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 "parse-options.h"
  27 #include "help.h"
  28
  29 #ifdef NO_FAST_WORKING_DIRECTORY
  30 #define FAST_WORKING_DIRECTORY 0
  31 #else
  32 #define FAST_WORKING_DIRECTORY 1
  33 #endif
  34
  35 static int diff_detect_rename_default;
  36 static int diff_indent_heuristic = 1;
  37 static int diff_rename_limit_default = 400;
  38 static int diff_suppress_blank_empty;
  39 static int diff_use_color_default = -1;
  40 static int diff_color_moved_default;
  41 static int diff_color_moved_ws_default;
  42 static int diff_context_default = 3;
  43 static int diff_interhunk_context_default;
  44 static const char *diff_word_regex_cfg;
  45 static const char *external_diff_cmd_cfg;
  46 static const char *diff_order_file_cfg;
  47 int diff_auto_refresh_index = 1;
  48 static int diff_mnemonic_prefix;
  49 static int diff_no_prefix;
  50 static int diff_stat_graph_width;
  51 static int diff_dirstat_permille_default = 30;
  52 static struct diff_options default_diff_options;
  53 static long diff_algorithm;
  54 static unsigned ws_error_highlight_default = WSEH_NEW;
  55
  56 static char diff_colors[][COLOR_MAXLEN] = {
  57         GIT_COLOR_RESET,
  58         GIT_COLOR_NORMAL,       /* CONTEXT */
  59         GIT_COLOR_BOLD,         /* METAINFO */
  60         GIT_COLOR_CYAN,         /* FRAGINFO */
  61         GIT_COLOR_RED,          /* OLD */
  62         GIT_COLOR_GREEN,        /* NEW */
  63         GIT_COLOR_YELLOW,       /* COMMIT */
  64         GIT_COLOR_BG_RED,       /* WHITESPACE */
  65         GIT_COLOR_NORMAL,       /* FUNCINFO */
  66         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  67         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  70         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  71         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  72         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  73         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  74         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  75         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  76         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  77         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  78         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  79         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  80 };
  81
  82 static const char *color_diff_slots[] = {
  83         [DIFF_CONTEXT]                = "context",
  84         [DIFF_METAINFO]               = "meta",
  85         [DIFF_FRAGINFO]               = "frag",
  86         [DIFF_FILE_OLD]               = "old",
  87         [DIFF_FILE_NEW]               = "new",
  88         [DIFF_COMMIT]                 = "commit",
  89         [DIFF_WHITESPACE]             = "whitespace",
  90         [DIFF_FUNCINFO]               = "func",
  91         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  92         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  93         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  94         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  95         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  96         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  97         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  98         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  99         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 100         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 101         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 102         [DIFF_CONTEXT_BOLD]           = "contextBold",
 103         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 104         [DIFF_FILE_NEW_BOLD]          = "newBold",
 105 };
 106
 107 static NORETURN void die_want_option(const char *option_name)
 108 {
 109         die(_("option '%s' requires a value"), option_name);
 110 }
 111
 112 define_list_config_array_extra(color_diff_slots, {"plain"});
 113
 114 static int parse_diff_color_slot(const char *var)
 115 {
 116         if (!strcasecmp(var, "plain"))
 117                 return DIFF_CONTEXT;
 118         return LOOKUP_CONFIG(color_diff_slots, var);
 119 }
 120
 121 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 122                                 struct strbuf *errmsg)
 123 {
 124         char *params_copy = xstrdup(params_string);
 125         struct string_list params = STRING_LIST_INIT_NODUP;
 126         int ret = 0;
 127         int i;
 128
 129         if (*params_copy)
 130                 string_list_split_in_place(&params, params_copy, ',', -1);
 131         for (i = 0; i < params.nr; i++) {
 132                 const char *p = params.items[i].string;
 133                 if (!strcmp(p, "changes")) {
 134                         options->flags.dirstat_by_line = 0;
 135                         options->flags.dirstat_by_file = 0;
 136                 } else if (!strcmp(p, "lines")) {
 137                         options->flags.dirstat_by_line = 1;
 138                         options->flags.dirstat_by_file = 0;
 139                 } else if (!strcmp(p, "files")) {
 140                         options->flags.dirstat_by_line = 0;
 141                         options->flags.dirstat_by_file = 1;
 142                 } else if (!strcmp(p, "noncumulative")) {
 143                         options->flags.dirstat_cumulative = 0;
 144                 } else if (!strcmp(p, "cumulative")) {
 145                         options->flags.dirstat_cumulative = 1;
 146                 } else if (isdigit(*p)) {
 147                         char *end;
 148                         int permille = strtoul(p, &end, 10) * 10;
 149                         if (*end == '.' && isdigit(*++end)) {
 150                                 /* only use first digit */
 151                                 permille += *end - '0';
 152                                 /* .. and ignore any further digits */
 153                                 while (isdigit(*++end))
 154                                         ; /* nothing */
 155                         }
 156                         if (!*end)
 157                                 options->dirstat_permille = permille;
 158                         else {
 159                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 160                                             p);
 161                                 ret++;
 162                         }
 163                 } else {
 164                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 165                         ret++;
 166                 }
 167
 168         }
 169         string_list_clear(&params, 0);
 170         free(params_copy);
 171         return ret;
 172 }
 173
 174 static int parse_submodule_params(struct diff_options *options, const char *value)
 175 {
 176         if (!strcmp(value, "log"))
 177                 options->submodule_format = DIFF_SUBMODULE_LOG;
 178         else if (!strcmp(value, "short"))
 179                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 180         else if (!strcmp(value, "diff"))
 181                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 182         else
 183                 return -1;
 184         return 0;
 185 }
 186
 187 int git_config_rename(const char *var, const char *value)
 188 {
 189         if (!value)
 190                 return DIFF_DETECT_RENAME;
 191         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 192                 return  DIFF_DETECT_COPY;
 193         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 194 }
 195
 196 long parse_algorithm_value(const char *value)
 197 {
 198         if (!value)
 199                 return -1;
 200         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 201                 return 0;
 202         else if (!strcasecmp(value, "minimal"))
 203                 return XDF_NEED_MINIMAL;
 204         else if (!strcasecmp(value, "patience"))
 205                 return XDF_PATIENCE_DIFF;
 206         else if (!strcasecmp(value, "histogram"))
 207                 return XDF_HISTOGRAM_DIFF;
 208         return -1;
 209 }
 210
 211 static int parse_one_token(const char **arg, const char *token)
 212 {
 213         const char *rest;
 214         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 215                 *arg = rest;
 216                 return 1;
 217         }
 218         return 0;
 219 }
 220
 221 static int parse_ws_error_highlight(const char *arg)
 222 {
 223         const char *orig_arg = arg;
 224         unsigned val = 0;
 225
 226         while (*arg) {
 227                 if (parse_one_token(&arg, "none"))
 228                         val = 0;
 229                 else if (parse_one_token(&arg, "default"))
 230                         val = WSEH_NEW;
 231                 else if (parse_one_token(&arg, "all"))
 232                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 233                 else if (parse_one_token(&arg, "new"))
 234                         val |= WSEH_NEW;
 235                 else if (parse_one_token(&arg, "old"))
 236                         val |= WSEH_OLD;
 237                 else if (parse_one_token(&arg, "context"))
 238                         val |= WSEH_CONTEXT;
 239                 else {
 240                         return -1 - (int)(arg - orig_arg);
 241                 }
 242                 if (*arg)
 243                         arg++;
 244         }
 245         return val;
 246 }
 247
 248 /*
 249  * These are to give UI layer defaults.
 250  * The core-level commands such as git-diff-files should
 251  * never be affected by the setting of diff.renames
 252  * the user happens to have in the configuration file.
 253  */
 254 void init_diff_ui_defaults(void)
 255 {
 256         diff_detect_rename_default = DIFF_DETECT_RENAME;
 257 }
 258
 259 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 260 {
 261         if (!strcmp(var, "diff.indentheuristic"))
 262                 diff_indent_heuristic = git_config_bool(var, value);
 263         return 0;
 264 }
 265
 266 static int parse_color_moved(const char *arg)
 267 {
 268         switch (git_parse_maybe_bool(arg)) {
 269         case 0:
 270                 return COLOR_MOVED_NO;
 271         case 1:
 272                 return COLOR_MOVED_DEFAULT;
 273         default:
 274                 break;
 275         }
 276
 277         if (!strcmp(arg, "no"))
 278                 return COLOR_MOVED_NO;
 279         else if (!strcmp(arg, "plain"))
 280                 return COLOR_MOVED_PLAIN;
 281         else if (!strcmp(arg, "blocks"))
 282                 return COLOR_MOVED_BLOCKS;
 283         else if (!strcmp(arg, "zebra"))
 284                 return COLOR_MOVED_ZEBRA;
 285         else if (!strcmp(arg, "default"))
 286                 return COLOR_MOVED_DEFAULT;
 287         else if (!strcmp(arg, "dimmed-zebra"))
 288                 return COLOR_MOVED_ZEBRA_DIM;
 289         else if (!strcmp(arg, "dimmed_zebra"))
 290                 return COLOR_MOVED_ZEBRA_DIM;
 291         else
 292                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 293 }
 294
 295 static unsigned parse_color_moved_ws(const char *arg)
 296 {
 297         int ret = 0;
 298         struct string_list l = STRING_LIST_INIT_DUP;
 299         struct string_list_item *i;
 300
 301         string_list_split(&l, arg, ',', -1);
 302
 303         for_each_string_list_item(i, &l) {
 304                 struct strbuf sb = STRBUF_INIT;
 305                 strbuf_addstr(&sb, i->string);
 306                 strbuf_trim(&sb);
 307
 308                 if (!strcmp(sb.buf, "no"))
 309                         ret = 0;
 310                 else if (!strcmp(sb.buf, "ignore-space-change"))
 311                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 312                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 313                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 314                 else if (!strcmp(sb.buf, "ignore-all-space"))
 315                         ret |= XDF_IGNORE_WHITESPACE;
 316                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 317                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 318                 else {
 319                         ret |= COLOR_MOVED_WS_ERROR;
 320                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 321                 }
 322
 323                 strbuf_release(&sb);
 324         }
 325
 326         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 327             (ret & XDF_WHITESPACE_FLAGS)) {
 328                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 329                 ret |= COLOR_MOVED_WS_ERROR;
 330         }
 331
 332         string_list_clear(&l, 0);
 333
 334         return ret;
 335 }
 336
 337 int git_diff_ui_config(const char *var, const char *value, void *cb)
 338 {
 339         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 340                 diff_use_color_default = git_config_colorbool(var, value);
 341                 return 0;
 342         }
 343         if (!strcmp(var, "diff.colormoved")) {
 344                 int cm = parse_color_moved(value);
 345                 if (cm < 0)
 346                         return -1;
 347                 diff_color_moved_default = cm;
 348                 return 0;
 349         }
 350         if (!strcmp(var, "diff.colormovedws")) {
 351                 unsigned cm = parse_color_moved_ws(value);
 352                 if (cm & COLOR_MOVED_WS_ERROR)
 353                         return -1;
 354                 diff_color_moved_ws_default = cm;
 355                 return 0;
 356         }
 357         if (!strcmp(var, "diff.context")) {
 358                 diff_context_default = git_config_int(var, value);
 359                 if (diff_context_default < 0)
 360                         return -1;
 361                 return 0;
 362         }
 363         if (!strcmp(var, "diff.interhunkcontext")) {
 364                 diff_interhunk_context_default = git_config_int(var, value);
 365                 if (diff_interhunk_context_default < 0)
 366                         return -1;
 367                 return 0;
 368         }
 369         if (!strcmp(var, "diff.renames")) {
 370                 diff_detect_rename_default = git_config_rename(var, value);
 371                 return 0;
 372         }
 373         if (!strcmp(var, "diff.autorefreshindex")) {
 374                 diff_auto_refresh_index = git_config_bool(var, value);
 375                 return 0;
 376         }
 377         if (!strcmp(var, "diff.mnemonicprefix")) {
 378                 diff_mnemonic_prefix = git_config_bool(var, value);
 379                 return 0;
 380         }
 381         if (!strcmp(var, "diff.noprefix")) {
 382                 diff_no_prefix = git_config_bool(var, value);
 383                 return 0;
 384         }
 385         if (!strcmp(var, "diff.statgraphwidth")) {
 386                 diff_stat_graph_width = git_config_int(var, value);
 387                 return 0;
 388         }
 389         if (!strcmp(var, "diff.external"))
 390                 return git_config_string(&external_diff_cmd_cfg, var, value);
 391         if (!strcmp(var, "diff.wordregex"))
 392                 return git_config_string(&diff_word_regex_cfg, var, value);
 393         if (!strcmp(var, "diff.orderfile"))
 394                 return git_config_pathname(&diff_order_file_cfg, var, value);
 395
 396         if (!strcmp(var, "diff.ignoresubmodules"))
 397                 handle_ignore_submodules_arg(&default_diff_options, value);
 398
 399         if (!strcmp(var, "diff.submodule")) {
 400                 if (parse_submodule_params(&default_diff_options, value))
 401                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 402                                 value);
 403                 return 0;
 404         }
 405
 406         if (!strcmp(var, "diff.algorithm")) {
 407                 diff_algorithm = parse_algorithm_value(value);
 408                 if (diff_algorithm < 0)
 409                         return -1;
 410                 return 0;
 411         }
 412
 413         if (!strcmp(var, "diff.wserrorhighlight")) {
 414                 int val = parse_ws_error_highlight(value);
 415                 if (val < 0)
 416                         return -1;
 417                 ws_error_highlight_default = val;
 418                 return 0;
 419         }
 420
 421         if (git_color_config(var, value, cb) < 0)
 422                 return -1;
 423
 424         return git_diff_basic_config(var, value, cb);
 425 }
 426
 427 int git_diff_basic_config(const char *var, const char *value, void *cb)
 428 {
 429         const char *name;
 430
 431         if (!strcmp(var, "diff.renamelimit")) {
 432                 diff_rename_limit_default = git_config_int(var, value);
 433                 return 0;
 434         }
 435
 436         if (userdiff_config(var, value) < 0)
 437                 return -1;
 438
 439         if (skip_prefix(var, "diff.color.", &name) ||
 440             skip_prefix(var, "color.diff.", &name)) {
 441                 int slot = parse_diff_color_slot(name);
 442                 if (slot < 0)
 443                         return 0;
 444                 if (!value)
 445                         return config_error_nonbool(var);
 446                 return color_parse(value, diff_colors[slot]);
 447         }
 448
 449         /* like GNU diff's --suppress-blank-empty option  */
 450         if (!strcmp(var, "diff.suppressblankempty") ||
 451                         /* for backwards compatibility */
 452                         !strcmp(var, "diff.suppress-blank-empty")) {
 453                 diff_suppress_blank_empty = git_config_bool(var, value);
 454                 return 0;
 455         }
 456
 457         if (!strcmp(var, "diff.dirstat")) {
 458                 struct strbuf errmsg = STRBUF_INIT;
 459                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 460                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 461                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 462                                 errmsg.buf);
 463                 strbuf_release(&errmsg);
 464                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 465                 return 0;
 466         }
 467
 468         if (git_diff_heuristic_config(var, value, cb) < 0)
 469                 return -1;
 470
 471         return git_default_config(var, value, cb);
 472 }
 473
 474 static char *quote_two(const char *one, const char *two)
 475 {
 476         int need_one = quote_c_style(one, NULL, NULL, 1);
 477         int need_two = quote_c_style(two, NULL, NULL, 1);
 478         struct strbuf res = STRBUF_INIT;
 479
 480         if (need_one + need_two) {
 481                 strbuf_addch(&res, '"');
 482                 quote_c_style(one, &res, NULL, 1);
 483                 quote_c_style(two, &res, NULL, 1);
 484                 strbuf_addch(&res, '"');
 485         } else {
 486                 strbuf_addstr(&res, one);
 487                 strbuf_addstr(&res, two);
 488         }
 489         return strbuf_detach(&res, NULL);
 490 }
 491
 492 static const char *external_diff(void)
 493 {
 494         static const char *external_diff_cmd = NULL;
 495         static int done_preparing = 0;
 496
 497         if (done_preparing)
 498                 return external_diff_cmd;
 499         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 500         if (!external_diff_cmd)
 501                 external_diff_cmd = external_diff_cmd_cfg;
 502         done_preparing = 1;
 503         return external_diff_cmd;
 504 }
 505
 506 /*
 507  * Keep track of files used for diffing. Sometimes such an entry
 508  * refers to a temporary file, sometimes to an existing file, and
 509  * sometimes to "/dev/null".
 510  */
 511 static struct diff_tempfile {
 512         /*
 513          * filename external diff should read from, or NULL if this
 514          * entry is currently not in use:
 515          */
 516         const char *name;
 517
 518         char hex[GIT_MAX_HEXSZ + 1];
 519         char mode[10];
 520
 521         /*
 522          * If this diff_tempfile instance refers to a temporary file,
 523          * this tempfile object is used to manage its lifetime.
 524          */
 525         struct tempfile *tempfile;
 526 } diff_temp[2];
 527
 528 struct emit_callback {
 529         int color_diff;
 530         unsigned ws_rule;
 531         int blank_at_eof_in_preimage;
 532         int blank_at_eof_in_postimage;
 533         int lno_in_preimage;
 534         int lno_in_postimage;
 535         const char **label_path;
 536         struct diff_words_data *diff_words;
 537         struct diff_options *opt;
 538         struct strbuf *header;
 539 };
 540
 541 static int count_lines(const char *data, int size)
 542 {
 543         int count, ch, completely_empty = 1, nl_just_seen = 0;
 544         count = 0;
 545         while (0 < size--) {
 546                 ch = *data++;
 547                 if (ch == '\n') {
 548                         count++;
 549                         nl_just_seen = 1;
 550                         completely_empty = 0;
 551                 }
 552                 else {
 553                         nl_just_seen = 0;
 554                         completely_empty = 0;
 555                 }
 556         }
 557         if (completely_empty)
 558                 return 0;
 559         if (!nl_just_seen)
 560                 count++; /* no trailing newline */
 561         return count;
 562 }
 563
 564 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 565                        struct diff_filespec *one)
 566 {
 567         if (!DIFF_FILE_VALID(one)) {
 568                 mf->ptr = (char *)""; /* does not matter */
 569                 mf->size = 0;
 570                 return 0;
 571         }
 572         else if (diff_populate_filespec(r, one, 0))
 573                 return -1;
 574
 575         mf->ptr = one->data;
 576         mf->size = one->size;
 577         return 0;
 578 }
 579
 580 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 581 static unsigned long diff_filespec_size(struct repository *r,
 582                                         struct diff_filespec *one)
 583 {
 584         if (!DIFF_FILE_VALID(one))
 585                 return 0;
 586         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 587         return one->size;
 588 }
 589
 590 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 591 {
 592         char *ptr = mf->ptr;
 593         long size = mf->size;
 594         int cnt = 0;
 595
 596         if (!size)
 597                 return cnt;
 598         ptr += size - 1; /* pointing at the very end */
 599         if (*ptr != '\n')
 600                 ; /* incomplete line */
 601         else
 602                 ptr--; /* skip the last LF */
 603         while (mf->ptr < ptr) {
 604                 char *prev_eol;
 605                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 606                         if (*prev_eol == '\n')
 607                                 break;
 608                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 609                         break;
 610                 cnt++;
 611                 ptr = prev_eol - 1;
 612         }
 613         return cnt;
 614 }
 615
 616 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 617                                struct emit_callback *ecbdata)
 618 {
 619         int l1, l2, at;
 620         unsigned ws_rule = ecbdata->ws_rule;
 621         l1 = count_trailing_blank(mf1, ws_rule);
 622         l2 = count_trailing_blank(mf2, ws_rule);
 623         if (l2 <= l1) {
 624                 ecbdata->blank_at_eof_in_preimage = 0;
 625                 ecbdata->blank_at_eof_in_postimage = 0;
 626                 return;
 627         }
 628         at = count_lines(mf1->ptr, mf1->size);
 629         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 630
 631         at = count_lines(mf2->ptr, mf2->size);
 632         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 633 }
 634
 635 static void emit_line_0(struct diff_options *o,
 636                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 637                         int first, const char *line, int len)
 638 {
 639         int has_trailing_newline, has_trailing_carriage_return;
 640         int needs_reset = 0; /* at the end of the line */
 641         FILE *file = o->file;
 642
 643         fputs(diff_line_prefix(o), file);
 644
 645         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 646         if (has_trailing_newline)
 647                 len--;
 648
 649         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 650         if (has_trailing_carriage_return)
 651                 len--;
 652
 653         if (!len && !first)
 654                 goto end_of_line;
 655
 656         if (reverse && want_color(o->use_color)) {
 657                 fputs(GIT_COLOR_REVERSE, file);
 658                 needs_reset = 1;
 659         }
 660
 661         if (set_sign) {
 662                 fputs(set_sign, file);
 663                 needs_reset = 1;
 664         }
 665
 666         if (first)
 667                 fputc(first, file);
 668
 669         if (!len)
 670                 goto end_of_line;
 671
 672         if (set) {
 673                 if (set_sign && set != set_sign)
 674                         fputs(reset, file);
 675                 fputs(set, file);
 676                 needs_reset = 1;
 677         }
 678         fwrite(line, len, 1, file);
 679         needs_reset = 1; /* 'line' may contain color codes. */
 680
 681 end_of_line:
 682         if (needs_reset)
 683                 fputs(reset, file);
 684         if (has_trailing_carriage_return)
 685                 fputc('\r', file);
 686         if (has_trailing_newline)
 687                 fputc('\n', file);
 688 }
 689
 690 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 691                       const char *line, int len)
 692 {
 693         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 694 }
 695
 696 enum diff_symbol {
 697         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 698         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 699         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 700         DIFF_SYMBOL_BINARY_DIFF_BODY,
 701         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 702         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 703         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 704         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 705         DIFF_SYMBOL_STATS_LINE,
 706         DIFF_SYMBOL_WORD_DIFF,
 707         DIFF_SYMBOL_STAT_SEP,
 708         DIFF_SYMBOL_SUMMARY,
 709         DIFF_SYMBOL_SUBMODULE_ADD,
 710         DIFF_SYMBOL_SUBMODULE_DEL,
 711         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 712         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 713         DIFF_SYMBOL_SUBMODULE_HEADER,
 714         DIFF_SYMBOL_SUBMODULE_ERROR,
 715         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 716         DIFF_SYMBOL_REWRITE_DIFF,
 717         DIFF_SYMBOL_BINARY_FILES,
 718         DIFF_SYMBOL_HEADER,
 719         DIFF_SYMBOL_FILEPAIR_PLUS,
 720         DIFF_SYMBOL_FILEPAIR_MINUS,
 721         DIFF_SYMBOL_WORDS_PORCELAIN,
 722         DIFF_SYMBOL_WORDS,
 723         DIFF_SYMBOL_CONTEXT,
 724         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 725         DIFF_SYMBOL_PLUS,
 726         DIFF_SYMBOL_MINUS,
 727         DIFF_SYMBOL_NO_LF_EOF,
 728         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 729         DIFF_SYMBOL_CONTEXT_MARKER,
 730         DIFF_SYMBOL_SEPARATOR
 731 };
 732 /*
 733  * Flags for content lines:
 734  * 0..12 are whitespace rules
 735  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 736  * 16 is marking if the line is blank at EOF
 737  */
 738 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 739 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 740 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 741 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 742 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 743
 744 /*
 745  * This struct is used when we need to buffer the output of the diff output.
 746  *
 747  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 748  * into the pre/post image file. This pointer could be a union with the
 749  * line pointer. By storing an offset into the file instead of the literal line,
 750  * we can decrease the memory footprint for the buffered output. At first we
 751  * may want to only have indirection for the content lines, but we could also
 752  * enhance the state for emitting prefabricated lines, e.g. the similarity
 753  * score line or hunk/file headers would only need to store a number or path
 754  * and then the output can be constructed later on depending on state.
 755  */
 756 struct emitted_diff_symbol {
 757         const char *line;
 758         int len;
 759         int flags;
 760         int indent_off;   /* Offset to first non-whitespace character */
 761         int indent_width; /* The visual width of the indentation */
 762         enum diff_symbol s;
 763 };
 764 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 765
 766 struct emitted_diff_symbols {
 767         struct emitted_diff_symbol *buf;
 768         int nr, alloc;
 769 };
 770 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 771
 772 static void append_emitted_diff_symbol(struct diff_options *o,
 773                                        struct emitted_diff_symbol *e)
 774 {
 775         struct emitted_diff_symbol *f;
 776
 777         ALLOC_GROW(o->emitted_symbols->buf,
 778                    o->emitted_symbols->nr + 1,
 779                    o->emitted_symbols->alloc);
 780         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 781
 782         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 783         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 784 }
 785
 786 struct moved_entry {
 787         struct hashmap_entry ent;
 788         const struct emitted_diff_symbol *es;
 789         struct moved_entry *next_line;
 790 };
 791
 792 struct moved_block {
 793         struct moved_entry *match;
 794         int wsd; /* The whitespace delta of this block */
 795 };
 796
 797 static void moved_block_clear(struct moved_block *b)
 798 {
 799         memset(b, 0, sizeof(*b));
 800 }
 801
 802 #define INDENT_BLANKLINE INT_MIN
 803
 804 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 805 {
 806         unsigned int off = 0, i;
 807         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 808         const char *s = es->line;
 809         const int len = es->len;
 810
 811         /* skip any \v \f \r at start of indentation */
 812         while (s[off] == '\f' || s[off] == '\v' ||
 813                (s[off] == '\r' && off < len - 1))
 814                 off++;
 815
 816         /* calculate the visual width of indentation */
 817         while(1) {
 818                 if (s[off] == ' ') {
 819                         width++;
 820                         off++;
 821                 } else if (s[off] == '\t') {
 822                         width += tab_width - (width % tab_width);
 823                         while (s[++off] == '\t')
 824                                 width += tab_width;
 825                 } else {
 826                         break;
 827                 }
 828         }
 829
 830         /* check if this line is blank */
 831         for (i = off; i < len; i++)
 832                 if (!isspace(s[i]))
 833                     break;
 834
 835         if (i == len) {
 836                 es->indent_width = INDENT_BLANKLINE;
 837                 es->indent_off = len;
 838         } else {
 839                 es->indent_off = off;
 840                 es->indent_width = width;
 841         }
 842 }
 843
 844 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 845                             const struct emitted_diff_symbol *b,
 846                             int *out)
 847 {
 848         int a_len = a->len,
 849             b_len = b->len,
 850             a_off = a->indent_off,
 851             a_width = a->indent_width,
 852             b_off = b->indent_off,
 853             b_width = b->indent_width;
 854         int delta;
 855
 856         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 857                 *out = INDENT_BLANKLINE;
 858                 return 1;
 859         }
 860
 861         if (a->s == DIFF_SYMBOL_PLUS)
 862                 delta = a_width - b_width;
 863         else
 864                 delta = b_width - a_width;
 865
 866         if (a_len - a_off != b_len - b_off ||
 867             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 868                 return 0;
 869
 870         *out = delta;
 871
 872         return 1;
 873 }
 874
 875 static int cmp_in_block_with_wsd(const struct diff_options *o,
 876                                  const struct moved_entry *cur,
 877                                  const struct moved_entry *match,
 878                                  struct moved_block *pmb,
 879                                  int n)
 880 {
 881         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 882         int al = cur->es->len, bl = match->es->len, cl = l->len;
 883         const char *a = cur->es->line,
 884                    *b = match->es->line,
 885                    *c = l->line;
 886         int a_off = cur->es->indent_off,
 887             a_width = cur->es->indent_width,
 888             c_off = l->indent_off,
 889             c_width = l->indent_width;
 890         int delta;
 891
 892         /*
 893          * We need to check if 'cur' is equal to 'match'.  As those
 894          * are from the same (+/-) side, we do not need to adjust for
 895          * indent changes. However these were found using fuzzy
 896          * matching so we do have to check if they are equal. Here we
 897          * just check the lengths. We delay calling memcmp() to check
 898          * the contents until later as if the length comparison for a
 899          * and c fails we can avoid the call all together.
