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, "no"))
 308                         ret = 0;
 309                 else if (!strcmp(sb.buf, "ignore-space-change"))
 310                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 311                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 312                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 313                 else if (!strcmp(sb.buf, "ignore-all-space"))
 314                         ret |= XDF_IGNORE_WHITESPACE;
 315                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 316                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 317                 else
 318                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 319
 320                 strbuf_release(&sb);
 321         }
 322
 323         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 324             (ret & XDF_WHITESPACE_FLAGS))
 325                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 326
 327         string_list_clear(&l, 0);
 328
 329         return ret;
 330 }
 331
 332 int git_diff_ui_config(const char *var, const char *value, void *cb)
 333 {
 334         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 335                 diff_use_color_default = git_config_colorbool(var, value);
 336                 return 0;
 337         }
 338         if (!strcmp(var, "diff.colormoved")) {
 339                 int cm = parse_color_moved(value);
 340                 if (cm < 0)
 341                         return -1;
 342                 diff_color_moved_default = cm;
 343                 return 0;
 344         }
 345         if (!strcmp(var, "diff.colormovedws")) {
 346                 int cm = parse_color_moved_ws(value);
 347                 if (cm < 0)
 348                         return -1;
 349                 diff_color_moved_ws_default = cm;
 350                 return 0;
 351         }
 352         if (!strcmp(var, "diff.context")) {
 353                 diff_context_default = git_config_int(var, value);
 354                 if (diff_context_default < 0)
 355                         return -1;
 356                 return 0;
 357         }
 358         if (!strcmp(var, "diff.interhunkcontext")) {
 359                 diff_interhunk_context_default = git_config_int(var, value);
 360                 if (diff_interhunk_context_default < 0)
 361                         return -1;
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.renames")) {
 365                 diff_detect_rename_default = git_config_rename(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.autorefreshindex")) {
 369                 diff_auto_refresh_index = git_config_bool(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.mnemonicprefix")) {
 373                 diff_mnemonic_prefix = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.noprefix")) {
 377                 diff_no_prefix = git_config_bool(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.statgraphwidth")) {
 381                 diff_stat_graph_width = git_config_int(var, value);
 382                 return 0;
 383         }
 384         if (!strcmp(var, "diff.external"))
 385                 return git_config_string(&external_diff_cmd_cfg, var, value);
 386         if (!strcmp(var, "diff.wordregex"))
 387                 return git_config_string(&diff_word_regex_cfg, var, value);
 388         if (!strcmp(var, "diff.orderfile"))
 389                 return git_config_pathname(&diff_order_file_cfg, var, value);
 390
 391         if (!strcmp(var, "diff.ignoresubmodules"))
 392                 handle_ignore_submodules_arg(&default_diff_options, value);
 393
 394         if (!strcmp(var, "diff.submodule")) {
 395                 if (parse_submodule_params(&default_diff_options, value))
 396                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 397                                 value);
 398                 return 0;
 399         }
 400
 401         if (!strcmp(var, "diff.algorithm")) {
 402                 diff_algorithm = parse_algorithm_value(value);
 403                 if (diff_algorithm < 0)
 404                         return -1;
 405                 return 0;
 406         }
 407
 408         if (!strcmp(var, "diff.wserrorhighlight")) {
 409                 int val = parse_ws_error_highlight(value);
 410                 if (val < 0)
 411                         return -1;
 412                 ws_error_highlight_default = val;
 413                 return 0;
 414         }
 415
 416         if (git_color_config(var, value, cb) < 0)
 417                 return -1;
 418
 419         return git_diff_basic_config(var, value, cb);
 420 }
 421
 422 int git_diff_basic_config(const char *var, const char *value, void *cb)
 423 {
 424         const char *name;
 425
 426         if (!strcmp(var, "diff.renamelimit")) {
 427                 diff_rename_limit_default = git_config_int(var, value);
 428                 return 0;
 429         }
 430
 431         if (userdiff_config(var, value) < 0)
 432                 return -1;
 433
 434         if (skip_prefix(var, "diff.color.", &name) ||
 435             skip_prefix(var, "color.diff.", &name)) {
 436                 int slot = parse_diff_color_slot(name);
 437                 if (slot < 0)
 438                         return 0;
 439                 if (!value)
 440                         return config_error_nonbool(var);
 441                 return color_parse(value, diff_colors[slot]);
 442         }
 443
 444         /* like GNU diff's --suppress-blank-empty option  */
 445         if (!strcmp(var, "diff.suppressblankempty") ||
 446                         /* for backwards compatibility */
 447                         !strcmp(var, "diff.suppress-blank-empty")) {
 448                 diff_suppress_blank_empty = git_config_bool(var, value);
 449                 return 0;
 450         }
 451
 452         if (!strcmp(var, "diff.dirstat")) {
 453                 struct strbuf errmsg = STRBUF_INIT;
 454                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 455                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 456                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 457                                 errmsg.buf);
 458                 strbuf_release(&errmsg);
 459                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 460                 return 0;
 461         }
 462
 463         if (git_diff_heuristic_config(var, value, cb) < 0)
 464                 return -1;
 465
 466         return git_default_config(var, value, cb);
 467 }
 468
 469 static char *quote_two(const char *one, const char *two)
 470 {
 471         int need_one = quote_c_style(one, NULL, NULL, 1);
 472         int need_two = quote_c_style(two, NULL, NULL, 1);
 473         struct strbuf res = STRBUF_INIT;
 474
 475         if (need_one + need_two) {
 476                 strbuf_addch(&res, '"');
 477                 quote_c_style(one, &res, NULL, 1);
 478                 quote_c_style(two, &res, NULL, 1);
 479                 strbuf_addch(&res, '"');
 480         } else {
 481                 strbuf_addstr(&res, one);
 482                 strbuf_addstr(&res, two);
 483         }
 484         return strbuf_detach(&res, NULL);
 485 }
 486
 487 static const char *external_diff(void)
 488 {
 489         static const char *external_diff_cmd = NULL;
 490         static int done_preparing = 0;
 491
 492         if (done_preparing)
 493                 return external_diff_cmd;
 494         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 495         if (!external_diff_cmd)
 496                 external_diff_cmd = external_diff_cmd_cfg;
 497         done_preparing = 1;
 498         return external_diff_cmd;
 499 }
 500
 501 /*
 502  * Keep track of files used for diffing. Sometimes such an entry
 503  * refers to a temporary file, sometimes to an existing file, and
 504  * sometimes to "/dev/null".
 505  */
 506 static struct diff_tempfile {
 507         /*
 508          * filename external diff should read from, or NULL if this
 509          * entry is currently not in use:
 510          */
 511         const char *name;
 512
 513         char hex[GIT_MAX_HEXSZ + 1];
 514         char mode[10];
 515
 516         /*
 517          * If this diff_tempfile instance refers to a temporary file,
 518          * this tempfile object is used to manage its lifetime.
 519          */
 520         struct tempfile *tempfile;
 521 } diff_temp[2];
 522
 523 struct emit_callback {
 524         int color_diff;
 525         unsigned ws_rule;
 526         int blank_at_eof_in_preimage;
 527         int blank_at_eof_in_postimage;
 528         int lno_in_preimage;
 529         int lno_in_postimage;
 530         const char **label_path;
 531         struct diff_words_data *diff_words;
 532         struct diff_options *opt;
 533         struct strbuf *header;
 534 };
 535
 536 static int count_lines(const char *data, int size)
 537 {
 538         int count, ch, completely_empty = 1, nl_just_seen = 0;
 539         count = 0;
 540         while (0 < size--) {
 541                 ch = *data++;
 542                 if (ch == '\n') {
 543                         count++;
 544                         nl_just_seen = 1;
 545                         completely_empty = 0;
 546                 }
 547                 else {
 548                         nl_just_seen = 0;
 549                         completely_empty = 0;
 550                 }
 551         }
 552         if (completely_empty)
 553                 return 0;
 554         if (!nl_just_seen)
 555                 count++; /* no trailing newline */
 556         return count;
 557 }
 558
 559 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 560                        struct diff_filespec *one)
 561 {
 562         if (!DIFF_FILE_VALID(one)) {
 563                 mf->ptr = (char *)""; /* does not matter */
 564                 mf->size = 0;
 565                 return 0;
 566         }
 567         else if (diff_populate_filespec(r, one, 0))
 568                 return -1;
 569
 570         mf->ptr = one->data;
 571         mf->size = one->size;
 572         return 0;
 573 }
 574
 575 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 576 static unsigned long diff_filespec_size(struct repository *r,
 577                                         struct diff_filespec *one)
 578 {
 579         if (!DIFF_FILE_VALID(one))
 580                 return 0;
 581         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 582         return one->size;
 583 }
 584
 585 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 586 {
 587         char *ptr = mf->ptr;
 588         long size = mf->size;
 589         int cnt = 0;
 590
 591         if (!size)
 592                 return cnt;
 593         ptr += size - 1; /* pointing at the very end */
 594         if (*ptr != '\n')
 595                 ; /* incomplete line */
 596         else
 597                 ptr--; /* skip the last LF */
 598         while (mf->ptr < ptr) {
 599                 char *prev_eol;
 600                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 601                         if (*prev_eol == '\n')
 602                                 break;
 603                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 604                         break;
 605                 cnt++;
 606                 ptr = prev_eol - 1;
 607         }
 608         return cnt;
 609 }
 610
 611 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 612                                struct emit_callback *ecbdata)
 613 {
 614         int l1, l2, at;
 615         unsigned ws_rule = ecbdata->ws_rule;
 616         l1 = count_trailing_blank(mf1, ws_rule);
 617         l2 = count_trailing_blank(mf2, ws_rule);
 618         if (l2 <= l1) {
 619                 ecbdata->blank_at_eof_in_preimage = 0;
 620                 ecbdata->blank_at_eof_in_postimage = 0;
 621                 return;
 622         }
 623         at = count_lines(mf1->ptr, mf1->size);
 624         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 625
 626         at = count_lines(mf2->ptr, mf2->size);
 627         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 628 }
 629
 630 static void emit_line_0(struct diff_options *o,
 631                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 632                         int first, const char *line, int len)
 633 {
 634         int has_trailing_newline, has_trailing_carriage_return;
 635         int needs_reset = 0; /* at the end of the line */
 636         FILE *file = o->file;
 637
 638         fputs(diff_line_prefix(o), file);
 639
 640         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 641         if (has_trailing_newline)
 642                 len--;
 643
 644         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 645         if (has_trailing_carriage_return)
 646                 len--;
 647
 648         if (!len && !first)
 649                 goto end_of_line;
 650
 651         if (reverse && want_color(o->use_color)) {
 652                 fputs(GIT_COLOR_REVERSE, file);
 653                 needs_reset = 1;
 654         }
 655
 656         if (set_sign) {
 657                 fputs(set_sign, file);
 658                 needs_reset = 1;
 659         }
 660
 661         if (first)
 662                 fputc(first, file);
 663
 664         if (!len)
 665                 goto end_of_line;
 666
 667         if (set) {
 668                 if (set_sign && set != set_sign)
 669                         fputs(reset, file);
 670                 fputs(set, file);
 671                 needs_reset = 1;
 672         }
 673         fwrite(line, len, 1, file);
 674         needs_reset = 1; /* 'line' may contain color codes. */
 675
 676 end_of_line:
 677         if (needs_reset)
 678                 fputs(reset, file);
 679         if (has_trailing_carriage_return)
 680                 fputc('\r', file);
 681         if (has_trailing_newline)
 682                 fputc('\n', file);
 683 }
 684
 685 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 686                       const char *line, int len)
 687 {
 688         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 689 }
 690
 691 enum diff_symbol {
 692         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 693         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 694         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 695         DIFF_SYMBOL_BINARY_DIFF_BODY,
 696         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 697         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 698         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 699         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 700         DIFF_SYMBOL_STATS_LINE,
 701         DIFF_SYMBOL_WORD_DIFF,
 702         DIFF_SYMBOL_STAT_SEP,
 703         DIFF_SYMBOL_SUMMARY,
 704         DIFF_SYMBOL_SUBMODULE_ADD,
 705         DIFF_SYMBOL_SUBMODULE_DEL,
 706         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 707         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 708         DIFF_SYMBOL_SUBMODULE_HEADER,
 709         DIFF_SYMBOL_SUBMODULE_ERROR,
 710         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 711         DIFF_SYMBOL_REWRITE_DIFF,
 712         DIFF_SYMBOL_BINARY_FILES,
 713         DIFF_SYMBOL_HEADER,
 714         DIFF_SYMBOL_FILEPAIR_PLUS,
 715         DIFF_SYMBOL_FILEPAIR_MINUS,
 716         DIFF_SYMBOL_WORDS_PORCELAIN,
 717         DIFF_SYMBOL_WORDS,
 718         DIFF_SYMBOL_CONTEXT,
 719         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 720         DIFF_SYMBOL_PLUS,
 721         DIFF_SYMBOL_MINUS,
 722         DIFF_SYMBOL_NO_LF_EOF,
 723         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 724         DIFF_SYMBOL_CONTEXT_MARKER,
 725         DIFF_SYMBOL_SEPARATOR
 726 };
 727 /*
 728  * Flags for content lines:
 729  * 0..12 are whitespace rules
 730  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 731  * 16 is marking if the line is blank at EOF
 732  */
 733 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 734 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 735 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 736 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 737 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 738
 739 /*
 740  * This struct is used when we need to buffer the output of the diff output.
 741  *
 742  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 743  * into the pre/post image file. This pointer could be a union with the
 744  * line pointer. By storing an offset into the file instead of the literal line,
 745  * we can decrease the memory footprint for the buffered output. At first we
 746  * may want to only have indirection for the content lines, but we could also
 747  * enhance the state for emitting prefabricated lines, e.g. the similarity
 748  * score line or hunk/file headers would only need to store a number or path
 749  * and then the output can be constructed later on depending on state.
 750  */
 751 struct emitted_diff_symbol {
 752         const char *line;
 753         int len;
 754         int flags;
 755         enum diff_symbol s;
 756 };
 757 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 758
 759 struct emitted_diff_symbols {
 760         struct emitted_diff_symbol *buf;
 761         int nr, alloc;
 762 };
 763 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 764
 765 static void append_emitted_diff_symbol(struct diff_options *o,
 766                                        struct emitted_diff_symbol *e)
 767 {
 768         struct emitted_diff_symbol *f;
 769
 770         ALLOC_GROW(o->emitted_symbols->buf,
 771                    o->emitted_symbols->nr + 1,
 772                    o->emitted_symbols->alloc);
 773         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 774
 775         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 776         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 777 }
 778
 779 struct moved_entry {
 780         struct hashmap_entry ent;
 781         const struct emitted_diff_symbol *es;
 782         struct moved_entry *next_line;
 783 };
 784
 785 /**
 786  * The struct ws_delta holds white space differences between moved lines, i.e.
 787  * between '+' and '-' lines that have been detected to be a move.
 788  * The string contains the difference in leading white spaces, before the
 789  * rest of the line is compared using the white space config for move
 790  * coloring. The current_longer indicates if the first string in the
 791  * comparision is longer than the second.
 792  */
 793 struct ws_delta {
 794         char *string;
 795         unsigned int current_longer : 1;
 796 };
 797 #define WS_DELTA_INIT { NULL, 0 }
 798
 799 struct moved_block {
 800         struct moved_entry *match;
 801         struct ws_delta wsd;
 802 };
 803
 804 static void moved_block_clear(struct moved_block *b)
 805 {
 806         FREE_AND_NULL(b->wsd.string);
 807         b->match = NULL;
 808 }
 809
 810 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 811                              const struct emitted_diff_symbol *b,
 812                              struct ws_delta *out)
 813 {
 814         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 815         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 816         int d = longer->len - shorter->len;
 817
 818         if (strncmp(longer->line + d, shorter->line, shorter->len))
 819                 return 0;
 820
 821         out->string = xmemdupz(longer->line, d);
 822         out->current_longer = (a == longer);
 823
 824         return 1;
 825 }
 826
 827 static int cmp_in_block_with_wsd(const struct diff_options *o,
 828                                  const struct moved_entry *cur,
 829                                  const struct moved_entry *match,
 830                                  struct moved_block *pmb,
 831                                  int n)
 832 {
 833         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 834         int al = cur->es->len, bl = match->es->len, cl = l->len;
 835         const char *a = cur->es->line,
 836                    *b = match->es->line,
 837                    *c = l->line;
 838         const char *orig_a = a;
 839         int wslen;
 840
 841         /*
 842          * We need to check if 'cur' is equal to 'match'.  As those
 843          * are from the same (+/-) side, we do not need to adjust for
 844          * indent changes. However these were found using fuzzy
 845          * matching so we do have to check if they are equal. Here we
 846          * just check the lengths. We delay calling memcmp() to check
 847          * the contents until later as if the length comparison for a
 848          * and c fails we can avoid the call all together.
 849          */
 850         if (al != bl)
 851                 return 1;
 852
 853         if (!pmb->wsd.string)
 854                 /*
 855                  * The white space delta is not active? This can happen
 856                  * when we exit early in this function.
 857                  */
 858                 return 1;
 859
 860         /*
 861          * The indent changes of the block are known and stored in
 862          * pmb->wsd; however we need to check if the indent changes of the
 863          * current line are still the same as before.
 864          *
 865          * To do so we need to compare 'l' to 'cur', adjusting the
 866          * one of them for the white spaces, depending which was longer.
 867          */
 868
 869         wslen = strlen(pmb->wsd.string);
 870         if (pmb->wsd.current_longer) {
 871                 c += wslen;
 872                 cl -= wslen;
 873         } else {
 874                 a += wslen;
 875                 al -= wslen;
 876         }
 877
 878         if (al != cl || memcmp(orig_a, b, bl) || memcmp(a, c, al))
 879                 return 1;
 880
 881         return 0;
 882 }
 883
 884 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 885                            const void *entry,
 886                            const void *entry_or_key,
 887                            const void *keydata)
 888 {
 889         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 890         const struct moved_entry *a = entry;
 891         const struct moved_entry *b = entry_or_key;
 892         unsigned flags = diffopt->color_moved_ws_handling
 893                          & XDF_WHITESPACE_FLAGS;
 894
 895         if (diffopt->color_moved_ws_handling &
 896             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 897                 /*
 898                  * As there is not specific white space config given,
 899                  * we'd need to check for a new block, so ignore all
 900                  * white space. The setup of the white space
 901                  * configuration for the next block is done else where
 902                  */
 903                 flags |= XDF_IGNORE_WHITESPACE;
 904
 905         return !xdiff_compare_lines(a->es->line, a->es->len,
 906                                     b->es->line, b->es->len,
 907                                     flags);
 908 }
 909
 910 static struct moved_entry *prepare_entry(struct diff_options *o,
 911                                          int line_no)
 912 {
 913         struct moved_entry *ret = xmalloc(sizeof(*ret));
 914         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 915         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 916
 917         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 918         ret->es = l;
 919         ret->next_line = NULL;
 920
 921         return ret;
 922 }
 923
 924 static void add_lines_to_move_detection(struct diff_options *o,
 925                                         struct hashmap *add_lines,
 926                                         struct hashmap *del_lines)
 927 {
 928         struct moved_entry *prev_line = NULL;
 929
 930         int n;
 931         for (n = 0; n < o->emitted_symbols->nr; n++) {
 932                 struct hashmap *hm;
 933                 struct moved_entry *key;
 934
 935                 switch (o->emitted_symbols->buf[n].s) {
 936                 case DIFF_SYMBOL_PLUS:
 937                         hm = add_lines;
 938                         break;
 939                 case DIFF_SYMBOL_MINUS:
 940                         hm = del_lines;
 941                         break;
 942                 default:
 943                         prev_line = NULL;
 944                         continue;
 945                 }
 946
 947                 key = prepare_entry(o, n);
 948                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 949                         prev_line->next_line = key;
 950
 951                 hashmap_add(hm, key);
 952                 prev_line = key;
 953         }
 954 }
 955
 956 static void pmb_advance_or_null(struct diff_options *o,
 957                                 struct moved_entry *match,
 958                                 struct hashmap *hm,
 959                                 struct moved_block *pmb,
 960                                 int pmb_nr)
 961 {
 962         int i;
 963         for (i = 0; i < pmb_nr; i++) {
 964                 struct moved_entry *prev = pmb[i].match;
 965                 struct moved_entry *cur = (prev && prev->next_line) ?
 966                                 prev->next_line : NULL;
 967                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 968                         pmb[i].match = cur;
 969                 } else {
 970                         pmb[i].match = NULL;
 971                 }
 972         }
 973 }
 974
 975 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 976                                             struct moved_entry *match,
 977                                             struct hashmap *hm,
 978                                             struct moved_block *pmb,
 979                                             int pmb_nr, int n)
 980 {
 981         int i;
 982         char *got_match = xcalloc(1, pmb_nr);
 983
 984         for (; match; match = hashmap_get_next(hm, match)) {
 985                 for (i = 0; i < pmb_nr; i++) {
 986                         struct moved_entry *prev = pmb[i].match;
 987                         struct moved_entry *cur = (prev && prev->next_line) ?
 988                                         prev->next_line : NULL;
 989                         if (!cur)
 990                                 continue;
 991                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
 992                                 got_match[i] |= 1;
 993                 }
 994         }
 995
 996         for (i = 0; i < pmb_nr; i++) {
 997                 if (got_match[i]) {
 998                         /* Advance to the next line */
 999                         pmb[i].match = pmb[i].match->next_line;
1000                 } else {
1001                         moved_block_clear(&pmb[i]);
1002                 }
1003         }
1004
1005         free(got_match);
1006 }
1007
1008 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1009                                          int pmb_nr)
1010 {
1011         int lp, rp;
1012
1013         /* Shrink the set of potential block to the remaining running */
1014         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1015                 while (lp < pmb_nr && pmb[lp].match)
1016                         lp++;
1017                 /* lp points at the first NULL now */
1018
1019                 while (rp > -1 && !pmb[rp].match)
1020                         rp--;
1021                 /* rp points at the last non-NULL */
1022
1023                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1024                         pmb[lp] = pmb[rp];
1025                         pmb[rp].match = NULL;
1026                         pmb[rp].wsd.string = NULL;
1027                         rp--;
1028                         lp++;
1029                 }
1030         }
1031
1032         /* Remember the number of running sets */
1033         return rp + 1;
1034 }
1035
1036 /*
1037  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1038  *
1039  * Otherwise, if the last block has fewer alphanumeric characters than
1040  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1041  * that block.
1042  *
1043  * The last block consists of the (n - block_length)'th line up to but not
1044  * including the nth line.
1045  *
1046  * Returns 0 if the last block is empty or is unset by this function, non zero
1047  * otherwise.
1048  *
1049  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1050  * Think of a way to unify them.
1051  */
1052 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1053 {
1054         int i, alnum_count = 0;
1055         if (o->color_moved == COLOR_MOVED_PLAIN)
1056                 return block_length;
1057         for (i = 1; i < block_length + 1; i++) {
1058                 const char *c = o->emitted_symbols->buf[n - i].line;
1059                 for (; *c; c++) {
1060                         if (!isalnum(*c))
1061                                 continue;
1062                         alnum_count++;
1063                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1064                                 return 1;
1065                 }
1066         }
1067         for (i = 1; i < block_length + 1; i++)
1068                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1069         return 0;
1070 }
1071
1072 /* Find blocks of moved code, delegate actual coloring decision to helper */
1073 static void mark_color_as_moved(struct diff_options *o,
1074                                 struct hashmap *add_lines,
1075                                 struct hashmap *del_lines)
1076 {
1077         struct moved_block *pmb = NULL; /* potentially moved blocks */
1078         int pmb_nr = 0, pmb_alloc = 0;
1079         int n, flipped_block = 0, block_length = 0;
1080
1081
1082         for (n = 0; n < o->emitted_symbols->nr; n++) {
1083                 struct hashmap *hm = NULL;
1084                 struct moved_entry *key;
1085                 struct moved_entry *match = NULL;
1086                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1087                 enum diff_symbol last_symbol = 0;
1088
1089                 switch (l->s) {
1090                 case DIFF_SYMBOL_PLUS:
1091                         hm = del_lines;
1092                         key = prepare_entry(o, n);
1093                         match = hashmap_get(hm, key, NULL);
1094                         free(key);
1095                         break;
1096                 case DIFF_SYMBOL_MINUS:
1097                         hm = add_lines;
1098                         key = prepare_entry(o, n);
1099                         match = hashmap_get(hm, key, NULL);
1100                         free(key);
1101                         break;
1102                 default:
1103                         flipped_block = 0;
1104                 }
1105
1106                 if (!match) {
1107                         int i;
1108
1109                         adjust_last_block(o, n, block_length);
1110                         for(i = 0; i < pmb_nr; i++)
1111                                 moved_block_clear(&pmb[i]);
1112                         pmb_nr = 0;
1113                         block_length = 0;
1114                         flipped_block = 0;
1115                         last_symbol = l->s;
1116                         continue;
1117                 }
1118
1119                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1120                         last_symbol = l->s;
1121                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1122                         continue;
1123                 }
1124
1125                 if (o->color_moved_ws_handling &
1126                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1127                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1128                 else
1129                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1130
1131                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1132
1133                 if (pmb_nr == 0) {
1134                         /*
1135                          * The current line is the start of a new block.
