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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         /*
 182          * Please update $__git_diff_submodule_formats in
 183          * git-completion.bash when you add new formats.
 184          */
 185         else
 186                 return -1;
 187         return 0;
 188 }
 189
 190 int git_config_rename(const char *var, const char *value)
 191 {
 192         if (!value)
 193                 return DIFF_DETECT_RENAME;
 194         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 195                 return  DIFF_DETECT_COPY;
 196         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 197 }
 198
 199 long parse_algorithm_value(const char *value)
 200 {
 201         if (!value)
 202                 return -1;
 203         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 204                 return 0;
 205         else if (!strcasecmp(value, "minimal"))
 206                 return XDF_NEED_MINIMAL;
 207         else if (!strcasecmp(value, "patience"))
 208                 return XDF_PATIENCE_DIFF;
 209         else if (!strcasecmp(value, "histogram"))
 210                 return XDF_HISTOGRAM_DIFF;
 211         /*
 212          * Please update $__git_diff_algorithms in git-completion.bash
 213          * when you add new algorithms.
 214          */
 215         return -1;
 216 }
 217
 218 static int parse_one_token(const char **arg, const char *token)
 219 {
 220         const char *rest;
 221         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 222                 *arg = rest;
 223                 return 1;
 224         }
 225         return 0;
 226 }
 227
 228 static int parse_ws_error_highlight(const char *arg)
 229 {
 230         const char *orig_arg = arg;
 231         unsigned val = 0;
 232
 233         while (*arg) {
 234                 if (parse_one_token(&arg, "none"))
 235                         val = 0;
 236                 else if (parse_one_token(&arg, "default"))
 237                         val = WSEH_NEW;
 238                 else if (parse_one_token(&arg, "all"))
 239                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 240                 else if (parse_one_token(&arg, "new"))
 241                         val |= WSEH_NEW;
 242                 else if (parse_one_token(&arg, "old"))
 243                         val |= WSEH_OLD;
 244                 else if (parse_one_token(&arg, "context"))
 245                         val |= WSEH_CONTEXT;
 246                 else {
 247                         return -1 - (int)(arg - orig_arg);
 248                 }
 249                 if (*arg)
 250                         arg++;
 251         }
 252         return val;
 253 }
 254
 255 /*
 256  * These are to give UI layer defaults.
 257  * The core-level commands such as git-diff-files should
 258  * never be affected by the setting of diff.renames
 259  * the user happens to have in the configuration file.
 260  */
 261 void init_diff_ui_defaults(void)
 262 {
 263         diff_detect_rename_default = DIFF_DETECT_RENAME;
 264 }
 265
 266 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 267 {
 268         if (!strcmp(var, "diff.indentheuristic"))
 269                 diff_indent_heuristic = git_config_bool(var, value);
 270         return 0;
 271 }
 272
 273 static int parse_color_moved(const char *arg)
 274 {
 275         switch (git_parse_maybe_bool(arg)) {
 276         case 0:
 277                 return COLOR_MOVED_NO;
 278         case 1:
 279                 return COLOR_MOVED_DEFAULT;
 280         default:
 281                 break;
 282         }
 283
 284         if (!strcmp(arg, "no"))
 285                 return COLOR_MOVED_NO;
 286         else if (!strcmp(arg, "plain"))
 287                 return COLOR_MOVED_PLAIN;
 288         else if (!strcmp(arg, "blocks"))
 289                 return COLOR_MOVED_BLOCKS;
 290         else if (!strcmp(arg, "zebra"))
 291                 return COLOR_MOVED_ZEBRA;
 292         else if (!strcmp(arg, "default"))
 293                 return COLOR_MOVED_DEFAULT;
 294         else if (!strcmp(arg, "dimmed-zebra"))
 295                 return COLOR_MOVED_ZEBRA_DIM;
 296         else if (!strcmp(arg, "dimmed_zebra"))
 297                 return COLOR_MOVED_ZEBRA_DIM;
 298         else
 299                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 300 }
 301
 302 static unsigned parse_color_moved_ws(const char *arg)
 303 {
 304         int ret = 0;
 305         struct string_list l = STRING_LIST_INIT_DUP;
 306         struct string_list_item *i;
 307
 308         string_list_split(&l, arg, ',', -1);
 309
 310         for_each_string_list_item(i, &l) {
 311                 struct strbuf sb = STRBUF_INIT;
 312                 strbuf_addstr(&sb, i->string);
 313                 strbuf_trim(&sb);
 314
 315                 if (!strcmp(sb.buf, "no"))
 316                         ret = 0;
 317                 else if (!strcmp(sb.buf, "ignore-space-change"))
 318                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 319                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 320                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 321                 else if (!strcmp(sb.buf, "ignore-all-space"))
 322                         ret |= XDF_IGNORE_WHITESPACE;
 323                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 324                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 325                 else {
 326                         ret |= COLOR_MOVED_WS_ERROR;
 327                         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);
 328                 }
 329
 330                 strbuf_release(&sb);
 331         }
 332
 333         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 334             (ret & XDF_WHITESPACE_FLAGS)) {
 335                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 336                 ret |= COLOR_MOVED_WS_ERROR;
 337         }
 338
 339         string_list_clear(&l, 0);
 340
 341         return ret;
 342 }
 343
 344 int git_diff_ui_config(const char *var, const char *value, void *cb)
 345 {
 346         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 347                 diff_use_color_default = git_config_colorbool(var, value);
 348                 return 0;
 349         }
 350         if (!strcmp(var, "diff.colormoved")) {
 351                 int cm = parse_color_moved(value);
 352                 if (cm < 0)
 353                         return -1;
 354                 diff_color_moved_default = cm;
 355                 return 0;
 356         }
 357         if (!strcmp(var, "diff.colormovedws")) {
 358                 unsigned cm = parse_color_moved_ws(value);
 359                 if (cm & COLOR_MOVED_WS_ERROR)
 360                         return -1;
 361                 diff_color_moved_ws_default = cm;
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.context")) {
 365                 diff_context_default = git_config_int(var, value);
 366                 if (diff_context_default < 0)
 367                         return -1;
 368                 return 0;
 369         }
 370         if (!strcmp(var, "diff.interhunkcontext")) {
 371                 diff_interhunk_context_default = git_config_int(var, value);
 372                 if (diff_interhunk_context_default < 0)
 373                         return -1;
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.renames")) {
 377                 diff_detect_rename_default = git_config_rename(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.autorefreshindex")) {
 381                 diff_auto_refresh_index = git_config_bool(var, value);
 382                 return 0;
 383         }
 384         if (!strcmp(var, "diff.mnemonicprefix")) {
 385                 diff_mnemonic_prefix = git_config_bool(var, value);
 386                 return 0;
 387         }
 388         if (!strcmp(var, "diff.noprefix")) {
 389                 diff_no_prefix = git_config_bool(var, value);
 390                 return 0;
 391         }
 392         if (!strcmp(var, "diff.statgraphwidth")) {
 393                 diff_stat_graph_width = git_config_int(var, value);
 394                 return 0;
 395         }
 396         if (!strcmp(var, "diff.external"))
 397                 return git_config_string(&external_diff_cmd_cfg, var, value);
 398         if (!strcmp(var, "diff.wordregex"))
 399                 return git_config_string(&diff_word_regex_cfg, var, value);
 400         if (!strcmp(var, "diff.orderfile"))
 401                 return git_config_pathname(&diff_order_file_cfg, var, value);
 402
 403         if (!strcmp(var, "diff.ignoresubmodules"))
 404                 handle_ignore_submodules_arg(&default_diff_options, value);
 405
 406         if (!strcmp(var, "diff.submodule")) {
 407                 if (parse_submodule_params(&default_diff_options, value))
 408                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 409                                 value);
 410                 return 0;
 411         }
 412
 413         if (!strcmp(var, "diff.algorithm")) {
 414                 diff_algorithm = parse_algorithm_value(value);
 415                 if (diff_algorithm < 0)
 416                         return -1;
 417                 return 0;
 418         }
 419
 420         if (!strcmp(var, "diff.wserrorhighlight")) {
 421                 int val = parse_ws_error_highlight(value);
 422                 if (val < 0)
 423                         return -1;
 424                 ws_error_highlight_default = val;
 425                 return 0;
 426         }
 427
 428         if (git_color_config(var, value, cb) < 0)
 429                 return -1;
 430
 431         return git_diff_basic_config(var, value, cb);
 432 }
 433
 434 int git_diff_basic_config(const char *var, const char *value, void *cb)
 435 {
 436         const char *name;
 437
 438         if (!strcmp(var, "diff.renamelimit")) {
 439                 diff_rename_limit_default = git_config_int(var, value);
 440                 return 0;
 441         }
 442
 443         if (userdiff_config(var, value) < 0)
 444                 return -1;
 445
 446         if (skip_prefix(var, "diff.color.", &name) ||
 447             skip_prefix(var, "color.diff.", &name)) {
 448                 int slot = parse_diff_color_slot(name);
 449                 if (slot < 0)
 450                         return 0;
 451                 if (!value)
 452                         return config_error_nonbool(var);
 453                 return color_parse(value, diff_colors[slot]);
 454         }
 455
 456         /* like GNU diff's --suppress-blank-empty option  */
 457         if (!strcmp(var, "diff.suppressblankempty") ||
 458                         /* for backwards compatibility */
 459                         !strcmp(var, "diff.suppress-blank-empty")) {
 460                 diff_suppress_blank_empty = git_config_bool(var, value);
 461                 return 0;
 462         }
 463
 464         if (!strcmp(var, "diff.dirstat")) {
 465                 struct strbuf errmsg = STRBUF_INIT;
 466                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 467                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 468                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 469                                 errmsg.buf);
 470                 strbuf_release(&errmsg);
 471                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 472                 return 0;
 473         }
 474
 475         if (git_diff_heuristic_config(var, value, cb) < 0)
 476                 return -1;
 477
 478         return git_default_config(var, value, cb);
 479 }
 480
 481 static char *quote_two(const char *one, const char *two)
 482 {
 483         int need_one = quote_c_style(one, NULL, NULL, 1);
 484         int need_two = quote_c_style(two, NULL, NULL, 1);
 485         struct strbuf res = STRBUF_INIT;
 486
 487         if (need_one + need_two) {
 488                 strbuf_addch(&res, '"');
 489                 quote_c_style(one, &res, NULL, 1);
 490                 quote_c_style(two, &res, NULL, 1);
 491                 strbuf_addch(&res, '"');
 492         } else {
 493                 strbuf_addstr(&res, one);
 494                 strbuf_addstr(&res, two);
 495         }
 496         return strbuf_detach(&res, NULL);
 497 }
 498
 499 static const char *external_diff(void)
 500 {
 501         static const char *external_diff_cmd = NULL;
 502         static int done_preparing = 0;
 503
 504         if (done_preparing)
 505                 return external_diff_cmd;
 506         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 507         if (!external_diff_cmd)
 508                 external_diff_cmd = external_diff_cmd_cfg;
 509         done_preparing = 1;
 510         return external_diff_cmd;
 511 }
 512
 513 /*
 514  * Keep track of files used for diffing. Sometimes such an entry
 515  * refers to a temporary file, sometimes to an existing file, and
 516  * sometimes to "/dev/null".
 517  */
 518 static struct diff_tempfile {
 519         /*
 520          * filename external diff should read from, or NULL if this
 521          * entry is currently not in use:
 522          */
 523         const char *name;
 524
 525         char hex[GIT_MAX_HEXSZ + 1];
 526         char mode[10];
 527
 528         /*
 529          * If this diff_tempfile instance refers to a temporary file,
 530          * this tempfile object is used to manage its lifetime.
 531          */
 532         struct tempfile *tempfile;
 533 } diff_temp[2];
 534
 535 struct emit_callback {
 536         int color_diff;
 537         unsigned ws_rule;
 538         int blank_at_eof_in_preimage;
 539         int blank_at_eof_in_postimage;
 540         int lno_in_preimage;
 541         int lno_in_postimage;
 542         const char **label_path;
 543         struct diff_words_data *diff_words;
 544         struct diff_options *opt;
 545         struct strbuf *header;
 546 };
 547
 548 static int count_lines(const char *data, int size)
 549 {
 550         int count, ch, completely_empty = 1, nl_just_seen = 0;
 551         count = 0;
 552         while (0 < size--) {
 553                 ch = *data++;
 554                 if (ch == '\n') {
 555                         count++;
 556                         nl_just_seen = 1;
 557                         completely_empty = 0;
 558                 }
 559                 else {
 560                         nl_just_seen = 0;
 561                         completely_empty = 0;
 562                 }
 563         }
 564         if (completely_empty)
 565                 return 0;
 566         if (!nl_just_seen)
 567                 count++; /* no trailing newline */
 568         return count;
 569 }
 570
 571 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 572                        struct diff_filespec *one)
 573 {
 574         if (!DIFF_FILE_VALID(one)) {
 575                 mf->ptr = (char *)""; /* does not matter */
 576                 mf->size = 0;
 577                 return 0;
 578         }
 579         else if (diff_populate_filespec(r, one, 0))
 580                 return -1;
 581
 582         mf->ptr = one->data;
 583         mf->size = one->size;
 584         return 0;
 585 }
 586
 587 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 588 static unsigned long diff_filespec_size(struct repository *r,
 589                                         struct diff_filespec *one)
 590 {
 591         if (!DIFF_FILE_VALID(one))
 592                 return 0;
 593         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 594         return one->size;
 595 }
 596
 597 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 598 {
 599         char *ptr = mf->ptr;
 600         long size = mf->size;
 601         int cnt = 0;
 602
 603         if (!size)
 604                 return cnt;
 605         ptr += size - 1; /* pointing at the very end */
 606         if (*ptr != '\n')
 607                 ; /* incomplete line */
 608         else
 609                 ptr--; /* skip the last LF */
 610         while (mf->ptr < ptr) {
 611                 char *prev_eol;
 612                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 613                         if (*prev_eol == '\n')
 614                                 break;
 615                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 616                         break;
 617                 cnt++;
 618                 ptr = prev_eol - 1;
 619         }
 620         return cnt;
 621 }
 622
 623 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 624                                struct emit_callback *ecbdata)
 625 {
 626         int l1, l2, at;
 627         unsigned ws_rule = ecbdata->ws_rule;
 628         l1 = count_trailing_blank(mf1, ws_rule);
 629         l2 = count_trailing_blank(mf2, ws_rule);
 630         if (l2 <= l1) {
 631                 ecbdata->blank_at_eof_in_preimage = 0;
 632                 ecbdata->blank_at_eof_in_postimage = 0;
 633                 return;
 634         }
 635         at = count_lines(mf1->ptr, mf1->size);
 636         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 637
 638         at = count_lines(mf2->ptr, mf2->size);
 639         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 640 }
 641
 642 static void emit_line_0(struct diff_options *o,
 643                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 644                         int first, const char *line, int len)
 645 {
 646         int has_trailing_newline, has_trailing_carriage_return;
 647         int needs_reset = 0; /* at the end of the line */
 648         FILE *file = o->file;
 649
 650         fputs(diff_line_prefix(o), file);
 651
 652         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 653         if (has_trailing_newline)
 654                 len--;
 655
 656         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 657         if (has_trailing_carriage_return)
 658                 len--;
 659
 660         if (!len && !first)
 661                 goto end_of_line;
 662
 663         if (reverse && want_color(o->use_color)) {
 664                 fputs(GIT_COLOR_REVERSE, file);
 665                 needs_reset = 1;
 666         }
 667
 668         if (set_sign) {
 669                 fputs(set_sign, file);
 670                 needs_reset = 1;
 671         }
 672
 673         if (first)
 674                 fputc(first, file);
 675
 676         if (!len)
 677                 goto end_of_line;
 678
 679         if (set) {
 680                 if (set_sign && set != set_sign)
 681                         fputs(reset, file);
 682                 fputs(set, file);
 683                 needs_reset = 1;
 684         }
 685         fwrite(line, len, 1, file);
 686         needs_reset = 1; /* 'line' may contain color codes. */
 687
 688 end_of_line:
 689         if (needs_reset)
 690                 fputs(reset, file);
 691         if (has_trailing_carriage_return)
 692                 fputc('\r', file);
 693         if (has_trailing_newline)
 694                 fputc('\n', file);
 695 }
 696
 697 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 698                       const char *line, int len)
 699 {
 700         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 701 }
 702
 703 enum diff_symbol {
 704         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 705         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 706         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 707         DIFF_SYMBOL_BINARY_DIFF_BODY,
 708         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 709         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 710         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 711         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 712         DIFF_SYMBOL_STATS_LINE,
 713         DIFF_SYMBOL_WORD_DIFF,
 714         DIFF_SYMBOL_STAT_SEP,
 715         DIFF_SYMBOL_SUMMARY,
 716         DIFF_SYMBOL_SUBMODULE_ADD,
 717         DIFF_SYMBOL_SUBMODULE_DEL,
 718         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 719         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 720         DIFF_SYMBOL_SUBMODULE_HEADER,
 721         DIFF_SYMBOL_SUBMODULE_ERROR,
 722         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 723         DIFF_SYMBOL_REWRITE_DIFF,
 724         DIFF_SYMBOL_BINARY_FILES,
 725         DIFF_SYMBOL_HEADER,
 726         DIFF_SYMBOL_FILEPAIR_PLUS,
 727         DIFF_SYMBOL_FILEPAIR_MINUS,
 728         DIFF_SYMBOL_WORDS_PORCELAIN,
 729         DIFF_SYMBOL_WORDS,
 730         DIFF_SYMBOL_CONTEXT,
 731         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 732         DIFF_SYMBOL_PLUS,
 733         DIFF_SYMBOL_MINUS,
 734         DIFF_SYMBOL_NO_LF_EOF,
 735         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 736         DIFF_SYMBOL_CONTEXT_MARKER,
 737         DIFF_SYMBOL_SEPARATOR
 738 };
 739 /*
 740  * Flags for content lines:
 741  * 0..12 are whitespace rules
 742  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 743  * 16 is marking if the line is blank at EOF
 744  */
 745 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 746 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 747 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 748 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 749 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 750
 751 /*
 752  * This struct is used when we need to buffer the output of the diff output.
 753  *
 754  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 755  * into the pre/post image file. This pointer could be a union with the
 756  * line pointer. By storing an offset into the file instead of the literal line,
 757  * we can decrease the memory footprint for the buffered output. At first we
 758  * may want to only have indirection for the content lines, but we could also
 759  * enhance the state for emitting prefabricated lines, e.g. the similarity
 760  * score line or hunk/file headers would only need to store a number or path
 761  * and then the output can be constructed later on depending on state.
 762  */
 763 struct emitted_diff_symbol {
 764         const char *line;
 765         int len;
 766         int flags;
 767         int indent_off;   /* Offset to first non-whitespace character */
 768         int indent_width; /* The visual width of the indentation */
 769         enum diff_symbol s;
 770 };
 771 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 772
 773 struct emitted_diff_symbols {
 774         struct emitted_diff_symbol *buf;
 775         int nr, alloc;
 776 };
 777 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 778
 779 static void append_emitted_diff_symbol(struct diff_options *o,
 780                                        struct emitted_diff_symbol *e)
 781 {
 782         struct emitted_diff_symbol *f;
 783
 784         ALLOC_GROW(o->emitted_symbols->buf,
 785                    o->emitted_symbols->nr + 1,
 786                    o->emitted_symbols->alloc);
 787         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 788
 789         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 790         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 791 }
 792
 793 struct moved_entry {
 794         struct hashmap_entry ent;
 795         const struct emitted_diff_symbol *es;
 796         struct moved_entry *next_line;
 797 };
 798
 799 struct moved_block {
 800         struct moved_entry *match;
 801         int wsd; /* The whitespace delta of this block */
 802 };
 803
 804 static void moved_block_clear(struct moved_block *b)
 805 {
 806         memset(b, 0, sizeof(*b));
 807 }
 808
 809 #define INDENT_BLANKLINE INT_MIN
 810
 811 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 812 {
 813         unsigned int off = 0, i;
 814         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 815         const char *s = es->line;
 816         const int len = es->len;
 817
 818         /* skip any \v \f \r at start of indentation */
 819         while (s[off] == '\f' || s[off] == '\v' ||
 820                (s[off] == '\r' && off < len - 1))
 821                 off++;
 822
 823         /* calculate the visual width of indentation */
 824         while(1) {
 825                 if (s[off] == ' ') {
 826                         width++;
 827                         off++;
 828                 } else if (s[off] == '\t') {
 829                         width += tab_width - (width % tab_width);
 830                         while (s[++off] == '\t')
 831                                 width += tab_width;
 832                 } else {
 833                         break;
 834                 }
 835         }
 836
 837         /* check if this line is blank */
 838         for (i = off; i < len; i++)
 839                 if (!isspace(s[i]))
 840                     break;
 841
 842         if (i == len) {
 843                 es->indent_width = INDENT_BLANKLINE;
 844                 es->indent_off = len;
 845         } else {
 846                 es->indent_off = off;
 847                 es->indent_width = width;
 848         }
 849 }
 850
 851 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 852                             const struct emitted_diff_symbol *b,
 853                             int *out)
 854 {
 855         int a_len = a->len,
 856             b_len = b->len,
 857             a_off = a->indent_off,
 858             a_width = a->indent_width,
 859             b_off = b->indent_off,
 860             b_width = b->indent_width;
 861         int delta;
 862
 863         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 864                 *out = INDENT_BLANKLINE;
 865                 return 1;
 866         }
 867
 868         if (a->s == DIFF_SYMBOL_PLUS)
 869                 delta = a_width - b_width;
 870         else
 871                 delta = b_width - a_width;
 872
 873         if (a_len - a_off != b_len - b_off ||
 874             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 875                 return 0;
 876
 877         *out = delta;
 878
 879         return 1;
 880 }
 881
 882 static int cmp_in_block_with_wsd(const struct diff_options *o,
 883                                  const struct moved_entry *cur,
 884                                  const struct moved_entry *match,
 885                                  struct moved_block *pmb,
 886                                  int n)
 887 {
 888         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 889         int al = cur->es->len, bl = match->es->len, cl = l->len;
 890         const char *a = cur->es->line,
 891                    *b = match->es->line,
 892                    *c = l->line;
 893         int a_off = cur->es->indent_off,
 894             a_width = cur->es->indent_width,
 895             c_off = l->indent_off,
 896             c_width = l->indent_width;
 897         int delta;
 898
 899         /*
 900          * We need to check if 'cur' is equal to 'match'.  As those
 901          * are from the same (+/-) side, we do not need to adjust for
 902          * indent changes. However these were found using fuzzy
 903          * matching so we do have to check if they are equal. Here we
 904          * just check the lengths. We delay calling memcmp() to check
 905          * the contents until later as if the length comparison for a
 906          * and c fails we can avoid the call all together.
 907          */
 908         if (al != bl)
 909                 return 1;
 910
 911         /* If 'l' and 'cur' are both blank then they match. */
 912         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 913                 return 0;
 914
 915         /*
 916          * The indent changes of the block are known and stored in pmb->wsd;
 917          * however we need to check if the indent changes of the current line
 918          * match those of the current block and that the text of 'l' and 'cur'
 919          * after the indentation match.
 920          */
 921         if (cur->es->s == DIFF_SYMBOL_PLUS)
 922                 delta = a_width - c_width;
 923         else
 924                 delta = c_width - a_width;
 925
 926         /*
 927          * If the previous lines of this block were all blank then set its
 928          * whitespace delta.
 929          */
 930         if (pmb->wsd == INDENT_BLANKLINE)
 931                 pmb->wsd = delta;
 932
 933         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 934                  !memcmp(a, b, al) && !
