git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed-zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else if (!strcmp(arg, "dimmed_zebra"))
 289                 return COLOR_MOVED_ZEBRA_DIM;
 290         else
 291                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 292 }
 293
 294 static int parse_color_moved_ws(const char *arg)
 295 {
 296         int ret = 0;
 297         struct string_list l = STRING_LIST_INIT_DUP;
 298         struct string_list_item *i;
 299
 300         string_list_split(&l, arg, ',', -1);
 301
 302         for_each_string_list_item(i, &l) {
 303                 struct strbuf sb = STRBUF_INIT;
 304                 strbuf_addstr(&sb, i->string);
 305                 strbuf_trim(&sb);
 306
 307                 if (!strcmp(sb.buf, "ignore-space-change"))
 308                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 309                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 310                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 311                 else if (!strcmp(sb.buf, "ignore-all-space"))
 312                         ret |= XDF_IGNORE_WHITESPACE;
 313                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 314                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 315                 else
 316                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 317
 318                 strbuf_release(&sb);
 319         }
 320
 321         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 322             (ret & XDF_WHITESPACE_FLAGS))
 323                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 324
 325         string_list_clear(&l, 0);
 326
 327         return ret;
 328 }
 329
 330 int git_diff_ui_config(const char *var, const char *value, void *cb)
 331 {
 332         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 333                 diff_use_color_default = git_config_colorbool(var, value);
 334                 return 0;
 335         }
 336         if (!strcmp(var, "diff.colormoved")) {
 337                 int cm = parse_color_moved(value);
 338                 if (cm < 0)
 339                         return -1;
 340                 diff_color_moved_default = cm;
 341                 return 0;
 342         }
 343         if (!strcmp(var, "diff.colormovedws")) {
 344                 int cm = parse_color_moved_ws(value);
 345                 if (cm < 0)
 346                         return -1;
 347                 diff_color_moved_ws_default = cm;
 348                 return 0;
 349         }
 350         if (!strcmp(var, "diff.context")) {
 351                 diff_context_default = git_config_int(var, value);
 352                 if (diff_context_default < 0)
 353                         return -1;
 354                 return 0;
 355         }
 356         if (!strcmp(var, "diff.interhunkcontext")) {
 357                 diff_interhunk_context_default = git_config_int(var, value);
 358                 if (diff_interhunk_context_default < 0)
 359                         return -1;
 360                 return 0;
 361         }
 362         if (!strcmp(var, "diff.renames")) {
 363                 diff_detect_rename_default = git_config_rename(var, value);
 364                 return 0;
 365         }
 366         if (!strcmp(var, "diff.autorefreshindex")) {
 367                 diff_auto_refresh_index = git_config_bool(var, value);
 368                 return 0;
 369         }
 370         if (!strcmp(var, "diff.mnemonicprefix")) {
 371                 diff_mnemonic_prefix = git_config_bool(var, value);
 372                 return 0;
 373         }
 374         if (!strcmp(var, "diff.noprefix")) {
 375                 diff_no_prefix = git_config_bool(var, value);
 376                 return 0;
 377         }
 378         if (!strcmp(var, "diff.statgraphwidth")) {
 379                 diff_stat_graph_width = git_config_int(var, value);
 380                 return 0;
 381         }
 382         if (!strcmp(var, "diff.external"))
 383                 return git_config_string(&external_diff_cmd_cfg, var, value);
 384         if (!strcmp(var, "diff.wordregex"))
 385                 return git_config_string(&diff_word_regex_cfg, var, value);
 386         if (!strcmp(var, "diff.orderfile"))
 387                 return git_config_pathname(&diff_order_file_cfg, var, value);
 388
 389         if (!strcmp(var, "diff.ignoresubmodules"))
 390                 handle_ignore_submodules_arg(&default_diff_options, value);
 391
 392         if (!strcmp(var, "diff.submodule")) {
 393                 if (parse_submodule_params(&default_diff_options, value))
 394                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 395                                 value);
 396                 return 0;
 397         }
 398
 399         if (!strcmp(var, "diff.algorithm")) {
 400                 diff_algorithm = parse_algorithm_value(value);
 401                 if (diff_algorithm < 0)
 402                         return -1;
 403                 return 0;
 404         }
 405
 406         if (!strcmp(var, "diff.wserrorhighlight")) {
 407                 int val = parse_ws_error_highlight(value);
 408                 if (val < 0)
 409                         return -1;
 410                 ws_error_highlight_default = val;
 411                 return 0;
 412         }
 413
 414         if (git_color_config(var, value, cb) < 0)
 415                 return -1;
 416
 417         return git_diff_basic_config(var, value, cb);
 418 }
 419
 420 int git_diff_basic_config(const char *var, const char *value, void *cb)
 421 {
 422         const char *name;
 423
 424         if (!strcmp(var, "diff.renamelimit")) {
 425                 diff_rename_limit_default = git_config_int(var, value);
 426                 return 0;
 427         }
 428
 429         if (userdiff_config(var, value) < 0)
 430                 return -1;
 431
 432         if (skip_prefix(var, "diff.color.", &name) ||
 433             skip_prefix(var, "color.diff.", &name)) {
 434                 int slot = parse_diff_color_slot(name);
 435                 if (slot < 0)
 436                         return 0;
 437                 if (!value)
 438                         return config_error_nonbool(var);
 439                 return color_parse(value, diff_colors[slot]);
 440         }
 441
 442         /* like GNU diff's --suppress-blank-empty option  */
 443         if (!strcmp(var, "diff.suppressblankempty") ||
 444                         /* for backwards compatibility */
 445                         !strcmp(var, "diff.suppress-blank-empty")) {
 446                 diff_suppress_blank_empty = git_config_bool(var, value);
 447                 return 0;
 448         }
 449
 450         if (!strcmp(var, "diff.dirstat")) {
 451                 struct strbuf errmsg = STRBUF_INIT;
 452                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 453                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 454                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 455                                 errmsg.buf);
 456                 strbuf_release(&errmsg);
 457                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 458                 return 0;
 459         }
 460
 461         if (git_diff_heuristic_config(var, value, cb) < 0)
 462                 return -1;
 463
 464         return git_default_config(var, value, cb);
 465 }
 466
 467 static char *quote_two(const char *one, const char *two)
 468 {
 469         int need_one = quote_c_style(one, NULL, NULL, 1);
 470         int need_two = quote_c_style(two, NULL, NULL, 1);
 471         struct strbuf res = STRBUF_INIT;
 472
 473         if (need_one + need_two) {
 474                 strbuf_addch(&res, '"');
 475                 quote_c_style(one, &res, NULL, 1);
 476                 quote_c_style(two, &res, NULL, 1);
 477                 strbuf_addch(&res, '"');
 478         } else {
 479                 strbuf_addstr(&res, one);
 480                 strbuf_addstr(&res, two);
 481         }
 482         return strbuf_detach(&res, NULL);
 483 }
 484
 485 static const char *external_diff(void)
 486 {
 487         static const char *external_diff_cmd = NULL;
 488         static int done_preparing = 0;
 489
 490         if (done_preparing)
 491                 return external_diff_cmd;
 492         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 493         if (!external_diff_cmd)
 494                 external_diff_cmd = external_diff_cmd_cfg;
 495         done_preparing = 1;
 496         return external_diff_cmd;
 497 }
 498
 499 /*
 500  * Keep track of files used for diffing. Sometimes such an entry
 501  * refers to a temporary file, sometimes to an existing file, and
 502  * sometimes to "/dev/null".
 503  */
 504 static struct diff_tempfile {
 505         /*
 506          * filename external diff should read from, or NULL if this
 507          * entry is currently not in use:
 508          */
 509         const char *name;
 510
 511         char hex[GIT_MAX_HEXSZ + 1];
 512         char mode[10];
 513
 514         /*
 515          * If this diff_tempfile instance refers to a temporary file,
 516          * this tempfile object is used to manage its lifetime.
 517          */
 518         struct tempfile *tempfile;
 519 } diff_temp[2];
 520
 521 struct emit_callback {
 522         int color_diff;
 523         unsigned ws_rule;
 524         int blank_at_eof_in_preimage;
 525         int blank_at_eof_in_postimage;
 526         int lno_in_preimage;
 527         int lno_in_postimage;
 528         const char **label_path;
 529         struct diff_words_data *diff_words;
 530         struct diff_options *opt;
 531         struct strbuf *header;
 532 };
 533
 534 static int count_lines(const char *data, int size)
 535 {
 536         int count, ch, completely_empty = 1, nl_just_seen = 0;
 537         count = 0;
 538         while (0 < size--) {
 539                 ch = *data++;
 540                 if (ch == '\n') {
 541                         count++;
 542                         nl_just_seen = 1;
 543                         completely_empty = 0;
 544                 }
 545                 else {
 546                         nl_just_seen = 0;
 547                         completely_empty = 0;
 548                 }
 549         }
 550         if (completely_empty)
 551                 return 0;
 552         if (!nl_just_seen)
 553                 count++; /* no trailing newline */
 554         return count;
 555 }
 556
 557 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 558 {
 559         if (!DIFF_FILE_VALID(one)) {
 560                 mf->ptr = (char *)""; /* does not matter */
 561                 mf->size = 0;
 562                 return 0;
 563         }
 564         else if (diff_populate_filespec(one, 0))
 565                 return -1;
 566
 567         mf->ptr = one->data;
 568         mf->size = one->size;
 569         return 0;
 570 }
 571
 572 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 573 static unsigned long diff_filespec_size(struct diff_filespec *one)
 574 {
 575         if (!DIFF_FILE_VALID(one))
 576                 return 0;
 577         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 578         return one->size;
 579 }
 580
 581 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 582 {
 583         char *ptr = mf->ptr;
 584         long size = mf->size;
 585         int cnt = 0;
 586
 587         if (!size)
 588                 return cnt;
 589         ptr += size - 1; /* pointing at the very end */
 590         if (*ptr != '\n')
 591                 ; /* incomplete line */
 592         else
 593                 ptr--; /* skip the last LF */
 594         while (mf->ptr < ptr) {
 595                 char *prev_eol;
 596                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 597                         if (*prev_eol == '\n')
 598                                 break;
 599                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 600                         break;
 601                 cnt++;
 602                 ptr = prev_eol - 1;
 603         }
 604         return cnt;
 605 }
 606
 607 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 608                                struct emit_callback *ecbdata)
 609 {
 610         int l1, l2, at;
 611         unsigned ws_rule = ecbdata->ws_rule;
 612         l1 = count_trailing_blank(mf1, ws_rule);
 613         l2 = count_trailing_blank(mf2, ws_rule);
 614         if (l2 <= l1) {
 615                 ecbdata->blank_at_eof_in_preimage = 0;
 616                 ecbdata->blank_at_eof_in_postimage = 0;
 617                 return;
 618         }
 619         at = count_lines(mf1->ptr, mf1->size);
 620         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 621
 622         at = count_lines(mf2->ptr, mf2->size);
 623         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 624 }
 625
 626 static void emit_line_0(struct diff_options *o,
 627                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 628                         int first, const char *line, int len)
 629 {
 630         int has_trailing_newline, has_trailing_carriage_return;
 631         int needs_reset = 0; /* at the end of the line */
 632         FILE *file = o->file;
 633
 634         fputs(diff_line_prefix(o), file);
 635
 636         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 637         if (has_trailing_newline)
 638                 len--;
 639
 640         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 641         if (has_trailing_carriage_return)
 642                 len--;
 643
 644         if (!len && !first)
 645                 goto end_of_line;
 646
 647         if (reverse && want_color(o->use_color)) {
 648                 fputs(GIT_COLOR_REVERSE, file);
 649                 needs_reset = 1;
 650         }
 651
 652         if (set_sign) {
 653                 fputs(set_sign, file);
 654                 needs_reset = 1;
 655         }
 656
 657         if (first)
 658                 fputc(first, file);
 659
 660         if (!len)
 661                 goto end_of_line;
 662
 663         if (set) {
 664                 if (set_sign && set != set_sign)
 665                         fputs(reset, file);
 666                 fputs(set, file);
 667                 needs_reset = 1;
 668         }
 669         fwrite(line, len, 1, file);
 670         needs_reset = 1; /* 'line' may contain color codes. */
 671
 672 end_of_line:
 673         if (needs_reset)
 674                 fputs(reset, file);
 675         if (has_trailing_carriage_return)
 676                 fputc('\r', file);
 677         if (has_trailing_newline)
 678                 fputc('\n', file);
 679 }
 680
 681 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 682                       const char *line, int len)
 683 {
 684         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 685 }
 686
 687 enum diff_symbol {
 688         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 689         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 690         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 691         DIFF_SYMBOL_BINARY_DIFF_BODY,
 692         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 693         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 694         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 695         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 696         DIFF_SYMBOL_STATS_LINE,
 697         DIFF_SYMBOL_WORD_DIFF,
 698         DIFF_SYMBOL_STAT_SEP,
 699         DIFF_SYMBOL_SUMMARY,
 700         DIFF_SYMBOL_SUBMODULE_ADD,
 701         DIFF_SYMBOL_SUBMODULE_DEL,
 702         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 703         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 704         DIFF_SYMBOL_SUBMODULE_HEADER,
 705         DIFF_SYMBOL_SUBMODULE_ERROR,
 706         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 707         DIFF_SYMBOL_REWRITE_DIFF,
 708         DIFF_SYMBOL_BINARY_FILES,
 709         DIFF_SYMBOL_HEADER,
 710         DIFF_SYMBOL_FILEPAIR_PLUS,
 711         DIFF_SYMBOL_FILEPAIR_MINUS,
 712         DIFF_SYMBOL_WORDS_PORCELAIN,
 713         DIFF_SYMBOL_WORDS,
 714         DIFF_SYMBOL_CONTEXT,
 715         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 716         DIFF_SYMBOL_PLUS,
 717         DIFF_SYMBOL_MINUS,
 718         DIFF_SYMBOL_NO_LF_EOF,
 719         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 720         DIFF_SYMBOL_CONTEXT_MARKER,
 721         DIFF_SYMBOL_SEPARATOR
 722 };
 723 /*
 724  * Flags for content lines:
 725  * 0..12 are whitespace rules
 726  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 727  * 16 is marking if the line is blank at EOF
 728  */
 729 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 730 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 731 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 732 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 733 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 734
 735 /*
 736  * This struct is used when we need to buffer the output of the diff output.
 737  *
 738  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 739  * into the pre/post image file. This pointer could be a union with the
 740  * line pointer. By storing an offset into the file instead of the literal line,
 741  * we can decrease the memory footprint for the buffered output. At first we
 742  * may want to only have indirection for the content lines, but we could also
 743  * enhance the state for emitting prefabricated lines, e.g. the similarity
 744  * score line or hunk/file headers would only need to store a number or path
 745  * and then the output can be constructed later on depending on state.
 746  */
 747 struct emitted_diff_symbol {
 748         const char *line;
 749         int len;
 750         int flags;
 751         enum diff_symbol s;
 752 };
 753 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 754
 755 struct emitted_diff_symbols {
 756         struct emitted_diff_symbol *buf;
 757         int nr, alloc;
 758 };
 759 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 760
 761 static void append_emitted_diff_symbol(struct diff_options *o,
 762                                        struct emitted_diff_symbol *e)
 763 {
 764         struct emitted_diff_symbol *f;
 765
 766         ALLOC_GROW(o->emitted_symbols->buf,
 767                    o->emitted_symbols->nr + 1,
 768                    o->emitted_symbols->alloc);
 769         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 770
 771         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 772         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 773 }
 774
 775 struct moved_entry {
 776         struct hashmap_entry ent;
 777         const struct emitted_diff_symbol *es;
 778         struct moved_entry *next_line;
 779         struct ws_delta *wsd;
 780 };
 781
 782 /**
 783  * The struct ws_delta holds white space differences between moved lines, i.e.
 784  * between '+' and '-' lines that have been detected to be a move.
 785  * The string contains the difference in leading white spaces, before the
 786  * rest of the line is compared using the white space config for move
 787  * coloring. The current_longer indicates if the first string in the
 788  * comparision is longer than the second.
 789  */
 790 struct ws_delta {
 791         char *string;
 792         unsigned int current_longer : 1;
 793 };
 794 #define WS_DELTA_INIT { NULL, 0 }
 795
 796 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 797                              const struct emitted_diff_symbol *b,
 798                              struct ws_delta *out)
 799 {
 800         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 801         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 802         int d = longer->len - shorter->len;
 803
 804         out->string = xmemdupz(longer->line, d);
 805         out->current_longer = (a == longer);
 806
 807         return !strncmp(longer->line + d, shorter->line, shorter->len);
 808 }
 809
 810 static int cmp_in_block_with_wsd(const struct diff_options *o,
 811                                  const struct moved_entry *cur,
 812                                  const struct moved_entry *match,
 813                                  struct moved_entry *pmb,
 814                                  int n)
 815 {
 816         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 817         int al = cur->es->len, cl = l->len;
 818         const char *a = cur->es->line,
 819                    *b = match->es->line,
 820                    *c = l->line;
 821
 822         int wslen;
 823
 824         /*
 825          * We need to check if 'cur' is equal to 'match'.
 826          * As those are from the same (+/-) side, we do not need to adjust for
 827          * indent changes. However these were found using fuzzy matching
 828          * so we do have to check if they are equal.
 829          */
 830         if (strcmp(a, b))
 831                 return 1;
 832
 833         if (!pmb->wsd)
 834                 /*
 835                  * No white space delta was carried forward? This can happen
 836                  * when we exit early in this function and do not carry
 837                  * forward ws.
 838                  */
 839                 return 1;
 840
 841         /*
 842          * The indent changes of the block are known and carried forward in
 843          * pmb->wsd; however we need to check if the indent changes of the
 844          * current line are still the same as before.
 845          *
 846          * To do so we need to compare 'l' to 'cur', adjusting the
 847          * one of them for the white spaces, depending which was longer.
 848          */
 849
 850         wslen = strlen(pmb->wsd->string);
 851         if (pmb->wsd->current_longer) {
 852                 c += wslen;
 853                 cl -= wslen;
 854         } else {
 855                 a += wslen;
 856                 al -= wslen;
 857         }
 858
 859         if (strcmp(a, c))
 860                 return 1;
 861
 862         return 0;
 863 }
 864
 865 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 866                            const void *entry,
 867                            const void *entry_or_key,
 868                            const void *keydata)
 869 {
 870         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 871         const struct moved_entry *a = entry;
 872         const struct moved_entry *b = entry_or_key;
 873         unsigned flags = diffopt->color_moved_ws_handling
 874                          & XDF_WHITESPACE_FLAGS;
 875
 876         if (diffopt->color_moved_ws_handling &
 877             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 878                 /*
 879                  * As there is not specific white space config given,
 880                  * we'd need to check for a new block, so ignore all
 881                  * white space. The setup of the white space
 882                  * configuration for the next block is done else where
 883                  */
 884                 flags |= XDF_IGNORE_WHITESPACE;
 885
 886         return !xdiff_compare_lines(a->es->line, a->es->len,
 887                                     b->es->line, b->es->len,
 888                                     flags);
 889 }
 890
 891 static struct moved_entry *prepare_entry(struct diff_options *o,
 892                                          int line_no)
 893 {
 894         struct moved_entry *ret = xmalloc(sizeof(*ret));
 895         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 896         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 897
 898         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 899         ret->es = l;
 900         ret->next_line = NULL;
 901         ret->wsd = NULL;
 902
 903         return ret;
 904 }
 905
 906 static void add_lines_to_move_detection(struct diff_options *o,
 907                                         struct hashmap *add_lines,
 908                                         struct hashmap *del_lines)
 909 {
 910         struct moved_entry *prev_line = NULL;
 911
 912         int n;
 913         for (n = 0; n < o->emitted_symbols->nr; n++) {
 914                 struct hashmap *hm;
 915                 struct moved_entry *key;
 916
 917                 switch (o->emitted_symbols->buf[n].s) {
 918                 case DIFF_SYMBOL_PLUS:
 919                         hm = add_lines;
 920                         break;
 921                 case DIFF_SYMBOL_MINUS:
 922                         hm = del_lines;
 923                         break;
 924                 default:
 925                         prev_line = NULL;
 926                         continue;
 927                 }
 928
 929                 key = prepare_entry(o, n);
 930                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 931                         prev_line->next_line = key;
 932
 933                 hashmap_add(hm, key);
 934                 prev_line = key;
 935         }
 936 }
 937
 938 static void pmb_advance_or_null(struct diff_options *o,
 939                                 struct moved_entry *match,
 940                                 struct hashmap *hm,
 941                                 struct moved_entry **pmb,
 942                                 int pmb_nr)
 943 {
 944         int i;
 945         for (i = 0; i < pmb_nr; i++) {
 946                 struct moved_entry *prev = pmb[i];
 947                 struct moved_entry *cur = (prev && prev->next_line) ?
 948                                 prev->next_line : NULL;
 949                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 950                         pmb[i] = cur;
 951                 } else {
 952                         pmb[i] = NULL;
 953                 }
 954         }
 955 }
 956
 957 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 958                                             struct moved_entry *match,
 959                                             struct hashmap *hm,
 960                                             struct moved_entry **pmb,
 961                                             int pmb_nr, int n)
 962 {
 963         int i;
 964         char *got_match = xcalloc(1, pmb_nr);
 965
 966         for (; match; match = hashmap_get_next(hm, match)) {
 967                 for (i = 0; i < pmb_nr; i++) {
 968                         struct moved_entry *prev = pmb[i];
 969                         struct moved_entry *cur = (prev && prev->next_line) ?
 970                                         prev->next_line : NULL;
 971                         if (!cur)
 972                                 continue;
 973                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 974                                 got_match[i] |= 1;
 975                 }
 976         }
 977
 978         for (i = 0; i < pmb_nr; i++) {
 979                 if (got_match[i]) {
 980                         /* Carry the white space delta forward */
 981                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 982                         pmb[i] = pmb[i]->next_line;
 983                 } else
 984                         pmb[i] = NULL;
 985         }
 986 }
 987
 988 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 989                                          int pmb_nr)
 990 {
 991         int lp, rp;
 992
 993         /* Shrink the set of potential block to the remaining running */
 994         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 995                 while (lp < pmb_nr && pmb[lp])
 996                         lp++;
 997                 /* lp points at the first NULL now */
 998
 999                 while (rp > -1 && !pmb[rp])
1000                         rp--;
1001                 /* rp points at the last non-NULL */
1002
1003                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1004                         pmb[lp] = pmb[rp];
1005                         if (pmb[rp]->wsd) {
1006                                 free(pmb[rp]->wsd->string);
1007                                 FREE_AND_NULL(pmb[rp]->wsd);
1008                         }
1009                         pmb[rp] = NULL;
1010                         rp--;
1011                         lp++;
1012                 }
1013         }
1014
1015         /* Remember the number of running sets */
1016         return rp + 1;
1017 }
1018
1019 /*
1020  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1021  *
1022  * Otherwise, if the last block has fewer alphanumeric characters than
1023  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1024  * that block.
1025  *
1026  * The last block consists of the (n - block_length)'th line up to but not
1027  * including the nth line.
1028  *
1029  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1030  * Think of a way to unify them.
1031  */
1032 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1033 {
1034         int i, alnum_count = 0;
1035         if (o->color_moved == COLOR_MOVED_PLAIN)
1036                 return;
1037         for (i = 1; i < block_length + 1; i++) {
1038                 const char *c = o->emitted_symbols->buf[n - i].line;
1039                 for (; *c; c++) {
1040                         if (!isalnum(*c))
1041                                 continue;
1042                         alnum_count++;
1043                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1044                                 return;
1045                 }
1046         }
1047         for (i = 1; i < block_length + 1; i++)
1048                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1049 }
1050
1051 /* Find blocks of moved code, delegate actual coloring decision to helper */
1052 static void mark_color_as_moved(struct diff_options *o,
1053                                 struct hashmap *add_lines,
1054                                 struct hashmap *del_lines)
1055 {
1056         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1057         int pmb_nr = 0, pmb_alloc = 0;
1058         int n, flipped_block = 1, block_length = 0;
1059
1060
1061         for (n = 0; n < o->emitted_symbols->nr; n++) {
1062                 struct hashmap *hm = NULL;
1063                 struct moved_entry *key;
1064                 struct moved_entry *match = NULL;
1065                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1066
1067                 switch (l->s) {
1068                 case DIFF_SYMBOL_PLUS:
1069                         hm = del_lines;
1070                         key = prepare_entry(o, n);
1071                         match = hashmap_get(hm, key, NULL);
1072                         free(key);
1073                         break;
1074                 case DIFF_SYMBOL_MINUS:
1075                         hm = add_lines;
1076                         key = prepare_entry(o, n);
1077                         match = hashmap_get(hm, key, NULL);
1078                         free(key);
1079                         break;
1080                 default:
1081                         flipped_block = 1;
1082                 }
1083
1084                 if (!match) {
1085                         adjust_last_block(o, n, block_length);
1086                         pmb_nr = 0;
1087                         block_length = 0;
1088                         continue;
1089                 }
1090
1091                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1092
1093                 if (o->color_moved == COLOR_MOVED_PLAIN)
1094                         continue;
1095
1096                 if (o->color_moved_ws_handling &
1097                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1098                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1099                 else
1100                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1101
1102                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1103
1104                 if (pmb_nr == 0) {
1105                         /*
1106                          * The current line is the start of a new block.