1136                          * Setup the set of potential blocks.
1137                          */
1138                         for (; match; match = hashmap_get_next(hm, match)) {
1139                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1140                                 if (o->color_moved_ws_handling &
1141                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1142                                         if (compute_ws_delta(l, match->es,
1143                                                              &pmb[pmb_nr].wsd))
1144                                                 pmb[pmb_nr++].match = match;
1145                                 } else {
1146                                         pmb[pmb_nr].wsd.string = NULL;
1147                                         pmb[pmb_nr++].match = match;
1148                                 }
1149                         }
1150
1151                         if (adjust_last_block(o, n, block_length) &&
1152                             pmb_nr && last_symbol != l->s)
1153                                 flipped_block = (flipped_block + 1) % 2;
1154                         else
1155                                 flipped_block = 0;
1156
1157                         block_length = 0;
1158                 }
1159
1160                 if (pmb_nr) {
1161                         block_length++;
1162                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1163                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1164                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1165                 }
1166                 last_symbol = l->s;
1167         }
1168         adjust_last_block(o, n, block_length);
1169
1170         for(n = 0; n < pmb_nr; n++)
1171                 moved_block_clear(&pmb[n]);
1172         free(pmb);
1173 }
1174
1175 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1176   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1177 static void dim_moved_lines(struct diff_options *o)
1178 {
1179         int n;
1180         for (n = 0; n < o->emitted_symbols->nr; n++) {
1181                 struct emitted_diff_symbol *prev = (n != 0) ?
1182                                 &o->emitted_symbols->buf[n - 1] : NULL;
1183                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1184                 struct emitted_diff_symbol *next =
1185                                 (n < o->emitted_symbols->nr - 1) ?
1186                                 &o->emitted_symbols->buf[n + 1] : NULL;
1187
1188                 /* Not a plus or minus line? */
1189                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1190                         continue;
1191
1192                 /* Not a moved line? */
1193                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1194                         continue;
1195
1196                 /*
1197                  * If prev or next are not a plus or minus line,
1198                  * pretend they don't exist
1199                  */
1200                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1201                             prev->s != DIFF_SYMBOL_MINUS)
1202                         prev = NULL;
1203                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1204                             next->s != DIFF_SYMBOL_MINUS)
1205                         next = NULL;
1206
1207                 /* Inside a block? */
1208                 if ((prev &&
1209                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1210                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1211                     (next &&
1212                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1213                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1214                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1215                         continue;
1216                 }
1217
1218                 /* Check if we are at an interesting bound: */
1219                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1220                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1221                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1222                         continue;
1223                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1224                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1225                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1226                         continue;
1227
1228                 /*
1229                  * The boundary to prev and next are not interesting,
1230                  * so this line is not interesting as a whole
1231                  */
1232                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1233         }
1234 }
1235
1236 static void emit_line_ws_markup(struct diff_options *o,
1237                                 const char *set_sign, const char *set,
1238                                 const char *reset,
1239                                 int sign_index, const char *line, int len,
1240                                 unsigned ws_rule, int blank_at_eof)
1241 {
1242         const char *ws = NULL;
1243         int sign = o->output_indicators[sign_index];
1244
1245         if (o->ws_error_highlight & ws_rule) {
1246                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1247                 if (!*ws)
1248                         ws = NULL;
1249         }
1250
1251         if (!ws && !set_sign)
1252                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1253         else if (!ws) {
1254                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1255         } else if (blank_at_eof)
1256                 /* Blank line at EOF - paint '+' as well */
1257                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1258         else {
1259                 /* Emit just the prefix, then the rest. */
1260                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1261                             sign, "", 0);
1262                 ws_check_emit(line, len, ws_rule,
1263                               o->file, set, reset, ws);
1264         }
1265 }
1266
1267 static void emit_diff_symbol_from_struct(struct diff_options *o,
1268                                          struct emitted_diff_symbol *eds)
1269 {
1270         static const char *nneof = " No newline at end of file\n";
1271         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1272         struct strbuf sb = STRBUF_INIT;
1273
1274         enum diff_symbol s = eds->s;
1275         const char *line = eds->line;
1276         int len = eds->len;
1277         unsigned flags = eds->flags;
1278
1279         switch (s) {
1280         case DIFF_SYMBOL_NO_LF_EOF:
1281                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1282                 reset = diff_get_color_opt(o, DIFF_RESET);
1283                 putc('\n', o->file);
1284                 emit_line_0(o, context, NULL, 0, reset, '\\',
1285                             nneof, strlen(nneof));
1286                 break;
1287         case DIFF_SYMBOL_SUBMODULE_HEADER:
1288         case DIFF_SYMBOL_SUBMODULE_ERROR:
1289         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1290         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1291         case DIFF_SYMBOL_SUMMARY:
1292         case DIFF_SYMBOL_STATS_LINE:
1293         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1294         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1295                 emit_line(o, "", "", line, len);
1296                 break;
1297         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1298         case DIFF_SYMBOL_CONTEXT_MARKER:
1299                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1300                 reset = diff_get_color_opt(o, DIFF_RESET);
1301                 emit_line(o, context, reset, line, len);
1302                 break;
1303         case DIFF_SYMBOL_SEPARATOR:
1304                 fprintf(o->file, "%s%c",
1305                         diff_line_prefix(o),
1306                         o->line_termination);
1307                 break;
1308         case DIFF_SYMBOL_CONTEXT:
1309                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1310                 reset = diff_get_color_opt(o, DIFF_RESET);
1311                 set_sign = NULL;
1312                 if (o->flags.dual_color_diffed_diffs) {
1313                         char c = !len ? 0 : line[0];
1314
1315                         if (c == '+')
1316                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1317                         else if (c == '@')
1318                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1319                         else if (c == '-')
1320                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1321                 }
1322                 emit_line_ws_markup(o, set_sign, set, reset,
1323                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1324                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1325                 break;
1326         case DIFF_SYMBOL_PLUS:
1327                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1328                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1329                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1330                 case DIFF_SYMBOL_MOVED_LINE |
1331                      DIFF_SYMBOL_MOVED_LINE_ALT |
1332                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1333                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1334                         break;
1335                 case DIFF_SYMBOL_MOVED_LINE |
1336                      DIFF_SYMBOL_MOVED_LINE_ALT:
1337                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1338                         break;
1339                 case DIFF_SYMBOL_MOVED_LINE |
1340                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1341                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1342                         break;
1343                 case DIFF_SYMBOL_MOVED_LINE:
1344                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1345                         break;
1346                 default:
1347                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1348                 }
1349                 reset = diff_get_color_opt(o, DIFF_RESET);
1350                 if (!o->flags.dual_color_diffed_diffs)
1351                         set_sign = NULL;
1352                 else {
1353                         char c = !len ? 0 : line[0];
1354
1355                         set_sign = set;
1356                         if (c == '-')
1357                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1358                         else if (c == '@')
1359                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1360                         else if (c == '+')
1361                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1362                         else
1363                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1364                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1365                 }
1366                 emit_line_ws_markup(o, set_sign, set, reset,
1367                                     OUTPUT_INDICATOR_NEW, line, len,
1368                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1369                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1370                 break;
1371         case DIFF_SYMBOL_MINUS:
1372                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1373                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1374                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1375                 case DIFF_SYMBOL_MOVED_LINE |
1376                      DIFF_SYMBOL_MOVED_LINE_ALT |
1377                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1378                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1379                         break;
1380                 case DIFF_SYMBOL_MOVED_LINE |
1381                      DIFF_SYMBOL_MOVED_LINE_ALT:
1382                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1383                         break;
1384                 case DIFF_SYMBOL_MOVED_LINE |
1385                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1386                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1387                         break;
1388                 case DIFF_SYMBOL_MOVED_LINE:
1389                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1390                         break;
1391                 default:
1392                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1393                 }
1394                 reset = diff_get_color_opt(o, DIFF_RESET);
1395                 if (!o->flags.dual_color_diffed_diffs)
1396                         set_sign = NULL;
1397                 else {
1398                         char c = !len ? 0 : line[0];
1399
1400                         set_sign = set;
1401                         if (c == '+')
1402                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1403                         else if (c == '@')
1404                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1405                         else if (c == '-')
1406                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1407                         else
1408                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1409                 }
1410                 emit_line_ws_markup(o, set_sign, set, reset,
1411                                     OUTPUT_INDICATOR_OLD, line, len,
1412                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1413                 break;
1414         case DIFF_SYMBOL_WORDS_PORCELAIN:
1415                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1416                 reset = diff_get_color_opt(o, DIFF_RESET);
1417                 emit_line(o, context, reset, line, len);
1418                 fputs("~\n", o->file);
1419                 break;
1420         case DIFF_SYMBOL_WORDS:
1421                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1422                 reset = diff_get_color_opt(o, DIFF_RESET);
1423                 /*
1424                  * Skip the prefix character, if any.  With
1425                  * diff_suppress_blank_empty, there may be
1426                  * none.
1427                  */
1428                 if (line[0] != '\n') {
1429                         line++;
1430                         len--;
1431                 }
1432                 emit_line(o, context, reset, line, len);
1433                 break;
1434         case DIFF_SYMBOL_FILEPAIR_PLUS:
1435                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1436                 reset = diff_get_color_opt(o, DIFF_RESET);
1437                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1438                         line, reset,
1439                         strchr(line, ' ') ? "\t" : "");
1440                 break;
1441         case DIFF_SYMBOL_FILEPAIR_MINUS:
1442                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1443                 reset = diff_get_color_opt(o, DIFF_RESET);
1444                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1445                         line, reset,
1446                         strchr(line, ' ') ? "\t" : "");
1447                 break;
1448         case DIFF_SYMBOL_BINARY_FILES:
1449         case DIFF_SYMBOL_HEADER:
1450                 fprintf(o->file, "%s", line);
1451                 break;
1452         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1453                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1454                 break;
1455         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1456                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1457                 break;
1458         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1459                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1460                 break;
1461         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1462                 fputs(diff_line_prefix(o), o->file);
1463                 fputc('\n', o->file);
1464                 break;
1465         case DIFF_SYMBOL_REWRITE_DIFF:
1466                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1467                 reset = diff_get_color_opt(o, DIFF_RESET);
1468                 emit_line(o, fraginfo, reset, line, len);
1469                 break;
1470         case DIFF_SYMBOL_SUBMODULE_ADD:
1471                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1472                 reset = diff_get_color_opt(o, DIFF_RESET);
1473                 emit_line(o, set, reset, line, len);
1474                 break;
1475         case DIFF_SYMBOL_SUBMODULE_DEL:
1476                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1477                 reset = diff_get_color_opt(o, DIFF_RESET);
1478                 emit_line(o, set, reset, line, len);
1479                 break;
1480         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1481                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1482                         diff_line_prefix(o), line);
1483                 break;
1484         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1485                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1486                         diff_line_prefix(o), line);
1487                 break;
1488         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1489                 emit_line(o, "", "", " 0 files changed\n",
1490                           strlen(" 0 files changed\n"));
1491                 break;
1492         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1493                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1494                 break;
1495         case DIFF_SYMBOL_WORD_DIFF:
1496                 fprintf(o->file, "%.*s", len, line);
1497                 break;
1498         case DIFF_SYMBOL_STAT_SEP:
1499                 fputs(o->stat_sep, o->file);
1500                 break;
1501         default:
1502                 BUG("unknown diff symbol");
1503         }
1504         strbuf_release(&sb);
1505 }
1506
1507 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1508                              const char *line, int len, unsigned flags)
1509 {
1510         struct emitted_diff_symbol e = {line, len, flags, s};
1511
1512         if (o->emitted_symbols)
1513                 append_emitted_diff_symbol(o, &e);
1514         else
1515                 emit_diff_symbol_from_struct(o, &e);
1516 }
1517
1518 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1519 {
1520         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1521 }
1522
1523 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1524 {
1525         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1526 }
1527
1528 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1529 {
1530         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1531                          path, strlen(path), 0);
1532 }
1533
1534 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1535 {
1536         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1537                          path, strlen(path), 0);
1538 }
1539
1540 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1541 {
1542         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1543                          header, strlen(header), 0);
1544 }
1545
1546 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1547 {
1548         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1549 }
1550
1551 void diff_emit_submodule_pipethrough(struct diff_options *o,
1552                                      const char *line, int len)
1553 {
1554         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1555 }
1556
1557 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1558 {
1559         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1560               ecbdata->blank_at_eof_in_preimage &&
1561               ecbdata->blank_at_eof_in_postimage &&
1562               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1563               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1564                 return 0;
1565         return ws_blank_line(line, len, ecbdata->ws_rule);
1566 }
1567
1568 static void emit_add_line(const char *reset,
1569                           struct emit_callback *ecbdata,
1570                           const char *line, int len)
1571 {
1572         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1573         if (new_blank_line_at_eof(ecbdata, line, len))
1574                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1575
1576         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1577 }
1578
1579 static void emit_del_line(const char *reset,
1580                           struct emit_callback *ecbdata,
1581                           const char *line, int len)
1582 {
1583         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1584         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1585 }
1586
1587 static void emit_context_line(const char *reset,
1588                               struct emit_callback *ecbdata,
1589                               const char *line, int len)
1590 {
1591         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1592         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1593 }
1594
1595 static void emit_hunk_header(struct emit_callback *ecbdata,
1596                              const char *line, int len)
1597 {
1598         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1599         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1600         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1601         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1602         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1603         static const char atat[2] = { '@', '@' };
1604         const char *cp, *ep;
1605         struct strbuf msgbuf = STRBUF_INIT;
1606         int org_len = len;
1607         int i = 1;
1608
1609         /*
1610          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1611          * it always is at least 10 bytes long.
1612          */
1613         if (len < 10 ||
1614             memcmp(line, atat, 2) ||
1615             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1616                 emit_diff_symbol(ecbdata->opt,
1617                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1618                 return;
1619         }
1620         ep += 2; /* skip over @@ */
1621
1622         /* The hunk header in fraginfo color */
1623         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1624                 strbuf_addstr(&msgbuf, reverse);
1625         strbuf_addstr(&msgbuf, frag);
1626         strbuf_add(&msgbuf, line, ep - line);
1627         strbuf_addstr(&msgbuf, reset);
1628
1629         /*
1630          * trailing "\r\n"
1631          */
1632         for ( ; i < 3; i++)
1633                 if (line[len - i] == '\r' || line[len - i] == '\n')
1634                         len--;
1635
1636         /* blank before the func header */
1637         for (cp = ep; ep - line < len; ep++)
1638                 if (*ep != ' ' && *ep != '\t')
1639                         break;
1640         if (ep != cp) {
1641                 strbuf_addstr(&msgbuf, context);
1642                 strbuf_add(&msgbuf, cp, ep - cp);
1643                 strbuf_addstr(&msgbuf, reset);
1644         }
1645
1646         if (ep < line + len) {
1647                 strbuf_addstr(&msgbuf, func);
1648                 strbuf_add(&msgbuf, ep, line + len - ep);
1649                 strbuf_addstr(&msgbuf, reset);
1650         }
1651
1652         strbuf_add(&msgbuf, line + len, org_len - len);
1653         strbuf_complete_line(&msgbuf);
1654         emit_diff_symbol(ecbdata->opt,
1655                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1656         strbuf_release(&msgbuf);
1657 }
1658
1659 static struct diff_tempfile *claim_diff_tempfile(void) {
1660         int i;
1661         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1662                 if (!diff_temp[i].name)
1663                         return diff_temp + i;
1664         BUG("diff is failing to clean up its tempfiles");
1665 }
1666
1667 static void remove_tempfile(void)
1668 {
1669         int i;
1670         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1671                 if (is_tempfile_active(diff_temp[i].tempfile))
1672                         delete_tempfile(&diff_temp[i].tempfile);
1673                 diff_temp[i].name = NULL;
1674         }
1675 }
1676
1677 static void add_line_count(struct strbuf *out, int count)
1678 {
1679         switch (count) {
1680         case 0:
1681                 strbuf_addstr(out, "0,0");
1682                 break;
1683         case 1:
1684                 strbuf_addstr(out, "1");
1685                 break;
1686         default:
1687                 strbuf_addf(out, "1,%d", count);
1688                 break;
1689         }
1690 }
1691
1692 static void emit_rewrite_lines(struct emit_callback *ecb,
1693                                int prefix, const char *data, int size)
1694 {
1695         const char *endp = NULL;
1696         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1697
1698         while (0 < size) {
1699                 int len;
1700
1701                 endp = memchr(data, '\n', size);
1702                 len = endp ? (endp - data + 1) : size;
1703                 if (prefix != '+') {
1704                         ecb->lno_in_preimage++;
1705                         emit_del_line(reset, ecb, data, len);
1706                 } else {
1707                         ecb->lno_in_postimage++;
1708                         emit_add_line(reset, ecb, data, len);
1709                 }
1710                 size -= len;
1711                 data += len;
1712         }
1713         if (!endp)
1714                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1715 }
1716
1717 static void emit_rewrite_diff(const char *name_a,
1718                               const char *name_b,
1719                               struct diff_filespec *one,
1720                               struct diff_filespec *two,
1721                               struct userdiff_driver *textconv_one,
1722                               struct userdiff_driver *textconv_two,
1723                               struct diff_options *o)
1724 {
1725         int lc_a, lc_b;
1726         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1727         const char *a_prefix, *b_prefix;
1728         char *data_one, *data_two;
1729         size_t size_one, size_two;
1730         struct emit_callback ecbdata;
1731         struct strbuf out = STRBUF_INIT;
1732
1733         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1734                 a_prefix = o->b_prefix;
1735                 b_prefix = o->a_prefix;
1736         } else {
1737                 a_prefix = o->a_prefix;
1738                 b_prefix = o->b_prefix;
1739         }
1740
1741         name_a += (*name_a == '/');
1742         name_b += (*name_b == '/');
1743
1744         strbuf_reset(&a_name);
1745         strbuf_reset(&b_name);
1746         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1747         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1748
1749         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1750         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1751
1752         memset(&ecbdata, 0, sizeof(ecbdata));
1753         ecbdata.color_diff = want_color(o->use_color);
1754         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1755         ecbdata.opt = o;
1756         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1757                 mmfile_t mf1, mf2;
1758                 mf1.ptr = (char *)data_one;
1759                 mf2.ptr = (char *)data_two;
1760                 mf1.size = size_one;
1761                 mf2.size = size_two;
1762                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1763         }
1764         ecbdata.lno_in_preimage = 1;
1765         ecbdata.lno_in_postimage = 1;
1766
1767         lc_a = count_lines(data_one, size_one);
1768         lc_b = count_lines(data_two, size_two);
1769
1770         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1771                          a_name.buf, a_name.len, 0);
1772         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1773                          b_name.buf, b_name.len, 0);
1774
1775         strbuf_addstr(&out, "@@ -");
1776         if (!o->irreversible_delete)
1777                 add_line_count(&out, lc_a);
1778         else
1779                 strbuf_addstr(&out, "?,?");
1780         strbuf_addstr(&out, " +");
1781         add_line_count(&out, lc_b);
1782         strbuf_addstr(&out, " @@\n");
1783         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1784         strbuf_release(&out);
1785
1786         if (lc_a && !o->irreversible_delete)
1787                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1788         if (lc_b)
1789                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1790         if (textconv_one)
1791                 free((char *)data_one);
1792         if (textconv_two)
1793                 free((char *)data_two);
1794 }
1795
1796 struct diff_words_buffer {
1797         mmfile_t text;
1798         unsigned long alloc;
1799         struct diff_words_orig {
1800                 const char *begin, *end;
1801         } *orig;
1802         int orig_nr, orig_alloc;
1803 };
1804
1805 static void diff_words_append(char *line, unsigned long len,
1806                 struct diff_words_buffer *buffer)
1807 {
1808         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1809         line++;
1810         len--;
1811         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1812         buffer->text.size += len;
1813         buffer->text.ptr[buffer->text.size] = '\0';
1814 }
1815
1816 struct diff_words_style_elem {
1817         const char *prefix;
1818         const char *suffix;
1819         const char *color; /* NULL; filled in by the setup code if
1820                             * color is enabled */
1821 };
1822
1823 struct diff_words_style {
1824         enum diff_words_type type;
1825         struct diff_words_style_elem new_word, old_word, ctx;
1826         const char *newline;
1827 };
1828
1829 static struct diff_words_style diff_words_styles[] = {
1830         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1831         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1832         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1833 };
1834
1835 struct diff_words_data {
1836         struct diff_words_buffer minus, plus;
1837         const char *current_plus;
1838         int last_minus;
1839         struct diff_options *opt;
1840         regex_t *word_regex;
1841         enum diff_words_type type;
1842         struct diff_words_style *style;
1843 };
1844
1845 static int fn_out_diff_words_write_helper(struct diff_options *o,
1846                                           struct diff_words_style_elem *st_el,
1847                                           const char *newline,
1848                                           size_t count, const char *buf)
1849 {
1850         int print = 0;
1851         struct strbuf sb = STRBUF_INIT;
1852
1853         while (count) {
1854                 char *p = memchr(buf, '\n', count);
1855                 if (print)
1856                         strbuf_addstr(&sb, diff_line_prefix(o));
1857
1858                 if (p != buf) {
1859                         const char *reset = st_el->color && *st_el->color ?
1860                                             GIT_COLOR_RESET : NULL;
1861                         if (st_el->color && *st_el->color)
1862                                 strbuf_addstr(&sb, st_el->color);
1863                         strbuf_addstr(&sb, st_el->prefix);
1864                         strbuf_add(&sb, buf, p ? p - buf : count);
1865                         strbuf_addstr(&sb, st_el->suffix);
1866                         if (reset)
1867                                 strbuf_addstr(&sb, reset);
1868                 }
1869                 if (!p)
1870                         goto out;
1871
1872                 strbuf_addstr(&sb, newline);
1873                 count -= p + 1 - buf;
1874                 buf = p + 1;
1875                 print = 1;
1876                 if (count) {
1877                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1878                                          sb.buf, sb.len, 0);
1879                         strbuf_reset(&sb);
1880                 }
1881         }
1882
1883 out:
1884         if (sb.len)
1885                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1886                                  sb.buf, sb.len, 0);
1887         strbuf_release(&sb);
1888         return 0;
1889 }
1890
1891 /*
1892  * '--color-words' algorithm can be described as:
1893  *
1894  *   1. collect the minus/plus lines of a diff hunk, divided into
1895  *      minus-lines and plus-lines;
1896  *
1897  *   2. break both minus-lines and plus-lines into words and
1898  *      place them into two mmfile_t with one word for each line;
1899  *
1900  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1901  *
1902  * And for the common parts of the both file, we output the plus side text.
1903  * diff_words->current_plus is used to trace the current position of the plus file
1904  * which printed. diff_words->last_minus is used to trace the last minus word
1905  * printed.
1906  *
1907  * For '--graph' to work with '--color-words', we need to output the graph prefix
1908  * on each line of color words output. Generally, there are two conditions on
1909  * which we should output the prefix.
1910  *
1911  *   1. diff_words->last_minus == 0 &&
1912  *      diff_words->current_plus == diff_words->plus.text.ptr
1913  *
1914  *      that is: the plus text must start as a new line, and if there is no minus
1915  *      word printed, a graph prefix must be printed.
