newbaselen = slash + 1 - f->name;
2883                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2884                         sources++;
2885                 } else {
2886                         changes = f->changed;
2887                         dir->files++;
2888                         dir->nr--;
2889                         sources += 2;
2890                 }
2891                 sum_changes += changes;
2892         }
2893
2894         /*
2895          * We don't report dirstat's for
2896          *  - the top level
2897          *  - or cases where everything came from a single directory
2898          *    under this directory (sources == 1).
2899          */
2900         if (baselen && sources != 1) {
2901                 if (sum_changes) {
2902                         int permille = sum_changes * 1000 / changed;
2903                         if (permille >= dir->permille) {
2904                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2905                                         permille / 10, permille % 10, baselen, base);
2906                                 if (!dir->cumulative)
2907                                         return 0;
2908                         }
2909                 }
2910         }
2911         return sum_changes;
2912 }
2913
2914 static int dirstat_compare(const void *_a, const void *_b)
2915 {
2916         const struct dirstat_file *a = _a;
2917         const struct dirstat_file *b = _b;
2918         return strcmp(a->name, b->name);
2919 }
2920
2921 static void show_dirstat(struct diff_options *options)
2922 {
2923         int i;
2924         unsigned long changed;
2925         struct dirstat_dir dir;
2926         struct diff_queue_struct *q = &diff_queued_diff;
2927
2928         dir.files = NULL;
2929         dir.alloc = 0;
2930         dir.nr = 0;
2931         dir.permille = options->dirstat_permille;
2932         dir.cumulative = options->flags.dirstat_cumulative;
2933
2934         changed = 0;
2935         for (i = 0; i < q->nr; i++) {
2936                 struct diff_filepair *p = q->queue[i];
2937                 const char *name;
2938                 unsigned long copied, added, damage;
2939                 int content_changed;
2940
2941                 name = p->two->path ? p->two->path : p->one->path;
2942
2943                 if (p->one->oid_valid && p->two->oid_valid)
2944                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2945                 else
2946                         content_changed = 1;
2947
2948                 if (!content_changed) {
2949                         /*
2950                          * The SHA1 has not changed, so pre-/post-content is
2951                          * identical. We can therefore skip looking at the
2952                          * file contents altogether.
2953                          */
2954                         damage = 0;
2955                         goto found_damage;
2956                 }
2957
2958                 if (options->flags.dirstat_by_file) {
2959                         /*
2960                          * In --dirstat-by-file mode, we don't really need to
2961                          * look at the actual file contents at all.
2962                          * The fact that the SHA1 changed is enough for us to
2963                          * add this file to the list of results
2964                          * (with each file contributing equal damage).
2965                          */
2966                         damage = 1;
2967                         goto found_damage;
2968                 }
2969
2970                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2971                         diff_populate_filespec(p->one, 0);
2972                         diff_populate_filespec(p->two, 0);
2973                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2974                                                &copied, &added);
2975                         diff_free_filespec_data(p->one);
2976                         diff_free_filespec_data(p->two);
2977                 } else if (DIFF_FILE_VALID(p->one)) {
2978                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2979                         copied = added = 0;
2980                         diff_free_filespec_data(p->one);
2981                 } else if (DIFF_FILE_VALID(p->two)) {
2982                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2983                         copied = 0;
2984                         added = p->two->size;
2985                         diff_free_filespec_data(p->two);
2986                 } else
2987                         continue;
2988
2989                 /*
2990                  * Original minus copied is the removed material,
2991                  * added is the new material.  They are both damages
2992                  * made to the preimage.
2993                  * If the resulting damage is zero, we know that
2994                  * diffcore_count_changes() considers the two entries to
2995                  * be identical, but since content_changed is true, we
2996                  * know that there must have been _some_ kind of change,
2997                  * so we force all entries to have damage > 0.
2998                  */
2999                 damage = (p->one->size - copied) + added;
3000                 if (!damage)
3001                         damage = 1;
3002
3003 found_damage:
3004                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3005                 dir.files[dir.nr].name = name;
3006                 dir.files[dir.nr].changed = damage;
3007                 changed += damage;
3008                 dir.nr++;
3009         }
3010
3011         /* This can happen even with many files, if everything was renames */
3012         if (!changed)
3013                 return;
3014
3015         /* Show all directories with more than x% of the changes */
3016         QSORT(dir.files, dir.nr, dirstat_compare);
3017         gather_dirstat(options, &dir, changed, "", 0);
3018 }
3019
3020 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3021 {
3022         int i;
3023         unsigned long changed;
3024         struct dirstat_dir dir;
3025
3026         if (data->nr == 0)
3027                 return;
3028
3029         dir.files = NULL;
3030         dir.alloc = 0;
3031         dir.nr = 0;
3032         dir.permille = options->dirstat_permille;
3033         dir.cumulative = options->flags.dirstat_cumulative;
3034
3035         changed = 0;
3036         for (i = 0; i < data->nr; i++) {
3037                 struct diffstat_file *file = data->files[i];
3038                 unsigned long damage = file->added + file->deleted;
3039                 if (file->is_binary)
3040                         /*
3041                          * binary files counts bytes, not lines. Must find some
3042                          * way to normalize binary bytes vs. textual lines.
3043                          * The following heuristic assumes that there are 64
3044                          * bytes per "line".
3045                          * This is stupid and ugly, but very cheap...
3046                          */
3047                         damage = DIV_ROUND_UP(damage, 64);
3048                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3049                 dir.files[dir.nr].name = file->name;
3050                 dir.files[dir.nr].changed = damage;
3051                 changed += damage;
3052                 dir.nr++;
3053         }
3054
3055         /* This can happen even with many files, if everything was renames */
3056         if (!changed)
3057                 return;
3058
3059         /* Show all directories with more than x% of the changes */
3060         QSORT(dir.files, dir.nr, dirstat_compare);
3061         gather_dirstat(options, &dir, changed, "", 0);
3062 }
3063
3064 static void free_diffstat_info(struct diffstat_t *diffstat)
3065 {
3066         int i;
3067         for (i = 0; i < diffstat->nr; i++) {
3068                 struct diffstat_file *f = diffstat->files[i];
3069                 free(f->print_name);
3070                 free(f->name);
3071                 free(f->from_name);
3072                 free(f);
3073         }
3074         free(diffstat->files);
3075 }
3076
3077 struct checkdiff_t {
3078         const char *filename;
3079         int lineno;
3080         int conflict_marker_size;
3081         struct diff_options *o;
3082         unsigned ws_rule;
3083         unsigned status;
3084 };
3085
3086 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3087 {
3088         char firstchar;
3089         int cnt;
3090
3091         if (len < marker_size + 1)
3092                 return 0;
3093         firstchar = line[0];
3094         switch (firstchar) {
3095         case '=': case '>': case '<': case '|':
3096                 break;
3097         default:
3098                 return 0;
3099         }
3100         for (cnt = 1; cnt < marker_size; cnt++)
3101                 if (line[cnt] != firstchar)
3102                         return 0;
3103         /* line[1] thru line[marker_size-1] are same as firstchar */
3104         if (len < marker_size + 1 || !isspace(line[marker_size]))
3105                 return 0;
3106         return 1;
3107 }
3108
3109 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3110 {
3111         struct checkdiff_t *data = priv;
3112         int marker_size = data->conflict_marker_size;
3113         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3114         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3115         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3116         char *err;
3117         const char *line_prefix;
3118
3119         assert(data->o);
3120         line_prefix = diff_line_prefix(data->o);
3121
3122         if (line[0] == '+') {
3123                 unsigned bad;
3124                 data->lineno++;
3125                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3126                         data->status |= 1;
3127                         fprintf(data->o->file,
3128                                 "%s%s:%d: leftover conflict marker\n",
3129                                 line_prefix, data->filename, data->lineno);
3130                 }
3131                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3132                 if (!bad)
3133                         return;
3134                 data->status |= bad;
3135                 err = whitespace_error_string(bad);
3136                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3137                         line_prefix, data->filename, data->lineno, err);
3138                 free(err);
3139                 emit_line(data->o, set, reset, line, 1);
3140                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3141                               data->o->file, set, reset, ws);
3142         } else if (line[0] == ' ') {
3143                 data->lineno++;
3144         } else if (line[0] == '@') {
3145                 char *plus = strchr(line, '+');
3146                 if (plus)
3147                         data->lineno = strtol(plus, NULL, 10) - 1;
3148                 else
3149                         die("invalid diff");
3150         }
3151 }
3152
3153 static unsigned char *deflate_it(char *data,
3154                                  unsigned long size,
3155                                  unsigned long *result_size)
3156 {
3157         int bound;
3158         unsigned char *deflated;
3159         git_zstream stream;
3160
3161         git_deflate_init(&stream, zlib_compression_level);
3162         bound = git_deflate_bound(&stream, size);
3163         deflated = xmalloc(bound);
3164         stream.next_out = deflated;
3165         stream.avail_out = bound;
3166
3167         stream.next_in = (unsigned char *)data;
3168         stream.avail_in = size;
3169         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3170                 ; /* nothing */
3171         git_deflate_end(&stream);
3172         *result_size = stream.total_out;
3173         return deflated;
3174 }
3175
3176 static void emit_binary_diff_body(struct diff_options *o,
3177                                   mmfile_t *one, mmfile_t *two)
3178 {
3179         void *cp;
3180         void *delta;
3181         void *deflated;
3182         void *data;
3183         unsigned long orig_size;
3184         unsigned long delta_size;
3185         unsigned long deflate_size;
3186         unsigned long data_size;
3187
3188         /* We could do deflated delta, or we could do just deflated two,
3189          * whichever is smaller.
3190          */
3191         delta = NULL;
3192         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3193         if (one->size && two->size) {
3194                 delta = diff_delta(one->ptr, one->size,
3195                                    two->ptr, two->size,
3196                                    &delta_size, deflate_size);
3197                 if (delta) {
3198                         void *to_free = delta;
3199                         orig_size = delta_size;
3200                         delta = deflate_it(delta, delta_size, &delta_size);
3201                         free(to_free);
3202                 }
3203         }
3204
3205         if (delta && delta_size < deflate_size) {
3206                 char *s = xstrfmt("%lu", orig_size);
3207                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3208                                  s, strlen(s), 0);
3209                 free(s);
3210                 free(deflated);
3211                 data = delta;
3212                 data_size = delta_size;
3213         } else {
3214                 char *s = xstrfmt("%lu", two->size);
3215                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3216                                  s, strlen(s), 0);
3217                 free(s);
3218                 free(delta);
3219                 data = deflated;
3220                 data_size = deflate_size;
3221         }
3222
3223         /* emit data encoded in base85 */
3224         cp = data;
3225         while (data_size) {
3226                 int len;
3227                 int bytes = (52 < data_size) ? 52 : data_size;
3228                 char line[71];
3229                 data_size -= bytes;
3230                 if (bytes <= 26)
3231                         line[0] = bytes + 'A' - 1;
3232                 else
3233                         line[0] = bytes - 26 + 'a' - 1;
3234                 encode_85(line + 1, cp, bytes);
3235                 cp = (char *) cp + bytes;
3236
3237                 len = strlen(line);
3238                 line[len++] = '\n';
3239                 line[len] = '\0';
3240
3241                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3242                                  line, len, 0);
3243         }
3244         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3245         free(data);
3246 }
3247
3248 static void emit_binary_diff(struct diff_options *o,
3249                              mmfile_t *one, mmfile_t *two)
3250 {
3251         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3252         emit_binary_diff_body(o, one, two);
3253         emit_binary_diff_body(o, two, one);
3254 }
3255
3256 int diff_filespec_is_binary(struct diff_filespec *one)
3257 {
3258         if (one->is_binary == -1) {
3259                 diff_filespec_load_driver(one);
3260                 if (one->driver->binary != -1)
3261                         one->is_binary = one->driver->binary;
3262                 else {
3263                         if (!one->data && DIFF_FILE_VALID(one))
3264                                 diff_populate_filespec(one, CHECK_BINARY);
3265                         if (one->is_binary == -1 && one->data)
3266                                 one->is_binary = buffer_is_binary(one->data,
3267                                                 one->size);
3268                         if (one->is_binary == -1)
3269                                 one->is_binary = 0;
3270                 }
3271         }
3272         return one->is_binary;
3273 }
3274
3275 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3276 {
3277         diff_filespec_load_driver(one);
3278         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3279 }
3280
3281 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3282 {
3283         if (!options->a_prefix)
3284                 options->a_prefix = a;
3285         if (!options->b_prefix)
3286                 options->b_prefix = b;
3287 }
3288
3289 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3290 {
3291         if (!DIFF_FILE_VALID(one))
3292                 return NULL;
3293
3294         diff_filespec_load_driver(one);
3295         return userdiff_get_textconv(one->driver);
3296 }
3297
3298 static void builtin_diff(const char *name_a,
3299                          const char *name_b,
3300                          struct diff_filespec *one,
3301                          struct diff_filespec *two,
3302                          const char *xfrm_msg,
3303                          int must_show_header,
3304                          struct diff_options *o,
3305                          int complete_rewrite)
3306 {
3307         mmfile_t mf1, mf2;
3308         const char *lbl[2];
3309         char *a_one, *b_two;
3310         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3311         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3312         const char *a_prefix, *b_prefix;
3313         struct userdiff_driver *textconv_one = NULL;
3314         struct userdiff_driver *textconv_two = NULL;
3315         struct strbuf header = STRBUF_INIT;
3316         const char *line_prefix = diff_line_prefix(o);
3317
3318         diff_set_mnemonic_prefix(o, "a/", "b/");
3319         if (o->flags.reverse_diff) {
3320                 a_prefix = o->b_prefix;
3321                 b_prefix = o->a_prefix;
3322         } else {
3323                 a_prefix = o->a_prefix;
3324                 b_prefix = o->b_prefix;
3325         }
3326
3327         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3328             (!one->mode || S_ISGITLINK(one->mode)) &&
3329             (!two->mode || S_ISGITLINK(two->mode))) {
3330                 show_submodule_summary(o, one->path ? one->path : two->path,
3331                                 &one->oid, &two->oid,
3332                                 two->dirty_submodule);
3333                 return;
3334         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3335                    (!one->mode || S_ISGITLINK(one->mode)) &&
3336                    (!two->mode || S_ISGITLINK(two->mode))) {
3337                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3338                                 &one->oid, &two->oid,
3339                                 two->dirty_submodule);
3340                 return;
3341         }
3342
3343         if (o->flags.allow_textconv) {
3344                 textconv_one = get_textconv(one);
3345                 textconv_two = get_textconv(two);
3346         }
3347
3348         /* Never use a non-valid filename anywhere if at all possible */
3349         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3350         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3351
3352         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3353         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3354         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3355         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3356         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3357         if (lbl[0][0] == '/') {
3358                 /* /dev/null */
3359                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3360                 if (xfrm_msg)
3361                         strbuf_addstr(&header, xfrm_msg);
3362                 must_show_header = 1;
3363         }
3364         else if (lbl[1][0] == '/') {
3365                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3366                 if (xfrm_msg)
3367                         strbuf_addstr(&header, xfrm_msg);
3368                 must_show_header = 1;
3369         }
3370         else {
3371                 if (one->mode != two->mode) {
3372                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3373                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3374                         must_show_header = 1;
3375                 }
3376                 if (xfrm_msg)
3377                         strbuf_addstr(&header, xfrm_msg);
3378
3379                 /*
3380                  * we do not run diff between different kind
3381                  * of objects.
