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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed-zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else if (!strcmp(arg, "dimmed_zebra"))
 289                 return COLOR_MOVED_ZEBRA_DIM;
 290         else
 291                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 292 }
 293
 294 static unsigned parse_color_moved_ws(const char *arg)
 295 {
 296         int ret = 0;
 297         struct string_list l = STRING_LIST_INIT_DUP;
 298         struct string_list_item *i;
 299
 300         string_list_split(&l, arg, ',', -1);
 301
 302         for_each_string_list_item(i, &l) {
 303                 struct strbuf sb = STRBUF_INIT;
 304                 strbuf_addstr(&sb, i->string);
 305                 strbuf_trim(&sb);
 306
 307                 if (!strcmp(sb.buf, "no"))
 308                         ret = 0;
 309                 else if (!strcmp(sb.buf, "ignore-space-change"))
 310                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 311                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 312                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 313                 else if (!strcmp(sb.buf, "ignore-all-space"))
 314                         ret |= XDF_IGNORE_WHITESPACE;
 315                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 316                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 317                 else {
 318                         ret |= COLOR_MOVED_WS_ERROR;
 319                         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);
 320                 }
 321
 322                 strbuf_release(&sb);
 323         }
 324
 325         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 326             (ret & XDF_WHITESPACE_FLAGS)) {
 327                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 328                 ret |= COLOR_MOVED_WS_ERROR;
 329         }
 330
 331         string_list_clear(&l, 0);
 332
 333         return ret;
 334 }
 335
 336 int git_diff_ui_config(const char *var, const char *value, void *cb)
 337 {
 338         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 339                 diff_use_color_default = git_config_colorbool(var, value);
 340                 return 0;
 341         }
 342         if (!strcmp(var, "diff.colormoved")) {
 343                 int cm = parse_color_moved(value);
 344                 if (cm < 0)
 345                         return -1;
 346                 diff_color_moved_default = cm;
 347                 return 0;
 348         }
 349         if (!strcmp(var, "diff.colormovedws")) {
 350                 unsigned cm = parse_color_moved_ws(value);
 351                 if (cm & COLOR_MOVED_WS_ERROR)
 352                         return -1;
 353                 diff_color_moved_ws_default = cm;
 354                 return 0;
 355         }
 356         if (!strcmp(var, "diff.context")) {
 357                 diff_context_default = git_config_int(var, value);
 358                 if (diff_context_default < 0)
 359                         return -1;
 360                 return 0;
 361         }
 362         if (!strcmp(var, "diff.interhunkcontext")) {
 363                 diff_interhunk_context_default = git_config_int(var, value);
 364                 if (diff_interhunk_context_default < 0)
 365                         return -1;
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.renames")) {
 369                 diff_detect_rename_default = git_config_rename(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.autorefreshindex")) {
 373                 diff_auto_refresh_index = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.mnemonicprefix")) {
 377                 diff_mnemonic_prefix = git_config_bool(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.noprefix")) {
 381                 diff_no_prefix = git_config_bool(var, value);
 382                 return 0;
 383         }
 384         if (!strcmp(var, "diff.statgraphwidth")) {
 385                 diff_stat_graph_width = git_config_int(var, value);
 386                 return 0;
 387         }
 388         if (!strcmp(var, "diff.external"))
 389                 return git_config_string(&external_diff_cmd_cfg, var, value);
 390         if (!strcmp(var, "diff.wordregex"))
 391                 return git_config_string(&diff_word_regex_cfg, var, value);
 392         if (!strcmp(var, "diff.orderfile"))
 393                 return git_config_pathname(&diff_order_file_cfg, var, value);
 394
 395         if (!strcmp(var, "diff.ignoresubmodules"))
 396                 handle_ignore_submodules_arg(&default_diff_options, value);
 397
 398         if (!strcmp(var, "diff.submodule")) {
 399                 if (parse_submodule_params(&default_diff_options, value))
 400                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 401                                 value);
 402                 return 0;
 403         }
 404
 405         if (!strcmp(var, "diff.algorithm")) {
 406                 diff_algorithm = parse_algorithm_value(value);
 407                 if (diff_algorithm < 0)
 408                         return -1;
 409                 return 0;
 410         }
 411
 412         if (!strcmp(var, "diff.wserrorhighlight")) {
 413                 int val = parse_ws_error_highlight(value);
 414                 if (val < 0)
 415                         return -1;
 416                 ws_error_highlight_default = val;
 417                 return 0;
 418         }
 419
 420         if (git_color_config(var, value, cb) < 0)
 421                 return -1;
 422
 423         return git_diff_basic_config(var, value, cb);
 424 }
 425
 426 int git_diff_basic_config(const char *var, const char *value, void *cb)
 427 {
 428         const char *name;
 429
 430         if (!strcmp(var, "diff.renamelimit")) {
 431                 diff_rename_limit_default = git_config_int(var, value);
 432                 return 0;
 433         }
 434
 435         if (userdiff_config(var, value) < 0)
 436                 return -1;
 437
 438         if (skip_prefix(var, "diff.color.", &name) ||
 439             skip_prefix(var, "color.diff.", &name)) {
 440                 int slot = parse_diff_color_slot(name);
 441                 if (slot < 0)
 442                         return 0;
 443                 if (!value)
 444                         return config_error_nonbool(var);
 445                 return color_parse(value, diff_colors[slot]);
 446         }
 447
 448         /* like GNU diff's --suppress-blank-empty option  */
 449         if (!strcmp(var, "diff.suppressblankempty") ||
 450                         /* for backwards compatibility */
 451                         !strcmp(var, "diff.suppress-blank-empty")) {
 452                 diff_suppress_blank_empty = git_config_bool(var, value);
 453                 return 0;
 454         }
 455
 456         if (!strcmp(var, "diff.dirstat")) {
 457                 struct strbuf errmsg = STRBUF_INIT;
 458                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 459                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 460                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 461                                 errmsg.buf);
 462                 strbuf_release(&errmsg);
 463                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 464                 return 0;
 465         }
 466
 467         if (git_diff_heuristic_config(var, value, cb) < 0)
 468                 return -1;
 469
 470         return git_default_config(var, value, cb);
 471 }
 472
 473 static char *quote_two(const char *one, const char *two)
 474 {
 475         int need_one = quote_c_style(one, NULL, NULL, 1);
 476         int need_two = quote_c_style(two, NULL, NULL, 1);
 477         struct strbuf res = STRBUF_INIT;
 478
 479         if (need_one + need_two) {
 480                 strbuf_addch(&res, '"');
 481                 quote_c_style(one, &res, NULL, 1);
 482                 quote_c_style(two, &res, NULL, 1);
 483                 strbuf_addch(&res, '"');
 484         } else {
 485                 strbuf_addstr(&res, one);
 486                 strbuf_addstr(&res, two);
 487         }
 488         return strbuf_detach(&res, NULL);
 489 }
 490
 491 static const char *external_diff(void)
 492 {
 493         static const char *external_diff_cmd = NULL;
 494         static int done_preparing = 0;
 495
 496         if (done_preparing)
 497                 return external_diff_cmd;
 498         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 499         if (!external_diff_cmd)
 500                 external_diff_cmd = external_diff_cmd_cfg;
 501         done_preparing = 1;
 502         return external_diff_cmd;
 503 }
 504
 505 /*
 506  * Keep track of files used for diffing. Sometimes such an entry
 507  * refers to a temporary file, sometimes to an existing file, and
 508  * sometimes to "/dev/null".
 509  */
 510 static struct diff_tempfile {
 511         /*
 512          * filename external diff should read from, or NULL if this
 513          * entry is currently not in use:
 514          */
 515         const char *name;
 516
 517         char hex[GIT_MAX_HEXSZ + 1];
 518         char mode[10];
 519
 520         /*
 521          * If this diff_tempfile instance refers to a temporary file,
 522          * this tempfile object is used to manage its lifetime.
 523          */
 524         struct tempfile *tempfile;
 525 } diff_temp[2];
 526
 527 struct emit_callback {
 528         int color_diff;
 529         unsigned ws_rule;
 530         int blank_at_eof_in_preimage;
 531         int blank_at_eof_in_postimage;
 532         int lno_in_preimage;
 533         int lno_in_postimage;
 534         const char **label_path;
 535         struct diff_words_data *diff_words;
 536         struct diff_options *opt;
 537         struct strbuf *header;
 538 };
 539
 540 static int count_lines(const char *data, int size)
 541 {
 542         int count, ch, completely_empty = 1, nl_just_seen = 0;
 543         count = 0;
 544         while (0 < size--) {
 545                 ch = *data++;
 546                 if (ch == '\n') {
 547                         count++;
 548                         nl_just_seen = 1;
 549                         completely_empty = 0;
 550                 }
 551                 else {
 552                         nl_just_seen = 0;
 553                         completely_empty = 0;
 554                 }
 555         }
 556         if (completely_empty)
 557                 return 0;
 558         if (!nl_just_seen)
 559                 count++; /* no trailing newline */
 560         return count;
 561 }
 562
 563 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 564                        struct diff_filespec *one)
 565 {
 566         if (!DIFF_FILE_VALID(one)) {
 567                 mf->ptr = (char *)""; /* does not matter */
 568                 mf->size = 0;
 569                 return 0;
 570         }
 571         else if (diff_populate_filespec(r, one, 0))
 572                 return -1;
 573
 574         mf->ptr = one->data;
 575         mf->size = one->size;
 576         return 0;
 577 }
 578
 579 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 580 static unsigned long diff_filespec_size(struct repository *r,
 581                                         struct diff_filespec *one)
 582 {
 583         if (!DIFF_FILE_VALID(one))
 584                 return 0;
 585         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 586         return one->size;
 587 }
 588
 589 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 590 {
 591         char *ptr = mf->ptr;
 592         long size = mf->size;
 593         int cnt = 0;
 594
 595         if (!size)
 596                 return cnt;
 597         ptr += size - 1; /* pointing at the very end */
 598         if (*ptr != '\n')
 599                 ; /* incomplete line */
 600         else
 601                 ptr--; /* skip the last LF */
 602         while (mf->ptr < ptr) {
 603                 char *prev_eol;
 604                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 605                         if (*prev_eol == '\n')
 606                                 break;
 607                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 608                         break;
 609                 cnt++;
 610                 ptr = prev_eol - 1;
 611         }
 612         return cnt;
 613 }
 614
 615 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 616                                struct emit_callback *ecbdata)
 617 {
 618         int l1, l2, at;
 619         unsigned ws_rule = ecbdata->ws_rule;
 620         l1 = count_trailing_blank(mf1, ws_rule);
 621         l2 = count_trailing_blank(mf2, ws_rule);
 622         if (l2 <= l1) {
 623                 ecbdata->blank_at_eof_in_preimage = 0;
 624                 ecbdata->blank_at_eof_in_postimage = 0;
 625                 return;
 626         }
 627         at = count_lines(mf1->ptr, mf1->size);
 628         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 629
 630         at = count_lines(mf2->ptr, mf2->size);
 631         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 632 }
 633
 634 static void emit_line_0(struct diff_options *o,
 635                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 636                         int first, const char *line, int len)
 637 {
 638         int has_trailing_newline, has_trailing_carriage_return;
 639         int needs_reset = 0; /* at the end of the line */
 640         FILE *file = o->file;
 641
 642         fputs(diff_line_prefix(o), file);
 643
 644         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 645         if (has_trailing_newline)
 646                 len--;
 647
 648         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 649         if (has_trailing_carriage_return)
 650                 len--;
 651
 652         if (!len && !first)
 653                 goto end_of_line;
 654
 655         if (reverse && want_color(o->use_color)) {
 656                 fputs(GIT_COLOR_REVERSE, file);
 657                 needs_reset = 1;
 658         }
 659
 660         if (set_sign) {
 661                 fputs(set_sign, file);
 662                 needs_reset = 1;
 663         }
 664
 665         if (first)
 666                 fputc(first, file);
 667
 668         if (!len)
 669                 goto end_of_line;
 670
 671         if (set) {
 672                 if (set_sign && set != set_sign)
 673                         fputs(reset, file);
 674                 fputs(set, file);
 675                 needs_reset = 1;
 676         }
 677         fwrite(line, len, 1, file);
 678         needs_reset = 1; /* 'line' may contain color codes. */
 679
 680 end_of_line:
 681         if (needs_reset)
 682                 fputs(reset, file);
 683         if (has_trailing_carriage_return)
 684                 fputc('\r', file);
 685         if (has_trailing_newline)
 686                 fputc('\n', file);
 687 }
 688
 689 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 690                       const char *line, int len)
 691 {
 692         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 693 }
 694
 695 enum diff_symbol {
 696         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 697         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 698         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 699         DIFF_SYMBOL_BINARY_DIFF_BODY,
 700         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 701         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 702         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 703         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 704         DIFF_SYMBOL_STATS_LINE,
 705         DIFF_SYMBOL_WORD_DIFF,
 706         DIFF_SYMBOL_STAT_SEP,
 707         DIFF_SYMBOL_SUMMARY,
 708         DIFF_SYMBOL_SUBMODULE_ADD,
 709         DIFF_SYMBOL_SUBMODULE_DEL,
 710         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 711         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 712         DIFF_SYMBOL_SUBMODULE_HEADER,
 713         DIFF_SYMBOL_SUBMODULE_ERROR,
 714         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 715         DIFF_SYMBOL_REWRITE_DIFF,
 716         DIFF_SYMBOL_BINARY_FILES,
 717         DIFF_SYMBOL_HEADER,
 718         DIFF_SYMBOL_FILEPAIR_PLUS,
 719         DIFF_SYMBOL_FILEPAIR_MINUS,
 720         DIFF_SYMBOL_WORDS_PORCELAIN,
 721         DIFF_SYMBOL_WORDS,
 722         DIFF_SYMBOL_CONTEXT,
 723         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 724         DIFF_SYMBOL_PLUS,
 725         DIFF_SYMBOL_MINUS,
 726         DIFF_SYMBOL_NO_LF_EOF,
 727         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 728         DIFF_SYMBOL_CONTEXT_MARKER,
 729         DIFF_SYMBOL_SEPARATOR
 730 };
 731 /*
 732  * Flags for content lines:
 733  * 0..12 are whitespace rules
 734  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 735  * 16 is marking if the line is blank at EOF
 736  */
 737 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 738 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 739 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 740 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 741 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 742
 743 /*
 744  * This struct is used when we need to buffer the output of the diff output.
 745  *
 746  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 747  * into the pre/post image file. This pointer could be a union with the
 748  * line pointer. By storing an offset into the file instead of the literal line,
 749  * we can decrease the memory footprint for the buffered output. At first we
 750  * may want to only have indirection for the content lines, but we could also
 751  * enhance the state for emitting prefabricated lines, e.g. the similarity
 752  * score line or hunk/file headers would only need to store a number or path
 753  * and then the output can be constructed later on depending on state.
 754  */
 755 struct emitted_diff_symbol {
 756         const char *line;
 757         int len;
 758         int flags;
 759         int indent_off;   /* Offset to first non-whitespace character */
 760         int indent_width; /* The visual width of the indentation */
 761         enum diff_symbol s;
 762 };
 763 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 764
 765 struct emitted_diff_symbols {
 766         struct emitted_diff_symbol *buf;
 767         int nr, alloc;
 768 };
 769 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 770
 771 static void append_emitted_diff_symbol(struct diff_options *o,
 772                                        struct emitted_diff_symbol *e)
 773 {
 774         struct emitted_diff_symbol *f;
 775
 776         ALLOC_GROW(o->emitted_symbols->buf,
 777                    o->emitted_symbols->nr + 1,
 778                    o->emitted_symbols->alloc);
 779         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 780
 781         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 782         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 783 }
 784
 785 struct moved_entry {
 786         struct hashmap_entry ent;
 787         const struct emitted_diff_symbol *es;
 788         struct moved_entry *next_line;
 789 };
 790
 791 struct moved_block {
 792         struct moved_entry *match;
 793         int wsd; /* The whitespace delta of this block */
 794 };
 795
 796 static void moved_block_clear(struct moved_block *b)
 797 {
 798         memset(b, 0, sizeof(*b));
 799 }
 800
 801 #define INDENT_BLANKLINE INT_MIN
 802
 803 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 804 {
 805         unsigned int off = 0, i;
 806         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 807         const char *s = es->line;
 808         const int len = es->len;
 809
 810         /* skip any \v \f \r at start of indentation */
 811         while (s[off] == '\f' || s[off] == '\v' ||
 812                (s[off] == '\r' && off < len - 1))
 813                 off++;
 814
 815         /* calculate the visual width of indentation */
 816         while(1) {
 817                 if (s[off] == ' ') {
 818                         width++;
 819                         off++;
 820                 } else if (s[off] == '\t') {
 821                         width += tab_width - (width % tab_width);
 822                         while (s[++off] == '\t')
 823                                 width += tab_width;
 824                 } else {
 825                         break;
 826                 }
 827         }
 828
 829         /* check if this line is blank */
 830         for (i = off; i < len; i++)
 831                 if (!isspace(s[i]))
 832                     break;
 833
 834         if (i == len) {
 835                 es->indent_width = INDENT_BLANKLINE;
 836                 es->indent_off = len;
 837         } else {
 838                 es->indent_off = off;
 839                 es->indent_width = width;
 840         }
 841 }
 842
 843 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 844                             const struct emitted_diff_symbol *b,
 845                             int *out)
 846 {
 847         int a_len = a->len,
 848             b_len = b->len,
 849             a_off = a->indent_off,
 850             a_width = a->indent_width,
 851             b_off = b->indent_off,
 852             b_width = b->indent_width;
 853         int delta;
 854
 855         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 856                 *out = INDENT_BLANKLINE;
 857                 return 1;
 858         }
 859
 860         if (a->s == DIFF_SYMBOL_PLUS)
 861                 delta = a_width - b_width;
 862         else
 863                 delta = b_width - a_width;
 864
 865         if (a_len - a_off != b_len - b_off ||
 866             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 867                 return 0;
 868
 869         *out = delta;
 870
 871         return 1;
 872 }
 873
 874 static int cmp_in_block_with_wsd(const struct diff_options *o,
 875                                  const struct moved_entry *cur,
 876                                  const struct moved_entry *match,
 877                                  struct moved_block *pmb,
 878                                  int n)
 879 {
 880         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 881         int al = cur->es->len, bl = match->es->len, cl = l->len;
 882         const char *a = cur->es->line,
 883                    *b = match->es->line,
 884                    *c = l->line;
 885         int a_off = cur->es->indent_off,
 886             a_width = cur->es->indent_width,
 887             c_off = l->indent_off,
 888             c_width = l->indent_width;
 889         int delta;
 890
 891         /*
 892          * We need to check if 'cur' is equal to 'match'.  As those
 893          * are from the same (+/-) side, we do not need to adjust for
 894          * indent changes. However these were found using fuzzy
 895          * matching so we do have to check if they are equal. Here we
 896          * just check the lengths. We delay calling memcmp() to check
 897          * the contents until later as if the length comparison for a
 898          * and c fails we can avoid the call all together.
 899          */
 900         if (al != bl)
 901                 return 1;
 902
 903         /* If 'l' and 'cur' are both blank then they match. */
 904         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 905                 return 0;
 906
 907         /*
 908          * The indent changes of the block are known and stored in pmb->wsd;
 909          * however we need to check if the indent changes of the current line
 910          * match those of the current block and that the text of 'l' and 'cur'
 911          * after the indentation match.
 912          */
 913         if (cur->es->s == DIFF_SYMBOL_PLUS)
 914                 delta = a_width - c_width;
 915         else
 916                 delta = c_width - a_width;
 917
 918         /*
 919          * If the previous lines of this block were all blank then set its
 920          * whitespace delta.
 921          */
 922         if (pmb->wsd == INDENT_BLANKLINE)
 923                 pmb->wsd = delta;
 924
 925         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 926                  !memcmp(a, b, al) && !
 927                  memcmp(a + a_off, c + c_off, al - a_off));
 928 }
 929
 930 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 931                            const void *entry,
 932                            const void *entry_or_key,
 933                            const void *keydata)
 934 {
 935         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 936         const struct moved_entry *a = entry;
 937         const struct moved_entry *b = entry_or_key;
 938         unsigned flags = diffopt->color_moved_ws_handling
 939                          & XDF_WHITESPACE_FLAGS;
 940
 941         if (diffopt->color_moved_ws_handling &
 942             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 943                 /*
 944                  * As there is not specific white space config given,
 945                  * we'd need to check for a new block, so ignore all
 946                  * white space. The setup of the white space
 947                  * configuration for the next block is done else where
 948                  */
 949                 flags |= XDF_IGNORE_WHITESPACE;
 950
 951         return !xdiff_compare_lines(a->es->line, a->es->len,
 952                                     b->es->line, b->es->len,
 953                                     flags);
 954 }
 955
 956 static struct moved_entry *prepare_entry(struct diff_options *o,
 957                                          int line_no)
 958 {
 959         struct moved_entry *ret = xmalloc(sizeof(*ret));
 960         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 961         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 962
 963         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 964         ret->es = l;
 965         ret->next_line = NULL;
 966
 967         return ret;
 968 }
 969
 970 static void add_lines_to_move_detection(struct diff_options *o,
 971                                         struct hashmap *add_lines,
 972                                         struct hashmap *del_lines)
 973 {
 974         struct moved_entry *prev_line = NULL;
 975
 976         int n;
 977         for (n = 0; n < o->emitted_symbols->nr; n++) {
 978                 struct hashmap *hm;
 979                 struct moved_entry *key;
 980
 981                 switch (o->emitted_symbols->buf[n].s) {
 982                 case DIFF_SYMBOL_PLUS:
 983                         hm = add_lines;
 984                         break;
 985                 case DIFF_SYMBOL_MINUS:
 986                         hm = del_lines;
 987                         break;
 988                 default:
 989                         prev_line = NULL;
 990                         continue;
 991                 }
 992
 993                 if (o->color_moved_ws_handling &
 994                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 995                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
 996                 key = prepare_entry(o, n);
 997                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 998                         prev_line->next_line = key;
 999
1000                 hashmap_add(hm, key);
1001                 prev_line = key;
1002         }
1003 }
1004
1005 static void pmb_advance_or_null(struct diff_options *o,
1006                                 struct moved_entry *match,
1007                                 struct hashmap *hm,
1008                                 struct moved_block *pmb,
1009                                 int pmb_nr)
1010 {
1011         int i;
1012         for (i = 0; i < pmb_nr; i++) {
1013                 struct moved_entry *prev = pmb[i].match;
1014                 struct moved_entry *cur = (prev && prev->next_line) ?
1015                                 prev->next_line : NULL;
1016                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1017                         pmb[i].match = cur;
1018                 } else {
1019                         pmb[i].match = NULL;
1020                 }
1021         }
1022 }
1023
1024 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1025                                             struct moved_entry *match,
1026                                             struct hashmap *hm,
1027                                             struct moved_block *pmb,
1028                                             int pmb_nr, int n)
1029 {
1030         int i;
1031         char *got_match = xcalloc(1, pmb_nr);
1032
1033         for (; match; match = hashmap_get_next(hm, match)) {
1034                 for (i = 0; i < pmb_nr; i++) {
1035                         struct moved_entry *prev = pmb[i].match;
1036                         struct moved_entry *cur = (prev && prev->next_line) ?
1037                                         prev->next_line : NULL;
1038                         if (!cur)
1039                                 continue;
1040                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1041                                 got_match[i] |= 1;
1042                 }
1043         }
1044
1045         for (i = 0; i < pmb_nr; i++) {
1046                 if (got_match[i]) {
1047                         /* Advance to the next line */
1048                         pmb[i].match = pmb[i].match->next_line;
1049                 } else {
1050                         moved_block_clear(&pmb[i]);
1051                 }
1052         }
1053
1054         free(got_match);
1055 }
1056
1057 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1058                                          int pmb_nr)
1059 {
1060         int lp, rp;
1061
1062         /* Shrink the set of potential block to the remaining running */
1063         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1064                 while (lp < pmb_nr && pmb[lp].match)
1065                         lp++;
1066                 /* lp points at the first NULL now */
1067
1068                 while (rp > -1 && !pmb[rp].match)
1069                         rp--;
1070                 /* rp points at the last non-NULL */
1071
1072                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1073                         pmb[lp] = pmb[rp];
1074                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1075                         rp--;
1076                         lp++;
1077                 }
1078         }
1079
1080         /* Remember the number of running sets */
1081         return rp + 1;
1082 }
1083
1084 /*
1085  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1086  *
1087  * Otherwise, if the last block has fewer alphanumeric characters than
1088  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1089  * that block.
1090  *
1091  * The last block consists of the (n - block_length)'th line up to but not
1092  * including the nth line.
1093  *
1094  * Returns 0 if the last block is empty or is unset by this function, non zero
1095  * otherwise.
1096  *
1097  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1098  * Think of a way to unify them.