1916  *
1917  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1918  *      *(diff_words->current_plus - 1) == '\n'
1919  *
1920  *      that is: a graph prefix must be printed following a '\n'
1921  */
1922 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1923 {
1924         if ((diff_words->last_minus == 0 &&
1925                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1926                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1927                 *(diff_words->current_plus - 1) == '\n')) {
1928                 return 1;
1929         } else {
1930                 return 0;
1931         }
1932 }
1933
1934 static void fn_out_diff_words_aux(void *priv,
1935                                   long minus_first, long minus_len,
1936                                   long plus_first, long plus_len,
1937                                   const char *func, long funclen)
1938 {
1939         struct diff_words_data *diff_words = priv;
1940         struct diff_words_style *style = diff_words->style;
1941         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1942         struct diff_options *opt = diff_words->opt;
1943         const char *line_prefix;
1944
1945         assert(opt);
1946         line_prefix = diff_line_prefix(opt);
1947
1948         /* POSIX requires that first be decremented by one if len == 0... */
1949         if (minus_len) {
1950                 minus_begin = diff_words->minus.orig[minus_first].begin;
1951                 minus_end =
1952                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1953         } else
1954                 minus_begin = minus_end =
1955                         diff_words->minus.orig[minus_first].end;
1956
1957         if (plus_len) {
1958                 plus_begin = diff_words->plus.orig[plus_first].begin;
1959                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1960         } else
1961                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1962
1963         if (color_words_output_graph_prefix(diff_words)) {
1964                 fputs(line_prefix, diff_words->opt->file);
1965         }
1966         if (diff_words->current_plus != plus_begin) {
1967                 fn_out_diff_words_write_helper(diff_words->opt,
1968                                 &style->ctx, style->newline,
1969                                 plus_begin - diff_words->current_plus,
1970                                 diff_words->current_plus);
1971         }
1972         if (minus_begin != minus_end) {
1973                 fn_out_diff_words_write_helper(diff_words->opt,
1974                                 &style->old_word, style->newline,
1975                                 minus_end - minus_begin, minus_begin);
1976         }
1977         if (plus_begin != plus_end) {
1978                 fn_out_diff_words_write_helper(diff_words->opt,
1979                                 &style->new_word, style->newline,
1980                                 plus_end - plus_begin, plus_begin);
1981         }
1982
1983         diff_words->current_plus = plus_end;
1984         diff_words->last_minus = minus_first;
1985 }
1986
1987 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1988 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1989                 int *begin, int *end)
1990 {
1991         if (word_regex && *begin < buffer->size) {
1992                 regmatch_t match[1];
1993                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1994                                  buffer->size - *begin, 1, match, 0)) {
1995                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1996                                         '\n', match[0].rm_eo - match[0].rm_so);
1997                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1998                         *begin += match[0].rm_so;
1999                         return *begin >= *end;
2000                 }
2001                 return -1;
2002         }
2003
2004         /* find the next word */
2005         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2006                 (*begin)++;
2007         if (*begin >= buffer->size)
2008                 return -1;
2009
2010         /* find the end of the word */
2011         *end = *begin + 1;
2012         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2013                 (*end)++;
2014
2015         return 0;
2016 }
2017
2018 /*
2019  * This function splits the words in buffer->text, stores the list with
2020  * newline separator into out, and saves the offsets of the original words
2021  * in buffer->orig.
2022  */
2023 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2024                 regex_t *word_regex)
2025 {
2026         int i, j;
2027         long alloc = 0;
2028
2029         out->size = 0;
2030         out->ptr = NULL;
2031
2032         /* fake an empty "0th" word */
2033         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2034         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2035         buffer->orig_nr = 1;
2036
2037         for (i = 0; i < buffer->text.size; i++) {
2038                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2039                         return;
2040
2041                 /* store original boundaries */
2042                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2043                                 buffer->orig_alloc);
2044                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2045                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2046                 buffer->orig_nr++;
2047
2048                 /* store one word */
2049                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2050                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2051                 out->ptr[out->size + j - i] = '\n';
2052                 out->size += j - i + 1;
2053
2054                 i = j - 1;
2055         }
2056 }
2057
2058 /* this executes the word diff on the accumulated buffers */
2059 static void diff_words_show(struct diff_words_data *diff_words)
2060 {
2061         xpparam_t xpp;
2062         xdemitconf_t xecfg;
2063         mmfile_t minus, plus;
2064         struct diff_words_style *style = diff_words->style;
2065
2066         struct diff_options *opt = diff_words->opt;
2067         const char *line_prefix;
2068
2069         assert(opt);
2070         line_prefix = diff_line_prefix(opt);
2071
2072         /* special case: only removal */
2073         if (!diff_words->plus.text.size) {
2074                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2075                                  line_prefix, strlen(line_prefix), 0);
2076                 fn_out_diff_words_write_helper(diff_words->opt,
2077                         &style->old_word, style->newline,
2078                         diff_words->minus.text.size,
2079                         diff_words->minus.text.ptr);
2080                 diff_words->minus.text.size = 0;
2081                 return;
2082         }
2083
2084         diff_words->current_plus = diff_words->plus.text.ptr;
2085         diff_words->last_minus = 0;
2086
2087         memset(&xpp, 0, sizeof(xpp));
2088         memset(&xecfg, 0, sizeof(xecfg));
2089         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2090         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2091         xpp.flags = 0;
2092         /* as only the hunk header will be parsed, we need a 0-context */
2093         xecfg.ctxlen = 0;
2094         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2095                           diff_words, &xpp, &xecfg))
2096                 die("unable to generate word diff");
2097         free(minus.ptr);
2098         free(plus.ptr);
2099         if (diff_words->current_plus != diff_words->plus.text.ptr +
2100                         diff_words->plus.text.size) {
2101                 if (color_words_output_graph_prefix(diff_words))
2102                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2103                                          line_prefix, strlen(line_prefix), 0);
2104                 fn_out_diff_words_write_helper(diff_words->opt,
2105                         &style->ctx, style->newline,
2106                         diff_words->plus.text.ptr + diff_words->plus.text.size
2107                         - diff_words->current_plus, diff_words->current_plus);
2108         }
2109         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2110 }
2111
2112 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2113 static void diff_words_flush(struct emit_callback *ecbdata)
2114 {
2115         struct diff_options *wo = ecbdata->diff_words->opt;
2116
2117         if (ecbdata->diff_words->minus.text.size ||
2118             ecbdata->diff_words->plus.text.size)
2119                 diff_words_show(ecbdata->diff_words);
2120
2121         if (wo->emitted_symbols) {
2122                 struct diff_options *o = ecbdata->opt;
2123                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2124                 int i;
2125
2126                 /*
2127                  * NEEDSWORK:
2128                  * Instead of appending each, concat all words to a line?
2129                  */
2130                 for (i = 0; i < wol->nr; i++)
2131                         append_emitted_diff_symbol(o, &wol->buf[i]);
2132
2133                 for (i = 0; i < wol->nr; i++)
2134                         free((void *)wol->buf[i].line);
2135
2136                 wol->nr = 0;
2137         }
2138 }
2139
2140 static void diff_filespec_load_driver(struct diff_filespec *one,
2141                                       struct index_state *istate)
2142 {
2143         /* Use already-loaded driver */
2144         if (one->driver)
2145                 return;
2146
2147         if (S_ISREG(one->mode))
2148                 one->driver = userdiff_find_by_path(istate, one->path);
2149
2150         /* Fallback to default settings */
2151         if (!one->driver)
2152                 one->driver = userdiff_find_by_name("default");
2153 }
2154
2155 static const char *userdiff_word_regex(struct diff_filespec *one,
2156                                        struct index_state *istate)
2157 {
2158         diff_filespec_load_driver(one, istate);
2159         return one->driver->word_regex;
2160 }
2161
2162 static void init_diff_words_data(struct emit_callback *ecbdata,
2163                                  struct diff_options *orig_opts,
2164                                  struct diff_filespec *one,
2165                                  struct diff_filespec *two)
2166 {
2167         int i;
2168         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2169         memcpy(o, orig_opts, sizeof(struct diff_options));
2170
2171         ecbdata->diff_words =
2172                 xcalloc(1, sizeof(struct diff_words_data));
2173         ecbdata->diff_words->type = o->word_diff;
2174         ecbdata->diff_words->opt = o;
2175
2176         if (orig_opts->emitted_symbols)
2177                 o->emitted_symbols =
2178                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2179
2180         if (!o->word_regex)
2181                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2182         if (!o->word_regex)
2183                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2184         if (!o->word_regex)
2185                 o->word_regex = diff_word_regex_cfg;
2186         if (o->word_regex) {
2187                 ecbdata->diff_words->word_regex = (regex_t *)
2188                         xmalloc(sizeof(regex_t));
2189                 if (regcomp(ecbdata->diff_words->word_regex,
2190                             o->word_regex,
2191                             REG_EXTENDED | REG_NEWLINE))
2192                         die("invalid regular expression: %s",
2193                             o->word_regex);
2194         }
2195         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2196                 if (o->word_diff == diff_words_styles[i].type) {
2197                         ecbdata->diff_words->style =
2198                                 &diff_words_styles[i];
2199                         break;
2200                 }
2201         }
2202         if (want_color(o->use_color)) {
2203                 struct diff_words_style *st = ecbdata->diff_words->style;
2204                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2205                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2206                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2207         }
2208 }
2209
2210 static void free_diff_words_data(struct emit_callback *ecbdata)
2211 {
2212         if (ecbdata->diff_words) {
2213                 diff_words_flush(ecbdata);
2214                 free (ecbdata->diff_words->opt->emitted_symbols);
2215                 free (ecbdata->diff_words->opt);
2216                 free (ecbdata->diff_words->minus.text.ptr);
2217                 free (ecbdata->diff_words->minus.orig);
2218                 free (ecbdata->diff_words->plus.text.ptr);
2219                 free (ecbdata->diff_words->plus.orig);
2220                 if (ecbdata->diff_words->word_regex) {
2221                         regfree(ecbdata->diff_words->word_regex);
2222                         free(ecbdata->diff_words->word_regex);
2223                 }
2224                 FREE_AND_NULL(ecbdata->diff_words);
2225         }
2226 }
2227
2228 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2229 {
2230         if (want_color(diff_use_color))
2231                 return diff_colors[ix];
2232         return "";
2233 }
2234
2235 const char *diff_line_prefix(struct diff_options *opt)
2236 {
2237         struct strbuf *msgbuf;
2238         if (!opt->output_prefix)
2239                 return "";
2240
2241         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2242         return msgbuf->buf;
2243 }
2244
2245 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2246 {
2247         const char *cp;
2248         unsigned long allot;
2249         size_t l = len;
2250
2251         cp = line;
2252         allot = l;
2253         while (0 < l) {
2254                 (void) utf8_width(&cp, &l);
2255                 if (!cp)
2256                         break; /* truncated in the middle? */
2257         }
2258         return allot - l;
2259 }
2260
2261 static void find_lno(const char *line, struct emit_callback *ecbdata)
2262 {
2263         const char *p;
2264         ecbdata->lno_in_preimage = 0;
2265         ecbdata->lno_in_postimage = 0;
2266         p = strchr(line, '-');
2267         if (!p)
2268                 return; /* cannot happen */
2269         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2270         p = strchr(p, '+');
2271         if (!p)
2272                 return; /* cannot happen */
2273         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2274 }
2275
2276 static void fn_out_consume(void *priv, char *line, unsigned long len)
2277 {
2278         struct emit_callback *ecbdata = priv;
2279         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2280         struct diff_options *o = ecbdata->opt;
2281
2282         o->found_changes = 1;
2283
2284         if (ecbdata->header) {
2285                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2286                                  ecbdata->header->buf, ecbdata->header->len, 0);
2287                 strbuf_reset(ecbdata->header);
2288                 ecbdata->header = NULL;
2289         }
2290
2291         if (ecbdata->label_path[0]) {
2292                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2293                                  ecbdata->label_path[0],
2294                                  strlen(ecbdata->label_path[0]), 0);
2295                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2296                                  ecbdata->label_path[1],
2297                                  strlen(ecbdata->label_path[1]), 0);
2298                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2299         }
2300
2301         if (diff_suppress_blank_empty
2302             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2303                 line[0] = '\n';
2304                 len = 1;
2305         }
2306
2307         if (line[0] == '@') {
2308                 if (ecbdata->diff_words)
2309                         diff_words_flush(ecbdata);
2310                 len = sane_truncate_line(ecbdata, line, len);
2311                 find_lno(line, ecbdata);
2312                 emit_hunk_header(ecbdata, line, len);
2313                 return;
2314         }
2315
2316         if (ecbdata->diff_words) {
2317                 enum diff_symbol s =
2318                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2319                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2320                 if (line[0] == '-') {
2321                         diff_words_append(line, len,
2322                                           &ecbdata->diff_words->minus);
2323                         return;
2324                 } else if (line[0] == '+') {
2325                         diff_words_append(line, len,
2326                                           &ecbdata->diff_words->plus);
2327                         return;
2328                 } else if (starts_with(line, "\\ ")) {
2329                         /*
2330                          * Eat the "no newline at eof" marker as if we
2331                          * saw a "+" or "-" line with nothing on it,
2332                          * and return without diff_words_flush() to
2333                          * defer processing. If this is the end of
2334                          * preimage, more "+" lines may come after it.
2335                          */
2336                         return;
2337                 }
2338                 diff_words_flush(ecbdata);
2339                 emit_diff_symbol(o, s, line, len, 0);
2340                 return;
2341         }
2342
2343         switch (line[0]) {
2344         case '+':
2345                 ecbdata->lno_in_postimage++;
2346                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2347                 break;
2348         case '-':
2349                 ecbdata->lno_in_preimage++;
2350                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2351                 break;
2352         case ' ':
2353                 ecbdata->lno_in_postimage++;
2354                 ecbdata->lno_in_preimage++;
2355                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2356                 break;
2357         default:
2358                 /* incomplete line at the end */
2359                 ecbdata->lno_in_preimage++;
2360                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2361                                  line, len, 0);
2362                 break;
2363         }
2364 }
2365
2366 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2367 {
2368         const char *old_name = a;
2369         const char *new_name = b;
2370         int pfx_length, sfx_length;
2371         int pfx_adjust_for_slash;
2372         int len_a = strlen(a);
2373         int len_b = strlen(b);
2374         int a_midlen, b_midlen;
2375         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2376         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2377
2378         if (qlen_a || qlen_b) {
2379                 quote_c_style(a, name, NULL, 0);
2380                 strbuf_addstr(name, " => ");
2381                 quote_c_style(b, name, NULL, 0);
2382                 return;
2383         }
2384
2385         /* Find common prefix */
2386         pfx_length = 0;
2387         while (*old_name && *new_name && *old_name == *new_name) {
2388                 if (*old_name == '/')
2389                         pfx_length = old_name - a + 1;
2390                 old_name++;
2391                 new_name++;
2392         }
2393
2394         /* Find common suffix */
2395         old_name = a + len_a;
2396         new_name = b + len_b;
2397         sfx_length = 0;
2398         /*
2399          * If there is a common prefix, it must end in a slash.  In
2400          * that case we let this loop run 1 into the prefix to see the
2401          * same slash.
2402          *
2403          * If there is no common prefix, we cannot do this as it would
2404          * underrun the input strings.
2405          */
2406         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2407         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2408                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2409                *old_name == *new_name) {
2410                 if (*old_name == '/')
2411                         sfx_length = len_a - (old_name - a);
2412                 old_name--;
2413                 new_name--;
2414         }
2415
2416         /*
2417          * pfx{mid-a => mid-b}sfx
2418          * {pfx-a => pfx-b}sfx
2419          * pfx{sfx-a => sfx-b}
2420          * name-a => name-b
2421          */
2422         a_midlen = len_a - pfx_length - sfx_length;
2423         b_midlen = len_b - pfx_length - sfx_length;
2424         if (a_midlen < 0)
2425                 a_midlen = 0;
2426         if (b_midlen < 0)
2427                 b_midlen = 0;
2428
2429         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2430         if (pfx_length + sfx_length) {
2431                 strbuf_add(name, a, pfx_length);
2432                 strbuf_addch(name, '{');
2433         }
2434         strbuf_add(name, a + pfx_length, a_midlen);
2435         strbuf_addstr(name, " => ");
2436         strbuf_add(name, b + pfx_length, b_midlen);
2437         if (pfx_length + sfx_length) {
2438                 strbuf_addch(name, '}');
2439                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2440         }
2441 }
2442
2443 struct diffstat_t {
2444         int nr;
2445         int alloc;
2446         struct diffstat_file {
2447                 char *from_name;
2448                 char *name;
2449                 char *print_name;
2450                 const char *comments;
2451                 unsigned is_unmerged:1;
2452                 unsigned is_binary:1;
2453                 unsigned is_renamed:1;
2454                 unsigned is_interesting:1;
2455                 uintmax_t added, deleted;
2456         } **files;
2457 };
2458
2459 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2460                                           const char *name_a,
2461                                           const char *name_b)
2462 {
2463         struct diffstat_file *x;
2464         x = xcalloc(1, sizeof(*x));
2465         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2466         diffstat->files[diffstat->nr++] = x;
2467         if (name_b) {
2468                 x->from_name = xstrdup(name_a);
2469                 x->name = xstrdup(name_b);
2470                 x->is_renamed = 1;
2471         }
2472         else {
2473                 x->from_name = NULL;
2474                 x->name = xstrdup(name_a);
2475         }
2476         return x;
2477 }
2478
2479 static void diffstat_consume(void *priv, char *line, unsigned long len)
2480 {
2481         struct diffstat_t *diffstat = priv;
2482         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2483
2484         if (line[0] == '+')
2485                 x->added++;
2486         else if (line[0] == '-')
2487                 x->deleted++;
2488 }
2489
2490 const char mime_boundary_leader[] = "------------";
2491
2492 static int scale_linear(int it, int width, int max_change)
2493 {
2494         if (!it)
2495                 return 0;
2496         /*
2497          * make sure that at least one '-' or '+' is printed if
2498          * there is any change to this path. The easiest way is to
2499          * scale linearly as if the alloted width is one column shorter
2500          * than it is, and then add 1 to the result.
2501          */
2502         return 1 + (it * (width - 1) / max_change);
2503 }
2504
2505 static void show_graph(struct strbuf *out, char ch, int cnt,
2506                        const char *set, const char *reset)
2507 {
2508         if (cnt <= 0)
2509                 return;
2510         strbuf_addstr(out, set);
2511         strbuf_addchars(out, ch, cnt);
2512         strbuf_addstr(out, reset);
2513 }
2514
2515 static void fill_print_name(struct diffstat_file *file)
2516 {
2517         struct strbuf pname = STRBUF_INIT;
2518
2519         if (file->print_name)
2520                 return;
2521
2522         if (file->is_renamed)
2523                 pprint_rename(&pname, file->from_name, file->name);
2524         else
2525                 quote_c_style(file->name, &pname, NULL, 0);
2526
2527         if (file->comments)
2528                 strbuf_addf(&pname, " (%s)", file->comments);
2529
2530         file->print_name = strbuf_detach(&pname, NULL);
2531 }
2532
2533 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2534                 int files, int insertions, int deletions)
2535 {
2536         struct strbuf sb = STRBUF_INIT;
2537
2538         if (!files) {
2539                 assert(insertions == 0 && deletions == 0);
2540                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2541                                  NULL, 0, 0);
2542                 return;
2543         }
2544
2545         strbuf_addf(&sb,
2546                     (files == 1) ? " %d file changed" : " %d files changed",
2547                     files);
2548
2549         /*
2550          * For binary diff, the caller may want to print "x files
2551          * changed" with insertions == 0 && deletions == 0.
2552          *
2553          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2554          * is probably less confusing (i.e skip over "2 files changed
2555          * but nothing about added/removed lines? Is this a bug in Git?").
2556          */
2557         if (insertions || deletions == 0) {
2558                 strbuf_addf(&sb,
2559                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2560                             insertions);
2561         }
2562
2563         if (deletions || insertions == 0) {
2564                 strbuf_addf(&sb,
2565                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2566                             deletions);
2567         }
2568         strbuf_addch(&sb, '\n');
2569         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2570                          sb.buf, sb.len, 0);
2571         strbuf_release(&sb);
2572 }
2573
2574 void print_stat_summary(FILE *fp, int files,
2575                         int insertions, int deletions)
2576 {
2577         struct diff_options o;
2578         memset(&o, 0, sizeof(o));
2579         o.file = fp;
2580
2581         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2582 }
2583
2584 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2585 {
2586         int i, len, add, del, adds = 0, dels = 0;
2587         uintmax_t max_change = 0, max_len = 0;
2588         int total_files = data->nr, count;
2589         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2590         const char *reset, *add_c, *del_c;
2591         int extra_shown = 0;
2592         const char *line_prefix = diff_line_prefix(options);
2593         struct strbuf out = STRBUF_INIT;
2594
2595         if (data->nr == 0)
2596                 return;
2597
2598         count = options->stat_count ? options->stat_count : data->nr;
2599
2600         reset = diff_get_color_opt(options, DIFF_RESET);
2601         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2602         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2603
2604         /*
2605          * Find the longest filename and max number of changes
2606          */
2607         for (i = 0; (i < count) && (i < data->nr); i++) {
2608                 struct diffstat_file *file = data->files[i];
2609                 uintmax_t change = file->added + file->deleted;
2610
2611                 if (!file->is_interesting && (change == 0)) {
2612                         count++; /* not shown == room for one more */
2613                         continue;
2614                 }
2615                 fill_print_name(file);
2616                 len = strlen(file->print_name);
2617                 if (max_len < len)
2618                         max_len = len;
2619
2620                 if (file->is_unmerged) {
2621                         /* "Unmerged" is 8 characters */
2622                         bin_width = bin_width < 8 ? 8 : bin_width;
2623                         continue;
2624                 }
2625                 if (file->is_binary) {
2626                         /* "Bin XXX -> YYY bytes" */
2627                         int w = 14 + decimal_width(file->added)
2628                                 + decimal_width(file->deleted);
2629                         bin_width = bin_width < w ? w : bin_width;
2630                         /* Display change counts aligned with "Bin" */
2631                         number_width = 3;
2632                         continue;
2633                 }
2634
2635                 if (max_change < change)
2636                         max_change = change;
2637         }
2638         count = i; /* where we can stop scanning in data->files[] */
2639
2640         /*
2641          * We have width = stat_width or term_columns() columns total.
2642          * We want a maximum of min(max_len, stat_name_width) for the name part.
2643          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2644          * We also need 1 for " " and 4 + decimal_width(max_change)
2645          * for " | NNNN " and one the empty column at the end, altogether
2646          * 6 + decimal_width(max_change).
2647          *
2648          * If there's not enough space, we will use the smaller of
2649          * stat_name_width (if set) and 5/8*width for the filename,
2650          * and the rest for constant elements + graph part, but no more
2651          * than stat_graph_width for the graph part.
2652          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2653          * for the standard terminal size).
2654          *
2655          * In other words: stat_width limits the maximum width, and
2656          * stat_name_width fixes the maximum width of the filename,
2657          * and is also used to divide available columns if there
2658          * aren't enough.
2659          *
2660          * Binary files are displayed with "Bin XXX -> YYY bytes"
2661          * instead of the change count and graph. This part is treated
2662          * similarly to the graph part, except that it is not
2663          * "scaled". If total width is too small to accommodate the
2664          * guaranteed minimum width of the filename part and the
2665          * separators and this message, this message will "overflow"
2666          * making the line longer than the maximum width.
2667          */
2668
2669         if (options->stat_width == -1)
2670                 width = term_columns() - strlen(line_prefix);
2671         else
2672                 width = options->stat_width ? options->stat_width : 80;
2673         number_width = decimal_width(max_change) > number_width ?
2674                 decimal_width(max_change) : number_width;
2675
2676         if (options->stat_graph_width == -1)
2677                 options->stat_graph_width = diff_stat_graph_width;
2678
2679         /*
2680          * Guarantee 3/8*16==6 for the graph part
2681          * and 5/8*16==10 for the filename part
2682          */
2683         if (width < 16 + 6 + number_width)
2684                 width = 16 + 6 + number_width;
2685
2686         /*
2687          * First assign sizes that are wanted, ignoring available width.
2688          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2689          * starting from "XXX" should fit in graph_width.
2690          */
2691         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2692         if (options->stat_graph_width &&
2693             options->stat_graph_width < graph_width)
2694                 graph_width = options->stat_graph_width;
2695
2696         name_width = (options->stat_name_width > 0 &&
2697                       options->stat_name_width < max_len) ?
2698                 options->stat_name_width : max_len;
2699
2700         /*
2701          * Adjust adjustable widths not to exceed maximum width
2702          */
2703         if (name_width + number_width + 6 + graph_width > width) {
2704                 if (graph_width > width * 3/8 - number_width - 6) {
2705                         graph_width = width * 3/8 - number_width - 6;
2706                         if (graph_width < 6)
2707                                 graph_width = 6;
2708                 }
2709
2710                 if (options->stat_graph_width &&
2711                     graph_width > options->stat_graph_width)
2712                         graph_width = options->stat_graph_width;
2713                 if (name_width > width - number_width - 6 - graph_width)
2714                         name_width = width - number_width - 6 - graph_width;
2715                 else
2716                         graph_width = width - number_width - 6 - name_width;
2717         }
2718
2719         /*
2720          * From here name_width is the width of the name area,
2721          * and graph_width is the width of the graph area.
