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 "parse-options.h"
  27 #include "help.h"
  28
  29 #ifdef NO_FAST_WORKING_DIRECTORY
  30 #define FAST_WORKING_DIRECTORY 0
  31 #else
  32 #define FAST_WORKING_DIRECTORY 1
  33 #endif
  34
  35 static int diff_detect_rename_default;
  36 static int diff_indent_heuristic = 1;
  37 static int diff_rename_limit_default = 400;
  38 static int diff_suppress_blank_empty;
  39 static int diff_use_color_default = -1;
  40 static int diff_color_moved_default;
  41 static int diff_color_moved_ws_default;
  42 static int diff_context_default = 3;
  43 static int diff_interhunk_context_default;
  44 static const char *diff_word_regex_cfg;
  45 static const char *external_diff_cmd_cfg;
  46 static const char *diff_order_file_cfg;
  47 int diff_auto_refresh_index = 1;
  48 static int diff_mnemonic_prefix;
  49 static int diff_no_prefix;
  50 static int diff_stat_graph_width;
  51 static int diff_dirstat_permille_default = 30;
  52 static struct diff_options default_diff_options;
  53 static long diff_algorithm;
  54 static unsigned ws_error_highlight_default = WSEH_NEW;
  55
  56 static char diff_colors[][COLOR_MAXLEN] = {
  57         GIT_COLOR_RESET,
  58         GIT_COLOR_NORMAL,       /* CONTEXT */
  59         GIT_COLOR_BOLD,         /* METAINFO */
  60         GIT_COLOR_CYAN,         /* FRAGINFO */
  61         GIT_COLOR_RED,          /* OLD */
  62         GIT_COLOR_GREEN,        /* NEW */
  63         GIT_COLOR_YELLOW,       /* COMMIT */
  64         GIT_COLOR_BG_RED,       /* WHITESPACE */
  65         GIT_COLOR_NORMAL,       /* FUNCINFO */
  66         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  67         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  70         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  71         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  72         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  73         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  74         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  75         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  76         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  77         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  78         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  79         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  80 };
  81
  82 static const char *color_diff_slots[] = {
  83         [DIFF_CONTEXT]                = "context",
  84         [DIFF_METAINFO]               = "meta",
  85         [DIFF_FRAGINFO]               = "frag",
  86         [DIFF_FILE_OLD]               = "old",
  87         [DIFF_FILE_NEW]               = "new",
  88         [DIFF_COMMIT]                 = "commit",
  89         [DIFF_WHITESPACE]             = "whitespace",
  90         [DIFF_FUNCINFO]               = "func",
  91         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  92         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  93         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  94         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  95         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  96         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  97         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  98         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  99         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 100         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 101         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 102         [DIFF_CONTEXT_BOLD]           = "contextBold",
 103         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 104         [DIFF_FILE_NEW_BOLD]          = "newBold",
 105 };
 106
 107 static NORETURN void die_want_option(const char *option_name)
 108 {
 109         die(_("option '%s' requires a value"), option_name);
 110 }
 111
 112 define_list_config_array_extra(color_diff_slots, {"plain"});
 113
 114 static int parse_diff_color_slot(const char *var)
 115 {
 116         if (!strcasecmp(var, "plain"))
 117                 return DIFF_CONTEXT;
 118         return LOOKUP_CONFIG(color_diff_slots, var);
 119 }
 120
 121 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 122                                 struct strbuf *errmsg)
 123 {
 124         char *params_copy = xstrdup(params_string);
 125         struct string_list params = STRING_LIST_INIT_NODUP;
 126         int ret = 0;
 127         int i;
 128
 129         if (*params_copy)
 130                 string_list_split_in_place(&params, params_copy, ',', -1);
 131         for (i = 0; i < params.nr; i++) {
 132                 const char *p = params.items[i].string;
 133                 if (!strcmp(p, "changes")) {
 134                         options->flags.dirstat_by_line = 0;
 135                         options->flags.dirstat_by_file = 0;
 136                 } else if (!strcmp(p, "lines")) {
 137                         options->flags.dirstat_by_line = 1;
 138                         options->flags.dirstat_by_file = 0;
 139                 } else if (!strcmp(p, "files")) {
 140                         options->flags.dirstat_by_line = 0;
 141                         options->flags.dirstat_by_file = 1;
 142                 } else if (!strcmp(p, "noncumulative")) {
 143                         options->flags.dirstat_cumulative = 0;
 144                 } else if (!strcmp(p, "cumulative")) {
 145                         options->flags.dirstat_cumulative = 1;
 146                 } else if (isdigit(*p)) {
 147                         char *end;
 148                         int permille = strtoul(p, &end, 10) * 10;
 149                         if (*end == '.' && isdigit(*++end)) {
 150                                 /* only use first digit */
 151                                 permille += *end - '0';
 152                                 /* .. and ignore any further digits */
 153                                 while (isdigit(*++end))
 154                                         ; /* nothing */
 155                         }
 156                         if (!*end)
 157                                 options->dirstat_permille = permille;
 158                         else {
 159                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 160                                             p);
 161                                 ret++;
 162                         }
 163                 } else {
 164                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 165                         ret++;
 166                 }
 167
 168         }
 169         string_list_clear(&params, 0);
 170         free(params_copy);
 171         return ret;
 172 }
 173
 174 static int parse_submodule_params(struct diff_options *options, const char *value)
 175 {
 176         if (!strcmp(value, "log"))
 177                 options->submodule_format = DIFF_SUBMODULE_LOG;
 178         else if (!strcmp(value, "short"))
 179                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 180         else if (!strcmp(value, "diff"))
 181                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 182         /*
 183          * Please update $__git_diff_submodule_formats in
 184          * git-completion.bash when you add new formats.
 185          */
 186         else
 187                 return -1;
 188         return 0;
 189 }
 190
 191 int git_config_rename(const char *var, const char *value)
 192 {
 193         if (!value)
 194                 return DIFF_DETECT_RENAME;
 195         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 196                 return  DIFF_DETECT_COPY;
 197         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 198 }
 199
 200 long parse_algorithm_value(const char *value)
 201 {
 202         if (!value)
 203                 return -1;
 204         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 205                 return 0;
 206         else if (!strcasecmp(value, "minimal"))
 207                 return XDF_NEED_MINIMAL;
 208         else if (!strcasecmp(value, "patience"))
 209                 return XDF_PATIENCE_DIFF;
 210         else if (!strcasecmp(value, "histogram"))
 211                 return XDF_HISTOGRAM_DIFF;
 212         /*
 213          * Please update $__git_diff_algorithms in git-completion.bash
 214          * when you add new algorithms.
 215          */
 216         return -1;
 217 }
 218
 219 static int parse_one_token(const char **arg, const char *token)
 220 {
 221         const char *rest;
 222         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 223                 *arg = rest;
 224                 return 1;
 225         }
 226         return 0;
 227 }
 228
 229 static int parse_ws_error_highlight(const char *arg)
 230 {
 231         const char *orig_arg = arg;
 232         unsigned val = 0;
 233
 234         while (*arg) {
 235                 if (parse_one_token(&arg, "none"))
 236                         val = 0;
 237                 else if (parse_one_token(&arg, "default"))
 238                         val = WSEH_NEW;
 239                 else if (parse_one_token(&arg, "all"))
 240                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 241                 else if (parse_one_token(&arg, "new"))
 242                         val |= WSEH_NEW;
 243                 else if (parse_one_token(&arg, "old"))
 244                         val |= WSEH_OLD;
 245                 else if (parse_one_token(&arg, "context"))
 246                         val |= WSEH_CONTEXT;
 247                 else {
 248                         return -1 - (int)(arg - orig_arg);
 249                 }
 250                 if (*arg)
 251                         arg++;
 252         }
 253         return val;
 254 }
 255
 256 /*
 257  * These are to give UI layer defaults.
 258  * The core-level commands such as git-diff-files should
 259  * never be affected by the setting of diff.renames
 260  * the user happens to have in the configuration file.
 261  */
 262 void init_diff_ui_defaults(void)
 263 {
 264         diff_detect_rename_default = DIFF_DETECT_RENAME;
 265 }
 266
 267 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 268 {
 269         if (!strcmp(var, "diff.indentheuristic"))
 270                 diff_indent_heuristic = git_config_bool(var, value);
 271         return 0;
 272 }
 273
 274 static int parse_color_moved(const char *arg)
 275 {
 276         switch (git_parse_maybe_bool(arg)) {
 277         case 0:
 278                 return COLOR_MOVED_NO;
 279         case 1:
 280                 return COLOR_MOVED_DEFAULT;
 281         default:
 282                 break;
 283         }
 284
 285         if (!strcmp(arg, "no"))
 286                 return COLOR_MOVED_NO;
 287         else if (!strcmp(arg, "plain"))
 288                 return COLOR_MOVED_PLAIN;
 289         else if (!strcmp(arg, "blocks"))
 290                 return COLOR_MOVED_BLOCKS;
 291         else if (!strcmp(arg, "zebra"))
 292                 return COLOR_MOVED_ZEBRA;
 293         else if (!strcmp(arg, "default"))
 294                 return COLOR_MOVED_DEFAULT;
 295         else if (!strcmp(arg, "dimmed-zebra"))
 296                 return COLOR_MOVED_ZEBRA_DIM;
 297         else if (!strcmp(arg, "dimmed_zebra"))
 298                 return COLOR_MOVED_ZEBRA_DIM;
 299         else
 300                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 301 }
 302
 303 static unsigned parse_color_moved_ws(const char *arg)
 304 {
 305         int ret = 0;
 306         struct string_list l = STRING_LIST_INIT_DUP;
 307         struct string_list_item *i;
 308
 309         string_list_split(&l, arg, ',', -1);
 310
 311         for_each_string_list_item(i, &l) {
 312                 struct strbuf sb = STRBUF_INIT;
 313                 strbuf_addstr(&sb, i->string);
 314                 strbuf_trim(&sb);
 315
 316                 if (!strcmp(sb.buf, "no"))
 317                         ret = 0;
 318                 else if (!strcmp(sb.buf, "ignore-space-change"))
 319                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 320                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 321                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 322                 else if (!strcmp(sb.buf, "ignore-all-space"))
 323                         ret |= XDF_IGNORE_WHITESPACE;
 324                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 325                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 326                 else {
 327                         ret |= COLOR_MOVED_WS_ERROR;
 328                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 329                 }
 330
 331                 strbuf_release(&sb);
 332         }
 333
 334         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 335             (ret & XDF_WHITESPACE_FLAGS)) {
 336                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 337                 ret |= COLOR_MOVED_WS_ERROR;
 338         }
 339
 340         string_list_clear(&l, 0);
 341
 342         return ret;
 343 }
 344
 345 int git_diff_ui_config(const char *var, const char *value, void *cb)
 346 {
 347         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 348                 diff_use_color_default = git_config_colorbool(var, value);
 349                 return 0;
 350         }
 351         if (!strcmp(var, "diff.colormoved")) {
 352                 int cm = parse_color_moved(value);
 353                 if (cm < 0)
 354                         return -1;
 355                 diff_color_moved_default = cm;
 356                 return 0;
 357         }
 358         if (!strcmp(var, "diff.colormovedws")) {
 359                 unsigned cm = parse_color_moved_ws(value);
 360                 if (cm & COLOR_MOVED_WS_ERROR)
 361                         return -1;
 362                 diff_color_moved_ws_default = cm;
 363                 return 0;
 364         }
 365         if (!strcmp(var, "diff.context")) {
 366                 diff_context_default = git_config_int(var, value);
 367                 if (diff_context_default < 0)
 368                         return -1;
 369                 return 0;
 370         }
 371         if (!strcmp(var, "diff.interhunkcontext")) {
 372                 diff_interhunk_context_default = git_config_int(var, value);
 373                 if (diff_interhunk_context_default < 0)
 374                         return -1;
 375                 return 0;
 376         }
 377         if (!strcmp(var, "diff.renames")) {
 378                 diff_detect_rename_default = git_config_rename(var, value);
 379                 return 0;
 380         }
 381         if (!strcmp(var, "diff.autorefreshindex")) {
 382                 diff_auto_refresh_index = git_config_bool(var, value);
 383                 return 0;
 384         }
 385         if (!strcmp(var, "diff.mnemonicprefix")) {
 386                 diff_mnemonic_prefix = git_config_bool(var, value);
 387                 return 0;
 388         }
 389         if (!strcmp(var, "diff.noprefix")) {
 390                 diff_no_prefix = git_config_bool(var, value);
 391                 return 0;
 392         }
 393         if (!strcmp(var, "diff.statgraphwidth")) {
 394                 diff_stat_graph_width = git_config_int(var, value);
 395                 return 0;
 396         }
 397         if (!strcmp(var, "diff.external"))
 398                 return git_config_string(&external_diff_cmd_cfg, var, value);
 399         if (!strcmp(var, "diff.wordregex"))
 400                 return git_config_string(&diff_word_regex_cfg, var, value);
 401         if (!strcmp(var, "diff.orderfile"))
 402                 return git_config_pathname(&diff_order_file_cfg, var, value);
 403
 404         if (!strcmp(var, "diff.ignoresubmodules"))
 405                 handle_ignore_submodules_arg(&default_diff_options, value);
 406
 407         if (!strcmp(var, "diff.submodule")) {
 408                 if (parse_submodule_params(&default_diff_options, value))
 409                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 410                                 value);
 411                 return 0;
 412         }
 413
 414         if (!strcmp(var, "diff.algorithm")) {
 415                 diff_algorithm = parse_algorithm_value(value);
 416                 if (diff_algorithm < 0)
 417                         return -1;
 418                 return 0;
 419         }
 420
 421         if (!strcmp(var, "diff.wserrorhighlight")) {
 422                 int val = parse_ws_error_highlight(value);
 423                 if (val < 0)
 424                         return -1;
 425                 ws_error_highlight_default = val;
 426                 return 0;
 427         }
 428
 429         if (git_color_config(var, value, cb) < 0)
 430                 return -1;
 431
 432         return git_diff_basic_config(var, value, cb);
 433 }
 434
 435 int git_diff_basic_config(const char *var, const char *value, void *cb)
 436 {
 437         const char *name;
 438
 439         if (!strcmp(var, "diff.renamelimit")) {
 440                 diff_rename_limit_default = git_config_int(var, value);
 441                 return 0;
 442         }
 443
 444         if (userdiff_config(var, value) < 0)
 445                 return -1;
 446
 447         if (skip_prefix(var, "diff.color.", &name) ||
 448             skip_prefix(var, "color.diff.", &name)) {
 449                 int slot = parse_diff_color_slot(name);
 450                 if (slot < 0)
 451                         return 0;
 452                 if (!value)
 453                         return config_error_nonbool(var);
 454                 return color_parse(value, diff_colors[slot]);
 455         }
 456
 457         /* like GNU diff's --suppress-blank-empty option  */
 458         if (!strcmp(var, "diff.suppressblankempty") ||
 459                         /* for backwards compatibility */
 460                         !strcmp(var, "diff.suppress-blank-empty")) {
 461                 diff_suppress_blank_empty = git_config_bool(var, value);
 462                 return 0;
 463         }
 464
 465         if (!strcmp(var, "diff.dirstat")) {
 466                 struct strbuf errmsg = STRBUF_INIT;
 467                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 468                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 469                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 470                                 errmsg.buf);
 471                 strbuf_release(&errmsg);
 472                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 473                 return 0;
 474         }
 475
 476         if (git_diff_heuristic_config(var, value, cb) < 0)
 477                 return -1;
 478
 479         return git_default_config(var, value, cb);
 480 }
 481
 482 static char *quote_two(const char *one, const char *two)
 483 {
 484         int need_one = quote_c_style(one, NULL, NULL, 1);
 485         int need_two = quote_c_style(two, NULL, NULL, 1);
 486         struct strbuf res = STRBUF_INIT;
 487
 488         if (need_one + need_two) {
 489                 strbuf_addch(&res, '"');
 490                 quote_c_style(one, &res, NULL, 1);
 491                 quote_c_style(two, &res, NULL, 1);
 492                 strbuf_addch(&res, '"');
 493         } else {
 494                 strbuf_addstr(&res, one);
 495                 strbuf_addstr(&res, two);
 496         }
 497         return strbuf_detach(&res, NULL);
 498 }
 499
 500 static const char *external_diff(void)
 501 {
 502         static const char *external_diff_cmd = NULL;
 503         static int done_preparing = 0;
 504
 505         if (done_preparing)
 506                 return external_diff_cmd;
 507         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 508         if (!external_diff_cmd)
 509                 external_diff_cmd = external_diff_cmd_cfg;
 510         done_preparing = 1;
 511         return external_diff_cmd;
 512 }
 513
 514 /*
 515  * Keep track of files used for diffing. Sometimes such an entry
 516  * refers to a temporary file, sometimes to an existing file, and
 517  * sometimes to "/dev/null".
 518  */
 519 static struct diff_tempfile {
 520         /*
 521          * filename external diff should read from, or NULL if this
 522          * entry is currently not in use:
 523          */
 524         const char *name;
 525
 526         char hex[GIT_MAX_HEXSZ + 1];
 527         char mode[10];
 528
 529         /*
 530          * If this diff_tempfile instance refers to a temporary file,
 531          * this tempfile object is used to manage its lifetime.
 532          */
 533         struct tempfile *tempfile;
 534 } diff_temp[2];
 535
 536 struct emit_callback {
 537         int color_diff;
 538         unsigned ws_rule;
 539         int blank_at_eof_in_preimage;
 540         int blank_at_eof_in_postimage;
 541         int lno_in_preimage;
 542         int lno_in_postimage;
 543         const char **label_path;
 544         struct diff_words_data *diff_words;
 545         struct diff_options *opt;
 546         struct strbuf *header;
 547 };
 548
 549 static int count_lines(const char *data, int size)
 550 {
 551         int count, ch, completely_empty = 1, nl_just_seen = 0;
 552         count = 0;
 553         while (0 < size--) {
 554                 ch = *data++;
 555                 if (ch == '\n') {
 556                         count++;
 557                         nl_just_seen = 1;
 558                         completely_empty = 0;
 559                 }
 560                 else {
 561                         nl_just_seen = 0;
 562                         completely_empty = 0;
 563                 }
 564         }
 565         if (completely_empty)
 566                 return 0;
 567         if (!nl_just_seen)
 568                 count++; /* no trailing newline */
 569         return count;
 570 }
 571
 572 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 573                        struct diff_filespec *one)
 574 {
 575         if (!DIFF_FILE_VALID(one)) {
 576                 mf->ptr = (char *)""; /* does not matter */
 577                 mf->size = 0;
 578                 return 0;
 579         }
 580         else if (diff_populate_filespec(r, one, 0))
 581                 return -1;
 582
 583         mf->ptr = one->data;
 584         mf->size = one->size;
 585         return 0;
 586 }
 587
 588 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 589 static unsigned long diff_filespec_size(struct repository *r,
 590                                         struct diff_filespec *one)
 591 {
 592         if (!DIFF_FILE_VALID(one))
 593                 return 0;
 594         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 595         return one->size;
 596 }
 597
 598 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 599 {
 600         char *ptr = mf->ptr;
 601         long size = mf->size;
 602         int cnt = 0;
 603
 604         if (!size)
 605                 return cnt;
 606         ptr += size - 1; /* pointing at the very end */
 607         if (*ptr != '\n')
 608                 ; /* incomplete line */
 609         else
 610                 ptr--; /* skip the last LF */
 611         while (mf->ptr < ptr) {
 612                 char *prev_eol;
 613                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 614                         if (*prev_eol == '\n')
 615                                 break;
 616                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 617                         break;
 618                 cnt++;
 619                 ptr = prev_eol - 1;
 620         }
 621         return cnt;
 622 }
 623
 624 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 625                                struct emit_callback *ecbdata)
 626 {
 627         int l1, l2, at;
 628         unsigned ws_rule = ecbdata->ws_rule;
 629         l1 = count_trailing_blank(mf1, ws_rule);
 630         l2 = count_trailing_blank(mf2, ws_rule);
 631         if (l2 <= l1) {
 632                 ecbdata->blank_at_eof_in_preimage = 0;
 633                 ecbdata->blank_at_eof_in_postimage = 0;
 634                 return;
 635         }
 636         at = count_lines(mf1->ptr, mf1->size);
 637         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 638
 639         at = count_lines(mf2->ptr, mf2->size);
 640         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 641 }
 642
 643 static void emit_line_0(struct diff_options *o,
 644                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 645                         int first, const char *line, int len)
 646 {
 647         int has_trailing_newline, has_trailing_carriage_return;
 648         int needs_reset = 0; /* at the end of the line */
 649         FILE *file = o->file;
 650
 651         fputs(diff_line_prefix(o), file);
 652
 653         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 654         if (has_trailing_newline)
 655                 len--;
 656
 657         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 658         if (has_trailing_carriage_return)
 659                 len--;
 660
 661         if (!len && !first)
 662                 goto end_of_line;
 663
 664         if (reverse && want_color(o->use_color)) {
 665                 fputs(GIT_COLOR_REVERSE, file);
 666                 needs_reset = 1;
 667         }
 668
 669         if (set_sign) {
 670                 fputs(set_sign, file);
 671                 needs_reset = 1;
 672         }
 673
 674         if (first)
 675                 fputc(first, file);
 676
 677         if (!len)
 678                 goto end_of_line;
 679
 680         if (set) {
 681                 if (set_sign && set != set_sign)
 682                         fputs(reset, file);
 683                 fputs(set, file);
 684                 needs_reset = 1;
 685         }
 686         fwrite(line, len, 1, file);
 687         needs_reset = 1; /* 'line' may contain color codes. */
 688
 689 end_of_line:
 690         if (needs_reset)
 691                 fputs(reset, file);
 692         if (has_trailing_carriage_return)
 693                 fputc('\r', file);
 694         if (has_trailing_newline)
 695                 fputc('\n', file);
 696 }
 697
 698 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 699                       const char *line, int len)
 700 {
 701         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 702 }
 703
 704 enum diff_symbol {
 705         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 706         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 707         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 708         DIFF_SYMBOL_BINARY_DIFF_BODY,
 709         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 710         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 711         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 712         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 713         DIFF_SYMBOL_STATS_LINE,
 714         DIFF_SYMBOL_WORD_DIFF,
 715         DIFF_SYMBOL_STAT_SEP,
 716         DIFF_SYMBOL_SUMMARY,
 717         DIFF_SYMBOL_SUBMODULE_ADD,
 718         DIFF_SYMBOL_SUBMODULE_DEL,
 719         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 720         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 721         DIFF_SYMBOL_SUBMODULE_HEADER,
 722         DIFF_SYMBOL_SUBMODULE_ERROR,
 723         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 724         DIFF_SYMBOL_REWRITE_DIFF,
 725         DIFF_SYMBOL_BINARY_FILES,
 726         DIFF_SYMBOL_HEADER,
 727         DIFF_SYMBOL_FILEPAIR_PLUS,
 728         DIFF_SYMBOL_FILEPAIR_MINUS,
 729         DIFF_SYMBOL_WORDS_PORCELAIN,
 730         DIFF_SYMBOL_WORDS,
 731         DIFF_SYMBOL_CONTEXT,
 732         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 733         DIFF_SYMBOL_PLUS,
 734         DIFF_SYMBOL_MINUS,
 735         DIFF_SYMBOL_NO_LF_EOF,
 736         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 737         DIFF_SYMBOL_CONTEXT_MARKER,
 738         DIFF_SYMBOL_SEPARATOR
 739 };
 740 /*
 741  * Flags for content lines:
 742  * 0..12 are whitespace rules
 743  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 744  * 16 is marking if the line is blank at EOF
 745  */
 746 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 747 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 748 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 749 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 750 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 751
 752 /*
 753  * This struct is used when we need to buffer the output of the diff output.
 754  *
 755  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 756  * into the pre/post image file. This pointer could be a union with the
 757  * line pointer. By storing an offset into the file instead of the literal line,
 758  * we can decrease the memory footprint for the buffered output. At first we
 759  * may want to only have indirection for the content lines, but we could also
 760  * enhance the state for emitting prefabricated lines, e.g. the similarity
 761  * score line or hunk/file headers would only need to store a number or path
 762  * and then the output can be constructed later on depending on state.
