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(const char *reset,
1618                           struct emit_callback *ecbdata,
1619                           const char *line, int len)
1620 {
1621         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1622         if (new_blank_line_at_eof(ecbdata, line, len))
1623                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1624
1625         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1626 }
1627
1628 static void emit_del_line(const char *reset,
1629                           struct emit_callback *ecbdata,
1630                           const char *line, int len)
1631 {
1632         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1633         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1634 }
1635
1636 static void emit_context_line(const char *reset,
1637                               struct emit_callback *ecbdata,
1638                               const char *line, int len)
1639 {
1640         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1641         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1642 }
1643
1644 static void emit_hunk_header(struct emit_callback *ecbdata,
1645                              const char *line, int len)
1646 {
1647         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1648         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1649         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1650         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1651         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1652         static const char atat[2] = { '@', '@' };
1653         const char *cp, *ep;
1654         struct strbuf msgbuf = STRBUF_INIT;
1655         int org_len = len;
1656         int i = 1;
1657
1658         /*
1659          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1660          * it always is at least 10 bytes long.
1661          */
1662         if (len < 10 ||
1663             memcmp(line, atat, 2) ||
1664             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1665                 emit_diff_symbol(ecbdata->opt,
1666                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1667                 return;
1668         }
1669         ep += 2; /* skip over @@ */
1670
1671         /* The hunk header in fraginfo color */
1672         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1673                 strbuf_addstr(&msgbuf, reverse);
1674         strbuf_addstr(&msgbuf, frag);
1675         strbuf_add(&msgbuf, line, ep - line);
1676         strbuf_addstr(&msgbuf, reset);
1677
1678         /*
1679          * trailing "\r\n"
1680          */
1681         for ( ; i < 3; i++)
1682                 if (line[len - i] == '\r' || line[len - i] == '\n')
1683                         len--;
1684
1685         /* blank before the func header */
1686         for (cp = ep; ep - line < len; ep++)
1687                 if (*ep != ' ' && *ep != '\t')
1688                         break;
1689         if (ep != cp) {
1690                 strbuf_addstr(&msgbuf, context);
1691                 strbuf_add(&msgbuf, cp, ep - cp);
1692                 strbuf_addstr(&msgbuf, reset);
1693         }
1694
1695         if (ep < line + len) {
1696                 strbuf_addstr(&msgbuf, func);
1697                 strbuf_add(&msgbuf, ep, line + len - ep);
1698                 strbuf_addstr(&msgbuf, reset);
1699         }
1700
1701         strbuf_add(&msgbuf, line + len, org_len - len);
1702         strbuf_complete_line(&msgbuf);
1703         emit_diff_symbol(ecbdata->opt,
1704                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1705         strbuf_release(&msgbuf);
1706 }
1707
1708 static struct diff_tempfile *claim_diff_tempfile(void)
1709 {
1710         int i;
1711         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1712                 if (!diff_temp[i].name)
1713                         return diff_temp + i;
1714         BUG("diff is failing to clean up its tempfiles");
1715 }
1716
1717 static void remove_tempfile(void)
1718 {
1719         int i;
1720         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1721                 if (is_tempfile_active(diff_temp[i].tempfile))
1722                         delete_tempfile(&diff_temp[i].tempfile);
1723                 diff_temp[i].name = NULL;
1724         }
1725 }
1726
1727 static void add_line_count(struct strbuf *out, int count)
1728 {
1729         switch (count) {
1730         case 0:
1731                 strbuf_addstr(out, "0,0");
1732                 break;
1733         case 1:
1734                 strbuf_addstr(out, "1");
1735                 break;
1736         default:
1737                 strbuf_addf(out, "1,%d", count);
1738                 break;
1739         }
1740 }
1741
1742 static void emit_rewrite_lines(struct emit_callback *ecb,
1743                                int prefix, const char *data, int size)
1744 {
1745         const char *endp = NULL;