1099  */
1100 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1101 {
1102         int i, alnum_count = 0;
1103         if (o->color_moved == COLOR_MOVED_PLAIN)
1104                 return block_length;
1105         for (i = 1; i < block_length + 1; i++) {
1106                 const char *c = o->emitted_symbols->buf[n - i].line;
1107                 for (; *c; c++) {
1108                         if (!isalnum(*c))
1109                                 continue;
1110                         alnum_count++;
1111                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1112                                 return 1;
1113                 }
1114         }
1115         for (i = 1; i < block_length + 1; i++)
1116                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1117         return 0;
1118 }
1119
1120 /* Find blocks of moved code, delegate actual coloring decision to helper */
1121 static void mark_color_as_moved(struct diff_options *o,
1122                                 struct hashmap *add_lines,
1123                                 struct hashmap *del_lines)
1124 {
1125         struct moved_block *pmb = NULL; /* potentially moved blocks */
1126         int pmb_nr = 0, pmb_alloc = 0;
1127         int n, flipped_block = 0, block_length = 0;
1128
1129
1130         for (n = 0; n < o->emitted_symbols->nr; n++) {
1131                 struct hashmap *hm = NULL;
1132                 struct moved_entry *key;
1133                 struct moved_entry *match = NULL;
1134                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1135                 enum diff_symbol last_symbol = 0;
1136
1137                 switch (l->s) {
1138                 case DIFF_SYMBOL_PLUS:
1139                         hm = del_lines;
1140                         key = prepare_entry(o, n);
1141                         match = hashmap_get(hm, key, NULL);
1142                         free(key);
1143                         break;
1144                 case DIFF_SYMBOL_MINUS:
1145                         hm = add_lines;
1146                         key = prepare_entry(o, n);
1147                         match = hashmap_get(hm, key, NULL);
1148                         free(key);
1149                         break;
1150                 default:
1151                         flipped_block = 0;
1152                 }
1153
1154                 if (!match) {
1155                         int i;
1156
1157                         adjust_last_block(o, n, block_length);
1158                         for(i = 0; i < pmb_nr; i++)
1159                                 moved_block_clear(&pmb[i]);
1160                         pmb_nr = 0;
1161                         block_length = 0;
1162                         flipped_block = 0;
1163                         last_symbol = l->s;
1164                         continue;
1165                 }
1166
1167                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1168                         last_symbol = l->s;
1169                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1170                         continue;
1171                 }
1172
1173                 if (o->color_moved_ws_handling &
1174                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1175                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1176                 else
1177                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1178
1179                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1180
1181                 if (pmb_nr == 0) {
1182                         /*
1183                          * The current line is the start of a new block.
1184                          * Setup the set of potential blocks.
1185                          */
1186                         for (; match; match = hashmap_get_next(hm, match)) {
1187                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1188                                 if (o->color_moved_ws_handling &
1189                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1190                                         if (compute_ws_delta(l, match->es,
1191                                                              &pmb[pmb_nr].wsd))
1192                                                 pmb[pmb_nr++].match = match;
1193                                 } else {
1194                                         pmb[pmb_nr].wsd = 0;
1195                                         pmb[pmb_nr++].match = match;
1196                                 }
1197                         }
1198
1199                         if (adjust_last_block(o, n, block_length) &&
1200                             pmb_nr && last_symbol != l->s)
1201                                 flipped_block = (flipped_block + 1) % 2;
1202                         else
1203                                 flipped_block = 0;
1204
1205                         block_length = 0;
1206                 }
1207
1208                 if (pmb_nr) {
1209                         block_length++;
1210                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1211                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1212                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1213                 }
1214                 last_symbol = l->s;
1215         }
1216         adjust_last_block(o, n, block_length);
1217
1218         for(n = 0; n < pmb_nr; n++)
1219                 moved_block_clear(&pmb[n]);
1220         free(pmb);
1221 }
1222
1223 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1224   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1225 static void dim_moved_lines(struct diff_options *o)
1226 {
1227         int n;
1228         for (n = 0; n < o->emitted_symbols->nr; n++) {
1229                 struct emitted_diff_symbol *prev = (n != 0) ?
1230                                 &o->emitted_symbols->buf[n - 1] : NULL;
1231                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1232                 struct emitted_diff_symbol *next =
1233                                 (n < o->emitted_symbols->nr - 1) ?
1234                                 &o->emitted_symbols->buf[n + 1] : NULL;
1235
1236                 /* Not a plus or minus line? */
1237                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1238                         continue;
1239
1240                 /* Not a moved line? */
1241                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1242                         continue;
1243
1244                 /*
1245                  * If prev or next are not a plus or minus line,
1246                  * pretend they don't exist
1247                  */
1248                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1249                             prev->s != DIFF_SYMBOL_MINUS)
1250                         prev = NULL;
1251                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1252                             next->s != DIFF_SYMBOL_MINUS)
1253                         next = NULL;
1254
1255                 /* Inside a block? */
1256                 if ((prev &&
1257                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1258                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1259                     (next &&
1260                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1261                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1262                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1263                         continue;
1264                 }
1265
1266                 /* Check if we are at an interesting bound: */
1267                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1268                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1269                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1270                         continue;
1271                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1272                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1273                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1274                         continue;
1275
1276                 /*
1277                  * The boundary to prev and next are not interesting,
1278                  * so this line is not interesting as a whole
1279                  */
1280                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1281         }
1282 }
1283
1284 static void emit_line_ws_markup(struct diff_options *o,
1285                                 const char *set_sign, const char *set,
1286                                 const char *reset,
1287                                 int sign_index, const char *line, int len,
1288                                 unsigned ws_rule, int blank_at_eof)
1289 {
1290         const char *ws = NULL;
1291         int sign = o->output_indicators[sign_index];
1292
1293         if (o->ws_error_highlight & ws_rule) {
1294                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1295                 if (!*ws)
1296                         ws = NULL;
1297         }
1298
1299         if (!ws && !set_sign)
1300                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1301         else if (!ws) {
1302                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1303         } else if (blank_at_eof)
1304                 /* Blank line at EOF - paint '+' as well */
1305                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1306         else {
1307                 /* Emit just the prefix, then the rest. */
1308                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1309                             sign, "", 0);
1310                 ws_check_emit(line, len, ws_rule,
1311                               o->file, set, reset, ws);
1312         }
1313 }
1314
1315 static void emit_diff_symbol_from_struct(struct diff_options *o,
1316                                          struct emitted_diff_symbol *eds)
1317 {
1318         static const char *nneof = " No newline at end of file\n";
1319         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1320         struct strbuf sb = STRBUF_INIT;
1321
1322         enum diff_symbol s = eds->s;
1323         const char *line = eds->line;
1324         int len = eds->len;
1325         unsigned flags = eds->flags;
1326
1327         switch (s) {
1328         case DIFF_SYMBOL_NO_LF_EOF:
1329                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1330                 reset = diff_get_color_opt(o, DIFF_RESET);
1331                 putc('\n', o->file);
1332                 emit_line_0(o, context, NULL, 0, reset, '\\',
1333                             nneof, strlen(nneof));
1334                 break;
1335         case DIFF_SYMBOL_SUBMODULE_HEADER:
1336         case DIFF_SYMBOL_SUBMODULE_ERROR:
1337         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1338         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1339         case DIFF_SYMBOL_SUMMARY:
1340         case DIFF_SYMBOL_STATS_LINE:
1341         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1342         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1343                 emit_line(o, "", "", line, len);
1344                 break;
1345         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1346         case DIFF_SYMBOL_CONTEXT_MARKER:
1347                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1348                 reset = diff_get_color_opt(o, DIFF_RESET);
1349                 emit_line(o, context, reset, line, len);
1350                 break;
1351         case DIFF_SYMBOL_SEPARATOR:
1352                 fprintf(o->file, "%s%c",
1353                         diff_line_prefix(o),
1354                         o->line_termination);
1355                 break;
1356         case DIFF_SYMBOL_CONTEXT:
1357                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1358                 reset = diff_get_color_opt(o, DIFF_RESET);
1359                 set_sign = NULL;
1360                 if (o->flags.dual_color_diffed_diffs) {
1361                         char c = !len ? 0 : line[0];
1362
1363                         if (c == '+')
1364                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1365                         else if (c == '@')
1366                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1367                         else if (c == '-')
1368                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1369                 }
1370                 emit_line_ws_markup(o, set_sign, set, reset,
1371                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1372                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1373                 break;
1374         case DIFF_SYMBOL_PLUS:
1375                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1376                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1377                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1378                 case DIFF_SYMBOL_MOVED_LINE |
1379                      DIFF_SYMBOL_MOVED_LINE_ALT |
1380                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1381                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1382                         break;
1383                 case DIFF_SYMBOL_MOVED_LINE |
1384                      DIFF_SYMBOL_MOVED_LINE_ALT:
1385                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1386                         break;
1387                 case DIFF_SYMBOL_MOVED_LINE |
1388                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1389                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1390                         break;
1391                 case DIFF_SYMBOL_MOVED_LINE:
1392                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1393                         break;
1394                 default:
1395                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1396                 }
1397                 reset = diff_get_color_opt(o, DIFF_RESET);
1398                 if (!o->flags.dual_color_diffed_diffs)
1399                         set_sign = NULL;
1400                 else {
1401                         char c = !len ? 0 : line[0];
1402
1403                         set_sign = set;
1404                         if (c == '-')
1405                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1406                         else if (c == '@')
1407                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1408                         else if (c == '+')
1409                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1410                         else
1411                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1412                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1413                 }
1414                 emit_line_ws_markup(o, set_sign, set, reset,
1415                                     OUTPUT_INDICATOR_NEW, line, len,
1416                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1417                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1418                 break;
1419         case DIFF_SYMBOL_MINUS:
1420                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1421                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1422                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1423                 case DIFF_SYMBOL_MOVED_LINE |
1424                      DIFF_SYMBOL_MOVED_LINE_ALT |
1425                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1426                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1427                         break;
1428                 case DIFF_SYMBOL_MOVED_LINE |
1429                      DIFF_SYMBOL_MOVED_LINE_ALT:
1430                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1431                         break;
1432                 case DIFF_SYMBOL_MOVED_LINE |
1433                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1434                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1435                         break;
1436                 case DIFF_SYMBOL_MOVED_LINE:
1437                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1438                         break;
1439                 default:
1440                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1441                 }
1442                 reset = diff_get_color_opt(o, DIFF_RESET);
1443                 if (!o->flags.dual_color_diffed_diffs)
1444                         set_sign = NULL;
1445                 else {
1446                         char c = !len ? 0 : line[0];
1447
1448                         set_sign = set;
1449                         if (c == '+')
1450                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1451                         else if (c == '@')
1452                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1453                         else if (c == '-')
1454                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1455                         else
1456                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1457                 }
1458                 emit_line_ws_markup(o, set_sign, set, reset,
1459                                     OUTPUT_INDICATOR_OLD, line, len,
1460                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1461                 break;
1462         case DIFF_SYMBOL_WORDS_PORCELAIN:
1463                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1464                 reset = diff_get_color_opt(o, DIFF_RESET);
1465                 emit_line(o, context, reset, line, len);
1466                 fputs("~\n", o->file);
1467                 break;
1468         case DIFF_SYMBOL_WORDS:
1469                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1470                 reset = diff_get_color_opt(o, DIFF_RESET);
1471                 /*
1472                  * Skip the prefix character, if any.  With
1473                  * diff_suppress_blank_empty, there may be
1474                  * none.
1475                  */
1476                 if (line[0] != '\n') {
1477                         line++;
1478                         len--;
1479                 }
1480                 emit_line(o, context, reset, line, len);
1481                 break;
1482         case DIFF_SYMBOL_FILEPAIR_PLUS:
1483                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1484                 reset = diff_get_color_opt(o, DIFF_RESET);
1485                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1486                         line, reset,
1487                         strchr(line, ' ') ? "\t" : "");
1488                 break;
1489         case DIFF_SYMBOL_FILEPAIR_MINUS:
1490                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1491                 reset = diff_get_color_opt(o, DIFF_RESET);
1492                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1493                         line, reset,
1494                         strchr(line, ' ') ? "\t" : "");
1495                 break;
1496         case DIFF_SYMBOL_BINARY_FILES:
1497         case DIFF_SYMBOL_HEADER:
1498                 fprintf(o->file, "%s", line);
1499                 break;
1500         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1501                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1502                 break;
1503         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1504                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1505                 break;
1506         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1507                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1508                 break;
1509         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1510                 fputs(diff_line_prefix(o), o->file);
1511                 fputc('\n', o->file);
1512                 break;
1513         case DIFF_SYMBOL_REWRITE_DIFF:
1514                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1515                 reset = diff_get_color_opt(o, DIFF_RESET);
1516                 emit_line(o, fraginfo, reset, line, len);
1517                 break;
1518         case DIFF_SYMBOL_SUBMODULE_ADD:
1519                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1520                 reset = diff_get_color_opt(o, DIFF_RESET);
1521                 emit_line(o, set, reset, line, len);
1522                 break;
1523         case DIFF_SYMBOL_SUBMODULE_DEL:
1524                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1525                 reset = diff_get_color_opt(o, DIFF_RESET);
1526                 emit_line(o, set, reset, line, len);
1527                 break;
1528         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1529                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1530                         diff_line_prefix(o), line);
1531                 break;
1532         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1533                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1534                         diff_line_prefix(o), line);
1535                 break;
1536         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1537                 emit_line(o, "", "", " 0 files changed\n",
1538                           strlen(" 0 files changed\n"));
1539                 break;
1540         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1541                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1542                 break;
1543         case DIFF_SYMBOL_WORD_DIFF:
1544                 fprintf(o->file, "%.*s", len, line);
1545                 break;
1546         case DIFF_SYMBOL_STAT_SEP:
1547                 fputs(o->stat_sep, o->file);
1548                 break;
1549         default:
1550                 BUG("unknown diff symbol");
1551         }
1552         strbuf_release(&sb);
1553 }
1554
1555 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1556                              const char *line, int len, unsigned flags)
1557 {
1558         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1559
1560         if (o->emitted_symbols)
1561                 append_emitted_diff_symbol(o, &e);
1562         else
1563                 emit_diff_symbol_from_struct(o, &e);
1564 }
1565
1566 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1567 {
1568         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1569 }
1570
1571 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1572 {
1573         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1574 }
1575
1576 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1577 {
1578         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1579                          path, strlen(path), 0);
1580 }
1581
1582 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1583 {
1584         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1585                          path, strlen(path), 0);
1586 }
1587
1588 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1589 {
1590         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1591                          header, strlen(header), 0);
1592 }
1593
1594 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1595 {
1596         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1597 }
1598
1599 void diff_emit_submodule_pipethrough(struct diff_options *o,
1600                                      const char *line, int len)
1601 {
1602         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1603 }
1604
1605 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1606 {
1607         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1608               ecbdata->blank_at_eof_in_preimage &&
1609               ecbdata->blank_at_eof_in_postimage &&
1610               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1611               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1612                 return 0;
1613         return ws_blank_line(line, len, ecbdata->ws_rule);
1614 }
1615
1616 static void emit_add_line(struct emit_callback *ecbdata,
1617                           const char *line, int len)
1618 {
1619         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1620         if (new_blank_line_at_eof(ecbdata, line, len))
1621                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1622
1623         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1624 }
1625
1626 static void emit_del_line(struct emit_callback *ecbdata,
1627                           const char *line, int len)
1628 {
1629         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1630         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1631 }
1632
1633 static void emit_context_line(struct emit_callback *ecbdata,
1634                               const char *line, int len)
1635 {
1636         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1637         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1638 }
1639
1640 static void emit_hunk_header(struct emit_callback *ecbdata,
1641                              const char *line, int len)
1642 {
1643         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1644         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1645         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1646         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1647         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1648         static const char atat[2] = { '@', '@' };
1649         const char *cp, *ep;
1650         struct strbuf msgbuf = STRBUF_INIT;
1651         int org_len = len;
1652         int i = 1;
1653
1654         /*
1655          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1656          * it always is at least 10 bytes long.
1657          */
1658         if (len < 10 ||
1659             memcmp(line, atat, 2) ||
1660             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1661                 emit_diff_symbol(ecbdata->opt,
1662                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1663                 return;
1664         }
1665         ep += 2; /* skip over @@ */
1666
1667         /* The hunk header in fraginfo color */
1668         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1669                 strbuf_addstr(&msgbuf, reverse);
1670         strbuf_addstr(&msgbuf, frag);
1671         strbuf_add(&msgbuf, line, ep - line);
1672         strbuf_addstr(&msgbuf, reset);
1673
1674         /*
1675          * trailing "\r\n"
1676          */
1677         for ( ; i < 3; i++)
1678                 if (line[len - i] == '\r' || line[len - i] == '\n')
1679                         len--;
1680
1681         /* blank before the func header */
1682         for (cp = ep; ep - line < len; ep++)
1683                 if (*ep != ' ' && *ep != '\t')
1684                         break;
1685         if (ep != cp) {
1686                 strbuf_addstr(&msgbuf, context);
1687                 strbuf_add(&msgbuf, cp, ep - cp);
1688                 strbuf_addstr(&msgbuf, reset);
1689         }
1690
1691         if (ep < line + len) {
1692                 strbuf_addstr(&msgbuf, func);
1693                 strbuf_add(&msgbuf, ep, line + len - ep);
1694                 strbuf_addstr(&msgbuf, reset);
1695         }
1696
1697         strbuf_add(&msgbuf, line + len, org_len - len);
1698         strbuf_complete_line(&msgbuf);
1699         emit_diff_symbol(ecbdata->opt,
1700                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1701         strbuf_release(&msgbuf);
1702 }
1703
1704 static struct diff_tempfile *claim_diff_tempfile(void)
1705 {
1706         int i;
1707         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1708                 if (!diff_temp[i].name)
1709                         return diff_temp + i;
1710         BUG("diff is failing to clean up its tempfiles");
1711 }
1712
1713 static void remove_tempfile(void)
1714 {
1715         int i;
1716         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1717                 if (is_tempfile_active(diff_temp[i].tempfile))
1718                         delete_tempfile(&diff_temp[i].tempfile);
1719                 diff_temp[i].name = NULL;
1720         }
1721 }
1722
1723 static void add_line_count(struct strbuf *out, int count)
1724 {
1725         switch (count) {
1726         case 0:
1727                 strbuf_addstr(out, "0,0");
1728                 break;
1729         case 1:
1730                 strbuf_addstr(out, "1");
1731                 break;
1732         default:
1733                 strbuf_addf(out, "1,%d", count);
1734                 break;
1735         }
1736 }
1737
1738 static void emit_rewrite_lines(struct emit_callback *ecb,
1739                                int prefix, const char *data, int size)
1740 {
1741         const char *endp = NULL;
1742
1743         while (0 < size) {
1744                 int len;
1745
1746                 endp = memchr(data, '\n', size);
1747                 len = endp ? (endp - data + 1) : size;
1748                 if (prefix != '+') {
1749                         ecb->lno_in_preimage++;
1750                         emit_del_line(ecb, data, len);
1751                 } else {
1752                         ecb->lno_in_postimage++;
1753                         emit_add_line(ecb, data, len);
1754                 }
1755                 size -= len;
1756                 data += len;
1757         }
1758         if (!endp)
1759                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1760 }
1761
1762 static void emit_rewrite_diff(const char *name_a,
1763                               const char *name_b,
1764                               struct diff_filespec *one,
1765                               struct diff_filespec *two,
1766                               struct userdiff_driver *textconv_one,
1767                               struct userdiff_driver *textconv_two,
1768                               struct diff_options *o)
1769 {
1770         int lc_a, lc_b;
1771         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1772         const char *a_prefix, *b_prefix;
1773         char *data_one, *data_two;
1774         size_t size_one, size_two;
1775         struct emit_callback ecbdata;
1776         struct strbuf out = STRBUF_INIT;
1777
1778         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1779                 a_prefix = o->b_prefix;
1780                 b_prefix = o->a_prefix;
1781         } else {
1782                 a_prefix = o->a_prefix;
1783                 b_prefix = o->b_prefix;
1784         }
1785
1786         name_a += (*name_a == '/');
1787         name_b += (*name_b == '/');
1788
1789         strbuf_reset(&a_name);
1790         strbuf_reset(&b_name);
1791         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1792         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1793
1794         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1795         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1796
1797         memset(&ecbdata, 0, sizeof(ecbdata));
1798         ecbdata.color_diff = want_color(o->use_color);
1799         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1800         ecbdata.opt = o;
1801         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1802                 mmfile_t mf1, mf2;
1803                 mf1.ptr = (char *)data_one;
1804                 mf2.ptr = (char *)data_two;
1805                 mf1.size = size_one;
1806                 mf2.size = size_two;
1807                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1808         }
1809         ecbdata.lno_in_preimage = 1;
1810         ecbdata.lno_in_postimage = 1;
1811
1812         lc_a = count_lines(data_one, size_one);
1813         lc_b = count_lines(data_two, size_two);
1814
1815         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1816                          a_name.buf, a_name.len, 0);
1817         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1818                          b_name.buf, b_name.len, 0);
1819
1820         strbuf_addstr(&out, "@@ -");
1821         if (!o->irreversible_delete)
1822                 add_line_count(&out, lc_a);
1823         else
1824                 strbuf_addstr(&out, "?,?");
1825         strbuf_addstr(&out, " +");
1826         add_line_count(&out, lc_b);
1827         strbuf_addstr(&out, " @@\n");
1828         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1829         strbuf_release(&out);
1830
1831         if (lc_a && !o->irreversible_delete)
1832                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1833         if (lc_b)
1834                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1835         if (textconv_one)
1836                 free((char *)data_one);
1837         if (textconv_two)
1838                 free((char *)data_two);
1839 }
1840
1841 struct diff_words_buffer {
1842         mmfile_t text;
1843         unsigned long alloc;
1844         struct diff_words_orig {
1845                 const char *begin, *end;
1846         } *orig;
1847         int orig_nr, orig_alloc;
1848 };
1849
1850 static void diff_words_append(char *line, unsigned long len,
1851                 struct diff_words_buffer *buffer)
1852 {
1853         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1854         line++;
1855         len--;
1856         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1857         buffer->text.size += len;
1858         buffer->text.ptr[buffer->text.size] = '\0';
1859 }
1860
1861 struct diff_words_style_elem {
1862         const char *prefix;
1863         const char *suffix;
1864         const char *color; /* NULL; filled in by the setup code if
1865                             * color is enabled */
1866 };
1867
1868 struct diff_words_style {
1869         enum diff_words_type type;
1870         struct diff_words_style_elem new_word, old_word, ctx;
1871         const char *newline;
1872 };
1873
1874 static struct diff_words_style diff_words_styles[] = {
1875         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1876         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1877         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1878 };
1879
1880 struct diff_words_data {
1881         struct diff_words_buffer minus, plus;
1882         const char *current_plus;
1883         int last_minus;
1884         struct diff_options *opt;
1885         regex_t *word_regex;
1886         enum diff_words_type type;
1887         struct diff_words_style *style;
1888 };
1889
1890 static int fn_out_diff_words_write_helper(struct diff_options *o,
1891                                           struct diff_words_style_elem *st_el,
1892                                           const char *newline,
1893                                           size_t count, const char *buf)
1894 {
1895         int print = 0;
1896         struct strbuf sb = STRBUF_INIT;
1897
1898         while (count) {
1899                 char *p = memchr(buf, '\n', count);
1900                 if (print)
1901                         strbuf_addstr(&sb, diff_line_prefix(o));
1902
1903                 if (p != buf) {
1904                         const char *reset = st_el->color && *st_el->color ?
1905                                             GIT_COLOR_RESET : NULL;
1906                         if (st_el->color && *st_el->color)
1907                                 strbuf_addstr(&sb, st_el->color);
1908                         strbuf_addstr(&sb, st_el->prefix);
1909                         strbuf_add(&sb, buf, p ? p - buf : count);
1910                         strbuf_addstr(&sb, st_el->suffix);
1911                         if (reset)
1912                                 strbuf_addstr(&sb, reset);
1913                 }
1914                 if (!p)
1915                         goto out;
1916
1917                 strbuf_addstr(&sb, newline);
1918                 count -= p + 1 - buf;
1919                 buf = p + 1;
1920                 print = 1;
1921                 if (count) {
1922                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1923                                          sb.buf, sb.len, 0);
1924                         strbuf_reset(&sb);
1925                 }
1926         }
1927
1928 out:
1929         if (sb.len)
1930                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1931                                  sb.buf, sb.len, 0);
1932         strbuf_release(&sb);
1933         return 0;
1934 }
1935
1936 /*
1937  * '--color-words' algorithm can be described as:
1938  *
1939  *   1. collect the minus/plus lines of a diff hunk, divided into
1940  *      minus-lines and plus-lines;
1941  *
1942  *   2. break both minus-lines and plus-lines into words and
1943  *      place them into two mmfile_t with one word for each line;
1944  *
1945  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1946  *
1947  * And for the common parts of the both file, we output the plus side text.