 763  */
 764 struct emitted_diff_symbol {
 765         const char *line;
 766         int len;
 767         int flags;
 768         int indent_off;   /* Offset to first non-whitespace character */
 769         int indent_width; /* The visual width of the indentation */
 770         enum diff_symbol s;
 771 };
 772 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 773
 774 struct emitted_diff_symbols {
 775         struct emitted_diff_symbol *buf;
 776         int nr, alloc;
 777 };
 778 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 779
 780 static void append_emitted_diff_symbol(struct diff_options *o,
 781                                        struct emitted_diff_symbol *e)
 782 {
 783         struct emitted_diff_symbol *f;
 784
 785         ALLOC_GROW(o->emitted_symbols->buf,
 786                    o->emitted_symbols->nr + 1,
 787                    o->emitted_symbols->alloc);
 788         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 789
 790         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 791         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 792 }
 793
 794 struct moved_entry {
 795         struct hashmap_entry ent;
 796         const struct emitted_diff_symbol *es;
 797         struct moved_entry *next_line;
 798 };
 799
 800 struct moved_block {
 801         struct moved_entry *match;
 802         int wsd; /* The whitespace delta of this block */
 803 };
 804
 805 static void moved_block_clear(struct moved_block *b)
 806 {
 807         memset(b, 0, sizeof(*b));
 808 }
 809
 810 #define INDENT_BLANKLINE INT_MIN
 811
 812 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 813 {
 814         unsigned int off = 0, i;
 815         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 816         const char *s = es->line;
 817         const int len = es->len;
 818
 819         /* skip any \v \f \r at start of indentation */
 820         while (s[off] == '\f' || s[off] == '\v' ||
 821                (s[off] == '\r' && off < len - 1))
 822                 off++;
 823
 824         /* calculate the visual width of indentation */
 825         while(1) {
 826                 if (s[off] == ' ') {
 827                         width++;
 828                         off++;
 829                 } else if (s[off] == '\t') {
 830                         width += tab_width - (width % tab_width);
 831                         while (s[++off] == '\t')
 832                                 width += tab_width;
 833                 } else {
 834                         break;
 835                 }
 836         }
 837
 838         /* check if this line is blank */
 839         for (i = off; i < len; i++)
 840                 if (!isspace(s[i]))
 841                     break;
 842
 843         if (i == len) {
 844                 es->indent_width = INDENT_BLANKLINE;
 845                 es->indent_off = len;
 846         } else {
 847                 es->indent_off = off;
 848                 es->indent_width = width;
 849         }
 850 }
 851
 852 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 853                             const struct emitted_diff_symbol *b,
 854                             int *out)
 855 {
 856         int a_len = a->len,
 857             b_len = b->len,
 858             a_off = a->indent_off,
 859             a_width = a->indent_width,
 860             b_off = b->indent_off,
 861             b_width = b->indent_width;
 862         int delta;
 863
 864         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 865                 *out = INDENT_BLANKLINE;
 866                 return 1;
 867         }
 868
 869         if (a->s == DIFF_SYMBOL_PLUS)
 870                 delta = a_width - b_width;
 871         else
 872                 delta = b_width - a_width;
 873
 874         if (a_len - a_off != b_len - b_off ||
 875             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 876                 return 0;
 877
 878         *out = delta;
 879
 880         return 1;
 881 }
 882
 883 static int cmp_in_block_with_wsd(const struct diff_options *o,
 884                                  const struct moved_entry *cur,
 885                                  const struct moved_entry *match,
 886                                  struct moved_block *pmb,
 887                                  int n)
 888 {
 889         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 890         int al = cur->es->len, bl = match->es->len, cl = l->len;
 891         const char *a = cur->es->line,
 892                    *b = match->es->line,
 893                    *c = l->line;
 894         int a_off = cur->es->indent_off,
 895             a_width = cur->es->indent_width,
 896             c_off = l->indent_off,
 897             c_width = l->indent_width;
 898         int delta;
 899
 900         /*
 901          * We need to check if 'cur' is equal to 'match'.  As those
 902          * are from the same (+/-) side, we do not need to adjust for
 903          * indent changes. However these were found using fuzzy
 904          * matching so we do have to check if they are equal. Here we
 905          * just check the lengths. We delay calling memcmp() to check
 906          * the contents until later as if the length comparison for a
 907          * and c fails we can avoid the call all together.
 908          */
 909         if (al != bl)
 910                 return 1;
 911
 912         /* If 'l' and 'cur' are both blank then they match. */
 913         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 914                 return 0;
 915
 916         /*
 917          * The indent changes of the block are known and stored in pmb->wsd;
 918          * however we need to check if the indent changes of the current line
 919          * match those of the current block and that the text of 'l' and 'cur'
 920          * after the indentation match.
 921          */
 922         if (cur->es->s == DIFF_SYMBOL_PLUS)
 923                 delta = a_width - c_width;
 924         else
 925                 delta = c_width - a_width;
 926
 927         /*
 928          * If the previous lines of this block were all blank then set its
 929          * whitespace delta.
 930          */
 931         if (pmb->wsd == INDENT_BLANKLINE)
 932                 pmb->wsd = delta;
 933
 934         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 935                  !memcmp(a, b, al) && !
 936                  memcmp(a + a_off, c + c_off, al - a_off));
 937 }
 938
 939 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 940                            const void *entry,
 941                            const void *entry_or_key,
 942                            const void *keydata)
 943 {
 944         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 945         const struct moved_entry *a = entry;
 946         const struct moved_entry *b = entry_or_key;
 947         unsigned flags = diffopt->color_moved_ws_handling
 948                          & XDF_WHITESPACE_FLAGS;
 949
 950         if (diffopt->color_moved_ws_handling &
 951             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 952                 /*
 953                  * As there is not specific white space config given,
 954                  * we'd need to check for a new block, so ignore all
 955                  * white space. The setup of the white space
 956                  * configuration for the next block is done else where
 957                  */
 958                 flags |= XDF_IGNORE_WHITESPACE;
 959
 960         return !xdiff_compare_lines(a->es->line, a->es->len,
 961                                     b->es->line, b->es->len,
 962                                     flags);
 963 }
 964
 965 static struct moved_entry *prepare_entry(struct diff_options *o,
 966                                          int line_no)
 967 {
 968         struct moved_entry *ret = xmalloc(sizeof(*ret));
 969         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 970         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 971
 972         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 973         ret->es = l;
 974         ret->next_line = NULL;
 975
 976         return ret;
 977 }
 978
 979 static void add_lines_to_move_detection(struct diff_options *o,
 980                                         struct hashmap *add_lines,
 981                                         struct hashmap *del_lines)
 982 {
 983         struct moved_entry *prev_line = NULL;
 984
 985         int n;
 986         for (n = 0; n < o->emitted_symbols->nr; n++) {
 987                 struct hashmap *hm;
 988                 struct moved_entry *key;
 989
 990                 switch (o->emitted_symbols->buf[n].s) {
 991                 case DIFF_SYMBOL_PLUS:
 992                         hm = add_lines;
 993                         break;
 994                 case DIFF_SYMBOL_MINUS:
 995                         hm = del_lines;
 996                         break;
 997                 default:
 998                         prev_line = NULL;
 999                         continue;
1000                 }
1001
1002                 if (o->color_moved_ws_handling &
1003                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1004                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1005                 key = prepare_entry(o, n);
1006                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1007                         prev_line->next_line = key;
1008
1009                 hashmap_add(hm, key);
1010                 prev_line = key;
1011         }
1012 }
1013
1014 static void pmb_advance_or_null(struct diff_options *o,
1015                                 struct moved_entry *match,
1016                                 struct hashmap *hm,
1017                                 struct moved_block *pmb,
1018                                 int pmb_nr)
1019 {
1020         int i;
1021         for (i = 0; i < pmb_nr; i++) {
1022                 struct moved_entry *prev = pmb[i].match;
1023                 struct moved_entry *cur = (prev && prev->next_line) ?
1024                                 prev->next_line : NULL;
1025                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1026                         pmb[i].match = cur;
1027                 } else {
1028                         pmb[i].match = NULL;
1029                 }
1030         }
1031 }
1032
1033 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1034                                             struct moved_entry *match,
1035                                             struct hashmap *hm,
1036                                             struct moved_block *pmb,
1037                                             int pmb_nr, int n)
1038 {
1039         int i;
1040         char *got_match = xcalloc(1, pmb_nr);
1041
1042         for (; match; match = hashmap_get_next(hm, match)) {
1043                 for (i = 0; i < pmb_nr; i++) {
1044                         struct moved_entry *prev = pmb[i].match;
1045                         struct moved_entry *cur = (prev && prev->next_line) ?
1046                                         prev->next_line : NULL;
1047                         if (!cur)
1048                                 continue;
1049                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1050                                 got_match[i] |= 1;
1051                 }
1052         }
1053
1054         for (i = 0; i < pmb_nr; i++) {
1055                 if (got_match[i]) {
1056                         /* Advance to the next line */
1057                         pmb[i].match = pmb[i].match->next_line;
1058                 } else {
1059                         moved_block_clear(&pmb[i]);
1060                 }
1061         }
1062
1063         free(got_match);
1064 }
1065
1066 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1067                                          int pmb_nr)
1068 {
1069         int lp, rp;
1070
1071         /* Shrink the set of potential block to the remaining running */
1072         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1073                 while (lp < pmb_nr && pmb[lp].match)
1074                         lp++;
1075                 /* lp points at the first NULL now */
1076
1077                 while (rp > -1 && !pmb[rp].match)
1078                         rp--;
1079                 /* rp points at the last non-NULL */
1080
1081                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1082                         pmb[lp] = pmb[rp];
1083                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1084                         rp--;
1085                         lp++;
1086                 }
1087         }
1088
1089         /* Remember the number of running sets */
1090         return rp + 1;
1091 }
1092
1093 /*
1094  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1095  *
1096  * Otherwise, if the last block has fewer alphanumeric characters than
1097  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1098  * that block.
1099  *
1100  * The last block consists of the (n - block_length)'th line up to but not
1101  * including the nth line.
1102  *
1103  * Returns 0 if the last block is empty or is unset by this function, non zero
1104  * otherwise.
1105  *
1106  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1107  * Think of a way to unify them.
1108  */
1109 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1110 {
1111         int i, alnum_count = 0;
1112         if (o->color_moved == COLOR_MOVED_PLAIN)
1113                 return block_length;
1114         for (i = 1; i < block_length + 1; i++) {
1115                 const char *c = o->emitted_symbols->buf[n - i].line;
1116                 for (; *c; c++) {
1117                         if (!isalnum(*c))
1118                                 continue;
1119                         alnum_count++;
1120                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1121                                 return 1;
1122                 }
1123         }
1124         for (i = 1; i < block_length + 1; i++)
1125                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1126         return 0;
1127 }
1128
1129 /* Find blocks of moved code, delegate actual coloring decision to helper */
1130 static void mark_color_as_moved(struct diff_options *o,
1131                                 struct hashmap *add_lines,
1132                                 struct hashmap *del_lines)
1133 {
1134         struct moved_block *pmb = NULL; /* potentially moved blocks */
1135         int pmb_nr = 0, pmb_alloc = 0;
1136         int n, flipped_block = 0, block_length = 0;
1137
1138
1139         for (n = 0; n < o->emitted_symbols->nr; n++) {
1140                 struct hashmap *hm = NULL;
1141                 struct moved_entry *key;
1142                 struct moved_entry *match = NULL;
1143                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1144                 enum diff_symbol last_symbol = 0;
1145
1146                 switch (l->s) {
1147                 case DIFF_SYMBOL_PLUS:
1148                         hm = del_lines;
1149                         key = prepare_entry(o, n);
1150                         match = hashmap_get(hm, key, NULL);
1151                         free(key);
1152                         break;
1153                 case DIFF_SYMBOL_MINUS:
1154                         hm = add_lines;
1155                         key = prepare_entry(o, n);
1156                         match = hashmap_get(hm, key, NULL);
1157                         free(key);
1158                         break;
1159                 default:
1160                         flipped_block = 0;
1161                 }
1162
1163                 if (!match) {
1164                         int i;
1165
1166                         adjust_last_block(o, n, block_length);
1167                         for(i = 0; i < pmb_nr; i++)
1168                                 moved_block_clear(&pmb[i]);
1169                         pmb_nr = 0;
1170                         block_length = 0;
1171                         flipped_block = 0;
1172                         last_symbol = l->s;
1173                         continue;
1174                 }
1175
1176                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1177                         last_symbol = l->s;
1178                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1179                         continue;
1180                 }
1181
1182                 if (o->color_moved_ws_handling &
1183                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1184                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1185                 else
1186                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1187
1188                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1189
1190                 if (pmb_nr == 0) {
1191                         /*
1192                          * The current line is the start of a new block.
1193                          * Setup the set of potential blocks.
1194                          */
1195                         for (; match; match = hashmap_get_next(hm, match)) {
1196                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1197                                 if (o->color_moved_ws_handling &
1198                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1199                                         if (compute_ws_delta(l, match->es,
1200                                                              &pmb[pmb_nr].wsd))
1201                                                 pmb[pmb_nr++].match = match;
1202                                 } else {
1203                                         pmb[pmb_nr].wsd = 0;
1204                                         pmb[pmb_nr++].match = match;
1205                                 }
1206                         }
1207
1208                         if (adjust_last_block(o, n, block_length) &&
1209                             pmb_nr && last_symbol != l->s)
1210                                 flipped_block = (flipped_block + 1) % 2;
1211                         else
1212                                 flipped_block = 0;
1213
1214                         block_length = 0;
1215                 }
1216
1217                 if (pmb_nr) {
1218                         block_length++;
1219                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1220                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1221                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1222                 }
1223                 last_symbol = l->s;
1224         }
1225         adjust_last_block(o, n, block_length);
1226
1227         for(n = 0; n < pmb_nr; n++)
1228                 moved_block_clear(&pmb[n]);
1229         free(pmb);
1230 }
1231
1232 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1233   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1234 static void dim_moved_lines(struct diff_options *o)
1235 {
1236         int n;
1237         for (n = 0; n < o->emitted_symbols->nr; n++) {
1238                 struct emitted_diff_symbol *prev = (n != 0) ?
1239                                 &o->emitted_symbols->buf[n - 1] : NULL;
1240                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1241                 struct emitted_diff_symbol *next =
1242                                 (n < o->emitted_symbols->nr - 1) ?
1243                                 &o->emitted_symbols->buf[n + 1] : NULL;
1244
1245                 /* Not a plus or minus line? */
1246                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1247                         continue;
1248
1249                 /* Not a moved line? */
1250                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1251                         continue;
1252
1253                 /*
1254                  * If prev or next are not a plus or minus line,
1255                  * pretend they don't exist
1256                  */
1257                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1258                             prev->s != DIFF_SYMBOL_MINUS)
1259                         prev = NULL;
1260                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1261                             next->s != DIFF_SYMBOL_MINUS)
1262                         next = NULL;
1263
1264                 /* Inside a block? */
1265                 if ((prev &&
1266                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1267                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1268                     (next &&
1269                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1270                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1271                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1272                         continue;
1273                 }
1274
1275                 /* Check if we are at an interesting bound: */
1276                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1277                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1278                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1279                         continue;
1280                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1281                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1282                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1283                         continue;
1284
1285                 /*
1286                  * The boundary to prev and next are not interesting,
1287                  * so this line is not interesting as a whole
1288                  */
1289                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1290         }
1291 }
1292
1293 static void emit_line_ws_markup(struct diff_options *o,
1294                                 const char *set_sign, const char *set,
1295                                 const char *reset,
1296                                 int sign_index, const char *line, int len,
1297                                 unsigned ws_rule, int blank_at_eof)
1298 {
1299         const char *ws = NULL;
1300         int sign = o->output_indicators[sign_index];
1301
1302         if (o->ws_error_highlight & ws_rule) {
1303                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1304                 if (!*ws)
1305                         ws = NULL;
1306         }
1307
1308         if (!ws && !set_sign)
1309                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1310         else if (!ws) {
1311                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1312         } else if (blank_at_eof)
1313                 /* Blank line at EOF - paint '+' as well */
1314                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1315         else {
1316                 /* Emit just the prefix, then the rest. */
1317                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1318                             sign, "", 0);
1319                 ws_check_emit(line, len, ws_rule,
1320                               o->file, set, reset, ws);
1321         }
1322 }
1323
1324 static void emit_diff_symbol_from_struct(struct diff_options *o,
1325                                          struct emitted_diff_symbol *eds)
1326 {
1327         static const char *nneof = " No newline at end of file\n";
1328         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1329         struct strbuf sb = STRBUF_INIT;
1330
1331         enum diff_symbol s = eds->s;
1332         const char *line = eds->line;
1333         int len = eds->len;
1334         unsigned flags = eds->flags;
1335
1336         switch (s) {
1337         case DIFF_SYMBOL_NO_LF_EOF:
1338                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1339                 reset = diff_get_color_opt(o, DIFF_RESET);
1340                 putc('\n', o->file);
1341                 emit_line_0(o, context, NULL, 0, reset, '\\',
1342                             nneof, strlen(nneof));
1343                 break;
1344         case DIFF_SYMBOL_SUBMODULE_HEADER:
1345         case DIFF_SYMBOL_SUBMODULE_ERROR:
1346         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1347         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1348         case DIFF_SYMBOL_SUMMARY:
1349         case DIFF_SYMBOL_STATS_LINE:
1350         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1351         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1352                 emit_line(o, "", "", line, len);
1353                 break;
1354         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1355         case DIFF_SYMBOL_CONTEXT_MARKER:
1356                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1357                 reset = diff_get_color_opt(o, DIFF_RESET);
1358                 emit_line(o, context, reset, line, len);
1359                 break;
1360         case DIFF_SYMBOL_SEPARATOR:
1361                 fprintf(o->file, "%s%c",
1362                         diff_line_prefix(o),
1363                         o->line_termination);
1364                 break;
1365         case DIFF_SYMBOL_CONTEXT:
1366                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1367                 reset = diff_get_color_opt(o, DIFF_RESET);
1368                 set_sign = NULL;
1369                 if (o->flags.dual_color_diffed_diffs) {
1370                         char c = !len ? 0 : line[0];
1371
1372                         if (c == '+')
1373                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1374                         else if (c == '@')
1375                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1376                         else if (c == '-')
1377                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1378                 }
1379                 emit_line_ws_markup(o, set_sign, set, reset,
1380                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1381                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1382                 break;
1383         case DIFF_SYMBOL_PLUS:
1384                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1385                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1386                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1387                 case DIFF_SYMBOL_MOVED_LINE |
1388                      DIFF_SYMBOL_MOVED_LINE_ALT |
1389                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1390                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1391                         break;
1392                 case DIFF_SYMBOL_MOVED_LINE |
1393                      DIFF_SYMBOL_MOVED_LINE_ALT:
1394                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1395                         break;
1396                 case DIFF_SYMBOL_MOVED_LINE |
1397                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1398                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1399                         break;
1400                 case DIFF_SYMBOL_MOVED_LINE:
1401                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1402                         break;
1403                 default:
1404                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1405                 }
1406                 reset = diff_get_color_opt(o, DIFF_RESET);
1407                 if (!o->flags.dual_color_diffed_diffs)
1408                         set_sign = NULL;
1409                 else {
1410                         char c = !len ? 0 : line[0];
1411
1412                         set_sign = set;
1413                         if (c == '-')
1414                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1415                         else if (c == '@')
1416                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1417                         else if (c == '+')
1418                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1419                         else
1420                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1421                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1422                 }
1423                 emit_line_ws_markup(o, set_sign, set, reset,
1424                                     OUTPUT_INDICATOR_NEW, line, len,
1425                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1426                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1427                 break;
1428         case DIFF_SYMBOL_MINUS:
1429                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1430                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1431                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1432                 case DIFF_SYMBOL_MOVED_LINE |
1433                      DIFF_SYMBOL_MOVED_LINE_ALT |
1434                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1435                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1436                         break;
1437                 case DIFF_SYMBOL_MOVED_LINE |
1438                      DIFF_SYMBOL_MOVED_LINE_ALT:
1439                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1440                         break;
1441                 case DIFF_SYMBOL_MOVED_LINE |
1442                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1443                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1444                         break;
1445                 case DIFF_SYMBOL_MOVED_LINE:
1446                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1447                         break;
1448                 default:
1449                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1450                 }
1451                 reset = diff_get_color_opt(o, DIFF_RESET);
1452                 if (!o->flags.dual_color_diffed_diffs)
1453                         set_sign = NULL;
1454                 else {
1455                         char c = !len ? 0 : line[0];
1456
1457                         set_sign = set;
1458                         if (c == '+')
1459                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1460                         else if (c == '@')
1461                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1462                         else if (c == '-')
1463                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1464                         else
1465                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1466                 }
1467                 emit_line_ws_markup(o, set_sign, set, reset,
1468                                     OUTPUT_INDICATOR_OLD, line, len,
1469                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1470                 break;
1471         case DIFF_SYMBOL_WORDS_PORCELAIN:
1472                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1473                 reset = diff_get_color_opt(o, DIFF_RESET);
1474                 emit_line(o, context, reset, line, len);
1475                 fputs("~\n", o->file);
1476                 break;
1477         case DIFF_SYMBOL_WORDS:
1478                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1479                 reset = diff_get_color_opt(o, DIFF_RESET);
1480                 /*
1481                  * Skip the prefix character, if any.  With
1482                  * diff_suppress_blank_empty, there may be
1483                  * none.
1484                  */
1485                 if (line[0] != '\n') {
1486                         line++;
1487                         len--;
1488                 }
1489                 emit_line(o, context, reset, line, len);
1490                 break;
1491         case DIFF_SYMBOL_FILEPAIR_PLUS:
1492                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1493                 reset = diff_get_color_opt(o, DIFF_RESET);
1494                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1495                         line, reset,
1496                         strchr(line, ' ') ? "\t" : "");
1497                 break;
1498         case DIFF_SYMBOL_FILEPAIR_MINUS:
1499                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1500                 reset = diff_get_color_opt(o, DIFF_RESET);
1501                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1502                         line, reset,
1503                         strchr(line, ' ') ? "\t" : "");
1504                 break;
1505         case DIFF_SYMBOL_BINARY_FILES:
1506         case DIFF_SYMBOL_HEADER:
1507                 fprintf(o->file, "%s", line);
1508                 break;
1509         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1510                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1511                 break;
1512         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1513                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1514                 break;
1515         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1516                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1517                 break;
1518         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1519                 fputs(diff_line_prefix(o), o->file);
1520                 fputc('\n', o->file);
1521                 break;
1522         case DIFF_SYMBOL_REWRITE_DIFF:
1523                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1524                 reset = diff_get_color_opt(o, DIFF_RESET);
1525                 emit_line(o, fraginfo, reset, line, len);
1526                 break;
1527         case DIFF_SYMBOL_SUBMODULE_ADD:
1528                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1529                 reset = diff_get_color_opt(o, DIFF_RESET);
1530                 emit_line(o, set, reset, line, len);
1531                 break;
1532         case DIFF_SYMBOL_SUBMODULE_DEL:
1533                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1534                 reset = diff_get_color_opt(o, DIFF_RESET);
1535                 emit_line(o, set, reset, line, len);
1536                 break;
1537         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1538                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1539                         diff_line_prefix(o), line);
1540                 break;
1541         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1542                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1543                         diff_line_prefix(o), line);
1544                 break;
1545         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1546                 emit_line(o, "", "", " 0 files changed\n",
1547                           strlen(" 0 files changed\n"));
1548                 break;
1549         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1550                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1551                 break;
1552         case DIFF_SYMBOL_WORD_DIFF:
1553                 fprintf(o->file, "%.*s", len, line);
1554                 break;
1555         case DIFF_SYMBOL_STAT_SEP:
1556                 fputs(o->stat_sep, o->file);
1557                 break;
1558         default:
1559                 BUG("unknown diff symbol");
1560         }
1561         strbuf_release(&sb);
1562 }
1563
1564 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1565                              const char *line, int len, unsigned flags)
1566 {
1567         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1568
1569         if (o->emitted_symbols)
1570                 append_emitted_diff_symbol(o, &e);
1571         else
1572                 emit_diff_symbol_from_struct(o, &e);
1573 }
1574
1575 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1576 {
1577         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1578 }
1579
1580 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1581 {
1582         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1583 }
1584
1585 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1586 {
1587         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1588                          path, strlen(path), 0);
1589 }
1590
1591 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1592 {
1593         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1594                          path, strlen(path), 0);
1595 }
1596
1597 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1598 {
1599         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1600                          header, strlen(header), 0);
1601 }
1602
1603 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1604 {
1605         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1606 }
1607
1608 void diff_emit_submodule_pipethrough(struct diff_options *o,
1609                                      const char *line, int len)
1610 {
1611         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1612 }
1613
1614 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1615 {
1616         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1617               ecbdata->blank_at_eof_in_preimage &&
1618               ecbdata->blank_at_eof_in_postimage &&
1619               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1620               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1621                 return 0;
1622         return ws_blank_line(line, len, ecbdata->ws_rule);
1623 }
1624
1625 static void emit_add_line(struct emit_callback *ecbdata,
1626                           const char *line, int len)
1627 {
1628         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1629         if (new_blank_line_at_eof(ecbdata, line, len))
1630                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1631
1632         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1633 }
1634
1635 static void emit_del_line(struct emit_callback *ecbdata,
1636                           const char *line, int len)
1637 {
1638         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1639         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1640 }
1641
1642 static void emit_context_line(struct emit_callback *ecbdata,
1643                               const char *line, int len)
1644 {
1645         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1646         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1647 }
1648
1649 static void emit_hunk_header(struct emit_callback *ecbdata,
1650                              const char *line, int len)
1651 {
1652         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1653         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1654         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1655         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1656         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1657         static const char atat[2] = { '@', '@' };
1658         const char *cp, *ep;
1659         struct strbuf msgbuf = STRBUF_INIT;
1660         int org_len = len;
1661         int i = 1;
1662
1663         /*
1664          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1665          * it always is at least 10 bytes long.