3382                  */
3383                 if ((one->mode ^ two->mode) & S_IFMT)
3384                         goto free_ab_and_return;
3385                 if (complete_rewrite &&
3386                     (textconv_one || !diff_filespec_is_binary(one)) &&
3387                     (textconv_two || !diff_filespec_is_binary(two))) {
3388                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3389                                          header.buf, header.len, 0);
3390                         strbuf_reset(&header);
3391                         emit_rewrite_diff(name_a, name_b, one, two,
3392                                                 textconv_one, textconv_two, o);
3393                         o->found_changes = 1;
3394                         goto free_ab_and_return;
3395                 }
3396         }
3397
3398         if (o->irreversible_delete && lbl[1][0] == '/') {
3399                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3400                                  header.len, 0);
3401                 strbuf_reset(&header);
3402                 goto free_ab_and_return;
3403         } else if (!o->flags.text &&
3404             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3405               (!textconv_two && diff_filespec_is_binary(two)) )) {
3406                 struct strbuf sb = STRBUF_INIT;
3407                 if (!one->data && !two->data &&
3408                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3409                     !o->flags.binary) {
3410                         if (!oidcmp(&one->oid, &two->oid)) {
3411                                 if (must_show_header)
3412                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3413                                                          header.buf, header.len,
3414                                                          0);
3415                                 goto free_ab_and_return;
3416                         }
3417                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3418                                          header.buf, header.len, 0);
3419                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3420                                     diff_line_prefix(o), lbl[0], lbl[1]);
3421                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3422                                          sb.buf, sb.len, 0);
3423                         strbuf_release(&sb);
3424                         goto free_ab_and_return;
3425                 }
3426                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3427                         die("unable to read files to diff");
3428                 /* Quite common confusing case */
3429                 if (mf1.size == mf2.size &&
3430                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3431                         if (must_show_header)
3432                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3433                                                  header.buf, header.len, 0);
3434                         goto free_ab_and_return;
3435                 }
3436                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3437                 strbuf_reset(&header);
3438                 if (o->flags.binary)
3439                         emit_binary_diff(o, &mf1, &mf2);
3440                 else {
3441                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3442                                     diff_line_prefix(o), lbl[0], lbl[1]);
3443                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3444                                          sb.buf, sb.len, 0);
3445                         strbuf_release(&sb);
3446                 }
3447                 o->found_changes = 1;
3448         } else {
3449                 /* Crazy xdl interfaces.. */
3450                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3451                 const char *v;
3452                 xpparam_t xpp;
3453                 xdemitconf_t xecfg;
3454                 struct emit_callback ecbdata;
3455                 const struct userdiff_funcname *pe;
3456
3457                 if (must_show_header) {
3458                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3459                                          header.buf, header.len, 0);
3460                         strbuf_reset(&header);
3461                 }
3462
3463                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3464                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3465
3466                 pe = diff_funcname_pattern(one);
3467                 if (!pe)
3468                         pe = diff_funcname_pattern(two);
3469
3470                 memset(&xpp, 0, sizeof(xpp));
3471                 memset(&xecfg, 0, sizeof(xecfg));
3472                 memset(&ecbdata, 0, sizeof(ecbdata));
3473                 if (o->flags.suppress_diff_headers)
3474                         lbl[0] = NULL;
3475                 ecbdata.label_path = lbl;
3476                 ecbdata.color_diff = want_color(o->use_color);
3477                 ecbdata.ws_rule = whitespace_rule(name_b);
3478                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3479                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3480                 ecbdata.opt = o;
3481                 if (header.len && !o->flags.suppress_diff_headers)
3482                         ecbdata.header = &header;
3483                 xpp.flags = o->xdl_opts;
3484                 xpp.anchors = o->anchors;
3485                 xpp.anchors_nr = o->anchors_nr;
3486                 xecfg.ctxlen = o->context;
3487                 xecfg.interhunkctxlen = o->interhunkcontext;
3488                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3489                 if (o->flags.funccontext)
3490                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3491                 if (pe)
3492                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3493                 if (!diffopts)
3494                         ;
3495                 else if (skip_prefix(diffopts, "--unified=", &v))
3496                         xecfg.ctxlen = strtoul(v, NULL, 10);
3497                 else if (skip_prefix(diffopts, "-u", &v))
3498                         xecfg.ctxlen = strtoul(v, NULL, 10);
3499                 if (o->word_diff)
3500                         init_diff_words_data(&ecbdata, o, one, two);
3501                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3502                                   &xpp, &xecfg))
3503                         die("unable to generate diff for %s", one->path);
3504                 if (o->word_diff)
3505                         free_diff_words_data(&ecbdata);
3506                 if (textconv_one)
3507                         free(mf1.ptr);
3508                 if (textconv_two)
3509                         free(mf2.ptr);
3510                 xdiff_clear_find_func(&xecfg);
3511         }
3512
3513  free_ab_and_return:
3514         strbuf_release(&header);
3515         diff_free_filespec_data(one);
3516         diff_free_filespec_data(two);
3517         free(a_one);
3518         free(b_two);
3519         return;
3520 }
3521
3522 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3523 {
3524         if (!is_renamed) {
3525                 if (p->status == DIFF_STATUS_ADDED) {
3526                         if (S_ISLNK(p->two->mode))
3527                                 return "new +l";
3528                         else if ((p->two->mode & 0777) == 0755)
3529                                 return "new +x";
3530                         else
3531                                 return "new";
3532                 } else if (p->status == DIFF_STATUS_DELETED)
3533                         return "gone";
3534         }
3535         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3536                 return "mode -l";
3537         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3538                 return "mode +l";
3539         else if ((p->one->mode & 0777) == 0644 &&
3540                  (p->two->mode & 0777) == 0755)
3541                 return "mode +x";
3542         else if ((p->one->mode & 0777) == 0755 &&
3543                  (p->two->mode & 0777) == 0644)
3544                 return "mode -x";
3545         return NULL;
3546 }
3547
3548 static void builtin_diffstat(const char *name_a, const char *name_b,
3549                              struct diff_filespec *one,
3550                              struct diff_filespec *two,
3551                              struct diffstat_t *diffstat,
3552                              struct diff_options *o,
3553                              struct diff_filepair *p)
3554 {
3555         mmfile_t mf1, mf2;
3556         struct diffstat_file *data;
3557         int same_contents;
3558         int complete_rewrite = 0;
3559
3560         if (!DIFF_PAIR_UNMERGED(p)) {
3561                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3562                         complete_rewrite = 1;
3563         }
3564
3565         data = diffstat_add(diffstat, name_a, name_b);
3566         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3567         if (o->flags.stat_with_summary)
3568                 data->comments = get_compact_summary(p, data->is_renamed);
3569
3570         if (!one || !two) {
3571                 data->is_unmerged = 1;
3572                 return;
3573         }
3574
3575         same_contents = !oidcmp(&one->oid, &two->oid);
3576
3577         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3578                 data->is_binary = 1;
3579                 if (same_contents) {
3580                         data->added = 0;
3581                         data->deleted = 0;
3582                 } else {
3583                         data->added = diff_filespec_size(two);
3584                         data->deleted = diff_filespec_size(one);
3585                 }
3586         }
3587
3588         else if (complete_rewrite) {
3589                 diff_populate_filespec(one, 0);
3590                 diff_populate_filespec(two, 0);
3591                 data->deleted = count_lines(one->data, one->size);
3592                 data->added = count_lines(two->data, two->size);
3593         }
3594
3595         else if (!same_contents) {
3596                 /* Crazy xdl interfaces.. */
3597                 xpparam_t xpp;
3598                 xdemitconf_t xecfg;
3599
3600                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3601                         die("unable to read files to diff");
3602
3603                 memset(&xpp, 0, sizeof(xpp));
3604                 memset(&xecfg, 0, sizeof(xecfg));
3605                 xpp.flags = o->xdl_opts;
3606                 xpp.anchors = o->anchors;
3607                 xpp.anchors_nr = o->anchors_nr;
3608                 xecfg.ctxlen = o->context;
3609                 xecfg.interhunkctxlen = o->interhunkcontext;
3610                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3611                                   &xpp, &xecfg))
3612                         die("unable to generate diffstat for %s", one->path);
3613         }
3614
3615         diff_free_filespec_data(one);
3616         diff_free_filespec_data(two);
3617 }
3618
3619 static void builtin_checkdiff(const char *name_a, const char *name_b,
3620                               const char *attr_path,
3621                               struct diff_filespec *one,
3622                               struct diff_filespec *two,
3623                               struct diff_options *o)
3624 {
3625         mmfile_t mf1, mf2;
3626         struct checkdiff_t data;
3627
3628         if (!two)
3629                 return;
3630
3631         memset(&data, 0, sizeof(data));
3632         data.filename = name_b ? name_b : name_a;
3633         data.lineno = 0;
3634         data.o = o;
3635         data.ws_rule = whitespace_rule(attr_path);
3636         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3637
3638         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3639                 die("unable to read files to diff");
3640
3641         /*
3642          * All the other codepaths check both sides, but not checking
3643          * the "old" side here is deliberate.  We are checking the newly
3644          * introduced changes, and as long as the "new" side is text, we
3645          * can and should check what it introduces.
3646          */
3647         if (diff_filespec_is_binary(two))
3648                 goto free_and_return;
3649         else {
3650                 /* Crazy xdl interfaces.. */
3651                 xpparam_t xpp;
3652                 xdemitconf_t xecfg;
3653
3654                 memset(&xpp, 0, sizeof(xpp));
3655                 memset(&xecfg, 0, sizeof(xecfg));
3656                 xecfg.ctxlen = 1; /* at least one context line */
3657                 xpp.flags = 0;
3658                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3659                                   &xpp, &xecfg))
3660                         die("unable to generate checkdiff for %s", one->path);
3661
3662                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3663                         struct emit_callback ecbdata;
3664                         int blank_at_eof;
3665
3666                         ecbdata.ws_rule = data.ws_rule;
3667                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3668                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3669
3670                         if (blank_at_eof) {
3671                                 static char *err;
3672                                 if (!err)
3673                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3674                                 fprintf(o->file, "%s:%d: %s.\n",
3675                                         data.filename, blank_at_eof, err);
3676                                 data.status = 1; /* report errors */
3677                         }
3678                 }
3679         }
3680  free_and_return:
3681         diff_free_filespec_data(one);
3682         diff_free_filespec_data(two);
3683         if (data.status)
3684                 o->flags.check_failed = 1;
3685 }
3686
3687 struct diff_filespec *alloc_filespec(const char *path)
3688 {
3689         struct diff_filespec *spec;
3690
3691         FLEXPTR_ALLOC_STR(spec, path, path);
3692         spec->count = 1;
3693         spec->is_binary = -1;
3694         return spec;
3695 }
3696
3697 void free_filespec(struct diff_filespec *spec)
3698 {
3699         if (!--spec->count) {
3700                 diff_free_filespec_data(spec);
3701                 free(spec);
3702         }
3703 }
3704
3705 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3706                    int oid_valid, unsigned short mode)
3707 {
3708         if (mode) {
3709                 spec->mode = canon_mode(mode);
3710                 oidcpy(&spec->oid, oid);
3711                 spec->oid_valid = oid_valid;
3712         }
3713 }
3714
3715 /*
3716  * Given a name and sha1 pair, if the index tells us the file in
3717  * the work tree has that object contents, return true, so that
3718  * prepare_temp_file() does not have to inflate and extract.
3719  */
3720 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3721 {
3722         const struct cache_entry *ce;
3723         struct stat st;
3724         int pos, len;
3725
3726         /*
3727          * We do not read the cache ourselves here, because the
3728          * benchmark with my previous version that always reads cache
3729          * shows that it makes things worse for diff-tree comparing
3730          * two linux-2.6 kernel trees in an already checked out work
3731          * tree.  This is because most diff-tree comparisons deal with
3732          * only a small number of files, while reading the cache is
3733          * expensive for a large project, and its cost outweighs the
3734          * savings we get by not inflating the object to a temporary
3735          * file.  Practically, this code only helps when we are used
3736          * by diff-cache --cached, which does read the cache before
3737          * calling us.
3738          */
3739         if (!active_cache)
3740                 return 0;
3741
3742         /* We want to avoid the working directory if our caller
3743          * doesn't need the data in a normal file, this system
3744          * is rather slow with its stat/open/mmap/close syscalls,
3745          * and the object is contained in a pack file.  The pack
3746          * is probably already open and will be faster to obtain
3747          * the data through than the working directory.  Loose
3748          * objects however would tend to be slower as they need
3749          * to be individually opened and inflated.
3750          */
3751         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3752                 return 0;
3753
3754         /*
3755          * Similarly, if we'd have to convert the file contents anyway, that
3756          * makes the optimization not worthwhile.
3757          */
3758         if (!want_file && would_convert_to_git(&the_index, name))
3759                 return 0;
3760
3761         len = strlen(name);
3762         pos = cache_name_pos(name, len);
3763         if (pos < 0)
3764                 return 0;
3765         ce = active_cache[pos];
3766
3767         /*
3768          * This is not the sha1 we are looking for, or
3769          * unreusable because it is not a regular file.
3770          */
3771         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3772                 return 0;
3773
3774         /*
3775          * If ce is marked as "assume unchanged", there is no
3776          * guarantee that work tree matches what we are looking for.
3777          */
3778         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3779                 return 0;
3780
3781         /*
3782          * If ce matches the file in the work tree, we can reuse it.
