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         int indent_off;   /* Offset to first non-whitespace character */
 756         int indent_width; /* The visual width of the indentation */
 757         enum diff_symbol s;
 758 };
 759 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 760
 761 struct emitted_diff_symbols {
 762         struct emitted_diff_symbol *buf;
 763         int nr, alloc;
 764 };
 765 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 766
 767 static void append_emitted_diff_symbol(struct diff_options *o,
 768                                        struct emitted_diff_symbol *e)
 769 {
 770         struct emitted_diff_symbol *f;
 771
 772         ALLOC_GROW(o->emitted_symbols->buf,
 773                    o->emitted_symbols->nr + 1,
 774                    o->emitted_symbols->alloc);
 775         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 776
 777         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 778         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 779 }
 780
 781 struct moved_entry {
 782         struct hashmap_entry ent;
 783         const struct emitted_diff_symbol *es;
 784         struct moved_entry *next_line;
 785 };
 786
 787 struct moved_block {
 788         struct moved_entry *match;
 789         int wsd; /* The whitespace delta of this block */
 790 };
 791
 792 static void moved_block_clear(struct moved_block *b)
 793 {
 794         memset(b, 0, sizeof(*b));
 795 }
 796
 797 #define INDENT_BLANKLINE INT_MIN
 798
 799 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 800 {
 801         unsigned int off = 0, i;
 802         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 803         const char *s = es->line;
 804         const int len = es->len;
 805
 806         /* skip any \v \f \r at start of indentation */
 807         while (s[off] == '\f' || s[off] == '\v' ||
 808                (s[off] == '\r' && off < len - 1))
 809                 off++;
 810
 811         /* calculate the visual width of indentation */
 812         while(1) {
 813                 if (s[off] == ' ') {
 814                         width++;
 815                         off++;
 816                 } else if (s[off] == '\t') {
 817                         width += tab_width - (width % tab_width);
 818                         while (s[++off] == '\t')
 819                                 width += tab_width;
 820                 } else {
 821                         break;
 822                 }
 823         }
 824
 825         /* check if this line is blank */
 826         for (i = off; i < len; i++)
 827                 if (!isspace(s[i]))
 828                     break;
 829
 830         if (i == len) {
 831                 es->indent_width = INDENT_BLANKLINE;
 832                 es->indent_off = len;
 833         } else {
 834                 es->indent_off = off;
 835                 es->indent_width = width;
 836         }
 837 }
 838
 839 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 840                             const struct emitted_diff_symbol *b,
 841                             int *out)
 842 {
 843         int a_len = a->len,
 844             b_len = b->len,
 845             a_off = a->indent_off,
 846             a_width = a->indent_width,
 847             b_off = b->indent_off,
 848             b_width = b->indent_width;
 849         int delta;
 850
 851         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 852                 *out = INDENT_BLANKLINE;
 853                 return 1;
 854         }
 855
 856         if (a->s == DIFF_SYMBOL_PLUS)
 857                 delta = a_width - b_width;
 858         else
 859                 delta = b_width - a_width;
 860
 861         if (a_len - a_off != b_len - b_off ||
 862             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 863                 return 0;
 864
 865         *out = delta;
 866
 867         return 1;
 868 }
 869
 870 static int cmp_in_block_with_wsd(const struct diff_options *o,
 871                                  const struct moved_entry *cur,
 872                                  const struct moved_entry *match,
 873                                  struct moved_block *pmb,
 874                                  int n)
 875 {
 876         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 877         int al = cur->es->len, bl = match->es->len, cl = l->len;
 878         const char *a = cur->es->line,
 879                    *b = match->es->line,
 880                    *c = l->line;
 881         int a_off = cur->es->indent_off,
 882             a_width = cur->es->indent_width,
 883             c_off = l->indent_off,
 884             c_width = l->indent_width;
 885         int delta;
 886
 887         /*
 888          * We need to check if 'cur' is equal to 'match'.  As those
 889          * are from the same (+/-) side, we do not need to adjust for
 890          * indent changes. However these were found using fuzzy
 891          * matching so we do have to check if they are equal. Here we
 892          * just check the lengths. We delay calling memcmp() to check
 893          * the contents until later as if the length comparison for a
 894          * and c fails we can avoid the call all together.
 895          */
 896         if (al != bl)
 897                 return 1;
 898
 899         /* If 'l' and 'cur' are both blank then they match. */
 900         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 901                 return 0;
 902
 903         /*
 904          * The indent changes of the block are known and stored in pmb->wsd;
 905          * however we need to check if the indent changes of the current line
 906          * match those of the current block and that the text of 'l' and 'cur'
 907          * after the indentation match.
 908          */
 909         if (cur->es->s == DIFF_SYMBOL_PLUS)
 910                 delta = a_width - c_width;
 911         else
 912                 delta = c_width - a_width;
 913
 914         /*
 915          * If the previous lines of this block were all blank then set its
 916          * whitespace delta.
 917          */
 918         if (pmb->wsd == INDENT_BLANKLINE)
 919                 pmb->wsd = delta;
 920
 921         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 922                  !memcmp(a, b, al) && !
 923                  memcmp(a + a_off, c + c_off, al - a_off));
 924 }
 925
 926 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 927                            const void *entry,
 928                            const void *entry_or_key,
 929                            const void *keydata)
 930 {
 931         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 932         const struct moved_entry *a = entry;
 933         const struct moved_entry *b = entry_or_key;
 934         unsigned flags = diffopt->color_moved_ws_handling
 935                          & XDF_WHITESPACE_FLAGS;
 936
 937         if (diffopt->color_moved_ws_handling &
 938             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 939                 /*
 940                  * As there is not specific white space config given,
 941                  * we'd need to check for a new block, so ignore all
 942                  * white space. The setup of the white space
 943                  * configuration for the next block is done else where
 944                  */
 945                 flags |= XDF_IGNORE_WHITESPACE;
 946
 947         return !xdiff_compare_lines(a->es->line, a->es->len,
 948                                     b->es->line, b->es->len,
 949                                     flags);
 950 }
 951
 952 static struct moved_entry *prepare_entry(struct diff_options *o,
 953                                          int line_no)
 954 {
 955         struct moved_entry *ret = xmalloc(sizeof(*ret));
 956         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 957         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 958
 959         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 960         ret->es = l;
 961         ret->next_line = NULL;
 962
 963         return ret;
 964 }
 965
 966 static void add_lines_to_move_detection(struct diff_options *o,
 967                                         struct hashmap *add_lines,
 968                                         struct hashmap *del_lines)
 969 {
 970         struct moved_entry *prev_line = NULL;
 971
 972         int n;
 973         for (n = 0; n < o->emitted_symbols->nr; n++) {
 974                 struct hashmap *hm;
 975                 struct moved_entry *key;
 976
 977                 switch (o->emitted_symbols->buf[n].s) {
 978                 case DIFF_SYMBOL_PLUS:
 979                         hm = add_lines;
 980                         break;
 981                 case DIFF_SYMBOL_MINUS:
 982                         hm = del_lines;
 983                         break;
 984                 default:
 985                         prev_line = NULL;
 986                         continue;
 987                 }
 988
 989                 if (o->color_moved_ws_handling &
 990                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 991                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
 992                 key = prepare_entry(o, n);
 993                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 994                         prev_line->next_line = key;
 995
 996                 hashmap_add(hm, key);
 997                 prev_line = key;
 998         }
 999 }
1000
1001 static void pmb_advance_or_null(struct diff_options *o,
1002                                 struct moved_entry *match,
1003                                 struct hashmap *hm,
1004                                 struct moved_block *pmb,
1005                                 int pmb_nr)
1006 {
1007         int i;
1008         for (i = 0; i < pmb_nr; i++) {
1009                 struct moved_entry *prev = pmb[i].match;
1010                 struct moved_entry *cur = (prev && prev->next_line) ?
1011                                 prev->next_line : NULL;
1012                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1013                         pmb[i].match = cur;
1014                 } else {
1015                         pmb[i].match = NULL;
1016                 }
1017         }
1018 }
1019
1020 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1021                                             struct moved_entry *match,
1022                                             struct hashmap *hm,
1023                                             struct moved_block *pmb,
1024                                             int pmb_nr, int n)
1025 {
1026         int i;
1027         char *got_match = xcalloc(1, pmb_nr);
1028
1029         for (; match; match = hashmap_get_next(hm, match)) {
1030                 for (i = 0; i < pmb_nr; i++) {
1031                         struct moved_entry *prev = pmb[i].match;
1032                         struct moved_entry *cur = (prev && prev->next_line) ?
1033                                         prev->next_line : NULL;
1034                         if (!cur)
1035                                 continue;
1036                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1037                                 got_match[i] |= 1;
1038                 }
1039         }
1040
1041         for (i = 0; i < pmb_nr; i++) {
1042                 if (got_match[i]) {
1043                         /* Advance to the next line */
1044                         pmb[i].match = pmb[i].match->next_line;
1045                 } else {
1046                         moved_block_clear(&pmb[i]);
1047                 }
1048         }
1049
1050         free(got_match);
1051 }
1052
1053 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1054                                          int pmb_nr)
1055 {
1056         int lp, rp;
1057
1058         /* Shrink the set of potential block to the remaining running */
1059         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1060                 while (lp < pmb_nr && pmb[lp].match)
1061                         lp++;
1062                 /* lp points at the first NULL now */
1063
1064                 while (rp > -1 && !pmb[rp].match)
1065                         rp--;
1066                 /* rp points at the last non-NULL */
1067
1068                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1069                         pmb[lp] = pmb[rp];
1070                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1071                         rp--;
1072                         lp++;
1073                 }
1074         }
1075
1076         /* Remember the number of running sets */
1077         return rp + 1;
1078 }
1079
1080 /*
1081  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1082  *
1083  * Otherwise, if the last block has fewer alphanumeric characters than
1084  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1085  * that block.
1086  *
1087  * The last block consists of the (n - block_length)'th line up to but not
1088  * including the nth line.
1089  *
1090  * Returns 0 if the last block is empty or is unset by this function, non zero
1091  * otherwise.
1092  *
1093  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1094  * Think of a way to unify them.
1095  */
1096 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1097 {
1098         int i, alnum_count = 0;
1099         if (o->color_moved == COLOR_MOVED_PLAIN)
1100                 return block_length;
1101         for (i = 1; i < block_length + 1; i++) {
1102                 const char *c = o->emitted_symbols->buf[n - i].line;
1103                 for (; *c; c++) {
1104                         if (!isalnum(*c))
1105                                 continue;
1106                         alnum_count++;
1107                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1108                                 return 1;
1109                 }
1110         }
1111         for (i = 1; i < block_length + 1; i++)
1112                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1113         return 0;
1114 }
1115
1116 /* Find blocks of moved code, delegate actual coloring decision to helper */
1117 static void mark_color_as_moved(struct diff_options *o,
1118                                 struct hashmap *add_lines,
1119                                 struct hashmap *del_lines)
1120 {
1121         struct moved_block *pmb = NULL; /* potentially moved blocks */
1122         int pmb_nr = 0, pmb_alloc = 0;
1123         int n, flipped_block = 0, block_length = 0;
1124
1125
1126         for (n = 0; n < o->emitted_symbols->nr; n++) {
1127                 struct hashmap *hm = NULL;
1128                 struct moved_entry *key;
1129                 struct moved_entry *match = NULL;
1130                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1131                 enum diff_symbol last_symbol = 0;
1132
1133                 switch (l->s) {
1134                 case DIFF_SYMBOL_PLUS:
1135                         hm = del_lines;
1136                         key = prepare_entry(o, n);
1137                         match = hashmap_get(hm, key, NULL);
1138                         free(key);
1139                         break;
1140                 case DIFF_SYMBOL_MINUS:
1141                         hm = add_lines;
1142                         key = prepare_entry(o, n);
1143                         match = hashmap_get(hm, key, NULL);
1144                         free(key);
1145                         break;
1146                 default:
1147                         flipped_block = 0;
1148                 }
1149
1150                 if (!match) {
1151                         int i;
1152
1153                         adjust_last_block(o, n, block_length);
1154                         for(i = 0; i < pmb_nr; i++)
1155                                 moved_block_clear(&pmb[i]);
1156                         pmb_nr = 0;
1157                         block_length = 0;
1158                         flipped_block = 0;
1159                         last_symbol = l->s;
1160                         continue;
1161                 }
1162
1163                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1164                         last_symbol = l->s;
1165                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1166                         continue;
1167                 }
1168
1169                 if (o->color_moved_ws_handling &
1170                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1171                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1172                 else
1173                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1174
1175                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1176
1177                 if (pmb_nr == 0) {
1178                         /*
1179                          * The current line is the start of a new block.
1180                          * Setup the set of potential blocks.
1181                          */
1182                         for (; match; match = hashmap_get_next(hm, match)) {
1183                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1184                                 if (o->color_moved_ws_handling &
1185                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1186                                         if (compute_ws_delta(l, match->es,
1187                                                              &pmb[pmb_nr].wsd))
1188                                                 pmb[pmb_nr++].match = match;
1189                                 } else {
1190                                         pmb[pmb_nr].wsd = 0;
1191                                         pmb[pmb_nr++].match = match;
1192                                 }
1193                         }
1194
1195                         if (adjust_last_block(o, n, block_length) &&
1196                             pmb_nr && last_symbol != l->s)
1197                                 flipped_block = (flipped_block + 1) % 2;
1198                         else
1199                                 flipped_block = 0;
1200
1201                         block_length = 0;
1202                 }
1203
1204                 if (pmb_nr) {
1205                         block_length++;
1206                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1207                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1208                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1209                 }
1210                 last_symbol = l->s;
1211         }
1212         adjust_last_block(o, n, block_length);
1213
1214         for(n = 0; n < pmb_nr; n++)
1215                 moved_block_clear(&pmb[n]);
1216         free(pmb);
1217 }
1218
1219 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1220   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1221 static void dim_moved_lines(struct diff_options *o)
1222 {
1223         int n;
1224         for (n = 0; n < o->emitted_symbols->nr; n++) {
1225                 struct emitted_diff_symbol *prev = (n != 0) ?
1226                                 &o->emitted_symbols->buf[n - 1] : NULL;
1227                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1228                 struct emitted_diff_symbol *next =
1229                                 (n < o->emitted_symbols->nr - 1) ?
1230                                 &o->emitted_symbols->buf[n + 1] : NULL;
1231
1232                 /* Not a plus or minus line? */
1233                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1234                         continue;
1235
1236                 /* Not a moved line? */
1237                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1238                         continue;
1239
1240                 /*
1241                  * If prev or next are not a plus or minus line,
1242                  * pretend they don't exist
1243                  */
1244                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1245                             prev->s != DIFF_SYMBOL_MINUS)
1246                         prev = NULL;
1247                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1248                             next->s != DIFF_SYMBOL_MINUS)
1249                         next = NULL;
1250
1251                 /* Inside a block? */
1252                 if ((prev &&
1253                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1254                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1255                     (next &&
1256                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1257                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1258                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1259                         continue;
1260                 }
1261
1262                 /* Check if we are at an interesting bound: */
1263                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1264                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1265                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1266                         continue;
1267                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1268                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1269                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1270                         continue;
1271
1272                 /*
1273                  * The boundary to prev and next are not interesting,
1274                  * so this line is not interesting as a whole
1275                  */
1276                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1277         }
1278 }
1279
1280 static void emit_line_ws_markup(struct diff_options *o,
1281                                 const char *set_sign, const char *set,
1282                                 const char *reset,
1283                                 int sign_index, const char *line, int len,
1284                                 unsigned ws_rule, int blank_at_eof)
1285 {
1286         const char *ws = NULL;
1287         int sign = o->output_indicators[sign_index];
1288
1289         if (o->ws_error_highlight & ws_rule) {
1290                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1291                 if (!*ws)
1292                         ws = NULL;
1293         }
1294
1295         if (!ws && !set_sign)
1296                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1297         else if (!ws) {
1298                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1299         } else if (blank_at_eof)
1300                 /* Blank line at EOF - paint '+' as well */
1301                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1302         else {
1303                 /* Emit just the prefix, then the rest. */
1304                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1305                             sign, "", 0);
1306                 ws_check_emit(line, len, ws_rule,
1307                               o->file, set, reset, ws);
1308         }
1309 }
1310
1311 static void emit_diff_symbol_from_struct(struct diff_options *o,
1312                                          struct emitted_diff_symbol *eds)
1313 {
1314         static const char *nneof = " No newline at end of file\n";
1315         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1316         struct strbuf sb = STRBUF_INIT;
1317
1318         enum diff_symbol s = eds->s;
1319         const char *line = eds->line;
1320         int len = eds->len;
1321         unsigned flags = eds->flags;
1322
1323         switch (s) {
1324         case DIFF_SYMBOL_NO_LF_EOF:
1325                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1326                 reset = diff_get_color_opt(o, DIFF_RESET);
1327                 putc('\n', o->file);
1328                 emit_line_0(o, context, NULL, 0, reset, '\\',
1329                             nneof, strlen(nneof));
1330                 break;
1331         case DIFF_SYMBOL_SUBMODULE_HEADER:
1332         case DIFF_SYMBOL_SUBMODULE_ERROR:
1333         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1334         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1335         case DIFF_SYMBOL_SUMMARY:
1336         case DIFF_SYMBOL_STATS_LINE:
1337         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1338         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1339                 emit_line(o, "", "", line, len);
1340                 break;
1341         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1342         case DIFF_SYMBOL_CONTEXT_MARKER:
1343                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1344                 reset = diff_get_color_opt(o, DIFF_RESET);
1345                 emit_line(o, context, reset, line, len);
1346                 break;
1347         case DIFF_SYMBOL_SEPARATOR:
1348                 fprintf(o->file, "%s%c",
1349                         diff_line_prefix(o),
1350                         o->line_termination);
1351                 break;
1352         case DIFF_SYMBOL_CONTEXT:
1353                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1354                 reset = diff_get_color_opt(o, DIFF_RESET);
1355                 set_sign = NULL;
1356                 if (o->flags.dual_color_diffed_diffs) {
1357                         char c = !len ? 0 : line[0];
1358
1359                         if (c == '+')
1360                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1361                         else if (c == '@')
1362                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1363                         else if (c == '-')
1364                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1365                 }
1366                 emit_line_ws_markup(o, set_sign, set, reset,
1367                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1368                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1369                 break;
1370         case DIFF_SYMBOL_PLUS:
1371                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1372                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1373                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1374                 case DIFF_SYMBOL_MOVED_LINE |
1375                      DIFF_SYMBOL_MOVED_LINE_ALT |
1376                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1377                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1378                         break;
1379                 case DIFF_SYMBOL_MOVED_LINE |
1380                      DIFF_SYMBOL_MOVED_LINE_ALT:
1381                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1382                         break;
1383                 case DIFF_SYMBOL_MOVED_LINE |
1384                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1385                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1386                         break;
1387                 case DIFF_SYMBOL_MOVED_LINE:
1388                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1389                         break;
1390                 default:
1391                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1392                 }
1393                 reset = diff_get_color_opt(o, DIFF_RESET);
1394                 if (!o->flags.dual_color_diffed_diffs)
1395                         set_sign = NULL;
1396                 else {
1397                         char c = !len ? 0 : line[0];
1398
1399                         set_sign = set;
1400                         if (c == '-')
1401                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1402                         else if (c == '@')
1403                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1404                         else if (c == '+')
1405                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1406                         else
1407                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1408                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1409                 }
1410                 emit_line_ws_markup(o, set_sign, set, reset,
1411                                     OUTPUT_INDICATOR_NEW, line, len,
1412                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1413                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1414                 break;
1415         case DIFF_SYMBOL_MINUS:
1416                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1417                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1418                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1419                 case DIFF_SYMBOL_MOVED_LINE |
1420                      DIFF_SYMBOL_MOVED_LINE_ALT |
1421                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1422                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1423                         break;
1424                 case DIFF_SYMBOL_MOVED_LINE |
1425                      DIFF_SYMBOL_MOVED_LINE_ALT:
1426                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1427                         break;
1428                 case DIFF_SYMBOL_MOVED_LINE |
1429                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1430                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1431                         break;
1432                 case DIFF_SYMBOL_MOVED_LINE:
1433                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1434                         break;
1435                 default:
1436                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1437                 }
1438                 reset = diff_get_color_opt(o, DIFF_RESET);
1439                 if (!o->flags.dual_color_diffed_diffs)
1440                         set_sign = NULL;
1441                 else {
1442                         char c = !len ? 0 : line[0];
1443
1444                         set_sign = set;
1445                         if (c == '+')
1446                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1447                         else if (c == '@')
1448                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1449                         else if (c == '-')
1450                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1451                         else
1452                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1453                 }
1454                 emit_line_ws_markup(o, set_sign, set, reset,
1455                                     OUTPUT_INDICATOR_OLD, line, len,
1456                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1457                 break;
1458         case DIFF_SYMBOL_WORDS_PORCELAIN:
1459                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1460                 reset = diff_get_color_opt(o, DIFF_RESET);
1461                 emit_line(o, context, reset, line, len);
1462                 fputs("~\n", o->file);
1463                 break;
1464         case DIFF_SYMBOL_WORDS:
1465                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1466                 reset = diff_get_color_opt(o, DIFF_RESET);
1467                 /*
1468                  * Skip the prefix character, if any.  With
1469                  * diff_suppress_blank_empty, there may be
1470                  * none.