1666          */
1667         if (len < 10 ||
1668             memcmp(line, atat, 2) ||
1669             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1670                 emit_diff_symbol(ecbdata->opt,
1671                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1672                 return;
1673         }
1674         ep += 2; /* skip over @@ */
1675
1676         /* The hunk header in fraginfo color */
1677         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1678                 strbuf_addstr(&msgbuf, reverse);
1679         strbuf_addstr(&msgbuf, frag);
1680         strbuf_add(&msgbuf, line, ep - line);
1681         strbuf_addstr(&msgbuf, reset);
1682
1683         /*
1684          * trailing "\r\n"
1685          */
1686         for ( ; i < 3; i++)
1687                 if (line[len - i] == '\r' || line[len - i] == '\n')
1688                         len--;
1689
1690         /* blank before the func header */
1691         for (cp = ep; ep - line < len; ep++)
1692                 if (*ep != ' ' && *ep != '\t')
1693                         break;
1694         if (ep != cp) {
1695                 strbuf_addstr(&msgbuf, context);
1696                 strbuf_add(&msgbuf, cp, ep - cp);
1697                 strbuf_addstr(&msgbuf, reset);
1698         }
1699
1700         if (ep < line + len) {
1701                 strbuf_addstr(&msgbuf, func);
1702                 strbuf_add(&msgbuf, ep, line + len - ep);
1703                 strbuf_addstr(&msgbuf, reset);
1704         }
1705
1706         strbuf_add(&msgbuf, line + len, org_len - len);
1707         strbuf_complete_line(&msgbuf);
1708         emit_diff_symbol(ecbdata->opt,
1709                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1710         strbuf_release(&msgbuf);
1711 }
1712
1713 static struct diff_tempfile *claim_diff_tempfile(void)
1714 {
1715         int i;
1716         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1717                 if (!diff_temp[i].name)
1718                         return diff_temp + i;
1719         BUG("diff is failing to clean up its tempfiles");
1720 }
1721
1722 static void remove_tempfile(void)
1723 {
1724         int i;
1725         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1726                 if (is_tempfile_active(diff_temp[i].tempfile))
1727                         delete_tempfile(&diff_temp[i].tempfile);
1728                 diff_temp[i].name = NULL;
1729         }
1730 }
1731
1732 static void add_line_count(struct strbuf *out, int count)
1733 {
1734         switch (count) {
1735         case 0:
1736                 strbuf_addstr(out, "0,0");
1737                 break;
1738         case 1:
1739                 strbuf_addstr(out, "1");
1740                 break;
1741         default:
1742                 strbuf_addf(out, "1,%d", count);
1743                 break;
1744         }
1745 }
1746
1747 static void emit_rewrite_lines(struct emit_callback *ecb,
1748                                int prefix, const char *data, int size)
1749 {
1750         const char *endp = NULL;
1751
1752         while (0 < size) {
1753                 int len;
1754
1755                 endp = memchr(data, '\n', size);
1756                 len = endp ? (endp - data + 1) : size;
1757                 if (prefix != '+') {
1758                         ecb->lno_in_preimage++;
1759                         emit_del_line(ecb, data, len);
1760                 } else {
1761                         ecb->lno_in_postimage++;
1762                         emit_add_line(ecb, data, len);
1763                 }
1764                 size -= len;
1765                 data += len;
1766         }
1767         if (!endp)
1768                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1769 }
1770
1771 static void emit_rewrite_diff(const char *name_a,
1772                               const char *name_b,
1773                               struct diff_filespec *one,
1774                               struct diff_filespec *two,
1775                               struct userdiff_driver *textconv_one,
1776                               struct userdiff_driver *textconv_two,
1777                               struct diff_options *o)
1778 {
1779         int lc_a, lc_b;
1780         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1781         const char *a_prefix, *b_prefix;
1782         char *data_one, *data_two;
1783         size_t size_one, size_two;
1784         struct emit_callback ecbdata;
1785         struct strbuf out = STRBUF_INIT;
1786
1787         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1788                 a_prefix = o->b_prefix;
1789                 b_prefix = o->a_prefix;
1790         } else {
1791                 a_prefix = o->a_prefix;
1792                 b_prefix = o->b_prefix;
1793         }
1794
1795         name_a += (*name_a == '/');
1796         name_b += (*name_b == '/');
1797
1798         strbuf_reset(&a_name);
1799         strbuf_reset(&b_name);
1800         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1801         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1802
1803         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1804         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1805
1806         memset(&ecbdata, 0, sizeof(ecbdata));
1807         ecbdata.color_diff = want_color(o->use_color);
1808         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1809         ecbdata.opt = o;
1810         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1811                 mmfile_t mf1, mf2;
1812                 mf1.ptr = (char *)data_one;
1813                 mf2.ptr = (char *)data_two;
1814                 mf1.size = size_one;
1815                 mf2.size = size_two;
1816                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1817         }
1818         ecbdata.lno_in_preimage = 1;
1819         ecbdata.lno_in_postimage = 1;
1820
1821         lc_a = count_lines(data_one, size_one);
1822         lc_b = count_lines(data_two, size_two);
1823
1824         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1825                          a_name.buf, a_name.len, 0);
1826         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1827                          b_name.buf, b_name.len, 0);
1828
1829         strbuf_addstr(&out, "@@ -");
1830         if (!o->irreversible_delete)
1831                 add_line_count(&out, lc_a);
1832         else
1833                 strbuf_addstr(&out, "?,?");
1834         strbuf_addstr(&out, " +");
1835         add_line_count(&out, lc_b);
1836         strbuf_addstr(&out, " @@\n");
1837         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1838         strbuf_release(&out);
1839
1840         if (lc_a && !o->irreversible_delete)
1841                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1842         if (lc_b)
1843                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1844         if (textconv_one)
1845                 free((char *)data_one);
1846         if (textconv_two)
1847                 free((char *)data_two);
1848 }
1849
1850 struct diff_words_buffer {
1851         mmfile_t text;
1852         unsigned long alloc;
1853         struct diff_words_orig {
1854                 const char *begin, *end;
1855         } *orig;
1856         int orig_nr, orig_alloc;
1857 };
1858
1859 static void diff_words_append(char *line, unsigned long len,
1860                 struct diff_words_buffer *buffer)
1861 {
1862         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1863         line++;
1864         len--;
1865         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1866         buffer->text.size += len;
1867         buffer->text.ptr[buffer->text.size] = '\0';
1868 }
1869
1870 struct diff_words_style_elem {
1871         const char *prefix;
1872         const char *suffix;
1873         const char *color; /* NULL; filled in by the setup code if
1874                             * color is enabled */
1875 };
1876
1877 struct diff_words_style {
1878         enum diff_words_type type;
1879         struct diff_words_style_elem new_word, old_word, ctx;
1880         const char *newline;
1881 };
1882
1883 static struct diff_words_style diff_words_styles[] = {
1884         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1885         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1886         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1887 };
1888
1889 struct diff_words_data {
1890         struct diff_words_buffer minus, plus;
1891         const char *current_plus;
1892         int last_minus;
1893         struct diff_options *opt;
1894         regex_t *word_regex;
1895         enum diff_words_type type;
1896         struct diff_words_style *style;
1897 };
1898
1899 static int fn_out_diff_words_write_helper(struct diff_options *o,
1900                                           struct diff_words_style_elem *st_el,
1901                                           const char *newline,
1902                                           size_t count, const char *buf)
1903 {
1904         int print = 0;
1905         struct strbuf sb = STRBUF_INIT;
1906
1907         while (count) {
1908                 char *p = memchr(buf, '\n', count);
1909                 if (print)
1910                         strbuf_addstr(&sb, diff_line_prefix(o));
1911
1912                 if (p != buf) {
1913                         const char *reset = st_el->color && *st_el->color ?
1914                                             GIT_COLOR_RESET : NULL;
1915                         if (st_el->color && *st_el->color)
1916                                 strbuf_addstr(&sb, st_el->color);
1917                         strbuf_addstr(&sb, st_el->prefix);
1918                         strbuf_add(&sb, buf, p ? p - buf : count);
1919                         strbuf_addstr(&sb, st_el->suffix);
1920                         if (reset)
1921                                 strbuf_addstr(&sb, reset);
1922                 }
1923                 if (!p)
1924                         goto out;
1925
1926                 strbuf_addstr(&sb, newline);
1927                 count -= p + 1 - buf;
1928                 buf = p + 1;
1929                 print = 1;
1930                 if (count) {
1931                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1932                                          sb.buf, sb.len, 0);
1933                         strbuf_reset(&sb);
1934                 }
1935         }
1936
1937 out:
1938         if (sb.len)
1939                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1940                                  sb.buf, sb.len, 0);
1941         strbuf_release(&sb);
1942         return 0;
1943 }
1944
1945 /*
1946  * '--color-words' algorithm can be described as:
1947  *
1948  *   1. collect the minus/plus lines of a diff hunk, divided into
1949  *      minus-lines and plus-lines;
1950  *
1951  *   2. break both minus-lines and plus-lines into words and
1952  *      place them into two mmfile_t with one word for each line;
1953  *
1954  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1955  *
1956  * And for the common parts of the both file, we output the plus side text.
1957  * diff_words->current_plus is used to trace the current position of the plus file
1958  * which printed. diff_words->last_minus is used to trace the last minus word
1959  * printed.
1960  *
1961  * For '--graph' to work with '--color-words', we need to output the graph prefix
1962  * on each line of color words output. Generally, there are two conditions on
1963  * which we should output the prefix.
1964  *
1965  *   1. diff_words->last_minus == 0 &&
1966  *      diff_words->current_plus == diff_words->plus.text.ptr
1967  *
1968  *      that is: the plus text must start as a new line, and if there is no minus
1969  *      word printed, a graph prefix must be printed.
1970  *
1971  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1972  *      *(diff_words->current_plus - 1) == '\n'
1973  *
1974  *      that is: a graph prefix must be printed following a '\n'
1975  */
1976 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1977 {
1978         if ((diff_words->last_minus == 0 &&
1979                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1980                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1981                 *(diff_words->current_plus - 1) == '\n')) {
1982                 return 1;
1983         } else {
1984                 return 0;
1985         }
1986 }
1987
1988 static void fn_out_diff_words_aux(void *priv,
1989                                   long minus_first, long minus_len,
1990                                   long plus_first, long plus_len,
1991                                   const char *func, long funclen)
1992 {
1993         struct diff_words_data *diff_words = priv;
1994         struct diff_words_style *style = diff_words->style;
1995         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1996         struct diff_options *opt = diff_words->opt;
1997         const char *line_prefix;
1998
1999         assert(opt);
2000         line_prefix = diff_line_prefix(opt);
2001
2002         /* POSIX requires that first be decremented by one if len == 0... */
2003         if (minus_len) {
2004                 minus_begin = diff_words->minus.orig[minus_first].begin;
2005                 minus_end =
2006                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2007         } else
2008                 minus_begin = minus_end =
2009                         diff_words->minus.orig[minus_first].end;
2010
2011         if (plus_len) {
2012                 plus_begin = diff_words->plus.orig[plus_first].begin;
2013                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2014         } else
2015                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2016
2017         if (color_words_output_graph_prefix(diff_words)) {
2018                 fputs(line_prefix, diff_words->opt->file);
2019         }
2020         if (diff_words->current_plus != plus_begin) {
2021                 fn_out_diff_words_write_helper(diff_words->opt,
2022                                 &style->ctx, style->newline,
2023                                 plus_begin - diff_words->current_plus,
2024                                 diff_words->current_plus);
2025         }
2026         if (minus_begin != minus_end) {
2027                 fn_out_diff_words_write_helper(diff_words->opt,
2028                                 &style->old_word, style->newline,
2029                                 minus_end - minus_begin, minus_begin);
2030         }
2031         if (plus_begin != plus_end) {
2032                 fn_out_diff_words_write_helper(diff_words->opt,
2033                                 &style->new_word, style->newline,
2034                                 plus_end - plus_begin, plus_begin);
2035         }
2036
2037         diff_words->current_plus = plus_end;
2038         diff_words->last_minus = minus_first;
2039 }
2040
2041 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2042 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2043                 int *begin, int *end)
2044 {
2045         if (word_regex && *begin < buffer->size) {
2046                 regmatch_t match[1];
2047                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2048                                  buffer->size - *begin, 1, match, 0)) {
2049                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2050                                         '\n', match[0].rm_eo - match[0].rm_so);
2051                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2052                         *begin += match[0].rm_so;
2053                         return *begin >= *end;
2054                 }
2055                 return -1;
2056         }
2057
2058         /* find the next word */
2059         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2060                 (*begin)++;
2061         if (*begin >= buffer->size)
2062                 return -1;
2063
2064         /* find the end of the word */
2065         *end = *begin + 1;
2066         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2067                 (*end)++;
2068
2069         return 0;
2070 }
2071
2072 /*
2073  * This function splits the words in buffer->text, stores the list with
2074  * newline separator into out, and saves the offsets of the original words
2075  * in buffer->orig.
2076  */
2077 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2078                 regex_t *word_regex)
2079 {
2080         int i, j;
2081         long alloc = 0;
2082
2083         out->size = 0;
2084         out->ptr = NULL;
2085
2086         /* fake an empty "0th" word */
2087         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2088         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2089         buffer->orig_nr = 1;
2090
2091         for (i = 0; i < buffer->text.size; i++) {
2092                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2093                         return;
2094
2095                 /* store original boundaries */
2096                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2097                                 buffer->orig_alloc);
2098                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2099                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2100                 buffer->orig_nr++;
2101
2102                 /* store one word */
2103                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2104                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2105                 out->ptr[out->size + j - i] = '\n';
2106                 out->size += j - i + 1;
2107
2108                 i = j - 1;
2109         }
2110 }
2111
2112 /* this executes the word diff on the accumulated buffers */
2113 static void diff_words_show(struct diff_words_data *diff_words)
2114 {
2115         xpparam_t xpp;
2116         xdemitconf_t xecfg;
2117         mmfile_t minus, plus;
2118         struct diff_words_style *style = diff_words->style;
2119
2120         struct diff_options *opt = diff_words->opt;
2121         const char *line_prefix;
2122
2123         assert(opt);
2124         line_prefix = diff_line_prefix(opt);
2125
2126         /* special case: only removal */
2127         if (!diff_words->plus.text.size) {
2128                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2129                                  line_prefix, strlen(line_prefix), 0);
2130                 fn_out_diff_words_write_helper(diff_words->opt,
2131                         &style->old_word, style->newline,
2132                         diff_words->minus.text.size,
2133                         diff_words->minus.text.ptr);
2134                 diff_words->minus.text.size = 0;
2135                 return;
2136         }
2137
2138         diff_words->current_plus = diff_words->plus.text.ptr;
2139         diff_words->last_minus = 0;
2140
2141         memset(&xpp, 0, sizeof(xpp));
2142         memset(&xecfg, 0, sizeof(xecfg));
2143         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2144         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2145         xpp.flags = 0;
2146         /* as only the hunk header will be parsed, we need a 0-context */
2147         xecfg.ctxlen = 0;
2148         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2149                           diff_words, &xpp, &xecfg))
2150                 die("unable to generate word diff");
2151         free(minus.ptr);
2152         free(plus.ptr);
2153         if (diff_words->current_plus != diff_words->plus.text.ptr +
2154                         diff_words->plus.text.size) {
2155                 if (color_words_output_graph_prefix(diff_words))
2156                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2157                                          line_prefix, strlen(line_prefix), 0);
2158                 fn_out_diff_words_write_helper(diff_words->opt,
2159                         &style->ctx, style->newline,
2160                         diff_words->plus.text.ptr + diff_words->plus.text.size
2161                         - diff_words->current_plus, diff_words->current_plus);
2162         }
2163         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2164 }
2165
2166 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2167 static void diff_words_flush(struct emit_callback *ecbdata)
2168 {
2169         struct diff_options *wo = ecbdata->diff_words->opt;
2170
2171         if (ecbdata->diff_words->minus.text.size ||
2172             ecbdata->diff_words->plus.text.size)
2173                 diff_words_show(ecbdata->diff_words);
2174
2175         if (wo->emitted_symbols) {
2176                 struct diff_options *o = ecbdata->opt;
2177                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2178                 int i;
2179
2180                 /*
2181                  * NEEDSWORK:
2182                  * Instead of appending each, concat all words to a line?
2183                  */
2184                 for (i = 0; i < wol->nr; i++)
2185                         append_emitted_diff_symbol(o, &wol->buf[i]);
2186
2187                 for (i = 0; i < wol->nr; i++)
2188                         free((void *)wol->buf[i].line);
2189
2190                 wol->nr = 0;
2191         }
2192 }
2193
2194 static void diff_filespec_load_driver(struct diff_filespec *one,
2195                                       struct index_state *istate)
2196 {
2197         /* Use already-loaded driver */
2198         if (one->driver)
2199                 return;
2200
2201         if (S_ISREG(one->mode))
2202                 one->driver = userdiff_find_by_path(istate, one->path);
2203
2204         /* Fallback to default settings */
2205         if (!one->driver)
2206                 one->driver = userdiff_find_by_name("default");
2207 }
2208
2209 static const char *userdiff_word_regex(struct diff_filespec *one,
2210                                        struct index_state *istate)
2211 {
2212         diff_filespec_load_driver(one, istate);
2213         return one->driver->word_regex;
2214 }
2215
2216 static void init_diff_words_data(struct emit_callback *ecbdata,
2217                                  struct diff_options *orig_opts,
2218                                  struct diff_filespec *one,
2219                                  struct diff_filespec *two)
2220 {
2221         int i;
2222         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2223         memcpy(o, orig_opts, sizeof(struct diff_options));
2224
2225         ecbdata->diff_words =
2226                 xcalloc(1, sizeof(struct diff_words_data));
2227         ecbdata->diff_words->type = o->word_diff;
2228         ecbdata->diff_words->opt = o;
2229
2230         if (orig_opts->emitted_symbols)
2231                 o->emitted_symbols =
2232                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2233
2234         if (!o->word_regex)
2235                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2236         if (!o->word_regex)
2237                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2238         if (!o->word_regex)
2239                 o->word_regex = diff_word_regex_cfg;
2240         if (o->word_regex) {
2241                 ecbdata->diff_words->word_regex = (regex_t *)
2242                         xmalloc(sizeof(regex_t));
2243                 if (regcomp(ecbdata->diff_words->word_regex,
2244                             o->word_regex,
2245                             REG_EXTENDED | REG_NEWLINE))
2246                         die("invalid regular expression: %s",
2247                             o->word_regex);
2248         }
2249         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2250                 if (o->word_diff == diff_words_styles[i].type) {
2251                         ecbdata->diff_words->style =
2252                                 &diff_words_styles[i];
2253                         break;
2254                 }
2255         }
2256         if (want_color(o->use_color)) {
2257                 struct diff_words_style *st = ecbdata->diff_words->style;
2258                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2259                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2260                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2261         }
2262 }
2263
2264 static void free_diff_words_data(struct emit_callback *ecbdata)
2265 {
2266         if (ecbdata->diff_words) {
2267                 diff_words_flush(ecbdata);
2268                 free (ecbdata->diff_words->opt->emitted_symbols);
2269                 free (ecbdata->diff_words->opt);
2270                 free (ecbdata->diff_words->minus.text.ptr);
2271                 free (ecbdata->diff_words->minus.orig);
2272                 free (ecbdata->diff_words->plus.text.ptr);
2273                 free (ecbdata->diff_words->plus.orig);
2274                 if (ecbdata->diff_words->word_regex) {
2275                         regfree(ecbdata->diff_words->word_regex);
2276                         free(ecbdata->diff_words->word_regex);
2277                 }
2278                 FREE_AND_NULL(ecbdata->diff_words);
2279         }
2280 }
2281
2282 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2283 {
2284         if (want_color(diff_use_color))
2285                 return diff_colors[ix];
2286         return "";
2287 }
2288
2289 const char *diff_line_prefix(struct diff_options *opt)
2290 {
2291         struct strbuf *msgbuf;
2292         if (!opt->output_prefix)
2293                 return "";
2294
2295         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2296         return msgbuf->buf;
2297 }
2298
2299 static unsigned long sane_truncate_line(char *line, unsigned long len)
2300 {
2301         const char *cp;
2302         unsigned long allot;
2303         size_t l = len;
2304
2305         cp = line;
2306         allot = l;
2307         while (0 < l) {
2308                 (void) utf8_width(&cp, &l);
2309                 if (!cp)
2310                         break; /* truncated in the middle? */
2311         }
2312         return allot - l;
2313 }
2314
2315 static void find_lno(const char *line, struct emit_callback *ecbdata)
2316 {
2317         const char *p;
2318         ecbdata->lno_in_preimage = 0;
2319         ecbdata->lno_in_postimage = 0;
2320         p = strchr(line, '-');
2321         if (!p)
2322                 return; /* cannot happen */
2323         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2324         p = strchr(p, '+');
2325         if (!p)
2326                 return; /* cannot happen */
2327         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2328 }
2329
2330 static void fn_out_consume(void *priv, char *line, unsigned long len)
2331 {
2332         struct emit_callback *ecbdata = priv;
2333         struct diff_options *o = ecbdata->opt;
2334
2335         o->found_changes = 1;
2336
2337         if (ecbdata->header) {
2338                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2339                                  ecbdata->header->buf, ecbdata->header->len, 0);
2340                 strbuf_reset(ecbdata->header);
2341                 ecbdata->header = NULL;
2342         }
2343
2344         if (ecbdata->label_path[0]) {
2345                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2346                                  ecbdata->label_path[0],
2347                                  strlen(ecbdata->label_path[0]), 0);
2348                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2349                                  ecbdata->label_path[1],
2350                                  strlen(ecbdata->label_path[1]), 0);
2351                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2352         }
2353
2354         if (diff_suppress_blank_empty
2355             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2356                 line[0] = '\n';
2357                 len = 1;
2358         }
2359
2360         if (line[0] == '@') {
2361                 if (ecbdata->diff_words)
2362                         diff_words_flush(ecbdata);
2363                 len = sane_truncate_line(line, len);
2364                 find_lno(line, ecbdata);
2365                 emit_hunk_header(ecbdata, line, len);
2366                 return;
2367         }
2368
2369         if (ecbdata->diff_words) {
2370                 enum diff_symbol s =
2371                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2372                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2373                 if (line[0] == '-') {
2374                         diff_words_append(line, len,
2375                                           &ecbdata->diff_words->minus);
2376                         return;
2377                 } else if (line[0] == '+') {
2378                         diff_words_append(line, len,
2379                                           &ecbdata->diff_words->plus);
2380                         return;
2381                 } else if (starts_with(line, "\\ ")) {
2382                         /*
2383                          * Eat the "no newline at eof" marker as if we
2384                          * saw a "+" or "-" line with nothing on it,
2385                          * and return without diff_words_flush() to
2386                          * defer processing. If this is the end of
2387                          * preimage, more "+" lines may come after it.
2388                          */
2389                         return;
2390                 }
2391                 diff_words_flush(ecbdata);
2392                 emit_diff_symbol(o, s, line, len, 0);
2393                 return;
2394         }
2395
2396         switch (line[0]) {
2397         case '+':
2398                 ecbdata->lno_in_postimage++;
2399                 emit_add_line(ecbdata, line + 1, len - 1);
2400                 break;
2401         case '-':
2402                 ecbdata->lno_in_preimage++;
2403                 emit_del_line(ecbdata, line + 1, len - 1);
2404                 break;
2405         case ' ':
2406                 ecbdata->lno_in_postimage++;
2407                 ecbdata->lno_in_preimage++;
2408                 emit_context_line(ecbdata, line + 1, len - 1);
2409                 break;
2410         default:
2411                 /* incomplete line at the end */
2412                 ecbdata->lno_in_preimage++;
2413                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2414                                  line, len, 0);
2415                 break;
2416         }
2417 }
2418
2419 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2420 {
2421         const char *old_name = a;
2422         const char *new_name = b;
2423         int pfx_length, sfx_length;
2424         int pfx_adjust_for_slash;
2425         int len_a = strlen(a);
2426         int len_b = strlen(b);
2427         int a_midlen, b_midlen;
2428         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2429         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2430
2431         if (qlen_a || qlen_b) {
2432                 quote_c_style(a, name, NULL, 0);
2433                 strbuf_addstr(name, " => ");
2434                 quote_c_style(b, name, NULL, 0);
2435                 return;
2436         }
2437
2438         /* Find common prefix */
2439         pfx_length = 0;
2440         while (*old_name && *new_name && *old_name == *new_name) {
2441                 if (*old_name == '/')
2442                         pfx_length = old_name - a + 1;
2443                 old_name++;
2444                 new_name++;
2445         }
2446
2447         /* Find common suffix */
2448         old_name = a + len_a;
2449         new_name = b + len_b;
2450         sfx_length = 0;
2451         /*
2452          * If there is a common prefix, it must end in a slash.  In
2453          * that case we let this loop run 1 into the prefix to see the
2454          * same slash.
2455          *
2456          * If there is no common prefix, we cannot do this as it would
2457          * underrun the input strings.
2458          */
2459         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2460         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2461                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2462                *old_name == *new_name) {
2463                 if (*old_name == '/')
2464                         sfx_length = len_a - (old_name - a);
2465                 old_name--;
2466                 new_name--;
2467         }
2468
2469         /*
2470          * pfx{mid-a => mid-b}sfx
2471          * {pfx-a => pfx-b}sfx
2472          * pfx{sfx-a => sfx-b}
2473          * name-a => name-b
2474          */
2475         a_midlen = len_a - pfx_length - sfx_length;
2476         b_midlen = len_b - pfx_length - sfx_length;
2477         if (a_midlen < 0)
2478                 a_midlen = 0;
2479         if (b_midlen < 0)
2480                 b_midlen = 0;
2481
2482         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2483         if (pfx_length + sfx_length) {
2484                 strbuf_add(name, a, pfx_length);
2485                 strbuf_addch(name, '{');
2486         }
2487         strbuf_add(name, a + pfx_length, a_midlen);
2488         strbuf_addstr(name, " => ");
2489         strbuf_add(name, b + pfx_length, b_midlen);
2490         if (pfx_length + sfx_length) {
2491                 strbuf_addch(name, '}');
2492                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2493         }
2494 }
2495
2496 struct diffstat_t {
2497         int nr;
2498         int alloc;
2499         struct diffstat_file {
2500                 char *from_name;
2501                 char *name;
2502                 char *print_name;
2503                 const char *comments;
2504                 unsigned is_unmerged:1;
2505                 unsigned is_binary:1;
2506                 unsigned is_renamed:1;
2507                 unsigned is_interesting:1;
2508                 uintmax_t added, deleted;
2509         } **files;
2510 };
2511
2512 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2513                                           const char *name_a,
2514                                           const char *name_b)
2515 {
2516         struct diffstat_file *x;
2517         x = xcalloc(1, sizeof(*x));
2518         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2519         diffstat->files[diffstat->nr++] = x;
2520         if (name_b) {
2521                 x->from_name = xstrdup(name_a);
2522                 x->name = xstrdup(name_b);
2523                 x->is_renamed = 1;
2524         }
2525         else {
2526                 x->from_name = NULL;
2527                 x->name = xstrdup(name_a);
2528         }
2529         return x;
2530 }
2531
2532 static void diffstat_consume(void *priv, char *line, unsigned long len)
2533 {
2534         struct diffstat_t *diffstat = priv;
2535         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2536
2537         if (line[0] == '+')
2538                 x->added++;
2539         else if (line[0] == '-')
2540                 x->deleted++;
2541 }
2542
2543 const char mime_boundary_leader[] = "------------";
2544
2545 static int scale_linear(int it, int width, int max_change)
2546 {
2547         if (!it)
2548                 return 0;
2549         /*
2550          * make sure that at least one '-' or '+' is printed if
2551          * there is any change to this path. The easiest way is to
2552          * scale linearly as if the alloted width is one column shorter
2553          * than it is, and then add 1 to the result.