 900          */
 901         if (al != bl)
 902                 return 1;
 903
 904         /* If 'l' and 'cur' are both blank then they match. */
 905         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 906                 return 0;
 907
 908         /*
 909          * The indent changes of the block are known and stored in pmb->wsd;
 910          * however we need to check if the indent changes of the current line
 911          * match those of the current block and that the text of 'l' and 'cur'
 912          * after the indentation match.
 913          */
 914         if (cur->es->s == DIFF_SYMBOL_PLUS)
 915                 delta = a_width - c_width;
 916         else
 917                 delta = c_width - a_width;
 918
 919         /*
 920          * If the previous lines of this block were all blank then set its
 921          * whitespace delta.
 922          */
 923         if (pmb->wsd == INDENT_BLANKLINE)
 924                 pmb->wsd = delta;
 925
 926         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 927                  !memcmp(a, b, al) && !
 928                  memcmp(a + a_off, c + c_off, al - a_off));
 929 }
 930
 931 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 932                            const void *entry,
 933                            const void *entry_or_key,
 934                            const void *keydata)
 935 {
 936         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 937         const struct moved_entry *a = entry;
 938         const struct moved_entry *b = entry_or_key;
 939         unsigned flags = diffopt->color_moved_ws_handling
 940                          & XDF_WHITESPACE_FLAGS;
 941
 942         if (diffopt->color_moved_ws_handling &
 943             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 944                 /*
 945                  * As there is not specific white space config given,
 946                  * we'd need to check for a new block, so ignore all
 947                  * white space. The setup of the white space
 948                  * configuration for the next block is done else where
 949                  */
 950                 flags |= XDF_IGNORE_WHITESPACE;
 951
 952         return !xdiff_compare_lines(a->es->line, a->es->len,
 953                                     b->es->line, b->es->len,
 954                                     flags);
 955 }
 956
 957 static struct moved_entry *prepare_entry(struct diff_options *o,
 958                                          int line_no)
 959 {
 960         struct moved_entry *ret = xmalloc(sizeof(*ret));
 961         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 962         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 963
 964         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 965         ret->es = l;
 966         ret->next_line = NULL;
 967
 968         return ret;
 969 }
 970
 971 static void add_lines_to_move_detection(struct diff_options *o,
 972                                         struct hashmap *add_lines,
 973                                         struct hashmap *del_lines)
 974 {
 975         struct moved_entry *prev_line = NULL;
 976
 977         int n;
 978         for (n = 0; n < o->emitted_symbols->nr; n++) {
 979                 struct hashmap *hm;
 980                 struct moved_entry *key;
 981
 982                 switch (o->emitted_symbols->buf[n].s) {
 983                 case DIFF_SYMBOL_PLUS:
 984                         hm = add_lines;
 985                         break;
 986                 case DIFF_SYMBOL_MINUS:
 987                         hm = del_lines;
 988                         break;
 989                 default:
 990                         prev_line = NULL;
 991                         continue;
 992                 }
 993
 994                 if (o->color_moved_ws_handling &
 995                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 996                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
 997                 key = prepare_entry(o, n);
 998                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 999                         prev_line->next_line = key;
1000
1001                 hashmap_add(hm, key);
1002                 prev_line = key;
1003         }
1004 }
1005
1006 static void pmb_advance_or_null(struct diff_options *o,
1007                                 struct moved_entry *match,
1008                                 struct hashmap *hm,
1009                                 struct moved_block *pmb,
1010                                 int pmb_nr)
1011 {
1012         int i;
1013         for (i = 0; i < pmb_nr; i++) {
1014                 struct moved_entry *prev = pmb[i].match;
1015                 struct moved_entry *cur = (prev && prev->next_line) ?
1016                                 prev->next_line : NULL;
1017                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1018                         pmb[i].match = cur;
1019                 } else {
1020                         pmb[i].match = NULL;
1021                 }
1022         }
1023 }
1024
1025 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1026                                             struct moved_entry *match,
1027                                             struct hashmap *hm,
1028                                             struct moved_block *pmb,
1029                                             int pmb_nr, int n)
1030 {
1031         int i;
1032         char *got_match = xcalloc(1, pmb_nr);
1033
1034         for (; match; match = hashmap_get_next(hm, match)) {
1035                 for (i = 0; i < pmb_nr; i++) {
1036                         struct moved_entry *prev = pmb[i].match;
1037                         struct moved_entry *cur = (prev && prev->next_line) ?
1038                                         prev->next_line : NULL;
1039                         if (!cur)
1040                                 continue;
1041                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1042                                 got_match[i] |= 1;
1043                 }
1044         }
1045
1046         for (i = 0; i < pmb_nr; i++) {
1047                 if (got_match[i]) {
1048                         /* Advance to the next line */
1049                         pmb[i].match = pmb[i].match->next_line;
1050                 } else {
1051                         moved_block_clear(&pmb[i]);
1052                 }
1053         }
1054
1055         free(got_match);
1056 }
1057
1058 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1059                                          int pmb_nr)
1060 {
1061         int lp, rp;
1062
1063         /* Shrink the set of potential block to the remaining running */
1064         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1065                 while (lp < pmb_nr && pmb[lp].match)
1066                         lp++;
1067                 /* lp points at the first NULL now */
1068
1069                 while (rp > -1 && !pmb[rp].match)
1070                         rp--;
1071                 /* rp points at the last non-NULL */
1072
1073                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1074                         pmb[lp] = pmb[rp];
1075                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1076                         rp--;
1077                         lp++;
1078                 }
1079         }
1080
1081         /* Remember the number of running sets */
1082         return rp + 1;
1083 }
1084
1085 /*
1086  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1087  *
1088  * Otherwise, if the last block has fewer alphanumeric characters than
1089  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1090  * that block.
1091  *
1092  * The last block consists of the (n - block_length)'th line up to but not
1093  * including the nth line.
1094  *
1095  * Returns 0 if the last block is empty or is unset by this function, non zero
1096  * otherwise.
1097  *
1098  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1099  * Think of a way to unify them.
1100  */
1101 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1102 {
1103         int i, alnum_count = 0;
1104         if (o->color_moved == COLOR_MOVED_PLAIN)
1105                 return block_length;
1106         for (i = 1; i < block_length + 1; i++) {
1107                 const char *c = o->emitted_symbols->buf[n - i].line;
1108                 for (; *c; c++) {
1109                         if (!isalnum(*c))
1110                                 continue;
1111                         alnum_count++;
1112                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1113                                 return 1;
1114                 }
1115         }
1116         for (i = 1; i < block_length + 1; i++)
1117                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1118         return 0;
1119 }
1120
1121 /* Find blocks of moved code, delegate actual coloring decision to helper */
1122 static void mark_color_as_moved(struct diff_options *o,
1123                                 struct hashmap *add_lines,
1124                                 struct hashmap *del_lines)
1125 {
1126         struct moved_block *pmb = NULL; /* potentially moved blocks */
1127         int pmb_nr = 0, pmb_alloc = 0;
1128         int n, flipped_block = 0, block_length = 0;
1129
1130
1131         for (n = 0; n < o->emitted_symbols->nr; n++) {
1132                 struct hashmap *hm = NULL;
1133                 struct moved_entry *key;
1134                 struct moved_entry *match = NULL;
1135                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1136                 enum diff_symbol last_symbol = 0;
1137
1138                 switch (l->s) {
1139                 case DIFF_SYMBOL_PLUS:
1140                         hm = del_lines;
1141                         key = prepare_entry(o, n);
1142                         match = hashmap_get(hm, key, NULL);
1143                         free(key);
1144                         break;
1145                 case DIFF_SYMBOL_MINUS:
1146                         hm = add_lines;
1147                         key = prepare_entry(o, n);
1148                         match = hashmap_get(hm, key, NULL);
1149                         free(key);
1150                         break;
1151                 default:
1152                         flipped_block = 0;
1153                 }
1154
1155                 if (!match) {
1156                         int i;
1157
1158                         adjust_last_block(o, n, block_length);
1159                         for(i = 0; i < pmb_nr; i++)
1160                                 moved_block_clear(&pmb[i]);
1161                         pmb_nr = 0;
1162                         block_length = 0;
1163                         flipped_block = 0;
1164                         last_symbol = l->s;
1165                         continue;
1166                 }
1167
1168                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1169                         last_symbol = l->s;
1170                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1171                         continue;
1172                 }
1173
1174                 if (o->color_moved_ws_handling &
1175                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1176                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1177                 else
1178                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1179
1180                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1181
1182                 if (pmb_nr == 0) {
1183                         /*
1184                          * The current line is the start of a new block.
1185                          * Setup the set of potential blocks.
1186                          */
1187                         for (; match; match = hashmap_get_next(hm, match)) {
1188                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1189                                 if (o->color_moved_ws_handling &
1190                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1191                                         if (compute_ws_delta(l, match->es,
1192                                                              &pmb[pmb_nr].wsd))
1193                                                 pmb[pmb_nr++].match = match;
1194                                 } else {
1195                                         pmb[pmb_nr].wsd = 0;
1196                                         pmb[pmb_nr++].match = match;
1197                                 }
1198                         }
1199
1200                         if (adjust_last_block(o, n, block_length) &&
1201                             pmb_nr && last_symbol != l->s)
1202                                 flipped_block = (flipped_block + 1) % 2;
1203                         else
1204                                 flipped_block = 0;
1205
1206                         block_length = 0;
1207                 }
1208
1209                 if (pmb_nr) {
1210                         block_length++;
1211                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1212                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1213                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1214                 }
1215                 last_symbol = l->s;
1216         }
1217         adjust_last_block(o, n, block_length);
1218
1219         for(n = 0; n < pmb_nr; n++)
1220                 moved_block_clear(&pmb[n]);
1221         free(pmb);
1222 }
1223
1224 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1225   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1226 static void dim_moved_lines(struct diff_options *o)
1227 {
1228         int n;
1229         for (n = 0; n < o->emitted_symbols->nr; n++) {
1230                 struct emitted_diff_symbol *prev = (n != 0) ?
1231                                 &o->emitted_symbols->buf[n - 1] : NULL;
1232                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1233                 struct emitted_diff_symbol *next =
1234                                 (n < o->emitted_symbols->nr - 1) ?
1235                                 &o->emitted_symbols->buf[n + 1] : NULL;
1236
1237                 /* Not a plus or minus line? */
1238                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1239                         continue;
1240
1241                 /* Not a moved line? */
1242                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1243                         continue;
1244
1245                 /*
1246                  * If prev or next are not a plus or minus line,
1247                  * pretend they don't exist
1248                  */
1249                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1250                             prev->s != DIFF_SYMBOL_MINUS)
1251                         prev = NULL;
1252                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1253                             next->s != DIFF_SYMBOL_MINUS)
1254                         next = NULL;
1255
1256                 /* Inside a block? */
1257                 if ((prev &&
1258                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1259                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1260                     (next &&
1261                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1262                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1263                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1264                         continue;
1265                 }
1266
1267                 /* Check if we are at an interesting bound: */
1268                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1269                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1270                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1271                         continue;
1272                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1273                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1274                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1275                         continue;
1276
1277                 /*
1278                  * The boundary to prev and next are not interesting,
1279                  * so this line is not interesting as a whole
1280                  */
1281                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1282         }
1283 }
1284
1285 static void emit_line_ws_markup(struct diff_options *o,
1286                                 const char *set_sign, const char *set,
1287                                 const char *reset,
1288                                 int sign_index, const char *line, int len,
1289                                 unsigned ws_rule, int blank_at_eof)
1290 {
1291         const char *ws = NULL;
1292         int sign = o->output_indicators[sign_index];
1293
1294         if (o->ws_error_highlight & ws_rule) {
1295                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1296                 if (!*ws)
1297                         ws = NULL;
1298         }
1299
1300         if (!ws && !set_sign)
1301                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1302         else if (!ws) {
1303                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1304         } else if (blank_at_eof)
1305                 /* Blank line at EOF - paint '+' as well */
1306                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1307         else {
1308                 /* Emit just the prefix, then the rest. */
1309                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1310                             sign, "", 0);
1311                 ws_check_emit(line, len, ws_rule,
1312                               o->file, set, reset, ws);
1313         }
1314 }
1315
1316 static void emit_diff_symbol_from_struct(struct diff_options *o,
1317                                          struct emitted_diff_symbol *eds)
1318 {
1319         static const char *nneof = " No newline at end of file\n";
1320         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1321         struct strbuf sb = STRBUF_INIT;
1322
1323         enum diff_symbol s = eds->s;
1324         const char *line = eds->line;
1325         int len = eds->len;
1326         unsigned flags = eds->flags;
1327
1328         switch (s) {
1329         case DIFF_SYMBOL_NO_LF_EOF:
1330                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1331                 reset = diff_get_color_opt(o, DIFF_RESET);
1332                 putc('\n', o->file);
1333                 emit_line_0(o, context, NULL, 0, reset, '\\',
1334                             nneof, strlen(nneof));
1335                 break;
1336         case DIFF_SYMBOL_SUBMODULE_HEADER:
1337         case DIFF_SYMBOL_SUBMODULE_ERROR:
1338         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1339         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1340         case DIFF_SYMBOL_SUMMARY:
1341         case DIFF_SYMBOL_STATS_LINE:
1342         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1343         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1344                 emit_line(o, "", "", line, len);
1345                 break;
1346         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1347         case DIFF_SYMBOL_CONTEXT_MARKER:
1348                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1349                 reset = diff_get_color_opt(o, DIFF_RESET);
1350                 emit_line(o, context, reset, line, len);
1351                 break;
1352         case DIFF_SYMBOL_SEPARATOR:
1353                 fprintf(o->file, "%s%c",
1354                         diff_line_prefix(o),
1355                         o->line_termination);
1356                 break;
1357         case DIFF_SYMBOL_CONTEXT:
1358                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1359                 reset = diff_get_color_opt(o, DIFF_RESET);
1360                 set_sign = NULL;
1361                 if (o->flags.dual_color_diffed_diffs) {
1362                         char c = !len ? 0 : line[0];
1363
1364                         if (c == '+')
1365                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1366                         else if (c == '@')
1367                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1368                         else if (c == '-')
1369                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1370                 }
1371                 emit_line_ws_markup(o, set_sign, set, reset,
1372                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1373                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1374                 break;
1375         case DIFF_SYMBOL_PLUS:
1376                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1377                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1378                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1379                 case DIFF_SYMBOL_MOVED_LINE |
1380                      DIFF_SYMBOL_MOVED_LINE_ALT |
1381                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1382                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1383                         break;
1384                 case DIFF_SYMBOL_MOVED_LINE |
1385                      DIFF_SYMBOL_MOVED_LINE_ALT:
1386                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1387                         break;
1388                 case DIFF_SYMBOL_MOVED_LINE |
1389                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1390                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1391                         break;
1392                 case DIFF_SYMBOL_MOVED_LINE:
1393                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1394                         break;
1395                 default:
1396                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1397                 }
1398                 reset = diff_get_color_opt(o, DIFF_RESET);
1399                 if (!o->flags.dual_color_diffed_diffs)
1400                         set_sign = NULL;
1401                 else {
1402                         char c = !len ? 0 : line[0];
1403
1404                         set_sign = set;
1405                         if (c == '-')
1406                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1407                         else if (c == '@')
1408                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1409                         else if (c == '+')
1410                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1411                         else
1412                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1413                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1414                 }
1415                 emit_line_ws_markup(o, set_sign, set, reset,
1416                                     OUTPUT_INDICATOR_NEW, line, len,
1417                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1418                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1419                 break;
1420         case DIFF_SYMBOL_MINUS:
1421                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1422                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1423                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1424                 case DIFF_SYMBOL_MOVED_LINE |
1425                      DIFF_SYMBOL_MOVED_LINE_ALT |
1426                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1427                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1428                         break;
1429                 case DIFF_SYMBOL_MOVED_LINE |
1430                      DIFF_SYMBOL_MOVED_LINE_ALT:
1431                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1432                         break;
1433                 case DIFF_SYMBOL_MOVED_LINE |
1434                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1435                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1436                         break;
1437                 case DIFF_SYMBOL_MOVED_LINE:
1438                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1439                         break;
1440                 default:
1441                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1442                 }
1443                 reset = diff_get_color_opt(o, DIFF_RESET);
1444                 if (!o->flags.dual_color_diffed_diffs)
1445                         set_sign = NULL;
1446                 else {
1447                         char c = !len ? 0 : line[0];
1448
1449                         set_sign = set;
1450                         if (c == '+')
1451                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1452                         else if (c == '@')
1453                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1454                         else if (c == '-')
1455                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1456                         else
1457                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1458                 }
1459                 emit_line_ws_markup(o, set_sign, set, reset,
1460                                     OUTPUT_INDICATOR_OLD, line, len,
1461                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1462                 break;
1463         case DIFF_SYMBOL_WORDS_PORCELAIN:
1464                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1465                 reset = diff_get_color_opt(o, DIFF_RESET);
1466                 emit_line(o, context, reset, line, len);
1467                 fputs("~\n", o->file);
1468                 break;
1469         case DIFF_SYMBOL_WORDS:
1470                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1471                 reset = diff_get_color_opt(o, DIFF_RESET);
1472                 /*
1473                  * Skip the prefix character, if any.  With
1474                  * diff_suppress_blank_empty, there may be
1475                  * none.
1476                  */
1477                 if (line[0] != '\n') {
1478                         line++;
1479                         len--;
1480                 }
1481                 emit_line(o, context, reset, line, len);
1482                 break;
1483         case DIFF_SYMBOL_FILEPAIR_PLUS:
1484                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1485                 reset = diff_get_color_opt(o, DIFF_RESET);
1486                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1487                         line, reset,
1488                         strchr(line, ' ') ? "\t" : "");
1489                 break;
1490         case DIFF_SYMBOL_FILEPAIR_MINUS:
1491                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1492                 reset = diff_get_color_opt(o, DIFF_RESET);
1493                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1494                         line, reset,
1495                         strchr(line, ' ') ? "\t" : "");
1496                 break;
1497         case DIFF_SYMBOL_BINARY_FILES:
1498         case DIFF_SYMBOL_HEADER:
1499                 fprintf(o->file, "%s", line);
1500                 break;
1501         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1502                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1503                 break;
1504         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1505                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1506                 break;
1507         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1508                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1509                 break;
1510         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1511                 fputs(diff_line_prefix(o), o->file);
1512                 fputc('\n', o->file);
1513                 break;
1514         case DIFF_SYMBOL_REWRITE_DIFF:
1515                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1516                 reset = diff_get_color_opt(o, DIFF_RESET);
1517                 emit_line(o, fraginfo, reset, line, len);
1518                 break;
1519         case DIFF_SYMBOL_SUBMODULE_ADD:
1520                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1521                 reset = diff_get_color_opt(o, DIFF_RESET);
1522                 emit_line(o, set, reset, line, len);
1523                 break;
1524         case DIFF_SYMBOL_SUBMODULE_DEL:
1525                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1526                 reset = diff_get_color_opt(o, DIFF_RESET);
1527                 emit_line(o, set, reset, line, len);
1528                 break;
1529         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1530                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1531                         diff_line_prefix(o), line);
1532                 break;
1533         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1534                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1535                         diff_line_prefix(o), line);
1536                 break;
1537         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1538                 emit_line(o, "", "", " 0 files changed\n",
1539                           strlen(" 0 files changed\n"));
1540                 break;
1541         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1542                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1543                 break;
1544         case DIFF_SYMBOL_WORD_DIFF:
1545                 fprintf(o->file, "%.*s", len, line);
1546                 break;
1547         case DIFF_SYMBOL_STAT_SEP:
1548                 fputs(o->stat_sep, o->file);
1549                 break;
1550         default:
1551                 BUG("unknown diff symbol");
1552         }
1553         strbuf_release(&sb);
1554 }
1555
1556 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1557                              const char *line, int len, unsigned flags)
1558 {
1559         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1560
1561         if (o->emitted_symbols)
1562                 append_emitted_diff_symbol(o, &e);
1563         else
1564                 emit_diff_symbol_from_struct(o, &e);
1565 }
1566
1567 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1568 {
1569         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1570 }
1571
1572 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1573 {
1574         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1575 }
1576
1577 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1578 {
1579         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1580                          path, strlen(path), 0);
1581 }
1582
1583 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1584 {
1585         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1586                          path, strlen(path), 0);
1587 }
1588
1589 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1590 {
1591         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1592                          header, strlen(header), 0);
1593 }
1594
1595 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1596 {
1597         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1598 }
1599
1600 void diff_emit_submodule_pipethrough(struct diff_options *o,
1601                                      const char *line, int len)
1602 {
1603         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1604 }
1605
1606 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1607 {
1608         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1609               ecbdata->blank_at_eof_in_preimage &&
1610               ecbdata->blank_at_eof_in_postimage &&
1611               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1612               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1613                 return 0;
1614         return ws_blank_line(line, len, ecbdata->ws_rule);
1615 }
1616
1617 static void emit_add_line(struct emit_callback *ecbdata,
1618                           const char *line, int len)
1619 {
1620         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1621         if (new_blank_line_at_eof(ecbdata, line, len))
1622                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1623
1624         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1625 }
1626
1627 static void emit_del_line(struct emit_callback *ecbdata,
1628                           const char *line, int len)
1629 {
1630         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1631         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1632 }
1633
1634 static void emit_context_line(struct emit_callback *ecbdata,
1635                               const char *line, int len)
1636 {
1637         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1638         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1639 }
1640
1641 static void emit_hunk_header(struct emit_callback *ecbdata,
1642                              const char *line, int len)
1643 {
1644         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1645         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1646         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1647         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1648         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1649         static const char atat[2] = { '@', '@' };
1650         const char *cp, *ep;
1651         struct strbuf msgbuf = STRBUF_INIT;
1652         int org_len = len;
1653         int i = 1;
1654
1655         /*
1656          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1657          * it always is at least 10 bytes long.