1948  * diff_words->current_plus is used to trace the current position of the plus file
1949  * which printed. diff_words->last_minus is used to trace the last minus word
1950  * printed.
1951  *
1952  * For '--graph' to work with '--color-words', we need to output the graph prefix
1953  * on each line of color words output. Generally, there are two conditions on
1954  * which we should output the prefix.
1955  *
1956  *   1. diff_words->last_minus == 0 &&
1957  *      diff_words->current_plus == diff_words->plus.text.ptr
1958  *
1959  *      that is: the plus text must start as a new line, and if there is no minus
1960  *      word printed, a graph prefix must be printed.
1961  *
1962  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1963  *      *(diff_words->current_plus - 1) == '\n'
1964  *
1965  *      that is: a graph prefix must be printed following a '\n'
1966  */
1967 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1968 {
1969         if ((diff_words->last_minus == 0 &&
1970                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1971                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1972                 *(diff_words->current_plus - 1) == '\n')) {
1973                 return 1;
1974         } else {
1975                 return 0;
1976         }
1977 }
1978
1979 static void fn_out_diff_words_aux(void *priv,
1980                                   long minus_first, long minus_len,
1981                                   long plus_first, long plus_len,
1982                                   const char *func, long funclen)
1983 {
1984         struct diff_words_data *diff_words = priv;
1985         struct diff_words_style *style = diff_words->style;
1986         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1987         struct diff_options *opt = diff_words->opt;
1988         const char *line_prefix;
1989
1990         assert(opt);
1991         line_prefix = diff_line_prefix(opt);
1992
1993         /* POSIX requires that first be decremented by one if len == 0... */
1994         if (minus_len) {
1995                 minus_begin = diff_words->minus.orig[minus_first].begin;
1996                 minus_end =
1997                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1998         } else
1999                 minus_begin = minus_end =
2000                         diff_words->minus.orig[minus_first].end;
2001
2002         if (plus_len) {
2003                 plus_begin = diff_words->plus.orig[plus_first].begin;
2004                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2005         } else
2006                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2007
2008         if (color_words_output_graph_prefix(diff_words)) {
2009                 fputs(line_prefix, diff_words->opt->file);
2010         }
2011         if (diff_words->current_plus != plus_begin) {
2012                 fn_out_diff_words_write_helper(diff_words->opt,
2013                                 &style->ctx, style->newline,
2014                                 plus_begin - diff_words->current_plus,
2015                                 diff_words->current_plus);
2016         }
2017         if (minus_begin != minus_end) {
2018                 fn_out_diff_words_write_helper(diff_words->opt,
2019                                 &style->old_word, style->newline,
2020                                 minus_end - minus_begin, minus_begin);
2021         }
2022         if (plus_begin != plus_end) {
2023                 fn_out_diff_words_write_helper(diff_words->opt,
2024                                 &style->new_word, style->newline,
2025                                 plus_end - plus_begin, plus_begin);
2026         }
2027
2028         diff_words->current_plus = plus_end;
2029         diff_words->last_minus = minus_first;
2030 }
2031
2032 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2033 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2034                 int *begin, int *end)
2035 {
2036         if (word_regex && *begin < buffer->size) {
2037                 regmatch_t match[1];
2038                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2039                                  buffer->size - *begin, 1, match, 0)) {
2040                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2041                                         '\n', match[0].rm_eo - match[0].rm_so);
2042                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2043                         *begin += match[0].rm_so;
2044                         return *begin >= *end;
2045                 }
2046                 return -1;
2047         }
2048
2049         /* find the next word */
2050         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2051                 (*begin)++;
2052         if (*begin >= buffer->size)
2053                 return -1;
2054
2055         /* find the end of the word */
2056         *end = *begin + 1;
2057         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2058                 (*end)++;
2059
2060         return 0;
2061 }
2062
2063 /*
2064  * This function splits the words in buffer->text, stores the list with
2065  * newline separator into out, and saves the offsets of the original words
2066  * in buffer->orig.
2067  */
2068 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2069                 regex_t *word_regex)
2070 {
2071         int i, j;
2072         long alloc = 0;
2073
2074         out->size = 0;
2075         out->ptr = NULL;
2076
2077         /* fake an empty "0th" word */
2078         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2079         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2080         buffer->orig_nr = 1;
2081
2082         for (i = 0; i < buffer->text.size; i++) {
2083                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2084                         return;
2085
2086                 /* store original boundaries */
2087                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2088                                 buffer->orig_alloc);
2089                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2090                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2091                 buffer->orig_nr++;
2092
2093                 /* store one word */
2094                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2095                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2096                 out->ptr[out->size + j - i] = '\n';
2097                 out->size += j - i + 1;
2098
2099                 i = j - 1;
2100         }
2101 }
2102
2103 /* this executes the word diff on the accumulated buffers */
2104 static void diff_words_show(struct diff_words_data *diff_words)
2105 {
2106         xpparam_t xpp;
2107         xdemitconf_t xecfg;
2108         mmfile_t minus, plus;
2109         struct diff_words_style *style = diff_words->style;
2110
2111         struct diff_options *opt = diff_words->opt;
2112         const char *line_prefix;
2113
2114         assert(opt);
2115         line_prefix = diff_line_prefix(opt);
2116
2117         /* special case: only removal */
2118         if (!diff_words->plus.text.size) {
2119                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2120                                  line_prefix, strlen(line_prefix), 0);
2121                 fn_out_diff_words_write_helper(diff_words->opt,
2122                         &style->old_word, style->newline,
2123                         diff_words->minus.text.size,
2124                         diff_words->minus.text.ptr);
2125                 diff_words->minus.text.size = 0;
2126                 return;
2127         }
2128
2129         diff_words->current_plus = diff_words->plus.text.ptr;
2130         diff_words->last_minus = 0;
2131
2132         memset(&xpp, 0, sizeof(xpp));
2133         memset(&xecfg, 0, sizeof(xecfg));
2134         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2135         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2136         xpp.flags = 0;
2137         /* as only the hunk header will be parsed, we need a 0-context */
2138         xecfg.ctxlen = 0;
2139         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2140                           diff_words, &xpp, &xecfg))
2141                 die("unable to generate word diff");
2142         free(minus.ptr);
2143         free(plus.ptr);
2144         if (diff_words->current_plus != diff_words->plus.text.ptr +
2145                         diff_words->plus.text.size) {
2146                 if (color_words_output_graph_prefix(diff_words))
2147                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2148                                          line_prefix, strlen(line_prefix), 0);
2149                 fn_out_diff_words_write_helper(diff_words->opt,
2150                         &style->ctx, style->newline,
2151                         diff_words->plus.text.ptr + diff_words->plus.text.size
2152                         - diff_words->current_plus, diff_words->current_plus);
2153         }
2154         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2155 }
2156
2157 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2158 static void diff_words_flush(struct emit_callback *ecbdata)
2159 {
2160         struct diff_options *wo = ecbdata->diff_words->opt;
2161
2162         if (ecbdata->diff_words->minus.text.size ||
2163             ecbdata->diff_words->plus.text.size)
2164                 diff_words_show(ecbdata->diff_words);
2165
2166         if (wo->emitted_symbols) {
2167                 struct diff_options *o = ecbdata->opt;
2168                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2169                 int i;
2170
2171                 /*
2172                  * NEEDSWORK:
2173                  * Instead of appending each, concat all words to a line?
2174                  */
2175                 for (i = 0; i < wol->nr; i++)
2176                         append_emitted_diff_symbol(o, &wol->buf[i]);
2177
2178                 for (i = 0; i < wol->nr; i++)
2179                         free((void *)wol->buf[i].line);
2180
2181                 wol->nr = 0;
2182         }
2183 }
2184
2185 static void diff_filespec_load_driver(struct diff_filespec *one,
2186                                       struct index_state *istate)
2187 {
2188         /* Use already-loaded driver */
2189         if (one->driver)
2190                 return;
2191
2192         if (S_ISREG(one->mode))
2193                 one->driver = userdiff_find_by_path(istate, one->path);
2194
2195         /* Fallback to default settings */
2196         if (!one->driver)
2197                 one->driver = userdiff_find_by_name("default");
2198 }
2199
2200 static const char *userdiff_word_regex(struct diff_filespec *one,
2201                                        struct index_state *istate)
2202 {
2203         diff_filespec_load_driver(one, istate);
2204         return one->driver->word_regex;
2205 }
2206
2207 static void init_diff_words_data(struct emit_callback *ecbdata,
2208                                  struct diff_options *orig_opts,
2209                                  struct diff_filespec *one,
2210                                  struct diff_filespec *two)
2211 {
2212         int i;
2213         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2214         memcpy(o, orig_opts, sizeof(struct diff_options));
2215
2216         ecbdata->diff_words =
2217                 xcalloc(1, sizeof(struct diff_words_data));
2218         ecbdata->diff_words->type = o->word_diff;
2219         ecbdata->diff_words->opt = o;
2220
2221         if (orig_opts->emitted_symbols)
2222                 o->emitted_symbols =
2223                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2224
2225         if (!o->word_regex)
2226                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2227         if (!o->word_regex)
2228                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2229         if (!o->word_regex)
2230                 o->word_regex = diff_word_regex_cfg;
2231         if (o->word_regex) {
2232                 ecbdata->diff_words->word_regex = (regex_t *)
2233                         xmalloc(sizeof(regex_t));
2234                 if (regcomp(ecbdata->diff_words->word_regex,
2235                             o->word_regex,
2236                             REG_EXTENDED | REG_NEWLINE))
2237                         die("invalid regular expression: %s",
2238                             o->word_regex);
2239         }
2240         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2241                 if (o->word_diff == diff_words_styles[i].type) {
2242                         ecbdata->diff_words->style =
2243                                 &diff_words_styles[i];
2244                         break;
2245                 }
2246         }
2247         if (want_color(o->use_color)) {
2248                 struct diff_words_style *st = ecbdata->diff_words->style;
2249                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2250                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2251                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2252         }
2253 }
2254
2255 static void free_diff_words_data(struct emit_callback *ecbdata)
2256 {
2257         if (ecbdata->diff_words) {
2258                 diff_words_flush(ecbdata);
2259                 free (ecbdata->diff_words->opt->emitted_symbols);
2260                 free (ecbdata->diff_words->opt);
2261                 free (ecbdata->diff_words->minus.text.ptr);
2262                 free (ecbdata->diff_words->minus.orig);
2263                 free (ecbdata->diff_words->plus.text.ptr);
2264                 free (ecbdata->diff_words->plus.orig);
2265                 if (ecbdata->diff_words->word_regex) {
2266                         regfree(ecbdata->diff_words->word_regex);
2267                         free(ecbdata->diff_words->word_regex);
2268                 }
2269                 FREE_AND_NULL(ecbdata->diff_words);
2270         }
2271 }
2272
2273 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2274 {
2275         if (want_color(diff_use_color))
2276                 return diff_colors[ix];
2277         return "";
2278 }
2279
2280 const char *diff_line_prefix(struct diff_options *opt)
2281 {
2282         struct strbuf *msgbuf;
2283         if (!opt->output_prefix)
2284                 return "";
2285
2286         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2287         return msgbuf->buf;
2288 }
2289
2290 static unsigned long sane_truncate_line(char *line, unsigned long len)
2291 {
2292         const char *cp;
2293         unsigned long allot;
2294         size_t l = len;
2295
2296         cp = line;
2297         allot = l;
2298         while (0 < l) {
2299                 (void) utf8_width(&cp, &l);
2300                 if (!cp)
2301                         break; /* truncated in the middle? */
2302         }
2303         return allot - l;
2304 }
2305
2306 static void find_lno(const char *line, struct emit_callback *ecbdata)
2307 {
2308         const char *p;
2309         ecbdata->lno_in_preimage = 0;
2310         ecbdata->lno_in_postimage = 0;
2311         p = strchr(line, '-');
2312         if (!p)
2313                 return; /* cannot happen */
2314         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2315         p = strchr(p, '+');
2316         if (!p)
2317                 return; /* cannot happen */
2318         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2319 }
2320
2321 static void fn_out_consume(void *priv, char *line, unsigned long len)
2322 {
2323         struct emit_callback *ecbdata = priv;
2324         struct diff_options *o = ecbdata->opt;
2325
2326         o->found_changes = 1;
2327
2328         if (ecbdata->header) {
2329                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2330                                  ecbdata->header->buf, ecbdata->header->len, 0);
2331                 strbuf_reset(ecbdata->header);
2332                 ecbdata->header = NULL;
2333         }
2334
2335         if (ecbdata->label_path[0]) {
2336                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2337                                  ecbdata->label_path[0],
2338                                  strlen(ecbdata->label_path[0]), 0);
2339                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2340                                  ecbdata->label_path[1],
2341                                  strlen(ecbdata->label_path[1]), 0);
2342                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2343         }
2344
2345         if (diff_suppress_blank_empty
2346             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2347                 line[0] = '\n';
2348                 len = 1;
2349         }
2350
2351         if (line[0] == '@') {
2352                 if (ecbdata->diff_words)
2353                         diff_words_flush(ecbdata);
2354                 len = sane_truncate_line(line, len);
2355                 find_lno(line, ecbdata);
2356                 emit_hunk_header(ecbdata, line, len);
2357                 return;
2358         }
2359
2360         if (ecbdata->diff_words) {
2361                 enum diff_symbol s =
2362                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2363                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2364                 if (line[0] == '-') {
2365                         diff_words_append(line, len,
2366                                           &ecbdata->diff_words->minus);
2367                         return;
2368                 } else if (line[0] == '+') {
2369                         diff_words_append(line, len,
2370                                           &ecbdata->diff_words->plus);
2371                         return;
2372                 } else if (starts_with(line, "\\ ")) {
2373                         /*
2374                          * Eat the "no newline at eof" marker as if we
2375                          * saw a "+" or "-" line with nothing on it,
2376                          * and return without diff_words_flush() to
2377                          * defer processing. If this is the end of
2378                          * preimage, more "+" lines may come after it.
2379                          */
2380                         return;
2381                 }
2382                 diff_words_flush(ecbdata);
2383                 emit_diff_symbol(o, s, line, len, 0);
2384                 return;
2385         }
2386
2387         switch (line[0]) {
2388         case '+':
2389                 ecbdata->lno_in_postimage++;
2390                 emit_add_line(ecbdata, line + 1, len - 1);
2391                 break;
2392         case '-':
2393                 ecbdata->lno_in_preimage++;
2394                 emit_del_line(ecbdata, line + 1, len - 1);
2395                 break;
2396         case ' ':
2397                 ecbdata->lno_in_postimage++;
2398                 ecbdata->lno_in_preimage++;
2399                 emit_context_line(ecbdata, line + 1, len - 1);
2400                 break;
2401         default:
2402                 /* incomplete line at the end */
2403                 ecbdata->lno_in_preimage++;
2404                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2405                                  line, len, 0);
2406                 break;
2407         }
2408 }
2409
2410 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2411 {
2412         const char *old_name = a;
2413         const char *new_name = b;
2414         int pfx_length, sfx_length;
2415         int pfx_adjust_for_slash;
2416         int len_a = strlen(a);
2417         int len_b = strlen(b);
2418         int a_midlen, b_midlen;
2419         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2420         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2421
2422         if (qlen_a || qlen_b) {
2423                 quote_c_style(a, name, NULL, 0);
2424                 strbuf_addstr(name, " => ");
2425                 quote_c_style(b, name, NULL, 0);
2426                 return;
2427         }
2428
2429         /* Find common prefix */
2430         pfx_length = 0;
2431         while (*old_name && *new_name && *old_name == *new_name) {
2432                 if (*old_name == '/')
2433                         pfx_length = old_name - a + 1;
2434                 old_name++;
2435                 new_name++;
2436         }
2437
2438         /* Find common suffix */
2439         old_name = a + len_a;
2440         new_name = b + len_b;
2441         sfx_length = 0;
2442         /*
2443          * If there is a common prefix, it must end in a slash.  In
2444          * that case we let this loop run 1 into the prefix to see the
2445          * same slash.
2446          *
2447          * If there is no common prefix, we cannot do this as it would
2448          * underrun the input strings.
2449          */
2450         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2451         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2452                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2453                *old_name == *new_name) {
2454                 if (*old_name == '/')
2455                         sfx_length = len_a - (old_name - a);
2456                 old_name--;
2457                 new_name--;
2458         }
2459
2460         /*
2461          * pfx{mid-a => mid-b}sfx
2462          * {pfx-a => pfx-b}sfx
2463          * pfx{sfx-a => sfx-b}
2464          * name-a => name-b
2465          */
2466         a_midlen = len_a - pfx_length - sfx_length;
2467         b_midlen = len_b - pfx_length - sfx_length;
2468         if (a_midlen < 0)
2469                 a_midlen = 0;
2470         if (b_midlen < 0)
2471                 b_midlen = 0;
2472
2473         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2474         if (pfx_length + sfx_length) {
2475                 strbuf_add(name, a, pfx_length);
2476                 strbuf_addch(name, '{');
2477         }
2478         strbuf_add(name, a + pfx_length, a_midlen);
2479         strbuf_addstr(name, " => ");
2480         strbuf_add(name, b + pfx_length, b_midlen);
2481         if (pfx_length + sfx_length) {
2482                 strbuf_addch(name, '}');
2483                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2484         }
2485 }
2486
2487 struct diffstat_t {
2488         int nr;
2489         int alloc;
2490         struct diffstat_file {
2491                 char *from_name;
2492                 char *name;
2493                 char *print_name;
2494                 const char *comments;
2495                 unsigned is_unmerged:1;
2496                 unsigned is_binary:1;
2497                 unsigned is_renamed:1;
2498                 unsigned is_interesting:1;
2499                 uintmax_t added, deleted;
2500         } **files;
2501 };
2502
2503 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2504                                           const char *name_a,
2505                                           const char *name_b)
2506 {
2507         struct diffstat_file *x;
2508         x = xcalloc(1, sizeof(*x));
2509         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2510         diffstat->files[diffstat->nr++] = x;
2511         if (name_b) {
2512                 x->from_name = xstrdup(name_a);
2513                 x->name = xstrdup(name_b);
2514                 x->is_renamed = 1;
2515         }
2516         else {
2517                 x->from_name = NULL;
2518                 x->name = xstrdup(name_a);
2519         }
2520         return x;
2521 }
2522
2523 static void diffstat_consume(void *priv, char *line, unsigned long len)
2524 {
2525         struct diffstat_t *diffstat = priv;
2526         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2527
2528         if (line[0] == '+')
2529                 x->added++;
2530         else if (line[0] == '-')
2531                 x->deleted++;
2532 }
2533
2534 const char mime_boundary_leader[] = "------------";
2535
2536 static int scale_linear(int it, int width, int max_change)
2537 {
2538         if (!it)
2539                 return 0;
2540         /*
2541          * make sure that at least one '-' or '+' is printed if
2542          * there is any change to this path. The easiest way is to
2543          * scale linearly as if the alloted width is one column shorter
2544          * than it is, and then add 1 to the result.
2545          */
2546         return 1 + (it * (width - 1) / max_change);
2547 }
2548
2549 static void show_graph(struct strbuf *out, char ch, int cnt,
2550                        const char *set, const char *reset)
2551 {
2552         if (cnt <= 0)
2553                 return;
2554         strbuf_addstr(out, set);
2555         strbuf_addchars(out, ch, cnt);
2556         strbuf_addstr(out, reset);
2557 }
2558
2559 static void fill_print_name(struct diffstat_file *file)
2560 {
2561         struct strbuf pname = STRBUF_INIT;
2562
2563         if (file->print_name)
2564                 return;
2565
2566         if (file->is_renamed)
2567                 pprint_rename(&pname, file->from_name, file->name);
2568         else
2569                 quote_c_style(file->name, &pname, NULL, 0);
2570
2571         if (file->comments)
2572                 strbuf_addf(&pname, " (%s)", file->comments);
2573
2574         file->print_name = strbuf_detach(&pname, NULL);
2575 }
2576
2577 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2578                 int files, int insertions, int deletions)
2579 {
2580         struct strbuf sb = STRBUF_INIT;
2581
2582         if (!files) {
2583                 assert(insertions == 0 && deletions == 0);
2584                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2585                                  NULL, 0, 0);
2586                 return;
2587         }
2588
2589         strbuf_addf(&sb,
2590                     (files == 1) ? " %d file changed" : " %d files changed",
2591                     files);
2592
2593         /*
2594          * For binary diff, the caller may want to print "x files
2595          * changed" with insertions == 0 && deletions == 0.
2596          *
2597          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2598          * is probably less confusing (i.e skip over "2 files changed
2599          * but nothing about added/removed lines? Is this a bug in Git?").
2600          */
2601         if (insertions || deletions == 0) {
2602                 strbuf_addf(&sb,
2603                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2604                             insertions);
2605         }
2606
2607         if (deletions || insertions == 0) {
2608                 strbuf_addf(&sb,
2609                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2610                             deletions);
2611         }
2612         strbuf_addch(&sb, '\n');
2613         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2614                          sb.buf, sb.len, 0);
2615         strbuf_release(&sb);
2616 }
2617
2618 void print_stat_summary(FILE *fp, int files,
2619                         int insertions, int deletions)
2620 {
2621         struct diff_options o;
2622         memset(&o, 0, sizeof(o));
2623         o.file = fp;
2624
2625         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2626 }
2627
2628 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2629 {
2630         int i, len, add, del, adds = 0, dels = 0;
2631         uintmax_t max_change = 0, max_len = 0;
2632         int total_files = data->nr, count;
2633         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2634         const char *reset, *add_c, *del_c;
2635         int extra_shown = 0;
2636         const char *line_prefix = diff_line_prefix(options);
2637         struct strbuf out = STRBUF_INIT;
2638
2639         if (data->nr == 0)
2640                 return;
2641
2642         count = options->stat_count ? options->stat_count : data->nr;
2643
2644         reset = diff_get_color_opt(options, DIFF_RESET);
2645         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2646         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2647
2648         /*
2649          * Find the longest filename and max number of changes
2650          */
2651         for (i = 0; (i < count) && (i < data->nr); i++) {
2652                 struct diffstat_file *file = data->files[i];
2653                 uintmax_t change = file->added + file->deleted;
2654
2655                 if (!file->is_interesting && (change == 0)) {
2656                         count++; /* not shown == room for one more */
2657                         continue;
2658                 }
2659                 fill_print_name(file);
2660                 len = strlen(file->print_name);
2661                 if (max_len < len)
2662                         max_len = len;
2663
2664                 if (file->is_unmerged) {
2665                         /* "Unmerged" is 8 characters */
2666                         bin_width = bin_width < 8 ? 8 : bin_width;
2667                         continue;
2668                 }
2669                 if (file->is_binary) {
2670                         /* "Bin XXX -> YYY bytes" */
2671                         int w = 14 + decimal_width(file->added)
2672                                 + decimal_width(file->deleted);
2673                         bin_width = bin_width < w ? w : bin_width;
2674                         /* Display change counts aligned with "Bin" */
2675                         number_width = 3;
2676                         continue;
2677                 }
2678
2679                 if (max_change < change)
2680                         max_change = change;
2681         }
2682         count = i; /* where we can stop scanning in data->files[] */
2683
2684         /*
2685          * We have width = stat_width or term_columns() columns total.
2686          * We want a maximum of min(max_len, stat_name_width) for the name part.
2687          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2688          * We also need 1 for " " and 4 + decimal_width(max_change)
2689          * for " | NNNN " and one the empty column at the end, altogether
2690          * 6 + decimal_width(max_change).
2691          *
2692          * If there's not enough space, we will use the smaller of
2693          * stat_name_width (if set) and 5/8*width for the filename,
2694          * and the rest for constant elements + graph part, but no more
2695          * than stat_graph_width for the graph part.
2696          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2697          * for the standard terminal size).
2698          *
2699          * In other words: stat_width limits the maximum width, and
2700          * stat_name_width fixes the maximum width of the filename,
2701          * and is also used to divide available columns if there
2702          * aren't enough.
2703          *
2704          * Binary files are displayed with "Bin XXX -> YYY bytes"
2705          * instead of the change count and graph. This part is treated
2706          * similarly to the graph part, except that it is not
2707          * "scaled". If total width is too small to accommodate the
2708          * guaranteed minimum width of the filename part and the
2709          * separators and this message, this message will "overflow"
2710          * making the line longer than the maximum width.
2711          */
2712
2713         if (options->stat_width == -1)
2714                 width = term_columns() - strlen(line_prefix);
2715         else
2716                 width = options->stat_width ? options->stat_width : 80;
2717         number_width = decimal_width(max_change) > number_width ?