1471                  */
1472                 if (line[0] != '\n') {
1473                         line++;
1474                         len--;
1475                 }
1476                 emit_line(o, context, reset, line, len);
1477                 break;
1478         case DIFF_SYMBOL_FILEPAIR_PLUS:
1479                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1480                 reset = diff_get_color_opt(o, DIFF_RESET);
1481                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1482                         line, reset,
1483                         strchr(line, ' ') ? "\t" : "");
1484                 break;
1485         case DIFF_SYMBOL_FILEPAIR_MINUS:
1486                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1487                 reset = diff_get_color_opt(o, DIFF_RESET);
1488                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1489                         line, reset,
1490                         strchr(line, ' ') ? "\t" : "");
1491                 break;
1492         case DIFF_SYMBOL_BINARY_FILES:
1493         case DIFF_SYMBOL_HEADER:
1494                 fprintf(o->file, "%s", line);
1495                 break;
1496         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1497                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1498                 break;
1499         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1500                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1501                 break;
1502         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1503                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1504                 break;
1505         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1506                 fputs(diff_line_prefix(o), o->file);
1507                 fputc('\n', o->file);
1508                 break;
1509         case DIFF_SYMBOL_REWRITE_DIFF:
1510                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1511                 reset = diff_get_color_opt(o, DIFF_RESET);
1512                 emit_line(o, fraginfo, reset, line, len);
1513                 break;
1514         case DIFF_SYMBOL_SUBMODULE_ADD:
1515                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1516                 reset = diff_get_color_opt(o, DIFF_RESET);
1517                 emit_line(o, set, reset, line, len);
1518                 break;
1519         case DIFF_SYMBOL_SUBMODULE_DEL:
1520                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1521                 reset = diff_get_color_opt(o, DIFF_RESET);
1522                 emit_line(o, set, reset, line, len);
1523                 break;
1524         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1525                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1526                         diff_line_prefix(o), line);
1527                 break;
1528         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1529                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1530                         diff_line_prefix(o), line);
1531                 break;
1532         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1533                 emit_line(o, "", "", " 0 files changed\n",
1534                           strlen(" 0 files changed\n"));
1535                 break;
1536         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1537                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1538                 break;
1539         case DIFF_SYMBOL_WORD_DIFF:
1540                 fprintf(o->file, "%.*s", len, line);
1541                 break;
1542         case DIFF_SYMBOL_STAT_SEP:
1543                 fputs(o->stat_sep, o->file);
1544                 break;
1545         default:
1546                 BUG("unknown diff symbol");
1547         }
1548         strbuf_release(&sb);
1549 }
1550
1551 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1552                              const char *line, int len, unsigned flags)
1553 {
1554         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1555
1556         if (o->emitted_symbols)
1557                 append_emitted_diff_symbol(o, &e);
1558         else
1559                 emit_diff_symbol_from_struct(o, &e);
1560 }
1561
1562 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1563 {
1564         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1565 }
1566
1567 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1568 {
1569         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1570 }
1571
1572 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1573 {
1574         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1575                          path, strlen(path), 0);
1576 }
1577
1578 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1579 {
1580         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1581                          path, strlen(path), 0);
1582 }
1583
1584 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1585 {
1586         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1587                          header, strlen(header), 0);
1588 }
1589
1590 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1591 {
1592         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1593 }
1594
1595 void diff_emit_submodule_pipethrough(struct diff_options *o,
1596                                      const char *line, int len)
1597 {
1598         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1599 }
1600
1601 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1602 {
1603         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1604               ecbdata->blank_at_eof_in_preimage &&
1605               ecbdata->blank_at_eof_in_postimage &&
1606               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1607               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1608                 return 0;
1609         return ws_blank_line(line, len, ecbdata->ws_rule);
1610 }
1611
1612 static void emit_add_line(const char *reset,
1613                           struct emit_callback *ecbdata,
1614                           const char *line, int len)
1615 {
1616         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1617         if (new_blank_line_at_eof(ecbdata, line, len))
1618                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1619
1620         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1621 }
1622
1623 static void emit_del_line(const char *reset,
1624                           struct emit_callback *ecbdata,
1625                           const char *line, int len)
1626 {
1627         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1628         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1629 }
1630
1631 static void emit_context_line(const char *reset,
1632                               struct emit_callback *ecbdata,
1633                               const char *line, int len)
1634 {
1635         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1636         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1637 }
1638
1639 static void emit_hunk_header(struct emit_callback *ecbdata,
1640                              const char *line, int len)
1641 {
1642         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1643         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1644         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1645         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1646         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1647         static const char atat[2] = { '@', '@' };
1648         const char *cp, *ep;
1649         struct strbuf msgbuf = STRBUF_INIT;
1650         int org_len = len;
1651         int i = 1;
1652
1653         /*
1654          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1655          * it always is at least 10 bytes long.
1656          */
1657         if (len < 10 ||
1658             memcmp(line, atat, 2) ||
1659             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1660                 emit_diff_symbol(ecbdata->opt,
1661                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1662                 return;
1663         }
1664         ep += 2; /* skip over @@ */
1665
1666         /* The hunk header in fraginfo color */
1667         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1668                 strbuf_addstr(&msgbuf, reverse);
1669         strbuf_addstr(&msgbuf, frag);
1670         strbuf_add(&msgbuf, line, ep - line);
1671         strbuf_addstr(&msgbuf, reset);
1672
1673         /*
1674          * trailing "\r\n"
1675          */
1676         for ( ; i < 3; i++)
1677                 if (line[len - i] == '\r' || line[len - i] == '\n')
1678                         len--;
1679
1680         /* blank before the func header */
1681         for (cp = ep; ep - line < len; ep++)
1682                 if (*ep != ' ' && *ep != '\t')
1683                         break;
1684         if (ep != cp) {
1685                 strbuf_addstr(&msgbuf, context);
1686                 strbuf_add(&msgbuf, cp, ep - cp);
1687                 strbuf_addstr(&msgbuf, reset);
1688         }
1689
1690         if (ep < line + len) {
1691                 strbuf_addstr(&msgbuf, func);
1692                 strbuf_add(&msgbuf, ep, line + len - ep);
1693                 strbuf_addstr(&msgbuf, reset);
1694         }
1695
1696         strbuf_add(&msgbuf, line + len, org_len - len);
1697         strbuf_complete_line(&msgbuf);
1698         emit_diff_symbol(ecbdata->opt,
1699                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1700         strbuf_release(&msgbuf);
1701 }
1702
1703 static struct diff_tempfile *claim_diff_tempfile(void) {
1704         int i;
1705         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1706                 if (!diff_temp[i].name)
1707                         return diff_temp + i;
1708         BUG("diff is failing to clean up its tempfiles");
1709 }
1710
1711 static void remove_tempfile(void)
1712 {
1713         int i;
1714         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1715                 if (is_tempfile_active(diff_temp[i].tempfile))
1716                         delete_tempfile(&diff_temp[i].tempfile);
1717                 diff_temp[i].name = NULL;
1718         }
1719 }
1720
1721 static void add_line_count(struct strbuf *out, int count)
1722 {
1723         switch (count) {
1724         case 0:
1725                 strbuf_addstr(out, "0,0");
1726                 break;
1727         case 1:
1728                 strbuf_addstr(out, "1");
1729                 break;
1730         default:
1731                 strbuf_addf(out, "1,%d", count);
1732                 break;
1733         }
1734 }
1735
1736 static void emit_rewrite_lines(struct emit_callback *ecb,
1737                                int prefix, const char *data, int size)
1738 {
1739         const char *endp = NULL;
1740         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1741
1742         while (0 < size) {
1743                 int len;
1744
1745                 endp = memchr(data, '\n', size);
1746                 len = endp ? (endp - data + 1) : size;
1747                 if (prefix != '+') {
1748                         ecb->lno_in_preimage++;
1749                         emit_del_line(reset, ecb, data, len);
1750                 } else {
1751                         ecb->lno_in_postimage++;
1752                         emit_add_line(reset, ecb, data, len);
1753                 }
1754                 size -= len;
1755                 data += len;
1756         }
1757         if (!endp)
1758                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1759 }
1760
1761 static void emit_rewrite_diff(const char *name_a,
1762                               const char *name_b,
1763                               struct diff_filespec *one,
1764                               struct diff_filespec *two,
1765                               struct userdiff_driver *textconv_one,
1766                               struct userdiff_driver *textconv_two,
1767                               struct diff_options *o)
1768 {
1769         int lc_a, lc_b;
1770         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1771         const char *a_prefix, *b_prefix;
1772         char *data_one, *data_two;
1773         size_t size_one, size_two;
1774         struct emit_callback ecbdata;
1775         struct strbuf out = STRBUF_INIT;
1776
1777         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1778                 a_prefix = o->b_prefix;
1779                 b_prefix = o->a_prefix;
1780         } else {
1781                 a_prefix = o->a_prefix;
1782                 b_prefix = o->b_prefix;
1783         }
1784
1785         name_a += (*name_a == '/');
1786         name_b += (*name_b == '/');
1787
1788         strbuf_reset(&a_name);
1789         strbuf_reset(&b_name);
1790         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1791         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1792
1793         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1794         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1795
1796         memset(&ecbdata, 0, sizeof(ecbdata));
1797         ecbdata.color_diff = want_color(o->use_color);
1798         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1799         ecbdata.opt = o;
1800         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1801                 mmfile_t mf1, mf2;
1802                 mf1.ptr = (char *)data_one;
1803                 mf2.ptr = (char *)data_two;
1804                 mf1.size = size_one;
1805                 mf2.size = size_two;
1806                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1807         }
1808         ecbdata.lno_in_preimage = 1;
1809         ecbdata.lno_in_postimage = 1;
1810
1811         lc_a = count_lines(data_one, size_one);
1812         lc_b = count_lines(data_two, size_two);
1813
1814         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1815                          a_name.buf, a_name.len, 0);
1816         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1817                          b_name.buf, b_name.len, 0);
1818
1819         strbuf_addstr(&out, "@@ -");
1820         if (!o->irreversible_delete)
1821                 add_line_count(&out, lc_a);
1822         else
1823                 strbuf_addstr(&out, "?,?");
1824         strbuf_addstr(&out, " +");
1825         add_line_count(&out, lc_b);
1826         strbuf_addstr(&out, " @@\n");
1827         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1828         strbuf_release(&out);
1829
1830         if (lc_a && !o->irreversible_delete)
1831                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1832         if (lc_b)
1833                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1834         if (textconv_one)
1835                 free((char *)data_one);
1836         if (textconv_two)
1837                 free((char *)data_two);
1838 }
1839
1840 struct diff_words_buffer {
1841         mmfile_t text;
1842         unsigned long alloc;
1843         struct diff_words_orig {
1844                 const char *begin, *end;
1845         } *orig;
1846         int orig_nr, orig_alloc;
1847 };
1848
1849 static void diff_words_append(char *line, unsigned long len,
1850                 struct diff_words_buffer *buffer)
1851 {
1852         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1853         line++;
1854         len--;
1855         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1856         buffer->text.size += len;
1857         buffer->text.ptr[buffer->text.size] = '\0';
1858 }
1859
1860 struct diff_words_style_elem {
1861         const char *prefix;
1862         const char *suffix;
1863         const char *color; /* NULL; filled in by the setup code if
1864                             * color is enabled */
1865 };
1866
1867 struct diff_words_style {
1868         enum diff_words_type type;
1869         struct diff_words_style_elem new_word, old_word, ctx;
1870         const char *newline;
1871 };
1872
1873 static struct diff_words_style diff_words_styles[] = {
1874         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1875         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1876         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1877 };
1878
1879 struct diff_words_data {
1880         struct diff_words_buffer minus, plus;
1881         const char *current_plus;
1882         int last_minus;
1883         struct diff_options *opt;
1884         regex_t *word_regex;
1885         enum diff_words_type type;
1886         struct diff_words_style *style;
1887 };
1888
1889 static int fn_out_diff_words_write_helper(struct diff_options *o,
1890                                           struct diff_words_style_elem *st_el,
1891                                           const char *newline,
1892                                           size_t count, const char *buf)
1893 {
1894         int print = 0;
1895         struct strbuf sb = STRBUF_INIT;
1896
1897         while (count) {
1898                 char *p = memchr(buf, '\n', count);
1899                 if (print)
1900                         strbuf_addstr(&sb, diff_line_prefix(o));
1901
1902                 if (p != buf) {
1903                         const char *reset = st_el->color && *st_el->color ?
1904                                             GIT_COLOR_RESET : NULL;
1905                         if (st_el->color && *st_el->color)
1906                                 strbuf_addstr(&sb, st_el->color);
1907                         strbuf_addstr(&sb, st_el->prefix);
1908                         strbuf_add(&sb, buf, p ? p - buf : count);
1909                         strbuf_addstr(&sb, st_el->suffix);
1910                         if (reset)
1911                                 strbuf_addstr(&sb, reset);
1912                 }
1913                 if (!p)
1914                         goto out;
1915
1916                 strbuf_addstr(&sb, newline);
1917                 count -= p + 1 - buf;
1918                 buf = p + 1;
1919                 print = 1;
1920                 if (count) {
1921                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1922                                          sb.buf, sb.len, 0);
1923                         strbuf_reset(&sb);
1924                 }
1925         }
1926
1927 out:
1928         if (sb.len)
1929                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1930                                  sb.buf, sb.len, 0);
1931         strbuf_release(&sb);
1932         return 0;
1933 }
1934
1935 /*
1936  * '--color-words' algorithm can be described as:
1937  *
1938  *   1. collect the minus/plus lines of a diff hunk, divided into
1939  *      minus-lines and plus-lines;
1940  *
1941  *   2. break both minus-lines and plus-lines into words and
1942  *      place them into two mmfile_t with one word for each line;
1943  *
1944  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1945  *
1946  * And for the common parts of the both file, we output the plus side text.
1947  * diff_words->current_plus is used to trace the current position of the plus file
1948  * which printed. diff_words->last_minus is used to trace the last minus word
1949  * printed.
1950  *
1951  * For '--graph' to work with '--color-words', we need to output the graph prefix
1952  * on each line of color words output. Generally, there are two conditions on
1953  * which we should output the prefix.
1954  *
1955  *   1. diff_words->last_minus == 0 &&
1956  *      diff_words->current_plus == diff_words->plus.text.ptr
1957  *
1958  *      that is: the plus text must start as a new line, and if there is no minus
1959  *      word printed, a graph prefix must be printed.
1960  *
1961  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1962  *      *(diff_words->current_plus - 1) == '\n'
1963  *
1964  *      that is: a graph prefix must be printed following a '\n'
1965  */
1966 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1967 {
1968         if ((diff_words->last_minus == 0 &&
1969                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1970                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1971                 *(diff_words->current_plus - 1) == '\n')) {
1972                 return 1;
1973         } else {
1974                 return 0;
1975         }
1976 }
1977
1978 static void fn_out_diff_words_aux(void *priv,
1979                                   long minus_first, long minus_len,
1980                                   long plus_first, long plus_len,
1981                                   const char *func, long funclen)
1982 {
1983         struct diff_words_data *diff_words = priv;
1984         struct diff_words_style *style = diff_words->style;
1985         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1986         struct diff_options *opt = diff_words->opt;
1987         const char *line_prefix;
1988
1989         assert(opt);
1990         line_prefix = diff_line_prefix(opt);
1991
1992         /* POSIX requires that first be decremented by one if len == 0... */
1993         if (minus_len) {
1994                 minus_begin = diff_words->minus.orig[minus_first].begin;
1995                 minus_end =
1996                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1997         } else
1998                 minus_begin = minus_end =
1999                         diff_words->minus.orig[minus_first].end;
2000
2001         if (plus_len) {
2002                 plus_begin = diff_words->plus.orig[plus_first].begin;
2003                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2004         } else
2005                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2006
2007         if (color_words_output_graph_prefix(diff_words)) {
2008                 fputs(line_prefix, diff_words->opt->file);
2009         }
2010         if (diff_words->current_plus != plus_begin) {
2011                 fn_out_diff_words_write_helper(diff_words->opt,
2012                                 &style->ctx, style->newline,
2013                                 plus_begin - diff_words->current_plus,
2014                                 diff_words->current_plus);
2015         }
2016         if (minus_begin != minus_end) {
2017                 fn_out_diff_words_write_helper(diff_words->opt,
2018                                 &style->old_word, style->newline,
2019                                 minus_end - minus_begin, minus_begin);
2020         }
2021         if (plus_begin != plus_end) {
2022                 fn_out_diff_words_write_helper(diff_words->opt,
2023                                 &style->new_word, style->newline,
2024                                 plus_end - plus_begin, plus_begin);
2025         }
2026
2027         diff_words->current_plus = plus_end;
2028         diff_words->last_minus = minus_first;
2029 }
2030
2031 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2032 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2033                 int *begin, int *end)
2034 {
2035         if (word_regex && *begin < buffer->size) {
2036                 regmatch_t match[1];
2037                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2038                                  buffer->size - *begin, 1, match, 0)) {
2039                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2040                                         '\n', match[0].rm_eo - match[0].rm_so);
2041                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2042                         *begin += match[0].rm_so;
2043                         return *begin >= *end;
2044                 }
2045                 return -1;
2046         }
2047
2048         /* find the next word */
2049         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2050                 (*begin)++;
2051         if (*begin >= buffer->size)
2052                 return -1;
2053
2054         /* find the end of the word */
2055         *end = *begin + 1;
2056         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2057                 (*end)++;
2058
2059         return 0;
2060 }
2061
2062 /*
2063  * This function splits the words in buffer->text, stores the list with
2064  * newline separator into out, and saves the offsets of the original words
2065  * in buffer->orig.
2066  */
2067 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2068                 regex_t *word_regex)
2069 {
2070         int i, j;
2071         long alloc = 0;
2072
2073         out->size = 0;
2074         out->ptr = NULL;
2075
2076         /* fake an empty "0th" word */
2077         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2078         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2079         buffer->orig_nr = 1;
2080
2081         for (i = 0; i < buffer->text.size; i++) {
2082                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2083                         return;
2084
2085                 /* store original boundaries */
2086                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2087                                 buffer->orig_alloc);
2088                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2089                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2090                 buffer->orig_nr++;
2091
2092                 /* store one word */
2093                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2094                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2095                 out->ptr[out->size + j - i] = '\n';
2096                 out->size += j - i + 1;
2097
2098                 i = j - 1;
2099         }
2100 }
2101
2102 /* this executes the word diff on the accumulated buffers */
2103 static void diff_words_show(struct diff_words_data *diff_words)
2104 {
2105         xpparam_t xpp;
2106         xdemitconf_t xecfg;
2107         mmfile_t minus, plus;
2108         struct diff_words_style *style = diff_words->style;
2109
2110         struct diff_options *opt = diff_words->opt;
2111         const char *line_prefix;
2112
2113         assert(opt);
2114         line_prefix = diff_line_prefix(opt);
2115
2116         /* special case: only removal */
2117         if (!diff_words->plus.text.size) {
2118                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2119                                  line_prefix, strlen(line_prefix), 0);
2120                 fn_out_diff_words_write_helper(diff_words->opt,
2121                         &style->old_word, style->newline,
2122                         diff_words->minus.text.size,
2123                         diff_words->minus.text.ptr);
2124                 diff_words->minus.text.size = 0;
2125                 return;
2126         }
2127
2128         diff_words->current_plus = diff_words->plus.text.ptr;
2129         diff_words->last_minus = 0;
2130
2131         memset(&xpp, 0, sizeof(xpp));
2132         memset(&xecfg, 0, sizeof(xecfg));
2133         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2134         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2135         xpp.flags = 0;
2136         /* as only the hunk header will be parsed, we need a 0-context */
2137         xecfg.ctxlen = 0;
2138         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2139                           diff_words, &xpp, &xecfg))
2140                 die("unable to generate word diff");
2141         free(minus.ptr);
2142         free(plus.ptr);
2143         if (diff_words->current_plus != diff_words->plus.text.ptr +
2144                         diff_words->plus.text.size) {
2145                 if (color_words_output_graph_prefix(diff_words))
2146                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2147                                          line_prefix, strlen(line_prefix), 0);
2148                 fn_out_diff_words_write_helper(diff_words->opt,
2149                         &style->ctx, style->newline,
2150                         diff_words->plus.text.ptr + diff_words->plus.text.size
2151                         - diff_words->current_plus, diff_words->current_plus);
2152         }
2153         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2154 }
2155
2156 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2157 static void diff_words_flush(struct emit_callback *ecbdata)
2158 {
2159         struct diff_options *wo = ecbdata->diff_words->opt;
2160
2161         if (ecbdata->diff_words->minus.text.size ||
2162             ecbdata->diff_words->plus.text.size)
2163                 diff_words_show(ecbdata->diff_words);
2164
2165         if (wo->emitted_symbols) {
2166                 struct diff_options *o = ecbdata->opt;
2167                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2168                 int i;
2169
2170                 /*
2171                  * NEEDSWORK:
2172                  * Instead of appending each, concat all words to a line?
2173                  */
2174                 for (i = 0; i < wol->nr; i++)
2175                         append_emitted_diff_symbol(o, &wol->buf[i]);
2176
2177                 for (i = 0; i < wol->nr; i++)
2178                         free((void *)wol->buf[i].line);
2179
2180                 wol->nr = 0;
2181         }
2182 }
2183
2184 static void diff_filespec_load_driver(struct diff_filespec *one,
2185                                       struct index_state *istate)
2186 {
2187         /* Use already-loaded driver */
2188         if (one->driver)
2189                 return;
2190
2191         if (S_ISREG(one->mode))
2192                 one->driver = userdiff_find_by_path(istate, one->path);
2193
2194         /* Fallback to default settings */
2195         if (!one->driver)
2196                 one->driver = userdiff_find_by_name("default");
2197 }
2198
2199 static const char *userdiff_word_regex(struct diff_filespec *one,
2200                                        struct index_state *istate)
2201 {
2202         diff_filespec_load_driver(one, istate);
2203         return one->driver->word_regex;
2204 }
2205
2206 static void init_diff_words_data(struct emit_callback *ecbdata,
2207                                  struct diff_options *orig_opts,
2208                                  struct diff_filespec *one,
2209                                  struct diff_filespec *two)
2210 {
2211         int i;
2212         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2213         memcpy(o, orig_opts, sizeof(struct diff_options));
2214
2215         ecbdata->diff_words =
2216                 xcalloc(1, sizeof(struct diff_words_data));
2217         ecbdata->diff_words->type = o->word_diff;
2218         ecbdata->diff_words->opt = o;
2219
2220         if (orig_opts->emitted_symbols)
2221                 o->emitted_symbols =
2222                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2223
2224         if (!o->word_regex)
2225                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2226         if (!o->word_regex)
2227                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2228         if (!o->word_regex)
2229                 o->word_regex = diff_word_regex_cfg;
2230         if (o->word_regex) {
2231                 ecbdata->diff_words->word_regex = (regex_t *)
2232                         xmalloc(sizeof(regex_t));
2233                 if (regcomp(ecbdata->diff_words->word_regex,
2234                             o->word_regex,
2235                             REG_EXTENDED | REG_NEWLINE))
2236                         die("invalid regular expression: %s",
2237                             o->word_regex);
2238         }
2239         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2240                 if (o->word_diff == diff_words_styles[i].type) {
2241                         ecbdata->diff_words->style =
2242                                 &diff_words_styles[i];
2243                         break;
2244                 }
2245         }
2246         if (want_color(o->use_color)) {
2247                 struct diff_words_style *st = ecbdata->diff_words->style;
2248                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2249                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2250                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2251         }
2252 }
2253
2254 static void free_diff_words_data(struct emit_callback *ecbdata)
2255 {
2256         if (ecbdata->diff_words) {
2257                 diff_words_flush(ecbdata);
2258                 free (ecbdata->diff_words->opt->emitted_symbols);
2259                 free (ecbdata->diff_words->opt);
2260                 free (ecbdata->diff_words->minus.text.ptr);
2261                 free (ecbdata->diff_words->minus.orig);
2262                 free (ecbdata->diff_words->plus.text.ptr);
2263                 free (ecbdata->diff_words->plus.orig);
2264                 if (ecbdata->diff_words->word_regex) {
2265                         regfree(ecbdata->diff_words->word_regex);
2266                         free(ecbdata->diff_words->word_regex);
2267                 }
2268                 FREE_AND_NULL(ecbdata->diff_words);
2269         }
2270 }
2271
2272 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2273 {
2274         if (want_color(diff_use_color))
2275                 return diff_colors[ix];
2276         return "";
2277 }
2278
2279 const char *diff_line_prefix(struct diff_options *opt)
2280 {
2281         struct strbuf *msgbuf;
2282         if (!opt->output_prefix)
2283                 return "";
2284
2285         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2286         return msgbuf->buf;
2287 }
2288
2289 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2290 {
2291         const char *cp;
2292         unsigned long allot;
2293         size_t l = len;
2294
2295         cp = line;
2296         allot = l;
2297         while (0 < l) {
2298                 (void) utf8_width(&cp, &l);
2299                 if (!cp)
2300                         break; /* truncated in the middle? */
2301         }
2302         return allot - l;
2303 }
2304
2305 static void find_lno(const char *line, struct emit_callback *ecbdata)
2306 {
2307         const char *p;
2308         ecbdata->lno_in_preimage = 0;
2309         ecbdata->lno_in_postimage = 0;
2310         p = strchr(line, '-');
2311         if (!p)
2312                 return; /* cannot happen */
2313         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2314         p = strchr(p, '+');
2315         if (!p)
2316                 return; /* cannot happen */
2317         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2318 }
2319
2320 static void fn_out_consume(void *priv, char *line, unsigned long len)
2321 {
2322         struct emit_callback *ecbdata = priv;
2323         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2324         struct diff_options *o = ecbdata->opt;
2325
2326         o->found_changes = 1;
2327
2328         if (ecbdata->header) {
2329                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2330                                  ecbdata->header->buf, ecbdata->header->len, 0);
2331                 strbuf_reset(ecbdata->header);
2332                 ecbdata->header = NULL;
2333         }
2334
2335         if (ecbdata->label_path[0]) {
2336                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2337                                  ecbdata->label_path[0],
2338                                  strlen(ecbdata->label_path[0]), 0);
2339                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2340                                  ecbdata->label_path[1],
2341                                  strlen(ecbdata->label_path[1]), 0);
2342                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2343         }
2344
2345         if (diff_suppress_blank_empty
2346             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2347                 line[0] = '\n';
2348                 len = 1;
2349         }
2350
2351         if (line[0] == '@') {
2352                 if (ecbdata->diff_words)
2353                         diff_words_flush(ecbdata);
2354                 len = sane_truncate_line(ecbdata, line, len);
2355                 find_lno(line, ecbdata);
2356                 emit_hunk_header(ecbdata, line, len);
2357                 return;
2358         }
2359
2360         if (ecbdata->diff_words) {
2361                 enum diff_symbol s =
2362                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2363                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2364                 if (line[0] == '-') {
2365                         diff_words_append(line, len,
2366                                           &ecbdata->diff_words->minus);
2367                         return;
2368                 } else if (line[0] == '+') {
2369                         diff_words_append(line, len,
2370                                           &ecbdata->diff_words->plus);
2371                         return;
2372                 } else if (starts_with(line, "\\ ")) {
2373                         /*
2374                          * Eat the "no newline at eof" marker as if we
2375                          * saw a "+" or "-" line with nothing on it,
2376                          * and return without diff_words_flush() to
2377                          * defer processing. If this is the end of
2378                          * preimage, more "+" lines may come after it.