3783          */
3784         if (ce_uptodate(ce) ||
3785             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3786                 return 1;
3787
3788         return 0;
3789 }
3790
3791 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3792 {
3793         struct strbuf buf = STRBUF_INIT;
3794         char *dirty = "";
3795
3796         /* Are we looking at the work tree? */
3797         if (s->dirty_submodule)
3798                 dirty = "-dirty";
3799
3800         strbuf_addf(&buf, "Subproject commit %s%s\n",
3801                     oid_to_hex(&s->oid), dirty);
3802         s->size = buf.len;
3803         if (size_only) {
3804                 s->data = NULL;
3805                 strbuf_release(&buf);
3806         } else {
3807                 s->data = strbuf_detach(&buf, NULL);
3808                 s->should_free = 1;
3809         }
3810         return 0;
3811 }
3812
3813 /*
3814  * While doing rename detection and pickaxe operation, we may need to
3815  * grab the data for the blob (or file) for our own in-core comparison.
3816  * diff_filespec has data and size fields for this purpose.
3817  */
3818 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3819 {
3820         int size_only = flags & CHECK_SIZE_ONLY;
3821         int err = 0;
3822         int conv_flags = global_conv_flags_eol;
3823         /*
3824          * demote FAIL to WARN to allow inspecting the situation
3825          * instead of refusing.
3826          */
3827         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3828                 conv_flags = CONV_EOL_RNDTRP_WARN;
3829
3830         if (!DIFF_FILE_VALID(s))
3831                 die("internal error: asking to populate invalid file.");
3832         if (S_ISDIR(s->mode))
3833                 return -1;
3834
3835         if (s->data)
3836                 return 0;
3837
3838         if (size_only && 0 < s->size)
3839                 return 0;
3840
3841         if (S_ISGITLINK(s->mode))
3842                 return diff_populate_gitlink(s, size_only);
3843
3844         if (!s->oid_valid ||
3845             reuse_worktree_file(s->path, &s->oid, 0)) {
3846                 struct strbuf buf = STRBUF_INIT;
3847                 struct stat st;
3848                 int fd;
3849
3850                 if (lstat(s->path, &st) < 0) {
3851                 err_empty:
3852                         err = -1;
3853                 empty:
3854                         s->data = (char *)"";
3855                         s->size = 0;
3856                         return err;
3857                 }
3858                 s->size = xsize_t(st.st_size);
3859                 if (!s->size)
3860                         goto empty;
3861                 if (S_ISLNK(st.st_mode)) {
3862                         struct strbuf sb = STRBUF_INIT;
3863
3864                         if (strbuf_readlink(&sb, s->path, s->size))
3865                                 goto err_empty;
3866                         s->size = sb.len;
3867                         s->data = strbuf_detach(&sb, NULL);
3868                         s->should_free = 1;
3869                         return 0;
3870                 }
3871
3872                 /*
3873                  * Even if the caller would be happy with getting
3874                  * only the size, we cannot return early at this
3875                  * point if the path requires us to run the content
3876                  * conversion.
3877                  */
3878                 if (size_only && !would_convert_to_git(&the_index, s->path))
3879                         return 0;
3880
3881                 /*
3882                  * Note: this check uses xsize_t(st.st_size) that may
3883                  * not be the true size of the blob after it goes
3884                  * through convert_to_git().  This may not strictly be
3885                  * correct, but the whole point of big_file_threshold
3886                  * and is_binary check being that we want to avoid
3887                  * opening the file and inspecting the contents, this
3888                  * is probably fine.
3889                  */
3890                 if ((flags & CHECK_BINARY) &&
3891                     s->size > big_file_threshold && s->is_binary == -1) {
3892                         s->is_binary = 1;
3893                         return 0;
3894                 }
3895                 fd = open(s->path, O_RDONLY);
3896                 if (fd < 0)
3897                         goto err_empty;
3898                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3899                 close(fd);
3900                 s->should_munmap = 1;
3901
3902                 /*
3903                  * Convert from working tree format to canonical git format
3904                  */
3905                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3906                         size_t size = 0;
3907                         munmap(s->data, s->size);
3908                         s->should_munmap = 0;
3909                         s->data = strbuf_detach(&buf, &size);
3910                         s->size = size;
3911                         s->should_free = 1;
3912                 }
3913         }
3914         else {
3915                 enum object_type type;
3916                 if (size_only || (flags & CHECK_BINARY)) {
3917                         type = oid_object_info(the_repository, &s->oid,
3918                                                &s->size);
3919                         if (type < 0)
3920                                 die("unable to read %s",
3921                                     oid_to_hex(&s->oid));
3922                         if (size_only)
3923                                 return 0;
3924                         if (s->size > big_file_threshold && s->is_binary == -1) {
3925                                 s->is_binary = 1;
3926                                 return 0;
3927                         }
3928                 }
3929                 s->data = read_object_file(&s->oid, &type, &s->size);
3930                 if (!s->data)
3931                         die("unable to read %s", oid_to_hex(&s->oid));
3932                 s->should_free = 1;
3933         }
3934         return 0;
3935 }
3936
3937 void diff_free_filespec_blob(struct diff_filespec *s)
3938 {
3939         if (s->should_free)
3940                 free(s->data);
3941         else if (s->should_munmap)
3942                 munmap(s->data, s->size);
3943
3944         if (s->should_free || s->should_munmap) {
3945                 s->should_free = s->should_munmap = 0;
3946                 s->data = NULL;
3947         }
3948 }
3949
3950 void diff_free_filespec_data(struct diff_filespec *s)
3951 {
3952         diff_free_filespec_blob(s);
3953         FREE_AND_NULL(s->cnt_data);
3954 }
3955
3956 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3957                            void *blob,
3958                            unsigned long size,
3959                            const struct object_id *oid,
3960                            int mode)
3961 {
3962         struct strbuf buf = STRBUF_INIT;
3963         struct strbuf tempfile = STRBUF_INIT;
3964         char *path_dup = xstrdup(path);
3965         const char *base = basename(path_dup);
3966
3967         /* Generate "XXXXXX_basename.ext" */
3968         strbuf_addstr(&tempfile, "XXXXXX_");
3969         strbuf_addstr(&tempfile, base);
3970
3971         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3972         if (!temp->tempfile)
3973                 die_errno("unable to create temp-file");
3974         if (convert_to_working_tree(path,
3975                         (const char *)blob, (size_t)size, &buf)) {
3976                 blob = buf.buf;
3977                 size = buf.len;
3978         }
3979         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3980             close_tempfile_gently(temp->tempfile))
3981                 die_errno("unable to write temp-file");
3982         temp->name = get_tempfile_path(temp->tempfile);
3983         oid_to_hex_r(temp->hex, oid);
3984         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3985         strbuf_release(&buf);
3986         strbuf_release(&tempfile);
3987         free(path_dup);
3988 }
3989
3990 static struct diff_tempfile *prepare_temp_file(const char *name,
3991                 struct diff_filespec *one)
3992 {
3993         struct diff_tempfile *temp = claim_diff_tempfile();
3994
3995         if (!DIFF_FILE_VALID(one)) {
3996         not_a_valid_file:
3997                 /* A '-' entry produces this for file-2, and
3998                  * a '+' entry produces this for file-1.
3999                  */
4000                 temp->name = "/dev/null";
4001                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4002                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4003                 return temp;
4004         }
4005
4006         if (!S_ISGITLINK(one->mode) &&
4007             (!one->oid_valid ||
4008              reuse_worktree_file(name, &one->oid, 1))) {
4009                 struct stat st;
4010                 if (lstat(name, &st) < 0) {
4011                         if (errno == ENOENT)
4012                                 goto not_a_valid_file;
4013                         die_errno("stat(%s)", name);
4014                 }
4015                 if (S_ISLNK(st.st_mode)) {
4016                         struct strbuf sb = STRBUF_INIT;
4017                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4018                                 die_errno("readlink(%s)", name);
4019                         prep_temp_blob(name, temp, sb.buf, sb.len,
4020                                        (one->oid_valid ?
4021                                         &one->oid : &null_oid),
4022                                        (one->oid_valid ?
4023                                         one->mode : S_IFLNK));
4024                         strbuf_release(&sb);
4025                 }
4026                 else {
4027                         /* we can borrow from the file in the work tree */
4028                         temp->name = name;
4029                         if (!one->oid_valid)
4030                                 oid_to_hex_r(temp->hex, &null_oid);
4031                         else
4032                                 oid_to_hex_r(temp->hex, &one->oid);
4033                         /* Even though we may sometimes borrow the
4034                          * contents from the work tree, we always want
4035                          * one->mode.  mode is trustworthy even when
4036                          * !(one->oid_valid), as long as
4037                          * DIFF_FILE_VALID(one).
4038                          */
4039                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4040                 }
4041                 return temp;
4042         }
4043         else {
4044                 if (diff_populate_filespec(one, 0))
4045                         die("cannot read data blob for %s", one->path);
4046                 prep_temp_blob(name, temp, one->data, one->size,
4047                                &one->oid, one->mode);
4048         }
4049         return temp;
4050 }
4051
4052 static void add_external_diff_name(struct argv_array *argv,
4053                                    const char *name,
4054                                    struct diff_filespec *df)
4055 {
4056         struct diff_tempfile *temp = prepare_temp_file(name, df);
4057         argv_array_push(argv, temp->name);
4058         argv_array_push(argv, temp->hex);
4059         argv_array_push(argv, temp->mode);
4060 }
4061
4062 /* An external diff command takes:
4063  *
4064  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4065  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4066  *
4067  */
4068 static void run_external_diff(const char *pgm,
4069                               const char *name,
4070                               const char *other,
4071                               struct diff_filespec *one,
4072                               struct diff_filespec *two,
4073                               const char *xfrm_msg,
4074                               int complete_rewrite,
4075                               struct diff_options *o)
4076 {
4077         struct argv_array argv = ARGV_ARRAY_INIT;
4078         struct argv_array env = ARGV_ARRAY_INIT;
4079         struct diff_queue_struct *q = &diff_queued_diff;
4080
4081         argv_array_push(&argv, pgm);
4082         argv_array_push(&argv, name);
4083
4084         if (one && two) {
4085                 add_external_diff_name(&argv, name, one);
4086                 if (!other)
4087                         add_external_diff_name(&argv, name, two);
4088                 else {
4089                         add_external_diff_name(&argv, other, two);
4090                         argv_array_push(&argv, other);
4091                         argv_array_push(&argv, xfrm_msg);
4092                 }
4093         }
4094
4095         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4096         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4097
4098         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4099                 die(_("external diff died, stopping at %s"), name);
4100
4101         remove_tempfile();
4102         argv_array_clear(&argv);
4103         argv_array_clear(&env);
4104 }
4105
4106 static int similarity_index(struct diff_filepair *p)
4107 {
4108         return p->score * 100 / MAX_SCORE;
4109 }
4110
4111 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4112 {
4113         if (startup_info->have_repository)
4114                 return find_unique_abbrev(oid, abbrev);
4115         else {
4116                 char *hex = oid_to_hex(oid);
4117                 if (abbrev < 0)
4118                         abbrev = FALLBACK_DEFAULT_ABBREV;
4119                 if (abbrev > the_hash_algo->hexsz)
4120                         BUG("oid abbreviation out of range: %d", abbrev);
4121                 if (abbrev)
4122                         hex[abbrev] = '\0';
4123                 return hex;
4124         }
4125 }
4126
4127 static void fill_metainfo(struct strbuf *msg,
4128                           const char *name,
4129                           const char *other,
4130                           struct diff_filespec *one,
4131                           struct diff_filespec *two,
4132                           struct diff_options *o,
4133                           struct diff_filepair *p,
4134                           int *must_show_header,
4135                           int use_color)
4136 {
4137         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4138         const char *reset = diff_get_color(use_color, DIFF_RESET);
4139         const char *line_prefix = diff_line_prefix(o);
4140
4141         *must_show_header = 1;
4142         strbuf_init(msg, PATH_MAX * 2 + 300);
4143         switch (p->status) {
4144         case DIFF_STATUS_COPIED:
4145                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4146                             line_prefix, set, similarity_index(p));
4147                 strbuf_addf(msg, "%s\n%s%scopy from ",
4148                             reset,  line_prefix, set);
4149                 quote_c_style(name, msg, NULL, 0);
4150                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4151                 quote_c_style(other, msg, NULL, 0);
4152                 strbuf_addf(msg, "%s\n", reset);
4153                 break;
4154         case DIFF_STATUS_RENAMED:
4155                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4156                             line_prefix, set, similarity_index(p));
4157                 strbuf_addf(msg, "%s\n%s%srename from ",
4158                             reset, line_prefix, set);
4159                 quote_c_style(name, msg, NULL, 0);
4160                 strbuf_addf(msg, "%s\n%s%srename to ",
4161                             reset, line_prefix, set);
4162                 quote_c_style(other, msg, NULL, 0);
4163                 strbuf_addf(msg, "%s\n", reset);
4164                 break;
4165         case DIFF_STATUS_MODIFIED:
4166                 if (p->score) {
4167                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4168                                     line_prefix,
4169                                     set, similarity_index(p), reset);
4170                         break;
4171                 }
4172                 /* fallthru */
4173         default:
4174                 *must_show_header = 0;
4175         }
4176         if (one && two && oidcmp(&one->oid, &two->oid)) {
4177                 const unsigned hexsz = the_hash_algo->hexsz;
4178                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4179
4180                 if (o->flags.binary) {
4181                         mmfile_t mf;
4182                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
4183                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
4184                                 abbrev = hexsz;
4185                 }
4186                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4187                             diff_abbrev_oid(&one->oid, abbrev),
4188                             diff_abbrev_oid(&two->oid, abbrev));
4189                 if (one->mode == two->mode)
4190                         strbuf_addf(msg, " %06o", one->mode);
4191                 strbuf_addf(msg, "%s\n", reset);
4192         }
4193 }
4194
4195 static void run_diff_cmd(const char *pgm,
4196                          const char *name,
4197                          const char *other,
4198                          const char *attr_path,
4199                          struct diff_filespec *one,
4200                          struct diff_filespec *two,
4201                          struct strbuf *msg,
4202                          struct diff_options *o,
4203                          struct diff_filepair *p)
4204 {
4205         const char *xfrm_msg = NULL;
4206         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4207         int must_show_header = 0;
4208
4209
4210         if (o->flags.allow_external) {
4211                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
4212                 if (drv && drv->external)
4213                         pgm = drv->external;
4214         }
4215
4216         if (msg) {
4217                 /*
4218                  * don't use colors when the header is intended for an
4219                  * external diff driver
4220                  */
4221                 fill_metainfo(msg, name, other, one, two, o, p,
4222                               &must_show_header,
4223                               want_color(o->use_color) && !pgm);
4224                 xfrm_msg = msg->len ? msg->buf : NULL;
4225         }
4226
4227         if (pgm) {
4228                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4229                                   complete_rewrite, o);
4230                 return;
4231         }
4232         if (one && two)
4233                 builtin_diff(name, other ? other : name,
4234                              one, two, xfrm_msg, must_show_header,
4235                              o, complete_rewrite);
4236         else
4237                 fprintf(o->file, "* Unmerged path %s\n", name);
4238 }
4239
4240 static void diff_fill_oid_info(struct diff_filespec *one)
4241 {
4242         if (DIFF_FILE_VALID(one)) {
4243                 if (!one->oid_valid) {
4244                         struct stat st;
4245                         if (one->is_stdin) {
4246                                 oidclr(&one->oid);
4247                                 return;
4248                         }
4249                         if (lstat(one->path, &st) < 0)
4250                                 die_errno("stat '%s'", one->path);
4251                         if (index_path(&one->oid, one->path, &st, 0))
4252                                 die("cannot hash %s", one->path);
4253                 }
4254         }
4255         else
4256                 oidclr(&one->oid);
4257 }
4258
4259 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4260 {
4261         /* Strip the prefix but do not molest /dev/null and absolute paths */
4262         if (*namep && **namep != '/') {
4263                 *namep += prefix_length;
4264                 if (**namep == '/')
4265                         ++*namep;
4266         }
4267         if (*otherp && **otherp != '/') {
4268                 *otherp += prefix_length;
4269                 if (**otherp == '/')
4270                         ++*otherp;
4271         }
4272 }
4273
4274 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4275 {
4276         const char *pgm = external_diff();
4277         struct strbuf msg;
4278         struct diff_filespec *one = p->one;
4279         struct diff_filespec *two = p->two;
4280         const char *name;
4281         const char *other;
4282         const char *attr_path;
4283
4284         name  = one->path;
4285         other = (strcmp(name, two->path) ? two->path : NULL);
4286         attr_path = name;
4287         if (o->prefix_length)
4288                 strip_prefix(o->prefix_length, &name, &other);
4289
4290         if (!o->flags.allow_external)
4291                 pgm = NULL;
4292
4293         if (DIFF_PAIR_UNMERGED(p)) {
4294                 run_diff_cmd(pgm, name, NULL, attr_path,
4295                              NULL, NULL, NULL, o, p);
4296                 return;
4297         }
4298
4299         diff_fill_oid_info(one);
4300         diff_fill_oid_info(two);
4301
4302         if (!pgm &&
4303             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4304             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4305                 /*
4306                  * a filepair that changes between file and symlink
4307                  * needs to be split into deletion and creation.