2718                 decimal_width(max_change) : number_width;
2719
2720         if (options->stat_graph_width == -1)
2721                 options->stat_graph_width = diff_stat_graph_width;
2722
2723         /*
2724          * Guarantee 3/8*16==6 for the graph part
2725          * and 5/8*16==10 for the filename part
2726          */
2727         if (width < 16 + 6 + number_width)
2728                 width = 16 + 6 + number_width;
2729
2730         /*
2731          * First assign sizes that are wanted, ignoring available width.
2732          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2733          * starting from "XXX" should fit in graph_width.
2734          */
2735         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2736         if (options->stat_graph_width &&
2737             options->stat_graph_width < graph_width)
2738                 graph_width = options->stat_graph_width;
2739
2740         name_width = (options->stat_name_width > 0 &&
2741                       options->stat_name_width < max_len) ?
2742                 options->stat_name_width : max_len;
2743
2744         /*
2745          * Adjust adjustable widths not to exceed maximum width
2746          */
2747         if (name_width + number_width + 6 + graph_width > width) {
2748                 if (graph_width > width * 3/8 - number_width - 6) {
2749                         graph_width = width * 3/8 - number_width - 6;
2750                         if (graph_width < 6)
2751                                 graph_width = 6;
2752                 }
2753
2754                 if (options->stat_graph_width &&
2755                     graph_width > options->stat_graph_width)
2756                         graph_width = options->stat_graph_width;
2757                 if (name_width > width - number_width - 6 - graph_width)
2758                         name_width = width - number_width - 6 - graph_width;
2759                 else
2760                         graph_width = width - number_width - 6 - name_width;
2761         }
2762
2763         /*
2764          * From here name_width is the width of the name area,
2765          * and graph_width is the width of the graph area.
2766          * max_change is used to scale graph properly.
2767          */
2768         for (i = 0; i < count; i++) {
2769                 const char *prefix = "";
2770                 struct diffstat_file *file = data->files[i];
2771                 char *name = file->print_name;
2772                 uintmax_t added = file->added;
2773                 uintmax_t deleted = file->deleted;
2774                 int name_len;
2775
2776                 if (!file->is_interesting && (added + deleted == 0))
2777                         continue;
2778
2779                 /*
2780                  * "scale" the filename
2781                  */
2782                 len = name_width;
2783                 name_len = strlen(name);
2784                 if (name_width < name_len) {
2785                         char *slash;
2786                         prefix = "...";
2787                         len -= 3;
2788                         name += name_len - len;
2789                         slash = strchr(name, '/');
2790                         if (slash)
2791                                 name = slash;
2792                 }
2793
2794                 if (file->is_binary) {
2795                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2796                         strbuf_addf(&out, " %*s", number_width, "Bin");
2797                         if (!added && !deleted) {
2798                                 strbuf_addch(&out, '\n');
2799                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2800                                                  out.buf, out.len, 0);
2801                                 strbuf_reset(&out);
2802                                 continue;
2803                         }
2804                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2805                                 del_c, deleted, reset);
2806                         strbuf_addstr(&out, " -> ");
2807                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2808                                 add_c, added, reset);
2809                         strbuf_addstr(&out, " bytes\n");
2810                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2811                                          out.buf, out.len, 0);
2812                         strbuf_reset(&out);
2813                         continue;
2814                 }
2815                 else if (file->is_unmerged) {
2816                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2817                         strbuf_addstr(&out, " Unmerged\n");
2818                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2819                                          out.buf, out.len, 0);
2820                         strbuf_reset(&out);
2821                         continue;
2822                 }
2823
2824                 /*
2825                  * scale the add/delete
2826                  */
2827                 add = added;
2828                 del = deleted;
2829
2830                 if (graph_width <= max_change) {
2831                         int total = scale_linear(add + del, graph_width, max_change);
2832                         if (total < 2 && add && del)
2833                                 /* width >= 2 due to the sanity check */
2834                                 total = 2;
2835                         if (add < del) {
2836                                 add = scale_linear(add, graph_width, max_change);
2837                                 del = total - add;
2838                         } else {
2839                                 del = scale_linear(del, graph_width, max_change);
2840                                 add = total - del;
2841                         }
2842                 }
2843                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2844                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2845                         number_width, added + deleted,
2846                         added + deleted ? " " : "");
2847                 show_graph(&out, '+', add, add_c, reset);
2848                 show_graph(&out, '-', del, del_c, reset);
2849                 strbuf_addch(&out, '\n');
2850                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2851                                  out.buf, out.len, 0);
2852                 strbuf_reset(&out);
2853         }
2854
2855         for (i = 0; i < data->nr; i++) {
2856                 struct diffstat_file *file = data->files[i];
2857                 uintmax_t added = file->added;
2858                 uintmax_t deleted = file->deleted;
2859
2860                 if (file->is_unmerged ||
2861                     (!file->is_interesting && (added + deleted == 0))) {
2862                         total_files--;
2863                         continue;
2864                 }
2865
2866                 if (!file->is_binary) {
2867                         adds += added;
2868                         dels += deleted;
2869                 }
2870                 if (i < count)
2871                         continue;
2872                 if (!extra_shown)
2873                         emit_diff_symbol(options,
2874                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2875                                          NULL, 0, 0);
2876                 extra_shown = 1;
2877         }
2878
2879         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2880         strbuf_release(&out);
2881 }
2882
2883 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2884 {
2885         int i, adds = 0, dels = 0, total_files = data->nr;
2886
2887         if (data->nr == 0)
2888                 return;
2889
2890         for (i = 0; i < data->nr; i++) {
2891                 int added = data->files[i]->added;
2892                 int deleted = data->files[i]->deleted;
2893
2894                 if (data->files[i]->is_unmerged ||
2895                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2896                         total_files--;
2897                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2898                         adds += added;
2899                         dels += deleted;
2900                 }
2901         }
2902         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2903 }
2904
2905 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2906 {
2907         int i;
2908
2909         if (data->nr == 0)
2910                 return;
2911
2912         for (i = 0; i < data->nr; i++) {
2913                 struct diffstat_file *file = data->files[i];
2914
2915                 fprintf(options->file, "%s", diff_line_prefix(options));
2916
2917                 if (file->is_binary)
2918                         fprintf(options->file, "-\t-\t");
2919                 else
2920                         fprintf(options->file,
2921                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2922                                 file->added, file->deleted);
2923                 if (options->line_termination) {
2924                         fill_print_name(file);
2925                         if (!file->is_renamed)
2926                                 write_name_quoted(file->name, options->file,
2927                                                   options->line_termination);
2928                         else {
2929                                 fputs(file->print_name, options->file);
2930                                 putc(options->line_termination, options->file);
2931                         }
2932                 } else {
2933                         if (file->is_renamed) {
2934                                 putc('\0', options->file);
2935                                 write_name_quoted(file->from_name, options->file, '\0');
2936                         }
2937                         write_name_quoted(file->name, options->file, '\0');
2938                 }
2939         }
2940 }
2941
2942 struct dirstat_file {
2943         const char *name;
2944         unsigned long changed;
2945 };
2946
2947 struct dirstat_dir {
2948         struct dirstat_file *files;
2949         int alloc, nr, permille, cumulative;
2950 };
2951
2952 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2953                 unsigned long changed, const char *base, int baselen)
2954 {
2955         unsigned long sum_changes = 0;
2956         unsigned int sources = 0;
2957         const char *line_prefix = diff_line_prefix(opt);
2958
2959         while (dir->nr) {
2960                 struct dirstat_file *f = dir->files;
2961                 int namelen = strlen(f->name);
2962                 unsigned long changes;
2963                 char *slash;
2964
2965                 if (namelen < baselen)
2966                         break;
2967                 if (memcmp(f->name, base, baselen))
2968                         break;
2969                 slash = strchr(f->name + baselen, '/');
2970                 if (slash) {
2971                         int newbaselen = slash + 1 - f->name;
2972                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2973                         sources++;
2974                 } else {
2975                         changes = f->changed;
2976                         dir->files++;
2977                         dir->nr--;
2978                         sources += 2;
2979                 }
2980                 sum_changes += changes;
2981         }
2982
2983         /*
2984          * We don't report dirstat's for
2985          *  - the top level
2986          *  - or cases where everything came from a single directory
2987          *    under this directory (sources == 1).
2988          */
2989         if (baselen && sources != 1) {
2990                 if (sum_changes) {
2991                         int permille = sum_changes * 1000 / changed;
2992                         if (permille >= dir->permille) {
2993                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2994                                         permille / 10, permille % 10, baselen, base);
2995                                 if (!dir->cumulative)
2996                                         return 0;
2997                         }
2998                 }
2999         }
3000         return sum_changes;
3001 }
3002
3003 static int dirstat_compare(const void *_a, const void *_b)
3004 {
3005         const struct dirstat_file *a = _a;
3006         const struct dirstat_file *b = _b;
3007         return strcmp(a->name, b->name);
3008 }
3009
3010 static void show_dirstat(struct diff_options *options)
3011 {
3012         int i;
3013         unsigned long changed;
3014         struct dirstat_dir dir;
3015         struct diff_queue_struct *q = &diff_queued_diff;
3016
3017         dir.files = NULL;
3018         dir.alloc = 0;
3019         dir.nr = 0;
3020         dir.permille = options->dirstat_permille;
3021         dir.cumulative = options->flags.dirstat_cumulative;
3022
3023         changed = 0;
3024         for (i = 0; i < q->nr; i++) {
3025                 struct diff_filepair *p = q->queue[i];
3026                 const char *name;
3027                 unsigned long copied, added, damage;
3028
3029                 name = p->two->path ? p->two->path : p->one->path;
3030
3031                 if (p->one->oid_valid && p->two->oid_valid &&
3032                     oideq(&p->one->oid, &p->two->oid)) {
3033                         /*
3034                          * The SHA1 has not changed, so pre-/post-content is
3035                          * identical. We can therefore skip looking at the
3036                          * file contents altogether.
3037                          */
3038                         damage = 0;
3039                         goto found_damage;
3040                 }
3041
3042                 if (options->flags.dirstat_by_file) {
3043                         /*
3044                          * In --dirstat-by-file mode, we don't really need to
3045                          * look at the actual file contents at all.
3046                          * The fact that the SHA1 changed is enough for us to
3047                          * add this file to the list of results
3048                          * (with each file contributing equal damage).
3049                          */
3050                         damage = 1;
3051                         goto found_damage;
3052                 }
3053
3054                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3055                         diff_populate_filespec(options->repo, p->one, 0);
3056                         diff_populate_filespec(options->repo, p->two, 0);
3057                         diffcore_count_changes(options->repo,
3058                                                p->one, p->two, NULL, NULL,
3059                                                &copied, &added);
3060                         diff_free_filespec_data(p->one);
3061                         diff_free_filespec_data(p->two);
3062                 } else if (DIFF_FILE_VALID(p->one)) {
3063                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3064                         copied = added = 0;
3065                         diff_free_filespec_data(p->one);
3066                 } else if (DIFF_FILE_VALID(p->two)) {
3067                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3068                         copied = 0;
3069                         added = p->two->size;
3070                         diff_free_filespec_data(p->two);
3071                 } else
3072                         continue;
3073
3074                 /*
3075                  * Original minus copied is the removed material,
3076                  * added is the new material.  They are both damages
3077                  * made to the preimage.
3078                  * If the resulting damage is zero, we know that
3079                  * diffcore_count_changes() considers the two entries to
3080                  * be identical, but since the oid changed, we
3081                  * know that there must have been _some_ kind of change,
3082                  * so we force all entries to have damage > 0.
3083                  */
3084                 damage = (p->one->size - copied) + added;
3085                 if (!damage)
3086                         damage = 1;
3087
3088 found_damage:
3089                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3090                 dir.files[dir.nr].name = name;
3091                 dir.files[dir.nr].changed = damage;
3092                 changed += damage;
3093                 dir.nr++;
3094         }
3095
3096         /* This can happen even with many files, if everything was renames */
3097         if (!changed)
3098                 return;
3099
3100         /* Show all directories with more than x% of the changes */
3101         QSORT(dir.files, dir.nr, dirstat_compare);
3102         gather_dirstat(options, &dir, changed, "", 0);
3103 }
3104
3105 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3106 {
3107         int i;
3108         unsigned long changed;
3109         struct dirstat_dir dir;
3110
3111         if (data->nr == 0)
3112                 return;
3113
3114         dir.files = NULL;
3115         dir.alloc = 0;
3116         dir.nr = 0;
3117         dir.permille = options->dirstat_permille;
3118         dir.cumulative = options->flags.dirstat_cumulative;
3119
3120         changed = 0;
3121         for (i = 0; i < data->nr; i++) {
3122                 struct diffstat_file *file = data->files[i];
3123                 unsigned long damage = file->added + file->deleted;
3124                 if (file->is_binary)
3125                         /*
3126                          * binary files counts bytes, not lines. Must find some
3127                          * way to normalize binary bytes vs. textual lines.
3128                          * The following heuristic assumes that there are 64
3129                          * bytes per "line".
3130                          * This is stupid and ugly, but very cheap...
3131                          */
3132                         damage = DIV_ROUND_UP(damage, 64);
3133                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3134                 dir.files[dir.nr].name = file->name;
3135                 dir.files[dir.nr].changed = damage;
3136                 changed += damage;
3137                 dir.nr++;
3138         }
3139
3140         /* This can happen even with many files, if everything was renames */
3141         if (!changed)
3142                 return;
3143
3144         /* Show all directories with more than x% of the changes */
3145         QSORT(dir.files, dir.nr, dirstat_compare);
3146         gather_dirstat(options, &dir, changed, "", 0);
3147 }
3148
3149 static void free_diffstat_info(struct diffstat_t *diffstat)
3150 {
3151         int i;
3152         for (i = 0; i < diffstat->nr; i++) {
3153                 struct diffstat_file *f = diffstat->files[i];
3154                 free(f->print_name);
3155                 free(f->name);
3156                 free(f->from_name);
3157                 free(f);
3158         }
3159         free(diffstat->files);
3160 }
3161
3162 struct checkdiff_t {
3163         const char *filename;
3164         int lineno;
3165         int conflict_marker_size;
3166         struct diff_options *o;
3167         unsigned ws_rule;
3168         unsigned status;
3169 };
3170
3171 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3172 {
3173         char firstchar;
3174         int cnt;
3175
3176         if (len < marker_size + 1)
3177                 return 0;
3178         firstchar = line[0];
3179         switch (firstchar) {
3180         case '=': case '>': case '<': case '|':
3181                 break;
3182         default:
3183                 return 0;
3184         }
3185         for (cnt = 1; cnt < marker_size; cnt++)
3186                 if (line[cnt] != firstchar)
3187                         return 0;
3188         /* line[1] thru line[marker_size-1] are same as firstchar */
3189         if (len < marker_size + 1 || !isspace(line[marker_size]))
3190                 return 0;
3191         return 1;
3192 }
3193
3194 static void checkdiff_consume_hunk(void *priv,
3195                                    long ob, long on, long nb, long nn,
3196                                    const char *func, long funclen)
3197
3198 {
3199         struct checkdiff_t *data = priv;
3200         data->lineno = nb - 1;
3201 }
3202
3203 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3204 {
3205         struct checkdiff_t *data = priv;
3206         int marker_size = data->conflict_marker_size;
3207         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3208         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3209         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3210         char *err;
3211         const char *line_prefix;
3212
3213         assert(data->o);
3214         line_prefix = diff_line_prefix(data->o);
3215
3216         if (line[0] == '+') {
3217                 unsigned bad;
3218                 data->lineno++;
3219                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3220                         data->status |= 1;
3221                         fprintf(data->o->file,
3222                                 "%s%s:%d: leftover conflict marker\n",
3223                                 line_prefix, data->filename, data->lineno);
3224                 }
3225                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3226                 if (!bad)
3227                         return;
3228                 data->status |= bad;
3229                 err = whitespace_error_string(bad);
3230                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3231                         line_prefix, data->filename, data->lineno, err);
3232                 free(err);
3233                 emit_line(data->o, set, reset, line, 1);
3234                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3235                               data->o->file, set, reset, ws);
3236         } else if (line[0] == ' ') {
3237                 data->lineno++;
3238         }
3239 }
3240
3241 static unsigned char *deflate_it(char *data,
3242                                  unsigned long size,
3243                                  unsigned long *result_size)
3244 {
3245         int bound;
3246         unsigned char *deflated;
3247         git_zstream stream;
3248
3249         git_deflate_init(&stream, zlib_compression_level);
3250         bound = git_deflate_bound(&stream, size);
3251         deflated = xmalloc(bound);
3252         stream.next_out = deflated;
3253         stream.avail_out = bound;
3254
3255         stream.next_in = (unsigned char *)data;
3256         stream.avail_in = size;
3257         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3258                 ; /* nothing */
3259         git_deflate_end(&stream);
3260         *result_size = stream.total_out;
3261         return deflated;
3262 }
3263
3264 static void emit_binary_diff_body(struct diff_options *o,
3265                                   mmfile_t *one, mmfile_t *two)
3266 {
3267         void *cp;
3268         void *delta;
3269         void *deflated;
3270         void *data;
3271         unsigned long orig_size;
3272         unsigned long delta_size;
3273         unsigned long deflate_size;
3274         unsigned long data_size;
3275
3276         /* We could do deflated delta, or we could do just deflated two,
3277          * whichever is smaller.
3278          */
3279         delta = NULL;
3280         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3281         if (one->size && two->size) {
3282                 delta = diff_delta(one->ptr, one->size,
3283                                    two->ptr, two->size,
3284                                    &delta_size, deflate_size);
3285                 if (delta) {
3286                         void *to_free = delta;
3287                         orig_size = delta_size;
3288                         delta = deflate_it(delta, delta_size, &delta_size);
3289                         free(to_free);
3290                 }
3291         }
3292
3293         if (delta && delta_size < deflate_size) {
3294                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3295                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3296                                  s, strlen(s), 0);
3297                 free(s);
3298                 free(deflated);
3299                 data = delta;
3300                 data_size = delta_size;
3301         } else {
3302                 char *s = xstrfmt("%lu", two->size);
3303                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3304                                  s, strlen(s), 0);
3305                 free(s);
3306                 free(delta);
3307                 data = deflated;
3308                 data_size = deflate_size;
3309         }
3310
3311         /* emit data encoded in base85 */
3312         cp = data;
3313         while (data_size) {
3314                 int len;
3315                 int bytes = (52 < data_size) ? 52 : data_size;
3316                 char line[71];
3317                 data_size -= bytes;
3318                 if (bytes <= 26)
3319                         line[0] = bytes + 'A' - 1;
3320                 else
3321                         line[0] = bytes - 26 + 'a' - 1;
3322                 encode_85(line + 1, cp, bytes);
3323                 cp = (char *) cp + bytes;
3324
3325                 len = strlen(line);
3326                 line[len++] = '\n';
3327                 line[len] = '\0';
3328
3329                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3330                                  line, len, 0);
3331         }
3332         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3333         free(data);
3334 }
3335
3336 static void emit_binary_diff(struct diff_options *o,
3337                              mmfile_t *one, mmfile_t *two)
3338 {
3339         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3340         emit_binary_diff_body(o, one, two);
3341         emit_binary_diff_body(o, two, one);
3342 }
3343
3344 int diff_filespec_is_binary(struct repository *r,
3345                             struct diff_filespec *one)
3346 {
3347         if (one->is_binary == -1) {
3348                 diff_filespec_load_driver(one, r->index);
3349                 if (one->driver->binary != -1)
3350                         one->is_binary = one->driver->binary;
3351                 else {
3352                         if (!one->data && DIFF_FILE_VALID(one))
3353                                 diff_populate_filespec(r, one, CHECK_BINARY);
3354                         if (one->is_binary == -1 && one->data)
3355                                 one->is_binary = buffer_is_binary(one->data,
3356                                                 one->size);
3357                         if (one->is_binary == -1)
3358                                 one->is_binary = 0;
3359                 }
3360         }
3361         return one->is_binary;
3362 }
3363
3364 static const struct userdiff_funcname *
3365 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3366 {
3367         diff_filespec_load_driver(one, o->repo->index);
3368         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3369 }
3370
3371 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3372 {
3373         if (!options->a_prefix)
3374                 options->a_prefix = a;
3375         if (!options->b_prefix)
3376                 options->b_prefix = b;
3377 }
3378
3379 struct userdiff_driver *get_textconv(struct repository *r,
3380                                      struct diff_filespec *one)
3381 {
3382         if (!DIFF_FILE_VALID(one))
3383                 return NULL;
3384
3385         diff_filespec_load_driver(one, r->index);
3386         return userdiff_get_textconv(r, one->driver);
3387 }
3388
3389 static void builtin_diff(const char *name_a,
3390                          const char *name_b,
3391                          struct diff_filespec *one,
3392                          struct diff_filespec *two,
3393                          const char *xfrm_msg,
3394                          int must_show_header,
3395                          struct diff_options *o,
3396                          int complete_rewrite)
3397 {
3398         mmfile_t mf1, mf2;
3399         const char *lbl[2];
3400         char *a_one, *b_two;
3401         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3402         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3403         const char *a_prefix, *b_prefix;
3404         struct userdiff_driver *textconv_one = NULL;
3405         struct userdiff_driver *textconv_two = NULL;
3406         struct strbuf header = STRBUF_INIT;
3407         const char *line_prefix = diff_line_prefix(o);
3408
3409         diff_set_mnemonic_prefix(o, "a/", "b/");
3410         if (o->flags.reverse_diff) {
3411                 a_prefix = o->b_prefix;
3412                 b_prefix = o->a_prefix;
3413         } else {
3414                 a_prefix = o->a_prefix;
3415                 b_prefix = o->b_prefix;
3416         }
3417
3418         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3419             (!one->mode || S_ISGITLINK(one->mode)) &&
3420             (!two->mode || S_ISGITLINK(two->mode))) {
3421                 show_submodule_summary(o, one->path ? one->path : two->path,
3422                                 &one->oid, &two->oid,
3423                                 two->dirty_submodule);
3424                 return;
3425         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3426                    (!one->mode || S_ISGITLINK(one->mode)) &&
3427                    (!two->mode || S_ISGITLINK(two->mode))) {
3428                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3429                                 &one->oid, &two->oid,
3430                                 two->dirty_submodule);
3431                 return;
3432         }
3433
3434         if (o->flags.allow_textconv) {
3435                 textconv_one = get_textconv(o->repo, one);
3436                 textconv_two = get_textconv(o->repo, two);
3437         }
3438
3439         /* Never use a non-valid filename anywhere if at all possible */
3440         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3441         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3442
3443         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3444         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3445         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3446         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3447         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3448         if (lbl[0][0] == '/') {
3449                 /* /dev/null */
3450                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3451                 if (xfrm_msg)
3452                         strbuf_addstr(&header, xfrm_msg);
3453                 must_show_header = 1;
3454         }
3455         else if (lbl[1][0] == '/') {
3456                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3457                 if (xfrm_msg)
3458                         strbuf_addstr(&header, xfrm_msg);
3459                 must_show_header = 1;
3460         }
3461         else {
3462                 if (one->mode != two->mode) {
3463                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3464                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3465                         must_show_header = 1;
3466                 }
3467                 if (xfrm_msg)
3468                         strbuf_addstr(&header, xfrm_msg);
3469
3470                 /*
3471                  * we do not run diff between different kind
3472                  * of objects.
