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