1107                          * Setup the set of potential blocks.
1108                          */
1109                         for (; match; match = hashmap_get_next(hm, match)) {
1110                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1111                                 if (o->color_moved_ws_handling &
1112                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1113                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1114                                         if (compute_ws_delta(l, match->es, wsd)) {
1115                                                 match->wsd = wsd;
1116                                                 pmb[pmb_nr++] = match;
1117                                         } else
1118                                                 free(wsd);
1119                                 } else {
1120                                         pmb[pmb_nr++] = match;
1121                                 }
1122                         }
1123
1124                         flipped_block = (flipped_block + 1) % 2;
1125
1126                         adjust_last_block(o, n, block_length);
1127                         block_length = 0;
1128                 }
1129
1130                 block_length++;
1131
1132                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1133                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1134         }
1135         adjust_last_block(o, n, block_length);
1136
1137         free(pmb);
1138 }
1139
1140 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1141   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1142 static void dim_moved_lines(struct diff_options *o)
1143 {
1144         int n;
1145         for (n = 0; n < o->emitted_symbols->nr; n++) {
1146                 struct emitted_diff_symbol *prev = (n != 0) ?
1147                                 &o->emitted_symbols->buf[n - 1] : NULL;
1148                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1149                 struct emitted_diff_symbol *next =
1150                                 (n < o->emitted_symbols->nr - 1) ?
1151                                 &o->emitted_symbols->buf[n + 1] : NULL;
1152
1153                 /* Not a plus or minus line? */
1154                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1155                         continue;
1156
1157                 /* Not a moved line? */
1158                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1159                         continue;
1160
1161                 /*
1162                  * If prev or next are not a plus or minus line,
1163                  * pretend they don't exist
1164                  */
1165                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1166                             prev->s != DIFF_SYMBOL_MINUS)
1167                         prev = NULL;
1168                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1169                             next->s != DIFF_SYMBOL_MINUS)
1170                         next = NULL;
1171
1172                 /* Inside a block? */
1173                 if ((prev &&
1174                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1175                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1176                     (next &&
1177                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1178                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1179                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1180                         continue;
1181                 }
1182
1183                 /* Check if we are at an interesting bound: */
1184                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1185                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1186                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1187                         continue;
1188                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1189                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1190                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1191                         continue;
1192
1193                 /*
1194                  * The boundary to prev and next are not interesting,
1195                  * so this line is not interesting as a whole
1196                  */
1197                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1198         }
1199 }
1200
1201 static void emit_line_ws_markup(struct diff_options *o,
1202                                 const char *set_sign, const char *set,
1203                                 const char *reset,
1204                                 char sign, const char *line, int len,
1205                                 unsigned ws_rule, int blank_at_eof)
1206 {
1207         const char *ws = NULL;
1208
1209         if (o->ws_error_highlight & ws_rule) {
1210                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1211                 if (!*ws)
1212                         ws = NULL;
1213         }
1214
1215         if (!ws && !set_sign)
1216                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1217         else if (!ws) {
1218                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1219         } else if (blank_at_eof)
1220                 /* Blank line at EOF - paint '+' as well */
1221                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1222         else {
1223                 /* Emit just the prefix, then the rest. */
1224                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1225                             sign, "", 0);
1226                 ws_check_emit(line, len, ws_rule,
1227                               o->file, set, reset, ws);
1228         }
1229 }
1230
1231 static void emit_diff_symbol_from_struct(struct diff_options *o,
1232                                          struct emitted_diff_symbol *eds)
1233 {
1234         static const char *nneof = " No newline at end of file\n";
1235         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1236         struct strbuf sb = STRBUF_INIT;
1237
1238         enum diff_symbol s = eds->s;
1239         const char *line = eds->line;
1240         int len = eds->len;
1241         unsigned flags = eds->flags;
1242
1243         switch (s) {
1244         case DIFF_SYMBOL_NO_LF_EOF:
1245                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1246                 reset = diff_get_color_opt(o, DIFF_RESET);
1247                 putc('\n', o->file);
1248                 emit_line_0(o, context, NULL, 0, reset, '\\',
1249                             nneof, strlen(nneof));
1250                 break;
1251         case DIFF_SYMBOL_SUBMODULE_HEADER:
1252         case DIFF_SYMBOL_SUBMODULE_ERROR:
1253         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1254         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1255         case DIFF_SYMBOL_SUMMARY:
1256         case DIFF_SYMBOL_STATS_LINE:
1257         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1258         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1259                 emit_line(o, "", "", line, len);
1260                 break;
1261         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1262         case DIFF_SYMBOL_CONTEXT_MARKER:
1263                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1264                 reset = diff_get_color_opt(o, DIFF_RESET);
1265                 emit_line(o, context, reset, line, len);
1266                 break;
1267         case DIFF_SYMBOL_SEPARATOR:
1268                 fprintf(o->file, "%s%c",
1269                         diff_line_prefix(o),
1270                         o->line_termination);
1271                 break;
1272         case DIFF_SYMBOL_CONTEXT:
1273                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1274                 reset = diff_get_color_opt(o, DIFF_RESET);
1275                 set_sign = NULL;
1276                 if (o->flags.dual_color_diffed_diffs) {
1277                         char c = !len ? 0 : line[0];
1278
1279                         if (c == '+')
1280                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1281                         else if (c == '@')
1282                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1283                         else if (c == '-')
1284                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1285                 }
1286                 emit_line_ws_markup(o, set_sign, set, reset,
1287                                     o->output_indicators[OUTPUT_INDICATOR_CONTEXT],
1288                                     line, len,
1289                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1290                 break;
1291         case DIFF_SYMBOL_PLUS:
1292                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1293                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1294                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1295                 case DIFF_SYMBOL_MOVED_LINE |
1296                      DIFF_SYMBOL_MOVED_LINE_ALT |
1297                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1298                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1299                         break;
1300                 case DIFF_SYMBOL_MOVED_LINE |
1301                      DIFF_SYMBOL_MOVED_LINE_ALT:
1302                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1303                         break;
1304                 case DIFF_SYMBOL_MOVED_LINE |
1305                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1306                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1307                         break;
1308                 case DIFF_SYMBOL_MOVED_LINE:
1309                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1310                         break;
1311                 default:
1312                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1313                 }
1314                 reset = diff_get_color_opt(o, DIFF_RESET);
1315                 if (!o->flags.dual_color_diffed_diffs)
1316                         set_sign = NULL;
1317                 else {
1318                         char c = !len ? 0 : line[0];
1319
1320                         set_sign = set;
1321                         if (c == '-')
1322                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1323                         else if (c == '@')
1324                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1325                         else if (c == '+')
1326                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1327                         else
1328                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1329                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1330                 }
1331                 emit_line_ws_markup(o, set_sign, set, reset,
1332                                     o->output_indicators[OUTPUT_INDICATOR_NEW],
1333                                     line, len,
1334                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1335                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1336                 break;
1337         case DIFF_SYMBOL_MINUS:
1338                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1339                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1340                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1341                 case DIFF_SYMBOL_MOVED_LINE |
1342                      DIFF_SYMBOL_MOVED_LINE_ALT |
1343                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1344                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1345                         break;
1346                 case DIFF_SYMBOL_MOVED_LINE |
1347                      DIFF_SYMBOL_MOVED_LINE_ALT:
1348                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1349                         break;
1350                 case DIFF_SYMBOL_MOVED_LINE |
1351                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1352                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1353                         break;
1354                 case DIFF_SYMBOL_MOVED_LINE:
1355                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1356                         break;
1357                 default:
1358                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1359                 }
1360                 reset = diff_get_color_opt(o, DIFF_RESET);
1361                 if (!o->flags.dual_color_diffed_diffs)
1362                         set_sign = NULL;
1363                 else {
1364                         char c = !len ? 0 : line[0];
1365
1366                         set_sign = set;
1367                         if (c == '+')
1368                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1369                         else if (c == '@')
1370                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1371                         else if (c == '-')
1372                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1373                         else
1374                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1375                 }
1376                 emit_line_ws_markup(o, set_sign, set, reset,
1377                                     o->output_indicators[OUTPUT_INDICATOR_OLD],
1378                                     line, len,
1379                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1380                 break;
1381         case DIFF_SYMBOL_WORDS_PORCELAIN:
1382                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1383                 reset = diff_get_color_opt(o, DIFF_RESET);
1384                 emit_line(o, context, reset, line, len);
1385                 fputs("~\n", o->file);
1386                 break;
1387         case DIFF_SYMBOL_WORDS:
1388                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1389                 reset = diff_get_color_opt(o, DIFF_RESET);
1390                 /*
1391                  * Skip the prefix character, if any.  With
1392                  * diff_suppress_blank_empty, there may be
1393                  * none.
1394                  */
1395                 if (line[0] != '\n') {
1396                         line++;
1397                         len--;
1398                 }
1399                 emit_line(o, context, reset, line, len);
1400                 break;
1401         case DIFF_SYMBOL_FILEPAIR_PLUS:
1402                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1403                 reset = diff_get_color_opt(o, DIFF_RESET);
1404                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1405                         line, reset,
1406                         strchr(line, ' ') ? "\t" : "");
1407                 break;
1408         case DIFF_SYMBOL_FILEPAIR_MINUS:
1409                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1410                 reset = diff_get_color_opt(o, DIFF_RESET);
1411                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1412                         line, reset,
1413                         strchr(line, ' ') ? "\t" : "");
1414                 break;
1415         case DIFF_SYMBOL_BINARY_FILES:
1416         case DIFF_SYMBOL_HEADER:
1417                 fprintf(o->file, "%s", line);
1418                 break;
1419         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1420                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1421                 break;
1422         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1423                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1424                 break;
1425         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1426                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1427                 break;
1428         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1429                 fputs(diff_line_prefix(o), o->file);
1430                 fputc('\n', o->file);
1431                 break;
1432         case DIFF_SYMBOL_REWRITE_DIFF:
1433                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1434                 reset = diff_get_color_opt(o, DIFF_RESET);
1435                 emit_line(o, fraginfo, reset, line, len);
1436                 break;
1437         case DIFF_SYMBOL_SUBMODULE_ADD:
1438                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1439                 reset = diff_get_color_opt(o, DIFF_RESET);
1440                 emit_line(o, set, reset, line, len);
1441                 break;
1442         case DIFF_SYMBOL_SUBMODULE_DEL:
1443                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1444                 reset = diff_get_color_opt(o, DIFF_RESET);
1445                 emit_line(o, set, reset, line, len);
1446                 break;
1447         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1448                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1449                         diff_line_prefix(o), line);
1450                 break;
1451         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1452                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1453                         diff_line_prefix(o), line);
1454                 break;
1455         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1456                 emit_line(o, "", "", " 0 files changed\n",
1457                           strlen(" 0 files changed\n"));
1458                 break;
1459         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1460                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1461                 break;
1462         case DIFF_SYMBOL_WORD_DIFF:
1463                 fprintf(o->file, "%.*s", len, line);
1464                 break;
1465         case DIFF_SYMBOL_STAT_SEP:
1466                 fputs(o->stat_sep, o->file);
1467                 break;
1468         default:
1469                 BUG("unknown diff symbol");
1470         }
1471         strbuf_release(&sb);
1472 }
1473
1474 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1475                              const char *line, int len, unsigned flags)
1476 {
1477         struct emitted_diff_symbol e = {line, len, flags, s};
1478
1479         if (o->emitted_symbols)
1480                 append_emitted_diff_symbol(o, &e);
1481         else
1482                 emit_diff_symbol_from_struct(o, &e);
1483 }
1484
1485 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1486 {
1487         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1488 }
1489
1490 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1491 {
1492         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1493 }
1494
1495 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1496 {
1497         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1498                          path, strlen(path), 0);
1499 }
1500
1501 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1502 {
1503         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1504                          path, strlen(path), 0);
1505 }
1506
1507 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1508 {
1509         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1510                          header, strlen(header), 0);
1511 }
1512
1513 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1514 {
1515         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1516 }
1517
1518 void diff_emit_submodule_pipethrough(struct diff_options *o,
1519                                      const char *line, int len)
1520 {
1521         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1522 }
1523
1524 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1525 {
1526         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1527               ecbdata->blank_at_eof_in_preimage &&
1528               ecbdata->blank_at_eof_in_postimage &&
1529               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1530               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1531                 return 0;
1532         return ws_blank_line(line, len, ecbdata->ws_rule);
1533 }
1534
1535 static void emit_add_line(const char *reset,
1536                           struct emit_callback *ecbdata,
1537                           const char *line, int len)
1538 {
1539         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1540         if (new_blank_line_at_eof(ecbdata, line, len))
1541                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1542
1543         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1544 }
1545
1546 static void emit_del_line(const char *reset,
1547                           struct emit_callback *ecbdata,
1548                           const char *line, int len)
1549 {
1550         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1551         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1552 }
1553
1554 static void emit_context_line(const char *reset,
1555                               struct emit_callback *ecbdata,
1556                               const char *line, int len)
1557 {
1558         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1559         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1560 }
1561
1562 static void emit_hunk_header(struct emit_callback *ecbdata,
1563                              const char *line, int len)
1564 {
1565         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1566         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1567         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1568         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1569         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1570         static const char atat[2] = { '@', '@' };
1571         const char *cp, *ep;
1572         struct strbuf msgbuf = STRBUF_INIT;
1573         int org_len = len;
1574         int i = 1;
1575
1576         /*
1577          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1578          * it always is at least 10 bytes long.
1579          */
1580         if (len < 10 ||
1581             memcmp(line, atat, 2) ||
1582             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1583                 emit_diff_symbol(ecbdata->opt,
1584                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1585                 return;
1586         }
1587         ep += 2; /* skip over @@ */
1588
1589         /* The hunk header in fraginfo color */
1590         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1591                 strbuf_addstr(&msgbuf, reverse);
1592         strbuf_addstr(&msgbuf, frag);
1593         strbuf_add(&msgbuf, line, ep - line);
1594         strbuf_addstr(&msgbuf, reset);
1595
1596         /*
1597          * trailing "\r\n"
1598          */
1599         for ( ; i < 3; i++)
1600                 if (line[len - i] == '\r' || line[len - i] == '\n')
1601                         len--;
1602
1603         /* blank before the func header */
1604         for (cp = ep; ep - line < len; ep++)
1605                 if (*ep != ' ' && *ep != '\t')
1606                         break;
1607         if (ep != cp) {
1608                 strbuf_addstr(&msgbuf, context);
1609                 strbuf_add(&msgbuf, cp, ep - cp);
1610                 strbuf_addstr(&msgbuf, reset);
1611         }
1612
1613         if (ep < line + len) {
1614                 strbuf_addstr(&msgbuf, func);
1615                 strbuf_add(&msgbuf, ep, line + len - ep);
1616                 strbuf_addstr(&msgbuf, reset);
1617         }
1618
1619         strbuf_add(&msgbuf, line + len, org_len - len);
1620         strbuf_complete_line(&msgbuf);
1621         emit_diff_symbol(ecbdata->opt,
1622                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1623         strbuf_release(&msgbuf);
1624 }
1625
1626 static struct diff_tempfile *claim_diff_tempfile(void) {
1627         int i;
1628         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1629                 if (!diff_temp[i].name)
1630                         return diff_temp + i;
1631         BUG("diff is failing to clean up its tempfiles");
1632 }
1633
1634 static void remove_tempfile(void)
1635 {
1636         int i;
1637         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1638                 if (is_tempfile_active(diff_temp[i].tempfile))
1639                         delete_tempfile(&diff_temp[i].tempfile);
1640                 diff_temp[i].name = NULL;
1641         }
1642 }
1643
1644 static void add_line_count(struct strbuf *out, int count)
1645 {
1646         switch (count) {
1647         case 0:
1648                 strbuf_addstr(out, "0,0");
1649                 break;
1650         case 1:
1651                 strbuf_addstr(out, "1");
1652                 break;
1653         default:
1654                 strbuf_addf(out, "1,%d", count);
1655                 break;
1656         }
1657 }
1658
1659 static void emit_rewrite_lines(struct emit_callback *ecb,
1660                                int prefix, const char *data, int size)
1661 {
1662         const char *endp = NULL;
1663         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1664
1665         while (0 < size) {
1666                 int len;
1667
1668                 endp = memchr(data, '\n', size);
1669                 len = endp ? (endp - data + 1) : size;
1670                 if (prefix != '+') {
1671                         ecb->lno_in_preimage++;
1672                         emit_del_line(reset, ecb, data, len);
1673                 } else {
1674                         ecb->lno_in_postimage++;
1675                         emit_add_line(reset, ecb, data, len);
1676                 }
1677                 size -= len;
1678                 data += len;
1679         }
1680         if (!endp)
1681                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1682 }
1683
1684 static void emit_rewrite_diff(const char *name_a,
1685                               const char *name_b,
1686                               struct diff_filespec *one,
1687                               struct diff_filespec *two,
1688                               struct userdiff_driver *textconv_one,
1689                               struct userdiff_driver *textconv_two,
1690                               struct diff_options *o)
1691 {
1692         int lc_a, lc_b;
1693         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1694         const char *a_prefix, *b_prefix;
1695         char *data_one, *data_two;
1696         size_t size_one, size_two;
1697         struct emit_callback ecbdata;
1698         struct strbuf out = STRBUF_INIT;
1699
1700         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1701                 a_prefix = o->b_prefix;
1702                 b_prefix = o->a_prefix;
1703         } else {
1704                 a_prefix = o->a_prefix;
1705                 b_prefix = o->b_prefix;
1706         }
1707
1708         name_a += (*name_a == '/');
1709         name_b += (*name_b == '/');
1710
1711         strbuf_reset(&a_name);
1712         strbuf_reset(&b_name);
1713         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1714         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1715
1716         size_one = fill_textconv(textconv_one, one, &data_one);
1717         size_two = fill_textconv(textconv_two, two, &data_two);
1718
1719         memset(&ecbdata, 0, sizeof(ecbdata));
1720         ecbdata.color_diff = want_color(o->use_color);
1721         ecbdata.ws_rule = whitespace_rule(name_b);
1722         ecbdata.opt = o;
1723         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1724                 mmfile_t mf1, mf2;
1725                 mf1.ptr = (char *)data_one;
1726                 mf2.ptr = (char *)data_two;
1727                 mf1.size = size_one;
1728                 mf2.size = size_two;
1729                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1730         }
1731         ecbdata.lno_in_preimage = 1;
1732         ecbdata.lno_in_postimage = 1;
1733
1734         lc_a = count_lines(data_one, size_one);
1735         lc_b = count_lines(data_two, size_two);
1736
1737         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1738                          a_name.buf, a_name.len, 0);
1739         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1740                          b_name.buf, b_name.len, 0);
1741
1742         strbuf_addstr(&out, "@@ -");
1743         if (!o->irreversible_delete)
1744                 add_line_count(&out, lc_a);
1745         else
1746                 strbuf_addstr(&out, "?,?");
1747         strbuf_addstr(&out, " +");
1748         add_line_count(&out, lc_b);
1749         strbuf_addstr(&out, " @@\n");
1750         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1751         strbuf_release(&out);
1752
1753         if (lc_a && !o->irreversible_delete)
1754                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1755         if (lc_b)
1756                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1757         if (textconv_one)
1758                 free((char *)data_one);
1759         if (textconv_two)
1760                 free((char *)data_two);
1761 }
1762
1763 struct diff_words_buffer {
1764         mmfile_t text;
1765         unsigned long alloc;
1766         struct diff_words_orig {
1767                 const char *begin, *end;
1768         } *orig;
1769         int orig_nr, orig_alloc;
1770 };
1771
1772 static void diff_words_append(char *line, unsigned long len,
1773                 struct diff_words_buffer *buffer)
1774 {
1775         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1776         line++;
1777         len--;
1778         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1779         buffer->text.size += len;
1780         buffer->text.ptr[buffer->text.size] = '\0';
1781 }
1782
1783 struct diff_words_style_elem {
1784         const char *prefix;
1785         const char *suffix;
1786         const char *color; /* NULL; filled in by the setup code if
1787                             * color is enabled */
1788 };
1789
1790 struct diff_words_style {
1791         enum diff_words_type type;
1792         struct diff_words_style_elem new_word, old_word, ctx;
1793         const char *newline;
1794 };
1795
1796 static struct diff_words_style diff_words_styles[] = {
1797         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1798         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1799         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1800 };
1801
1802 struct diff_words_data {
1803         struct diff_words_buffer minus, plus;
1804         const char *current_plus;
1805         int last_minus;
1806         struct diff_options *opt;
1807         regex_t *word_regex;
1808         enum diff_words_type type;
1809         struct diff_words_style *style;
1810 };
1811
1812 static int fn_out_diff_words_write_helper(struct diff_options *o,
1813                                           struct diff_words_style_elem *st_el,
1814                                           const char *newline,
1815                                           size_t count, const char *buf)
1816 {
1817         int print = 0;
1818         struct strbuf sb = STRBUF_INIT;
1819
1820         while (count) {
1821                 char *p = memchr(buf, '\n', count);
1822                 if (print)
1823                         strbuf_addstr(&sb, diff_line_prefix(o));
1824
1825                 if (p != buf) {
1826                         const char *reset = st_el->color && *st_el->color ?
1827                                             GIT_COLOR_RESET : NULL;
1828                         if (st_el->color && *st_el->color)
1829                                 strbuf_addstr(&sb, st_el->color);
1830                         strbuf_addstr(&sb, st_el->prefix);
1831                         strbuf_add(&sb, buf, p ? p - buf : count);
1832                         strbuf_addstr(&sb, st_el->suffix);
1833                         if (reset)
1834                                 strbuf_addstr(&sb, reset);
1835                 }
1836                 if (!p)
1837                         goto out;
1838
1839                 strbuf_addstr(&sb, newline);
1840                 count -= p + 1 - buf;
1841                 buf = p + 1;
1842                 print = 1;
1843                 if (count) {
1844                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1845                                          sb.buf, sb.len, 0);
1846                         strbuf_reset(&sb);
1847                 }
1848         }
1849
1850 out:
1851         if (sb.len)
1852                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1853                                  sb.buf, sb.len, 0);
1854         strbuf_release(&sb);
1855         return 0;
1856 }
1857
1858 /*
1859  * '--color-words' algorithm can be described as:
1860  *
1861  *   1. collect the minus/plus lines of a diff hunk, divided into
1862  *      minus-lines and plus-lines;
1863  *
1864  *   2. break both minus-lines and plus-lines into words and
1865  *      place them into two mmfile_t with one word for each line;
1866  *
1867  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1868  *
1869  * And for the common parts of the both file, we output the plus side text.
1870  * diff_words->current_plus is used to trace the current position of the plus file
1871  * which printed. diff_words->last_minus is used to trace the last minus word
1872  * printed.
1873  *
1874  * For '--graph' to work with '--color-words', we need to output the graph prefix
1875  * on each line of color words output. Generally, there are two conditions on
1876  * which we should output the prefix.
1877  *
1878  *   1. diff_words->last_minus == 0 &&
1879  *      diff_words->current_plus == diff_words->plus.text.ptr
1880  *
1881  *      that is: the plus text must start as a new line, and if there is no minus
1882  *      word printed, a graph prefix must be printed.