3473                  */
3474                 if ((one->mode ^ two->mode) & S_IFMT)
3475                         goto free_ab_and_return;
3476                 if (complete_rewrite &&
3477                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3478                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3479                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3480                                          header.buf, header.len, 0);
3481                         strbuf_reset(&header);
3482                         emit_rewrite_diff(name_a, name_b, one, two,
3483                                           textconv_one, textconv_two, o);
3484                         o->found_changes = 1;
3485                         goto free_ab_and_return;
3486                 }
3487         }
3488
3489         if (o->irreversible_delete && lbl[1][0] == '/') {
3490                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3491                                  header.len, 0);
3492                 strbuf_reset(&header);
3493                 goto free_ab_and_return;
3494         } else if (!o->flags.text &&
3495                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3496                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3497                 struct strbuf sb = STRBUF_INIT;
3498                 if (!one->data && !two->data &&
3499                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3500                     !o->flags.binary) {
3501                         if (oideq(&one->oid, &two->oid)) {
3502                                 if (must_show_header)
3503                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3504                                                          header.buf, header.len,
3505                                                          0);
3506                                 goto free_ab_and_return;
3507                         }
3508                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3509                                          header.buf, header.len, 0);
3510                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3511                                     diff_line_prefix(o), lbl[0], lbl[1]);
3512                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3513                                          sb.buf, sb.len, 0);
3514                         strbuf_release(&sb);
3515                         goto free_ab_and_return;
3516                 }
3517                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3518                     fill_mmfile(o->repo, &mf2, two) < 0)
3519                         die("unable to read files to diff");
3520                 /* Quite common confusing case */
3521                 if (mf1.size == mf2.size &&
3522                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3523                         if (must_show_header)
3524                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3525                                                  header.buf, header.len, 0);
3526                         goto free_ab_and_return;
3527                 }
3528                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3529                 strbuf_reset(&header);
3530                 if (o->flags.binary)
3531                         emit_binary_diff(o, &mf1, &mf2);
3532                 else {
3533                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3534                                     diff_line_prefix(o), lbl[0], lbl[1]);
3535                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3536                                          sb.buf, sb.len, 0);
3537                         strbuf_release(&sb);
3538                 }
3539                 o->found_changes = 1;
3540         } else {
3541                 /* Crazy xdl interfaces.. */
3542                 const char *diffopts;
3543                 const char *v;
3544                 xpparam_t xpp;
3545                 xdemitconf_t xecfg;
3546                 struct emit_callback ecbdata;
3547                 const struct userdiff_funcname *pe;
3548
3549                 if (must_show_header) {
3550                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3551                                          header.buf, header.len, 0);
3552                         strbuf_reset(&header);
3553                 }
3554
3555                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3556                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3557
3558                 pe = diff_funcname_pattern(o, one);
3559                 if (!pe)
3560                         pe = diff_funcname_pattern(o, two);
3561
3562                 memset(&xpp, 0, sizeof(xpp));
3563                 memset(&xecfg, 0, sizeof(xecfg));
3564                 memset(&ecbdata, 0, sizeof(ecbdata));
3565                 if (o->flags.suppress_diff_headers)
3566                         lbl[0] = NULL;
3567                 ecbdata.label_path = lbl;
3568                 ecbdata.color_diff = want_color(o->use_color);
3569                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3570                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3571                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3572                 ecbdata.opt = o;
3573                 if (header.len && !o->flags.suppress_diff_headers)
3574                         ecbdata.header = &header;
3575                 xpp.flags = o->xdl_opts;
3576                 xpp.anchors = o->anchors;
3577                 xpp.anchors_nr = o->anchors_nr;
3578                 xecfg.ctxlen = o->context;
3579                 xecfg.interhunkctxlen = o->interhunkcontext;
3580                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3581                 if (o->flags.funccontext)
3582                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3583                 if (pe)
3584                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3585
3586                 diffopts = getenv("GIT_DIFF_OPTS");
3587                 if (!diffopts)
3588                         ;
3589                 else if (skip_prefix(diffopts, "--unified=", &v))
3590                         xecfg.ctxlen = strtoul(v, NULL, 10);
3591                 else if (skip_prefix(diffopts, "-u", &v))
3592                         xecfg.ctxlen = strtoul(v, NULL, 10);
3593
3594                 if (o->word_diff)
3595                         init_diff_words_data(&ecbdata, o, one, two);
3596                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3597                                   &ecbdata, &xpp, &xecfg))
3598                         die("unable to generate diff for %s", one->path);
3599                 if (o->word_diff)
3600                         free_diff_words_data(&ecbdata);
3601                 if (textconv_one)
3602                         free(mf1.ptr);
3603                 if (textconv_two)
3604                         free(mf2.ptr);
3605                 xdiff_clear_find_func(&xecfg);
3606         }
3607
3608  free_ab_and_return:
3609         strbuf_release(&header);
3610         diff_free_filespec_data(one);
3611         diff_free_filespec_data(two);
3612         free(a_one);
3613         free(b_two);
3614         return;
3615 }
3616
3617 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3618 {
3619         if (!is_renamed) {
3620                 if (p->status == DIFF_STATUS_ADDED) {
3621                         if (S_ISLNK(p->two->mode))
3622                                 return "new +l";
3623                         else if ((p->two->mode & 0777) == 0755)
3624                                 return "new +x";
3625                         else
3626                                 return "new";
3627                 } else if (p->status == DIFF_STATUS_DELETED)
3628                         return "gone";
3629         }
3630         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3631                 return "mode -l";
3632         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3633                 return "mode +l";
3634         else if ((p->one->mode & 0777) == 0644 &&
3635                  (p->two->mode & 0777) == 0755)
3636                 return "mode +x";
3637         else if ((p->one->mode & 0777) == 0755 &&
3638                  (p->two->mode & 0777) == 0644)
3639                 return "mode -x";
3640         return NULL;
3641 }
3642
3643 static void builtin_diffstat(const char *name_a, const char *name_b,
3644                              struct diff_filespec *one,
3645                              struct diff_filespec *two,
3646                              struct diffstat_t *diffstat,
3647                              struct diff_options *o,
3648                              struct diff_filepair *p)
3649 {
3650         mmfile_t mf1, mf2;
3651         struct diffstat_file *data;
3652         int same_contents;
3653         int complete_rewrite = 0;
3654
3655         if (!DIFF_PAIR_UNMERGED(p)) {
3656                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3657                         complete_rewrite = 1;
3658         }
3659
3660         data = diffstat_add(diffstat, name_a, name_b);
3661         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3662         if (o->flags.stat_with_summary)
3663                 data->comments = get_compact_summary(p, data->is_renamed);
3664
3665         if (!one || !two) {
3666                 data->is_unmerged = 1;
3667                 return;
3668         }
3669
3670         same_contents = oideq(&one->oid, &two->oid);
3671
3672         if (diff_filespec_is_binary(o->repo, one) ||
3673             diff_filespec_is_binary(o->repo, two)) {
3674                 data->is_binary = 1;
3675                 if (same_contents) {
3676                         data->added = 0;
3677                         data->deleted = 0;
3678                 } else {
3679                         data->added = diff_filespec_size(o->repo, two);
3680                         data->deleted = diff_filespec_size(o->repo, one);
3681                 }
3682         }
3683
3684         else if (complete_rewrite) {
3685                 diff_populate_filespec(o->repo, one, 0);
3686                 diff_populate_filespec(o->repo, two, 0);
3687                 data->deleted = count_lines(one->data, one->size);
3688                 data->added = count_lines(two->data, two->size);
3689         }
3690
3691         else if (!same_contents) {
3692                 /* Crazy xdl interfaces.. */
3693                 xpparam_t xpp;
3694                 xdemitconf_t xecfg;
3695
3696                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3697                     fill_mmfile(o->repo, &mf2, two) < 0)
3698                         die("unable to read files to diff");
3699
3700                 memset(&xpp, 0, sizeof(xpp));
3701                 memset(&xecfg, 0, sizeof(xecfg));
3702                 xpp.flags = o->xdl_opts;
3703                 xpp.anchors = o->anchors;
3704                 xpp.anchors_nr = o->anchors_nr;
3705                 xecfg.ctxlen = o->context;
3706                 xecfg.interhunkctxlen = o->interhunkcontext;
3707                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3708                                   diffstat_consume, diffstat, &xpp, &xecfg))
3709                         die("unable to generate diffstat for %s", one->path);
3710         }
3711
3712         diff_free_filespec_data(one);
3713         diff_free_filespec_data(two);
3714 }
3715
3716 static void builtin_checkdiff(const char *name_a, const char *name_b,
3717                               const char *attr_path,
3718                               struct diff_filespec *one,
3719                               struct diff_filespec *two,
3720                               struct diff_options *o)
3721 {
3722         mmfile_t mf1, mf2;
3723         struct checkdiff_t data;
3724
3725         if (!two)
3726                 return;
3727
3728         memset(&data, 0, sizeof(data));
3729         data.filename = name_b ? name_b : name_a;
3730         data.lineno = 0;
3731         data.o = o;
3732         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3733         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3734
3735         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3736             fill_mmfile(o->repo, &mf2, two) < 0)
3737                 die("unable to read files to diff");
3738
3739         /*
3740          * All the other codepaths check both sides, but not checking
3741          * the "old" side here is deliberate.  We are checking the newly
3742          * introduced changes, and as long as the "new" side is text, we
3743          * can and should check what it introduces.
3744          */
3745         if (diff_filespec_is_binary(o->repo, two))
3746                 goto free_and_return;
3747         else {
3748                 /* Crazy xdl interfaces.. */
3749                 xpparam_t xpp;
3750                 xdemitconf_t xecfg;
3751
3752                 memset(&xpp, 0, sizeof(xpp));
3753                 memset(&xecfg, 0, sizeof(xecfg));
3754                 xecfg.ctxlen = 1; /* at least one context line */
3755                 xpp.flags = 0;
3756                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3757                                   checkdiff_consume, &data,
3758                                   &xpp, &xecfg))
3759                         die("unable to generate checkdiff for %s", one->path);
3760
3761                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3762                         struct emit_callback ecbdata;
3763                         int blank_at_eof;
3764
3765                         ecbdata.ws_rule = data.ws_rule;
3766                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3767                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3768
3769                         if (blank_at_eof) {
3770                                 static char *err;
3771                                 if (!err)
3772                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3773                                 fprintf(o->file, "%s:%d: %s.\n",
3774                                         data.filename, blank_at_eof, err);
3775                                 data.status = 1; /* report errors */
3776                         }
3777                 }
3778         }
3779  free_and_return:
3780         diff_free_filespec_data(one);
3781         diff_free_filespec_data(two);
3782         if (data.status)
3783                 o->flags.check_failed = 1;
3784 }
3785
3786 struct diff_filespec *alloc_filespec(const char *path)
3787 {
3788         struct diff_filespec *spec;
3789
3790         FLEXPTR_ALLOC_STR(spec, path, path);
3791         spec->count = 1;
3792         spec->is_binary = -1;
3793         return spec;
3794 }
3795
3796 void free_filespec(struct diff_filespec *spec)
3797 {
3798         if (!--spec->count) {
3799                 diff_free_filespec_data(spec);
3800                 free(spec);
3801         }
3802 }
3803
3804 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3805                    int oid_valid, unsigned short mode)
3806 {
3807         if (mode) {
3808                 spec->mode = canon_mode(mode);
3809                 oidcpy(&spec->oid, oid);
3810                 spec->oid_valid = oid_valid;
3811         }
3812 }
3813
3814 /*
3815  * Given a name and sha1 pair, if the index tells us the file in
3816  * the work tree has that object contents, return true, so that
3817  * prepare_temp_file() does not have to inflate and extract.
3818  */
3819 static int reuse_worktree_file(struct index_state *istate,
3820                                const char *name,
3821                                const struct object_id *oid,
3822                                int want_file)
3823 {
3824         const struct cache_entry *ce;
3825         struct stat st;
3826         int pos, len;
3827
3828         /*
3829          * We do not read the cache ourselves here, because the
3830          * benchmark with my previous version that always reads cache
3831          * shows that it makes things worse for diff-tree comparing
3832          * two linux-2.6 kernel trees in an already checked out work
3833          * tree.  This is because most diff-tree comparisons deal with
3834          * only a small number of files, while reading the cache is
3835          * expensive for a large project, and its cost outweighs the
3836          * savings we get by not inflating the object to a temporary
3837          * file.  Practically, this code only helps when we are used
3838          * by diff-cache --cached, which does read the cache before
3839          * calling us.
3840          */
3841         if (!istate->cache)
3842                 return 0;
3843
3844         /* We want to avoid the working directory if our caller
3845          * doesn't need the data in a normal file, this system
3846          * is rather slow with its stat/open/mmap/close syscalls,
3847          * and the object is contained in a pack file.  The pack
3848          * is probably already open and will be faster to obtain
3849          * the data through than the working directory.  Loose
3850          * objects however would tend to be slower as they need
3851          * to be individually opened and inflated.
3852          */
3853         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3854                 return 0;
3855
3856         /*
3857          * Similarly, if we'd have to convert the file contents anyway, that
3858          * makes the optimization not worthwhile.
3859          */
3860         if (!want_file && would_convert_to_git(istate, name))
3861                 return 0;
3862
3863         len = strlen(name);
3864         pos = index_name_pos(istate, name, len);
3865         if (pos < 0)
3866                 return 0;
3867         ce = istate->cache[pos];
3868
3869         /*
3870          * This is not the sha1 we are looking for, or
3871          * unreusable because it is not a regular file.
3872          */
3873         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3874                 return 0;
3875
3876         /*
3877          * If ce is marked as "assume unchanged", there is no
3878          * guarantee that work tree matches what we are looking for.
3879          */
3880         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3881                 return 0;
3882
3883         /*
3884          * If ce matches the file in the work tree, we can reuse it.
3885          */
3886         if (ce_uptodate(ce) ||
3887             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3888                 return 1;
3889
3890         return 0;
3891 }
3892
3893 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3894 {
3895         struct strbuf buf = STRBUF_INIT;
3896         char *dirty = "";
3897
3898         /* Are we looking at the work tree? */
3899         if (s->dirty_submodule)
3900                 dirty = "-dirty";
3901
3902         strbuf_addf(&buf, "Subproject commit %s%s\n",
3903                     oid_to_hex(&s->oid), dirty);
3904         s->size = buf.len;
3905         if (size_only) {
3906                 s->data = NULL;
3907                 strbuf_release(&buf);
3908         } else {
3909                 s->data = strbuf_detach(&buf, NULL);
3910                 s->should_free = 1;
3911         }
3912         return 0;
3913 }
3914
3915 /*
3916  * While doing rename detection and pickaxe operation, we may need to
3917  * grab the data for the blob (or file) for our own in-core comparison.
3918  * diff_filespec has data and size fields for this purpose.
3919  */
3920 int diff_populate_filespec(struct repository *r,
3921                            struct diff_filespec *s,
3922                            unsigned int flags)
3923 {
3924         int size_only = flags & CHECK_SIZE_ONLY;
3925         int err = 0;
3926         int conv_flags = global_conv_flags_eol;
3927         /*
3928          * demote FAIL to WARN to allow inspecting the situation
3929          * instead of refusing.
3930          */
3931         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3932                 conv_flags = CONV_EOL_RNDTRP_WARN;
3933
3934         if (!DIFF_FILE_VALID(s))
3935                 die("internal error: asking to populate invalid file.");
3936         if (S_ISDIR(s->mode))
3937                 return -1;
3938
3939         if (s->data)
3940                 return 0;
3941
3942         if (size_only && 0 < s->size)
3943                 return 0;
3944
3945         if (S_ISGITLINK(s->mode))
3946                 return diff_populate_gitlink(s, size_only);
3947
3948         if (!s->oid_valid ||
3949             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3950                 struct strbuf buf = STRBUF_INIT;
3951                 struct stat st;
3952                 int fd;
3953
3954                 if (lstat(s->path, &st) < 0) {
3955                 err_empty:
3956                         err = -1;
3957                 empty:
3958                         s->data = (char *)"";
3959                         s->size = 0;
3960                         return err;
3961                 }
3962                 s->size = xsize_t(st.st_size);
3963                 if (!s->size)
3964                         goto empty;
3965                 if (S_ISLNK(st.st_mode)) {
3966                         struct strbuf sb = STRBUF_INIT;
3967
3968                         if (strbuf_readlink(&sb, s->path, s->size))
3969                                 goto err_empty;
3970                         s->size = sb.len;
3971                         s->data = strbuf_detach(&sb, NULL);
3972                         s->should_free = 1;
3973                         return 0;
3974                 }
3975
3976                 /*
3977                  * Even if the caller would be happy with getting
3978                  * only the size, we cannot return early at this
3979                  * point if the path requires us to run the content
3980                  * conversion.
3981                  */
3982                 if (size_only && !would_convert_to_git(r->index, s->path))
3983                         return 0;
3984
3985                 /*
3986                  * Note: this check uses xsize_t(st.st_size) that may
3987                  * not be the true size of the blob after it goes
3988                  * through convert_to_git().  This may not strictly be
3989                  * correct, but the whole point of big_file_threshold
3990                  * and is_binary check being that we want to avoid
3991                  * opening the file and inspecting the contents, this
3992                  * is probably fine.
3993                  */
3994                 if ((flags & CHECK_BINARY) &&
3995                     s->size > big_file_threshold && s->is_binary == -1) {
3996                         s->is_binary = 1;
3997                         return 0;
3998                 }
3999                 fd = open(s->path, O_RDONLY);
4000                 if (fd < 0)
4001                         goto err_empty;
4002                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4003                 close(fd);
4004                 s->should_munmap = 1;
4005
4006                 /*
4007                  * Convert from working tree format to canonical git format
4008                  */
4009                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4010                         size_t size = 0;
4011                         munmap(s->data, s->size);
4012                         s->should_munmap = 0;
4013                         s->data = strbuf_detach(&buf, &size);
4014                         s->size = size;
4015                         s->should_free = 1;
4016                 }
4017         }
4018         else {
4019                 enum object_type type;
4020                 if (size_only || (flags & CHECK_BINARY)) {
4021                         type = oid_object_info(r, &s->oid, &s->size);
4022                         if (type < 0)
4023                                 die("unable to read %s",
4024                                     oid_to_hex(&s->oid));
4025                         if (size_only)
4026                                 return 0;
4027                         if (s->size > big_file_threshold && s->is_binary == -1) {
4028                                 s->is_binary = 1;
4029                                 return 0;
4030                         }
4031                 }
4032                 s->data = read_object_file(&s->oid, &type, &s->size);
4033                 if (!s->data)
4034                         die("unable to read %s", oid_to_hex(&s->oid));
4035                 s->should_free = 1;
4036         }
4037         return 0;
4038 }
4039
4040 void diff_free_filespec_blob(struct diff_filespec *s)
4041 {
4042         if (s->should_free)
4043                 free(s->data);
4044         else if (s->should_munmap)
4045                 munmap(s->data, s->size);
4046
4047         if (s->should_free || s->should_munmap) {
4048                 s->should_free = s->should_munmap = 0;
4049                 s->data = NULL;
4050         }
4051 }
4052
4053 void diff_free_filespec_data(struct diff_filespec *s)
4054 {
4055         diff_free_filespec_blob(s);
4056         FREE_AND_NULL(s->cnt_data);
4057 }
4058
4059 static void prep_temp_blob(struct index_state *istate,
4060                            const char *path, struct diff_tempfile *temp,
4061                            void *blob,
4062                            unsigned long size,
4063                            const struct object_id *oid,
4064                            int mode)
4065 {
4066         struct strbuf buf = STRBUF_INIT;
4067         struct strbuf tempfile = STRBUF_INIT;
4068         char *path_dup = xstrdup(path);
4069         const char *base = basename(path_dup);
4070
4071         /* Generate "XXXXXX_basename.ext" */
4072         strbuf_addstr(&tempfile, "XXXXXX_");
4073         strbuf_addstr(&tempfile, base);
4074
4075         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4076         if (!temp->tempfile)
4077                 die_errno("unable to create temp-file");
4078         if (convert_to_working_tree(istate, path,
4079                         (const char *)blob, (size_t)size, &buf)) {
4080                 blob = buf.buf;
4081                 size = buf.len;
4082         }
4083         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4084             close_tempfile_gently(temp->tempfile))
4085                 die_errno("unable to write temp-file");
4086         temp->name = get_tempfile_path(temp->tempfile);
4087         oid_to_hex_r(temp->hex, oid);
4088         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4089         strbuf_release(&buf);
4090         strbuf_release(&tempfile);
4091         free(path_dup);
4092 }
4093
4094 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4095                                                const char *name,
4096                                                struct diff_filespec *one)
4097 {
4098         struct diff_tempfile *temp = claim_diff_tempfile();
4099
4100         if (!DIFF_FILE_VALID(one)) {
4101         not_a_valid_file:
4102                 /* A '-' entry produces this for file-2, and
4103                  * a '+' entry produces this for file-1.
4104                  */
4105                 temp->name = "/dev/null";
4106                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4107                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4108                 return temp;
4109         }
4110
4111         if (!S_ISGITLINK(one->mode) &&
4112             (!one->oid_valid ||
4113              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4114                 struct stat st;
4115                 if (lstat(name, &st) < 0) {
4116                         if (errno == ENOENT)
4117                                 goto not_a_valid_file;
4118                         die_errno("stat(%s)", name);
4119                 }
4120                 if (S_ISLNK(st.st_mode)) {
4121                         struct strbuf sb = STRBUF_INIT;
4122                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4123                                 die_errno("readlink(%s)", name);
4124                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4125                                        (one->oid_valid ?
4126                                         &one->oid : &null_oid),
4127                                        (one->oid_valid ?
4128                                         one->mode : S_IFLNK));
4129                         strbuf_release(&sb);
4130                 }
4131                 else {
4132                         /* we can borrow from the file in the work tree */
4133                         temp->name = name;
4134                         if (!one->oid_valid)
4135                                 oid_to_hex_r(temp->hex, &null_oid);
4136                         else
4137                                 oid_to_hex_r(temp->hex, &one->oid);
4138                         /* Even though we may sometimes borrow the
4139                          * contents from the work tree, we always want
4140                          * one->mode.  mode is trustworthy even when
4141                          * !(one->oid_valid), as long as
4142                          * DIFF_FILE_VALID(one).