4308                  */
4309                 struct diff_filespec *null = alloc_filespec(two->path);
4310                 run_diff_cmd(NULL, name, other, attr_path,
4311                              one, null, &msg, o, p);
4312                 free(null);
4313                 strbuf_release(&msg);
4314
4315                 null = alloc_filespec(one->path);
4316                 run_diff_cmd(NULL, name, other, attr_path,
4317                              null, two, &msg, o, p);
4318                 free(null);
4319         }
4320         else
4321                 run_diff_cmd(pgm, name, other, attr_path,
4322                              one, two, &msg, o, p);
4323
4324         strbuf_release(&msg);
4325 }
4326
4327 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4328                          struct diffstat_t *diffstat)
4329 {
4330         const char *name;
4331         const char *other;
4332
4333         if (DIFF_PAIR_UNMERGED(p)) {
4334                 /* unmerged */
4335                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4336                 return;
4337         }
4338
4339         name = p->one->path;
4340         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4341
4342         if (o->prefix_length)
4343                 strip_prefix(o->prefix_length, &name, &other);
4344
4345         diff_fill_oid_info(p->one);
4346         diff_fill_oid_info(p->two);
4347
4348         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4349 }
4350
4351 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4352 {
4353         const char *name;
4354         const char *other;
4355         const char *attr_path;
4356
4357         if (DIFF_PAIR_UNMERGED(p)) {
4358                 /* unmerged */
4359                 return;
4360         }
4361
4362         name = p->one->path;
4363         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4364         attr_path = other ? other : name;
4365
4366         if (o->prefix_length)
4367                 strip_prefix(o->prefix_length, &name, &other);
4368
4369         diff_fill_oid_info(p->one);
4370         diff_fill_oid_info(p->two);
4371
4372         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4373 }
4374
4375 void diff_setup(struct diff_options *options)
4376 {
4377         memcpy(options, &default_diff_options, sizeof(*options));
4378
4379         options->file = stdout;
4380
4381         options->abbrev = DEFAULT_ABBREV;
4382         options->line_termination = '\n';
4383         options->break_opt = -1;
4384         options->rename_limit = -1;
4385         options->dirstat_permille = diff_dirstat_permille_default;
4386         options->context = diff_context_default;
4387         options->interhunkcontext = diff_interhunk_context_default;
4388         options->ws_error_highlight = ws_error_highlight_default;
4389         options->flags.rename_empty = 1;
4390         options->objfind = NULL;
4391
4392         /* pathchange left =NULL by default */
4393         options->change = diff_change;
4394         options->add_remove = diff_addremove;
4395         options->use_color = diff_use_color_default;
4396         options->detect_rename = diff_detect_rename_default;
4397         options->xdl_opts |= diff_algorithm;
4398         if (diff_indent_heuristic)
4399                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4400
4401         options->orderfile = diff_order_file_cfg;
4402
4403         if (diff_no_prefix) {
4404                 options->a_prefix = options->b_prefix = "";
4405         } else if (!diff_mnemonic_prefix) {
4406                 options->a_prefix = "a/";
4407                 options->b_prefix = "b/";
4408         }
4409
4410         options->color_moved = diff_color_moved_default;
4411         options->color_moved_ws_handling = diff_color_moved_ws_default;
4412 }
4413
4414 void diff_setup_done(struct diff_options *options)
4415 {
4416         unsigned check_mask = DIFF_FORMAT_NAME |
4417                               DIFF_FORMAT_NAME_STATUS |
4418                               DIFF_FORMAT_CHECKDIFF |
4419                               DIFF_FORMAT_NO_OUTPUT;
4420         /*
4421          * This must be signed because we're comparing against a potentially
4422          * negative value.
4423          */
4424         const int hexsz = the_hash_algo->hexsz;
4425
4426         if (options->set_default)
4427                 options->set_default(options);
4428
4429         if (HAS_MULTI_BITS(options->output_format & check_mask))
4430                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4431
4432         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4433                 die(_("-G, -S and --find-object are mutually exclusive"));
4434
4435         /*
4436          * Most of the time we can say "there are changes"
4437          * only by checking if there are changed paths, but
4438          * --ignore-whitespace* options force us to look
4439          * inside contents.
4440          */
4441
4442         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4443                 options->flags.diff_from_contents = 1;
4444         else
4445                 options->flags.diff_from_contents = 0;
4446
4447         if (options->flags.find_copies_harder)
4448                 options->detect_rename = DIFF_DETECT_COPY;
4449
4450         if (!options->flags.relative_name)
4451                 options->prefix = NULL;
4452         if (options->prefix)
4453                 options->prefix_length = strlen(options->prefix);
4454         else
4455                 options->prefix_length = 0;
4456
4457         if (options->output_format & (DIFF_FORMAT_NAME |
4458                                       DIFF_FORMAT_NAME_STATUS |
4459                                       DIFF_FORMAT_CHECKDIFF |
4460                                       DIFF_FORMAT_NO_OUTPUT))
4461                 options->output_format &= ~(DIFF_FORMAT_RAW |
4462                                             DIFF_FORMAT_NUMSTAT |
4463                                             DIFF_FORMAT_DIFFSTAT |
4464                                             DIFF_FORMAT_SHORTSTAT |
4465                                             DIFF_FORMAT_DIRSTAT |
4466                                             DIFF_FORMAT_SUMMARY |
4467                                             DIFF_FORMAT_PATCH);
4468
4469         /*
4470          * These cases always need recursive; we do not drop caller-supplied
4471          * recursive bits for other formats here.
4472          */
4473         if (options->output_format & (DIFF_FORMAT_PATCH |
4474                                       DIFF_FORMAT_NUMSTAT |
4475                                       DIFF_FORMAT_DIFFSTAT |
4476                                       DIFF_FORMAT_SHORTSTAT |
4477                                       DIFF_FORMAT_DIRSTAT |
4478                                       DIFF_FORMAT_SUMMARY |
4479                                       DIFF_FORMAT_CHECKDIFF))
4480                 options->flags.recursive = 1;
4481         /*
4482          * Also pickaxe would not work very well if you do not say recursive
4483          */
4484         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4485                 options->flags.recursive = 1;
4486         /*
4487          * When patches are generated, submodules diffed against the work tree
4488          * must be checked for dirtiness too so it can be shown in the output
4489          */
4490         if (options->output_format & DIFF_FORMAT_PATCH)
4491                 options->flags.dirty_submodules = 1;
4492
4493         if (options->detect_rename && options->rename_limit < 0)
4494                 options->rename_limit = diff_rename_limit_default;
4495         if (options->setup & DIFF_SETUP_USE_CACHE) {
4496                 if (!active_cache)
4497                         /* read-cache does not die even when it fails
4498                          * so it is safe for us to do this here.  Also
4499                          * it does not smudge active_cache or active_nr
4500                          * when it fails, so we do not have to worry about
4501                          * cleaning it up ourselves either.
4502                          */
4503                         read_cache();
4504         }
4505         if (hexsz < options->abbrev)
4506                 options->abbrev = hexsz; /* full */
4507
4508         /*
4509          * It does not make sense to show the first hit we happened
4510          * to have found.  It does not make sense not to return with
4511          * exit code in such a case either.
4512          */
4513         if (options->flags.quick) {
4514                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4515                 options->flags.exit_with_status = 1;
4516         }
4517
4518         options->diff_path_counter = 0;
4519
4520         if (options->flags.follow_renames && options->pathspec.nr != 1)
4521                 die(_("--follow requires exactly one pathspec"));
4522
4523         if (!options->use_color || external_diff())
4524                 options->color_moved = 0;
4525 }
4526
4527 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4528 {
4529         char c, *eq;
4530         int len;
4531
4532         if (*arg != '-')
4533                 return 0;
4534         c = *++arg;
4535         if (!c)
4536                 return 0;
4537         if (c == arg_short) {
4538                 c = *++arg;
4539                 if (!c)
4540                         return 1;
4541                 if (val && isdigit(c)) {
4542                         char *end;
4543                         int n = strtoul(arg, &end, 10);
4544                         if (*end)
4545                                 return 0;
4546                         *val = n;
4547                         return 1;
4548                 }
4549                 return 0;
4550         }
4551         if (c != '-')
4552                 return 0;
4553         arg++;
4554         eq = strchrnul(arg, '=');
4555         len = eq - arg;
4556         if (!len || strncmp(arg, arg_long, len))
4557                 return 0;
4558         if (*eq) {
4559                 int n;
4560                 char *end;
4561                 if (!isdigit(*++eq))
4562                         return 0;
4563                 n = strtoul(eq, &end, 10);
4564                 if (*end)
4565                         return 0;
4566                 *val = n;
4567         }
4568         return 1;
4569 }
4570
4571 static int diff_scoreopt_parse(const char *opt);
4572
4573 static inline int short_opt(char opt, const char **argv,
4574                             const char **optarg)
4575 {
4576         const char *arg = argv[0];
4577         if (arg[0] != '-' || arg[1] != opt)
4578                 return 0;
4579         if (arg[2] != '\0') {
4580                 *optarg = arg + 2;
4581                 return 1;
4582         }
4583         if (!argv[1])
4584                 die("Option '%c' requires a value", opt);
4585         *optarg = argv[1];
4586         return 2;
4587 }
4588
4589 int parse_long_opt(const char *opt, const char **argv,
4590                    const char **optarg)
4591 {
4592         const char *arg = argv[0];
4593         if (!skip_prefix(arg, "--", &arg))
4594                 return 0;
4595         if (!skip_prefix(arg, opt, &arg))
4596                 return 0;
4597         if (*arg == '=') { /* stuck form: --option=value */
4598                 *optarg = arg + 1;
4599                 return 1;
4600         }
4601         if (*arg != '\0')
4602                 return 0;
4603         /* separate form: --option value */
4604         if (!argv[1])
4605                 die("Option '--%s' requires a value", opt);
4606         *optarg = argv[1];
4607         return 2;
4608 }
4609
4610 static int stat_opt(struct diff_options *options, const char **av)
4611 {
4612         const char *arg = av[0];
4613         char *end;
4614         int width = options->stat_width;
4615         int name_width = options->stat_name_width;
4616         int graph_width = options->stat_graph_width;
4617         int count = options->stat_count;
4618         int argcount = 1;
4619
4620         if (!skip_prefix(arg, "--stat", &arg))
4621                 BUG("stat option does not begin with --stat: %s", arg);
4622         end = (char *)arg;
4623
4624         switch (*arg) {
4625         case '-':
4626                 if (skip_prefix(arg, "-width", &arg)) {
4627                         if (*arg == '=')
4628                                 width = strtoul(arg + 1, &end, 10);
4629                         else if (!*arg && !av[1])
4630                                 die_want_option("--stat-width");
4631                         else if (!*arg) {
4632                                 width = strtoul(av[1], &end, 10);
4633                                 argcount = 2;
4634                         }
4635                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4636                         if (*arg == '=')
4637                                 name_width = strtoul(arg + 1, &end, 10);
4638                         else if (!*arg && !av[1])
4639                                 die_want_option("--stat-name-width");
4640                         else if (!*arg) {
4641                                 name_width = strtoul(av[1], &end, 10);
4642                                 argcount = 2;
4643                         }
4644                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4645                         if (*arg == '=')
4646                                 graph_width = strtoul(arg + 1, &end, 10);
4647                         else if (!*arg && !av[1])
4648                                 die_want_option("--stat-graph-width");
4649                         else if (!*arg) {
4650                                 graph_width = strtoul(av[1], &end, 10);
4651                                 argcount = 2;
4652                         }
4653                 } else if (skip_prefix(arg, "-count", &arg)) {
4654                         if (*arg == '=')
4655                                 count = strtoul(arg + 1, &end, 10);
4656                         else if (!*arg && !av[1])
4657                                 die_want_option("--stat-count");
4658                         else if (!*arg) {
4659                                 count = strtoul(av[1], &end, 10);
4660                                 argcount = 2;
4661                         }
4662                 }
4663                 break;
4664         case '=':
4665                 width = strtoul(arg+1, &end, 10);
4666                 if (*end == ',')
4667                         name_width = strtoul(end+1, &end, 10);
4668                 if (*end == ',')
4669                         count = strtoul(end+1, &end, 10);
4670         }
4671
4672         /* Important! This checks all the error cases! */
4673         if (*end)
4674                 return 0;
4675         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4676         options->stat_name_width = name_width;
4677         options->stat_graph_width = graph_width;
4678         options->stat_width = width;
4679         options->stat_count = count;
4680         return argcount;
4681 }
4682
4683 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4684 {
4685         struct strbuf errmsg = STRBUF_INIT;
4686         if (parse_dirstat_params(options, params, &errmsg))
4687                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4688                     errmsg.buf);
4689         strbuf_release(&errmsg);
4690         /*
4691          * The caller knows a dirstat-related option is given from the command
4692          * line; allow it to say "return this_function();"
4693          */
4694         options->output_format |= DIFF_FORMAT_DIRSTAT;
4695         return 1;
4696 }
4697
4698 static int parse_submodule_opt(struct diff_options *options, const char *value)
4699 {
4700         if (parse_submodule_params(options, value))
4701                 die(_("Failed to parse --submodule option parameter: '%s'"),
4702                         value);
4703         return 1;
4704 }
4705
4706 static const char diff_status_letters[] = {
4707         DIFF_STATUS_ADDED,
4708         DIFF_STATUS_COPIED,
4709         DIFF_STATUS_DELETED,
4710         DIFF_STATUS_MODIFIED,
4711         DIFF_STATUS_RENAMED,
4712         DIFF_STATUS_TYPE_CHANGED,
4713         DIFF_STATUS_UNKNOWN,
4714         DIFF_STATUS_UNMERGED,
4715         DIFF_STATUS_FILTER_AON,
4716         DIFF_STATUS_FILTER_BROKEN,
4717         '\0',
4718 };
4719
4720 static unsigned int filter_bit['Z' + 1];
4721
4722 static void prepare_filter_bits(void)
4723 {
4724         int i;
4725
4726         if (!filter_bit[DIFF_STATUS_ADDED]) {
4727                 for (i = 0; diff_status_letters[i]; i++)
4728                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4729         }
4730 }
4731
4732 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4733 {
4734         return opt->filter & filter_bit[(int) status];
4735 }
4736
4737 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4738 {
4739         int i, optch;
4740
4741         prepare_filter_bits();
4742
4743         /*
4744          * If there is a negation e.g. 'd' in the input, and we haven't
4745          * initialized the filter field with another --diff-filter, start
4746          * from full set of bits, except for AON.