 935                  memcmp(a + a_off, c + c_off, al - a_off));
 936 }
 937
 938 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 939                            const void *entry,
 940                            const void *entry_or_key,
 941                            const void *keydata)
 942 {
 943         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 944         const struct moved_entry *a = entry;
 945         const struct moved_entry *b = entry_or_key;
 946         unsigned flags = diffopt->color_moved_ws_handling
 947                          & XDF_WHITESPACE_FLAGS;
 948
 949         if (diffopt->color_moved_ws_handling &
 950             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 951                 /*
 952                  * As there is not specific white space config given,
 953                  * we'd need to check for a new block, so ignore all
 954                  * white space. The setup of the white space
 955                  * configuration for the next block is done else where
 956                  */
 957                 flags |= XDF_IGNORE_WHITESPACE;
 958
 959         return !xdiff_compare_lines(a->es->line, a->es->len,
 960                                     b->es->line, b->es->len,
 961                                     flags);
 962 }
 963
 964 static struct moved_entry *prepare_entry(struct diff_options *o,
 965                                          int line_no)
 966 {
 967         struct moved_entry *ret = xmalloc(sizeof(*ret));
 968         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 969         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 970
 971         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 972         ret->es = l;
 973         ret->next_line = NULL;
 974
 975         return ret;
 976 }
 977
 978 static void add_lines_to_move_detection(struct diff_options *o,
 979                                         struct hashmap *add_lines,
 980                                         struct hashmap *del_lines)
 981 {
 982         struct moved_entry *prev_line = NULL;
 983
 984         int n;
 985         for (n = 0; n < o->emitted_symbols->nr; n++) {
 986                 struct hashmap *hm;
 987                 struct moved_entry *key;
 988
 989                 switch (o->emitted_symbols->buf[n].s) {
 990                 case DIFF_SYMBOL_PLUS:
 991                         hm = add_lines;
 992                         break;
 993                 case DIFF_SYMBOL_MINUS:
 994                         hm = del_lines;
 995                         break;
 996                 default:
 997                         prev_line = NULL;
 998                         continue;
 999                 }
1000
1001                 if (o->color_moved_ws_handling &
1002                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1003                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1004                 key = prepare_entry(o, n);
1005                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1006                         prev_line->next_line = key;
1007
1008                 hashmap_add(hm, key);
1009                 prev_line = key;
1010         }
1011 }
1012
1013 static void pmb_advance_or_null(struct diff_options *o,
1014                                 struct moved_entry *match,
1015                                 struct hashmap *hm,
1016                                 struct moved_block *pmb,
1017                                 int pmb_nr)
1018 {
1019         int i;
1020         for (i = 0; i < pmb_nr; i++) {
1021                 struct moved_entry *prev = pmb[i].match;
1022                 struct moved_entry *cur = (prev && prev->next_line) ?
1023                                 prev->next_line : NULL;
1024                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1025                         pmb[i].match = cur;
1026                 } else {
1027                         pmb[i].match = NULL;
1028                 }
1029         }
1030 }
1031
1032 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1033                                             struct moved_entry *match,
1034                                             struct hashmap *hm,
1035                                             struct moved_block *pmb,
1036                                             int pmb_nr, int n)
1037 {
1038         int i;
1039         char *got_match = xcalloc(1, pmb_nr);
1040
1041         for (; match; match = hashmap_get_next(hm, match)) {
1042                 for (i = 0; i < pmb_nr; i++) {
1043                         struct moved_entry *prev = pmb[i].match;
1044                         struct moved_entry *cur = (prev && prev->next_line) ?
1045                                         prev->next_line : NULL;
1046                         if (!cur)
1047                                 continue;
1048                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1049                                 got_match[i] |= 1;
1050                 }
1051         }
1052
1053         for (i = 0; i < pmb_nr; i++) {
1054                 if (got_match[i]) {
1055                         /* Advance to the next line */
1056                         pmb[i].match = pmb[i].match->next_line;
1057                 } else {
1058                         moved_block_clear(&pmb[i]);
1059                 }
1060         }
1061
1062         free(got_match);
1063 }
1064
1065 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1066                                          int pmb_nr)
1067 {
1068         int lp, rp;
1069
1070         /* Shrink the set of potential block to the remaining running */
1071         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1072                 while (lp < pmb_nr && pmb[lp].match)
1073                         lp++;
1074                 /* lp points at the first NULL now */
1075
1076                 while (rp > -1 && !pmb[rp].match)
1077                         rp--;
1078                 /* rp points at the last non-NULL */
1079
1080                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1081                         pmb[lp] = pmb[rp];
1082                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1083                         rp--;
1084                         lp++;
1085                 }
1086         }
1087
1088         /* Remember the number of running sets */
1089         return rp + 1;
1090 }
1091
1092 /*
1093  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1094  *
1095  * Otherwise, if the last block has fewer alphanumeric characters than
1096  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1097  * that block.
1098  *
1099  * The last block consists of the (n - block_length)'th line up to but not
1100  * including the nth line.
1101  *
1102  * Returns 0 if the last block is empty or is unset by this function, non zero
1103  * otherwise.
1104  *
1105  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1106  * Think of a way to unify them.
1107  */
1108 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1109 {
1110         int i, alnum_count = 0;
1111         if (o->color_moved == COLOR_MOVED_PLAIN)
1112                 return block_length;
1113         for (i = 1; i < block_length + 1; i++) {
1114                 const char *c = o->emitted_symbols->buf[n - i].line;
1115                 for (; *c; c++) {
1116                         if (!isalnum(*c))
1117                                 continue;
1118                         alnum_count++;
1119                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1120                                 return 1;
1121                 }
1122         }
1123         for (i = 1; i < block_length + 1; i++)
1124                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1125         return 0;
1126 }
1127
1128 /* Find blocks of moved code, delegate actual coloring decision to helper */
1129 static void mark_color_as_moved(struct diff_options *o,
1130                                 struct hashmap *add_lines,
1131                                 struct hashmap *del_lines)
1132 {
1133         struct moved_block *pmb = NULL; /* potentially moved blocks */
1134         int pmb_nr = 0, pmb_alloc = 0;
1135         int n, flipped_block = 0, block_length = 0;
1136
1137
1138         for (n = 0; n < o->emitted_symbols->nr; n++) {
1139                 struct hashmap *hm = NULL;
1140                 struct moved_entry *key;
1141                 struct moved_entry *match = NULL;
1142                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1143                 enum diff_symbol last_symbol = 0;
1144
1145                 switch (l->s) {
1146                 case DIFF_SYMBOL_PLUS:
1147                         hm = del_lines;
1148                         key = prepare_entry(o, n);
1149                         match = hashmap_get(hm, key, NULL);
1150                         free(key);
1151                         break;
1152                 case DIFF_SYMBOL_MINUS:
1153                         hm = add_lines;
1154                         key = prepare_entry(o, n);
1155                         match = hashmap_get(hm, key, NULL);
1156                         free(key);
1157                         break;
1158                 default:
1159                         flipped_block = 0;
1160                 }
1161
1162                 if (!match) {
1163                         int i;
1164
1165                         adjust_last_block(o, n, block_length);
1166                         for(i = 0; i < pmb_nr; i++)
1167                                 moved_block_clear(&pmb[i]);
1168                         pmb_nr = 0;
1169                         block_length = 0;
1170                         flipped_block = 0;
1171                         last_symbol = l->s;
1172                         continue;
1173                 }
1174
1175                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1176                         last_symbol = l->s;
1177                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1178                         continue;
1179                 }
1180
1181                 if (o->color_moved_ws_handling &
1182                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1183                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1184                 else
1185                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1186
1187                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1188
1189                 if (pmb_nr == 0) {
1190                         /*
1191                          * The current line is the start of a new block.
1192                          * Setup the set of potential blocks.
1193                          */
1194                         for (; match; match = hashmap_get_next(hm, match)) {
1195                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1196                                 if (o->color_moved_ws_handling &
1197                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1198                                         if (compute_ws_delta(l, match->es,
1199                                                              &pmb[pmb_nr].wsd))
1200                                                 pmb[pmb_nr++].match = match;
1201                                 } else {
1202                                         pmb[pmb_nr].wsd = 0;
1203                                         pmb[pmb_nr++].match = match;
1204                                 }
1205                         }
1206
1207                         if (adjust_last_block(o, n, block_length) &&
1208                             pmb_nr && last_symbol != l->s)
1209                                 flipped_block = (flipped_block + 1) % 2;
1210                         else
1211                                 flipped_block = 0;
1212
1213                         block_length = 0;
1214                 }
1215
1216                 if (pmb_nr) {
1217                         block_length++;
1218                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1219                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1220                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1221                 }
1222                 last_symbol = l->s;
1223         }
1224         adjust_last_block(o, n, block_length);
1225
1226         for(n = 0; n < pmb_nr; n++)
1227                 moved_block_clear(&pmb[n]);
1228         free(pmb);
1229 }
1230
1231 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1232   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1233 static void dim_moved_lines(struct diff_options *o)
1234 {
1235         int n;
1236         for (n = 0; n < o->emitted_symbols->nr; n++) {
1237                 struct emitted_diff_symbol *prev = (n != 0) ?
1238                                 &o->emitted_symbols->buf[n - 1] : NULL;
1239                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1240                 struct emitted_diff_symbol *next =
1241                                 (n < o->emitted_symbols->nr - 1) ?
1242                                 &o->emitted_symbols->buf[n + 1] : NULL;
1243
1244                 /* Not a plus or minus line? */
1245                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1246                         continue;
1247
1248                 /* Not a moved line? */
1249                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1250                         continue;
1251
1252                 /*
1253                  * If prev or next are not a plus or minus line,
1254                  * pretend they don't exist
1255                  */
1256                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1257                             prev->s != DIFF_SYMBOL_MINUS)
1258                         prev = NULL;
1259                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1260                             next->s != DIFF_SYMBOL_MINUS)
1261                         next = NULL;
1262
1263                 /* Inside a block? */
1264                 if ((prev &&
1265                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1266                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1267                     (next &&
1268                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1269                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1270                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1271                         continue;
1272                 }
1273
1274                 /* Check if we are at an interesting bound: */
1275                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1276                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1277                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1278                         continue;
1279                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1280                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1281                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1282                         continue;
1283
1284                 /*
1285                  * The boundary to prev and next are not interesting,
1286                  * so this line is not interesting as a whole
1287                  */
1288                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1289         }
1290 }
1291
1292 static void emit_line_ws_markup(struct diff_options *o,
1293                                 const char *set_sign, const char *set,
1294                                 const char *reset,
1295                                 int sign_index, const char *line, int len,
1296                                 unsigned ws_rule, int blank_at_eof)
1297 {
1298         const char *ws = NULL;
1299         int sign = o->output_indicators[sign_index];
1300
1301         if (o->ws_error_highlight & ws_rule) {
1302                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1303                 if (!*ws)
1304                         ws = NULL;
1305         }
1306
1307         if (!ws && !set_sign)
1308                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1309         else if (!ws) {
1310                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1311         } else if (blank_at_eof)
1312                 /* Blank line at EOF - paint '+' as well */
1313                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1314         else {
1315                 /* Emit just the prefix, then the rest. */
1316                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1317                             sign, "", 0);
1318                 ws_check_emit(line, len, ws_rule,
1319                               o->file, set, reset, ws);
1320         }
1321 }
1322
1323 static void emit_diff_symbol_from_struct(struct diff_options *o,
1324                                          struct emitted_diff_symbol *eds)
1325 {
1326         static const char *nneof = " No newline at end of file\n";
1327         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1328         struct strbuf sb = STRBUF_INIT;
1329
1330         enum diff_symbol s = eds->s;
1331         const char *line = eds->line;
1332         int len = eds->len;
1333         unsigned flags = eds->flags;
1334
1335         switch (s) {
1336         case DIFF_SYMBOL_NO_LF_EOF:
1337                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1338                 reset = diff_get_color_opt(o, DIFF_RESET);
1339                 putc('\n', o->file);
1340                 emit_line_0(o, context, NULL, 0, reset, '\\',
1341                             nneof, strlen(nneof));
1342                 break;
1343         case DIFF_SYMBOL_SUBMODULE_HEADER:
1344         case DIFF_SYMBOL_SUBMODULE_ERROR:
1345         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1346         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1347         case DIFF_SYMBOL_SUMMARY:
1348         case DIFF_SYMBOL_STATS_LINE:
1349         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1350         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1351                 emit_line(o, "", "", line, len);
1352                 break;
1353         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1354         case DIFF_SYMBOL_CONTEXT_MARKER:
1355                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1356                 reset = diff_get_color_opt(o, DIFF_RESET);
1357                 emit_line(o, context, reset, line, len);
1358                 break;
1359         case DIFF_SYMBOL_SEPARATOR:
1360                 fprintf(o->file, "%s%c",
1361                         diff_line_prefix(o),
1362                         o->line_termination);
1363                 break;
1364         case DIFF_SYMBOL_CONTEXT:
1365                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1366                 reset = diff_get_color_opt(o, DIFF_RESET);
1367                 set_sign = NULL;
1368                 if (o->flags.dual_color_diffed_diffs) {
1369                         char c = !len ? 0 : line[0];
1370
1371                         if (c == '+')
1372                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1373                         else if (c == '@')
1374                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1375                         else if (c == '-')
1376                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1377                 }
1378                 emit_line_ws_markup(o, set_sign, set, reset,
1379                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1380                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1381                 break;
1382         case DIFF_SYMBOL_PLUS:
1383                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1384                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1385                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1386                 case DIFF_SYMBOL_MOVED_LINE |
1387                      DIFF_SYMBOL_MOVED_LINE_ALT |
1388                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1389                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1390                         break;
1391                 case DIFF_SYMBOL_MOVED_LINE |
1392                      DIFF_SYMBOL_MOVED_LINE_ALT:
1393                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1394                         break;
1395                 case DIFF_SYMBOL_MOVED_LINE |
1396                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1397                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1398                         break;
1399                 case DIFF_SYMBOL_MOVED_LINE:
1400                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1401                         break;
1402                 default:
1403                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1404                 }
1405                 reset = diff_get_color_opt(o, DIFF_RESET);
1406                 if (!o->flags.dual_color_diffed_diffs)
1407                         set_sign = NULL;
1408                 else {
1409                         char c = !len ? 0 : line[0];
1410
1411                         set_sign = set;
1412                         if (c == '-')
1413                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1414                         else if (c == '@')
1415                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1416                         else if (c == '+')
1417                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1418                         else
1419                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1420                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1421                 }
1422                 emit_line_ws_markup(o, set_sign, set, reset,
1423                                     OUTPUT_INDICATOR_NEW, line, len,
1424                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1425                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1426                 break;
1427         case DIFF_SYMBOL_MINUS:
1428                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1429                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1430                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1431                 case DIFF_SYMBOL_MOVED_LINE |
1432                      DIFF_SYMBOL_MOVED_LINE_ALT |
1433                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1434                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1435                         break;
1436                 case DIFF_SYMBOL_MOVED_LINE |
1437                      DIFF_SYMBOL_MOVED_LINE_ALT:
1438                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1439                         break;
1440                 case DIFF_SYMBOL_MOVED_LINE |
1441                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1442                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1443                         break;
1444                 case DIFF_SYMBOL_MOVED_LINE:
1445                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1446                         break;
1447                 default:
1448                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1449                 }
1450                 reset = diff_get_color_opt(o, DIFF_RESET);
1451                 if (!o->flags.dual_color_diffed_diffs)
1452                         set_sign = NULL;
1453                 else {
1454                         char c = !len ? 0 : line[0];
1455
1456                         set_sign = set;
1457                         if (c == '+')
1458                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1459                         else if (c == '@')
1460                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1461                         else if (c == '-')
1462                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1463                         else
1464                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1465                 }
1466                 emit_line_ws_markup(o, set_sign, set, reset,
1467                                     OUTPUT_INDICATOR_OLD, line, len,
1468                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1469                 break;
1470         case DIFF_SYMBOL_WORDS_PORCELAIN:
1471                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1472                 reset = diff_get_color_opt(o, DIFF_RESET);
1473                 emit_line(o, context, reset, line, len);
1474                 fputs("~\n", o->file);
1475                 break;
1476         case DIFF_SYMBOL_WORDS:
1477                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1478                 reset = diff_get_color_opt(o, DIFF_RESET);
1479                 /*
1480                  * Skip the prefix character, if any.  With
1481                  * diff_suppress_blank_empty, there may be
1482                  * none.
1483                  */
1484                 if (line[0] != '\n') {
1485                         line++;
1486                         len--;
1487                 }
1488                 emit_line(o, context, reset, line, len);
1489                 break;
1490         case DIFF_SYMBOL_FILEPAIR_PLUS:
1491                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1492                 reset = diff_get_color_opt(o, DIFF_RESET);
1493                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1494                         line, reset,
1495                         strchr(line, ' ') ? "\t" : "");
1496                 break;
1497         case DIFF_SYMBOL_FILEPAIR_MINUS:
1498                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1499                 reset = diff_get_color_opt(o, DIFF_RESET);
1500                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1501                         line, reset,
1502                         strchr(line, ' ') ? "\t" : "");
1503                 break;
1504         case DIFF_SYMBOL_BINARY_FILES:
1505         case DIFF_SYMBOL_HEADER:
1506                 fprintf(o->file, "%s", line);
1507                 break;
1508         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1509                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1510                 break;
1511         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1512                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1513                 break;
1514         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1515                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1516                 break;
1517         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1518                 fputs(diff_line_prefix(o), o->file);
1519                 fputc('\n', o->file);
1520                 break;
1521         case DIFF_SYMBOL_REWRITE_DIFF:
1522                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1523                 reset = diff_get_color_opt(o, DIFF_RESET);
1524                 emit_line(o, fraginfo, reset, line, len);
1525                 break;
1526         case DIFF_SYMBOL_SUBMODULE_ADD:
1527                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1528                 reset = diff_get_color_opt(o, DIFF_RESET);
1529                 emit_line(o, set, reset, line, len);
1530                 break;
1531         case DIFF_SYMBOL_SUBMODULE_DEL:
1532                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1533                 reset = diff_get_color_opt(o, DIFF_RESET);
1534                 emit_line(o, set, reset, line, len);
1535                 break;
1536         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1537                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1538                         diff_line_prefix(o), line);
1539                 break;
1540         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1541                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1542                         diff_line_prefix(o), line);
1543                 break;
1544         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1545                 emit_line(o, "", "", " 0 files changed\n",
1546                           strlen(" 0 files changed\n"));
1547                 break;
1548         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1549                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1550                 break;
1551         case DIFF_SYMBOL_WORD_DIFF:
1552                 fprintf(o->file, "%.*s", len, line);
1553                 break;
1554         case DIFF_SYMBOL_STAT_SEP:
1555                 fputs(o->stat_sep, o->file);
1556                 break;
1557         default:
1558                 BUG("unknown diff symbol");
1559         }
1560         strbuf_release(&sb);
1561 }
1562
1563 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1564                              const char *line, int len, unsigned flags)
1565 {
1566         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1567
1568         if (o->emitted_symbols)
1569                 append_emitted_diff_symbol(o, &e);
1570         else
1571                 emit_diff_symbol_from_struct(o, &e);
1572 }
1573
1574 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1575 {
1576         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1577 }
1578
1579 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1580 {
1581         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1582 }
1583
1584 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1585 {
1586         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1587                          path, strlen(path), 0);
1588 }
1589
1590 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1591 {
1592         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1593                          path, strlen(path), 0);
1594 }
1595
1596 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1597 {
1598         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1599                          header, strlen(header), 0);
1600 }
1601
1602 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1603 {
1604         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1605 }
1606
1607 void diff_emit_submodule_pipethrough(struct diff_options *o,
1608                                      const char *line, int len)
1609 {
1610         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1611 }
1612
1613 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1614 {
1615         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1616               ecbdata->blank_at_eof_in_preimage &&
1617               ecbdata->blank_at_eof_in_postimage &&
1618               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1619               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1620                 return 0;
1621         return ws_blank_line(line, len, ecbdata->ws_rule);
1622 }
1623
1624 static void emit_add_line(const char *reset,
1625                           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(const char *reset,
1636                           struct emit_callback *ecbdata,
1637                           const char *line, int len)
1638 {
1639         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1640         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1641 }
1642
1643 static void emit_context_line(const char *reset,
1644                               struct emit_callback *ecbdata,
1645                               const char *line, int len)
1646 {
1647         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1648         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1649 }
1650
1651 static void emit_hunk_header(struct emit_callback *ecbdata,
1652                              const char *line, int len)
1653 {
1654         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1655         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1656         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1657         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1658         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1659         static const char atat[2] = { '@', '@' };
1660         const char *cp, *ep;
1661         struct strbuf msgbuf = STRBUF_INIT;
1662         int org_len = len;
1663         int i = 1;
1664
1665         /*
1666          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1667          * it always is at least 10 bytes long.
1668          */
1669         if (len < 10 ||
1670             memcmp(line, atat, 2) ||
1671             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1672                 emit_diff_symbol(ecbdata->opt,
1673                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1674                 return;
1675         }
1676         ep += 2; /* skip over @@ */
1677
1678         /* The hunk header in fraginfo color */
1679         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1680                 strbuf_addstr(&msgbuf, reverse);
1681         strbuf_addstr(&msgbuf, frag);
1682         strbuf_add(&msgbuf, line, ep - line);
1683         strbuf_addstr(&msgbuf, reset);
1684
1685         /*
1686          * trailing "\r\n"
1687          */
1688         for ( ; i < 3; i++)
1689                 if (line[len - i] == '\r' || line[len - i] == '\n')
1690                         len--;
1691
1692         /* blank before the func header */
1693         for (cp = ep; ep - line < len; ep++)
1694                 if (*ep != ' ' && *ep != '\t')
1695                         break;
1696         if (ep != cp) {
1697                 strbuf_addstr(&msgbuf, context);
1698                 strbuf_add(&msgbuf, cp, ep - cp);
1699                 strbuf_addstr(&msgbuf, reset);
1700         }
1701
1702         if (ep < line + len) {
1703                 strbuf_addstr(&msgbuf, func);
1704                 strbuf_add(&msgbuf, ep, line + len - ep);
1705                 strbuf_addstr(&msgbuf, reset);
1706         }
1707
1708         strbuf_add(&msgbuf, line + len, org_len - len);
1709         strbuf_complete_line(&msgbuf);
1710         emit_diff_symbol(ecbdata->opt,
1711                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1712         strbuf_release(&msgbuf);
1713 }
1714
1715 static struct diff_tempfile *claim_diff_tempfile(void)
1716 {
1717         int i;
1718         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1719                 if (!diff_temp[i].name)
1720                         return diff_temp + i;
1721         BUG("diff is failing to clean up its tempfiles");
1722 }
1723
1724 static void remove_tempfile(void)
1725 {
1726         int i;
1727         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1728                 if (is_tempfile_active(diff_temp[i].tempfile))
1729                         delete_tempfile(&diff_temp[i].tempfile);
1730                 diff_temp[i].name = NULL;
1731         }
1732 }
1733
1734 static void add_line_count(struct strbuf *out, int count)
1735 {
1736         switch (count) {
1737         case 0:
1738                 strbuf_addstr(out, "0,0");
1739                 break;
1740         case 1:
1741                 strbuf_addstr(out, "1");
1742                 break;
1743         default:
1744                 strbuf_addf(out, "1,%d", count);
1745                 break;
1746         }
1747 }
1748
1749 static void emit_rewrite_lines(struct emit_callback *ecb,
1750                                int prefix, const char *data, int size)
1751 {
1752         const char *endp = NULL;
1753         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1754
1755         while (0 < size) {
1756                 int len;
1757
1758                 endp = memchr(data, '\n', size);
1759                 len = endp ? (endp - data + 1) : size;
1760                 if (prefix != '+') {
1761                         ecb->lno_in_preimage++;
1762                         emit_del_line(reset, ecb, data, len);
1763                 } else {
1764                         ecb->lno_in_postimage++;
1765                         emit_add_line(reset, ecb, data, len);
1766                 }
1767                 size -= len;
1768                 data += len;
1769         }
1770         if (!endp)
1771                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1772 }
1773
1774 static void emit_rewrite_diff(const char *name_a,
1775                               const char *name_b,
1776                               struct diff_filespec *one,
1777                               struct diff_filespec *two,
1778                               struct userdiff_driver *textconv_one,
1779                               struct userdiff_driver *textconv_two,
1780                               struct diff_options *o)
1781 {
1782         int lc_a, lc_b;
1783         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1784         const char *a_prefix, *b_prefix;
1785         char *data_one, *data_two;
1786         size_t size_one, size_two;
1787         struct emit_callback ecbdata;
1788         struct strbuf out = STRBUF_INIT;
1789
1790         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1791                 a_prefix = o->b_prefix;
1792                 b_prefix = o->a_prefix;
1793         } else {
1794                 a_prefix = o->a_prefix;
1795                 b_prefix = o->b_prefix;
1796         }
1797
1798         name_a += (*name_a == '/');
1799         name_b += (*name_b == '/');
1800
1801         strbuf_reset(&a_name);
1802         strbuf_reset(&b_name);
1803         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1804         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1805
1806         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1807         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1808
1809         memset(&ecbdata, 0, sizeof(ecbdata));
1810         ecbdata.color_diff = want_color(o->use_color);
1811         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1812         ecbdata.opt = o;
1813         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1814                 mmfile_t mf1, mf2;
1815                 mf1.ptr = (char *)data_one;
1816                 mf2.ptr = (char *)data_two;
1817                 mf1.size = size_one;
1818                 mf2.size = size_two;
1819                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1820         }
1821         ecbdata.lno_in_preimage = 1;
1822         ecbdata.lno_in_postimage = 1;
1823
1824         lc_a = count_lines(data_one, size_one);
1825         lc_b = count_lines(data_two, size_two);
1826
1827         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1828                          a_name.buf, a_name.len, 0);
1829         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1830                          b_name.buf, b_name.len, 0);
1831
1832         strbuf_addstr(&out, "@@ -");
1833         if (!o->irreversible_delete)
1834                 add_line_count(&out, lc_a);
1835         else
1836                 strbuf_addstr(&out, "?,?");
1837         strbuf_addstr(&out, " +");
1838         add_line_count(&out, lc_b);
1839         strbuf_addstr(&out, " @@\n");
1840         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1841         strbuf_release(&out);
1842
1843         if (lc_a && !o->irreversible_delete)
1844                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1845         if (lc_b)
1846                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1847         if (textconv_one)
1848                 free((char *)data_one);
1849         if (textconv_two)
1850                 free((char *)data_two);
1851 }
1852
1853 struct diff_words_buffer {
1854         mmfile_t text;
1855         unsigned long alloc;
1856         struct diff_words_orig {
1857                 const char *begin, *end;
1858         } *orig;
1859         int orig_nr, orig_alloc;
1860 };
1861
1862 static void diff_words_append(char *line, unsigned long len,
1863                 struct diff_words_buffer *buffer)
1864 {
1865         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1866         line++;
1867         len--;
1868         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1869         buffer->text.size += len;
1870         buffer->text.ptr[buffer->text.size] = '\0';
1871 }
1872
1873 struct diff_words_style_elem {
1874         const char *prefix;
1875         const char *suffix;
1876         const char *color; /* NULL; filled in by the setup code if
1877                             * color is enabled */
1878 };
1879
1880 struct diff_words_style {
1881         enum diff_words_type type;
1882         struct diff_words_style_elem new_word, old_word, ctx;
1883         const char *newline;
1884 };
1885
1886 static struct diff_words_style diff_words_styles[] = {
1887         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1888         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1889         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1890 };
1891
1892 struct diff_words_data {
1893         struct diff_words_buffer minus, plus;
1894         const char *current_plus;
1895         int last_minus;
1896         struct diff_options *opt;
1897         regex_t *word_regex;
1898         enum diff_words_type type;
1899         struct diff_words_style *style;
1900 };
1901
1902 static int fn_out_diff_words_write_helper(struct diff_options *o,
1903                                           struct diff_words_style_elem *st_el,
1904                                           const char *newline,
1905                                           size_t count, const char *buf)
1906 {
1907         int print = 0;
1908         struct strbuf sb = STRBUF_INIT;
1909
1910         while (count) {
1911                 char *p = memchr(buf, '\n', count);
1912                 if (print)
1913                         strbuf_addstr(&sb, diff_line_prefix(o));
1914
1915                 if (p != buf) {
1916                         const char *reset = st_el->color && *st_el->color ?
