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