4747          */
4748         if (!opt->filter) {
4749                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4750                         if (optch < 'a' || 'z' < optch)
4751                                 continue;
4752                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4753                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4754                         break;
4755                 }
4756         }
4757
4758         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4759                 unsigned int bit;
4760                 int negate;
4761
4762                 if ('a' <= optch && optch <= 'z') {
4763                         negate = 1;
4764                         optch = toupper(optch);
4765                 } else {
4766                         negate = 0;
4767                 }
4768
4769                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4770                 if (!bit)
4771                         return optarg[i];
4772                 if (negate)
4773                         opt->filter &= ~bit;
4774                 else
4775                         opt->filter |= bit;
4776         }
4777         return 0;
4778 }
4779
4780 static void enable_patch_output(int *fmt) {
4781         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4782         *fmt |= DIFF_FORMAT_PATCH;
4783 }
4784
4785 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4786 {
4787         int val = parse_ws_error_highlight(arg);
4788
4789         if (val < 0) {
4790                 error("unknown value after ws-error-highlight=%.*s",
4791                       -1 - val, arg);
4792                 return 0;
4793         }
4794         opt->ws_error_highlight = val;
4795         return 1;
4796 }
4797
4798 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4799 {
4800         struct object_id oid;
4801
4802         if (get_oid(arg, &oid))
4803                 return error("unable to resolve '%s'", arg);
4804
4805         if (!opt->objfind)
4806                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4807
4808         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4809         opt->flags.recursive = 1;
4810         opt->flags.tree_in_recursive = 1;
4811         oidset_insert(opt->objfind, &oid);
4812         return 1;
4813 }
4814
4815 int diff_opt_parse(struct diff_options *options,
4816                    const char **av, int ac, const char *prefix)
4817 {
4818         const char *arg = av[0];
4819         const char *optarg;
4820         int argcount;
4821
4822         if (!prefix)
4823                 prefix = "";
4824
4825         /* Output format options */
4826         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4827             || opt_arg(arg, 'U', "unified", &options->context))
4828                 enable_patch_output(&options->output_format);
4829         else if (!strcmp(arg, "--raw"))
4830                 options->output_format |= DIFF_FORMAT_RAW;
4831         else if (!strcmp(arg, "--patch-with-raw")) {
4832                 enable_patch_output(&options->output_format);
4833                 options->output_format |= DIFF_FORMAT_RAW;
4834         } else if (!strcmp(arg, "--numstat"))
4835                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4836         else if (!strcmp(arg, "--shortstat"))
4837                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4838         else if (skip_prefix(arg, "-X", &arg) ||
4839                  skip_to_optional_arg(arg, "--dirstat", &arg))
4840                 return parse_dirstat_opt(options, arg);
4841         else if (!strcmp(arg, "--cumulative"))
4842                 return parse_dirstat_opt(options, "cumulative");
4843         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4844                 parse_dirstat_opt(options, "files");
4845                 return parse_dirstat_opt(options, arg);
4846         }
4847         else if (!strcmp(arg, "--check"))
4848                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4849         else if (!strcmp(arg, "--summary"))
4850                 options->output_format |= DIFF_FORMAT_SUMMARY;
4851         else if (!strcmp(arg, "--patch-with-stat")) {
4852                 enable_patch_output(&options->output_format);
4853                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4854         } else if (!strcmp(arg, "--name-only"))
4855                 options->output_format |= DIFF_FORMAT_NAME;
4856         else if (!strcmp(arg, "--name-status"))
4857                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4858         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4859                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4860         else if (starts_with(arg, "--stat"))
4861                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4862                 return stat_opt(options, av);
4863         else if (!strcmp(arg, "--compact-summary")) {
4864                  options->flags.stat_with_summary = 1;
4865                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4866         } else if (!strcmp(arg, "--no-compact-summary"))
4867                  options->flags.stat_with_summary = 0;
4868
4869         /* renames options */
4870         else if (starts_with(arg, "-B") ||
4871                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4872                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4873                         return error("invalid argument to -B: %s", arg+2);
4874         }
4875         else if (starts_with(arg, "-M") ||
4876                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4877                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4878                         return error("invalid argument to -M: %s", arg+2);
4879                 options->detect_rename = DIFF_DETECT_RENAME;
4880         }
4881         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4882                 options->irreversible_delete = 1;
4883         }
4884         else if (starts_with(arg, "-C") ||
4885                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4886                 if (options->detect_rename == DIFF_DETECT_COPY)
4887                         options->flags.find_copies_harder = 1;
4888                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4889                         return error("invalid argument to -C: %s", arg+2);
4890                 options->detect_rename = DIFF_DETECT_COPY;
4891         }
4892         else if (!strcmp(arg, "--no-renames"))
4893                 options->detect_rename = 0;
4894         else if (!strcmp(arg, "--rename-empty"))
4895                 options->flags.rename_empty = 1;
4896         else if (!strcmp(arg, "--no-rename-empty"))
4897                 options->flags.rename_empty = 0;
4898         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4899                 options->flags.relative_name = 1;
4900                 if (arg)
4901                         options->prefix = arg;
4902         }
4903
4904         /* xdiff options */
4905         else if (!strcmp(arg, "--minimal"))
4906                 DIFF_XDL_SET(options, NEED_MINIMAL);
4907         else if (!strcmp(arg, "--no-minimal"))
4908                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4909         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4910                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4911         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4912                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4913         else if (!strcmp(arg, "--ignore-space-at-eol"))
4914                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4915         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4916                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4917         else if (!strcmp(arg, "--ignore-blank-lines"))
4918                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4919         else if (!strcmp(arg, "--indent-heuristic"))
4920                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4921         else if (!strcmp(arg, "--no-indent-heuristic"))
4922                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4923         else if (!strcmp(arg, "--patience")) {
4924                 int i;
4925                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4926                 /*
4927                  * Both --patience and --anchored use PATIENCE_DIFF
4928                  * internally, so remove any anchors previously
4929                  * specified.
4930                  */
4931                 for (i = 0; i < options->anchors_nr; i++)
4932                         free(options->anchors[i]);
4933                 options->anchors_nr = 0;
4934         } else if (!strcmp(arg, "--histogram"))
4935                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4936         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4937                 long value = parse_algorithm_value(optarg);
4938                 if (value < 0)
4939                         return error("option diff-algorithm accepts \"myers\", "
4940                                      "\"minimal\", \"patience\" and \"histogram\"");
4941                 /* clear out previous settings */
4942                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4943                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4944                 options->xdl_opts |= value;
4945                 return argcount;
4946         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4947                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4948                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4949                            options->anchors_alloc);
4950                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4951         }
4952
4953         /* flags options */
4954         else if (!strcmp(arg, "--binary")) {
4955                 enable_patch_output(&options->output_format);
4956                 options->flags.binary = 1;
4957         }
4958         else if (!strcmp(arg, "--full-index"))
4959                 options->flags.full_index = 1;
4960         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4961                 options->flags.text = 1;
4962         else if (!strcmp(arg, "-R"))
4963                 options->flags.reverse_diff = 1;
4964         else if (!strcmp(arg, "--find-copies-harder"))
4965                 options->flags.find_copies_harder = 1;
4966         else if (!strcmp(arg, "--follow"))
4967                 options->flags.follow_renames = 1;
4968         else if (!strcmp(arg, "--no-follow")) {
4969                 options->flags.follow_renames = 0;
4970                 options->flags.default_follow_renames = 0;
4971         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4972                 int value = git_config_colorbool(NULL, arg);
4973                 if (value < 0)
4974                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4975                 options->use_color = value;
4976         }
4977         else if (!strcmp(arg, "--no-color"))
4978                 options->use_color = 0;
4979         else if (!strcmp(arg, "--color-moved")) {
4980                 if (diff_color_moved_default)
4981                         options->color_moved = diff_color_moved_default;
4982                 if (options->color_moved == COLOR_MOVED_NO)
4983                         options->color_moved = COLOR_MOVED_DEFAULT;
4984         } else if (!strcmp(arg, "--no-color-moved"))
4985                 options->color_moved = COLOR_MOVED_NO;
4986         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4987                 int cm = parse_color_moved(arg);
4988                 if (cm < 0)
4989                         die("bad --color-moved argument: %s", arg);
4990                 options->color_moved = cm;
4991         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
4992                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
4993         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4994                 options->use_color = 1;
4995                 options->word_diff = DIFF_WORDS_COLOR;
4996         }
4997         else if (!strcmp(arg, "--word-diff")) {
4998                 if (options->word_diff == DIFF_WORDS_NONE)
4999                         options->word_diff = DIFF_WORDS_PLAIN;
5000         }
5001         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5002                 if (!strcmp(arg, "plain"))
5003                         options->word_diff = DIFF_WORDS_PLAIN;
5004                 else if (!strcmp(arg, "color")) {
5005                         options->use_color = 1;
5006                         options->word_diff = DIFF_WORDS_COLOR;
5007                 }
5008                 else if (!strcmp(arg, "porcelain"))
5009                         options->word_diff = DIFF_WORDS_PORCELAIN;
5010                 else if (!strcmp(arg, "none"))
5011                         options->word_diff = DIFF_WORDS_NONE;
5012                 else
5013                         die("bad --word-diff argument: %s", arg);
5014         }
5015         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5016                 if (options->word_diff == DIFF_WORDS_NONE)
5017                         options->word_diff = DIFF_WORDS_PLAIN;
5018                 options->word_regex = optarg;
5019                 return argcount;
5020         }
5021         else if (!strcmp(arg, "--exit-code"))
5022                 options->flags.exit_with_status = 1;
5023         else if (!strcmp(arg, "--quiet"))
5024                 options->flags.quick = 1;
5025         else if (!strcmp(arg, "--ext-diff"))
5026                 options->flags.allow_external = 1;
5027         else if (!strcmp(arg, "--no-ext-diff"))
5028                 options->flags.allow_external = 0;
5029         else if (!strcmp(arg, "--textconv")) {
5030                 options->flags.allow_textconv = 1;
5031                 options->flags.textconv_set_via_cmdline = 1;
5032         } else if (!strcmp(arg, "--no-textconv"))
5033                 options->flags.allow_textconv = 0;
5034         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5035                 options->flags.override_submodule_config = 1;
5036                 handle_ignore_submodules_arg(options, arg);
5037         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5038                 return parse_submodule_opt(options, arg);
5039         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5040                 return parse_ws_error_highlight_opt(options, arg);
5041         else if (!strcmp(arg, "--ita-invisible-in-index"))
5042                 options->ita_invisible_in_index = 1;
5043         else if (!strcmp(arg, "--ita-visible-in-index"))
5044                 options->ita_invisible_in_index = 0;
5045
5046         /* misc options */
5047         else if (!strcmp(arg, "-z"))
5048                 options->line_termination = 0;
5049         else if ((argcount = short_opt('l', av, &optarg))) {
5050                 options->rename_limit = strtoul(optarg, NULL, 10);
5051                 return argcount;
5052         }
5053         else if ((argcount = short_opt('S', av, &optarg))) {
5054                 options->pickaxe = optarg;
5055                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5056                 return argcount;
5057         } else if ((argcount = short_opt('G', av, &optarg))) {
5058                 options->pickaxe = optarg;
5059                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5060                 return argcount;
5061         }
5062         else if (!strcmp(arg, "--pickaxe-all"))
5063                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5064         else if (!strcmp(arg, "--pickaxe-regex"))
5065                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5066         else if ((argcount = short_opt('O', av, &optarg))) {
5067                 options->orderfile = prefix_filename(prefix, optarg);
5068                 return argcount;
5069         } else if (skip_prefix(arg, "--find-object=", &arg))
5070                 return parse_objfind_opt(options, arg);
5071         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5072                 int offending = parse_diff_filter_opt(optarg, options);
5073                 if (offending)
5074                         die("unknown change class '%c' in --diff-filter=%s",
5075                             offending, optarg);
5076                 return argcount;
5077         }
5078         else if (!strcmp(arg, "--no-abbrev"))
5079                 options->abbrev = 0;
5080         else if (!strcmp(arg, "--abbrev"))
5081                 options->abbrev = DEFAULT_ABBREV;
5082         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5083                 options->abbrev = strtoul(arg, NULL, 10);
5084                 if (options->abbrev < MINIMUM_ABBREV)
5085                         options->abbrev = MINIMUM_ABBREV;
5086                 else if (the_hash_algo->hexsz < options->abbrev)
5087                         options->abbrev = the_hash_algo->hexsz;
5088         }
5089         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5090                 options->a_prefix = optarg;
5091                 return argcount;
5092         }
5093         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5094                 options->line_prefix = optarg;
5095                 options->line_prefix_length = strlen(options->line_prefix);
5096                 graph_setup_line_prefix(options);
5097                 return argcount;
5098         }
5099         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5100                 options->b_prefix = optarg;
5101                 return argcount;
5102         }
5103         else if (!strcmp(arg, "--no-prefix"))
5104                 options->a_prefix = options->b_prefix = "";
5105         else if (opt_arg(arg, '\0', "inter-hunk-context",
5106                          &options->interhunkcontext))
5107                 ;
5108         else if (!strcmp(arg, "-W"))
5109                 options->flags.funccontext = 1;
5110         else if (!strcmp(arg, "--function-context"))
5111                 options->flags.funccontext = 1;
5112         else if (!strcmp(arg, "--no-function-context"))
5113                 options->flags.funccontext = 0;
5114         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5115                 char *path = prefix_filename(prefix, optarg);
5116                 options->file = xfopen(path, "w");
5117                 options->close_file = 1;
5118                 if (options->use_color != GIT_COLOR_ALWAYS)
5119                         options->use_color = GIT_COLOR_NEVER;
5120                 free(path);
5121                 return argcount;
5122         } else
5123                 return 0;
5124         return 1;
5125 }
5126
5127 int parse_rename_score(const char **cp_p)
5128 {
5129         unsigned long num, scale;
5130         int ch, dot;
5131         const char *cp = *cp_p;
5132
5133         num = 0;
5134         scale = 1;
5135         dot = 0;
5136         for (;;) {
5137                 ch = *cp;
5138                 if ( !dot && ch == '.' ) {
5139                         scale = 1;
5140                         dot = 1;
5141                 } else if ( ch == '%' ) {
5142                         scale = dot ? scale*100 : 100;
5143                         cp++;   /* % is always at the end */
5144                         break;
5145                 } else if ( ch >= '0' && ch <= '9' ) {
5146                         if ( scale < 100000 ) {
5147                                 scale *= 10;
5148                                 num = (num*10) + (ch-'0');
5149                         }
5150                 } else {
5151                         break;
5152                 }
5153                 cp++;
5154         }
5155         *cp_p = cp;
5156
5157         /* user says num divided by scale and we say internally that
5158          * is MAX_SCORE * num / scale.