1883  *
1884  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1885  *      *(diff_words->current_plus - 1) == '\n'
1886  *
1887  *      that is: a graph prefix must be printed following a '\n'
1888  */
1889 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1890 {
1891         if ((diff_words->last_minus == 0 &&
1892                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1893                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1894                 *(diff_words->current_plus - 1) == '\n')) {
1895                 return 1;
1896         } else {
1897                 return 0;
1898         }
1899 }
1900
1901 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1902 {
1903         struct diff_words_data *diff_words = priv;
1904         struct diff_words_style *style = diff_words->style;
1905         int minus_first, minus_len, plus_first, plus_len;
1906         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1907         struct diff_options *opt = diff_words->opt;
1908         const char *line_prefix;
1909
1910         if (line[0] != '@' || parse_hunk_header(line, len,
1911                         &minus_first, &minus_len, &plus_first, &plus_len))
1912                 return;
1913
1914         assert(opt);
1915         line_prefix = diff_line_prefix(opt);
1916
1917         /* POSIX requires that first be decremented by one if len == 0... */
1918         if (minus_len) {
1919                 minus_begin = diff_words->minus.orig[minus_first].begin;
1920                 minus_end =
1921                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1922         } else
1923                 minus_begin = minus_end =
1924                         diff_words->minus.orig[minus_first].end;
1925
1926         if (plus_len) {
1927                 plus_begin = diff_words->plus.orig[plus_first].begin;
1928                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1929         } else
1930                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1931
1932         if (color_words_output_graph_prefix(diff_words)) {
1933                 fputs(line_prefix, diff_words->opt->file);
1934         }
1935         if (diff_words->current_plus != plus_begin) {
1936                 fn_out_diff_words_write_helper(diff_words->opt,
1937                                 &style->ctx, style->newline,
1938                                 plus_begin - diff_words->current_plus,
1939                                 diff_words->current_plus);
1940         }
1941         if (minus_begin != minus_end) {
1942                 fn_out_diff_words_write_helper(diff_words->opt,
1943                                 &style->old_word, style->newline,
1944                                 minus_end - minus_begin, minus_begin);
1945         }
1946         if (plus_begin != plus_end) {
1947                 fn_out_diff_words_write_helper(diff_words->opt,
1948                                 &style->new_word, style->newline,
1949                                 plus_end - plus_begin, plus_begin);
1950         }
1951
1952         diff_words->current_plus = plus_end;
1953         diff_words->last_minus = minus_first;
1954 }
1955
1956 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1957 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1958                 int *begin, int *end)
1959 {
1960         if (word_regex && *begin < buffer->size) {
1961                 regmatch_t match[1];
1962                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1963                                  buffer->size - *begin, 1, match, 0)) {
1964                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1965                                         '\n', match[0].rm_eo - match[0].rm_so);
1966                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1967                         *begin += match[0].rm_so;
1968                         return *begin >= *end;
1969                 }
1970                 return -1;
1971         }
1972
1973         /* find the next word */
1974         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1975                 (*begin)++;
1976         if (*begin >= buffer->size)
1977                 return -1;
1978
1979         /* find the end of the word */
1980         *end = *begin + 1;
1981         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1982                 (*end)++;
1983
1984         return 0;
1985 }
1986
1987 /*
1988  * This function splits the words in buffer->text, stores the list with
1989  * newline separator into out, and saves the offsets of the original words
1990  * in buffer->orig.
1991  */
1992 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1993                 regex_t *word_regex)
1994 {
1995         int i, j;
1996         long alloc = 0;
1997
1998         out->size = 0;
1999         out->ptr = NULL;
2000
2001         /* fake an empty "0th" word */
2002         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2003         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2004         buffer->orig_nr = 1;
2005
2006         for (i = 0; i < buffer->text.size; i++) {
2007                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2008                         return;
2009
2010                 /* store original boundaries */
2011                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2012                                 buffer->orig_alloc);
2013                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2014                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2015                 buffer->orig_nr++;
2016
2017                 /* store one word */
2018                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2019                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2020                 out->ptr[out->size + j - i] = '\n';
2021                 out->size += j - i + 1;
2022
2023                 i = j - 1;
2024         }
2025 }
2026
2027 /* this executes the word diff on the accumulated buffers */
2028 static void diff_words_show(struct diff_words_data *diff_words)
2029 {
2030         xpparam_t xpp;
2031         xdemitconf_t xecfg;
2032         mmfile_t minus, plus;
2033         struct diff_words_style *style = diff_words->style;
2034
2035         struct diff_options *opt = diff_words->opt;
2036         const char *line_prefix;
2037
2038         assert(opt);
2039         line_prefix = diff_line_prefix(opt);
2040
2041         /* special case: only removal */
2042         if (!diff_words->plus.text.size) {
2043                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2044                                  line_prefix, strlen(line_prefix), 0);
2045                 fn_out_diff_words_write_helper(diff_words->opt,
2046                         &style->old_word, style->newline,
2047                         diff_words->minus.text.size,
2048                         diff_words->minus.text.ptr);
2049                 diff_words->minus.text.size = 0;
2050                 return;
2051         }
2052
2053         diff_words->current_plus = diff_words->plus.text.ptr;
2054         diff_words->last_minus = 0;
2055
2056         memset(&xpp, 0, sizeof(xpp));
2057         memset(&xecfg, 0, sizeof(xecfg));
2058         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2059         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2060         xpp.flags = 0;
2061         /* as only the hunk header will be parsed, we need a 0-context */
2062         xecfg.ctxlen = 0;
2063         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2064                           &xpp, &xecfg))
2065                 die("unable to generate word diff");
2066         free(minus.ptr);
2067         free(plus.ptr);
2068         if (diff_words->current_plus != diff_words->plus.text.ptr +
2069                         diff_words->plus.text.size) {
2070                 if (color_words_output_graph_prefix(diff_words))
2071                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2072                                          line_prefix, strlen(line_prefix), 0);
2073                 fn_out_diff_words_write_helper(diff_words->opt,
2074                         &style->ctx, style->newline,
2075                         diff_words->plus.text.ptr + diff_words->plus.text.size
2076                         - diff_words->current_plus, diff_words->current_plus);
2077         }
2078         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2079 }
2080
2081 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2082 static void diff_words_flush(struct emit_callback *ecbdata)
2083 {
2084         struct diff_options *wo = ecbdata->diff_words->opt;
2085
2086         if (ecbdata->diff_words->minus.text.size ||
2087             ecbdata->diff_words->plus.text.size)
2088                 diff_words_show(ecbdata->diff_words);
2089
2090         if (wo->emitted_symbols) {
2091                 struct diff_options *o = ecbdata->opt;
2092                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2093                 int i;
2094
2095                 /*
2096                  * NEEDSWORK:
2097                  * Instead of appending each, concat all words to a line?
2098                  */
2099                 for (i = 0; i < wol->nr; i++)
2100                         append_emitted_diff_symbol(o, &wol->buf[i]);
2101
2102                 for (i = 0; i < wol->nr; i++)
2103                         free((void *)wol->buf[i].line);
2104
2105                 wol->nr = 0;
2106         }
2107 }
2108
2109 static void diff_filespec_load_driver(struct diff_filespec *one)
2110 {
2111         /* Use already-loaded driver */
2112         if (one->driver)
2113                 return;
2114
2115         if (S_ISREG(one->mode))
2116                 one->driver = userdiff_find_by_path(one->path);
2117
2118         /* Fallback to default settings */
2119         if (!one->driver)
2120                 one->driver = userdiff_find_by_name("default");
2121 }
2122
2123 static const char *userdiff_word_regex(struct diff_filespec *one)
2124 {
2125         diff_filespec_load_driver(one);
2126         return one->driver->word_regex;
2127 }
2128
2129 static void init_diff_words_data(struct emit_callback *ecbdata,
2130                                  struct diff_options *orig_opts,
2131                                  struct diff_filespec *one,
2132                                  struct diff_filespec *two)
2133 {
2134         int i;
2135         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2136         memcpy(o, orig_opts, sizeof(struct diff_options));
2137
2138         ecbdata->diff_words =
2139                 xcalloc(1, sizeof(struct diff_words_data));
2140         ecbdata->diff_words->type = o->word_diff;
2141         ecbdata->diff_words->opt = o;
2142
2143         if (orig_opts->emitted_symbols)
2144                 o->emitted_symbols =
2145                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2146
2147         if (!o->word_regex)
2148                 o->word_regex = userdiff_word_regex(one);
2149         if (!o->word_regex)
2150                 o->word_regex = userdiff_word_regex(two);
2151         if (!o->word_regex)
2152                 o->word_regex = diff_word_regex_cfg;
2153         if (o->word_regex) {
2154                 ecbdata->diff_words->word_regex = (regex_t *)
2155                         xmalloc(sizeof(regex_t));
2156                 if (regcomp(ecbdata->diff_words->word_regex,
2157                             o->word_regex,
2158                             REG_EXTENDED | REG_NEWLINE))
2159                         die("invalid regular expression: %s",
2160                             o->word_regex);
2161         }
2162         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2163                 if (o->word_diff == diff_words_styles[i].type) {
2164                         ecbdata->diff_words->style =
2165                                 &diff_words_styles[i];
2166                         break;
2167                 }
2168         }
2169         if (want_color(o->use_color)) {
2170                 struct diff_words_style *st = ecbdata->diff_words->style;
2171                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2172                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2173                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2174         }
2175 }
2176
2177 static void free_diff_words_data(struct emit_callback *ecbdata)
2178 {
2179         if (ecbdata->diff_words) {
2180                 diff_words_flush(ecbdata);
2181                 free (ecbdata->diff_words->opt->emitted_symbols);
2182                 free (ecbdata->diff_words->opt);
2183                 free (ecbdata->diff_words->minus.text.ptr);
2184                 free (ecbdata->diff_words->minus.orig);
2185                 free (ecbdata->diff_words->plus.text.ptr);
2186                 free (ecbdata->diff_words->plus.orig);
2187                 if (ecbdata->diff_words->word_regex) {
2188                         regfree(ecbdata->diff_words->word_regex);
2189                         free(ecbdata->diff_words->word_regex);
2190                 }
2191                 FREE_AND_NULL(ecbdata->diff_words);
2192         }
2193 }
2194
2195 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2196 {
2197         if (want_color(diff_use_color))
2198                 return diff_colors[ix];
2199         return "";
2200 }
2201
2202 const char *diff_line_prefix(struct diff_options *opt)
2203 {
2204         struct strbuf *msgbuf;
2205         if (!opt->output_prefix)
2206                 return "";
2207
2208         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2209         return msgbuf->buf;
2210 }
2211
2212 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2213 {
2214         const char *cp;
2215         unsigned long allot;
2216         size_t l = len;
2217
2218         cp = line;
2219         allot = l;
2220         while (0 < l) {
2221                 (void) utf8_width(&cp, &l);
2222                 if (!cp)
2223                         break; /* truncated in the middle? */
2224         }
2225         return allot - l;
2226 }
2227
2228 static void find_lno(const char *line, struct emit_callback *ecbdata)
2229 {
2230         const char *p;
2231         ecbdata->lno_in_preimage = 0;
2232         ecbdata->lno_in_postimage = 0;
2233         p = strchr(line, '-');
2234         if (!p)
2235                 return; /* cannot happen */
2236         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2237         p = strchr(p, '+');
2238         if (!p)
2239                 return; /* cannot happen */
2240         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2241 }
2242
2243 static void fn_out_consume(void *priv, char *line, unsigned long len)
2244 {
2245         struct emit_callback *ecbdata = priv;
2246         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2247         struct diff_options *o = ecbdata->opt;
2248
2249         o->found_changes = 1;
2250
2251         if (ecbdata->header) {
2252                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2253                                  ecbdata->header->buf, ecbdata->header->len, 0);
2254                 strbuf_reset(ecbdata->header);
2255                 ecbdata->header = NULL;
2256         }
2257
2258         if (ecbdata->label_path[0]) {
2259                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2260                                  ecbdata->label_path[0],
2261                                  strlen(ecbdata->label_path[0]), 0);
2262                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2263                                  ecbdata->label_path[1],
2264                                  strlen(ecbdata->label_path[1]), 0);
2265                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2266         }
2267
2268         if (diff_suppress_blank_empty
2269             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2270                 line[0] = '\n';
2271                 len = 1;
2272         }
2273
2274         if (line[0] == '@') {
2275                 if (ecbdata->diff_words)
2276                         diff_words_flush(ecbdata);
2277                 len = sane_truncate_line(ecbdata, line, len);
2278                 find_lno(line, ecbdata);
2279                 emit_hunk_header(ecbdata, line, len);
2280                 return;
2281         }
2282
2283         if (ecbdata->diff_words) {
2284                 enum diff_symbol s =
2285                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2286                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2287                 if (line[0] == '-') {
2288                         diff_words_append(line, len,
2289                                           &ecbdata->diff_words->minus);
2290                         return;
2291                 } else if (line[0] == '+') {
2292                         diff_words_append(line, len,
2293                                           &ecbdata->diff_words->plus);
2294                         return;
2295                 } else if (starts_with(line, "\\ ")) {
2296                         /*
2297                          * Eat the "no newline at eof" marker as if we
2298                          * saw a "+" or "-" line with nothing on it,
2299                          * and return without diff_words_flush() to
2300                          * defer processing. If this is the end of
2301                          * preimage, more "+" lines may come after it.
2302                          */
2303                         return;
2304                 }
2305                 diff_words_flush(ecbdata);
2306                 emit_diff_symbol(o, s, line, len, 0);
2307                 return;
2308         }
2309
2310         switch (line[0]) {
2311         case '+':
2312                 ecbdata->lno_in_postimage++;
2313                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2314                 break;
2315         case '-':
2316                 ecbdata->lno_in_preimage++;
2317                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2318                 break;
2319         case ' ':
2320                 ecbdata->lno_in_postimage++;
2321                 ecbdata->lno_in_preimage++;
2322                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2323                 break;
2324         default:
2325                 /* incomplete line at the end */
2326                 ecbdata->lno_in_preimage++;
2327                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2328                                  line, len, 0);
2329                 break;
2330         }
2331 }
2332
2333 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2334 {
2335         const char *old_name = a;
2336         const char *new_name = b;
2337         int pfx_length, sfx_length;
2338         int pfx_adjust_for_slash;
2339         int len_a = strlen(a);
2340         int len_b = strlen(b);
2341         int a_midlen, b_midlen;
2342         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2343         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2344
2345         if (qlen_a || qlen_b) {
2346                 quote_c_style(a, name, NULL, 0);
2347                 strbuf_addstr(name, " => ");
2348                 quote_c_style(b, name, NULL, 0);
2349                 return;
2350         }
2351
2352         /* Find common prefix */
2353         pfx_length = 0;
2354         while (*old_name && *new_name && *old_name == *new_name) {
2355                 if (*old_name == '/')
2356                         pfx_length = old_name - a + 1;
2357                 old_name++;
2358                 new_name++;
2359         }
2360
2361         /* Find common suffix */
2362         old_name = a + len_a;
2363         new_name = b + len_b;
2364         sfx_length = 0;
2365         /*
2366          * If there is a common prefix, it must end in a slash.  In
2367          * that case we let this loop run 1 into the prefix to see the
2368          * same slash.
2369          *
2370          * If there is no common prefix, we cannot do this as it would
2371          * underrun the input strings.
2372          */
2373         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2374         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2375                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2376                *old_name == *new_name) {
2377                 if (*old_name == '/')
2378                         sfx_length = len_a - (old_name - a);
2379                 old_name--;
2380                 new_name--;
2381         }
2382
2383         /*
2384          * pfx{mid-a => mid-b}sfx
2385          * {pfx-a => pfx-b}sfx
2386          * pfx{sfx-a => sfx-b}
2387          * name-a => name-b
2388          */
2389         a_midlen = len_a - pfx_length - sfx_length;
2390         b_midlen = len_b - pfx_length - sfx_length;
2391         if (a_midlen < 0)
2392                 a_midlen = 0;
2393         if (b_midlen < 0)
2394                 b_midlen = 0;
2395
2396         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2397         if (pfx_length + sfx_length) {
2398                 strbuf_add(name, a, pfx_length);
2399                 strbuf_addch(name, '{');
2400         }
2401         strbuf_add(name, a + pfx_length, a_midlen);
2402         strbuf_addstr(name, " => ");
2403         strbuf_add(name, b + pfx_length, b_midlen);
2404         if (pfx_length + sfx_length) {
2405                 strbuf_addch(name, '}');
2406                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2407         }
2408 }
2409
2410 struct diffstat_t {
2411         int nr;
2412         int alloc;
2413         struct diffstat_file {
2414                 char *from_name;
2415                 char *name;
2416                 char *print_name;
2417                 const char *comments;
2418                 unsigned is_unmerged:1;
2419                 unsigned is_binary:1;
2420                 unsigned is_renamed:1;
2421                 unsigned is_interesting:1;
2422                 uintmax_t added, deleted;
2423         } **files;
2424 };
2425
2426 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2427                                           const char *name_a,
2428                                           const char *name_b)
2429 {
2430         struct diffstat_file *x;
2431         x = xcalloc(1, sizeof(*x));
2432         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2433         diffstat->files[diffstat->nr++] = x;
2434         if (name_b) {
2435                 x->from_name = xstrdup(name_a);
2436                 x->name = xstrdup(name_b);
2437                 x->is_renamed = 1;
2438         }
2439         else {
2440                 x->from_name = NULL;
2441                 x->name = xstrdup(name_a);
2442         }
2443         return x;
2444 }
2445
2446 static void diffstat_consume(void *priv, char *line, unsigned long len)
2447 {
2448         struct diffstat_t *diffstat = priv;
2449         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2450
2451         if (line[0] == '+')
2452                 x->added++;
2453         else if (line[0] == '-')
2454                 x->deleted++;
2455 }
2456
2457 const char mime_boundary_leader[] = "------------";
2458
2459 static int scale_linear(int it, int width, int max_change)
2460 {
2461         if (!it)
2462                 return 0;
2463         /*
2464          * make sure that at least one '-' or '+' is printed if
2465          * there is any change to this path. The easiest way is to
2466          * scale linearly as if the alloted width is one column shorter
2467          * than it is, and then add 1 to the result.
2468          */
2469         return 1 + (it * (width - 1) / max_change);
2470 }
2471
2472 static void show_graph(struct strbuf *out, char ch, int cnt,
2473                        const char *set, const char *reset)
2474 {
2475         if (cnt <= 0)
2476                 return;
2477         strbuf_addstr(out, set);
2478         strbuf_addchars(out, ch, cnt);
2479         strbuf_addstr(out, reset);
2480 }
2481
2482 static void fill_print_name(struct diffstat_file *file)
2483 {
2484         struct strbuf pname = STRBUF_INIT;
2485
2486         if (file->print_name)
2487                 return;
2488
2489         if (file->is_renamed)
2490                 pprint_rename(&pname, file->from_name, file->name);
2491         else
2492                 quote_c_style(file->name, &pname, NULL, 0);
2493
2494         if (file->comments)
2495                 strbuf_addf(&pname, " (%s)", file->comments);
2496
2497         file->print_name = strbuf_detach(&pname, NULL);
2498 }
2499
2500 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2501                 int files, int insertions, int deletions)
2502 {
2503         struct strbuf sb = STRBUF_INIT;
2504
2505         if (!files) {
2506                 assert(insertions == 0 && deletions == 0);
2507                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2508                                  NULL, 0, 0);
2509                 return;
2510         }
2511
2512         strbuf_addf(&sb,
2513                     (files == 1) ? " %d file changed" : " %d files changed",
2514                     files);
2515
2516         /*
2517          * For binary diff, the caller may want to print "x files
2518          * changed" with insertions == 0 && deletions == 0.
2519          *
2520          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2521          * is probably less confusing (i.e skip over "2 files changed
2522          * but nothing about added/removed lines? Is this a bug in Git?").
2523          */
2524         if (insertions || deletions == 0) {
2525                 strbuf_addf(&sb,
2526                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2527                             insertions);
2528         }
2529
2530         if (deletions || insertions == 0) {
2531                 strbuf_addf(&sb,
2532                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2533                             deletions);
2534         }
2535         strbuf_addch(&sb, '\n');
2536         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2537                          sb.buf, sb.len, 0);
2538         strbuf_release(&sb);
2539 }
2540
2541 void print_stat_summary(FILE *fp, int files,
2542                         int insertions, int deletions)
2543 {
2544         struct diff_options o;
2545         memset(&o, 0, sizeof(o));
2546         o.file = fp;
2547
2548         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2549 }
2550
2551 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2552 {
2553         int i, len, add, del, adds = 0, dels = 0;
2554         uintmax_t max_change = 0, max_len = 0;
2555         int total_files = data->nr, count;
2556         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2557         const char *reset, *add_c, *del_c;
2558         int extra_shown = 0;
2559         const char *line_prefix = diff_line_prefix(options);
2560         struct strbuf out = STRBUF_INIT;
2561
2562         if (data->nr == 0)
2563                 return;
2564
2565         count = options->stat_count ? options->stat_count : data->nr;
2566
2567         reset = diff_get_color_opt(options, DIFF_RESET);
2568         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2569         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2570
2571         /*
2572          * Find the longest filename and max number of changes
2573          */
2574         for (i = 0; (i < count) && (i < data->nr); i++) {
2575                 struct diffstat_file *file = data->files[i];
2576                 uintmax_t change = file->added + file->deleted;
2577
2578                 if (!file->is_interesting && (change == 0)) {
2579                         count++; /* not shown == room for one more */
2580                         continue;
2581                 }
2582                 fill_print_name(file);
2583                 len = strlen(file->print_name);
2584                 if (max_len < len)
2585                         max_len = len;
2586
2587                 if (file->is_unmerged) {
2588                         /* "Unmerged" is 8 characters */
2589                         bin_width = bin_width < 8 ? 8 : bin_width;
2590                         continue;
2591                 }
2592                 if (file->is_binary) {
2593                         /* "Bin XXX -> YYY bytes" */
2594                         int w = 14 + decimal_width(file->added)
2595                                 + decimal_width(file->deleted);
2596                         bin_width = bin_width < w ? w : bin_width;
2597                         /* Display change counts aligned with "Bin" */
2598                         number_width = 3;
2599                         continue;
2600                 }
2601
2602                 if (max_change < change)
2603                         max_change = change;
2604         }
2605         count = i; /* where we can stop scanning in data->files[] */
2606
2607         /*
2608          * We have width = stat_width or term_columns() columns total.
2609          * We want a maximum of min(max_len, stat_name_width) for the name part.
2610          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2611          * We also need 1 for " " and 4 + decimal_width(max_change)
2612          * for " | NNNN " and one the empty column at the end, altogether
2613          * 6 + decimal_width(max_change).
2614          *
2615          * If there's not enough space, we will use the smaller of
2616          * stat_name_width (if set) and 5/8*width for the filename,
2617          * and the rest for constant elements + graph part, but no more
2618          * than stat_graph_width for the graph part.
2619          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2620          * for the standard terminal size).
2621          *
2622          * In other words: stat_width limits the maximum width, and
2623          * stat_name_width fixes the maximum width of the filename,
2624          * and is also used to divide available columns if there
2625          * aren't enough.
2626          *
2627          * Binary files are displayed with "Bin XXX -> YYY bytes"
2628          * instead of the change count and graph. This part is treated
2629          * similarly to the graph part, except that it is not
2630          * "scaled". If total width is too small to accommodate the
2631          * guaranteed minimum width of the filename part and the
2632          * separators and this message, this message will "overflow"
2633          * making the line longer than the maximum width.
2634          */
2635
2636         if (options->stat_width == -1)
2637                 width = term_columns() - strlen(line_prefix);
2638         else
2639                 width = options->stat_width ? options->stat_width : 80;
2640         number_width = decimal_width(max_change) > number_width ?
2641                 decimal_width(max_change) : number_width;
2642
2643         if (options->stat_graph_width == -1)
2644                 options->stat_graph_width = diff_stat_graph_width;
2645
2646         /*
2647          * Guarantee 3/8*16==6 for the graph part
2648          * and 5/8*16==10 for the filename part
2649          */
2650         if (width < 16 + 6 + number_width)
2651                 width = 16 + 6 + number_width;
2652
2653         /*
2654          * First assign sizes that are wanted, ignoring available width.
2655          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2656          * starting from "XXX" should fit in graph_width.
2657          */
2658         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2659         if (options->stat_graph_width &&
2660             options->stat_graph_width < graph_width)
2661                 graph_width = options->stat_graph_width;
2662
2663         name_width = (options->stat_name_width > 0 &&
2664                       options->stat_name_width < max_len) ?