1658          */
1659         if (len < 10 ||
1660             memcmp(line, atat, 2) ||
1661             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1662                 emit_diff_symbol(ecbdata->opt,
1663                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1664                 return;
1665         }
1666         ep += 2; /* skip over @@ */
1667
1668         /* The hunk header in fraginfo color */
1669         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1670                 strbuf_addstr(&msgbuf, reverse);
1671         strbuf_addstr(&msgbuf, frag);
1672         strbuf_add(&msgbuf, line, ep - line);
1673         strbuf_addstr(&msgbuf, reset);
1674
1675         /*
1676          * trailing "\r\n"
1677          */
1678         for ( ; i < 3; i++)
1679                 if (line[len - i] == '\r' || line[len - i] == '\n')
1680                         len--;
1681
1682         /* blank before the func header */
1683         for (cp = ep; ep - line < len; ep++)
1684                 if (*ep != ' ' && *ep != '\t')
1685                         break;
1686         if (ep != cp) {
1687                 strbuf_addstr(&msgbuf, context);
1688                 strbuf_add(&msgbuf, cp, ep - cp);
1689                 strbuf_addstr(&msgbuf, reset);
1690         }
1691
1692         if (ep < line + len) {
1693                 strbuf_addstr(&msgbuf, func);
1694                 strbuf_add(&msgbuf, ep, line + len - ep);
1695                 strbuf_addstr(&msgbuf, reset);
1696         }
1697
1698         strbuf_add(&msgbuf, line + len, org_len - len);
1699         strbuf_complete_line(&msgbuf);
1700         emit_diff_symbol(ecbdata->opt,
1701                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1702         strbuf_release(&msgbuf);
1703 }
1704
1705 static struct diff_tempfile *claim_diff_tempfile(void)
1706 {
1707         int i;
1708         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1709                 if (!diff_temp[i].name)
1710                         return diff_temp + i;
1711         BUG("diff is failing to clean up its tempfiles");
1712 }
1713
1714 static void remove_tempfile(void)
1715 {
1716         int i;
1717         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1718                 if (is_tempfile_active(diff_temp[i].tempfile))
1719                         delete_tempfile(&diff_temp[i].tempfile);
1720                 diff_temp[i].name = NULL;
1721         }
1722 }
1723
1724 static void add_line_count(struct strbuf *out, int count)
1725 {
1726         switch (count) {
1727         case 0:
1728                 strbuf_addstr(out, "0,0");
1729                 break;
1730         case 1:
1731                 strbuf_addstr(out, "1");
1732                 break;
1733         default:
1734                 strbuf_addf(out, "1,%d", count);
1735                 break;
1736         }
1737 }
1738
1739 static void emit_rewrite_lines(struct emit_callback *ecb,
1740                                int prefix, const char *data, int size)
1741 {
1742         const char *endp = NULL;
1743
1744         while (0 < size) {
1745                 int len;
1746
1747                 endp = memchr(data, '\n', size);
1748                 len = endp ? (endp - data + 1) : size;
1749                 if (prefix != '+') {
1750                         ecb->lno_in_preimage++;
1751                         emit_del_line(ecb, data, len);
1752                 } else {
1753                         ecb->lno_in_postimage++;
1754                         emit_add_line(ecb, data, len);
1755                 }
1756                 size -= len;
1757                 data += len;
1758         }
1759         if (!endp)
1760                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1761 }
1762
1763 static void emit_rewrite_diff(const char *name_a,
1764                               const char *name_b,
1765                               struct diff_filespec *one,
1766                               struct diff_filespec *two,
1767                               struct userdiff_driver *textconv_one,
1768                               struct userdiff_driver *textconv_two,
1769                               struct diff_options *o)
1770 {
1771         int lc_a, lc_b;
1772         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1773         const char *a_prefix, *b_prefix;
1774         char *data_one, *data_two;
1775         size_t size_one, size_two;
1776         struct emit_callback ecbdata;
1777         struct strbuf out = STRBUF_INIT;
1778
1779         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1780                 a_prefix = o->b_prefix;
1781                 b_prefix = o->a_prefix;
1782         } else {
1783                 a_prefix = o->a_prefix;
1784                 b_prefix = o->b_prefix;
1785         }
1786
1787         name_a += (*name_a == '/');
1788         name_b += (*name_b == '/');
1789
1790         strbuf_reset(&a_name);
1791         strbuf_reset(&b_name);
1792         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1793         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1794
1795         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1796         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1797
1798         memset(&ecbdata, 0, sizeof(ecbdata));
1799         ecbdata.color_diff = want_color(o->use_color);
1800         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1801         ecbdata.opt = o;
1802         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1803                 mmfile_t mf1, mf2;
1804                 mf1.ptr = (char *)data_one;
1805                 mf2.ptr = (char *)data_two;
1806                 mf1.size = size_one;
1807                 mf2.size = size_two;
1808                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1809         }
1810         ecbdata.lno_in_preimage = 1;
1811         ecbdata.lno_in_postimage = 1;
1812
1813         lc_a = count_lines(data_one, size_one);
1814         lc_b = count_lines(data_two, size_two);
1815
1816         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1817                          a_name.buf, a_name.len, 0);
1818         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1819                          b_name.buf, b_name.len, 0);
1820
1821         strbuf_addstr(&out, "@@ -");
1822         if (!o->irreversible_delete)
1823                 add_line_count(&out, lc_a);
1824         else
1825                 strbuf_addstr(&out, "?,?");
1826         strbuf_addstr(&out, " +");
1827         add_line_count(&out, lc_b);
1828         strbuf_addstr(&out, " @@\n");
1829         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1830         strbuf_release(&out);
1831
1832         if (lc_a && !o->irreversible_delete)
1833                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1834         if (lc_b)
1835                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1836         if (textconv_one)
1837                 free((char *)data_one);
1838         if (textconv_two)
1839                 free((char *)data_two);
1840 }
1841
1842 struct diff_words_buffer {
1843         mmfile_t text;
1844         unsigned long alloc;
1845         struct diff_words_orig {
1846                 const char *begin, *end;
1847         } *orig;
1848         int orig_nr, orig_alloc;
1849 };
1850
1851 static void diff_words_append(char *line, unsigned long len,
1852                 struct diff_words_buffer *buffer)
1853 {
1854         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1855         line++;
1856         len--;
1857         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1858         buffer->text.size += len;
1859         buffer->text.ptr[buffer->text.size] = '\0';
1860 }
1861
1862 struct diff_words_style_elem {
1863         const char *prefix;
1864         const char *suffix;
1865         const char *color; /* NULL; filled in by the setup code if
1866                             * color is enabled */
1867 };
1868
1869 struct diff_words_style {
1870         enum diff_words_type type;
1871         struct diff_words_style_elem new_word, old_word, ctx;
1872         const char *newline;
1873 };
1874
1875 static struct diff_words_style diff_words_styles[] = {
1876         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1877         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1878         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1879 };
1880
1881 struct diff_words_data {
1882         struct diff_words_buffer minus, plus;
1883         const char *current_plus;
1884         int last_minus;
1885         struct diff_options *opt;
1886         regex_t *word_regex;
1887         enum diff_words_type type;
1888         struct diff_words_style *style;
1889 };
1890
1891 static int fn_out_diff_words_write_helper(struct diff_options *o,
1892                                           struct diff_words_style_elem *st_el,
1893                                           const char *newline,
1894                                           size_t count, const char *buf)
1895 {
1896         int print = 0;
1897         struct strbuf sb = STRBUF_INIT;
1898
1899         while (count) {
1900                 char *p = memchr(buf, '\n', count);
1901                 if (print)
1902                         strbuf_addstr(&sb, diff_line_prefix(o));
1903
1904                 if (p != buf) {
1905                         const char *reset = st_el->color && *st_el->color ?
1906                                             GIT_COLOR_RESET : NULL;
1907                         if (st_el->color && *st_el->color)
1908                                 strbuf_addstr(&sb, st_el->color);
1909                         strbuf_addstr(&sb, st_el->prefix);
1910                         strbuf_add(&sb, buf, p ? p - buf : count);
1911                         strbuf_addstr(&sb, st_el->suffix);
1912                         if (reset)
1913                                 strbuf_addstr(&sb, reset);
1914                 }
1915                 if (!p)
1916                         goto out;
1917
1918                 strbuf_addstr(&sb, newline);
1919                 count -= p + 1 - buf;
1920                 buf = p + 1;
1921                 print = 1;
1922                 if (count) {
1923                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1924                                          sb.buf, sb.len, 0);
1925                         strbuf_reset(&sb);
1926                 }
1927         }
1928
1929 out:
1930         if (sb.len)
1931                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1932                                  sb.buf, sb.len, 0);
1933         strbuf_release(&sb);
1934         return 0;
1935 }
1936
1937 /*
1938  * '--color-words' algorithm can be described as:
1939  *
1940  *   1. collect the minus/plus lines of a diff hunk, divided into
1941  *      minus-lines and plus-lines;
1942  *
1943  *   2. break both minus-lines and plus-lines into words and
1944  *      place them into two mmfile_t with one word for each line;
1945  *
1946  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1947  *
1948  * And for the common parts of the both file, we output the plus side text.
1949  * diff_words->current_plus is used to trace the current position of the plus file
1950  * which printed. diff_words->last_minus is used to trace the last minus word
1951  * printed.
1952  *
1953  * For '--graph' to work with '--color-words', we need to output the graph prefix
1954  * on each line of color words output. Generally, there are two conditions on
1955  * which we should output the prefix.
1956  *
1957  *   1. diff_words->last_minus == 0 &&
1958  *      diff_words->current_plus == diff_words->plus.text.ptr
1959  *
1960  *      that is: the plus text must start as a new line, and if there is no minus
1961  *      word printed, a graph prefix must be printed.
1962  *
1963  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1964  *      *(diff_words->current_plus - 1) == '\n'
1965  *
1966  *      that is: a graph prefix must be printed following a '\n'
1967  */
1968 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1969 {
1970         if ((diff_words->last_minus == 0 &&
1971                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1972                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1973                 *(diff_words->current_plus - 1) == '\n')) {
1974                 return 1;
1975         } else {
1976                 return 0;
1977         }
1978 }
1979
1980 static void fn_out_diff_words_aux(void *priv,
1981                                   long minus_first, long minus_len,
1982                                   long plus_first, long plus_len,
1983                                   const char *func, long funclen)
1984 {
1985         struct diff_words_data *diff_words = priv;
1986         struct diff_words_style *style = diff_words->style;
1987         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1988         struct diff_options *opt = diff_words->opt;
1989         const char *line_prefix;
1990
1991         assert(opt);
1992         line_prefix = diff_line_prefix(opt);
1993
1994         /* POSIX requires that first be decremented by one if len == 0... */
1995         if (minus_len) {
1996                 minus_begin = diff_words->minus.orig[minus_first].begin;
1997                 minus_end =
1998                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1999         } else
2000                 minus_begin = minus_end =
2001                         diff_words->minus.orig[minus_first].end;
2002
2003         if (plus_len) {
2004                 plus_begin = diff_words->plus.orig[plus_first].begin;
2005                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2006         } else
2007                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2008
2009         if (color_words_output_graph_prefix(diff_words)) {
2010                 fputs(line_prefix, diff_words->opt->file);
2011         }
2012         if (diff_words->current_plus != plus_begin) {
2013                 fn_out_diff_words_write_helper(diff_words->opt,
2014                                 &style->ctx, style->newline,
2015                                 plus_begin - diff_words->current_plus,
2016                                 diff_words->current_plus);
2017         }
2018         if (minus_begin != minus_end) {
2019                 fn_out_diff_words_write_helper(diff_words->opt,
2020                                 &style->old_word, style->newline,
2021                                 minus_end - minus_begin, minus_begin);
2022         }
2023         if (plus_begin != plus_end) {
2024                 fn_out_diff_words_write_helper(diff_words->opt,
2025                                 &style->new_word, style->newline,
2026                                 plus_end - plus_begin, plus_begin);
2027         }
2028
2029         diff_words->current_plus = plus_end;
2030         diff_words->last_minus = minus_first;
2031 }
2032
2033 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2034 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2035                 int *begin, int *end)
2036 {
2037         if (word_regex && *begin < buffer->size) {
2038                 regmatch_t match[1];
2039                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2040                                  buffer->size - *begin, 1, match, 0)) {
2041                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2042                                         '\n', match[0].rm_eo - match[0].rm_so);
2043                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2044                         *begin += match[0].rm_so;
2045                         return *begin >= *end;
2046                 }
2047                 return -1;
2048         }
2049
2050         /* find the next word */
2051         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2052                 (*begin)++;
2053         if (*begin >= buffer->size)
2054                 return -1;
2055
2056         /* find the end of the word */
2057         *end = *begin + 1;
2058         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2059                 (*end)++;
2060
2061         return 0;
2062 }
2063
2064 /*
2065  * This function splits the words in buffer->text, stores the list with
2066  * newline separator into out, and saves the offsets of the original words
2067  * in buffer->orig.
2068  */
2069 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2070                 regex_t *word_regex)
2071 {
2072         int i, j;
2073         long alloc = 0;
2074
2075         out->size = 0;
2076         out->ptr = NULL;
2077
2078         /* fake an empty "0th" word */
2079         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2080         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2081         buffer->orig_nr = 1;
2082
2083         for (i = 0; i < buffer->text.size; i++) {
2084                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2085                         return;
2086
2087                 /* store original boundaries */
2088                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2089                                 buffer->orig_alloc);
2090                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2091                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2092                 buffer->orig_nr++;
2093
2094                 /* store one word */
2095                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2096                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2097                 out->ptr[out->size + j - i] = '\n';
2098                 out->size += j - i + 1;
2099
2100                 i = j - 1;
2101         }
2102 }
2103
2104 /* this executes the word diff on the accumulated buffers */
2105 static void diff_words_show(struct diff_words_data *diff_words)
2106 {
2107         xpparam_t xpp;
2108         xdemitconf_t xecfg;
2109         mmfile_t minus, plus;
2110         struct diff_words_style *style = diff_words->style;
2111
2112         struct diff_options *opt = diff_words->opt;
2113         const char *line_prefix;
2114
2115         assert(opt);
2116         line_prefix = diff_line_prefix(opt);
2117
2118         /* special case: only removal */
2119         if (!diff_words->plus.text.size) {
2120                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2121                                  line_prefix, strlen(line_prefix), 0);
2122                 fn_out_diff_words_write_helper(diff_words->opt,
2123                         &style->old_word, style->newline,
2124                         diff_words->minus.text.size,
2125                         diff_words->minus.text.ptr);
2126                 diff_words->minus.text.size = 0;
2127                 return;
2128         }
2129
2130         diff_words->current_plus = diff_words->plus.text.ptr;
2131         diff_words->last_minus = 0;
2132
2133         memset(&xpp, 0, sizeof(xpp));
2134         memset(&xecfg, 0, sizeof(xecfg));
2135         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2136         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2137         xpp.flags = 0;
2138         /* as only the hunk header will be parsed, we need a 0-context */
2139         xecfg.ctxlen = 0;
2140         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2141                           diff_words, &xpp, &xecfg))
2142                 die("unable to generate word diff");
2143         free(minus.ptr);
2144         free(plus.ptr);
2145         if (diff_words->current_plus != diff_words->plus.text.ptr +
2146                         diff_words->plus.text.size) {
2147                 if (color_words_output_graph_prefix(diff_words))
2148                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2149                                          line_prefix, strlen(line_prefix), 0);
2150                 fn_out_diff_words_write_helper(diff_words->opt,
2151                         &style->ctx, style->newline,
2152                         diff_words->plus.text.ptr + diff_words->plus.text.size
2153                         - diff_words->current_plus, diff_words->current_plus);
2154         }
2155         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2156 }
2157
2158 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2159 static void diff_words_flush(struct emit_callback *ecbdata)
2160 {
2161         struct diff_options *wo = ecbdata->diff_words->opt;
2162
2163         if (ecbdata->diff_words->minus.text.size ||
2164             ecbdata->diff_words->plus.text.size)
2165                 diff_words_show(ecbdata->diff_words);
2166
2167         if (wo->emitted_symbols) {
2168                 struct diff_options *o = ecbdata->opt;
2169                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2170                 int i;
2171
2172                 /*
2173                  * NEEDSWORK:
2174                  * Instead of appending each, concat all words to a line?
2175                  */
2176                 for (i = 0; i < wol->nr; i++)
2177                         append_emitted_diff_symbol(o, &wol->buf[i]);
2178
2179                 for (i = 0; i < wol->nr; i++)
2180                         free((void *)wol->buf[i].line);
2181
2182                 wol->nr = 0;
2183         }
2184 }
2185
2186 static void diff_filespec_load_driver(struct diff_filespec *one,
2187                                       struct index_state *istate)
2188 {
2189         /* Use already-loaded driver */
2190         if (one->driver)
2191                 return;
2192
2193         if (S_ISREG(one->mode))
2194                 one->driver = userdiff_find_by_path(istate, one->path);
2195
2196         /* Fallback to default settings */
2197         if (!one->driver)
2198                 one->driver = userdiff_find_by_name("default");
2199 }
2200
2201 static const char *userdiff_word_regex(struct diff_filespec *one,
2202                                        struct index_state *istate)
2203 {
2204         diff_filespec_load_driver(one, istate);
2205         return one->driver->word_regex;
2206 }
2207
2208 static void init_diff_words_data(struct emit_callback *ecbdata,
2209                                  struct diff_options *orig_opts,
2210                                  struct diff_filespec *one,
2211                                  struct diff_filespec *two)
2212 {
2213         int i;
2214         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2215         memcpy(o, orig_opts, sizeof(struct diff_options));
2216
2217         ecbdata->diff_words =
2218                 xcalloc(1, sizeof(struct diff_words_data));
2219         ecbdata->diff_words->type = o->word_diff;
2220         ecbdata->diff_words->opt = o;
2221
2222         if (orig_opts->emitted_symbols)
2223                 o->emitted_symbols =
2224                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2225
2226         if (!o->word_regex)
2227                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2228         if (!o->word_regex)
2229                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2230         if (!o->word_regex)
2231                 o->word_regex = diff_word_regex_cfg;
2232         if (o->word_regex) {
2233                 ecbdata->diff_words->word_regex = (regex_t *)
2234                         xmalloc(sizeof(regex_t));
2235                 if (regcomp(ecbdata->diff_words->word_regex,
2236                             o->word_regex,
2237                             REG_EXTENDED | REG_NEWLINE))
2238                         die("invalid regular expression: %s",
2239                             o->word_regex);
2240         }
2241         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2242                 if (o->word_diff == diff_words_styles[i].type) {
2243                         ecbdata->diff_words->style =
2244                                 &diff_words_styles[i];
2245                         break;
2246                 }
2247         }
2248         if (want_color(o->use_color)) {
2249                 struct diff_words_style *st = ecbdata->diff_words->style;
2250                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2251                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2252                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2253         }
2254 }
2255
2256 static void free_diff_words_data(struct emit_callback *ecbdata)
2257 {
2258         if (ecbdata->diff_words) {
2259                 diff_words_flush(ecbdata);
2260                 free (ecbdata->diff_words->opt->emitted_symbols);
2261                 free (ecbdata->diff_words->opt);
2262                 free (ecbdata->diff_words->minus.text.ptr);
2263                 free (ecbdata->diff_words->minus.orig);
2264                 free (ecbdata->diff_words->plus.text.ptr);
2265                 free (ecbdata->diff_words->plus.orig);
2266                 if (ecbdata->diff_words->word_regex) {
2267                         regfree(ecbdata->diff_words->word_regex);
2268                         free(ecbdata->diff_words->word_regex);
2269                 }
2270                 FREE_AND_NULL(ecbdata->diff_words);
2271         }
2272 }
2273
2274 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2275 {
2276         if (want_color(diff_use_color))
2277                 return diff_colors[ix];
2278         return "";
2279 }
2280
2281 const char *diff_line_prefix(struct diff_options *opt)
2282 {
2283         struct strbuf *msgbuf;
2284         if (!opt->output_prefix)
2285                 return "";
2286
2287         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2288         return msgbuf->buf;
2289 }
2290
2291 static unsigned long sane_truncate_line(char *line, unsigned long len)
2292 {
2293         const char *cp;
2294         unsigned long allot;
2295         size_t l = len;
2296
2297         cp = line;
2298         allot = l;
2299         while (0 < l) {
2300                 (void) utf8_width(&cp, &l);
2301                 if (!cp)
2302                         break; /* truncated in the middle? */
2303         }
2304         return allot - l;
2305 }
2306
2307 static void find_lno(const char *line, struct emit_callback *ecbdata)
2308 {
2309         const char *p;
2310         ecbdata->lno_in_preimage = 0;
2311         ecbdata->lno_in_postimage = 0;
2312         p = strchr(line, '-');
2313         if (!p)
2314                 return; /* cannot happen */
2315         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2316         p = strchr(p, '+');
2317         if (!p)
2318                 return; /* cannot happen */
2319         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2320 }
2321
2322 static void fn_out_consume(void *priv, char *line, unsigned long len)
2323 {
2324         struct emit_callback *ecbdata = priv;
2325         struct diff_options *o = ecbdata->opt;
2326
2327         o->found_changes = 1;
2328
2329         if (ecbdata->header) {
2330                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2331                                  ecbdata->header->buf, ecbdata->header->len, 0);
2332                 strbuf_reset(ecbdata->header);
2333                 ecbdata->header = NULL;
2334         }
2335
2336         if (ecbdata->label_path[0]) {
2337                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2338                                  ecbdata->label_path[0],
2339                                  strlen(ecbdata->label_path[0]), 0);
2340                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2341                                  ecbdata->label_path[1],
2342                                  strlen(ecbdata->label_path[1]), 0);
2343                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2344         }
2345
2346         if (diff_suppress_blank_empty
2347             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2348                 line[0] = '\n';
2349                 len = 1;
2350         }
2351
2352         if (line[0] == '@') {
2353                 if (ecbdata->diff_words)
2354                         diff_words_flush(ecbdata);
2355                 len = sane_truncate_line(line, len);
2356                 find_lno(line, ecbdata);
2357                 emit_hunk_header(ecbdata, line, len);
2358                 return;
2359         }
2360
2361         if (ecbdata->diff_words) {
2362                 enum diff_symbol s =
2363                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2364                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2365                 if (line[0] == '-') {
2366                         diff_words_append(line, len,
2367                                           &ecbdata->diff_words->minus);
2368                         return;
2369                 } else if (line[0] == '+') {
2370                         diff_words_append(line, len,
2371                                           &ecbdata->diff_words->plus);
2372                         return;
2373                 } else if (starts_with(line, "\\ ")) {
2374                         /*
2375                          * Eat the "no newline at eof" marker as if we
2376                          * saw a "+" or "-" line with nothing on it,
2377                          * and return without diff_words_flush() to
2378                          * defer processing. If this is the end of
2379                          * preimage, more "+" lines may come after it.
2380                          */
2381                         return;
2382                 }
2383                 diff_words_flush(ecbdata);
2384                 emit_diff_symbol(o, s, line, len, 0);
2385                 return;
2386         }
2387
2388         switch (line[0]) {
2389         case '+':
2390                 ecbdata->lno_in_postimage++;
2391                 emit_add_line(ecbdata, line + 1, len - 1);
2392                 break;
2393         case '-':
2394                 ecbdata->lno_in_preimage++;
2395                 emit_del_line(ecbdata, line + 1, len - 1);
2396                 break;
2397         case ' ':
2398                 ecbdata->lno_in_postimage++;
2399                 ecbdata->lno_in_preimage++;
2400                 emit_context_line(ecbdata, line + 1, len - 1);
2401                 break;
2402         default:
2403                 /* incomplete line at the end */
2404                 ecbdata->lno_in_preimage++;
2405                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2406                                  line, len, 0);
2407                 break;
2408         }
2409 }
2410
2411 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2412 {
2413         const char *old_name = a;
2414         const char *new_name = b;
2415         int pfx_length, sfx_length;
2416         int pfx_adjust_for_slash;
2417         int len_a = strlen(a);
2418         int len_b = strlen(b);
2419         int a_midlen, b_midlen;
2420         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2421         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2422
2423         if (qlen_a || qlen_b) {
2424                 quote_c_style(a, name, NULL, 0);
2425                 strbuf_addstr(name, " => ");
2426                 quote_c_style(b, name, NULL, 0);
2427                 return;
2428         }
2429
2430         /* Find common prefix */
2431         pfx_length = 0;
2432         while (*old_name && *new_name && *old_name == *new_name) {
2433                 if (*old_name == '/')
2434                         pfx_length = old_name - a + 1;
2435                 old_name++;
2436                 new_name++;
2437         }
2438
2439         /* Find common suffix */
2440         old_name = a + len_a;
2441         new_name = b + len_b;
2442         sfx_length = 0;
2443         /*
2444          * If there is a common prefix, it must end in a slash.  In
2445          * that case we let this loop run 1 into the prefix to see the
2446          * same slash.
2447          *
2448          * If there is no common prefix, we cannot do this as it would
2449          * underrun the input strings.
2450          */
2451         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2452         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2453                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2454                *old_name == *new_name) {
2455                 if (*old_name == '/')
2456                         sfx_length = len_a - (old_name - a);
2457                 old_name--;
2458                 new_name--;
2459         }
2460
2461         /*
2462          * pfx{mid-a => mid-b}sfx
2463          * {pfx-a => pfx-b}sfx
2464          * pfx{sfx-a => sfx-b}
2465          * name-a => name-b
2466          */
2467         a_midlen = len_a - pfx_length - sfx_length;
2468         b_midlen = len_b - pfx_length - sfx_length;
2469         if (a_midlen < 0)
2470                 a_midlen = 0;
2471         if (b_midlen < 0)
2472                 b_midlen = 0;
2473
2474         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2475         if (pfx_length + sfx_length) {
2476                 strbuf_add(name, a, pfx_length);
2477                 strbuf_addch(name, '{');
2478         }
2479         strbuf_add(name, a + pfx_length, a_midlen);
2480         strbuf_addstr(name, " => ");
2481         strbuf_add(name, b + pfx_length, b_midlen);
2482         if (pfx_length + sfx_length) {
2483                 strbuf_addch(name, '}');
2484                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2485         }
2486 }
2487
2488 struct diffstat_t {
2489         int nr;
2490         int alloc;
2491         struct diffstat_file {
2492                 char *from_name;
2493                 char *name;
2494                 char *print_name;
2495                 const char *comments;
2496                 unsigned is_unmerged:1;
2497                 unsigned is_binary:1;
2498                 unsigned is_renamed:1;
2499                 unsigned is_interesting:1;
2500                 uintmax_t added, deleted;
2501         } **files;
2502 };
2503
2504 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2505                                           const char *name_a,
2506                                           const char *name_b)
2507 {
2508         struct diffstat_file *x;
2509         x = xcalloc(1, sizeof(*x));
2510         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2511         diffstat->files[diffstat->nr++] = x;
2512         if (name_b) {
2513                 x->from_name = xstrdup(name_a);
2514                 x->name = xstrdup(name_b);
2515                 x->is_renamed = 1;
2516         }
2517         else {
2518                 x->from_name = NULL;
2519                 x->name = xstrdup(name_a);
2520         }
2521         return x;
2522 }
2523
2524 static void diffstat_consume(void *priv, char *line, unsigned long len)
2525 {
2526         struct diffstat_t *diffstat = priv;
2527         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2528
2529         if (line[0] == '+')
2530                 x->added++;
2531         else if (line[0] == '-')
2532                 x->deleted++;
2533 }
2534
2535 const char mime_boundary_leader[] = "------------";
2536
2537 static int scale_linear(int it, int width, int max_change)
2538 {
2539         if (!it)
2540                 return 0;
2541         /*
2542          * make sure that at least one '-' or '+' is printed if
2543          * there is any change to this path. The easiest way is to
2544          * scale linearly as if the alloted width is one column shorter
2545          * than it is, and then add 1 to the result.