5159          */
5160         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5161 }
5162
5163 static int diff_scoreopt_parse(const char *opt)
5164 {
5165         int opt1, opt2, cmd;
5166
5167         if (*opt++ != '-')
5168                 return -1;
5169         cmd = *opt++;
5170         if (cmd == '-') {
5171                 /* convert the long-form arguments into short-form versions */
5172                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5173                         if (*opt == 0 || *opt++ == '=')
5174                                 cmd = 'B';
5175                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5176                         if (*opt == 0 || *opt++ == '=')
5177                                 cmd = 'C';
5178                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5179                         if (*opt == 0 || *opt++ == '=')
5180                                 cmd = 'M';
5181                 }
5182         }
5183         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5184                 return -1; /* that is not a -M, -C, or -B option */
5185
5186         opt1 = parse_rename_score(&opt);
5187         if (cmd != 'B')
5188                 opt2 = 0;
5189         else {
5190                 if (*opt == 0)
5191                         opt2 = 0;
5192                 else if (*opt != '/')
5193                         return -1; /* we expect -B80/99 or -B80 */
5194                 else {
5195                         opt++;
5196                         opt2 = parse_rename_score(&opt);
5197                 }
5198         }
5199         if (*opt != 0)
5200                 return -1;
5201         return opt1 | (opt2 << 16);
5202 }
5203
5204 struct diff_queue_struct diff_queued_diff;
5205
5206 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5207 {
5208         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5209         queue->queue[queue->nr++] = dp;
5210 }
5211
5212 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5213                                  struct diff_filespec *one,
5214                                  struct diff_filespec *two)
5215 {
5216         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5217         dp->one = one;
5218         dp->two = two;
5219         if (queue)
5220                 diff_q(queue, dp);
5221         return dp;
5222 }
5223
5224 void diff_free_filepair(struct diff_filepair *p)
5225 {
5226         free_filespec(p->one);
5227         free_filespec(p->two);
5228         free(p);
5229 }
5230
5231 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5232 {
5233         int abblen;
5234         const char *abbrev;
5235
5236         /* Do we want all 40 hex characters? */
5237         if (len == the_hash_algo->hexsz)
5238                 return oid_to_hex(oid);
5239
5240         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5241         abbrev = diff_abbrev_oid(oid, len);
5242
5243         if (!print_sha1_ellipsis())
5244                 return abbrev;
5245
5246         abblen = strlen(abbrev);
5247
5248         /*
5249          * In well-behaved cases, where the abbreviated result is the
5250          * same as the requested length, append three dots after the
5251          * abbreviation (hence the whole logic is limited to the case
5252          * where abblen < 37); when the actual abbreviated result is a
5253          * bit longer than the requested length, we reduce the number
5254          * of dots so that they match the well-behaved ones.  However,
5255          * if the actual abbreviation is longer than the requested
5256          * length by more than three, we give up on aligning, and add
5257          * three dots anyway, to indicate that the output is not the
5258          * full object name.  Yes, this may be suboptimal, but this
5259          * appears only in "diff --raw --abbrev" output and it is not
5260          * worth the effort to change it now.  Note that this would
5261          * likely to work fine when the automatic sizing of default
5262          * abbreviation length is used--we would be fed -1 in "len" in
5263          * that case, and will end up always appending three-dots, but
5264          * the automatic sizing is supposed to give abblen that ensures
5265          * uniqueness across all objects (statistically speaking).
5266          */
5267         if (abblen < the_hash_algo->hexsz - 3) {
5268                 static char hex[GIT_MAX_HEXSZ + 1];
5269                 if (len < abblen && abblen <= len + 2)
5270                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5271                 else
5272                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5273                 return hex;
5274         }
5275
5276         return oid_to_hex(oid);
5277 }
5278
5279 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5280 {
5281         int line_termination = opt->line_termination;
5282         int inter_name_termination = line_termination ? '\t' : '\0';
5283
5284         fprintf(opt->file, "%s", diff_line_prefix(opt));
5285         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5286                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5287                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5288                 fprintf(opt->file, "%s ",
5289                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5290         }
5291         if (p->score) {
5292                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5293                         inter_name_termination);
5294         } else {
5295                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5296         }
5297
5298         if (p->status == DIFF_STATUS_COPIED ||
5299             p->status == DIFF_STATUS_RENAMED) {
5300                 const char *name_a, *name_b;
5301                 name_a = p->one->path;
5302                 name_b = p->two->path;
5303                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5304                 write_name_quoted(name_a, opt->file, inter_name_termination);
5305                 write_name_quoted(name_b, opt->file, line_termination);
5306         } else {
5307                 const char *name_a, *name_b;
5308                 name_a = p->one->mode ? p->one->path : p->two->path;
5309                 name_b = NULL;
5310                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5311                 write_name_quoted(name_a, opt->file, line_termination);
5312         }
5313 }
5314
5315 int diff_unmodified_pair(struct diff_filepair *p)
5316 {
5317         /* This function is written stricter than necessary to support
5318          * the currently implemented transformers, but the idea is to
5319          * let transformers to produce diff_filepairs any way they want,
5320          * and filter and clean them up here before producing the output.
5321          */
5322         struct diff_filespec *one = p->one, *two = p->two;
5323
5324         if (DIFF_PAIR_UNMERGED(p))
5325                 return 0; /* unmerged is interesting */
5326
5327         /* deletion, addition, mode or type change
5328          * and rename are all interesting.
5329          */
5330         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5331             DIFF_PAIR_MODE_CHANGED(p) ||
5332             strcmp(one->path, two->path))
5333                 return 0;
5334
5335         /* both are valid and point at the same path.  that is, we are
5336          * dealing with a change.
5337          */
5338         if (one->oid_valid && two->oid_valid &&
5339             !oidcmp(&one->oid, &two->oid) &&
5340             !one->dirty_submodule && !two->dirty_submodule)
5341                 return 1; /* no change */
5342         if (!one->oid_valid && !two->oid_valid)
5343                 return 1; /* both look at the same file on the filesystem. */
5344         return 0;
5345 }
5346
5347 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5348 {
5349         if (diff_unmodified_pair(p))
5350                 return;
5351
5352         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5353             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5354                 return; /* no tree diffs in patch format */
5355
5356         run_diff(p, o);
5357 }
5358
5359 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5360                             struct diffstat_t *diffstat)
5361 {
5362         if (diff_unmodified_pair(p))
5363                 return;
5364
5365         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5366             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5367                 return; /* no useful stat for tree diffs */
5368
5369         run_diffstat(p, o, diffstat);
5370 }
5371
5372 static void diff_flush_checkdiff(struct diff_filepair *p,
5373                 struct diff_options *o)
5374 {
5375         if (diff_unmodified_pair(p))
5376                 return;
5377
5378         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5379             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5380                 return; /* nothing to check in tree diffs */
5381
5382         run_checkdiff(p, o);
5383 }
5384
5385 int diff_queue_is_empty(void)
5386 {
5387         struct diff_queue_struct *q = &diff_queued_diff;
5388         int i;
5389         for (i = 0; i < q->nr; i++)
5390                 if (!diff_unmodified_pair(q->queue[i]))
5391                         return 0;
5392         return 1;
5393 }
5394
5395 #if DIFF_DEBUG
5396 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5397 {
5398         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5399                 x, one ? one : "",
5400                 s->path,
5401                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5402                 s->mode,
5403                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5404         fprintf(stderr, "queue[%d] %s size %lu\n",
5405                 x, one ? one : "",
5406                 s->size);
5407 }
5408
5409 void diff_debug_filepair(const struct diff_filepair *p, int i)
5410 {
5411         diff_debug_filespec(p->one, i, "one");
5412         diff_debug_filespec(p->two, i, "two");
5413         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5414                 p->score, p->status ? p->status : '?',
5415                 p->one->rename_used, p->broken_pair);
5416 }
5417
5418 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5419 {
5420         int i;
5421         if (msg)
5422                 fprintf(stderr, "%s\n", msg);
5423         fprintf(stderr, "q->nr = %d\n", q->nr);
5424         for (i = 0; i < q->nr; i++) {
5425                 struct diff_filepair *p = q->queue[i];
5426                 diff_debug_filepair(p, i);
5427         }
5428 }
5429 #endif
5430
5431 static void diff_resolve_rename_copy(void)
5432 {
5433         int i;
5434         struct diff_filepair *p;
5435         struct diff_queue_struct *q = &diff_queued_diff;
5436
5437         diff_debug_queue("resolve-rename-copy", q);
5438
5439         for (i = 0; i < q->nr; i++) {
5440                 p = q->queue[i];
5441                 p->status = 0; /* undecided */
5442                 if (DIFF_PAIR_UNMERGED(p))
5443                         p->status = DIFF_STATUS_UNMERGED;
5444                 else if (!DIFF_FILE_VALID(p->one))
5445                         p->status = DIFF_STATUS_ADDED;
5446                 else if (!DIFF_FILE_VALID(p->two))
5447                         p->status = DIFF_STATUS_DELETED;
5448                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5449                         p->status = DIFF_STATUS_TYPE_CHANGED;
5450
5451                 /* from this point on, we are dealing with a pair
5452                  * whose both sides are valid and of the same type, i.e.
5453                  * either in-place edit or rename/copy edit.
5454                  */
5455                 else if (DIFF_PAIR_RENAME(p)) {
5456                         /*
5457                          * A rename might have re-connected a broken
5458                          * pair up, causing the pathnames to be the
5459                          * same again. If so, that's not a rename at
5460                          * all, just a modification..
5461                          *
5462                          * Otherwise, see if this source was used for
5463                          * multiple renames, in which case we decrement
5464                          * the count, and call it a copy.
5465                          */
5466                         if (!strcmp(p->one->path, p->two->path))
5467                                 p->status = DIFF_STATUS_MODIFIED;
5468                         else if (--p->one->rename_used > 0)
5469                                 p->status = DIFF_STATUS_COPIED;
5470                         else
5471                                 p->status = DIFF_STATUS_RENAMED;
5472                 }
5473                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5474                          p->one->mode != p->two->mode ||
5475                          p->one->dirty_submodule ||
5476                          p->two->dirty_submodule ||
5477                          is_null_oid(&p->one->oid))
5478                         p->status = DIFF_STATUS_MODIFIED;
5479                 else {
5480                         /* This is a "no-change" entry and should not
5481                          * happen anymore, but prepare for broken callers.
