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