2722          * max_change is used to scale graph properly.
2723          */
2724         for (i = 0; i < count; i++) {
2725                 const char *prefix = "";
2726                 struct diffstat_file *file = data->files[i];
2727                 char *name = file->print_name;
2728                 uintmax_t added = file->added;
2729                 uintmax_t deleted = file->deleted;
2730                 int name_len;
2731
2732                 if (!file->is_interesting && (added + deleted == 0))
2733                         continue;
2734
2735                 /*
2736                  * "scale" the filename
2737                  */
2738                 len = name_width;
2739                 name_len = strlen(name);
2740                 if (name_width < name_len) {
2741                         char *slash;
2742                         prefix = "...";
2743                         len -= 3;
2744                         name += name_len - len;
2745                         slash = strchr(name, '/');
2746                         if (slash)
2747                                 name = slash;
2748                 }
2749
2750                 if (file->is_binary) {
2751                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2752                         strbuf_addf(&out, " %*s", number_width, "Bin");
2753                         if (!added && !deleted) {
2754                                 strbuf_addch(&out, '\n');
2755                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2756                                                  out.buf, out.len, 0);
2757                                 strbuf_reset(&out);
2758                                 continue;
2759                         }
2760                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2761                                 del_c, deleted, reset);
2762                         strbuf_addstr(&out, " -> ");
2763                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2764                                 add_c, added, reset);
2765                         strbuf_addstr(&out, " bytes\n");
2766                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2767                                          out.buf, out.len, 0);
2768                         strbuf_reset(&out);
2769                         continue;
2770                 }
2771                 else if (file->is_unmerged) {
2772                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2773                         strbuf_addstr(&out, " Unmerged\n");
2774                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2775                                          out.buf, out.len, 0);
2776                         strbuf_reset(&out);
2777                         continue;
2778                 }
2779
2780                 /*
2781                  * scale the add/delete
2782                  */
2783                 add = added;
2784                 del = deleted;
2785
2786                 if (graph_width <= max_change) {
2787                         int total = scale_linear(add + del, graph_width, max_change);
2788                         if (total < 2 && add && del)
2789                                 /* width >= 2 due to the sanity check */
2790                                 total = 2;
2791                         if (add < del) {
2792                                 add = scale_linear(add, graph_width, max_change);
2793                                 del = total - add;
2794                         } else {
2795                                 del = scale_linear(del, graph_width, max_change);
2796                                 add = total - del;
2797                         }
2798                 }
2799                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2800                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2801                         number_width, added + deleted,
2802                         added + deleted ? " " : "");
2803                 show_graph(&out, '+', add, add_c, reset);
2804                 show_graph(&out, '-', del, del_c, reset);
2805                 strbuf_addch(&out, '\n');
2806                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2807                                  out.buf, out.len, 0);
2808                 strbuf_reset(&out);
2809         }
2810
2811         for (i = 0; i < data->nr; i++) {
2812                 struct diffstat_file *file = data->files[i];
2813                 uintmax_t added = file->added;
2814                 uintmax_t deleted = file->deleted;
2815
2816                 if (file->is_unmerged ||
2817                     (!file->is_interesting && (added + deleted == 0))) {
2818                         total_files--;
2819                         continue;
2820                 }
2821
2822                 if (!file->is_binary) {
2823                         adds += added;
2824                         dels += deleted;
2825                 }
2826                 if (i < count)
2827                         continue;
2828                 if (!extra_shown)
2829                         emit_diff_symbol(options,
2830                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2831                                          NULL, 0, 0);
2832                 extra_shown = 1;
2833         }
2834
2835         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2836         strbuf_release(&out);
2837 }
2838
2839 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2840 {
2841         int i, adds = 0, dels = 0, total_files = data->nr;
2842
2843         if (data->nr == 0)
2844                 return;
2845
2846         for (i = 0; i < data->nr; i++) {
2847                 int added = data->files[i]->added;
2848                 int deleted = data->files[i]->deleted;
2849
2850                 if (data->files[i]->is_unmerged ||
2851                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2852                         total_files--;
2853                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2854                         adds += added;
2855                         dels += deleted;
2856                 }
2857         }
2858         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2859 }
2860
2861 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2862 {
2863         int i;
2864
2865         if (data->nr == 0)
2866                 return;
2867
2868         for (i = 0; i < data->nr; i++) {
2869                 struct diffstat_file *file = data->files[i];
2870
2871                 fprintf(options->file, "%s", diff_line_prefix(options));
2872
2873                 if (file->is_binary)
2874                         fprintf(options->file, "-\t-\t");
2875                 else
2876                         fprintf(options->file,
2877                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2878                                 file->added, file->deleted);
2879                 if (options->line_termination) {
2880                         fill_print_name(file);
2881                         if (!file->is_renamed)
2882                                 write_name_quoted(file->name, options->file,
2883                                                   options->line_termination);
2884                         else {
2885                                 fputs(file->print_name, options->file);
2886                                 putc(options->line_termination, options->file);
2887                         }
2888                 } else {
2889                         if (file->is_renamed) {
2890                                 putc('\0', options->file);
2891                                 write_name_quoted(file->from_name, options->file, '\0');
2892                         }
2893                         write_name_quoted(file->name, options->file, '\0');
2894                 }
2895         }
2896 }
2897
2898 struct dirstat_file {
2899         const char *name;
2900         unsigned long changed;
2901 };
2902
2903 struct dirstat_dir {
2904         struct dirstat_file *files;
2905         int alloc, nr, permille, cumulative;
2906 };
2907
2908 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2909                 unsigned long changed, const char *base, int baselen)
2910 {
2911         unsigned long sum_changes = 0;
2912         unsigned int sources = 0;
2913         const char *line_prefix = diff_line_prefix(opt);
2914
2915         while (dir->nr) {
2916                 struct dirstat_file *f = dir->files;
2917                 int namelen = strlen(f->name);
2918                 unsigned long changes;
2919                 char *slash;
2920
2921                 if (namelen < baselen)
2922                         break;
2923                 if (memcmp(f->name, base, baselen))
2924                         break;
2925                 slash = strchr(f->name + baselen, '/');
2926                 if (slash) {
2927                         int newbaselen = slash + 1 - f->name;
2928                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2929                         sources++;
2930                 } else {
2931                         changes = f->changed;
2932                         dir->files++;
2933                         dir->nr--;
2934                         sources += 2;
2935                 }
2936                 sum_changes += changes;
2937         }
2938
2939         /*
2940          * We don't report dirstat's for
2941          *  - the top level
2942          *  - or cases where everything came from a single directory
2943          *    under this directory (sources == 1).
2944          */
2945         if (baselen && sources != 1) {
2946                 if (sum_changes) {
2947                         int permille = sum_changes * 1000 / changed;
2948                         if (permille >= dir->permille) {
2949                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2950                                         permille / 10, permille % 10, baselen, base);
2951                                 if (!dir->cumulative)
2952                                         return 0;
2953                         }
2954                 }
2955         }
2956         return sum_changes;
2957 }
2958
2959 static int dirstat_compare(const void *_a, const void *_b)
2960 {
2961         const struct dirstat_file *a = _a;
2962         const struct dirstat_file *b = _b;
2963         return strcmp(a->name, b->name);
2964 }
2965
2966 static void show_dirstat(struct diff_options *options)
2967 {
2968         int i;
2969         unsigned long changed;
2970         struct dirstat_dir dir;
2971         struct diff_queue_struct *q = &diff_queued_diff;
2972
2973         dir.files = NULL;
2974         dir.alloc = 0;
2975         dir.nr = 0;
2976         dir.permille = options->dirstat_permille;
2977         dir.cumulative = options->flags.dirstat_cumulative;
2978
2979         changed = 0;
2980         for (i = 0; i < q->nr; i++) {
2981                 struct diff_filepair *p = q->queue[i];
2982                 const char *name;
2983                 unsigned long copied, added, damage;
2984
2985                 name = p->two->path ? p->two->path : p->one->path;
2986
2987                 if (p->one->oid_valid && p->two->oid_valid &&
2988                     oideq(&p->one->oid, &p->two->oid)) {
2989                         /*
2990                          * The SHA1 has not changed, so pre-/post-content is
2991                          * identical. We can therefore skip looking at the
2992                          * file contents altogether.
2993                          */
2994                         damage = 0;
2995                         goto found_damage;
2996                 }
2997
2998                 if (options->flags.dirstat_by_file) {
2999                         /*
3000                          * In --dirstat-by-file mode, we don't really need to
3001                          * look at the actual file contents at all.
3002                          * The fact that the SHA1 changed is enough for us to
3003                          * add this file to the list of results
3004                          * (with each file contributing equal damage).
3005                          */
3006                         damage = 1;
3007                         goto found_damage;
3008                 }
3009
3010                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3011                         diff_populate_filespec(options->repo, p->one, 0);
3012                         diff_populate_filespec(options->repo, p->two, 0);
3013                         diffcore_count_changes(options->repo,
3014                                                p->one, p->two, NULL, NULL,
3015                                                &copied, &added);
3016                         diff_free_filespec_data(p->one);
3017                         diff_free_filespec_data(p->two);
3018                 } else if (DIFF_FILE_VALID(p->one)) {
3019                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3020                         copied = added = 0;
3021                         diff_free_filespec_data(p->one);
3022                 } else if (DIFF_FILE_VALID(p->two)) {
3023                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3024                         copied = 0;
3025                         added = p->two->size;
3026                         diff_free_filespec_data(p->two);
3027                 } else
3028                         continue;
3029
3030                 /*
3031                  * Original minus copied is the removed material,
3032                  * added is the new material.  They are both damages
3033                  * made to the preimage.
3034                  * If the resulting damage is zero, we know that
3035                  * diffcore_count_changes() considers the two entries to
3036                  * be identical, but since the oid changed, we
3037                  * know that there must have been _some_ kind of change,
3038                  * so we force all entries to have damage > 0.
3039                  */
3040                 damage = (p->one->size - copied) + added;
3041                 if (!damage)
3042                         damage = 1;
3043
3044 found_damage:
3045                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3046                 dir.files[dir.nr].name = name;
3047                 dir.files[dir.nr].changed = damage;
3048                 changed += damage;
3049                 dir.nr++;
3050         }
3051
3052         /* This can happen even with many files, if everything was renames */
3053         if (!changed)
3054                 return;
3055
3056         /* Show all directories with more than x% of the changes */
3057         QSORT(dir.files, dir.nr, dirstat_compare);
3058         gather_dirstat(options, &dir, changed, "", 0);
3059 }
3060
3061 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3062 {
3063         int i;
3064         unsigned long changed;
3065         struct dirstat_dir dir;
3066
3067         if (data->nr == 0)
3068                 return;
3069
3070         dir.files = NULL;
3071         dir.alloc = 0;
3072         dir.nr = 0;
3073         dir.permille = options->dirstat_permille;
3074         dir.cumulative = options->flags.dirstat_cumulative;
3075
3076         changed = 0;
3077         for (i = 0; i < data->nr; i++) {
3078                 struct diffstat_file *file = data->files[i];
3079                 unsigned long damage = file->added + file->deleted;
3080                 if (file->is_binary)
3081                         /*
3082                          * binary files counts bytes, not lines. Must find some
3083                          * way to normalize binary bytes vs. textual lines.
3084                          * The following heuristic assumes that there are 64
3085                          * bytes per "line".
3086                          * This is stupid and ugly, but very cheap...
3087                          */
3088                         damage = DIV_ROUND_UP(damage, 64);
3089                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3090                 dir.files[dir.nr].name = file->name;
3091                 dir.files[dir.nr].changed = damage;
3092                 changed += damage;
3093                 dir.nr++;
3094         }
3095
3096         /* This can happen even with many files, if everything was renames */
3097         if (!changed)
3098                 return;
3099
3100         /* Show all directories with more than x% of the changes */
3101         QSORT(dir.files, dir.nr, dirstat_compare);
3102         gather_dirstat(options, &dir, changed, "", 0);
3103 }
3104
3105 static void free_diffstat_info(struct diffstat_t *diffstat)
3106 {
3107         int i;
3108         for (i = 0; i < diffstat->nr; i++) {
3109                 struct diffstat_file *f = diffstat->files[i];
3110                 free(f->print_name);
3111                 free(f->name);
3112                 free(f->from_name);
3113                 free(f);
3114         }
3115         free(diffstat->files);
3116 }
3117
3118 struct checkdiff_t {
3119         const char *filename;
3120         int lineno;
3121         int conflict_marker_size;
3122         struct diff_options *o;
3123         unsigned ws_rule;
3124         unsigned status;
3125 };
3126
3127 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3128 {
3129         char firstchar;
3130         int cnt;
3131
3132         if (len < marker_size + 1)
3133                 return 0;
3134         firstchar = line[0];
3135         switch (firstchar) {
3136         case '=': case '>': case '<': case '|':
3137                 break;
3138         default:
3139                 return 0;
3140         }
3141         for (cnt = 1; cnt < marker_size; cnt++)
3142                 if (line[cnt] != firstchar)
3143                         return 0;
3144         /* line[1] thru line[marker_size-1] are same as firstchar */
3145         if (len < marker_size + 1 || !isspace(line[marker_size]))
3146                 return 0;
3147         return 1;
3148 }
3149
3150 static void checkdiff_consume_hunk(void *priv,
3151                                    long ob, long on, long nb, long nn,
3152                                    const char *func, long funclen)
3153
3154 {
3155         struct checkdiff_t *data = priv;
3156         data->lineno = nb - 1;
3157 }
3158
3159 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3160 {
3161         struct checkdiff_t *data = priv;
3162         int marker_size = data->conflict_marker_size;
3163         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3164         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3165         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3166         char *err;
3167         const char *line_prefix;
3168
3169         assert(data->o);
3170         line_prefix = diff_line_prefix(data->o);
3171
3172         if (line[0] == '+') {
3173                 unsigned bad;
3174                 data->lineno++;
3175                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3176                         data->status |= 1;
3177                         fprintf(data->o->file,
3178                                 "%s%s:%d: leftover conflict marker\n",
3179                                 line_prefix, data->filename, data->lineno);
3180                 }
3181                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3182                 if (!bad)
3183                         return;
3184                 data->status |= bad;
3185                 err = whitespace_error_string(bad);
3186                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3187                         line_prefix, data->filename, data->lineno, err);
3188                 free(err);
3189                 emit_line(data->o, set, reset, line, 1);
3190                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3191                               data->o->file, set, reset, ws);
3192         } else if (line[0] == ' ') {
3193                 data->lineno++;
3194         }
3195 }
3196
3197 static unsigned char *deflate_it(char *data,
3198                                  unsigned long size,
3199                                  unsigned long *result_size)
3200 {
3201         int bound;
3202         unsigned char *deflated;
3203         git_zstream stream;
3204
3205         git_deflate_init(&stream, zlib_compression_level);
3206         bound = git_deflate_bound(&stream, size);
3207         deflated = xmalloc(bound);
3208         stream.next_out = deflated;
3209         stream.avail_out = bound;
3210
3211         stream.next_in = (unsigned char *)data;
3212         stream.avail_in = size;
3213         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3214                 ; /* nothing */
3215         git_deflate_end(&stream);
3216         *result_size = stream.total_out;
3217         return deflated;
3218 }
3219
3220 static void emit_binary_diff_body(struct diff_options *o,
3221                                   mmfile_t *one, mmfile_t *two)
3222 {
3223         void *cp;
3224         void *delta;
3225         void *deflated;
3226         void *data;
3227         unsigned long orig_size;
3228         unsigned long delta_size;
3229         unsigned long deflate_size;
3230         unsigned long data_size;
3231
3232         /* We could do deflated delta, or we could do just deflated two,
3233          * whichever is smaller.
3234          */
3235         delta = NULL;
3236         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3237         if (one->size && two->size) {
3238                 delta = diff_delta(one->ptr, one->size,
3239                                    two->ptr, two->size,
3240                                    &delta_size, deflate_size);
3241                 if (delta) {
3242                         void *to_free = delta;
3243                         orig_size = delta_size;
3244                         delta = deflate_it(delta, delta_size, &delta_size);
3245                         free(to_free);
3246                 }
3247         }
3248
3249         if (delta && delta_size < deflate_size) {
3250                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3251                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3252                                  s, strlen(s), 0);
3253                 free(s);
3254                 free(deflated);
3255                 data = delta;
3256                 data_size = delta_size;
3257         } else {
3258                 char *s = xstrfmt("%lu", two->size);
3259                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3260                                  s, strlen(s), 0);
3261                 free(s);
3262                 free(delta);
3263                 data = deflated;
3264                 data_size = deflate_size;
3265         }
3266
3267         /* emit data encoded in base85 */
3268         cp = data;
3269         while (data_size) {
3270                 int len;
3271                 int bytes = (52 < data_size) ? 52 : data_size;
3272                 char line[71];
3273                 data_size -= bytes;
3274                 if (bytes <= 26)
3275                         line[0] = bytes + 'A' - 1;
3276                 else
3277                         line[0] = bytes - 26 + 'a' - 1;
3278                 encode_85(line + 1, cp, bytes);
3279                 cp = (char *) cp + bytes;
3280
3281                 len = strlen(line);
3282                 line[len++] = '\n';
3283                 line[len] = '\0';
3284
3285                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3286                                  line, len, 0);
3287         }
3288         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3289         free(data);
3290 }
3291
3292 static void emit_binary_diff(struct diff_options *o,
3293                              mmfile_t *one, mmfile_t *two)
3294 {
3295         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3296         emit_binary_diff_body(o, one, two);
3297         emit_binary_diff_body(o, two, one);
3298 }
3299
3300 int diff_filespec_is_binary(struct repository *r,
3301                             struct diff_filespec *one)
3302 {
3303         if (one->is_binary == -1) {
3304                 diff_filespec_load_driver(one, r->index);
3305                 if (one->driver->binary != -1)
3306                         one->is_binary = one->driver->binary;
3307                 else {
3308                         if (!one->data && DIFF_FILE_VALID(one))
3309                                 diff_populate_filespec(r, one, CHECK_BINARY);
3310                         if (one->is_binary == -1 && one->data)
3311                                 one->is_binary = buffer_is_binary(one->data,
3312                                                 one->size);
3313                         if (one->is_binary == -1)
3314                                 one->is_binary = 0;
3315                 }
3316         }
3317         return one->is_binary;
3318 }
3319
3320 static const struct userdiff_funcname *
3321 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3322 {
3323         diff_filespec_load_driver(one, o->repo->index);
3324         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3325 }
3326
3327 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3328 {
3329         if (!options->a_prefix)
3330                 options->a_prefix = a;
3331         if (!options->b_prefix)
3332                 options->b_prefix = b;
3333 }
3334
3335 struct userdiff_driver *get_textconv(struct index_state *istate,
3336                                      struct diff_filespec *one)
3337 {
3338         if (!DIFF_FILE_VALID(one))
3339                 return NULL;
3340
3341         diff_filespec_load_driver(one, istate);
3342         return userdiff_get_textconv(one->driver);
3343 }
3344
3345 static void builtin_diff(const char *name_a,
3346                          const char *name_b,
3347                          struct diff_filespec *one,
3348                          struct diff_filespec *two,
3349                          const char *xfrm_msg,
3350                          int must_show_header,
3351                          struct diff_options *o,
3352                          int complete_rewrite)
3353 {
3354         mmfile_t mf1, mf2;
3355         const char *lbl[2];
3356         char *a_one, *b_two;
3357         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3358         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3359         const char *a_prefix, *b_prefix;
3360         struct userdiff_driver *textconv_one = NULL;
3361         struct userdiff_driver *textconv_two = NULL;
3362         struct strbuf header = STRBUF_INIT;
3363         const char *line_prefix = diff_line_prefix(o);
3364
3365         diff_set_mnemonic_prefix(o, "a/", "b/");
3366         if (o->flags.reverse_diff) {
3367                 a_prefix = o->b_prefix;
3368                 b_prefix = o->a_prefix;
3369         } else {
3370                 a_prefix = o->a_prefix;
3371                 b_prefix = o->b_prefix;
3372         }
3373
3374         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3375             (!one->mode || S_ISGITLINK(one->mode)) &&
3376             (!two->mode || S_ISGITLINK(two->mode))) {
3377                 show_submodule_summary(o, one->path ? one->path : two->path,
3378                                 &one->oid, &two->oid,
3379                                 two->dirty_submodule);
3380                 return;
3381         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3382                    (!one->mode || S_ISGITLINK(one->mode)) &&
3383                    (!two->mode || S_ISGITLINK(two->mode))) {
3384                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3385                                 &one->oid, &two->oid,
3386                                 two->dirty_submodule);
3387                 return;
3388         }
3389
3390         if (o->flags.allow_textconv) {
3391                 textconv_one = get_textconv(o->repo->index, one);
3392                 textconv_two = get_textconv(o->repo->index, two);
3393         }
3394
3395         /* Never use a non-valid filename anywhere if at all possible */
3396         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3397         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3398
3399         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3400         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3401         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3402         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3403         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3404         if (lbl[0][0] == '/') {
3405                 /* /dev/null */
3406                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3407                 if (xfrm_msg)
3408                         strbuf_addstr(&header, xfrm_msg);
3409                 must_show_header = 1;
3410         }
3411         else if (lbl[1][0] == '/') {
3412                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3413                 if (xfrm_msg)
3414                         strbuf_addstr(&header, xfrm_msg);
3415                 must_show_header = 1;
3416         }
3417         else {
3418                 if (one->mode != two->mode) {
3419                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3420                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3421                         must_show_header = 1;
3422                 }
3423                 if (xfrm_msg)
3424                         strbuf_addstr(&header, xfrm_msg);
3425
3426                 /*
3427                  * we do not run diff between different kind
3428                  * of objects.