2546          */
2547         return 1 + (it * (width - 1) / max_change);
2548 }
2549
2550 static void show_graph(struct strbuf *out, char ch, int cnt,
2551                        const char *set, const char *reset)
2552 {
2553         if (cnt <= 0)
2554                 return;
2555         strbuf_addstr(out, set);
2556         strbuf_addchars(out, ch, cnt);
2557         strbuf_addstr(out, reset);
2558 }
2559
2560 static void fill_print_name(struct diffstat_file *file)
2561 {
2562         struct strbuf pname = STRBUF_INIT;
2563
2564         if (file->print_name)
2565                 return;
2566
2567         if (file->is_renamed)
2568                 pprint_rename(&pname, file->from_name, file->name);
2569         else
2570                 quote_c_style(file->name, &pname, NULL, 0);
2571
2572         if (file->comments)
2573                 strbuf_addf(&pname, " (%s)", file->comments);
2574
2575         file->print_name = strbuf_detach(&pname, NULL);
2576 }
2577
2578 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2579                 int files, int insertions, int deletions)
2580 {
2581         struct strbuf sb = STRBUF_INIT;
2582
2583         if (!files) {
2584                 assert(insertions == 0 && deletions == 0);
2585                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2586                                  NULL, 0, 0);
2587                 return;
2588         }
2589
2590         strbuf_addf(&sb,
2591                     (files == 1) ? " %d file changed" : " %d files changed",
2592                     files);
2593
2594         /*
2595          * For binary diff, the caller may want to print "x files
2596          * changed" with insertions == 0 && deletions == 0.
2597          *
2598          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2599          * is probably less confusing (i.e skip over "2 files changed
2600          * but nothing about added/removed lines? Is this a bug in Git?").
2601          */
2602         if (insertions || deletions == 0) {
2603                 strbuf_addf(&sb,
2604                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2605                             insertions);
2606         }
2607
2608         if (deletions || insertions == 0) {
2609                 strbuf_addf(&sb,
2610                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2611                             deletions);
2612         }
2613         strbuf_addch(&sb, '\n');
2614         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2615                          sb.buf, sb.len, 0);
2616         strbuf_release(&sb);
2617 }
2618
2619 void print_stat_summary(FILE *fp, int files,
2620                         int insertions, int deletions)
2621 {
2622         struct diff_options o;
2623         memset(&o, 0, sizeof(o));
2624         o.file = fp;
2625
2626         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2627 }
2628
2629 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2630 {
2631         int i, len, add, del, adds = 0, dels = 0;
2632         uintmax_t max_change = 0, max_len = 0;
2633         int total_files = data->nr, count;
2634         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2635         const char *reset, *add_c, *del_c;
2636         int extra_shown = 0;
2637         const char *line_prefix = diff_line_prefix(options);
2638         struct strbuf out = STRBUF_INIT;
2639
2640         if (data->nr == 0)
2641                 return;
2642
2643         count = options->stat_count ? options->stat_count : data->nr;
2644
2645         reset = diff_get_color_opt(options, DIFF_RESET);
2646         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2647         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2648
2649         /*
2650          * Find the longest filename and max number of changes
2651          */
2652         for (i = 0; (i < count) && (i < data->nr); i++) {
2653                 struct diffstat_file *file = data->files[i];
2654                 uintmax_t change = file->added + file->deleted;
2655
2656                 if (!file->is_interesting && (change == 0)) {
2657                         count++; /* not shown == room for one more */
2658                         continue;
2659                 }
2660                 fill_print_name(file);
2661                 len = strlen(file->print_name);
2662                 if (max_len < len)
2663                         max_len = len;
2664
2665                 if (file->is_unmerged) {
2666                         /* "Unmerged" is 8 characters */
2667                         bin_width = bin_width < 8 ? 8 : bin_width;
2668                         continue;
2669                 }
2670                 if (file->is_binary) {
2671                         /* "Bin XXX -> YYY bytes" */
2672                         int w = 14 + decimal_width(file->added)
2673                                 + decimal_width(file->deleted);
2674                         bin_width = bin_width < w ? w : bin_width;
2675                         /* Display change counts aligned with "Bin" */
2676                         number_width = 3;
2677                         continue;
2678                 }
2679
2680                 if (max_change < change)
2681                         max_change = change;
2682         }
2683         count = i; /* where we can stop scanning in data->files[] */
2684
2685         /*
2686          * We have width = stat_width or term_columns() columns total.
2687          * We want a maximum of min(max_len, stat_name_width) for the name part.
2688          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2689          * We also need 1 for " " and 4 + decimal_width(max_change)
2690          * for " | NNNN " and one the empty column at the end, altogether
2691          * 6 + decimal_width(max_change).
2692          *
2693          * If there's not enough space, we will use the smaller of
2694          * stat_name_width (if set) and 5/8*width for the filename,
2695          * and the rest for constant elements + graph part, but no more
2696          * than stat_graph_width for the graph part.
2697          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2698          * for the standard terminal size).
2699          *
2700          * In other words: stat_width limits the maximum width, and
2701          * stat_name_width fixes the maximum width of the filename,
2702          * and is also used to divide available columns if there
2703          * aren't enough.
2704          *
2705          * Binary files are displayed with "Bin XXX -> YYY bytes"
2706          * instead of the change count and graph. This part is treated
2707          * similarly to the graph part, except that it is not
2708          * "scaled". If total width is too small to accommodate the
2709          * guaranteed minimum width of the filename part and the
2710          * separators and this message, this message will "overflow"
2711          * making the line longer than the maximum width.
2712          */
2713
2714         if (options->stat_width == -1)
2715                 width = term_columns() - strlen(line_prefix);
2716         else
2717                 width = options->stat_width ? options->stat_width : 80;
2718         number_width = decimal_width(max_change) > number_width ?
2719                 decimal_width(max_change) : number_width;
2720
2721         if (options->stat_graph_width == -1)
2722                 options->stat_graph_width = diff_stat_graph_width;
2723
2724         /*
2725          * Guarantee 3/8*16==6 for the graph part
2726          * and 5/8*16==10 for the filename part
2727          */
2728         if (width < 16 + 6 + number_width)
2729                 width = 16 + 6 + number_width;
2730
2731         /*
2732          * First assign sizes that are wanted, ignoring available width.
2733          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2734          * starting from "XXX" should fit in graph_width.
2735          */
2736         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2737         if (options->stat_graph_width &&
2738             options->stat_graph_width < graph_width)
2739                 graph_width = options->stat_graph_width;
2740
2741         name_width = (options->stat_name_width > 0 &&
2742                       options->stat_name_width < max_len) ?
2743                 options->stat_name_width : max_len;
2744
2745         /*
2746          * Adjust adjustable widths not to exceed maximum width
2747          */
2748         if (name_width + number_width + 6 + graph_width > width) {
2749                 if (graph_width > width * 3/8 - number_width - 6) {
2750                         graph_width = width * 3/8 - number_width - 6;
2751                         if (graph_width < 6)
2752                                 graph_width = 6;
2753                 }
2754
2755                 if (options->stat_graph_width &&
2756                     graph_width > options->stat_graph_width)
2757                         graph_width = options->stat_graph_width;
2758                 if (name_width > width - number_width - 6 - graph_width)
2759                         name_width = width - number_width - 6 - graph_width;
2760                 else
2761                         graph_width = width - number_width - 6 - name_width;
2762         }
2763
2764         /*
2765          * From here name_width is the width of the name area,
2766          * and graph_width is the width of the graph area.
2767          * max_change is used to scale graph properly.
2768          */
2769         for (i = 0; i < count; i++) {
2770                 const char *prefix = "";
2771                 struct diffstat_file *file = data->files[i];
2772                 char *name = file->print_name;
2773                 uintmax_t added = file->added;
2774                 uintmax_t deleted = file->deleted;
2775                 int name_len;
2776
2777                 if (!file->is_interesting && (added + deleted == 0))
2778                         continue;
2779
2780                 /*
2781                  * "scale" the filename
2782                  */
2783                 len = name_width;
2784                 name_len = strlen(name);
2785                 if (name_width < name_len) {
2786                         char *slash;
2787                         prefix = "...";
2788                         len -= 3;
2789                         name += name_len - len;
2790                         slash = strchr(name, '/');
2791                         if (slash)
2792                                 name = slash;
2793                 }
2794
2795                 if (file->is_binary) {
2796                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2797                         strbuf_addf(&out, " %*s", number_width, "Bin");
2798                         if (!added && !deleted) {
2799                                 strbuf_addch(&out, '\n');
2800                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2801                                                  out.buf, out.len, 0);
2802                                 strbuf_reset(&out);
2803                                 continue;
2804                         }
2805                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2806                                 del_c, deleted, reset);
2807                         strbuf_addstr(&out, " -> ");
2808                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2809                                 add_c, added, reset);
2810                         strbuf_addstr(&out, " bytes\n");
2811                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2812                                          out.buf, out.len, 0);
2813                         strbuf_reset(&out);
2814                         continue;
2815                 }
2816                 else if (file->is_unmerged) {
2817                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2818                         strbuf_addstr(&out, " Unmerged\n");
2819                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2820                                          out.buf, out.len, 0);
2821                         strbuf_reset(&out);
2822                         continue;
2823                 }
2824
2825                 /*
2826                  * scale the add/delete
2827                  */
2828                 add = added;
2829                 del = deleted;
2830
2831                 if (graph_width <= max_change) {
2832                         int total = scale_linear(add + del, graph_width, max_change);
2833                         if (total < 2 && add && del)
2834                                 /* width >= 2 due to the sanity check */
2835                                 total = 2;
2836                         if (add < del) {
2837                                 add = scale_linear(add, graph_width, max_change);
2838                                 del = total - add;
2839                         } else {
2840                                 del = scale_linear(del, graph_width, max_change);
2841                                 add = total - del;
2842                         }
2843                 }
2844                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2845                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2846                         number_width, added + deleted,
2847                         added + deleted ? " " : "");
2848                 show_graph(&out, '+', add, add_c, reset);
2849                 show_graph(&out, '-', del, del_c, reset);
2850                 strbuf_addch(&out, '\n');
2851                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2852                                  out.buf, out.len, 0);
2853                 strbuf_reset(&out);
2854         }
2855
2856         for (i = 0; i < data->nr; i++) {
2857                 struct diffstat_file *file = data->files[i];
2858                 uintmax_t added = file->added;
2859                 uintmax_t deleted = file->deleted;
2860
2861                 if (file->is_unmerged ||
2862                     (!file->is_interesting && (added + deleted == 0))) {
2863                         total_files--;
2864                         continue;
2865                 }
2866
2867                 if (!file->is_binary) {
2868                         adds += added;
2869                         dels += deleted;
2870                 }
2871                 if (i < count)
2872                         continue;
2873                 if (!extra_shown)
2874                         emit_diff_symbol(options,
2875                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2876                                          NULL, 0, 0);
2877                 extra_shown = 1;
2878         }
2879
2880         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2881         strbuf_release(&out);
2882 }
2883
2884 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2885 {
2886         int i, adds = 0, dels = 0, total_files = data->nr;
2887
2888         if (data->nr == 0)
2889                 return;
2890
2891         for (i = 0; i < data->nr; i++) {
2892                 int added = data->files[i]->added;
2893                 int deleted = data->files[i]->deleted;
2894
2895                 if (data->files[i]->is_unmerged ||
2896                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2897                         total_files--;
2898                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2899                         adds += added;
2900                         dels += deleted;
2901                 }
2902         }
2903         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2904 }
2905
2906 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2907 {
2908         int i;
2909
2910         if (data->nr == 0)
2911                 return;
2912
2913         for (i = 0; i < data->nr; i++) {
2914                 struct diffstat_file *file = data->files[i];
2915
2916                 fprintf(options->file, "%s", diff_line_prefix(options));
2917
2918                 if (file->is_binary)
2919                         fprintf(options->file, "-\t-\t");
2920                 else
2921                         fprintf(options->file,
2922                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2923                                 file->added, file->deleted);
2924                 if (options->line_termination) {
2925                         fill_print_name(file);
2926                         if (!file->is_renamed)
2927                                 write_name_quoted(file->name, options->file,
2928                                                   options->line_termination);
2929                         else {
2930                                 fputs(file->print_name, options->file);
2931                                 putc(options->line_termination, options->file);
2932                         }
2933                 } else {
2934                         if (file->is_renamed) {
2935                                 putc('\0', options->file);
2936                                 write_name_quoted(file->from_name, options->file, '\0');
2937                         }
2938                         write_name_quoted(file->name, options->file, '\0');
2939                 }
2940         }
2941 }
2942
2943 struct dirstat_file {
2944         const char *name;
2945         unsigned long changed;
2946 };
2947
2948 struct dirstat_dir {
2949         struct dirstat_file *files;
2950         int alloc, nr, permille, cumulative;
2951 };
2952
2953 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2954                 unsigned long changed, const char *base, int baselen)
2955 {
2956         unsigned long sum_changes = 0;
2957         unsigned int sources = 0;
2958         const char *line_prefix = diff_line_prefix(opt);
2959
2960         while (dir->nr) {
2961                 struct dirstat_file *f = dir->files;
2962                 int namelen = strlen(f->name);
2963                 unsigned long changes;
2964                 char *slash;
2965
2966                 if (namelen < baselen)
2967                         break;
2968                 if (memcmp(f->name, base, baselen))
2969                         break;
2970                 slash = strchr(f->name + baselen, '/');
2971                 if (slash) {
2972                         int newbaselen = slash + 1 - f->name;
2973                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2974                         sources++;
2975                 } else {
2976                         changes = f->changed;
2977                         dir->files++;
2978                         dir->nr--;
2979                         sources += 2;
2980                 }
2981                 sum_changes += changes;
2982         }
2983
2984         /*
2985          * We don't report dirstat's for
2986          *  - the top level
2987          *  - or cases where everything came from a single directory
2988          *    under this directory (sources == 1).
2989          */
2990         if (baselen && sources != 1) {
2991                 if (sum_changes) {
2992                         int permille = sum_changes * 1000 / changed;
2993                         if (permille >= dir->permille) {
2994                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2995                                         permille / 10, permille % 10, baselen, base);
2996                                 if (!dir->cumulative)
2997                                         return 0;
2998                         }
2999                 }
3000         }
3001         return sum_changes;
3002 }
3003
3004 static int dirstat_compare(const void *_a, const void *_b)
3005 {
3006         const struct dirstat_file *a = _a;
3007         const struct dirstat_file *b = _b;
3008         return strcmp(a->name, b->name);
3009 }
3010
3011 static void show_dirstat(struct diff_options *options)
3012 {
3013         int i;
3014         unsigned long changed;
3015         struct dirstat_dir dir;
3016         struct diff_queue_struct *q = &diff_queued_diff;
3017
3018         dir.files = NULL;
3019         dir.alloc = 0;
3020         dir.nr = 0;
3021         dir.permille = options->dirstat_permille;
3022         dir.cumulative = options->flags.dirstat_cumulative;
3023
3024         changed = 0;
3025         for (i = 0; i < q->nr; i++) {
3026                 struct diff_filepair *p = q->queue[i];
3027                 const char *name;
3028                 unsigned long copied, added, damage;
3029
3030                 name = p->two->path ? p->two->path : p->one->path;
3031
3032                 if (p->one->oid_valid && p->two->oid_valid &&
3033                     oideq(&p->one->oid, &p->two->oid)) {
3034                         /*
3035                          * The SHA1 has not changed, so pre-/post-content is
3036                          * identical. We can therefore skip looking at the
3037                          * file contents altogether.
3038                          */
3039                         damage = 0;
3040                         goto found_damage;
3041                 }
3042
3043                 if (options->flags.dirstat_by_file) {
3044                         /*
3045                          * In --dirstat-by-file mode, we don't really need to
3046                          * look at the actual file contents at all.
3047                          * The fact that the SHA1 changed is enough for us to
3048                          * add this file to the list of results
3049                          * (with each file contributing equal damage).
3050                          */
3051                         damage = 1;
3052                         goto found_damage;
3053                 }
3054
3055                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3056                         diff_populate_filespec(options->repo, p->one, 0);
3057                         diff_populate_filespec(options->repo, p->two, 0);
3058                         diffcore_count_changes(options->repo,
3059                                                p->one, p->two, NULL, NULL,
3060                                                &copied, &added);
3061                         diff_free_filespec_data(p->one);
3062                         diff_free_filespec_data(p->two);
3063                 } else if (DIFF_FILE_VALID(p->one)) {
3064                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3065                         copied = added = 0;
3066                         diff_free_filespec_data(p->one);
3067                 } else if (DIFF_FILE_VALID(p->two)) {
3068                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3069                         copied = 0;
3070                         added = p->two->size;
3071                         diff_free_filespec_data(p->two);
3072                 } else
3073                         continue;
3074
3075                 /*
3076                  * Original minus copied is the removed material,
3077                  * added is the new material.  They are both damages
3078                  * made to the preimage.
3079                  * If the resulting damage is zero, we know that
3080                  * diffcore_count_changes() considers the two entries to
3081                  * be identical, but since the oid changed, we
3082                  * know that there must have been _some_ kind of change,
3083                  * so we force all entries to have damage > 0.
3084                  */
3085                 damage = (p->one->size - copied) + added;
3086                 if (!damage)
3087                         damage = 1;
3088
3089 found_damage:
3090                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3091                 dir.files[dir.nr].name = name;
3092                 dir.files[dir.nr].changed = damage;
3093                 changed += damage;
3094                 dir.nr++;
3095         }
3096
3097         /* This can happen even with many files, if everything was renames */
3098         if (!changed)
3099                 return;
3100
3101         /* Show all directories with more than x% of the changes */
3102         QSORT(dir.files, dir.nr, dirstat_compare);
3103         gather_dirstat(options, &dir, changed, "", 0);
3104 }
3105
3106 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3107 {
3108         int i;
3109         unsigned long changed;
3110         struct dirstat_dir dir;
3111
3112         if (data->nr == 0)
3113                 return;
3114
3115         dir.files = NULL;
3116         dir.alloc = 0;
3117         dir.nr = 0;
3118         dir.permille = options->dirstat_permille;
3119         dir.cumulative = options->flags.dirstat_cumulative;
3120
3121         changed = 0;
3122         for (i = 0; i < data->nr; i++) {
3123                 struct diffstat_file *file = data->files[i];
3124                 unsigned long damage = file->added + file->deleted;
3125                 if (file->is_binary)
3126                         /*
3127                          * binary files counts bytes, not lines. Must find some
3128                          * way to normalize binary bytes vs. textual lines.
3129                          * The following heuristic assumes that there are 64
3130                          * bytes per "line".
3131                          * This is stupid and ugly, but very cheap...
3132                          */
3133                         damage = DIV_ROUND_UP(damage, 64);
3134                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3135                 dir.files[dir.nr].name = file->name;
3136                 dir.files[dir.nr].changed = damage;
3137                 changed += damage;
3138                 dir.nr++;
3139         }
3140
3141         /* This can happen even with many files, if everything was renames */
3142         if (!changed)
3143                 return;
3144
3145         /* Show all directories with more than x% of the changes */
3146         QSORT(dir.files, dir.nr, dirstat_compare);
3147         gather_dirstat(options, &dir, changed, "", 0);
3148 }
3149
3150 static void free_diffstat_info(struct diffstat_t *diffstat)
3151 {
3152         int i;
3153         for (i = 0; i < diffstat->nr; i++) {
3154                 struct diffstat_file *f = diffstat->files[i];
3155                 free(f->print_name);
3156                 free(f->name);
3157                 free(f->from_name);
3158                 free(f);
3159         }
3160         free(diffstat->files);
3161 }
3162
3163 struct checkdiff_t {
3164         const char *filename;
3165         int lineno;
3166         int conflict_marker_size;
3167         struct diff_options *o;
3168         unsigned ws_rule;
3169         unsigned status;
3170 };
3171
3172 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3173 {
3174         char firstchar;
3175         int cnt;
3176
3177         if (len < marker_size + 1)
3178                 return 0;
3179         firstchar = line[0];
3180         switch (firstchar) {
3181         case '=': case '>': case '<': case '|':
3182                 break;
3183         default:
3184                 return 0;
3185         }
3186         for (cnt = 1; cnt < marker_size; cnt++)
3187                 if (line[cnt] != firstchar)
3188                         return 0;
3189         /* line[1] thru line[marker_size-1] are same as firstchar */
3190         if (len < marker_size + 1 || !isspace(line[marker_size]))
3191                 return 0;
3192         return 1;
3193 }
3194
3195 static void checkdiff_consume_hunk(void *priv,
3196                                    long ob, long on, long nb, long nn,
3197                                    const char *func, long funclen)
3198
3199 {
3200         struct checkdiff_t *data = priv;
3201         data->lineno = nb - 1;
3202 }
3203
3204 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3205 {
3206         struct checkdiff_t *data = priv;
3207         int marker_size = data->conflict_marker_size;
3208         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3209         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3210         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3211         char *err;
3212         const char *line_prefix;
3213
3214         assert(data->o);
3215         line_prefix = diff_line_prefix(data->o);
3216
3217         if (line[0] == '+') {
3218                 unsigned bad;
3219                 data->lineno++;
3220                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3221                         data->status |= 1;
3222                         fprintf(data->o->file,
3223                                 "%s%s:%d: leftover conflict marker\n",
3224                                 line_prefix, data->filename, data->lineno);
3225                 }
3226                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3227                 if (!bad)
3228                         return;
3229                 data->status |= bad;
3230                 err = whitespace_error_string(bad);
3231                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3232                         line_prefix, data->filename, data->lineno, err);
3233                 free(err);
3234                 emit_line(data->o, set, reset, line, 1);
3235                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3236                               data->o->file, set, reset, ws);
3237         } else if (line[0] == ' ') {
3238                 data->lineno++;
3239         }
3240 }
3241
3242 static unsigned char *deflate_it(char *data,
3243                                  unsigned long size,
3244                                  unsigned long *result_size)
3245 {
3246         int bound;
3247         unsigned char *deflated;
3248         git_zstream stream;
3249
3250         git_deflate_init(&stream, zlib_compression_level);
3251         bound = git_deflate_bound(&stream, size);
3252         deflated = xmalloc(bound);
3253         stream.next_out = deflated;
3254         stream.avail_out = bound;
3255
3256         stream.next_in = (unsigned char *)data;
3257         stream.avail_in = size;
3258         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3259                 ; /* nothing */
3260         git_deflate_end(&stream);
3261         *result_size = stream.total_out;
3262         return deflated;
3263 }
3264
3265 static void emit_binary_diff_body(struct diff_options *o,
3266                                   mmfile_t *one, mmfile_t *two)
3267 {
3268         void *cp;
3269         void *delta;
3270         void *deflated;
3271         void *data;
3272         unsigned long orig_size;
3273         unsigned long delta_size;
3274         unsigned long deflate_size;
3275         unsigned long data_size;
3276
3277         /* We could do deflated delta, or we could do just deflated two,
3278          * whichever is smaller.
3279          */
3280         delta = NULL;
3281         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3282         if (one->size && two->size) {
3283                 delta = diff_delta(one->ptr, one->size,
3284                                    two->ptr, two->size,
3285                                    &delta_size, deflate_size);
3286                 if (delta) {
3287                         void *to_free = delta;
3288                         orig_size = delta_size;
3289                         delta = deflate_it(delta, delta_size, &delta_size);
3290                         free(to_free);
3291                 }
3292         }
3293
3294         if (delta && delta_size < deflate_size) {
3295                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3296                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3297                                  s, strlen(s), 0);
3298                 free(s);
3299                 free(deflated);
3300                 data = delta;
3301                 data_size = delta_size;
3302         } else {
3303                 char *s = xstrfmt("%lu", two->size);
3304                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3305                                  s, strlen(s), 0);
3306                 free(s);
3307                 free(delta);
3308                 data = deflated;
3309                 data_size = deflate_size;
3310         }
3311
3312         /* emit data encoded in base85 */
3313         cp = data;
3314         while (data_size) {
3315                 int len;
3316                 int bytes = (52 < data_size) ? 52 : data_size;
3317                 char line[71];
3318                 data_size -= bytes;
3319                 if (bytes <= 26)
3320                         line[0] = bytes + 'A' - 1;
3321                 else
3322                         line[0] = bytes - 26 + 'a' - 1;
3323                 encode_85(line + 1, cp, bytes);
3324                 cp = (char *) cp + bytes;
3325
3326                 len = strlen(line);
3327                 line[len++] = '\n';
3328                 line[len] = '\0';
3329
3330                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3331                                  line, len, 0);
3332         }
3333         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3334         free(data);
3335 }
3336
3337 static void emit_binary_diff(struct diff_options *o,
3338                              mmfile_t *one, mmfile_t *two)
3339 {
3340         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3341         emit_binary_diff_body(o, one, two);
3342         emit_binary_diff_body(o, two, one);
3343 }
3344
3345 int diff_filespec_is_binary(struct repository *r,
3346                             struct diff_filespec *one)
3347 {
3348         if (one->is_binary == -1) {
3349                 diff_filespec_load_driver(one, r->index);
3350                 if (one->driver->binary != -1)
3351                         one->is_binary = one->driver->binary;
3352                 else {
3353                         if (!one->data && DIFF_FILE_VALID(one))
3354                                 diff_populate_filespec(r, one, CHECK_BINARY);
3355                         if (one->is_binary == -1 && one->data)
3356                                 one->is_binary = buffer_is_binary(one->data,
3357                                                 one->size);
3358                         if (one->is_binary == -1)
3359                                 one->is_binary = 0;
3360                 }
3361         }
3362         return one->is_binary;
3363 }
3364
3365 static const struct userdiff_funcname *
3366 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3367 {
3368         diff_filespec_load_driver(one, o->repo->index);
3369         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3370 }
3371
3372 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3373 {
3374         if (!options->a_prefix)
3375                 options->a_prefix = a;
3376         if (!options->b_prefix)
3377                 options->b_prefix = b;
3378 }
3379
3380 struct userdiff_driver *get_textconv(struct repository *r,
3381                                      struct diff_filespec *one)
3382 {
3383         if (!DIFF_FILE_VALID(one))
3384                 return NULL;
3385
3386         diff_filespec_load_driver(one, r->index);
3387         return userdiff_get_textconv(r, one->driver);
3388 }
3389
3390 static void builtin_diff(const char *name_a,
3391                          const char *name_b,
3392                          struct diff_filespec *one,
3393                          struct diff_filespec *two,
3394                          const char *xfrm_msg,
3395                          int must_show_header,
3396                          struct diff_options *o,
3397                          int complete_rewrite)
3398 {
3399         mmfile_t mf1, mf2;
3400         const char *lbl[2];
3401         char *a_one, *b_two;
3402         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3403         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3404         const char *a_prefix, *b_prefix;
3405         struct userdiff_driver *textconv_one = NULL;
3406         struct userdiff_driver *textconv_two = NULL;
3407         struct strbuf header = STRBUF_INIT;
3408         const char *line_prefix = diff_line_prefix(o);
3409
3410         diff_set_mnemonic_prefix(o, "a/", "b/");
3411         if (o->flags.reverse_diff) {
3412                 a_prefix = o->b_prefix;
3413                 b_prefix = o->a_prefix;
3414         } else {
3415                 a_prefix = o->a_prefix;
3416                 b_prefix = o->b_prefix;
3417         }
3418
3419         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3420             (!one->mode || S_ISGITLINK(one->mode)) &&
3421             (!two->mode || S_ISGITLINK(two->mode))) {
3422                 show_submodule_summary(o, one->path ? one->path : two->path,
3423                                 &one->oid, &two->oid,
3424                                 two->dirty_submodule);
3425                 return;
3426         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3427                    (!one->mode || S_ISGITLINK(one->mode)) &&
3428                    (!two->mode || S_ISGITLINK(two->mode))) {
3429                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3430                                 &one->oid, &two->oid,
3431                                 two->dirty_submodule);
3432                 return;
3433         }
3434
3435         if (o->flags.allow_textconv) {
3436                 textconv_one = get_textconv(o->repo, one);
3437                 textconv_two = get_textconv(o->repo, two);
3438         }
3439
3440         /* Never use a non-valid filename anywhere if at all possible */
3441         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3442         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3443
3444         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3445         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3446         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3447         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3448         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3449         if (lbl[0][0] == '/') {
3450                 /* /dev/null */
3451                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3452                 if (xfrm_msg)
3453                         strbuf_addstr(&header, xfrm_msg);
3454                 must_show_header = 1;
3455         }
3456         else if (lbl[1][0] == '/') {
3457                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3458                 if (xfrm_msg)
3459                         strbuf_addstr(&header, xfrm_msg);
3460                 must_show_header = 1;
3461         }
3462         else {
3463                 if (one->mode != two->mode) {
3464                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3465                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3466                         must_show_header = 1;
3467                 }
3468                 if (xfrm_msg)
3469                         strbuf_addstr(&header, xfrm_msg);
3470
3471                 /*
3472                  * we do not run diff between different kind
3473                  * of objects.
