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