4143                          */
4144                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4145                 }
4146                 return temp;
4147         }
4148         else {
4149                 if (diff_populate_filespec(r, one, 0))
4150                         die("cannot read data blob for %s", one->path);
4151                 prep_temp_blob(r->index, name, temp,
4152                                one->data, one->size,
4153                                &one->oid, one->mode);
4154         }
4155         return temp;
4156 }
4157
4158 static void add_external_diff_name(struct repository *r,
4159                                    struct argv_array *argv,
4160                                    const char *name,
4161                                    struct diff_filespec *df)
4162 {
4163         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4164         argv_array_push(argv, temp->name);
4165         argv_array_push(argv, temp->hex);
4166         argv_array_push(argv, temp->mode);
4167 }
4168
4169 /* An external diff command takes:
4170  *
4171  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4172  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4173  *
4174  */
4175 static void run_external_diff(const char *pgm,
4176                               const char *name,
4177                               const char *other,
4178                               struct diff_filespec *one,
4179                               struct diff_filespec *two,
4180                               const char *xfrm_msg,
4181                               struct diff_options *o)
4182 {
4183         struct argv_array argv = ARGV_ARRAY_INIT;
4184         struct argv_array env = ARGV_ARRAY_INIT;
4185         struct diff_queue_struct *q = &diff_queued_diff;
4186
4187         argv_array_push(&argv, pgm);
4188         argv_array_push(&argv, name);
4189
4190         if (one && two) {
4191                 add_external_diff_name(o->repo, &argv, name, one);
4192                 if (!other)
4193                         add_external_diff_name(o->repo, &argv, name, two);
4194                 else {
4195                         add_external_diff_name(o->repo, &argv, other, two);
4196                         argv_array_push(&argv, other);
4197                         argv_array_push(&argv, xfrm_msg);
4198                 }
4199         }
4200
4201         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4202         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4203
4204         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4205                 die(_("external diff died, stopping at %s"), name);
4206
4207         remove_tempfile();
4208         argv_array_clear(&argv);
4209         argv_array_clear(&env);
4210 }
4211
4212 static int similarity_index(struct diff_filepair *p)
4213 {
4214         return p->score * 100 / MAX_SCORE;
4215 }
4216
4217 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4218 {
4219         if (startup_info->have_repository)
4220                 return find_unique_abbrev(oid, abbrev);
4221         else {
4222                 char *hex = oid_to_hex(oid);
4223                 if (abbrev < 0)
4224                         abbrev = FALLBACK_DEFAULT_ABBREV;
4225                 if (abbrev > the_hash_algo->hexsz)
4226                         BUG("oid abbreviation out of range: %d", abbrev);
4227                 if (abbrev)
4228                         hex[abbrev] = '\0';
4229                 return hex;
4230         }
4231 }
4232
4233 static void fill_metainfo(struct strbuf *msg,
4234                           const char *name,
4235                           const char *other,
4236                           struct diff_filespec *one,
4237                           struct diff_filespec *two,
4238                           struct diff_options *o,
4239                           struct diff_filepair *p,
4240                           int *must_show_header,
4241                           int use_color)
4242 {
4243         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4244         const char *reset = diff_get_color(use_color, DIFF_RESET);
4245         const char *line_prefix = diff_line_prefix(o);
4246
4247         *must_show_header = 1;
4248         strbuf_init(msg, PATH_MAX * 2 + 300);
4249         switch (p->status) {
4250         case DIFF_STATUS_COPIED:
4251                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4252                             line_prefix, set, similarity_index(p));
4253                 strbuf_addf(msg, "%s\n%s%scopy from ",
4254                             reset,  line_prefix, set);
4255                 quote_c_style(name, msg, NULL, 0);
4256                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4257                 quote_c_style(other, msg, NULL, 0);
4258                 strbuf_addf(msg, "%s\n", reset);
4259                 break;
4260         case DIFF_STATUS_RENAMED:
4261                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4262                             line_prefix, set, similarity_index(p));
4263                 strbuf_addf(msg, "%s\n%s%srename from ",
4264                             reset, line_prefix, set);
4265                 quote_c_style(name, msg, NULL, 0);
4266                 strbuf_addf(msg, "%s\n%s%srename to ",
4267                             reset, line_prefix, set);
4268                 quote_c_style(other, msg, NULL, 0);
4269                 strbuf_addf(msg, "%s\n", reset);
4270                 break;
4271         case DIFF_STATUS_MODIFIED:
4272                 if (p->score) {
4273                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4274                                     line_prefix,
4275                                     set, similarity_index(p), reset);
4276                         break;
4277                 }
4278                 /* fallthru */
4279         default:
4280                 *must_show_header = 0;
4281         }
4282         if (one && two && !oideq(&one->oid, &two->oid)) {
4283                 const unsigned hexsz = the_hash_algo->hexsz;
4284                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4285
4286                 if (o->flags.binary) {
4287                         mmfile_t mf;
4288                         if ((!fill_mmfile(o->repo, &mf, one) &&
4289                              diff_filespec_is_binary(o->repo, one)) ||
4290                             (!fill_mmfile(o->repo, &mf, two) &&
4291                              diff_filespec_is_binary(o->repo, two)))
4292                                 abbrev = hexsz;
4293                 }
4294                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4295                             diff_abbrev_oid(&one->oid, abbrev),
4296                             diff_abbrev_oid(&two->oid, abbrev));
4297                 if (one->mode == two->mode)
4298                         strbuf_addf(msg, " %06o", one->mode);
4299                 strbuf_addf(msg, "%s\n", reset);
4300         }
4301 }
4302
4303 static void run_diff_cmd(const char *pgm,
4304                          const char *name,
4305                          const char *other,
4306                          const char *attr_path,
4307                          struct diff_filespec *one,
4308                          struct diff_filespec *two,
4309                          struct strbuf *msg,
4310                          struct diff_options *o,
4311                          struct diff_filepair *p)
4312 {
4313         const char *xfrm_msg = NULL;
4314         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4315         int must_show_header = 0;
4316
4317
4318         if (o->flags.allow_external) {
4319                 struct userdiff_driver *drv;
4320
4321                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4322                 if (drv && drv->external)
4323                         pgm = drv->external;
4324         }
4325
4326         if (msg) {
4327                 /*
4328                  * don't use colors when the header is intended for an
4329                  * external diff driver
4330                  */
4331                 fill_metainfo(msg, name, other, one, two, o, p,
4332                               &must_show_header,
4333                               want_color(o->use_color) && !pgm);
4334                 xfrm_msg = msg->len ? msg->buf : NULL;
4335         }
4336
4337         if (pgm) {
4338                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4339                 return;
4340         }
4341         if (one && two)
4342                 builtin_diff(name, other ? other : name,
4343                              one, two, xfrm_msg, must_show_header,
4344                              o, complete_rewrite);
4345         else
4346                 fprintf(o->file, "* Unmerged path %s\n", name);
4347 }
4348
4349 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4350 {
4351         if (DIFF_FILE_VALID(one)) {
4352                 if (!one->oid_valid) {
4353                         struct stat st;
4354                         if (one->is_stdin) {
4355                                 oidclr(&one->oid);
4356                                 return;
4357                         }
4358                         if (lstat(one->path, &st) < 0)
4359                                 die_errno("stat '%s'", one->path);
4360                         if (index_path(istate, &one->oid, one->path, &st, 0))
4361                                 die("cannot hash %s", one->path);
4362                 }
4363         }
4364         else
4365                 oidclr(&one->oid);
4366 }
4367
4368 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4369 {
4370         /* Strip the prefix but do not molest /dev/null and absolute paths */
4371         if (*namep && !is_absolute_path(*namep)) {
4372                 *namep += prefix_length;
4373                 if (**namep == '/')
4374                         ++*namep;
4375         }
4376         if (*otherp && !is_absolute_path(*otherp)) {
4377                 *otherp += prefix_length;
4378                 if (**otherp == '/')
4379                         ++*otherp;
4380         }
4381 }
4382
4383 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4384 {
4385         const char *pgm = external_diff();
4386         struct strbuf msg;
4387         struct diff_filespec *one = p->one;
4388         struct diff_filespec *two = p->two;
4389         const char *name;
4390         const char *other;
4391         const char *attr_path;
4392
4393         name  = one->path;
4394         other = (strcmp(name, two->path) ? two->path : NULL);
4395         attr_path = name;
4396         if (o->prefix_length)
4397                 strip_prefix(o->prefix_length, &name, &other);
4398
4399         if (!o->flags.allow_external)
4400                 pgm = NULL;
4401
4402         if (DIFF_PAIR_UNMERGED(p)) {
4403                 run_diff_cmd(pgm, name, NULL, attr_path,
4404                              NULL, NULL, NULL, o, p);
4405                 return;
4406         }
4407
4408         diff_fill_oid_info(one, o->repo->index);
4409         diff_fill_oid_info(two, o->repo->index);
4410
4411         if (!pgm &&
4412             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4413             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4414                 /*
4415                  * a filepair that changes between file and symlink
4416                  * needs to be split into deletion and creation.
4417                  */
4418                 struct diff_filespec *null = alloc_filespec(two->path);
4419                 run_diff_cmd(NULL, name, other, attr_path,
4420                              one, null, &msg,
4421                              o, p);
4422                 free(null);
4423                 strbuf_release(&msg);
4424
4425                 null = alloc_filespec(one->path);
4426                 run_diff_cmd(NULL, name, other, attr_path,
4427                              null, two, &msg, o, p);
4428                 free(null);
4429         }
4430         else
4431                 run_diff_cmd(pgm, name, other, attr_path,
4432                              one, two, &msg, o, p);
4433
4434         strbuf_release(&msg);
4435 }
4436
4437 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4438                          struct diffstat_t *diffstat)
4439 {
4440         const char *name;
4441         const char *other;
4442
4443         if (DIFF_PAIR_UNMERGED(p)) {
4444                 /* unmerged */
4445                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4446                                  diffstat, o, p);
4447                 return;
4448         }
4449
4450         name = p->one->path;
4451         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4452
4453         if (o->prefix_length)
4454                 strip_prefix(o->prefix_length, &name, &other);
4455
4456         diff_fill_oid_info(p->one, o->repo->index);
4457         diff_fill_oid_info(p->two, o->repo->index);
4458
4459         builtin_diffstat(name, other, p->one, p->two,
4460                          diffstat, o, p);
4461 }
4462
4463 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4464 {
4465         const char *name;
4466         const char *other;
4467         const char *attr_path;
4468
4469         if (DIFF_PAIR_UNMERGED(p)) {
4470                 /* unmerged */
4471                 return;
4472         }
4473
4474         name = p->one->path;
4475         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4476         attr_path = other ? other : name;
4477
4478         if (o->prefix_length)
4479                 strip_prefix(o->prefix_length, &name, &other);
4480
4481         diff_fill_oid_info(p->one, o->repo->index);
4482         diff_fill_oid_info(p->two, o->repo->index);
4483
4484         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4485 }
4486
4487 void repo_diff_setup(struct repository *r, struct diff_options *options)
4488 {
4489         memcpy(options, &default_diff_options, sizeof(*options));
4490
4491         options->file = stdout;
4492         options->repo = r;
4493
4494         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4495         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4496         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4497         options->abbrev = DEFAULT_ABBREV;
4498         options->line_termination = '\n';
4499         options->break_opt = -1;
4500         options->rename_limit = -1;
4501         options->dirstat_permille = diff_dirstat_permille_default;
4502         options->context = diff_context_default;
4503         options->interhunkcontext = diff_interhunk_context_default;
4504         options->ws_error_highlight = ws_error_highlight_default;
4505         options->flags.rename_empty = 1;
4506         options->objfind = NULL;
4507
4508         /* pathchange left =NULL by default */
4509         options->change = diff_change;
4510         options->add_remove = diff_addremove;
4511         options->use_color = diff_use_color_default;
4512         options->detect_rename = diff_detect_rename_default;
4513         options->xdl_opts |= diff_algorithm;
4514         if (diff_indent_heuristic)
4515                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4516
4517         options->orderfile = diff_order_file_cfg;
4518
4519         if (diff_no_prefix) {
4520                 options->a_prefix = options->b_prefix = "";
4521         } else if (!diff_mnemonic_prefix) {
4522                 options->a_prefix = "a/";
4523                 options->b_prefix = "b/";
4524         }
4525
4526         options->color_moved = diff_color_moved_default;
4527         options->color_moved_ws_handling = diff_color_moved_ws_default;
4528 }
4529
4530 void diff_setup_done(struct diff_options *options)
4531 {
4532         unsigned check_mask = DIFF_FORMAT_NAME |
4533                               DIFF_FORMAT_NAME_STATUS |
4534                               DIFF_FORMAT_CHECKDIFF |
4535                               DIFF_FORMAT_NO_OUTPUT;
4536         /*
4537          * This must be signed because we're comparing against a potentially
4538          * negative value.
4539          */
4540         const int hexsz = the_hash_algo->hexsz;
4541
4542         if (options->set_default)
4543                 options->set_default(options);
4544
4545         if (HAS_MULTI_BITS(options->output_format & check_mask))
4546                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4547
4548         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4549                 die(_("-G, -S and --find-object are mutually exclusive"));
4550
4551         /*
4552          * Most of the time we can say "there are changes"
4553          * only by checking if there are changed paths, but
4554          * --ignore-whitespace* options force us to look
4555          * inside contents.
4556          */
4557
4558         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4559                 options->flags.diff_from_contents = 1;
4560         else
4561                 options->flags.diff_from_contents = 0;
4562
4563         if (options->flags.find_copies_harder)
4564                 options->detect_rename = DIFF_DETECT_COPY;
4565
4566         if (!options->flags.relative_name)
4567                 options->prefix = NULL;
4568         if (options->prefix)
4569                 options->prefix_length = strlen(options->prefix);
4570         else
4571                 options->prefix_length = 0;
4572
4573         if (options->output_format & (DIFF_FORMAT_NAME |
4574                                       DIFF_FORMAT_NAME_STATUS |
4575                                       DIFF_FORMAT_CHECKDIFF |
4576                                       DIFF_FORMAT_NO_OUTPUT))
4577                 options->output_format &= ~(DIFF_FORMAT_RAW |
4578                                             DIFF_FORMAT_NUMSTAT |
4579                                             DIFF_FORMAT_DIFFSTAT |
4580                                             DIFF_FORMAT_SHORTSTAT |
4581                                             DIFF_FORMAT_DIRSTAT |
4582                                             DIFF_FORMAT_SUMMARY |
4583                                             DIFF_FORMAT_PATCH);
4584
4585         /*
4586          * These cases always need recursive; we do not drop caller-supplied
4587          * recursive bits for other formats here.
4588          */
4589         if (options->output_format & (DIFF_FORMAT_PATCH |
4590                                       DIFF_FORMAT_NUMSTAT |
4591                                       DIFF_FORMAT_DIFFSTAT |
4592                                       DIFF_FORMAT_SHORTSTAT |
4593                                       DIFF_FORMAT_DIRSTAT |
4594                                       DIFF_FORMAT_SUMMARY |
4595                                       DIFF_FORMAT_CHECKDIFF))
4596                 options->flags.recursive = 1;
4597         /*
4598          * Also pickaxe would not work very well if you do not say recursive
4599          */
4600         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4601                 options->flags.recursive = 1;
4602         /*
4603          * When patches are generated, submodules diffed against the work tree
4604          * must be checked for dirtiness too so it can be shown in the output
4605          */
4606         if (options->output_format & DIFF_FORMAT_PATCH)
4607                 options->flags.dirty_submodules = 1;
4608
4609         if (options->detect_rename && options->rename_limit < 0)
4610                 options->rename_limit = diff_rename_limit_default;
4611         if (hexsz < options->abbrev)
4612                 options->abbrev = hexsz; /* full */
4613
4614         /*
4615          * It does not make sense to show the first hit we happened
4616          * to have found.  It does not make sense not to return with
4617          * exit code in such a case either.
4618          */
4619         if (options->flags.quick) {
4620                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4621                 options->flags.exit_with_status = 1;
4622         }
4623
4624         options->diff_path_counter = 0;
4625
4626         if (options->flags.follow_renames && options->pathspec.nr != 1)
4627                 die(_("--follow requires exactly one pathspec"));
4628
4629         if (!options->use_color || external_diff())
4630                 options->color_moved = 0;
4631 }
4632
4633 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4634 {
4635         char c, *eq;
4636         int len;
4637
4638         if (*arg != '-')
4639                 return 0;
4640         c = *++arg;
4641         if (!c)
4642                 return 0;
4643         if (c == arg_short) {
4644                 c = *++arg;
4645                 if (!c)
4646                         return 1;
4647                 if (val && isdigit(c)) {
4648                         char *end;
4649                         int n = strtoul(arg, &end, 10);
4650                         if (*end)
4651                                 return 0;
4652                         *val = n;
4653                         return 1;
4654                 }
4655                 return 0;
4656         }
4657         if (c != '-')
4658                 return 0;
4659         arg++;
4660         eq = strchrnul(arg, '=');
4661         len = eq - arg;
4662         if (!len || strncmp(arg, arg_long, len))
4663                 return 0;
4664         if (*eq) {
4665                 int n;
4666                 char *end;
4667                 if (!isdigit(*++eq))
4668                         return 0;
4669                 n = strtoul(eq, &end, 10);
4670                 if (*end)
4671                         return 0;
4672                 *val = n;
4673         }
4674         return 1;
4675 }
4676
4677 static int diff_scoreopt_parse(const char *opt);
4678
4679 static inline int short_opt(char opt, const char **argv,
4680                             const char **optarg)
4681 {
4682         const char *arg = argv[0];
4683         if (arg[0] != '-' || arg[1] != opt)
4684                 return 0;
4685         if (arg[2] != '\0') {
4686                 *optarg = arg + 2;
4687                 return 1;
4688         }
4689         if (!argv[1])
4690                 die("Option '%c' requires a value", opt);
4691         *optarg = argv[1];
4692         return 2;
4693 }
4694
4695 int parse_long_opt(const char *opt, const char **argv,
4696                    const char **optarg)
4697 {
4698         const char *arg = argv[0];
4699         if (!skip_prefix(arg, "--", &arg))
4700                 return 0;
4701         if (!skip_prefix(arg, opt, &arg))
4702                 return 0;
4703         if (*arg == '=') { /* stuck form: --option=value */
4704                 *optarg = arg + 1;
4705                 return 1;
4706         }
4707         if (*arg != '\0')
4708                 return 0;
4709         /* separate form: --option value */
4710         if (!argv[1])
4711                 die("Option '--%s' requires a value", opt);
4712         *optarg = argv[1];
4713         return 2;
4714 }
4715
4716 static int stat_opt(struct diff_options *options, const char **av)
4717 {
4718         const char *arg = av[0];
4719         char *end;
4720         int width = options->stat_width;
4721         int name_width = options->stat_name_width;
4722         int graph_width = options->stat_graph_width;
4723         int count = options->stat_count;
4724         int argcount = 1;
4725
4726         if (!skip_prefix(arg, "--stat", &arg))
4727                 BUG("stat option does not begin with --stat: %s", arg);
4728         end = (char *)arg;
4729
4730         switch (*arg) {
4731         case '-':
4732                 if (skip_prefix(arg, "-width", &arg)) {
4733                         if (*arg == '=')
4734                                 width = strtoul(arg + 1, &end, 10);
4735                         else if (!*arg && !av[1])
4736                                 die_want_option("--stat-width");
4737                         else if (!*arg) {
4738                                 width = strtoul(av[1], &end, 10);
4739                                 argcount = 2;
4740                         }
4741                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4742                         if (*arg == '=')
4743                                 name_width = strtoul(arg + 1, &end, 10);
4744                         else if (!*arg && !av[1])
4745                                 die_want_option("--stat-name-width");
4746                         else if (!*arg) {
4747                                 name_width = strtoul(av[1], &end, 10);
4748                                 argcount = 2;
4749                         }
4750                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4751                         if (*arg == '=')
4752                                 graph_width = strtoul(arg + 1, &end, 10);
4753                         else if (!*arg && !av[1])
4754                                 die_want_option("--stat-graph-width");
4755                         else if (!*arg) {
4756                                 graph_width = strtoul(av[1], &end, 10);
4757                                 argcount = 2;
4758                         }
4759                 } else if (skip_prefix(arg, "-count", &arg)) {
4760                         if (*arg == '=')
4761                                 count = strtoul(arg + 1, &end, 10);
4762                         else if (!*arg && !av[1])
4763                                 die_want_option("--stat-count");
4764                         else if (!*arg) {
4765                                 count = strtoul(av[1], &end, 10);
4766                                 argcount = 2;
4767                         }
4768                 }
4769                 break;
4770         case '=':
4771                 width = strtoul(arg+1, &end, 10);
4772                 if (*end == ',')
4773                         name_width = strtoul(end+1, &end, 10);
4774                 if (*end == ',')
4775                         count = strtoul(end+1, &end, 10);
4776         }
4777
4778         /* Important! This checks all the error cases! */
4779         if (*end)
4780                 return 0;
4781         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4782         options->stat_name_width = name_width;
4783         options->stat_graph_width = graph_width;
4784         options->stat_width = width;
4785         options->stat_count = count;
4786         return argcount;
4787 }
4788
4789 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4790 {
4791         struct strbuf errmsg = STRBUF_INIT;
4792         if (parse_dirstat_params(options, params, &errmsg))
4793                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4794                     errmsg.buf);
4795         strbuf_release(&errmsg);
4796         /*
4797          * The caller knows a dirstat-related option is given from the command
4798          * line; allow it to say "return this_function();"
4799          */
4800         options->output_format |= DIFF_FORMAT_DIRSTAT;
4801         return 1;
4802 }
4803
4804 static int parse_submodule_opt(struct diff_options *options, const char *value)
4805 {
4806         if (parse_submodule_params(options, value))
4807                 die(_("Failed to parse --submodule option parameter: '%s'"),
4808                         value);
4809         return 1;
4810 }
4811
4812 static const char diff_status_letters[] = {
4813         DIFF_STATUS_ADDED,
4814         DIFF_STATUS_COPIED,
4815         DIFF_STATUS_DELETED,
4816         DIFF_STATUS_MODIFIED,
4817         DIFF_STATUS_RENAMED,
4818         DIFF_STATUS_TYPE_CHANGED,
4819         DIFF_STATUS_UNKNOWN,
4820         DIFF_STATUS_UNMERGED,
4821         DIFF_STATUS_FILTER_AON,
4822         DIFF_STATUS_FILTER_BROKEN,
4823         '\0',
4824 };
4825
4826 static unsigned int filter_bit['Z' + 1];
4827
4828 static void prepare_filter_bits(void)
4829 {
4830         int i;
4831
4832         if (!filter_bit[DIFF_STATUS_ADDED]) {
4833                 for (i = 0; diff_status_letters[i]; i++)
4834                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4835         }
4836 }
4837
4838 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4839 {
4840         return opt->filter & filter_bit[(int) status];
4841 }
4842
4843 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4844 {
4845         int i, optch;
4846
4847         prepare_filter_bits();
4848
4849         /*
4850          * If there is a negation e.g. 'd' in the input, and we haven't
4851          * initialized the filter field with another --diff-filter, start
4852          * from full set of bits, except for AON.
4853          */
4854         if (!opt->filter) {
4855                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4856                         if (optch < 'a' || 'z' < optch)
4857                                 continue;
4858                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4859                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4860                         break;
4861                 }
4862         }
4863
4864         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4865                 unsigned int bit;
4866                 int negate;
4867
4868                 if ('a' <= optch && optch <= 'z') {
4869                         negate = 1;
4870                         optch = toupper(optch);
4871                 } else {
4872                         negate = 0;
4873                 }
4874
4875                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4876                 if (!bit)
4877                         return optarg[i];
4878                 if (negate)
4879                         opt->filter &= ~bit;
4880                 else
4881                         opt->filter |= bit;
4882         }
4883         return 0;
4884 }
4885
4886 static void enable_patch_output(int *fmt)
4887 {
4888         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4889         *fmt |= DIFF_FORMAT_PATCH;
4890 }
4891
4892 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4893 {
4894         int val = parse_ws_error_highlight(arg);
4895
4896         if (val < 0) {
4897                 error("unknown value after ws-error-highlight=%.*s",
4898                       -1 - val, arg);
4899                 return 0;
4900         }
4901         opt->ws_error_highlight = val;
4902         return 1;
4903 }
4904
4905 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4906 {
4907         struct object_id oid;
4908
4909         if (get_oid(arg, &oid))
4910                 return error("unable to resolve '%s'", arg);
4911
4912         if (!opt->objfind)
4913                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4914
4915         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4916         opt->flags.recursive = 1;
4917         opt->flags.tree_in_recursive = 1;
4918         oidset_insert(opt->objfind, &oid);
4919         return 1;
4920 }
4921
4922 int diff_opt_parse(struct diff_options *options,
4923                    const char **av, int ac, const char *prefix)
4924 {
4925         const char *arg = av[0];
4926         const char *optarg;
4927         int argcount;
4928
4929         if (!prefix)
4930                 prefix = "";
4931
4932         /* Output format options */
4933         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4934             || opt_arg(arg, 'U', "unified", &options->context))
4935                 enable_patch_output(&options->output_format);
4936         else if (!strcmp(arg, "--raw"))
4937                 options->output_format |= DIFF_FORMAT_RAW;
4938         else if (!strcmp(arg, "--patch-with-raw")) {
4939                 enable_patch_output(&options->output_format);
4940                 options->output_format |= DIFF_FORMAT_RAW;
4941         } else if (!strcmp(arg, "--numstat"))
4942                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4943         else if (!strcmp(arg, "--shortstat"))
4944                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4945         else if (skip_prefix(arg, "-X", &arg) ||
4946                  skip_to_optional_arg(arg, "--dirstat", &arg))
4947                 return parse_dirstat_opt(options, arg);
4948         else if (!strcmp(arg, "--cumulative"))
4949                 return parse_dirstat_opt(options, "cumulative");
4950         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4951                 parse_dirstat_opt(options, "files");
4952                 return parse_dirstat_opt(options, arg);
4953         }
4954         else if (!strcmp(arg, "--check"))
4955                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4956         else if (!strcmp(arg, "--summary"))
4957                 options->output_format |= DIFF_FORMAT_SUMMARY;
4958         else if (!strcmp(arg, "--patch-with-stat")) {
4959                 enable_patch_output(&options->output_format);
4960                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4961         } else if (!strcmp(arg, "--name-only"))
4962                 options->output_format |= DIFF_FORMAT_NAME;
4963         else if (!strcmp(arg, "--name-status"))
4964                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4965         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4966                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4967         else if (starts_with(arg, "--stat"))
4968                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4969                 return stat_opt(options, av);
4970         else if (!strcmp(arg, "--compact-summary")) {
4971                  options->flags.stat_with_summary = 1;
4972                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4973         } else if (!strcmp(arg, "--no-compact-summary"))
4974                  options->flags.stat_with_summary = 0;
4975         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4976                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4977         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4978                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4979         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4980                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4981
4982         /* renames options */
4983         else if (starts_with(arg, "-B") ||
4984                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4985                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4986                         return error("invalid argument to -B: %s", arg+2);
4987         }
4988         else if (starts_with(arg, "-M") ||
4989                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4990                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4991                         return error("invalid argument to -M: %s", arg+2);
4992                 options->detect_rename = DIFF_DETECT_RENAME;
4993         }
4994         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4995                 options->irreversible_delete = 1;
4996         }
4997         else if (starts_with(arg, "-C") ||
4998                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4999                 if (options->detect_rename == DIFF_DETECT_COPY)
5000                         options->flags.find_copies_harder = 1;
5001                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5002                         return error("invalid argument to -C: %s", arg+2);
5003                 options->detect_rename = DIFF_DETECT_COPY;
5004         }
5005         else if (!strcmp(arg, "--no-renames"))
5006                 options->detect_rename = 0;
5007         else if (!strcmp(arg, "--rename-empty"))
5008                 options->flags.rename_empty = 1;
5009         else if (!strcmp(arg, "--no-rename-empty"))
5010                 options->flags.rename_empty = 0;
5011         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5012                 options->flags.relative_name = 1;
5013                 if (arg)
5014                         options->prefix = arg;
5015         }
5016
5017         /* xdiff options */
5018         else if (!strcmp(arg, "--minimal"))
5019                 DIFF_XDL_SET(options, NEED_MINIMAL);
5020         else if (!strcmp(arg, "--no-minimal"))
5021                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5022         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5023                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5024         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5025                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5026         else if (!strcmp(arg, "--ignore-space-at-eol"))
5027                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5028         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5029                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5030         else if (!strcmp(arg, "--ignore-blank-lines"))
5031                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5032         else if (!strcmp(arg, "--indent-heuristic"))
5033                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5034         else if (!strcmp(arg, "--no-indent-heuristic"))
5035                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5036         else if (!strcmp(arg, "--patience")) {
5037                 int i;
5038                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5039                 /*
5040                  * Both --patience and --anchored use PATIENCE_DIFF
5041                  * internally, so remove any anchors previously
5042                  * specified.