2554          */
2555         return 1 + (it * (width - 1) / max_change);
2556 }
2557
2558 static void show_graph(struct strbuf *out, char ch, int cnt,
2559                        const char *set, const char *reset)
2560 {
2561         if (cnt <= 0)
2562                 return;
2563         strbuf_addstr(out, set);
2564         strbuf_addchars(out, ch, cnt);
2565         strbuf_addstr(out, reset);
2566 }
2567
2568 static void fill_print_name(struct diffstat_file *file)
2569 {
2570         struct strbuf pname = STRBUF_INIT;
2571
2572         if (file->print_name)
2573                 return;
2574
2575         if (file->is_renamed)
2576                 pprint_rename(&pname, file->from_name, file->name);
2577         else
2578                 quote_c_style(file->name, &pname, NULL, 0);
2579
2580         if (file->comments)
2581                 strbuf_addf(&pname, " (%s)", file->comments);
2582
2583         file->print_name = strbuf_detach(&pname, NULL);
2584 }
2585
2586 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2587                 int files, int insertions, int deletions)
2588 {
2589         struct strbuf sb = STRBUF_INIT;
2590
2591         if (!files) {
2592                 assert(insertions == 0 && deletions == 0);
2593                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2594                                  NULL, 0, 0);
2595                 return;
2596         }
2597
2598         strbuf_addf(&sb,
2599                     (files == 1) ? " %d file changed" : " %d files changed",
2600                     files);
2601
2602         /*
2603          * For binary diff, the caller may want to print "x files
2604          * changed" with insertions == 0 && deletions == 0.
2605          *
2606          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2607          * is probably less confusing (i.e skip over "2 files changed
2608          * but nothing about added/removed lines? Is this a bug in Git?").
2609          */
2610         if (insertions || deletions == 0) {
2611                 strbuf_addf(&sb,
2612                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2613                             insertions);
2614         }
2615
2616         if (deletions || insertions == 0) {
2617                 strbuf_addf(&sb,
2618                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2619                             deletions);
2620         }
2621         strbuf_addch(&sb, '\n');
2622         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2623                          sb.buf, sb.len, 0);
2624         strbuf_release(&sb);
2625 }
2626
2627 void print_stat_summary(FILE *fp, int files,
2628                         int insertions, int deletions)
2629 {
2630         struct diff_options o;
2631         memset(&o, 0, sizeof(o));
2632         o.file = fp;
2633
2634         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2635 }
2636
2637 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2638 {
2639         int i, len, add, del, adds = 0, dels = 0;
2640         uintmax_t max_change = 0, max_len = 0;
2641         int total_files = data->nr, count;
2642         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2643         const char *reset, *add_c, *del_c;
2644         int extra_shown = 0;
2645         const char *line_prefix = diff_line_prefix(options);
2646         struct strbuf out = STRBUF_INIT;
2647
2648         if (data->nr == 0)
2649                 return;
2650
2651         count = options->stat_count ? options->stat_count : data->nr;
2652
2653         reset = diff_get_color_opt(options, DIFF_RESET);
2654         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2655         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2656
2657         /*
2658          * Find the longest filename and max number of changes
2659          */
2660         for (i = 0; (i < count) && (i < data->nr); i++) {
2661                 struct diffstat_file *file = data->files[i];
2662                 uintmax_t change = file->added + file->deleted;
2663
2664                 if (!file->is_interesting && (change == 0)) {
2665                         count++; /* not shown == room for one more */
2666                         continue;
2667                 }
2668                 fill_print_name(file);
2669                 len = strlen(file->print_name);
2670                 if (max_len < len)
2671                         max_len = len;
2672
2673                 if (file->is_unmerged) {
2674                         /* "Unmerged" is 8 characters */
2675                         bin_width = bin_width < 8 ? 8 : bin_width;
2676                         continue;
2677                 }
2678                 if (file->is_binary) {
2679                         /* "Bin XXX -> YYY bytes" */
2680                         int w = 14 + decimal_width(file->added)
2681                                 + decimal_width(file->deleted);
2682                         bin_width = bin_width < w ? w : bin_width;
2683                         /* Display change counts aligned with "Bin" */
2684                         number_width = 3;
2685                         continue;
2686                 }
2687
2688                 if (max_change < change)
2689                         max_change = change;
2690         }
2691         count = i; /* where we can stop scanning in data->files[] */
2692
2693         /*
2694          * We have width = stat_width or term_columns() columns total.
2695          * We want a maximum of min(max_len, stat_name_width) for the name part.
2696          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2697          * We also need 1 for " " and 4 + decimal_width(max_change)
2698          * for " | NNNN " and one the empty column at the end, altogether
2699          * 6 + decimal_width(max_change).
2700          *
2701          * If there's not enough space, we will use the smaller of
2702          * stat_name_width (if set) and 5/8*width for the filename,
2703          * and the rest for constant elements + graph part, but no more
2704          * than stat_graph_width for the graph part.
2705          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2706          * for the standard terminal size).
2707          *
2708          * In other words: stat_width limits the maximum width, and
2709          * stat_name_width fixes the maximum width of the filename,
2710          * and is also used to divide available columns if there
2711          * aren't enough.
2712          *
2713          * Binary files are displayed with "Bin XXX -> YYY bytes"
2714          * instead of the change count and graph. This part is treated
2715          * similarly to the graph part, except that it is not
2716          * "scaled". If total width is too small to accommodate the
2717          * guaranteed minimum width of the filename part and the
2718          * separators and this message, this message will "overflow"
2719          * making the line longer than the maximum width.
2720          */
2721
2722         if (options->stat_width == -1)
2723                 width = term_columns() - strlen(line_prefix);
2724         else
2725                 width = options->stat_width ? options->stat_width : 80;
2726         number_width = decimal_width(max_change) > number_width ?
2727                 decimal_width(max_change) : number_width;
2728
2729         if (options->stat_graph_width == -1)
2730                 options->stat_graph_width = diff_stat_graph_width;
2731
2732         /*
2733          * Guarantee 3/8*16==6 for the graph part
2734          * and 5/8*16==10 for the filename part
2735          */
2736         if (width < 16 + 6 + number_width)
2737                 width = 16 + 6 + number_width;
2738
2739         /*
2740          * First assign sizes that are wanted, ignoring available width.
2741          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2742          * starting from "XXX" should fit in graph_width.
2743          */
2744         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2745         if (options->stat_graph_width &&
2746             options->stat_graph_width < graph_width)
2747                 graph_width = options->stat_graph_width;
2748
2749         name_width = (options->stat_name_width > 0 &&
2750                       options->stat_name_width < max_len) ?
2751                 options->stat_name_width : max_len;
2752
2753         /*
2754          * Adjust adjustable widths not to exceed maximum width
2755          */
2756         if (name_width + number_width + 6 + graph_width > width) {
2757                 if (graph_width > width * 3/8 - number_width - 6) {
2758                         graph_width = width * 3/8 - number_width - 6;
2759                         if (graph_width < 6)
2760                                 graph_width = 6;
2761                 }
2762
2763                 if (options->stat_graph_width &&
2764                     graph_width > options->stat_graph_width)
2765                         graph_width = options->stat_graph_width;
2766                 if (name_width > width - number_width - 6 - graph_width)
2767                         name_width = width - number_width - 6 - graph_width;
2768                 else
2769                         graph_width = width - number_width - 6 - name_width;
2770         }
2771
2772         /*
2773          * From here name_width is the width of the name area,
2774          * and graph_width is the width of the graph area.
2775          * max_change is used to scale graph properly.
2776          */
2777         for (i = 0; i < count; i++) {
2778                 const char *prefix = "";
2779                 struct diffstat_file *file = data->files[i];
2780                 char *name = file->print_name;
2781                 uintmax_t added = file->added;
2782                 uintmax_t deleted = file->deleted;
2783                 int name_len;
2784
2785                 if (!file->is_interesting && (added + deleted == 0))
2786                         continue;
2787
2788                 /*
2789                  * "scale" the filename
2790                  */
2791                 len = name_width;
2792                 name_len = strlen(name);
2793                 if (name_width < name_len) {
2794                         char *slash;
2795                         prefix = "...";
2796                         len -= 3;
2797                         name += name_len - len;
2798                         slash = strchr(name, '/');
2799                         if (slash)
2800                                 name = slash;
2801                 }
2802
2803                 if (file->is_binary) {
2804                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2805                         strbuf_addf(&out, " %*s", number_width, "Bin");
2806                         if (!added && !deleted) {
2807                                 strbuf_addch(&out, '\n');
2808                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2809                                                  out.buf, out.len, 0);
2810                                 strbuf_reset(&out);
2811                                 continue;
2812                         }
2813                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2814                                 del_c, deleted, reset);
2815                         strbuf_addstr(&out, " -> ");
2816                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2817                                 add_c, added, reset);
2818                         strbuf_addstr(&out, " bytes\n");
2819                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2820                                          out.buf, out.len, 0);
2821                         strbuf_reset(&out);
2822                         continue;
2823                 }
2824                 else if (file->is_unmerged) {
2825                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2826                         strbuf_addstr(&out, " Unmerged\n");
2827                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2828                                          out.buf, out.len, 0);
2829                         strbuf_reset(&out);
2830                         continue;
2831                 }
2832
2833                 /*
2834                  * scale the add/delete
2835                  */
2836                 add = added;
2837                 del = deleted;
2838
2839                 if (graph_width <= max_change) {
2840                         int total = scale_linear(add + del, graph_width, max_change);
2841                         if (total < 2 && add && del)
2842                                 /* width >= 2 due to the sanity check */
2843                                 total = 2;
2844                         if (add < del) {
2845                                 add = scale_linear(add, graph_width, max_change);
2846                                 del = total - add;
2847                         } else {
2848                                 del = scale_linear(del, graph_width, max_change);
2849                                 add = total - del;
2850                         }
2851                 }
2852                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2853                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2854                         number_width, added + deleted,
2855                         added + deleted ? " " : "");
2856                 show_graph(&out, '+', add, add_c, reset);
2857                 show_graph(&out, '-', del, del_c, reset);
2858                 strbuf_addch(&out, '\n');
2859                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2860                                  out.buf, out.len, 0);
2861                 strbuf_reset(&out);
2862         }
2863
2864         for (i = 0; i < data->nr; i++) {
2865                 struct diffstat_file *file = data->files[i];
2866                 uintmax_t added = file->added;
2867                 uintmax_t deleted = file->deleted;
2868
2869                 if (file->is_unmerged ||
2870                     (!file->is_interesting && (added + deleted == 0))) {
2871                         total_files--;
2872                         continue;
2873                 }
2874
2875                 if (!file->is_binary) {
2876                         adds += added;
2877                         dels += deleted;
2878                 }
2879                 if (i < count)
2880                         continue;
2881                 if (!extra_shown)
2882                         emit_diff_symbol(options,
2883                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2884                                          NULL, 0, 0);
2885                 extra_shown = 1;
2886         }
2887
2888         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2889         strbuf_release(&out);
2890 }
2891
2892 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2893 {
2894         int i, adds = 0, dels = 0, total_files = data->nr;
2895
2896         if (data->nr == 0)
2897                 return;
2898
2899         for (i = 0; i < data->nr; i++) {
2900                 int added = data->files[i]->added;
2901                 int deleted = data->files[i]->deleted;
2902
2903                 if (data->files[i]->is_unmerged ||
2904                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2905                         total_files--;
2906                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2907                         adds += added;
2908                         dels += deleted;
2909                 }
2910         }
2911         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2912 }
2913
2914 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2915 {
2916         int i;
2917
2918         if (data->nr == 0)
2919                 return;
2920
2921         for (i = 0; i < data->nr; i++) {
2922                 struct diffstat_file *file = data->files[i];
2923
2924                 fprintf(options->file, "%s", diff_line_prefix(options));
2925
2926                 if (file->is_binary)
2927                         fprintf(options->file, "-\t-\t");
2928                 else
2929                         fprintf(options->file,
2930                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2931                                 file->added, file->deleted);
2932                 if (options->line_termination) {
2933                         fill_print_name(file);
2934                         if (!file->is_renamed)
2935                                 write_name_quoted(file->name, options->file,
2936                                                   options->line_termination);
2937                         else {
2938                                 fputs(file->print_name, options->file);
2939                                 putc(options->line_termination, options->file);
2940                         }
2941                 } else {
2942                         if (file->is_renamed) {
2943                                 putc('\0', options->file);
2944                                 write_name_quoted(file->from_name, options->file, '\0');
2945                         }
2946                         write_name_quoted(file->name, options->file, '\0');
2947                 }
2948         }
2949 }
2950
2951 struct dirstat_file {
2952         const char *name;
2953         unsigned long changed;
2954 };
2955
2956 struct dirstat_dir {
2957         struct dirstat_file *files;
2958         int alloc, nr, permille, cumulative;
2959 };
2960
2961 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2962                 unsigned long changed, const char *base, int baselen)
2963 {
2964         unsigned long sum_changes = 0;
2965         unsigned int sources = 0;
2966         const char *line_prefix = diff_line_prefix(opt);
2967
2968         while (dir->nr) {
2969                 struct dirstat_file *f = dir->files;
2970                 int namelen = strlen(f->name);
2971                 unsigned long changes;
2972                 char *slash;
2973
2974                 if (namelen < baselen)
2975                         break;
2976                 if (memcmp(f->name, base, baselen))
2977                         break;
2978                 slash = strchr(f->name + baselen, '/');
2979                 if (slash) {
2980                         int newbaselen = slash + 1 - f->name;
2981                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2982                         sources++;
2983                 } else {
2984                         changes = f->changed;
2985                         dir->files++;
2986                         dir->nr--;
2987                         sources += 2;
2988                 }
2989                 sum_changes += changes;
2990         }
2991
2992         /*
2993          * We don't report dirstat's for
2994          *  - the top level
2995          *  - or cases where everything came from a single directory
2996          *    under this directory (sources == 1).
2997          */
2998         if (baselen && sources != 1) {
2999                 if (sum_changes) {
3000                         int permille = sum_changes * 1000 / changed;
3001                         if (permille >= dir->permille) {
3002                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3003                                         permille / 10, permille % 10, baselen, base);
3004                                 if (!dir->cumulative)
3005                                         return 0;
3006                         }
3007                 }
3008         }
3009         return sum_changes;
3010 }
3011
3012 static int dirstat_compare(const void *_a, const void *_b)
3013 {
3014         const struct dirstat_file *a = _a;
3015         const struct dirstat_file *b = _b;
3016         return strcmp(a->name, b->name);
3017 }
3018
3019 static void show_dirstat(struct diff_options *options)
3020 {
3021         int i;
3022         unsigned long changed;
3023         struct dirstat_dir dir;
3024         struct diff_queue_struct *q = &diff_queued_diff;
3025
3026         dir.files = NULL;
3027         dir.alloc = 0;
3028         dir.nr = 0;
3029         dir.permille = options->dirstat_permille;
3030         dir.cumulative = options->flags.dirstat_cumulative;
3031
3032         changed = 0;
3033         for (i = 0; i < q->nr; i++) {
3034                 struct diff_filepair *p = q->queue[i];
3035                 const char *name;
3036                 unsigned long copied, added, damage;
3037
3038                 name = p->two->path ? p->two->path : p->one->path;
3039
3040                 if (p->one->oid_valid && p->two->oid_valid &&
3041                     oideq(&p->one->oid, &p->two->oid)) {
3042                         /*
3043                          * The SHA1 has not changed, so pre-/post-content is
3044                          * identical. We can therefore skip looking at the
3045                          * file contents altogether.
3046                          */
3047                         damage = 0;
3048                         goto found_damage;
3049                 }
3050
3051                 if (options->flags.dirstat_by_file) {
3052                         /*
3053                          * In --dirstat-by-file mode, we don't really need to
3054                          * look at the actual file contents at all.
3055                          * The fact that the SHA1 changed is enough for us to
3056                          * add this file to the list of results
3057                          * (with each file contributing equal damage).
3058                          */
3059                         damage = 1;
3060                         goto found_damage;
3061                 }
3062
3063                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3064                         diff_populate_filespec(options->repo, p->one, 0);
3065                         diff_populate_filespec(options->repo, p->two, 0);
3066                         diffcore_count_changes(options->repo,
3067                                                p->one, p->two, NULL, NULL,
3068                                                &copied, &added);
3069                         diff_free_filespec_data(p->one);
3070                         diff_free_filespec_data(p->two);
3071                 } else if (DIFF_FILE_VALID(p->one)) {
3072                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3073                         copied = added = 0;
3074                         diff_free_filespec_data(p->one);
3075                 } else if (DIFF_FILE_VALID(p->two)) {
3076                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3077                         copied = 0;
3078                         added = p->two->size;
3079                         diff_free_filespec_data(p->two);
3080                 } else
3081                         continue;
3082
3083                 /*
3084                  * Original minus copied is the removed material,
3085                  * added is the new material.  They are both damages
3086                  * made to the preimage.
3087                  * If the resulting damage is zero, we know that
3088                  * diffcore_count_changes() considers the two entries to
3089                  * be identical, but since the oid changed, we
3090                  * know that there must have been _some_ kind of change,
3091                  * so we force all entries to have damage > 0.
3092                  */
3093                 damage = (p->one->size - copied) + added;
3094                 if (!damage)
3095                         damage = 1;
3096
3097 found_damage:
3098                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3099                 dir.files[dir.nr].name = name;
3100                 dir.files[dir.nr].changed = damage;
3101                 changed += damage;
3102                 dir.nr++;
3103         }
3104
3105         /* This can happen even with many files, if everything was renames */
3106         if (!changed)
3107                 return;
3108
3109         /* Show all directories with more than x% of the changes */
3110         QSORT(dir.files, dir.nr, dirstat_compare);
3111         gather_dirstat(options, &dir, changed, "", 0);
3112 }
3113
3114 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3115 {
3116         int i;
3117         unsigned long changed;
3118         struct dirstat_dir dir;
3119
3120         if (data->nr == 0)
3121                 return;
3122
3123         dir.files = NULL;
3124         dir.alloc = 0;
3125         dir.nr = 0;
3126         dir.permille = options->dirstat_permille;
3127         dir.cumulative = options->flags.dirstat_cumulative;
3128
3129         changed = 0;
3130         for (i = 0; i < data->nr; i++) {
3131                 struct diffstat_file *file = data->files[i];
3132                 unsigned long damage = file->added + file->deleted;
3133                 if (file->is_binary)
3134                         /*
3135                          * binary files counts bytes, not lines. Must find some
3136                          * way to normalize binary bytes vs. textual lines.
3137                          * The following heuristic assumes that there are 64
3138                          * bytes per "line".
3139                          * This is stupid and ugly, but very cheap...
3140                          */
3141                         damage = DIV_ROUND_UP(damage, 64);
3142                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3143                 dir.files[dir.nr].name = file->name;
3144                 dir.files[dir.nr].changed = damage;
3145                 changed += damage;
3146                 dir.nr++;
3147         }
3148
3149         /* This can happen even with many files, if everything was renames */
3150         if (!changed)
3151                 return;
3152
3153         /* Show all directories with more than x% of the changes */
3154         QSORT(dir.files, dir.nr, dirstat_compare);
3155         gather_dirstat(options, &dir, changed, "", 0);
3156 }
3157
3158 static void free_diffstat_info(struct diffstat_t *diffstat)
3159 {
3160         int i;
3161         for (i = 0; i < diffstat->nr; i++) {
3162                 struct diffstat_file *f = diffstat->files[i];
3163                 free(f->print_name);
3164                 free(f->name);
3165                 free(f->from_name);
3166                 free(f);
3167         }
3168         free(diffstat->files);
3169 }
3170
3171 struct checkdiff_t {
3172         const char *filename;
3173         int lineno;
3174         int conflict_marker_size;
3175         struct diff_options *o;
3176         unsigned ws_rule;
3177         unsigned status;
3178 };
3179
3180 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3181 {
3182         char firstchar;
3183         int cnt;
3184
3185         if (len < marker_size + 1)
3186                 return 0;
3187         firstchar = line[0];
3188         switch (firstchar) {
3189         case '=': case '>': case '<': case '|':
3190                 break;
3191         default:
3192                 return 0;
3193         }
3194         for (cnt = 1; cnt < marker_size; cnt++)
3195                 if (line[cnt] != firstchar)
3196                         return 0;
3197         /* line[1] thru line[marker_size-1] are same as firstchar */
3198         if (len < marker_size + 1 || !isspace(line[marker_size]))
3199                 return 0;
3200         return 1;
3201 }
3202
3203 static void checkdiff_consume_hunk(void *priv,
3204                                    long ob, long on, long nb, long nn,
3205                                    const char *func, long funclen)
3206
3207 {
3208         struct checkdiff_t *data = priv;
3209         data->lineno = nb - 1;
3210 }
3211
3212 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3213 {
3214         struct checkdiff_t *data = priv;
3215         int marker_size = data->conflict_marker_size;
3216         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3217         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3218         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3219         char *err;
3220         const char *line_prefix;
3221
3222         assert(data->o);
3223         line_prefix = diff_line_prefix(data->o);
3224
3225         if (line[0] == '+') {
3226                 unsigned bad;
3227                 data->lineno++;
3228                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3229                         data->status |= 1;
3230                         fprintf(data->o->file,
3231                                 "%s%s:%d: leftover conflict marker\n",
3232                                 line_prefix, data->filename, data->lineno);
3233                 }
3234                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3235                 if (!bad)
3236                         return;
3237                 data->status |= bad;
3238                 err = whitespace_error_string(bad);
3239                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3240                         line_prefix, data->filename, data->lineno, err);
3241                 free(err);
3242                 emit_line(data->o, set, reset, line, 1);
3243                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3244                               data->o->file, set, reset, ws);
3245         } else if (line[0] == ' ') {
3246                 data->lineno++;
3247         }
3248 }
3249
3250 static unsigned char *deflate_it(char *data,
3251                                  unsigned long size,
3252                                  unsigned long *result_size)
3253 {
3254         int bound;
3255         unsigned char *deflated;
3256         git_zstream stream;
3257
3258         git_deflate_init(&stream, zlib_compression_level);
3259         bound = git_deflate_bound(&stream, size);
3260         deflated = xmalloc(bound);
3261         stream.next_out = deflated;
3262         stream.avail_out = bound;
3263
3264         stream.next_in = (unsigned char *)data;
3265         stream.avail_in = size;
3266         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3267                 ; /* nothing */
3268         git_deflate_end(&stream);
3269         *result_size = stream.total_out;
3270         return deflated;
3271 }
3272
3273 static void emit_binary_diff_body(struct diff_options *o,
3274                                   mmfile_t *one, mmfile_t *two)
3275 {
3276         void *cp;
3277         void *delta;
3278         void *deflated;
3279         void *data;
3280         unsigned long orig_size;
3281         unsigned long delta_size;
3282         unsigned long deflate_size;
3283         unsigned long data_size;
3284
3285         /* We could do deflated delta, or we could do just deflated two,
3286          * whichever is smaller.
3287          */
3288         delta = NULL;
3289         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3290         if (one->size && two->size) {
3291                 delta = diff_delta(one->ptr, one->size,
3292                                    two->ptr, two->size,
3293                                    &delta_size, deflate_size);
3294                 if (delta) {
3295                         void *to_free = delta;
3296                         orig_size = delta_size;
3297                         delta = deflate_it(delta, delta_size, &delta_size);
3298                         free(to_free);
3299                 }
3300         }
3301
3302         if (delta && delta_size < deflate_size) {
3303                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3304                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3305                                  s, strlen(s), 0);
3306                 free(s);
3307                 free(deflated);
3308                 data = delta;
3309                 data_size = delta_size;
3310         } else {
3311                 char *s = xstrfmt("%lu", two->size);
3312                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3313                                  s, strlen(s), 0);
3314                 free(s);
3315                 free(delta);
3316                 data = deflated;
3317                 data_size = deflate_size;
3318         }
3319
3320         /* emit data encoded in base85 */
3321         cp = data;
3322         while (data_size) {
3323                 int len;
3324                 int bytes = (52 < data_size) ? 52 : data_size;
3325                 char line[71];
3326                 data_size -= bytes;
3327                 if (bytes <= 26)
3328                         line[0] = bytes + 'A' - 1;
3329                 else
3330                         line[0] = bytes - 26 + 'a' - 1;
3331                 encode_85(line + 1, cp, bytes);
3332                 cp = (char *) cp + bytes;
3333
3334                 len = strlen(line);
3335                 line[len++] = '\n';
3336                 line[len] = '\0';
3337
3338                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3339                                  line, len, 0);
3340         }
3341         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3342         free(data);
3343 }
3344
3345 static void emit_binary_diff(struct diff_options *o,
3346                              mmfile_t *one, mmfile_t *two)
3347 {
3348         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3349         emit_binary_diff_body(o, one, two);
3350         emit_binary_diff_body(o, two, one);
3351 }
3352
3353 int diff_filespec_is_binary(struct repository *r,
3354                             struct diff_filespec *one)
3355 {
3356         if (one->is_binary == -1) {
3357                 diff_filespec_load_driver(one, r->index);
3358                 if (one->driver->binary != -1)
3359                         one->is_binary = one->driver->binary;
3360                 else {
3361                         if (!one->data && DIFF_FILE_VALID(one))
3362                                 diff_populate_filespec(r, one, CHECK_BINARY);
3363                         if (one->is_binary == -1 && one->data)
3364                                 one->is_binary = buffer_is_binary(one->data,
3365                                                 one->size);
3366                         if (one->is_binary == -1)
3367                                 one->is_binary = 0;
3368                 }
3369         }
3370         return one->is_binary;
3371 }
3372
3373 static const struct userdiff_funcname *
3374 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3375 {
3376         diff_filespec_load_driver(one, o->repo->index);
3377         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3378 }
3379
3380 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3381 {
3382         if (!options->a_prefix)
3383                 options->a_prefix = a;
3384         if (!options->b_prefix)
3385                 options->b_prefix = b;
3386 }
3387
3388 struct userdiff_driver *get_textconv(struct repository *r,
3389                                      struct diff_filespec *one)
3390 {
3391         if (!DIFF_FILE_VALID(one))
3392                 return NULL;
3393
3394         diff_filespec_load_driver(one, r->index);
3395         return userdiff_get_textconv(r, one->driver);
3396 }
3397
3398 static void builtin_diff(const char *name_a,
3399                          const char *name_b,
3400                          struct diff_filespec *one,
3401                          struct diff_filespec *two,
3402                          const char *xfrm_msg,
3403                          int must_show_header,
3404                          struct diff_options *o,
3405                          int complete_rewrite)
3406 {
3407         mmfile_t mf1, mf2;
3408         const char *lbl[2];
3409         char *a_one, *b_two;
3410         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3411         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3412         const char *a_prefix, *b_prefix;
3413         struct userdiff_driver *textconv_one = NULL;
3414         struct userdiff_driver *textconv_two = NULL;
3415         struct strbuf header = STRBUF_INIT;
3416         const char *line_prefix = diff_line_prefix(o);
3417
3418         diff_set_mnemonic_prefix(o, "a/", "b/");
3419         if (o->flags.reverse_diff) {
3420                 a_prefix = o->b_prefix;
3421                 b_prefix = o->a_prefix;
3422         } else {
3423                 a_prefix = o->a_prefix;
3424                 b_prefix = o->b_prefix;
3425         }
3426
3427         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3428             (!one->mode || S_ISGITLINK(one->mode)) &&
3429             (!two->mode || S_ISGITLINK(two->mode))) {
3430                 show_submodule_summary(o, one->path ? one->path : two->path,
3431                                 &one->oid, &two->oid,
3432                                 two->dirty_submodule);
3433                 return;
3434         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3435                    (!one->mode || S_ISGITLINK(one->mode)) &&
3436                    (!two->mode || S_ISGITLINK(two->mode))) {
3437                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3438                                 &one->oid, &two->oid,
3439                                 two->dirty_submodule);
3440                 return;
3441         }
3442
3443         if (o->flags.allow_textconv) {
3444                 textconv_one = get_textconv(o->repo, one);
3445                 textconv_two = get_textconv(o->repo, two);
3446         }
3447
3448         /* Never use a non-valid filename anywhere if at all possible */
3449         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3450         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3451
3452         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3453         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3454         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3455         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3456         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3457         if (lbl[0][0] == '/') {
3458                 /* /dev/null */
3459                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3460                 if (xfrm_msg)
3461                         strbuf_addstr(&header, xfrm_msg);
3462                 must_show_header = 1;
3463         }
3464         else if (lbl[1][0] == '/') {
3465                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3466                 if (xfrm_msg)
3467                         strbuf_addstr(&header, xfrm_msg);
3468                 must_show_header = 1;
3469         }
3470         else {
3471                 if (one->mode != two->mode) {
3472                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3473                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3474                         must_show_header = 1;
3475                 }
3476                 if (xfrm_msg)
3477                         strbuf_addstr(&header, xfrm_msg);
3478
3479                 /*
3480                  * we do not run diff between different kind
3481                  * of objects.