2379                          */
2380                         return;
2381                 }
2382                 diff_words_flush(ecbdata);
2383                 emit_diff_symbol(o, s, line, len, 0);
2384                 return;
2385         }
2386
2387         switch (line[0]) {
2388         case '+':
2389                 ecbdata->lno_in_postimage++;
2390                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2391                 break;
2392         case '-':
2393                 ecbdata->lno_in_preimage++;
2394                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2395                 break;
2396         case ' ':
2397                 ecbdata->lno_in_postimage++;
2398                 ecbdata->lno_in_preimage++;
2399                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2400                 break;
2401         default:
2402                 /* incomplete line at the end */
2403                 ecbdata->lno_in_preimage++;
2404                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2405                                  line, len, 0);
2406                 break;
2407         }
2408 }
2409
2410 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2411 {
2412         const char *old_name = a;
2413         const char *new_name = b;
2414         int pfx_length, sfx_length;
2415         int pfx_adjust_for_slash;
2416         int len_a = strlen(a);
2417         int len_b = strlen(b);
2418         int a_midlen, b_midlen;
2419         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2420         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2421
2422         if (qlen_a || qlen_b) {
2423                 quote_c_style(a, name, NULL, 0);
2424                 strbuf_addstr(name, " => ");
2425                 quote_c_style(b, name, NULL, 0);
2426                 return;
2427         }
2428
2429         /* Find common prefix */
2430         pfx_length = 0;
2431         while (*old_name && *new_name && *old_name == *new_name) {
2432                 if (*old_name == '/')
2433                         pfx_length = old_name - a + 1;
2434                 old_name++;
2435                 new_name++;
2436         }
2437
2438         /* Find common suffix */
2439         old_name = a + len_a;
2440         new_name = b + len_b;
2441         sfx_length = 0;
2442         /*
2443          * If there is a common prefix, it must end in a slash.  In
2444          * that case we let this loop run 1 into the prefix to see the
2445          * same slash.
2446          *
2447          * If there is no common prefix, we cannot do this as it would
2448          * underrun the input strings.
2449          */
2450         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2451         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2452                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2453                *old_name == *new_name) {
2454                 if (*old_name == '/')
2455                         sfx_length = len_a - (old_name - a);
2456                 old_name--;
2457                 new_name--;
2458         }
2459
2460         /*
2461          * pfx{mid-a => mid-b}sfx
2462          * {pfx-a => pfx-b}sfx
2463          * pfx{sfx-a => sfx-b}
2464          * name-a => name-b
2465          */
2466         a_midlen = len_a - pfx_length - sfx_length;
2467         b_midlen = len_b - pfx_length - sfx_length;
2468         if (a_midlen < 0)
2469                 a_midlen = 0;
2470         if (b_midlen < 0)
2471                 b_midlen = 0;
2472
2473         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2474         if (pfx_length + sfx_length) {
2475                 strbuf_add(name, a, pfx_length);
2476                 strbuf_addch(name, '{');
2477         }
2478         strbuf_add(name, a + pfx_length, a_midlen);
2479         strbuf_addstr(name, " => ");
2480         strbuf_add(name, b + pfx_length, b_midlen);
2481         if (pfx_length + sfx_length) {
2482                 strbuf_addch(name, '}');
2483                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2484         }
2485 }
2486
2487 struct diffstat_t {
2488         int nr;
2489         int alloc;
2490         struct diffstat_file {
2491                 char *from_name;
2492                 char *name;
2493                 char *print_name;
2494                 const char *comments;
2495                 unsigned is_unmerged:1;
2496                 unsigned is_binary:1;
2497                 unsigned is_renamed:1;
2498                 unsigned is_interesting:1;
2499                 uintmax_t added, deleted;
2500         } **files;
2501 };
2502
2503 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2504                                           const char *name_a,
2505                                           const char *name_b)
2506 {
2507         struct diffstat_file *x;
2508         x = xcalloc(1, sizeof(*x));
2509         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2510         diffstat->files[diffstat->nr++] = x;
2511         if (name_b) {
2512                 x->from_name = xstrdup(name_a);
2513                 x->name = xstrdup(name_b);
2514                 x->is_renamed = 1;
2515         }
2516         else {
2517                 x->from_name = NULL;
2518                 x->name = xstrdup(name_a);
2519         }
2520         return x;
2521 }
2522
2523 static void diffstat_consume(void *priv, char *line, unsigned long len)
2524 {
2525         struct diffstat_t *diffstat = priv;
2526         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2527
2528         if (line[0] == '+')
2529                 x->added++;
2530         else if (line[0] == '-')
2531                 x->deleted++;
2532 }
2533
2534 const char mime_boundary_leader[] = "------------";
2535
2536 static int scale_linear(int it, int width, int max_change)
2537 {
2538         if (!it)
2539                 return 0;
2540         /*
2541          * make sure that at least one '-' or '+' is printed if
2542          * there is any change to this path. The easiest way is to
2543          * scale linearly as if the alloted width is one column shorter
2544          * than it is, and then add 1 to the result.
2545          */
2546         return 1 + (it * (width - 1) / max_change);
2547 }
2548
2549 static void show_graph(struct strbuf *out, char ch, int cnt,
2550                        const char *set, const char *reset)
2551 {
2552         if (cnt <= 0)
2553                 return;
2554         strbuf_addstr(out, set);
2555         strbuf_addchars(out, ch, cnt);
2556         strbuf_addstr(out, reset);
2557 }
2558
2559 static void fill_print_name(struct diffstat_file *file)
2560 {
2561         struct strbuf pname = STRBUF_INIT;
2562
2563         if (file->print_name)
2564                 return;
2565
2566         if (file->is_renamed)
2567                 pprint_rename(&pname, file->from_name, file->name);
2568         else
2569                 quote_c_style(file->name, &pname, NULL, 0);
2570
2571         if (file->comments)
2572                 strbuf_addf(&pname, " (%s)", file->comments);
2573
2574         file->print_name = strbuf_detach(&pname, NULL);
2575 }
2576
2577 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2578                 int files, int insertions, int deletions)
2579 {
2580         struct strbuf sb = STRBUF_INIT;
2581
2582         if (!files) {
2583                 assert(insertions == 0 && deletions == 0);
2584                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2585                                  NULL, 0, 0);
2586                 return;
2587         }
2588
2589         strbuf_addf(&sb,
2590                     (files == 1) ? " %d file changed" : " %d files changed",
2591                     files);
2592
2593         /*
2594          * For binary diff, the caller may want to print "x files
2595          * changed" with insertions == 0 && deletions == 0.
2596          *
2597          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2598          * is probably less confusing (i.e skip over "2 files changed
2599          * but nothing about added/removed lines? Is this a bug in Git?").
2600          */
2601         if (insertions || deletions == 0) {
2602                 strbuf_addf(&sb,
2603                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2604                             insertions);
2605         }
2606
2607         if (deletions || insertions == 0) {
2608                 strbuf_addf(&sb,
2609                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2610                             deletions);
2611         }
2612         strbuf_addch(&sb, '\n');
2613         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2614                          sb.buf, sb.len, 0);
2615         strbuf_release(&sb);
2616 }
2617
2618 void print_stat_summary(FILE *fp, int files,
2619                         int insertions, int deletions)
2620 {
2621         struct diff_options o;
2622         memset(&o, 0, sizeof(o));
2623         o.file = fp;
2624
2625         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2626 }
2627
2628 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2629 {
2630         int i, len, add, del, adds = 0, dels = 0;
2631         uintmax_t max_change = 0, max_len = 0;
2632         int total_files = data->nr, count;
2633         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2634         const char *reset, *add_c, *del_c;
2635         int extra_shown = 0;
2636         const char *line_prefix = diff_line_prefix(options);
2637         struct strbuf out = STRBUF_INIT;
2638
2639         if (data->nr == 0)
2640                 return;
2641
2642         count = options->stat_count ? options->stat_count : data->nr;
2643
2644         reset = diff_get_color_opt(options, DIFF_RESET);
2645         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2646         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2647
2648         /*
2649          * Find the longest filename and max number of changes
2650          */
2651         for (i = 0; (i < count) && (i < data->nr); i++) {
2652                 struct diffstat_file *file = data->files[i];
2653                 uintmax_t change = file->added + file->deleted;
2654
2655                 if (!file->is_interesting && (change == 0)) {
2656                         count++; /* not shown == room for one more */
2657                         continue;
2658                 }
2659                 fill_print_name(file);
2660                 len = strlen(file->print_name);
2661                 if (max_len < len)
2662                         max_len = len;
2663
2664                 if (file->is_unmerged) {
2665                         /* "Unmerged" is 8 characters */
2666                         bin_width = bin_width < 8 ? 8 : bin_width;
2667                         continue;
2668                 }
2669                 if (file->is_binary) {
2670                         /* "Bin XXX -> YYY bytes" */
2671                         int w = 14 + decimal_width(file->added)
2672                                 + decimal_width(file->deleted);
2673                         bin_width = bin_width < w ? w : bin_width;
2674                         /* Display change counts aligned with "Bin" */
2675                         number_width = 3;
2676                         continue;
2677                 }
2678
2679                 if (max_change < change)
2680                         max_change = change;
2681         }
2682         count = i; /* where we can stop scanning in data->files[] */
2683
2684         /*
2685          * We have width = stat_width or term_columns() columns total.
2686          * We want a maximum of min(max_len, stat_name_width) for the name part.
2687          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2688          * We also need 1 for " " and 4 + decimal_width(max_change)
2689          * for " | NNNN " and one the empty column at the end, altogether
2690          * 6 + decimal_width(max_change).
2691          *
2692          * If there's not enough space, we will use the smaller of
2693          * stat_name_width (if set) and 5/8*width for the filename,
2694          * and the rest for constant elements + graph part, but no more
2695          * than stat_graph_width for the graph part.
2696          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2697          * for the standard terminal size).
2698          *
2699          * In other words: stat_width limits the maximum width, and
2700          * stat_name_width fixes the maximum width of the filename,
2701          * and is also used to divide available columns if there
2702          * aren't enough.
2703          *
2704          * Binary files are displayed with "Bin XXX -> YYY bytes"
2705          * instead of the change count and graph. This part is treated
2706          * similarly to the graph part, except that it is not
2707          * "scaled". If total width is too small to accommodate the
2708          * guaranteed minimum width of the filename part and the
2709          * separators and this message, this message will "overflow"
2710          * making the line longer than the maximum width.
2711          */
2712
2713         if (options->stat_width == -1)
2714                 width = term_columns() - strlen(line_prefix);
2715         else
2716                 width = options->stat_width ? options->stat_width : 80;
2717         number_width = decimal_width(max_change) > number_width ?
2718                 decimal_width(max_change) : number_width;
2719
2720         if (options->stat_graph_width == -1)
2721                 options->stat_graph_width = diff_stat_graph_width;
2722
2723         /*
2724          * Guarantee 3/8*16==6 for the graph part
2725          * and 5/8*16==10 for the filename part
2726          */
2727         if (width < 16 + 6 + number_width)
2728                 width = 16 + 6 + number_width;
2729
2730         /*
2731          * First assign sizes that are wanted, ignoring available width.
2732          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2733          * starting from "XXX" should fit in graph_width.
2734          */
2735         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2736         if (options->stat_graph_width &&
2737             options->stat_graph_width < graph_width)
2738                 graph_width = options->stat_graph_width;
2739
2740         name_width = (options->stat_name_width > 0 &&
2741                       options->stat_name_width < max_len) ?
2742                 options->stat_name_width : max_len;
2743
2744         /*
2745          * Adjust adjustable widths not to exceed maximum width
2746          */
2747         if (name_width + number_width + 6 + graph_width > width) {
2748                 if (graph_width > width * 3/8 - number_width - 6) {
2749                         graph_width = width * 3/8 - number_width - 6;
2750                         if (graph_width < 6)
2751                                 graph_width = 6;
2752                 }
2753
2754                 if (options->stat_graph_width &&
2755                     graph_width > options->stat_graph_width)
2756                         graph_width = options->stat_graph_width;
2757                 if (name_width > width - number_width - 6 - graph_width)
2758                         name_width = width - number_width - 6 - graph_width;
2759                 else
2760                         graph_width = width - number_width - 6 - name_width;
2761         }
2762
2763         /*
2764          * From here name_width is the width of the name area,
2765          * and graph_width is the width of the graph area.
2766          * max_change is used to scale graph properly.
2767          */
2768         for (i = 0; i < count; i++) {
2769                 const char *prefix = "";
2770                 struct diffstat_file *file = data->files[i];
2771                 char *name = file->print_name;
2772                 uintmax_t added = file->added;
2773                 uintmax_t deleted = file->deleted;
2774                 int name_len;
2775
2776                 if (!file->is_interesting && (added + deleted == 0))
2777                         continue;
2778
2779                 /*
2780                  * "scale" the filename
2781                  */
2782                 len = name_width;
2783                 name_len = strlen(name);
2784                 if (name_width < name_len) {
2785                         char *slash;
2786                         prefix = "...";
2787                         len -= 3;
2788                         name += name_len - len;
2789                         slash = strchr(name, '/');
2790                         if (slash)
2791                                 name = slash;
2792                 }
2793
2794                 if (file->is_binary) {
2795                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2796                         strbuf_addf(&out, " %*s", number_width, "Bin");
2797                         if (!added && !deleted) {
2798                                 strbuf_addch(&out, '\n');
2799                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2800                                                  out.buf, out.len, 0);
2801                                 strbuf_reset(&out);
2802                                 continue;
2803                         }
2804                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2805                                 del_c, deleted, reset);
2806                         strbuf_addstr(&out, " -> ");
2807                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2808                                 add_c, added, reset);
2809                         strbuf_addstr(&out, " bytes\n");
2810                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2811                                          out.buf, out.len, 0);
2812                         strbuf_reset(&out);
2813                         continue;
2814                 }
2815                 else if (file->is_unmerged) {
2816                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2817                         strbuf_addstr(&out, " Unmerged\n");
2818                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2819                                          out.buf, out.len, 0);
2820                         strbuf_reset(&out);
2821                         continue;
2822                 }
2823
2824                 /*
2825                  * scale the add/delete
2826                  */
2827                 add = added;
2828                 del = deleted;
2829
2830                 if (graph_width <= max_change) {
2831                         int total = scale_linear(add + del, graph_width, max_change);
2832                         if (total < 2 && add && del)
2833                                 /* width >= 2 due to the sanity check */
2834                                 total = 2;
2835                         if (add < del) {
2836                                 add = scale_linear(add, graph_width, max_change);
2837                                 del = total - add;
2838                         } else {
2839                                 del = scale_linear(del, graph_width, max_change);
2840                                 add = total - del;
2841                         }
2842                 }
2843                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2844                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2845                         number_width, added + deleted,
2846                         added + deleted ? " " : "");
2847                 show_graph(&out, '+', add, add_c, reset);
2848                 show_graph(&out, '-', del, del_c, reset);
2849                 strbuf_addch(&out, '\n');
2850                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2851                                  out.buf, out.len, 0);
2852                 strbuf_reset(&out);
2853         }
2854
2855         for (i = 0; i < data->nr; i++) {
2856                 struct diffstat_file *file = data->files[i];
2857                 uintmax_t added = file->added;
2858                 uintmax_t deleted = file->deleted;
2859
2860                 if (file->is_unmerged ||
2861                     (!file->is_interesting && (added + deleted == 0))) {
2862                         total_files--;
2863                         continue;
2864                 }
2865
2866                 if (!file->is_binary) {
2867                         adds += added;
2868                         dels += deleted;
2869                 }
2870                 if (i < count)
2871                         continue;
2872                 if (!extra_shown)
2873                         emit_diff_symbol(options,
2874                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2875                                          NULL, 0, 0);
2876                 extra_shown = 1;
2877         }
2878
2879         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2880         strbuf_release(&out);
2881 }
2882
2883 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2884 {
2885         int i, adds = 0, dels = 0, total_files = data->nr;
2886
2887         if (data->nr == 0)
2888                 return;
2889
2890         for (i = 0; i < data->nr; i++) {
2891                 int added = data->files[i]->added;
2892                 int deleted = data->files[i]->deleted;
2893
2894                 if (data->files[i]->is_unmerged ||
2895                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2896                         total_files--;
2897                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2898                         adds += added;
2899                         dels += deleted;
2900                 }
2901         }
2902         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2903 }
2904
2905 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2906 {
2907         int i;
2908
2909         if (data->nr == 0)
2910                 return;
2911
2912         for (i = 0; i < data->nr; i++) {
2913                 struct diffstat_file *file = data->files[i];
2914
2915                 fprintf(options->file, "%s", diff_line_prefix(options));
2916
2917                 if (file->is_binary)
2918                         fprintf(options->file, "-\t-\t");
2919                 else
2920                         fprintf(options->file,
2921                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2922                                 file->added, file->deleted);
2923                 if (options->line_termination) {
2924                         fill_print_name(file);
2925                         if (!file->is_renamed)
2926                                 write_name_quoted(file->name, options->file,
2927                                                   options->line_termination);
2928                         else {
2929                                 fputs(file->print_name, options->file);
2930                                 putc(options->line_termination, options->file);
2931                         }
2932                 } else {
2933                         if (file->is_renamed) {
2934                                 putc('\0', options->file);
2935                                 write_name_quoted(file->from_name, options->file, '\0');
2936                         }
2937                         write_name_quoted(file->name, options->file, '\0');
2938                 }
2939         }
2940 }
2941
2942 struct dirstat_file {
2943         const char *name;
2944         unsigned long changed;
2945 };
2946
2947 struct dirstat_dir {
2948         struct dirstat_file *files;
2949         int alloc, nr, permille, cumulative;
2950 };
2951
2952 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2953                 unsigned long changed, const char *base, int baselen)
2954 {
2955         unsigned long sum_changes = 0;
2956         unsigned int sources = 0;
2957         const char *line_prefix = diff_line_prefix(opt);
2958
2959         while (dir->nr) {
2960                 struct dirstat_file *f = dir->files;
2961                 int namelen = strlen(f->name);
2962                 unsigned long changes;
2963                 char *slash;
2964
2965                 if (namelen < baselen)
2966                         break;
2967                 if (memcmp(f->name, base, baselen))
2968                         break;
2969                 slash = strchr(f->name + baselen, '/');
2970                 if (slash) {
2971                         int newbaselen = slash + 1 - f->name;
2972                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2973                         sources++;
2974                 } else {
2975                         changes = f->changed;
2976                         dir->files++;
2977                         dir->nr--;
2978                         sources += 2;
2979                 }
2980                 sum_changes += changes;
2981         }
2982
2983         /*
2984          * We don't report dirstat's for
2985          *  - the top level
2986          *  - or cases where everything came from a single directory
2987          *    under this directory (sources == 1).
2988          */
2989         if (baselen && sources != 1) {
2990                 if (sum_changes) {
2991                         int permille = sum_changes * 1000 / changed;
2992                         if (permille >= dir->permille) {
2993                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2994                                         permille / 10, permille % 10, baselen, base);
2995                                 if (!dir->cumulative)
2996                                         return 0;
2997                         }
2998                 }
2999         }
3000         return sum_changes;
3001 }
3002
3003 static int dirstat_compare(const void *_a, const void *_b)
3004 {
3005         const struct dirstat_file *a = _a;
3006         const struct dirstat_file *b = _b;
3007         return strcmp(a->name, b->name);
3008 }
3009
3010 static void show_dirstat(struct diff_options *options)
3011 {
3012         int i;
3013         unsigned long changed;
3014         struct dirstat_dir dir;
3015         struct diff_queue_struct *q = &diff_queued_diff;
3016
3017         dir.files = NULL;
3018         dir.alloc = 0;
3019         dir.nr = 0;
3020         dir.permille = options->dirstat_permille;
3021         dir.cumulative = options->flags.dirstat_cumulative;
3022
3023         changed = 0;
3024         for (i = 0; i < q->nr; i++) {
3025                 struct diff_filepair *p = q->queue[i];
3026                 const char *name;
3027                 unsigned long copied, added, damage;
3028
3029                 name = p->two->path ? p->two->path : p->one->path;
3030
3031                 if (p->one->oid_valid && p->two->oid_valid &&
3032                     oideq(&p->one->oid, &p->two->oid)) {
3033                         /*
3034                          * The SHA1 has not changed, so pre-/post-content is
3035                          * identical. We can therefore skip looking at the
3036                          * file contents altogether.