5043                  */
5044                 for (i = 0; i < options->anchors_nr; i++)
5045                         free(options->anchors[i]);
5046                 options->anchors_nr = 0;
5047         } else if (!strcmp(arg, "--histogram"))
5048                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5049         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5050                 long value = parse_algorithm_value(optarg);
5051                 if (value < 0)
5052                         return error("option diff-algorithm accepts \"myers\", "
5053                                      "\"minimal\", \"patience\" and \"histogram\"");
5054                 /* clear out previous settings */
5055                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5056                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5057                 options->xdl_opts |= value;
5058                 return argcount;
5059         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5060                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5061                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5062                            options->anchors_alloc);
5063                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5064         }
5065
5066         /* flags options */
5067         else if (!strcmp(arg, "--binary")) {
5068                 enable_patch_output(&options->output_format);
5069                 options->flags.binary = 1;
5070         }
5071         else if (!strcmp(arg, "--full-index"))
5072                 options->flags.full_index = 1;
5073         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5074                 options->flags.text = 1;
5075         else if (!strcmp(arg, "-R"))
5076                 options->flags.reverse_diff = 1;
5077         else if (!strcmp(arg, "--find-copies-harder"))
5078                 options->flags.find_copies_harder = 1;
5079         else if (!strcmp(arg, "--follow"))
5080                 options->flags.follow_renames = 1;
5081         else if (!strcmp(arg, "--no-follow")) {
5082                 options->flags.follow_renames = 0;
5083                 options->flags.default_follow_renames = 0;
5084         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5085                 int value = git_config_colorbool(NULL, arg);
5086                 if (value < 0)
5087                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5088                 options->use_color = value;
5089         }
5090         else if (!strcmp(arg, "--no-color"))
5091                 options->use_color = 0;
5092         else if (!strcmp(arg, "--color-moved")) {
5093                 if (diff_color_moved_default)
5094                         options->color_moved = diff_color_moved_default;
5095                 if (options->color_moved == COLOR_MOVED_NO)
5096                         options->color_moved = COLOR_MOVED_DEFAULT;
5097         } else if (!strcmp(arg, "--no-color-moved"))
5098                 options->color_moved = COLOR_MOVED_NO;
5099         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5100                 int cm = parse_color_moved(arg);
5101                 if (cm < 0)
5102                         return error("bad --color-moved argument: %s", arg);
5103                 options->color_moved = cm;
5104         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5105                 options->color_moved_ws_handling = 0;
5106         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5107                 unsigned cm = parse_color_moved_ws(arg);
5108                 if (cm & COLOR_MOVED_WS_ERROR)
5109                         return -1;
5110                 options->color_moved_ws_handling = cm;
5111         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5112                 options->use_color = 1;
5113                 options->word_diff = DIFF_WORDS_COLOR;
5114         }
5115         else if (!strcmp(arg, "--word-diff")) {
5116                 if (options->word_diff == DIFF_WORDS_NONE)
5117                         options->word_diff = DIFF_WORDS_PLAIN;
5118         }
5119         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5120                 if (!strcmp(arg, "plain"))
5121                         options->word_diff = DIFF_WORDS_PLAIN;
5122                 else if (!strcmp(arg, "color")) {
5123                         options->use_color = 1;
5124                         options->word_diff = DIFF_WORDS_COLOR;
5125                 }
5126                 else if (!strcmp(arg, "porcelain"))
5127                         options->word_diff = DIFF_WORDS_PORCELAIN;
5128                 else if (!strcmp(arg, "none"))
5129                         options->word_diff = DIFF_WORDS_NONE;
5130                 else
5131                         die("bad --word-diff argument: %s", arg);
5132         }
5133         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5134                 if (options->word_diff == DIFF_WORDS_NONE)
5135                         options->word_diff = DIFF_WORDS_PLAIN;
5136                 options->word_regex = optarg;
5137                 return argcount;
5138         }
5139         else if (!strcmp(arg, "--exit-code"))
5140                 options->flags.exit_with_status = 1;
5141         else if (!strcmp(arg, "--quiet"))
5142                 options->flags.quick = 1;
5143         else if (!strcmp(arg, "--ext-diff"))
5144                 options->flags.allow_external = 1;
5145         else if (!strcmp(arg, "--no-ext-diff"))
5146                 options->flags.allow_external = 0;
5147         else if (!strcmp(arg, "--textconv")) {
5148                 options->flags.allow_textconv = 1;
5149                 options->flags.textconv_set_via_cmdline = 1;
5150         } else if (!strcmp(arg, "--no-textconv"))
5151                 options->flags.allow_textconv = 0;
5152         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5153                 options->flags.override_submodule_config = 1;
5154                 handle_ignore_submodules_arg(options, arg);
5155         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5156                 return parse_submodule_opt(options, arg);
5157         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5158                 return parse_ws_error_highlight_opt(options, arg);
5159         else if (!strcmp(arg, "--ita-invisible-in-index"))
5160                 options->ita_invisible_in_index = 1;
5161         else if (!strcmp(arg, "--ita-visible-in-index"))
5162                 options->ita_invisible_in_index = 0;
5163
5164         /* misc options */
5165         else if (!strcmp(arg, "-z"))
5166                 options->line_termination = 0;
5167         else if ((argcount = short_opt('l', av, &optarg))) {
5168                 options->rename_limit = strtoul(optarg, NULL, 10);
5169                 return argcount;
5170         }
5171         else if ((argcount = short_opt('S', av, &optarg))) {
5172                 options->pickaxe = optarg;
5173                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5174                 return argcount;
5175         } else if ((argcount = short_opt('G', av, &optarg))) {
5176                 options->pickaxe = optarg;
5177                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5178                 return argcount;
5179         }
5180         else if (!strcmp(arg, "--pickaxe-all"))
5181                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5182         else if (!strcmp(arg, "--pickaxe-regex"))
5183                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5184         else if ((argcount = short_opt('O', av, &optarg))) {
5185                 options->orderfile = prefix_filename(prefix, optarg);
5186                 return argcount;
5187         } else if (skip_prefix(arg, "--find-object=", &arg))
5188                 return parse_objfind_opt(options, arg);
5189         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5190                 int offending = parse_diff_filter_opt(optarg, options);
5191                 if (offending)
5192                         die("unknown change class '%c' in --diff-filter=%s",
5193                             offending, optarg);
5194                 return argcount;
5195         }
5196         else if (!strcmp(arg, "--no-abbrev"))
5197                 options->abbrev = 0;
5198         else if (!strcmp(arg, "--abbrev"))
5199                 options->abbrev = DEFAULT_ABBREV;
5200         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5201                 options->abbrev = strtoul(arg, NULL, 10);
5202                 if (options->abbrev < MINIMUM_ABBREV)
5203                         options->abbrev = MINIMUM_ABBREV;
5204                 else if (the_hash_algo->hexsz < options->abbrev)
5205                         options->abbrev = the_hash_algo->hexsz;
5206         }
5207         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5208                 options->a_prefix = optarg;
5209                 return argcount;
5210         }
5211         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5212                 options->line_prefix = optarg;
5213                 options->line_prefix_length = strlen(options->line_prefix);
5214                 graph_setup_line_prefix(options);
5215                 return argcount;
5216         }
5217         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5218                 options->b_prefix = optarg;
5219                 return argcount;
5220         }
5221         else if (!strcmp(arg, "--no-prefix"))
5222                 options->a_prefix = options->b_prefix = "";
5223         else if (opt_arg(arg, '\0', "inter-hunk-context",
5224                          &options->interhunkcontext))
5225                 ;
5226         else if (!strcmp(arg, "-W"))
5227                 options->flags.funccontext = 1;
5228         else if (!strcmp(arg, "--function-context"))
5229                 options->flags.funccontext = 1;
5230         else if (!strcmp(arg, "--no-function-context"))
5231                 options->flags.funccontext = 0;
5232         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5233                 char *path = prefix_filename(prefix, optarg);
5234                 options->file = xfopen(path, "w");
5235                 options->close_file = 1;
5236                 if (options->use_color != GIT_COLOR_ALWAYS)
5237                         options->use_color = GIT_COLOR_NEVER;
5238                 free(path);
5239                 return argcount;
5240         } else
5241                 return 0;
5242         return 1;
5243 }
5244
5245 int parse_rename_score(const char **cp_p)
5246 {
5247         unsigned long num, scale;
5248         int ch, dot;
5249         const char *cp = *cp_p;
5250
5251         num = 0;
5252         scale = 1;
5253         dot = 0;
5254         for (;;) {
5255                 ch = *cp;
5256                 if ( !dot && ch == '.' ) {
5257                         scale = 1;
5258                         dot = 1;
5259                 } else if ( ch == '%' ) {
5260                         scale = dot ? scale*100 : 100;
5261                         cp++;   /* % is always at the end */
5262                         break;
5263                 } else if ( ch >= '0' && ch <= '9' ) {
5264                         if ( scale < 100000 ) {
5265                                 scale *= 10;
5266                                 num = (num*10) + (ch-'0');
5267                         }
5268                 } else {
5269                         break;
5270                 }
5271                 cp++;
5272         }
5273         *cp_p = cp;
5274
5275         /* user says num divided by scale and we say internally that
5276          * is MAX_SCORE * num / scale.
5277          */
5278         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5279 }
5280
5281 static int diff_scoreopt_parse(const char *opt)
5282 {
5283         int opt1, opt2, cmd;
5284
5285         if (*opt++ != '-')
5286                 return -1;
5287         cmd = *opt++;
5288         if (cmd == '-') {
5289                 /* convert the long-form arguments into short-form versions */
5290                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5291                         if (*opt == 0 || *opt++ == '=')
5292                                 cmd = 'B';
5293                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5294                         if (*opt == 0 || *opt++ == '=')
5295                                 cmd = 'C';
5296                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5297                         if (*opt == 0 || *opt++ == '=')
5298                                 cmd = 'M';
5299                 }
5300         }
5301         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5302                 return -1; /* that is not a -M, -C, or -B option */
5303
5304         opt1 = parse_rename_score(&opt);
5305         if (cmd != 'B')
5306                 opt2 = 0;
5307         else {
5308                 if (*opt == 0)
5309                         opt2 = 0;
5310                 else if (*opt != '/')
5311                         return -1; /* we expect -B80/99 or -B80 */
5312                 else {
5313                         opt++;
5314                         opt2 = parse_rename_score(&opt);
5315                 }
5316         }
5317         if (*opt != 0)
5318                 return -1;
5319         return opt1 | (opt2 << 16);
5320 }
5321
5322 struct diff_queue_struct diff_queued_diff;
5323
5324 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5325 {
5326         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5327         queue->queue[queue->nr++] = dp;
5328 }
5329
5330 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5331                                  struct diff_filespec *one,
5332                                  struct diff_filespec *two)
5333 {
5334         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5335         dp->one = one;
5336         dp->two = two;
5337         if (queue)
5338                 diff_q(queue, dp);
5339         return dp;
5340 }
5341
5342 void diff_free_filepair(struct diff_filepair *p)
5343 {
5344         free_filespec(p->one);
5345         free_filespec(p->two);
5346         free(p);
5347 }
5348
5349 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5350 {
5351         int abblen;
5352         const char *abbrev;
5353
5354         /* Do we want all 40 hex characters? */
5355         if (len == the_hash_algo->hexsz)
5356                 return oid_to_hex(oid);
5357
5358         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5359         abbrev = diff_abbrev_oid(oid, len);
5360
5361         if (!print_sha1_ellipsis())
5362                 return abbrev;
5363
5364         abblen = strlen(abbrev);
5365
5366         /*
5367          * In well-behaved cases, where the abbreviated result is the
5368          * same as the requested length, append three dots after the
5369          * abbreviation (hence the whole logic is limited to the case
5370          * where abblen < 37); when the actual abbreviated result is a
5371          * bit longer than the requested length, we reduce the number
5372          * of dots so that they match the well-behaved ones.  However,
5373          * if the actual abbreviation is longer than the requested
5374          * length by more than three, we give up on aligning, and add
5375          * three dots anyway, to indicate that the output is not the
5376          * full object name.  Yes, this may be suboptimal, but this
5377          * appears only in "diff --raw --abbrev" output and it is not
5378          * worth the effort to change it now.  Note that this would
5379          * likely to work fine when the automatic sizing of default
5380          * abbreviation length is used--we would be fed -1 in "len" in
5381          * that case, and will end up always appending three-dots, but
5382          * the automatic sizing is supposed to give abblen that ensures
5383          * uniqueness across all objects (statistically speaking).
5384          */
5385         if (abblen < the_hash_algo->hexsz - 3) {
5386                 static char hex[GIT_MAX_HEXSZ + 1];
5387                 if (len < abblen && abblen <= len + 2)
5388                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5389                 else
5390                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5391                 return hex;
5392         }
5393
5394         return oid_to_hex(oid);
5395 }
5396
5397 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5398 {
5399         int line_termination = opt->line_termination;
5400         int inter_name_termination = line_termination ? '\t' : '\0';
5401
5402         fprintf(opt->file, "%s", diff_line_prefix(opt));
5403         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5404                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5405                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5406                 fprintf(opt->file, "%s ",
5407                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5408         }
5409         if (p->score) {
5410                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5411                         inter_name_termination);
5412         } else {
5413                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5414         }
5415
5416         if (p->status == DIFF_STATUS_COPIED ||
5417             p->status == DIFF_STATUS_RENAMED) {
5418                 const char *name_a, *name_b;
5419                 name_a = p->one->path;
5420                 name_b = p->two->path;
5421                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5422                 write_name_quoted(name_a, opt->file, inter_name_termination);
5423                 write_name_quoted(name_b, opt->file, line_termination);
5424         } else {
5425                 const char *name_a, *name_b;
5426                 name_a = p->one->mode ? p->one->path : p->two->path;
5427                 name_b = NULL;
5428                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5429                 write_name_quoted(name_a, opt->file, line_termination);
5430         }
5431 }
5432
5433 int diff_unmodified_pair(struct diff_filepair *p)
5434 {
5435         /* This function is written stricter than necessary to support
5436          * the currently implemented transformers, but the idea is to
5437          * let transformers to produce diff_filepairs any way they want,
5438          * and filter and clean them up here before producing the output.
5439          */
5440         struct diff_filespec *one = p->one, *two = p->two;
5441
5442         if (DIFF_PAIR_UNMERGED(p))
5443                 return 0; /* unmerged is interesting */
5444
5445         /* deletion, addition, mode or type change
5446          * and rename are all interesting.
5447          */
5448         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5449             DIFF_PAIR_MODE_CHANGED(p) ||
5450             strcmp(one->path, two->path))
5451                 return 0;
5452
5453         /* both are valid and point at the same path.  that is, we are
5454          * dealing with a change.
5455          */
5456         if (one->oid_valid && two->oid_valid &&
5457             oideq(&one->oid, &two->oid) &&
5458             !one->dirty_submodule && !two->dirty_submodule)
5459                 return 1; /* no change */
5460         if (!one->oid_valid && !two->oid_valid)
5461                 return 1; /* both look at the same file on the filesystem. */
5462         return 0;
5463 }
5464
5465 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5466 {
5467         if (diff_unmodified_pair(p))
5468                 return;
5469
5470         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5471             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5472                 return; /* no tree diffs in patch format */
5473
5474         run_diff(p, o);
5475 }
5476
5477 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5478                             struct diffstat_t *diffstat)
5479 {
5480         if (diff_unmodified_pair(p))
5481                 return;
5482
5483         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5484             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5485                 return; /* no useful stat for tree diffs */
5486
5487         run_diffstat(p, o, diffstat);
5488 }
5489
5490 static void diff_flush_checkdiff(struct diff_filepair *p,
5491                 struct diff_options *o)
5492 {
5493         if (diff_unmodified_pair(p))
5494                 return;
5495
5496         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5497             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5498                 return; /* nothing to check in tree diffs */
5499
5500         run_checkdiff(p, o);
5501 }
5502
5503 int diff_queue_is_empty(void)
5504 {
5505         struct diff_queue_struct *q = &diff_queued_diff;
5506         int i;
5507         for (i = 0; i < q->nr; i++)
5508                 if (!diff_unmodified_pair(q->queue[i]))
5509                         return 0;
5510         return 1;
5511 }
5512
5513 #if DIFF_DEBUG
5514 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5515 {
5516         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5517                 x, one ? one : "",
5518                 s->path,
5519                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5520                 s->mode,
5521                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5522         fprintf(stderr, "queue[%d] %s size %lu\n",
5523                 x, one ? one : "",
5524                 s->size);
5525 }
5526
5527 void diff_debug_filepair(const struct diff_filepair *p, int i)
5528 {
5529         diff_debug_filespec(p->one, i, "one");
5530         diff_debug_filespec(p->two, i, "two");
5531         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5532                 p->score, p->status ? p->status : '?',
5533                 p->one->rename_used, p->broken_pair);
5534 }
5535
5536 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5537 {
5538         int i;
5539         if (msg)
5540                 fprintf(stderr, "%s\n", msg);
5541         fprintf(stderr, "q->nr = %d\n", q->nr);
5542         for (i = 0; i < q->nr; i++) {
5543                 struct diff_filepair *p = q->queue[i];
5544                 diff_debug_filepair(p, i);
5545         }
5546 }
5547 #endif
5548
5549 static void diff_resolve_rename_copy(void)
5550 {
5551         int i;
5552         struct diff_filepair *p;
5553         struct diff_queue_struct *q = &diff_queued_diff;
5554
5555         diff_debug_queue("resolve-rename-copy", q);
5556
5557         for (i = 0; i < q->nr; i++) {
5558                 p = q->queue[i];
5559                 p->status = 0; /* undecided */
5560                 if (DIFF_PAIR_UNMERGED(p))
5561                         p->status = DIFF_STATUS_UNMERGED;
5562                 else if (!DIFF_FILE_VALID(p->one))
5563                         p->status = DIFF_STATUS_ADDED;
5564                 else if (!DIFF_FILE_VALID(p->two))
5565                         p->status = DIFF_STATUS_DELETED;
5566                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5567                         p->status = DIFF_STATUS_TYPE_CHANGED;
5568
5569                 /* from this point on, we are dealing with a pair
5570                  * whose both sides are valid and of the same type, i.e.
5571                  * either in-place edit or rename/copy edit.
5572                  */
5573                 else if (DIFF_PAIR_RENAME(p)) {
5574                         /*
5575                          * A rename might have re-connected a broken
5576                          * pair up, causing the pathnames to be the
5577                          * same again. If so, that's not a rename at
5578                          * all, just a modification..
5579                          *
5580                          * Otherwise, see if this source was used for
5581                          * multiple renames, in which case we decrement
5582                          * the count, and call it a copy.
5583                          */
5584                         if (!strcmp(p->one->path, p->two->path))
5585                                 p->status = DIFF_STATUS_MODIFIED;
5586                         else if (--p->one->rename_used > 0)
5587                                 p->status = DIFF_STATUS_COPIED;
5588                         else
5589                                 p->status = DIFF_STATUS_RENAMED;
5590                 }
5591                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5592                          p->one->mode != p->two->mode ||
5593                          p->one->dirty_submodule ||
5594                          p->two->dirty_submodule ||
5595                          is_null_oid(&p->one->oid))
5596                         p->status = DIFF_STATUS_MODIFIED;
5597                 else {
5598                         /* This is a "no-change" entry and should not
5599                          * happen anymore, but prepare for broken callers.