1917                                             GIT_COLOR_RESET : NULL;
1918                         if (st_el->color && *st_el->color)
1919                                 strbuf_addstr(&sb, st_el->color);
1920                         strbuf_addstr(&sb, st_el->prefix);
1921                         strbuf_add(&sb, buf, p ? p - buf : count);
1922                         strbuf_addstr(&sb, st_el->suffix);
1923                         if (reset)
1924                                 strbuf_addstr(&sb, reset);
1925                 }
1926                 if (!p)
1927                         goto out;
1928
1929                 strbuf_addstr(&sb, newline);
1930                 count -= p + 1 - buf;
1931                 buf = p + 1;
1932                 print = 1;
1933                 if (count) {
1934                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1935                                          sb.buf, sb.len, 0);
1936                         strbuf_reset(&sb);
1937                 }
1938         }
1939
1940 out:
1941         if (sb.len)
1942                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1943                                  sb.buf, sb.len, 0);
1944         strbuf_release(&sb);
1945         return 0;
1946 }
1947
1948 /*
1949  * '--color-words' algorithm can be described as:
1950  *
1951  *   1. collect the minus/plus lines of a diff hunk, divided into
1952  *      minus-lines and plus-lines;
1953  *
1954  *   2. break both minus-lines and plus-lines into words and
1955  *      place them into two mmfile_t with one word for each line;
1956  *
1957  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1958  *
1959  * And for the common parts of the both file, we output the plus side text.
1960  * diff_words->current_plus is used to trace the current position of the plus file
1961  * which printed. diff_words->last_minus is used to trace the last minus word
1962  * printed.
1963  *
1964  * For '--graph' to work with '--color-words', we need to output the graph prefix
1965  * on each line of color words output. Generally, there are two conditions on
1966  * which we should output the prefix.
1967  *
1968  *   1. diff_words->last_minus == 0 &&
1969  *      diff_words->current_plus == diff_words->plus.text.ptr
1970  *
1971  *      that is: the plus text must start as a new line, and if there is no minus
1972  *      word printed, a graph prefix must be printed.
1973  *
1974  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1975  *      *(diff_words->current_plus - 1) == '\n'
1976  *
1977  *      that is: a graph prefix must be printed following a '\n'
1978  */
1979 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1980 {
1981         if ((diff_words->last_minus == 0 &&
1982                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1983                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1984                 *(diff_words->current_plus - 1) == '\n')) {
1985                 return 1;
1986         } else {
1987                 return 0;
1988         }
1989 }
1990
1991 static void fn_out_diff_words_aux(void *priv,
1992                                   long minus_first, long minus_len,
1993                                   long plus_first, long plus_len,
1994                                   const char *func, long funclen)
1995 {
1996         struct diff_words_data *diff_words = priv;
1997         struct diff_words_style *style = diff_words->style;
1998         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1999         struct diff_options *opt = diff_words->opt;
2000         const char *line_prefix;
2001
2002         assert(opt);
2003         line_prefix = diff_line_prefix(opt);
2004
2005         /* POSIX requires that first be decremented by one if len == 0... */
2006         if (minus_len) {
2007                 minus_begin = diff_words->minus.orig[minus_first].begin;
2008                 minus_end =
2009                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2010         } else
2011                 minus_begin = minus_end =
2012                         diff_words->minus.orig[minus_first].end;
2013
2014         if (plus_len) {
2015                 plus_begin = diff_words->plus.orig[plus_first].begin;
2016                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2017         } else
2018                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2019
2020         if (color_words_output_graph_prefix(diff_words)) {
2021                 fputs(line_prefix, diff_words->opt->file);
2022         }
2023         if (diff_words->current_plus != plus_begin) {
2024                 fn_out_diff_words_write_helper(diff_words->opt,
2025                                 &style->ctx, style->newline,
2026                                 plus_begin - diff_words->current_plus,
2027                                 diff_words->current_plus);
2028         }
2029         if (minus_begin != minus_end) {
2030                 fn_out_diff_words_write_helper(diff_words->opt,
2031                                 &style->old_word, style->newline,
2032                                 minus_end - minus_begin, minus_begin);
2033         }
2034         if (plus_begin != plus_end) {
2035                 fn_out_diff_words_write_helper(diff_words->opt,
2036                                 &style->new_word, style->newline,
2037                                 plus_end - plus_begin, plus_begin);
2038         }
2039
2040         diff_words->current_plus = plus_end;
2041         diff_words->last_minus = minus_first;
2042 }
2043
2044 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2045 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2046                 int *begin, int *end)
2047 {
2048         if (word_regex && *begin < buffer->size) {
2049                 regmatch_t match[1];
2050                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2051                                  buffer->size - *begin, 1, match, 0)) {
2052                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2053                                         '\n', match[0].rm_eo - match[0].rm_so);
2054                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2055                         *begin += match[0].rm_so;
2056                         return *begin >= *end;
2057                 }
2058                 return -1;
2059         }
2060
2061         /* find the next word */
2062         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2063                 (*begin)++;
2064         if (*begin >= buffer->size)
2065                 return -1;
2066
2067         /* find the end of the word */
2068         *end = *begin + 1;
2069         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2070                 (*end)++;
2071
2072         return 0;
2073 }
2074
2075 /*
2076  * This function splits the words in buffer->text, stores the list with
2077  * newline separator into out, and saves the offsets of the original words
2078  * in buffer->orig.
2079  */
2080 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2081                 regex_t *word_regex)
2082 {
2083         int i, j;
2084         long alloc = 0;
2085
2086         out->size = 0;
2087         out->ptr = NULL;
2088
2089         /* fake an empty "0th" word */
2090         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2091         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2092         buffer->orig_nr = 1;
2093
2094         for (i = 0; i < buffer->text.size; i++) {
2095                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2096                         return;
2097
2098                 /* store original boundaries */
2099                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2100                                 buffer->orig_alloc);
2101                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2102                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2103                 buffer->orig_nr++;
2104
2105                 /* store one word */
2106                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2107                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2108                 out->ptr[out->size + j - i] = '\n';
2109                 out->size += j - i + 1;
2110
2111                 i = j - 1;
2112         }
2113 }
2114
2115 /* this executes the word diff on the accumulated buffers */
2116 static void diff_words_show(struct diff_words_data *diff_words)
2117 {
2118         xpparam_t xpp;
2119         xdemitconf_t xecfg;
2120         mmfile_t minus, plus;
2121         struct diff_words_style *style = diff_words->style;
2122
2123         struct diff_options *opt = diff_words->opt;
2124         const char *line_prefix;
2125
2126         assert(opt);
2127         line_prefix = diff_line_prefix(opt);
2128
2129         /* special case: only removal */
2130         if (!diff_words->plus.text.size) {
2131                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2132                                  line_prefix, strlen(line_prefix), 0);
2133                 fn_out_diff_words_write_helper(diff_words->opt,
2134                         &style->old_word, style->newline,
2135                         diff_words->minus.text.size,
2136                         diff_words->minus.text.ptr);
2137                 diff_words->minus.text.size = 0;
2138                 return;
2139         }
2140
2141         diff_words->current_plus = diff_words->plus.text.ptr;
2142         diff_words->last_minus = 0;
2143
2144         memset(&xpp, 0, sizeof(xpp));
2145         memset(&xecfg, 0, sizeof(xecfg));
2146         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2147         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2148         xpp.flags = 0;
2149         /* as only the hunk header will be parsed, we need a 0-context */
2150         xecfg.ctxlen = 0;
2151         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2152                           diff_words, &xpp, &xecfg))
2153                 die("unable to generate word diff");
2154         free(minus.ptr);
2155         free(plus.ptr);
2156         if (diff_words->current_plus != diff_words->plus.text.ptr +
2157                         diff_words->plus.text.size) {
2158                 if (color_words_output_graph_prefix(diff_words))
2159                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2160                                          line_prefix, strlen(line_prefix), 0);
2161                 fn_out_diff_words_write_helper(diff_words->opt,
2162                         &style->ctx, style->newline,
2163                         diff_words->plus.text.ptr + diff_words->plus.text.size
2164                         - diff_words->current_plus, diff_words->current_plus);
2165         }
2166         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2167 }
2168
2169 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2170 static void diff_words_flush(struct emit_callback *ecbdata)
2171 {
2172         struct diff_options *wo = ecbdata->diff_words->opt;
2173
2174         if (ecbdata->diff_words->minus.text.size ||
2175             ecbdata->diff_words->plus.text.size)
2176                 diff_words_show(ecbdata->diff_words);
2177
2178         if (wo->emitted_symbols) {
2179                 struct diff_options *o = ecbdata->opt;
2180                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2181                 int i;
2182
2183                 /*
2184                  * NEEDSWORK:
2185                  * Instead of appending each, concat all words to a line?
2186                  */
2187                 for (i = 0; i < wol->nr; i++)
2188                         append_emitted_diff_symbol(o, &wol->buf[i]);
2189
2190                 for (i = 0; i < wol->nr; i++)
2191                         free((void *)wol->buf[i].line);
2192
2193                 wol->nr = 0;
2194         }
2195 }
2196
2197 static void diff_filespec_load_driver(struct diff_filespec *one,
2198                                       struct index_state *istate)
2199 {
2200         /* Use already-loaded driver */
2201         if (one->driver)
2202                 return;
2203
2204         if (S_ISREG(one->mode))
2205                 one->driver = userdiff_find_by_path(istate, one->path);
2206
2207         /* Fallback to default settings */
2208         if (!one->driver)
2209                 one->driver = userdiff_find_by_name("default");
2210 }
2211
2212 static const char *userdiff_word_regex(struct diff_filespec *one,
2213                                        struct index_state *istate)
2214 {
2215         diff_filespec_load_driver(one, istate);
2216         return one->driver->word_regex;
2217 }
2218
2219 static void init_diff_words_data(struct emit_callback *ecbdata,
2220                                  struct diff_options *orig_opts,
2221                                  struct diff_filespec *one,
2222                                  struct diff_filespec *two)
2223 {
2224         int i;
2225         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2226         memcpy(o, orig_opts, sizeof(struct diff_options));
2227
2228         ecbdata->diff_words =
2229                 xcalloc(1, sizeof(struct diff_words_data));
2230         ecbdata->diff_words->type = o->word_diff;
2231         ecbdata->diff_words->opt = o;
2232
2233         if (orig_opts->emitted_symbols)
2234                 o->emitted_symbols =
2235                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2236
2237         if (!o->word_regex)
2238                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2239         if (!o->word_regex)
2240                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2241         if (!o->word_regex)
2242                 o->word_regex = diff_word_regex_cfg;
2243         if (o->word_regex) {
2244                 ecbdata->diff_words->word_regex = (regex_t *)
2245                         xmalloc(sizeof(regex_t));
2246                 if (regcomp(ecbdata->diff_words->word_regex,
2247                             o->word_regex,
2248                             REG_EXTENDED | REG_NEWLINE))
2249                         die("invalid regular expression: %s",
2250                             o->word_regex);
2251         }
2252         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2253                 if (o->word_diff == diff_words_styles[i].type) {
2254                         ecbdata->diff_words->style =
2255                                 &diff_words_styles[i];
2256                         break;
2257                 }
2258         }
2259         if (want_color(o->use_color)) {
2260                 struct diff_words_style *st = ecbdata->diff_words->style;
2261                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2262                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2263                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2264         }
2265 }
2266
2267 static void free_diff_words_data(struct emit_callback *ecbdata)
2268 {
2269         if (ecbdata->diff_words) {
2270                 diff_words_flush(ecbdata);
2271                 free (ecbdata->diff_words->opt->emitted_symbols);
2272                 free (ecbdata->diff_words->opt);
2273                 free (ecbdata->diff_words->minus.text.ptr);
2274                 free (ecbdata->diff_words->minus.orig);
2275                 free (ecbdata->diff_words->plus.text.ptr);
2276                 free (ecbdata->diff_words->plus.orig);
2277                 if (ecbdata->diff_words->word_regex) {
2278                         regfree(ecbdata->diff_words->word_regex);
2279                         free(ecbdata->diff_words->word_regex);
2280                 }
2281                 FREE_AND_NULL(ecbdata->diff_words);
2282         }
2283 }
2284
2285 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2286 {
2287         if (want_color(diff_use_color))
2288                 return diff_colors[ix];
2289         return "";
2290 }
2291
2292 const char *diff_line_prefix(struct diff_options *opt)
2293 {
2294         struct strbuf *msgbuf;
2295         if (!opt->output_prefix)
2296                 return "";
2297
2298         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2299         return msgbuf->buf;
2300 }
2301
2302 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2303 {
2304         const char *cp;
2305         unsigned long allot;
2306         size_t l = len;
2307
2308         cp = line;
2309         allot = l;
2310         while (0 < l) {
2311                 (void) utf8_width(&cp, &l);
2312                 if (!cp)
2313                         break; /* truncated in the middle? */
2314         }
2315         return allot - l;
2316 }
2317
2318 static void find_lno(const char *line, struct emit_callback *ecbdata)
2319 {
2320         const char *p;
2321         ecbdata->lno_in_preimage = 0;
2322         ecbdata->lno_in_postimage = 0;
2323         p = strchr(line, '-');
2324         if (!p)
2325                 return; /* cannot happen */
2326         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2327         p = strchr(p, '+');
2328         if (!p)
2329                 return; /* cannot happen */
2330         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2331 }
2332
2333 static void fn_out_consume(void *priv, char *line, unsigned long len)
2334 {
2335         struct emit_callback *ecbdata = priv;
2336         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2337         struct diff_options *o = ecbdata->opt;
2338
2339         o->found_changes = 1;
2340
2341         if (ecbdata->header) {
2342                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2343                                  ecbdata->header->buf, ecbdata->header->len, 0);
2344                 strbuf_reset(ecbdata->header);
2345                 ecbdata->header = NULL;
2346         }
2347
2348         if (ecbdata->label_path[0]) {
2349                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2350                                  ecbdata->label_path[0],
2351                                  strlen(ecbdata->label_path[0]), 0);
2352                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2353                                  ecbdata->label_path[1],
2354                                  strlen(ecbdata->label_path[1]), 0);
2355                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2356         }
2357
2358         if (diff_suppress_blank_empty
2359             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2360                 line[0] = '\n';
2361                 len = 1;
2362         }
2363
2364         if (line[0] == '@') {
2365                 if (ecbdata->diff_words)
2366                         diff_words_flush(ecbdata);
2367                 len = sane_truncate_line(ecbdata, line, len);
2368                 find_lno(line, ecbdata);
2369                 emit_hunk_header(ecbdata, line, len);
2370                 return;
2371         }
2372
2373         if (ecbdata->diff_words) {
2374                 enum diff_symbol s =
2375                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2376                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2377                 if (line[0] == '-') {
2378                         diff_words_append(line, len,
2379                                           &ecbdata->diff_words->minus);
2380                         return;
2381                 } else if (line[0] == '+') {
2382                         diff_words_append(line, len,
2383                                           &ecbdata->diff_words->plus);
2384                         return;
2385                 } else if (starts_with(line, "\\ ")) {
2386                         /*
2387                          * Eat the "no newline at eof" marker as if we
2388                          * saw a "+" or "-" line with nothing on it,
2389                          * and return without diff_words_flush() to
2390                          * defer processing. If this is the end of
2391                          * preimage, more "+" lines may come after it.
2392                          */
2393                         return;
2394                 }
2395                 diff_words_flush(ecbdata);
2396                 emit_diff_symbol(o, s, line, len, 0);
2397                 return;
2398         }
2399
2400         switch (line[0]) {
2401         case '+':
2402                 ecbdata->lno_in_postimage++;
2403                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2404                 break;
2405         case '-':
2406                 ecbdata->lno_in_preimage++;
2407                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2408                 break;
2409         case ' ':
2410                 ecbdata->lno_in_postimage++;
2411                 ecbdata->lno_in_preimage++;
2412                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2413                 break;
2414         default:
2415                 /* incomplete line at the end */
2416                 ecbdata->lno_in_preimage++;
2417                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2418                                  line, len, 0);
2419                 break;
2420         }
2421 }
2422
2423 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2424 {
2425         const char *old_name = a;
2426         const char *new_name = b;
2427         int pfx_length, sfx_length;
2428         int pfx_adjust_for_slash;
2429         int len_a = strlen(a);
2430         int len_b = strlen(b);
2431         int a_midlen, b_midlen;
2432         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2433         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2434
2435         if (qlen_a || qlen_b) {
2436                 quote_c_style(a, name, NULL, 0);
2437                 strbuf_addstr(name, " => ");
2438                 quote_c_style(b, name, NULL, 0);
2439                 return;
2440         }
2441
2442         /* Find common prefix */
2443         pfx_length = 0;
2444         while (*old_name && *new_name && *old_name == *new_name) {
2445                 if (*old_name == '/')
2446                         pfx_length = old_name - a + 1;
2447                 old_name++;
2448                 new_name++;
2449         }
2450
2451         /* Find common suffix */
2452         old_name = a + len_a;
2453         new_name = b + len_b;
2454         sfx_length = 0;
2455         /*
2456          * If there is a common prefix, it must end in a slash.  In
2457          * that case we let this loop run 1 into the prefix to see the
2458          * same slash.
2459          *
2460          * If there is no common prefix, we cannot do this as it would
2461          * underrun the input strings.
2462          */
2463         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2464         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2465                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2466                *old_name == *new_name) {
2467                 if (*old_name == '/')
2468                         sfx_length = len_a - (old_name - a);
2469                 old_name--;
2470                 new_name--;
2471         }
2472
2473         /*
2474          * pfx{mid-a => mid-b}sfx
2475          * {pfx-a => pfx-b}sfx
2476          * pfx{sfx-a => sfx-b}
2477          * name-a => name-b
2478          */
2479         a_midlen = len_a - pfx_length - sfx_length;
2480         b_midlen = len_b - pfx_length - sfx_length;
2481         if (a_midlen < 0)
2482                 a_midlen = 0;
2483         if (b_midlen < 0)
2484                 b_midlen = 0;
2485
2486         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2487         if (pfx_length + sfx_length) {
2488                 strbuf_add(name, a, pfx_length);
2489                 strbuf_addch(name, '{');
2490         }
2491         strbuf_add(name, a + pfx_length, a_midlen);
2492         strbuf_addstr(name, " => ");
2493         strbuf_add(name, b + pfx_length, b_midlen);
2494         if (pfx_length + sfx_length) {
2495                 strbuf_addch(name, '}');
2496                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2497         }
2498 }
2499
2500 struct diffstat_t {
2501         int nr;
2502         int alloc;
2503         struct diffstat_file {
2504                 char *from_name;
2505                 char *name;
2506                 char *print_name;
2507                 const char *comments;
2508                 unsigned is_unmerged:1;
2509                 unsigned is_binary:1;
2510                 unsigned is_renamed:1;
2511                 unsigned is_interesting:1;
2512                 uintmax_t added, deleted;
2513         } **files;
2514 };
2515
2516 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2517                                           const char *name_a,
2518                                           const char *name_b)
2519 {
2520         struct diffstat_file *x;
2521         x = xcalloc(1, sizeof(*x));
2522         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2523         diffstat->files[diffstat->nr++] = x;
2524         if (name_b) {
2525                 x->from_name = xstrdup(name_a);
2526                 x->name = xstrdup(name_b);
2527                 x->is_renamed = 1;
2528         }
2529         else {
2530                 x->from_name = NULL;
2531                 x->name = xstrdup(name_a);
2532         }
2533         return x;
2534 }
2535
2536 static void diffstat_consume(void *priv, char *line, unsigned long len)
2537 {
2538         struct diffstat_t *diffstat = priv;
2539         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2540
2541         if (line[0] == '+')
2542                 x->added++;
2543         else if (line[0] == '-')
2544                 x->deleted++;
2545 }
2546
2547 const char mime_boundary_leader[] = "------------";
2548
2549 static int scale_linear(int it, int width, int max_change)
2550 {
2551         if (!it)
2552                 return 0;
2553         /*
2554          * make sure that at least one '-' or '+' is printed if
2555          * there is any change to this path. The easiest way is to
2556          * scale linearly as if the alloted width is one column shorter
2557          * than it is, and then add 1 to the result.