3474                  */
3475                 if ((one->mode ^ two->mode) & S_IFMT)
3476                         goto free_ab_and_return;
3477                 if (complete_rewrite &&
3478                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3479                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3480                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3481                                          header.buf, header.len, 0);
3482                         strbuf_reset(&header);
3483                         emit_rewrite_diff(name_a, name_b, one, two,
3484                                           textconv_one, textconv_two, o);
3485                         o->found_changes = 1;
3486                         goto free_ab_and_return;
3487                 }
3488         }
3489
3490         if (o->irreversible_delete && lbl[1][0] == '/') {
3491                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3492                                  header.len, 0);
3493                 strbuf_reset(&header);
3494                 goto free_ab_and_return;
3495         } else if (!o->flags.text &&
3496                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3497                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3498                 struct strbuf sb = STRBUF_INIT;
3499                 if (!one->data && !two->data &&
3500                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3501                     !o->flags.binary) {
3502                         if (oideq(&one->oid, &two->oid)) {
3503                                 if (must_show_header)
3504                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3505                                                          header.buf, header.len,
3506                                                          0);
3507                                 goto free_ab_and_return;
3508                         }
3509                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3510                                          header.buf, header.len, 0);
3511                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3512                                     diff_line_prefix(o), lbl[0], lbl[1]);
3513                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3514                                          sb.buf, sb.len, 0);
3515                         strbuf_release(&sb);
3516                         goto free_ab_and_return;
3517                 }
3518                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3519                     fill_mmfile(o->repo, &mf2, two) < 0)
3520                         die("unable to read files to diff");
3521                 /* Quite common confusing case */
3522                 if (mf1.size == mf2.size &&
3523                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3524                         if (must_show_header)
3525                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3526                                                  header.buf, header.len, 0);
3527                         goto free_ab_and_return;
3528                 }
3529                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3530                 strbuf_reset(&header);
3531                 if (o->flags.binary)
3532                         emit_binary_diff(o, &mf1, &mf2);
3533                 else {
3534                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3535                                     diff_line_prefix(o), lbl[0], lbl[1]);
3536                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3537                                          sb.buf, sb.len, 0);
3538                         strbuf_release(&sb);
3539                 }
3540                 o->found_changes = 1;
3541         } else {
3542                 /* Crazy xdl interfaces.. */
3543                 const char *diffopts;
3544                 const char *v;
3545                 xpparam_t xpp;
3546                 xdemitconf_t xecfg;
3547                 struct emit_callback ecbdata;
3548                 const struct userdiff_funcname *pe;
3549
3550                 if (must_show_header) {
3551                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3552                                          header.buf, header.len, 0);
3553                         strbuf_reset(&header);
3554                 }
3555
3556                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3557                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3558
3559                 pe = diff_funcname_pattern(o, one);
3560                 if (!pe)
3561                         pe = diff_funcname_pattern(o, two);
3562
3563                 memset(&xpp, 0, sizeof(xpp));
3564                 memset(&xecfg, 0, sizeof(xecfg));
3565                 memset(&ecbdata, 0, sizeof(ecbdata));
3566                 if (o->flags.suppress_diff_headers)
3567                         lbl[0] = NULL;
3568                 ecbdata.label_path = lbl;
3569                 ecbdata.color_diff = want_color(o->use_color);
3570                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3571                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3572                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3573                 ecbdata.opt = o;
3574                 if (header.len && !o->flags.suppress_diff_headers)
3575                         ecbdata.header = &header;
3576                 xpp.flags = o->xdl_opts;
3577                 xpp.anchors = o->anchors;
3578                 xpp.anchors_nr = o->anchors_nr;
3579                 xecfg.ctxlen = o->context;
3580                 xecfg.interhunkctxlen = o->interhunkcontext;
3581                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3582                 if (o->flags.funccontext)
3583                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3584                 if (pe)
3585                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3586
3587                 diffopts = getenv("GIT_DIFF_OPTS");
3588                 if (!diffopts)
3589                         ;
3590                 else if (skip_prefix(diffopts, "--unified=", &v))
3591                         xecfg.ctxlen = strtoul(v, NULL, 10);
3592                 else if (skip_prefix(diffopts, "-u", &v))
3593                         xecfg.ctxlen = strtoul(v, NULL, 10);
3594
3595                 if (o->word_diff)
3596                         init_diff_words_data(&ecbdata, o, one, two);
3597                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3598                                   &ecbdata, &xpp, &xecfg))
3599                         die("unable to generate diff for %s", one->path);
3600                 if (o->word_diff)
3601                         free_diff_words_data(&ecbdata);
3602                 if (textconv_one)
3603                         free(mf1.ptr);
3604                 if (textconv_two)
3605                         free(mf2.ptr);
3606                 xdiff_clear_find_func(&xecfg);
3607         }
3608
3609  free_ab_and_return:
3610         strbuf_release(&header);
3611         diff_free_filespec_data(one);
3612         diff_free_filespec_data(two);
3613         free(a_one);
3614         free(b_two);
3615         return;
3616 }
3617
3618 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3619 {
3620         if (!is_renamed) {
3621                 if (p->status == DIFF_STATUS_ADDED) {
3622                         if (S_ISLNK(p->two->mode))
3623                                 return "new +l";
3624                         else if ((p->two->mode & 0777) == 0755)
3625                                 return "new +x";
3626                         else
3627                                 return "new";
3628                 } else if (p->status == DIFF_STATUS_DELETED)
3629                         return "gone";
3630         }
3631         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3632                 return "mode -l";
3633         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3634                 return "mode +l";
3635         else if ((p->one->mode & 0777) == 0644 &&
3636                  (p->two->mode & 0777) == 0755)
3637                 return "mode +x";
3638         else if ((p->one->mode & 0777) == 0755 &&
3639                  (p->two->mode & 0777) == 0644)
3640                 return "mode -x";
3641         return NULL;
3642 }
3643
3644 static void builtin_diffstat(const char *name_a, const char *name_b,
3645                              struct diff_filespec *one,
3646                              struct diff_filespec *two,
3647                              struct diffstat_t *diffstat,
3648                              struct diff_options *o,
3649                              struct diff_filepair *p)
3650 {
3651         mmfile_t mf1, mf2;
3652         struct diffstat_file *data;
3653         int same_contents;
3654         int complete_rewrite = 0;
3655
3656         if (!DIFF_PAIR_UNMERGED(p)) {
3657                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3658                         complete_rewrite = 1;
3659         }
3660
3661         data = diffstat_add(diffstat, name_a, name_b);
3662         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3663         if (o->flags.stat_with_summary)
3664                 data->comments = get_compact_summary(p, data->is_renamed);
3665
3666         if (!one || !two) {
3667                 data->is_unmerged = 1;
3668                 return;
3669         }
3670
3671         same_contents = oideq(&one->oid, &two->oid);
3672
3673         if (diff_filespec_is_binary(o->repo, one) ||
3674             diff_filespec_is_binary(o->repo, two)) {
3675                 data->is_binary = 1;
3676                 if (same_contents) {
3677                         data->added = 0;
3678                         data->deleted = 0;
3679                 } else {
3680                         data->added = diff_filespec_size(o->repo, two);
3681                         data->deleted = diff_filespec_size(o->repo, one);
3682                 }
3683         }
3684
3685         else if (complete_rewrite) {
3686                 diff_populate_filespec(o->repo, one, 0);
3687                 diff_populate_filespec(o->repo, two, 0);
3688                 data->deleted = count_lines(one->data, one->size);
3689                 data->added = count_lines(two->data, two->size);
3690         }
3691
3692         else if (!same_contents) {
3693                 /* Crazy xdl interfaces.. */
3694                 xpparam_t xpp;
3695                 xdemitconf_t xecfg;
3696
3697                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3698                     fill_mmfile(o->repo, &mf2, two) < 0)
3699                         die("unable to read files to diff");
3700
3701                 memset(&xpp, 0, sizeof(xpp));
3702                 memset(&xecfg, 0, sizeof(xecfg));
3703                 xpp.flags = o->xdl_opts;
3704                 xpp.anchors = o->anchors;
3705                 xpp.anchors_nr = o->anchors_nr;
3706                 xecfg.ctxlen = o->context;
3707                 xecfg.interhunkctxlen = o->interhunkcontext;
3708                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3709                                   diffstat_consume, diffstat, &xpp, &xecfg))
3710                         die("unable to generate diffstat for %s", one->path);
3711         }
3712
3713         diff_free_filespec_data(one);
3714         diff_free_filespec_data(two);
3715 }
3716
3717 static void builtin_checkdiff(const char *name_a, const char *name_b,
3718                               const char *attr_path,
3719                               struct diff_filespec *one,
3720                               struct diff_filespec *two,
3721                               struct diff_options *o)
3722 {
3723         mmfile_t mf1, mf2;
3724         struct checkdiff_t data;
3725
3726         if (!two)
3727                 return;
3728
3729         memset(&data, 0, sizeof(data));
3730         data.filename = name_b ? name_b : name_a;
3731         data.lineno = 0;
3732         data.o = o;
3733         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3734         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3735
3736         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3737             fill_mmfile(o->repo, &mf2, two) < 0)
3738                 die("unable to read files to diff");
3739
3740         /*
3741          * All the other codepaths check both sides, but not checking
3742          * the "old" side here is deliberate.  We are checking the newly
3743          * introduced changes, and as long as the "new" side is text, we
3744          * can and should check what it introduces.
3745          */
3746         if (diff_filespec_is_binary(o->repo, two))
3747                 goto free_and_return;
3748         else {
3749                 /* Crazy xdl interfaces.. */
3750                 xpparam_t xpp;
3751                 xdemitconf_t xecfg;
3752
3753                 memset(&xpp, 0, sizeof(xpp));
3754                 memset(&xecfg, 0, sizeof(xecfg));
3755                 xecfg.ctxlen = 1; /* at least one context line */
3756                 xpp.flags = 0;
3757                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3758                                   checkdiff_consume, &data,
3759                                   &xpp, &xecfg))
3760                         die("unable to generate checkdiff for %s", one->path);
3761
3762                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3763                         struct emit_callback ecbdata;
3764                         int blank_at_eof;
3765
3766                         ecbdata.ws_rule = data.ws_rule;
3767                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3768                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3769
3770                         if (blank_at_eof) {
3771                                 static char *err;
3772                                 if (!err)
3773                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3774                                 fprintf(o->file, "%s:%d: %s.\n",
3775                                         data.filename, blank_at_eof, err);
3776                                 data.status = 1; /* report errors */
3777                         }
3778                 }
3779         }
3780  free_and_return:
3781         diff_free_filespec_data(one);
3782         diff_free_filespec_data(two);
3783         if (data.status)
3784                 o->flags.check_failed = 1;
3785 }
3786
3787 struct diff_filespec *alloc_filespec(const char *path)
3788 {
3789         struct diff_filespec *spec;
3790
3791         FLEXPTR_ALLOC_STR(spec, path, path);
3792         spec->count = 1;
3793         spec->is_binary = -1;
3794         return spec;
3795 }
3796
3797 void free_filespec(struct diff_filespec *spec)
3798 {
3799         if (!--spec->count) {
3800                 diff_free_filespec_data(spec);
3801                 free(spec);
3802         }
3803 }
3804
3805 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3806                    int oid_valid, unsigned short mode)
3807 {
3808         if (mode) {
3809                 spec->mode = canon_mode(mode);
3810                 oidcpy(&spec->oid, oid);
3811                 spec->oid_valid = oid_valid;
3812         }
3813 }
3814
3815 /*
3816  * Given a name and sha1 pair, if the index tells us the file in
3817  * the work tree has that object contents, return true, so that
3818  * prepare_temp_file() does not have to inflate and extract.
3819  */
3820 static int reuse_worktree_file(struct index_state *istate,
3821                                const char *name,
3822                                const struct object_id *oid,
3823                                int want_file)
3824 {
3825         const struct cache_entry *ce;
3826         struct stat st;
3827         int pos, len;
3828
3829         /*
3830          * We do not read the cache ourselves here, because the
3831          * benchmark with my previous version that always reads cache
3832          * shows that it makes things worse for diff-tree comparing
3833          * two linux-2.6 kernel trees in an already checked out work
3834          * tree.  This is because most diff-tree comparisons deal with
3835          * only a small number of files, while reading the cache is
3836          * expensive for a large project, and its cost outweighs the
3837          * savings we get by not inflating the object to a temporary
3838          * file.  Practically, this code only helps when we are used
3839          * by diff-cache --cached, which does read the cache before
3840          * calling us.
3841          */
3842         if (!istate->cache)
3843                 return 0;
3844
3845         /* We want to avoid the working directory if our caller
3846          * doesn't need the data in a normal file, this system
3847          * is rather slow with its stat/open/mmap/close syscalls,
3848          * and the object is contained in a pack file.  The pack
3849          * is probably already open and will be faster to obtain
3850          * the data through than the working directory.  Loose
3851          * objects however would tend to be slower as they need
3852          * to be individually opened and inflated.
3853          */
3854         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3855                 return 0;
3856
3857         /*
3858          * Similarly, if we'd have to convert the file contents anyway, that
3859          * makes the optimization not worthwhile.
3860          */
3861         if (!want_file && would_convert_to_git(istate, name))
3862                 return 0;
3863
3864         len = strlen(name);
3865         pos = index_name_pos(istate, name, len);
3866         if (pos < 0)
3867                 return 0;
3868         ce = istate->cache[pos];
3869
3870         /*
3871          * This is not the sha1 we are looking for, or
3872          * unreusable because it is not a regular file.
3873          */
3874         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3875                 return 0;
3876
3877         /*
3878          * If ce is marked as "assume unchanged", there is no
3879          * guarantee that work tree matches what we are looking for.
3880          */
3881         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3882                 return 0;
3883
3884         /*
3885          * If ce matches the file in the work tree, we can reuse it.
3886          */
3887         if (ce_uptodate(ce) ||
3888             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3889                 return 1;
3890
3891         return 0;
3892 }
3893
3894 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3895 {
3896         struct strbuf buf = STRBUF_INIT;
3897         char *dirty = "";
3898
3899         /* Are we looking at the work tree? */
3900         if (s->dirty_submodule)
3901                 dirty = "-dirty";
3902
3903         strbuf_addf(&buf, "Subproject commit %s%s\n",
3904                     oid_to_hex(&s->oid), dirty);
3905         s->size = buf.len;
3906         if (size_only) {
3907                 s->data = NULL;
3908                 strbuf_release(&buf);
3909         } else {
3910                 s->data = strbuf_detach(&buf, NULL);
3911                 s->should_free = 1;
3912         }
3913         return 0;
3914 }
3915
3916 /*
3917  * While doing rename detection and pickaxe operation, we may need to
3918  * grab the data for the blob (or file) for our own in-core comparison.
3919  * diff_filespec has data and size fields for this purpose.
3920  */
3921 int diff_populate_filespec(struct repository *r,
3922                            struct diff_filespec *s,
3923                            unsigned int flags)
3924 {
3925         int size_only = flags & CHECK_SIZE_ONLY;
3926         int err = 0;
3927         int conv_flags = global_conv_flags_eol;
3928         /*
3929          * demote FAIL to WARN to allow inspecting the situation
3930          * instead of refusing.
3931          */
3932         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3933                 conv_flags = CONV_EOL_RNDTRP_WARN;
3934
3935         if (!DIFF_FILE_VALID(s))
3936                 die("internal error: asking to populate invalid file.");
3937         if (S_ISDIR(s->mode))
3938                 return -1;
3939
3940         if (s->data)
3941                 return 0;
3942
3943         if (size_only && 0 < s->size)
3944                 return 0;
3945
3946         if (S_ISGITLINK(s->mode))
3947                 return diff_populate_gitlink(s, size_only);
3948
3949         if (!s->oid_valid ||
3950             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3951                 struct strbuf buf = STRBUF_INIT;
3952                 struct stat st;
3953                 int fd;
3954
3955                 if (lstat(s->path, &st) < 0) {
3956                 err_empty:
3957                         err = -1;
3958                 empty:
3959                         s->data = (char *)"";
3960                         s->size = 0;
3961                         return err;
3962                 }
3963                 s->size = xsize_t(st.st_size);
3964                 if (!s->size)
3965                         goto empty;
3966                 if (S_ISLNK(st.st_mode)) {
3967                         struct strbuf sb = STRBUF_INIT;
3968
3969                         if (strbuf_readlink(&sb, s->path, s->size))
3970                                 goto err_empty;
3971                         s->size = sb.len;
3972                         s->data = strbuf_detach(&sb, NULL);
3973                         s->should_free = 1;
3974                         return 0;
3975                 }
3976
3977                 /*
3978                  * Even if the caller would be happy with getting
3979                  * only the size, we cannot return early at this
3980                  * point if the path requires us to run the content
3981                  * conversion.
3982                  */
3983                 if (size_only && !would_convert_to_git(r->index, s->path))
3984                         return 0;
3985
3986                 /*
3987                  * Note: this check uses xsize_t(st.st_size) that may
3988                  * not be the true size of the blob after it goes
3989                  * through convert_to_git().  This may not strictly be
3990                  * correct, but the whole point of big_file_threshold
3991                  * and is_binary check being that we want to avoid
3992                  * opening the file and inspecting the contents, this
3993                  * is probably fine.
3994                  */
3995                 if ((flags & CHECK_BINARY) &&
3996                     s->size > big_file_threshold && s->is_binary == -1) {
3997                         s->is_binary = 1;
3998                         return 0;
3999                 }
4000                 fd = open(s->path, O_RDONLY);
4001                 if (fd < 0)
4002                         goto err_empty;
4003                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4004                 close(fd);
4005                 s->should_munmap = 1;
4006
4007                 /*
4008                  * Convert from working tree format to canonical git format
4009                  */
4010                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4011                         size_t size = 0;
4012                         munmap(s->data, s->size);
4013                         s->should_munmap = 0;
4014                         s->data = strbuf_detach(&buf, &size);
4015                         s->size = size;
4016                         s->should_free = 1;
4017                 }
4018         }
4019         else {
4020                 enum object_type type;
4021                 if (size_only || (flags & CHECK_BINARY)) {
4022                         type = oid_object_info(r, &s->oid, &s->size);
4023                         if (type < 0)
4024                                 die("unable to read %s",
4025                                     oid_to_hex(&s->oid));
4026                         if (size_only)
4027                                 return 0;
4028                         if (s->size > big_file_threshold && s->is_binary == -1) {
4029                                 s->is_binary = 1;
4030                                 return 0;
4031                         }
4032                 }
4033                 s->data = read_object_file(&s->oid, &type, &s->size);
4034                 if (!s->data)
4035                         die("unable to read %s", oid_to_hex(&s->oid));
4036                 s->should_free = 1;
4037         }
4038         return 0;
4039 }
4040
4041 void diff_free_filespec_blob(struct diff_filespec *s)
4042 {
4043         if (s->should_free)
4044                 free(s->data);
4045         else if (s->should_munmap)
4046                 munmap(s->data, s->size);
4047
4048         if (s->should_free || s->should_munmap) {
4049                 s->should_free = s->should_munmap = 0;
4050                 s->data = NULL;
4051         }
4052 }
4053
4054 void diff_free_filespec_data(struct diff_filespec *s)
4055 {
4056         diff_free_filespec_blob(s);
4057         FREE_AND_NULL(s->cnt_data);
4058 }
4059
4060 static void prep_temp_blob(struct index_state *istate,
4061                            const char *path, struct diff_tempfile *temp,
4062                            void *blob,
4063                            unsigned long size,
4064                            const struct object_id *oid,
4065                            int mode)
4066 {
4067         struct strbuf buf = STRBUF_INIT;
4068         struct strbuf tempfile = STRBUF_INIT;
4069         char *path_dup = xstrdup(path);
4070         const char *base = basename(path_dup);
4071
4072         /* Generate "XXXXXX_basename.ext" */
4073         strbuf_addstr(&tempfile, "XXXXXX_");
4074         strbuf_addstr(&tempfile, base);
4075
4076         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4077         if (!temp->tempfile)
4078                 die_errno("unable to create temp-file");
4079         if (convert_to_working_tree(istate, path,
4080                         (const char *)blob, (size_t)size, &buf)) {
4081                 blob = buf.buf;
4082                 size = buf.len;
4083         }
4084         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4085             close_tempfile_gently(temp->tempfile))
4086                 die_errno("unable to write temp-file");
4087         temp->name = get_tempfile_path(temp->tempfile);
4088         oid_to_hex_r(temp->hex, oid);
4089         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4090         strbuf_release(&buf);
4091         strbuf_release(&tempfile);
4092         free(path_dup);
4093 }
4094
4095 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4096                                                const char *name,
4097                                                struct diff_filespec *one)
4098 {
4099         struct diff_tempfile *temp = claim_diff_tempfile();
4100
4101         if (!DIFF_FILE_VALID(one)) {
4102         not_a_valid_file:
4103                 /* A '-' entry produces this for file-2, and
4104                  * a '+' entry produces this for file-1.
4105                  */
4106                 temp->name = "/dev/null";
4107                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4108                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4109                 return temp;
4110         }
4111
4112         if (!S_ISGITLINK(one->mode) &&
4113             (!one->oid_valid ||
4114              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4115                 struct stat st;
4116                 if (lstat(name, &st) < 0) {
4117                         if (errno == ENOENT)
4118                                 goto not_a_valid_file;
4119                         die_errno("stat(%s)", name);
4120                 }
4121                 if (S_ISLNK(st.st_mode)) {
4122                         struct strbuf sb = STRBUF_INIT;
4123                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4124                                 die_errno("readlink(%s)", name);
4125                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4126                                        (one->oid_valid ?
4127                                         &one->oid : &null_oid),
4128                                        (one->oid_valid ?
4129                                         one->mode : S_IFLNK));
4130                         strbuf_release(&sb);
4131                 }
4132                 else {
4133                         /* we can borrow from the file in the work tree */
4134                         temp->name = name;
4135                         if (!one->oid_valid)
4136                                 oid_to_hex_r(temp->hex, &null_oid);
4137                         else
4138                                 oid_to_hex_r(temp->hex, &one->oid);
4139                         /* Even though we may sometimes borrow the
4140                          * contents from the work tree, we always want
4141                          * one->mode.  mode is trustworthy even when
4142                          * !(one->oid_valid), as long as
4143                          * DIFF_FILE_VALID(one).