3037                          */
3038                         damage = 0;
3039                         goto found_damage;
3040                 }
3041
3042                 if (options->flags.dirstat_by_file) {
3043                         /*
3044                          * In --dirstat-by-file mode, we don't really need to
3045                          * look at the actual file contents at all.
3046                          * The fact that the SHA1 changed is enough for us to
3047                          * add this file to the list of results
3048                          * (with each file contributing equal damage).
3049                          */
3050                         damage = 1;
3051                         goto found_damage;
3052                 }
3053
3054                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3055                         diff_populate_filespec(options->repo, p->one, 0);
3056                         diff_populate_filespec(options->repo, p->two, 0);
3057                         diffcore_count_changes(options->repo,
3058                                                p->one, p->two, NULL, NULL,
3059                                                &copied, &added);
3060                         diff_free_filespec_data(p->one);
3061                         diff_free_filespec_data(p->two);
3062                 } else if (DIFF_FILE_VALID(p->one)) {
3063                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3064                         copied = added = 0;
3065                         diff_free_filespec_data(p->one);
3066                 } else if (DIFF_FILE_VALID(p->two)) {
3067                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3068                         copied = 0;
3069                         added = p->two->size;
3070                         diff_free_filespec_data(p->two);
3071                 } else
3072                         continue;
3073
3074                 /*
3075                  * Original minus copied is the removed material,
3076                  * added is the new material.  They are both damages
3077                  * made to the preimage.
3078                  * If the resulting damage is zero, we know that
3079                  * diffcore_count_changes() considers the two entries to
3080                  * be identical, but since the oid changed, we
3081                  * know that there must have been _some_ kind of change,
3082                  * so we force all entries to have damage > 0.
3083                  */
3084                 damage = (p->one->size - copied) + added;
3085                 if (!damage)
3086                         damage = 1;
3087
3088 found_damage:
3089                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3090                 dir.files[dir.nr].name = name;
3091                 dir.files[dir.nr].changed = damage;
3092                 changed += damage;
3093                 dir.nr++;
3094         }
3095
3096         /* This can happen even with many files, if everything was renames */
3097         if (!changed)
3098                 return;
3099
3100         /* Show all directories with more than x% of the changes */
3101         QSORT(dir.files, dir.nr, dirstat_compare);
3102         gather_dirstat(options, &dir, changed, "", 0);
3103 }
3104
3105 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3106 {
3107         int i;
3108         unsigned long changed;
3109         struct dirstat_dir dir;
3110
3111         if (data->nr == 0)
3112                 return;
3113
3114         dir.files = NULL;
3115         dir.alloc = 0;
3116         dir.nr = 0;
3117         dir.permille = options->dirstat_permille;
3118         dir.cumulative = options->flags.dirstat_cumulative;
3119
3120         changed = 0;
3121         for (i = 0; i < data->nr; i++) {
3122                 struct diffstat_file *file = data->files[i];
3123                 unsigned long damage = file->added + file->deleted;
3124                 if (file->is_binary)
3125                         /*
3126                          * binary files counts bytes, not lines. Must find some
3127                          * way to normalize binary bytes vs. textual lines.
3128                          * The following heuristic assumes that there are 64
3129                          * bytes per "line".
3130                          * This is stupid and ugly, but very cheap...
3131                          */
3132                         damage = DIV_ROUND_UP(damage, 64);
3133                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3134                 dir.files[dir.nr].name = file->name;
3135                 dir.files[dir.nr].changed = damage;
3136                 changed += damage;
3137                 dir.nr++;
3138         }
3139
3140         /* This can happen even with many files, if everything was renames */
3141         if (!changed)
3142                 return;
3143
3144         /* Show all directories with more than x% of the changes */
3145         QSORT(dir.files, dir.nr, dirstat_compare);
3146         gather_dirstat(options, &dir, changed, "", 0);
3147 }
3148
3149 static void free_diffstat_info(struct diffstat_t *diffstat)
3150 {
3151         int i;
3152         for (i = 0; i < diffstat->nr; i++) {
3153                 struct diffstat_file *f = diffstat->files[i];
3154                 free(f->print_name);
3155                 free(f->name);
3156                 free(f->from_name);
3157                 free(f);
3158         }
3159         free(diffstat->files);
3160 }
3161
3162 struct checkdiff_t {
3163         const char *filename;
3164         int lineno;
3165         int conflict_marker_size;
3166         struct diff_options *o;
3167         unsigned ws_rule;
3168         unsigned status;
3169 };
3170
3171 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3172 {
3173         char firstchar;
3174         int cnt;
3175
3176         if (len < marker_size + 1)
3177                 return 0;
3178         firstchar = line[0];
3179         switch (firstchar) {
3180         case '=': case '>': case '<': case '|':
3181                 break;
3182         default:
3183                 return 0;
3184         }
3185         for (cnt = 1; cnt < marker_size; cnt++)
3186                 if (line[cnt] != firstchar)
3187                         return 0;
3188         /* line[1] thru line[marker_size-1] are same as firstchar */
3189         if (len < marker_size + 1 || !isspace(line[marker_size]))
3190                 return 0;
3191         return 1;
3192 }
3193
3194 static void checkdiff_consume_hunk(void *priv,
3195                                    long ob, long on, long nb, long nn,
3196                                    const char *func, long funclen)
3197
3198 {
3199         struct checkdiff_t *data = priv;
3200         data->lineno = nb - 1;
3201 }
3202
3203 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3204 {
3205         struct checkdiff_t *data = priv;
3206         int marker_size = data->conflict_marker_size;
3207         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3208         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3209         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3210         char *err;
3211         const char *line_prefix;
3212
3213         assert(data->o);
3214         line_prefix = diff_line_prefix(data->o);
3215
3216         if (line[0] == '+') {
3217                 unsigned bad;
3218                 data->lineno++;
3219                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3220                         data->status |= 1;
3221                         fprintf(data->o->file,
3222                                 "%s%s:%d: leftover conflict marker\n",
3223                                 line_prefix, data->filename, data->lineno);
3224                 }
3225                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3226                 if (!bad)
3227                         return;
3228                 data->status |= bad;
3229                 err = whitespace_error_string(bad);
3230                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3231                         line_prefix, data->filename, data->lineno, err);
3232                 free(err);
3233                 emit_line(data->o, set, reset, line, 1);
3234                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3235                               data->o->file, set, reset, ws);
3236         } else if (line[0] == ' ') {
3237                 data->lineno++;
3238         }
3239 }
3240
3241 static unsigned char *deflate_it(char *data,
3242                                  unsigned long size,
3243                                  unsigned long *result_size)
3244 {
3245         int bound;
3246         unsigned char *deflated;
3247         git_zstream stream;
3248
3249         git_deflate_init(&stream, zlib_compression_level);
3250         bound = git_deflate_bound(&stream, size);
3251         deflated = xmalloc(bound);
3252         stream.next_out = deflated;
3253         stream.avail_out = bound;
3254
3255         stream.next_in = (unsigned char *)data;
3256         stream.avail_in = size;
3257         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3258                 ; /* nothing */
3259         git_deflate_end(&stream);
3260         *result_size = stream.total_out;
3261         return deflated;
3262 }
3263
3264 static void emit_binary_diff_body(struct diff_options *o,
3265                                   mmfile_t *one, mmfile_t *two)
3266 {
3267         void *cp;
3268         void *delta;
3269         void *deflated;
3270         void *data;
3271         unsigned long orig_size;
3272         unsigned long delta_size;
3273         unsigned long deflate_size;
3274         unsigned long data_size;
3275
3276         /* We could do deflated delta, or we could do just deflated two,
3277          * whichever is smaller.
3278          */
3279         delta = NULL;
3280         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3281         if (one->size && two->size) {
3282                 delta = diff_delta(one->ptr, one->size,
3283                                    two->ptr, two->size,
3284                                    &delta_size, deflate_size);
3285                 if (delta) {
3286                         void *to_free = delta;
3287                         orig_size = delta_size;
3288                         delta = deflate_it(delta, delta_size, &delta_size);
3289                         free(to_free);
3290                 }
3291         }
3292
3293         if (delta && delta_size < deflate_size) {
3294                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3295                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3296                                  s, strlen(s), 0);
3297                 free(s);
3298                 free(deflated);
3299                 data = delta;
3300                 data_size = delta_size;
3301         } else {
3302                 char *s = xstrfmt("%lu", two->size);
3303                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3304                                  s, strlen(s), 0);
3305                 free(s);
3306                 free(delta);
3307                 data = deflated;
3308                 data_size = deflate_size;
3309         }
3310
3311         /* emit data encoded in base85 */
3312         cp = data;
3313         while (data_size) {
3314                 int len;
3315                 int bytes = (52 < data_size) ? 52 : data_size;
3316                 char line[71];
3317                 data_size -= bytes;
3318                 if (bytes <= 26)
3319                         line[0] = bytes + 'A' - 1;
3320                 else
3321                         line[0] = bytes - 26 + 'a' - 1;
3322                 encode_85(line + 1, cp, bytes);
3323                 cp = (char *) cp + bytes;
3324
3325                 len = strlen(line);
3326                 line[len++] = '\n';
3327                 line[len] = '\0';
3328
3329                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3330                                  line, len, 0);
3331         }
3332         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3333         free(data);
3334 }
3335
3336 static void emit_binary_diff(struct diff_options *o,
3337                              mmfile_t *one, mmfile_t *two)
3338 {
3339         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3340         emit_binary_diff_body(o, one, two);
3341         emit_binary_diff_body(o, two, one);
3342 }
3343
3344 int diff_filespec_is_binary(struct repository *r,
3345                             struct diff_filespec *one)
3346 {
3347         if (one->is_binary == -1) {
3348                 diff_filespec_load_driver(one, r->index);
3349                 if (one->driver->binary != -1)
3350                         one->is_binary = one->driver->binary;
3351                 else {
3352                         if (!one->data && DIFF_FILE_VALID(one))
3353                                 diff_populate_filespec(r, one, CHECK_BINARY);
3354                         if (one->is_binary == -1 && one->data)
3355                                 one->is_binary = buffer_is_binary(one->data,
3356                                                 one->size);
3357                         if (one->is_binary == -1)
3358                                 one->is_binary = 0;
3359                 }
3360         }
3361         return one->is_binary;
3362 }
3363
3364 static const struct userdiff_funcname *
3365 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3366 {
3367         diff_filespec_load_driver(one, o->repo->index);
3368         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3369 }
3370
3371 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3372 {
3373         if (!options->a_prefix)
3374                 options->a_prefix = a;
3375         if (!options->b_prefix)
3376                 options->b_prefix = b;
3377 }
3378
3379 struct userdiff_driver *get_textconv(struct index_state *istate,
3380                                      struct diff_filespec *one)
3381 {
3382         if (!DIFF_FILE_VALID(one))
3383                 return NULL;
3384
3385         diff_filespec_load_driver(one, istate);
3386         return userdiff_get_textconv(one->driver);
3387 }
3388
3389 static void builtin_diff(const char *name_a,
3390                          const char *name_b,
3391                          struct diff_filespec *one,
3392                          struct diff_filespec *two,
3393                          const char *xfrm_msg,
3394                          int must_show_header,
3395                          struct diff_options *o,
3396                          int complete_rewrite)
3397 {
3398         mmfile_t mf1, mf2;
3399         const char *lbl[2];
3400         char *a_one, *b_two;
3401         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3402         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3403         const char *a_prefix, *b_prefix;
3404         struct userdiff_driver *textconv_one = NULL;
3405         struct userdiff_driver *textconv_two = NULL;
3406         struct strbuf header = STRBUF_INIT;
3407         const char *line_prefix = diff_line_prefix(o);
3408
3409         diff_set_mnemonic_prefix(o, "a/", "b/");
3410         if (o->flags.reverse_diff) {
3411                 a_prefix = o->b_prefix;
3412                 b_prefix = o->a_prefix;
3413         } else {
3414                 a_prefix = o->a_prefix;
3415                 b_prefix = o->b_prefix;
3416         }
3417
3418         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3419             (!one->mode || S_ISGITLINK(one->mode)) &&
3420             (!two->mode || S_ISGITLINK(two->mode))) {
3421                 show_submodule_summary(o, one->path ? one->path : two->path,
3422                                 &one->oid, &two->oid,
3423                                 two->dirty_submodule);
3424                 return;
3425         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3426                    (!one->mode || S_ISGITLINK(one->mode)) &&
3427                    (!two->mode || S_ISGITLINK(two->mode))) {
3428                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3429                                 &one->oid, &two->oid,
3430                                 two->dirty_submodule);
3431                 return;
3432         }
3433
3434         if (o->flags.allow_textconv) {
3435                 textconv_one = get_textconv(o->repo->index, one);
3436                 textconv_two = get_textconv(o->repo->index, two);
3437         }
3438
3439         /* Never use a non-valid filename anywhere if at all possible */
3440         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3441         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3442
3443         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3444         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3445         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3446         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3447         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3448         if (lbl[0][0] == '/') {
3449                 /* /dev/null */
3450                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3451                 if (xfrm_msg)
3452                         strbuf_addstr(&header, xfrm_msg);
3453                 must_show_header = 1;
3454         }
3455         else if (lbl[1][0] == '/') {
3456                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3457                 if (xfrm_msg)
3458                         strbuf_addstr(&header, xfrm_msg);
3459                 must_show_header = 1;
3460         }
3461         else {
3462                 if (one->mode != two->mode) {
3463                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3464                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3465                         must_show_header = 1;
3466                 }
3467                 if (xfrm_msg)
3468                         strbuf_addstr(&header, xfrm_msg);
3469
3470                 /*
3471                  * we do not run diff between different kind
3472                  * of objects.
3473                  */
3474                 if ((one->mode ^ two->mode) & S_IFMT)
3475                         goto free_ab_and_return;
3476                 if (complete_rewrite &&
3477                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3478                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3479                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3480                                          header.buf, header.len, 0);
3481                         strbuf_reset(&header);
3482                         emit_rewrite_diff(name_a, name_b, one, two,
3483                                           textconv_one, textconv_two, o);
3484                         o->found_changes = 1;
3485                         goto free_ab_and_return;
3486                 }
3487         }
3488
3489         if (o->irreversible_delete && lbl[1][0] == '/') {
3490                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3491                                  header.len, 0);
3492                 strbuf_reset(&header);
3493                 goto free_ab_and_return;
3494         } else if (!o->flags.text &&
3495                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3496                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3497                 struct strbuf sb = STRBUF_INIT;
3498                 if (!one->data && !two->data &&
3499                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3500                     !o->flags.binary) {
3501                         if (oideq(&one->oid, &two->oid)) {
3502                                 if (must_show_header)
3503                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3504                                                          header.buf, header.len,
3505                                                          0);
3506                                 goto free_ab_and_return;
3507                         }
3508                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3509                                          header.buf, header.len, 0);
3510                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3511                                     diff_line_prefix(o), lbl[0], lbl[1]);
3512                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3513                                          sb.buf, sb.len, 0);
3514                         strbuf_release(&sb);
3515                         goto free_ab_and_return;
3516                 }
3517                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3518                     fill_mmfile(o->repo, &mf2, two) < 0)
3519                         die("unable to read files to diff");
3520                 /* Quite common confusing case */
3521                 if (mf1.size == mf2.size &&
3522                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3523                         if (must_show_header)
3524                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3525                                                  header.buf, header.len, 0);
3526                         goto free_ab_and_return;
3527                 }
3528                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3529                 strbuf_reset(&header);
3530                 if (o->flags.binary)
3531                         emit_binary_diff(o, &mf1, &mf2);
3532                 else {
3533                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3534                                     diff_line_prefix(o), lbl[0], lbl[1]);
3535                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3536                                          sb.buf, sb.len, 0);
3537                         strbuf_release(&sb);
3538                 }
3539                 o->found_changes = 1;
3540         } else {
3541                 /* Crazy xdl interfaces.. */
3542                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3543                 const char *v;
3544                 xpparam_t xpp;
3545                 xdemitconf_t xecfg;
3546                 struct emit_callback ecbdata;
3547                 const struct userdiff_funcname *pe;
3548
3549                 if (must_show_header) {
3550                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3551                                          header.buf, header.len, 0);
3552                         strbuf_reset(&header);
3553                 }
3554
3555                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3556                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3557
3558                 pe = diff_funcname_pattern(o, one);
3559                 if (!pe)
3560                         pe = diff_funcname_pattern(o, two);
3561
3562                 memset(&xpp, 0, sizeof(xpp));
3563                 memset(&xecfg, 0, sizeof(xecfg));
3564                 memset(&ecbdata, 0, sizeof(ecbdata));
3565                 if (o->flags.suppress_diff_headers)
3566                         lbl[0] = NULL;
3567                 ecbdata.label_path = lbl;
3568                 ecbdata.color_diff = want_color(o->use_color);
3569                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3570                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3571                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3572                 ecbdata.opt = o;
3573                 if (header.len && !o->flags.suppress_diff_headers)
3574                         ecbdata.header = &header;
3575                 xpp.flags = o->xdl_opts;
3576                 xpp.anchors = o->anchors;
3577                 xpp.anchors_nr = o->anchors_nr;
3578                 xecfg.ctxlen = o->context;
3579                 xecfg.interhunkctxlen = o->interhunkcontext;
3580                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3581                 if (o->flags.funccontext)
3582                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3583                 if (pe)
3584                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3585                 if (!diffopts)
3586                         ;
3587                 else if (skip_prefix(diffopts, "--unified=", &v))
3588                         xecfg.ctxlen = strtoul(v, NULL, 10);
3589                 else if (skip_prefix(diffopts, "-u", &v))
3590                         xecfg.ctxlen = strtoul(v, NULL, 10);
3591                 if (o->word_diff)
3592                         init_diff_words_data(&ecbdata, o, one, two);
3593                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3594                                   &ecbdata, &xpp, &xecfg))
3595                         die("unable to generate diff for %s", one->path);
3596                 if (o->word_diff)
3597                         free_diff_words_data(&ecbdata);
3598                 if (textconv_one)
3599                         free(mf1.ptr);
3600                 if (textconv_two)
3601                         free(mf2.ptr);
3602                 xdiff_clear_find_func(&xecfg);
3603         }
3604
3605  free_ab_and_return:
3606         strbuf_release(&header);
3607         diff_free_filespec_data(one);
3608         diff_free_filespec_data(two);
3609         free(a_one);
3610         free(b_two);
3611         return;
3612 }
3613
3614 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3615 {
3616         if (!is_renamed) {
3617                 if (p->status == DIFF_STATUS_ADDED) {
3618                         if (S_ISLNK(p->two->mode))
3619                                 return "new +l";
3620                         else if ((p->two->mode & 0777) == 0755)
3621                                 return "new +x";
3622                         else
3623                                 return "new";
3624                 } else if (p->status == DIFF_STATUS_DELETED)
3625                         return "gone";
3626         }
3627         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3628                 return "mode -l";
3629         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3630                 return "mode +l";
3631         else if ((p->one->mode & 0777) == 0644 &&
3632                  (p->two->mode & 0777) == 0755)
3633                 return "mode +x";
3634         else if ((p->one->mode & 0777) == 0755 &&
3635                  (p->two->mode & 0777) == 0644)
3636                 return "mode -x";
3637         return NULL;
3638 }
3639
3640 static void builtin_diffstat(const char *name_a, const char *name_b,
3641                              struct diff_filespec *one,
3642                              struct diff_filespec *two,
3643                              struct diffstat_t *diffstat,
3644                              struct diff_options *o,
3645                              struct diff_filepair *p)
3646 {
3647         mmfile_t mf1, mf2;
3648         struct diffstat_file *data;
3649         int same_contents;
3650         int complete_rewrite = 0;
3651
3652         if (!DIFF_PAIR_UNMERGED(p)) {
3653                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3654                         complete_rewrite = 1;
3655         }
3656
3657         data = diffstat_add(diffstat, name_a, name_b);
3658         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3659         if (o->flags.stat_with_summary)
3660                 data->comments = get_compact_summary(p, data->is_renamed);
3661
3662         if (!one || !two) {
3663                 data->is_unmerged = 1;
3664                 return;
3665         }
3666
3667         same_contents = oideq(&one->oid, &two->oid);
3668
3669         if (diff_filespec_is_binary(o->repo, one) ||
3670             diff_filespec_is_binary(o->repo, two)) {
3671                 data->is_binary = 1;
3672                 if (same_contents) {
3673                         data->added = 0;
3674                         data->deleted = 0;
3675                 } else {
3676                         data->added = diff_filespec_size(o->repo, two);
3677                         data->deleted = diff_filespec_size(o->repo, one);
3678                 }
3679         }
3680
3681         else if (complete_rewrite) {
3682                 diff_populate_filespec(o->repo, one, 0);
3683                 diff_populate_filespec(o->repo, two, 0);
3684                 data->deleted = count_lines(one->data, one->size);
3685                 data->added = count_lines(two->data, two->size);
3686         }
3687
3688         else if (!same_contents) {
3689                 /* Crazy xdl interfaces.. */
3690                 xpparam_t xpp;
3691                 xdemitconf_t xecfg;
3692
3693                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3694                     fill_mmfile(o->repo, &mf2, two) < 0)
3695                         die("unable to read files to diff");
3696
3697                 memset(&xpp, 0, sizeof(xpp));
3698                 memset(&xecfg, 0, sizeof(xecfg));
3699                 xpp.flags = o->xdl_opts;
3700                 xpp.anchors = o->anchors;
3701                 xpp.anchors_nr = o->anchors_nr;
3702                 xecfg.ctxlen = o->context;
3703                 xecfg.interhunkctxlen = o->interhunkcontext;
3704                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3705                                   diffstat_consume, diffstat, &xpp, &xecfg))
3706                         die("unable to generate diffstat for %s", one->path);
3707         }
3708
3709         diff_free_filespec_data(one);
3710         diff_free_filespec_data(two);
3711 }
3712
3713 static void builtin_checkdiff(const char *name_a, const char *name_b,
3714                               const char *attr_path,
3715                               struct diff_filespec *one,
3716                               struct diff_filespec *two,
3717                               struct diff_options *o)
3718 {
3719         mmfile_t mf1, mf2;
3720         struct checkdiff_t data;
3721
3722         if (!two)
3723                 return;
3724
3725         memset(&data, 0, sizeof(data));
3726         data.filename = name_b ? name_b : name_a;
3727         data.lineno = 0;
3728         data.o = o;
3729         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3730         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3731
3732         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3733             fill_mmfile(o->repo, &mf2, two) < 0)
3734                 die("unable to read files to diff");
3735
3736         /*
3737          * All the other codepaths check both sides, but not checking
3738          * the "old" side here is deliberate.  We are checking the newly
3739          * introduced changes, and as long as the "new" side is text, we
3740          * can and should check what it introduces.