2558          */
2559         return 1 + (it * (width - 1) / max_change);
2560 }
2561
2562 static void show_graph(struct strbuf *out, char ch, int cnt,
2563                        const char *set, const char *reset)
2564 {
2565         if (cnt <= 0)
2566                 return;
2567         strbuf_addstr(out, set);
2568         strbuf_addchars(out, ch, cnt);
2569         strbuf_addstr(out, reset);
2570 }
2571
2572 static void fill_print_name(struct diffstat_file *file)
2573 {
2574         struct strbuf pname = STRBUF_INIT;
2575
2576         if (file->print_name)
2577                 return;
2578
2579         if (file->is_renamed)
2580                 pprint_rename(&pname, file->from_name, file->name);
2581         else
2582                 quote_c_style(file->name, &pname, NULL, 0);
2583
2584         if (file->comments)
2585                 strbuf_addf(&pname, " (%s)", file->comments);
2586
2587         file->print_name = strbuf_detach(&pname, NULL);
2588 }
2589
2590 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2591                 int files, int insertions, int deletions)
2592 {
2593         struct strbuf sb = STRBUF_INIT;
2594
2595         if (!files) {
2596                 assert(insertions == 0 && deletions == 0);
2597                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2598                                  NULL, 0, 0);
2599                 return;
2600         }
2601
2602         strbuf_addf(&sb,
2603                     (files == 1) ? " %d file changed" : " %d files changed",
2604                     files);
2605
2606         /*
2607          * For binary diff, the caller may want to print "x files
2608          * changed" with insertions == 0 && deletions == 0.
2609          *
2610          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2611          * is probably less confusing (i.e skip over "2 files changed
2612          * but nothing about added/removed lines? Is this a bug in Git?").
2613          */
2614         if (insertions || deletions == 0) {
2615                 strbuf_addf(&sb,
2616                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2617                             insertions);
2618         }
2619
2620         if (deletions || insertions == 0) {
2621                 strbuf_addf(&sb,
2622                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2623                             deletions);
2624         }
2625         strbuf_addch(&sb, '\n');
2626         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2627                          sb.buf, sb.len, 0);
2628         strbuf_release(&sb);
2629 }
2630
2631 void print_stat_summary(FILE *fp, int files,
2632                         int insertions, int deletions)
2633 {
2634         struct diff_options o;
2635         memset(&o, 0, sizeof(o));
2636         o.file = fp;
2637
2638         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2639 }
2640
2641 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2642 {
2643         int i, len, add, del, adds = 0, dels = 0;
2644         uintmax_t max_change = 0, max_len = 0;
2645         int total_files = data->nr, count;
2646         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2647         const char *reset, *add_c, *del_c;
2648         int extra_shown = 0;
2649         const char *line_prefix = diff_line_prefix(options);
2650         struct strbuf out = STRBUF_INIT;
2651
2652         if (data->nr == 0)
2653                 return;
2654
2655         count = options->stat_count ? options->stat_count : data->nr;
2656
2657         reset = diff_get_color_opt(options, DIFF_RESET);
2658         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2659         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2660
2661         /*
2662          * Find the longest filename and max number of changes
2663          */
2664         for (i = 0; (i < count) && (i < data->nr); i++) {
2665                 struct diffstat_file *file = data->files[i];
2666                 uintmax_t change = file->added + file->deleted;
2667
2668                 if (!file->is_interesting && (change == 0)) {
2669                         count++; /* not shown == room for one more */
2670                         continue;
2671                 }
2672                 fill_print_name(file);
2673                 len = strlen(file->print_name);
2674                 if (max_len < len)
2675                         max_len = len;
2676
2677                 if (file->is_unmerged) {
2678                         /* "Unmerged" is 8 characters */
2679                         bin_width = bin_width < 8 ? 8 : bin_width;
2680                         continue;
2681                 }
2682                 if (file->is_binary) {
2683                         /* "Bin XXX -> YYY bytes" */
2684                         int w = 14 + decimal_width(file->added)
2685                                 + decimal_width(file->deleted);
2686                         bin_width = bin_width < w ? w : bin_width;
2687                         /* Display change counts aligned with "Bin" */
2688                         number_width = 3;
2689                         continue;
2690                 }
2691
2692                 if (max_change < change)
2693                         max_change = change;
2694         }
2695         count = i; /* where we can stop scanning in data->files[] */
2696
2697         /*
2698          * We have width = stat_width or term_columns() columns total.
2699          * We want a maximum of min(max_len, stat_name_width) for the name part.
2700          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2701          * We also need 1 for " " and 4 + decimal_width(max_change)
2702          * for " | NNNN " and one the empty column at the end, altogether
2703          * 6 + decimal_width(max_change).
2704          *
2705          * If there's not enough space, we will use the smaller of
2706          * stat_name_width (if set) and 5/8*width for the filename,
2707          * and the rest for constant elements + graph part, but no more
2708          * than stat_graph_width for the graph part.
2709          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2710          * for the standard terminal size).
2711          *
2712          * In other words: stat_width limits the maximum width, and
2713          * stat_name_width fixes the maximum width of the filename,
2714          * and is also used to divide available columns if there
2715          * aren't enough.
2716          *
2717          * Binary files are displayed with "Bin XXX -> YYY bytes"
2718          * instead of the change count and graph. This part is treated
2719          * similarly to the graph part, except that it is not
2720          * "scaled". If total width is too small to accommodate the
2721          * guaranteed minimum width of the filename part and the
2722          * separators and this message, this message will "overflow"
2723          * making the line longer than the maximum width.
2724          */
2725
2726         if (options->stat_width == -1)
2727                 width = term_columns() - strlen(line_prefix);
2728         else
2729                 width = options->stat_width ? options->stat_width : 80;
2730         number_width = decimal_width(max_change) > number_width ?
2731                 decimal_width(max_change) : number_width;
2732
2733         if (options->stat_graph_width == -1)
2734                 options->stat_graph_width = diff_stat_graph_width;
2735
2736         /*
2737          * Guarantee 3/8*16==6 for the graph part
2738          * and 5/8*16==10 for the filename part
2739          */
2740         if (width < 16 + 6 + number_width)
2741                 width = 16 + 6 + number_width;
2742
2743         /*
2744          * First assign sizes that are wanted, ignoring available width.
2745          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2746          * starting from "XXX" should fit in graph_width.
2747          */
2748         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2749         if (options->stat_graph_width &&
2750             options->stat_graph_width < graph_width)
2751                 graph_width = options->stat_graph_width;
2752
2753         name_width = (options->stat_name_width > 0 &&
2754                       options->stat_name_width < max_len) ?
2755                 options->stat_name_width : max_len;
2756
2757         /*
2758          * Adjust adjustable widths not to exceed maximum width
2759          */
2760         if (name_width + number_width + 6 + graph_width > width) {
2761                 if (graph_width > width * 3/8 - number_width - 6) {
2762                         graph_width = width * 3/8 - number_width - 6;
2763                         if (graph_width < 6)
2764                                 graph_width = 6;
2765                 }
2766
2767                 if (options->stat_graph_width &&
2768                     graph_width > options->stat_graph_width)
2769                         graph_width = options->stat_graph_width;
2770                 if (name_width > width - number_width - 6 - graph_width)
2771                         name_width = width - number_width - 6 - graph_width;
2772                 else
2773                         graph_width = width - number_width - 6 - name_width;
2774         }
2775
2776         /*
2777          * From here name_width is the width of the name area,
2778          * and graph_width is the width of the graph area.
2779          * max_change is used to scale graph properly.
2780          */
2781         for (i = 0; i < count; i++) {
2782                 const char *prefix = "";
2783                 struct diffstat_file *file = data->files[i];
2784                 char *name = file->print_name;
2785                 uintmax_t added = file->added;
2786                 uintmax_t deleted = file->deleted;
2787                 int name_len;
2788
2789                 if (!file->is_interesting && (added + deleted == 0))
2790                         continue;
2791
2792                 /*
2793                  * "scale" the filename
2794                  */
2795                 len = name_width;
2796                 name_len = strlen(name);
2797                 if (name_width < name_len) {
2798                         char *slash;
2799                         prefix = "...";
2800                         len -= 3;
2801                         name += name_len - len;
2802                         slash = strchr(name, '/');
2803                         if (slash)
2804                                 name = slash;
2805                 }
2806
2807                 if (file->is_binary) {
2808                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2809                         strbuf_addf(&out, " %*s", number_width, "Bin");
2810                         if (!added && !deleted) {
2811                                 strbuf_addch(&out, '\n');
2812                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2813                                                  out.buf, out.len, 0);
2814                                 strbuf_reset(&out);
2815                                 continue;
2816                         }
2817                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2818                                 del_c, deleted, reset);
2819                         strbuf_addstr(&out, " -> ");
2820                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2821                                 add_c, added, reset);
2822                         strbuf_addstr(&out, " bytes\n");
2823                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2824                                          out.buf, out.len, 0);
2825                         strbuf_reset(&out);
2826                         continue;
2827                 }
2828                 else if (file->is_unmerged) {
2829                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2830                         strbuf_addstr(&out, " Unmerged\n");
2831                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2832                                          out.buf, out.len, 0);
2833                         strbuf_reset(&out);
2834                         continue;
2835                 }
2836
2837                 /*
2838                  * scale the add/delete
2839                  */
2840                 add = added;
2841                 del = deleted;
2842
2843                 if (graph_width <= max_change) {
2844                         int total = scale_linear(add + del, graph_width, max_change);
2845                         if (total < 2 && add && del)
2846                                 /* width >= 2 due to the sanity check */
2847                                 total = 2;
2848                         if (add < del) {
2849                                 add = scale_linear(add, graph_width, max_change);
2850                                 del = total - add;
2851                         } else {
2852                                 del = scale_linear(del, graph_width, max_change);
2853                                 add = total - del;
2854                         }
2855                 }
2856                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2857                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2858                         number_width, added + deleted,
2859                         added + deleted ? " " : "");
2860                 show_graph(&out, '+', add, add_c, reset);
2861                 show_graph(&out, '-', del, del_c, reset);
2862                 strbuf_addch(&out, '\n');
2863                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2864                                  out.buf, out.len, 0);
2865                 strbuf_reset(&out);
2866         }
2867
2868         for (i = 0; i < data->nr; i++) {
2869                 struct diffstat_file *file = data->files[i];
2870                 uintmax_t added = file->added;
2871                 uintmax_t deleted = file->deleted;
2872
2873                 if (file->is_unmerged ||
2874                     (!file->is_interesting && (added + deleted == 0))) {
2875                         total_files--;
2876                         continue;
2877                 }
2878
2879                 if (!file->is_binary) {
2880                         adds += added;
2881                         dels += deleted;
2882                 }
2883                 if (i < count)
2884                         continue;
2885                 if (!extra_shown)
2886                         emit_diff_symbol(options,
2887                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2888                                          NULL, 0, 0);
2889                 extra_shown = 1;
2890         }
2891
2892         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2893         strbuf_release(&out);
2894 }
2895
2896 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2897 {
2898         int i, adds = 0, dels = 0, total_files = data->nr;
2899
2900         if (data->nr == 0)
2901                 return;
2902
2903         for (i = 0; i < data->nr; i++) {
2904                 int added = data->files[i]->added;
2905                 int deleted = data->files[i]->deleted;
2906
2907                 if (data->files[i]->is_unmerged ||
2908                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2909                         total_files--;
2910                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2911                         adds += added;
2912                         dels += deleted;
2913                 }
2914         }
2915         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2916 }
2917
2918 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2919 {
2920         int i;
2921
2922         if (data->nr == 0)
2923                 return;
2924
2925         for (i = 0; i < data->nr; i++) {
2926                 struct diffstat_file *file = data->files[i];
2927
2928                 fprintf(options->file, "%s", diff_line_prefix(options));
2929
2930                 if (file->is_binary)
2931                         fprintf(options->file, "-\t-\t");
2932                 else
2933                         fprintf(options->file,
2934                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2935                                 file->added, file->deleted);
2936                 if (options->line_termination) {
2937                         fill_print_name(file);
2938                         if (!file->is_renamed)
2939                                 write_name_quoted(file->name, options->file,
2940                                                   options->line_termination);
2941                         else {
2942                                 fputs(file->print_name, options->file);
2943                                 putc(options->line_termination, options->file);
2944                         }
2945                 } else {
2946                         if (file->is_renamed) {
2947                                 putc('\0', options->file);
2948                                 write_name_quoted(file->from_name, options->file, '\0');
2949                         }
2950                         write_name_quoted(file->name, options->file, '\0');
2951                 }
2952         }
2953 }
2954
2955 struct dirstat_file {
2956         const char *name;
2957         unsigned long changed;
2958 };
2959
2960 struct dirstat_dir {
2961         struct dirstat_file *files;
2962         int alloc, nr, permille, cumulative;
2963 };
2964
2965 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2966                 unsigned long changed, const char *base, int baselen)
2967 {
2968         unsigned long sum_changes = 0;
2969         unsigned int sources = 0;
2970         const char *line_prefix = diff_line_prefix(opt);
2971
2972         while (dir->nr) {
2973                 struct dirstat_file *f = dir->files;
2974                 int namelen = strlen(f->name);
2975                 unsigned long changes;
2976                 char *slash;
2977
2978                 if (namelen < baselen)
2979                         break;
2980                 if (memcmp(f->name, base, baselen))
2981                         break;
2982                 slash = strchr(f->name + baselen, '/');
2983                 if (slash) {
2984                         int newbaselen = slash + 1 - f->name;
2985                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2986                         sources++;
2987                 } else {
2988                         changes = f->changed;
2989                         dir->files++;
2990                         dir->nr--;
2991                         sources += 2;
2992                 }
2993                 sum_changes += changes;
2994         }
2995
2996         /*
2997          * We don't report dirstat's for
2998          *  - the top level
2999          *  - or cases where everything came from a single directory
3000          *    under this directory (sources == 1).
3001          */
3002         if (baselen && sources != 1) {
3003                 if (sum_changes) {
3004                         int permille = sum_changes * 1000 / changed;
3005                         if (permille >= dir->permille) {
3006                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3007                                         permille / 10, permille % 10, baselen, base);
3008                                 if (!dir->cumulative)
3009                                         return 0;
3010                         }
3011                 }
3012         }
3013         return sum_changes;
3014 }
3015
3016 static int dirstat_compare(const void *_a, const void *_b)
3017 {
3018         const struct dirstat_file *a = _a;
3019         const struct dirstat_file *b = _b;
3020         return strcmp(a->name, b->name);
3021 }
3022
3023 static void show_dirstat(struct diff_options *options)
3024 {
3025         int i;
3026         unsigned long changed;
3027         struct dirstat_dir dir;
3028         struct diff_queue_struct *q = &diff_queued_diff;
3029
3030         dir.files = NULL;
3031         dir.alloc = 0;
3032         dir.nr = 0;
3033         dir.permille = options->dirstat_permille;
3034         dir.cumulative = options->flags.dirstat_cumulative;
3035
3036         changed = 0;
3037         for (i = 0; i < q->nr; i++) {
3038                 struct diff_filepair *p = q->queue[i];
3039                 const char *name;
3040                 unsigned long copied, added, damage;
3041
3042                 name = p->two->path ? p->two->path : p->one->path;
3043
3044                 if (p->one->oid_valid && p->two->oid_valid &&
3045                     oideq(&p->one->oid, &p->two->oid)) {
3046                         /*
3047                          * The SHA1 has not changed, so pre-/post-content is
3048                          * identical. We can therefore skip looking at the
3049                          * file contents altogether.
3050                          */
3051                         damage = 0;
3052                         goto found_damage;
3053                 }
3054
3055                 if (options->flags.dirstat_by_file) {
3056                         /*
3057                          * In --dirstat-by-file mode, we don't really need to
3058                          * look at the actual file contents at all.
3059                          * The fact that the SHA1 changed is enough for us to
3060                          * add this file to the list of results
3061                          * (with each file contributing equal damage).
3062                          */
3063                         damage = 1;
3064                         goto found_damage;
3065                 }
3066
3067                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3068                         diff_populate_filespec(options->repo, p->one, 0);
3069                         diff_populate_filespec(options->repo, p->two, 0);
3070                         diffcore_count_changes(options->repo,
3071                                                p->one, p->two, NULL, NULL,
3072                                                &copied, &added);
3073                         diff_free_filespec_data(p->one);
3074                         diff_free_filespec_data(p->two);
3075                 } else if (DIFF_FILE_VALID(p->one)) {
3076                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3077                         copied = added = 0;
3078                         diff_free_filespec_data(p->one);
3079                 } else if (DIFF_FILE_VALID(p->two)) {
3080                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3081                         copied = 0;
3082                         added = p->two->size;
3083                         diff_free_filespec_data(p->two);
3084                 } else
3085                         continue;
3086
3087                 /*
3088                  * Original minus copied is the removed material,
3089                  * added is the new material.  They are both damages
3090                  * made to the preimage.
3091                  * If the resulting damage is zero, we know that
3092                  * diffcore_count_changes() considers the two entries to
3093                  * be identical, but since the oid changed, we
3094                  * know that there must have been _some_ kind of change,
3095                  * so we force all entries to have damage > 0.
3096                  */
3097                 damage = (p->one->size - copied) + added;
3098                 if (!damage)
3099                         damage = 1;
3100
3101 found_damage:
3102                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3103                 dir.files[dir.nr].name = name;
3104                 dir.files[dir.nr].changed = damage;
3105                 changed += damage;
3106                 dir.nr++;
3107         }
3108
3109         /* This can happen even with many files, if everything was renames */
3110         if (!changed)
3111                 return;
3112
3113         /* Show all directories with more than x% of the changes */
3114         QSORT(dir.files, dir.nr, dirstat_compare);
3115         gather_dirstat(options, &dir, changed, "", 0);
3116 }
3117
3118 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3119 {
3120         int i;
3121         unsigned long changed;
3122         struct dirstat_dir dir;
3123
3124         if (data->nr == 0)
3125                 return;
3126
3127         dir.files = NULL;
3128         dir.alloc = 0;
3129         dir.nr = 0;
3130         dir.permille = options->dirstat_permille;
3131         dir.cumulative = options->flags.dirstat_cumulative;
3132
3133         changed = 0;
3134         for (i = 0; i < data->nr; i++) {
3135                 struct diffstat_file *file = data->files[i];
3136                 unsigned long damage = file->added + file->deleted;
3137                 if (file->is_binary)
3138                         /*
3139                          * binary files counts bytes, not lines. Must find some
3140                          * way to normalize binary bytes vs. textual lines.
3141                          * The following heuristic assumes that there are 64
3142                          * bytes per "line".
3143                          * This is stupid and ugly, but very cheap...
3144                          */
3145                         damage = DIV_ROUND_UP(damage, 64);
3146                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3147                 dir.files[dir.nr].name = file->name;
3148                 dir.files[dir.nr].changed = damage;
3149                 changed += damage;
3150                 dir.nr++;
3151         }
3152
3153         /* This can happen even with many files, if everything was renames */
3154         if (!changed)
3155                 return;
3156
3157         /* Show all directories with more than x% of the changes */
3158         QSORT(dir.files, dir.nr, dirstat_compare);
3159         gather_dirstat(options, &dir, changed, "", 0);
3160 }
3161
3162 static void free_diffstat_info(struct diffstat_t *diffstat)
3163 {
3164         int i;
3165         for (i = 0; i < diffstat->nr; i++) {
3166                 struct diffstat_file *f = diffstat->files[i];
3167                 free(f->print_name);
3168                 free(f->name);
3169                 free(f->from_name);
3170                 free(f);
3171         }
3172         free(diffstat->files);
3173 }
3174
3175 struct checkdiff_t {
3176         const char *filename;
3177         int lineno;
3178         int conflict_marker_size;
3179         struct diff_options *o;
3180         unsigned ws_rule;
3181         unsigned status;
3182 };
3183
3184 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3185 {
3186         char firstchar;
3187         int cnt;
3188
3189         if (len < marker_size + 1)
3190                 return 0;
3191         firstchar = line[0];
3192         switch (firstchar) {
3193         case '=': case '>': case '<': case '|':
3194                 break;
3195         default:
3196                 return 0;
3197         }
3198         for (cnt = 1; cnt < marker_size; cnt++)
3199                 if (line[cnt] != firstchar)
3200                         return 0;
3201         /* line[1] thru line[marker_size-1] are same as firstchar */
3202         if (len < marker_size + 1 || !isspace(line[marker_size]))
3203                 return 0;
3204         return 1;
3205 }
3206
3207 static void checkdiff_consume_hunk(void *priv,
3208                                    long ob, long on, long nb, long nn,
3209                                    const char *func, long funclen)
3210
3211 {
3212         struct checkdiff_t *data = priv;
3213         data->lineno = nb - 1;
3214 }
3215
3216 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3217 {
3218         struct checkdiff_t *data = priv;
3219         int marker_size = data->conflict_marker_size;
3220         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3221         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3222         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3223         char *err;
3224         const char *line_prefix;
3225
3226         assert(data->o);
3227         line_prefix = diff_line_prefix(data->o);
3228
3229         if (line[0] == '+') {
3230                 unsigned bad;
3231                 data->lineno++;
3232                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3233                         data->status |= 1;
3234                         fprintf(data->o->file,
3235                                 "%s%s:%d: leftover conflict marker\n",
3236                                 line_prefix, data->filename, data->lineno);
3237                 }
3238                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3239                 if (!bad)
3240                         return;
3241                 data->status |= bad;
3242                 err = whitespace_error_string(bad);
3243                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3244                         line_prefix, data->filename, data->lineno, err);
3245                 free(err);
3246                 emit_line(data->o, set, reset, line, 1);
3247                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3248                               data->o->file, set, reset, ws);
3249         } else if (line[0] == ' ') {
3250                 data->lineno++;
3251         }
3252 }
3253
3254 static unsigned char *deflate_it(char *data,
3255                                  unsigned long size,
3256                                  unsigned long *result_size)
3257 {
3258         int bound;
3259         unsigned char *deflated;
3260         git_zstream stream;
3261
3262         git_deflate_init(&stream, zlib_compression_level);
3263         bound = git_deflate_bound(&stream, size);
3264         deflated = xmalloc(bound);
3265         stream.next_out = deflated;
3266         stream.avail_out = bound;
3267
3268         stream.next_in = (unsigned char *)data;
3269         stream.avail_in = size;
3270         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3271                 ; /* nothing */
3272         git_deflate_end(&stream);
3273         *result_size = stream.total_out;
3274         return deflated;
3275 }
3276
3277 static void emit_binary_diff_body(struct diff_options *o,
3278                                   mmfile_t *one, mmfile_t *two)
3279 {
3280         void *cp;
3281         void *delta;
3282         void *deflated;
3283         void *data;
3284         unsigned long orig_size;
3285         unsigned long delta_size;
3286         unsigned long deflate_size;
3287         unsigned long data_size;
3288
3289         /* We could do deflated delta, or we could do just deflated two,
3290          * whichever is smaller.
