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