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