3291          */
3292         delta = NULL;
3293         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3294         if (one->size && two->size) {
3295                 delta = diff_delta(one->ptr, one->size,
3296                                    two->ptr, two->size,
3297                                    &delta_size, deflate_size);
3298                 if (delta) {
3299                         void *to_free = delta;
3300                         orig_size = delta_size;
3301                         delta = deflate_it(delta, delta_size, &delta_size);
3302                         free(to_free);
3303                 }
3304         }
3305
3306         if (delta && delta_size < deflate_size) {
3307                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3308                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3309                                  s, strlen(s), 0);
3310                 free(s);
3311                 free(deflated);
3312                 data = delta;
3313                 data_size = delta_size;
3314         } else {
3315                 char *s = xstrfmt("%lu", two->size);
3316                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3317                                  s, strlen(s), 0);
3318                 free(s);
3319                 free(delta);
3320                 data = deflated;
3321                 data_size = deflate_size;
3322         }
3323
3324         /* emit data encoded in base85 */
3325         cp = data;
3326         while (data_size) {
3327                 int len;
3328                 int bytes = (52 < data_size) ? 52 : data_size;
3329                 char line[71];
3330                 data_size -= bytes;
3331                 if (bytes <= 26)
3332                         line[0] = bytes + 'A' - 1;
3333                 else
3334                         line[0] = bytes - 26 + 'a' - 1;
3335                 encode_85(line + 1, cp, bytes);
3336                 cp = (char *) cp + bytes;
3337
3338                 len = strlen(line);
3339                 line[len++] = '\n';
3340                 line[len] = '\0';
3341
3342                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3343                                  line, len, 0);
3344         }
3345         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3346         free(data);
3347 }
3348
3349 static void emit_binary_diff(struct diff_options *o,
3350                              mmfile_t *one, mmfile_t *two)
3351 {
3352         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3353         emit_binary_diff_body(o, one, two);
3354         emit_binary_diff_body(o, two, one);
3355 }
3356
3357 int diff_filespec_is_binary(struct repository *r,
3358                             struct diff_filespec *one)
3359 {
3360         if (one->is_binary == -1) {
3361                 diff_filespec_load_driver(one, r->index);
3362                 if (one->driver->binary != -1)
3363                         one->is_binary = one->driver->binary;
3364                 else {
3365                         if (!one->data && DIFF_FILE_VALID(one))
3366                                 diff_populate_filespec(r, one, CHECK_BINARY);
3367                         if (one->is_binary == -1 && one->data)
3368                                 one->is_binary = buffer_is_binary(one->data,
3369                                                 one->size);
3370                         if (one->is_binary == -1)
3371                                 one->is_binary = 0;
3372                 }
3373         }
3374         return one->is_binary;
3375 }
3376
3377 static const struct userdiff_funcname *
3378 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3379 {
3380         diff_filespec_load_driver(one, o->repo->index);
3381         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3382 }
3383
3384 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3385 {
3386         if (!options->a_prefix)
3387                 options->a_prefix = a;
3388         if (!options->b_prefix)
3389                 options->b_prefix = b;
3390 }
3391
3392 struct userdiff_driver *get_textconv(struct repository *r,
3393                                      struct diff_filespec *one)
3394 {
3395         if (!DIFF_FILE_VALID(one))
3396                 return NULL;
3397
3398         diff_filespec_load_driver(one, r->index);
3399         return userdiff_get_textconv(r, one->driver);
3400 }
3401
3402 static void builtin_diff(const char *name_a,
3403                          const char *name_b,
3404                          struct diff_filespec *one,
3405                          struct diff_filespec *two,
3406                          const char *xfrm_msg,
3407                          int must_show_header,
3408                          struct diff_options *o,
3409                          int complete_rewrite)
3410 {
3411         mmfile_t mf1, mf2;
3412         const char *lbl[2];
3413         char *a_one, *b_two;
3414         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3415         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3416         const char *a_prefix, *b_prefix;
3417         struct userdiff_driver *textconv_one = NULL;
3418         struct userdiff_driver *textconv_two = NULL;
3419         struct strbuf header = STRBUF_INIT;
3420         const char *line_prefix = diff_line_prefix(o);
3421
3422         diff_set_mnemonic_prefix(o, "a/", "b/");
3423         if (o->flags.reverse_diff) {
3424                 a_prefix = o->b_prefix;
3425                 b_prefix = o->a_prefix;
3426         } else {
3427                 a_prefix = o->a_prefix;
3428                 b_prefix = o->b_prefix;
3429         }
3430
3431         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3432             (!one->mode || S_ISGITLINK(one->mode)) &&
3433             (!two->mode || S_ISGITLINK(two->mode))) {
3434                 show_submodule_summary(o, one->path ? one->path : two->path,
3435                                 &one->oid, &two->oid,
3436                                 two->dirty_submodule);
3437                 return;
3438         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3439                    (!one->mode || S_ISGITLINK(one->mode)) &&
3440                    (!two->mode || S_ISGITLINK(two->mode))) {
3441                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3442                                 &one->oid, &two->oid,
3443                                 two->dirty_submodule);
3444                 return;
3445         }
3446
3447         if (o->flags.allow_textconv) {
3448                 textconv_one = get_textconv(o->repo, one);
3449                 textconv_two = get_textconv(o->repo, two);
3450         }
3451
3452         /* Never use a non-valid filename anywhere if at all possible */
3453         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3454         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3455
3456         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3457         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3458         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3459         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3460         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3461         if (lbl[0][0] == '/') {
3462                 /* /dev/null */
3463                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3464                 if (xfrm_msg)
3465                         strbuf_addstr(&header, xfrm_msg);
3466                 must_show_header = 1;
3467         }
3468         else if (lbl[1][0] == '/') {
3469                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3470                 if (xfrm_msg)
3471                         strbuf_addstr(&header, xfrm_msg);
3472                 must_show_header = 1;
3473         }
3474         else {
3475                 if (one->mode != two->mode) {
3476                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3477                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3478                         must_show_header = 1;
3479                 }
3480                 if (xfrm_msg)
3481                         strbuf_addstr(&header, xfrm_msg);
3482
3483                 /*
3484                  * we do not run diff between different kind
3485                  * of objects.
3486                  */
3487                 if ((one->mode ^ two->mode) & S_IFMT)
3488                         goto free_ab_and_return;
3489                 if (complete_rewrite &&
3490                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3491                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3492                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3493                                          header.buf, header.len, 0);
3494                         strbuf_reset(&header);
3495                         emit_rewrite_diff(name_a, name_b, one, two,
3496                                           textconv_one, textconv_two, o);
3497                         o->found_changes = 1;
3498                         goto free_ab_and_return;
3499                 }
3500         }
3501
3502         if (o->irreversible_delete && lbl[1][0] == '/') {
3503                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3504                                  header.len, 0);
3505                 strbuf_reset(&header);
3506                 goto free_ab_and_return;
3507         } else if (!o->flags.text &&
3508                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3509                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3510                 struct strbuf sb = STRBUF_INIT;
3511                 if (!one->data && !two->data &&
3512                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3513                     !o->flags.binary) {
3514                         if (oideq(&one->oid, &two->oid)) {
3515                                 if (must_show_header)
3516                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3517                                                          header.buf, header.len,
3518                                                          0);
3519                                 goto free_ab_and_return;
3520                         }
3521                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3522                                          header.buf, header.len, 0);
3523                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3524                                     diff_line_prefix(o), lbl[0], lbl[1]);
3525                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3526                                          sb.buf, sb.len, 0);
3527                         strbuf_release(&sb);
3528                         goto free_ab_and_return;
3529                 }
3530                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3531                     fill_mmfile(o->repo, &mf2, two) < 0)
3532                         die("unable to read files to diff");
3533                 /* Quite common confusing case */
3534                 if (mf1.size == mf2.size &&
3535                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3536                         if (must_show_header)
3537                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3538                                                  header.buf, header.len, 0);
3539                         goto free_ab_and_return;
3540                 }
3541                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3542                 strbuf_reset(&header);
3543                 if (o->flags.binary)
3544                         emit_binary_diff(o, &mf1, &mf2);
3545                 else {
3546                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3547                                     diff_line_prefix(o), lbl[0], lbl[1]);
3548                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3549                                          sb.buf, sb.len, 0);
3550                         strbuf_release(&sb);
3551                 }
3552                 o->found_changes = 1;
3553         } else {
3554                 /* Crazy xdl interfaces.. */
3555                 const char *diffopts;
3556                 const char *v;
3557                 xpparam_t xpp;
3558                 xdemitconf_t xecfg;
3559                 struct emit_callback ecbdata;
3560                 const struct userdiff_funcname *pe;
3561
3562                 if (must_show_header) {
3563                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3564                                          header.buf, header.len, 0);
3565                         strbuf_reset(&header);
3566                 }
3567
3568                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3569                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3570
3571                 pe = diff_funcname_pattern(o, one);
3572                 if (!pe)
3573                         pe = diff_funcname_pattern(o, two);
3574
3575                 memset(&xpp, 0, sizeof(xpp));
3576                 memset(&xecfg, 0, sizeof(xecfg));
3577                 memset(&ecbdata, 0, sizeof(ecbdata));
3578                 if (o->flags.suppress_diff_headers)
3579                         lbl[0] = NULL;
3580                 ecbdata.label_path = lbl;
3581                 ecbdata.color_diff = want_color(o->use_color);
3582                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3583                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3584                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3585                 ecbdata.opt = o;
3586                 if (header.len && !o->flags.suppress_diff_headers)
3587                         ecbdata.header = &header;
3588                 xpp.flags = o->xdl_opts;
3589                 xpp.anchors = o->anchors;
3590                 xpp.anchors_nr = o->anchors_nr;
3591                 xecfg.ctxlen = o->context;
3592                 xecfg.interhunkctxlen = o->interhunkcontext;
3593                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3594                 if (o->flags.funccontext)
3595                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3596                 if (pe)
3597                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3598
3599                 diffopts = getenv("GIT_DIFF_OPTS");
3600                 if (!diffopts)
3601                         ;
3602                 else if (skip_prefix(diffopts, "--unified=", &v))
3603                         xecfg.ctxlen = strtoul(v, NULL, 10);
3604                 else if (skip_prefix(diffopts, "-u", &v))
3605                         xecfg.ctxlen = strtoul(v, NULL, 10);
3606
3607                 if (o->word_diff)
3608                         init_diff_words_data(&ecbdata, o, one, two);
3609                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3610                                   &ecbdata, &xpp, &xecfg))
3611                         die("unable to generate diff for %s", one->path);
3612                 if (o->word_diff)
3613                         free_diff_words_data(&ecbdata);
3614                 if (textconv_one)
3615                         free(mf1.ptr);
3616                 if (textconv_two)
3617                         free(mf2.ptr);
3618                 xdiff_clear_find_func(&xecfg);
3619         }
3620
3621  free_ab_and_return:
3622         strbuf_release(&header);
3623         diff_free_filespec_data(one);
3624         diff_free_filespec_data(two);
3625         free(a_one);
3626         free(b_two);
3627         return;
3628 }
3629
3630 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3631 {
3632         if (!is_renamed) {
3633                 if (p->status == DIFF_STATUS_ADDED) {
3634                         if (S_ISLNK(p->two->mode))
3635                                 return "new +l";
3636                         else if ((p->two->mode & 0777) == 0755)
3637                                 return "new +x";
3638                         else
3639                                 return "new";
3640                 } else if (p->status == DIFF_STATUS_DELETED)
3641                         return "gone";
3642         }
3643         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3644                 return "mode -l";
3645         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3646                 return "mode +l";
3647         else if ((p->one->mode & 0777) == 0644 &&
3648                  (p->two->mode & 0777) == 0755)
3649                 return "mode +x";
3650         else if ((p->one->mode & 0777) == 0755 &&
3651                  (p->two->mode & 0777) == 0644)
3652                 return "mode -x";
3653         return NULL;
3654 }
3655
3656 static void builtin_diffstat(const char *name_a, const char *name_b,
3657                              struct diff_filespec *one,
3658                              struct diff_filespec *two,
3659                              struct diffstat_t *diffstat,
3660                              struct diff_options *o,
3661                              struct diff_filepair *p)
3662 {
3663         mmfile_t mf1, mf2;
3664         struct diffstat_file *data;
3665         int same_contents;
3666         int complete_rewrite = 0;
3667
3668         if (!DIFF_PAIR_UNMERGED(p)) {
3669                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3670                         complete_rewrite = 1;
3671         }
3672
3673         data = diffstat_add(diffstat, name_a, name_b);
3674         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3675         if (o->flags.stat_with_summary)
3676                 data->comments = get_compact_summary(p, data->is_renamed);
3677
3678         if (!one || !two) {
3679                 data->is_unmerged = 1;
3680                 return;
3681         }
3682
3683         same_contents = oideq(&one->oid, &two->oid);
3684
3685         if (diff_filespec_is_binary(o->repo, one) ||
3686             diff_filespec_is_binary(o->repo, two)) {
3687                 data->is_binary = 1;
3688                 if (same_contents) {
3689                         data->added = 0;
3690                         data->deleted = 0;
3691                 } else {
3692                         data->added = diff_filespec_size(o->repo, two);
3693                         data->deleted = diff_filespec_size(o->repo, one);
3694                 }
3695         }
3696
3697         else if (complete_rewrite) {
3698                 diff_populate_filespec(o->repo, one, 0);
3699                 diff_populate_filespec(o->repo, two, 0);
3700                 data->deleted = count_lines(one->data, one->size);
3701                 data->added = count_lines(two->data, two->size);
3702         }
3703
3704         else if (!same_contents) {
3705                 /* Crazy xdl interfaces.. */
3706                 xpparam_t xpp;
3707                 xdemitconf_t xecfg;
3708
3709                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3710                     fill_mmfile(o->repo, &mf2, two) < 0)
3711                         die("unable to read files to diff");
3712
3713                 memset(&xpp, 0, sizeof(xpp));
3714                 memset(&xecfg, 0, sizeof(xecfg));
3715                 xpp.flags = o->xdl_opts;
3716                 xpp.anchors = o->anchors;
3717                 xpp.anchors_nr = o->anchors_nr;
3718                 xecfg.ctxlen = o->context;
3719                 xecfg.interhunkctxlen = o->interhunkcontext;
3720                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3721                                   diffstat_consume, diffstat, &xpp, &xecfg))
3722                         die("unable to generate diffstat for %s", one->path);
3723         }
3724
3725         diff_free_filespec_data(one);
3726         diff_free_filespec_data(two);
3727 }
3728
3729 static void builtin_checkdiff(const char *name_a, const char *name_b,
3730                               const char *attr_path,
3731                               struct diff_filespec *one,
3732                               struct diff_filespec *two,
3733                               struct diff_options *o)
3734 {
3735         mmfile_t mf1, mf2;
3736         struct checkdiff_t data;
3737
3738         if (!two)
3739                 return;
3740
3741         memset(&data, 0, sizeof(data));
3742         data.filename = name_b ? name_b : name_a;
3743         data.lineno = 0;
3744         data.o = o;
3745         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3746         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3747
3748         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3749             fill_mmfile(o->repo, &mf2, two) < 0)
3750                 die("unable to read files to diff");
3751
3752         /*
3753          * All the other codepaths check both sides, but not checking
3754          * the "old" side here is deliberate.  We are checking the newly
3755          * introduced changes, and as long as the "new" side is text, we
3756          * can and should check what it introduces.
3757          */
3758         if (diff_filespec_is_binary(o->repo, two))
3759                 goto free_and_return;
3760         else {
3761                 /* Crazy xdl interfaces.. */
3762                 xpparam_t xpp;
3763                 xdemitconf_t xecfg;
3764
3765                 memset(&xpp, 0, sizeof(xpp));
3766                 memset(&xecfg, 0, sizeof(xecfg));
3767                 xecfg.ctxlen = 1; /* at least one context line */
3768                 xpp.flags = 0;
3769                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3770                                   checkdiff_consume, &data,
3771                                   &xpp, &xecfg))
3772                         die("unable to generate checkdiff for %s", one->path);
3773
3774                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3775                         struct emit_callback ecbdata;
3776                         int blank_at_eof;
3777
3778                         ecbdata.ws_rule = data.ws_rule;
3779                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3780                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3781
3782                         if (blank_at_eof) {
3783                                 static char *err;
3784                                 if (!err)
3785                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3786                                 fprintf(o->file, "%s:%d: %s.\n",
3787                                         data.filename, blank_at_eof, err);
3788                                 data.status = 1; /* report errors */
3789                         }
3790                 }
3791         }
3792  free_and_return:
3793         diff_free_filespec_data(one);
3794         diff_free_filespec_data(two);
3795         if (data.status)
3796                 o->flags.check_failed = 1;
3797 }
3798
3799 struct diff_filespec *alloc_filespec(const char *path)
3800 {
3801         struct diff_filespec *spec;
3802
3803         FLEXPTR_ALLOC_STR(spec, path, path);
3804         spec->count = 1;
3805         spec->is_binary = -1;
3806         return spec;
3807 }
3808
3809 void free_filespec(struct diff_filespec *spec)
3810 {
3811         if (!--spec->count) {
3812                 diff_free_filespec_data(spec);
3813                 free(spec);
3814         }
3815 }
3816
3817 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3818                    int oid_valid, unsigned short mode)
3819 {
3820         if (mode) {
3821                 spec->mode = canon_mode(mode);
3822                 oidcpy(&spec->oid, oid);
3823                 spec->oid_valid = oid_valid;
3824         }
3825 }
3826
3827 /*
3828  * Given a name and sha1 pair, if the index tells us the file in
3829  * the work tree has that object contents, return true, so that
3830  * prepare_temp_file() does not have to inflate and extract.
3831  */
3832 static int reuse_worktree_file(struct index_state *istate,
3833                                const char *name,
3834                                const struct object_id *oid,
3835                                int want_file)
3836 {
3837         const struct cache_entry *ce;
3838         struct stat st;
3839         int pos, len;
3840
3841         /*
3842          * We do not read the cache ourselves here, because the
3843          * benchmark with my previous version that always reads cache
3844          * shows that it makes things worse for diff-tree comparing
3845          * two linux-2.6 kernel trees in an already checked out work
3846          * tree.  This is because most diff-tree comparisons deal with
3847          * only a small number of files, while reading the cache is
3848          * expensive for a large project, and its cost outweighs the
3849          * savings we get by not inflating the object to a temporary
3850          * file.  Practically, this code only helps when we are used
3851          * by diff-cache --cached, which does read the cache before
3852          * calling us.
3853          */
3854         if (!istate->cache)
3855                 return 0;
3856
3857         /* We want to avoid the working directory if our caller
3858          * doesn't need the data in a normal file, this system
3859          * is rather slow with its stat/open/mmap/close syscalls,
3860          * and the object is contained in a pack file.  The pack
3861          * is probably already open and will be faster to obtain
3862          * the data through than the working directory.  Loose
3863          * objects however would tend to be slower as they need
3864          * to be individually opened and inflated.
3865          */
3866         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3867                 return 0;
3868
3869         /*
3870          * Similarly, if we'd have to convert the file contents anyway, that
3871          * makes the optimization not worthwhile.
3872          */
3873         if (!want_file && would_convert_to_git(istate, name))
3874                 return 0;
3875
3876         len = strlen(name);
3877         pos = index_name_pos(istate, name, len);
3878         if (pos < 0)
3879                 return 0;
3880         ce = istate->cache[pos];
3881
3882         /*
3883          * This is not the sha1 we are looking for, or
3884          * unreusable because it is not a regular file.
3885          */
3886         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3887                 return 0;
3888
3889         /*
3890          * If ce is marked as "assume unchanged", there is no
3891          * guarantee that work tree matches what we are looking for.
3892          */
3893         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3894                 return 0;
3895
3896         /*
3897          * If ce matches the file in the work tree, we can reuse it.
3898          */
3899         if (ce_uptodate(ce) ||
3900             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3901                 return 1;
3902
3903         return 0;
3904 }
3905
3906 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3907 {
3908         struct strbuf buf = STRBUF_INIT;
3909         char *dirty = "";
3910
3911         /* Are we looking at the work tree? */
3912         if (s->dirty_submodule)
3913                 dirty = "-dirty";
3914
3915         strbuf_addf(&buf, "Subproject commit %s%s\n",
3916                     oid_to_hex(&s->oid), dirty);
3917         s->size = buf.len;
3918         if (size_only) {
3919                 s->data = NULL;
3920                 strbuf_release(&buf);
3921         } else {
3922                 s->data = strbuf_detach(&buf, NULL);
3923                 s->should_free = 1;
3924         }
3925         return 0;
3926 }
3927
3928 /*
3929  * While doing rename detection and pickaxe operation, we may need to
3930  * grab the data for the blob (or file) for our own in-core comparison.
3931  * diff_filespec has data and size fields for this purpose.
3932  */
3933 int diff_populate_filespec(struct repository *r,
3934                            struct diff_filespec *s,
3935                            unsigned int flags)
3936 {
3937         int size_only = flags & CHECK_SIZE_ONLY;
3938         int err = 0;
3939         int conv_flags = global_conv_flags_eol;
3940         /*
3941          * demote FAIL to WARN to allow inspecting the situation
3942          * instead of refusing.
3943          */
3944         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3945                 conv_flags = CONV_EOL_RNDTRP_WARN;
3946
3947         if (!DIFF_FILE_VALID(s))
3948                 die("internal error: asking to populate invalid file.");
3949         if (S_ISDIR(s->mode))
3950                 return -1;
3951
3952         if (s->data)
3953                 return 0;
3954
3955         if (size_only && 0 < s->size)
3956                 return 0;
3957
3958         if (S_ISGITLINK(s->mode))
3959                 return diff_populate_gitlink(s, size_only);
3960
3961         if (!s->oid_valid ||
3962             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3963                 struct strbuf buf = STRBUF_INIT;
3964                 struct stat st;
3965                 int fd;
3966
3967                 if (lstat(s->path, &st) < 0) {
3968                 err_empty:
3969                         err = -1;
3970                 empty:
3971                         s->data = (char *)"";
3972                         s->size = 0;
3973                         return err;
3974                 }
3975                 s->size = xsize_t(st.st_size);
3976                 if (!s->size)
3977                         goto empty;
3978                 if (S_ISLNK(st.st_mode)) {
3979                         struct strbuf sb = STRBUF_INIT;
3980
3981                         if (strbuf_readlink(&sb, s->path, s->size))
3982                                 goto err_empty;
3983                         s->size = sb.len;
3984                         s->data = strbuf_detach(&sb, NULL);
3985                         s->should_free = 1;
3986                         return 0;
3987                 }
3988
3989                 /*
3990                  * Even if the caller would be happy with getting
3991                  * only the size, we cannot return early at this
3992                  * point if the path requires us to run the content
3993                  * conversion.
3994                  */
3995                 if (size_only && !would_convert_to_git(r->index, s->path))
3996                         return 0;
3997
3998                 /*
3999                  * Note: this check uses xsize_t(st.st_size) that may
4000                  * not be the true size of the blob after it goes
4001                  * through convert_to_git().  This may not strictly be
4002                  * correct, but the whole point of big_file_threshold
4003                  * and is_binary check being that we want to avoid
4004                  * opening the file and inspecting the contents, this
4005                  * is probably fine.
4006                  */
4007                 if ((flags & CHECK_BINARY) &&
4008                     s->size > big_file_threshold && s->is_binary == -1) {
4009                         s->is_binary = 1;
4010                         return 0;
4011                 }
4012                 fd = open(s->path, O_RDONLY);
4013                 if (fd < 0)
4014                         goto err_empty;
4015                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4016                 close(fd);
4017                 s->should_munmap = 1;
4018
4019                 /*
4020                  * Convert from working tree format to canonical git format
4021                  */
4022                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4023                         size_t size = 0;
4024                         munmap(s->data, s->size);
4025                         s->should_munmap = 0;
4026                         s->data = strbuf_detach(&buf, &size);
4027                         s->size = size;
4028                         s->should_free = 1;
4029                 }
4030         }
4031         else {
4032                 enum object_type type;
4033                 if (size_only || (flags & CHECK_BINARY)) {
4034                         type = oid_object_info(r, &s->oid, &s->size);
4035                         if (type < 0)
4036                                 die("unable to read %s",
4037                                     oid_to_hex(&s->oid));
4038                         if (size_only)
4039                                 return 0;
4040                         if (s->size > big_file_threshold && s->is_binary == -1) {
4041                                 s->is_binary = 1;
4042                                 return 0;
4043                         }
4044                 }
4045                 s->data = read_object_file(&s->oid, &type, &s->size);
4046                 if (!s->data)
4047                         die("unable to read %s", oid_to_hex(&s->oid));
4048                 s->should_free = 1;
4049         }
4050         return 0;
4051 }
4052
4053 void diff_free_filespec_blob(struct diff_filespec *s)
4054 {
4055         if (s->should_free)
4056                 free(s->data);
4057         else if (s->should_munmap)
4058                 munmap(s->data, s->size);
4059
4060         if (s->should_free || s->should_munmap) {
4061                 s->should_free = s->should_munmap = 0;
4062                 s->data = NULL;
4063         }
4064 }
4065
4066 void diff_free_filespec_data(struct diff_filespec *s)
4067 {
4068         diff_free_filespec_blob(s);
4069         FREE_AND_NULL(s->cnt_data);
4070 }
4071
4072 static void prep_temp_blob(struct index_state *istate,
4073                            const char *path, struct diff_tempfile *temp,
4074                            void *blob,
4075                            unsigned long size,
4076                            const struct object_id *oid,
4077                            int mode)
4078 {
4079         struct strbuf buf = STRBUF_INIT;
4080         struct strbuf tempfile = STRBUF_INIT;
4081         char *path_dup = xstrdup(path);
4082         const char *base = basename(path_dup);
4083
4084         /* Generate "XXXXXX_basename.ext" */
4085         strbuf_addstr(&tempfile, "XXXXXX_");
4086         strbuf_addstr(&tempfile, base);
4087
4088         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4089         if (!temp->tempfile)
4090                 die_errno("unable to create temp-file");
4091         if (convert_to_working_tree(istate, path,
4092                         (const char *)blob, (size_t)size, &buf)) {
4093                 blob = buf.buf;
4094                 size = buf.len;
4095         }
4096         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4097             close_tempfile_gently(temp->tempfile))
4098                 die_errno("unable to write temp-file");
4099         temp->name = get_tempfile_path(temp->tempfile);
4100         oid_to_hex_r(temp->hex, oid);
4101         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4102         strbuf_release(&buf);
4103         strbuf_release(&tempfile);
4104         free(path_dup);
4105 }
4106
4107 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4108                                                const char *name,
4109                                                struct diff_filespec *one)
4110 {
4111         struct diff_tempfile *temp = claim_diff_tempfile();
4112
4113         if (!DIFF_FILE_VALID(one)) {
4114         not_a_valid_file:
4115                 /* A '-' entry produces this for file-2, and
4116                  * a '+' entry produces this for file-1.