2665                 options->stat_name_width : max_len;
2666
2667         /*
2668          * Adjust adjustable widths not to exceed maximum width
2669          */
2670         if (name_width + number_width + 6 + graph_width > width) {
2671                 if (graph_width > width * 3/8 - number_width - 6) {
2672                         graph_width = width * 3/8 - number_width - 6;
2673                         if (graph_width < 6)
2674                                 graph_width = 6;
2675                 }
2676
2677                 if (options->stat_graph_width &&
2678                     graph_width > options->stat_graph_width)
2679                         graph_width = options->stat_graph_width;
2680                 if (name_width > width - number_width - 6 - graph_width)
2681                         name_width = width - number_width - 6 - graph_width;
2682                 else
2683                         graph_width = width - number_width - 6 - name_width;
2684         }
2685
2686         /*
2687          * From here name_width is the width of the name area,
2688          * and graph_width is the width of the graph area.
2689          * max_change is used to scale graph properly.
2690          */
2691         for (i = 0; i < count; i++) {
2692                 const char *prefix = "";
2693                 struct diffstat_file *file = data->files[i];
2694                 char *name = file->print_name;
2695                 uintmax_t added = file->added;
2696                 uintmax_t deleted = file->deleted;
2697                 int name_len;
2698
2699                 if (!file->is_interesting && (added + deleted == 0))
2700                         continue;
2701
2702                 /*
2703                  * "scale" the filename
2704                  */
2705                 len = name_width;
2706                 name_len = strlen(name);
2707                 if (name_width < name_len) {
2708                         char *slash;
2709                         prefix = "...";
2710                         len -= 3;
2711                         name += name_len - len;
2712                         slash = strchr(name, '/');
2713                         if (slash)
2714                                 name = slash;
2715                 }
2716
2717                 if (file->is_binary) {
2718                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2719                         strbuf_addf(&out, " %*s", number_width, "Bin");
2720                         if (!added && !deleted) {
2721                                 strbuf_addch(&out, '\n');
2722                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2723                                                  out.buf, out.len, 0);
2724                                 strbuf_reset(&out);
2725                                 continue;
2726                         }
2727                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2728                                 del_c, deleted, reset);
2729                         strbuf_addstr(&out, " -> ");
2730                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2731                                 add_c, added, reset);
2732                         strbuf_addstr(&out, " bytes\n");
2733                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2734                                          out.buf, out.len, 0);
2735                         strbuf_reset(&out);
2736                         continue;
2737                 }
2738                 else if (file->is_unmerged) {
2739                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2740                         strbuf_addstr(&out, " Unmerged\n");
2741                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2742                                          out.buf, out.len, 0);
2743                         strbuf_reset(&out);
2744                         continue;
2745                 }
2746
2747                 /*
2748                  * scale the add/delete
2749                  */
2750                 add = added;
2751                 del = deleted;
2752
2753                 if (graph_width <= max_change) {
2754                         int total = scale_linear(add + del, graph_width, max_change);
2755                         if (total < 2 && add && del)
2756                                 /* width >= 2 due to the sanity check */
2757                                 total = 2;
2758                         if (add < del) {
2759                                 add = scale_linear(add, graph_width, max_change);
2760                                 del = total - add;
2761                         } else {
2762                                 del = scale_linear(del, graph_width, max_change);
2763                                 add = total - del;
2764                         }
2765                 }
2766                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2767                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2768                         number_width, added + deleted,
2769                         added + deleted ? " " : "");
2770                 show_graph(&out, '+', add, add_c, reset);
2771                 show_graph(&out, '-', del, del_c, reset);
2772                 strbuf_addch(&out, '\n');
2773                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2774                                  out.buf, out.len, 0);
2775                 strbuf_reset(&out);
2776         }
2777
2778         for (i = 0; i < data->nr; i++) {
2779                 struct diffstat_file *file = data->files[i];
2780                 uintmax_t added = file->added;
2781                 uintmax_t deleted = file->deleted;
2782
2783                 if (file->is_unmerged ||
2784                     (!file->is_interesting && (added + deleted == 0))) {
2785                         total_files--;
2786                         continue;
2787                 }
2788
2789                 if (!file->is_binary) {
2790                         adds += added;
2791                         dels += deleted;
2792                 }
2793                 if (i < count)
2794                         continue;
2795                 if (!extra_shown)
2796                         emit_diff_symbol(options,
2797                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2798                                          NULL, 0, 0);
2799                 extra_shown = 1;
2800         }
2801
2802         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2803         strbuf_release(&out);
2804 }
2805
2806 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2807 {
2808         int i, adds = 0, dels = 0, total_files = data->nr;
2809
2810         if (data->nr == 0)
2811                 return;
2812
2813         for (i = 0; i < data->nr; i++) {
2814                 int added = data->files[i]->added;
2815                 int deleted = data->files[i]->deleted;
2816
2817                 if (data->files[i]->is_unmerged ||
2818                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2819                         total_files--;
2820                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2821                         adds += added;
2822                         dels += deleted;
2823                 }
2824         }
2825         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2826 }
2827
2828 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2829 {
2830         int i;
2831
2832         if (data->nr == 0)
2833                 return;
2834
2835         for (i = 0; i < data->nr; i++) {
2836                 struct diffstat_file *file = data->files[i];
2837
2838                 fprintf(options->file, "%s", diff_line_prefix(options));
2839
2840                 if (file->is_binary)
2841                         fprintf(options->file, "-\t-\t");
2842                 else
2843                         fprintf(options->file,
2844                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2845                                 file->added, file->deleted);
2846                 if (options->line_termination) {
2847                         fill_print_name(file);
2848                         if (!file->is_renamed)
2849                                 write_name_quoted(file->name, options->file,
2850                                                   options->line_termination);
2851                         else {
2852                                 fputs(file->print_name, options->file);
2853                                 putc(options->line_termination, options->file);
2854                         }
2855                 } else {
2856                         if (file->is_renamed) {
2857                                 putc('\0', options->file);
2858                                 write_name_quoted(file->from_name, options->file, '\0');
2859                         }
2860                         write_name_quoted(file->name, options->file, '\0');
2861                 }
2862         }
2863 }
2864
2865 struct dirstat_file {
2866         const char *name;
2867         unsigned long changed;
2868 };
2869
2870 struct dirstat_dir {
2871         struct dirstat_file *files;
2872         int alloc, nr, permille, cumulative;
2873 };
2874
2875 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2876                 unsigned long changed, const char *base, int baselen)
2877 {
2878         unsigned long sum_changes = 0;
2879         unsigned int sources = 0;
2880         const char *line_prefix = diff_line_prefix(opt);
2881
2882         while (dir->nr) {
2883                 struct dirstat_file *f = dir->files;
2884                 int namelen = strlen(f->name);
2885                 unsigned long changes;
2886                 char *slash;
2887
2888                 if (namelen < baselen)
2889                         break;
2890                 if (memcmp(f->name, base, baselen))
2891                         break;
2892                 slash = strchr(f->name + baselen, '/');
2893                 if (slash) {
2894                         int newbaselen = slash + 1 - f->name;
2895                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2896                         sources++;
2897                 } else {
2898                         changes = f->changed;
2899                         dir->files++;
2900                         dir->nr--;
2901                         sources += 2;
2902                 }
2903                 sum_changes += changes;
2904         }
2905
2906         /*
2907          * We don't report dirstat's for
2908          *  - the top level
2909          *  - or cases where everything came from a single directory
2910          *    under this directory (sources == 1).
2911          */
2912         if (baselen && sources != 1) {
2913                 if (sum_changes) {
2914                         int permille = sum_changes * 1000 / changed;
2915                         if (permille >= dir->permille) {
2916                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2917                                         permille / 10, permille % 10, baselen, base);
2918                                 if (!dir->cumulative)
2919                                         return 0;
2920                         }
2921                 }
2922         }
2923         return sum_changes;
2924 }
2925
2926 static int dirstat_compare(const void *_a, const void *_b)
2927 {
2928         const struct dirstat_file *a = _a;
2929         const struct dirstat_file *b = _b;
2930         return strcmp(a->name, b->name);
2931 }
2932
2933 static void show_dirstat(struct diff_options *options)
2934 {
2935         int i;
2936         unsigned long changed;
2937         struct dirstat_dir dir;
2938         struct diff_queue_struct *q = &diff_queued_diff;
2939
2940         dir.files = NULL;
2941         dir.alloc = 0;
2942         dir.nr = 0;
2943         dir.permille = options->dirstat_permille;
2944         dir.cumulative = options->flags.dirstat_cumulative;
2945
2946         changed = 0;
2947         for (i = 0; i < q->nr; i++) {
2948                 struct diff_filepair *p = q->queue[i];
2949                 const char *name;
2950                 unsigned long copied, added, damage;
2951                 int content_changed;
2952
2953                 name = p->two->path ? p->two->path : p->one->path;
2954
2955                 if (p->one->oid_valid && p->two->oid_valid)
2956                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2957                 else
2958                         content_changed = 1;
2959
2960                 if (!content_changed) {
2961                         /*
2962                          * The SHA1 has not changed, so pre-/post-content is
2963                          * identical. We can therefore skip looking at the
2964                          * file contents altogether.
2965                          */
2966                         damage = 0;
2967                         goto found_damage;
2968                 }
2969
2970                 if (options->flags.dirstat_by_file) {
2971                         /*
2972                          * In --dirstat-by-file mode, we don't really need to
2973                          * look at the actual file contents at all.
2974                          * The fact that the SHA1 changed is enough for us to
2975                          * add this file to the list of results
2976                          * (with each file contributing equal damage).
2977                          */
2978                         damage = 1;
2979                         goto found_damage;
2980                 }
2981
2982                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2983                         diff_populate_filespec(p->one, 0);
2984                         diff_populate_filespec(p->two, 0);
2985                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2986                                                &copied, &added);
2987                         diff_free_filespec_data(p->one);
2988                         diff_free_filespec_data(p->two);
2989                 } else if (DIFF_FILE_VALID(p->one)) {
2990                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2991                         copied = added = 0;
2992                         diff_free_filespec_data(p->one);
2993                 } else if (DIFF_FILE_VALID(p->two)) {
2994                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2995                         copied = 0;
2996                         added = p->two->size;
2997                         diff_free_filespec_data(p->two);
2998                 } else
2999                         continue;
3000
3001                 /*
3002                  * Original minus copied is the removed material,
3003                  * added is the new material.  They are both damages
3004                  * made to the preimage.
3005                  * If the resulting damage is zero, we know that
3006                  * diffcore_count_changes() considers the two entries to
3007                  * be identical, but since content_changed is true, we
3008                  * know that there must have been _some_ kind of change,
3009                  * so we force all entries to have damage > 0.
3010                  */
3011                 damage = (p->one->size - copied) + added;
3012                 if (!damage)
3013                         damage = 1;
3014
3015 found_damage:
3016                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3017                 dir.files[dir.nr].name = name;
3018                 dir.files[dir.nr].changed = damage;
3019                 changed += damage;
3020                 dir.nr++;
3021         }
3022
3023         /* This can happen even with many files, if everything was renames */
3024         if (!changed)
3025                 return;
3026
3027         /* Show all directories with more than x% of the changes */
3028         QSORT(dir.files, dir.nr, dirstat_compare);
3029         gather_dirstat(options, &dir, changed, "", 0);
3030 }
3031
3032 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3033 {
3034         int i;
3035         unsigned long changed;
3036         struct dirstat_dir dir;
3037
3038         if (data->nr == 0)
3039                 return;
3040
3041         dir.files = NULL;
3042         dir.alloc = 0;
3043         dir.nr = 0;
3044         dir.permille = options->dirstat_permille;
3045         dir.cumulative = options->flags.dirstat_cumulative;
3046
3047         changed = 0;
3048         for (i = 0; i < data->nr; i++) {
3049                 struct diffstat_file *file = data->files[i];
3050                 unsigned long damage = file->added + file->deleted;
3051                 if (file->is_binary)
3052                         /*
3053                          * binary files counts bytes, not lines. Must find some
3054                          * way to normalize binary bytes vs. textual lines.
3055                          * The following heuristic assumes that there are 64
3056                          * bytes per "line".
3057                          * This is stupid and ugly, but very cheap...
3058                          */
3059                         damage = DIV_ROUND_UP(damage, 64);
3060                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3061                 dir.files[dir.nr].name = file->name;
3062                 dir.files[dir.nr].changed = damage;
3063                 changed += damage;
3064                 dir.nr++;
3065         }
3066
3067         /* This can happen even with many files, if everything was renames */
3068         if (!changed)
3069                 return;
3070
3071         /* Show all directories with more than x% of the changes */
3072         QSORT(dir.files, dir.nr, dirstat_compare);
3073         gather_dirstat(options, &dir, changed, "", 0);
3074 }
3075
3076 static void free_diffstat_info(struct diffstat_t *diffstat)
3077 {
3078         int i;
3079         for (i = 0; i < diffstat->nr; i++) {
3080                 struct diffstat_file *f = diffstat->files[i];
3081                 free(f->print_name);
3082                 free(f->name);
3083                 free(f->from_name);
3084                 free(f);
3085         }
3086         free(diffstat->files);
3087 }
3088
3089 struct checkdiff_t {
3090         const char *filename;
3091         int lineno;
3092         int conflict_marker_size;
3093         struct diff_options *o;
3094         unsigned ws_rule;
3095         unsigned status;
3096 };
3097
3098 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3099 {
3100         char firstchar;
3101         int cnt;
3102
3103         if (len < marker_size + 1)
3104                 return 0;
3105         firstchar = line[0];
3106         switch (firstchar) {
3107         case '=': case '>': case '<': case '|':
3108                 break;
3109         default:
3110                 return 0;
3111         }
3112         for (cnt = 1; cnt < marker_size; cnt++)
3113                 if (line[cnt] != firstchar)
3114                         return 0;
3115         /* line[1] thru line[marker_size-1] are same as firstchar */
3116         if (len < marker_size + 1 || !isspace(line[marker_size]))
3117                 return 0;
3118         return 1;
3119 }
3120
3121 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3122 {
3123         struct checkdiff_t *data = priv;
3124         int marker_size = data->conflict_marker_size;
3125         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3126         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3127         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3128         char *err;
3129         const char *line_prefix;
3130
3131         assert(data->o);
3132         line_prefix = diff_line_prefix(data->o);
3133
3134         if (line[0] == '+') {
3135                 unsigned bad;
3136                 data->lineno++;
3137                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3138                         data->status |= 1;
3139                         fprintf(data->o->file,
3140                                 "%s%s:%d: leftover conflict marker\n",
3141                                 line_prefix, data->filename, data->lineno);
3142                 }
3143                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3144                 if (!bad)
3145                         return;
3146                 data->status |= bad;
3147                 err = whitespace_error_string(bad);
3148                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3149                         line_prefix, data->filename, data->lineno, err);
3150                 free(err);
3151                 emit_line(data->o, set, reset, line, 1);
3152                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3153                               data->o->file, set, reset, ws);
3154         } else if (line[0] == ' ') {
3155                 data->lineno++;
3156         } else if (line[0] == '@') {
3157                 char *plus = strchr(line, '+');
3158                 if (plus)
3159                         data->lineno = strtol(plus, NULL, 10) - 1;
3160                 else
3161                         die("invalid diff");
3162         }
3163 }
3164
3165 static unsigned char *deflate_it(char *data,
3166                                  unsigned long size,
3167                                  unsigned long *result_size)
3168 {
3169         int bound;
3170         unsigned char *deflated;
3171         git_zstream stream;
3172
3173         git_deflate_init(&stream, zlib_compression_level);
3174         bound = git_deflate_bound(&stream, size);
3175         deflated = xmalloc(bound);
3176         stream.next_out = deflated;
3177         stream.avail_out = bound;
3178
3179         stream.next_in = (unsigned char *)data;
3180         stream.avail_in = size;
3181         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3182                 ; /* nothing */
3183         git_deflate_end(&stream);
3184         *result_size = stream.total_out;
3185         return deflated;
3186 }
3187
3188 static void emit_binary_diff_body(struct diff_options *o,
3189                                   mmfile_t *one, mmfile_t *two)
3190 {
3191         void *cp;
3192         void *delta;
3193         void *deflated;
3194         void *data;
3195         unsigned long orig_size;
3196         unsigned long delta_size;
3197         unsigned long deflate_size;
3198         unsigned long data_size;
3199
3200         /* We could do deflated delta, or we could do just deflated two,
3201          * whichever is smaller.
3202          */
3203         delta = NULL;
3204         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3205         if (one->size && two->size) {
3206                 delta = diff_delta(one->ptr, one->size,
3207                                    two->ptr, two->size,
3208                                    &delta_size, deflate_size);
3209                 if (delta) {
3210                         void *to_free = delta;
3211                         orig_size = delta_size;
3212                         delta = deflate_it(delta, delta_size, &delta_size);
3213                         free(to_free);
3214                 }
3215         }
3216
3217         if (delta && delta_size < deflate_size) {
3218                 char *s = xstrfmt("%lu", orig_size);
3219                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3220                                  s, strlen(s), 0);
3221                 free(s);
3222                 free(deflated);
3223                 data = delta;
3224                 data_size = delta_size;
3225         } else {
3226                 char *s = xstrfmt("%lu", two->size);
3227                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3228                                  s, strlen(s), 0);
3229                 free(s);
3230                 free(delta);
3231                 data = deflated;
3232                 data_size = deflate_size;
3233         }
3234
3235         /* emit data encoded in base85 */
3236         cp = data;
3237         while (data_size) {
3238                 int len;
3239                 int bytes = (52 < data_size) ? 52 : data_size;
3240                 char line[71];
3241                 data_size -= bytes;
3242                 if (bytes <= 26)
3243                         line[0] = bytes + 'A' - 1;
3244                 else
3245                         line[0] = bytes - 26 + 'a' - 1;
3246                 encode_85(line + 1, cp, bytes);
3247                 cp = (char *) cp + bytes;
3248
3249                 len = strlen(line);
3250                 line[len++] = '\n';
3251                 line[len] = '\0';
3252
3253                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3254                                  line, len, 0);
3255         }
3256         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3257         free(data);
3258 }
3259
3260 static void emit_binary_diff(struct diff_options *o,
3261                              mmfile_t *one, mmfile_t *two)
3262 {
3263         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3264         emit_binary_diff_body(o, one, two);
3265         emit_binary_diff_body(o, two, one);
3266 }
3267
3268 int diff_filespec_is_binary(struct diff_filespec *one)
3269 {
3270         if (one->is_binary == -1) {
3271                 diff_filespec_load_driver(one);
3272                 if (one->driver->binary != -1)
3273                         one->is_binary = one->driver->binary;
3274                 else {
3275                         if (!one->data && DIFF_FILE_VALID(one))
3276                                 diff_populate_filespec(one, CHECK_BINARY);
3277                         if (one->is_binary == -1 && one->data)
3278                                 one->is_binary = buffer_is_binary(one->data,
3279                                                 one->size);
3280                         if (one->is_binary == -1)
3281                                 one->is_binary = 0;
3282                 }
3283         }
3284         return one->is_binary;
3285 }
3286
3287 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3288 {
3289         diff_filespec_load_driver(one);
3290         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3291 }
3292
3293 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3294 {
3295         if (!options->a_prefix)
3296                 options->a_prefix = a;
3297         if (!options->b_prefix)
3298                 options->b_prefix = b;
3299 }
3300
3301 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3302 {
3303         if (!DIFF_FILE_VALID(one))
3304                 return NULL;
3305
3306         diff_filespec_load_driver(one);
3307         return userdiff_get_textconv(one->driver);
3308 }
3309
3310 static void builtin_diff(const char *name_a,
3311                          const char *name_b,
3312                          struct diff_filespec *one,
3313                          struct diff_filespec *two,
3314                          const char *xfrm_msg,
3315                          int must_show_header,
3316                          struct diff_options *o,
3317                          int complete_rewrite)
3318 {
3319         mmfile_t mf1, mf2;
3320         const char *lbl[2];
3321         char *a_one, *b_two;
3322         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3323         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3324         const char *a_prefix, *b_prefix;
3325         struct userdiff_driver *textconv_one = NULL;
3326         struct userdiff_driver *textconv_two = NULL;
3327         struct strbuf header = STRBUF_INIT;
3328         const char *line_prefix = diff_line_prefix(o);
3329
3330         diff_set_mnemonic_prefix(o, "a/", "b/");
3331         if (o->flags.reverse_diff) {
3332                 a_prefix = o->b_prefix;
3333                 b_prefix = o->a_prefix;
3334         } else {
3335                 a_prefix = o->a_prefix;
3336                 b_prefix = o->b_prefix;
3337         }
3338
3339         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3340             (!one->mode || S_ISGITLINK(one->mode)) &&
3341             (!two->mode || S_ISGITLINK(two->mode))) {
3342                 show_submodule_summary(o, one->path ? one->path : two->path,
3343                                 &one->oid, &two->oid,
3344                                 two->dirty_submodule);
3345                 return;
3346         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3347                    (!one->mode || S_ISGITLINK(one->mode)) &&
3348                    (!two->mode || S_ISGITLINK(two->mode))) {
3349                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3350                                 &one->oid, &two->oid,
3351                                 two->dirty_submodule);
3352                 return;
3353         }
3354
3355         if (o->flags.allow_textconv) {
3356                 textconv_one = get_textconv(one);
3357                 textconv_two = get_textconv(two);
3358         }
3359
3360         /* Never use a non-valid filename anywhere if at all possible */
3361         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3362         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3363
3364         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3365         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3366         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3367         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3368         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3369         if (lbl[0][0] == '/') {
3370                 /* /dev/null */
3371                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3372                 if (xfrm_msg)
3373                         strbuf_addstr(&header, xfrm_msg);
3374                 must_show_header = 1;
3375         }
3376         else if (lbl[1][0] == '/') {
3377                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3378                 if (xfrm_msg)
3379                         strbuf_addstr(&header, xfrm_msg);
3380                 must_show_header = 1;
3381         }
3382         else {
3383                 if (one->mode != two->mode) {
3384                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3385                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3386                         must_show_header = 1;
3387                 }
3388                 if (xfrm_msg)
3389                         strbuf_addstr(&header, xfrm_msg);
3390
3391                 /*
3392                  * we do not run diff between different kind
3393                  * of objects.