3482                  */
3483                 if ((one->mode ^ two->mode) & S_IFMT)
3484                         goto free_ab_and_return;
3485                 if (complete_rewrite &&
3486                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3487                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3488                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3489                                          header.buf, header.len, 0);
3490                         strbuf_reset(&header);
3491                         emit_rewrite_diff(name_a, name_b, one, two,
3492                                           textconv_one, textconv_two, o);
3493                         o->found_changes = 1;
3494                         goto free_ab_and_return;
3495                 }
3496         }
3497
3498         if (o->irreversible_delete && lbl[1][0] == '/') {
3499                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3500                                  header.len, 0);
3501                 strbuf_reset(&header);
3502                 goto free_ab_and_return;
3503         } else if (!o->flags.text &&
3504                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3505                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3506                 struct strbuf sb = STRBUF_INIT;
3507                 if (!one->data && !two->data &&
3508                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3509                     !o->flags.binary) {
3510                         if (oideq(&one->oid, &two->oid)) {
3511                                 if (must_show_header)
3512                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3513                                                          header.buf, header.len,
3514                                                          0);
3515                                 goto free_ab_and_return;
3516                         }
3517                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3518                                          header.buf, header.len, 0);
3519                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3520                                     diff_line_prefix(o), lbl[0], lbl[1]);
3521                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3522                                          sb.buf, sb.len, 0);
3523                         strbuf_release(&sb);
3524                         goto free_ab_and_return;
3525                 }
3526                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3527                     fill_mmfile(o->repo, &mf2, two) < 0)
3528                         die("unable to read files to diff");
3529                 /* Quite common confusing case */
3530                 if (mf1.size == mf2.size &&
3531                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3532                         if (must_show_header)
3533                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3534                                                  header.buf, header.len, 0);
3535                         goto free_ab_and_return;
3536                 }
3537                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3538                 strbuf_reset(&header);
3539                 if (o->flags.binary)
3540                         emit_binary_diff(o, &mf1, &mf2);
3541                 else {
3542                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3543                                     diff_line_prefix(o), lbl[0], lbl[1]);
3544                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3545                                          sb.buf, sb.len, 0);
3546                         strbuf_release(&sb);
3547                 }
3548                 o->found_changes = 1;
3549         } else {
3550                 /* Crazy xdl interfaces.. */
3551                 const char *diffopts;
3552                 const char *v;
3553                 xpparam_t xpp;
3554                 xdemitconf_t xecfg;
3555                 struct emit_callback ecbdata;
3556                 const struct userdiff_funcname *pe;
3557
3558                 if (must_show_header) {
3559                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3560                                          header.buf, header.len, 0);
3561                         strbuf_reset(&header);
3562                 }
3563
3564                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3565                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3566
3567                 pe = diff_funcname_pattern(o, one);
3568                 if (!pe)
3569                         pe = diff_funcname_pattern(o, two);
3570
3571                 memset(&xpp, 0, sizeof(xpp));
3572                 memset(&xecfg, 0, sizeof(xecfg));
3573                 memset(&ecbdata, 0, sizeof(ecbdata));
3574                 if (o->flags.suppress_diff_headers)
3575                         lbl[0] = NULL;
3576                 ecbdata.label_path = lbl;
3577                 ecbdata.color_diff = want_color(o->use_color);
3578                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3579                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3580                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3581                 ecbdata.opt = o;
3582                 if (header.len && !o->flags.suppress_diff_headers)
3583                         ecbdata.header = &header;
3584                 xpp.flags = o->xdl_opts;
3585                 xpp.anchors = o->anchors;
3586                 xpp.anchors_nr = o->anchors_nr;
3587                 xecfg.ctxlen = o->context;
3588                 xecfg.interhunkctxlen = o->interhunkcontext;
3589                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3590                 if (o->flags.funccontext)
3591                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3592                 if (pe)
3593                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3594
3595                 diffopts = getenv("GIT_DIFF_OPTS");
3596                 if (!diffopts)
3597                         ;
3598                 else if (skip_prefix(diffopts, "--unified=", &v))
3599                         xecfg.ctxlen = strtoul(v, NULL, 10);
3600                 else if (skip_prefix(diffopts, "-u", &v))
3601                         xecfg.ctxlen = strtoul(v, NULL, 10);
3602
3603                 if (o->word_diff)
3604                         init_diff_words_data(&ecbdata, o, one, two);
3605                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3606                                   &ecbdata, &xpp, &xecfg))
3607                         die("unable to generate diff for %s", one->path);
3608                 if (o->word_diff)
3609                         free_diff_words_data(&ecbdata);
3610                 if (textconv_one)
3611                         free(mf1.ptr);
3612                 if (textconv_two)
3613                         free(mf2.ptr);
3614                 xdiff_clear_find_func(&xecfg);
3615         }
3616
3617  free_ab_and_return:
3618         strbuf_release(&header);
3619         diff_free_filespec_data(one);
3620         diff_free_filespec_data(two);
3621         free(a_one);
3622         free(b_two);
3623         return;
3624 }
3625
3626 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3627 {
3628         if (!is_renamed) {
3629                 if (p->status == DIFF_STATUS_ADDED) {
3630                         if (S_ISLNK(p->two->mode))
3631                                 return "new +l";
3632                         else if ((p->two->mode & 0777) == 0755)
3633                                 return "new +x";
3634                         else
3635                                 return "new";
3636                 } else if (p->status == DIFF_STATUS_DELETED)
3637                         return "gone";
3638         }
3639         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3640                 return "mode -l";
3641         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3642                 return "mode +l";
3643         else if ((p->one->mode & 0777) == 0644 &&
3644                  (p->two->mode & 0777) == 0755)
3645                 return "mode +x";
3646         else if ((p->one->mode & 0777) == 0755 &&
3647                  (p->two->mode & 0777) == 0644)
3648                 return "mode -x";
3649         return NULL;
3650 }
3651
3652 static void builtin_diffstat(const char *name_a, const char *name_b,
3653                              struct diff_filespec *one,
3654                              struct diff_filespec *two,
3655                              struct diffstat_t *diffstat,
3656                              struct diff_options *o,
3657                              struct diff_filepair *p)
3658 {
3659         mmfile_t mf1, mf2;
3660         struct diffstat_file *data;
3661         int same_contents;
3662         int complete_rewrite = 0;
3663
3664         if (!DIFF_PAIR_UNMERGED(p)) {
3665                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3666                         complete_rewrite = 1;
3667         }
3668
3669         data = diffstat_add(diffstat, name_a, name_b);
3670         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3671         if (o->flags.stat_with_summary)
3672                 data->comments = get_compact_summary(p, data->is_renamed);
3673
3674         if (!one || !two) {
3675                 data->is_unmerged = 1;
3676                 return;
3677         }
3678
3679         same_contents = oideq(&one->oid, &two->oid);
3680
3681         if (diff_filespec_is_binary(o->repo, one) ||
3682             diff_filespec_is_binary(o->repo, two)) {
3683                 data->is_binary = 1;
3684                 if (same_contents) {
3685                         data->added = 0;
3686                         data->deleted = 0;
3687                 } else {
3688                         data->added = diff_filespec_size(o->repo, two);
3689                         data->deleted = diff_filespec_size(o->repo, one);
3690                 }
3691         }
3692
3693         else if (complete_rewrite) {
3694                 diff_populate_filespec(o->repo, one, 0);
3695                 diff_populate_filespec(o->repo, two, 0);
3696                 data->deleted = count_lines(one->data, one->size);
3697                 data->added = count_lines(two->data, two->size);
3698         }
3699
3700         else if (!same_contents) {
3701                 /* Crazy xdl interfaces.. */
3702                 xpparam_t xpp;
3703                 xdemitconf_t xecfg;
3704
3705                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3706                     fill_mmfile(o->repo, &mf2, two) < 0)
3707                         die("unable to read files to diff");
3708
3709                 memset(&xpp, 0, sizeof(xpp));
3710                 memset(&xecfg, 0, sizeof(xecfg));
3711                 xpp.flags = o->xdl_opts;
3712                 xpp.anchors = o->anchors;
3713                 xpp.anchors_nr = o->anchors_nr;
3714                 xecfg.ctxlen = o->context;
3715                 xecfg.interhunkctxlen = o->interhunkcontext;
3716                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3717                                   diffstat_consume, diffstat, &xpp, &xecfg))
3718                         die("unable to generate diffstat for %s", one->path);
3719         }
3720
3721         diff_free_filespec_data(one);
3722         diff_free_filespec_data(two);
3723 }
3724
3725 static void builtin_checkdiff(const char *name_a, const char *name_b,
3726                               const char *attr_path,
3727                               struct diff_filespec *one,
3728                               struct diff_filespec *two,
3729                               struct diff_options *o)
3730 {
3731         mmfile_t mf1, mf2;
3732         struct checkdiff_t data;
3733
3734         if (!two)
3735                 return;
3736
3737         memset(&data, 0, sizeof(data));
3738         data.filename = name_b ? name_b : name_a;
3739         data.lineno = 0;
3740         data.o = o;
3741         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3742         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3743
3744         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3745             fill_mmfile(o->repo, &mf2, two) < 0)
3746                 die("unable to read files to diff");
3747
3748         /*
3749          * All the other codepaths check both sides, but not checking
3750          * the "old" side here is deliberate.  We are checking the newly
3751          * introduced changes, and as long as the "new" side is text, we
3752          * can and should check what it introduces.
3753          */
3754         if (diff_filespec_is_binary(o->repo, two))
3755                 goto free_and_return;
3756         else {
3757                 /* Crazy xdl interfaces.. */
3758                 xpparam_t xpp;
3759                 xdemitconf_t xecfg;
3760
3761                 memset(&xpp, 0, sizeof(xpp));
3762                 memset(&xecfg, 0, sizeof(xecfg));
3763                 xecfg.ctxlen = 1; /* at least one context line */
3764                 xpp.flags = 0;
3765                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3766                                   checkdiff_consume, &data,
3767                                   &xpp, &xecfg))
3768                         die("unable to generate checkdiff for %s", one->path);
3769
3770                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3771                         struct emit_callback ecbdata;
3772                         int blank_at_eof;
3773
3774                         ecbdata.ws_rule = data.ws_rule;
3775                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3776                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3777
3778                         if (blank_at_eof) {
3779                                 static char *err;
3780                                 if (!err)
3781                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3782                                 fprintf(o->file, "%s:%d: %s.\n",
3783                                         data.filename, blank_at_eof, err);
3784                                 data.status = 1; /* report errors */
3785                         }
3786                 }
3787         }
3788  free_and_return:
3789         diff_free_filespec_data(one);
3790         diff_free_filespec_data(two);
3791         if (data.status)
3792                 o->flags.check_failed = 1;
3793 }
3794
3795 struct diff_filespec *alloc_filespec(const char *path)
3796 {
3797         struct diff_filespec *spec;
3798
3799         FLEXPTR_ALLOC_STR(spec, path, path);
3800         spec->count = 1;
3801         spec->is_binary = -1;
3802         return spec;
3803 }
3804
3805 void free_filespec(struct diff_filespec *spec)
3806 {
3807         if (!--spec->count) {
3808                 diff_free_filespec_data(spec);
3809                 free(spec);
3810         }
3811 }
3812
3813 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3814                    int oid_valid, unsigned short mode)
3815 {
3816         if (mode) {
3817                 spec->mode = canon_mode(mode);
3818                 oidcpy(&spec->oid, oid);
3819                 spec->oid_valid = oid_valid;
3820         }
3821 }
3822
3823 /*
3824  * Given a name and sha1 pair, if the index tells us the file in
3825  * the work tree has that object contents, return true, so that
3826  * prepare_temp_file() does not have to inflate and extract.
3827  */
3828 static int reuse_worktree_file(struct index_state *istate,
3829                                const char *name,
3830                                const struct object_id *oid,
3831                                int want_file)
3832 {
3833         const struct cache_entry *ce;
3834         struct stat st;
3835         int pos, len;
3836
3837         /*
3838          * We do not read the cache ourselves here, because the
3839          * benchmark with my previous version that always reads cache
3840          * shows that it makes things worse for diff-tree comparing
3841          * two linux-2.6 kernel trees in an already checked out work
3842          * tree.  This is because most diff-tree comparisons deal with
3843          * only a small number of files, while reading the cache is
3844          * expensive for a large project, and its cost outweighs the
3845          * savings we get by not inflating the object to a temporary
3846          * file.  Practically, this code only helps when we are used
3847          * by diff-cache --cached, which does read the cache before
3848          * calling us.
3849          */
3850         if (!istate->cache)
3851                 return 0;
3852
3853         /* We want to avoid the working directory if our caller
3854          * doesn't need the data in a normal file, this system
3855          * is rather slow with its stat/open/mmap/close syscalls,
3856          * and the object is contained in a pack file.  The pack
3857          * is probably already open and will be faster to obtain
3858          * the data through than the working directory.  Loose
3859          * objects however would tend to be slower as they need
3860          * to be individually opened and inflated.
3861          */
3862         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3863                 return 0;
3864
3865         /*
3866          * Similarly, if we'd have to convert the file contents anyway, that
3867          * makes the optimization not worthwhile.
3868          */
3869         if (!want_file && would_convert_to_git(istate, name))
3870                 return 0;
3871
3872         len = strlen(name);
3873         pos = index_name_pos(istate, name, len);
3874         if (pos < 0)
3875                 return 0;
3876         ce = istate->cache[pos];
3877
3878         /*
3879          * This is not the sha1 we are looking for, or
3880          * unreusable because it is not a regular file.
3881          */
3882         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3883                 return 0;
3884
3885         /*
3886          * If ce is marked as "assume unchanged", there is no
3887          * guarantee that work tree matches what we are looking for.
3888          */
3889         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3890                 return 0;
3891
3892         /*
3893          * If ce matches the file in the work tree, we can reuse it.
3894          */
3895         if (ce_uptodate(ce) ||
3896             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3897                 return 1;
3898
3899         return 0;
3900 }
3901
3902 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3903 {
3904         struct strbuf buf = STRBUF_INIT;
3905         char *dirty = "";
3906
3907         /* Are we looking at the work tree? */
3908         if (s->dirty_submodule)
3909                 dirty = "-dirty";
3910
3911         strbuf_addf(&buf, "Subproject commit %s%s\n",
3912                     oid_to_hex(&s->oid), dirty);
3913         s->size = buf.len;
3914         if (size_only) {
3915                 s->data = NULL;
3916                 strbuf_release(&buf);
3917         } else {
3918                 s->data = strbuf_detach(&buf, NULL);
3919                 s->should_free = 1;
3920         }
3921         return 0;
3922 }
3923
3924 /*
3925  * While doing rename detection and pickaxe operation, we may need to
3926  * grab the data for the blob (or file) for our own in-core comparison.
3927  * diff_filespec has data and size fields for this purpose.
3928  */
3929 int diff_populate_filespec(struct repository *r,
3930                            struct diff_filespec *s,
3931                            unsigned int flags)
3932 {
3933         int size_only = flags & CHECK_SIZE_ONLY;
3934         int err = 0;
3935         int conv_flags = global_conv_flags_eol;
3936         /*
3937          * demote FAIL to WARN to allow inspecting the situation
3938          * instead of refusing.
3939          */
3940         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3941                 conv_flags = CONV_EOL_RNDTRP_WARN;
3942
3943         if (!DIFF_FILE_VALID(s))
3944                 die("internal error: asking to populate invalid file.");
3945         if (S_ISDIR(s->mode))
3946                 return -1;
3947
3948         if (s->data)
3949                 return 0;
3950
3951         if (size_only && 0 < s->size)
3952                 return 0;
3953
3954         if (S_ISGITLINK(s->mode))
3955                 return diff_populate_gitlink(s, size_only);
3956
3957         if (!s->oid_valid ||
3958             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3959                 struct strbuf buf = STRBUF_INIT;
3960                 struct stat st;
3961                 int fd;
3962
3963                 if (lstat(s->path, &st) < 0) {
3964                 err_empty:
3965                         err = -1;
3966                 empty:
3967                         s->data = (char *)"";
3968                         s->size = 0;
3969                         return err;
3970                 }
3971                 s->size = xsize_t(st.st_size);
3972                 if (!s->size)
3973                         goto empty;
3974                 if (S_ISLNK(st.st_mode)) {
3975                         struct strbuf sb = STRBUF_INIT;
3976
3977                         if (strbuf_readlink(&sb, s->path, s->size))
3978                                 goto err_empty;
3979                         s->size = sb.len;
3980                         s->data = strbuf_detach(&sb, NULL);
3981                         s->should_free = 1;
3982                         return 0;
3983                 }
3984
3985                 /*
3986                  * Even if the caller would be happy with getting
3987                  * only the size, we cannot return early at this
3988                  * point if the path requires us to run the content
3989                  * conversion.
3990                  */
3991                 if (size_only && !would_convert_to_git(r->index, s->path))
3992                         return 0;
3993
3994                 /*
3995                  * Note: this check uses xsize_t(st.st_size) that may
3996                  * not be the true size of the blob after it goes
3997                  * through convert_to_git().  This may not strictly be
3998                  * correct, but the whole point of big_file_threshold
3999                  * and is_binary check being that we want to avoid
4000                  * opening the file and inspecting the contents, this
4001                  * is probably fine.
4002                  */
4003                 if ((flags & CHECK_BINARY) &&
4004                     s->size > big_file_threshold && s->is_binary == -1) {
4005                         s->is_binary = 1;
4006                         return 0;
4007                 }
4008                 fd = open(s->path, O_RDONLY);
4009                 if (fd < 0)
4010                         goto err_empty;
4011                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4012                 close(fd);
4013                 s->should_munmap = 1;
4014
4015                 /*
4016                  * Convert from working tree format to canonical git format
4017                  */
4018                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4019                         size_t size = 0;
4020                         munmap(s->data, s->size);
4021                         s->should_munmap = 0;
4022                         s->data = strbuf_detach(&buf, &size);
4023                         s->size = size;
4024                         s->should_free = 1;
4025                 }
4026         }
4027         else {
4028                 enum object_type type;
4029                 if (size_only || (flags & CHECK_BINARY)) {
4030                         type = oid_object_info(r, &s->oid, &s->size);
4031                         if (type < 0)
4032                                 die("unable to read %s",
4033                                     oid_to_hex(&s->oid));
4034                         if (size_only)
4035                                 return 0;
4036                         if (s->size > big_file_threshold && s->is_binary == -1) {
4037                                 s->is_binary = 1;
4038                                 return 0;
4039                         }
4040                 }
4041                 s->data = read_object_file(&s->oid, &type, &s->size);
4042                 if (!s->data)
4043                         die("unable to read %s", oid_to_hex(&s->oid));
4044                 s->should_free = 1;
4045         }
4046         return 0;
4047 }
4048
4049 void diff_free_filespec_blob(struct diff_filespec *s)
4050 {
4051         if (s->should_free)
4052                 free(s->data);
4053         else if (s->should_munmap)
4054                 munmap(s->data, s->size);
4055
4056         if (s->should_free || s->should_munmap) {
4057                 s->should_free = s->should_munmap = 0;
4058                 s->data = NULL;
4059         }
4060 }
4061
4062 void diff_free_filespec_data(struct diff_filespec *s)
4063 {
4064         diff_free_filespec_blob(s);
4065         FREE_AND_NULL(s->cnt_data);
4066 }
4067
4068 static void prep_temp_blob(struct index_state *istate,
4069                            const char *path, struct diff_tempfile *temp,
4070                            void *blob,
4071                            unsigned long size,
4072                            const struct object_id *oid,
4073                            int mode)
4074 {
4075         struct strbuf buf = STRBUF_INIT;
4076         struct strbuf tempfile = STRBUF_INIT;
4077         char *path_dup = xstrdup(path);
4078         const char *base = basename(path_dup);
4079
4080         /* Generate "XXXXXX_basename.ext" */
4081         strbuf_addstr(&tempfile, "XXXXXX_");
4082         strbuf_addstr(&tempfile, base);
4083
4084         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4085         if (!temp->tempfile)
4086                 die_errno("unable to create temp-file");
4087         if (convert_to_working_tree(istate, path,
4088                         (const char *)blob, (size_t)size, &buf)) {
4089                 blob = buf.buf;
4090                 size = buf.len;
4091         }
4092         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4093             close_tempfile_gently(temp->tempfile))
4094                 die_errno("unable to write temp-file");
4095         temp->name = get_tempfile_path(temp->tempfile);
4096         oid_to_hex_r(temp->hex, oid);
4097         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4098         strbuf_release(&buf);
4099         strbuf_release(&tempfile);
4100         free(path_dup);
4101 }
4102
4103 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4104                                                const char *name,
4105                                                struct diff_filespec *one)
4106 {
4107         struct diff_tempfile *temp = claim_diff_tempfile();
4108
4109         if (!DIFF_FILE_VALID(one)) {
4110         not_a_valid_file:
4111                 /* A '-' entry produces this for file-2, and
4112                  * a '+' entry produces this for file-1.
4113                  */
4114                 temp->name = "/dev/null";
4115                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4116                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4117                 return temp;
4118         }
4119
4120         if (!S_ISGITLINK(one->mode) &&
4121             (!one->oid_valid ||
4122              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4123                 struct stat st;
4124                 if (lstat(name, &st) < 0) {
4125                         if (errno == ENOENT)
4126                                 goto not_a_valid_file;
4127                         die_errno("stat(%s)", name);
4128                 }
4129                 if (S_ISLNK(st.st_mode)) {
4130                         struct strbuf sb = STRBUF_INIT;
4131                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4132                                 die_errno("readlink(%s)", name);
4133                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4134                                        (one->oid_valid ?
4135                                         &one->oid : &null_oid),
4136                                        (one->oid_valid ?
4137                                         one->mode : S_IFLNK));
4138                         strbuf_release(&sb);
4139                 }
4140                 else {
4141                         /* we can borrow from the file in the work tree */
4142                         temp->name = name;
4143                         if (!one->oid_valid)
4144                                 oid_to_hex_r(temp->hex, &null_oid);
4145                         else
4146                                 oid_to_hex_r(temp->hex, &one->oid);
4147                         /* Even though we may sometimes borrow the
4148                          * contents from the work tree, we always want
4149                          * one->mode.  mode is trustworthy even when
4150                          * !(one->oid_valid), as long as
4151                          * DIFF_FILE_VALID(one).