4117                  */
4118                 temp->name = "/dev/null";
4119                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4120                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4121                 return temp;
4122         }
4123
4124         if (!S_ISGITLINK(one->mode) &&
4125             (!one->oid_valid ||
4126              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4127                 struct stat st;
4128                 if (lstat(name, &st) < 0) {
4129                         if (errno == ENOENT)
4130                                 goto not_a_valid_file;
4131                         die_errno("stat(%s)", name);
4132                 }
4133                 if (S_ISLNK(st.st_mode)) {
4134                         struct strbuf sb = STRBUF_INIT;
4135                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4136                                 die_errno("readlink(%s)", name);
4137                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4138                                        (one->oid_valid ?
4139                                         &one->oid : &null_oid),
4140                                        (one->oid_valid ?
4141                                         one->mode : S_IFLNK));
4142                         strbuf_release(&sb);
4143                 }
4144                 else {
4145                         /* we can borrow from the file in the work tree */
4146                         temp->name = name;
4147                         if (!one->oid_valid)
4148                                 oid_to_hex_r(temp->hex, &null_oid);
4149                         else
4150                                 oid_to_hex_r(temp->hex, &one->oid);
4151                         /* Even though we may sometimes borrow the
4152                          * contents from the work tree, we always want
4153                          * one->mode.  mode is trustworthy even when
4154                          * !(one->oid_valid), as long as
4155                          * DIFF_FILE_VALID(one).
4156                          */
4157                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4158                 }
4159                 return temp;
4160         }
4161         else {
4162                 if (diff_populate_filespec(r, one, 0))
4163                         die("cannot read data blob for %s", one->path);
4164                 prep_temp_blob(r->index, name, temp,
4165                                one->data, one->size,
4166                                &one->oid, one->mode);
4167         }
4168         return temp;
4169 }
4170
4171 static void add_external_diff_name(struct repository *r,
4172                                    struct argv_array *argv,
4173                                    const char *name,
4174                                    struct diff_filespec *df)
4175 {
4176         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4177         argv_array_push(argv, temp->name);
4178         argv_array_push(argv, temp->hex);
4179         argv_array_push(argv, temp->mode);
4180 }
4181
4182 /* An external diff command takes:
4183  *
4184  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4185  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4186  *
4187  */
4188 static void run_external_diff(const char *pgm,
4189                               const char *name,
4190                               const char *other,
4191                               struct diff_filespec *one,
4192                               struct diff_filespec *two,
4193                               const char *xfrm_msg,
4194                               int complete_rewrite,
4195                               struct diff_options *o)
4196 {
4197         struct argv_array argv = ARGV_ARRAY_INIT;
4198         struct argv_array env = ARGV_ARRAY_INIT;
4199         struct diff_queue_struct *q = &diff_queued_diff;
4200
4201         argv_array_push(&argv, pgm);
4202         argv_array_push(&argv, name);
4203
4204         if (one && two) {
4205                 add_external_diff_name(o->repo, &argv, name, one);
4206                 if (!other)
4207                         add_external_diff_name(o->repo, &argv, name, two);
4208                 else {
4209                         add_external_diff_name(o->repo, &argv, other, two);
4210                         argv_array_push(&argv, other);
4211                         argv_array_push(&argv, xfrm_msg);
4212                 }
4213         }
4214
4215         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4216         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4217
4218         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4219                 die(_("external diff died, stopping at %s"), name);
4220
4221         remove_tempfile();
4222         argv_array_clear(&argv);
4223         argv_array_clear(&env);
4224 }
4225
4226 static int similarity_index(struct diff_filepair *p)
4227 {
4228         return p->score * 100 / MAX_SCORE;
4229 }
4230
4231 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4232 {
4233         if (startup_info->have_repository)
4234                 return find_unique_abbrev(oid, abbrev);
4235         else {
4236                 char *hex = oid_to_hex(oid);
4237                 if (abbrev < 0)
4238                         abbrev = FALLBACK_DEFAULT_ABBREV;
4239                 if (abbrev > the_hash_algo->hexsz)
4240                         BUG("oid abbreviation out of range: %d", abbrev);
4241                 if (abbrev)
4242                         hex[abbrev] = '\0';
4243                 return hex;
4244         }
4245 }
4246
4247 static void fill_metainfo(struct strbuf *msg,
4248                           const char *name,
4249                           const char *other,
4250                           struct diff_filespec *one,
4251                           struct diff_filespec *two,
4252                           struct diff_options *o,
4253                           struct diff_filepair *p,
4254                           int *must_show_header,
4255                           int use_color)
4256 {
4257         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4258         const char *reset = diff_get_color(use_color, DIFF_RESET);
4259         const char *line_prefix = diff_line_prefix(o);
4260
4261         *must_show_header = 1;
4262         strbuf_init(msg, PATH_MAX * 2 + 300);
4263         switch (p->status) {
4264         case DIFF_STATUS_COPIED:
4265                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4266                             line_prefix, set, similarity_index(p));
4267                 strbuf_addf(msg, "%s\n%s%scopy from ",
4268                             reset,  line_prefix, set);
4269                 quote_c_style(name, msg, NULL, 0);
4270                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4271                 quote_c_style(other, msg, NULL, 0);
4272                 strbuf_addf(msg, "%s\n", reset);
4273                 break;
4274         case DIFF_STATUS_RENAMED:
4275                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4276                             line_prefix, set, similarity_index(p));
4277                 strbuf_addf(msg, "%s\n%s%srename from ",
4278                             reset, line_prefix, set);
4279                 quote_c_style(name, msg, NULL, 0);
4280                 strbuf_addf(msg, "%s\n%s%srename to ",
4281                             reset, line_prefix, set);
4282                 quote_c_style(other, msg, NULL, 0);
4283                 strbuf_addf(msg, "%s\n", reset);
4284                 break;
4285         case DIFF_STATUS_MODIFIED:
4286                 if (p->score) {
4287                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4288                                     line_prefix,
4289                                     set, similarity_index(p), reset);
4290                         break;
4291                 }
4292                 /* fallthru */
4293         default:
4294                 *must_show_header = 0;
4295         }
4296         if (one && two && !oideq(&one->oid, &two->oid)) {
4297                 const unsigned hexsz = the_hash_algo->hexsz;
4298                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4299
4300                 if (o->flags.binary) {
4301                         mmfile_t mf;
4302                         if ((!fill_mmfile(o->repo, &mf, one) &&
4303                              diff_filespec_is_binary(o->repo, one)) ||
4304                             (!fill_mmfile(o->repo, &mf, two) &&
4305                              diff_filespec_is_binary(o->repo, two)))
4306                                 abbrev = hexsz;
4307                 }
4308                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4309                             diff_abbrev_oid(&one->oid, abbrev),
4310                             diff_abbrev_oid(&two->oid, abbrev));
4311                 if (one->mode == two->mode)
4312                         strbuf_addf(msg, " %06o", one->mode);
4313                 strbuf_addf(msg, "%s\n", reset);
4314         }
4315 }
4316
4317 static void run_diff_cmd(const char *pgm,
4318                          const char *name,
4319                          const char *other,
4320                          const char *attr_path,
4321                          struct diff_filespec *one,
4322                          struct diff_filespec *two,
4323                          struct strbuf *msg,
4324                          struct diff_options *o,
4325                          struct diff_filepair *p)
4326 {
4327         const char *xfrm_msg = NULL;
4328         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4329         int must_show_header = 0;
4330
4331
4332         if (o->flags.allow_external) {
4333                 struct userdiff_driver *drv;
4334
4335                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4336                 if (drv && drv->external)
4337                         pgm = drv->external;
4338         }
4339
4340         if (msg) {
4341                 /*
4342                  * don't use colors when the header is intended for an
4343                  * external diff driver
4344                  */
4345                 fill_metainfo(msg, name, other, one, two, o, p,
4346                               &must_show_header,
4347                               want_color(o->use_color) && !pgm);
4348                 xfrm_msg = msg->len ? msg->buf : NULL;
4349         }
4350
4351         if (pgm) {
4352                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4353                                   complete_rewrite, o);
4354                 return;
4355         }
4356         if (one && two)
4357                 builtin_diff(name, other ? other : name,
4358                              one, two, xfrm_msg, must_show_header,
4359                              o, complete_rewrite);
4360         else
4361                 fprintf(o->file, "* Unmerged path %s\n", name);
4362 }
4363
4364 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4365 {
4366         if (DIFF_FILE_VALID(one)) {
4367                 if (!one->oid_valid) {
4368                         struct stat st;
4369                         if (one->is_stdin) {
4370                                 oidclr(&one->oid);
4371                                 return;
4372                         }
4373                         if (lstat(one->path, &st) < 0)
4374                                 die_errno("stat '%s'", one->path);
4375                         if (index_path(istate, &one->oid, one->path, &st, 0))
4376                                 die("cannot hash %s", one->path);
4377                 }
4378         }
4379         else
4380                 oidclr(&one->oid);
4381 }
4382
4383 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4384 {
4385         /* Strip the prefix but do not molest /dev/null and absolute paths */
4386         if (*namep && !is_absolute_path(*namep)) {
4387                 *namep += prefix_length;
4388                 if (**namep == '/')
4389                         ++*namep;
4390         }
4391         if (*otherp && !is_absolute_path(*otherp)) {
4392                 *otherp += prefix_length;
4393                 if (**otherp == '/')
4394                         ++*otherp;
4395         }
4396 }
4397
4398 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4399 {
4400         const char *pgm = external_diff();
4401         struct strbuf msg;
4402         struct diff_filespec *one = p->one;
4403         struct diff_filespec *two = p->two;
4404         const char *name;
4405         const char *other;
4406         const char *attr_path;
4407
4408         name  = one->path;
4409         other = (strcmp(name, two->path) ? two->path : NULL);
4410         attr_path = name;
4411         if (o->prefix_length)
4412                 strip_prefix(o->prefix_length, &name, &other);
4413
4414         if (!o->flags.allow_external)
4415                 pgm = NULL;
4416
4417         if (DIFF_PAIR_UNMERGED(p)) {
4418                 run_diff_cmd(pgm, name, NULL, attr_path,
4419                              NULL, NULL, NULL, o, p);
4420                 return;
4421         }
4422
4423         diff_fill_oid_info(one, o->repo->index);
4424         diff_fill_oid_info(two, o->repo->index);
4425
4426         if (!pgm &&
4427             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4428             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4429                 /*
4430                  * a filepair that changes between file and symlink
4431                  * needs to be split into deletion and creation.
4432                  */
4433                 struct diff_filespec *null = alloc_filespec(two->path);
4434                 run_diff_cmd(NULL, name, other, attr_path,
4435                              one, null, &msg,
4436                              o, p);
4437                 free(null);
4438                 strbuf_release(&msg);
4439
4440                 null = alloc_filespec(one->path);
4441                 run_diff_cmd(NULL, name, other, attr_path,
4442                              null, two, &msg, o, p);
4443                 free(null);
4444         }
4445         else
4446                 run_diff_cmd(pgm, name, other, attr_path,
4447                              one, two, &msg, o, p);
4448
4449         strbuf_release(&msg);
4450 }
4451
4452 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4453                          struct diffstat_t *diffstat)
4454 {
4455         const char *name;
4456         const char *other;
4457
4458         if (DIFF_PAIR_UNMERGED(p)) {
4459                 /* unmerged */
4460                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4461                                  diffstat, o, p);
4462                 return;
4463         }
4464
4465         name = p->one->path;
4466         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4467
4468         if (o->prefix_length)
4469                 strip_prefix(o->prefix_length, &name, &other);
4470
4471         diff_fill_oid_info(p->one, o->repo->index);
4472         diff_fill_oid_info(p->two, o->repo->index);
4473
4474         builtin_diffstat(name, other, p->one, p->two,
4475                          diffstat, o, p);
4476 }
4477
4478 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4479 {
4480         const char *name;
4481         const char *other;
4482         const char *attr_path;
4483
4484         if (DIFF_PAIR_UNMERGED(p)) {
4485                 /* unmerged */
4486                 return;
4487         }
4488
4489         name = p->one->path;
4490         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4491         attr_path = other ? other : name;
4492
4493         if (o->prefix_length)
4494                 strip_prefix(o->prefix_length, &name, &other);
4495
4496         diff_fill_oid_info(p->one, o->repo->index);
4497         diff_fill_oid_info(p->two, o->repo->index);
4498
4499         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4500 }
4501
4502 void repo_diff_setup(struct repository *r, struct diff_options *options)
4503 {
4504         memcpy(options, &default_diff_options, sizeof(*options));
4505
4506         options->file = stdout;
4507         options->repo = r;
4508
4509         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4510         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4511         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4512         options->abbrev = DEFAULT_ABBREV;
4513         options->line_termination = '\n';
4514         options->break_opt = -1;
4515         options->rename_limit = -1;
4516         options->dirstat_permille = diff_dirstat_permille_default;
4517         options->context = diff_context_default;
4518         options->interhunkcontext = diff_interhunk_context_default;
4519         options->ws_error_highlight = ws_error_highlight_default;
4520         options->flags.rename_empty = 1;
4521         options->objfind = NULL;
4522
4523         /* pathchange left =NULL by default */
4524         options->change = diff_change;
4525         options->add_remove = diff_addremove;
4526         options->use_color = diff_use_color_default;
4527         options->detect_rename = diff_detect_rename_default;
4528         options->xdl_opts |= diff_algorithm;
4529         if (diff_indent_heuristic)
4530                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4531
4532         options->orderfile = diff_order_file_cfg;
4533
4534         if (diff_no_prefix) {
4535                 options->a_prefix = options->b_prefix = "";
4536         } else if (!diff_mnemonic_prefix) {
4537                 options->a_prefix = "a/";
4538                 options->b_prefix = "b/";
4539         }
4540
4541         options->color_moved = diff_color_moved_default;
4542         options->color_moved_ws_handling = diff_color_moved_ws_default;
4543 }
4544
4545 void diff_setup_done(struct diff_options *options)
4546 {
4547         unsigned check_mask = DIFF_FORMAT_NAME |
4548                               DIFF_FORMAT_NAME_STATUS |
4549                               DIFF_FORMAT_CHECKDIFF |
4550                               DIFF_FORMAT_NO_OUTPUT;
4551         /*
4552          * This must be signed because we're comparing against a potentially
4553          * negative value.
4554          */
4555         const int hexsz = the_hash_algo->hexsz;
4556
4557         if (options->set_default)
4558                 options->set_default(options);
4559
4560         if (HAS_MULTI_BITS(options->output_format & check_mask))
4561                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4562
4563         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4564                 die(_("-G, -S and --find-object are mutually exclusive"));
4565
4566         /*
4567          * Most of the time we can say "there are changes"
4568          * only by checking if there are changed paths, but
4569          * --ignore-whitespace* options force us to look
4570          * inside contents.
4571          */
4572
4573         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4574                 options->flags.diff_from_contents = 1;
4575         else
4576                 options->flags.diff_from_contents = 0;
4577
4578         if (options->flags.find_copies_harder)
4579                 options->detect_rename = DIFF_DETECT_COPY;
4580
4581         if (!options->flags.relative_name)
4582                 options->prefix = NULL;
4583         if (options->prefix)
4584                 options->prefix_length = strlen(options->prefix);
4585         else
4586                 options->prefix_length = 0;
4587
4588         if (options->output_format & (DIFF_FORMAT_NAME |
4589                                       DIFF_FORMAT_NAME_STATUS |
4590                                       DIFF_FORMAT_CHECKDIFF |
4591                                       DIFF_FORMAT_NO_OUTPUT))
4592                 options->output_format &= ~(DIFF_FORMAT_RAW |
4593                                             DIFF_FORMAT_NUMSTAT |
4594                                             DIFF_FORMAT_DIFFSTAT |
4595                                             DIFF_FORMAT_SHORTSTAT |
4596                                             DIFF_FORMAT_DIRSTAT |
4597                                             DIFF_FORMAT_SUMMARY |
4598                                             DIFF_FORMAT_PATCH);
4599
4600         /*
4601          * These cases always need recursive; we do not drop caller-supplied
4602          * recursive bits for other formats here.
4603          */
4604         if (options->output_format & (DIFF_FORMAT_PATCH |
4605                                       DIFF_FORMAT_NUMSTAT |
4606                                       DIFF_FORMAT_DIFFSTAT |
4607                                       DIFF_FORMAT_SHORTSTAT |
4608                                       DIFF_FORMAT_DIRSTAT |
4609                                       DIFF_FORMAT_SUMMARY |
4610                                       DIFF_FORMAT_CHECKDIFF))
4611                 options->flags.recursive = 1;
4612         /*
4613          * Also pickaxe would not work very well if you do not say recursive
4614          */
4615         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4616                 options->flags.recursive = 1;
4617         /*
4618          * When patches are generated, submodules diffed against the work tree
4619          * must be checked for dirtiness too so it can be shown in the output
4620          */
4621         if (options->output_format & DIFF_FORMAT_PATCH)
4622                 options->flags.dirty_submodules = 1;
4623
4624         if (options->detect_rename && options->rename_limit < 0)
4625                 options->rename_limit = diff_rename_limit_default;
4626         if (hexsz < options->abbrev)
4627                 options->abbrev = hexsz; /* full */
4628
4629         /*
4630          * It does not make sense to show the first hit we happened
4631          * to have found.  It does not make sense not to return with
4632          * exit code in such a case either.
4633          */
4634         if (options->flags.quick) {
4635                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4636                 options->flags.exit_with_status = 1;
4637         }
4638
4639         options->diff_path_counter = 0;
4640
4641         if (options->flags.follow_renames && options->pathspec.nr != 1)
4642                 die(_("--follow requires exactly one pathspec"));
4643
4644         if (!options->use_color || external_diff())
4645                 options->color_moved = 0;
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 int diff_opt_parse(struct diff_options *options,
4938                    const char **av, int ac, const char *prefix)
4939 {
4940         const char *arg = av[0];
4941         const char *optarg;
4942         int argcount;
4943
4944         if (!prefix)
4945                 prefix = "";
4946
4947         /* Output format options */
4948         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4949             || opt_arg(arg, 'U', "unified", &options->context))
4950                 enable_patch_output(&options->output_format);
4951         else if (!strcmp(arg, "--raw"))
4952                 options->output_format |= DIFF_FORMAT_RAW;
4953         else if (!strcmp(arg, "--patch-with-raw")) {
4954                 enable_patch_output(&options->output_format);
4955                 options->output_format |= DIFF_FORMAT_RAW;
4956         } else if (!strcmp(arg, "--numstat"))
4957                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4958         else if (!strcmp(arg, "--shortstat"))
4959                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4960         else if (skip_prefix(arg, "-X", &arg) ||
4961                  skip_to_optional_arg(arg, "--dirstat", &arg))
4962                 return parse_dirstat_opt(options, arg);
4963         else if (!strcmp(arg, "--cumulative"))
4964                 return parse_dirstat_opt(options, "cumulative");
4965         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4966                 parse_dirstat_opt(options, "files");
4967                 return parse_dirstat_opt(options, arg);
4968         }
4969         else if (!strcmp(arg, "--check"))
4970                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4971         else if (!strcmp(arg, "--summary"))
4972                 options->output_format |= DIFF_FORMAT_SUMMARY;
4973         else if (!strcmp(arg, "--patch-with-stat")) {
4974                 enable_patch_output(&options->output_format);
4975                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4976         } else if (!strcmp(arg, "--name-only"))
4977                 options->output_format |= DIFF_FORMAT_NAME;
4978         else if (!strcmp(arg, "--name-status"))
4979                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4980         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4981                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4982         else if (starts_with(arg, "--stat"))
4983                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4984                 return stat_opt(options, av);
4985         else if (!strcmp(arg, "--compact-summary")) {
4986                  options->flags.stat_with_summary = 1;
4987                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4988         } else if (!strcmp(arg, "--no-compact-summary"))
4989                  options->flags.stat_with_summary = 0;
4990         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4991                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4992         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4993                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4994         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4995                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4996
4997         /* renames options */
4998         else if (starts_with(arg, "-B") ||
4999                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
5000                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
5001                         return error("invalid argument to -B: %s", arg+2);
5002         }
5003         else if (starts_with(arg, "-M") ||
5004                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
5005                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5006                         return error("invalid argument to -M: %s", arg+2);
5007                 options->detect_rename = DIFF_DETECT_RENAME;
5008         }
5009         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
5010                 options->irreversible_delete = 1;
5011         }
5012         else if (starts_with(arg, "-C") ||
5013                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
5014                 if (options->detect_rename == DIFF_DETECT_COPY)
5015                         options->flags.find_copies_harder = 1;
5016                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5017                         return error("invalid argument to -C: %s", arg+2);
5018                 options->detect_rename = DIFF_DETECT_COPY;
5019         }
5020         else if (!strcmp(arg, "--no-renames"))
5021                 options->detect_rename = 0;
5022         else if (!strcmp(arg, "--rename-empty"))
5023                 options->flags.rename_empty = 1;
5024         else if (!strcmp(arg, "--no-rename-empty"))
5025                 options->flags.rename_empty = 0;
5026         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5027                 options->flags.relative_name = 1;
5028                 if (arg)
5029                         options->prefix = arg;
5030         }
5031
5032         /* xdiff options */
5033         else if (!strcmp(arg, "--minimal"))
5034                 DIFF_XDL_SET(options, NEED_MINIMAL);
5035         else if (!strcmp(arg, "--no-minimal"))
5036                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5037         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5038                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5039         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5040                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5041         else if (!strcmp(arg, "--ignore-space-at-eol"))
5042                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5043         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5044                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5045         else if (!strcmp(arg, "--ignore-blank-lines"))
5046                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5047         else if (!strcmp(arg, "--indent-heuristic"))
5048                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5049         else if (!strcmp(arg, "--no-indent-heuristic"))
5050                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5051         else if (!strcmp(arg, "--patience")) {
5052                 int i;
5053                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5054                 /*
5055                  * Both --patience and --anchored use PATIENCE_DIFF
5056                  * internally, so remove any anchors previously
5057                  * specified.