5482                          */
5483                         error("feeding unmodified %s to diffcore",
5484                               p->one->path);
5485                         p->status = DIFF_STATUS_UNKNOWN;
5486                 }
5487         }
5488         diff_debug_queue("resolve-rename-copy done", q);
5489 }
5490
5491 static int check_pair_status(struct diff_filepair *p)
5492 {
5493         switch (p->status) {
5494         case DIFF_STATUS_UNKNOWN:
5495                 return 0;
5496         case 0:
5497                 die("internal error in diff-resolve-rename-copy");
5498         default:
5499                 return 1;
5500         }
5501 }
5502
5503 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5504 {
5505         int fmt = opt->output_format;
5506
5507         if (fmt & DIFF_FORMAT_CHECKDIFF)
5508                 diff_flush_checkdiff(p, opt);
5509         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5510                 diff_flush_raw(p, opt);
5511         else if (fmt & DIFF_FORMAT_NAME) {
5512                 const char *name_a, *name_b;
5513                 name_a = p->two->path;
5514                 name_b = NULL;
5515                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5516                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5517                 write_name_quoted(name_a, opt->file, opt->line_termination);
5518         }
5519 }
5520
5521 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5522 {
5523         struct strbuf sb = STRBUF_INIT;
5524         if (fs->mode)
5525                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5526         else
5527                 strbuf_addf(&sb, " %s ", newdelete);
5528
5529         quote_c_style(fs->path, &sb, NULL, 0);
5530         strbuf_addch(&sb, '\n');
5531         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5532                          sb.buf, sb.len, 0);
5533         strbuf_release(&sb);
5534 }
5535
5536 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5537                 int show_name)
5538 {
5539         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5540                 struct strbuf sb = STRBUF_INIT;
5541                 strbuf_addf(&sb, " mode change %06o => %06o",
5542                             p->one->mode, p->two->mode);
5543                 if (show_name) {
5544                         strbuf_addch(&sb, ' ');
5545                         quote_c_style(p->two->path, &sb, NULL, 0);
5546                 }
5547                 strbuf_addch(&sb, '\n');
5548                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5549                                  sb.buf, sb.len, 0);
5550                 strbuf_release(&sb);
5551         }
5552 }
5553
5554 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5555                 struct diff_filepair *p)
5556 {
5557         struct strbuf sb = STRBUF_INIT;
5558         struct strbuf names = STRBUF_INIT;
5559
5560         pprint_rename(&names, p->one->path, p->two->path);
5561         strbuf_addf(&sb, " %s %s (%d%%)\n",
5562                     renamecopy, names.buf, similarity_index(p));
5563         strbuf_release(&names);
5564         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5565                                  sb.buf, sb.len, 0);
5566         show_mode_change(opt, p, 0);
5567         strbuf_release(&sb);
5568 }
5569
5570 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5571 {
5572         switch(p->status) {
5573         case DIFF_STATUS_DELETED:
5574                 show_file_mode_name(opt, "delete", p->one);
5575                 break;
5576         case DIFF_STATUS_ADDED:
5577                 show_file_mode_name(opt, "create", p->two);
5578                 break;
5579         case DIFF_STATUS_COPIED:
5580                 show_rename_copy(opt, "copy", p);
5581                 break;
5582         case DIFF_STATUS_RENAMED:
5583                 show_rename_copy(opt, "rename", p);
5584                 break;
5585         default:
5586                 if (p->score) {
5587                         struct strbuf sb = STRBUF_INIT;
5588                         strbuf_addstr(&sb, " rewrite ");
5589                         quote_c_style(p->two->path, &sb, NULL, 0);
5590                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5591                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5592                                          sb.buf, sb.len, 0);
5593                         strbuf_release(&sb);
5594                 }
5595                 show_mode_change(opt, p, !p->score);
5596                 break;
5597         }
5598 }
5599
5600 struct patch_id_t {
5601         git_SHA_CTX *ctx;
5602         int patchlen;
5603 };
5604
5605 static int remove_space(char *line, int len)
5606 {
5607         int i;
5608         char *dst = line;
5609         unsigned char c;
5610
5611         for (i = 0; i < len; i++)
5612                 if (!isspace((c = line[i])))
5613                         *dst++ = c;
5614
5615         return dst - line;
5616 }
5617
5618 static void patch_id_consume(void *priv, char *line, unsigned long len)
5619 {
5620         struct patch_id_t *data = priv;
5621         int new_len;
5622
5623         /* Ignore line numbers when computing the SHA1 of the patch */
5624         if (starts_with(line, "@@ -"))
5625                 return;
5626
5627         new_len = remove_space(line, len);
5628
5629         git_SHA1_Update(data->ctx, line, new_len);
5630         data->patchlen += new_len;
5631 }
5632
5633 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5634 {
5635         git_SHA1_Update(ctx, str, strlen(str));
5636 }
5637
5638 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5639 {
5640         /* large enough for 2^32 in octal */
5641         char buf[12];
5642         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5643         git_SHA1_Update(ctx, buf, len);
5644 }
5645
5646 /* returns 0 upon success, and writes result into sha1 */
5647 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5648 {
5649         struct diff_queue_struct *q = &diff_queued_diff;
5650         int i;
5651         git_SHA_CTX ctx;
5652         struct patch_id_t data;
5653
5654         git_SHA1_Init(&ctx);
5655         memset(&data, 0, sizeof(struct patch_id_t));
5656         data.ctx = &ctx;
5657
5658         for (i = 0; i < q->nr; i++) {
5659                 xpparam_t xpp;
5660                 xdemitconf_t xecfg;
5661                 mmfile_t mf1, mf2;
5662                 struct diff_filepair *p = q->queue[i];
5663                 int len1, len2;
5664
5665                 memset(&xpp, 0, sizeof(xpp));
5666                 memset(&xecfg, 0, sizeof(xecfg));
5667                 if (p->status == 0)
5668                         return error("internal diff status error");
5669                 if (p->status == DIFF_STATUS_UNKNOWN)
5670                         continue;
5671                 if (diff_unmodified_pair(p))
5672                         continue;
5673                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5674                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5675                         continue;
5676                 if (DIFF_PAIR_UNMERGED(p))
5677                         continue;
5678
5679                 diff_fill_oid_info(p->one);
5680                 diff_fill_oid_info(p->two);
5681
5682                 len1 = remove_space(p->one->path, strlen(p->one->path));
5683                 len2 = remove_space(p->two->path, strlen(p->two->path));
5684                 patch_id_add_string(&ctx, "diff--git");
5685                 patch_id_add_string(&ctx, "a/");
5686                 git_SHA1_Update(&ctx, p->one->path, len1);
5687                 patch_id_add_string(&ctx, "b/");
5688                 git_SHA1_Update(&ctx, p->two->path, len2);
5689
5690                 if (p->one->mode == 0) {
5691                         patch_id_add_string(&ctx, "newfilemode");
5692                         patch_id_add_mode(&ctx, p->two->mode);
5693                         patch_id_add_string(&ctx, "---/dev/null");
5694                         patch_id_add_string(&ctx, "+++b/");
5695                         git_SHA1_Update(&ctx, p->two->path, len2);
5696                 } else if (p->two->mode == 0) {
5697                         patch_id_add_string(&ctx, "deletedfilemode");
5698                         patch_id_add_mode(&ctx, p->one->mode);
5699                         patch_id_add_string(&ctx, "---a/");
5700                         git_SHA1_Update(&ctx, p->one->path, len1);
5701                         patch_id_add_string(&ctx, "+++/dev/null");
5702                 } else {
5703                         patch_id_add_string(&ctx, "---a/");
5704                         git_SHA1_Update(&ctx, p->one->path, len1);
5705                         patch_id_add_string(&ctx, "+++b/");
5706                         git_SHA1_Update(&ctx, p->two->path, len2);
5707                 }
5708
5709                 if (diff_header_only)
5710                         continue;
5711
5712                 if (fill_mmfile(&mf1, p->one) < 0 ||
5713                     fill_mmfile(&mf2, p->two) < 0)
5714                         return error("unable to read files to diff");
5715
5716                 if (diff_filespec_is_binary(p->one) ||
5717                     diff_filespec_is_binary(p->two)) {
5718                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5719                                         GIT_SHA1_HEXSZ);
5720                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5721                                         GIT_SHA1_HEXSZ);
5722                         continue;
5723                 }
5724
5725                 xpp.flags = 0;
5726                 xecfg.ctxlen = 3;
5727                 xecfg.flags = 0;
5728                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5729                                   &xpp, &xecfg))
5730                         return error("unable to generate patch-id diff for %s",
5731                                      p->one->path);
5732         }
5733
5734         git_SHA1_Final(oid->hash, &ctx);
5735         return 0;
5736 }
5737
5738 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5739 {
5740         struct diff_queue_struct *q = &diff_queued_diff;
5741         int i;
5742         int result = diff_get_patch_id(options, oid, diff_header_only);
5743
5744         for (i = 0; i < q->nr; i++)
5745                 diff_free_filepair(q->queue[i]);
5746
5747         free(q->queue);
5748         DIFF_QUEUE_CLEAR(q);
5749
5750         return result;
5751 }
5752
5753 static int is_summary_empty(const struct diff_queue_struct *q)
5754 {
5755         int i;
5756
5757         for (i = 0; i < q->nr; i++) {
5758                 const struct diff_filepair *p = q->queue[i];
5759
5760                 switch (p->status) {
5761                 case DIFF_STATUS_DELETED:
5762                 case DIFF_STATUS_ADDED:
5763                 case DIFF_STATUS_COPIED:
5764                 case DIFF_STATUS_RENAMED:
5765                         return 0;
5766                 default:
5767                         if (p->score)
5768                                 return 0;
5769                         if (p->one->mode && p->two->mode &&
5770                             p->one->mode != p->two->mode)
5771                                 return 0;
5772                         break;
5773                 }
5774         }
5775         return 1;
5776 }
5777
5778 static const char rename_limit_warning[] =
5779 N_("inexact rename detection was skipped due to too many files.");
5780
5781 static const char degrade_cc_to_c_warning[] =
5782 N_("only found copies from modified paths due to too many files.");
5783
5784 static const char rename_limit_advice[] =
5785 N_("you may want to set your %s variable to at least "
5786    "%d and retry the command.");
5787
5788 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5789 {
5790         fflush(stdout);
5791         if (degraded_cc)
5792                 warning(_(degrade_cc_to_c_warning));
5793         else if (needed)
5794                 warning(_(rename_limit_warning));
5795         else
5796                 return;
5797         if (0 < needed)
5798                 warning(_(rename_limit_advice), varname, needed);
5799 }
5800
5801 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5802 {
5803         int i;
5804         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5805         struct diff_queue_struct *q = &diff_queued_diff;
5806
5807         if (WSEH_NEW & WS_RULE_MASK)
5808                 BUG("WS rules bit mask overlaps with diff symbol flags");
5809
5810         if (o->color_moved)
5811                 o->emitted_symbols = &esm;
5812
5813         for (i = 0; i < q->nr; i++) {
5814                 struct diff_filepair *p = q->queue[i];
5815                 if (check_pair_status(p))
5816                         diff_flush_patch(p, o);
5817         }
5818
5819         if (o->emitted_symbols) {
5820                 if (o->color_moved) {
5821                         struct hashmap add_lines, del_lines;
5822
5823                         if (o->color_moved_ws_handling &
5824                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5825                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5826
5827                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5828                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5829
5830                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5831                         mark_color_as_moved(o, &add_lines, &del_lines);
5832                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5833                                 dim_moved_lines(o);
5834
5835                         hashmap_free(&add_lines, 0);
5836                         hashmap_free(&del_lines, 0);
5837                 }
5838
5839                 for (i = 0; i < esm.nr; i++)
5840                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5841
5842                 for (i = 0; i < esm.nr; i++)
5843                         free((void *)esm.buf[i].line);
5844         }
5845         esm.nr = 0;
5846 }
5847
5848 void diff_flush(struct diff_options *options)
5849 {
5850         struct diff_queue_struct *q = &diff_queued_diff;
5851         int i, output_format = options->output_format;
5852         int separator = 0;
5853         int dirstat_by_line = 0;
5854
5855         /*
5856          * Order: raw, stat, summary, patch
5857          * or:    name/name-status/checkdiff (other bits clear)
5858          */
5859         if (!q->nr)
5860                 goto free_queue;
5861
5862         if (output_format & (DIFF_FORMAT_RAW |
5863                              DIFF_FORMAT_NAME |
5864                              DIFF_FORMAT_NAME_STATUS |
5865                              DIFF_FORMAT_CHECKDIFF)) {
5866                 for (i = 0; i < q->nr; i++) {
5867                         struct diff_filepair *p = q->queue[i];
5868                         if (check_pair_status(p))
5869                                 flush_one_pair(p, options);
5870                 }
5871                 separator++;
5872         }
5873
5874         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5875                 dirstat_by_line = 1;
5876
5877         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5878             dirstat_by_line) {
5879                 struct diffstat_t diffstat;
5880
5881                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5882                 for (i = 0; i < q->nr; i++) {
5883                         struct diff_filepair *p = q->queue[i];
5884                         if (check_pair_status(p))
5885                                 diff_flush_stat(p, options, &diffstat);
5886                 }
5887                 if (output_format & DIFF_FORMAT_NUMSTAT)
5888                         show_numstat(&diffstat, options);
5889                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5890                         show_stats(&diffstat, options);
5891                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5892                         show_shortstats(&diffstat, options);
5893                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5894                         show_dirstat_by_line(&diffstat, options);
5895                 free_diffstat_info(&diffstat);
5896                 separator++;
5897         }
5898         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5899                 show_dirstat(options);
5900
5901         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5902                 for (i = 0; i < q->nr; i++) {
5903                         diff_summary(options, q->queue[i]);
5904                 }
5905                 separator++;
5906         }
5907
5908         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5909             options->flags.exit_with_status &&
5910             options->flags.diff_from_contents) {
5911                 /*
5912                  * run diff_flush_patch for the exit status. setting
5913                  * options->file to /dev/null should be safe, because we
5914                  * aren't supposed to produce any output anyway.
5915                  */
5916                 if (options->close_file)
5917                         fclose(options->file);
5918                 options->file = xfopen("/dev/null", "w");
5919                 options->close_file = 1;
5920                 options->color_moved = 0;
5921                 for (i = 0; i < q->nr; i++) {
5922                         struct diff_filepair *p = q->queue[i];
5923                         if (check_pair_status(p))
5924                                 diff_flush_patch(p, options);
5925                         if (options->found_changes)
5926                                 break;
5927                 }
5928         }
5929
5930         if (output_format & DIFF_FORMAT_PATCH) {
5931                 if (separator) {
5932                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5933                         if (options->stat_sep)
5934                                 /* attach patch instead of inline */
5935                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5936                                                  NULL, 0, 0);
5937                 }
5938
5939                 diff_flush_patch_all_file_pairs(options);
5940         }
5941
5942         if (output_format & DIFF_FORMAT_CALLBACK)
5943                 options->format_callback(q, options, options->format_callback_data);
5944
5945         for (i = 0; i < q->nr; i++)
5946                 diff_free_filepair(q->queue[i]);
5947 free_queue:
5948         free(q->queue);
5949         DIFF_QUEUE_CLEAR(q);
5950         if (options->close_file)
5951                 fclose(options->file);
5952
5953         /*
5954          * Report the content-level differences with HAS_CHANGES;
5955          * diff_addremove/diff_change does not set the bit when
5956          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5957          */
5958         if (options->flags.diff_from_contents) {
5959                 if (options->found_changes)
5960                         options->flags.has_changes = 1;
5961                 else
5962                         options->flags.has_changes = 0;
5963         }
5964 }
5965
5966 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5967 {
5968         return (((p->status == DIFF_STATUS_MODIFIED) &&
5969                  ((p->score &&
5970                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5971                   (!p->score &&
5972                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5973                 ((p->status != DIFF_STATUS_MODIFIED) &&
5974                  filter_bit_tst(p->status, options)));
5975 }
5976
5977 static void diffcore_apply_filter(struct diff_options *options)
5978 {
5979         int i;
5980         struct diff_queue_struct *q = &diff_queued_diff;
5981         struct diff_queue_struct outq;
5982
5983         DIFF_QUEUE_CLEAR(&outq);
5984
5985         if (!options->filter)
5986                 return;
5987
5988         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5989                 int found;
5990                 for (i = found = 0; !found && i < q->nr; i++) {
5991                         if (match_filter(options, q->queue[i]))
5992                                 found++;
5993                 }
5994                 if (found)
5995                         return;
5996
5997                 /* otherwise we will clear the whole queue
5998                  * by copying the empty outq at the end of this
5999                  * function, but first clear the current entries
6000                  * in the queue.