3429                  */
3430                 if ((one->mode ^ two->mode) & S_IFMT)
3431                         goto free_ab_and_return;
3432                 if (complete_rewrite &&
3433                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3434                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3435                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3436                                          header.buf, header.len, 0);
3437                         strbuf_reset(&header);
3438                         emit_rewrite_diff(name_a, name_b, one, two,
3439                                           textconv_one, textconv_two, o);
3440                         o->found_changes = 1;
3441                         goto free_ab_and_return;
3442                 }
3443         }
3444
3445         if (o->irreversible_delete && lbl[1][0] == '/') {
3446                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3447                                  header.len, 0);
3448                 strbuf_reset(&header);
3449                 goto free_ab_and_return;
3450         } else if (!o->flags.text &&
3451                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3452                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3453                 struct strbuf sb = STRBUF_INIT;
3454                 if (!one->data && !two->data &&
3455                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3456                     !o->flags.binary) {
3457                         if (oideq(&one->oid, &two->oid)) {
3458                                 if (must_show_header)
3459                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3460                                                          header.buf, header.len,
3461                                                          0);
3462                                 goto free_ab_and_return;
3463                         }
3464                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3465                                          header.buf, header.len, 0);
3466                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3467                                     diff_line_prefix(o), lbl[0], lbl[1]);
3468                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3469                                          sb.buf, sb.len, 0);
3470                         strbuf_release(&sb);
3471                         goto free_ab_and_return;
3472                 }
3473                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3474                     fill_mmfile(o->repo, &mf2, two) < 0)
3475                         die("unable to read files to diff");
3476                 /* Quite common confusing case */
3477                 if (mf1.size == mf2.size &&
3478                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3479                         if (must_show_header)
3480                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3481                                                  header.buf, header.len, 0);
3482                         goto free_ab_and_return;
3483                 }
3484                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3485                 strbuf_reset(&header);
3486                 if (o->flags.binary)
3487                         emit_binary_diff(o, &mf1, &mf2);
3488                 else {
3489                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3490                                     diff_line_prefix(o), lbl[0], lbl[1]);
3491                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3492                                          sb.buf, sb.len, 0);
3493                         strbuf_release(&sb);
3494                 }
3495                 o->found_changes = 1;
3496         } else {
3497                 /* Crazy xdl interfaces.. */
3498                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3499                 const char *v;
3500                 xpparam_t xpp;
3501                 xdemitconf_t xecfg;
3502                 struct emit_callback ecbdata;
3503                 const struct userdiff_funcname *pe;
3504
3505                 if (must_show_header) {
3506                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3507                                          header.buf, header.len, 0);
3508                         strbuf_reset(&header);
3509                 }
3510
3511                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3512                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3513
3514                 pe = diff_funcname_pattern(o, one);
3515                 if (!pe)
3516                         pe = diff_funcname_pattern(o, two);
3517
3518                 memset(&xpp, 0, sizeof(xpp));
3519                 memset(&xecfg, 0, sizeof(xecfg));
3520                 memset(&ecbdata, 0, sizeof(ecbdata));
3521                 if (o->flags.suppress_diff_headers)
3522                         lbl[0] = NULL;
3523                 ecbdata.label_path = lbl;
3524                 ecbdata.color_diff = want_color(o->use_color);
3525                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3526                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3527                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3528                 ecbdata.opt = o;
3529                 if (header.len && !o->flags.suppress_diff_headers)
3530                         ecbdata.header = &header;
3531                 xpp.flags = o->xdl_opts;
3532                 xpp.anchors = o->anchors;
3533                 xpp.anchors_nr = o->anchors_nr;
3534                 xecfg.ctxlen = o->context;
3535                 xecfg.interhunkctxlen = o->interhunkcontext;
3536                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3537                 if (o->flags.funccontext)
3538                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3539                 if (pe)
3540                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3541                 if (!diffopts)
3542                         ;
3543                 else if (skip_prefix(diffopts, "--unified=", &v))
3544                         xecfg.ctxlen = strtoul(v, NULL, 10);
3545                 else if (skip_prefix(diffopts, "-u", &v))
3546                         xecfg.ctxlen = strtoul(v, NULL, 10);
3547                 if (o->word_diff)
3548                         init_diff_words_data(&ecbdata, o, one, two);
3549                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3550                                   &ecbdata, &xpp, &xecfg))
3551                         die("unable to generate diff for %s", one->path);
3552                 if (o->word_diff)
3553                         free_diff_words_data(&ecbdata);
3554                 if (textconv_one)
3555                         free(mf1.ptr);
3556                 if (textconv_two)
3557                         free(mf2.ptr);
3558                 xdiff_clear_find_func(&xecfg);
3559         }
3560
3561  free_ab_and_return:
3562         strbuf_release(&header);
3563         diff_free_filespec_data(one);
3564         diff_free_filespec_data(two);
3565         free(a_one);
3566         free(b_two);
3567         return;
3568 }
3569
3570 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3571 {
3572         if (!is_renamed) {
3573                 if (p->status == DIFF_STATUS_ADDED) {
3574                         if (S_ISLNK(p->two->mode))
3575                                 return "new +l";
3576                         else if ((p->two->mode & 0777) == 0755)
3577                                 return "new +x";
3578                         else
3579                                 return "new";
3580                 } else if (p->status == DIFF_STATUS_DELETED)
3581                         return "gone";
3582         }
3583         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3584                 return "mode -l";
3585         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3586                 return "mode +l";
3587         else if ((p->one->mode & 0777) == 0644 &&
3588                  (p->two->mode & 0777) == 0755)
3589                 return "mode +x";
3590         else if ((p->one->mode & 0777) == 0755 &&
3591                  (p->two->mode & 0777) == 0644)
3592                 return "mode -x";
3593         return NULL;
3594 }
3595
3596 static void builtin_diffstat(const char *name_a, const char *name_b,
3597                              struct diff_filespec *one,
3598                              struct diff_filespec *two,
3599                              struct diffstat_t *diffstat,
3600                              struct diff_options *o,
3601                              struct diff_filepair *p)
3602 {
3603         mmfile_t mf1, mf2;
3604         struct diffstat_file *data;
3605         int same_contents;
3606         int complete_rewrite = 0;
3607
3608         if (!DIFF_PAIR_UNMERGED(p)) {
3609                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3610                         complete_rewrite = 1;
3611         }
3612
3613         data = diffstat_add(diffstat, name_a, name_b);
3614         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3615         if (o->flags.stat_with_summary)
3616                 data->comments = get_compact_summary(p, data->is_renamed);
3617
3618         if (!one || !two) {
3619                 data->is_unmerged = 1;
3620                 return;
3621         }
3622
3623         same_contents = oideq(&one->oid, &two->oid);
3624
3625         if (diff_filespec_is_binary(o->repo, one) ||
3626             diff_filespec_is_binary(o->repo, two)) {
3627                 data->is_binary = 1;
3628                 if (same_contents) {
3629                         data->added = 0;
3630                         data->deleted = 0;
3631                 } else {
3632                         data->added = diff_filespec_size(o->repo, two);
3633                         data->deleted = diff_filespec_size(o->repo, one);
3634                 }
3635         }
3636
3637         else if (complete_rewrite) {
3638                 diff_populate_filespec(o->repo, one, 0);
3639                 diff_populate_filespec(o->repo, two, 0);
3640                 data->deleted = count_lines(one->data, one->size);
3641                 data->added = count_lines(two->data, two->size);
3642         }
3643
3644         else if (!same_contents) {
3645                 /* Crazy xdl interfaces.. */
3646                 xpparam_t xpp;
3647                 xdemitconf_t xecfg;
3648
3649                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3650                     fill_mmfile(o->repo, &mf2, two) < 0)
3651                         die("unable to read files to diff");
3652
3653                 memset(&xpp, 0, sizeof(xpp));
3654                 memset(&xecfg, 0, sizeof(xecfg));
3655                 xpp.flags = o->xdl_opts;
3656                 xpp.anchors = o->anchors;
3657                 xpp.anchors_nr = o->anchors_nr;
3658                 xecfg.ctxlen = o->context;
3659                 xecfg.interhunkctxlen = o->interhunkcontext;
3660                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3661                                   diffstat_consume, diffstat, &xpp, &xecfg))
3662                         die("unable to generate diffstat for %s", one->path);
3663         }
3664
3665         diff_free_filespec_data(one);
3666         diff_free_filespec_data(two);
3667 }
3668
3669 static void builtin_checkdiff(const char *name_a, const char *name_b,
3670                               const char *attr_path,
3671                               struct diff_filespec *one,
3672                               struct diff_filespec *two,
3673                               struct diff_options *o)
3674 {
3675         mmfile_t mf1, mf2;
3676         struct checkdiff_t data;
3677
3678         if (!two)
3679                 return;
3680
3681         memset(&data, 0, sizeof(data));
3682         data.filename = name_b ? name_b : name_a;
3683         data.lineno = 0;
3684         data.o = o;
3685         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3686         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3687
3688         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3689             fill_mmfile(o->repo, &mf2, two) < 0)
3690                 die("unable to read files to diff");
3691
3692         /*
3693          * All the other codepaths check both sides, but not checking
3694          * the "old" side here is deliberate.  We are checking the newly
3695          * introduced changes, and as long as the "new" side is text, we
3696          * can and should check what it introduces.
3697          */
3698         if (diff_filespec_is_binary(o->repo, two))
3699                 goto free_and_return;
3700         else {
3701                 /* Crazy xdl interfaces.. */
3702                 xpparam_t xpp;
3703                 xdemitconf_t xecfg;
3704
3705                 memset(&xpp, 0, sizeof(xpp));
3706                 memset(&xecfg, 0, sizeof(xecfg));
3707                 xecfg.ctxlen = 1; /* at least one context line */
3708                 xpp.flags = 0;
3709                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3710                                   checkdiff_consume, &data,
3711                                   &xpp, &xecfg))
3712                         die("unable to generate checkdiff for %s", one->path);
3713
3714                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3715                         struct emit_callback ecbdata;
3716                         int blank_at_eof;
3717
3718                         ecbdata.ws_rule = data.ws_rule;
3719                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3720                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3721
3722                         if (blank_at_eof) {
3723                                 static char *err;
3724                                 if (!err)
3725                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3726                                 fprintf(o->file, "%s:%d: %s.\n",
3727                                         data.filename, blank_at_eof, err);
3728                                 data.status = 1; /* report errors */
3729                         }
3730                 }
3731         }
3732  free_and_return:
3733         diff_free_filespec_data(one);
3734         diff_free_filespec_data(two);
3735         if (data.status)
3736                 o->flags.check_failed = 1;
3737 }
3738
3739 struct diff_filespec *alloc_filespec(const char *path)
3740 {
3741         struct diff_filespec *spec;
3742
3743         FLEXPTR_ALLOC_STR(spec, path, path);
3744         spec->count = 1;
3745         spec->is_binary = -1;
3746         return spec;
3747 }
3748
3749 void free_filespec(struct diff_filespec *spec)
3750 {
3751         if (!--spec->count) {
3752                 diff_free_filespec_data(spec);
3753                 free(spec);
3754         }
3755 }
3756
3757 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3758                    int oid_valid, unsigned short mode)
3759 {
3760         if (mode) {
3761                 spec->mode = canon_mode(mode);
3762                 oidcpy(&spec->oid, oid);
3763                 spec->oid_valid = oid_valid;
3764         }
3765 }
3766
3767 /*
3768  * Given a name and sha1 pair, if the index tells us the file in
3769  * the work tree has that object contents, return true, so that
3770  * prepare_temp_file() does not have to inflate and extract.
3771  */
3772 static int reuse_worktree_file(struct index_state *istate,
3773                                const char *name,
3774                                const struct object_id *oid,
3775                                int want_file)
3776 {
3777         const struct cache_entry *ce;
3778         struct stat st;
3779         int pos, len;
3780
3781         /*
3782          * We do not read the cache ourselves here, because the
3783          * benchmark with my previous version that always reads cache
3784          * shows that it makes things worse for diff-tree comparing
3785          * two linux-2.6 kernel trees in an already checked out work
3786          * tree.  This is because most diff-tree comparisons deal with
3787          * only a small number of files, while reading the cache is
3788          * expensive for a large project, and its cost outweighs the
3789          * savings we get by not inflating the object to a temporary
3790          * file.  Practically, this code only helps when we are used
3791          * by diff-cache --cached, which does read the cache before
3792          * calling us.
3793          */
3794         if (!istate->cache)
3795                 return 0;
3796
3797         /* We want to avoid the working directory if our caller
3798          * doesn't need the data in a normal file, this system
3799          * is rather slow with its stat/open/mmap/close syscalls,
3800          * and the object is contained in a pack file.  The pack
3801          * is probably already open and will be faster to obtain
3802          * the data through than the working directory.  Loose
3803          * objects however would tend to be slower as they need
3804          * to be individually opened and inflated.
3805          */
3806         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3807                 return 0;
3808
3809         /*
3810          * Similarly, if we'd have to convert the file contents anyway, that
3811          * makes the optimization not worthwhile.
3812          */
3813         if (!want_file && would_convert_to_git(istate, name))
3814                 return 0;
3815
3816         len = strlen(name);
3817         pos = index_name_pos(istate, name, len);
3818         if (pos < 0)
3819                 return 0;
3820         ce = istate->cache[pos];
3821
3822         /*
3823          * This is not the sha1 we are looking for, or
3824          * unreusable because it is not a regular file.
3825          */
3826         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3827                 return 0;
3828
3829         /*
3830          * If ce is marked as "assume unchanged", there is no
3831          * guarantee that work tree matches what we are looking for.
3832          */
3833         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3834                 return 0;
3835
3836         /*
3837          * If ce matches the file in the work tree, we can reuse it.
3838          */
3839         if (ce_uptodate(ce) ||
3840             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3841                 return 1;
3842
3843         return 0;
3844 }
3845
3846 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3847 {
3848         struct strbuf buf = STRBUF_INIT;
3849         char *dirty = "";
3850
3851         /* Are we looking at the work tree? */
3852         if (s->dirty_submodule)
3853                 dirty = "-dirty";
3854
3855         strbuf_addf(&buf, "Subproject commit %s%s\n",
3856                     oid_to_hex(&s->oid), dirty);
3857         s->size = buf.len;
3858         if (size_only) {
3859                 s->data = NULL;
3860                 strbuf_release(&buf);
3861         } else {
3862                 s->data = strbuf_detach(&buf, NULL);
3863                 s->should_free = 1;
3864         }
3865         return 0;
3866 }
3867
3868 /*
3869  * While doing rename detection and pickaxe operation, we may need to
3870  * grab the data for the blob (or file) for our own in-core comparison.
3871  * diff_filespec has data and size fields for this purpose.
3872  */
3873 int diff_populate_filespec(struct repository *r,
3874                            struct diff_filespec *s,
3875                            unsigned int flags)
3876 {
3877         int size_only = flags & CHECK_SIZE_ONLY;
3878         int err = 0;
3879         int conv_flags = global_conv_flags_eol;
3880         /*
3881          * demote FAIL to WARN to allow inspecting the situation
3882          * instead of refusing.
3883          */
3884         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3885                 conv_flags = CONV_EOL_RNDTRP_WARN;
3886
3887         if (!DIFF_FILE_VALID(s))
3888                 die("internal error: asking to populate invalid file.");
3889         if (S_ISDIR(s->mode))
3890                 return -1;
3891
3892         if (s->data)
3893                 return 0;
3894
3895         if (size_only && 0 < s->size)
3896                 return 0;
3897
3898         if (S_ISGITLINK(s->mode))
3899                 return diff_populate_gitlink(s, size_only);
3900
3901         if (!s->oid_valid ||
3902             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3903                 struct strbuf buf = STRBUF_INIT;
3904                 struct stat st;
3905                 int fd;
3906
3907                 if (lstat(s->path, &st) < 0) {
3908                 err_empty:
3909                         err = -1;
3910                 empty:
3911                         s->data = (char *)"";
3912                         s->size = 0;
3913                         return err;
3914                 }
3915                 s->size = xsize_t(st.st_size);
3916                 if (!s->size)
3917                         goto empty;
3918                 if (S_ISLNK(st.st_mode)) {
3919                         struct strbuf sb = STRBUF_INIT;
3920
3921                         if (strbuf_readlink(&sb, s->path, s->size))
3922                                 goto err_empty;
3923                         s->size = sb.len;
3924                         s->data = strbuf_detach(&sb, NULL);
3925                         s->should_free = 1;
3926                         return 0;
3927                 }
3928
3929                 /*
3930                  * Even if the caller would be happy with getting
3931                  * only the size, we cannot return early at this
3932                  * point if the path requires us to run the content
3933                  * conversion.
3934                  */
3935                 if (size_only && !would_convert_to_git(r->index, s->path))
3936                         return 0;
3937
3938                 /*
3939                  * Note: this check uses xsize_t(st.st_size) that may
3940                  * not be the true size of the blob after it goes
3941                  * through convert_to_git().  This may not strictly be
3942                  * correct, but the whole point of big_file_threshold
3943                  * and is_binary check being that we want to avoid
3944                  * opening the file and inspecting the contents, this
3945                  * is probably fine.
3946                  */
3947                 if ((flags & CHECK_BINARY) &&
3948                     s->size > big_file_threshold && s->is_binary == -1) {
3949                         s->is_binary = 1;
3950                         return 0;
3951                 }
3952                 fd = open(s->path, O_RDONLY);
3953                 if (fd < 0)
3954                         goto err_empty;
3955                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3956                 close(fd);
3957                 s->should_munmap = 1;
3958
3959                 /*
3960                  * Convert from working tree format to canonical git format
3961                  */
3962                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3963                         size_t size = 0;
3964                         munmap(s->data, s->size);
3965                         s->should_munmap = 0;
3966                         s->data = strbuf_detach(&buf, &size);
3967                         s->size = size;
3968                         s->should_free = 1;
3969                 }
3970         }
3971         else {
3972                 enum object_type type;
3973                 if (size_only || (flags & CHECK_BINARY)) {
3974                         type = oid_object_info(r, &s->oid, &s->size);
3975                         if (type < 0)
3976                                 die("unable to read %s",
3977                                     oid_to_hex(&s->oid));
3978                         if (size_only)
3979                                 return 0;
3980                         if (s->size > big_file_threshold && s->is_binary == -1) {
3981                                 s->is_binary = 1;
3982                                 return 0;
3983                         }
3984                 }
3985                 s->data = read_object_file(&s->oid, &type, &s->size);
3986                 if (!s->data)
3987                         die("unable to read %s", oid_to_hex(&s->oid));
3988                 s->should_free = 1;
3989         }
3990         return 0;
3991 }
3992
3993 void diff_free_filespec_blob(struct diff_filespec *s)
3994 {
3995         if (s->should_free)
3996                 free(s->data);
3997         else if (s->should_munmap)
3998                 munmap(s->data, s->size);
3999
4000         if (s->should_free || s->should_munmap) {
4001                 s->should_free = s->should_munmap = 0;
4002                 s->data = NULL;
4003         }
4004 }
4005
4006 void diff_free_filespec_data(struct diff_filespec *s)
4007 {
4008         diff_free_filespec_blob(s);
4009         FREE_AND_NULL(s->cnt_data);
4010 }
4011
4012 static void prep_temp_blob(struct index_state *istate,
4013                            const char *path, struct diff_tempfile *temp,
4014                            void *blob,
4015                            unsigned long size,
4016                            const struct object_id *oid,
4017                            int mode)
4018 {
4019         struct strbuf buf = STRBUF_INIT;
4020         struct strbuf tempfile = STRBUF_INIT;
4021         char *path_dup = xstrdup(path);
4022         const char *base = basename(path_dup);
4023
4024         /* Generate "XXXXXX_basename.ext" */
4025         strbuf_addstr(&tempfile, "XXXXXX_");
4026         strbuf_addstr(&tempfile, base);
4027
4028         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4029         if (!temp->tempfile)
4030                 die_errno("unable to create temp-file");
4031         if (convert_to_working_tree(istate, path,
4032                         (const char *)blob, (size_t)size, &buf)) {
4033                 blob = buf.buf;
4034                 size = buf.len;
4035         }
4036         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4037             close_tempfile_gently(temp->tempfile))
4038                 die_errno("unable to write temp-file");
4039         temp->name = get_tempfile_path(temp->tempfile);
4040         oid_to_hex_r(temp->hex, oid);
4041         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4042         strbuf_release(&buf);
4043         strbuf_release(&tempfile);
4044         free(path_dup);
4045 }
4046
4047 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4048                                                const char *name,
4049                                                struct diff_filespec *one)
4050 {
4051         struct diff_tempfile *temp = claim_diff_tempfile();
4052
4053         if (!DIFF_FILE_VALID(one)) {
4054         not_a_valid_file:
4055                 /* A '-' entry produces this for file-2, and
4056                  * a '+' entry produces this for file-1.
4057                  */
4058                 temp->name = "/dev/null";
4059                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4060                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4061                 return temp;
4062         }
4063
4064         if (!S_ISGITLINK(one->mode) &&
4065             (!one->oid_valid ||
4066              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4067                 struct stat st;
4068                 if (lstat(name, &st) < 0) {
4069                         if (errno == ENOENT)
4070                                 goto not_a_valid_file;
4071                         die_errno("stat(%s)", name);
4072                 }
4073                 if (S_ISLNK(st.st_mode)) {
4074                         struct strbuf sb = STRBUF_INIT;
4075                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4076                                 die_errno("readlink(%s)", name);
4077                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4078                                        (one->oid_valid ?
4079                                         &one->oid : &null_oid),
4080                                        (one->oid_valid ?
4081                                         one->mode : S_IFLNK));
4082                         strbuf_release(&sb);
4083                 }
4084                 else {
4085                         /* we can borrow from the file in the work tree */
4086                         temp->name = name;
4087                         if (!one->oid_valid)
4088                                 oid_to_hex_r(temp->hex, &null_oid);
4089                         else
4090                                 oid_to_hex_r(temp->hex, &one->oid);
4091                         /* Even though we may sometimes borrow the
4092                          * contents from the work tree, we always want
4093                          * one->mode.  mode is trustworthy even when
4094                          * !(one->oid_valid), as long as
4095                          * DIFF_FILE_VALID(one).
4096                          */
4097                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4098                 }
4099                 return temp;
4100         }
4101         else {
4102                 if (diff_populate_filespec(r, one, 0))
4103                         die("cannot read data blob for %s", one->path);
4104                 prep_temp_blob(r->index, name, temp,
4105                                one->data, one->size,
4106                                &one->oid, one->mode);
4107         }
4108         return temp;
4109 }
4110
4111 static void add_external_diff_name(struct repository *r,
4112                                    struct argv_array *argv,
4113                                    const char *name,
4114                                    struct diff_filespec *df)
4115 {
4116         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4117         argv_array_push(argv, temp->name);
4118         argv_array_push(argv, temp->hex);
4119         argv_array_push(argv, temp->mode);
4120 }
4121
4122 /* An external diff command takes:
4123  *
4124  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4125  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4126  *
4127  */
4128 static void run_external_diff(const char *pgm,
4129                               const char *name,
4130                               const char *other,
4131                               struct diff_filespec *one,
4132                               struct diff_filespec *two,
4133                               const char *xfrm_msg,
4134                               int complete_rewrite,
4135                               struct diff_options *o)
4136 {
4137         struct argv_array argv = ARGV_ARRAY_INIT;
4138         struct argv_array env = ARGV_ARRAY_INIT;
4139         struct diff_queue_struct *q = &diff_queued_diff;
4140
4141         argv_array_push(&argv, pgm);
4142         argv_array_push(&argv, name);
4143
4144         if (one && two) {
4145                 add_external_diff_name(o->repo, &argv, name, one);
4146                 if (!other)
4147                         add_external_diff_name(o->repo, &argv, name, two);
4148                 else {
4149                         add_external_diff_name(o->repo, &argv, other, two);
4150                         argv_array_push(&argv, other);
4151                         argv_array_push(&argv, xfrm_msg);
4152                 }
4153         }
4154
4155         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4156         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4157
4158         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4159                 die(_("external diff died, stopping at %s"), name);
4160
4161         remove_tempfile();
4162         argv_array_clear(&argv);
4163         argv_array_clear(&env);
4164 }
4165
4166 static int similarity_index(struct diff_filepair *p)
4167 {
4168         return p->score * 100 / MAX_SCORE;
4169 }
4170
4171 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4172 {
4173         if (startup_info->have_repository)
4174                 return find_unique_abbrev(oid, abbrev);
4175         else {
4176                 char *hex = oid_to_hex(oid);
4177                 if (abbrev < 0)
4178                         abbrev = FALLBACK_DEFAULT_ABBREV;
4179                 if (abbrev > the_hash_algo->hexsz)
4180                         BUG("oid abbreviation out of range: %d", abbrev);
4181                 if (abbrev)
4182                         hex[abbrev] = '\0';
4183                 return hex;
4184         }
4185 }
4186
4187 static void fill_metainfo(struct strbuf *msg,
4188                           const char *name,
4189                           const char *other,
4190                           struct diff_filespec *one,
4191                           struct diff_filespec *two,
4192                           struct diff_options *o,
4193                           struct diff_filepair *p,
4194                           int *must_show_header,
4195                           int use_color)
4196 {
4197         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4198         const char *reset = diff_get_color(use_color, DIFF_RESET);
4199         const char *line_prefix = diff_line_prefix(o);
4200
4201         *must_show_header = 1;
4202         strbuf_init(msg, PATH_MAX * 2 + 300);
4203         switch (p->status) {
4204         case DIFF_STATUS_COPIED:
4205                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4206                             line_prefix, set, similarity_index(p));
4207                 strbuf_addf(msg, "%s\n%s%scopy from ",
4208                             reset,  line_prefix, set);
4209                 quote_c_style(name, msg, NULL, 0);
4210                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4211                 quote_c_style(other, msg, NULL, 0);
4212                 strbuf_addf(msg, "%s\n", reset);
4213                 break;
4214         case DIFF_STATUS_RENAMED:
4215                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4216                             line_prefix, set, similarity_index(p));
4217                 strbuf_addf(msg, "%s\n%s%srename from ",
4218                             reset, line_prefix, set);
4219                 quote_c_style(name, msg, NULL, 0);
4220                 strbuf_addf(msg, "%s\n%s%srename to ",
4221                             reset, line_prefix, set);
4222                 quote_c_style(other, msg, NULL, 0);
4223                 strbuf_addf(msg, "%s\n", reset);
4224                 break;
4225         case DIFF_STATUS_MODIFIED:
4226                 if (p->score) {
4227                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4228                                     line_prefix,
4229                                     set, similarity_index(p), reset);
4230                         break;
4231                 }
4232                 /* fallthru */
4233         default:
4234                 *must_show_header = 0;
4235         }
4236         if (one && two && !oideq(&one->oid, &two->oid)) {
4237                 const unsigned hexsz = the_hash_algo->hexsz;
4238                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4239
4240                 if (o->flags.binary) {
4241                         mmfile_t mf;
4242                         if ((!fill_mmfile(o->repo, &mf, one) &&
4243                              diff_filespec_is_binary(o->repo, one)) ||
4244                             (!fill_mmfile(o->repo, &mf, two) &&
4245                              diff_filespec_is_binary(o->repo, two)))
4246                                 abbrev = hexsz;
4247                 }
4248                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4249                             diff_abbrev_oid(&one->oid, abbrev),
4250                             diff_abbrev_oid(&two->oid, abbrev));
4251                 if (one->mode == two->mode)
4252                         strbuf_addf(msg, " %06o", one->mode);
4253                 strbuf_addf(msg, "%s\n", reset);
4254         }
4255 }
4256
4257 static void run_diff_cmd(const char *pgm,
4258                          const char *name,
4259                          const char *other,
4260                          const char *attr_path,
4261                          struct diff_filespec *one,
4262                          struct diff_filespec *two,
4263                          struct strbuf *msg,
4264                          struct diff_options *o,
4265                          struct diff_filepair *p)
4266 {
4267         const char *xfrm_msg = NULL;
4268         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4269         int must_show_header = 0;
4270
4271
4272         if (o->flags.allow_external) {
4273                 struct userdiff_driver *drv;
4274
4275                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4276                 if (drv && drv->external)
4277                         pgm = drv->external;
4278         }
4279
4280         if (msg) {
4281                 /*
4282                  * don't use colors when the header is intended for an
4283                  * external diff driver
4284                  */
4285                 fill_metainfo(msg, name, other, one, two, o, p,
4286                               &must_show_header,
4287                               want_color(o->use_color) && !pgm);
4288                 xfrm_msg = msg->len ? msg->buf : NULL;
4289         }
4290
4291         if (pgm) {
4292                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4293                                   complete_rewrite, o);
4294                 return;
4295         }
4296         if (one && two)
4297                 builtin_diff(name, other ? other : name,
4298                              one, two, xfrm_msg, must_show_header,
4299                              o, complete_rewrite);
4300         else
4301                 fprintf(o->file, "* Unmerged path %s\n", name);
4302 }
4303
4304 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4305 {
4306         if (DIFF_FILE_VALID(one)) {
4307                 if (!one->oid_valid) {
4308                         struct stat st;
4309                         if (one->is_stdin) {
4310                                 oidclr(&one->oid);
4311                                 return;
4312                         }
4313                         if (lstat(one->path, &st) < 0)
4314                                 die_errno("stat '%s'", one->path);
4315                         if (index_path(istate, &one->oid, one->path, &st, 0))
4316                                 die("cannot hash %s", one->path);
4317                 }
4318         }
4319         else
4320                 oidclr(&one->oid);
4321 }
4322
4323 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4324 {
4325         /* Strip the prefix but do not molest /dev/null and absolute paths */
4326         if (*namep && !is_absolute_path(*namep)) {
4327                 *namep += prefix_length;
4328                 if (**namep == '/')
4329                         ++*namep;
4330         }
4331         if (*otherp && !is_absolute_path(*otherp)) {
4332                 *otherp += prefix_length;
4333                 if (**otherp == '/')
4334                         ++*otherp;
4335         }
4336 }
4337
4338 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4339 {
4340         const char *pgm = external_diff();
4341         struct strbuf msg;
4342         struct diff_filespec *one = p->one;
4343         struct diff_filespec *two = p->two;
4344         const char *name;
4345         const char *other;
4346         const char *attr_path;
4347
4348         name  = one->path;
4349         other = (strcmp(name, two->path) ? two->path : NULL);
4350         attr_path = name;
4351         if (o->prefix_length)
4352                 strip_prefix(o->prefix_length, &name, &other);
4353
4354         if (!o->flags.allow_external)
4355                 pgm = NULL;
4356
4357         if (DIFF_PAIR_UNMERGED(p)) {
4358                 run_diff_cmd(pgm, name, NULL, attr_path,
4359                              NULL, NULL, NULL, o, p);
4360                 return;
4361         }
4362
4363         diff_fill_oid_info(one, o->repo->index);
4364         diff_fill_oid_info(two, o->repo->index);
4365
4366         if (!pgm &&
4367             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4368             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4369                 /*
4370                  * a filepair that changes between file and symlink
4371                  * needs to be split into deletion and creation.