4144                          */
4145                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4146                 }
4147                 return temp;
4148         }
4149         else {
4150                 if (diff_populate_filespec(r, one, 0))
4151                         die("cannot read data blob for %s", one->path);
4152                 prep_temp_blob(r->index, name, temp,
4153                                one->data, one->size,
4154                                &one->oid, one->mode);
4155         }
4156         return temp;
4157 }
4158
4159 static void add_external_diff_name(struct repository *r,
4160                                    struct argv_array *argv,
4161                                    const char *name,
4162                                    struct diff_filespec *df)
4163 {
4164         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4165         argv_array_push(argv, temp->name);
4166         argv_array_push(argv, temp->hex);
4167         argv_array_push(argv, temp->mode);
4168 }
4169
4170 /* An external diff command takes:
4171  *
4172  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4173  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4174  *
4175  */
4176 static void run_external_diff(const char *pgm,
4177                               const char *name,
4178                               const char *other,
4179                               struct diff_filespec *one,
4180                               struct diff_filespec *two,
4181                               const char *xfrm_msg,
4182                               struct diff_options *o)
4183 {
4184         struct argv_array argv = ARGV_ARRAY_INIT;
4185         struct argv_array env = ARGV_ARRAY_INIT;
4186         struct diff_queue_struct *q = &diff_queued_diff;
4187
4188         argv_array_push(&argv, pgm);
4189         argv_array_push(&argv, name);
4190
4191         if (one && two) {
4192                 add_external_diff_name(o->repo, &argv, name, one);
4193                 if (!other)
4194                         add_external_diff_name(o->repo, &argv, name, two);
4195                 else {
4196                         add_external_diff_name(o->repo, &argv, other, two);
4197                         argv_array_push(&argv, other);
4198                         argv_array_push(&argv, xfrm_msg);
4199                 }
4200         }
4201
4202         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4203         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4204
4205         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4206                 die(_("external diff died, stopping at %s"), name);
4207
4208         remove_tempfile();
4209         argv_array_clear(&argv);
4210         argv_array_clear(&env);
4211 }
4212
4213 static int similarity_index(struct diff_filepair *p)
4214 {
4215         return p->score * 100 / MAX_SCORE;
4216 }
4217
4218 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4219 {
4220         if (startup_info->have_repository)
4221                 return find_unique_abbrev(oid, abbrev);
4222         else {
4223                 char *hex = oid_to_hex(oid);
4224                 if (abbrev < 0)
4225                         abbrev = FALLBACK_DEFAULT_ABBREV;
4226                 if (abbrev > the_hash_algo->hexsz)
4227                         BUG("oid abbreviation out of range: %d", abbrev);
4228                 if (abbrev)
4229                         hex[abbrev] = '\0';
4230                 return hex;
4231         }
4232 }
4233
4234 static void fill_metainfo(struct strbuf *msg,
4235                           const char *name,
4236                           const char *other,
4237                           struct diff_filespec *one,
4238                           struct diff_filespec *two,
4239                           struct diff_options *o,
4240                           struct diff_filepair *p,
4241                           int *must_show_header,
4242                           int use_color)
4243 {
4244         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4245         const char *reset = diff_get_color(use_color, DIFF_RESET);
4246         const char *line_prefix = diff_line_prefix(o);
4247
4248         *must_show_header = 1;
4249         strbuf_init(msg, PATH_MAX * 2 + 300);
4250         switch (p->status) {
4251         case DIFF_STATUS_COPIED:
4252                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4253                             line_prefix, set, similarity_index(p));
4254                 strbuf_addf(msg, "%s\n%s%scopy from ",
4255                             reset,  line_prefix, set);
4256                 quote_c_style(name, msg, NULL, 0);
4257                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4258                 quote_c_style(other, msg, NULL, 0);
4259                 strbuf_addf(msg, "%s\n", reset);
4260                 break;
4261         case DIFF_STATUS_RENAMED:
4262                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4263                             line_prefix, set, similarity_index(p));
4264                 strbuf_addf(msg, "%s\n%s%srename from ",
4265                             reset, line_prefix, set);
4266                 quote_c_style(name, msg, NULL, 0);
4267                 strbuf_addf(msg, "%s\n%s%srename to ",
4268                             reset, line_prefix, set);
4269                 quote_c_style(other, msg, NULL, 0);
4270                 strbuf_addf(msg, "%s\n", reset);
4271                 break;
4272         case DIFF_STATUS_MODIFIED:
4273                 if (p->score) {
4274                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4275                                     line_prefix,
4276                                     set, similarity_index(p), reset);
4277                         break;
4278                 }
4279                 /* fallthru */
4280         default:
4281                 *must_show_header = 0;
4282         }
4283         if (one && two && !oideq(&one->oid, &two->oid)) {
4284                 const unsigned hexsz = the_hash_algo->hexsz;
4285                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4286
4287                 if (o->flags.binary) {
4288                         mmfile_t mf;
4289                         if ((!fill_mmfile(o->repo, &mf, one) &&
4290                              diff_filespec_is_binary(o->repo, one)) ||
4291                             (!fill_mmfile(o->repo, &mf, two) &&
4292                              diff_filespec_is_binary(o->repo, two)))
4293                                 abbrev = hexsz;
4294                 }
4295                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4296                             diff_abbrev_oid(&one->oid, abbrev),
4297                             diff_abbrev_oid(&two->oid, abbrev));
4298                 if (one->mode == two->mode)
4299                         strbuf_addf(msg, " %06o", one->mode);
4300                 strbuf_addf(msg, "%s\n", reset);
4301         }
4302 }
4303
4304 static void run_diff_cmd(const char *pgm,
4305                          const char *name,
4306                          const char *other,
4307                          const char *attr_path,
4308                          struct diff_filespec *one,
4309                          struct diff_filespec *two,
4310                          struct strbuf *msg,
4311                          struct diff_options *o,
4312                          struct diff_filepair *p)
4313 {
4314         const char *xfrm_msg = NULL;
4315         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4316         int must_show_header = 0;
4317
4318
4319         if (o->flags.allow_external) {
4320                 struct userdiff_driver *drv;
4321
4322                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4323                 if (drv && drv->external)
4324                         pgm = drv->external;
4325         }
4326
4327         if (msg) {
4328                 /*
4329                  * don't use colors when the header is intended for an
4330                  * external diff driver
4331                  */
4332                 fill_metainfo(msg, name, other, one, two, o, p,
4333                               &must_show_header,
4334                               want_color(o->use_color) && !pgm);
4335                 xfrm_msg = msg->len ? msg->buf : NULL;
4336         }
4337
4338         if (pgm) {
4339                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4340                 return;
4341         }
4342         if (one && two)
4343                 builtin_diff(name, other ? other : name,
4344                              one, two, xfrm_msg, must_show_header,
4345                              o, complete_rewrite);
4346         else
4347                 fprintf(o->file, "* Unmerged path %s\n", name);
4348 }
4349
4350 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4351 {
4352         if (DIFF_FILE_VALID(one)) {
4353                 if (!one->oid_valid) {
4354                         struct stat st;
4355                         if (one->is_stdin) {
4356                                 oidclr(&one->oid);
4357                                 return;
4358                         }
4359                         if (lstat(one->path, &st) < 0)
4360                                 die_errno("stat '%s'", one->path);
4361                         if (index_path(istate, &one->oid, one->path, &st, 0))
4362                                 die("cannot hash %s", one->path);
4363                 }
4364         }
4365         else
4366                 oidclr(&one->oid);
4367 }
4368
4369 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4370 {
4371         /* Strip the prefix but do not molest /dev/null and absolute paths */
4372         if (*namep && !is_absolute_path(*namep)) {
4373                 *namep += prefix_length;
4374                 if (**namep == '/')
4375                         ++*namep;
4376         }
4377         if (*otherp && !is_absolute_path(*otherp)) {
4378                 *otherp += prefix_length;
4379                 if (**otherp == '/')
4380                         ++*otherp;
4381         }
4382 }
4383
4384 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4385 {
4386         const char *pgm = external_diff();
4387         struct strbuf msg;
4388         struct diff_filespec *one = p->one;
4389         struct diff_filespec *two = p->two;
4390         const char *name;
4391         const char *other;
4392         const char *attr_path;
4393
4394         name  = one->path;
4395         other = (strcmp(name, two->path) ? two->path : NULL);
4396         attr_path = name;
4397         if (o->prefix_length)
4398                 strip_prefix(o->prefix_length, &name, &other);
4399
4400         if (!o->flags.allow_external)
4401                 pgm = NULL;
4402
4403         if (DIFF_PAIR_UNMERGED(p)) {
4404                 run_diff_cmd(pgm, name, NULL, attr_path,
4405                              NULL, NULL, NULL, o, p);
4406                 return;
4407         }
4408
4409         diff_fill_oid_info(one, o->repo->index);
4410         diff_fill_oid_info(two, o->repo->index);
4411
4412         if (!pgm &&
4413             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4414             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4415                 /*
4416                  * a filepair that changes between file and symlink
4417                  * needs to be split into deletion and creation.
4418                  */
4419                 struct diff_filespec *null = alloc_filespec(two->path);
4420                 run_diff_cmd(NULL, name, other, attr_path,
4421                              one, null, &msg,
4422                              o, p);
4423                 free(null);
4424                 strbuf_release(&msg);
4425
4426                 null = alloc_filespec(one->path);
4427                 run_diff_cmd(NULL, name, other, attr_path,
4428                              null, two, &msg, o, p);
4429                 free(null);
4430         }
4431         else
4432                 run_diff_cmd(pgm, name, other, attr_path,
4433                              one, two, &msg, o, p);
4434
4435         strbuf_release(&msg);
4436 }
4437
4438 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4439                          struct diffstat_t *diffstat)
4440 {
4441         const char *name;
4442         const char *other;
4443
4444         if (DIFF_PAIR_UNMERGED(p)) {
4445                 /* unmerged */
4446                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4447                                  diffstat, o, p);
4448                 return;
4449         }
4450
4451         name = p->one->path;
4452         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4453
4454         if (o->prefix_length)
4455                 strip_prefix(o->prefix_length, &name, &other);
4456
4457         diff_fill_oid_info(p->one, o->repo->index);
4458         diff_fill_oid_info(p->two, o->repo->index);
4459
4460         builtin_diffstat(name, other, p->one, p->two,
4461                          diffstat, o, p);
4462 }
4463
4464 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4465 {
4466         const char *name;
4467         const char *other;
4468         const char *attr_path;
4469
4470         if (DIFF_PAIR_UNMERGED(p)) {
4471                 /* unmerged */
4472                 return;
4473         }
4474
4475         name = p->one->path;
4476         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4477         attr_path = other ? other : name;
4478
4479         if (o->prefix_length)
4480                 strip_prefix(o->prefix_length, &name, &other);
4481
4482         diff_fill_oid_info(p->one, o->repo->index);
4483         diff_fill_oid_info(p->two, o->repo->index);
4484
4485         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4486 }
4487
4488 static void prep_parse_options(struct diff_options *options);
4489
4490 void repo_diff_setup(struct repository *r, struct diff_options *options)
4491 {
4492         memcpy(options, &default_diff_options, sizeof(*options));
4493
4494         options->file = stdout;
4495         options->repo = r;
4496
4497         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4498         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4499         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4500         options->abbrev = DEFAULT_ABBREV;
4501         options->line_termination = '\n';
4502         options->break_opt = -1;
4503         options->rename_limit = -1;
4504         options->dirstat_permille = diff_dirstat_permille_default;
4505         options->context = diff_context_default;
4506         options->interhunkcontext = diff_interhunk_context_default;
4507         options->ws_error_highlight = ws_error_highlight_default;
4508         options->flags.rename_empty = 1;
4509         options->objfind = NULL;
4510
4511         /* pathchange left =NULL by default */
4512         options->change = diff_change;
4513         options->add_remove = diff_addremove;
4514         options->use_color = diff_use_color_default;
4515         options->detect_rename = diff_detect_rename_default;
4516         options->xdl_opts |= diff_algorithm;
4517         if (diff_indent_heuristic)
4518                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4519
4520         options->orderfile = diff_order_file_cfg;
4521
4522         if (diff_no_prefix) {
4523                 options->a_prefix = options->b_prefix = "";
4524         } else if (!diff_mnemonic_prefix) {
4525                 options->a_prefix = "a/";
4526                 options->b_prefix = "b/";
4527         }
4528
4529         options->color_moved = diff_color_moved_default;
4530         options->color_moved_ws_handling = diff_color_moved_ws_default;
4531
4532         prep_parse_options(options);
4533 }
4534
4535 void diff_setup_done(struct diff_options *options)
4536 {
4537         unsigned check_mask = DIFF_FORMAT_NAME |
4538                               DIFF_FORMAT_NAME_STATUS |
4539                               DIFF_FORMAT_CHECKDIFF |
4540                               DIFF_FORMAT_NO_OUTPUT;
4541         /*
4542          * This must be signed because we're comparing against a potentially
4543          * negative value.
4544          */
4545         const int hexsz = the_hash_algo->hexsz;
4546
4547         if (options->set_default)
4548                 options->set_default(options);
4549
4550         if (HAS_MULTI_BITS(options->output_format & check_mask))
4551                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4552
4553         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4554                 die(_("-G, -S and --find-object are mutually exclusive"));
4555
4556         /*
4557          * Most of the time we can say "there are changes"
4558          * only by checking if there are changed paths, but
4559          * --ignore-whitespace* options force us to look
4560          * inside contents.
4561          */
4562
4563         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4564                 options->flags.diff_from_contents = 1;
4565         else
4566                 options->flags.diff_from_contents = 0;
4567
4568         if (options->flags.find_copies_harder)
4569                 options->detect_rename = DIFF_DETECT_COPY;
4570
4571         if (!options->flags.relative_name)
4572                 options->prefix = NULL;
4573         if (options->prefix)
4574                 options->prefix_length = strlen(options->prefix);
4575         else
4576                 options->prefix_length = 0;
4577
4578         if (options->output_format & (DIFF_FORMAT_NAME |
4579                                       DIFF_FORMAT_NAME_STATUS |
4580                                       DIFF_FORMAT_CHECKDIFF |
4581                                       DIFF_FORMAT_NO_OUTPUT))
4582                 options->output_format &= ~(DIFF_FORMAT_RAW |
4583                                             DIFF_FORMAT_NUMSTAT |
4584                                             DIFF_FORMAT_DIFFSTAT |
4585                                             DIFF_FORMAT_SHORTSTAT |
4586                                             DIFF_FORMAT_DIRSTAT |
4587                                             DIFF_FORMAT_SUMMARY |
4588                                             DIFF_FORMAT_PATCH);
4589
4590         /*
4591          * These cases always need recursive; we do not drop caller-supplied
4592          * recursive bits for other formats here.
4593          */
4594         if (options->output_format & (DIFF_FORMAT_PATCH |
4595                                       DIFF_FORMAT_NUMSTAT |
4596                                       DIFF_FORMAT_DIFFSTAT |
4597                                       DIFF_FORMAT_SHORTSTAT |
4598                                       DIFF_FORMAT_DIRSTAT |
4599                                       DIFF_FORMAT_SUMMARY |
4600                                       DIFF_FORMAT_CHECKDIFF))
4601                 options->flags.recursive = 1;
4602         /*
4603          * Also pickaxe would not work very well if you do not say recursive
4604          */
4605         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4606                 options->flags.recursive = 1;
4607         /*
4608          * When patches are generated, submodules diffed against the work tree
4609          * must be checked for dirtiness too so it can be shown in the output
4610          */
4611         if (options->output_format & DIFF_FORMAT_PATCH)
4612                 options->flags.dirty_submodules = 1;
4613
4614         if (options->detect_rename && options->rename_limit < 0)
4615                 options->rename_limit = diff_rename_limit_default;
4616         if (hexsz < options->abbrev)
4617                 options->abbrev = hexsz; /* full */
4618
4619         /*
4620          * It does not make sense to show the first hit we happened
4621          * to have found.  It does not make sense not to return with
4622          * exit code in such a case either.
4623          */
4624         if (options->flags.quick) {
4625                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4626                 options->flags.exit_with_status = 1;
4627         }
4628
4629         options->diff_path_counter = 0;
4630
4631         if (options->flags.follow_renames && options->pathspec.nr != 1)
4632                 die(_("--follow requires exactly one pathspec"));
4633
4634         if (!options->use_color || external_diff())
4635                 options->color_moved = 0;
4636
4637         FREE_AND_NULL(options->parseopts);
4638 }
4639
4640 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4641 {
4642         char c, *eq;
4643         int len;
4644
4645         if (*arg != '-')
4646                 return 0;
4647         c = *++arg;
4648         if (!c)
4649                 return 0;
4650         if (c == arg_short) {
4651                 c = *++arg;
4652                 if (!c)
4653                         return 1;
4654                 if (val && isdigit(c)) {
4655                         char *end;
4656                         int n = strtoul(arg, &end, 10);
4657                         if (*end)
4658                                 return 0;
4659                         *val = n;
4660                         return 1;
4661                 }
4662                 return 0;
4663         }
4664         if (c != '-')
4665                 return 0;
4666         arg++;
4667         eq = strchrnul(arg, '=');
4668         len = eq - arg;
4669         if (!len || strncmp(arg, arg_long, len))
4670                 return 0;
4671         if (*eq) {
4672                 int n;
4673                 char *end;
4674                 if (!isdigit(*++eq))
4675                         return 0;
4676                 n = strtoul(eq, &end, 10);
4677                 if (*end)
4678                         return 0;
4679                 *val = n;
4680         }
4681         return 1;
4682 }
4683
4684 static int diff_scoreopt_parse(const char *opt);
4685
4686 static inline int short_opt(char opt, const char **argv,
4687                             const char **optarg)
4688 {
4689         const char *arg = argv[0];
4690         if (arg[0] != '-' || arg[1] != opt)
4691                 return 0;
4692         if (arg[2] != '\0') {
4693                 *optarg = arg + 2;
4694                 return 1;
4695         }
4696         if (!argv[1])
4697                 die("Option '%c' requires a value", opt);
4698         *optarg = argv[1];
4699         return 2;
4700 }
4701
4702 int parse_long_opt(const char *opt, const char **argv,
4703                    const char **optarg)
4704 {
4705         const char *arg = argv[0];
4706         if (!skip_prefix(arg, "--", &arg))
4707                 return 0;
4708         if (!skip_prefix(arg, opt, &arg))
4709                 return 0;
4710         if (*arg == '=') { /* stuck form: --option=value */
4711                 *optarg = arg + 1;
4712                 return 1;
4713         }
4714         if (*arg != '\0')
4715                 return 0;
4716         /* separate form: --option value */
4717         if (!argv[1])
4718                 die("Option '--%s' requires a value", opt);
4719         *optarg = argv[1];
4720         return 2;
4721 }
4722
4723 static int stat_opt(struct diff_options *options, const char **av)
4724 {
4725         const char *arg = av[0];
4726         char *end;
4727         int width = options->stat_width;
4728         int name_width = options->stat_name_width;
4729         int graph_width = options->stat_graph_width;
4730         int count = options->stat_count;
4731         int argcount = 1;
4732
4733         if (!skip_prefix(arg, "--stat", &arg))
4734                 BUG("stat option does not begin with --stat: %s", arg);
4735         end = (char *)arg;
4736
4737         switch (*arg) {
4738         case '-':
4739                 if (skip_prefix(arg, "-width", &arg)) {
4740                         if (*arg == '=')
4741                                 width = strtoul(arg + 1, &end, 10);
4742                         else if (!*arg && !av[1])
4743                                 die_want_option("--stat-width");
4744                         else if (!*arg) {
4745                                 width = strtoul(av[1], &end, 10);
4746                                 argcount = 2;
4747                         }
4748                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4749                         if (*arg == '=')
4750                                 name_width = strtoul(arg + 1, &end, 10);
4751                         else if (!*arg && !av[1])
4752                                 die_want_option("--stat-name-width");
4753                         else if (!*arg) {
4754                                 name_width = strtoul(av[1], &end, 10);
4755                                 argcount = 2;
4756                         }
4757                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4758                         if (*arg == '=')
4759                                 graph_width = strtoul(arg + 1, &end, 10);
4760                         else if (!*arg && !av[1])
4761                                 die_want_option("--stat-graph-width");
4762                         else if (!*arg) {
4763                                 graph_width = strtoul(av[1], &end, 10);
4764                                 argcount = 2;
4765                         }
4766                 } else if (skip_prefix(arg, "-count", &arg)) {
4767                         if (*arg == '=')
4768                                 count = strtoul(arg + 1, &end, 10);
4769                         else if (!*arg && !av[1])
4770                                 die_want_option("--stat-count");
4771                         else if (!*arg) {
4772                                 count = strtoul(av[1], &end, 10);
4773                                 argcount = 2;
4774                         }
4775                 }
4776                 break;
4777         case '=':
4778                 width = strtoul(arg+1, &end, 10);
4779                 if (*end == ',')
4780                         name_width = strtoul(end+1, &end, 10);
4781                 if (*end == ',')
4782                         count = strtoul(end+1, &end, 10);
4783         }
4784
4785         /* Important! This checks all the error cases! */
4786         if (*end)
4787                 return 0;
4788         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4789         options->stat_name_width = name_width;
4790         options->stat_graph_width = graph_width;
4791         options->stat_width = width;
4792         options->stat_count = count;
4793         return argcount;
4794 }
4795
4796 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4797 {
4798         struct strbuf errmsg = STRBUF_INIT;
4799         if (parse_dirstat_params(options, params, &errmsg))
4800                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4801                     errmsg.buf);
4802         strbuf_release(&errmsg);
4803         /*
4804          * The caller knows a dirstat-related option is given from the command
4805          * line; allow it to say "return this_function();"
4806          */
4807         options->output_format |= DIFF_FORMAT_DIRSTAT;
4808         return 1;
4809 }
4810
4811 static int parse_submodule_opt(struct diff_options *options, const char *value)
4812 {
4813         if (parse_submodule_params(options, value))
4814                 die(_("Failed to parse --submodule option parameter: '%s'"),
4815                         value);
4816         return 1;
4817 }
4818
4819 static const char diff_status_letters[] = {
4820         DIFF_STATUS_ADDED,
4821         DIFF_STATUS_COPIED,
4822         DIFF_STATUS_DELETED,
4823         DIFF_STATUS_MODIFIED,
4824         DIFF_STATUS_RENAMED,
4825         DIFF_STATUS_TYPE_CHANGED,
4826         DIFF_STATUS_UNKNOWN,
4827         DIFF_STATUS_UNMERGED,
4828         DIFF_STATUS_FILTER_AON,
4829         DIFF_STATUS_FILTER_BROKEN,
4830         '\0',
4831 };
4832
4833 static unsigned int filter_bit['Z' + 1];
4834
4835 static void prepare_filter_bits(void)
4836 {
4837         int i;
4838
4839         if (!filter_bit[DIFF_STATUS_ADDED]) {
4840                 for (i = 0; diff_status_letters[i]; i++)
4841                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4842         }
4843 }
4844
4845 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4846 {
4847         return opt->filter & filter_bit[(int) status];
4848 }
4849
4850 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4851 {
4852         int i, optch;
4853
4854         prepare_filter_bits();
4855
4856         /*
4857          * If there is a negation e.g. 'd' in the input, and we haven't
4858          * initialized the filter field with another --diff-filter, start
4859          * from full set of bits, except for AON.
4860          */
4861         if (!opt->filter) {
4862                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4863                         if (optch < 'a' || 'z' < optch)
4864                                 continue;
4865                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4866                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4867                         break;
4868                 }
4869         }
4870
4871         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4872                 unsigned int bit;
4873                 int negate;
4874
4875                 if ('a' <= optch && optch <= 'z') {
4876                         negate = 1;
4877                         optch = toupper(optch);
4878                 } else {
4879                         negate = 0;
4880                 }
4881
4882                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4883                 if (!bit)
4884                         return optarg[i];
4885                 if (negate)
4886                         opt->filter &= ~bit;
4887                 else
4888                         opt->filter |= bit;
4889         }
4890         return 0;
4891 }
4892
4893 static void enable_patch_output(int *fmt)
4894 {
4895         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4896         *fmt |= DIFF_FORMAT_PATCH;
4897 }
4898
4899 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4900 {
4901         int val = parse_ws_error_highlight(arg);
4902
4903         if (val < 0) {
4904                 error("unknown value after ws-error-highlight=%.*s",
4905                       -1 - val, arg);
4906                 return 0;
4907         }
4908         opt->ws_error_highlight = val;
4909         return 1;
4910 }
4911
4912 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4913 {
4914         struct object_id oid;
4915
4916         if (get_oid(arg, &oid))
4917                 return error("unable to resolve '%s'", arg);
4918
4919         if (!opt->objfind)
4920                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4921
4922         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4923         opt->flags.recursive = 1;
4924         opt->flags.tree_in_recursive = 1;
4925         oidset_insert(opt->objfind, &oid);
4926         return 1;
4927 }
4928
4929 static int diff_opt_unified(const struct option *opt,
4930                             const char *arg, int unset)
4931 {
4932         struct diff_options *options = opt->value;
4933         char *s;
4934
4935         BUG_ON_OPT_NEG(unset);
4936
4937         options->context = strtol(arg, &s, 10);
4938         if (*s)
4939                 return error(_("%s expects a numerical value"), "--unified");
4940         enable_patch_output(&options->output_format);
4941
4942         return 0;
4943 }
4944
4945 static void prep_parse_options(struct diff_options *options)
4946 {
4947         struct option parseopts[] = {
4948                 OPT_GROUP(N_("Diff output format options")),
4949                 OPT_BITOP('p', "patch", &options->output_format,
4950                           N_("generate patch"),
4951                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4952                 OPT_BITOP('u', NULL, &options->output_format,
4953                           N_("generate patch"),
4954                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4955                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
4956                                N_("generate diffs with <n> lines context"),
4957                                PARSE_OPT_NONEG, diff_opt_unified),
4958                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
4959                          N_("generate diffs with <n> lines context")),
4960                 OPT_BIT_F(0, "raw", &options->output_format,
4961                           N_("generate the diff in raw format"),
4962                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
4963                 OPT_END()
4964         };
4965
4966         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
4967         memcpy(options->parseopts, parseopts, sizeof(parseopts));
4968 }
4969
4970 int diff_opt_parse(struct diff_options *options,
4971                    const char **av, int ac, const char *prefix)
4972 {
4973         const char *arg = av[0];
4974         const char *optarg;
4975         int argcount;
4976
4977         if (!prefix)
4978                 prefix = "";
4979
4980         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
4981                            PARSE_OPT_KEEP_DASHDASH |
4982                            PARSE_OPT_KEEP_UNKNOWN |
4983                            PARSE_OPT_NO_INTERNAL_HELP |
4984                            PARSE_OPT_ONE_SHOT |
4985                            PARSE_OPT_STOP_AT_NON_OPTION);
4986
4987         if (ac)
4988                 return ac;
4989
4990         /* Output format options */
4991         if (!strcmp(arg, "--patch-with-raw")) {
4992                 enable_patch_output(&options->output_format);
4993                 options->output_format |= DIFF_FORMAT_RAW;
4994         } else if (!strcmp(arg, "--numstat"))
4995                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4996         else if (!strcmp(arg, "--shortstat"))
4997                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4998         else if (skip_prefix(arg, "-X", &arg) ||
4999                  skip_to_optional_arg(arg, "--dirstat", &arg))
5000                 return parse_dirstat_opt(options, arg);
5001         else if (!strcmp(arg, "--cumulative"))
5002                 return parse_dirstat_opt(options, "cumulative");
5003         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
5004                 parse_dirstat_opt(options, "files");
5005                 return parse_dirstat_opt(options, arg);
5006         }
5007         else if (!strcmp(arg, "--check"))
5008                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
5009         else if (!strcmp(arg, "--summary"))
5010                 options->output_format |= DIFF_FORMAT_SUMMARY;
5011         else if (!strcmp(arg, "--patch-with-stat")) {
5012                 enable_patch_output(&options->output_format);
5013                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5014         } else if (!strcmp(arg, "--name-only"))
5015                 options->output_format |= DIFF_FORMAT_NAME;
5016         else if (!strcmp(arg, "--name-status"))
5017                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
5018         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
5019                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
5020         else if (starts_with(arg, "--stat"))
5021                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
5022                 return stat_opt(options, av);
5023         else if (!strcmp(arg, "--compact-summary")) {
5024                  options->flags.stat_with_summary = 1;
5025                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
5026         } else if (!strcmp(arg, "--no-compact-summary"))
5027                  options->flags.stat_with_summary = 0;
5028         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
5029                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
5030         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
5031                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
5032         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
5033                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
5034
5035         /* renames options */
5036         else if (starts_with(arg, "-B") ||
5037                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
5038                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
5039                         return error("invalid argument to -B: %s", arg+2);
5040         }
5041         else if (starts_with(arg, "-M") ||
5042                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
5043                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5044                         return error("invalid argument to -M: %s", arg+2);
5045                 options->detect_rename = DIFF_DETECT_RENAME;
5046         }
5047         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
5048                 options->irreversible_delete = 1;
5049         }
5050         else if (starts_with(arg, "-C") ||
5051                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
5052                 if (options->detect_rename == DIFF_DETECT_COPY)
5053                         options->flags.find_copies_harder = 1;
5054                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5055                         return error("invalid argument to -C: %s", arg+2);
5056                 options->detect_rename = DIFF_DETECT_COPY;
5057         }
5058         else if (!strcmp(arg, "--no-renames"))
5059                 options->detect_rename = 0;
5060         else if (!strcmp(arg, "--rename-empty"))
5061                 options->flags.rename_empty = 1;
5062         else if (!strcmp(arg, "--no-rename-empty"))
5063                 options->flags.rename_empty = 0;
5064         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5065                 options->flags.relative_name = 1;
5066                 if (arg)
5067                         options->prefix = arg;
5068         }
5069
5070         /* xdiff options */
5071         else if (!strcmp(arg, "--minimal"))
5072                 DIFF_XDL_SET(options, NEED_MINIMAL);
5073         else if (!strcmp(arg, "--no-minimal"))
5074                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5075         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5076                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5077         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5078                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5079         else if (!strcmp(arg, "--ignore-space-at-eol"))
5080                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5081         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5082                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5083         else if (!strcmp(arg, "--ignore-blank-lines"))
5084                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5085         else if (!strcmp(arg, "--indent-heuristic"))
5086                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5087         else if (!strcmp(arg, "--no-indent-heuristic"))
5088                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5089         else if (!strcmp(arg, "--patience")) {
5090                 int i;
5091                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5092                 /*
5093                  * Both --patience and --anchored use PATIENCE_DIFF
5094                  * internally, so remove any anchors previously
5095                  * specified.