4152                          */
4153                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4154                 }
4155                 return temp;
4156         }
4157         else {
4158                 if (diff_populate_filespec(r, one, 0))
4159                         die("cannot read data blob for %s", one->path);
4160                 prep_temp_blob(r->index, name, temp,
4161                                one->data, one->size,
4162                                &one->oid, one->mode);
4163         }
4164         return temp;
4165 }
4166
4167 static void add_external_diff_name(struct repository *r,
4168                                    struct argv_array *argv,
4169                                    const char *name,
4170                                    struct diff_filespec *df)
4171 {
4172         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4173         argv_array_push(argv, temp->name);
4174         argv_array_push(argv, temp->hex);
4175         argv_array_push(argv, temp->mode);
4176 }
4177
4178 /* An external diff command takes:
4179  *
4180  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4181  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4182  *
4183  */
4184 static void run_external_diff(const char *pgm,
4185                               const char *name,
4186                               const char *other,
4187                               struct diff_filespec *one,
4188                               struct diff_filespec *two,
4189                               const char *xfrm_msg,
4190                               struct diff_options *o)
4191 {
4192         struct argv_array argv = ARGV_ARRAY_INIT;
4193         struct argv_array env = ARGV_ARRAY_INIT;
4194         struct diff_queue_struct *q = &diff_queued_diff;
4195
4196         argv_array_push(&argv, pgm);
4197         argv_array_push(&argv, name);
4198
4199         if (one && two) {
4200                 add_external_diff_name(o->repo, &argv, name, one);
4201                 if (!other)
4202                         add_external_diff_name(o->repo, &argv, name, two);
4203                 else {
4204                         add_external_diff_name(o->repo, &argv, other, two);
4205                         argv_array_push(&argv, other);
4206                         argv_array_push(&argv, xfrm_msg);
4207                 }
4208         }
4209
4210         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4211         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4212
4213         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4214                 die(_("external diff died, stopping at %s"), name);
4215
4216         remove_tempfile();
4217         argv_array_clear(&argv);
4218         argv_array_clear(&env);
4219 }
4220
4221 static int similarity_index(struct diff_filepair *p)
4222 {
4223         return p->score * 100 / MAX_SCORE;
4224 }
4225
4226 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4227 {
4228         if (startup_info->have_repository)
4229                 return find_unique_abbrev(oid, abbrev);
4230         else {
4231                 char *hex = oid_to_hex(oid);
4232                 if (abbrev < 0)
4233                         abbrev = FALLBACK_DEFAULT_ABBREV;
4234                 if (abbrev > the_hash_algo->hexsz)
4235                         BUG("oid abbreviation out of range: %d", abbrev);
4236                 if (abbrev)
4237                         hex[abbrev] = '\0';
4238                 return hex;
4239         }
4240 }
4241
4242 static void fill_metainfo(struct strbuf *msg,
4243                           const char *name,
4244                           const char *other,
4245                           struct diff_filespec *one,
4246                           struct diff_filespec *two,
4247                           struct diff_options *o,
4248                           struct diff_filepair *p,
4249                           int *must_show_header,
4250                           int use_color)
4251 {
4252         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4253         const char *reset = diff_get_color(use_color, DIFF_RESET);
4254         const char *line_prefix = diff_line_prefix(o);
4255
4256         *must_show_header = 1;
4257         strbuf_init(msg, PATH_MAX * 2 + 300);
4258         switch (p->status) {
4259         case DIFF_STATUS_COPIED:
4260                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4261                             line_prefix, set, similarity_index(p));
4262                 strbuf_addf(msg, "%s\n%s%scopy from ",
4263                             reset,  line_prefix, set);
4264                 quote_c_style(name, msg, NULL, 0);
4265                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4266                 quote_c_style(other, msg, NULL, 0);
4267                 strbuf_addf(msg, "%s\n", reset);
4268                 break;
4269         case DIFF_STATUS_RENAMED:
4270                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4271                             line_prefix, set, similarity_index(p));
4272                 strbuf_addf(msg, "%s\n%s%srename from ",
4273                             reset, line_prefix, set);
4274                 quote_c_style(name, msg, NULL, 0);
4275                 strbuf_addf(msg, "%s\n%s%srename to ",
4276                             reset, line_prefix, set);
4277                 quote_c_style(other, msg, NULL, 0);
4278                 strbuf_addf(msg, "%s\n", reset);
4279                 break;
4280         case DIFF_STATUS_MODIFIED:
4281                 if (p->score) {
4282                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4283                                     line_prefix,
4284                                     set, similarity_index(p), reset);
4285                         break;
4286                 }
4287                 /* fallthru */
4288         default:
4289                 *must_show_header = 0;
4290         }
4291         if (one && two && !oideq(&one->oid, &two->oid)) {
4292                 const unsigned hexsz = the_hash_algo->hexsz;
4293                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4294
4295                 if (o->flags.binary) {
4296                         mmfile_t mf;
4297                         if ((!fill_mmfile(o->repo, &mf, one) &&
4298                              diff_filespec_is_binary(o->repo, one)) ||
4299                             (!fill_mmfile(o->repo, &mf, two) &&
4300                              diff_filespec_is_binary(o->repo, two)))
4301                                 abbrev = hexsz;
4302                 }
4303                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4304                             diff_abbrev_oid(&one->oid, abbrev),
4305                             diff_abbrev_oid(&two->oid, abbrev));
4306                 if (one->mode == two->mode)
4307                         strbuf_addf(msg, " %06o", one->mode);
4308                 strbuf_addf(msg, "%s\n", reset);
4309         }
4310 }
4311
4312 static void run_diff_cmd(const char *pgm,
4313                          const char *name,
4314                          const char *other,
4315                          const char *attr_path,
4316                          struct diff_filespec *one,
4317                          struct diff_filespec *two,
4318                          struct strbuf *msg,
4319                          struct diff_options *o,
4320                          struct diff_filepair *p)
4321 {
4322         const char *xfrm_msg = NULL;
4323         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4324         int must_show_header = 0;
4325
4326
4327         if (o->flags.allow_external) {
4328                 struct userdiff_driver *drv;
4329
4330                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4331                 if (drv && drv->external)
4332                         pgm = drv->external;
4333         }
4334
4335         if (msg) {
4336                 /*
4337                  * don't use colors when the header is intended for an
4338                  * external diff driver
4339                  */
4340                 fill_metainfo(msg, name, other, one, two, o, p,
4341                               &must_show_header,
4342                               want_color(o->use_color) && !pgm);
4343                 xfrm_msg = msg->len ? msg->buf : NULL;
4344         }
4345
4346         if (pgm) {
4347                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4348                 return;
4349         }
4350         if (one && two)
4351                 builtin_diff(name, other ? other : name,
4352                              one, two, xfrm_msg, must_show_header,
4353                              o, complete_rewrite);
4354         else
4355                 fprintf(o->file, "* Unmerged path %s\n", name);
4356 }
4357
4358 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4359 {
4360         if (DIFF_FILE_VALID(one)) {
4361                 if (!one->oid_valid) {
4362                         struct stat st;
4363                         if (one->is_stdin) {
4364                                 oidclr(&one->oid);
4365                                 return;
4366                         }
4367                         if (lstat(one->path, &st) < 0)
4368                                 die_errno("stat '%s'", one->path);
4369                         if (index_path(istate, &one->oid, one->path, &st, 0))
4370                                 die("cannot hash %s", one->path);
4371                 }
4372         }
4373         else
4374                 oidclr(&one->oid);
4375 }
4376
4377 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4378 {
4379         /* Strip the prefix but do not molest /dev/null and absolute paths */
4380         if (*namep && !is_absolute_path(*namep)) {
4381                 *namep += prefix_length;
4382                 if (**namep == '/')
4383                         ++*namep;
4384         }
4385         if (*otherp && !is_absolute_path(*otherp)) {
4386                 *otherp += prefix_length;
4387                 if (**otherp == '/')
4388                         ++*otherp;
4389         }
4390 }
4391
4392 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4393 {
4394         const char *pgm = external_diff();
4395         struct strbuf msg;
4396         struct diff_filespec *one = p->one;
4397         struct diff_filespec *two = p->two;
4398         const char *name;
4399         const char *other;
4400         const char *attr_path;
4401
4402         name  = one->path;
4403         other = (strcmp(name, two->path) ? two->path : NULL);
4404         attr_path = name;
4405         if (o->prefix_length)
4406                 strip_prefix(o->prefix_length, &name, &other);
4407
4408         if (!o->flags.allow_external)
4409                 pgm = NULL;
4410
4411         if (DIFF_PAIR_UNMERGED(p)) {
4412                 run_diff_cmd(pgm, name, NULL, attr_path,
4413                              NULL, NULL, NULL, o, p);
4414                 return;
4415         }
4416
4417         diff_fill_oid_info(one, o->repo->index);
4418         diff_fill_oid_info(two, o->repo->index);
4419
4420         if (!pgm &&
4421             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4422             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4423                 /*
4424                  * a filepair that changes between file and symlink
4425                  * needs to be split into deletion and creation.
4426                  */
4427                 struct diff_filespec *null = alloc_filespec(two->path);
4428                 run_diff_cmd(NULL, name, other, attr_path,
4429                              one, null, &msg,
4430                              o, p);
4431                 free(null);
4432                 strbuf_release(&msg);
4433
4434                 null = alloc_filespec(one->path);
4435                 run_diff_cmd(NULL, name, other, attr_path,
4436                              null, two, &msg, o, p);
4437                 free(null);
4438         }
4439         else
4440                 run_diff_cmd(pgm, name, other, attr_path,
4441                              one, two, &msg, o, p);
4442
4443         strbuf_release(&msg);
4444 }
4445
4446 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4447                          struct diffstat_t *diffstat)
4448 {
4449         const char *name;
4450         const char *other;
4451
4452         if (DIFF_PAIR_UNMERGED(p)) {
4453                 /* unmerged */
4454                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4455                                  diffstat, o, p);
4456                 return;
4457         }
4458
4459         name = p->one->path;
4460         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4461
4462         if (o->prefix_length)
4463                 strip_prefix(o->prefix_length, &name, &other);
4464
4465         diff_fill_oid_info(p->one, o->repo->index);
4466         diff_fill_oid_info(p->two, o->repo->index);
4467
4468         builtin_diffstat(name, other, p->one, p->two,
4469                          diffstat, o, p);
4470 }
4471
4472 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4473 {
4474         const char *name;
4475         const char *other;
4476         const char *attr_path;
4477
4478         if (DIFF_PAIR_UNMERGED(p)) {
4479                 /* unmerged */
4480                 return;
4481         }
4482
4483         name = p->one->path;
4484         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4485         attr_path = other ? other : name;
4486
4487         if (o->prefix_length)
4488                 strip_prefix(o->prefix_length, &name, &other);
4489
4490         diff_fill_oid_info(p->one, o->repo->index);
4491         diff_fill_oid_info(p->two, o->repo->index);
4492
4493         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4494 }
4495
4496 static void prep_parse_options(struct diff_options *options);
4497
4498 void repo_diff_setup(struct repository *r, struct diff_options *options)
4499 {
4500         memcpy(options, &default_diff_options, sizeof(*options));
4501
4502         options->file = stdout;
4503         options->repo = r;
4504
4505         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4506         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4507         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4508         options->abbrev = DEFAULT_ABBREV;
4509         options->line_termination = '\n';
4510         options->break_opt = -1;
4511         options->rename_limit = -1;
4512         options->dirstat_permille = diff_dirstat_permille_default;
4513         options->context = diff_context_default;
4514         options->interhunkcontext = diff_interhunk_context_default;
4515         options->ws_error_highlight = ws_error_highlight_default;
4516         options->flags.rename_empty = 1;
4517         options->objfind = NULL;
4518
4519         /* pathchange left =NULL by default */
4520         options->change = diff_change;
4521         options->add_remove = diff_addremove;
4522         options->use_color = diff_use_color_default;
4523         options->detect_rename = diff_detect_rename_default;
4524         options->xdl_opts |= diff_algorithm;
4525         if (diff_indent_heuristic)
4526                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4527
4528         options->orderfile = diff_order_file_cfg;
4529
4530         if (diff_no_prefix) {
4531                 options->a_prefix = options->b_prefix = "";
4532         } else if (!diff_mnemonic_prefix) {
4533                 options->a_prefix = "a/";
4534                 options->b_prefix = "b/";
4535         }
4536
4537         options->color_moved = diff_color_moved_default;
4538         options->color_moved_ws_handling = diff_color_moved_ws_default;
4539
4540         prep_parse_options(options);
4541 }
4542
4543 void diff_setup_done(struct diff_options *options)
4544 {
4545         unsigned check_mask = DIFF_FORMAT_NAME |
4546                               DIFF_FORMAT_NAME_STATUS |
4547                               DIFF_FORMAT_CHECKDIFF |
4548                               DIFF_FORMAT_NO_OUTPUT;
4549         /*
4550          * This must be signed because we're comparing against a potentially
4551          * negative value.
4552          */
4553         const int hexsz = the_hash_algo->hexsz;
4554
4555         if (options->set_default)
4556                 options->set_default(options);
4557
4558         if (HAS_MULTI_BITS(options->output_format & check_mask))
4559                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4560
4561         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4562                 die(_("-G, -S and --find-object are mutually exclusive"));
4563
4564         /*
4565          * Most of the time we can say "there are changes"
4566          * only by checking if there are changed paths, but
4567          * --ignore-whitespace* options force us to look
4568          * inside contents.
4569          */
4570
4571         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4572                 options->flags.diff_from_contents = 1;
4573         else
4574                 options->flags.diff_from_contents = 0;
4575
4576         if (options->flags.find_copies_harder)
4577                 options->detect_rename = DIFF_DETECT_COPY;
4578
4579         if (!options->flags.relative_name)
4580                 options->prefix = NULL;
4581         if (options->prefix)
4582                 options->prefix_length = strlen(options->prefix);
4583         else
4584                 options->prefix_length = 0;
4585
4586         if (options->output_format & (DIFF_FORMAT_NAME |
4587                                       DIFF_FORMAT_NAME_STATUS |
4588                                       DIFF_FORMAT_CHECKDIFF |
4589                                       DIFF_FORMAT_NO_OUTPUT))
4590                 options->output_format &= ~(DIFF_FORMAT_RAW |
4591                                             DIFF_FORMAT_NUMSTAT |
4592                                             DIFF_FORMAT_DIFFSTAT |
4593                                             DIFF_FORMAT_SHORTSTAT |
4594                                             DIFF_FORMAT_DIRSTAT |
4595                                             DIFF_FORMAT_SUMMARY |
4596                                             DIFF_FORMAT_PATCH);
4597
4598         /*
4599          * These cases always need recursive; we do not drop caller-supplied
4600          * recursive bits for other formats here.
4601          */
4602         if (options->output_format & (DIFF_FORMAT_PATCH |
4603                                       DIFF_FORMAT_NUMSTAT |
4604                                       DIFF_FORMAT_DIFFSTAT |
4605                                       DIFF_FORMAT_SHORTSTAT |
4606                                       DIFF_FORMAT_DIRSTAT |
4607                                       DIFF_FORMAT_SUMMARY |
4608                                       DIFF_FORMAT_CHECKDIFF))
4609                 options->flags.recursive = 1;
4610         /*
4611          * Also pickaxe would not work very well if you do not say recursive
4612          */
4613         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4614                 options->flags.recursive = 1;
4615         /*
4616          * When patches are generated, submodules diffed against the work tree
4617          * must be checked for dirtiness too so it can be shown in the output
4618          */
4619         if (options->output_format & DIFF_FORMAT_PATCH)
4620                 options->flags.dirty_submodules = 1;
4621
4622         if (options->detect_rename && options->rename_limit < 0)
4623                 options->rename_limit = diff_rename_limit_default;
4624         if (hexsz < options->abbrev)
4625                 options->abbrev = hexsz; /* full */
4626
4627         /*
4628          * It does not make sense to show the first hit we happened
4629          * to have found.  It does not make sense not to return with
4630          * exit code in such a case either.
4631          */
4632         if (options->flags.quick) {
4633                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4634                 options->flags.exit_with_status = 1;
4635         }
4636
4637         options->diff_path_counter = 0;
4638
4639         if (options->flags.follow_renames && options->pathspec.nr != 1)
4640                 die(_("--follow requires exactly one pathspec"));
4641
4642         if (!options->use_color || external_diff())
4643                 options->color_moved = 0;
4644
4645         FREE_AND_NULL(options->parseopts);
4646 }
4647
4648 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4649 {
4650         char c, *eq;
4651         int len;
4652
4653         if (*arg != '-')
4654                 return 0;
4655         c = *++arg;
4656         if (!c)
4657                 return 0;
4658         if (c == arg_short) {
4659                 c = *++arg;
4660                 if (!c)
4661                         return 1;
4662                 if (val && isdigit(c)) {
4663                         char *end;
4664                         int n = strtoul(arg, &end, 10);
4665                         if (*end)
4666                                 return 0;
4667                         *val = n;
4668                         return 1;
4669                 }
4670                 return 0;
4671         }
4672         if (c != '-')
4673                 return 0;
4674         arg++;
4675         eq = strchrnul(arg, '=');
4676         len = eq - arg;
4677         if (!len || strncmp(arg, arg_long, len))
4678                 return 0;
4679         if (*eq) {
4680                 int n;
4681                 char *end;
4682                 if (!isdigit(*++eq))
4683                         return 0;
4684                 n = strtoul(eq, &end, 10);
4685                 if (*end)
4686                         return 0;
4687                 *val = n;
4688         }
4689         return 1;
4690 }
4691
4692 static int diff_scoreopt_parse(const char *opt);
4693
4694 static inline int short_opt(char opt, const char **argv,
4695                             const char **optarg)
4696 {
4697         const char *arg = argv[0];
4698         if (arg[0] != '-' || arg[1] != opt)
4699                 return 0;
4700         if (arg[2] != '\0') {
4701                 *optarg = arg + 2;
4702                 return 1;
4703         }
4704         if (!argv[1])
4705                 die("Option '%c' requires a value", opt);
4706         *optarg = argv[1];
4707         return 2;
4708 }
4709
4710 int parse_long_opt(const char *opt, const char **argv,
4711                    const char **optarg)
4712 {
4713         const char *arg = argv[0];
4714         if (!skip_prefix(arg, "--", &arg))
4715                 return 0;
4716         if (!skip_prefix(arg, opt, &arg))
4717                 return 0;
4718         if (*arg == '=') { /* stuck form: --option=value */
4719                 *optarg = arg + 1;
4720                 return 1;
4721         }
4722         if (*arg != '\0')
4723                 return 0;
4724         /* separate form: --option value */
4725         if (!argv[1])
4726                 die("Option '--%s' requires a value", opt);
4727         *optarg = argv[1];
4728         return 2;
4729 }
4730
4731 static int stat_opt(struct diff_options *options, const char **av)
4732 {
4733         const char *arg = av[0];
4734         char *end;
4735         int width = options->stat_width;
4736         int name_width = options->stat_name_width;
4737         int graph_width = options->stat_graph_width;
4738         int count = options->stat_count;
4739         int argcount = 1;
4740
4741         if (!skip_prefix(arg, "--stat", &arg))
4742                 BUG("stat option does not begin with --stat: %s", arg);
4743         end = (char *)arg;
4744
4745         switch (*arg) {
4746         case '-':
4747                 if (skip_prefix(arg, "-width", &arg)) {
4748                         if (*arg == '=')
4749                                 width = strtoul(arg + 1, &end, 10);
4750                         else if (!*arg && !av[1])
4751                                 die_want_option("--stat-width");
4752                         else if (!*arg) {
4753                                 width = strtoul(av[1], &end, 10);
4754                                 argcount = 2;
4755                         }
4756                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4757                         if (*arg == '=')
4758                                 name_width = strtoul(arg + 1, &end, 10);
4759                         else if (!*arg && !av[1])
4760                                 die_want_option("--stat-name-width");
4761                         else if (!*arg) {
4762                                 name_width = strtoul(av[1], &end, 10);
4763                                 argcount = 2;
4764                         }
4765                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4766                         if (*arg == '=')
4767                                 graph_width = strtoul(arg + 1, &end, 10);
4768                         else if (!*arg && !av[1])
4769                                 die_want_option("--stat-graph-width");
4770                         else if (!*arg) {
4771                                 graph_width = strtoul(av[1], &end, 10);
4772                                 argcount = 2;
4773                         }
4774                 } else if (skip_prefix(arg, "-count", &arg)) {
4775                         if (*arg == '=')
4776                                 count = strtoul(arg + 1, &end, 10);
4777                         else if (!*arg && !av[1])
4778                                 die_want_option("--stat-count");
4779                         else if (!*arg) {
4780                                 count = strtoul(av[1], &end, 10);
4781                                 argcount = 2;
4782                         }
4783                 }
4784                 break;
4785         case '=':
4786                 width = strtoul(arg+1, &end, 10);
4787                 if (*end == ',')
4788                         name_width = strtoul(end+1, &end, 10);
4789                 if (*end == ',')
4790                         count = strtoul(end+1, &end, 10);
4791         }
4792
4793         /* Important! This checks all the error cases! */
4794         if (*end)
4795                 return 0;
4796         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4797         options->stat_name_width = name_width;
4798         options->stat_graph_width = graph_width;
4799         options->stat_width = width;
4800         options->stat_count = count;
4801         return argcount;
4802 }
4803
4804 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4805 {
4806         struct strbuf errmsg = STRBUF_INIT;
4807         if (parse_dirstat_params(options, params, &errmsg))
4808                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4809                     errmsg.buf);
4810         strbuf_release(&errmsg);
4811         /*
4812          * The caller knows a dirstat-related option is given from the command
4813          * line; allow it to say "return this_function();"
4814          */
4815         options->output_format |= DIFF_FORMAT_DIRSTAT;
4816         return 1;
4817 }
4818
4819 static int parse_submodule_opt(struct diff_options *options, const char *value)
4820 {
4821         if (parse_submodule_params(options, value))
4822                 die(_("Failed to parse --submodule option parameter: '%s'"),
4823                         value);
4824         return 1;
4825 }
4826
4827 static const char diff_status_letters[] = {
4828         DIFF_STATUS_ADDED,
4829         DIFF_STATUS_COPIED,
4830         DIFF_STATUS_DELETED,
4831         DIFF_STATUS_MODIFIED,
4832         DIFF_STATUS_RENAMED,
4833         DIFF_STATUS_TYPE_CHANGED,
4834         DIFF_STATUS_UNKNOWN,
4835         DIFF_STATUS_UNMERGED,
4836         DIFF_STATUS_FILTER_AON,
4837         DIFF_STATUS_FILTER_BROKEN,
4838         '\0',
4839 };
4840
4841 static unsigned int filter_bit['Z' + 1];
4842
4843 static void prepare_filter_bits(void)
4844 {
4845         int i;
4846
4847         if (!filter_bit[DIFF_STATUS_ADDED]) {
4848                 for (i = 0; diff_status_letters[i]; i++)
4849                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4850         }
4851 }
4852
4853 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4854 {
4855         return opt->filter & filter_bit[(int) status];
4856 }
4857
4858 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4859 {
4860         int i, optch;
4861
4862         prepare_filter_bits();
4863
4864         /*
4865          * If there is a negation e.g. 'd' in the input, and we haven't
4866          * initialized the filter field with another --diff-filter, start
4867          * from full set of bits, except for AON.
4868          */
4869         if (!opt->filter) {
4870                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4871                         if (optch < 'a' || 'z' < optch)
4872                                 continue;
4873                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4874                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4875                         break;
4876                 }
4877         }
4878
4879         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4880                 unsigned int bit;
4881                 int negate;
4882
4883                 if ('a' <= optch && optch <= 'z') {
4884                         negate = 1;
4885                         optch = toupper(optch);
4886                 } else {
4887                         negate = 0;
4888                 }
4889
4890                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4891                 if (!bit)
4892                         return optarg[i];
4893                 if (negate)
4894                         opt->filter &= ~bit;
4895                 else
4896                         opt->filter |= bit;
4897         }
4898         return 0;
4899 }
4900
4901 static void enable_patch_output(int *fmt)
4902 {
4903         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4904         *fmt |= DIFF_FORMAT_PATCH;
4905 }
4906
4907 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4908 {
4909         int val = parse_ws_error_highlight(arg);
4910
4911         if (val < 0) {
4912                 error("unknown value after ws-error-highlight=%.*s",
4913                       -1 - val, arg);
4914                 return 0;
4915         }
4916         opt->ws_error_highlight = val;
4917         return 1;
4918 }
4919
4920 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4921 {
4922         struct object_id oid;
4923
4924         if (get_oid(arg, &oid))
4925                 return error("unable to resolve '%s'", arg);
4926
4927         if (!opt->objfind)
4928                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4929
4930         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4931         opt->flags.recursive = 1;
4932         opt->flags.tree_in_recursive = 1;
4933         oidset_insert(opt->objfind, &oid);
4934         return 1;
4935 }
4936
4937 static int diff_opt_unified(const struct option *opt,
4938                             const char *arg, int unset)
4939 {
4940         struct diff_options *options = opt->value;
4941         char *s;
4942
4943         BUG_ON_OPT_NEG(unset);
4944
4945         options->context = strtol(arg, &s, 10);
4946         if (*s)
4947                 return error(_("%s expects a numerical value"), "--unified");
4948         enable_patch_output(&options->output_format);
4949
4950         return 0;
4951 }
4952
4953 static void prep_parse_options(struct diff_options *options)
4954 {
4955         struct option parseopts[] = {
4956                 OPT_GROUP(N_("Diff output format options")),
4957                 OPT_BITOP('p', "patch", &options->output_format,
4958                           N_("generate patch"),
4959                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4960                 OPT_BITOP('u', NULL, &options->output_format,
4961                           N_("generate patch"),
4962                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4963                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
4964                                N_("generate diffs with <n> lines context"),
4965                                PARSE_OPT_NONEG, diff_opt_unified),
4966                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
4967                          N_("generate diffs with <n> lines context")),
4968                 OPT_BIT_F(0, "raw", &options->output_format,
4969                           N_("generate the diff in raw format"),
4970                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
4971                 OPT_END()
4972         };
4973
4974         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
4975         memcpy(options->parseopts, parseopts, sizeof(parseopts));
4976 }
4977
4978 int diff_opt_parse(struct diff_options *options,
4979                    const char **av, int ac, const char *prefix)
4980 {
4981         const char *arg = av[0];
4982         const char *optarg;
4983         int argcount;
4984
4985         if (!prefix)
4986                 prefix = "";
4987
4988         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
4989                            PARSE_OPT_KEEP_DASHDASH |
4990                            PARSE_OPT_KEEP_UNKNOWN |
4991                            PARSE_OPT_NO_INTERNAL_HELP |
4992                            PARSE_OPT_ONE_SHOT |
4993                            PARSE_OPT_STOP_AT_NON_OPTION);
4994
4995         if (ac)
4996                 return ac;
4997
4998         /* Output format options */
4999         if (!strcmp(arg, "--patch-with-raw")) {
5000                 enable_patch_output(&options->output_format);
5001                 options->output_format |= DIFF_FORMAT_RAW;
5002         } else if (!strcmp(arg, "--numstat"))
5003                 options->output_format |= DIFF_FORMAT_NUMSTAT;
5004         else if (!strcmp(arg, "--shortstat"))
5005                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
5006         else if (skip_prefix(arg, "-X", &arg) ||
5007                  skip_to_optional_arg(arg, "--dirstat", &arg))
5008                 return parse_dirstat_opt(options, arg);
5009         else if (!strcmp(arg, "--cumulative"))
5010                 return parse_dirstat_opt(options, "cumulative");
5011         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
5012                 parse_dirstat_opt(options, "files");
5013                 return parse_dirstat_opt(options, arg);
5014         }
5015         else if (!strcmp(arg, "--check"))
5016                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
5017         else if (!strcmp(arg, "--summary"))
5018                 options->output_format |= DIFF_FORMAT_SUMMARY;
5019         else if (!strcmp(arg, "--patch-with-stat")) {
5020                 enable_patch_output(&options->output_format);
5021                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5022         } else if (!strcmp(arg, "--name-only"))
5023                 options->output_format |= DIFF_FORMAT_NAME;
5024         else if (!strcmp(arg, "--name-status"))
5025                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
5026         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
5027                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
5028         else if (starts_with(arg, "--stat"))
5029                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
5030                 return stat_opt(options, av);
5031         else if (!strcmp(arg, "--compact-summary")) {
5032                  options->flags.stat_with_summary = 1;
5033                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
5034         } else if (!strcmp(arg, "--no-compact-summary"))
5035                  options->flags.stat_with_summary = 0;
5036         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
5037                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
5038         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
5039                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
5040         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
5041                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
5042
5043         /* renames options */
5044         else if (starts_with(arg, "-B") ||
5045                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
5046                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
5047                         return error("invalid argument to -B: %s", arg+2);
5048         }
5049         else if (starts_with(arg, "-M") ||
5050                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
5051                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5052                         return error("invalid argument to -M: %s", arg+2);
5053                 options->detect_rename = DIFF_DETECT_RENAME;
5054         }
5055         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
5056                 options->irreversible_delete = 1;
5057         }
5058         else if (starts_with(arg, "-C") ||
5059                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
5060                 if (options->detect_rename == DIFF_DETECT_COPY)
5061                         options->flags.find_copies_harder = 1;
5062                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5063                         return error("invalid argument to -C: %s", arg+2);
5064                 options->detect_rename = DIFF_DETECT_COPY;
5065         }
5066         else if (!strcmp(arg, "--no-renames"))
5067                 options->detect_rename = 0;
5068         else if (!strcmp(arg, "--rename-empty"))
5069                 options->flags.rename_empty = 1;
5070         else if (!strcmp(arg, "--no-rename-empty"))
5071                 options->flags.rename_empty = 0;
5072         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5073                 options->flags.relative_name = 1;
5074                 if (arg)
5075                         options->prefix = arg;
5076         }
5077
5078         /* xdiff options */
5079         else if (!strcmp(arg, "--minimal"))
5080                 DIFF_XDL_SET(options, NEED_MINIMAL);
5081         else if (!strcmp(arg, "--no-minimal"))
5082                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5083         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5084                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5085         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5086                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5087         else if (!strcmp(arg, "--ignore-space-at-eol"))
5088                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5089         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5090                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5091         else if (!strcmp(arg, "--ignore-blank-lines"))
5092                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5093         else if (!strcmp(arg, "--indent-heuristic"))
5094                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5095         else if (!strcmp(arg, "--no-indent-heuristic"))
5096                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5097         else if (!strcmp(arg, "--patience")) {
5098                 int i;
5099                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5100                 /*
5101                  * Both --patience and --anchored use PATIENCE_DIFF
5102                  * internally, so remove any anchors previously
5103                  * specified.
