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