3394                  */
3395                 if ((one->mode ^ two->mode) & S_IFMT)
3396                         goto free_ab_and_return;
3397                 if (complete_rewrite &&
3398                     (textconv_one || !diff_filespec_is_binary(one)) &&
3399                     (textconv_two || !diff_filespec_is_binary(two))) {
3400                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3401                                          header.buf, header.len, 0);
3402                         strbuf_reset(&header);
3403                         emit_rewrite_diff(name_a, name_b, one, two,
3404                                                 textconv_one, textconv_two, o);
3405                         o->found_changes = 1;
3406                         goto free_ab_and_return;
3407                 }
3408         }
3409
3410         if (o->irreversible_delete && lbl[1][0] == '/') {
3411                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3412                                  header.len, 0);
3413                 strbuf_reset(&header);
3414                 goto free_ab_and_return;
3415         } else if (!o->flags.text &&
3416             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3417               (!textconv_two && diff_filespec_is_binary(two)) )) {
3418                 struct strbuf sb = STRBUF_INIT;
3419                 if (!one->data && !two->data &&
3420                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3421                     !o->flags.binary) {
3422                         if (!oidcmp(&one->oid, &two->oid)) {
3423                                 if (must_show_header)
3424                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3425                                                          header.buf, header.len,
3426                                                          0);
3427                                 goto free_ab_and_return;
3428                         }
3429                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3430                                          header.buf, header.len, 0);
3431                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3432                                     diff_line_prefix(o), lbl[0], lbl[1]);
3433                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3434                                          sb.buf, sb.len, 0);
3435                         strbuf_release(&sb);
3436                         goto free_ab_and_return;
3437                 }
3438                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3439                         die("unable to read files to diff");
3440                 /* Quite common confusing case */
3441                 if (mf1.size == mf2.size &&
3442                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3443                         if (must_show_header)
3444                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3445                                                  header.buf, header.len, 0);
3446                         goto free_ab_and_return;
3447                 }
3448                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3449                 strbuf_reset(&header);
3450                 if (o->flags.binary)
3451                         emit_binary_diff(o, &mf1, &mf2);
3452                 else {
3453                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3454                                     diff_line_prefix(o), lbl[0], lbl[1]);
3455                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3456                                          sb.buf, sb.len, 0);
3457                         strbuf_release(&sb);
3458                 }
3459                 o->found_changes = 1;
3460         } else {
3461                 /* Crazy xdl interfaces.. */
3462                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3463                 const char *v;
3464                 xpparam_t xpp;
3465                 xdemitconf_t xecfg;
3466                 struct emit_callback ecbdata;
3467                 const struct userdiff_funcname *pe;
3468
3469                 if (must_show_header) {
3470                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3471                                          header.buf, header.len, 0);
3472                         strbuf_reset(&header);
3473                 }
3474
3475                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3476                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3477
3478                 pe = diff_funcname_pattern(one);
3479                 if (!pe)
3480                         pe = diff_funcname_pattern(two);
3481
3482                 memset(&xpp, 0, sizeof(xpp));
3483                 memset(&xecfg, 0, sizeof(xecfg));
3484                 memset(&ecbdata, 0, sizeof(ecbdata));
3485                 if (o->flags.suppress_diff_headers)
3486                         lbl[0] = NULL;
3487                 ecbdata.label_path = lbl;
3488                 ecbdata.color_diff = want_color(o->use_color);
3489                 ecbdata.ws_rule = whitespace_rule(name_b);
3490                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3491                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3492                 ecbdata.opt = o;
3493                 if (header.len && !o->flags.suppress_diff_headers)
3494                         ecbdata.header = &header;
3495                 xpp.flags = o->xdl_opts;
3496                 xpp.anchors = o->anchors;
3497                 xpp.anchors_nr = o->anchors_nr;
3498                 xecfg.ctxlen = o->context;
3499                 xecfg.interhunkctxlen = o->interhunkcontext;
3500                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3501                 if (o->flags.funccontext)
3502                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3503                 if (pe)
3504                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3505                 if (!diffopts)
3506                         ;
3507                 else if (skip_prefix(diffopts, "--unified=", &v))
3508                         xecfg.ctxlen = strtoul(v, NULL, 10);
3509                 else if (skip_prefix(diffopts, "-u", &v))
3510                         xecfg.ctxlen = strtoul(v, NULL, 10);
3511                 if (o->word_diff)
3512                         init_diff_words_data(&ecbdata, o, one, two);
3513                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3514                                   &xpp, &xecfg))
3515                         die("unable to generate diff for %s", one->path);
3516                 if (o->word_diff)
3517                         free_diff_words_data(&ecbdata);
3518                 if (textconv_one)
3519                         free(mf1.ptr);
3520                 if (textconv_two)
3521                         free(mf2.ptr);
3522                 xdiff_clear_find_func(&xecfg);
3523         }
3524
3525  free_ab_and_return:
3526         strbuf_release(&header);
3527         diff_free_filespec_data(one);
3528         diff_free_filespec_data(two);
3529         free(a_one);
3530         free(b_two);
3531         return;
3532 }
3533
3534 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3535 {
3536         if (!is_renamed) {
3537                 if (p->status == DIFF_STATUS_ADDED) {
3538                         if (S_ISLNK(p->two->mode))
3539                                 return "new +l";
3540                         else if ((p->two->mode & 0777) == 0755)
3541                                 return "new +x";
3542                         else
3543                                 return "new";
3544                 } else if (p->status == DIFF_STATUS_DELETED)
3545                         return "gone";
3546         }
3547         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3548                 return "mode -l";
3549         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3550                 return "mode +l";
3551         else if ((p->one->mode & 0777) == 0644 &&
3552                  (p->two->mode & 0777) == 0755)
3553                 return "mode +x";
3554         else if ((p->one->mode & 0777) == 0755 &&
3555                  (p->two->mode & 0777) == 0644)
3556                 return "mode -x";
3557         return NULL;
3558 }
3559
3560 static void builtin_diffstat(const char *name_a, const char *name_b,
3561                              struct diff_filespec *one,
3562                              struct diff_filespec *two,
3563                              struct diffstat_t *diffstat,
3564                              struct diff_options *o,
3565                              struct diff_filepair *p)
3566 {
3567         mmfile_t mf1, mf2;
3568         struct diffstat_file *data;
3569         int same_contents;
3570         int complete_rewrite = 0;
3571
3572         if (!DIFF_PAIR_UNMERGED(p)) {
3573                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3574                         complete_rewrite = 1;
3575         }
3576
3577         data = diffstat_add(diffstat, name_a, name_b);
3578         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3579         if (o->flags.stat_with_summary)
3580                 data->comments = get_compact_summary(p, data->is_renamed);
3581
3582         if (!one || !two) {
3583                 data->is_unmerged = 1;
3584                 return;
3585         }
3586
3587         same_contents = !oidcmp(&one->oid, &two->oid);
3588
3589         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3590                 data->is_binary = 1;
3591                 if (same_contents) {
3592                         data->added = 0;
3593                         data->deleted = 0;
3594                 } else {
3595                         data->added = diff_filespec_size(two);
3596                         data->deleted = diff_filespec_size(one);
3597                 }
3598         }
3599
3600         else if (complete_rewrite) {
3601                 diff_populate_filespec(one, 0);
3602                 diff_populate_filespec(two, 0);
3603                 data->deleted = count_lines(one->data, one->size);
3604                 data->added = count_lines(two->data, two->size);
3605         }
3606
3607         else if (!same_contents) {
3608                 /* Crazy xdl interfaces.. */
3609                 xpparam_t xpp;
3610                 xdemitconf_t xecfg;
3611
3612                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3613                         die("unable to read files to diff");
3614
3615                 memset(&xpp, 0, sizeof(xpp));
3616                 memset(&xecfg, 0, sizeof(xecfg));
3617                 xpp.flags = o->xdl_opts;
3618                 xpp.anchors = o->anchors;
3619                 xpp.anchors_nr = o->anchors_nr;
3620                 xecfg.ctxlen = o->context;
3621                 xecfg.interhunkctxlen = o->interhunkcontext;
3622                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3623                                   &xpp, &xecfg))
3624                         die("unable to generate diffstat for %s", one->path);
3625         }
3626
3627         diff_free_filespec_data(one);
3628         diff_free_filespec_data(two);
3629 }
3630
3631 static void builtin_checkdiff(const char *name_a, const char *name_b,
3632                               const char *attr_path,
3633                               struct diff_filespec *one,
3634                               struct diff_filespec *two,
3635                               struct diff_options *o)
3636 {
3637         mmfile_t mf1, mf2;
3638         struct checkdiff_t data;
3639
3640         if (!two)
3641                 return;
3642
3643         memset(&data, 0, sizeof(data));
3644         data.filename = name_b ? name_b : name_a;
3645         data.lineno = 0;
3646         data.o = o;
3647         data.ws_rule = whitespace_rule(attr_path);
3648         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3649
3650         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3651                 die("unable to read files to diff");
3652
3653         /*
3654          * All the other codepaths check both sides, but not checking
3655          * the "old" side here is deliberate.  We are checking the newly
3656          * introduced changes, and as long as the "new" side is text, we
3657          * can and should check what it introduces.
3658          */
3659         if (diff_filespec_is_binary(two))
3660                 goto free_and_return;
3661         else {
3662                 /* Crazy xdl interfaces.. */
3663                 xpparam_t xpp;
3664                 xdemitconf_t xecfg;
3665
3666                 memset(&xpp, 0, sizeof(xpp));
3667                 memset(&xecfg, 0, sizeof(xecfg));
3668                 xecfg.ctxlen = 1; /* at least one context line */
3669                 xpp.flags = 0;
3670                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3671                                   &xpp, &xecfg))
3672                         die("unable to generate checkdiff for %s", one->path);
3673
3674                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3675                         struct emit_callback ecbdata;
3676                         int blank_at_eof;
3677
3678                         ecbdata.ws_rule = data.ws_rule;
3679                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3680                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3681
3682                         if (blank_at_eof) {
3683                                 static char *err;
3684                                 if (!err)
3685                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3686                                 fprintf(o->file, "%s:%d: %s.\n",
3687                                         data.filename, blank_at_eof, err);
3688                                 data.status = 1; /* report errors */
3689                         }
3690                 }
3691         }
3692  free_and_return:
3693         diff_free_filespec_data(one);
3694         diff_free_filespec_data(two);
3695         if (data.status)
3696                 o->flags.check_failed = 1;
3697 }
3698
3699 struct diff_filespec *alloc_filespec(const char *path)
3700 {
3701         struct diff_filespec *spec;
3702
3703         FLEXPTR_ALLOC_STR(spec, path, path);
3704         spec->count = 1;
3705         spec->is_binary = -1;
3706         return spec;
3707 }
3708
3709 void free_filespec(struct diff_filespec *spec)
3710 {
3711         if (!--spec->count) {
3712                 diff_free_filespec_data(spec);
3713                 free(spec);
3714         }
3715 }
3716
3717 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3718                    int oid_valid, unsigned short mode)
3719 {
3720         if (mode) {
3721                 spec->mode = canon_mode(mode);
3722                 oidcpy(&spec->oid, oid);
3723                 spec->oid_valid = oid_valid;
3724         }
3725 }
3726
3727 /*
3728  * Given a name and sha1 pair, if the index tells us the file in
3729  * the work tree has that object contents, return true, so that
3730  * prepare_temp_file() does not have to inflate and extract.
3731  */
3732 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3733 {
3734         const struct cache_entry *ce;
3735         struct stat st;
3736         int pos, len;
3737
3738         /*
3739          * We do not read the cache ourselves here, because the
3740          * benchmark with my previous version that always reads cache
3741          * shows that it makes things worse for diff-tree comparing
3742          * two linux-2.6 kernel trees in an already checked out work
3743          * tree.  This is because most diff-tree comparisons deal with
3744          * only a small number of files, while reading the cache is
3745          * expensive for a large project, and its cost outweighs the
3746          * savings we get by not inflating the object to a temporary
3747          * file.  Practically, this code only helps when we are used
3748          * by diff-cache --cached, which does read the cache before
3749          * calling us.
3750          */
3751         if (!active_cache)
3752                 return 0;
3753
3754         /* We want to avoid the working directory if our caller
3755          * doesn't need the data in a normal file, this system
3756          * is rather slow with its stat/open/mmap/close syscalls,
3757          * and the object is contained in a pack file.  The pack
3758          * is probably already open and will be faster to obtain
3759          * the data through than the working directory.  Loose
3760          * objects however would tend to be slower as they need
3761          * to be individually opened and inflated.
3762          */
3763         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3764                 return 0;
3765
3766         /*
3767          * Similarly, if we'd have to convert the file contents anyway, that
3768          * makes the optimization not worthwhile.
3769          */
3770         if (!want_file && would_convert_to_git(&the_index, name))
3771                 return 0;
3772
3773         len = strlen(name);
3774         pos = cache_name_pos(name, len);
3775         if (pos < 0)
3776                 return 0;
3777         ce = active_cache[pos];
3778
3779         /*
3780          * This is not the sha1 we are looking for, or
3781          * unreusable because it is not a regular file.
3782          */
3783         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3784                 return 0;
3785
3786         /*
3787          * If ce is marked as "assume unchanged", there is no
3788          * guarantee that work tree matches what we are looking for.
3789          */
3790         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3791                 return 0;
3792
3793         /*
3794          * If ce matches the file in the work tree, we can reuse it.
3795          */
3796         if (ce_uptodate(ce) ||
3797             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3798                 return 1;
3799
3800         return 0;
3801 }
3802
3803 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3804 {
3805         struct strbuf buf = STRBUF_INIT;
3806         char *dirty = "";
3807
3808         /* Are we looking at the work tree? */
3809         if (s->dirty_submodule)
3810                 dirty = "-dirty";
3811
3812         strbuf_addf(&buf, "Subproject commit %s%s\n",
3813                     oid_to_hex(&s->oid), dirty);
3814         s->size = buf.len;
3815         if (size_only) {
3816                 s->data = NULL;
3817                 strbuf_release(&buf);
3818         } else {
3819                 s->data = strbuf_detach(&buf, NULL);
3820                 s->should_free = 1;
3821         }
3822         return 0;
3823 }
3824
3825 /*
3826  * While doing rename detection and pickaxe operation, we may need to
3827  * grab the data for the blob (or file) for our own in-core comparison.
3828  * diff_filespec has data and size fields for this purpose.
3829  */
3830 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3831 {
3832         int size_only = flags & CHECK_SIZE_ONLY;
3833         int err = 0;
3834         int conv_flags = global_conv_flags_eol;
3835         /*
3836          * demote FAIL to WARN to allow inspecting the situation
3837          * instead of refusing.
3838          */
3839         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3840                 conv_flags = CONV_EOL_RNDTRP_WARN;
3841
3842         if (!DIFF_FILE_VALID(s))
3843                 die("internal error: asking to populate invalid file.");
3844         if (S_ISDIR(s->mode))
3845                 return -1;
3846
3847         if (s->data)
3848                 return 0;
3849
3850         if (size_only && 0 < s->size)
3851                 return 0;
3852
3853         if (S_ISGITLINK(s->mode))
3854                 return diff_populate_gitlink(s, size_only);
3855
3856         if (!s->oid_valid ||
3857             reuse_worktree_file(s->path, &s->oid, 0)) {
3858                 struct strbuf buf = STRBUF_INIT;
3859                 struct stat st;
3860                 int fd;
3861
3862                 if (lstat(s->path, &st) < 0) {
3863                 err_empty:
3864                         err = -1;
3865                 empty:
3866                         s->data = (char *)"";
3867                         s->size = 0;
3868                         return err;
3869                 }
3870                 s->size = xsize_t(st.st_size);
3871                 if (!s->size)
3872                         goto empty;
3873                 if (S_ISLNK(st.st_mode)) {
3874                         struct strbuf sb = STRBUF_INIT;
3875
3876                         if (strbuf_readlink(&sb, s->path, s->size))
3877                                 goto err_empty;
3878                         s->size = sb.len;
3879                         s->data = strbuf_detach(&sb, NULL);
3880                         s->should_free = 1;
3881                         return 0;
3882                 }
3883
3884                 /*
3885                  * Even if the caller would be happy with getting
3886                  * only the size, we cannot return early at this
3887                  * point if the path requires us to run the content
3888                  * conversion.
3889                  */
3890                 if (size_only && !would_convert_to_git(&the_index, s->path))
3891                         return 0;
3892
3893                 /*
3894                  * Note: this check uses xsize_t(st.st_size) that may
3895                  * not be the true size of the blob after it goes
3896                  * through convert_to_git().  This may not strictly be
3897                  * correct, but the whole point of big_file_threshold
3898                  * and is_binary check being that we want to avoid
3899                  * opening the file and inspecting the contents, this
3900                  * is probably fine.
3901                  */
3902                 if ((flags & CHECK_BINARY) &&
3903                     s->size > big_file_threshold && s->is_binary == -1) {
3904                         s->is_binary = 1;
3905                         return 0;
3906                 }
3907                 fd = open(s->path, O_RDONLY);
3908                 if (fd < 0)
3909                         goto err_empty;
3910                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3911                 close(fd);
3912                 s->should_munmap = 1;
3913
3914                 /*
3915                  * Convert from working tree format to canonical git format
3916                  */
3917                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3918                         size_t size = 0;
3919                         munmap(s->data, s->size);
3920                         s->should_munmap = 0;
3921                         s->data = strbuf_detach(&buf, &size);
3922                         s->size = size;
3923                         s->should_free = 1;
3924                 }
3925         }
3926         else {
3927                 enum object_type type;
3928                 if (size_only || (flags & CHECK_BINARY)) {
3929                         type = oid_object_info(the_repository, &s->oid,
3930                                                &s->size);
3931                         if (type < 0)
3932                                 die("unable to read %s",
3933                                     oid_to_hex(&s->oid));
3934                         if (size_only)
3935                                 return 0;
3936                         if (s->size > big_file_threshold && s->is_binary == -1) {
3937                                 s->is_binary = 1;
3938                                 return 0;
3939                         }
3940                 }
3941                 s->data = read_object_file(&s->oid, &type, &s->size);
3942                 if (!s->data)
3943                         die("unable to read %s", oid_to_hex(&s->oid));
3944                 s->should_free = 1;
3945         }
3946         return 0;
3947 }
3948
3949 void diff_free_filespec_blob(struct diff_filespec *s)
3950 {
3951         if (s->should_free)
3952                 free(s->data);
3953         else if (s->should_munmap)
3954                 munmap(s->data, s->size);
3955
3956         if (s->should_free || s->should_munmap) {
3957                 s->should_free = s->should_munmap = 0;
3958                 s->data = NULL;
3959         }
3960 }
3961
3962 void diff_free_filespec_data(struct diff_filespec *s)
3963 {
3964         diff_free_filespec_blob(s);
3965         FREE_AND_NULL(s->cnt_data);
3966 }
3967
3968 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3969                            void *blob,
3970                            unsigned long size,
3971                            const struct object_id *oid,
3972                            int mode)
3973 {
3974         struct strbuf buf = STRBUF_INIT;
3975         struct strbuf tempfile = STRBUF_INIT;
3976         char *path_dup = xstrdup(path);
3977         const char *base = basename(path_dup);
3978
3979         /* Generate "XXXXXX_basename.ext" */
3980         strbuf_addstr(&tempfile, "XXXXXX_");
3981         strbuf_addstr(&tempfile, base);
3982
3983         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3984         if (!temp->tempfile)
3985                 die_errno("unable to create temp-file");
3986         if (convert_to_working_tree(&the_index, path,
3987                         (const char *)blob, (size_t)size, &buf)) {
3988                 blob = buf.buf;
3989                 size = buf.len;
3990         }
3991         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3992             close_tempfile_gently(temp->tempfile))
3993                 die_errno("unable to write temp-file");
3994         temp->name = get_tempfile_path(temp->tempfile);
3995         oid_to_hex_r(temp->hex, oid);
3996         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3997         strbuf_release(&buf);
3998         strbuf_release(&tempfile);
3999         free(path_dup);
4000 }
4001
4002 static struct diff_tempfile *prepare_temp_file(const char *name,
4003                 struct diff_filespec *one)
4004 {
4005         struct diff_tempfile *temp = claim_diff_tempfile();
4006
4007         if (!DIFF_FILE_VALID(one)) {
4008         not_a_valid_file:
4009                 /* A '-' entry produces this for file-2, and
4010                  * a '+' entry produces this for file-1.
4011                  */
4012                 temp->name = "/dev/null";
4013                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4014                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4015                 return temp;
4016         }
4017
4018         if (!S_ISGITLINK(one->mode) &&
4019             (!one->oid_valid ||
4020              reuse_worktree_file(name, &one->oid, 1))) {
4021                 struct stat st;
4022                 if (lstat(name, &st) < 0) {
4023                         if (errno == ENOENT)
4024                                 goto not_a_valid_file;
4025                         die_errno("stat(%s)", name);
4026                 }
4027                 if (S_ISLNK(st.st_mode)) {
4028                         struct strbuf sb = STRBUF_INIT;
4029                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4030                                 die_errno("readlink(%s)", name);
4031                         prep_temp_blob(name, temp, sb.buf, sb.len,
4032                                        (one->oid_valid ?
4033                                         &one->oid : &null_oid),
4034                                        (one->oid_valid ?
4035                                         one->mode : S_IFLNK));
4036                         strbuf_release(&sb);
4037                 }
4038                 else {
4039                         /* we can borrow from the file in the work tree */
4040                         temp->name = name;
4041                         if (!one->oid_valid)
4042                                 oid_to_hex_r(temp->hex, &null_oid);
4043                         else
4044                                 oid_to_hex_r(temp->hex, &one->oid);
4045                         /* Even though we may sometimes borrow the
4046                          * contents from the work tree, we always want
4047                          * one->mode.  mode is trustworthy even when
4048                          * !(one->oid_valid), as long as
4049                          * DIFF_FILE_VALID(one).
4050                          */
4051                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4052                 }
4053                 return temp;
4054         }
4055         else {
4056                 if (diff_populate_filespec(one, 0))
4057                         die("cannot read data blob for %s", one->path);
4058                 prep_temp_blob(name, temp, one->data, one->size,
4059                                &one->oid, one->mode);
4060         }
4061         return temp;
4062 }
4063
4064 static void add_external_diff_name(struct argv_array *argv,
4065                                    const char *name,
4066                                    struct diff_filespec *df)
4067 {
4068         struct diff_tempfile *temp = prepare_temp_file(name, df);
4069         argv_array_push(argv, temp->name);
4070         argv_array_push(argv, temp->hex);
4071         argv_array_push(argv, temp->mode);
4072 }
4073
4074 /* An external diff command takes:
4075  *
4076  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4077  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4078  *
4079  */
4080 static void run_external_diff(const char *pgm,
4081                               const char *name,
4082                               const char *other,
4083                               struct diff_filespec *one,
4084                               struct diff_filespec *two,
4085                               const char *xfrm_msg,
4086                               int complete_rewrite,
4087                               struct diff_options *o)
4088 {
4089         struct argv_array argv = ARGV_ARRAY_INIT;
4090         struct argv_array env = ARGV_ARRAY_INIT;
4091         struct diff_queue_struct *q = &diff_queued_diff;
4092
4093         argv_array_push(&argv, pgm);
4094         argv_array_push(&argv, name);
4095
4096         if (one && two) {
4097                 add_external_diff_name(&argv, name, one);
4098                 if (!other)
4099                         add_external_diff_name(&argv, name, two);
4100                 else {
4101                         add_external_diff_name(&argv, other, two);
4102                         argv_array_push(&argv, other);
4103                         argv_array_push(&argv, xfrm_msg);
4104                 }
4105         }
4106
4107         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4108         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4109
4110         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4111                 die(_("external diff died, stopping at %s"), name);
4112
4113         remove_tempfile();
4114         argv_array_clear(&argv);
4115         argv_array_clear(&env);
4116 }
4117
4118 static int similarity_index(struct diff_filepair *p)
4119 {
4120         return p->score * 100 / MAX_SCORE;
4121 }
4122
4123 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4124 {
4125         if (startup_info->have_repository)
4126                 return find_unique_abbrev(oid, abbrev);
4127         else {
4128                 char *hex = oid_to_hex(oid);
4129                 if (abbrev < 0)
4130                         abbrev = FALLBACK_DEFAULT_ABBREV;
4131                 if (abbrev > the_hash_algo->hexsz)
4132                         BUG("oid abbreviation out of range: %d", abbrev);
4133                 if (abbrev)
4134                         hex[abbrev] = '\0';
4135                 return hex;
4136         }
4137 }
4138
4139 static void fill_metainfo(struct strbuf *msg,
4140                           const char *name,
4141                           const char *other,
4142                           struct diff_filespec *one,
4143                           struct diff_filespec *two,
4144                           struct diff_options *o,
4145                           struct diff_filepair *p,
4146                           int *must_show_header,
4147                           int use_color)
4148 {
4149         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4150         const char *reset = diff_get_color(use_color, DIFF_RESET);
4151         const char *line_prefix = diff_line_prefix(o);
4152
4153         *must_show_header = 1;
4154         strbuf_init(msg, PATH_MAX * 2 + 300);
4155         switch (p->status) {
4156         case DIFF_STATUS_COPIED:
4157                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4158                             line_prefix, set, similarity_index(p));
4159                 strbuf_addf(msg, "%s\n%s%scopy from ",
4160                             reset,  line_prefix, set);
4161                 quote_c_style(name, msg, NULL, 0);
4162                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4163                 quote_c_style(other, msg, NULL, 0);
4164                 strbuf_addf(msg, "%s\n", reset);
4165                 break;
4166         case DIFF_STATUS_RENAMED:
4167                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4168                             line_prefix, set, similarity_index(p));
4169                 strbuf_addf(msg, "%s\n%s%srename from ",
4170                             reset, line_prefix, set);
4171                 quote_c_style(name, msg, NULL, 0);
4172                 strbuf_addf(msg, "%s\n%s%srename to ",
4173                             reset, line_prefix, set);
4174                 quote_c_style(other, msg, NULL, 0);
4175                 strbuf_addf(msg, "%s\n", reset);
4176                 break;
4177         case DIFF_STATUS_MODIFIED:
4178                 if (p->score) {
4179                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4180                                     line_prefix,
4181                                     set, similarity_index(p), reset);
4182                         break;
4183                 }
4184                 /* fallthru */
4185         default:
4186                 *must_show_header = 0;
4187         }
4188         if (one && two && oidcmp(&one->oid, &two->oid)) {
4189                 const unsigned hexsz = the_hash_algo->hexsz;
4190                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4191
4192                 if (o->flags.binary) {
4193                         mmfile_t mf;
4194                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
4195                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
4196                                 abbrev = hexsz;
4197                 }
4198                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4199                             diff_abbrev_oid(&one->oid, abbrev),
4200                             diff_abbrev_oid(&two->oid, abbrev));
4201                 if (one->mode == two->mode)
4202                         strbuf_addf(msg, " %06o", one->mode);
4203                 strbuf_addf(msg, "%s\n", reset);
4204         }
4205 }
4206
4207 static void run_diff_cmd(const char *pgm,
4208                          const char *name,
4209                          const char *other,
4210                          const char *attr_path,
4211                          struct diff_filespec *one,
4212                          struct diff_filespec *two,
4213                          struct strbuf *msg,
4214                          struct diff_options *o,
4215                          struct diff_filepair *p)
4216 {
4217         const char *xfrm_msg = NULL;
4218         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4219         int must_show_header = 0;
4220
4221
4222         if (o->flags.allow_external) {
4223                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
4224                 if (drv && drv->external)
4225                         pgm = drv->external;
4226         }
4227
4228         if (msg) {
4229                 /*
4230                  * don't use colors when the header is intended for an
4231                  * external diff driver
4232                  */
4233                 fill_metainfo(msg, name, other, one, two, o, p,
4234                               &must_show_header,
4235                               want_color(o->use_color) && !pgm);
4236                 xfrm_msg = msg->len ? msg->buf : NULL;
4237         }
4238
4239         if (pgm) {
4240                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4241                                   complete_rewrite, o);
4242                 return;
4243         }
4244         if (one && two)
4245                 builtin_diff(name, other ? other : name,
4246                              one, two, xfrm_msg, must_show_header,
4247                              o, complete_rewrite);
4248         else
4249                 fprintf(o->file, "* Unmerged path %s\n", name);
4250 }
4251
4252 static void diff_fill_oid_info(struct diff_filespec *one)
4253 {
4254         if (DIFF_FILE_VALID(one)) {
4255                 if (!one->oid_valid) {
4256                         struct stat st;
4257                         if (one->is_stdin) {
4258                                 oidclr(&one->oid);
4259                                 return;
4260                         }
4261                         if (lstat(one->path, &st) < 0)
4262                                 die_errno("stat '%s'", one->path);
4263                         if (index_path(&one->oid, one->path, &st, 0))
4264                                 die("cannot hash %s", one->path);
4265                 }
4266         }
4267         else
4268                 oidclr(&one->oid);
4269 }
4270
4271 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4272 {
4273         /* Strip the prefix but do not molest /dev/null and absolute paths */
4274         if (*namep && **namep != '/') {
4275                 *namep += prefix_length;
4276                 if (**namep == '/')
4277                         ++*namep;
4278         }
4279         if (*otherp && **otherp != '/') {
4280                 *otherp += prefix_length;
4281                 if (**otherp == '/')
4282                         ++*otherp;
4283         }
4284 }
4285
4286 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4287 {
4288         const char *pgm = external_diff();
4289         struct strbuf msg;
4290         struct diff_filespec *one = p->one;
4291         struct diff_filespec *two = p->two;
4292         const char *name;
4293         const char *other;
4294         const char *attr_path;
4295
4296         name  = one->path;
4297         other = (strcmp(name, two->path) ? two->path : NULL);
4298         attr_path = name;
4299         if (o->prefix_length)
4300                 strip_prefix(o->prefix_length, &name, &other);
4301
4302         if (!o->flags.allow_external)
4303                 pgm = NULL;
4304
4305         if (DIFF_PAIR_UNMERGED(p)) {
4306                 run_diff_cmd(pgm, name, NULL, attr_path,
4307                              NULL, NULL, NULL, o, p);
4308                 return;
4309         }
4310
4311         diff_fill_oid_info(one);
4312         diff_fill_oid_info(two);
4313
4314         if (!pgm &&
4315             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4316             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4317                 /*
4318                  * a filepair that changes between file and symlink
4319                  * needs to be split into deletion and creation.