5058                  */
5059                 for (i = 0; i < options->anchors_nr; i++)
5060                         free(options->anchors[i]);
5061                 options->anchors_nr = 0;
5062         } else if (!strcmp(arg, "--histogram"))
5063                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5064         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5065                 long value = parse_algorithm_value(optarg);
5066                 if (value < 0)
5067                         return error("option diff-algorithm accepts \"myers\", "
5068                                      "\"minimal\", \"patience\" and \"histogram\"");
5069                 /* clear out previous settings */
5070                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5071                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5072                 options->xdl_opts |= value;
5073                 return argcount;
5074         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5075                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5076                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5077                            options->anchors_alloc);
5078                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5079         }
5080
5081         /* flags options */
5082         else if (!strcmp(arg, "--binary")) {
5083                 enable_patch_output(&options->output_format);
5084                 options->flags.binary = 1;
5085         }
5086         else if (!strcmp(arg, "--full-index"))
5087                 options->flags.full_index = 1;
5088         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5089                 options->flags.text = 1;
5090         else if (!strcmp(arg, "-R"))
5091                 options->flags.reverse_diff = 1;
5092         else if (!strcmp(arg, "--find-copies-harder"))
5093                 options->flags.find_copies_harder = 1;
5094         else if (!strcmp(arg, "--follow"))
5095                 options->flags.follow_renames = 1;
5096         else if (!strcmp(arg, "--no-follow")) {
5097                 options->flags.follow_renames = 0;
5098                 options->flags.default_follow_renames = 0;
5099         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5100                 int value = git_config_colorbool(NULL, arg);
5101                 if (value < 0)
5102                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5103                 options->use_color = value;
5104         }
5105         else if (!strcmp(arg, "--no-color"))
5106                 options->use_color = 0;
5107         else if (!strcmp(arg, "--color-moved")) {
5108                 if (diff_color_moved_default)
5109                         options->color_moved = diff_color_moved_default;
5110                 if (options->color_moved == COLOR_MOVED_NO)
5111                         options->color_moved = COLOR_MOVED_DEFAULT;
5112         } else if (!strcmp(arg, "--no-color-moved"))
5113                 options->color_moved = COLOR_MOVED_NO;
5114         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5115                 int cm = parse_color_moved(arg);
5116                 if (cm < 0)
5117                         return error("bad --color-moved argument: %s", arg);
5118                 options->color_moved = cm;
5119         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5120                 options->color_moved_ws_handling = 0;
5121         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5122                 unsigned cm = parse_color_moved_ws(arg);
5123                 if (cm & COLOR_MOVED_WS_ERROR)
5124                         return -1;
5125                 options->color_moved_ws_handling = cm;
5126         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5127                 options->use_color = 1;
5128                 options->word_diff = DIFF_WORDS_COLOR;
5129         }
5130         else if (!strcmp(arg, "--word-diff")) {
5131                 if (options->word_diff == DIFF_WORDS_NONE)
5132                         options->word_diff = DIFF_WORDS_PLAIN;
5133         }
5134         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5135                 if (!strcmp(arg, "plain"))
5136                         options->word_diff = DIFF_WORDS_PLAIN;
5137                 else if (!strcmp(arg, "color")) {
5138                         options->use_color = 1;
5139                         options->word_diff = DIFF_WORDS_COLOR;
5140                 }
5141                 else if (!strcmp(arg, "porcelain"))
5142                         options->word_diff = DIFF_WORDS_PORCELAIN;
5143                 else if (!strcmp(arg, "none"))
5144                         options->word_diff = DIFF_WORDS_NONE;
5145                 else
5146                         die("bad --word-diff argument: %s", arg);
5147         }
5148         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5149                 if (options->word_diff == DIFF_WORDS_NONE)
5150                         options->word_diff = DIFF_WORDS_PLAIN;
5151                 options->word_regex = optarg;
5152                 return argcount;
5153         }
5154         else if (!strcmp(arg, "--exit-code"))
5155                 options->flags.exit_with_status = 1;
5156         else if (!strcmp(arg, "--quiet"))
5157                 options->flags.quick = 1;
5158         else if (!strcmp(arg, "--ext-diff"))
5159                 options->flags.allow_external = 1;
5160         else if (!strcmp(arg, "--no-ext-diff"))
5161                 options->flags.allow_external = 0;
5162         else if (!strcmp(arg, "--textconv")) {
5163                 options->flags.allow_textconv = 1;
5164                 options->flags.textconv_set_via_cmdline = 1;
5165         } else if (!strcmp(arg, "--no-textconv"))
5166                 options->flags.allow_textconv = 0;
5167         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5168                 options->flags.override_submodule_config = 1;
5169                 handle_ignore_submodules_arg(options, arg);
5170         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5171                 return parse_submodule_opt(options, arg);
5172         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5173                 return parse_ws_error_highlight_opt(options, arg);
5174         else if (!strcmp(arg, "--ita-invisible-in-index"))
5175                 options->ita_invisible_in_index = 1;
5176         else if (!strcmp(arg, "--ita-visible-in-index"))
5177                 options->ita_invisible_in_index = 0;
5178
5179         /* misc options */
5180         else if (!strcmp(arg, "-z"))
5181                 options->line_termination = 0;
5182         else if ((argcount = short_opt('l', av, &optarg))) {
5183                 options->rename_limit = strtoul(optarg, NULL, 10);
5184                 return argcount;
5185         }
5186         else if ((argcount = short_opt('S', av, &optarg))) {
5187                 options->pickaxe = optarg;
5188                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5189                 return argcount;
5190         } else if ((argcount = short_opt('G', av, &optarg))) {
5191                 options->pickaxe = optarg;
5192                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5193                 return argcount;
5194         }
5195         else if (!strcmp(arg, "--pickaxe-all"))
5196                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5197         else if (!strcmp(arg, "--pickaxe-regex"))
5198                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5199         else if ((argcount = short_opt('O', av, &optarg))) {
5200                 options->orderfile = prefix_filename(prefix, optarg);
5201                 return argcount;
5202         } else if (skip_prefix(arg, "--find-object=", &arg))
5203                 return parse_objfind_opt(options, arg);
5204         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5205                 int offending = parse_diff_filter_opt(optarg, options);
5206                 if (offending)
5207                         die("unknown change class '%c' in --diff-filter=%s",
5208                             offending, optarg);
5209                 return argcount;
5210         }
5211         else if (!strcmp(arg, "--no-abbrev"))
5212                 options->abbrev = 0;
5213         else if (!strcmp(arg, "--abbrev"))
5214                 options->abbrev = DEFAULT_ABBREV;
5215         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5216                 options->abbrev = strtoul(arg, NULL, 10);
5217                 if (options->abbrev < MINIMUM_ABBREV)
5218                         options->abbrev = MINIMUM_ABBREV;
5219                 else if (the_hash_algo->hexsz < options->abbrev)
5220                         options->abbrev = the_hash_algo->hexsz;
5221         }
5222         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5223                 options->a_prefix = optarg;
5224                 return argcount;
5225         }
5226         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5227                 options->line_prefix = optarg;
5228                 options->line_prefix_length = strlen(options->line_prefix);
5229                 graph_setup_line_prefix(options);
5230                 return argcount;
5231         }
5232         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5233                 options->b_prefix = optarg;
5234                 return argcount;
5235         }
5236         else if (!strcmp(arg, "--no-prefix"))
5237                 options->a_prefix = options->b_prefix = "";
5238         else if (opt_arg(arg, '\0', "inter-hunk-context",
5239                          &options->interhunkcontext))
5240                 ;
5241         else if (!strcmp(arg, "-W"))
5242                 options->flags.funccontext = 1;
5243         else if (!strcmp(arg, "--function-context"))
5244                 options->flags.funccontext = 1;
5245         else if (!strcmp(arg, "--no-function-context"))
5246                 options->flags.funccontext = 0;
5247         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5248                 char *path = prefix_filename(prefix, optarg);
5249                 options->file = xfopen(path, "w");
5250                 options->close_file = 1;
5251                 if (options->use_color != GIT_COLOR_ALWAYS)
5252                         options->use_color = GIT_COLOR_NEVER;
5253                 free(path);
5254                 return argcount;
5255         } else
5256                 return 0;
5257         return 1;
5258 }
5259
5260 int parse_rename_score(const char **cp_p)
5261 {
5262         unsigned long num, scale;
5263         int ch, dot;
5264         const char *cp = *cp_p;
5265
5266         num = 0;
5267         scale = 1;
5268         dot = 0;
5269         for (;;) {
5270                 ch = *cp;
5271                 if ( !dot && ch == '.' ) {
5272                         scale = 1;
5273                         dot = 1;
5274                 } else if ( ch == '%' ) {
5275                         scale = dot ? scale*100 : 100;
5276                         cp++;   /* % is always at the end */
5277                         break;
5278                 } else if ( ch >= '0' && ch <= '9' ) {
5279                         if ( scale < 100000 ) {
5280                                 scale *= 10;
5281                                 num = (num*10) + (ch-'0');
5282                         }
5283                 } else {
5284                         break;
5285                 }
5286                 cp++;
5287         }
5288         *cp_p = cp;
5289
5290         /* user says num divided by scale and we say internally that
5291          * is MAX_SCORE * num / scale.
5292          */
5293         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5294 }
5295
5296 static int diff_scoreopt_parse(const char *opt)
5297 {
5298         int opt1, opt2, cmd;
5299
5300         if (*opt++ != '-')
5301                 return -1;
5302         cmd = *opt++;
5303         if (cmd == '-') {
5304                 /* convert the long-form arguments into short-form versions */
5305                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5306                         if (*opt == 0 || *opt++ == '=')
5307                                 cmd = 'B';
5308                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5309                         if (*opt == 0 || *opt++ == '=')
5310                                 cmd = 'C';
5311                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5312                         if (*opt == 0 || *opt++ == '=')
5313                                 cmd = 'M';
5314                 }
5315         }
5316         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5317                 return -1; /* that is not a -M, -C, or -B option */
5318
5319         opt1 = parse_rename_score(&opt);
5320         if (cmd != 'B')
5321                 opt2 = 0;
5322         else {
5323                 if (*opt == 0)
5324                         opt2 = 0;
5325                 else if (*opt != '/')
5326                         return -1; /* we expect -B80/99 or -B80 */
5327                 else {
5328                         opt++;
5329                         opt2 = parse_rename_score(&opt);
5330                 }
5331         }
5332         if (*opt != 0)
5333                 return -1;
5334         return opt1 | (opt2 << 16);
5335 }
5336
5337 struct diff_queue_struct diff_queued_diff;
5338
5339 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5340 {
5341         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5342         queue->queue[queue->nr++] = dp;
5343 }
5344
5345 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5346                                  struct diff_filespec *one,
5347                                  struct diff_filespec *two)
5348 {
5349         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5350         dp->one = one;
5351         dp->two = two;
5352         if (queue)
5353                 diff_q(queue, dp);
5354         return dp;
5355 }
5356
5357 void diff_free_filepair(struct diff_filepair *p)
5358 {
5359         free_filespec(p->one);
5360         free_filespec(p->two);
5361         free(p);
5362 }
5363
5364 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5365 {
5366         int abblen;
5367         const char *abbrev;
5368
5369         /* Do we want all 40 hex characters? */
5370         if (len == the_hash_algo->hexsz)
5371                 return oid_to_hex(oid);
5372
5373         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5374         abbrev = diff_abbrev_oid(oid, len);
5375
5376         if (!print_sha1_ellipsis())
5377                 return abbrev;
5378
5379         abblen = strlen(abbrev);
5380
5381         /*
5382          * In well-behaved cases, where the abbreviated result is the
5383          * same as the requested length, append three dots after the
5384          * abbreviation (hence the whole logic is limited to the case
5385          * where abblen < 37); when the actual abbreviated result is a
5386          * bit longer than the requested length, we reduce the number
5387          * of dots so that they match the well-behaved ones.  However,
5388          * if the actual abbreviation is longer than the requested
5389          * length by more than three, we give up on aligning, and add
5390          * three dots anyway, to indicate that the output is not the
5391          * full object name.  Yes, this may be suboptimal, but this
5392          * appears only in "diff --raw --abbrev" output and it is not
5393          * worth the effort to change it now.  Note that this would
5394          * likely to work fine when the automatic sizing of default
5395          * abbreviation length is used--we would be fed -1 in "len" in
5396          * that case, and will end up always appending three-dots, but
5397          * the automatic sizing is supposed to give abblen that ensures
5398          * uniqueness across all objects (statistically speaking).
5399          */
5400         if (abblen < the_hash_algo->hexsz - 3) {
5401                 static char hex[GIT_MAX_HEXSZ + 1];
5402                 if (len < abblen && abblen <= len + 2)
5403                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5404                 else
5405                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5406                 return hex;
5407         }
5408
5409         return oid_to_hex(oid);
5410 }
5411
5412 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5413 {
5414         int line_termination = opt->line_termination;
5415         int inter_name_termination = line_termination ? '\t' : '\0';
5416
5417         fprintf(opt->file, "%s", diff_line_prefix(opt));
5418         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5419                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5420                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5421                 fprintf(opt->file, "%s ",
5422                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5423         }
5424         if (p->score) {
5425                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5426                         inter_name_termination);
5427         } else {
5428                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5429         }
5430
5431         if (p->status == DIFF_STATUS_COPIED ||
5432             p->status == DIFF_STATUS_RENAMED) {
5433                 const char *name_a, *name_b;
5434                 name_a = p->one->path;
5435                 name_b = p->two->path;
5436                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5437                 write_name_quoted(name_a, opt->file, inter_name_termination);
5438                 write_name_quoted(name_b, opt->file, line_termination);
5439         } else {
5440                 const char *name_a, *name_b;
5441                 name_a = p->one->mode ? p->one->path : p->two->path;
5442                 name_b = NULL;
5443                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5444                 write_name_quoted(name_a, opt->file, line_termination);
5445         }
5446 }
5447
5448 int diff_unmodified_pair(struct diff_filepair *p)
5449 {
5450         /* This function is written stricter than necessary to support
5451          * the currently implemented transformers, but the idea is to
5452          * let transformers to produce diff_filepairs any way they want,
5453          * and filter and clean them up here before producing the output.
5454          */
5455         struct diff_filespec *one = p->one, *two = p->two;
5456
5457         if (DIFF_PAIR_UNMERGED(p))
5458                 return 0; /* unmerged is interesting */
5459
5460         /* deletion, addition, mode or type change
5461          * and rename are all interesting.
5462          */
5463         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5464             DIFF_PAIR_MODE_CHANGED(p) ||
5465             strcmp(one->path, two->path))
5466                 return 0;
5467
5468         /* both are valid and point at the same path.  that is, we are
5469          * dealing with a change.
5470          */
5471         if (one->oid_valid && two->oid_valid &&
5472             oideq(&one->oid, &two->oid) &&
5473             !one->dirty_submodule && !two->dirty_submodule)
5474                 return 1; /* no change */
5475         if (!one->oid_valid && !two->oid_valid)
5476                 return 1; /* both look at the same file on the filesystem. */
5477         return 0;
5478 }
5479
5480 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5481 {
5482         if (diff_unmodified_pair(p))
5483                 return;
5484
5485         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5486             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5487                 return; /* no tree diffs in patch format */
5488
5489         run_diff(p, o);
5490 }
5491
5492 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5493                             struct diffstat_t *diffstat)
5494 {
5495         if (diff_unmodified_pair(p))
5496                 return;
5497
5498         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5499             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5500                 return; /* no useful stat for tree diffs */
5501
5502         run_diffstat(p, o, diffstat);
5503 }
5504
5505 static void diff_flush_checkdiff(struct diff_filepair *p,
5506                 struct diff_options *o)
5507 {
5508         if (diff_unmodified_pair(p))
5509                 return;
5510
5511         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5512             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5513                 return; /* nothing to check in tree diffs */
5514
5515         run_checkdiff(p, o);
5516 }
5517
5518 int diff_queue_is_empty(void)
5519 {
5520         struct diff_queue_struct *q = &diff_queued_diff;
5521         int i;
5522         for (i = 0; i < q->nr; i++)
5523                 if (!diff_unmodified_pair(q->queue[i]))
5524                         return 0;
5525         return 1;
5526 }
5527
5528 #if DIFF_DEBUG
5529 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5530 {
5531         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5532                 x, one ? one : "",
5533                 s->path,
5534                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5535                 s->mode,
5536                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5537         fprintf(stderr, "queue[%d] %s size %lu\n",
5538                 x, one ? one : "",
5539                 s->size);
5540 }
5541
5542 void diff_debug_filepair(const struct diff_filepair *p, int i)
5543 {
5544         diff_debug_filespec(p->one, i, "one");
5545         diff_debug_filespec(p->two, i, "two");
5546         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5547                 p->score, p->status ? p->status : '?',
5548                 p->one->rename_used, p->broken_pair);
5549 }
5550
5551 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5552 {
5553         int i;
5554         if (msg)
5555                 fprintf(stderr, "%s\n", msg);
5556         fprintf(stderr, "q->nr = %d\n", q->nr);
5557         for (i = 0; i < q->nr; i++) {
5558                 struct diff_filepair *p = q->queue[i];
5559                 diff_debug_filepair(p, i);
5560         }
5561 }
5562 #endif
5563
5564 static void diff_resolve_rename_copy(void)
5565 {
5566         int i;
5567         struct diff_filepair *p;
5568         struct diff_queue_struct *q = &diff_queued_diff;
5569
5570         diff_debug_queue("resolve-rename-copy", q);
5571
5572         for (i = 0; i < q->nr; i++) {
5573                 p = q->queue[i];
5574                 p->status = 0; /* undecided */
5575                 if (DIFF_PAIR_UNMERGED(p))
5576                         p->status = DIFF_STATUS_UNMERGED;
5577                 else if (!DIFF_FILE_VALID(p->one))
5578                         p->status = DIFF_STATUS_ADDED;
5579                 else if (!DIFF_FILE_VALID(p->two))
5580                         p->status = DIFF_STATUS_DELETED;
5581                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5582                         p->status = DIFF_STATUS_TYPE_CHANGED;
5583
5584                 /* from this point on, we are dealing with a pair
5585                  * whose both sides are valid and of the same type, i.e.
5586                  * either in-place edit or rename/copy edit.
5587                  */
5588                 else if (DIFF_PAIR_RENAME(p)) {
5589                         /*
5590                          * A rename might have re-connected a broken
5591                          * pair up, causing the pathnames to be the
5592                          * same again. If so, that's not a rename at
5593                          * all, just a modification..
5594                          *
5595                          * Otherwise, see if this source was used for
5596                          * multiple renames, in which case we decrement
5597                          * the count, and call it a copy.
5598                          */
5599                         if (!strcmp(p->one->path, p->two->path))
5600                                 p->status = DIFF_STATUS_MODIFIED;
5601                         else if (--p->one->rename_used > 0)
5602                                 p->status = DIFF_STATUS_COPIED;
5603                         else
5604                                 p->status = DIFF_STATUS_RENAMED;
5605                 }
5606                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5607                          p->one->mode != p->two->mode ||
5608                          p->one->dirty_submodule ||
5609                          p->two->dirty_submodule ||
5610                          is_null_oid(&p->one->oid))
5611                         p->status = DIFF_STATUS_MODIFIED;
5612                 else {
5613                         /* This is a "no-change" entry and should not
5614                          * happen anymore, but prepare for broken callers.