5600                          */
5601                         error("feeding unmodified %s to diffcore",
5602                               p->one->path);
5603                         p->status = DIFF_STATUS_UNKNOWN;
5604                 }
5605         }
5606         diff_debug_queue("resolve-rename-copy done", q);
5607 }
5608
5609 static int check_pair_status(struct diff_filepair *p)
5610 {
5611         switch (p->status) {
5612         case DIFF_STATUS_UNKNOWN:
5613                 return 0;
5614         case 0:
5615                 die("internal error in diff-resolve-rename-copy");
5616         default:
5617                 return 1;
5618         }
5619 }
5620
5621 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5622 {
5623         int fmt = opt->output_format;
5624
5625         if (fmt & DIFF_FORMAT_CHECKDIFF)
5626                 diff_flush_checkdiff(p, opt);
5627         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5628                 diff_flush_raw(p, opt);
5629         else if (fmt & DIFF_FORMAT_NAME) {
5630                 const char *name_a, *name_b;
5631                 name_a = p->two->path;
5632                 name_b = NULL;
5633                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5634                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5635                 write_name_quoted(name_a, opt->file, opt->line_termination);
5636         }
5637 }
5638
5639 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5640 {
5641         struct strbuf sb = STRBUF_INIT;
5642         if (fs->mode)
5643                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5644         else
5645                 strbuf_addf(&sb, " %s ", newdelete);
5646
5647         quote_c_style(fs->path, &sb, NULL, 0);
5648         strbuf_addch(&sb, '\n');
5649         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5650                          sb.buf, sb.len, 0);
5651         strbuf_release(&sb);
5652 }
5653
5654 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5655                 int show_name)
5656 {
5657         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5658                 struct strbuf sb = STRBUF_INIT;
5659                 strbuf_addf(&sb, " mode change %06o => %06o",
5660                             p->one->mode, p->two->mode);
5661                 if (show_name) {
5662                         strbuf_addch(&sb, ' ');
5663                         quote_c_style(p->two->path, &sb, NULL, 0);
5664                 }
5665                 strbuf_addch(&sb, '\n');
5666                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5667                                  sb.buf, sb.len, 0);
5668                 strbuf_release(&sb);
5669         }
5670 }
5671
5672 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5673                 struct diff_filepair *p)
5674 {
5675         struct strbuf sb = STRBUF_INIT;
5676         struct strbuf names = STRBUF_INIT;
5677
5678         pprint_rename(&names, p->one->path, p->two->path);
5679         strbuf_addf(&sb, " %s %s (%d%%)\n",
5680                     renamecopy, names.buf, similarity_index(p));
5681         strbuf_release(&names);
5682         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5683                                  sb.buf, sb.len, 0);
5684         show_mode_change(opt, p, 0);
5685         strbuf_release(&sb);
5686 }
5687
5688 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5689 {
5690         switch(p->status) {
5691         case DIFF_STATUS_DELETED:
5692                 show_file_mode_name(opt, "delete", p->one);
5693                 break;
5694         case DIFF_STATUS_ADDED:
5695                 show_file_mode_name(opt, "create", p->two);
5696                 break;
5697         case DIFF_STATUS_COPIED:
5698                 show_rename_copy(opt, "copy", p);
5699                 break;
5700         case DIFF_STATUS_RENAMED:
5701                 show_rename_copy(opt, "rename", p);
5702                 break;
5703         default:
5704                 if (p->score) {
5705                         struct strbuf sb = STRBUF_INIT;
5706                         strbuf_addstr(&sb, " rewrite ");
5707                         quote_c_style(p->two->path, &sb, NULL, 0);
5708                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5709                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5710                                          sb.buf, sb.len, 0);
5711                         strbuf_release(&sb);
5712                 }
5713                 show_mode_change(opt, p, !p->score);
5714                 break;
5715         }
5716 }
5717
5718 struct patch_id_t {
5719         git_SHA_CTX *ctx;
5720         int patchlen;
5721 };
5722
5723 static int remove_space(char *line, int len)
5724 {
5725         int i;
5726         char *dst = line;
5727         unsigned char c;
5728
5729         for (i = 0; i < len; i++)
5730                 if (!isspace((c = line[i])))
5731                         *dst++ = c;
5732
5733         return dst - line;
5734 }
5735
5736 static void patch_id_consume(void *priv, char *line, unsigned long len)
5737 {
5738         struct patch_id_t *data = priv;
5739         int new_len;
5740
5741         new_len = remove_space(line, len);
5742
5743         git_SHA1_Update(data->ctx, line, new_len);
5744         data->patchlen += new_len;
5745 }
5746
5747 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5748 {
5749         git_SHA1_Update(ctx, str, strlen(str));
5750 }
5751
5752 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5753 {
5754         /* large enough for 2^32 in octal */
5755         char buf[12];
5756         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5757         git_SHA1_Update(ctx, buf, len);
5758 }
5759
5760 /* returns 0 upon success, and writes result into sha1 */
5761 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5762 {
5763         struct diff_queue_struct *q = &diff_queued_diff;
5764         int i;
5765         git_SHA_CTX ctx;
5766         struct patch_id_t data;
5767
5768         git_SHA1_Init(&ctx);
5769         memset(&data, 0, sizeof(struct patch_id_t));
5770         data.ctx = &ctx;
5771
5772         for (i = 0; i < q->nr; i++) {
5773                 xpparam_t xpp;
5774                 xdemitconf_t xecfg;
5775                 mmfile_t mf1, mf2;
5776                 struct diff_filepair *p = q->queue[i];
5777                 int len1, len2;
5778
5779                 memset(&xpp, 0, sizeof(xpp));
5780                 memset(&xecfg, 0, sizeof(xecfg));
5781                 if (p->status == 0)
5782                         return error("internal diff status error");
5783                 if (p->status == DIFF_STATUS_UNKNOWN)
5784                         continue;
5785                 if (diff_unmodified_pair(p))
5786                         continue;
5787                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5788                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5789                         continue;
5790                 if (DIFF_PAIR_UNMERGED(p))
5791                         continue;
5792
5793                 diff_fill_oid_info(p->one, options->repo->index);
5794                 diff_fill_oid_info(p->two, options->repo->index);
5795
5796                 len1 = remove_space(p->one->path, strlen(p->one->path));
5797                 len2 = remove_space(p->two->path, strlen(p->two->path));
5798                 patch_id_add_string(&ctx, "diff--git");
5799                 patch_id_add_string(&ctx, "a/");
5800                 git_SHA1_Update(&ctx, p->one->path, len1);
5801                 patch_id_add_string(&ctx, "b/");
5802                 git_SHA1_Update(&ctx, p->two->path, len2);
5803
5804                 if (p->one->mode == 0) {
5805                         patch_id_add_string(&ctx, "newfilemode");
5806                         patch_id_add_mode(&ctx, p->two->mode);
5807                         patch_id_add_string(&ctx, "---/dev/null");
5808                         patch_id_add_string(&ctx, "+++b/");
5809                         git_SHA1_Update(&ctx, p->two->path, len2);
5810                 } else if (p->two->mode == 0) {
5811                         patch_id_add_string(&ctx, "deletedfilemode");
5812                         patch_id_add_mode(&ctx, p->one->mode);
5813                         patch_id_add_string(&ctx, "---a/");
5814                         git_SHA1_Update(&ctx, p->one->path, len1);
5815                         patch_id_add_string(&ctx, "+++/dev/null");
5816                 } else {
5817                         patch_id_add_string(&ctx, "---a/");
5818                         git_SHA1_Update(&ctx, p->one->path, len1);
5819                         patch_id_add_string(&ctx, "+++b/");
5820                         git_SHA1_Update(&ctx, p->two->path, len2);
5821                 }
5822
5823                 if (diff_header_only)
5824                         continue;
5825
5826                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5827                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5828                         return error("unable to read files to diff");
5829
5830                 if (diff_filespec_is_binary(options->repo, p->one) ||
5831                     diff_filespec_is_binary(options->repo, p->two)) {
5832                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5833                                         GIT_SHA1_HEXSZ);
5834                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5835                                         GIT_SHA1_HEXSZ);
5836                         continue;
5837                 }
5838
5839                 xpp.flags = 0;
5840                 xecfg.ctxlen = 3;
5841                 xecfg.flags = 0;
5842                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5843                                   patch_id_consume, &data, &xpp, &xecfg))
5844                         return error("unable to generate patch-id diff for %s",
5845                                      p->one->path);
5846         }
5847
5848         git_SHA1_Final(oid->hash, &ctx);
5849         return 0;
5850 }
5851
5852 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5853 {
5854         struct diff_queue_struct *q = &diff_queued_diff;
5855         int i;
5856         int result = diff_get_patch_id(options, oid, diff_header_only);
5857
5858         for (i = 0; i < q->nr; i++)
5859                 diff_free_filepair(q->queue[i]);
5860
5861         free(q->queue);
5862         DIFF_QUEUE_CLEAR(q);
5863
5864         return result;
5865 }
5866
5867 static int is_summary_empty(const struct diff_queue_struct *q)
5868 {
5869         int i;
5870
5871         for (i = 0; i < q->nr; i++) {
5872                 const struct diff_filepair *p = q->queue[i];
5873
5874                 switch (p->status) {
5875                 case DIFF_STATUS_DELETED:
5876                 case DIFF_STATUS_ADDED:
5877                 case DIFF_STATUS_COPIED:
5878                 case DIFF_STATUS_RENAMED:
5879                         return 0;
5880                 default:
5881                         if (p->score)
5882                                 return 0;
5883                         if (p->one->mode && p->two->mode &&
5884                             p->one->mode != p->two->mode)
5885                                 return 0;
5886                         break;
5887                 }
5888         }
5889         return 1;
5890 }
5891
5892 static const char rename_limit_warning[] =
5893 N_("inexact rename detection was skipped due to too many files.");
5894
5895 static const char degrade_cc_to_c_warning[] =
5896 N_("only found copies from modified paths due to too many files.");
5897
5898 static const char rename_limit_advice[] =
5899 N_("you may want to set your %s variable to at least "
5900    "%d and retry the command.");
5901
5902 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5903 {
5904         fflush(stdout);
5905         if (degraded_cc)
5906                 warning(_(degrade_cc_to_c_warning));
5907         else if (needed)
5908                 warning(_(rename_limit_warning));
5909         else
5910                 return;
5911         if (0 < needed)
5912                 warning(_(rename_limit_advice), varname, needed);
5913 }
5914
5915 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5916 {
5917         int i;
5918         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5919         struct diff_queue_struct *q = &diff_queued_diff;
5920
5921         if (WSEH_NEW & WS_RULE_MASK)
5922                 BUG("WS rules bit mask overlaps with diff symbol flags");
5923
5924         if (o->color_moved)
5925                 o->emitted_symbols = &esm;
5926
5927         for (i = 0; i < q->nr; i++) {
5928                 struct diff_filepair *p = q->queue[i];
5929                 if (check_pair_status(p))
5930                         diff_flush_patch(p, o);
5931         }
5932
5933         if (o->emitted_symbols) {
5934                 if (o->color_moved) {
5935                         struct hashmap add_lines, del_lines;
5936
5937                         if (o->color_moved_ws_handling &
5938                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5939                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5940
5941                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5942                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5943
5944                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5945                         mark_color_as_moved(o, &add_lines, &del_lines);
5946                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5947                                 dim_moved_lines(o);
5948
5949                         hashmap_free(&add_lines, 1);
5950                         hashmap_free(&del_lines, 1);
5951                 }
5952
5953                 for (i = 0; i < esm.nr; i++)
5954                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5955
5956                 for (i = 0; i < esm.nr; i++)
5957                         free((void *)esm.buf[i].line);
5958                 esm.nr = 0;
5959
5960                 o->emitted_symbols = NULL;
5961         }
5962 }
5963
5964 void diff_flush(struct diff_options *options)
5965 {
5966         struct diff_queue_struct *q = &diff_queued_diff;
5967         int i, output_format = options->output_format;
5968         int separator = 0;
5969         int dirstat_by_line = 0;
5970
5971         /*
5972          * Order: raw, stat, summary, patch
5973          * or:    name/name-status/checkdiff (other bits clear)
5974          */
5975         if (!q->nr)
5976                 goto free_queue;
5977
5978         if (output_format & (DIFF_FORMAT_RAW |
5979                              DIFF_FORMAT_NAME |
5980                              DIFF_FORMAT_NAME_STATUS |
5981                              DIFF_FORMAT_CHECKDIFF)) {
5982                 for (i = 0; i < q->nr; i++) {
5983                         struct diff_filepair *p = q->queue[i];
5984                         if (check_pair_status(p))
5985                                 flush_one_pair(p, options);
5986                 }
5987                 separator++;
5988         }
5989
5990         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5991                 dirstat_by_line = 1;
5992
5993         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5994             dirstat_by_line) {
5995                 struct diffstat_t diffstat;
5996
5997                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5998                 for (i = 0; i < q->nr; i++) {
5999                         struct diff_filepair *p = q->queue[i];
6000                         if (check_pair_status(p))
6001                                 diff_flush_stat(p, options, &diffstat);
6002                 }
6003                 if (output_format & DIFF_FORMAT_NUMSTAT)
6004                         show_numstat(&diffstat, options);
6005                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6006                         show_stats(&diffstat, options);
6007                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6008                         show_shortstats(&diffstat, options);
6009                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6010                         show_dirstat_by_line(&diffstat, options);
6011                 free_diffstat_info(&diffstat);
6012                 separator++;
6013         }
6014         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6015                 show_dirstat(options);
6016
6017         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6018                 for (i = 0; i < q->nr; i++) {
6019                         diff_summary(options, q->queue[i]);
6020                 }
6021                 separator++;
6022         }
6023
6024         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6025             options->flags.exit_with_status &&
6026             options->flags.diff_from_contents) {
6027                 /*
6028                  * run diff_flush_patch for the exit status. setting
6029                  * options->file to /dev/null should be safe, because we
6030                  * aren't supposed to produce any output anyway.
6031                  */
6032                 if (options->close_file)
6033                         fclose(options->file);
6034                 options->file = xfopen("/dev/null", "w");
6035                 options->close_file = 1;
6036                 options->color_moved = 0;
6037                 for (i = 0; i < q->nr; i++) {
6038                         struct diff_filepair *p = q->queue[i];
6039                         if (check_pair_status(p))
6040                                 diff_flush_patch(p, options);
6041                         if (options->found_changes)
6042                                 break;
6043                 }
6044         }
6045
6046         if (output_format & DIFF_FORMAT_PATCH) {
6047                 if (separator) {
6048                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6049                         if (options->stat_sep)
6050                                 /* attach patch instead of inline */
6051                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6052                                                  NULL, 0, 0);
6053                 }
6054
6055                 diff_flush_patch_all_file_pairs(options);
6056         }
6057
6058         if (output_format & DIFF_FORMAT_CALLBACK)
6059                 options->format_callback(q, options, options->format_callback_data);
6060
6061         for (i = 0; i < q->nr; i++)
6062                 diff_free_filepair(q->queue[i]);
6063 free_queue:
6064         free(q->queue);
6065         DIFF_QUEUE_CLEAR(q);
6066         if (options->close_file)
6067                 fclose(options->file);
6068
6069         /*
6070          * Report the content-level differences with HAS_CHANGES;
6071          * diff_addremove/diff_change does not set the bit when
6072          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6073          */
6074         if (options->flags.diff_from_contents) {
6075                 if (options->found_changes)
6076                         options->flags.has_changes = 1;
6077                 else
6078                         options->flags.has_changes = 0;
6079         }
6080 }
6081
6082 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6083 {
6084         return (((p->status == DIFF_STATUS_MODIFIED) &&
6085                  ((p->score &&
6086                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6087                   (!p->score &&
6088                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6089                 ((p->status != DIFF_STATUS_MODIFIED) &&
6090                  filter_bit_tst(p->status, options)));
6091 }
6092
6093 static void diffcore_apply_filter(struct diff_options *options)
6094 {
6095         int i;
6096         struct diff_queue_struct *q = &diff_queued_diff;
6097         struct diff_queue_struct outq;
6098
6099         DIFF_QUEUE_CLEAR(&outq);
6100
6101         if (!options->filter)
6102                 return;
6103
6104         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6105                 int found;
6106                 for (i = found = 0; !found && i < q->nr; i++) {
6107                         if (match_filter(options, q->queue[i]))
6108                                 found++;
6109                 }
6110                 if (found)
6111                         return;
6112
6113                 /* otherwise we will clear the whole queue
6114                  * by copying the empty outq at the end of this
6115                  * function, but first clear the current entries
6116                  * in the queue.
6117                  */
6118                 for (i = 0; i < q->nr; i++)
6119                         diff_free_filepair(q->queue[i]);
6120         }
6121         else {
6122                 /* Only the matching ones */
6123                 for (i = 0; i < q->nr; i++) {
6124                         struct diff_filepair *p = q->queue[i];
6125                         if (match_filter(options, p))
6126                                 diff_q(&outq, p);
6127                         else
6128                                 diff_free_filepair(p);
6129                 }
6130         }
6131         free(q->queue);
6132         *q = outq;
6133 }
6134
6135 /* Check whether two filespecs with the same mode and size are identical */
6136 static int diff_filespec_is_identical(struct repository *r,
6137                                       struct diff_filespec *one,
6138                                       struct diff_filespec *two)
6139 {
6140         if (S_ISGITLINK(one->mode))
6141                 return 0;
6142         if (diff_populate_filespec(r, one, 0))
6143                 return 0;
6144         if (diff_populate_filespec(r, two, 0))
6145                 return 0;
6146         return !memcmp(one->data, two->data, one->size);
6147 }
6148
6149 static int diff_filespec_check_stat_unmatch(struct repository *r,
6150                                             struct diff_filepair *p)
6151 {
6152         if (p->done_skip_stat_unmatch)
6153                 return p->skip_stat_unmatch_result;
6154
6155         p->done_skip_stat_unmatch = 1;
6156         p->skip_stat_unmatch_result = 0;
6157         /*
6158          * 1. Entries that come from stat info dirtiness
6159          *    always have both sides (iow, not create/delete),
6160          *    one side of the object name is unknown, with
6161          *    the same mode and size.  Keep the ones that
6162          *    do not match these criteria.  They have real
6163          *    differences.
6164          *
6165          * 2. At this point, the file is known to be modified,
6166          *    with the same mode and size, and the object
6167          *    name of one side is unknown.  Need to inspect
6168          *    the identical contents.
6169          */
6170         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6171             !DIFF_FILE_VALID(p->two) ||
6172             (p->one->oid_valid && p->two->oid_valid) ||
6173             (p->one->mode != p->two->mode) ||
6174             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6175             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6176             (p->one->size != p->two->size) ||
6177             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6178                 p->skip_stat_unmatch_result = 1;
6179         return p->skip_stat_unmatch_result;
6180 }
6181
6182 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6183 {
6184         int i;
6185         struct diff_queue_struct *q = &diff_queued_diff;
6186         struct diff_queue_struct outq;
6187         DIFF_QUEUE_CLEAR(&outq);
6188
6189         for (i = 0; i < q->nr; i++) {
6190                 struct diff_filepair *p = q->queue[i];
6191
6192                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6193                         diff_q(&outq, p);
6194                 else {
6195                         /*
6196                          * The caller can subtract 1 from skip_stat_unmatch
6197                          * to determine how many paths were dirty only
6198                          * due to stat info mismatch.
6199                          */
6200                         if (!diffopt->flags.no_index)
6201                                 diffopt->skip_stat_unmatch++;
6202                         diff_free_filepair(p);
6203                 }
6204         }
6205         free(q->queue);
6206         *q = outq;
6207 }
6208
6209 static int diffnamecmp(const void *a_, const void *b_)
6210 {
6211         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6212         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6213         const char *name_a, *name_b;
6214
6215         name_a = a->one ? a->one->path : a->two->path;
6216         name_b = b->one ? b->one->path : b->two->path;
6217         return strcmp(name_a, name_b);
6218 }
6219
6220 void diffcore_fix_diff_index(void)
6221 {
6222         struct diff_queue_struct *q = &diff_queued_diff;
6223         QSORT(q->queue, q->nr, diffnamecmp);
6224 }
6225
6226 void diffcore_std(struct diff_options *options)
6227 {
6228         /* NOTE please keep the following in sync with diff_tree_combined() */
6229         if (options->skip_stat_unmatch)
6230                 diffcore_skip_stat_unmatch(options);
6231         if (!options->found_follow) {
6232                 /* See try_to_follow_renames() in tree-diff.c */
6233                 if (options->break_opt != -1)
6234                         diffcore_break(options->repo,
6235                                        options->break_opt);
6236                 if (options->detect_rename)
6237                         diffcore_rename(options);
6238                 if (options->break_opt != -1)
6239                         diffcore_merge_broken();
6240         }
6241         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6242                 diffcore_pickaxe(options);
6243         if (options->orderfile)
6244                 diffcore_order(options->orderfile);
6245         if (!options->found_follow)
6246                 /* See try_to_follow_renames() in tree-diff.c */
6247                 diff_resolve_rename_copy();
6248         diffcore_apply_filter(options);
6249
6250         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6251                 options->flags.has_changes = 1;
6252         else
6253                 options->flags.has_changes = 0;
6254
6255         options->found_follow = 0;
6256 }
6257
6258 int diff_result_code(struct diff_options *opt, int status)
6259 {
6260         int result = 0;
6261
6262         diff_warn_rename_limit("diff.renameLimit",
6263                                opt->needed_rename_limit,
6264                                opt->degraded_cc_to_c);
6265         if (!opt->flags.exit_with_status &&
6266             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6267                 return status;
6268         if (opt->flags.exit_with_status &&
6269             opt->flags.has_changes)
6270                 result |= 01;
6271         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6272             opt->flags.check_failed)
6273                 result |= 02;
6274         return result;
6275 }
6276
6277 int diff_can_quit_early(struct diff_options *opt)
6278 {
6279         return (opt->flags.quick &&
6280                 !opt->filter &&
6281                 opt->flags.has_changes);
6282 }
6283
6284 /*
6285  * Shall changes to this submodule be ignored?
6286  *
6287  * Submodule changes can be configured to be ignored separately for each path,
6288  * but that configuration can be overridden from the command line.
6289  */
6290 static int is_submodule_ignored(const char *path, struct diff_options *options)
6291 {
6292         int ignored = 0;
6293         struct diff_flags orig_flags = options->flags;
6294         if (!options->flags.override_submodule_config)
6295                 set_diffopt_flags_from_submodule_config(options, path);
6296         if (options->flags.ignore_submodules)
6297                 ignored = 1;
6298         options->flags = orig_flags;
6299         return ignored;
6300 }
6301
6302 void diff_addremove(struct diff_options *options,
6303                     int addremove, unsigned mode,
6304                     const struct object_id *oid,
6305                     int oid_valid,
6306                     const char *concatpath, unsigned dirty_submodule)
6307 {
6308         struct diff_filespec *one, *two;
6309
6310         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6311                 return;
6312
6313         /* This may look odd, but it is a preparation for
6314          * feeding "there are unchanged files which should
6315          * not produce diffs, but when you are doing copy
6316          * detection you would need them, so here they are"
6317          * entries to the diff-core.  They will be prefixed
6318          * with something like '=' or '*' (I haven't decided
6319          * which but should not make any difference).
6320          * Feeding the same new and old to diff_change()
6321          * also has the same effect.
6322          * Before the final output happens, they are pruned after
6323          * merged into rename/copy pairs as appropriate.
6324          */
6325         if (options->flags.reverse_diff)
6326                 addremove = (addremove == '+' ? '-' :
6327                              addremove == '-' ? '+' : addremove);
6328
6329         if (options->prefix &&
6330             strncmp(concatpath, options->prefix, options->prefix_length))
6331                 return;
6332
6333         one = alloc_filespec(concatpath);
6334         two = alloc_filespec(concatpath);
6335
6336         if (addremove != '+')
6337                 fill_filespec(one, oid, oid_valid, mode);
6338         if (addremove != '-') {
6339                 fill_filespec(two, oid, oid_valid, mode);
6340                 two->dirty_submodule = dirty_submodule;
6341         }
6342
6343         diff_queue(&diff_queued_diff, one, two);
6344         if (!options->flags.diff_from_contents)
6345                 options->flags.has_changes = 1;
6346 }
6347
6348 void diff_change(struct diff_options *options,
6349                  unsigned old_mode, unsigned new_mode,
6350                  const struct object_id *old_oid,
6351                  const struct object_id *new_oid,
6352                  int old_oid_valid, int new_oid_valid,
6353                  const char *concatpath,
6354                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6355 {
6356         struct diff_filespec *one, *two;
6357         struct diff_filepair *p;
6358
6359         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6360             is_submodule_ignored(concatpath, options))
6361                 return;
6362
6363         if (options->flags.reverse_diff) {
6364                 SWAP(old_mode, new_mode);
6365                 SWAP(old_oid, new_oid);
6366                 SWAP(old_oid_valid, new_oid_valid);
6367                 SWAP(old_dirty_submodule, new_dirty_submodule);
6368         }
6369
6370         if (options->prefix &&
6371             strncmp(concatpath, options->prefix, options->prefix_length))
6372                 return;
6373
6374         one = alloc_filespec(concatpath);
6375         two = alloc_filespec(concatpath);
6376         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6377         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6378         one->dirty_submodule = old_dirty_submodule;
6379         two->dirty_submodule = new_dirty_submodule;
6380         p = diff_queue(&diff_queued_diff, one, two);
6381
6382         if (options->flags.diff_from_contents)
6383                 return;
6384
6385         if (options->flags.quick && options->skip_stat_unmatch &&
6386             !diff_filespec_check_stat_unmatch(options->repo, p))
6387                 return;
6388
6389         options->flags.has_changes = 1;
6390 }
6391
6392 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6393 {
6394         struct diff_filepair *pair;
6395         struct diff_filespec *one, *two;
6396
6397         if (options->prefix &&
6398             strncmp(path, options->prefix, options->prefix_length))
6399                 return NULL;
6400
6401         one = alloc_filespec(path);
6402         two = alloc_filespec(path);
6403         pair = diff_queue(&diff_queued_diff, one, two);
6404         pair->is_unmerged = 1;
6405         return pair;
6406 }
6407
6408 static char *run_textconv(struct repository *r,
6409                           const char *pgm,
6410                           struct diff_filespec *spec,
6411                           size_t *outsize)
6412 {
6413         struct diff_tempfile *temp;
6414         const char *argv[3];
6415         const char **arg = argv;
6416         struct child_process child = CHILD_PROCESS_INIT;
6417         struct strbuf buf = STRBUF_INIT;
6418         int err = 0;
6419
6420         temp = prepare_temp_file(r, spec->path, spec);
6421         *arg++ = pgm;
6422         *arg++ = temp->name;
6423         *arg = NULL;
6424
6425         child.use_shell = 1;
6426         child.argv = argv;
6427         child.out = -1;
6428         if (start_command(&child)) {
6429                 remove_tempfile();
6430                 return NULL;
6431         }
6432
6433         if (strbuf_read(&buf, child.out, 0) < 0)
6434                 err = error("error reading from textconv command '%s'", pgm);
6435         close(child.out);
6436
6437         if (finish_command(&child) || err) {
6438                 strbuf_release(&buf);
6439                 remove_tempfile();
6440                 return NULL;
6441         }
6442         remove_tempfile();
6443
6444         return strbuf_detach(&buf, outsize);
6445 }
6446
6447 size_t fill_textconv(struct repository *r,
6448                      struct userdiff_driver *driver,
6449                      struct diff_filespec *df,
6450                      char **outbuf)
6451 {
6452         size_t size;
6453
6454         if (!driver) {
6455                 if (!DIFF_FILE_VALID(df)) {
6456                         *outbuf = "";
6457                         return 0;
6458                 }
6459                 if (diff_populate_filespec(r, df, 0))
6460                         die("unable to read files to diff");
6461                 *outbuf = df->data;
6462                 return df->size;
6463         }
6464
6465         if (!driver->textconv)
6466                 BUG("fill_textconv called with non-textconv driver");
6467
6468         if (driver->textconv_cache && df->oid_valid) {
6469                 *outbuf = notes_cache_get(driver->textconv_cache,
6470                                           &df->oid,
6471                                           &size);
6472                 if (*outbuf)
6473                         return size;
6474         }
6475
6476         *outbuf = run_textconv(r, driver->textconv, df, &size);
6477         if (!*outbuf)
6478                 die("unable to read files to diff");
6479
6480         if (driver->textconv_cache && df->oid_valid) {
6481                 /* ignore errors, as we might be in a readonly repository */
6482                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6483                                 size);
6484                 /*
6485                  * we could save up changes and flush them all at the end,
6486                  * but we would need an extra call after all diffing is done.
6487                  * Since generating a cache entry is the slow path anyway,
6488                  * this extra overhead probably isn't a big deal.
6489                  */
6490                 notes_cache_write(driver->textconv_cache);
6491         }
6492
6493         return size;
6494 }
6495
6496 int textconv_object(struct repository *r,
6497                     const char *path,
6498                     unsigned mode,
6499                     const struct object_id *oid,
6500                     int oid_valid,
6501                     char **buf,
6502                     unsigned long *buf_size)
6503 {
6504         struct diff_filespec *df;
6505         struct userdiff_driver *textconv;
6506
6507         df = alloc_filespec(path);
6508         fill_filespec(df, oid, oid_valid, mode);
6509         textconv = get_textconv(r, df);
6510         if (!textconv) {
6511                 free_filespec(df);
6512                 return 0;
6513         }
6514
6515         *buf_size = fill_textconv(r, textconv, df, buf);
6516         free_filespec(df);
6517         return 1;
6518 }
6519
6520 void setup_diff_pager(struct diff_options *opt)
6521 {
6522         /*
6523          * If the user asked for our exit code, then either they want --quiet
6524          * or --exit-code. We should definitely not bother with a pager in the
6525          * former case, as we will generate no output. Since we still properly
6526          * report our exit code even when a pager is run, we _could_ run a
6527          * pager with --exit-code. But since we have not done so historically,
6528          * and because it is easy to find people oneline advising "git diff
6529          * --exit-code" in hooks and other scripts, we do not do so.
6530          */
6531         if (!opt->flags.exit_with_status &&
6532             check_pager_config("diff") != 0)
6533                 setup_pager();
6534 }