4320                  */
4321                 struct diff_filespec *null = alloc_filespec(two->path);
4322                 run_diff_cmd(NULL, name, other, attr_path,
4323                              one, null, &msg, o, p);
4324                 free(null);
4325                 strbuf_release(&msg);
4326
4327                 null = alloc_filespec(one->path);
4328                 run_diff_cmd(NULL, name, other, attr_path,
4329                              null, two, &msg, o, p);
4330                 free(null);
4331         }
4332         else
4333                 run_diff_cmd(pgm, name, other, attr_path,
4334                              one, two, &msg, o, p);
4335
4336         strbuf_release(&msg);
4337 }
4338
4339 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4340                          struct diffstat_t *diffstat)
4341 {
4342         const char *name;
4343         const char *other;
4344
4345         if (DIFF_PAIR_UNMERGED(p)) {
4346                 /* unmerged */
4347                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4348                 return;
4349         }
4350
4351         name = p->one->path;
4352         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4353
4354         if (o->prefix_length)
4355                 strip_prefix(o->prefix_length, &name, &other);
4356
4357         diff_fill_oid_info(p->one);
4358         diff_fill_oid_info(p->two);
4359
4360         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4361 }
4362
4363 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4364 {
4365         const char *name;
4366         const char *other;
4367         const char *attr_path;
4368
4369         if (DIFF_PAIR_UNMERGED(p)) {
4370                 /* unmerged */
4371                 return;
4372         }
4373
4374         name = p->one->path;
4375         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4376         attr_path = other ? other : name;
4377
4378         if (o->prefix_length)
4379                 strip_prefix(o->prefix_length, &name, &other);
4380
4381         diff_fill_oid_info(p->one);
4382         diff_fill_oid_info(p->two);
4383
4384         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4385 }
4386
4387 void diff_setup(struct diff_options *options)
4388 {
4389         memcpy(options, &default_diff_options, sizeof(*options));
4390
4391         options->file = stdout;
4392
4393         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4394         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4395         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4396         options->abbrev = DEFAULT_ABBREV;
4397         options->line_termination = '\n';
4398         options->break_opt = -1;
4399         options->rename_limit = -1;
4400         options->dirstat_permille = diff_dirstat_permille_default;
4401         options->context = diff_context_default;
4402         options->interhunkcontext = diff_interhunk_context_default;
4403         options->ws_error_highlight = ws_error_highlight_default;
4404         options->flags.rename_empty = 1;
4405         options->objfind = NULL;
4406
4407         /* pathchange left =NULL by default */
4408         options->change = diff_change;
4409         options->add_remove = diff_addremove;
4410         options->use_color = diff_use_color_default;
4411         options->detect_rename = diff_detect_rename_default;
4412         options->xdl_opts |= diff_algorithm;
4413         if (diff_indent_heuristic)
4414                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4415
4416         options->orderfile = diff_order_file_cfg;
4417
4418         if (diff_no_prefix) {
4419                 options->a_prefix = options->b_prefix = "";
4420         } else if (!diff_mnemonic_prefix) {
4421                 options->a_prefix = "a/";
4422                 options->b_prefix = "b/";
4423         }
4424
4425         options->color_moved = diff_color_moved_default;
4426         options->color_moved_ws_handling = diff_color_moved_ws_default;
4427 }
4428
4429 void diff_setup_done(struct diff_options *options)
4430 {
4431         unsigned check_mask = DIFF_FORMAT_NAME |
4432                               DIFF_FORMAT_NAME_STATUS |
4433                               DIFF_FORMAT_CHECKDIFF |
4434                               DIFF_FORMAT_NO_OUTPUT;
4435         /*
4436          * This must be signed because we're comparing against a potentially
4437          * negative value.
4438          */
4439         const int hexsz = the_hash_algo->hexsz;
4440
4441         if (options->set_default)
4442                 options->set_default(options);
4443
4444         if (HAS_MULTI_BITS(options->output_format & check_mask))
4445                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4446
4447         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4448                 die(_("-G, -S and --find-object are mutually exclusive"));
4449
4450         /*
4451          * Most of the time we can say "there are changes"
4452          * only by checking if there are changed paths, but
4453          * --ignore-whitespace* options force us to look
4454          * inside contents.
4455          */
4456
4457         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4458                 options->flags.diff_from_contents = 1;
4459         else
4460                 options->flags.diff_from_contents = 0;
4461
4462         if (options->flags.find_copies_harder)
4463                 options->detect_rename = DIFF_DETECT_COPY;
4464
4465         if (!options->flags.relative_name)
4466                 options->prefix = NULL;
4467         if (options->prefix)
4468                 options->prefix_length = strlen(options->prefix);
4469         else
4470                 options->prefix_length = 0;
4471
4472         if (options->output_format & (DIFF_FORMAT_NAME |
4473                                       DIFF_FORMAT_NAME_STATUS |
4474                                       DIFF_FORMAT_CHECKDIFF |
4475                                       DIFF_FORMAT_NO_OUTPUT))
4476                 options->output_format &= ~(DIFF_FORMAT_RAW |
4477                                             DIFF_FORMAT_NUMSTAT |
4478                                             DIFF_FORMAT_DIFFSTAT |
4479                                             DIFF_FORMAT_SHORTSTAT |
4480                                             DIFF_FORMAT_DIRSTAT |
4481                                             DIFF_FORMAT_SUMMARY |
4482                                             DIFF_FORMAT_PATCH);
4483
4484         /*
4485          * These cases always need recursive; we do not drop caller-supplied
4486          * recursive bits for other formats here.
4487          */
4488         if (options->output_format & (DIFF_FORMAT_PATCH |
4489                                       DIFF_FORMAT_NUMSTAT |
4490                                       DIFF_FORMAT_DIFFSTAT |
4491                                       DIFF_FORMAT_SHORTSTAT |
4492                                       DIFF_FORMAT_DIRSTAT |
4493                                       DIFF_FORMAT_SUMMARY |
4494                                       DIFF_FORMAT_CHECKDIFF))
4495                 options->flags.recursive = 1;
4496         /*
4497          * Also pickaxe would not work very well if you do not say recursive
4498          */
4499         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4500                 options->flags.recursive = 1;
4501         /*
4502          * When patches are generated, submodules diffed against the work tree
4503          * must be checked for dirtiness too so it can be shown in the output
4504          */
4505         if (options->output_format & DIFF_FORMAT_PATCH)
4506                 options->flags.dirty_submodules = 1;
4507
4508         if (options->detect_rename && options->rename_limit < 0)
4509                 options->rename_limit = diff_rename_limit_default;
4510         if (hexsz < options->abbrev)
4511                 options->abbrev = hexsz; /* full */
4512
4513         /*
4514          * It does not make sense to show the first hit we happened
4515          * to have found.  It does not make sense not to return with
4516          * exit code in such a case either.
4517          */
4518         if (options->flags.quick) {
4519                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4520                 options->flags.exit_with_status = 1;
4521         }
4522
4523         options->diff_path_counter = 0;
4524
4525         if (options->flags.follow_renames && options->pathspec.nr != 1)
4526                 die(_("--follow requires exactly one pathspec"));
4527
4528         if (!options->use_color || external_diff())
4529                 options->color_moved = 0;
4530 }
4531
4532 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4533 {
4534         char c, *eq;
4535         int len;
4536
4537         if (*arg != '-')
4538                 return 0;
4539         c = *++arg;
4540         if (!c)
4541                 return 0;
4542         if (c == arg_short) {
4543                 c = *++arg;
4544                 if (!c)
4545                         return 1;
4546                 if (val && isdigit(c)) {
4547                         char *end;
4548                         int n = strtoul(arg, &end, 10);
4549                         if (*end)
4550                                 return 0;
4551                         *val = n;
4552                         return 1;
4553                 }
4554                 return 0;
4555         }
4556         if (c != '-')
4557                 return 0;
4558         arg++;
4559         eq = strchrnul(arg, '=');
4560         len = eq - arg;
4561         if (!len || strncmp(arg, arg_long, len))
4562                 return 0;
4563         if (*eq) {
4564                 int n;
4565                 char *end;
4566                 if (!isdigit(*++eq))
4567                         return 0;
4568                 n = strtoul(eq, &end, 10);
4569                 if (*end)
4570                         return 0;
4571                 *val = n;
4572         }
4573         return 1;
4574 }
4575
4576 static int diff_scoreopt_parse(const char *opt);
4577
4578 static inline int short_opt(char opt, const char **argv,
4579                             const char **optarg)
4580 {
4581         const char *arg = argv[0];
4582         if (arg[0] != '-' || arg[1] != opt)
4583                 return 0;
4584         if (arg[2] != '\0') {
4585                 *optarg = arg + 2;
4586                 return 1;
4587         }
4588         if (!argv[1])
4589                 die("Option '%c' requires a value", opt);
4590         *optarg = argv[1];
4591         return 2;
4592 }
4593
4594 int parse_long_opt(const char *opt, const char **argv,
4595                    const char **optarg)
4596 {
4597         const char *arg = argv[0];
4598         if (!skip_prefix(arg, "--", &arg))
4599                 return 0;
4600         if (!skip_prefix(arg, opt, &arg))
4601                 return 0;
4602         if (*arg == '=') { /* stuck form: --option=value */
4603                 *optarg = arg + 1;
4604                 return 1;
4605         }
4606         if (*arg != '\0')
4607                 return 0;
4608         /* separate form: --option value */
4609         if (!argv[1])
4610                 die("Option '--%s' requires a value", opt);
4611         *optarg = argv[1];
4612         return 2;
4613 }
4614
4615 static int stat_opt(struct diff_options *options, const char **av)
4616 {
4617         const char *arg = av[0];
4618         char *end;
4619         int width = options->stat_width;
4620         int name_width = options->stat_name_width;
4621         int graph_width = options->stat_graph_width;
4622         int count = options->stat_count;
4623         int argcount = 1;
4624
4625         if (!skip_prefix(arg, "--stat", &arg))
4626                 BUG("stat option does not begin with --stat: %s", arg);
4627         end = (char *)arg;
4628
4629         switch (*arg) {
4630         case '-':
4631                 if (skip_prefix(arg, "-width", &arg)) {
4632                         if (*arg == '=')
4633                                 width = strtoul(arg + 1, &end, 10);
4634                         else if (!*arg && !av[1])
4635                                 die_want_option("--stat-width");
4636                         else if (!*arg) {
4637                                 width = strtoul(av[1], &end, 10);
4638                                 argcount = 2;
4639                         }
4640                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4641                         if (*arg == '=')
4642                                 name_width = strtoul(arg + 1, &end, 10);
4643                         else if (!*arg && !av[1])
4644                                 die_want_option("--stat-name-width");
4645                         else if (!*arg) {
4646                                 name_width = strtoul(av[1], &end, 10);
4647                                 argcount = 2;
4648                         }
4649                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4650                         if (*arg == '=')
4651                                 graph_width = strtoul(arg + 1, &end, 10);
4652                         else if (!*arg && !av[1])
4653                                 die_want_option("--stat-graph-width");
4654                         else if (!*arg) {
4655                                 graph_width = strtoul(av[1], &end, 10);
4656                                 argcount = 2;
4657                         }
4658                 } else if (skip_prefix(arg, "-count", &arg)) {
4659                         if (*arg == '=')
4660                                 count = strtoul(arg + 1, &end, 10);
4661                         else if (!*arg && !av[1])
4662                                 die_want_option("--stat-count");
4663                         else if (!*arg) {
4664                                 count = strtoul(av[1], &end, 10);
4665                                 argcount = 2;
4666                         }
4667                 }
4668                 break;
4669         case '=':
4670                 width = strtoul(arg+1, &end, 10);
4671                 if (*end == ',')
4672                         name_width = strtoul(end+1, &end, 10);
4673                 if (*end == ',')
4674                         count = strtoul(end+1, &end, 10);
4675         }
4676
4677         /* Important! This checks all the error cases! */
4678         if (*end)
4679                 return 0;
4680         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4681         options->stat_name_width = name_width;
4682         options->stat_graph_width = graph_width;
4683         options->stat_width = width;
4684         options->stat_count = count;
4685         return argcount;
4686 }
4687
4688 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4689 {
4690         struct strbuf errmsg = STRBUF_INIT;
4691         if (parse_dirstat_params(options, params, &errmsg))
4692                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4693                     errmsg.buf);
4694         strbuf_release(&errmsg);
4695         /*
4696          * The caller knows a dirstat-related option is given from the command
4697          * line; allow it to say "return this_function();"
4698          */
4699         options->output_format |= DIFF_FORMAT_DIRSTAT;
4700         return 1;
4701 }
4702
4703 static int parse_submodule_opt(struct diff_options *options, const char *value)
4704 {
4705         if (parse_submodule_params(options, value))
4706                 die(_("Failed to parse --submodule option parameter: '%s'"),
4707                         value);
4708         return 1;
4709 }
4710
4711 static const char diff_status_letters[] = {
4712         DIFF_STATUS_ADDED,
4713         DIFF_STATUS_COPIED,
4714         DIFF_STATUS_DELETED,
4715         DIFF_STATUS_MODIFIED,
4716         DIFF_STATUS_RENAMED,
4717         DIFF_STATUS_TYPE_CHANGED,
4718         DIFF_STATUS_UNKNOWN,
4719         DIFF_STATUS_UNMERGED,
4720         DIFF_STATUS_FILTER_AON,
4721         DIFF_STATUS_FILTER_BROKEN,
4722         '\0',
4723 };
4724
4725 static unsigned int filter_bit['Z' + 1];
4726
4727 static void prepare_filter_bits(void)
4728 {
4729         int i;
4730
4731         if (!filter_bit[DIFF_STATUS_ADDED]) {
4732                 for (i = 0; diff_status_letters[i]; i++)
4733                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4734         }
4735 }
4736
4737 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4738 {
4739         return opt->filter & filter_bit[(int) status];
4740 }
4741
4742 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4743 {
4744         int i, optch;
4745
4746         prepare_filter_bits();
4747
4748         /*
4749          * If there is a negation e.g. 'd' in the input, and we haven't
4750          * initialized the filter field with another --diff-filter, start
4751          * from full set of bits, except for AON.
4752          */
4753         if (!opt->filter) {
4754                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4755                         if (optch < 'a' || 'z' < optch)
4756                                 continue;
4757                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4758                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4759                         break;
4760                 }
4761         }
4762
4763         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4764                 unsigned int bit;
4765                 int negate;
4766
4767                 if ('a' <= optch && optch <= 'z') {
4768                         negate = 1;
4769                         optch = toupper(optch);
4770                 } else {
4771                         negate = 0;
4772                 }
4773
4774                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4775                 if (!bit)
4776                         return optarg[i];
4777                 if (negate)
4778                         opt->filter &= ~bit;
4779                 else
4780                         opt->filter |= bit;
4781         }
4782         return 0;
4783 }
4784
4785 static void enable_patch_output(int *fmt) {
4786         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4787         *fmt |= DIFF_FORMAT_PATCH;
4788 }
4789
4790 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4791 {
4792         int val = parse_ws_error_highlight(arg);
4793
4794         if (val < 0) {
4795                 error("unknown value after ws-error-highlight=%.*s",
4796                       -1 - val, arg);
4797                 return 0;
4798         }
4799         opt->ws_error_highlight = val;
4800         return 1;
4801 }
4802
4803 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4804 {
4805         struct object_id oid;
4806
4807         if (get_oid(arg, &oid))
4808                 return error("unable to resolve '%s'", arg);
4809
4810         if (!opt->objfind)
4811                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4812
4813         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4814         opt->flags.recursive = 1;
4815         opt->flags.tree_in_recursive = 1;
4816         oidset_insert(opt->objfind, &oid);
4817         return 1;
4818 }
4819
4820 int diff_opt_parse(struct diff_options *options,
4821                    const char **av, int ac, const char *prefix)
4822 {
4823         const char *arg = av[0];
4824         const char *optarg;
4825         int argcount;
4826
4827         if (!prefix)
4828                 prefix = "";
4829
4830         /* Output format options */
4831         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4832             || opt_arg(arg, 'U', "unified", &options->context))
4833                 enable_patch_output(&options->output_format);
4834         else if (!strcmp(arg, "--raw"))
4835                 options->output_format |= DIFF_FORMAT_RAW;
4836         else if (!strcmp(arg, "--patch-with-raw")) {
4837                 enable_patch_output(&options->output_format);
4838                 options->output_format |= DIFF_FORMAT_RAW;
4839         } else if (!strcmp(arg, "--numstat"))
4840                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4841         else if (!strcmp(arg, "--shortstat"))
4842                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4843         else if (skip_prefix(arg, "-X", &arg) ||
4844                  skip_to_optional_arg(arg, "--dirstat", &arg))
4845                 return parse_dirstat_opt(options, arg);
4846         else if (!strcmp(arg, "--cumulative"))
4847                 return parse_dirstat_opt(options, "cumulative");
4848         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4849                 parse_dirstat_opt(options, "files");
4850                 return parse_dirstat_opt(options, arg);
4851         }
4852         else if (!strcmp(arg, "--check"))
4853                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4854         else if (!strcmp(arg, "--summary"))
4855                 options->output_format |= DIFF_FORMAT_SUMMARY;
4856         else if (!strcmp(arg, "--patch-with-stat")) {
4857                 enable_patch_output(&options->output_format);
4858                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4859         } else if (!strcmp(arg, "--name-only"))
4860                 options->output_format |= DIFF_FORMAT_NAME;
4861         else if (!strcmp(arg, "--name-status"))
4862                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4863         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4864                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4865         else if (starts_with(arg, "--stat"))
4866                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4867                 return stat_opt(options, av);
4868         else if (!strcmp(arg, "--compact-summary")) {
4869                  options->flags.stat_with_summary = 1;
4870                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4871         } else if (!strcmp(arg, "--no-compact-summary"))
4872                  options->flags.stat_with_summary = 0;
4873         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4874                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4875         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4876                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4877         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4878                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4879
4880         /* renames options */
4881         else if (starts_with(arg, "-B") ||
4882                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4883                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4884                         return error("invalid argument to -B: %s", arg+2);
4885         }
4886         else if (starts_with(arg, "-M") ||
4887                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4888                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4889                         return error("invalid argument to -M: %s", arg+2);
4890                 options->detect_rename = DIFF_DETECT_RENAME;
4891         }
4892         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4893                 options->irreversible_delete = 1;
4894         }
4895         else if (starts_with(arg, "-C") ||
4896                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4897                 if (options->detect_rename == DIFF_DETECT_COPY)
4898                         options->flags.find_copies_harder = 1;
4899                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4900                         return error("invalid argument to -C: %s", arg+2);
4901                 options->detect_rename = DIFF_DETECT_COPY;
4902         }
4903         else if (!strcmp(arg, "--no-renames"))
4904                 options->detect_rename = 0;
4905         else if (!strcmp(arg, "--rename-empty"))
4906                 options->flags.rename_empty = 1;
4907         else if (!strcmp(arg, "--no-rename-empty"))
4908                 options->flags.rename_empty = 0;
4909         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4910                 options->flags.relative_name = 1;
4911                 if (arg)
4912                         options->prefix = arg;
4913         }
4914
4915         /* xdiff options */
4916         else if (!strcmp(arg, "--minimal"))
4917                 DIFF_XDL_SET(options, NEED_MINIMAL);
4918         else if (!strcmp(arg, "--no-minimal"))
4919                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4920         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4921                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4922         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4923                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4924         else if (!strcmp(arg, "--ignore-space-at-eol"))
4925                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4926         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4927                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4928         else if (!strcmp(arg, "--ignore-blank-lines"))
4929                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4930         else if (!strcmp(arg, "--indent-heuristic"))
4931                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4932         else if (!strcmp(arg, "--no-indent-heuristic"))
4933                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4934         else if (!strcmp(arg, "--patience")) {
4935                 int i;
4936                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4937                 /*
4938                  * Both --patience and --anchored use PATIENCE_DIFF
4939                  * internally, so remove any anchors previously
4940                  * specified.
