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