5615                          */
5616                         error("feeding unmodified %s to diffcore",
5617                               p->one->path);
5618                         p->status = DIFF_STATUS_UNKNOWN;
5619                 }
5620         }
5621         diff_debug_queue("resolve-rename-copy done", q);
5622 }
5623
5624 static int check_pair_status(struct diff_filepair *p)
5625 {
5626         switch (p->status) {
5627         case DIFF_STATUS_UNKNOWN:
5628                 return 0;
5629         case 0:
5630                 die("internal error in diff-resolve-rename-copy");
5631         default:
5632                 return 1;
5633         }
5634 }
5635
5636 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5637 {
5638         int fmt = opt->output_format;
5639
5640         if (fmt & DIFF_FORMAT_CHECKDIFF)
5641                 diff_flush_checkdiff(p, opt);
5642         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5643                 diff_flush_raw(p, opt);
5644         else if (fmt & DIFF_FORMAT_NAME) {
5645                 const char *name_a, *name_b;
5646                 name_a = p->two->path;
5647                 name_b = NULL;
5648                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5649                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5650                 write_name_quoted(name_a, opt->file, opt->line_termination);
5651         }
5652 }
5653
5654 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5655 {
5656         struct strbuf sb = STRBUF_INIT;
5657         if (fs->mode)
5658                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5659         else
5660                 strbuf_addf(&sb, " %s ", newdelete);
5661
5662         quote_c_style(fs->path, &sb, NULL, 0);
5663         strbuf_addch(&sb, '\n');
5664         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5665                          sb.buf, sb.len, 0);
5666         strbuf_release(&sb);
5667 }
5668
5669 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5670                 int show_name)
5671 {
5672         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5673                 struct strbuf sb = STRBUF_INIT;
5674                 strbuf_addf(&sb, " mode change %06o => %06o",
5675                             p->one->mode, p->two->mode);
5676                 if (show_name) {
5677                         strbuf_addch(&sb, ' ');
5678                         quote_c_style(p->two->path, &sb, NULL, 0);
5679                 }
5680                 strbuf_addch(&sb, '\n');
5681                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5682                                  sb.buf, sb.len, 0);
5683                 strbuf_release(&sb);
5684         }
5685 }
5686
5687 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5688                 struct diff_filepair *p)
5689 {
5690         struct strbuf sb = STRBUF_INIT;
5691         struct strbuf names = STRBUF_INIT;
5692
5693         pprint_rename(&names, p->one->path, p->two->path);
5694         strbuf_addf(&sb, " %s %s (%d%%)\n",
5695                     renamecopy, names.buf, similarity_index(p));
5696         strbuf_release(&names);
5697         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5698                                  sb.buf, sb.len, 0);
5699         show_mode_change(opt, p, 0);
5700         strbuf_release(&sb);
5701 }
5702
5703 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5704 {
5705         switch(p->status) {
5706         case DIFF_STATUS_DELETED:
5707                 show_file_mode_name(opt, "delete", p->one);
5708                 break;
5709         case DIFF_STATUS_ADDED:
5710                 show_file_mode_name(opt, "create", p->two);
5711                 break;
5712         case DIFF_STATUS_COPIED:
5713                 show_rename_copy(opt, "copy", p);
5714                 break;
5715         case DIFF_STATUS_RENAMED:
5716                 show_rename_copy(opt, "rename", p);
5717                 break;
5718         default:
5719                 if (p->score) {
5720                         struct strbuf sb = STRBUF_INIT;
5721                         strbuf_addstr(&sb, " rewrite ");
5722                         quote_c_style(p->two->path, &sb, NULL, 0);
5723                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5724                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5725                                          sb.buf, sb.len, 0);
5726                         strbuf_release(&sb);
5727                 }
5728                 show_mode_change(opt, p, !p->score);
5729                 break;
5730         }
5731 }
5732
5733 struct patch_id_t {
5734         git_SHA_CTX *ctx;
5735         int patchlen;
5736 };
5737
5738 static int remove_space(char *line, int len)
5739 {
5740         int i;
5741         char *dst = line;
5742         unsigned char c;
5743
5744         for (i = 0; i < len; i++)
5745                 if (!isspace((c = line[i])))
5746                         *dst++ = c;
5747
5748         return dst - line;
5749 }
5750
5751 static void patch_id_consume(void *priv, char *line, unsigned long len)
5752 {
5753         struct patch_id_t *data = priv;
5754         int new_len;
5755
5756         new_len = remove_space(line, len);
5757
5758         git_SHA1_Update(data->ctx, line, new_len);
5759         data->patchlen += new_len;
5760 }
5761
5762 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5763 {
5764         git_SHA1_Update(ctx, str, strlen(str));
5765 }
5766
5767 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5768 {
5769         /* large enough for 2^32 in octal */
5770         char buf[12];
5771         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5772         git_SHA1_Update(ctx, buf, len);
5773 }
5774
5775 /* returns 0 upon success, and writes result into sha1 */
5776 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5777 {
5778         struct diff_queue_struct *q = &diff_queued_diff;
5779         int i;
5780         git_SHA_CTX ctx;
5781         struct patch_id_t data;
5782
5783         git_SHA1_Init(&ctx);
5784         memset(&data, 0, sizeof(struct patch_id_t));
5785         data.ctx = &ctx;
5786
5787         for (i = 0; i < q->nr; i++) {
5788                 xpparam_t xpp;
5789                 xdemitconf_t xecfg;
5790                 mmfile_t mf1, mf2;
5791                 struct diff_filepair *p = q->queue[i];
5792                 int len1, len2;
5793
5794                 memset(&xpp, 0, sizeof(xpp));
5795                 memset(&xecfg, 0, sizeof(xecfg));
5796                 if (p->status == 0)
5797                         return error("internal diff status error");
5798                 if (p->status == DIFF_STATUS_UNKNOWN)
5799                         continue;
5800                 if (diff_unmodified_pair(p))
5801                         continue;
5802                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5803                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5804                         continue;
5805                 if (DIFF_PAIR_UNMERGED(p))
5806                         continue;
5807
5808                 diff_fill_oid_info(p->one, options->repo->index);
5809                 diff_fill_oid_info(p->two, options->repo->index);
5810
5811                 len1 = remove_space(p->one->path, strlen(p->one->path));
5812                 len2 = remove_space(p->two->path, strlen(p->two->path));
5813                 patch_id_add_string(&ctx, "diff--git");
5814                 patch_id_add_string(&ctx, "a/");
5815                 git_SHA1_Update(&ctx, p->one->path, len1);
5816                 patch_id_add_string(&ctx, "b/");
5817                 git_SHA1_Update(&ctx, p->two->path, len2);
5818
5819                 if (p->one->mode == 0) {
5820                         patch_id_add_string(&ctx, "newfilemode");
5821                         patch_id_add_mode(&ctx, p->two->mode);
5822                         patch_id_add_string(&ctx, "---/dev/null");
5823                         patch_id_add_string(&ctx, "+++b/");
5824                         git_SHA1_Update(&ctx, p->two->path, len2);
5825                 } else if (p->two->mode == 0) {
5826                         patch_id_add_string(&ctx, "deletedfilemode");
5827                         patch_id_add_mode(&ctx, p->one->mode);
5828                         patch_id_add_string(&ctx, "---a/");
5829                         git_SHA1_Update(&ctx, p->one->path, len1);
5830                         patch_id_add_string(&ctx, "+++/dev/null");
5831                 } else {
5832                         patch_id_add_string(&ctx, "---a/");
5833                         git_SHA1_Update(&ctx, p->one->path, len1);
5834                         patch_id_add_string(&ctx, "+++b/");
5835                         git_SHA1_Update(&ctx, p->two->path, len2);
5836                 }
5837
5838                 if (diff_header_only)
5839                         continue;
5840
5841                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5842                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5843                         return error("unable to read files to diff");
5844
5845                 if (diff_filespec_is_binary(options->repo, p->one) ||
5846                     diff_filespec_is_binary(options->repo, p->two)) {
5847                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5848                                         GIT_SHA1_HEXSZ);
5849                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5850                                         GIT_SHA1_HEXSZ);
5851                         continue;
5852                 }
5853
5854                 xpp.flags = 0;
5855                 xecfg.ctxlen = 3;
5856                 xecfg.flags = 0;
5857                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5858                                   patch_id_consume, &data, &xpp, &xecfg))
5859                         return error("unable to generate patch-id diff for %s",
5860                                      p->one->path);
5861         }
5862
5863         git_SHA1_Final(oid->hash, &ctx);
5864         return 0;
5865 }
5866
5867 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5868 {
5869         struct diff_queue_struct *q = &diff_queued_diff;
5870         int i;
5871         int result = diff_get_patch_id(options, oid, diff_header_only);
5872
5873         for (i = 0; i < q->nr; i++)
5874                 diff_free_filepair(q->queue[i]);
5875
5876         free(q->queue);
5877         DIFF_QUEUE_CLEAR(q);
5878
5879         return result;
5880 }
5881
5882 static int is_summary_empty(const struct diff_queue_struct *q)
5883 {
5884         int i;
5885
5886         for (i = 0; i < q->nr; i++) {
5887                 const struct diff_filepair *p = q->queue[i];
5888
5889                 switch (p->status) {
5890                 case DIFF_STATUS_DELETED:
5891                 case DIFF_STATUS_ADDED:
5892                 case DIFF_STATUS_COPIED:
5893                 case DIFF_STATUS_RENAMED:
5894                         return 0;
5895                 default:
5896                         if (p->score)
5897                                 return 0;
5898                         if (p->one->mode && p->two->mode &&
5899                             p->one->mode != p->two->mode)
5900                                 return 0;
5901                         break;
5902                 }
5903         }
5904         return 1;
5905 }
5906
5907 static const char rename_limit_warning[] =
5908 N_("inexact rename detection was skipped due to too many files.");
5909
5910 static const char degrade_cc_to_c_warning[] =
5911 N_("only found copies from modified paths due to too many files.");
5912
5913 static const char rename_limit_advice[] =
5914 N_("you may want to set your %s variable to at least "
5915    "%d and retry the command.");
5916
5917 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5918 {
5919         fflush(stdout);
5920         if (degraded_cc)
5921                 warning(_(degrade_cc_to_c_warning));
5922         else if (needed)
5923                 warning(_(rename_limit_warning));
5924         else
5925                 return;
5926         if (0 < needed)
5927                 warning(_(rename_limit_advice), varname, needed);
5928 }
5929
5930 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5931 {
5932         int i;
5933         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5934         struct diff_queue_struct *q = &diff_queued_diff;
5935
5936         if (WSEH_NEW & WS_RULE_MASK)
5937                 BUG("WS rules bit mask overlaps with diff symbol flags");
5938
5939         if (o->color_moved)
5940                 o->emitted_symbols = &esm;
5941
5942         for (i = 0; i < q->nr; i++) {
5943                 struct diff_filepair *p = q->queue[i];
5944                 if (check_pair_status(p))
5945                         diff_flush_patch(p, o);
5946         }
5947
5948         if (o->emitted_symbols) {
5949                 if (o->color_moved) {
5950                         struct hashmap add_lines, del_lines;
5951
5952                         if (o->color_moved_ws_handling &
5953                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5954                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5955
5956                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5957                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5958
5959                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5960                         mark_color_as_moved(o, &add_lines, &del_lines);
5961                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5962                                 dim_moved_lines(o);
5963
5964                         hashmap_free(&add_lines, 1);
5965                         hashmap_free(&del_lines, 1);
5966                 }
5967
5968                 for (i = 0; i < esm.nr; i++)
5969                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5970
5971                 for (i = 0; i < esm.nr; i++)
5972                         free((void *)esm.buf[i].line);
5973                 esm.nr = 0;
5974
5975                 o->emitted_symbols = NULL;
5976         }
5977 }
5978
5979 void diff_flush(struct diff_options *options)
5980 {
5981         struct diff_queue_struct *q = &diff_queued_diff;
5982         int i, output_format = options->output_format;
5983         int separator = 0;
5984         int dirstat_by_line = 0;
5985
5986         /*
5987          * Order: raw, stat, summary, patch
5988          * or:    name/name-status/checkdiff (other bits clear)
5989          */
5990         if (!q->nr)
5991                 goto free_queue;
5992
5993         if (output_format & (DIFF_FORMAT_RAW |
5994                              DIFF_FORMAT_NAME |
5995                              DIFF_FORMAT_NAME_STATUS |
5996                              DIFF_FORMAT_CHECKDIFF)) {
5997                 for (i = 0; i < q->nr; i++) {
5998                         struct diff_filepair *p = q->queue[i];
5999                         if (check_pair_status(p))
6000                                 flush_one_pair(p, options);
6001                 }
6002                 separator++;
6003         }
6004
6005         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6006                 dirstat_by_line = 1;
6007
6008         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6009             dirstat_by_line) {
6010                 struct diffstat_t diffstat;
6011
6012                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6013                 for (i = 0; i < q->nr; i++) {
6014                         struct diff_filepair *p = q->queue[i];
6015                         if (check_pair_status(p))
6016                                 diff_flush_stat(p, options, &diffstat);
6017                 }
6018                 if (output_format & DIFF_FORMAT_NUMSTAT)
6019                         show_numstat(&diffstat, options);
6020                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6021                         show_stats(&diffstat, options);
6022                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6023                         show_shortstats(&diffstat, options);
6024                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6025                         show_dirstat_by_line(&diffstat, options);
6026                 free_diffstat_info(&diffstat);
6027                 separator++;
6028         }
6029         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6030                 show_dirstat(options);
6031
6032         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6033                 for (i = 0; i < q->nr; i++) {
6034                         diff_summary(options, q->queue[i]);
6035                 }
6036                 separator++;
6037         }
6038
6039         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6040             options->flags.exit_with_status &&
6041             options->flags.diff_from_contents) {
6042                 /*
6043                  * run diff_flush_patch for the exit status. setting
6044                  * options->file to /dev/null should be safe, because we
6045                  * aren't supposed to produce any output anyway.
6046                  */
6047                 if (options->close_file)
6048                         fclose(options->file);
6049                 options->file = xfopen("/dev/null", "w");
6050                 options->close_file = 1;
6051                 options->color_moved = 0;
6052                 for (i = 0; i < q->nr; i++) {
6053                         struct diff_filepair *p = q->queue[i];
6054                         if (check_pair_status(p))
6055                                 diff_flush_patch(p, options);
6056                         if (options->found_changes)
6057                                 break;
6058                 }
6059         }
6060
6061         if (output_format & DIFF_FORMAT_PATCH) {
6062                 if (separator) {
6063                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6064                         if (options->stat_sep)
6065                                 /* attach patch instead of inline */
6066                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6067                                                  NULL, 0, 0);
6068                 }
6069
6070                 diff_flush_patch_all_file_pairs(options);
6071         }
6072
6073         if (output_format & DIFF_FORMAT_CALLBACK)
6074                 options->format_callback(q, options, options->format_callback_data);
6075
6076         for (i = 0; i < q->nr; i++)
6077                 diff_free_filepair(q->queue[i]);
6078 free_queue:
6079         free(q->queue);
6080         DIFF_QUEUE_CLEAR(q);
6081         if (options->close_file)
6082                 fclose(options->file);
6083
6084         /*
6085          * Report the content-level differences with HAS_CHANGES;
6086          * diff_addremove/diff_change does not set the bit when
6087          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6088          */
6089         if (options->flags.diff_from_contents) {
6090                 if (options->found_changes)
6091                         options->flags.has_changes = 1;
6092                 else
6093                         options->flags.has_changes = 0;
6094         }
6095 }
6096
6097 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6098 {
6099         return (((p->status == DIFF_STATUS_MODIFIED) &&
6100                  ((p->score &&
6101                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6102                   (!p->score &&
6103                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6104                 ((p->status != DIFF_STATUS_MODIFIED) &&
6105                  filter_bit_tst(p->status, options)));
6106 }
6107
6108 static void diffcore_apply_filter(struct diff_options *options)
6109 {
6110         int i;
6111         struct diff_queue_struct *q = &diff_queued_diff;
6112         struct diff_queue_struct outq;
6113
6114         DIFF_QUEUE_CLEAR(&outq);
6115
6116         if (!options->filter)
6117                 return;
6118
6119         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6120                 int found;
6121                 for (i = found = 0; !found && i < q->nr; i++) {
6122                         if (match_filter(options, q->queue[i]))
6123                                 found++;
6124                 }
6125                 if (found)
6126                         return;
6127
6128                 /* otherwise we will clear the whole queue
6129                  * by copying the empty outq at the end of this
6130                  * function, but first clear the current entries
6131                  * in the queue.
6132                  */
6133                 for (i = 0; i < q->nr; i++)
6134                         diff_free_filepair(q->queue[i]);
6135         }
6136         else {
6137                 /* Only the matching ones */
6138                 for (i = 0; i < q->nr; i++) {
6139                         struct diff_filepair *p = q->queue[i];
6140                         if (match_filter(options, p))
6141                                 diff_q(&outq, p);
6142                         else
6143                                 diff_free_filepair(p);
6144                 }
6145         }
6146         free(q->queue);
6147         *q = outq;
6148 }
6149
6150 /* Check whether two filespecs with the same mode and size are identical */
6151 static int diff_filespec_is_identical(struct repository *r,
6152                                       struct diff_filespec *one,
6153                                       struct diff_filespec *two)
6154 {
6155         if (S_ISGITLINK(one->mode))
6156                 return 0;
6157         if (diff_populate_filespec(r, one, 0))
6158                 return 0;
6159         if (diff_populate_filespec(r, two, 0))
6160                 return 0;
6161         return !memcmp(one->data, two->data, one->size);
6162 }
6163
6164 static int diff_filespec_check_stat_unmatch(struct repository *r,
6165                                             struct diff_filepair *p)
6166 {
6167         if (p->done_skip_stat_unmatch)
6168                 return p->skip_stat_unmatch_result;
6169
6170         p->done_skip_stat_unmatch = 1;
6171         p->skip_stat_unmatch_result = 0;
6172         /*
6173          * 1. Entries that come from stat info dirtiness
6174          *    always have both sides (iow, not create/delete),
6175          *    one side of the object name is unknown, with
6176          *    the same mode and size.  Keep the ones that
6177          *    do not match these criteria.  They have real
6178          *    differences.
6179          *
6180          * 2. At this point, the file is known to be modified,
6181          *    with the same mode and size, and the object
6182          *    name of one side is unknown.  Need to inspect
6183          *    the identical contents.
6184          */
6185         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6186             !DIFF_FILE_VALID(p->two) ||
6187             (p->one->oid_valid && p->two->oid_valid) ||
6188             (p->one->mode != p->two->mode) ||
6189             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6190             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6191             (p->one->size != p->two->size) ||
6192             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6193                 p->skip_stat_unmatch_result = 1;
6194         return p->skip_stat_unmatch_result;
6195 }
6196
6197 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6198 {
6199         int i;
6200         struct diff_queue_struct *q = &diff_queued_diff;
6201         struct diff_queue_struct outq;
6202         DIFF_QUEUE_CLEAR(&outq);
6203
6204         for (i = 0; i < q->nr; i++) {
6205                 struct diff_filepair *p = q->queue[i];
6206
6207                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6208                         diff_q(&outq, p);
6209                 else {
6210                         /*
6211                          * The caller can subtract 1 from skip_stat_unmatch
6212                          * to determine how many paths were dirty only
6213                          * due to stat info mismatch.
6214                          */
6215                         if (!diffopt->flags.no_index)
6216                                 diffopt->skip_stat_unmatch++;
6217                         diff_free_filepair(p);
6218                 }
6219         }
6220         free(q->queue);
6221         *q = outq;
6222 }
6223
6224 static int diffnamecmp(const void *a_, const void *b_)
6225 {
6226         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6227         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6228         const char *name_a, *name_b;
6229
6230         name_a = a->one ? a->one->path : a->two->path;
6231         name_b = b->one ? b->one->path : b->two->path;
6232         return strcmp(name_a, name_b);
6233 }
6234
6235 void diffcore_fix_diff_index(struct diff_options *options)
6236 {
6237         struct diff_queue_struct *q = &diff_queued_diff;
6238         QSORT(q->queue, q->nr, diffnamecmp);
6239 }
6240
6241 void diffcore_std(struct diff_options *options)
6242 {
6243         /* NOTE please keep the following in sync with diff_tree_combined() */
6244         if (options->skip_stat_unmatch)
6245                 diffcore_skip_stat_unmatch(options);
6246         if (!options->found_follow) {
6247                 /* See try_to_follow_renames() in tree-diff.c */
6248                 if (options->break_opt != -1)
6249                         diffcore_break(options->repo,
6250                                        options->break_opt);
6251                 if (options->detect_rename)
6252                         diffcore_rename(options);
6253                 if (options->break_opt != -1)
6254                         diffcore_merge_broken();
6255         }
6256         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6257                 diffcore_pickaxe(options);
6258         if (options->orderfile)
6259                 diffcore_order(options->orderfile);
6260         if (!options->found_follow)
6261                 /* See try_to_follow_renames() in tree-diff.c */
6262                 diff_resolve_rename_copy();
6263         diffcore_apply_filter(options);
6264
6265         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6266                 options->flags.has_changes = 1;
6267         else
6268                 options->flags.has_changes = 0;
6269
6270         options->found_follow = 0;
6271 }
6272
6273 int diff_result_code(struct diff_options *opt, int status)
6274 {
6275         int result = 0;
6276
6277         diff_warn_rename_limit("diff.renameLimit",
6278                                opt->needed_rename_limit,
6279                                opt->degraded_cc_to_c);
6280         if (!opt->flags.exit_with_status &&
6281             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6282                 return status;
6283         if (opt->flags.exit_with_status &&
6284             opt->flags.has_changes)
6285                 result |= 01;
6286         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6287             opt->flags.check_failed)
6288                 result |= 02;
6289         return result;
6290 }
6291
6292 int diff_can_quit_early(struct diff_options *opt)
6293 {
6294         return (opt->flags.quick &&
6295                 !opt->filter &&
6296                 opt->flags.has_changes);
6297 }
6298
6299 /*
6300  * Shall changes to this submodule be ignored?
6301  *
6302  * Submodule changes can be configured to be ignored separately for each path,
6303  * but that configuration can be overridden from the command line.
6304  */
6305 static int is_submodule_ignored(const char *path, struct diff_options *options)
6306 {
6307         int ignored = 0;
6308         struct diff_flags orig_flags = options->flags;
6309         if (!options->flags.override_submodule_config)
6310                 set_diffopt_flags_from_submodule_config(options, path);
6311         if (options->flags.ignore_submodules)
6312                 ignored = 1;
6313         options->flags = orig_flags;
6314         return ignored;
6315 }
6316
6317 void diff_addremove(struct diff_options *options,
6318                     int addremove, unsigned mode,
6319                     const struct object_id *oid,
6320                     int oid_valid,
6321                     const char *concatpath, unsigned dirty_submodule)
6322 {
6323         struct diff_filespec *one, *two;
6324
6325         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6326                 return;
6327
6328         /* This may look odd, but it is a preparation for
6329          * feeding "there are unchanged files which should
6330          * not produce diffs, but when you are doing copy
6331          * detection you would need them, so here they are"
6332          * entries to the diff-core.  They will be prefixed
6333          * with something like '=' or '*' (I haven't decided
6334          * which but should not make any difference).
6335          * Feeding the same new and old to diff_change()
6336          * also has the same effect.
6337          * Before the final output happens, they are pruned after
6338          * merged into rename/copy pairs as appropriate.
6339          */
6340         if (options->flags.reverse_diff)
6341                 addremove = (addremove == '+' ? '-' :
6342                              addremove == '-' ? '+' : addremove);
6343
6344         if (options->prefix &&
6345             strncmp(concatpath, options->prefix, options->prefix_length))
6346                 return;
6347
6348         one = alloc_filespec(concatpath);
6349         two = alloc_filespec(concatpath);
6350
6351         if (addremove != '+')
6352                 fill_filespec(one, oid, oid_valid, mode);
6353         if (addremove != '-') {
6354                 fill_filespec(two, oid, oid_valid, mode);
6355                 two->dirty_submodule = dirty_submodule;
6356         }
6357
6358         diff_queue(&diff_queued_diff, one, two);
6359         if (!options->flags.diff_from_contents)
6360                 options->flags.has_changes = 1;
6361 }
6362
6363 void diff_change(struct diff_options *options,
6364                  unsigned old_mode, unsigned new_mode,
6365                  const struct object_id *old_oid,
6366                  const struct object_id *new_oid,
6367                  int old_oid_valid, int new_oid_valid,
6368                  const char *concatpath,
6369                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6370 {
6371         struct diff_filespec *one, *two;
6372         struct diff_filepair *p;
6373
6374         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6375             is_submodule_ignored(concatpath, options))
6376                 return;
6377
6378         if (options->flags.reverse_diff) {
6379                 SWAP(old_mode, new_mode);
6380                 SWAP(old_oid, new_oid);
6381                 SWAP(old_oid_valid, new_oid_valid);
6382                 SWAP(old_dirty_submodule, new_dirty_submodule);
6383         }
6384
6385         if (options->prefix &&
6386             strncmp(concatpath, options->prefix, options->prefix_length))
6387                 return;
6388
6389         one = alloc_filespec(concatpath);
6390         two = alloc_filespec(concatpath);
6391         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6392         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6393         one->dirty_submodule = old_dirty_submodule;
6394         two->dirty_submodule = new_dirty_submodule;
6395         p = diff_queue(&diff_queued_diff, one, two);
6396
6397         if (options->flags.diff_from_contents)
6398                 return;
6399
6400         if (options->flags.quick && options->skip_stat_unmatch &&
6401             !diff_filespec_check_stat_unmatch(options->repo, p))
6402                 return;
6403
6404         options->flags.has_changes = 1;
6405 }
6406
6407 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6408 {
6409         struct diff_filepair *pair;
6410         struct diff_filespec *one, *two;
6411
6412         if (options->prefix &&
6413             strncmp(path, options->prefix, options->prefix_length))
6414                 return NULL;
6415
6416         one = alloc_filespec(path);
6417         two = alloc_filespec(path);
6418         pair = diff_queue(&diff_queued_diff, one, two);
6419         pair->is_unmerged = 1;
6420         return pair;
6421 }
6422
6423 static char *run_textconv(struct repository *r,
6424                           const char *pgm,
6425                           struct diff_filespec *spec,
6426                           size_t *outsize)
6427 {
6428         struct diff_tempfile *temp;
6429         const char *argv[3];
6430         const char **arg = argv;
6431         struct child_process child = CHILD_PROCESS_INIT;
6432         struct strbuf buf = STRBUF_INIT;
6433         int err = 0;
6434
6435         temp = prepare_temp_file(r, spec->path, spec);
6436         *arg++ = pgm;
6437         *arg++ = temp->name;
6438         *arg = NULL;
6439
6440         child.use_shell = 1;
6441         child.argv = argv;
6442         child.out = -1;
6443         if (start_command(&child)) {
6444                 remove_tempfile();
6445                 return NULL;
6446         }
6447
6448         if (strbuf_read(&buf, child.out, 0) < 0)
6449                 err = error("error reading from textconv command '%s'", pgm);
6450         close(child.out);
6451
6452         if (finish_command(&child) || err) {
6453                 strbuf_release(&buf);
6454                 remove_tempfile();
6455                 return NULL;
6456         }
6457         remove_tempfile();
6458
6459         return strbuf_detach(&buf, outsize);
6460 }
6461
6462 size_t fill_textconv(struct repository *r,
6463                      struct userdiff_driver *driver,
6464                      struct diff_filespec *df,
6465                      char **outbuf)
6466 {
6467         size_t size;
6468
6469         if (!driver) {
6470                 if (!DIFF_FILE_VALID(df)) {
6471                         *outbuf = "";
6472                         return 0;
6473                 }
6474                 if (diff_populate_filespec(r, df, 0))
6475                         die("unable to read files to diff");
6476                 *outbuf = df->data;
6477                 return df->size;
6478         }
6479
6480         if (!driver->textconv)
6481                 BUG("fill_textconv called with non-textconv driver");
6482
6483         if (driver->textconv_cache && df->oid_valid) {
6484                 *outbuf = notes_cache_get(driver->textconv_cache,
6485                                           &df->oid,
6486                                           &size);
6487                 if (*outbuf)
6488                         return size;
6489         }
6490
6491         *outbuf = run_textconv(r, driver->textconv, df, &size);
6492         if (!*outbuf)
6493                 die("unable to read files to diff");
6494
6495         if (driver->textconv_cache && df->oid_valid) {
6496                 /* ignore errors, as we might be in a readonly repository */
6497                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6498                                 size);
6499                 /*
6500                  * we could save up changes and flush them all at the end,
6501                  * but we would need an extra call after all diffing is done.
6502                  * Since generating a cache entry is the slow path anyway,
6503                  * this extra overhead probably isn't a big deal.
6504                  */
6505                 notes_cache_write(driver->textconv_cache);
6506         }
6507
6508         return size;
6509 }
6510
6511 int textconv_object(struct repository *r,
6512                     const char *path,
6513                     unsigned mode,
6514                     const struct object_id *oid,
6515                     int oid_valid,
6516                     char **buf,
6517                     unsigned long *buf_size)
6518 {
6519         struct diff_filespec *df;
6520         struct userdiff_driver *textconv;
6521
6522         df = alloc_filespec(path);
6523         fill_filespec(df, oid, oid_valid, mode);
6524         textconv = get_textconv(r, df);
6525         if (!textconv) {
6526                 free_filespec(df);
6527                 return 0;
6528         }
6529
6530         *buf_size = fill_textconv(r, textconv, df, buf);
6531         free_filespec(df);
6532         return 1;
6533 }
6534
6535 void setup_diff_pager(struct diff_options *opt)
6536 {
6537         /*
6538          * If the user asked for our exit code, then either they want --quiet
6539          * or --exit-code. We should definitely not bother with a pager in the
6540          * former case, as we will generate no output. Since we still properly
6541          * report our exit code even when a pager is run, we _could_ run a
6542          * pager with --exit-code. But since we have not done so historically,
6543          * and because it is easy to find people oneline advising "git diff
6544          * --exit-code" in hooks and other scripts, we do not do so.
6545          */
6546         if (!opt->flags.exit_with_status &&
6547             check_pager_config("diff") != 0)
6548                 setup_pager();
6549 }