5104                  */
5105                 for (i = 0; i < options->anchors_nr; i++)
5106                         free(options->anchors[i]);
5107                 options->anchors_nr = 0;
5108         } else if (!strcmp(arg, "--histogram"))
5109                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5110         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5111                 long value = parse_algorithm_value(optarg);
5112                 if (value < 0)
5113                         return error("option diff-algorithm accepts \"myers\", "
5114                                      "\"minimal\", \"patience\" and \"histogram\"");
5115                 /* clear out previous settings */
5116                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5117                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5118                 options->xdl_opts |= value;
5119                 return argcount;
5120         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5121                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5122                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5123                            options->anchors_alloc);
5124                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5125         }
5126
5127         /* flags options */
5128         else if (!strcmp(arg, "--binary")) {
5129                 enable_patch_output(&options->output_format);
5130                 options->flags.binary = 1;
5131         }
5132         else if (!strcmp(arg, "--full-index"))
5133                 options->flags.full_index = 1;
5134         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5135                 options->flags.text = 1;
5136         else if (!strcmp(arg, "-R"))
5137                 options->flags.reverse_diff = 1;
5138         else if (!strcmp(arg, "--find-copies-harder"))
5139                 options->flags.find_copies_harder = 1;
5140         else if (!strcmp(arg, "--follow"))
5141                 options->flags.follow_renames = 1;
5142         else if (!strcmp(arg, "--no-follow")) {
5143                 options->flags.follow_renames = 0;
5144                 options->flags.default_follow_renames = 0;
5145         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5146                 int value = git_config_colorbool(NULL, arg);
5147                 if (value < 0)
5148                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5149                 options->use_color = value;
5150         }
5151         else if (!strcmp(arg, "--no-color"))
5152                 options->use_color = 0;
5153         else if (!strcmp(arg, "--color-moved")) {
5154                 if (diff_color_moved_default)
5155                         options->color_moved = diff_color_moved_default;
5156                 if (options->color_moved == COLOR_MOVED_NO)
5157                         options->color_moved = COLOR_MOVED_DEFAULT;
5158         } else if (!strcmp(arg, "--no-color-moved"))
5159                 options->color_moved = COLOR_MOVED_NO;
5160         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5161                 int cm = parse_color_moved(arg);
5162                 if (cm < 0)
5163                         return error("bad --color-moved argument: %s", arg);
5164                 options->color_moved = cm;
5165         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5166                 options->color_moved_ws_handling = 0;
5167         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5168                 unsigned cm = parse_color_moved_ws(arg);
5169                 if (cm & COLOR_MOVED_WS_ERROR)
5170                         return -1;
5171                 options->color_moved_ws_handling = cm;
5172         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5173                 options->use_color = 1;
5174                 options->word_diff = DIFF_WORDS_COLOR;
5175         }
5176         else if (!strcmp(arg, "--word-diff")) {
5177                 if (options->word_diff == DIFF_WORDS_NONE)
5178                         options->word_diff = DIFF_WORDS_PLAIN;
5179         }
5180         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5181                 if (!strcmp(arg, "plain"))
5182                         options->word_diff = DIFF_WORDS_PLAIN;
5183                 else if (!strcmp(arg, "color")) {
5184                         options->use_color = 1;
5185                         options->word_diff = DIFF_WORDS_COLOR;
5186                 }
5187                 else if (!strcmp(arg, "porcelain"))
5188                         options->word_diff = DIFF_WORDS_PORCELAIN;
5189                 else if (!strcmp(arg, "none"))
5190                         options->word_diff = DIFF_WORDS_NONE;
5191                 else
5192                         die("bad --word-diff argument: %s", arg);
5193         }
5194         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5195                 if (options->word_diff == DIFF_WORDS_NONE)
5196                         options->word_diff = DIFF_WORDS_PLAIN;
5197                 options->word_regex = optarg;
5198                 return argcount;
5199         }
5200         else if (!strcmp(arg, "--exit-code"))
5201                 options->flags.exit_with_status = 1;
5202         else if (!strcmp(arg, "--quiet"))
5203                 options->flags.quick = 1;
5204         else if (!strcmp(arg, "--ext-diff"))
5205                 options->flags.allow_external = 1;
5206         else if (!strcmp(arg, "--no-ext-diff"))
5207                 options->flags.allow_external = 0;
5208         else if (!strcmp(arg, "--textconv")) {
5209                 options->flags.allow_textconv = 1;
5210                 options->flags.textconv_set_via_cmdline = 1;
5211         } else if (!strcmp(arg, "--no-textconv"))
5212                 options->flags.allow_textconv = 0;
5213         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5214                 options->flags.override_submodule_config = 1;
5215                 handle_ignore_submodules_arg(options, arg);
5216         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5217                 return parse_submodule_opt(options, arg);
5218         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5219                 return parse_ws_error_highlight_opt(options, arg);
5220         else if (!strcmp(arg, "--ita-invisible-in-index"))
5221                 options->ita_invisible_in_index = 1;
5222         else if (!strcmp(arg, "--ita-visible-in-index"))
5223                 options->ita_invisible_in_index = 0;
5224
5225         /* misc options */
5226         else if (!strcmp(arg, "-z"))
5227                 options->line_termination = 0;
5228         else if ((argcount = short_opt('l', av, &optarg))) {
5229                 options->rename_limit = strtoul(optarg, NULL, 10);
5230                 return argcount;
5231         }
5232         else if ((argcount = short_opt('S', av, &optarg))) {
5233                 options->pickaxe = optarg;
5234                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5235                 return argcount;
5236         } else if ((argcount = short_opt('G', av, &optarg))) {
5237                 options->pickaxe = optarg;
5238                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5239                 return argcount;
5240         }
5241         else if (!strcmp(arg, "--pickaxe-all"))
5242                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5243         else if (!strcmp(arg, "--pickaxe-regex"))
5244                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5245         else if ((argcount = short_opt('O', av, &optarg))) {
5246                 options->orderfile = prefix_filename(prefix, optarg);
5247                 return argcount;
5248         } else if (skip_prefix(arg, "--find-object=", &arg))
5249                 return parse_objfind_opt(options, arg);
5250         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5251                 int offending = parse_diff_filter_opt(optarg, options);
5252                 if (offending)
5253                         die("unknown change class '%c' in --diff-filter=%s",
5254                             offending, optarg);
5255                 return argcount;
5256         }
5257         else if (!strcmp(arg, "--no-abbrev"))
5258                 options->abbrev = 0;
5259         else if (!strcmp(arg, "--abbrev"))
5260                 options->abbrev = DEFAULT_ABBREV;
5261         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5262                 options->abbrev = strtoul(arg, NULL, 10);
5263                 if (options->abbrev < MINIMUM_ABBREV)
5264                         options->abbrev = MINIMUM_ABBREV;
5265                 else if (the_hash_algo->hexsz < options->abbrev)
5266                         options->abbrev = the_hash_algo->hexsz;
5267         }
5268         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5269                 options->a_prefix = optarg;
5270                 return argcount;
5271         }
5272         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5273                 options->line_prefix = optarg;
5274                 options->line_prefix_length = strlen(options->line_prefix);
5275                 graph_setup_line_prefix(options);
5276                 return argcount;
5277         }
5278         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5279                 options->b_prefix = optarg;
5280                 return argcount;
5281         }
5282         else if (!strcmp(arg, "--no-prefix"))
5283                 options->a_prefix = options->b_prefix = "";
5284         else if (opt_arg(arg, '\0', "inter-hunk-context",
5285                          &options->interhunkcontext))
5286                 ;
5287         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5288                 char *path = prefix_filename(prefix, optarg);
5289                 options->file = xfopen(path, "w");
5290                 options->close_file = 1;
5291                 if (options->use_color != GIT_COLOR_ALWAYS)
5292                         options->use_color = GIT_COLOR_NEVER;
5293                 free(path);
5294                 return argcount;
5295         } else
5296                 return 0;
5297         return 1;
5298 }
5299
5300 int parse_rename_score(const char **cp_p)
5301 {
5302         unsigned long num, scale;
5303         int ch, dot;
5304         const char *cp = *cp_p;
5305
5306         num = 0;
5307         scale = 1;
5308         dot = 0;
5309         for (;;) {
5310                 ch = *cp;
5311                 if ( !dot && ch == '.' ) {
5312                         scale = 1;
5313                         dot = 1;
5314                 } else if ( ch == '%' ) {
5315                         scale = dot ? scale*100 : 100;
5316                         cp++;   /* % is always at the end */
5317                         break;
5318                 } else if ( ch >= '0' && ch <= '9' ) {
5319                         if ( scale < 100000 ) {
5320                                 scale *= 10;
5321                                 num = (num*10) + (ch-'0');
5322                         }
5323                 } else {
5324                         break;
5325                 }
5326                 cp++;
5327         }
5328         *cp_p = cp;
5329
5330         /* user says num divided by scale and we say internally that
5331          * is MAX_SCORE * num / scale.
5332          */
5333         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5334 }
5335
5336 static int diff_scoreopt_parse(const char *opt)
5337 {
5338         int opt1, opt2, cmd;
5339
5340         if (*opt++ != '-')
5341                 return -1;
5342         cmd = *opt++;
5343         if (cmd == '-') {
5344                 /* convert the long-form arguments into short-form versions */
5345                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5346                         if (*opt == 0 || *opt++ == '=')
5347                                 cmd = 'B';
5348                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5349                         if (*opt == 0 || *opt++ == '=')
5350                                 cmd = 'C';
5351                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5352                         if (*opt == 0 || *opt++ == '=')
5353                                 cmd = 'M';
5354                 }
5355         }
5356         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5357                 return -1; /* that is not a -M, -C, or -B option */
5358
5359         opt1 = parse_rename_score(&opt);
5360         if (cmd != 'B')
5361                 opt2 = 0;
5362         else {
5363                 if (*opt == 0)
5364                         opt2 = 0;
5365                 else if (*opt != '/')
5366                         return -1; /* we expect -B80/99 or -B80 */
5367                 else {
5368                         opt++;
5369                         opt2 = parse_rename_score(&opt);
5370                 }
5371         }
5372         if (*opt != 0)
5373                 return -1;
5374         return opt1 | (opt2 << 16);
5375 }
5376
5377 struct diff_queue_struct diff_queued_diff;
5378
5379 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5380 {
5381         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5382         queue->queue[queue->nr++] = dp;
5383 }
5384
5385 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5386                                  struct diff_filespec *one,
5387                                  struct diff_filespec *two)
5388 {
5389         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5390         dp->one = one;
5391         dp->two = two;
5392         if (queue)
5393                 diff_q(queue, dp);
5394         return dp;
5395 }
5396
5397 void diff_free_filepair(struct diff_filepair *p)
5398 {
5399         free_filespec(p->one);
5400         free_filespec(p->two);
5401         free(p);
5402 }
5403
5404 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5405 {
5406         int abblen;
5407         const char *abbrev;
5408
5409         /* Do we want all 40 hex characters? */
5410         if (len == the_hash_algo->hexsz)
5411                 return oid_to_hex(oid);
5412
5413         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5414         abbrev = diff_abbrev_oid(oid, len);
5415
5416         if (!print_sha1_ellipsis())
5417                 return abbrev;
5418
5419         abblen = strlen(abbrev);
5420
5421         /*
5422          * In well-behaved cases, where the abbreviated result is the
5423          * same as the requested length, append three dots after the
5424          * abbreviation (hence the whole logic is limited to the case
5425          * where abblen < 37); when the actual abbreviated result is a
5426          * bit longer than the requested length, we reduce the number
5427          * of dots so that they match the well-behaved ones.  However,
5428          * if the actual abbreviation is longer than the requested
5429          * length by more than three, we give up on aligning, and add
5430          * three dots anyway, to indicate that the output is not the
5431          * full object name.  Yes, this may be suboptimal, but this
5432          * appears only in "diff --raw --abbrev" output and it is not
5433          * worth the effort to change it now.  Note that this would
5434          * likely to work fine when the automatic sizing of default
5435          * abbreviation length is used--we would be fed -1 in "len" in
5436          * that case, and will end up always appending three-dots, but
5437          * the automatic sizing is supposed to give abblen that ensures
5438          * uniqueness across all objects (statistically speaking).
5439          */
5440         if (abblen < the_hash_algo->hexsz - 3) {
5441                 static char hex[GIT_MAX_HEXSZ + 1];
5442                 if (len < abblen && abblen <= len + 2)
5443                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5444                 else
5445                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5446                 return hex;
5447         }
5448
5449         return oid_to_hex(oid);
5450 }
5451
5452 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5453 {
5454         int line_termination = opt->line_termination;
5455         int inter_name_termination = line_termination ? '\t' : '\0';
5456
5457         fprintf(opt->file, "%s", diff_line_prefix(opt));
5458         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5459                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5460                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5461                 fprintf(opt->file, "%s ",
5462                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5463         }
5464         if (p->score) {
5465                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5466                         inter_name_termination);
5467         } else {
5468                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5469         }
5470
5471         if (p->status == DIFF_STATUS_COPIED ||
5472             p->status == DIFF_STATUS_RENAMED) {
5473                 const char *name_a, *name_b;
5474                 name_a = p->one->path;
5475                 name_b = p->two->path;
5476                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5477                 write_name_quoted(name_a, opt->file, inter_name_termination);
5478                 write_name_quoted(name_b, opt->file, line_termination);
5479         } else {
5480                 const char *name_a, *name_b;
5481                 name_a = p->one->mode ? p->one->path : p->two->path;
5482                 name_b = NULL;
5483                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5484                 write_name_quoted(name_a, opt->file, line_termination);
5485         }
5486 }
5487
5488 int diff_unmodified_pair(struct diff_filepair *p)
5489 {
5490         /* This function is written stricter than necessary to support
5491          * the currently implemented transformers, but the idea is to
5492          * let transformers to produce diff_filepairs any way they want,
5493          * and filter and clean them up here before producing the output.
5494          */
5495         struct diff_filespec *one = p->one, *two = p->two;
5496
5497         if (DIFF_PAIR_UNMERGED(p))
5498                 return 0; /* unmerged is interesting */
5499
5500         /* deletion, addition, mode or type change
5501          * and rename are all interesting.
5502          */
5503         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5504             DIFF_PAIR_MODE_CHANGED(p) ||
5505             strcmp(one->path, two->path))
5506                 return 0;
5507
5508         /* both are valid and point at the same path.  that is, we are
5509          * dealing with a change.
5510          */
5511         if (one->oid_valid && two->oid_valid &&
5512             oideq(&one->oid, &two->oid) &&
5513             !one->dirty_submodule && !two->dirty_submodule)
5514                 return 1; /* no change */
5515         if (!one->oid_valid && !two->oid_valid)
5516                 return 1; /* both look at the same file on the filesystem. */
5517         return 0;
5518 }
5519
5520 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5521 {
5522         if (diff_unmodified_pair(p))
5523                 return;
5524
5525         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5526             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5527                 return; /* no tree diffs in patch format */
5528
5529         run_diff(p, o);
5530 }
5531
5532 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5533                             struct diffstat_t *diffstat)
5534 {
5535         if (diff_unmodified_pair(p))
5536                 return;
5537
5538         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5539             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5540                 return; /* no useful stat for tree diffs */
5541
5542         run_diffstat(p, o, diffstat);
5543 }
5544
5545 static void diff_flush_checkdiff(struct diff_filepair *p,
5546                 struct diff_options *o)
5547 {
5548         if (diff_unmodified_pair(p))
5549                 return;
5550
5551         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5552             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5553                 return; /* nothing to check in tree diffs */
5554
5555         run_checkdiff(p, o);
5556 }
5557
5558 int diff_queue_is_empty(void)
5559 {
5560         struct diff_queue_struct *q = &diff_queued_diff;
5561         int i;
5562         for (i = 0; i < q->nr; i++)
5563                 if (!diff_unmodified_pair(q->queue[i]))
5564                         return 0;
5565         return 1;
5566 }
5567
5568 #if DIFF_DEBUG
5569 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5570 {
5571         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5572                 x, one ? one : "",
5573                 s->path,
5574                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5575                 s->mode,
5576                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5577         fprintf(stderr, "queue[%d] %s size %lu\n",
5578                 x, one ? one : "",
5579                 s->size);
5580 }
5581
5582 void diff_debug_filepair(const struct diff_filepair *p, int i)
5583 {
5584         diff_debug_filespec(p->one, i, "one");
5585         diff_debug_filespec(p->two, i, "two");
5586         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5587                 p->score, p->status ? p->status : '?',
5588                 p->one->rename_used, p->broken_pair);
5589 }
5590
5591 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5592 {
5593         int i;
5594         if (msg)
5595                 fprintf(stderr, "%s\n", msg);
5596         fprintf(stderr, "q->nr = %d\n", q->nr);
5597         for (i = 0; i < q->nr; i++) {
5598                 struct diff_filepair *p = q->queue[i];
5599                 diff_debug_filepair(p, i);
5600         }
5601 }
5602 #endif
5603
5604 static void diff_resolve_rename_copy(void)
5605 {
5606         int i;
5607         struct diff_filepair *p;
5608         struct diff_queue_struct *q = &diff_queued_diff;
5609
5610         diff_debug_queue("resolve-rename-copy", q);
5611
5612         for (i = 0; i < q->nr; i++) {
5613                 p = q->queue[i];
5614                 p->status = 0; /* undecided */
5615                 if (DIFF_PAIR_UNMERGED(p))
5616                         p->status = DIFF_STATUS_UNMERGED;
5617                 else if (!DIFF_FILE_VALID(p->one))
5618                         p->status = DIFF_STATUS_ADDED;
5619                 else if (!DIFF_FILE_VALID(p->two))
5620                         p->status = DIFF_STATUS_DELETED;
5621                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5622                         p->status = DIFF_STATUS_TYPE_CHANGED;
5623
5624                 /* from this point on, we are dealing with a pair
5625                  * whose both sides are valid and of the same type, i.e.
5626                  * either in-place edit or rename/copy edit.
5627                  */
5628                 else if (DIFF_PAIR_RENAME(p)) {
5629                         /*
5630                          * A rename might have re-connected a broken
5631                          * pair up, causing the pathnames to be the
5632                          * same again. If so, that's not a rename at
5633                          * all, just a modification..
5634                          *
5635                          * Otherwise, see if this source was used for
5636                          * multiple renames, in which case we decrement
5637                          * the count, and call it a copy.
5638                          */
5639                         if (!strcmp(p->one->path, p->two->path))
5640                                 p->status = DIFF_STATUS_MODIFIED;
5641                         else if (--p->one->rename_used > 0)
5642                                 p->status = DIFF_STATUS_COPIED;
5643                         else
5644                                 p->status = DIFF_STATUS_RENAMED;
5645                 }
5646                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5647                          p->one->mode != p->two->mode ||
5648                          p->one->dirty_submodule ||
5649                          p->two->dirty_submodule ||
5650                          is_null_oid(&p->one->oid))
5651                         p->status = DIFF_STATUS_MODIFIED;
5652                 else {
5653                         /* This is a "no-change" entry and should not
5654                          * happen anymore, but prepare for broken callers.