3741          */
3742         if (diff_filespec_is_binary(o->repo, two))
3743                 goto free_and_return;
3744         else {
3745                 /* Crazy xdl interfaces.. */
3746                 xpparam_t xpp;
3747                 xdemitconf_t xecfg;
3748
3749                 memset(&xpp, 0, sizeof(xpp));
3750                 memset(&xecfg, 0, sizeof(xecfg));
3751                 xecfg.ctxlen = 1; /* at least one context line */
3752                 xpp.flags = 0;
3753                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3754                                   checkdiff_consume, &data,
3755                                   &xpp, &xecfg))
3756                         die("unable to generate checkdiff for %s", one->path);
3757
3758                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3759                         struct emit_callback ecbdata;
3760                         int blank_at_eof;
3761
3762                         ecbdata.ws_rule = data.ws_rule;
3763                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3764                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3765
3766                         if (blank_at_eof) {
3767                                 static char *err;
3768                                 if (!err)
3769                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3770                                 fprintf(o->file, "%s:%d: %s.\n",
3771                                         data.filename, blank_at_eof, err);
3772                                 data.status = 1; /* report errors */
3773                         }
3774                 }
3775         }
3776  free_and_return:
3777         diff_free_filespec_data(one);
3778         diff_free_filespec_data(two);
3779         if (data.status)
3780                 o->flags.check_failed = 1;
3781 }
3782
3783 struct diff_filespec *alloc_filespec(const char *path)
3784 {
3785         struct diff_filespec *spec;
3786
3787         FLEXPTR_ALLOC_STR(spec, path, path);
3788         spec->count = 1;
3789         spec->is_binary = -1;
3790         return spec;
3791 }
3792
3793 void free_filespec(struct diff_filespec *spec)
3794 {
3795         if (!--spec->count) {
3796                 diff_free_filespec_data(spec);
3797                 free(spec);
3798         }
3799 }
3800
3801 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3802                    int oid_valid, unsigned short mode)
3803 {
3804         if (mode) {
3805                 spec->mode = canon_mode(mode);
3806                 oidcpy(&spec->oid, oid);
3807                 spec->oid_valid = oid_valid;
3808         }
3809 }
3810
3811 /*
3812  * Given a name and sha1 pair, if the index tells us the file in
3813  * the work tree has that object contents, return true, so that
3814  * prepare_temp_file() does not have to inflate and extract.
3815  */
3816 static int reuse_worktree_file(struct index_state *istate,
3817                                const char *name,
3818                                const struct object_id *oid,
3819                                int want_file)
3820 {
3821         const struct cache_entry *ce;
3822         struct stat st;
3823         int pos, len;
3824
3825         /*
3826          * We do not read the cache ourselves here, because the
3827          * benchmark with my previous version that always reads cache
3828          * shows that it makes things worse for diff-tree comparing
3829          * two linux-2.6 kernel trees in an already checked out work
3830          * tree.  This is because most diff-tree comparisons deal with
3831          * only a small number of files, while reading the cache is
3832          * expensive for a large project, and its cost outweighs the
3833          * savings we get by not inflating the object to a temporary
3834          * file.  Practically, this code only helps when we are used
3835          * by diff-cache --cached, which does read the cache before
3836          * calling us.
3837          */
3838         if (!istate->cache)
3839                 return 0;
3840
3841         /* We want to avoid the working directory if our caller
3842          * doesn't need the data in a normal file, this system
3843          * is rather slow with its stat/open/mmap/close syscalls,
3844          * and the object is contained in a pack file.  The pack
3845          * is probably already open and will be faster to obtain
3846          * the data through than the working directory.  Loose
3847          * objects however would tend to be slower as they need
3848          * to be individually opened and inflated.
3849          */
3850         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3851                 return 0;
3852
3853         /*
3854          * Similarly, if we'd have to convert the file contents anyway, that
3855          * makes the optimization not worthwhile.
3856          */
3857         if (!want_file && would_convert_to_git(istate, name))
3858                 return 0;
3859
3860         len = strlen(name);
3861         pos = index_name_pos(istate, name, len);
3862         if (pos < 0)
3863                 return 0;
3864         ce = istate->cache[pos];
3865
3866         /*
3867          * This is not the sha1 we are looking for, or
3868          * unreusable because it is not a regular file.
3869          */
3870         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3871                 return 0;
3872
3873         /*
3874          * If ce is marked as "assume unchanged", there is no
3875          * guarantee that work tree matches what we are looking for.
3876          */
3877         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3878                 return 0;
3879
3880         /*
3881          * If ce matches the file in the work tree, we can reuse it.
3882          */
3883         if (ce_uptodate(ce) ||
3884             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3885                 return 1;
3886
3887         return 0;
3888 }
3889
3890 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3891 {
3892         struct strbuf buf = STRBUF_INIT;
3893         char *dirty = "";
3894
3895         /* Are we looking at the work tree? */
3896         if (s->dirty_submodule)
3897                 dirty = "-dirty";
3898
3899         strbuf_addf(&buf, "Subproject commit %s%s\n",
3900                     oid_to_hex(&s->oid), dirty);
3901         s->size = buf.len;
3902         if (size_only) {
3903                 s->data = NULL;
3904                 strbuf_release(&buf);
3905         } else {
3906                 s->data = strbuf_detach(&buf, NULL);
3907                 s->should_free = 1;
3908         }
3909         return 0;
3910 }
3911
3912 /*
3913  * While doing rename detection and pickaxe operation, we may need to
3914  * grab the data for the blob (or file) for our own in-core comparison.
3915  * diff_filespec has data and size fields for this purpose.
3916  */
3917 int diff_populate_filespec(struct repository *r,
3918                            struct diff_filespec *s,
3919                            unsigned int flags)
3920 {
3921         int size_only = flags & CHECK_SIZE_ONLY;
3922         int err = 0;
3923         int conv_flags = global_conv_flags_eol;
3924         /*
3925          * demote FAIL to WARN to allow inspecting the situation
3926          * instead of refusing.
3927          */
3928         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3929                 conv_flags = CONV_EOL_RNDTRP_WARN;
3930
3931         if (!DIFF_FILE_VALID(s))
3932                 die("internal error: asking to populate invalid file.");
3933         if (S_ISDIR(s->mode))
3934                 return -1;
3935
3936         if (s->data)
3937                 return 0;
3938
3939         if (size_only && 0 < s->size)
3940                 return 0;
3941
3942         if (S_ISGITLINK(s->mode))
3943                 return diff_populate_gitlink(s, size_only);
3944
3945         if (!s->oid_valid ||
3946             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3947                 struct strbuf buf = STRBUF_INIT;
3948                 struct stat st;
3949                 int fd;
3950
3951                 if (lstat(s->path, &st) < 0) {
3952                 err_empty:
3953                         err = -1;
3954                 empty:
3955                         s->data = (char *)"";
3956                         s->size = 0;
3957                         return err;
3958                 }
3959                 s->size = xsize_t(st.st_size);
3960                 if (!s->size)
3961                         goto empty;
3962                 if (S_ISLNK(st.st_mode)) {
3963                         struct strbuf sb = STRBUF_INIT;
3964
3965                         if (strbuf_readlink(&sb, s->path, s->size))
3966                                 goto err_empty;
3967                         s->size = sb.len;
3968                         s->data = strbuf_detach(&sb, NULL);
3969                         s->should_free = 1;
3970                         return 0;
3971                 }
3972
3973                 /*
3974                  * Even if the caller would be happy with getting
3975                  * only the size, we cannot return early at this
3976                  * point if the path requires us to run the content
3977                  * conversion.
3978                  */
3979                 if (size_only && !would_convert_to_git(r->index, s->path))
3980                         return 0;
3981
3982                 /*
3983                  * Note: this check uses xsize_t(st.st_size) that may
3984                  * not be the true size of the blob after it goes
3985                  * through convert_to_git().  This may not strictly be
3986                  * correct, but the whole point of big_file_threshold
3987                  * and is_binary check being that we want to avoid
3988                  * opening the file and inspecting the contents, this
3989                  * is probably fine.
3990                  */
3991                 if ((flags & CHECK_BINARY) &&
3992                     s->size > big_file_threshold && s->is_binary == -1) {
3993                         s->is_binary = 1;
3994                         return 0;
3995                 }
3996                 fd = open(s->path, O_RDONLY);
3997                 if (fd < 0)
3998                         goto err_empty;
3999                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4000                 close(fd);
4001                 s->should_munmap = 1;
4002
4003                 /*
4004                  * Convert from working tree format to canonical git format
4005                  */
4006                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4007                         size_t size = 0;
4008                         munmap(s->data, s->size);
4009                         s->should_munmap = 0;
4010                         s->data = strbuf_detach(&buf, &size);
4011                         s->size = size;
4012                         s->should_free = 1;
4013                 }
4014         }
4015         else {
4016                 enum object_type type;
4017                 if (size_only || (flags & CHECK_BINARY)) {
4018                         type = oid_object_info(r, &s->oid, &s->size);
4019                         if (type < 0)
4020                                 die("unable to read %s",
4021                                     oid_to_hex(&s->oid));
4022                         if (size_only)
4023                                 return 0;
4024                         if (s->size > big_file_threshold && s->is_binary == -1) {
4025                                 s->is_binary = 1;
4026                                 return 0;
4027                         }
4028                 }
4029                 s->data = read_object_file(&s->oid, &type, &s->size);
4030                 if (!s->data)
4031                         die("unable to read %s", oid_to_hex(&s->oid));
4032                 s->should_free = 1;
4033         }
4034         return 0;
4035 }
4036
4037 void diff_free_filespec_blob(struct diff_filespec *s)
4038 {
4039         if (s->should_free)
4040                 free(s->data);
4041         else if (s->should_munmap)
4042                 munmap(s->data, s->size);
4043
4044         if (s->should_free || s->should_munmap) {
4045                 s->should_free = s->should_munmap = 0;
4046                 s->data = NULL;
4047         }
4048 }
4049
4050 void diff_free_filespec_data(struct diff_filespec *s)
4051 {
4052         diff_free_filespec_blob(s);
4053         FREE_AND_NULL(s->cnt_data);
4054 }
4055
4056 static void prep_temp_blob(struct index_state *istate,
4057                            const char *path, struct diff_tempfile *temp,
4058                            void *blob,
4059                            unsigned long size,
4060                            const struct object_id *oid,
4061                            int mode)
4062 {
4063         struct strbuf buf = STRBUF_INIT;
4064         struct strbuf tempfile = STRBUF_INIT;
4065         char *path_dup = xstrdup(path);
4066         const char *base = basename(path_dup);
4067
4068         /* Generate "XXXXXX_basename.ext" */
4069         strbuf_addstr(&tempfile, "XXXXXX_");
4070         strbuf_addstr(&tempfile, base);
4071
4072         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4073         if (!temp->tempfile)
4074                 die_errno("unable to create temp-file");
4075         if (convert_to_working_tree(istate, path,
4076                         (const char *)blob, (size_t)size, &buf)) {
4077                 blob = buf.buf;
4078                 size = buf.len;
4079         }
4080         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4081             close_tempfile_gently(temp->tempfile))
4082                 die_errno("unable to write temp-file");
4083         temp->name = get_tempfile_path(temp->tempfile);
4084         oid_to_hex_r(temp->hex, oid);
4085         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4086         strbuf_release(&buf);
4087         strbuf_release(&tempfile);
4088         free(path_dup);
4089 }
4090
4091 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4092                                                const char *name,
4093                                                struct diff_filespec *one)
4094 {
4095         struct diff_tempfile *temp = claim_diff_tempfile();
4096
4097         if (!DIFF_FILE_VALID(one)) {
4098         not_a_valid_file:
4099                 /* A '-' entry produces this for file-2, and
4100                  * a '+' entry produces this for file-1.
4101                  */
4102                 temp->name = "/dev/null";
4103                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4104                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4105                 return temp;
4106         }
4107
4108         if (!S_ISGITLINK(one->mode) &&
4109             (!one->oid_valid ||
4110              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4111                 struct stat st;
4112                 if (lstat(name, &st) < 0) {
4113                         if (errno == ENOENT)
4114                                 goto not_a_valid_file;
4115                         die_errno("stat(%s)", name);
4116                 }
4117                 if (S_ISLNK(st.st_mode)) {
4118                         struct strbuf sb = STRBUF_INIT;
4119                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4120                                 die_errno("readlink(%s)", name);
4121                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4122                                        (one->oid_valid ?
4123                                         &one->oid : &null_oid),
4124                                        (one->oid_valid ?
4125                                         one->mode : S_IFLNK));
4126                         strbuf_release(&sb);
4127                 }
4128                 else {
4129                         /* we can borrow from the file in the work tree */
4130                         temp->name = name;
4131                         if (!one->oid_valid)
4132                                 oid_to_hex_r(temp->hex, &null_oid);
4133                         else
4134                                 oid_to_hex_r(temp->hex, &one->oid);
4135                         /* Even though we may sometimes borrow the
4136                          * contents from the work tree, we always want
4137                          * one->mode.  mode is trustworthy even when
4138                          * !(one->oid_valid), as long as
4139                          * DIFF_FILE_VALID(one).
4140                          */
4141                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4142                 }
4143                 return temp;
4144         }
4145         else {
4146                 if (diff_populate_filespec(r, one, 0))
4147                         die("cannot read data blob for %s", one->path);
4148                 prep_temp_blob(r->index, name, temp,
4149                                one->data, one->size,
4150                                &one->oid, one->mode);
4151         }
4152         return temp;
4153 }
4154
4155 static void add_external_diff_name(struct repository *r,
4156                                    struct argv_array *argv,
4157                                    const char *name,
4158                                    struct diff_filespec *df)
4159 {
4160         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4161         argv_array_push(argv, temp->name);
4162         argv_array_push(argv, temp->hex);
4163         argv_array_push(argv, temp->mode);
4164 }
4165
4166 /* An external diff command takes:
4167  *
4168  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4169  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4170  *
4171  */
4172 static void run_external_diff(const char *pgm,
4173                               const char *name,
4174                               const char *other,
4175                               struct diff_filespec *one,
4176                               struct diff_filespec *two,
4177                               const char *xfrm_msg,
4178                               int complete_rewrite,
4179                               struct diff_options *o)
4180 {
4181         struct argv_array argv = ARGV_ARRAY_INIT;
4182         struct argv_array env = ARGV_ARRAY_INIT;
4183         struct diff_queue_struct *q = &diff_queued_diff;
4184
4185         argv_array_push(&argv, pgm);
4186         argv_array_push(&argv, name);
4187
4188         if (one && two) {
4189                 add_external_diff_name(o->repo, &argv, name, one);
4190                 if (!other)
4191                         add_external_diff_name(o->repo, &argv, name, two);
4192                 else {
4193                         add_external_diff_name(o->repo, &argv, other, two);
4194                         argv_array_push(&argv, other);
4195                         argv_array_push(&argv, xfrm_msg);
4196                 }
4197         }
4198
4199         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4200         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4201
4202         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4203                 die(_("external diff died, stopping at %s"), name);
4204
4205         remove_tempfile();
4206         argv_array_clear(&argv);
4207         argv_array_clear(&env);
4208 }
4209
4210 static int similarity_index(struct diff_filepair *p)
4211 {
4212         return p->score * 100 / MAX_SCORE;
4213 }
4214
4215 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4216 {
4217         if (startup_info->have_repository)
4218                 return find_unique_abbrev(oid, abbrev);
4219         else {
4220                 char *hex = oid_to_hex(oid);
4221                 if (abbrev < 0)
4222                         abbrev = FALLBACK_DEFAULT_ABBREV;
4223                 if (abbrev > the_hash_algo->hexsz)
4224                         BUG("oid abbreviation out of range: %d", abbrev);
4225                 if (abbrev)
4226                         hex[abbrev] = '\0';
4227                 return hex;
4228         }
4229 }
4230
4231 static void fill_metainfo(struct strbuf *msg,
4232                           const char *name,
4233                           const char *other,
4234                           struct diff_filespec *one,
4235                           struct diff_filespec *two,
4236                           struct diff_options *o,
4237                           struct diff_filepair *p,
4238                           int *must_show_header,
4239                           int use_color)
4240 {
4241         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4242         const char *reset = diff_get_color(use_color, DIFF_RESET);
4243         const char *line_prefix = diff_line_prefix(o);
4244
4245         *must_show_header = 1;
4246         strbuf_init(msg, PATH_MAX * 2 + 300);
4247         switch (p->status) {
4248         case DIFF_STATUS_COPIED:
4249                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4250                             line_prefix, set, similarity_index(p));
4251                 strbuf_addf(msg, "%s\n%s%scopy from ",
4252                             reset,  line_prefix, set);
4253                 quote_c_style(name, msg, NULL, 0);
4254                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4255                 quote_c_style(other, msg, NULL, 0);
4256                 strbuf_addf(msg, "%s\n", reset);
4257                 break;
4258         case DIFF_STATUS_RENAMED:
4259                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4260                             line_prefix, set, similarity_index(p));
4261                 strbuf_addf(msg, "%s\n%s%srename from ",
4262                             reset, line_prefix, set);
4263                 quote_c_style(name, msg, NULL, 0);
4264                 strbuf_addf(msg, "%s\n%s%srename to ",
4265                             reset, line_prefix, set);
4266                 quote_c_style(other, msg, NULL, 0);
4267                 strbuf_addf(msg, "%s\n", reset);
4268                 break;
4269         case DIFF_STATUS_MODIFIED:
4270                 if (p->score) {
4271                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4272                                     line_prefix,
4273                                     set, similarity_index(p), reset);
4274                         break;
4275                 }
4276                 /* fallthru */
4277         default:
4278                 *must_show_header = 0;
4279         }
4280         if (one && two && !oideq(&one->oid, &two->oid)) {
4281                 const unsigned hexsz = the_hash_algo->hexsz;
4282                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4283
4284                 if (o->flags.binary) {
4285                         mmfile_t mf;
4286                         if ((!fill_mmfile(o->repo, &mf, one) &&
4287                              diff_filespec_is_binary(o->repo, one)) ||
4288                             (!fill_mmfile(o->repo, &mf, two) &&
4289                              diff_filespec_is_binary(o->repo, two)))
4290                                 abbrev = hexsz;
4291                 }
4292                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4293                             diff_abbrev_oid(&one->oid, abbrev),
4294                             diff_abbrev_oid(&two->oid, abbrev));
4295                 if (one->mode == two->mode)
4296                         strbuf_addf(msg, " %06o", one->mode);
4297                 strbuf_addf(msg, "%s\n", reset);
4298         }
4299 }
4300
4301 static void run_diff_cmd(const char *pgm,
4302                          const char *name,
4303                          const char *other,
4304                          const char *attr_path,
4305                          struct diff_filespec *one,
4306                          struct diff_filespec *two,
4307                          struct strbuf *msg,
4308                          struct diff_options *o,
4309                          struct diff_filepair *p)
4310 {
4311         const char *xfrm_msg = NULL;
4312         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4313         int must_show_header = 0;
4314
4315
4316         if (o->flags.allow_external) {
4317                 struct userdiff_driver *drv;
4318
4319                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4320                 if (drv && drv->external)
4321                         pgm = drv->external;
4322         }
4323
4324         if (msg) {
4325                 /*
4326                  * don't use colors when the header is intended for an
4327                  * external diff driver
4328                  */
4329                 fill_metainfo(msg, name, other, one, two, o, p,
4330                               &must_show_header,
4331                               want_color(o->use_color) && !pgm);
4332                 xfrm_msg = msg->len ? msg->buf : NULL;
4333         }
4334
4335         if (pgm) {
4336                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4337                                   complete_rewrite, o);
4338                 return;
4339         }
4340         if (one && two)
4341                 builtin_diff(name, other ? other : name,
4342                              one, two, xfrm_msg, must_show_header,
4343                              o, complete_rewrite);
4344         else
4345                 fprintf(o->file, "* Unmerged path %s\n", name);
4346 }
4347
4348 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4349 {
4350         if (DIFF_FILE_VALID(one)) {
4351                 if (!one->oid_valid) {
4352                         struct stat st;
4353                         if (one->is_stdin) {
4354                                 oidclr(&one->oid);
4355                                 return;
4356                         }
4357                         if (lstat(one->path, &st) < 0)
4358                                 die_errno("stat '%s'", one->path);
4359                         if (index_path(istate, &one->oid, one->path, &st, 0))
4360                                 die("cannot hash %s", one->path);
4361                 }
4362         }
4363         else
4364                 oidclr(&one->oid);
4365 }
4366
4367 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4368 {
4369         /* Strip the prefix but do not molest /dev/null and absolute paths */
4370         if (*namep && !is_absolute_path(*namep)) {
4371                 *namep += prefix_length;
4372                 if (**namep == '/')
4373                         ++*namep;
4374         }
4375         if (*otherp && !is_absolute_path(*otherp)) {
4376                 *otherp += prefix_length;
4377                 if (**otherp == '/')
4378                         ++*otherp;
4379         }
4380 }
4381
4382 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4383 {
4384         const char *pgm = external_diff();
4385         struct strbuf msg;
4386         struct diff_filespec *one = p->one;
4387         struct diff_filespec *two = p->two;
4388         const char *name;
4389         const char *other;
4390         const char *attr_path;
4391
4392         name  = one->path;
4393         other = (strcmp(name, two->path) ? two->path : NULL);
4394         attr_path = name;
4395         if (o->prefix_length)
4396                 strip_prefix(o->prefix_length, &name, &other);
4397
4398         if (!o->flags.allow_external)
4399                 pgm = NULL;
4400
4401         if (DIFF_PAIR_UNMERGED(p)) {
4402                 run_diff_cmd(pgm, name, NULL, attr_path,
4403                              NULL, NULL, NULL, o, p);
4404                 return;
4405         }
4406
4407         diff_fill_oid_info(one, o->repo->index);
4408         diff_fill_oid_info(two, o->repo->index);
4409
4410         if (!pgm &&
4411             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4412             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4413                 /*
4414                  * a filepair that changes between file and symlink
4415                  * needs to be split into deletion and creation.