5096                  */
5097                 for (i = 0; i < options->anchors_nr; i++)
5098                         free(options->anchors[i]);
5099                 options->anchors_nr = 0;
5100         } else if (!strcmp(arg, "--histogram"))
5101                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5102         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5103                 long value = parse_algorithm_value(optarg);
5104                 if (value < 0)
5105                         return error("option diff-algorithm accepts \"myers\", "
5106                                      "\"minimal\", \"patience\" and \"histogram\"");
5107                 /* clear out previous settings */
5108                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5109                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5110                 options->xdl_opts |= value;
5111                 return argcount;
5112         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5113                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5114                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5115                            options->anchors_alloc);
5116                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5117         }
5118
5119         /* flags options */
5120         else if (!strcmp(arg, "--binary")) {
5121                 enable_patch_output(&options->output_format);
5122                 options->flags.binary = 1;
5123         }
5124         else if (!strcmp(arg, "--full-index"))
5125                 options->flags.full_index = 1;
5126         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5127                 options->flags.text = 1;
5128         else if (!strcmp(arg, "-R"))
5129                 options->flags.reverse_diff = 1;
5130         else if (!strcmp(arg, "--find-copies-harder"))
5131                 options->flags.find_copies_harder = 1;
5132         else if (!strcmp(arg, "--follow"))
5133                 options->flags.follow_renames = 1;
5134         else if (!strcmp(arg, "--no-follow")) {
5135                 options->flags.follow_renames = 0;
5136                 options->flags.default_follow_renames = 0;
5137         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5138                 int value = git_config_colorbool(NULL, arg);
5139                 if (value < 0)
5140                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5141                 options->use_color = value;
5142         }
5143         else if (!strcmp(arg, "--no-color"))
5144                 options->use_color = 0;
5145         else if (!strcmp(arg, "--color-moved")) {
5146                 if (diff_color_moved_default)
5147                         options->color_moved = diff_color_moved_default;
5148                 if (options->color_moved == COLOR_MOVED_NO)
5149                         options->color_moved = COLOR_MOVED_DEFAULT;
5150         } else if (!strcmp(arg, "--no-color-moved"))
5151                 options->color_moved = COLOR_MOVED_NO;
5152         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5153                 int cm = parse_color_moved(arg);
5154                 if (cm < 0)
5155                         return error("bad --color-moved argument: %s", arg);
5156                 options->color_moved = cm;
5157         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5158                 options->color_moved_ws_handling = 0;
5159         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5160                 unsigned cm = parse_color_moved_ws(arg);
5161                 if (cm & COLOR_MOVED_WS_ERROR)
5162                         return -1;
5163                 options->color_moved_ws_handling = cm;
5164         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5165                 options->use_color = 1;
5166                 options->word_diff = DIFF_WORDS_COLOR;
5167         }
5168         else if (!strcmp(arg, "--word-diff")) {
5169                 if (options->word_diff == DIFF_WORDS_NONE)
5170                         options->word_diff = DIFF_WORDS_PLAIN;
5171         }
5172         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5173                 if (!strcmp(arg, "plain"))
5174                         options->word_diff = DIFF_WORDS_PLAIN;
5175                 else if (!strcmp(arg, "color")) {
5176                         options->use_color = 1;
5177                         options->word_diff = DIFF_WORDS_COLOR;
5178                 }
5179                 else if (!strcmp(arg, "porcelain"))
5180                         options->word_diff = DIFF_WORDS_PORCELAIN;
5181                 else if (!strcmp(arg, "none"))
5182                         options->word_diff = DIFF_WORDS_NONE;
5183                 else
5184                         die("bad --word-diff argument: %s", arg);
5185         }
5186         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5187                 if (options->word_diff == DIFF_WORDS_NONE)
5188                         options->word_diff = DIFF_WORDS_PLAIN;
5189                 options->word_regex = optarg;
5190                 return argcount;
5191         }
5192         else if (!strcmp(arg, "--exit-code"))
5193                 options->flags.exit_with_status = 1;
5194         else if (!strcmp(arg, "--quiet"))
5195                 options->flags.quick = 1;
5196         else if (!strcmp(arg, "--ext-diff"))
5197                 options->flags.allow_external = 1;
5198         else if (!strcmp(arg, "--no-ext-diff"))
5199                 options->flags.allow_external = 0;
5200         else if (!strcmp(arg, "--textconv")) {
5201                 options->flags.allow_textconv = 1;
5202                 options->flags.textconv_set_via_cmdline = 1;
5203         } else if (!strcmp(arg, "--no-textconv"))
5204                 options->flags.allow_textconv = 0;
5205         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5206                 options->flags.override_submodule_config = 1;
5207                 handle_ignore_submodules_arg(options, arg);
5208         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5209                 return parse_submodule_opt(options, arg);
5210         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5211                 return parse_ws_error_highlight_opt(options, arg);
5212         else if (!strcmp(arg, "--ita-invisible-in-index"))
5213                 options->ita_invisible_in_index = 1;
5214         else if (!strcmp(arg, "--ita-visible-in-index"))
5215                 options->ita_invisible_in_index = 0;
5216
5217         /* misc options */
5218         else if (!strcmp(arg, "-z"))
5219                 options->line_termination = 0;
5220         else if ((argcount = short_opt('l', av, &optarg))) {
5221                 options->rename_limit = strtoul(optarg, NULL, 10);
5222                 return argcount;
5223         }
5224         else if ((argcount = short_opt('S', av, &optarg))) {
5225                 options->pickaxe = optarg;
5226                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5227                 return argcount;
5228         } else if ((argcount = short_opt('G', av, &optarg))) {
5229                 options->pickaxe = optarg;
5230                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5231                 return argcount;
5232         }
5233         else if (!strcmp(arg, "--pickaxe-all"))
5234                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5235         else if (!strcmp(arg, "--pickaxe-regex"))
5236                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5237         else if ((argcount = short_opt('O', av, &optarg))) {
5238                 options->orderfile = prefix_filename(prefix, optarg);
5239                 return argcount;
5240         } else if (skip_prefix(arg, "--find-object=", &arg))
5241                 return parse_objfind_opt(options, arg);
5242         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5243                 int offending = parse_diff_filter_opt(optarg, options);
5244                 if (offending)
5245                         die("unknown change class '%c' in --diff-filter=%s",
5246                             offending, optarg);
5247                 return argcount;
5248         }
5249         else if (!strcmp(arg, "--no-abbrev"))
5250                 options->abbrev = 0;
5251         else if (!strcmp(arg, "--abbrev"))
5252                 options->abbrev = DEFAULT_ABBREV;
5253         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5254                 options->abbrev = strtoul(arg, NULL, 10);
5255                 if (options->abbrev < MINIMUM_ABBREV)
5256                         options->abbrev = MINIMUM_ABBREV;
5257                 else if (the_hash_algo->hexsz < options->abbrev)
5258                         options->abbrev = the_hash_algo->hexsz;
5259         }
5260         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5261                 options->a_prefix = optarg;
5262                 return argcount;
5263         }
5264         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5265                 options->line_prefix = optarg;
5266                 options->line_prefix_length = strlen(options->line_prefix);
5267                 graph_setup_line_prefix(options);
5268                 return argcount;
5269         }
5270         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5271                 options->b_prefix = optarg;
5272                 return argcount;
5273         }
5274         else if (!strcmp(arg, "--no-prefix"))
5275                 options->a_prefix = options->b_prefix = "";
5276         else if (opt_arg(arg, '\0', "inter-hunk-context",
5277                          &options->interhunkcontext))
5278                 ;
5279         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5280                 char *path = prefix_filename(prefix, optarg);
5281                 options->file = xfopen(path, "w");
5282                 options->close_file = 1;
5283                 if (options->use_color != GIT_COLOR_ALWAYS)
5284                         options->use_color = GIT_COLOR_NEVER;
5285                 free(path);
5286                 return argcount;
5287         } else
5288                 return 0;
5289         return 1;
5290 }
5291
5292 int parse_rename_score(const char **cp_p)
5293 {
5294         unsigned long num, scale;
5295         int ch, dot;
5296         const char *cp = *cp_p;
5297
5298         num = 0;
5299         scale = 1;
5300         dot = 0;
5301         for (;;) {
5302                 ch = *cp;
5303                 if ( !dot && ch == '.' ) {
5304                         scale = 1;
5305                         dot = 1;
5306                 } else if ( ch == '%' ) {
5307                         scale = dot ? scale*100 : 100;
5308                         cp++;   /* % is always at the end */
5309                         break;
5310                 } else if ( ch >= '0' && ch <= '9' ) {
5311                         if ( scale < 100000 ) {
5312                                 scale *= 10;
5313                                 num = (num*10) + (ch-'0');
5314                         }
5315                 } else {
5316                         break;
5317                 }
5318                 cp++;
5319         }
5320         *cp_p = cp;
5321
5322         /* user says num divided by scale and we say internally that
5323          * is MAX_SCORE * num / scale.
5324          */
5325         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5326 }
5327
5328 static int diff_scoreopt_parse(const char *opt)
5329 {
5330         int opt1, opt2, cmd;
5331
5332         if (*opt++ != '-')
5333                 return -1;
5334         cmd = *opt++;
5335         if (cmd == '-') {
5336                 /* convert the long-form arguments into short-form versions */
5337                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5338                         if (*opt == 0 || *opt++ == '=')
5339                                 cmd = 'B';
5340                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5341                         if (*opt == 0 || *opt++ == '=')
5342                                 cmd = 'C';
5343                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5344                         if (*opt == 0 || *opt++ == '=')
5345                                 cmd = 'M';
5346                 }
5347         }
5348         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5349                 return -1; /* that is not a -M, -C, or -B option */
5350
5351         opt1 = parse_rename_score(&opt);
5352         if (cmd != 'B')
5353                 opt2 = 0;
5354         else {
5355                 if (*opt == 0)
5356                         opt2 = 0;
5357                 else if (*opt != '/')
5358                         return -1; /* we expect -B80/99 or -B80 */
5359                 else {
5360                         opt++;
5361                         opt2 = parse_rename_score(&opt);
5362                 }
5363         }
5364         if (*opt != 0)
5365                 return -1;
5366         return opt1 | (opt2 << 16);
5367 }
5368
5369 struct diff_queue_struct diff_queued_diff;
5370
5371 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5372 {
5373         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5374         queue->queue[queue->nr++] = dp;
5375 }
5376
5377 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5378                                  struct diff_filespec *one,
5379                                  struct diff_filespec *two)
5380 {
5381         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5382         dp->one = one;
5383         dp->two = two;
5384         if (queue)
5385                 diff_q(queue, dp);
5386         return dp;
5387 }
5388
5389 void diff_free_filepair(struct diff_filepair *p)
5390 {
5391         free_filespec(p->one);
5392         free_filespec(p->two);
5393         free(p);
5394 }
5395
5396 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5397 {
5398         int abblen;
5399         const char *abbrev;
5400
5401         /* Do we want all 40 hex characters? */
5402         if (len == the_hash_algo->hexsz)
5403                 return oid_to_hex(oid);
5404
5405         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5406         abbrev = diff_abbrev_oid(oid, len);
5407
5408         if (!print_sha1_ellipsis())
5409                 return abbrev;
5410
5411         abblen = strlen(abbrev);
5412
5413         /*
5414          * In well-behaved cases, where the abbreviated result is the
5415          * same as the requested length, append three dots after the
5416          * abbreviation (hence the whole logic is limited to the case
5417          * where abblen < 37); when the actual abbreviated result is a
5418          * bit longer than the requested length, we reduce the number
5419          * of dots so that they match the well-behaved ones.  However,
5420          * if the actual abbreviation is longer than the requested
5421          * length by more than three, we give up on aligning, and add
5422          * three dots anyway, to indicate that the output is not the
5423          * full object name.  Yes, this may be suboptimal, but this
5424          * appears only in "diff --raw --abbrev" output and it is not
5425          * worth the effort to change it now.  Note that this would
5426          * likely to work fine when the automatic sizing of default
5427          * abbreviation length is used--we would be fed -1 in "len" in
5428          * that case, and will end up always appending three-dots, but
5429          * the automatic sizing is supposed to give abblen that ensures
5430          * uniqueness across all objects (statistically speaking).
5431          */
5432         if (abblen < the_hash_algo->hexsz - 3) {
5433                 static char hex[GIT_MAX_HEXSZ + 1];
5434                 if (len < abblen && abblen <= len + 2)
5435                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5436                 else
5437                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5438                 return hex;
5439         }
5440
5441         return oid_to_hex(oid);
5442 }
5443
5444 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5445 {
5446         int line_termination = opt->line_termination;
5447         int inter_name_termination = line_termination ? '\t' : '\0';
5448
5449         fprintf(opt->file, "%s", diff_line_prefix(opt));
5450         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5451                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5452                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5453                 fprintf(opt->file, "%s ",
5454                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5455         }
5456         if (p->score) {
5457                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5458                         inter_name_termination);
5459         } else {
5460                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5461         }
5462
5463         if (p->status == DIFF_STATUS_COPIED ||
5464             p->status == DIFF_STATUS_RENAMED) {
5465                 const char *name_a, *name_b;
5466                 name_a = p->one->path;
5467                 name_b = p->two->path;
5468                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5469                 write_name_quoted(name_a, opt->file, inter_name_termination);
5470                 write_name_quoted(name_b, opt->file, line_termination);
5471         } else {
5472                 const char *name_a, *name_b;
5473                 name_a = p->one->mode ? p->one->path : p->two->path;
5474                 name_b = NULL;
5475                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5476                 write_name_quoted(name_a, opt->file, line_termination);
5477         }
5478 }
5479
5480 int diff_unmodified_pair(struct diff_filepair *p)
5481 {
5482         /* This function is written stricter than necessary to support
5483          * the currently implemented transformers, but the idea is to
5484          * let transformers to produce diff_filepairs any way they want,
5485          * and filter and clean them up here before producing the output.
5486          */
5487         struct diff_filespec *one = p->one, *two = p->two;
5488
5489         if (DIFF_PAIR_UNMERGED(p))
5490                 return 0; /* unmerged is interesting */
5491
5492         /* deletion, addition, mode or type change
5493          * and rename are all interesting.
5494          */
5495         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5496             DIFF_PAIR_MODE_CHANGED(p) ||
5497             strcmp(one->path, two->path))
5498                 return 0;
5499
5500         /* both are valid and point at the same path.  that is, we are
5501          * dealing with a change.
5502          */
5503         if (one->oid_valid && two->oid_valid &&
5504             oideq(&one->oid, &two->oid) &&
5505             !one->dirty_submodule && !two->dirty_submodule)
5506                 return 1; /* no change */
5507         if (!one->oid_valid && !two->oid_valid)
5508                 return 1; /* both look at the same file on the filesystem. */
5509         return 0;
5510 }
5511
5512 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5513 {
5514         if (diff_unmodified_pair(p))
5515                 return;
5516
5517         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5518             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5519                 return; /* no tree diffs in patch format */
5520
5521         run_diff(p, o);
5522 }
5523
5524 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5525                             struct diffstat_t *diffstat)
5526 {
5527         if (diff_unmodified_pair(p))
5528                 return;
5529
5530         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5531             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5532                 return; /* no useful stat for tree diffs */
5533
5534         run_diffstat(p, o, diffstat);
5535 }
5536
5537 static void diff_flush_checkdiff(struct diff_filepair *p,
5538                 struct diff_options *o)
5539 {
5540         if (diff_unmodified_pair(p))
5541                 return;
5542
5543         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5544             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5545                 return; /* nothing to check in tree diffs */
5546
5547         run_checkdiff(p, o);
5548 }
5549
5550 int diff_queue_is_empty(void)
5551 {
5552         struct diff_queue_struct *q = &diff_queued_diff;
5553         int i;
5554         for (i = 0; i < q->nr; i++)
5555                 if (!diff_unmodified_pair(q->queue[i]))
5556                         return 0;
5557         return 1;
5558 }
5559
5560 #if DIFF_DEBUG
5561 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5562 {
5563         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5564                 x, one ? one : "",
5565                 s->path,
5566                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5567                 s->mode,
5568                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5569         fprintf(stderr, "queue[%d] %s size %lu\n",
5570                 x, one ? one : "",
5571                 s->size);
5572 }
5573
5574 void diff_debug_filepair(const struct diff_filepair *p, int i)
5575 {
5576         diff_debug_filespec(p->one, i, "one");
5577         diff_debug_filespec(p->two, i, "two");
5578         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5579                 p->score, p->status ? p->status : '?',
5580                 p->one->rename_used, p->broken_pair);
5581 }
5582
5583 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5584 {
5585         int i;
5586         if (msg)
5587                 fprintf(stderr, "%s\n", msg);
5588         fprintf(stderr, "q->nr = %d\n", q->nr);
5589         for (i = 0; i < q->nr; i++) {
5590                 struct diff_filepair *p = q->queue[i];
5591                 diff_debug_filepair(p, i);
5592         }
5593 }
5594 #endif
5595
5596 static void diff_resolve_rename_copy(void)
5597 {
5598         int i;
5599         struct diff_filepair *p;
5600         struct diff_queue_struct *q = &diff_queued_diff;
5601
5602         diff_debug_queue("resolve-rename-copy", q);
5603
5604         for (i = 0; i < q->nr; i++) {
5605                 p = q->queue[i];
5606                 p->status = 0; /* undecided */
5607                 if (DIFF_PAIR_UNMERGED(p))
5608                         p->status = DIFF_STATUS_UNMERGED;
5609                 else if (!DIFF_FILE_VALID(p->one))
5610                         p->status = DIFF_STATUS_ADDED;
5611                 else if (!DIFF_FILE_VALID(p->two))
5612                         p->status = DIFF_STATUS_DELETED;
5613                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5614                         p->status = DIFF_STATUS_TYPE_CHANGED;
5615
5616                 /* from this point on, we are dealing with a pair
5617                  * whose both sides are valid and of the same type, i.e.
5618                  * either in-place edit or rename/copy edit.
5619                  */
5620                 else if (DIFF_PAIR_RENAME(p)) {
5621                         /*
5622                          * A rename might have re-connected a broken
5623                          * pair up, causing the pathnames to be the
5624                          * same again. If so, that's not a rename at
5625                          * all, just a modification..
5626                          *
5627                          * Otherwise, see if this source was used for
5628                          * multiple renames, in which case we decrement
5629                          * the count, and call it a copy.
5630                          */
5631                         if (!strcmp(p->one->path, p->two->path))
5632                                 p->status = DIFF_STATUS_MODIFIED;
5633                         else if (--p->one->rename_used > 0)
5634                                 p->status = DIFF_STATUS_COPIED;
5635                         else
5636                                 p->status = DIFF_STATUS_RENAMED;
5637                 }
5638                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5639                          p->one->mode != p->two->mode ||
5640                          p->one->dirty_submodule ||
5641                          p->two->dirty_submodule ||
5642                          is_null_oid(&p->one->oid))
5643                         p->status = DIFF_STATUS_MODIFIED;
5644                 else {
5645                         /* This is a "no-change" entry and should not
5646                          * happen anymore, but prepare for broken callers.