4941                  */
4942                 for (i = 0; i < options->anchors_nr; i++)
4943                         free(options->anchors[i]);
4944                 options->anchors_nr = 0;
4945         } else if (!strcmp(arg, "--histogram"))
4946                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4947         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4948                 long value = parse_algorithm_value(optarg);
4949                 if (value < 0)
4950                         return error("option diff-algorithm accepts \"myers\", "
4951                                      "\"minimal\", \"patience\" and \"histogram\"");
4952                 /* clear out previous settings */
4953                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4954                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4955                 options->xdl_opts |= value;
4956                 return argcount;
4957         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4958                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4959                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4960                            options->anchors_alloc);
4961                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4962         }
4963
4964         /* flags options */
4965         else if (!strcmp(arg, "--binary")) {
4966                 enable_patch_output(&options->output_format);
4967                 options->flags.binary = 1;
4968         }
4969         else if (!strcmp(arg, "--full-index"))
4970                 options->flags.full_index = 1;
4971         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4972                 options->flags.text = 1;
4973         else if (!strcmp(arg, "-R"))
4974                 options->flags.reverse_diff = 1;
4975         else if (!strcmp(arg, "--find-copies-harder"))
4976                 options->flags.find_copies_harder = 1;
4977         else if (!strcmp(arg, "--follow"))
4978                 options->flags.follow_renames = 1;
4979         else if (!strcmp(arg, "--no-follow")) {
4980                 options->flags.follow_renames = 0;
4981                 options->flags.default_follow_renames = 0;
4982         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4983                 int value = git_config_colorbool(NULL, arg);
4984                 if (value < 0)
4985                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4986                 options->use_color = value;
4987         }
4988         else if (!strcmp(arg, "--no-color"))
4989                 options->use_color = 0;
4990         else if (!strcmp(arg, "--color-moved")) {
4991                 if (diff_color_moved_default)
4992                         options->color_moved = diff_color_moved_default;
4993                 if (options->color_moved == COLOR_MOVED_NO)
4994                         options->color_moved = COLOR_MOVED_DEFAULT;
4995         } else if (!strcmp(arg, "--no-color-moved"))
4996                 options->color_moved = COLOR_MOVED_NO;
4997         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4998                 int cm = parse_color_moved(arg);
4999                 if (cm < 0)
5000                         die("bad --color-moved argument: %s", arg);
5001                 options->color_moved = cm;
5002         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5003                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5004         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5005                 options->use_color = 1;
5006                 options->word_diff = DIFF_WORDS_COLOR;
5007         }
5008         else if (!strcmp(arg, "--word-diff")) {
5009                 if (options->word_diff == DIFF_WORDS_NONE)
5010                         options->word_diff = DIFF_WORDS_PLAIN;
5011         }
5012         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5013                 if (!strcmp(arg, "plain"))
5014                         options->word_diff = DIFF_WORDS_PLAIN;
5015                 else if (!strcmp(arg, "color")) {
5016                         options->use_color = 1;
5017                         options->word_diff = DIFF_WORDS_COLOR;
5018                 }
5019                 else if (!strcmp(arg, "porcelain"))
5020                         options->word_diff = DIFF_WORDS_PORCELAIN;
5021                 else if (!strcmp(arg, "none"))
5022                         options->word_diff = DIFF_WORDS_NONE;
5023                 else
5024                         die("bad --word-diff argument: %s", arg);
5025         }
5026         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5027                 if (options->word_diff == DIFF_WORDS_NONE)
5028                         options->word_diff = DIFF_WORDS_PLAIN;
5029                 options->word_regex = optarg;
5030                 return argcount;
5031         }
5032         else if (!strcmp(arg, "--exit-code"))
5033                 options->flags.exit_with_status = 1;
5034         else if (!strcmp(arg, "--quiet"))
5035                 options->flags.quick = 1;
5036         else if (!strcmp(arg, "--ext-diff"))
5037                 options->flags.allow_external = 1;
5038         else if (!strcmp(arg, "--no-ext-diff"))
5039                 options->flags.allow_external = 0;
5040         else if (!strcmp(arg, "--textconv")) {
5041                 options->flags.allow_textconv = 1;
5042                 options->flags.textconv_set_via_cmdline = 1;
5043         } else if (!strcmp(arg, "--no-textconv"))
5044                 options->flags.allow_textconv = 0;
5045         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5046                 options->flags.override_submodule_config = 1;
5047                 handle_ignore_submodules_arg(options, arg);
5048         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5049                 return parse_submodule_opt(options, arg);
5050         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5051                 return parse_ws_error_highlight_opt(options, arg);
5052         else if (!strcmp(arg, "--ita-invisible-in-index"))
5053                 options->ita_invisible_in_index = 1;
5054         else if (!strcmp(arg, "--ita-visible-in-index"))
5055                 options->ita_invisible_in_index = 0;
5056
5057         /* misc options */
5058         else if (!strcmp(arg, "-z"))
5059                 options->line_termination = 0;
5060         else if ((argcount = short_opt('l', av, &optarg))) {
5061                 options->rename_limit = strtoul(optarg, NULL, 10);
5062                 return argcount;
5063         }
5064         else if ((argcount = short_opt('S', av, &optarg))) {
5065                 options->pickaxe = optarg;
5066                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5067                 return argcount;
5068         } else if ((argcount = short_opt('G', av, &optarg))) {
5069                 options->pickaxe = optarg;
5070                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5071                 return argcount;
5072         }
5073         else if (!strcmp(arg, "--pickaxe-all"))
5074                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5075         else if (!strcmp(arg, "--pickaxe-regex"))
5076                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5077         else if ((argcount = short_opt('O', av, &optarg))) {
5078                 options->orderfile = prefix_filename(prefix, optarg);
5079                 return argcount;
5080         } else if (skip_prefix(arg, "--find-object=", &arg))
5081                 return parse_objfind_opt(options, arg);
5082         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5083                 int offending = parse_diff_filter_opt(optarg, options);
5084                 if (offending)
5085                         die("unknown change class '%c' in --diff-filter=%s",
5086                             offending, optarg);
5087                 return argcount;
5088         }
5089         else if (!strcmp(arg, "--no-abbrev"))
5090                 options->abbrev = 0;
5091         else if (!strcmp(arg, "--abbrev"))
5092                 options->abbrev = DEFAULT_ABBREV;
5093         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5094                 options->abbrev = strtoul(arg, NULL, 10);
5095                 if (options->abbrev < MINIMUM_ABBREV)
5096                         options->abbrev = MINIMUM_ABBREV;
5097                 else if (the_hash_algo->hexsz < options->abbrev)
5098                         options->abbrev = the_hash_algo->hexsz;
5099         }
5100         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5101                 options->a_prefix = optarg;
5102                 return argcount;
5103         }
5104         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5105                 options->line_prefix = optarg;
5106                 options->line_prefix_length = strlen(options->line_prefix);
5107                 graph_setup_line_prefix(options);
5108                 return argcount;
5109         }
5110         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5111                 options->b_prefix = optarg;
5112                 return argcount;
5113         }
5114         else if (!strcmp(arg, "--no-prefix"))
5115                 options->a_prefix = options->b_prefix = "";
5116         else if (opt_arg(arg, '\0', "inter-hunk-context",
5117                          &options->interhunkcontext))
5118                 ;
5119         else if (!strcmp(arg, "-W"))
5120                 options->flags.funccontext = 1;
5121         else if (!strcmp(arg, "--function-context"))
5122                 options->flags.funccontext = 1;
5123         else if (!strcmp(arg, "--no-function-context"))
5124                 options->flags.funccontext = 0;
5125         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5126                 char *path = prefix_filename(prefix, optarg);
5127                 options->file = xfopen(path, "w");
5128                 options->close_file = 1;
5129                 if (options->use_color != GIT_COLOR_ALWAYS)
5130                         options->use_color = GIT_COLOR_NEVER;
5131                 free(path);
5132                 return argcount;
5133         } else
5134                 return 0;
5135         return 1;
5136 }
5137
5138 int parse_rename_score(const char **cp_p)
5139 {
5140         unsigned long num, scale;
5141         int ch, dot;
5142         const char *cp = *cp_p;
5143
5144         num = 0;
5145         scale = 1;
5146         dot = 0;
5147         for (;;) {
5148                 ch = *cp;
5149                 if ( !dot && ch == '.' ) {
5150                         scale = 1;
5151                         dot = 1;
5152                 } else if ( ch == '%' ) {
5153                         scale = dot ? scale*100 : 100;
5154                         cp++;   /* % is always at the end */
5155                         break;
5156                 } else if ( ch >= '0' && ch <= '9' ) {
5157                         if ( scale < 100000 ) {
5158                                 scale *= 10;
5159                                 num = (num*10) + (ch-'0');
5160                         }
5161                 } else {
5162                         break;
5163                 }
5164                 cp++;
5165         }
5166         *cp_p = cp;
5167
5168         /* user says num divided by scale and we say internally that
5169          * is MAX_SCORE * num / scale.
5170          */
5171         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5172 }
5173
5174 static int diff_scoreopt_parse(const char *opt)
5175 {
5176         int opt1, opt2, cmd;
5177
5178         if (*opt++ != '-')
5179                 return -1;
5180         cmd = *opt++;
5181         if (cmd == '-') {
5182                 /* convert the long-form arguments into short-form versions */
5183                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5184                         if (*opt == 0 || *opt++ == '=')
5185                                 cmd = 'B';
5186                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5187                         if (*opt == 0 || *opt++ == '=')
5188                                 cmd = 'C';
5189                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5190                         if (*opt == 0 || *opt++ == '=')
5191                                 cmd = 'M';
5192                 }
5193         }
5194         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5195                 return -1; /* that is not a -M, -C, or -B option */
5196
5197         opt1 = parse_rename_score(&opt);
5198         if (cmd != 'B')
5199                 opt2 = 0;
5200         else {
5201                 if (*opt == 0)
5202                         opt2 = 0;
5203                 else if (*opt != '/')
5204                         return -1; /* we expect -B80/99 or -B80 */
5205                 else {
5206                         opt++;
5207                         opt2 = parse_rename_score(&opt);
5208                 }
5209         }
5210         if (*opt != 0)
5211                 return -1;
5212         return opt1 | (opt2 << 16);
5213 }
5214
5215 struct diff_queue_struct diff_queued_diff;
5216
5217 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5218 {
5219         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5220         queue->queue[queue->nr++] = dp;
5221 }
5222
5223 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5224                                  struct diff_filespec *one,
5225                                  struct diff_filespec *two)
5226 {
5227         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5228         dp->one = one;
5229         dp->two = two;
5230         if (queue)
5231                 diff_q(queue, dp);
5232         return dp;
5233 }
5234
5235 void diff_free_filepair(struct diff_filepair *p)
5236 {
5237         free_filespec(p->one);
5238         free_filespec(p->two);
5239         free(p);
5240 }
5241
5242 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5243 {
5244         int abblen;
5245         const char *abbrev;
5246
5247         /* Do we want all 40 hex characters? */
5248         if (len == the_hash_algo->hexsz)
5249                 return oid_to_hex(oid);
5250
5251         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5252         abbrev = diff_abbrev_oid(oid, len);
5253
5254         if (!print_sha1_ellipsis())
5255                 return abbrev;
5256
5257         abblen = strlen(abbrev);
5258
5259         /*
5260          * In well-behaved cases, where the abbreviated result is the
5261          * same as the requested length, append three dots after the
5262          * abbreviation (hence the whole logic is limited to the case
5263          * where abblen < 37); when the actual abbreviated result is a
5264          * bit longer than the requested length, we reduce the number
5265          * of dots so that they match the well-behaved ones.  However,
5266          * if the actual abbreviation is longer than the requested
5267          * length by more than three, we give up on aligning, and add
5268          * three dots anyway, to indicate that the output is not the
5269          * full object name.  Yes, this may be suboptimal, but this
5270          * appears only in "diff --raw --abbrev" output and it is not
5271          * worth the effort to change it now.  Note that this would
5272          * likely to work fine when the automatic sizing of default
5273          * abbreviation length is used--we would be fed -1 in "len" in
5274          * that case, and will end up always appending three-dots, but
5275          * the automatic sizing is supposed to give abblen that ensures
5276          * uniqueness across all objects (statistically speaking).
5277          */
5278         if (abblen < the_hash_algo->hexsz - 3) {
5279                 static char hex[GIT_MAX_HEXSZ + 1];
5280                 if (len < abblen && abblen <= len + 2)
5281                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5282                 else
5283                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5284                 return hex;
5285         }
5286
5287         return oid_to_hex(oid);
5288 }
5289
5290 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5291 {
5292         int line_termination = opt->line_termination;
5293         int inter_name_termination = line_termination ? '\t' : '\0';
5294
5295         fprintf(opt->file, "%s", diff_line_prefix(opt));
5296         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5297                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5298                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5299                 fprintf(opt->file, "%s ",
5300                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5301         }
5302         if (p->score) {
5303                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5304                         inter_name_termination);
5305         } else {
5306                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5307         }
5308
5309         if (p->status == DIFF_STATUS_COPIED ||
5310             p->status == DIFF_STATUS_RENAMED) {
5311                 const char *name_a, *name_b;
5312                 name_a = p->one->path;
5313                 name_b = p->two->path;
5314                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5315                 write_name_quoted(name_a, opt->file, inter_name_termination);
5316                 write_name_quoted(name_b, opt->file, line_termination);
5317         } else {
5318                 const char *name_a, *name_b;
5319                 name_a = p->one->mode ? p->one->path : p->two->path;
5320                 name_b = NULL;
5321                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5322                 write_name_quoted(name_a, opt->file, line_termination);
5323         }
5324 }
5325
5326 int diff_unmodified_pair(struct diff_filepair *p)
5327 {
5328         /* This function is written stricter than necessary to support
5329          * the currently implemented transformers, but the idea is to
5330          * let transformers to produce diff_filepairs any way they want,
5331          * and filter and clean them up here before producing the output.
5332          */
5333         struct diff_filespec *one = p->one, *two = p->two;
5334
5335         if (DIFF_PAIR_UNMERGED(p))
5336                 return 0; /* unmerged is interesting */
5337
5338         /* deletion, addition, mode or type change
5339          * and rename are all interesting.
5340          */
5341         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5342             DIFF_PAIR_MODE_CHANGED(p) ||
5343             strcmp(one->path, two->path))
5344                 return 0;
5345
5346         /* both are valid and point at the same path.  that is, we are
5347          * dealing with a change.
5348          */
5349         if (one->oid_valid && two->oid_valid &&
5350             !oidcmp(&one->oid, &two->oid) &&
5351             !one->dirty_submodule && !two->dirty_submodule)
5352                 return 1; /* no change */
5353         if (!one->oid_valid && !two->oid_valid)
5354                 return 1; /* both look at the same file on the filesystem. */
5355         return 0;
5356 }
5357
5358 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5359 {
5360         if (diff_unmodified_pair(p))
5361                 return;
5362
5363         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5364             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5365                 return; /* no tree diffs in patch format */
5366
5367         run_diff(p, o);
5368 }
5369
5370 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5371                             struct diffstat_t *diffstat)
5372 {
5373         if (diff_unmodified_pair(p))
5374                 return;
5375
5376         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5377             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5378                 return; /* no useful stat for tree diffs */
5379
5380         run_diffstat(p, o, diffstat);
5381 }
5382
5383 static void diff_flush_checkdiff(struct diff_filepair *p,
5384                 struct diff_options *o)
5385 {
5386         if (diff_unmodified_pair(p))
5387                 return;
5388
5389         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5390             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5391                 return; /* nothing to check in tree diffs */
5392
5393         run_checkdiff(p, o);
5394 }
5395
5396 int diff_queue_is_empty(void)
5397 {
5398         struct diff_queue_struct *q = &diff_queued_diff;
5399         int i;
5400         for (i = 0; i < q->nr; i++)
5401                 if (!diff_unmodified_pair(q->queue[i]))
5402                         return 0;
5403         return 1;
5404 }
5405
5406 #if DIFF_DEBUG
5407 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5408 {
5409         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5410                 x, one ? one : "",
5411                 s->path,
5412                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5413                 s->mode,
5414                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5415         fprintf(stderr, "queue[%d] %s size %lu\n",
5416                 x, one ? one : "",
5417                 s->size);
5418 }
5419
5420 void diff_debug_filepair(const struct diff_filepair *p, int i)
5421 {
5422         diff_debug_filespec(p->one, i, "one");
5423         diff_debug_filespec(p->two, i, "two");
5424         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5425                 p->score, p->status ? p->status : '?',
5426                 p->one->rename_used, p->broken_pair);
5427 }
5428
5429 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5430 {
5431         int i;
5432         if (msg)
5433                 fprintf(stderr, "%s\n", msg);
5434         fprintf(stderr, "q->nr = %d\n", q->nr);
5435         for (i = 0; i < q->nr; i++) {
5436                 struct diff_filepair *p = q->queue[i];
5437                 diff_debug_filepair(p, i);
5438         }
5439 }
5440 #endif
5441
5442 static void diff_resolve_rename_copy(void)
5443 {
5444         int i;
5445         struct diff_filepair *p;
5446         struct diff_queue_struct *q = &diff_queued_diff;
5447
5448         diff_debug_queue("resolve-rename-copy", q);
5449
5450         for (i = 0; i < q->nr; i++) {
5451                 p = q->queue[i];
5452                 p->status = 0; /* undecided */
5453                 if (DIFF_PAIR_UNMERGED(p))
5454                         p->status = DIFF_STATUS_UNMERGED;
5455                 else if (!DIFF_FILE_VALID(p->one))
5456                         p->status = DIFF_STATUS_ADDED;
5457                 else if (!DIFF_FILE_VALID(p->two))
5458                         p->status = DIFF_STATUS_DELETED;
5459                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5460                         p->status = DIFF_STATUS_TYPE_CHANGED;
5461
5462                 /* from this point on, we are dealing with a pair
5463                  * whose both sides are valid and of the same type, i.e.
5464                  * either in-place edit or rename/copy edit.
5465                  */
5466                 else if (DIFF_PAIR_RENAME(p)) {
5467                         /*
5468                          * A rename might have re-connected a broken
5469                          * pair up, causing the pathnames to be the
5470                          * same again. If so, that's not a rename at
5471                          * all, just a modification..
5472                          *
5473                          * Otherwise, see if this source was used for
5474                          * multiple renames, in which case we decrement
5475                          * the count, and call it a copy.
5476                          */
5477                         if (!strcmp(p->one->path, p->two->path))
5478                                 p->status = DIFF_STATUS_MODIFIED;
5479                         else if (--p->one->rename_used > 0)
5480                                 p->status = DIFF_STATUS_COPIED;
5481                         else
5482                                 p->status = DIFF_STATUS_RENAMED;
5483                 }
5484                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5485                          p->one->mode != p->two->mode ||
5486                          p->one->dirty_submodule ||
5487                          p->two->dirty_submodule ||
5488                          is_null_oid(&p->one->oid))
5489                         p->status = DIFF_STATUS_MODIFIED;
5490                 else {
5491                         /* This is a "no-change" entry and should not
5492                          * happen anymore, but prepare for broken callers.