5655                          */
5656                         error("feeding unmodified %s to diffcore",
5657                               p->one->path);
5658                         p->status = DIFF_STATUS_UNKNOWN;
5659                 }
5660         }
5661         diff_debug_queue("resolve-rename-copy done", q);
5662 }
5663
5664 static int check_pair_status(struct diff_filepair *p)
5665 {
5666         switch (p->status) {
5667         case DIFF_STATUS_UNKNOWN:
5668                 return 0;
5669         case 0:
5670                 die("internal error in diff-resolve-rename-copy");
5671         default:
5672                 return 1;
5673         }
5674 }
5675
5676 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5677 {
5678         int fmt = opt->output_format;
5679
5680         if (fmt & DIFF_FORMAT_CHECKDIFF)
5681                 diff_flush_checkdiff(p, opt);
5682         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5683                 diff_flush_raw(p, opt);
5684         else if (fmt & DIFF_FORMAT_NAME) {
5685                 const char *name_a, *name_b;
5686                 name_a = p->two->path;
5687                 name_b = NULL;
5688                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5689                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5690                 write_name_quoted(name_a, opt->file, opt->line_termination);
5691         }
5692 }
5693
5694 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5695 {
5696         struct strbuf sb = STRBUF_INIT;
5697         if (fs->mode)
5698                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5699         else
5700                 strbuf_addf(&sb, " %s ", newdelete);
5701
5702         quote_c_style(fs->path, &sb, NULL, 0);
5703         strbuf_addch(&sb, '\n');
5704         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5705                          sb.buf, sb.len, 0);
5706         strbuf_release(&sb);
5707 }
5708
5709 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5710                 int show_name)
5711 {
5712         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5713                 struct strbuf sb = STRBUF_INIT;
5714                 strbuf_addf(&sb, " mode change %06o => %06o",
5715                             p->one->mode, p->two->mode);
5716                 if (show_name) {
5717                         strbuf_addch(&sb, ' ');
5718                         quote_c_style(p->two->path, &sb, NULL, 0);
5719                 }
5720                 strbuf_addch(&sb, '\n');
5721                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5722                                  sb.buf, sb.len, 0);
5723                 strbuf_release(&sb);
5724         }
5725 }
5726
5727 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5728                 struct diff_filepair *p)
5729 {
5730         struct strbuf sb = STRBUF_INIT;
5731         struct strbuf names = STRBUF_INIT;
5732
5733         pprint_rename(&names, p->one->path, p->two->path);
5734         strbuf_addf(&sb, " %s %s (%d%%)\n",
5735                     renamecopy, names.buf, similarity_index(p));
5736         strbuf_release(&names);
5737         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5738                                  sb.buf, sb.len, 0);
5739         show_mode_change(opt, p, 0);
5740         strbuf_release(&sb);
5741 }
5742
5743 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5744 {
5745         switch(p->status) {
5746         case DIFF_STATUS_DELETED:
5747                 show_file_mode_name(opt, "delete", p->one);
5748                 break;
5749         case DIFF_STATUS_ADDED:
5750                 show_file_mode_name(opt, "create", p->two);
5751                 break;
5752         case DIFF_STATUS_COPIED:
5753                 show_rename_copy(opt, "copy", p);
5754                 break;
5755         case DIFF_STATUS_RENAMED:
5756                 show_rename_copy(opt, "rename", p);
5757                 break;
5758         default:
5759                 if (p->score) {
5760                         struct strbuf sb = STRBUF_INIT;
5761                         strbuf_addstr(&sb, " rewrite ");
5762                         quote_c_style(p->two->path, &sb, NULL, 0);
5763                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5764                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5765                                          sb.buf, sb.len, 0);
5766                         strbuf_release(&sb);
5767                 }
5768                 show_mode_change(opt, p, !p->score);
5769                 break;
5770         }
5771 }
5772
5773 struct patch_id_t {
5774         git_SHA_CTX *ctx;
5775         int patchlen;
5776 };
5777
5778 static int remove_space(char *line, int len)
5779 {
5780         int i;
5781         char *dst = line;
5782         unsigned char c;
5783
5784         for (i = 0; i < len; i++)
5785                 if (!isspace((c = line[i])))
5786                         *dst++ = c;
5787
5788         return dst - line;
5789 }
5790
5791 static void patch_id_consume(void *priv, char *line, unsigned long len)
5792 {
5793         struct patch_id_t *data = priv;
5794         int new_len;
5795
5796         new_len = remove_space(line, len);
5797
5798         git_SHA1_Update(data->ctx, line, new_len);
5799         data->patchlen += new_len;
5800 }
5801
5802 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5803 {
5804         git_SHA1_Update(ctx, str, strlen(str));
5805 }
5806
5807 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5808 {
5809         /* large enough for 2^32 in octal */
5810         char buf[12];
5811         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5812         git_SHA1_Update(ctx, buf, len);
5813 }
5814
5815 /* returns 0 upon success, and writes result into sha1 */
5816 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5817 {
5818         struct diff_queue_struct *q = &diff_queued_diff;
5819         int i;
5820         git_SHA_CTX ctx;
5821         struct patch_id_t data;
5822
5823         git_SHA1_Init(&ctx);
5824         memset(&data, 0, sizeof(struct patch_id_t));
5825         data.ctx = &ctx;
5826
5827         for (i = 0; i < q->nr; i++) {
5828                 xpparam_t xpp;
5829                 xdemitconf_t xecfg;
5830                 mmfile_t mf1, mf2;
5831                 struct diff_filepair *p = q->queue[i];
5832                 int len1, len2;
5833
5834                 memset(&xpp, 0, sizeof(xpp));
5835                 memset(&xecfg, 0, sizeof(xecfg));
5836                 if (p->status == 0)
5837                         return error("internal diff status error");
5838                 if (p->status == DIFF_STATUS_UNKNOWN)
5839                         continue;
5840                 if (diff_unmodified_pair(p))
5841                         continue;
5842                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5843                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5844                         continue;
5845                 if (DIFF_PAIR_UNMERGED(p))
5846                         continue;
5847
5848                 diff_fill_oid_info(p->one, options->repo->index);
5849                 diff_fill_oid_info(p->two, options->repo->index);
5850
5851                 len1 = remove_space(p->one->path, strlen(p->one->path));
5852                 len2 = remove_space(p->two->path, strlen(p->two->path));
5853                 patch_id_add_string(&ctx, "diff--git");
5854                 patch_id_add_string(&ctx, "a/");
5855                 git_SHA1_Update(&ctx, p->one->path, len1);
5856                 patch_id_add_string(&ctx, "b/");
5857                 git_SHA1_Update(&ctx, p->two->path, len2);
5858
5859                 if (p->one->mode == 0) {
5860                         patch_id_add_string(&ctx, "newfilemode");
5861                         patch_id_add_mode(&ctx, p->two->mode);
5862                         patch_id_add_string(&ctx, "---/dev/null");
5863                         patch_id_add_string(&ctx, "+++b/");
5864                         git_SHA1_Update(&ctx, p->two->path, len2);
5865                 } else if (p->two->mode == 0) {
5866                         patch_id_add_string(&ctx, "deletedfilemode");
5867                         patch_id_add_mode(&ctx, p->one->mode);
5868                         patch_id_add_string(&ctx, "---a/");
5869                         git_SHA1_Update(&ctx, p->one->path, len1);
5870                         patch_id_add_string(&ctx, "+++/dev/null");
5871                 } else {
5872                         patch_id_add_string(&ctx, "---a/");
5873                         git_SHA1_Update(&ctx, p->one->path, len1);
5874                         patch_id_add_string(&ctx, "+++b/");
5875                         git_SHA1_Update(&ctx, p->two->path, len2);
5876                 }
5877
5878                 if (diff_header_only)
5879                         continue;
5880
5881                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5882                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5883                         return error("unable to read files to diff");
5884
5885                 if (diff_filespec_is_binary(options->repo, p->one) ||
5886                     diff_filespec_is_binary(options->repo, p->two)) {
5887                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5888                                         GIT_SHA1_HEXSZ);
5889                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5890                                         GIT_SHA1_HEXSZ);
5891                         continue;
5892                 }
5893
5894                 xpp.flags = 0;
5895                 xecfg.ctxlen = 3;
5896                 xecfg.flags = 0;
5897                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5898                                   patch_id_consume, &data, &xpp, &xecfg))
5899                         return error("unable to generate patch-id diff for %s",
5900                                      p->one->path);
5901         }
5902
5903         git_SHA1_Final(oid->hash, &ctx);
5904         return 0;
5905 }
5906
5907 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5908 {
5909         struct diff_queue_struct *q = &diff_queued_diff;
5910         int i;
5911         int result = diff_get_patch_id(options, oid, diff_header_only);
5912
5913         for (i = 0; i < q->nr; i++)
5914                 diff_free_filepair(q->queue[i]);
5915
5916         free(q->queue);
5917         DIFF_QUEUE_CLEAR(q);
5918
5919         return result;
5920 }
5921
5922 static int is_summary_empty(const struct diff_queue_struct *q)
5923 {
5924         int i;
5925
5926         for (i = 0; i < q->nr; i++) {
5927                 const struct diff_filepair *p = q->queue[i];
5928
5929                 switch (p->status) {
5930                 case DIFF_STATUS_DELETED:
5931                 case DIFF_STATUS_ADDED:
5932                 case DIFF_STATUS_COPIED:
5933                 case DIFF_STATUS_RENAMED:
5934                         return 0;
5935                 default:
5936                         if (p->score)
5937                                 return 0;
5938                         if (p->one->mode && p->two->mode &&
5939                             p->one->mode != p->two->mode)
5940                                 return 0;
5941                         break;
5942                 }
5943         }
5944         return 1;
5945 }
5946
5947 static const char rename_limit_warning[] =
5948 N_("inexact rename detection was skipped due to too many files.");
5949
5950 static const char degrade_cc_to_c_warning[] =
5951 N_("only found copies from modified paths due to too many files.");
5952
5953 static const char rename_limit_advice[] =
5954 N_("you may want to set your %s variable to at least "
5955    "%d and retry the command.");
5956
5957 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5958 {
5959         fflush(stdout);
5960         if (degraded_cc)
5961                 warning(_(degrade_cc_to_c_warning));
5962         else if (needed)
5963                 warning(_(rename_limit_warning));
5964         else
5965                 return;
5966         if (0 < needed)
5967                 warning(_(rename_limit_advice), varname, needed);
5968 }
5969
5970 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5971 {
5972         int i;
5973         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5974         struct diff_queue_struct *q = &diff_queued_diff;
5975
5976         if (WSEH_NEW & WS_RULE_MASK)
5977                 BUG("WS rules bit mask overlaps with diff symbol flags");
5978
5979         if (o->color_moved)
5980                 o->emitted_symbols = &esm;
5981
5982         for (i = 0; i < q->nr; i++) {
5983                 struct diff_filepair *p = q->queue[i];
5984                 if (check_pair_status(p))
5985                         diff_flush_patch(p, o);
5986         }
5987
5988         if (o->emitted_symbols) {
5989                 if (o->color_moved) {
5990                         struct hashmap add_lines, del_lines;
5991
5992                         if (o->color_moved_ws_handling &
5993                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5994                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5995
5996                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5997                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5998
5999                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6000                         mark_color_as_moved(o, &add_lines, &del_lines);
6001                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6002                                 dim_moved_lines(o);
6003
6004                         hashmap_free(&add_lines, 1);
6005                         hashmap_free(&del_lines, 1);
6006                 }
6007
6008                 for (i = 0; i < esm.nr; i++)
6009                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6010
6011                 for (i = 0; i < esm.nr; i++)
6012                         free((void *)esm.buf[i].line);
6013                 esm.nr = 0;
6014
6015                 o->emitted_symbols = NULL;
6016         }
6017 }
6018
6019 void diff_flush(struct diff_options *options)
6020 {
6021         struct diff_queue_struct *q = &diff_queued_diff;
6022         int i, output_format = options->output_format;
6023         int separator = 0;
6024         int dirstat_by_line = 0;
6025
6026         /*
6027          * Order: raw, stat, summary, patch
6028          * or:    name/name-status/checkdiff (other bits clear)
6029          */
6030         if (!q->nr)
6031                 goto free_queue;
6032
6033         if (output_format & (DIFF_FORMAT_RAW |
6034                              DIFF_FORMAT_NAME |
6035                              DIFF_FORMAT_NAME_STATUS |
6036                              DIFF_FORMAT_CHECKDIFF)) {
6037                 for (i = 0; i < q->nr; i++) {
6038                         struct diff_filepair *p = q->queue[i];
6039                         if (check_pair_status(p))
6040                                 flush_one_pair(p, options);
6041                 }
6042                 separator++;
6043         }
6044
6045         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6046                 dirstat_by_line = 1;
6047
6048         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6049             dirstat_by_line) {
6050                 struct diffstat_t diffstat;
6051
6052                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6053                 for (i = 0; i < q->nr; i++) {
6054                         struct diff_filepair *p = q->queue[i];
6055                         if (check_pair_status(p))
6056                                 diff_flush_stat(p, options, &diffstat);
6057                 }
6058                 if (output_format & DIFF_FORMAT_NUMSTAT)
6059                         show_numstat(&diffstat, options);
6060                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6061                         show_stats(&diffstat, options);
6062                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6063                         show_shortstats(&diffstat, options);
6064                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6065                         show_dirstat_by_line(&diffstat, options);
6066                 free_diffstat_info(&diffstat);
6067                 separator++;
6068         }
6069         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6070                 show_dirstat(options);
6071
6072         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6073                 for (i = 0; i < q->nr; i++) {
6074                         diff_summary(options, q->queue[i]);
6075                 }
6076                 separator++;
6077         }
6078
6079         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6080             options->flags.exit_with_status &&
6081             options->flags.diff_from_contents) {
6082                 /*
6083                  * run diff_flush_patch for the exit status. setting
6084                  * options->file to /dev/null should be safe, because we
6085                  * aren't supposed to produce any output anyway.
6086                  */
6087                 if (options->close_file)
6088                         fclose(options->file);
6089                 options->file = xfopen("/dev/null", "w");
6090                 options->close_file = 1;
6091                 options->color_moved = 0;
6092                 for (i = 0; i < q->nr; i++) {
6093                         struct diff_filepair *p = q->queue[i];
6094                         if (check_pair_status(p))
6095                                 diff_flush_patch(p, options);
6096                         if (options->found_changes)
6097                                 break;
6098                 }
6099         }
6100
6101         if (output_format & DIFF_FORMAT_PATCH) {
6102                 if (separator) {
6103                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6104                         if (options->stat_sep)
6105                                 /* attach patch instead of inline */
6106                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6107                                                  NULL, 0, 0);
6108                 }
6109
6110                 diff_flush_patch_all_file_pairs(options);
6111         }
6112
6113         if (output_format & DIFF_FORMAT_CALLBACK)
6114                 options->format_callback(q, options, options->format_callback_data);
6115
6116         for (i = 0; i < q->nr; i++)
6117                 diff_free_filepair(q->queue[i]);
6118 free_queue:
6119         free(q->queue);
6120         DIFF_QUEUE_CLEAR(q);
6121         if (options->close_file)
6122                 fclose(options->file);
6123
6124         /*
6125          * Report the content-level differences with HAS_CHANGES;
6126          * diff_addremove/diff_change does not set the bit when
6127          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6128          */
6129         if (options->flags.diff_from_contents) {
6130                 if (options->found_changes)
6131                         options->flags.has_changes = 1;
6132                 else
6133                         options->flags.has_changes = 0;
6134         }
6135 }
6136
6137 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6138 {
6139         return (((p->status == DIFF_STATUS_MODIFIED) &&
6140                  ((p->score &&
6141                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6142                   (!p->score &&
6143                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6144                 ((p->status != DIFF_STATUS_MODIFIED) &&
6145                  filter_bit_tst(p->status, options)));
6146 }
6147
6148 static void diffcore_apply_filter(struct diff_options *options)
6149 {
6150         int i;
6151         struct diff_queue_struct *q = &diff_queued_diff;
6152         struct diff_queue_struct outq;
6153
6154         DIFF_QUEUE_CLEAR(&outq);
6155
6156         if (!options->filter)
6157                 return;
6158
6159         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6160                 int found;
6161                 for (i = found = 0; !found && i < q->nr; i++) {
6162                         if (match_filter(options, q->queue[i]))
6163                                 found++;
6164                 }
6165                 if (found)
6166                         return;
6167
6168                 /* otherwise we will clear the whole queue
6169                  * by copying the empty outq at the end of this
6170                  * function, but first clear the current entries
6171                  * in the queue.
6172                  */
6173                 for (i = 0; i < q->nr; i++)
6174                         diff_free_filepair(q->queue[i]);
6175         }
6176         else {
6177                 /* Only the matching ones */
6178                 for (i = 0; i < q->nr; i++) {
6179                         struct diff_filepair *p = q->queue[i];
6180                         if (match_filter(options, p))
6181                                 diff_q(&outq, p);
6182                         else
6183                                 diff_free_filepair(p);
6184                 }
6185         }
6186         free(q->queue);
6187         *q = outq;
6188 }
6189
6190 /* Check whether two filespecs with the same mode and size are identical */
6191 static int diff_filespec_is_identical(struct repository *r,
6192                                       struct diff_filespec *one,
6193                                       struct diff_filespec *two)
6194 {
6195         if (S_ISGITLINK(one->mode))
6196                 return 0;
6197         if (diff_populate_filespec(r, one, 0))
6198                 return 0;
6199         if (diff_populate_filespec(r, two, 0))
6200                 return 0;
6201         return !memcmp(one->data, two->data, one->size);
6202 }
6203
6204 static int diff_filespec_check_stat_unmatch(struct repository *r,
6205                                             struct diff_filepair *p)
6206 {
6207         if (p->done_skip_stat_unmatch)
6208                 return p->skip_stat_unmatch_result;
6209
6210         p->done_skip_stat_unmatch = 1;
6211         p->skip_stat_unmatch_result = 0;
6212         /*
6213          * 1. Entries that come from stat info dirtiness
6214          *    always have both sides (iow, not create/delete),
6215          *    one side of the object name is unknown, with
6216          *    the same mode and size.  Keep the ones that
6217          *    do not match these criteria.  They have real
6218          *    differences.
6219          *
6220          * 2. At this point, the file is known to be modified,
6221          *    with the same mode and size, and the object
6222          *    name of one side is unknown.  Need to inspect
6223          *    the identical contents.
6224          */
6225         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6226             !DIFF_FILE_VALID(p->two) ||
6227             (p->one->oid_valid && p->two->oid_valid) ||
6228             (p->one->mode != p->two->mode) ||
6229             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6230             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6231             (p->one->size != p->two->size) ||
6232             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6233                 p->skip_stat_unmatch_result = 1;
6234         return p->skip_stat_unmatch_result;
6235 }
6236
6237 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6238 {
6239         int i;
6240         struct diff_queue_struct *q = &diff_queued_diff;
6241         struct diff_queue_struct outq;
6242         DIFF_QUEUE_CLEAR(&outq);
6243
6244         for (i = 0; i < q->nr; i++) {
6245                 struct diff_filepair *p = q->queue[i];
6246
6247                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6248                         diff_q(&outq, p);
6249                 else {
6250                         /*
6251                          * The caller can subtract 1 from skip_stat_unmatch
6252                          * to determine how many paths were dirty only
6253                          * due to stat info mismatch.
6254                          */
6255                         if (!diffopt->flags.no_index)
6256                                 diffopt->skip_stat_unmatch++;
6257                         diff_free_filepair(p);
6258                 }
6259         }
6260         free(q->queue);
6261         *q = outq;
6262 }
6263
6264 static int diffnamecmp(const void *a_, const void *b_)
6265 {
6266         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6267         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6268         const char *name_a, *name_b;
6269
6270         name_a = a->one ? a->one->path : a->two->path;
6271         name_b = b->one ? b->one->path : b->two->path;
6272         return strcmp(name_a, name_b);
6273 }
6274
6275 void diffcore_fix_diff_index(void)
6276 {
6277         struct diff_queue_struct *q = &diff_queued_diff;
6278         QSORT(q->queue, q->nr, diffnamecmp);
6279 }
6280
6281 void diffcore_std(struct diff_options *options)
6282 {
6283         /* NOTE please keep the following in sync with diff_tree_combined() */
6284         if (options->skip_stat_unmatch)
6285                 diffcore_skip_stat_unmatch(options);
6286         if (!options->found_follow) {
6287                 /* See try_to_follow_renames() in tree-diff.c */
6288                 if (options->break_opt != -1)
6289                         diffcore_break(options->repo,
6290                                        options->break_opt);
6291                 if (options->detect_rename)
6292                         diffcore_rename(options);
6293                 if (options->break_opt != -1)
6294                         diffcore_merge_broken();
6295         }
6296         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6297                 diffcore_pickaxe(options);
6298         if (options->orderfile)
6299                 diffcore_order(options->orderfile);
6300         if (!options->found_follow)
6301                 /* See try_to_follow_renames() in tree-diff.c */
6302                 diff_resolve_rename_copy();
6303         diffcore_apply_filter(options);
6304
6305         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6306                 options->flags.has_changes = 1;
6307         else
6308                 options->flags.has_changes = 0;
6309
6310         options->found_follow = 0;
6311 }
6312
6313 int diff_result_code(struct diff_options *opt, int status)
6314 {
6315         int result = 0;
6316
6317         diff_warn_rename_limit("diff.renameLimit",
6318                                opt->needed_rename_limit,
6319                                opt->degraded_cc_to_c);
6320         if (!opt->flags.exit_with_status &&
6321             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6322                 return status;
6323         if (opt->flags.exit_with_status &&
6324             opt->flags.has_changes)
6325                 result |= 01;
6326         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6327             opt->flags.check_failed)
6328                 result |= 02;
6329         return result;
6330 }
6331
6332 int diff_can_quit_early(struct diff_options *opt)
6333 {
6334         return (opt->flags.quick &&
6335                 !opt->filter &&
6336                 opt->flags.has_changes);
6337 }
6338
6339 /*
6340  * Shall changes to this submodule be ignored?
6341  *
6342  * Submodule changes can be configured to be ignored separately for each path,
6343  * but that configuration can be overridden from the command line.
6344  */
6345 static int is_submodule_ignored(const char *path, struct diff_options *options)
6346 {
6347         int ignored = 0;
6348         struct diff_flags orig_flags = options->flags;
6349         if (!options->flags.override_submodule_config)
6350                 set_diffopt_flags_from_submodule_config(options, path);
6351         if (options->flags.ignore_submodules)
6352                 ignored = 1;
6353         options->flags = orig_flags;
6354         return ignored;
6355 }
6356
6357 void diff_addremove(struct diff_options *options,
6358                     int addremove, unsigned mode,
6359                     const struct object_id *oid,
6360                     int oid_valid,
6361                     const char *concatpath, unsigned dirty_submodule)
6362 {
6363         struct diff_filespec *one, *two;
6364
6365         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6366                 return;
6367
6368         /* This may look odd, but it is a preparation for
6369          * feeding "there are unchanged files which should
6370          * not produce diffs, but when you are doing copy
6371          * detection you would need them, so here they are"
6372          * entries to the diff-core.  They will be prefixed
6373          * with something like '=' or '*' (I haven't decided
6374          * which but should not make any difference).
6375          * Feeding the same new and old to diff_change()
6376          * also has the same effect.
6377          * Before the final output happens, they are pruned after
6378          * merged into rename/copy pairs as appropriate.
6379          */
6380         if (options->flags.reverse_diff)
6381                 addremove = (addremove == '+' ? '-' :
6382                              addremove == '-' ? '+' : addremove);
6383
6384         if (options->prefix &&
6385             strncmp(concatpath, options->prefix, options->prefix_length))
6386                 return;
6387
6388         one = alloc_filespec(concatpath);
6389         two = alloc_filespec(concatpath);
6390
6391         if (addremove != '+')
6392                 fill_filespec(one, oid, oid_valid, mode);
6393         if (addremove != '-') {
6394                 fill_filespec(two, oid, oid_valid, mode);
6395                 two->dirty_submodule = dirty_submodule;
6396         }
6397
6398         diff_queue(&diff_queued_diff, one, two);
6399         if (!options->flags.diff_from_contents)
6400                 options->flags.has_changes = 1;
6401 }
6402
6403 void diff_change(struct diff_options *options,
6404                  unsigned old_mode, unsigned new_mode,
6405                  const struct object_id *old_oid,
6406                  const struct object_id *new_oid,
6407                  int old_oid_valid, int new_oid_valid,
6408                  const char *concatpath,
6409                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6410 {
6411         struct diff_filespec *one, *two;
6412         struct diff_filepair *p;
6413
6414         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6415             is_submodule_ignored(concatpath, options))
6416                 return;
6417
6418         if (options->flags.reverse_diff) {
6419                 SWAP(old_mode, new_mode);
6420                 SWAP(old_oid, new_oid);
6421                 SWAP(old_oid_valid, new_oid_valid);
6422                 SWAP(old_dirty_submodule, new_dirty_submodule);
6423         }
6424
6425         if (options->prefix &&
6426             strncmp(concatpath, options->prefix, options->prefix_length))
6427                 return;
6428
6429         one = alloc_filespec(concatpath);
6430         two = alloc_filespec(concatpath);
6431         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6432         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6433         one->dirty_submodule = old_dirty_submodule;
6434         two->dirty_submodule = new_dirty_submodule;
6435         p = diff_queue(&diff_queued_diff, one, two);
6436
6437         if (options->flags.diff_from_contents)
6438                 return;
6439
6440         if (options->flags.quick && options->skip_stat_unmatch &&
6441             !diff_filespec_check_stat_unmatch(options->repo, p))
6442                 return;
6443
6444         options->flags.has_changes = 1;
6445 }
6446
6447 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6448 {
6449         struct diff_filepair *pair;
6450         struct diff_filespec *one, *two;
6451
6452         if (options->prefix &&
6453             strncmp(path, options->prefix, options->prefix_length))
6454                 return NULL;
6455
6456         one = alloc_filespec(path);
6457         two = alloc_filespec(path);
6458         pair = diff_queue(&diff_queued_diff, one, two);
6459         pair->is_unmerged = 1;
6460         return pair;
6461 }
6462
6463 static char *run_textconv(struct repository *r,
6464                           const char *pgm,
6465                           struct diff_filespec *spec,
6466                           size_t *outsize)
6467 {
6468         struct diff_tempfile *temp;
6469         const char *argv[3];
6470         const char **arg = argv;
6471         struct child_process child = CHILD_PROCESS_INIT;
6472         struct strbuf buf = STRBUF_INIT;
6473         int err = 0;
6474
6475         temp = prepare_temp_file(r, spec->path, spec);
6476         *arg++ = pgm;
6477         *arg++ = temp->name;
6478         *arg = NULL;
6479
6480         child.use_shell = 1;
6481         child.argv = argv;
6482         child.out = -1;
6483         if (start_command(&child)) {
6484                 remove_tempfile();
6485                 return NULL;
6486         }
6487
6488         if (strbuf_read(&buf, child.out, 0) < 0)
6489                 err = error("error reading from textconv command '%s'", pgm);
6490         close(child.out);
6491
6492         if (finish_command(&child) || err) {
6493                 strbuf_release(&buf);
6494                 remove_tempfile();
6495                 return NULL;
6496         }
6497         remove_tempfile();
6498
6499         return strbuf_detach(&buf, outsize);
6500 }
6501
6502 size_t fill_textconv(struct repository *r,
6503                      struct userdiff_driver *driver,
6504                      struct diff_filespec *df,
6505                      char **outbuf)
6506 {
6507         size_t size;
6508
6509         if (!driver) {
6510                 if (!DIFF_FILE_VALID(df)) {
6511                         *outbuf = "";
6512                         return 0;
6513                 }
6514                 if (diff_populate_filespec(r, df, 0))
6515                         die("unable to read files to diff");
6516                 *outbuf = df->data;
6517                 return df->size;
6518         }
6519
6520         if (!driver->textconv)
6521                 BUG("fill_textconv called with non-textconv driver");
6522
6523         if (driver->textconv_cache && df->oid_valid) {
6524                 *outbuf = notes_cache_get(driver->textconv_cache,
6525                                           &df->oid,
6526                                           &size);
6527                 if (*outbuf)
6528                         return size;
6529         }
6530
6531         *outbuf = run_textconv(r, driver->textconv, df, &size);
6532         if (!*outbuf)
6533                 die("unable to read files to diff");
6534
6535         if (driver->textconv_cache && df->oid_valid) {
6536                 /* ignore errors, as we might be in a readonly repository */
6537                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6538                                 size);
6539                 /*
6540                  * we could save up changes and flush them all at the end,
6541                  * but we would need an extra call after all diffing is done.
6542                  * Since generating a cache entry is the slow path anyway,
6543                  * this extra overhead probably isn't a big deal.
6544                  */
6545                 notes_cache_write(driver->textconv_cache);
6546         }
6547
6548         return size;
6549 }
6550
6551 int textconv_object(struct repository *r,
6552                     const char *path,
6553                     unsigned mode,
6554                     const struct object_id *oid,
6555                     int oid_valid,
6556                     char **buf,
6557                     unsigned long *buf_size)
6558 {
6559         struct diff_filespec *df;
6560         struct userdiff_driver *textconv;
6561
6562         df = alloc_filespec(path);
6563         fill_filespec(df, oid, oid_valid, mode);
6564         textconv = get_textconv(r, df);
6565         if (!textconv) {
6566                 free_filespec(df);
6567                 return 0;
6568         }
6569
6570         *buf_size = fill_textconv(r, textconv, df, buf);
6571         free_filespec(df);
6572         return 1;
6573 }
6574
6575 void setup_diff_pager(struct diff_options *opt)
6576 {
6577         /*
6578          * If the user asked for our exit code, then either they want --quiet
6579          * or --exit-code. We should definitely not bother with a pager in the
6580          * former case, as we will generate no output. Since we still properly
6581          * report our exit code even when a pager is run, we _could_ run a
6582          * pager with --exit-code. But since we have not done so historically,
6583          * and because it is easy to find people oneline advising "git diff
6584          * --exit-code" in hooks and other scripts, we do not do so.
6585          */
6586         if (!opt->flags.exit_with_status &&
6587             check_pager_config("diff") != 0)
6588                 setup_pager();
6589 }