6001                  */
6002                 for (i = 0; i < q->nr; i++)
6003                         diff_free_filepair(q->queue[i]);
6004         }
6005         else {
6006                 /* Only the matching ones */
6007                 for (i = 0; i < q->nr; i++) {
6008                         struct diff_filepair *p = q->queue[i];
6009                         if (match_filter(options, p))
6010                                 diff_q(&outq, p);
6011                         else
6012                                 diff_free_filepair(p);
6013                 }
6014         }
6015         free(q->queue);
6016         *q = outq;
6017 }
6018
6019 /* Check whether two filespecs with the same mode and size are identical */
6020 static int diff_filespec_is_identical(struct diff_filespec *one,
6021                                       struct diff_filespec *two)
6022 {
6023         if (S_ISGITLINK(one->mode))
6024                 return 0;
6025         if (diff_populate_filespec(one, 0))
6026                 return 0;
6027         if (diff_populate_filespec(two, 0))
6028                 return 0;
6029         return !memcmp(one->data, two->data, one->size);
6030 }
6031
6032 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
6033 {
6034         if (p->done_skip_stat_unmatch)
6035                 return p->skip_stat_unmatch_result;
6036
6037         p->done_skip_stat_unmatch = 1;
6038         p->skip_stat_unmatch_result = 0;
6039         /*
6040          * 1. Entries that come from stat info dirtiness
6041          *    always have both sides (iow, not create/delete),
6042          *    one side of the object name is unknown, with
6043          *    the same mode and size.  Keep the ones that
6044          *    do not match these criteria.  They have real
6045          *    differences.
6046          *
6047          * 2. At this point, the file is known to be modified,
6048          *    with the same mode and size, and the object
6049          *    name of one side is unknown.  Need to inspect
6050          *    the identical contents.
6051          */
6052         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6053             !DIFF_FILE_VALID(p->two) ||
6054             (p->one->oid_valid && p->two->oid_valid) ||
6055             (p->one->mode != p->two->mode) ||
6056             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
6057             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
6058             (p->one->size != p->two->size) ||
6059             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
6060                 p->skip_stat_unmatch_result = 1;
6061         return p->skip_stat_unmatch_result;
6062 }
6063
6064 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6065 {
6066         int i;
6067         struct diff_queue_struct *q = &diff_queued_diff;
6068         struct diff_queue_struct outq;
6069         DIFF_QUEUE_CLEAR(&outq);
6070
6071         for (i = 0; i < q->nr; i++) {
6072                 struct diff_filepair *p = q->queue[i];
6073
6074                 if (diff_filespec_check_stat_unmatch(p))
6075                         diff_q(&outq, p);
6076                 else {
6077                         /*
6078                          * The caller can subtract 1 from skip_stat_unmatch
6079                          * to determine how many paths were dirty only
6080                          * due to stat info mismatch.
6081                          */
6082                         if (!diffopt->flags.no_index)
6083                                 diffopt->skip_stat_unmatch++;
6084                         diff_free_filepair(p);
6085                 }
6086         }
6087         free(q->queue);
6088         *q = outq;
6089 }
6090
6091 static int diffnamecmp(const void *a_, const void *b_)
6092 {
6093         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6094         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6095         const char *name_a, *name_b;
6096
6097         name_a = a->one ? a->one->path : a->two->path;
6098         name_b = b->one ? b->one->path : b->two->path;
6099         return strcmp(name_a, name_b);
6100 }
6101
6102 void diffcore_fix_diff_index(struct diff_options *options)
6103 {
6104         struct diff_queue_struct *q = &diff_queued_diff;
6105         QSORT(q->queue, q->nr, diffnamecmp);
6106 }
6107
6108 void diffcore_std(struct diff_options *options)
6109 {
6110         /* NOTE please keep the following in sync with diff_tree_combined() */
6111         if (options->skip_stat_unmatch)
6112                 diffcore_skip_stat_unmatch(options);
6113         if (!options->found_follow) {
6114                 /* See try_to_follow_renames() in tree-diff.c */
6115                 if (options->break_opt != -1)
6116                         diffcore_break(options->break_opt);
6117                 if (options->detect_rename)
6118                         diffcore_rename(options);
6119                 if (options->break_opt != -1)
6120                         diffcore_merge_broken();
6121         }
6122         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6123                 diffcore_pickaxe(options);
6124         if (options->orderfile)
6125                 diffcore_order(options->orderfile);
6126         if (!options->found_follow)
6127                 /* See try_to_follow_renames() in tree-diff.c */
6128                 diff_resolve_rename_copy();
6129         diffcore_apply_filter(options);
6130
6131         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6132                 options->flags.has_changes = 1;
6133         else
6134                 options->flags.has_changes = 0;
6135
6136         options->found_follow = 0;
6137 }
6138
6139 int diff_result_code(struct diff_options *opt, int status)
6140 {
6141         int result = 0;
6142
6143         diff_warn_rename_limit("diff.renameLimit",
6144                                opt->needed_rename_limit,
6145                                opt->degraded_cc_to_c);
6146         if (!opt->flags.exit_with_status &&
6147             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6148                 return status;
6149         if (opt->flags.exit_with_status &&
6150             opt->flags.has_changes)
6151                 result |= 01;
6152         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6153             opt->flags.check_failed)
6154                 result |= 02;
6155         return result;
6156 }
6157
6158 int diff_can_quit_early(struct diff_options *opt)
6159 {
6160         return (opt->flags.quick &&
6161                 !opt->filter &&
6162                 opt->flags.has_changes);
6163 }
6164
6165 /*
6166  * Shall changes to this submodule be ignored?
6167  *
6168  * Submodule changes can be configured to be ignored separately for each path,
6169  * but that configuration can be overridden from the command line.
6170  */
6171 static int is_submodule_ignored(const char *path, struct diff_options *options)
6172 {
6173         int ignored = 0;
6174         struct diff_flags orig_flags = options->flags;
6175         if (!options->flags.override_submodule_config)
6176                 set_diffopt_flags_from_submodule_config(options, path);
6177         if (options->flags.ignore_submodules)
6178                 ignored = 1;
6179         options->flags = orig_flags;
6180         return ignored;
6181 }
6182
6183 void diff_addremove(struct diff_options *options,
6184                     int addremove, unsigned mode,
6185                     const struct object_id *oid,
6186                     int oid_valid,
6187                     const char *concatpath, unsigned dirty_submodule)
6188 {
6189         struct diff_filespec *one, *two;
6190
6191         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6192                 return;
6193
6194         /* This may look odd, but it is a preparation for
6195          * feeding "there are unchanged files which should
6196          * not produce diffs, but when you are doing copy
6197          * detection you would need them, so here they are"
6198          * entries to the diff-core.  They will be prefixed
6199          * with something like '=' or '*' (I haven't decided
6200          * which but should not make any difference).
6201          * Feeding the same new and old to diff_change()
6202          * also has the same effect.
6203          * Before the final output happens, they are pruned after
6204          * merged into rename/copy pairs as appropriate.
6205          */
6206         if (options->flags.reverse_diff)
6207                 addremove = (addremove == '+' ? '-' :
6208                              addremove == '-' ? '+' : addremove);
6209
6210         if (options->prefix &&
6211             strncmp(concatpath, options->prefix, options->prefix_length))
6212                 return;
6213
6214         one = alloc_filespec(concatpath);
6215         two = alloc_filespec(concatpath);
6216
6217         if (addremove != '+')
6218                 fill_filespec(one, oid, oid_valid, mode);
6219         if (addremove != '-') {
6220                 fill_filespec(two, oid, oid_valid, mode);
6221                 two->dirty_submodule = dirty_submodule;
6222         }
6223
6224         diff_queue(&diff_queued_diff, one, two);
6225         if (!options->flags.diff_from_contents)
6226                 options->flags.has_changes = 1;
6227 }
6228
6229 void diff_change(struct diff_options *options,
6230                  unsigned old_mode, unsigned new_mode,
6231                  const struct object_id *old_oid,
6232                  const struct object_id *new_oid,
6233                  int old_oid_valid, int new_oid_valid,
6234                  const char *concatpath,
6235                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6236 {
6237         struct diff_filespec *one, *two;
6238         struct diff_filepair *p;
6239
6240         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6241             is_submodule_ignored(concatpath, options))
6242                 return;
6243
6244         if (options->flags.reverse_diff) {
6245                 SWAP(old_mode, new_mode);
6246                 SWAP(old_oid, new_oid);
6247                 SWAP(old_oid_valid, new_oid_valid);
6248                 SWAP(old_dirty_submodule, new_dirty_submodule);
6249         }
6250
6251         if (options->prefix &&
6252             strncmp(concatpath, options->prefix, options->prefix_length))
6253                 return;
6254
6255         one = alloc_filespec(concatpath);
6256         two = alloc_filespec(concatpath);
6257         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6258         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6259         one->dirty_submodule = old_dirty_submodule;
6260         two->dirty_submodule = new_dirty_submodule;
6261         p = diff_queue(&diff_queued_diff, one, two);
6262
6263         if (options->flags.diff_from_contents)
6264                 return;
6265
6266         if (options->flags.quick && options->skip_stat_unmatch &&
6267             !diff_filespec_check_stat_unmatch(p))
6268                 return;
6269
6270         options->flags.has_changes = 1;
6271 }
6272
6273 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6274 {
6275         struct diff_filepair *pair;
6276         struct diff_filespec *one, *two;
6277
6278         if (options->prefix &&
6279             strncmp(path, options->prefix, options->prefix_length))
6280                 return NULL;
6281
6282         one = alloc_filespec(path);
6283         two = alloc_filespec(path);
6284         pair = diff_queue(&diff_queued_diff, one, two);
6285         pair->is_unmerged = 1;
6286         return pair;
6287 }
6288
6289 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6290                 size_t *outsize)
6291 {
6292         struct diff_tempfile *temp;
6293         const char *argv[3];
6294         const char **arg = argv;
6295         struct child_process child = CHILD_PROCESS_INIT;
6296         struct strbuf buf = STRBUF_INIT;
6297         int err = 0;
6298
6299         temp = prepare_temp_file(spec->path, spec);
6300         *arg++ = pgm;
6301         *arg++ = temp->name;
6302         *arg = NULL;
6303
6304         child.use_shell = 1;
6305         child.argv = argv;
6306         child.out = -1;
6307         if (start_command(&child)) {
6308                 remove_tempfile();
6309                 return NULL;
6310         }
6311
6312         if (strbuf_read(&buf, child.out, 0) < 0)
6313                 err = error("error reading from textconv command '%s'", pgm);
6314         close(child.out);
6315
6316         if (finish_command(&child) || err) {
6317                 strbuf_release(&buf);
6318                 remove_tempfile();
6319                 return NULL;
6320         }
6321         remove_tempfile();
6322
6323         return strbuf_detach(&buf, outsize);
6324 }
6325
6326 size_t fill_textconv(struct userdiff_driver *driver,
6327                      struct diff_filespec *df,
6328                      char **outbuf)
6329 {
6330         size_t size;
6331
6332         if (!driver) {
6333                 if (!DIFF_FILE_VALID(df)) {
6334                         *outbuf = "";
6335                         return 0;
6336                 }
6337                 if (diff_populate_filespec(df, 0))
6338                         die("unable to read files to diff");
6339                 *outbuf = df->data;
6340                 return df->size;
6341         }
6342
6343         if (!driver->textconv)
6344                 BUG("fill_textconv called with non-textconv driver");
6345
6346         if (driver->textconv_cache && df->oid_valid) {
6347                 *outbuf = notes_cache_get(driver->textconv_cache,
6348                                           &df->oid,
6349                                           &size);
6350                 if (*outbuf)
6351                         return size;
6352         }
6353
6354         *outbuf = run_textconv(driver->textconv, df, &size);
6355         if (!*outbuf)
6356                 die("unable to read files to diff");
6357
6358         if (driver->textconv_cache && df->oid_valid) {
6359                 /* ignore errors, as we might be in a readonly repository */
6360                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6361                                 size);
6362                 /*
6363                  * we could save up changes and flush them all at the end,
6364                  * but we would need an extra call after all diffing is done.
6365                  * Since generating a cache entry is the slow path anyway,
6366                  * this extra overhead probably isn't a big deal.
6367                  */
6368                 notes_cache_write(driver->textconv_cache);
6369         }
6370
6371         return size;
6372 }
6373
6374 int textconv_object(const char *path,
6375                     unsigned mode,
6376                     const struct object_id *oid,
6377                     int oid_valid,
6378                     char **buf,
6379                     unsigned long *buf_size)
6380 {
6381         struct diff_filespec *df;
6382         struct userdiff_driver *textconv;
6383
6384         df = alloc_filespec(path);
6385         fill_filespec(df, oid, oid_valid, mode);
6386         textconv = get_textconv(df);
6387         if (!textconv) {
6388                 free_filespec(df);
6389                 return 0;
6390         }
6391
6392         *buf_size = fill_textconv(textconv, df, buf);
6393         free_filespec(df);
6394         return 1;
6395 }
6396
6397 void setup_diff_pager(struct diff_options *opt)
6398 {
6399         /*
6400          * If the user asked for our exit code, then either they want --quiet
6401          * or --exit-code. We should definitely not bother with a pager in the
6402          * former case, as we will generate no output. Since we still properly
6403          * report our exit code even when a pager is run, we _could_ run a
6404          * pager with --exit-code. But since we have not done so historically,
6405          * and because it is easy to find people oneline advising "git diff
6406          * --exit-code" in hooks and other scripts, we do not do so.
6407          */
6408         if (!opt->flags.exit_with_status &&
6409             check_pager_config("diff") != 0)
6410                 setup_pager();
6411 }