5647                          */
5648                         error("feeding unmodified %s to diffcore",
5649                               p->one->path);
5650                         p->status = DIFF_STATUS_UNKNOWN;
5651                 }
5652         }
5653         diff_debug_queue("resolve-rename-copy done", q);
5654 }
5655
5656 static int check_pair_status(struct diff_filepair *p)
5657 {
5658         switch (p->status) {
5659         case DIFF_STATUS_UNKNOWN:
5660                 return 0;
5661         case 0:
5662                 die("internal error in diff-resolve-rename-copy");
5663         default:
5664                 return 1;
5665         }
5666 }
5667
5668 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5669 {
5670         int fmt = opt->output_format;
5671
5672         if (fmt & DIFF_FORMAT_CHECKDIFF)
5673                 diff_flush_checkdiff(p, opt);
5674         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5675                 diff_flush_raw(p, opt);
5676         else if (fmt & DIFF_FORMAT_NAME) {
5677                 const char *name_a, *name_b;
5678                 name_a = p->two->path;
5679                 name_b = NULL;
5680                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5681                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5682                 write_name_quoted(name_a, opt->file, opt->line_termination);
5683         }
5684 }
5685
5686 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5687 {
5688         struct strbuf sb = STRBUF_INIT;
5689         if (fs->mode)
5690                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5691         else
5692                 strbuf_addf(&sb, " %s ", newdelete);
5693
5694         quote_c_style(fs->path, &sb, NULL, 0);
5695         strbuf_addch(&sb, '\n');
5696         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5697                          sb.buf, sb.len, 0);
5698         strbuf_release(&sb);
5699 }
5700
5701 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5702                 int show_name)
5703 {
5704         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5705                 struct strbuf sb = STRBUF_INIT;
5706                 strbuf_addf(&sb, " mode change %06o => %06o",
5707                             p->one->mode, p->two->mode);
5708                 if (show_name) {
5709                         strbuf_addch(&sb, ' ');
5710                         quote_c_style(p->two->path, &sb, NULL, 0);
5711                 }
5712                 strbuf_addch(&sb, '\n');
5713                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5714                                  sb.buf, sb.len, 0);
5715                 strbuf_release(&sb);
5716         }
5717 }
5718
5719 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5720                 struct diff_filepair *p)
5721 {
5722         struct strbuf sb = STRBUF_INIT;
5723         struct strbuf names = STRBUF_INIT;
5724
5725         pprint_rename(&names, p->one->path, p->two->path);
5726         strbuf_addf(&sb, " %s %s (%d%%)\n",
5727                     renamecopy, names.buf, similarity_index(p));
5728         strbuf_release(&names);
5729         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5730                                  sb.buf, sb.len, 0);
5731         show_mode_change(opt, p, 0);
5732         strbuf_release(&sb);
5733 }
5734
5735 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5736 {
5737         switch(p->status) {
5738         case DIFF_STATUS_DELETED:
5739                 show_file_mode_name(opt, "delete", p->one);
5740                 break;
5741         case DIFF_STATUS_ADDED:
5742                 show_file_mode_name(opt, "create", p->two);
5743                 break;
5744         case DIFF_STATUS_COPIED:
5745                 show_rename_copy(opt, "copy", p);
5746                 break;
5747         case DIFF_STATUS_RENAMED:
5748                 show_rename_copy(opt, "rename", p);
5749                 break;
5750         default:
5751                 if (p->score) {
5752                         struct strbuf sb = STRBUF_INIT;
5753                         strbuf_addstr(&sb, " rewrite ");
5754                         quote_c_style(p->two->path, &sb, NULL, 0);
5755                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5756                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5757                                          sb.buf, sb.len, 0);
5758                         strbuf_release(&sb);
5759                 }
5760                 show_mode_change(opt, p, !p->score);
5761                 break;
5762         }
5763 }
5764
5765 struct patch_id_t {
5766         git_SHA_CTX *ctx;
5767         int patchlen;
5768 };
5769
5770 static int remove_space(char *line, int len)
5771 {
5772         int i;
5773         char *dst = line;
5774         unsigned char c;
5775
5776         for (i = 0; i < len; i++)
5777                 if (!isspace((c = line[i])))
5778                         *dst++ = c;
5779
5780         return dst - line;
5781 }
5782
5783 static void patch_id_consume(void *priv, char *line, unsigned long len)
5784 {
5785         struct patch_id_t *data = priv;
5786         int new_len;
5787
5788         new_len = remove_space(line, len);
5789
5790         git_SHA1_Update(data->ctx, line, new_len);
5791         data->patchlen += new_len;
5792 }
5793
5794 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5795 {
5796         git_SHA1_Update(ctx, str, strlen(str));
5797 }
5798
5799 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5800 {
5801         /* large enough for 2^32 in octal */
5802         char buf[12];
5803         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5804         git_SHA1_Update(ctx, buf, len);
5805 }
5806
5807 /* returns 0 upon success, and writes result into sha1 */
5808 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5809 {
5810         struct diff_queue_struct *q = &diff_queued_diff;
5811         int i;
5812         git_SHA_CTX ctx;
5813         struct patch_id_t data;
5814
5815         git_SHA1_Init(&ctx);
5816         memset(&data, 0, sizeof(struct patch_id_t));
5817         data.ctx = &ctx;
5818
5819         for (i = 0; i < q->nr; i++) {
5820                 xpparam_t xpp;
5821                 xdemitconf_t xecfg;
5822                 mmfile_t mf1, mf2;
5823                 struct diff_filepair *p = q->queue[i];
5824                 int len1, len2;
5825
5826                 memset(&xpp, 0, sizeof(xpp));
5827                 memset(&xecfg, 0, sizeof(xecfg));
5828                 if (p->status == 0)
5829                         return error("internal diff status error");
5830                 if (p->status == DIFF_STATUS_UNKNOWN)
5831                         continue;
5832                 if (diff_unmodified_pair(p))
5833                         continue;
5834                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5835                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5836                         continue;
5837                 if (DIFF_PAIR_UNMERGED(p))
5838                         continue;
5839
5840                 diff_fill_oid_info(p->one, options->repo->index);
5841                 diff_fill_oid_info(p->two, options->repo->index);
5842
5843                 len1 = remove_space(p->one->path, strlen(p->one->path));
5844                 len2 = remove_space(p->two->path, strlen(p->two->path));
5845                 patch_id_add_string(&ctx, "diff--git");
5846                 patch_id_add_string(&ctx, "a/");
5847                 git_SHA1_Update(&ctx, p->one->path, len1);
5848                 patch_id_add_string(&ctx, "b/");
5849                 git_SHA1_Update(&ctx, p->two->path, len2);
5850
5851                 if (p->one->mode == 0) {
5852                         patch_id_add_string(&ctx, "newfilemode");
5853                         patch_id_add_mode(&ctx, p->two->mode);
5854                         patch_id_add_string(&ctx, "---/dev/null");
5855                         patch_id_add_string(&ctx, "+++b/");
5856                         git_SHA1_Update(&ctx, p->two->path, len2);
5857                 } else if (p->two->mode == 0) {
5858                         patch_id_add_string(&ctx, "deletedfilemode");
5859                         patch_id_add_mode(&ctx, p->one->mode);
5860                         patch_id_add_string(&ctx, "---a/");
5861                         git_SHA1_Update(&ctx, p->one->path, len1);
5862                         patch_id_add_string(&ctx, "+++/dev/null");
5863                 } else {
5864                         patch_id_add_string(&ctx, "---a/");
5865                         git_SHA1_Update(&ctx, p->one->path, len1);
5866                         patch_id_add_string(&ctx, "+++b/");
5867                         git_SHA1_Update(&ctx, p->two->path, len2);
5868                 }
5869
5870                 if (diff_header_only)
5871                         continue;
5872
5873                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5874                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5875                         return error("unable to read files to diff");
5876
5877                 if (diff_filespec_is_binary(options->repo, p->one) ||
5878                     diff_filespec_is_binary(options->repo, p->two)) {
5879                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5880                                         GIT_SHA1_HEXSZ);
5881                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5882                                         GIT_SHA1_HEXSZ);
5883                         continue;
5884                 }
5885
5886                 xpp.flags = 0;
5887                 xecfg.ctxlen = 3;
5888                 xecfg.flags = 0;
5889                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5890                                   patch_id_consume, &data, &xpp, &xecfg))
5891                         return error("unable to generate patch-id diff for %s",
5892                                      p->one->path);
5893         }
5894
5895         git_SHA1_Final(oid->hash, &ctx);
5896         return 0;
5897 }
5898
5899 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5900 {
5901         struct diff_queue_struct *q = &diff_queued_diff;
5902         int i;
5903         int result = diff_get_patch_id(options, oid, diff_header_only);
5904
5905         for (i = 0; i < q->nr; i++)
5906                 diff_free_filepair(q->queue[i]);
5907
5908         free(q->queue);
5909         DIFF_QUEUE_CLEAR(q);
5910
5911         return result;
5912 }
5913
5914 static int is_summary_empty(const struct diff_queue_struct *q)
5915 {
5916         int i;
5917
5918         for (i = 0; i < q->nr; i++) {
5919                 const struct diff_filepair *p = q->queue[i];
5920
5921                 switch (p->status) {
5922                 case DIFF_STATUS_DELETED:
5923                 case DIFF_STATUS_ADDED:
5924                 case DIFF_STATUS_COPIED:
5925                 case DIFF_STATUS_RENAMED:
5926                         return 0;
5927                 default:
5928                         if (p->score)
5929                                 return 0;
5930                         if (p->one->mode && p->two->mode &&
5931                             p->one->mode != p->two->mode)
5932                                 return 0;
5933                         break;
5934                 }
5935         }
5936         return 1;
5937 }
5938
5939 static const char rename_limit_warning[] =
5940 N_("inexact rename detection was skipped due to too many files.");
5941
5942 static const char degrade_cc_to_c_warning[] =
5943 N_("only found copies from modified paths due to too many files.");
5944
5945 static const char rename_limit_advice[] =
5946 N_("you may want to set your %s variable to at least "
5947    "%d and retry the command.");
5948
5949 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5950 {
5951         fflush(stdout);
5952         if (degraded_cc)
5953                 warning(_(degrade_cc_to_c_warning));
5954         else if (needed)
5955                 warning(_(rename_limit_warning));
5956         else
5957                 return;
5958         if (0 < needed)
5959                 warning(_(rename_limit_advice), varname, needed);
5960 }
5961
5962 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5963 {
5964         int i;
5965         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5966         struct diff_queue_struct *q = &diff_queued_diff;
5967
5968         if (WSEH_NEW & WS_RULE_MASK)
5969                 BUG("WS rules bit mask overlaps with diff symbol flags");
5970
5971         if (o->color_moved)
5972                 o->emitted_symbols = &esm;
5973
5974         for (i = 0; i < q->nr; i++) {
5975                 struct diff_filepair *p = q->queue[i];
5976                 if (check_pair_status(p))
5977                         diff_flush_patch(p, o);
5978         }
5979
5980         if (o->emitted_symbols) {
5981                 if (o->color_moved) {
5982                         struct hashmap add_lines, del_lines;
5983
5984                         if (o->color_moved_ws_handling &
5985                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5986                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5987
5988                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5989                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5990
5991                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5992                         mark_color_as_moved(o, &add_lines, &del_lines);
5993                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5994                                 dim_moved_lines(o);
5995
5996                         hashmap_free(&add_lines, 1);
5997                         hashmap_free(&del_lines, 1);
5998                 }
5999
6000                 for (i = 0; i < esm.nr; i++)
6001                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6002
6003                 for (i = 0; i < esm.nr; i++)
6004                         free((void *)esm.buf[i].line);
6005                 esm.nr = 0;
6006
6007                 o->emitted_symbols = NULL;
6008         }
6009 }
6010
6011 void diff_flush(struct diff_options *options)
6012 {
6013         struct diff_queue_struct *q = &diff_queued_diff;
6014         int i, output_format = options->output_format;
6015         int separator = 0;
6016         int dirstat_by_line = 0;
6017
6018         /*
6019          * Order: raw, stat, summary, patch
6020          * or:    name/name-status/checkdiff (other bits clear)
6021          */
6022         if (!q->nr)
6023                 goto free_queue;
6024
6025         if (output_format & (DIFF_FORMAT_RAW |
6026                              DIFF_FORMAT_NAME |
6027                              DIFF_FORMAT_NAME_STATUS |
6028                              DIFF_FORMAT_CHECKDIFF)) {
6029                 for (i = 0; i < q->nr; i++) {
6030                         struct diff_filepair *p = q->queue[i];
6031                         if (check_pair_status(p))
6032                                 flush_one_pair(p, options);
6033                 }
6034                 separator++;
6035         }
6036
6037         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6038                 dirstat_by_line = 1;
6039
6040         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6041             dirstat_by_line) {
6042                 struct diffstat_t diffstat;
6043
6044                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6045                 for (i = 0; i < q->nr; i++) {
6046                         struct diff_filepair *p = q->queue[i];
6047                         if (check_pair_status(p))
6048                                 diff_flush_stat(p, options, &diffstat);
6049                 }
6050                 if (output_format & DIFF_FORMAT_NUMSTAT)
6051                         show_numstat(&diffstat, options);
6052                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6053                         show_stats(&diffstat, options);
6054                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6055                         show_shortstats(&diffstat, options);
6056                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6057                         show_dirstat_by_line(&diffstat, options);
6058                 free_diffstat_info(&diffstat);
6059                 separator++;
6060         }
6061         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6062                 show_dirstat(options);
6063
6064         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6065                 for (i = 0; i < q->nr; i++) {
6066                         diff_summary(options, q->queue[i]);
6067                 }
6068                 separator++;
6069         }
6070
6071         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6072             options->flags.exit_with_status &&
6073             options->flags.diff_from_contents) {
6074                 /*
6075                  * run diff_flush_patch for the exit status. setting
6076                  * options->file to /dev/null should be safe, because we
6077                  * aren't supposed to produce any output anyway.
6078                  */
6079                 if (options->close_file)
6080                         fclose(options->file);
6081                 options->file = xfopen("/dev/null", "w");
6082                 options->close_file = 1;
6083                 options->color_moved = 0;
6084                 for (i = 0; i < q->nr; i++) {
6085                         struct diff_filepair *p = q->queue[i];
6086                         if (check_pair_status(p))
6087                                 diff_flush_patch(p, options);
6088                         if (options->found_changes)
6089                                 break;
6090                 }
6091         }
6092
6093         if (output_format & DIFF_FORMAT_PATCH) {
6094                 if (separator) {
6095                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6096                         if (options->stat_sep)
6097                                 /* attach patch instead of inline */
6098                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6099                                                  NULL, 0, 0);
6100                 }
6101
6102                 diff_flush_patch_all_file_pairs(options);
6103         }
6104
6105         if (output_format & DIFF_FORMAT_CALLBACK)
6106                 options->format_callback(q, options, options->format_callback_data);
6107
6108         for (i = 0; i < q->nr; i++)
6109                 diff_free_filepair(q->queue[i]);
6110 free_queue:
6111         free(q->queue);
6112         DIFF_QUEUE_CLEAR(q);
6113         if (options->close_file)
6114                 fclose(options->file);
6115
6116         /*
6117          * Report the content-level differences with HAS_CHANGES;
6118          * diff_addremove/diff_change does not set the bit when
6119          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6120          */
6121         if (options->flags.diff_from_contents) {
6122                 if (options->found_changes)
6123                         options->flags.has_changes = 1;
6124                 else
6125                         options->flags.has_changes = 0;
6126         }
6127 }
6128
6129 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6130 {
6131         return (((p->status == DIFF_STATUS_MODIFIED) &&
6132                  ((p->score &&
6133                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6134                   (!p->score &&
6135                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6136                 ((p->status != DIFF_STATUS_MODIFIED) &&
6137                  filter_bit_tst(p->status, options)));
6138 }
6139
6140 static void diffcore_apply_filter(struct diff_options *options)
6141 {
6142         int i;
6143         struct diff_queue_struct *q = &diff_queued_diff;
6144         struct diff_queue_struct outq;
6145
6146         DIFF_QUEUE_CLEAR(&outq);
6147
6148         if (!options->filter)
6149                 return;
6150
6151         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6152                 int found;
6153                 for (i = found = 0; !found && i < q->nr; i++) {
6154                         if (match_filter(options, q->queue[i]))
6155                                 found++;
6156                 }
6157                 if (found)
6158                         return;
6159
6160                 /* otherwise we will clear the whole queue
6161                  * by copying the empty outq at the end of this
6162                  * function, but first clear the current entries
6163                  * in the queue.
6164                  */
6165                 for (i = 0; i < q->nr; i++)
6166                         diff_free_filepair(q->queue[i]);
6167         }
6168         else {
6169                 /* Only the matching ones */
6170                 for (i = 0; i < q->nr; i++) {
6171                         struct diff_filepair *p = q->queue[i];
6172                         if (match_filter(options, p))
6173                                 diff_q(&outq, p);
6174                         else
6175                                 diff_free_filepair(p);
6176                 }
6177         }
6178         free(q->queue);
6179         *q = outq;
6180 }
6181
6182 /* Check whether two filespecs with the same mode and size are identical */
6183 static int diff_filespec_is_identical(struct repository *r,
6184                                       struct diff_filespec *one,
6185                                       struct diff_filespec *two)
6186 {
6187         if (S_ISGITLINK(one->mode))
6188                 return 0;
6189         if (diff_populate_filespec(r, one, 0))
6190                 return 0;
6191         if (diff_populate_filespec(r, two, 0))
6192                 return 0;
6193         return !memcmp(one->data, two->data, one->size);
6194 }
6195
6196 static int diff_filespec_check_stat_unmatch(struct repository *r,
6197                                             struct diff_filepair *p)
6198 {
6199         if (p->done_skip_stat_unmatch)
6200                 return p->skip_stat_unmatch_result;
6201
6202         p->done_skip_stat_unmatch = 1;
6203         p->skip_stat_unmatch_result = 0;
6204         /*
6205          * 1. Entries that come from stat info dirtiness
6206          *    always have both sides (iow, not create/delete),
6207          *    one side of the object name is unknown, with
6208          *    the same mode and size.  Keep the ones that
6209          *    do not match these criteria.  They have real
6210          *    differences.
6211          *
6212          * 2. At this point, the file is known to be modified,
6213          *    with the same mode and size, and the object
6214          *    name of one side is unknown.  Need to inspect
6215          *    the identical contents.
6216          */
6217         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6218             !DIFF_FILE_VALID(p->two) ||
6219             (p->one->oid_valid && p->two->oid_valid) ||
6220             (p->one->mode != p->two->mode) ||
6221             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6222             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6223             (p->one->size != p->two->size) ||
6224             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6225                 p->skip_stat_unmatch_result = 1;
6226         return p->skip_stat_unmatch_result;
6227 }
6228
6229 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6230 {
6231         int i;
6232         struct diff_queue_struct *q = &diff_queued_diff;
6233         struct diff_queue_struct outq;
6234         DIFF_QUEUE_CLEAR(&outq);
6235
6236         for (i = 0; i < q->nr; i++) {
6237                 struct diff_filepair *p = q->queue[i];
6238
6239                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6240                         diff_q(&outq, p);
6241                 else {
6242                         /*
6243                          * The caller can subtract 1 from skip_stat_unmatch
6244                          * to determine how many paths were dirty only
6245                          * due to stat info mismatch.
6246                          */
6247                         if (!diffopt->flags.no_index)
6248                                 diffopt->skip_stat_unmatch++;
6249                         diff_free_filepair(p);
6250                 }
6251         }
6252         free(q->queue);
6253         *q = outq;
6254 }
6255
6256 static int diffnamecmp(const void *a_, const void *b_)
6257 {
6258         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6259         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6260         const char *name_a, *name_b;
6261
6262         name_a = a->one ? a->one->path : a->two->path;
6263         name_b = b->one ? b->one->path : b->two->path;
6264         return strcmp(name_a, name_b);
6265 }
6266
6267 void diffcore_fix_diff_index(void)
6268 {
6269         struct diff_queue_struct *q = &diff_queued_diff;
6270         QSORT(q->queue, q->nr, diffnamecmp);
6271 }
6272
6273 void diffcore_std(struct diff_options *options)
6274 {
6275         /* NOTE please keep the following in sync with diff_tree_combined() */
6276         if (options->skip_stat_unmatch)
6277                 diffcore_skip_stat_unmatch(options);
6278         if (!options->found_follow) {
6279                 /* See try_to_follow_renames() in tree-diff.c */
6280                 if (options->break_opt != -1)
6281                         diffcore_break(options->repo,
6282                                        options->break_opt);
6283                 if (options->detect_rename)
6284                         diffcore_rename(options);
6285                 if (options->break_opt != -1)
6286                         diffcore_merge_broken();
6287         }
6288         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6289                 diffcore_pickaxe(options);
6290         if (options->orderfile)
6291                 diffcore_order(options->orderfile);
6292         if (!options->found_follow)
6293                 /* See try_to_follow_renames() in tree-diff.c */
6294                 diff_resolve_rename_copy();
6295         diffcore_apply_filter(options);
6296
6297         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6298                 options->flags.has_changes = 1;
6299         else
6300                 options->flags.has_changes = 0;
6301
6302         options->found_follow = 0;
6303 }
6304
6305 int diff_result_code(struct diff_options *opt, int status)
6306 {
6307         int result = 0;
6308
6309         diff_warn_rename_limit("diff.renameLimit",
6310                                opt->needed_rename_limit,
6311                                opt->degraded_cc_to_c);
6312         if (!opt->flags.exit_with_status &&
6313             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6314                 return status;
6315         if (opt->flags.exit_with_status &&
6316             opt->flags.has_changes)
6317                 result |= 01;
6318         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6319             opt->flags.check_failed)
6320                 result |= 02;
6321         return result;
6322 }
6323
6324 int diff_can_quit_early(struct diff_options *opt)
6325 {
6326         return (opt->flags.quick &&
6327                 !opt->filter &&
6328                 opt->flags.has_changes);
6329 }
6330
6331 /*
6332  * Shall changes to this submodule be ignored?
6333  *
6334  * Submodule changes can be configured to be ignored separately for each path,
6335  * but that configuration can be overridden from the command line.
6336  */
6337 static int is_submodule_ignored(const char *path, struct diff_options *options)
6338 {
6339         int ignored = 0;
6340         struct diff_flags orig_flags = options->flags;
6341         if (!options->flags.override_submodule_config)
6342                 set_diffopt_flags_from_submodule_config(options, path);
6343         if (options->flags.ignore_submodules)
6344                 ignored = 1;
6345         options->flags = orig_flags;
6346         return ignored;
6347 }
6348
6349 void diff_addremove(struct diff_options *options,
6350                     int addremove, unsigned mode,
6351                     const struct object_id *oid,
6352                     int oid_valid,
6353                     const char *concatpath, unsigned dirty_submodule)
6354 {
6355         struct diff_filespec *one, *two;
6356
6357         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6358                 return;
6359
6360         /* This may look odd, but it is a preparation for
6361          * feeding "there are unchanged files which should
6362          * not produce diffs, but when you are doing copy
6363          * detection you would need them, so here they are"
6364          * entries to the diff-core.  They will be prefixed
6365          * with something like '=' or '*' (I haven't decided
6366          * which but should not make any difference).
6367          * Feeding the same new and old to diff_change()
6368          * also has the same effect.
6369          * Before the final output happens, they are pruned after
6370          * merged into rename/copy pairs as appropriate.
6371          */
6372         if (options->flags.reverse_diff)
6373                 addremove = (addremove == '+' ? '-' :
6374                              addremove == '-' ? '+' : addremove);
6375
6376         if (options->prefix &&
6377             strncmp(concatpath, options->prefix, options->prefix_length))
6378                 return;
6379
6380         one = alloc_filespec(concatpath);
6381         two = alloc_filespec(concatpath);
6382
6383         if (addremove != '+')
6384                 fill_filespec(one, oid, oid_valid, mode);
6385         if (addremove != '-') {
6386                 fill_filespec(two, oid, oid_valid, mode);
6387                 two->dirty_submodule = dirty_submodule;
6388         }
6389
6390         diff_queue(&diff_queued_diff, one, two);
6391         if (!options->flags.diff_from_contents)
6392                 options->flags.has_changes = 1;
6393 }
6394
6395 void diff_change(struct diff_options *options,
6396                  unsigned old_mode, unsigned new_mode,
6397                  const struct object_id *old_oid,
6398                  const struct object_id *new_oid,
6399                  int old_oid_valid, int new_oid_valid,
6400                  const char *concatpath,
6401                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6402 {
6403         struct diff_filespec *one, *two;
6404         struct diff_filepair *p;
6405
6406         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6407             is_submodule_ignored(concatpath, options))
6408                 return;
6409
6410         if (options->flags.reverse_diff) {
6411                 SWAP(old_mode, new_mode);
6412                 SWAP(old_oid, new_oid);
6413                 SWAP(old_oid_valid, new_oid_valid);
6414                 SWAP(old_dirty_submodule, new_dirty_submodule);
6415         }
6416
6417         if (options->prefix &&
6418             strncmp(concatpath, options->prefix, options->prefix_length))
6419                 return;
6420
6421         one = alloc_filespec(concatpath);
6422         two = alloc_filespec(concatpath);
6423         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6424         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6425         one->dirty_submodule = old_dirty_submodule;
6426         two->dirty_submodule = new_dirty_submodule;
6427         p = diff_queue(&diff_queued_diff, one, two);
6428
6429         if (options->flags.diff_from_contents)
6430                 return;
6431
6432         if (options->flags.quick && options->skip_stat_unmatch &&
6433             !diff_filespec_check_stat_unmatch(options->repo, p))
6434                 return;
6435
6436         options->flags.has_changes = 1;
6437 }
6438
6439 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6440 {
6441         struct diff_filepair *pair;
6442         struct diff_filespec *one, *two;
6443
6444         if (options->prefix &&
6445             strncmp(path, options->prefix, options->prefix_length))
6446                 return NULL;
6447
6448         one = alloc_filespec(path);
6449         two = alloc_filespec(path);
6450         pair = diff_queue(&diff_queued_diff, one, two);
6451         pair->is_unmerged = 1;
6452         return pair;
6453 }
6454
6455 static char *run_textconv(struct repository *r,
6456                           const char *pgm,
6457                           struct diff_filespec *spec,
6458                           size_t *outsize)
6459 {
6460         struct diff_tempfile *temp;
6461         const char *argv[3];
6462         const char **arg = argv;
6463         struct child_process child = CHILD_PROCESS_INIT;
6464         struct strbuf buf = STRBUF_INIT;
6465         int err = 0;
6466
6467         temp = prepare_temp_file(r, spec->path, spec);
6468         *arg++ = pgm;
6469         *arg++ = temp->name;
6470         *arg = NULL;
6471
6472         child.use_shell = 1;
6473         child.argv = argv;
6474         child.out = -1;
6475         if (start_command(&child)) {
6476                 remove_tempfile();
6477                 return NULL;
6478         }
6479
6480         if (strbuf_read(&buf, child.out, 0) < 0)
6481                 err = error("error reading from textconv command '%s'", pgm);
6482         close(child.out);
6483
6484         if (finish_command(&child) || err) {
6485                 strbuf_release(&buf);
6486                 remove_tempfile();
6487                 return NULL;
6488         }
6489         remove_tempfile();
6490
6491         return strbuf_detach(&buf, outsize);
6492 }
6493
6494 size_t fill_textconv(struct repository *r,
6495                      struct userdiff_driver *driver,
6496                      struct diff_filespec *df,
6497                      char **outbuf)
6498 {
6499         size_t size;
6500
6501         if (!driver) {
6502                 if (!DIFF_FILE_VALID(df)) {
6503                         *outbuf = "";
6504                         return 0;
6505                 }
6506                 if (diff_populate_filespec(r, df, 0))
6507                         die("unable to read files to diff");
6508                 *outbuf = df->data;
6509                 return df->size;
6510         }
6511
6512         if (!driver->textconv)
6513                 BUG("fill_textconv called with non-textconv driver");
6514
6515         if (driver->textconv_cache && df->oid_valid) {
6516                 *outbuf = notes_cache_get(driver->textconv_cache,
6517                                           &df->oid,
6518                                           &size);
6519                 if (*outbuf)
6520                         return size;
6521         }
6522
6523         *outbuf = run_textconv(r, driver->textconv, df, &size);
6524         if (!*outbuf)
6525                 die("unable to read files to diff");
6526
6527         if (driver->textconv_cache && df->oid_valid) {
6528                 /* ignore errors, as we might be in a readonly repository */
6529                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6530                                 size);
6531                 /*
6532                  * we could save up changes and flush them all at the end,
6533                  * but we would need an extra call after all diffing is done.
6534                  * Since generating a cache entry is the slow path anyway,
6535                  * this extra overhead probably isn't a big deal.
6536                  */
6537                 notes_cache_write(driver->textconv_cache);
6538         }
6539
6540         return size;
6541 }
6542
6543 int textconv_object(struct repository *r,
6544                     const char *path,
6545                     unsigned mode,
6546                     const struct object_id *oid,
6547                     int oid_valid,
6548                     char **buf,
6549                     unsigned long *buf_size)
6550 {
6551         struct diff_filespec *df;
6552         struct userdiff_driver *textconv;
6553
6554         df = alloc_filespec(path);
6555         fill_filespec(df, oid, oid_valid, mode);
6556         textconv = get_textconv(r, df);
6557         if (!textconv) {
6558                 free_filespec(df);
6559                 return 0;
6560         }
6561
6562         *buf_size = fill_textconv(r, textconv, df, buf);
6563         free_filespec(df);
6564         return 1;
6565 }
6566
6567 void setup_diff_pager(struct diff_options *opt)
6568 {
6569         /*
6570          * If the user asked for our exit code, then either they want --quiet
6571          * or --exit-code. We should definitely not bother with a pager in the
6572          * former case, as we will generate no output. Since we still properly
6573          * report our exit code even when a pager is run, we _could_ run a
6574          * pager with --exit-code. But since we have not done so historically,
6575          * and because it is easy to find people oneline advising "git diff
6576          * --exit-code" in hooks and other scripts, we do not do so.
6577          */
6578         if (!opt->flags.exit_with_status &&
6579             check_pager_config("diff") != 0)
6580                 setup_pager();
6581 }