4416                  */
4417                 struct diff_filespec *null = alloc_filespec(two->path);
4418                 run_diff_cmd(NULL, name, other, attr_path,
4419                              one, null, &msg,
4420                              o, p);
4421                 free(null);
4422                 strbuf_release(&msg);
4423
4424                 null = alloc_filespec(one->path);
4425                 run_diff_cmd(NULL, name, other, attr_path,
4426                              null, two, &msg, o, p);
4427                 free(null);
4428         }
4429         else
4430                 run_diff_cmd(pgm, name, other, attr_path,
4431                              one, two, &msg, o, p);
4432
4433         strbuf_release(&msg);
4434 }
4435
4436 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4437                          struct diffstat_t *diffstat)
4438 {
4439         const char *name;
4440         const char *other;
4441
4442         if (DIFF_PAIR_UNMERGED(p)) {
4443                 /* unmerged */
4444                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4445                                  diffstat, o, p);
4446                 return;
4447         }
4448
4449         name = p->one->path;
4450         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4451
4452         if (o->prefix_length)
4453                 strip_prefix(o->prefix_length, &name, &other);
4454
4455         diff_fill_oid_info(p->one, o->repo->index);
4456         diff_fill_oid_info(p->two, o->repo->index);
4457
4458         builtin_diffstat(name, other, p->one, p->two,
4459                          diffstat, o, p);
4460 }
4461
4462 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4463 {
4464         const char *name;
4465         const char *other;
4466         const char *attr_path;
4467
4468         if (DIFF_PAIR_UNMERGED(p)) {
4469                 /* unmerged */
4470                 return;
4471         }
4472
4473         name = p->one->path;
4474         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4475         attr_path = other ? other : name;
4476
4477         if (o->prefix_length)
4478                 strip_prefix(o->prefix_length, &name, &other);
4479
4480         diff_fill_oid_info(p->one, o->repo->index);
4481         diff_fill_oid_info(p->two, o->repo->index);
4482
4483         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4484 }
4485
4486 void repo_diff_setup(struct repository *r, struct diff_options *options)
4487 {
4488         memcpy(options, &default_diff_options, sizeof(*options));
4489
4490         options->file = stdout;
4491         options->repo = r;
4492
4493         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4494         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4495         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4496         options->abbrev = DEFAULT_ABBREV;
4497         options->line_termination = '\n';
4498         options->break_opt = -1;
4499         options->rename_limit = -1;
4500         options->dirstat_permille = diff_dirstat_permille_default;
4501         options->context = diff_context_default;
4502         options->interhunkcontext = diff_interhunk_context_default;
4503         options->ws_error_highlight = ws_error_highlight_default;
4504         options->flags.rename_empty = 1;
4505         options->objfind = NULL;
4506
4507         /* pathchange left =NULL by default */
4508         options->change = diff_change;
4509         options->add_remove = diff_addremove;
4510         options->use_color = diff_use_color_default;
4511         options->detect_rename = diff_detect_rename_default;
4512         options->xdl_opts |= diff_algorithm;
4513         if (diff_indent_heuristic)
4514                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4515
4516         options->orderfile = diff_order_file_cfg;
4517
4518         if (diff_no_prefix) {
4519                 options->a_prefix = options->b_prefix = "";
4520         } else if (!diff_mnemonic_prefix) {
4521                 options->a_prefix = "a/";
4522                 options->b_prefix = "b/";
4523         }
4524
4525         options->color_moved = diff_color_moved_default;
4526         options->color_moved_ws_handling = diff_color_moved_ws_default;
4527 }
4528
4529 void diff_setup_done(struct diff_options *options)
4530 {
4531         unsigned check_mask = DIFF_FORMAT_NAME |
4532                               DIFF_FORMAT_NAME_STATUS |
4533                               DIFF_FORMAT_CHECKDIFF |
4534                               DIFF_FORMAT_NO_OUTPUT;
4535         /*
4536          * This must be signed because we're comparing against a potentially
4537          * negative value.
4538          */
4539         const int hexsz = the_hash_algo->hexsz;
4540
4541         if (options->set_default)
4542                 options->set_default(options);
4543
4544         if (HAS_MULTI_BITS(options->output_format & check_mask))
4545                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4546
4547         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4548                 die(_("-G, -S and --find-object are mutually exclusive"));
4549
4550         /*
4551          * Most of the time we can say "there are changes"
4552          * only by checking if there are changed paths, but
4553          * --ignore-whitespace* options force us to look
4554          * inside contents.
4555          */
4556
4557         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4558                 options->flags.diff_from_contents = 1;
4559         else
4560                 options->flags.diff_from_contents = 0;
4561
4562         if (options->flags.find_copies_harder)
4563                 options->detect_rename = DIFF_DETECT_COPY;
4564
4565         if (!options->flags.relative_name)
4566                 options->prefix = NULL;
4567         if (options->prefix)
4568                 options->prefix_length = strlen(options->prefix);
4569         else
4570                 options->prefix_length = 0;
4571
4572         if (options->output_format & (DIFF_FORMAT_NAME |
4573                                       DIFF_FORMAT_NAME_STATUS |
4574                                       DIFF_FORMAT_CHECKDIFF |
4575                                       DIFF_FORMAT_NO_OUTPUT))
4576                 options->output_format &= ~(DIFF_FORMAT_RAW |
4577                                             DIFF_FORMAT_NUMSTAT |
4578                                             DIFF_FORMAT_DIFFSTAT |
4579                                             DIFF_FORMAT_SHORTSTAT |
4580                                             DIFF_FORMAT_DIRSTAT |
4581                                             DIFF_FORMAT_SUMMARY |
4582                                             DIFF_FORMAT_PATCH);
4583
4584         /*
4585          * These cases always need recursive; we do not drop caller-supplied
4586          * recursive bits for other formats here.
4587          */
4588         if (options->output_format & (DIFF_FORMAT_PATCH |
4589                                       DIFF_FORMAT_NUMSTAT |
4590                                       DIFF_FORMAT_DIFFSTAT |
4591                                       DIFF_FORMAT_SHORTSTAT |
4592                                       DIFF_FORMAT_DIRSTAT |
4593                                       DIFF_FORMAT_SUMMARY |
4594                                       DIFF_FORMAT_CHECKDIFF))
4595                 options->flags.recursive = 1;
4596         /*
4597          * Also pickaxe would not work very well if you do not say recursive
4598          */
4599         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4600                 options->flags.recursive = 1;
4601         /*
4602          * When patches are generated, submodules diffed against the work tree
4603          * must be checked for dirtiness too so it can be shown in the output
4604          */
4605         if (options->output_format & DIFF_FORMAT_PATCH)
4606                 options->flags.dirty_submodules = 1;
4607
4608         if (options->detect_rename && options->rename_limit < 0)
4609                 options->rename_limit = diff_rename_limit_default;
4610         if (hexsz < options->abbrev)
4611                 options->abbrev = hexsz; /* full */
4612
4613         /*
4614          * It does not make sense to show the first hit we happened
4615          * to have found.  It does not make sense not to return with
4616          * exit code in such a case either.
4617          */
4618         if (options->flags.quick) {
4619                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4620                 options->flags.exit_with_status = 1;
4621         }
4622
4623         options->diff_path_counter = 0;
4624
4625         if (options->flags.follow_renames && options->pathspec.nr != 1)
4626                 die(_("--follow requires exactly one pathspec"));
4627
4628         if (!options->use_color || external_diff())
4629                 options->color_moved = 0;
4630 }
4631
4632 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4633 {
4634         char c, *eq;
4635         int len;
4636
4637         if (*arg != '-')
4638                 return 0;
4639         c = *++arg;
4640         if (!c)
4641                 return 0;
4642         if (c == arg_short) {
4643                 c = *++arg;
4644                 if (!c)
4645                         return 1;
4646                 if (val && isdigit(c)) {
4647                         char *end;
4648                         int n = strtoul(arg, &end, 10);
4649                         if (*end)
4650                                 return 0;
4651                         *val = n;
4652                         return 1;
4653                 }
4654                 return 0;
4655         }
4656         if (c != '-')
4657                 return 0;
4658         arg++;
4659         eq = strchrnul(arg, '=');
4660         len = eq - arg;
4661         if (!len || strncmp(arg, arg_long, len))
4662                 return 0;
4663         if (*eq) {
4664                 int n;
4665                 char *end;
4666                 if (!isdigit(*++eq))
4667                         return 0;
4668                 n = strtoul(eq, &end, 10);
4669                 if (*end)
4670                         return 0;
4671                 *val = n;
4672         }
4673         return 1;
4674 }
4675
4676 static int diff_scoreopt_parse(const char *opt);
4677
4678 static inline int short_opt(char opt, const char **argv,
4679                             const char **optarg)
4680 {
4681         const char *arg = argv[0];
4682         if (arg[0] != '-' || arg[1] != opt)
4683                 return 0;
4684         if (arg[2] != '\0') {
4685                 *optarg = arg + 2;
4686                 return 1;
4687         }
4688         if (!argv[1])
4689                 die("Option '%c' requires a value", opt);
4690         *optarg = argv[1];
4691         return 2;
4692 }
4693
4694 int parse_long_opt(const char *opt, const char **argv,
4695                    const char **optarg)
4696 {
4697         const char *arg = argv[0];
4698         if (!skip_prefix(arg, "--", &arg))
4699                 return 0;
4700         if (!skip_prefix(arg, opt, &arg))
4701                 return 0;
4702         if (*arg == '=') { /* stuck form: --option=value */
4703                 *optarg = arg + 1;
4704                 return 1;
4705         }
4706         if (*arg != '\0')
4707                 return 0;
4708         /* separate form: --option value */
4709         if (!argv[1])
4710                 die("Option '--%s' requires a value", opt);
4711         *optarg = argv[1];
4712         return 2;
4713 }
4714
4715 static int stat_opt(struct diff_options *options, const char **av)
4716 {
4717         const char *arg = av[0];
4718         char *end;
4719         int width = options->stat_width;
4720         int name_width = options->stat_name_width;
4721         int graph_width = options->stat_graph_width;
4722         int count = options->stat_count;
4723         int argcount = 1;
4724
4725         if (!skip_prefix(arg, "--stat", &arg))
4726                 BUG("stat option does not begin with --stat: %s", arg);
4727         end = (char *)arg;
4728
4729         switch (*arg) {
4730         case '-':
4731                 if (skip_prefix(arg, "-width", &arg)) {
4732                         if (*arg == '=')
4733                                 width = strtoul(arg + 1, &end, 10);
4734                         else if (!*arg && !av[1])
4735                                 die_want_option("--stat-width");
4736                         else if (!*arg) {
4737                                 width = strtoul(av[1], &end, 10);
4738                                 argcount = 2;
4739                         }
4740                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4741                         if (*arg == '=')
4742                                 name_width = strtoul(arg + 1, &end, 10);
4743                         else if (!*arg && !av[1])
4744                                 die_want_option("--stat-name-width");
4745                         else if (!*arg) {
4746                                 name_width = strtoul(av[1], &end, 10);
4747                                 argcount = 2;
4748                         }
4749                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4750                         if (*arg == '=')
4751                                 graph_width = strtoul(arg + 1, &end, 10);
4752                         else if (!*arg && !av[1])
4753                                 die_want_option("--stat-graph-width");
4754                         else if (!*arg) {
4755                                 graph_width = strtoul(av[1], &end, 10);
4756                                 argcount = 2;
4757                         }
4758                 } else if (skip_prefix(arg, "-count", &arg)) {
4759                         if (*arg == '=')
4760                                 count = strtoul(arg + 1, &end, 10);
4761                         else if (!*arg && !av[1])
4762                                 die_want_option("--stat-count");
4763                         else if (!*arg) {
4764                                 count = strtoul(av[1], &end, 10);
4765                                 argcount = 2;
4766                         }
4767                 }
4768                 break;
4769         case '=':
4770                 width = strtoul(arg+1, &end, 10);
4771                 if (*end == ',')
4772                         name_width = strtoul(end+1, &end, 10);
4773                 if (*end == ',')
4774                         count = strtoul(end+1, &end, 10);
4775         }
4776
4777         /* Important! This checks all the error cases! */
4778         if (*end)
4779                 return 0;
4780         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4781         options->stat_name_width = name_width;
4782         options->stat_graph_width = graph_width;
4783         options->stat_width = width;
4784         options->stat_count = count;
4785         return argcount;
4786 }
4787
4788 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4789 {
4790         struct strbuf errmsg = STRBUF_INIT;
4791         if (parse_dirstat_params(options, params, &errmsg))
4792                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4793                     errmsg.buf);
4794         strbuf_release(&errmsg);
4795         /*
4796          * The caller knows a dirstat-related option is given from the command
4797          * line; allow it to say "return this_function();"
4798          */
4799         options->output_format |= DIFF_FORMAT_DIRSTAT;
4800         return 1;
4801 }
4802
4803 static int parse_submodule_opt(struct diff_options *options, const char *value)
4804 {
4805         if (parse_submodule_params(options, value))
4806                 die(_("Failed to parse --submodule option parameter: '%s'"),
4807                         value);
4808         return 1;
4809 }
4810
4811 static const char diff_status_letters[] = {
4812         DIFF_STATUS_ADDED,
4813         DIFF_STATUS_COPIED,
4814         DIFF_STATUS_DELETED,
4815         DIFF_STATUS_MODIFIED,
4816         DIFF_STATUS_RENAMED,
4817         DIFF_STATUS_TYPE_CHANGED,
4818         DIFF_STATUS_UNKNOWN,
4819         DIFF_STATUS_UNMERGED,
4820         DIFF_STATUS_FILTER_AON,
4821         DIFF_STATUS_FILTER_BROKEN,
4822         '\0',
4823 };
4824
4825 static unsigned int filter_bit['Z' + 1];
4826
4827 static void prepare_filter_bits(void)
4828 {
4829         int i;
4830
4831         if (!filter_bit[DIFF_STATUS_ADDED]) {
4832                 for (i = 0; diff_status_letters[i]; i++)
4833                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4834         }
4835 }
4836
4837 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4838 {
4839         return opt->filter & filter_bit[(int) status];
4840 }
4841
4842 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4843 {
4844         int i, optch;
4845
4846         prepare_filter_bits();
4847
4848         /*
4849          * If there is a negation e.g. 'd' in the input, and we haven't
4850          * initialized the filter field with another --diff-filter, start
4851          * from full set of bits, except for AON.
4852          */
4853         if (!opt->filter) {
4854                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4855                         if (optch < 'a' || 'z' < optch)
4856                                 continue;
4857                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4858                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4859                         break;
4860                 }
4861         }
4862
4863         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4864                 unsigned int bit;
4865                 int negate;
4866
4867                 if ('a' <= optch && optch <= 'z') {
4868                         negate = 1;
4869                         optch = toupper(optch);
4870                 } else {
4871                         negate = 0;
4872                 }
4873
4874                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4875                 if (!bit)
4876                         return optarg[i];
4877                 if (negate)
4878                         opt->filter &= ~bit;
4879                 else
4880                         opt->filter |= bit;
4881         }
4882         return 0;
4883 }
4884
4885 static void enable_patch_output(int *fmt) {
4886         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4887         *fmt |= DIFF_FORMAT_PATCH;
4888 }
4889
4890 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4891 {
4892         int val = parse_ws_error_highlight(arg);
4893
4894         if (val < 0) {
4895                 error("unknown value after ws-error-highlight=%.*s",
4896                       -1 - val, arg);
4897                 return 0;
4898         }
4899         opt->ws_error_highlight = val;
4900         return 1;
4901 }
4902
4903 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4904 {
4905         struct object_id oid;
4906
4907         if (get_oid(arg, &oid))
4908                 return error("unable to resolve '%s'", arg);
4909
4910         if (!opt->objfind)
4911                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4912
4913         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4914         opt->flags.recursive = 1;
4915         opt->flags.tree_in_recursive = 1;
4916         oidset_insert(opt->objfind, &oid);
4917         return 1;
4918 }
4919
4920 int diff_opt_parse(struct diff_options *options,
4921                    const char **av, int ac, const char *prefix)
4922 {
4923         const char *arg = av[0];
4924         const char *optarg;
4925         int argcount;
4926
4927         if (!prefix)
4928                 prefix = "";
4929
4930         /* Output format options */
4931         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4932             || opt_arg(arg, 'U', "unified", &options->context))
4933                 enable_patch_output(&options->output_format);
4934         else if (!strcmp(arg, "--raw"))
4935                 options->output_format |= DIFF_FORMAT_RAW;
4936         else if (!strcmp(arg, "--patch-with-raw")) {
4937                 enable_patch_output(&options->output_format);
4938                 options->output_format |= DIFF_FORMAT_RAW;
4939         } else if (!strcmp(arg, "--numstat"))
4940                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4941         else if (!strcmp(arg, "--shortstat"))
4942                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4943         else if (skip_prefix(arg, "-X", &arg) ||
4944                  skip_to_optional_arg(arg, "--dirstat", &arg))
4945                 return parse_dirstat_opt(options, arg);
4946         else if (!strcmp(arg, "--cumulative"))
4947                 return parse_dirstat_opt(options, "cumulative");
4948         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4949                 parse_dirstat_opt(options, "files");
4950                 return parse_dirstat_opt(options, arg);
4951         }
4952         else if (!strcmp(arg, "--check"))
4953                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4954         else if (!strcmp(arg, "--summary"))
4955                 options->output_format |= DIFF_FORMAT_SUMMARY;
4956         else if (!strcmp(arg, "--patch-with-stat")) {
4957                 enable_patch_output(&options->output_format);
4958                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4959         } else if (!strcmp(arg, "--name-only"))
4960                 options->output_format |= DIFF_FORMAT_NAME;
4961         else if (!strcmp(arg, "--name-status"))
4962                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4963         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4964                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4965         else if (starts_with(arg, "--stat"))
4966                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4967                 return stat_opt(options, av);
4968         else if (!strcmp(arg, "--compact-summary")) {
4969                  options->flags.stat_with_summary = 1;
4970                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4971         } else if (!strcmp(arg, "--no-compact-summary"))
4972                  options->flags.stat_with_summary = 0;
4973         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4974                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4975         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4976                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4977         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4978                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4979
4980         /* renames options */
4981         else if (starts_with(arg, "-B") ||
4982                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4983                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4984                         return error("invalid argument to -B: %s", arg+2);
4985         }
4986         else if (starts_with(arg, "-M") ||
4987                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4988                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4989                         return error("invalid argument to -M: %s", arg+2);
4990                 options->detect_rename = DIFF_DETECT_RENAME;
4991         }
4992         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4993                 options->irreversible_delete = 1;
4994         }
4995         else if (starts_with(arg, "-C") ||
4996                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4997                 if (options->detect_rename == DIFF_DETECT_COPY)
4998                         options->flags.find_copies_harder = 1;
4999                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5000                         return error("invalid argument to -C: %s", arg+2);
5001                 options->detect_rename = DIFF_DETECT_COPY;
5002         }
5003         else if (!strcmp(arg, "--no-renames"))
5004                 options->detect_rename = 0;
5005         else if (!strcmp(arg, "--rename-empty"))
5006                 options->flags.rename_empty = 1;
5007         else if (!strcmp(arg, "--no-rename-empty"))
5008                 options->flags.rename_empty = 0;
5009         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5010                 options->flags.relative_name = 1;
5011                 if (arg)
5012                         options->prefix = arg;
5013         }
5014
5015         /* xdiff options */
5016         else if (!strcmp(arg, "--minimal"))
5017                 DIFF_XDL_SET(options, NEED_MINIMAL);
5018         else if (!strcmp(arg, "--no-minimal"))
5019                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5020         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5021                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5022         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5023                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5024         else if (!strcmp(arg, "--ignore-space-at-eol"))
5025                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5026         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5027                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5028         else if (!strcmp(arg, "--ignore-blank-lines"))
5029                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5030         else if (!strcmp(arg, "--indent-heuristic"))
5031                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5032         else if (!strcmp(arg, "--no-indent-heuristic"))
5033                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5034         else if (!strcmp(arg, "--patience")) {
5035                 int i;
5036                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5037                 /*
5038                  * Both --patience and --anchored use PATIENCE_DIFF
5039                  * internally, so remove any anchors previously
5040                  * specified.