5493                          */
5494                         error("feeding unmodified %s to diffcore",
5495                               p->one->path);
5496                         p->status = DIFF_STATUS_UNKNOWN;
5497                 }
5498         }
5499         diff_debug_queue("resolve-rename-copy done", q);
5500 }
5501
5502 static int check_pair_status(struct diff_filepair *p)
5503 {
5504         switch (p->status) {
5505         case DIFF_STATUS_UNKNOWN:
5506                 return 0;
5507         case 0:
5508                 die("internal error in diff-resolve-rename-copy");
5509         default:
5510                 return 1;
5511         }
5512 }
5513
5514 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5515 {
5516         int fmt = opt->output_format;
5517
5518         if (fmt & DIFF_FORMAT_CHECKDIFF)
5519                 diff_flush_checkdiff(p, opt);
5520         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5521                 diff_flush_raw(p, opt);
5522         else if (fmt & DIFF_FORMAT_NAME) {
5523                 const char *name_a, *name_b;
5524                 name_a = p->two->path;
5525                 name_b = NULL;
5526                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5527                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5528                 write_name_quoted(name_a, opt->file, opt->line_termination);
5529         }
5530 }
5531
5532 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5533 {
5534         struct strbuf sb = STRBUF_INIT;
5535         if (fs->mode)
5536                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5537         else
5538                 strbuf_addf(&sb, " %s ", newdelete);
5539
5540         quote_c_style(fs->path, &sb, NULL, 0);
5541         strbuf_addch(&sb, '\n');
5542         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5543                          sb.buf, sb.len, 0);
5544         strbuf_release(&sb);
5545 }
5546
5547 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5548                 int show_name)
5549 {
5550         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5551                 struct strbuf sb = STRBUF_INIT;
5552                 strbuf_addf(&sb, " mode change %06o => %06o",
5553                             p->one->mode, p->two->mode);
5554                 if (show_name) {
5555                         strbuf_addch(&sb, ' ');
5556                         quote_c_style(p->two->path, &sb, NULL, 0);
5557                 }
5558                 strbuf_addch(&sb, '\n');
5559                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5560                                  sb.buf, sb.len, 0);
5561                 strbuf_release(&sb);
5562         }
5563 }
5564
5565 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5566                 struct diff_filepair *p)
5567 {
5568         struct strbuf sb = STRBUF_INIT;
5569         struct strbuf names = STRBUF_INIT;
5570
5571         pprint_rename(&names, p->one->path, p->two->path);
5572         strbuf_addf(&sb, " %s %s (%d%%)\n",
5573                     renamecopy, names.buf, similarity_index(p));
5574         strbuf_release(&names);
5575         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5576                                  sb.buf, sb.len, 0);
5577         show_mode_change(opt, p, 0);
5578         strbuf_release(&sb);
5579 }
5580
5581 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5582 {
5583         switch(p->status) {
5584         case DIFF_STATUS_DELETED:
5585                 show_file_mode_name(opt, "delete", p->one);
5586                 break;
5587         case DIFF_STATUS_ADDED:
5588                 show_file_mode_name(opt, "create", p->two);
5589                 break;
5590         case DIFF_STATUS_COPIED:
5591                 show_rename_copy(opt, "copy", p);
5592                 break;
5593         case DIFF_STATUS_RENAMED:
5594                 show_rename_copy(opt, "rename", p);
5595                 break;
5596         default:
5597                 if (p->score) {
5598                         struct strbuf sb = STRBUF_INIT;
5599                         strbuf_addstr(&sb, " rewrite ");
5600                         quote_c_style(p->two->path, &sb, NULL, 0);
5601                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5602                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5603                                          sb.buf, sb.len, 0);
5604                         strbuf_release(&sb);
5605                 }
5606                 show_mode_change(opt, p, !p->score);
5607                 break;
5608         }
5609 }
5610
5611 struct patch_id_t {
5612         git_SHA_CTX *ctx;
5613         int patchlen;
5614 };
5615
5616 static int remove_space(char *line, int len)
5617 {
5618         int i;
5619         char *dst = line;
5620         unsigned char c;
5621
5622         for (i = 0; i < len; i++)
5623                 if (!isspace((c = line[i])))
5624                         *dst++ = c;
5625
5626         return dst - line;
5627 }
5628
5629 static void patch_id_consume(void *priv, char *line, unsigned long len)
5630 {
5631         struct patch_id_t *data = priv;
5632         int new_len;
5633
5634         /* Ignore line numbers when computing the SHA1 of the patch */
5635         if (starts_with(line, "@@ -"))
5636                 return;
5637
5638         new_len = remove_space(line, len);
5639
5640         git_SHA1_Update(data->ctx, line, new_len);
5641         data->patchlen += new_len;
5642 }
5643
5644 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5645 {
5646         git_SHA1_Update(ctx, str, strlen(str));
5647 }
5648
5649 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5650 {
5651         /* large enough for 2^32 in octal */
5652         char buf[12];
5653         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5654         git_SHA1_Update(ctx, buf, len);
5655 }
5656
5657 /* returns 0 upon success, and writes result into sha1 */
5658 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5659 {
5660         struct diff_queue_struct *q = &diff_queued_diff;
5661         int i;
5662         git_SHA_CTX ctx;
5663         struct patch_id_t data;
5664
5665         git_SHA1_Init(&ctx);
5666         memset(&data, 0, sizeof(struct patch_id_t));
5667         data.ctx = &ctx;
5668
5669         for (i = 0; i < q->nr; i++) {
5670                 xpparam_t xpp;
5671                 xdemitconf_t xecfg;
5672                 mmfile_t mf1, mf2;
5673                 struct diff_filepair *p = q->queue[i];
5674                 int len1, len2;
5675
5676                 memset(&xpp, 0, sizeof(xpp));
5677                 memset(&xecfg, 0, sizeof(xecfg));
5678                 if (p->status == 0)
5679                         return error("internal diff status error");
5680                 if (p->status == DIFF_STATUS_UNKNOWN)
5681                         continue;
5682                 if (diff_unmodified_pair(p))
5683                         continue;
5684                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5685                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5686                         continue;
5687                 if (DIFF_PAIR_UNMERGED(p))
5688                         continue;
5689
5690                 diff_fill_oid_info(p->one);
5691                 diff_fill_oid_info(p->two);
5692
5693                 len1 = remove_space(p->one->path, strlen(p->one->path));
5694                 len2 = remove_space(p->two->path, strlen(p->two->path));
5695                 patch_id_add_string(&ctx, "diff--git");
5696                 patch_id_add_string(&ctx, "a/");
5697                 git_SHA1_Update(&ctx, p->one->path, len1);
5698                 patch_id_add_string(&ctx, "b/");
5699                 git_SHA1_Update(&ctx, p->two->path, len2);
5700
5701                 if (p->one->mode == 0) {
5702                         patch_id_add_string(&ctx, "newfilemode");
5703                         patch_id_add_mode(&ctx, p->two->mode);
5704                         patch_id_add_string(&ctx, "---/dev/null");
5705                         patch_id_add_string(&ctx, "+++b/");
5706                         git_SHA1_Update(&ctx, p->two->path, len2);
5707                 } else if (p->two->mode == 0) {
5708                         patch_id_add_string(&ctx, "deletedfilemode");
5709                         patch_id_add_mode(&ctx, p->one->mode);
5710                         patch_id_add_string(&ctx, "---a/");
5711                         git_SHA1_Update(&ctx, p->one->path, len1);
5712                         patch_id_add_string(&ctx, "+++/dev/null");
5713                 } else {
5714                         patch_id_add_string(&ctx, "---a/");
5715                         git_SHA1_Update(&ctx, p->one->path, len1);
5716                         patch_id_add_string(&ctx, "+++b/");
5717                         git_SHA1_Update(&ctx, p->two->path, len2);
5718                 }
5719
5720                 if (diff_header_only)
5721                         continue;
5722
5723                 if (fill_mmfile(&mf1, p->one) < 0 ||
5724                     fill_mmfile(&mf2, p->two) < 0)
5725                         return error("unable to read files to diff");
5726
5727                 if (diff_filespec_is_binary(p->one) ||
5728                     diff_filespec_is_binary(p->two)) {
5729                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5730                                         GIT_SHA1_HEXSZ);
5731                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5732                                         GIT_SHA1_HEXSZ);
5733                         continue;
5734                 }
5735
5736                 xpp.flags = 0;
5737                 xecfg.ctxlen = 3;
5738                 xecfg.flags = 0;
5739                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5740                                   &xpp, &xecfg))
5741                         return error("unable to generate patch-id diff for %s",
5742                                      p->one->path);
5743         }
5744
5745         git_SHA1_Final(oid->hash, &ctx);
5746         return 0;
5747 }
5748
5749 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5750 {
5751         struct diff_queue_struct *q = &diff_queued_diff;
5752         int i;
5753         int result = diff_get_patch_id(options, oid, diff_header_only);
5754
5755         for (i = 0; i < q->nr; i++)
5756                 diff_free_filepair(q->queue[i]);
5757
5758         free(q->queue);
5759         DIFF_QUEUE_CLEAR(q);
5760
5761         return result;
5762 }
5763
5764 static int is_summary_empty(const struct diff_queue_struct *q)
5765 {
5766         int i;
5767
5768         for (i = 0; i < q->nr; i++) {
5769                 const struct diff_filepair *p = q->queue[i];
5770
5771                 switch (p->status) {
5772                 case DIFF_STATUS_DELETED:
5773                 case DIFF_STATUS_ADDED:
5774                 case DIFF_STATUS_COPIED:
5775                 case DIFF_STATUS_RENAMED:
5776                         return 0;
5777                 default:
5778                         if (p->score)
5779                                 return 0;
5780                         if (p->one->mode && p->two->mode &&
5781                             p->one->mode != p->two->mode)
5782                                 return 0;
5783                         break;
5784                 }
5785         }
5786         return 1;
5787 }
5788
5789 static const char rename_limit_warning[] =
5790 N_("inexact rename detection was skipped due to too many files.");
5791
5792 static const char degrade_cc_to_c_warning[] =
5793 N_("only found copies from modified paths due to too many files.");
5794
5795 static const char rename_limit_advice[] =
5796 N_("you may want to set your %s variable to at least "
5797    "%d and retry the command.");
5798
5799 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5800 {
5801         fflush(stdout);
5802         if (degraded_cc)
5803                 warning(_(degrade_cc_to_c_warning));
5804         else if (needed)
5805                 warning(_(rename_limit_warning));
5806         else
5807                 return;
5808         if (0 < needed)
5809                 warning(_(rename_limit_advice), varname, needed);
5810 }
5811
5812 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5813 {
5814         int i;
5815         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5816         struct diff_queue_struct *q = &diff_queued_diff;
5817
5818         if (WSEH_NEW & WS_RULE_MASK)
5819                 BUG("WS rules bit mask overlaps with diff symbol flags");
5820
5821         if (o->color_moved)
5822                 o->emitted_symbols = &esm;
5823
5824         for (i = 0; i < q->nr; i++) {
5825                 struct diff_filepair *p = q->queue[i];
5826                 if (check_pair_status(p))
5827                         diff_flush_patch(p, o);
5828         }
5829
5830         if (o->emitted_symbols) {
5831                 if (o->color_moved) {
5832                         struct hashmap add_lines, del_lines;
5833
5834                         if (o->color_moved_ws_handling &
5835                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5836                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5837
5838                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5839                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5840
5841                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5842                         mark_color_as_moved(o, &add_lines, &del_lines);
5843                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5844                                 dim_moved_lines(o);
5845
5846                         hashmap_free(&add_lines, 0);
5847                         hashmap_free(&del_lines, 0);
5848                 }
5849
5850                 for (i = 0; i < esm.nr; i++)
5851                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5852
5853                 for (i = 0; i < esm.nr; i++)
5854                         free((void *)esm.buf[i].line);
5855         }
5856         esm.nr = 0;
5857 }
5858
5859 void diff_flush(struct diff_options *options)
5860 {
5861         struct diff_queue_struct *q = &diff_queued_diff;
5862         int i, output_format = options->output_format;
5863         int separator = 0;
5864         int dirstat_by_line = 0;
5865
5866         /*
5867          * Order: raw, stat, summary, patch
5868          * or:    name/name-status/checkdiff (other bits clear)
5869          */
5870         if (!q->nr)
5871                 goto free_queue;
5872
5873         if (output_format & (DIFF_FORMAT_RAW |
5874                              DIFF_FORMAT_NAME |
5875                              DIFF_FORMAT_NAME_STATUS |
5876                              DIFF_FORMAT_CHECKDIFF)) {
5877                 for (i = 0; i < q->nr; i++) {
5878                         struct diff_filepair *p = q->queue[i];
5879                         if (check_pair_status(p))
5880                                 flush_one_pair(p, options);
5881                 }
5882                 separator++;
5883         }
5884
5885         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5886                 dirstat_by_line = 1;
5887
5888         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5889             dirstat_by_line) {
5890                 struct diffstat_t diffstat;
5891
5892                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5893                 for (i = 0; i < q->nr; i++) {
5894                         struct diff_filepair *p = q->queue[i];
5895                         if (check_pair_status(p))
5896                                 diff_flush_stat(p, options, &diffstat);
5897                 }
5898                 if (output_format & DIFF_FORMAT_NUMSTAT)
5899                         show_numstat(&diffstat, options);
5900                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5901                         show_stats(&diffstat, options);
5902                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5903                         show_shortstats(&diffstat, options);
5904                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5905                         show_dirstat_by_line(&diffstat, options);
5906                 free_diffstat_info(&diffstat);
5907                 separator++;
5908         }
5909         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5910                 show_dirstat(options);
5911
5912         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5913                 for (i = 0; i < q->nr; i++) {
5914                         diff_summary(options, q->queue[i]);
5915                 }
5916                 separator++;
5917         }
5918
5919         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5920             options->flags.exit_with_status &&
5921             options->flags.diff_from_contents) {
5922                 /*
5923                  * run diff_flush_patch for the exit status. setting
5924                  * options->file to /dev/null should be safe, because we
5925                  * aren't supposed to produce any output anyway.
5926                  */
5927                 if (options->close_file)
5928                         fclose(options->file);
5929                 options->file = xfopen("/dev/null", "w");
5930                 options->close_file = 1;
5931                 options->color_moved = 0;
5932                 for (i = 0; i < q->nr; i++) {
5933                         struct diff_filepair *p = q->queue[i];
5934                         if (check_pair_status(p))
5935                                 diff_flush_patch(p, options);
5936                         if (options->found_changes)
5937                                 break;
5938                 }
5939         }
5940
5941         if (output_format & DIFF_FORMAT_PATCH) {
5942                 if (separator) {
5943                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5944                         if (options->stat_sep)
5945                                 /* attach patch instead of inline */
5946                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5947                                                  NULL, 0, 0);
5948                 }
5949
5950                 diff_flush_patch_all_file_pairs(options);
5951         }
5952
5953         if (output_format & DIFF_FORMAT_CALLBACK)
5954                 options->format_callback(q, options, options->format_callback_data);
5955
5956         for (i = 0; i < q->nr; i++)
5957                 diff_free_filepair(q->queue[i]);
5958 free_queue:
5959         free(q->queue);
5960         DIFF_QUEUE_CLEAR(q);
5961         if (options->close_file)
5962                 fclose(options->file);
5963
5964         /*
5965          * Report the content-level differences with HAS_CHANGES;
5966          * diff_addremove/diff_change does not set the bit when
5967          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5968          */
5969         if (options->flags.diff_from_contents) {
5970                 if (options->found_changes)
5971                         options->flags.has_changes = 1;
5972                 else
5973                         options->flags.has_changes = 0;
5974         }
5975 }
5976
5977 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5978 {
5979         return (((p->status == DIFF_STATUS_MODIFIED) &&
5980                  ((p->score &&
5981                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5982                   (!p->score &&
5983                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5984                 ((p->status != DIFF_STATUS_MODIFIED) &&
5985                  filter_bit_tst(p->status, options)));
5986 }
5987
5988 static void diffcore_apply_filter(struct diff_options *options)
5989 {
5990         int i;
5991         struct diff_queue_struct *q = &diff_queued_diff;
5992         struct diff_queue_struct outq;
5993
5994         DIFF_QUEUE_CLEAR(&outq);
5995
5996         if (!options->filter)
5997                 return;
5998
5999         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6000                 int found;
6001                 for (i = found = 0; !found && i < q->nr; i++) {
6002                         if (match_filter(options, q->queue[i]))
6003                                 found++;
6004                 }
6005                 if (found)
6006                         return;
6007
6008                 /* otherwise we will clear the whole queue
6009                  * by copying the empty outq at the end of this
6010                  * function, but first clear the current entries
6011                  * in the queue.
6012                  */
6013                 for (i = 0; i < q->nr; i++)
6014                         diff_free_filepair(q->queue[i]);
6015         }
6016         else {
6017                 /* Only the matching ones */
6018                 for (i = 0; i < q->nr; i++) {
6019                         struct diff_filepair *p = q->queue[i];
6020                         if (match_filter(options, p))
6021                                 diff_q(&outq, p);
6022                         else
6023                                 diff_free_filepair(p);
6024                 }
6025         }
6026         free(q->queue);
6027         *q = outq;
6028 }
6029
6030 /* Check whether two filespecs with the same mode and size are identical */
6031 static int diff_filespec_is_identical(struct diff_filespec *one,
6032                                       struct diff_filespec *two)
6033 {
6034         if (S_ISGITLINK(one->mode))
6035                 return 0;
6036         if (diff_populate_filespec(one, 0))
6037                 return 0;
6038         if (diff_populate_filespec(two, 0))
6039                 return 0;
6040         return !memcmp(one->data, two->data, one->size);
6041 }
6042
6043 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
6044 {
6045         if (p->done_skip_stat_unmatch)
6046                 return p->skip_stat_unmatch_result;
6047
6048         p->done_skip_stat_unmatch = 1;
6049         p->skip_stat_unmatch_result = 0;
6050         /*
6051          * 1. Entries that come from stat info dirtiness
6052          *    always have both sides (iow, not create/delete),
6053          *    one side of the object name is unknown, with
6054          *    the same mode and size.  Keep the ones that
6055          *    do not match these criteria.  They have real
6056          *    differences.
6057          *
6058          * 2. At this point, the file is known to be modified,
6059          *    with the same mode and size, and the object
6060          *    name of one side is unknown.  Need to inspect
6061          *    the identical contents.
6062          */
6063         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6064             !DIFF_FILE_VALID(p->two) ||
6065             (p->one->oid_valid && p->two->oid_valid) ||
6066             (p->one->mode != p->two->mode) ||
6067             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
6068             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
6069             (p->one->size != p->two->size) ||
6070             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
6071                 p->skip_stat_unmatch_result = 1;
6072         return p->skip_stat_unmatch_result;
6073 }
6074
6075 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6076 {
6077         int i;
6078         struct diff_queue_struct *q = &diff_queued_diff;
6079         struct diff_queue_struct outq;
6080         DIFF_QUEUE_CLEAR(&outq);
6081
6082         for (i = 0; i < q->nr; i++) {
6083                 struct diff_filepair *p = q->queue[i];
6084
6085                 if (diff_filespec_check_stat_unmatch(p))
6086                         diff_q(&outq, p);
6087                 else {
6088                         /*
6089                          * The caller can subtract 1 from skip_stat_unmatch
6090                          * to determine how many paths were dirty only
6091                          * due to stat info mismatch.
6092                          */
6093                         if (!diffopt->flags.no_index)
6094                                 diffopt->skip_stat_unmatch++;
6095                         diff_free_filepair(p);
6096                 }
6097         }
6098         free(q->queue);
6099         *q = outq;
6100 }
6101
6102 static int diffnamecmp(const void *a_, const void *b_)
6103 {
6104         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6105         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6106         const char *name_a, *name_b;
6107
6108         name_a = a->one ? a->one->path : a->two->path;
6109         name_b = b->one ? b->one->path : b->two->path;
6110         return strcmp(name_a, name_b);
6111 }
6112
6113 void diffcore_fix_diff_index(struct diff_options *options)
6114 {
6115         struct diff_queue_struct *q = &diff_queued_diff;
6116         QSORT(q->queue, q->nr, diffnamecmp);
6117 }
6118
6119 void diffcore_std(struct diff_options *options)
6120 {
6121         /* NOTE please keep the following in sync with diff_tree_combined() */
6122         if (options->skip_stat_unmatch)
6123                 diffcore_skip_stat_unmatch(options);
6124         if (!options->found_follow) {
6125                 /* See try_to_follow_renames() in tree-diff.c */
6126                 if (options->break_opt != -1)
6127                         diffcore_break(options->break_opt);
6128                 if (options->detect_rename)
6129                         diffcore_rename(options);
6130                 if (options->break_opt != -1)
6131                         diffcore_merge_broken();
6132         }
6133         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6134                 diffcore_pickaxe(options);
6135         if (options->orderfile)
6136                 diffcore_order(options->orderfile);
6137         if (!options->found_follow)
6138                 /* See try_to_follow_renames() in tree-diff.c */
6139                 diff_resolve_rename_copy();
6140         diffcore_apply_filter(options);
6141
6142         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6143                 options->flags.has_changes = 1;
6144         else
6145                 options->flags.has_changes = 0;
6146
6147         options->found_follow = 0;
6148 }
6149
6150 int diff_result_code(struct diff_options *opt, int status)
6151 {
6152         int result = 0;
6153
6154         diff_warn_rename_limit("diff.renameLimit",
6155                                opt->needed_rename_limit,
6156                                opt->degraded_cc_to_c);
6157         if (!opt->flags.exit_with_status &&
6158             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6159                 return status;
6160         if (opt->flags.exit_with_status &&
6161             opt->flags.has_changes)
6162                 result |= 01;
6163         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6164             opt->flags.check_failed)
6165                 result |= 02;
6166         return result;
6167 }
6168
6169 int diff_can_quit_early(struct diff_options *opt)
6170 {
6171         return (opt->flags.quick &&
6172                 !opt->filter &&
6173                 opt->flags.has_changes);
6174 }
6175
6176 /*
6177  * Shall changes to this submodule be ignored?
6178  *
6179  * Submodule changes can be configured to be ignored separately for each path,
6180  * but that configuration can be overridden from the command line.
6181  */
6182 static int is_submodule_ignored(const char *path, struct diff_options *options)
6183 {
6184         int ignored = 0;
6185         struct diff_flags orig_flags = options->flags;
6186         if (!options->flags.override_submodule_config)
6187                 set_diffopt_flags_from_submodule_config(options, path);
6188         if (options->flags.ignore_submodules)
6189                 ignored = 1;
6190         options->flags = orig_flags;
6191         return ignored;
6192 }
6193
6194 void diff_addremove(struct diff_options *options,
6195                     int addremove, unsigned mode,
6196                     const struct object_id *oid,
6197                     int oid_valid,
6198                     const char *concatpath, unsigned dirty_submodule)
6199 {
6200         struct diff_filespec *one, *two;
6201
6202         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6203                 return;
6204
6205         /* This may look odd, but it is a preparation for
6206          * feeding "there are unchanged files which should
6207          * not produce diffs, but when you are doing copy
6208          * detection you would need them, so here they are"
6209          * entries to the diff-core.  They will be prefixed
6210          * with something like '=' or '*' (I haven't decided
6211          * which but should not make any difference).
6212          * Feeding the same new and old to diff_change()
6213          * also has the same effect.
6214          * Before the final output happens, they are pruned after
6215          * merged into rename/copy pairs as appropriate.
6216          */
6217         if (options->flags.reverse_diff)
6218                 addremove = (addremove == '+' ? '-' :
6219                              addremove == '-' ? '+' : addremove);
6220
6221         if (options->prefix &&
6222             strncmp(concatpath, options->prefix, options->prefix_length))
6223                 return;
6224
6225         one = alloc_filespec(concatpath);
6226         two = alloc_filespec(concatpath);
6227
6228         if (addremove != '+')
6229                 fill_filespec(one, oid, oid_valid, mode);
6230         if (addremove != '-') {
6231                 fill_filespec(two, oid, oid_valid, mode);
6232                 two->dirty_submodule = dirty_submodule;
6233         }
6234
6235         diff_queue(&diff_queued_diff, one, two);
6236         if (!options->flags.diff_from_contents)
6237                 options->flags.has_changes = 1;
6238 }
6239
6240 void diff_change(struct diff_options *options,
6241                  unsigned old_mode, unsigned new_mode,
6242                  const struct object_id *old_oid,
6243                  const struct object_id *new_oid,
6244                  int old_oid_valid, int new_oid_valid,
6245                  const char *concatpath,
6246                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6247 {
6248         struct diff_filespec *one, *two;
6249         struct diff_filepair *p;
6250
6251         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6252             is_submodule_ignored(concatpath, options))
6253                 return;
6254
6255         if (options->flags.reverse_diff) {
6256                 SWAP(old_mode, new_mode);
6257                 SWAP(old_oid, new_oid);
6258                 SWAP(old_oid_valid, new_oid_valid);
6259                 SWAP(old_dirty_submodule, new_dirty_submodule);
6260         }
6261
6262         if (options->prefix &&
6263             strncmp(concatpath, options->prefix, options->prefix_length))
6264                 return;
6265
6266         one = alloc_filespec(concatpath);
6267         two = alloc_filespec(concatpath);
6268         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6269         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6270         one->dirty_submodule = old_dirty_submodule;
6271         two->dirty_submodule = new_dirty_submodule;
6272         p = diff_queue(&diff_queued_diff, one, two);
6273
6274         if (options->flags.diff_from_contents)
6275                 return;
6276
6277         if (options->flags.quick && options->skip_stat_unmatch &&
6278             !diff_filespec_check_stat_unmatch(p))
6279                 return;
6280
6281         options->flags.has_changes = 1;
6282 }
6283
6284 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6285 {
6286         struct diff_filepair *pair;
6287         struct diff_filespec *one, *two;
6288
6289         if (options->prefix &&
6290             strncmp(path, options->prefix, options->prefix_length))
6291                 return NULL;
6292
6293         one = alloc_filespec(path);
6294         two = alloc_filespec(path);
6295         pair = diff_queue(&diff_queued_diff, one, two);
6296         pair->is_unmerged = 1;
6297         return pair;
6298 }
6299
6300 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6301                 size_t *outsize)
6302 {
6303         struct diff_tempfile *temp;
6304         const char *argv[3];
6305         const char **arg = argv;
6306         struct child_process child = CHILD_PROCESS_INIT;
6307         struct strbuf buf = STRBUF_INIT;
6308         int err = 0;
6309
6310         temp = prepare_temp_file(spec->path, spec);
6311         *arg++ = pgm;
6312         *arg++ = temp->name;
6313         *arg = NULL;
6314
6315         child.use_shell = 1;
6316         child.argv = argv;
6317         child.out = -1;
6318         if (start_command(&child)) {
6319                 remove_tempfile();
6320                 return NULL;
6321         }
6322
6323         if (strbuf_read(&buf, child.out, 0) < 0)
6324                 err = error("error reading from textconv command '%s'", pgm);
6325         close(child.out);
6326
6327         if (finish_command(&child) || err) {
6328                 strbuf_release(&buf);
6329                 remove_tempfile();
6330                 return NULL;
6331         }
6332         remove_tempfile();
6333
6334         return strbuf_detach(&buf, outsize);
6335 }
6336
6337 size_t fill_textconv(struct userdiff_driver *driver,
6338                      struct diff_filespec *df,
6339                      char **outbuf)
6340 {
6341         size_t size;
6342
6343         if (!driver) {
6344                 if (!DIFF_FILE_VALID(df)) {
6345                         *outbuf = "";
6346                         return 0;
6347                 }
6348                 if (diff_populate_filespec(df, 0))
6349                         die("unable to read files to diff");
6350                 *outbuf = df->data;
6351                 return df->size;
6352         }
6353
6354         if (!driver->textconv)
6355                 BUG("fill_textconv called with non-textconv driver");
6356
6357         if (driver->textconv_cache && df->oid_valid) {
6358                 *outbuf = notes_cache_get(driver->textconv_cache,
6359                                           &df->oid,
6360                                           &size);
6361                 if (*outbuf)
6362                         return size;
6363         }
6364
6365         *outbuf = run_textconv(driver->textconv, df, &size);
6366         if (!*outbuf)
6367                 die("unable to read files to diff");
6368
6369         if (driver->textconv_cache && df->oid_valid) {
6370                 /* ignore errors, as we might be in a readonly repository */
6371                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6372                                 size);
6373                 /*
6374                  * we could save up changes and flush them all at the end,
6375                  * but we would need an extra call after all diffing is done.
6376                  * Since generating a cache entry is the slow path anyway,
6377                  * this extra overhead probably isn't a big deal.
6378                  */
6379                 notes_cache_write(driver->textconv_cache);
6380         }
6381
6382         return size;
6383 }
6384
6385 int textconv_object(const char *path,
6386                     unsigned mode,
6387                     const struct object_id *oid,
6388                     int oid_valid,
6389                     char **buf,
6390                     unsigned long *buf_size)
6391 {
6392         struct diff_filespec *df;
6393         struct userdiff_driver *textconv;
6394
6395         df = alloc_filespec(path);
6396         fill_filespec(df, oid, oid_valid, mode);
6397         textconv = get_textconv(df);
6398         if (!textconv) {
6399                 free_filespec(df);
6400                 return 0;
6401         }
6402
6403         *buf_size = fill_textconv(textconv, df, buf);
6404         free_filespec(df);
6405         return 1;
6406 }
6407
6408 void setup_diff_pager(struct diff_options *opt)
6409 {
6410         /*
6411          * If the user asked for our exit code, then either they want --quiet
6412          * or --exit-code. We should definitely not bother with a pager in the
6413          * former case, as we will generate no output. Since we still properly
6414          * report our exit code even when a pager is run, we _could_ run a
6415          * pager with --exit-code. But since we have not done so historically,
6416          * and because it is easy to find people oneline advising "git diff
6417          * --exit-code" in hooks and other scripts, we do not do so.
6418          */
6419         if (!opt->flags.exit_with_status &&
6420             check_pager_config("diff") != 0)
6421                 setup_pager();
6422 }