4372                  */
4373                 struct diff_filespec *null = alloc_filespec(two->path);
4374                 run_diff_cmd(NULL, name, other, attr_path,
4375                              one, null, &msg,
4376                              o, p);
4377                 free(null);
4378                 strbuf_release(&msg);
4379
4380                 null = alloc_filespec(one->path);
4381                 run_diff_cmd(NULL, name, other, attr_path,
4382                              null, two, &msg, o, p);
4383                 free(null);
4384         }
4385         else
4386                 run_diff_cmd(pgm, name, other, attr_path,
4387                              one, two, &msg, o, p);
4388
4389         strbuf_release(&msg);
4390 }
4391
4392 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4393                          struct diffstat_t *diffstat)
4394 {
4395         const char *name;
4396         const char *other;
4397
4398         if (DIFF_PAIR_UNMERGED(p)) {
4399                 /* unmerged */
4400                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4401                                  diffstat, o, p);
4402                 return;
4403         }
4404
4405         name = p->one->path;
4406         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4407
4408         if (o->prefix_length)
4409                 strip_prefix(o->prefix_length, &name, &other);
4410
4411         diff_fill_oid_info(p->one, o->repo->index);
4412         diff_fill_oid_info(p->two, o->repo->index);
4413
4414         builtin_diffstat(name, other, p->one, p->two,
4415                          diffstat, o, p);
4416 }
4417
4418 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4419 {
4420         const char *name;
4421         const char *other;
4422         const char *attr_path;
4423
4424         if (DIFF_PAIR_UNMERGED(p)) {
4425                 /* unmerged */
4426                 return;
4427         }
4428
4429         name = p->one->path;
4430         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4431         attr_path = other ? other : name;
4432
4433         if (o->prefix_length)
4434                 strip_prefix(o->prefix_length, &name, &other);
4435
4436         diff_fill_oid_info(p->one, o->repo->index);
4437         diff_fill_oid_info(p->two, o->repo->index);
4438
4439         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4440 }
4441
4442 void repo_diff_setup(struct repository *r, struct diff_options *options)
4443 {
4444         memcpy(options, &default_diff_options, sizeof(*options));
4445
4446         options->file = stdout;
4447         options->repo = r;
4448
4449         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4450         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4451         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4452         options->abbrev = DEFAULT_ABBREV;
4453         options->line_termination = '\n';
4454         options->break_opt = -1;
4455         options->rename_limit = -1;
4456         options->dirstat_permille = diff_dirstat_permille_default;
4457         options->context = diff_context_default;
4458         options->interhunkcontext = diff_interhunk_context_default;
4459         options->ws_error_highlight = ws_error_highlight_default;
4460         options->flags.rename_empty = 1;
4461         options->objfind = NULL;
4462
4463         /* pathchange left =NULL by default */
4464         options->change = diff_change;
4465         options->add_remove = diff_addremove;
4466         options->use_color = diff_use_color_default;
4467         options->detect_rename = diff_detect_rename_default;
4468         options->xdl_opts |= diff_algorithm;
4469         if (diff_indent_heuristic)
4470                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4471
4472         options->orderfile = diff_order_file_cfg;
4473
4474         if (diff_no_prefix) {
4475                 options->a_prefix = options->b_prefix = "";
4476         } else if (!diff_mnemonic_prefix) {
4477                 options->a_prefix = "a/";
4478                 options->b_prefix = "b/";
4479         }
4480
4481         options->color_moved = diff_color_moved_default;
4482         options->color_moved_ws_handling = diff_color_moved_ws_default;
4483 }
4484
4485 void diff_setup_done(struct diff_options *options)
4486 {
4487         unsigned check_mask = DIFF_FORMAT_NAME |
4488                               DIFF_FORMAT_NAME_STATUS |
4489                               DIFF_FORMAT_CHECKDIFF |
4490                               DIFF_FORMAT_NO_OUTPUT;
4491         /*
4492          * This must be signed because we're comparing against a potentially
4493          * negative value.
4494          */
4495         const int hexsz = the_hash_algo->hexsz;
4496
4497         if (options->set_default)
4498                 options->set_default(options);
4499
4500         if (HAS_MULTI_BITS(options->output_format & check_mask))
4501                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4502
4503         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4504                 die(_("-G, -S and --find-object are mutually exclusive"));
4505
4506         /*
4507          * Most of the time we can say "there are changes"
4508          * only by checking if there are changed paths, but
4509          * --ignore-whitespace* options force us to look
4510          * inside contents.
4511          */
4512
4513         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4514                 options->flags.diff_from_contents = 1;
4515         else
4516                 options->flags.diff_from_contents = 0;
4517
4518         if (options->flags.find_copies_harder)
4519                 options->detect_rename = DIFF_DETECT_COPY;
4520
4521         if (!options->flags.relative_name)
4522                 options->prefix = NULL;
4523         if (options->prefix)
4524                 options->prefix_length = strlen(options->prefix);
4525         else
4526                 options->prefix_length = 0;
4527
4528         if (options->output_format & (DIFF_FORMAT_NAME |
4529                                       DIFF_FORMAT_NAME_STATUS |
4530                                       DIFF_FORMAT_CHECKDIFF |
4531                                       DIFF_FORMAT_NO_OUTPUT))
4532                 options->output_format &= ~(DIFF_FORMAT_RAW |
4533                                             DIFF_FORMAT_NUMSTAT |
4534                                             DIFF_FORMAT_DIFFSTAT |
4535                                             DIFF_FORMAT_SHORTSTAT |
4536                                             DIFF_FORMAT_DIRSTAT |
4537                                             DIFF_FORMAT_SUMMARY |
4538                                             DIFF_FORMAT_PATCH);
4539
4540         /*
4541          * These cases always need recursive; we do not drop caller-supplied
4542          * recursive bits for other formats here.
4543          */
4544         if (options->output_format & (DIFF_FORMAT_PATCH |
4545                                       DIFF_FORMAT_NUMSTAT |
4546                                       DIFF_FORMAT_DIFFSTAT |
4547                                       DIFF_FORMAT_SHORTSTAT |
4548                                       DIFF_FORMAT_DIRSTAT |
4549                                       DIFF_FORMAT_SUMMARY |
4550                                       DIFF_FORMAT_CHECKDIFF))
4551                 options->flags.recursive = 1;
4552         /*
4553          * Also pickaxe would not work very well if you do not say recursive
4554          */
4555         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4556                 options->flags.recursive = 1;
4557         /*
4558          * When patches are generated, submodules diffed against the work tree
4559          * must be checked for dirtiness too so it can be shown in the output
4560          */
4561         if (options->output_format & DIFF_FORMAT_PATCH)
4562                 options->flags.dirty_submodules = 1;
4563
4564         if (options->detect_rename && options->rename_limit < 0)
4565                 options->rename_limit = diff_rename_limit_default;
4566         if (hexsz < options->abbrev)
4567                 options->abbrev = hexsz; /* full */
4568
4569         /*
4570          * It does not make sense to show the first hit we happened
4571          * to have found.  It does not make sense not to return with
4572          * exit code in such a case either.
4573          */
4574         if (options->flags.quick) {
4575                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4576                 options->flags.exit_with_status = 1;
4577         }
4578
4579         options->diff_path_counter = 0;
4580
4581         if (options->flags.follow_renames && options->pathspec.nr != 1)
4582                 die(_("--follow requires exactly one pathspec"));
4583
4584         if (!options->use_color || external_diff())
4585                 options->color_moved = 0;
4586 }
4587
4588 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4589 {
4590         char c, *eq;
4591         int len;
4592
4593         if (*arg != '-')
4594                 return 0;
4595         c = *++arg;
4596         if (!c)
4597                 return 0;
4598         if (c == arg_short) {
4599                 c = *++arg;
4600                 if (!c)
4601                         return 1;
4602                 if (val && isdigit(c)) {
4603                         char *end;
4604                         int n = strtoul(arg, &end, 10);
4605                         if (*end)
4606                                 return 0;
4607                         *val = n;
4608                         return 1;
4609                 }
4610                 return 0;
4611         }
4612         if (c != '-')
4613                 return 0;
4614         arg++;
4615         eq = strchrnul(arg, '=');
4616         len = eq - arg;
4617         if (!len || strncmp(arg, arg_long, len))
4618                 return 0;
4619         if (*eq) {
4620                 int n;
4621                 char *end;
4622                 if (!isdigit(*++eq))
4623                         return 0;
4624                 n = strtoul(eq, &end, 10);
4625                 if (*end)
4626                         return 0;
4627                 *val = n;
4628         }
4629         return 1;
4630 }
4631
4632 static int diff_scoreopt_parse(const char *opt);
4633
4634 static inline int short_opt(char opt, const char **argv,
4635                             const char **optarg)
4636 {
4637         const char *arg = argv[0];
4638         if (arg[0] != '-' || arg[1] != opt)
4639                 return 0;
4640         if (arg[2] != '\0') {
4641                 *optarg = arg + 2;
4642                 return 1;
4643         }
4644         if (!argv[1])
4645                 die("Option '%c' requires a value", opt);
4646         *optarg = argv[1];
4647         return 2;
4648 }
4649
4650 int parse_long_opt(const char *opt, const char **argv,
4651                    const char **optarg)
4652 {
4653         const char *arg = argv[0];
4654         if (!skip_prefix(arg, "--", &arg))
4655                 return 0;
4656         if (!skip_prefix(arg, opt, &arg))
4657                 return 0;
4658         if (*arg == '=') { /* stuck form: --option=value */
4659                 *optarg = arg + 1;
4660                 return 1;
4661         }
4662         if (*arg != '\0')
4663                 return 0;
4664         /* separate form: --option value */
4665         if (!argv[1])
4666                 die("Option '--%s' requires a value", opt);
4667         *optarg = argv[1];
4668         return 2;
4669 }
4670
4671 static int stat_opt(struct diff_options *options, const char **av)
4672 {
4673         const char *arg = av[0];
4674         char *end;
4675         int width = options->stat_width;
4676         int name_width = options->stat_name_width;
4677         int graph_width = options->stat_graph_width;
4678         int count = options->stat_count;
4679         int argcount = 1;
4680
4681         if (!skip_prefix(arg, "--stat", &arg))
4682                 BUG("stat option does not begin with --stat: %s", arg);
4683         end = (char *)arg;
4684
4685         switch (*arg) {
4686         case '-':
4687                 if (skip_prefix(arg, "-width", &arg)) {
4688                         if (*arg == '=')
4689                                 width = strtoul(arg + 1, &end, 10);
4690                         else if (!*arg && !av[1])
4691                                 die_want_option("--stat-width");
4692                         else if (!*arg) {
4693                                 width = strtoul(av[1], &end, 10);
4694                                 argcount = 2;
4695                         }
4696                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4697                         if (*arg == '=')
4698                                 name_width = strtoul(arg + 1, &end, 10);
4699                         else if (!*arg && !av[1])
4700                                 die_want_option("--stat-name-width");
4701                         else if (!*arg) {
4702                                 name_width = strtoul(av[1], &end, 10);
4703                                 argcount = 2;
4704                         }
4705                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4706                         if (*arg == '=')
4707                                 graph_width = strtoul(arg + 1, &end, 10);
4708                         else if (!*arg && !av[1])
4709                                 die_want_option("--stat-graph-width");
4710                         else if (!*arg) {
4711                                 graph_width = strtoul(av[1], &end, 10);
4712                                 argcount = 2;
4713                         }
4714                 } else if (skip_prefix(arg, "-count", &arg)) {
4715                         if (*arg == '=')
4716                                 count = strtoul(arg + 1, &end, 10);
4717                         else if (!*arg && !av[1])
4718                                 die_want_option("--stat-count");
4719                         else if (!*arg) {
4720                                 count = strtoul(av[1], &end, 10);
4721                                 argcount = 2;
4722                         }
4723                 }
4724                 break;
4725         case '=':
4726                 width = strtoul(arg+1, &end, 10);
4727                 if (*end == ',')
4728                         name_width = strtoul(end+1, &end, 10);
4729                 if (*end == ',')
4730                         count = strtoul(end+1, &end, 10);
4731         }
4732
4733         /* Important! This checks all the error cases! */
4734         if (*end)
4735                 return 0;
4736         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4737         options->stat_name_width = name_width;
4738         options->stat_graph_width = graph_width;
4739         options->stat_width = width;
4740         options->stat_count = count;
4741         return argcount;
4742 }
4743
4744 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4745 {
4746         struct strbuf errmsg = STRBUF_INIT;
4747         if (parse_dirstat_params(options, params, &errmsg))
4748                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4749                     errmsg.buf);
4750         strbuf_release(&errmsg);
4751         /*
4752          * The caller knows a dirstat-related option is given from the command
4753          * line; allow it to say "return this_function();"
4754          */
4755         options->output_format |= DIFF_FORMAT_DIRSTAT;
4756         return 1;
4757 }
4758
4759 static int parse_submodule_opt(struct diff_options *options, const char *value)
4760 {
4761         if (parse_submodule_params(options, value))
4762                 die(_("Failed to parse --submodule option parameter: '%s'"),
4763                         value);
4764         return 1;
4765 }
4766
4767 static const char diff_status_letters[] = {
4768         DIFF_STATUS_ADDED,
4769         DIFF_STATUS_COPIED,
4770         DIFF_STATUS_DELETED,
4771         DIFF_STATUS_MODIFIED,
4772         DIFF_STATUS_RENAMED,
4773         DIFF_STATUS_TYPE_CHANGED,
4774         DIFF_STATUS_UNKNOWN,
4775         DIFF_STATUS_UNMERGED,
4776         DIFF_STATUS_FILTER_AON,
4777         DIFF_STATUS_FILTER_BROKEN,
4778         '\0',
4779 };
4780
4781 static unsigned int filter_bit['Z' + 1];
4782
4783 static void prepare_filter_bits(void)
4784 {
4785         int i;
4786
4787         if (!filter_bit[DIFF_STATUS_ADDED]) {
4788                 for (i = 0; diff_status_letters[i]; i++)
4789                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4790         }
4791 }
4792
4793 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4794 {
4795         return opt->filter & filter_bit[(int) status];
4796 }
4797
4798 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4799 {
4800         int i, optch;
4801
4802         prepare_filter_bits();
4803
4804         /*
4805          * If there is a negation e.g. 'd' in the input, and we haven't
4806          * initialized the filter field with another --diff-filter, start
4807          * from full set of bits, except for AON.
4808          */
4809         if (!opt->filter) {
4810                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4811                         if (optch < 'a' || 'z' < optch)
4812                                 continue;
4813                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4814                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4815                         break;
4816                 }
4817         }
4818
4819         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4820                 unsigned int bit;
4821                 int negate;
4822
4823                 if ('a' <= optch && optch <= 'z') {
4824                         negate = 1;
4825                         optch = toupper(optch);
4826                 } else {
4827                         negate = 0;
4828                 }
4829
4830                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4831                 if (!bit)
4832                         return optarg[i];
4833                 if (negate)
4834                         opt->filter &= ~bit;
4835                 else
4836                         opt->filter |= bit;
4837         }
4838         return 0;
4839 }
4840
4841 static void enable_patch_output(int *fmt) {
4842         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4843         *fmt |= DIFF_FORMAT_PATCH;
4844 }
4845
4846 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4847 {
4848         int val = parse_ws_error_highlight(arg);
4849
4850         if (val < 0) {
4851                 error("unknown value after ws-error-highlight=%.*s",
4852                       -1 - val, arg);
4853                 return 0;
4854         }
4855         opt->ws_error_highlight = val;
4856         return 1;
4857 }
4858
4859 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4860 {
4861         struct object_id oid;
4862
4863         if (get_oid(arg, &oid))
4864                 return error("unable to resolve '%s'", arg);
4865
4866         if (!opt->objfind)
4867                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4868
4869         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4870         opt->flags.recursive = 1;
4871         opt->flags.tree_in_recursive = 1;
4872         oidset_insert(opt->objfind, &oid);
4873         return 1;
4874 }
4875
4876 int diff_opt_parse(struct diff_options *options,
4877                    const char **av, int ac, const char *prefix)
4878 {
4879         const char *arg = av[0];
4880         const char *optarg;
4881         int argcount;
4882
4883         if (!prefix)
4884                 prefix = "";
4885
4886         /* Output format options */
4887         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4888             || opt_arg(arg, 'U', "unified", &options->context))
4889                 enable_patch_output(&options->output_format);
4890         else if (!strcmp(arg, "--raw"))
4891                 options->output_format |= DIFF_FORMAT_RAW;
4892         else if (!strcmp(arg, "--patch-with-raw")) {
4893                 enable_patch_output(&options->output_format);
4894                 options->output_format |= DIFF_FORMAT_RAW;
4895         } else if (!strcmp(arg, "--numstat"))
4896                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4897         else if (!strcmp(arg, "--shortstat"))
4898                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4899         else if (skip_prefix(arg, "-X", &arg) ||
4900                  skip_to_optional_arg(arg, "--dirstat", &arg))
4901                 return parse_dirstat_opt(options, arg);
4902         else if (!strcmp(arg, "--cumulative"))
4903                 return parse_dirstat_opt(options, "cumulative");
4904         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4905                 parse_dirstat_opt(options, "files");
4906                 return parse_dirstat_opt(options, arg);
4907         }
4908         else if (!strcmp(arg, "--check"))
4909                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4910         else if (!strcmp(arg, "--summary"))
4911                 options->output_format |= DIFF_FORMAT_SUMMARY;
4912         else if (!strcmp(arg, "--patch-with-stat")) {
4913                 enable_patch_output(&options->output_format);
4914                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4915         } else if (!strcmp(arg, "--name-only"))
4916                 options->output_format |= DIFF_FORMAT_NAME;
4917         else if (!strcmp(arg, "--name-status"))
4918                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4919         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4920                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4921         else if (starts_with(arg, "--stat"))
4922                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4923                 return stat_opt(options, av);
4924         else if (!strcmp(arg, "--compact-summary")) {
4925                  options->flags.stat_with_summary = 1;
4926                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4927         } else if (!strcmp(arg, "--no-compact-summary"))
4928                  options->flags.stat_with_summary = 0;
4929         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4930                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4931         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4932                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4933         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4934                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4935
4936         /* renames options */
4937         else if (starts_with(arg, "-B") ||
4938                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4939                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4940                         return error("invalid argument to -B: %s", arg+2);
4941         }
4942         else if (starts_with(arg, "-M") ||
4943                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4944                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4945                         return error("invalid argument to -M: %s", arg+2);
4946                 options->detect_rename = DIFF_DETECT_RENAME;
4947         }
4948         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4949                 options->irreversible_delete = 1;
4950         }
4951         else if (starts_with(arg, "-C") ||
4952                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4953                 if (options->detect_rename == DIFF_DETECT_COPY)
4954                         options->flags.find_copies_harder = 1;
4955                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4956                         return error("invalid argument to -C: %s", arg+2);
4957                 options->detect_rename = DIFF_DETECT_COPY;
4958         }
4959         else if (!strcmp(arg, "--no-renames"))
4960                 options->detect_rename = 0;
4961         else if (!strcmp(arg, "--rename-empty"))
4962                 options->flags.rename_empty = 1;
4963         else if (!strcmp(arg, "--no-rename-empty"))
4964                 options->flags.rename_empty = 0;
4965         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4966                 options->flags.relative_name = 1;
4967                 if (arg)
4968                         options->prefix = arg;
4969         }
4970
4971         /* xdiff options */
4972         else if (!strcmp(arg, "--minimal"))
4973                 DIFF_XDL_SET(options, NEED_MINIMAL);
4974         else if (!strcmp(arg, "--no-minimal"))
4975                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4976         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4977                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4978         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4979                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4980         else if (!strcmp(arg, "--ignore-space-at-eol"))
4981                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4982         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4983                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4984         else if (!strcmp(arg, "--ignore-blank-lines"))
4985                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4986         else if (!strcmp(arg, "--indent-heuristic"))
4987                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4988         else if (!strcmp(arg, "--no-indent-heuristic"))
4989                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4990         else if (!strcmp(arg, "--patience")) {
4991                 int i;
4992                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4993                 /*
4994                  * Both --patience and --anchored use PATIENCE_DIFF
4995                  * internally, so remove any anchors previously
4996                  * specified.