5041                  */
5042                 for (i = 0; i < options->anchors_nr; i++)
5043                         free(options->anchors[i]);
5044                 options->anchors_nr = 0;
5045         } else if (!strcmp(arg, "--histogram"))
5046                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5047         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5048                 long value = parse_algorithm_value(optarg);
5049                 if (value < 0)
5050                         return error("option diff-algorithm accepts \"myers\", "
5051                                      "\"minimal\", \"patience\" and \"histogram\"");
5052                 /* clear out previous settings */
5053                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5054                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5055                 options->xdl_opts |= value;
5056                 return argcount;
5057         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5058                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5059                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5060                            options->anchors_alloc);
5061                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5062         }
5063
5064         /* flags options */
5065         else if (!strcmp(arg, "--binary")) {
5066                 enable_patch_output(&options->output_format);
5067                 options->flags.binary = 1;
5068         }
5069         else if (!strcmp(arg, "--full-index"))
5070                 options->flags.full_index = 1;
5071         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5072                 options->flags.text = 1;
5073         else if (!strcmp(arg, "-R"))
5074                 options->flags.reverse_diff = 1;
5075         else if (!strcmp(arg, "--find-copies-harder"))
5076                 options->flags.find_copies_harder = 1;
5077         else if (!strcmp(arg, "--follow"))
5078                 options->flags.follow_renames = 1;
5079         else if (!strcmp(arg, "--no-follow")) {
5080                 options->flags.follow_renames = 0;
5081                 options->flags.default_follow_renames = 0;
5082         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5083                 int value = git_config_colorbool(NULL, arg);
5084                 if (value < 0)
5085                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5086                 options->use_color = value;
5087         }
5088         else if (!strcmp(arg, "--no-color"))
5089                 options->use_color = 0;
5090         else if (!strcmp(arg, "--color-moved")) {
5091                 if (diff_color_moved_default)
5092                         options->color_moved = diff_color_moved_default;
5093                 if (options->color_moved == COLOR_MOVED_NO)
5094                         options->color_moved = COLOR_MOVED_DEFAULT;
5095         } else if (!strcmp(arg, "--no-color-moved"))
5096                 options->color_moved = COLOR_MOVED_NO;
5097         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5098                 int cm = parse_color_moved(arg);
5099                 if (cm < 0)
5100                         die("bad --color-moved argument: %s", arg);
5101                 options->color_moved = cm;
5102         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5103                 options->color_moved_ws_handling = 0;
5104         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5105                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5106         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5107                 options->use_color = 1;
5108                 options->word_diff = DIFF_WORDS_COLOR;
5109         }
5110         else if (!strcmp(arg, "--word-diff")) {
5111                 if (options->word_diff == DIFF_WORDS_NONE)
5112                         options->word_diff = DIFF_WORDS_PLAIN;
5113         }
5114         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5115                 if (!strcmp(arg, "plain"))
5116                         options->word_diff = DIFF_WORDS_PLAIN;
5117                 else if (!strcmp(arg, "color")) {
5118                         options->use_color = 1;
5119                         options->word_diff = DIFF_WORDS_COLOR;
5120                 }
5121                 else if (!strcmp(arg, "porcelain"))
5122                         options->word_diff = DIFF_WORDS_PORCELAIN;
5123                 else if (!strcmp(arg, "none"))
5124                         options->word_diff = DIFF_WORDS_NONE;
5125                 else
5126                         die("bad --word-diff argument: %s", arg);
5127         }
5128         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5129                 if (options->word_diff == DIFF_WORDS_NONE)
5130                         options->word_diff = DIFF_WORDS_PLAIN;
5131                 options->word_regex = optarg;
5132                 return argcount;
5133         }
5134         else if (!strcmp(arg, "--exit-code"))
5135                 options->flags.exit_with_status = 1;
5136         else if (!strcmp(arg, "--quiet"))
5137                 options->flags.quick = 1;
5138         else if (!strcmp(arg, "--ext-diff"))
5139                 options->flags.allow_external = 1;
5140         else if (!strcmp(arg, "--no-ext-diff"))
5141                 options->flags.allow_external = 0;
5142         else if (!strcmp(arg, "--textconv")) {
5143                 options->flags.allow_textconv = 1;
5144                 options->flags.textconv_set_via_cmdline = 1;
5145         } else if (!strcmp(arg, "--no-textconv"))
5146                 options->flags.allow_textconv = 0;
5147         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5148                 options->flags.override_submodule_config = 1;
5149                 handle_ignore_submodules_arg(options, arg);
5150         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5151                 return parse_submodule_opt(options, arg);
5152         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5153                 return parse_ws_error_highlight_opt(options, arg);
5154         else if (!strcmp(arg, "--ita-invisible-in-index"))
5155                 options->ita_invisible_in_index = 1;
5156         else if (!strcmp(arg, "--ita-visible-in-index"))
5157                 options->ita_invisible_in_index = 0;
5158
5159         /* misc options */
5160         else if (!strcmp(arg, "-z"))
5161                 options->line_termination = 0;
5162         else if ((argcount = short_opt('l', av, &optarg))) {
5163                 options->rename_limit = strtoul(optarg, NULL, 10);
5164                 return argcount;
5165         }
5166         else if ((argcount = short_opt('S', av, &optarg))) {
5167                 options->pickaxe = optarg;
5168                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5169                 return argcount;
5170         } else if ((argcount = short_opt('G', av, &optarg))) {
5171                 options->pickaxe = optarg;
5172                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5173                 return argcount;
5174         }
5175         else if (!strcmp(arg, "--pickaxe-all"))
5176                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5177         else if (!strcmp(arg, "--pickaxe-regex"))
5178                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5179         else if ((argcount = short_opt('O', av, &optarg))) {
5180                 options->orderfile = prefix_filename(prefix, optarg);
5181                 return argcount;
5182         } else if (skip_prefix(arg, "--find-object=", &arg))
5183                 return parse_objfind_opt(options, arg);
5184         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5185                 int offending = parse_diff_filter_opt(optarg, options);
5186                 if (offending)
5187                         die("unknown change class '%c' in --diff-filter=%s",
5188                             offending, optarg);
5189                 return argcount;
5190         }
5191         else if (!strcmp(arg, "--no-abbrev"))
5192                 options->abbrev = 0;
5193         else if (!strcmp(arg, "--abbrev"))
5194                 options->abbrev = DEFAULT_ABBREV;
5195         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5196                 options->abbrev = strtoul(arg, NULL, 10);
5197                 if (options->abbrev < MINIMUM_ABBREV)
5198                         options->abbrev = MINIMUM_ABBREV;
5199                 else if (the_hash_algo->hexsz < options->abbrev)
5200                         options->abbrev = the_hash_algo->hexsz;
5201         }
5202         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5203                 options->a_prefix = optarg;
5204                 return argcount;
5205         }
5206         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5207                 options->line_prefix = optarg;
5208                 options->line_prefix_length = strlen(options->line_prefix);
5209                 graph_setup_line_prefix(options);
5210                 return argcount;
5211         }
5212         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5213                 options->b_prefix = optarg;
5214                 return argcount;
5215         }
5216         else if (!strcmp(arg, "--no-prefix"))
5217                 options->a_prefix = options->b_prefix = "";
5218         else if (opt_arg(arg, '\0', "inter-hunk-context",
5219                          &options->interhunkcontext))
5220                 ;
5221         else if (!strcmp(arg, "-W"))
5222                 options->flags.funccontext = 1;
5223         else if (!strcmp(arg, "--function-context"))
5224                 options->flags.funccontext = 1;
5225         else if (!strcmp(arg, "--no-function-context"))
5226                 options->flags.funccontext = 0;
5227         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5228                 char *path = prefix_filename(prefix, optarg);
5229                 options->file = xfopen(path, "w");
5230                 options->close_file = 1;
5231                 if (options->use_color != GIT_COLOR_ALWAYS)
5232                         options->use_color = GIT_COLOR_NEVER;
5233                 free(path);
5234                 return argcount;
5235         } else
5236                 return 0;
5237         return 1;
5238 }
5239
5240 int parse_rename_score(const char **cp_p)
5241 {
5242         unsigned long num, scale;
5243         int ch, dot;
5244         const char *cp = *cp_p;
5245
5246         num = 0;
5247         scale = 1;
5248         dot = 0;
5249         for (;;) {
5250                 ch = *cp;
5251                 if ( !dot && ch == '.' ) {
5252                         scale = 1;
5253                         dot = 1;
5254                 } else if ( ch == '%' ) {
5255                         scale = dot ? scale*100 : 100;
5256                         cp++;   /* % is always at the end */
5257                         break;
5258                 } else if ( ch >= '0' && ch <= '9' ) {
5259                         if ( scale < 100000 ) {
5260                                 scale *= 10;
5261                                 num = (num*10) + (ch-'0');
5262                         }
5263                 } else {
5264                         break;
5265                 }
5266                 cp++;
5267         }
5268         *cp_p = cp;
5269
5270         /* user says num divided by scale and we say internally that
5271          * is MAX_SCORE * num / scale.
5272          */
5273         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5274 }
5275
5276 static int diff_scoreopt_parse(const char *opt)
5277 {
5278         int opt1, opt2, cmd;
5279
5280         if (*opt++ != '-')
5281                 return -1;
5282         cmd = *opt++;
5283         if (cmd == '-') {
5284                 /* convert the long-form arguments into short-form versions */
5285                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5286                         if (*opt == 0 || *opt++ == '=')
5287                                 cmd = 'B';
5288                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5289                         if (*opt == 0 || *opt++ == '=')
5290                                 cmd = 'C';
5291                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5292                         if (*opt == 0 || *opt++ == '=')
5293                                 cmd = 'M';
5294                 }
5295         }
5296         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5297                 return -1; /* that is not a -M, -C, or -B option */
5298
5299         opt1 = parse_rename_score(&opt);
5300         if (cmd != 'B')
5301                 opt2 = 0;
5302         else {
5303                 if (*opt == 0)
5304                         opt2 = 0;
5305                 else if (*opt != '/')
5306                         return -1; /* we expect -B80/99 or -B80 */
5307                 else {
5308                         opt++;
5309                         opt2 = parse_rename_score(&opt);
5310                 }
5311         }
5312         if (*opt != 0)
5313                 return -1;
5314         return opt1 | (opt2 << 16);
5315 }
5316
5317 struct diff_queue_struct diff_queued_diff;
5318
5319 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5320 {
5321         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5322         queue->queue[queue->nr++] = dp;
5323 }
5324
5325 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5326                                  struct diff_filespec *one,
5327                                  struct diff_filespec *two)
5328 {
5329         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5330         dp->one = one;
5331         dp->two = two;
5332         if (queue)
5333                 diff_q(queue, dp);
5334         return dp;
5335 }
5336
5337 void diff_free_filepair(struct diff_filepair *p)
5338 {
5339         free_filespec(p->one);
5340         free_filespec(p->two);
5341         free(p);
5342 }
5343
5344 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5345 {
5346         int abblen;
5347         const char *abbrev;
5348
5349         /* Do we want all 40 hex characters? */
5350         if (len == the_hash_algo->hexsz)
5351                 return oid_to_hex(oid);
5352
5353         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5354         abbrev = diff_abbrev_oid(oid, len);
5355
5356         if (!print_sha1_ellipsis())
5357                 return abbrev;
5358
5359         abblen = strlen(abbrev);
5360
5361         /*
5362          * In well-behaved cases, where the abbreviated result is the
5363          * same as the requested length, append three dots after the
5364          * abbreviation (hence the whole logic is limited to the case
5365          * where abblen < 37); when the actual abbreviated result is a
5366          * bit longer than the requested length, we reduce the number
5367          * of dots so that they match the well-behaved ones.  However,
5368          * if the actual abbreviation is longer than the requested
5369          * length by more than three, we give up on aligning, and add
5370          * three dots anyway, to indicate that the output is not the
5371          * full object name.  Yes, this may be suboptimal, but this
5372          * appears only in "diff --raw --abbrev" output and it is not
5373          * worth the effort to change it now.  Note that this would
5374          * likely to work fine when the automatic sizing of default
5375          * abbreviation length is used--we would be fed -1 in "len" in
5376          * that case, and will end up always appending three-dots, but
5377          * the automatic sizing is supposed to give abblen that ensures
5378          * uniqueness across all objects (statistically speaking).
5379          */
5380         if (abblen < the_hash_algo->hexsz - 3) {
5381                 static char hex[GIT_MAX_HEXSZ + 1];
5382                 if (len < abblen && abblen <= len + 2)
5383                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5384                 else
5385                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5386                 return hex;
5387         }
5388
5389         return oid_to_hex(oid);
5390 }
5391
5392 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5393 {
5394         int line_termination = opt->line_termination;
5395         int inter_name_termination = line_termination ? '\t' : '\0';
5396
5397         fprintf(opt->file, "%s", diff_line_prefix(opt));
5398         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5399                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5400                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5401                 fprintf(opt->file, "%s ",
5402                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5403         }
5404         if (p->score) {
5405                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5406                         inter_name_termination);
5407         } else {
5408                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5409         }
5410
5411         if (p->status == DIFF_STATUS_COPIED ||
5412             p->status == DIFF_STATUS_RENAMED) {
5413                 const char *name_a, *name_b;
5414                 name_a = p->one->path;
5415                 name_b = p->two->path;
5416                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5417                 write_name_quoted(name_a, opt->file, inter_name_termination);
5418                 write_name_quoted(name_b, opt->file, line_termination);
5419         } else {
5420                 const char *name_a, *name_b;
5421                 name_a = p->one->mode ? p->one->path : p->two->path;
5422                 name_b = NULL;
5423                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5424                 write_name_quoted(name_a, opt->file, line_termination);
5425         }
5426 }
5427
5428 int diff_unmodified_pair(struct diff_filepair *p)
5429 {
5430         /* This function is written stricter than necessary to support
5431          * the currently implemented transformers, but the idea is to
5432          * let transformers to produce diff_filepairs any way they want,
5433          * and filter and clean them up here before producing the output.
5434          */
5435         struct diff_filespec *one = p->one, *two = p->two;
5436
5437         if (DIFF_PAIR_UNMERGED(p))
5438                 return 0; /* unmerged is interesting */
5439
5440         /* deletion, addition, mode or type change
5441          * and rename are all interesting.
5442          */
5443         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5444             DIFF_PAIR_MODE_CHANGED(p) ||
5445             strcmp(one->path, two->path))
5446                 return 0;
5447
5448         /* both are valid and point at the same path.  that is, we are
5449          * dealing with a change.
5450          */
5451         if (one->oid_valid && two->oid_valid &&
5452             oideq(&one->oid, &two->oid) &&
5453             !one->dirty_submodule && !two->dirty_submodule)
5454                 return 1; /* no change */
5455         if (!one->oid_valid && !two->oid_valid)
5456                 return 1; /* both look at the same file on the filesystem. */
5457         return 0;
5458 }
5459
5460 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5461 {
5462         if (diff_unmodified_pair(p))
5463                 return;
5464
5465         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5466             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5467                 return; /* no tree diffs in patch format */
5468
5469         run_diff(p, o);
5470 }
5471
5472 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5473                             struct diffstat_t *diffstat)
5474 {
5475         if (diff_unmodified_pair(p))
5476                 return;
5477
5478         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5479             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5480                 return; /* no useful stat for tree diffs */
5481
5482         run_diffstat(p, o, diffstat);
5483 }
5484
5485 static void diff_flush_checkdiff(struct diff_filepair *p,
5486                 struct diff_options *o)
5487 {
5488         if (diff_unmodified_pair(p))
5489                 return;
5490
5491         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5492             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5493                 return; /* nothing to check in tree diffs */
5494
5495         run_checkdiff(p, o);
5496 }
5497
5498 int diff_queue_is_empty(void)
5499 {
5500         struct diff_queue_struct *q = &diff_queued_diff;
5501         int i;
5502         for (i = 0; i < q->nr; i++)
5503                 if (!diff_unmodified_pair(q->queue[i]))
5504                         return 0;
5505         return 1;
5506 }
5507
5508 #if DIFF_DEBUG
5509 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5510 {
5511         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5512                 x, one ? one : "",
5513                 s->path,
5514                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5515                 s->mode,
5516                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5517         fprintf(stderr, "queue[%d] %s size %lu\n",
5518                 x, one ? one : "",
5519                 s->size);
5520 }
5521
5522 void diff_debug_filepair(const struct diff_filepair *p, int i)
5523 {
5524         diff_debug_filespec(p->one, i, "one");
5525         diff_debug_filespec(p->two, i, "two");
5526         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5527                 p->score, p->status ? p->status : '?',
5528                 p->one->rename_used, p->broken_pair);
5529 }
5530
5531 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5532 {
5533         int i;
5534         if (msg)
5535                 fprintf(stderr, "%s\n", msg);
5536         fprintf(stderr, "q->nr = %d\n", q->nr);
5537         for (i = 0; i < q->nr; i++) {
5538                 struct diff_filepair *p = q->queue[i];
5539                 diff_debug_filepair(p, i);
5540         }
5541 }
5542 #endif
5543
5544 static void diff_resolve_rename_copy(void)
5545 {
5546         int i;
5547         struct diff_filepair *p;
5548         struct diff_queue_struct *q = &diff_queued_diff;
5549
5550         diff_debug_queue("resolve-rename-copy", q);
5551
5552         for (i = 0; i < q->nr; i++) {
5553                 p = q->queue[i];
5554                 p->status = 0; /* undecided */
5555                 if (DIFF_PAIR_UNMERGED(p))
5556                         p->status = DIFF_STATUS_UNMERGED;
5557                 else if (!DIFF_FILE_VALID(p->one))
5558                         p->status = DIFF_STATUS_ADDED;
5559                 else if (!DIFF_FILE_VALID(p->two))
5560                         p->status = DIFF_STATUS_DELETED;
5561                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5562                         p->status = DIFF_STATUS_TYPE_CHANGED;
5563
5564                 /* from this point on, we are dealing with a pair
5565                  * whose both sides are valid and of the same type, i.e.
5566                  * either in-place edit or rename/copy edit.
5567                  */
5568                 else if (DIFF_PAIR_RENAME(p)) {
5569                         /*
5570                          * A rename might have re-connected a broken
5571                          * pair up, causing the pathnames to be the
5572                          * same again. If so, that's not a rename at
5573                          * all, just a modification..
5574                          *
5575                          * Otherwise, see if this source was used for
5576                          * multiple renames, in which case we decrement
5577                          * the count, and call it a copy.
5578                          */
5579                         if (!strcmp(p->one->path, p->two->path))
5580                                 p->status = DIFF_STATUS_MODIFIED;
5581                         else if (--p->one->rename_used > 0)
5582                                 p->status = DIFF_STATUS_COPIED;
5583                         else
5584                                 p->status = DIFF_STATUS_RENAMED;
5585                 }
5586                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5587                          p->one->mode != p->two->mode ||
5588                          p->one->dirty_submodule ||
5589                          p->two->dirty_submodule ||
5590                          is_null_oid(&p->one->oid))
5591                         p->status = DIFF_STATUS_MODIFIED;
5592                 else {
5593                         /* This is a "no-change" entry and should not
5594                          * happen anymore, but prepare for broken callers.
5595                          */
5596                         error("feeding unmodified %s to diffcore",
5597                               p->one->path);
5598                         p->status = DIFF_STATUS_UNKNOWN;
5599                 }
5600         }
5601         diff_debug_queue("resolve-rename-copy done", q);
5602 }
5603
5604 static int check_pair_status(struct diff_filepair *p)
5605 {
5606         switch (p->status) {
5607         case DIFF_STATUS_UNKNOWN:
5608                 return 0;
5609         case 0:
5610                 die("internal error in diff-resolve-rename-copy");
5611         default:
5612                 return 1;
5613         }
5614 }
5615
5616 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5617 {
5618         int fmt = opt->output_format;
5619
5620         if (fmt & DIFF_FORMAT_CHECKDIFF)
5621                 diff_flush_checkdiff(p, opt);
5622         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5623                 diff_flush_raw(p, opt);
5624         else if (fmt & DIFF_FORMAT_NAME) {
5625                 const char *name_a, *name_b;
5626                 name_a = p->two->path;
5627                 name_b = NULL;
5628                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5629                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5630                 write_name_quoted(name_a, opt->file, opt->line_termination);
5631         }
5632 }
5633
5634 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5635 {
5636         struct strbuf sb = STRBUF_INIT;
5637         if (fs->mode)
5638                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5639         else
5640                 strbuf_addf(&sb, " %s ", newdelete);
5641
5642         quote_c_style(fs->path, &sb, NULL, 0);
5643         strbuf_addch(&sb, '\n');
5644         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5645                          sb.buf, sb.len, 0);
5646         strbuf_release(&sb);
5647 }
5648
5649 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5650                 int show_name)
5651 {
5652         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5653                 struct strbuf sb = STRBUF_INIT;
5654                 strbuf_addf(&sb, " mode change %06o => %06o",
5655                             p->one->mode, p->two->mode);
5656                 if (show_name) {
5657                         strbuf_addch(&sb, ' ');
5658                         quote_c_style(p->two->path, &sb, NULL, 0);
5659                 }
5660                 strbuf_addch(&sb, '\n');
5661                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5662                                  sb.buf, sb.len, 0);
5663                 strbuf_release(&sb);
5664         }
5665 }
5666
5667 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5668                 struct diff_filepair *p)
5669 {
5670         struct strbuf sb = STRBUF_INIT;
5671         struct strbuf names = STRBUF_INIT;
5672
5673         pprint_rename(&names, p->one->path, p->two->path);
5674         strbuf_addf(&sb, " %s %s (%d%%)\n",
5675                     renamecopy, names.buf, similarity_index(p));
5676         strbuf_release(&names);
5677         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5678                                  sb.buf, sb.len, 0);
5679         show_mode_change(opt, p, 0);
5680         strbuf_release(&sb);
5681 }
5682
5683 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5684 {
5685         switch(p->status) {
5686         case DIFF_STATUS_DELETED:
5687                 show_file_mode_name(opt, "delete", p->one);
5688                 break;
5689         case DIFF_STATUS_ADDED:
5690                 show_file_mode_name(opt, "create", p->two);
5691                 break;
5692         case DIFF_STATUS_COPIED:
5693                 show_rename_copy(opt, "copy", p);
5694                 break;
5695         case DIFF_STATUS_RENAMED:
5696                 show_rename_copy(opt, "rename", p);
5697                 break;
5698         default:
5699                 if (p->score) {
5700                         struct strbuf sb = STRBUF_INIT;
5701                         strbuf_addstr(&sb, " rewrite ");
5702                         quote_c_style(p->two->path, &sb, NULL, 0);
5703                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5704                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5705                                          sb.buf, sb.len, 0);
5706                         strbuf_release(&sb);
5707                 }
5708                 show_mode_change(opt, p, !p->score);
5709                 break;
5710         }
5711 }
5712
5713 struct patch_id_t {
5714         git_SHA_CTX *ctx;
5715         int patchlen;
5716 };
5717
5718 static int remove_space(char *line, int len)
5719 {
5720         int i;
5721         char *dst = line;
5722         unsigned char c;
5723
5724         for (i = 0; i < len; i++)
5725                 if (!isspace((c = line[i])))
5726                         *dst++ = c;
5727
5728         return dst - line;
5729 }
5730
5731 static void patch_id_consume(void *priv, char *line, unsigned long len)
5732 {
5733         struct patch_id_t *data = priv;
5734         int new_len;
5735
5736         new_len = remove_space(line, len);
5737
5738         git_SHA1_Update(data->ctx, line, new_len);
5739         data->patchlen += new_len;
5740 }
5741
5742 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5743 {
5744         git_SHA1_Update(ctx, str, strlen(str));
5745 }
5746
5747 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5748 {
5749         /* large enough for 2^32 in octal */
5750         char buf[12];
5751         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5752         git_SHA1_Update(ctx, buf, len);
5753 }
5754
5755 /* returns 0 upon success, and writes result into sha1 */
5756 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5757 {
5758         struct diff_queue_struct *q = &diff_queued_diff;
5759         int i;
5760         git_SHA_CTX ctx;
5761         struct patch_id_t data;
5762
5763         git_SHA1_Init(&ctx);
5764         memset(&data, 0, sizeof(struct patch_id_t));
5765         data.ctx = &ctx;
5766
5767         for (i = 0; i < q->nr; i++) {
5768                 xpparam_t xpp;
5769                 xdemitconf_t xecfg;
5770                 mmfile_t mf1, mf2;
5771                 struct diff_filepair *p = q->queue[i];
5772                 int len1, len2;
5773
5774                 memset(&xpp, 0, sizeof(xpp));
5775                 memset(&xecfg, 0, sizeof(xecfg));
5776                 if (p->status == 0)
5777                         return error("internal diff status error");
5778                 if (p->status == DIFF_STATUS_UNKNOWN)
5779                         continue;
5780                 if (diff_unmodified_pair(p))
5781                         continue;
5782                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5783                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5784                         continue;
5785                 if (DIFF_PAIR_UNMERGED(p))
5786                         continue;
5787
5788                 diff_fill_oid_info(p->one, options->repo->index);
5789                 diff_fill_oid_info(p->two, options->repo->index);
5790
5791                 len1 = remove_space(p->one->path, strlen(p->one->path));
5792                 len2 = remove_space(p->two->path, strlen(p->two->path));
5793                 patch_id_add_string(&ctx, "diff--git");
5794                 patch_id_add_string(&ctx, "a/");
5795                 git_SHA1_Update(&ctx, p->one->path, len1);
5796                 patch_id_add_string(&ctx, "b/");
5797                 git_SHA1_Update(&ctx, p->two->path, len2);
5798
5799                 if (p->one->mode == 0) {
5800                         patch_id_add_string(&ctx, "newfilemode");
5801                         patch_id_add_mode(&ctx, p->two->mode);
5802                         patch_id_add_string(&ctx, "---/dev/null");
5803                         patch_id_add_string(&ctx, "+++b/");
5804                         git_SHA1_Update(&ctx, p->two->path, len2);
5805                 } else if (p->two->mode == 0) {
5806                         patch_id_add_string(&ctx, "deletedfilemode");
5807                         patch_id_add_mode(&ctx, p->one->mode);
5808                         patch_id_add_string(&ctx, "---a/");
5809                         git_SHA1_Update(&ctx, p->one->path, len1);
5810                         patch_id_add_string(&ctx, "+++/dev/null");
5811                 } else {
5812                         patch_id_add_string(&ctx, "---a/");
5813                         git_SHA1_Update(&ctx, p->one->path, len1);
5814                         patch_id_add_string(&ctx, "+++b/");
5815                         git_SHA1_Update(&ctx, p->two->path, len2);
5816                 }
5817
5818                 if (diff_header_only)
5819                         continue;
5820
5821                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5822                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5823                         return error("unable to read files to diff");
5824
5825                 if (diff_filespec_is_binary(options->repo, p->one) ||
5826                     diff_filespec_is_binary(options->repo, p->two)) {
5827                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5828                                         GIT_SHA1_HEXSZ);
5829                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5830                                         GIT_SHA1_HEXSZ);
5831                         continue;
5832                 }
5833
5834                 xpp.flags = 0;
5835                 xecfg.ctxlen = 3;
5836                 xecfg.flags = 0;
5837                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5838                                   patch_id_consume, &data, &xpp, &xecfg))
5839                         return error("unable to generate patch-id diff for %s",
5840                                      p->one->path);
5841         }
5842
5843         git_SHA1_Final(oid->hash, &ctx);
5844         return 0;
5845 }
5846
5847 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5848 {
5849         struct diff_queue_struct *q = &diff_queued_diff;
5850         int i;
5851         int result = diff_get_patch_id(options, oid, diff_header_only);
5852
5853         for (i = 0; i < q->nr; i++)
5854                 diff_free_filepair(q->queue[i]);
5855
5856         free(q->queue);
5857         DIFF_QUEUE_CLEAR(q);
5858
5859         return result;
5860 }
5861
5862 static int is_summary_empty(const struct diff_queue_struct *q)
5863 {
5864         int i;
5865
5866         for (i = 0; i < q->nr; i++) {
5867                 const struct diff_filepair *p = q->queue[i];
5868
5869                 switch (p->status) {
5870                 case DIFF_STATUS_DELETED:
5871                 case DIFF_STATUS_ADDED:
5872                 case DIFF_STATUS_COPIED:
5873                 case DIFF_STATUS_RENAMED:
5874                         return 0;
5875                 default:
5876                         if (p->score)
5877                                 return 0;
5878                         if (p->one->mode && p->two->mode &&
5879                             p->one->mode != p->two->mode)
5880                                 return 0;
5881                         break;
5882                 }
5883         }
5884         return 1;
5885 }
5886
5887 static const char rename_limit_warning[] =
5888 N_("inexact rename detection was skipped due to too many files.");
5889
5890 static const char degrade_cc_to_c_warning[] =
5891 N_("only found copies from modified paths due to too many files.");
5892
5893 static const char rename_limit_advice[] =
5894 N_("you may want to set your %s variable to at least "
5895    "%d and retry the command.");
5896
5897 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5898 {
5899         fflush(stdout);
5900         if (degraded_cc)
5901                 warning(_(degrade_cc_to_c_warning));
5902         else if (needed)
5903                 warning(_(rename_limit_warning));
5904         else
5905                 return;
5906         if (0 < needed)
5907                 warning(_(rename_limit_advice), varname, needed);
5908 }
5909
5910 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5911 {
5912         int i;
5913         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5914         struct diff_queue_struct *q = &diff_queued_diff;
5915
5916         if (WSEH_NEW & WS_RULE_MASK)
5917                 BUG("WS rules bit mask overlaps with diff symbol flags");
5918
5919         if (o->color_moved)
5920                 o->emitted_symbols = &esm;
5921
5922         for (i = 0; i < q->nr; i++) {
5923                 struct diff_filepair *p = q->queue[i];
5924                 if (check_pair_status(p))
5925                         diff_flush_patch(p, o);
5926         }
5927
5928         if (o->emitted_symbols) {
5929                 if (o->color_moved) {
5930                         struct hashmap add_lines, del_lines;
5931
5932                         if (o->color_moved_ws_handling &
5933                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5934                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5935
5936                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5937                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5938
5939                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5940                         mark_color_as_moved(o, &add_lines, &del_lines);
5941                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5942                                 dim_moved_lines(o);
5943
5944                         hashmap_free(&add_lines, 1);
5945                         hashmap_free(&del_lines, 1);
5946                 }
5947
5948                 for (i = 0; i < esm.nr; i++)
5949                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5950
5951                 for (i = 0; i < esm.nr; i++)
5952                         free((void *)esm.buf[i].line);
5953         }
5954         esm.nr = 0;
5955 }
5956
5957 void diff_flush(struct diff_options *options)
5958 {
5959         struct diff_queue_struct *q = &diff_queued_diff;
5960         int i, output_format = options->output_format;
5961         int separator = 0;
5962         int dirstat_by_line = 0;
5963
5964         /*
5965          * Order: raw, stat, summary, patch
5966          * or:    name/name-status/checkdiff (other bits clear)
5967          */
5968         if (!q->nr)
5969                 goto free_queue;
5970
5971         if (output_format & (DIFF_FORMAT_RAW |
5972                              DIFF_FORMAT_NAME |
5973                              DIFF_FORMAT_NAME_STATUS |
5974                              DIFF_FORMAT_CHECKDIFF)) {
5975                 for (i = 0; i < q->nr; i++) {
5976                         struct diff_filepair *p = q->queue[i];
5977                         if (check_pair_status(p))
5978                                 flush_one_pair(p, options);
5979                 }
5980                 separator++;
5981         }
5982
5983         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5984                 dirstat_by_line = 1;
5985
5986         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5987             dirstat_by_line) {
5988                 struct diffstat_t diffstat;
5989
5990                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5991                 for (i = 0; i < q->nr; i++) {
5992                         struct diff_filepair *p = q->queue[i];
5993                         if (check_pair_status(p))
5994                                 diff_flush_stat(p, options, &diffstat);
5995                 }
5996                 if (output_format & DIFF_FORMAT_NUMSTAT)
5997                         show_numstat(&diffstat, options);
5998                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5999                         show_stats(&diffstat, options);
6000                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6001                         show_shortstats(&diffstat, options);
6002                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6003                         show_dirstat_by_line(&diffstat, options);
6004                 free_diffstat_info(&diffstat);
6005                 separator++;
6006         }
6007         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6008                 show_dirstat(options);
6009
6010         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6011                 for (i = 0; i < q->nr; i++) {
6012                         diff_summary(options, q->queue[i]);
6013                 }
6014                 separator++;
6015         }
6016
6017         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6018             options->flags.exit_with_status &&
6019             options->flags.diff_from_contents) {
6020                 /*
6021                  * run diff_flush_patch for the exit status. setting
6022                  * options->file to /dev/null should be safe, because we
6023                  * aren't supposed to produce any output anyway.