4997                  */
4998                 for (i = 0; i < options->anchors_nr; i++)
4999                         free(options->anchors[i]);
5000                 options->anchors_nr = 0;
5001         } else if (!strcmp(arg, "--histogram"))
5002                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5003         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5004                 long value = parse_algorithm_value(optarg);
5005                 if (value < 0)
5006                         return error("option diff-algorithm accepts \"myers\", "
5007                                      "\"minimal\", \"patience\" and \"histogram\"");
5008                 /* clear out previous settings */
5009                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5010                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5011                 options->xdl_opts |= value;
5012                 return argcount;
5013         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5014                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5015                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5016                            options->anchors_alloc);
5017                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5018         }
5019
5020         /* flags options */
5021         else if (!strcmp(arg, "--binary")) {
5022                 enable_patch_output(&options->output_format);
5023                 options->flags.binary = 1;
5024         }
5025         else if (!strcmp(arg, "--full-index"))
5026                 options->flags.full_index = 1;
5027         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5028                 options->flags.text = 1;
5029         else if (!strcmp(arg, "-R"))
5030                 options->flags.reverse_diff = 1;
5031         else if (!strcmp(arg, "--find-copies-harder"))
5032                 options->flags.find_copies_harder = 1;
5033         else if (!strcmp(arg, "--follow"))
5034                 options->flags.follow_renames = 1;
5035         else if (!strcmp(arg, "--no-follow")) {
5036                 options->flags.follow_renames = 0;
5037                 options->flags.default_follow_renames = 0;
5038         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5039                 int value = git_config_colorbool(NULL, arg);
5040                 if (value < 0)
5041                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5042                 options->use_color = value;
5043         }
5044         else if (!strcmp(arg, "--no-color"))
5045                 options->use_color = 0;
5046         else if (!strcmp(arg, "--color-moved")) {
5047                 if (diff_color_moved_default)
5048                         options->color_moved = diff_color_moved_default;
5049                 if (options->color_moved == COLOR_MOVED_NO)
5050                         options->color_moved = COLOR_MOVED_DEFAULT;
5051         } else if (!strcmp(arg, "--no-color-moved"))
5052                 options->color_moved = COLOR_MOVED_NO;
5053         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5054                 int cm = parse_color_moved(arg);
5055                 if (cm < 0)
5056                         die("bad --color-moved argument: %s", arg);
5057                 options->color_moved = cm;
5058         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5059                 options->color_moved_ws_handling = 0;
5060         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5061                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5062         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5063                 options->use_color = 1;
5064                 options->word_diff = DIFF_WORDS_COLOR;
5065         }
5066         else if (!strcmp(arg, "--word-diff")) {
5067                 if (options->word_diff == DIFF_WORDS_NONE)
5068                         options->word_diff = DIFF_WORDS_PLAIN;
5069         }
5070         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5071                 if (!strcmp(arg, "plain"))
5072                         options->word_diff = DIFF_WORDS_PLAIN;
5073                 else if (!strcmp(arg, "color")) {
5074                         options->use_color = 1;
5075                         options->word_diff = DIFF_WORDS_COLOR;
5076                 }
5077                 else if (!strcmp(arg, "porcelain"))
5078                         options->word_diff = DIFF_WORDS_PORCELAIN;
5079                 else if (!strcmp(arg, "none"))
5080                         options->word_diff = DIFF_WORDS_NONE;
5081                 else
5082                         die("bad --word-diff argument: %s", arg);
5083         }
5084         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5085                 if (options->word_diff == DIFF_WORDS_NONE)
5086                         options->word_diff = DIFF_WORDS_PLAIN;
5087                 options->word_regex = optarg;
5088                 return argcount;
5089         }
5090         else if (!strcmp(arg, "--exit-code"))
5091                 options->flags.exit_with_status = 1;
5092         else if (!strcmp(arg, "--quiet"))
5093                 options->flags.quick = 1;
5094         else if (!strcmp(arg, "--ext-diff"))
5095                 options->flags.allow_external = 1;
5096         else if (!strcmp(arg, "--no-ext-diff"))
5097                 options->flags.allow_external = 0;
5098         else if (!strcmp(arg, "--textconv")) {
5099                 options->flags.allow_textconv = 1;
5100                 options->flags.textconv_set_via_cmdline = 1;
5101         } else if (!strcmp(arg, "--no-textconv"))
5102                 options->flags.allow_textconv = 0;
5103         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5104                 options->flags.override_submodule_config = 1;
5105                 handle_ignore_submodules_arg(options, arg);
5106         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5107                 return parse_submodule_opt(options, arg);
5108         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5109                 return parse_ws_error_highlight_opt(options, arg);
5110         else if (!strcmp(arg, "--ita-invisible-in-index"))
5111                 options->ita_invisible_in_index = 1;
5112         else if (!strcmp(arg, "--ita-visible-in-index"))
5113                 options->ita_invisible_in_index = 0;
5114
5115         /* misc options */
5116         else if (!strcmp(arg, "-z"))
5117                 options->line_termination = 0;
5118         else if ((argcount = short_opt('l', av, &optarg))) {
5119                 options->rename_limit = strtoul(optarg, NULL, 10);
5120                 return argcount;
5121         }
5122         else if ((argcount = short_opt('S', av, &optarg))) {
5123                 options->pickaxe = optarg;
5124                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5125                 return argcount;
5126         } else if ((argcount = short_opt('G', av, &optarg))) {
5127                 options->pickaxe = optarg;
5128                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5129                 return argcount;
5130         }
5131         else if (!strcmp(arg, "--pickaxe-all"))
5132                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5133         else if (!strcmp(arg, "--pickaxe-regex"))
5134                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5135         else if ((argcount = short_opt('O', av, &optarg))) {
5136                 options->orderfile = prefix_filename(prefix, optarg);
5137                 return argcount;
5138         } else if (skip_prefix(arg, "--find-object=", &arg))
5139                 return parse_objfind_opt(options, arg);
5140         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5141                 int offending = parse_diff_filter_opt(optarg, options);
5142                 if (offending)
5143                         die("unknown change class '%c' in --diff-filter=%s",
5144                             offending, optarg);
5145                 return argcount;
5146         }
5147         else if (!strcmp(arg, "--no-abbrev"))
5148                 options->abbrev = 0;
5149         else if (!strcmp(arg, "--abbrev"))
5150                 options->abbrev = DEFAULT_ABBREV;
5151         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5152                 options->abbrev = strtoul(arg, NULL, 10);
5153                 if (options->abbrev < MINIMUM_ABBREV)
5154                         options->abbrev = MINIMUM_ABBREV;
5155                 else if (the_hash_algo->hexsz < options->abbrev)
5156                         options->abbrev = the_hash_algo->hexsz;
5157         }
5158         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5159                 options->a_prefix = optarg;
5160                 return argcount;
5161         }
5162         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5163                 options->line_prefix = optarg;
5164                 options->line_prefix_length = strlen(options->line_prefix);
5165                 graph_setup_line_prefix(options);
5166                 return argcount;
5167         }
5168         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5169                 options->b_prefix = optarg;
5170                 return argcount;
5171         }
5172         else if (!strcmp(arg, "--no-prefix"))
5173                 options->a_prefix = options->b_prefix = "";
5174         else if (opt_arg(arg, '\0', "inter-hunk-context",
5175                          &options->interhunkcontext))
5176                 ;
5177         else if (!strcmp(arg, "-W"))
5178                 options->flags.funccontext = 1;
5179         else if (!strcmp(arg, "--function-context"))
5180                 options->flags.funccontext = 1;
5181         else if (!strcmp(arg, "--no-function-context"))
5182                 options->flags.funccontext = 0;
5183         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5184                 char *path = prefix_filename(prefix, optarg);
5185                 options->file = xfopen(path, "w");
5186                 options->close_file = 1;
5187                 if (options->use_color != GIT_COLOR_ALWAYS)
5188                         options->use_color = GIT_COLOR_NEVER;
5189                 free(path);
5190                 return argcount;
5191         } else
5192                 return 0;
5193         return 1;
5194 }
5195
5196 int parse_rename_score(const char **cp_p)
5197 {
5198         unsigned long num, scale;
5199         int ch, dot;
5200         const char *cp = *cp_p;
5201
5202         num = 0;
5203         scale = 1;
5204         dot = 0;
5205         for (;;) {
5206                 ch = *cp;
5207                 if ( !dot && ch == '.' ) {
5208                         scale = 1;
5209                         dot = 1;
5210                 } else if ( ch == '%' ) {
5211                         scale = dot ? scale*100 : 100;
5212                         cp++;   /* % is always at the end */
5213                         break;
5214                 } else if ( ch >= '0' && ch <= '9' ) {
5215                         if ( scale < 100000 ) {
5216                                 scale *= 10;
5217                                 num = (num*10) + (ch-'0');
5218                         }
5219                 } else {
5220                         break;
5221                 }
5222                 cp++;
5223         }
5224         *cp_p = cp;
5225
5226         /* user says num divided by scale and we say internally that
5227          * is MAX_SCORE * num / scale.
5228          */
5229         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5230 }
5231
5232 static int diff_scoreopt_parse(const char *opt)
5233 {
5234         int opt1, opt2, cmd;
5235
5236         if (*opt++ != '-')
5237                 return -1;
5238         cmd = *opt++;
5239         if (cmd == '-') {
5240                 /* convert the long-form arguments into short-form versions */
5241                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5242                         if (*opt == 0 || *opt++ == '=')
5243                                 cmd = 'B';
5244                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5245                         if (*opt == 0 || *opt++ == '=')
5246                                 cmd = 'C';
5247                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5248                         if (*opt == 0 || *opt++ == '=')
5249                                 cmd = 'M';
5250                 }
5251         }
5252         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5253                 return -1; /* that is not a -M, -C, or -B option */
5254
5255         opt1 = parse_rename_score(&opt);
5256         if (cmd != 'B')
5257                 opt2 = 0;
5258         else {
5259                 if (*opt == 0)
5260                         opt2 = 0;
5261                 else if (*opt != '/')
5262                         return -1; /* we expect -B80/99 or -B80 */
5263                 else {
5264                         opt++;
5265                         opt2 = parse_rename_score(&opt);
5266                 }
5267         }
5268         if (*opt != 0)
5269                 return -1;
5270         return opt1 | (opt2 << 16);
5271 }
5272
5273 struct diff_queue_struct diff_queued_diff;
5274
5275 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5276 {
5277         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5278         queue->queue[queue->nr++] = dp;
5279 }
5280
5281 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5282                                  struct diff_filespec *one,
5283                                  struct diff_filespec *two)
5284 {
5285         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5286         dp->one = one;
5287         dp->two = two;
5288         if (queue)
5289                 diff_q(queue, dp);
5290         return dp;
5291 }
5292
5293 void diff_free_filepair(struct diff_filepair *p)
5294 {
5295         free_filespec(p->one);
5296         free_filespec(p->two);
5297         free(p);
5298 }
5299
5300 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5301 {
5302         int abblen;
5303         const char *abbrev;
5304
5305         /* Do we want all 40 hex characters? */
5306         if (len == the_hash_algo->hexsz)
5307                 return oid_to_hex(oid);
5308
5309         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5310         abbrev = diff_abbrev_oid(oid, len);
5311
5312         if (!print_sha1_ellipsis())
5313                 return abbrev;
5314
5315         abblen = strlen(abbrev);
5316
5317         /*
5318          * In well-behaved cases, where the abbreviated result is the
5319          * same as the requested length, append three dots after the
5320          * abbreviation (hence the whole logic is limited to the case
5321          * where abblen < 37); when the actual abbreviated result is a
5322          * bit longer than the requested length, we reduce the number
5323          * of dots so that they match the well-behaved ones.  However,
5324          * if the actual abbreviation is longer than the requested
5325          * length by more than three, we give up on aligning, and add
5326          * three dots anyway, to indicate that the output is not the
5327          * full object name.  Yes, this may be suboptimal, but this
5328          * appears only in "diff --raw --abbrev" output and it is not
5329          * worth the effort to change it now.  Note that this would
5330          * likely to work fine when the automatic sizing of default
5331          * abbreviation length is used--we would be fed -1 in "len" in
5332          * that case, and will end up always appending three-dots, but
5333          * the automatic sizing is supposed to give abblen that ensures
5334          * uniqueness across all objects (statistically speaking).
5335          */
5336         if (abblen < the_hash_algo->hexsz - 3) {
5337                 static char hex[GIT_MAX_HEXSZ + 1];
5338                 if (len < abblen && abblen <= len + 2)
5339                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5340                 else
5341                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5342                 return hex;
5343         }
5344
5345         return oid_to_hex(oid);
5346 }
5347
5348 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5349 {
5350         int line_termination = opt->line_termination;
5351         int inter_name_termination = line_termination ? '\t' : '\0';
5352
5353         fprintf(opt->file, "%s", diff_line_prefix(opt));
5354         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5355                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5356                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5357                 fprintf(opt->file, "%s ",
5358                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5359         }
5360         if (p->score) {
5361                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5362                         inter_name_termination);
5363         } else {
5364                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5365         }
5366
5367         if (p->status == DIFF_STATUS_COPIED ||
5368             p->status == DIFF_STATUS_RENAMED) {
5369                 const char *name_a, *name_b;
5370                 name_a = p->one->path;
5371                 name_b = p->two->path;
5372                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5373                 write_name_quoted(name_a, opt->file, inter_name_termination);
5374                 write_name_quoted(name_b, opt->file, line_termination);
5375         } else {
5376                 const char *name_a, *name_b;
5377                 name_a = p->one->mode ? p->one->path : p->two->path;
5378                 name_b = NULL;
5379                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5380                 write_name_quoted(name_a, opt->file, line_termination);
5381         }
5382 }
5383
5384 int diff_unmodified_pair(struct diff_filepair *p)
5385 {
5386         /* This function is written stricter than necessary to support
5387          * the currently implemented transformers, but the idea is to
5388          * let transformers to produce diff_filepairs any way they want,
5389          * and filter and clean them up here before producing the output.
5390          */
5391         struct diff_filespec *one = p->one, *two = p->two;
5392
5393         if (DIFF_PAIR_UNMERGED(p))
5394                 return 0; /* unmerged is interesting */
5395
5396         /* deletion, addition, mode or type change
5397          * and rename are all interesting.
5398          */
5399         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5400             DIFF_PAIR_MODE_CHANGED(p) ||
5401             strcmp(one->path, two->path))
5402                 return 0;
5403
5404         /* both are valid and point at the same path.  that is, we are
5405          * dealing with a change.
5406          */
5407         if (one->oid_valid && two->oid_valid &&
5408             oideq(&one->oid, &two->oid) &&
5409             !one->dirty_submodule && !two->dirty_submodule)
5410                 return 1; /* no change */
5411         if (!one->oid_valid && !two->oid_valid)
5412                 return 1; /* both look at the same file on the filesystem. */
5413         return 0;
5414 }
5415
5416 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5417 {
5418         if (diff_unmodified_pair(p))
5419                 return;
5420
5421         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5422             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5423                 return; /* no tree diffs in patch format */
5424
5425         run_diff(p, o);
5426 }
5427
5428 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5429                             struct diffstat_t *diffstat)
5430 {
5431         if (diff_unmodified_pair(p))
5432                 return;
5433
5434         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5435             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5436                 return; /* no useful stat for tree diffs */
5437
5438         run_diffstat(p, o, diffstat);
5439 }
5440
5441 static void diff_flush_checkdiff(struct diff_filepair *p,
5442                 struct diff_options *o)
5443 {
5444         if (diff_unmodified_pair(p))
5445                 return;
5446
5447         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5448             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5449                 return; /* nothing to check in tree diffs */
5450
5451         run_checkdiff(p, o);
5452 }
5453
5454 int diff_queue_is_empty(void)
5455 {
5456         struct diff_queue_struct *q = &diff_queued_diff;
5457         int i;
5458         for (i = 0; i < q->nr; i++)
5459                 if (!diff_unmodified_pair(q->queue[i]))
5460                         return 0;
5461         return 1;
5462 }
5463
5464 #if DIFF_DEBUG
5465 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5466 {
5467         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5468                 x, one ? one : "",
5469                 s->path,
5470                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5471                 s->mode,
5472                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5473         fprintf(stderr, "queue[%d] %s size %lu\n",
5474                 x, one ? one : "",
5475                 s->size);
5476 }
5477
5478 void diff_debug_filepair(const struct diff_filepair *p, int i)
5479 {
5480         diff_debug_filespec(p->one, i, "one");
5481         diff_debug_filespec(p->two, i, "two");
5482         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5483                 p->score, p->status ? p->status : '?',
5484                 p->one->rename_used, p->broken_pair);
5485 }
5486
5487 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5488 {
5489         int i;
5490         if (msg)
5491                 fprintf(stderr, "%s\n", msg);
5492         fprintf(stderr, "q->nr = %d\n", q->nr);
5493         for (i = 0; i < q->nr; i++) {
5494                 struct diff_filepair *p = q->queue[i];
5495                 diff_debug_filepair(p, i);
5496         }
5497 }
5498 #endif
5499
5500 static void diff_resolve_rename_copy(void)
5501 {
5502         int i;
5503         struct diff_filepair *p;
5504         struct diff_queue_struct *q = &diff_queued_diff;
5505
5506         diff_debug_queue("resolve-rename-copy", q);
5507
5508         for (i = 0; i < q->nr; i++) {
5509                 p = q->queue[i];
5510                 p->status = 0; /* undecided */
5511                 if (DIFF_PAIR_UNMERGED(p))
5512                         p->status = DIFF_STATUS_UNMERGED;
5513                 else if (!DIFF_FILE_VALID(p->one))
5514                         p->status = DIFF_STATUS_ADDED;
5515                 else if (!DIFF_FILE_VALID(p->two))
5516                         p->status = DIFF_STATUS_DELETED;
5517                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5518                         p->status = DIFF_STATUS_TYPE_CHANGED;
5519
5520                 /* from this point on, we are dealing with a pair
5521                  * whose both sides are valid and of the same type, i.e.
5522                  * either in-place edit or rename/copy edit.
5523                  */
5524                 else if (DIFF_PAIR_RENAME(p)) {
5525                         /*
5526                          * A rename might have re-connected a broken
5527                          * pair up, causing the pathnames to be the
5528                          * same again. If so, that's not a rename at
5529                          * all, just a modification..
5530                          *
5531                          * Otherwise, see if this source was used for
5532                          * multiple renames, in which case we decrement
5533                          * the count, and call it a copy.
5534                          */
5535                         if (!strcmp(p->one->path, p->two->path))
5536                                 p->status = DIFF_STATUS_MODIFIED;
5537                         else if (--p->one->rename_used > 0)
5538                                 p->status = DIFF_STATUS_COPIED;
5539                         else
5540                                 p->status = DIFF_STATUS_RENAMED;
5541                 }
5542                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5543                          p->one->mode != p->two->mode ||
5544                          p->one->dirty_submodule ||
5545                          p->two->dirty_submodule ||
5546                          is_null_oid(&p->one->oid))
5547                         p->status = DIFF_STATUS_MODIFIED;
5548                 else {
5549                         /* This is a "no-change" entry and should not
5550                          * happen anymore, but prepare for broken callers.