6024                  */
6025                 if (options->close_file)
6026                         fclose(options->file);
6027                 options->file = xfopen("/dev/null", "w");
6028                 options->close_file = 1;
6029                 options->color_moved = 0;
6030                 for (i = 0; i < q->nr; i++) {
6031                         struct diff_filepair *p = q->queue[i];
6032                         if (check_pair_status(p))
6033                                 diff_flush_patch(p, options);
6034                         if (options->found_changes)
6035                                 break;
6036                 }
6037         }
6038
6039         if (output_format & DIFF_FORMAT_PATCH) {
6040                 if (separator) {
6041                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6042                         if (options->stat_sep)
6043                                 /* attach patch instead of inline */
6044                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6045                                                  NULL, 0, 0);
6046                 }
6047
6048                 diff_flush_patch_all_file_pairs(options);
6049         }
6050
6051         if (output_format & DIFF_FORMAT_CALLBACK)
6052                 options->format_callback(q, options, options->format_callback_data);
6053
6054         for (i = 0; i < q->nr; i++)
6055                 diff_free_filepair(q->queue[i]);
6056 free_queue:
6057         free(q->queue);
6058         DIFF_QUEUE_CLEAR(q);
6059         if (options->close_file)
6060                 fclose(options->file);
6061
6062         /*
6063          * Report the content-level differences with HAS_CHANGES;
6064          * diff_addremove/diff_change does not set the bit when
6065          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6066          */
6067         if (options->flags.diff_from_contents) {
6068                 if (options->found_changes)
6069                         options->flags.has_changes = 1;
6070                 else
6071                         options->flags.has_changes = 0;
6072         }
6073 }
6074
6075 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6076 {
6077         return (((p->status == DIFF_STATUS_MODIFIED) &&
6078                  ((p->score &&
6079                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6080                   (!p->score &&
6081                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6082                 ((p->status != DIFF_STATUS_MODIFIED) &&
6083                  filter_bit_tst(p->status, options)));
6084 }
6085
6086 static void diffcore_apply_filter(struct diff_options *options)
6087 {
6088         int i;
6089         struct diff_queue_struct *q = &diff_queued_diff;
6090         struct diff_queue_struct outq;
6091
6092         DIFF_QUEUE_CLEAR(&outq);
6093
6094         if (!options->filter)
6095                 return;
6096
6097         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6098                 int found;
6099                 for (i = found = 0; !found && i < q->nr; i++) {
6100                         if (match_filter(options, q->queue[i]))
6101                                 found++;
6102                 }
6103                 if (found)
6104                         return;
6105
6106                 /* otherwise we will clear the whole queue
6107                  * by copying the empty outq at the end of this
6108                  * function, but first clear the current entries
6109                  * in the queue.
6110                  */
6111                 for (i = 0; i < q->nr; i++)
6112                         diff_free_filepair(q->queue[i]);
6113         }
6114         else {
6115                 /* Only the matching ones */
6116                 for (i = 0; i < q->nr; i++) {
6117                         struct diff_filepair *p = q->queue[i];
6118                         if (match_filter(options, p))
6119                                 diff_q(&outq, p);
6120                         else
6121                                 diff_free_filepair(p);
6122                 }
6123         }
6124         free(q->queue);
6125         *q = outq;
6126 }
6127
6128 /* Check whether two filespecs with the same mode and size are identical */
6129 static int diff_filespec_is_identical(struct repository *r,
6130                                       struct diff_filespec *one,
6131                                       struct diff_filespec *two)
6132 {
6133         if (S_ISGITLINK(one->mode))
6134                 return 0;
6135         if (diff_populate_filespec(r, one, 0))
6136                 return 0;
6137         if (diff_populate_filespec(r, two, 0))
6138                 return 0;
6139         return !memcmp(one->data, two->data, one->size);
6140 }
6141
6142 static int diff_filespec_check_stat_unmatch(struct repository *r,
6143                                             struct diff_filepair *p)
6144 {
6145         if (p->done_skip_stat_unmatch)
6146                 return p->skip_stat_unmatch_result;
6147
6148         p->done_skip_stat_unmatch = 1;
6149         p->skip_stat_unmatch_result = 0;
6150         /*
6151          * 1. Entries that come from stat info dirtiness
6152          *    always have both sides (iow, not create/delete),
6153          *    one side of the object name is unknown, with
6154          *    the same mode and size.  Keep the ones that
6155          *    do not match these criteria.  They have real
6156          *    differences.
6157          *
6158          * 2. At this point, the file is known to be modified,
6159          *    with the same mode and size, and the object
6160          *    name of one side is unknown.  Need to inspect
6161          *    the identical contents.
6162          */
6163         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6164             !DIFF_FILE_VALID(p->two) ||
6165             (p->one->oid_valid && p->two->oid_valid) ||
6166             (p->one->mode != p->two->mode) ||
6167             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6168             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6169             (p->one->size != p->two->size) ||
6170             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6171                 p->skip_stat_unmatch_result = 1;
6172         return p->skip_stat_unmatch_result;
6173 }
6174
6175 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6176 {
6177         int i;
6178         struct diff_queue_struct *q = &diff_queued_diff;
6179         struct diff_queue_struct outq;
6180         DIFF_QUEUE_CLEAR(&outq);
6181
6182         for (i = 0; i < q->nr; i++) {
6183                 struct diff_filepair *p = q->queue[i];
6184
6185                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6186                         diff_q(&outq, p);
6187                 else {
6188                         /*
6189                          * The caller can subtract 1 from skip_stat_unmatch
6190                          * to determine how many paths were dirty only
6191                          * due to stat info mismatch.
6192                          */
6193                         if (!diffopt->flags.no_index)
6194                                 diffopt->skip_stat_unmatch++;
6195                         diff_free_filepair(p);
6196                 }
6197         }
6198         free(q->queue);
6199         *q = outq;
6200 }
6201
6202 static int diffnamecmp(const void *a_, const void *b_)
6203 {
6204         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6205         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6206         const char *name_a, *name_b;
6207
6208         name_a = a->one ? a->one->path : a->two->path;
6209         name_b = b->one ? b->one->path : b->two->path;
6210         return strcmp(name_a, name_b);
6211 }
6212
6213 void diffcore_fix_diff_index(struct diff_options *options)
6214 {
6215         struct diff_queue_struct *q = &diff_queued_diff;
6216         QSORT(q->queue, q->nr, diffnamecmp);
6217 }
6218
6219 void diffcore_std(struct diff_options *options)
6220 {
6221         /* NOTE please keep the following in sync with diff_tree_combined() */
6222         if (options->skip_stat_unmatch)
6223                 diffcore_skip_stat_unmatch(options);
6224         if (!options->found_follow) {
6225                 /* See try_to_follow_renames() in tree-diff.c */
6226                 if (options->break_opt != -1)
6227                         diffcore_break(options->repo,
6228                                        options->break_opt);
6229                 if (options->detect_rename)
6230                         diffcore_rename(options);
6231                 if (options->break_opt != -1)
6232                         diffcore_merge_broken();
6233         }
6234         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6235                 diffcore_pickaxe(options);
6236         if (options->orderfile)
6237                 diffcore_order(options->orderfile);
6238         if (!options->found_follow)
6239                 /* See try_to_follow_renames() in tree-diff.c */
6240                 diff_resolve_rename_copy();
6241         diffcore_apply_filter(options);
6242
6243         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6244                 options->flags.has_changes = 1;
6245         else
6246                 options->flags.has_changes = 0;
6247
6248         options->found_follow = 0;
6249 }
6250
6251 int diff_result_code(struct diff_options *opt, int status)
6252 {
6253         int result = 0;
6254
6255         diff_warn_rename_limit("diff.renameLimit",
6256                                opt->needed_rename_limit,
6257                                opt->degraded_cc_to_c);
6258         if (!opt->flags.exit_with_status &&
6259             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6260                 return status;
6261         if (opt->flags.exit_with_status &&
6262             opt->flags.has_changes)
6263                 result |= 01;
6264         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6265             opt->flags.check_failed)
6266                 result |= 02;
6267         return result;
6268 }
6269
6270 int diff_can_quit_early(struct diff_options *opt)
6271 {
6272         return (opt->flags.quick &&
6273                 !opt->filter &&
6274                 opt->flags.has_changes);
6275 }
6276
6277 /*
6278  * Shall changes to this submodule be ignored?
6279  *
6280  * Submodule changes can be configured to be ignored separately for each path,
6281  * but that configuration can be overridden from the command line.
6282  */
6283 static int is_submodule_ignored(const char *path, struct diff_options *options)
6284 {
6285         int ignored = 0;
6286         struct diff_flags orig_flags = options->flags;
6287         if (!options->flags.override_submodule_config)
6288                 set_diffopt_flags_from_submodule_config(options, path);
6289         if (options->flags.ignore_submodules)
6290                 ignored = 1;
6291         options->flags = orig_flags;
6292         return ignored;
6293 }
6294
6295 void diff_addremove(struct diff_options *options,
6296                     int addremove, unsigned mode,
6297                     const struct object_id *oid,
6298                     int oid_valid,
6299                     const char *concatpath, unsigned dirty_submodule)
6300 {
6301         struct diff_filespec *one, *two;
6302
6303         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6304                 return;
6305
6306         /* This may look odd, but it is a preparation for
6307          * feeding "there are unchanged files which should
6308          * not produce diffs, but when you are doing copy
6309          * detection you would need them, so here they are"
6310          * entries to the diff-core.  They will be prefixed
6311          * with something like '=' or '*' (I haven't decided
6312          * which but should not make any difference).
6313          * Feeding the same new and old to diff_change()
6314          * also has the same effect.
6315          * Before the final output happens, they are pruned after
6316          * merged into rename/copy pairs as appropriate.
6317          */
6318         if (options->flags.reverse_diff)
6319                 addremove = (addremove == '+' ? '-' :
6320                              addremove == '-' ? '+' : addremove);
6321
6322         if (options->prefix &&
6323             strncmp(concatpath, options->prefix, options->prefix_length))
6324                 return;
6325
6326         one = alloc_filespec(concatpath);
6327         two = alloc_filespec(concatpath);
6328
6329         if (addremove != '+')
6330                 fill_filespec(one, oid, oid_valid, mode);
6331         if (addremove != '-') {
6332                 fill_filespec(two, oid, oid_valid, mode);
6333                 two->dirty_submodule = dirty_submodule;
6334         }
6335
6336         diff_queue(&diff_queued_diff, one, two);
6337         if (!options->flags.diff_from_contents)
6338                 options->flags.has_changes = 1;
6339 }
6340
6341 void diff_change(struct diff_options *options,
6342                  unsigned old_mode, unsigned new_mode,
6343                  const struct object_id *old_oid,
6344                  const struct object_id *new_oid,
6345                  int old_oid_valid, int new_oid_valid,
6346                  const char *concatpath,
6347                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6348 {
6349         struct diff_filespec *one, *two;
6350         struct diff_filepair *p;
6351
6352         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6353             is_submodule_ignored(concatpath, options))
6354                 return;
6355
6356         if (options->flags.reverse_diff) {
6357                 SWAP(old_mode, new_mode);
6358                 SWAP(old_oid, new_oid);
6359                 SWAP(old_oid_valid, new_oid_valid);
6360                 SWAP(old_dirty_submodule, new_dirty_submodule);
6361         }
6362
6363         if (options->prefix &&
6364             strncmp(concatpath, options->prefix, options->prefix_length))
6365                 return;
6366
6367         one = alloc_filespec(concatpath);
6368         two = alloc_filespec(concatpath);
6369         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6370         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6371         one->dirty_submodule = old_dirty_submodule;
6372         two->dirty_submodule = new_dirty_submodule;
6373         p = diff_queue(&diff_queued_diff, one, two);
6374
6375         if (options->flags.diff_from_contents)
6376                 return;
6377
6378         if (options->flags.quick && options->skip_stat_unmatch &&
6379             !diff_filespec_check_stat_unmatch(options->repo, p))
6380                 return;
6381
6382         options->flags.has_changes = 1;
6383 }
6384
6385 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6386 {
6387         struct diff_filepair *pair;
6388         struct diff_filespec *one, *two;
6389
6390         if (options->prefix &&
6391             strncmp(path, options->prefix, options->prefix_length))
6392                 return NULL;
6393
6394         one = alloc_filespec(path);
6395         two = alloc_filespec(path);
6396         pair = diff_queue(&diff_queued_diff, one, two);
6397         pair->is_unmerged = 1;
6398         return pair;
6399 }
6400
6401 static char *run_textconv(struct repository *r,
6402                           const char *pgm,
6403                           struct diff_filespec *spec,
6404                           size_t *outsize)
6405 {
6406         struct diff_tempfile *temp;
6407         const char *argv[3];
6408         const char **arg = argv;
6409         struct child_process child = CHILD_PROCESS_INIT;
6410         struct strbuf buf = STRBUF_INIT;
6411         int err = 0;
6412
6413         temp = prepare_temp_file(r, spec->path, spec);
6414         *arg++ = pgm;
6415         *arg++ = temp->name;
6416         *arg = NULL;
6417
6418         child.use_shell = 1;
6419         child.argv = argv;
6420         child.out = -1;
6421         if (start_command(&child)) {
6422                 remove_tempfile();
6423                 return NULL;
6424         }
6425
6426         if (strbuf_read(&buf, child.out, 0) < 0)
6427                 err = error("error reading from textconv command '%s'", pgm);
6428         close(child.out);
6429
6430         if (finish_command(&child) || err) {
6431                 strbuf_release(&buf);
6432                 remove_tempfile();
6433                 return NULL;
6434         }
6435         remove_tempfile();
6436
6437         return strbuf_detach(&buf, outsize);
6438 }
6439
6440 size_t fill_textconv(struct repository *r,
6441                      struct userdiff_driver *driver,
6442                      struct diff_filespec *df,
6443                      char **outbuf)
6444 {
6445         size_t size;
6446
6447         if (!driver) {
6448                 if (!DIFF_FILE_VALID(df)) {
6449                         *outbuf = "";
6450                         return 0;
6451                 }
6452                 if (diff_populate_filespec(r, df, 0))
6453                         die("unable to read files to diff");
6454                 *outbuf = df->data;
6455                 return df->size;
6456         }
6457
6458         if (!driver->textconv)
6459                 BUG("fill_textconv called with non-textconv driver");
6460
6461         if (driver->textconv_cache && df->oid_valid) {
6462                 *outbuf = notes_cache_get(driver->textconv_cache,
6463                                           &df->oid,
6464                                           &size);
6465                 if (*outbuf)
6466                         return size;
6467         }
6468
6469         *outbuf = run_textconv(r, driver->textconv, df, &size);
6470         if (!*outbuf)
6471                 die("unable to read files to diff");
6472
6473         if (driver->textconv_cache && df->oid_valid) {
6474                 /* ignore errors, as we might be in a readonly repository */
6475                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6476                                 size);
6477                 /*
6478                  * we could save up changes and flush them all at the end,
6479                  * but we would need an extra call after all diffing is done.
6480                  * Since generating a cache entry is the slow path anyway,
6481                  * this extra overhead probably isn't a big deal.
6482                  */
6483                 notes_cache_write(driver->textconv_cache);
6484         }
6485
6486         return size;
6487 }
6488
6489 int textconv_object(struct repository *r,
6490                     const char *path,
6491                     unsigned mode,
6492                     const struct object_id *oid,
6493                     int oid_valid,
6494                     char **buf,
6495                     unsigned long *buf_size)
6496 {
6497         struct diff_filespec *df;
6498         struct userdiff_driver *textconv;
6499
6500         df = alloc_filespec(path);
6501         fill_filespec(df, oid, oid_valid, mode);
6502         textconv = get_textconv(r->index, df);
6503         if (!textconv) {
6504                 free_filespec(df);
6505                 return 0;
6506         }
6507
6508         *buf_size = fill_textconv(r, textconv, df, buf);
6509         free_filespec(df);
6510         return 1;
6511 }
6512
6513 void setup_diff_pager(struct diff_options *opt)
6514 {
6515         /*
6516          * If the user asked for our exit code, then either they want --quiet
6517          * or --exit-code. We should definitely not bother with a pager in the
6518          * former case, as we will generate no output. Since we still properly
6519          * report our exit code even when a pager is run, we _could_ run a
6520          * pager with --exit-code. But since we have not done so historically,
6521          * and because it is easy to find people oneline advising "git diff
6522          * --exit-code" in hooks and other scripts, we do not do so.
6523          */
6524         if (!opt->flags.exit_with_status &&
6525             check_pager_config("diff") != 0)
6526                 setup_pager();
6527 }