5551                          */
5552                         error("feeding unmodified %s to diffcore",
5553                               p->one->path);
5554                         p->status = DIFF_STATUS_UNKNOWN;
5555                 }
5556         }
5557         diff_debug_queue("resolve-rename-copy done", q);
5558 }
5559
5560 static int check_pair_status(struct diff_filepair *p)
5561 {
5562         switch (p->status) {
5563         case DIFF_STATUS_UNKNOWN:
5564                 return 0;
5565         case 0:
5566                 die("internal error in diff-resolve-rename-copy");
5567         default:
5568                 return 1;
5569         }
5570 }
5571
5572 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5573 {
5574         int fmt = opt->output_format;
5575
5576         if (fmt & DIFF_FORMAT_CHECKDIFF)
5577                 diff_flush_checkdiff(p, opt);
5578         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5579                 diff_flush_raw(p, opt);
5580         else if (fmt & DIFF_FORMAT_NAME) {
5581                 const char *name_a, *name_b;
5582                 name_a = p->two->path;
5583                 name_b = NULL;
5584                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5585                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5586                 write_name_quoted(name_a, opt->file, opt->line_termination);
5587         }
5588 }
5589
5590 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5591 {
5592         struct strbuf sb = STRBUF_INIT;
5593         if (fs->mode)
5594                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5595         else
5596                 strbuf_addf(&sb, " %s ", newdelete);
5597
5598         quote_c_style(fs->path, &sb, NULL, 0);
5599         strbuf_addch(&sb, '\n');
5600         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5601                          sb.buf, sb.len, 0);
5602         strbuf_release(&sb);
5603 }
5604
5605 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5606                 int show_name)
5607 {
5608         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5609                 struct strbuf sb = STRBUF_INIT;
5610                 strbuf_addf(&sb, " mode change %06o => %06o",
5611                             p->one->mode, p->two->mode);
5612                 if (show_name) {
5613                         strbuf_addch(&sb, ' ');
5614                         quote_c_style(p->two->path, &sb, NULL, 0);
5615                 }
5616                 strbuf_addch(&sb, '\n');
5617                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5618                                  sb.buf, sb.len, 0);
5619                 strbuf_release(&sb);
5620         }
5621 }
5622
5623 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5624                 struct diff_filepair *p)
5625 {
5626         struct strbuf sb = STRBUF_INIT;
5627         struct strbuf names = STRBUF_INIT;
5628
5629         pprint_rename(&names, p->one->path, p->two->path);
5630         strbuf_addf(&sb, " %s %s (%d%%)\n",
5631                     renamecopy, names.buf, similarity_index(p));
5632         strbuf_release(&names);
5633         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5634                                  sb.buf, sb.len, 0);
5635         show_mode_change(opt, p, 0);
5636         strbuf_release(&sb);
5637 }
5638
5639 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5640 {
5641         switch(p->status) {
5642         case DIFF_STATUS_DELETED:
5643                 show_file_mode_name(opt, "delete", p->one);
5644                 break;
5645         case DIFF_STATUS_ADDED:
5646                 show_file_mode_name(opt, "create", p->two);
5647                 break;
5648         case DIFF_STATUS_COPIED:
5649                 show_rename_copy(opt, "copy", p);
5650                 break;
5651         case DIFF_STATUS_RENAMED:
5652                 show_rename_copy(opt, "rename", p);
5653                 break;
5654         default:
5655                 if (p->score) {
5656                         struct strbuf sb = STRBUF_INIT;
5657                         strbuf_addstr(&sb, " rewrite ");
5658                         quote_c_style(p->two->path, &sb, NULL, 0);
5659                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5660                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5661                                          sb.buf, sb.len, 0);
5662                         strbuf_release(&sb);
5663                 }
5664                 show_mode_change(opt, p, !p->score);
5665                 break;
5666         }
5667 }
5668
5669 struct patch_id_t {
5670         git_SHA_CTX *ctx;
5671         int patchlen;
5672 };
5673
5674 static int remove_space(char *line, int len)
5675 {
5676         int i;
5677         char *dst = line;
5678         unsigned char c;
5679
5680         for (i = 0; i < len; i++)
5681                 if (!isspace((c = line[i])))
5682                         *dst++ = c;
5683
5684         return dst - line;
5685 }
5686
5687 static void patch_id_consume(void *priv, char *line, unsigned long len)
5688 {
5689         struct patch_id_t *data = priv;
5690         int new_len;
5691
5692         new_len = remove_space(line, len);
5693
5694         git_SHA1_Update(data->ctx, line, new_len);
5695         data->patchlen += new_len;
5696 }
5697
5698 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5699 {
5700         git_SHA1_Update(ctx, str, strlen(str));
5701 }
5702
5703 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5704 {
5705         /* large enough for 2^32 in octal */
5706         char buf[12];
5707         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5708         git_SHA1_Update(ctx, buf, len);
5709 }
5710
5711 /* returns 0 upon success, and writes result into sha1 */
5712 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5713 {
5714         struct diff_queue_struct *q = &diff_queued_diff;
5715         int i;
5716         git_SHA_CTX ctx;
5717         struct patch_id_t data;
5718
5719         git_SHA1_Init(&ctx);
5720         memset(&data, 0, sizeof(struct patch_id_t));
5721         data.ctx = &ctx;
5722
5723         for (i = 0; i < q->nr; i++) {
5724                 xpparam_t xpp;
5725                 xdemitconf_t xecfg;
5726                 mmfile_t mf1, mf2;
5727                 struct diff_filepair *p = q->queue[i];
5728                 int len1, len2;
5729
5730                 memset(&xpp, 0, sizeof(xpp));
5731                 memset(&xecfg, 0, sizeof(xecfg));
5732                 if (p->status == 0)
5733                         return error("internal diff status error");
5734                 if (p->status == DIFF_STATUS_UNKNOWN)
5735                         continue;
5736                 if (diff_unmodified_pair(p))
5737                         continue;
5738                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5739                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5740                         continue;
5741                 if (DIFF_PAIR_UNMERGED(p))
5742                         continue;
5743
5744                 diff_fill_oid_info(p->one, options->repo->index);
5745                 diff_fill_oid_info(p->two, options->repo->index);
5746
5747                 len1 = remove_space(p->one->path, strlen(p->one->path));
5748                 len2 = remove_space(p->two->path, strlen(p->two->path));
5749                 patch_id_add_string(&ctx, "diff--git");
5750                 patch_id_add_string(&ctx, "a/");
5751                 git_SHA1_Update(&ctx, p->one->path, len1);
5752                 patch_id_add_string(&ctx, "b/");
5753                 git_SHA1_Update(&ctx, p->two->path, len2);
5754
5755                 if (p->one->mode == 0) {
5756                         patch_id_add_string(&ctx, "newfilemode");
5757                         patch_id_add_mode(&ctx, p->two->mode);
5758                         patch_id_add_string(&ctx, "---/dev/null");
5759                         patch_id_add_string(&ctx, "+++b/");
5760                         git_SHA1_Update(&ctx, p->two->path, len2);
5761                 } else if (p->two->mode == 0) {
5762                         patch_id_add_string(&ctx, "deletedfilemode");
5763                         patch_id_add_mode(&ctx, p->one->mode);
5764                         patch_id_add_string(&ctx, "---a/");
5765                         git_SHA1_Update(&ctx, p->one->path, len1);
5766                         patch_id_add_string(&ctx, "+++/dev/null");
5767                 } else {
5768                         patch_id_add_string(&ctx, "---a/");
5769                         git_SHA1_Update(&ctx, p->one->path, len1);
5770                         patch_id_add_string(&ctx, "+++b/");
5771                         git_SHA1_Update(&ctx, p->two->path, len2);
5772                 }
5773
5774                 if (diff_header_only)
5775                         continue;
5776
5777                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5778                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5779                         return error("unable to read files to diff");
5780
5781                 if (diff_filespec_is_binary(options->repo, p->one) ||
5782                     diff_filespec_is_binary(options->repo, p->two)) {
5783                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5784                                         GIT_SHA1_HEXSZ);
5785                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5786                                         GIT_SHA1_HEXSZ);
5787                         continue;
5788                 }
5789
5790                 xpp.flags = 0;
5791                 xecfg.ctxlen = 3;
5792                 xecfg.flags = 0;
5793                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5794                                   patch_id_consume, &data, &xpp, &xecfg))
5795                         return error("unable to generate patch-id diff for %s",
5796                                      p->one->path);
5797         }
5798
5799         git_SHA1_Final(oid->hash, &ctx);
5800         return 0;
5801 }
5802
5803 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5804 {
5805         struct diff_queue_struct *q = &diff_queued_diff;
5806         int i;
5807         int result = diff_get_patch_id(options, oid, diff_header_only);
5808
5809         for (i = 0; i < q->nr; i++)
5810                 diff_free_filepair(q->queue[i]);
5811
5812         free(q->queue);
5813         DIFF_QUEUE_CLEAR(q);
5814
5815         return result;
5816 }
5817
5818 static int is_summary_empty(const struct diff_queue_struct *q)
5819 {
5820         int i;
5821
5822         for (i = 0; i < q->nr; i++) {
5823                 const struct diff_filepair *p = q->queue[i];
5824
5825                 switch (p->status) {
5826                 case DIFF_STATUS_DELETED:
5827                 case DIFF_STATUS_ADDED:
5828                 case DIFF_STATUS_COPIED:
5829                 case DIFF_STATUS_RENAMED:
5830                         return 0;
5831                 default:
5832                         if (p->score)
5833                                 return 0;
5834                         if (p->one->mode && p->two->mode &&
5835                             p->one->mode != p->two->mode)
5836                                 return 0;
5837                         break;
5838                 }
5839         }
5840         return 1;
5841 }
5842
5843 static const char rename_limit_warning[] =
5844 N_("inexact rename detection was skipped due to too many files.");
5845
5846 static const char degrade_cc_to_c_warning[] =
5847 N_("only found copies from modified paths due to too many files.");
5848
5849 static const char rename_limit_advice[] =
5850 N_("you may want to set your %s variable to at least "
5851    "%d and retry the command.");
5852
5853 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5854 {
5855         fflush(stdout);
5856         if (degraded_cc)
5857                 warning(_(degrade_cc_to_c_warning));
5858         else if (needed)
5859                 warning(_(rename_limit_warning));
5860         else
5861                 return;
5862         if (0 < needed)
5863                 warning(_(rename_limit_advice), varname, needed);
5864 }
5865
5866 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5867 {
5868         int i;
5869         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5870         struct diff_queue_struct *q = &diff_queued_diff;
5871
5872         if (WSEH_NEW & WS_RULE_MASK)
5873                 BUG("WS rules bit mask overlaps with diff symbol flags");
5874
5875         if (o->color_moved)
5876                 o->emitted_symbols = &esm;
5877
5878         for (i = 0; i < q->nr; i++) {
5879                 struct diff_filepair *p = q->queue[i];
5880                 if (check_pair_status(p))
5881                         diff_flush_patch(p, o);
5882         }
5883
5884         if (o->emitted_symbols) {
5885                 if (o->color_moved) {
5886                         struct hashmap add_lines, del_lines;
5887
5888                         if (o->color_moved_ws_handling &
5889                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5890                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5891
5892                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5893                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5894
5895                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5896                         mark_color_as_moved(o, &add_lines, &del_lines);
5897                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5898                                 dim_moved_lines(o);
5899
5900                         hashmap_free(&add_lines, 1);
5901                         hashmap_free(&del_lines, 1);
5902                 }
5903
5904                 for (i = 0; i < esm.nr; i++)
5905                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5906
5907                 for (i = 0; i < esm.nr; i++)
5908                         free((void *)esm.buf[i].line);
5909         }
5910         esm.nr = 0;
5911 }
5912
5913 void diff_flush(struct diff_options *options)
5914 {
5915         struct diff_queue_struct *q = &diff_queued_diff;
5916         int i, output_format = options->output_format;
5917         int separator = 0;
5918         int dirstat_by_line = 0;
5919
5920         /*
5921          * Order: raw, stat, summary, patch
5922          * or:    name/name-status/checkdiff (other bits clear)
5923          */
5924         if (!q->nr)
5925                 goto free_queue;
5926
5927         if (output_format & (DIFF_FORMAT_RAW |
5928                              DIFF_FORMAT_NAME |
5929                              DIFF_FORMAT_NAME_STATUS |
5930                              DIFF_FORMAT_CHECKDIFF)) {
5931                 for (i = 0; i < q->nr; i++) {
5932                         struct diff_filepair *p = q->queue[i];
5933                         if (check_pair_status(p))
5934                                 flush_one_pair(p, options);
5935                 }
5936                 separator++;
5937         }
5938
5939         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5940                 dirstat_by_line = 1;
5941
5942         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5943             dirstat_by_line) {
5944                 struct diffstat_t diffstat;
5945
5946                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5947                 for (i = 0; i < q->nr; i++) {
5948                         struct diff_filepair *p = q->queue[i];
5949                         if (check_pair_status(p))
5950                                 diff_flush_stat(p, options, &diffstat);
5951                 }
5952                 if (output_format & DIFF_FORMAT_NUMSTAT)
5953                         show_numstat(&diffstat, options);
5954                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5955                         show_stats(&diffstat, options);
5956                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5957                         show_shortstats(&diffstat, options);
5958                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5959                         show_dirstat_by_line(&diffstat, options);
5960                 free_diffstat_info(&diffstat);
5961                 separator++;
5962         }
5963         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5964                 show_dirstat(options);
5965
5966         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5967                 for (i = 0; i < q->nr; i++) {
5968                         diff_summary(options, q->queue[i]);
5969                 }
5970                 separator++;
5971         }
5972
5973         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5974             options->flags.exit_with_status &&
5975             options->flags.diff_from_contents) {
5976                 /*
5977                  * run diff_flush_patch for the exit status. setting
5978                  * options->file to /dev/null should be safe, because we
5979                  * aren't supposed to produce any output anyway.
5980                  */
5981                 if (options->close_file)
5982                         fclose(options->file);
5983                 options->file = xfopen("/dev/null", "w");
5984                 options->close_file = 1;
5985                 options->color_moved = 0;
5986                 for (i = 0; i < q->nr; i++) {
5987                         struct diff_filepair *p = q->queue[i];
5988                         if (check_pair_status(p))
5989                                 diff_flush_patch(p, options);
5990                         if (options->found_changes)
5991                                 break;
5992                 }
5993         }
5994
5995         if (output_format & DIFF_FORMAT_PATCH) {
5996                 if (separator) {
5997                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5998                         if (options->stat_sep)
5999                                 /* attach patch instead of inline */
6000                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6001                                                  NULL, 0, 0);
6002                 }
6003
6004                 diff_flush_patch_all_file_pairs(options);
6005         }
6006
6007         if (output_format & DIFF_FORMAT_CALLBACK)
6008                 options->format_callback(q, options, options->format_callback_data);
6009
6010         for (i = 0; i < q->nr; i++)
6011                 diff_free_filepair(q->queue[i]);
6012 free_queue:
6013         free(q->queue);
6014         DIFF_QUEUE_CLEAR(q);
6015         if (options->close_file)
6016                 fclose(options->file);
6017
6018         /*
6019          * Report the content-level differences with HAS_CHANGES;
6020          * diff_addremove/diff_change does not set the bit when
6021          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6022          */
6023         if (options->flags.diff_from_contents) {
6024                 if (options->found_changes)
6025                         options->flags.has_changes = 1;
6026                 else
6027                         options->flags.has_changes = 0;
6028         }
6029 }
6030
6031 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6032 {
6033         return (((p->status == DIFF_STATUS_MODIFIED) &&
6034                  ((p->score &&
6035                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6036                   (!p->score &&
6037                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6038                 ((p->status != DIFF_STATUS_MODIFIED) &&
6039                  filter_bit_tst(p->status, options)));
6040 }
6041
6042 static void diffcore_apply_filter(struct diff_options *options)
6043 {
6044         int i;
6045         struct diff_queue_struct *q = &diff_queued_diff;
6046         struct diff_queue_struct outq;
6047
6048         DIFF_QUEUE_CLEAR(&outq);
6049
6050         if (!options->filter)
6051                 return;
6052
6053         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6054                 int found;
6055                 for (i = found = 0; !found && i < q->nr; i++) {
6056                         if (match_filter(options, q->queue[i]))
6057                                 found++;
6058                 }
6059                 if (found)
6060                         return;
6061
6062                 /* otherwise we will clear the whole queue
6063                  * by copying the empty outq at the end of this
6064                  * function, but first clear the current entries
6065                  * in the queue.
6066                  */
6067                 for (i = 0; i < q->nr; i++)
6068                         diff_free_filepair(q->queue[i]);
6069         }
6070         else {
6071                 /* Only the matching ones */
6072                 for (i = 0; i < q->nr; i++) {
6073                         struct diff_filepair *p = q->queue[i];
6074                         if (match_filter(options, p))
6075                                 diff_q(&outq, p);
6076                         else
6077                                 diff_free_filepair(p);
6078                 }
6079         }
6080         free(q->queue);
6081         *q = outq;
6082 }
6083
6084 /* Check whether two filespecs with the same mode and size are identical */
6085 static int diff_filespec_is_identical(struct repository *r,
6086                                       struct diff_filespec *one,
6087                                       struct diff_filespec *two)
6088 {
6089         if (S_ISGITLINK(one->mode))
6090                 return 0;
6091         if (diff_populate_filespec(r, one, 0))
6092                 return 0;
6093         if (diff_populate_filespec(r, two, 0))
6094                 return 0;
6095         return !memcmp(one->data, two->data, one->size);
6096 }
6097
6098 static int diff_filespec_check_stat_unmatch(struct repository *r,
6099                                             struct diff_filepair *p)
6100 {
6101         if (p->done_skip_stat_unmatch)
6102                 return p->skip_stat_unmatch_result;
6103
6104         p->done_skip_stat_unmatch = 1;
6105         p->skip_stat_unmatch_result = 0;
6106         /*
6107          * 1. Entries that come from stat info dirtiness
6108          *    always have both sides (iow, not create/delete),
6109          *    one side of the object name is unknown, with
6110          *    the same mode and size.  Keep the ones that
6111          *    do not match these criteria.  They have real
6112          *    differences.
6113          *
6114          * 2. At this point, the file is known to be modified,
6115          *    with the same mode and size, and the object
6116          *    name of one side is unknown.  Need to inspect
6117          *    the identical contents.
6118          */
6119         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6120             !DIFF_FILE_VALID(p->two) ||
6121             (p->one->oid_valid && p->two->oid_valid) ||
6122             (p->one->mode != p->two->mode) ||
6123             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6124             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6125             (p->one->size != p->two->size) ||
6126             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6127                 p->skip_stat_unmatch_result = 1;
6128         return p->skip_stat_unmatch_result;
6129 }
6130
6131 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6132 {
6133         int i;
6134         struct diff_queue_struct *q = &diff_queued_diff;
6135         struct diff_queue_struct outq;
6136         DIFF_QUEUE_CLEAR(&outq);
6137
6138         for (i = 0; i < q->nr; i++) {
6139                 struct diff_filepair *p = q->queue[i];
6140
6141                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6142                         diff_q(&outq, p);
6143                 else {
6144                         /*
6145                          * The caller can subtract 1 from skip_stat_unmatch
6146                          * to determine how many paths were dirty only
6147                          * due to stat info mismatch.
6148                          */
6149                         if (!diffopt->flags.no_index)
6150                                 diffopt->skip_stat_unmatch++;
6151                         diff_free_filepair(p);
6152                 }
6153         }
6154         free(q->queue);
6155         *q = outq;
6156 }
6157
6158 static int diffnamecmp(const void *a_, const void *b_)
6159 {
6160         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6161         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6162         const char *name_a, *name_b;
6163
6164         name_a = a->one ? a->one->path : a->two->path;
6165         name_b = b->one ? b->one->path : b->two->path;
6166         return strcmp(name_a, name_b);
6167 }
6168
6169 void diffcore_fix_diff_index(struct diff_options *options)
6170 {
6171         struct diff_queue_struct *q = &diff_queued_diff;
6172         QSORT(q->queue, q->nr, diffnamecmp);
6173 }
6174
6175 void diffcore_std(struct diff_options *options)
6176 {
6177         /* NOTE please keep the following in sync with diff_tree_combined() */
6178         if (options->skip_stat_unmatch)
6179                 diffcore_skip_stat_unmatch(options);
6180         if (!options->found_follow) {
6181                 /* See try_to_follow_renames() in tree-diff.c */
6182                 if (options->break_opt != -1)
6183                         diffcore_break(options->repo,
6184                                        options->break_opt);
6185                 if (options->detect_rename)
6186                         diffcore_rename(options);
6187                 if (options->break_opt != -1)
6188                         diffcore_merge_broken();
6189         }
6190         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6191                 diffcore_pickaxe(options);
6192         if (options->orderfile)
6193                 diffcore_order(options->orderfile);
6194         if (!options->found_follow)
6195                 /* See try_to_follow_renames() in tree-diff.c */
6196                 diff_resolve_rename_copy();
6197         diffcore_apply_filter(options);
6198
6199         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6200                 options->flags.has_changes = 1;
6201         else
6202                 options->flags.has_changes = 0;
6203
6204         options->found_follow = 0;
6205 }
6206
6207 int diff_result_code(struct diff_options *opt, int status)
6208 {
6209         int result = 0;
6210
6211         diff_warn_rename_limit("diff.renameLimit",
6212                                opt->needed_rename_limit,
6213                                opt->degraded_cc_to_c);
6214         if (!opt->flags.exit_with_status &&
6215             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6216                 return status;
6217         if (opt->flags.exit_with_status &&
6218             opt->flags.has_changes)
6219                 result |= 01;
6220         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6221             opt->flags.check_failed)
6222                 result |= 02;
6223         return result;
6224 }
6225
6226 int diff_can_quit_early(struct diff_options *opt)
6227 {
6228         return (opt->flags.quick &&
6229                 !opt->filter &&
6230                 opt->flags.has_changes);
6231 }
6232
6233 /*
6234  * Shall changes to this submodule be ignored?
6235  *
6236  * Submodule changes can be configured to be ignored separately for each path,
6237  * but that configuration can be overridden from the command line.
6238  */
6239 static int is_submodule_ignored(const char *path, struct diff_options *options)
6240 {
6241         int ignored = 0;
6242         struct diff_flags orig_flags = options->flags;
6243         if (!options->flags.override_submodule_config)
6244                 set_diffopt_flags_from_submodule_config(options, path);
6245         if (options->flags.ignore_submodules)
6246                 ignored = 1;
6247         options->flags = orig_flags;
6248         return ignored;
6249 }
6250
6251 void diff_addremove(struct diff_options *options,
6252                     int addremove, unsigned mode,
6253                     const struct object_id *oid,
6254                     int oid_valid,
6255                     const char *concatpath, unsigned dirty_submodule)
6256 {
6257         struct diff_filespec *one, *two;
6258
6259         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6260                 return;
6261
6262         /* This may look odd, but it is a preparation for
6263          * feeding "there are unchanged files which should
6264          * not produce diffs, but when you are doing copy
6265          * detection you would need them, so here they are"
6266          * entries to the diff-core.  They will be prefixed
6267          * with something like '=' or '*' (I haven't decided
6268          * which but should not make any difference).
6269          * Feeding the same new and old to diff_change()
6270          * also has the same effect.
6271          * Before the final output happens, they are pruned after
6272          * merged into rename/copy pairs as appropriate.
6273          */
6274         if (options->flags.reverse_diff)
6275                 addremove = (addremove == '+' ? '-' :
6276                              addremove == '-' ? '+' : addremove);
6277
6278         if (options->prefix &&
6279             strncmp(concatpath, options->prefix, options->prefix_length))
6280                 return;
6281
6282         one = alloc_filespec(concatpath);
6283         two = alloc_filespec(concatpath);
6284
6285         if (addremove != '+')
6286                 fill_filespec(one, oid, oid_valid, mode);
6287         if (addremove != '-') {
6288                 fill_filespec(two, oid, oid_valid, mode);
6289                 two->dirty_submodule = dirty_submodule;
6290         }
6291
6292         diff_queue(&diff_queued_diff, one, two);
6293         if (!options->flags.diff_from_contents)
6294                 options->flags.has_changes = 1;
6295 }
6296
6297 void diff_change(struct diff_options *options,
6298                  unsigned old_mode, unsigned new_mode,
6299                  const struct object_id *old_oid,
6300                  const struct object_id *new_oid,
6301                  int old_oid_valid, int new_oid_valid,
6302                  const char *concatpath,
6303                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6304 {
6305         struct diff_filespec *one, *two;
6306         struct diff_filepair *p;
6307
6308         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6309             is_submodule_ignored(concatpath, options))
6310                 return;
6311
6312         if (options->flags.reverse_diff) {
6313                 SWAP(old_mode, new_mode);
6314                 SWAP(old_oid, new_oid);
6315                 SWAP(old_oid_valid, new_oid_valid);
6316                 SWAP(old_dirty_submodule, new_dirty_submodule);
6317         }
6318
6319         if (options->prefix &&
6320             strncmp(concatpath, options->prefix, options->prefix_length))
6321                 return;
6322
6323         one = alloc_filespec(concatpath);
6324         two = alloc_filespec(concatpath);
6325         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6326         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6327         one->dirty_submodule = old_dirty_submodule;
6328         two->dirty_submodule = new_dirty_submodule;
6329         p = diff_queue(&diff_queued_diff, one, two);
6330
6331         if (options->flags.diff_from_contents)
6332                 return;
6333
6334         if (options->flags.quick && options->skip_stat_unmatch &&
6335             !diff_filespec_check_stat_unmatch(options->repo, p))
6336                 return;
6337
6338         options->flags.has_changes = 1;
6339 }
6340
6341 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6342 {
6343         struct diff_filepair *pair;
6344         struct diff_filespec *one, *two;
6345
6346         if (options->prefix &&
6347             strncmp(path, options->prefix, options->prefix_length))
6348                 return NULL;
6349
6350         one = alloc_filespec(path);
6351         two = alloc_filespec(path);
6352         pair = diff_queue(&diff_queued_diff, one, two);
6353         pair->is_unmerged = 1;
6354         return pair;
6355 }
6356
6357 static char *run_textconv(struct repository *r,
6358                           const char *pgm,
6359                           struct diff_filespec *spec,
6360                           size_t *outsize)
6361 {
6362         struct diff_tempfile *temp;
6363         const char *argv[3];
6364         const char **arg = argv;
6365         struct child_process child = CHILD_PROCESS_INIT;
6366         struct strbuf buf = STRBUF_INIT;
6367         int err = 0;
6368
6369         temp = prepare_temp_file(r, spec->path, spec);
6370         *arg++ = pgm;
6371         *arg++ = temp->name;
6372         *arg = NULL;
6373
6374         child.use_shell = 1;
6375         child.argv = argv;
6376         child.out = -1;
6377         if (start_command(&child)) {
6378                 remove_tempfile();
6379                 return NULL;
6380         }
6381
6382         if (strbuf_read(&buf, child.out, 0) < 0)
6383                 err = error("error reading from textconv command '%s'", pgm);
6384         close(child.out);
6385
6386         if (finish_command(&child) || err) {
6387                 strbuf_release(&buf);
6388                 remove_tempfile();
6389                 return NULL;
6390         }
6391         remove_tempfile();
6392
6393         return strbuf_detach(&buf, outsize);
6394 }
6395
6396 size_t fill_textconv(struct repository *r,
6397                      struct userdiff_driver *driver,
6398                      struct diff_filespec *df,
6399                      char **outbuf)
6400 {
6401         size_t size;
6402
6403         if (!driver) {
6404                 if (!DIFF_FILE_VALID(df)) {
6405                         *outbuf = "";
6406                         return 0;
6407                 }
6408                 if (diff_populate_filespec(r, df, 0))
6409                         die("unable to read files to diff");
6410                 *outbuf = df->data;
6411                 return df->size;
6412         }
6413
6414         if (!driver->textconv)
6415                 BUG("fill_textconv called with non-textconv driver");
6416
6417         if (driver->textconv_cache && df->oid_valid) {
6418                 *outbuf = notes_cache_get(driver->textconv_cache,
6419                                           &df->oid,
6420                                           &size);
6421                 if (*outbuf)
6422                         return size;
6423         }
6424
6425         *outbuf = run_textconv(r, driver->textconv, df, &size);
6426         if (!*outbuf)
6427                 die("unable to read files to diff");
6428
6429         if (driver->textconv_cache && df->oid_valid) {
6430                 /* ignore errors, as we might be in a readonly repository */
6431                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6432                                 size);
6433                 /*
6434                  * we could save up changes and flush them all at the end,
6435                  * but we would need an extra call after all diffing is done.
6436                  * Since generating a cache entry is the slow path anyway,
6437                  * this extra overhead probably isn't a big deal.
6438                  */
6439                 notes_cache_write(driver->textconv_cache);
6440         }
6441
6442         return size;
6443 }
6444
6445 int textconv_object(struct repository *r,
6446                     const char *path,
6447                     unsigned mode,
6448                     const struct object_id *oid,
6449                     int oid_valid,
6450                     char **buf,
6451                     unsigned long *buf_size)
6452 {
6453         struct diff_filespec *df;
6454         struct userdiff_driver *textconv;
6455
6456         df = alloc_filespec(path);
6457         fill_filespec(df, oid, oid_valid, mode);
6458         textconv = get_textconv(r->index, df);
6459         if (!textconv) {
6460                 free_filespec(df);
6461                 return 0;
6462         }
6463
6464         *buf_size = fill_textconv(r, textconv, df, buf);
6465         free_filespec(df);
6466         return 1;
6467 }
6468
6469 void setup_diff_pager(struct diff_options *opt)
6470 {
6471         /*
6472          * If the user asked for our exit code, then either they want --quiet
6473          * or --exit-code. We should definitely not bother with a pager in the
6474          * former case, as we will generate no output. Since we still properly
6475          * report our exit code even when a pager is run, we _could_ run a
6476          * pager with --exit-code. But since we have not done so historically,
6477          * and because it is easy to find people oneline advising "git diff
6478          * --exit-code" in hooks and other scripts, we do not do so.
6479          */
6480         if (!opt->flags.exit_with_status &&
6481             check_pager_config("diff") != 0)
6482                 setup_pager();
6483 }