1746         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1747
1748         while (0 < size) {
1749                 int len;
1750
1751                 endp = memchr(data, '\n', size);
1752                 len = endp ? (endp - data + 1) : size;
1753                 if (prefix != '+') {
1754                         ecb->lno_in_preimage++;
1755                         emit_del_line(reset, ecb, data, len);
1756                 } else {
1757                         ecb->lno_in_postimage++;
1758                         emit_add_line(reset, ecb, data, len);
1759                 }
1760                 size -= len;
1761                 data += len;
1762         }
1763         if (!endp)
1764                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1765 }
1766
1767 static void emit_rewrite_diff(const char *name_a,
1768                               const char *name_b,
1769                               struct diff_filespec *one,
1770                               struct diff_filespec *two,
1771                               struct userdiff_driver *textconv_one,
1772                               struct userdiff_driver *textconv_two,
1773                               struct diff_options *o)
1774 {
1775         int lc_a, lc_b;
1776         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1777         const char *a_prefix, *b_prefix;
1778         char *data_one, *data_two;
1779         size_t size_one, size_two;
1780         struct emit_callback ecbdata;
1781         struct strbuf out = STRBUF_INIT;
1782
1783         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1784                 a_prefix = o->b_prefix;
1785                 b_prefix = o->a_prefix;
1786         } else {
1787                 a_prefix = o->a_prefix;
1788                 b_prefix = o->b_prefix;
1789         }
1790
1791         name_a += (*name_a == '/');
1792         name_b += (*name_b == '/');
1793
1794         strbuf_reset(&a_name);
1795         strbuf_reset(&b_name);
1796         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1797         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1798
1799         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1800         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1801
1802         memset(&ecbdata, 0, sizeof(ecbdata));
1803         ecbdata.color_diff = want_color(o->use_color);
1804         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1805         ecbdata.opt = o;
1806         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1807                 mmfile_t mf1, mf2;
1808                 mf1.ptr = (char *)data_one;
1809                 mf2.ptr = (char *)data_two;
1810                 mf1.size = size_one;
1811                 mf2.size = size_two;
1812                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1813         }
1814         ecbdata.lno_in_preimage = 1;
1815         ecbdata.lno_in_postimage = 1;
1816
1817         lc_a = count_lines(data_one, size_one);
1818         lc_b = count_lines(data_two, size_two);
1819
1820         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1821                          a_name.buf, a_name.len, 0);
1822         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1823                          b_name.buf, b_name.len, 0);
1824
1825         strbuf_addstr(&out, "@@ -");
1826         if (!o->irreversible_delete)
1827                 add_line_count(&out, lc_a);
1828         else
1829                 strbuf_addstr(&out, "?,?");
1830         strbuf_addstr(&out, " +");
1831         add_line_count(&out, lc_b);
1832         strbuf_addstr(&out, " @@\n");
1833         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1834         strbuf_release(&out);
1835
1836         if (lc_a && !o->irreversible_delete)
1837                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1838         if (lc_b)
1839                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1840         if (textconv_one)
1841                 free((char *)data_one);
1842         if (textconv_two)
1843                 free((char *)data_two);
1844 }
1845
1846 struct diff_words_buffer {
1847         mmfile_t text;
1848         unsigned long alloc;
1849         struct diff_words_orig {
1850                 const char *begin, *end;
1851         } *orig;
1852         int orig_nr, orig_alloc;
1853 };
1854
1855 static void diff_words_append(char *line, unsigned long len,
1856                 struct diff_words_buffer *buffer)
1857 {
1858         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1859         line++;
1860         len--;
1861         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1862         buffer->text.size += len;
1863         buffer->text.ptr[buffer->text.size] = '\0';
1864 }
1865
1866 struct diff_words_style_elem {
1867         const char *prefix;
1868         const char *suffix;
1869         const char *color; /* NULL; filled in by the setup code if
1870                             * color is enabled */
1871 };
1872
1873 struct diff_words_style {
1874         enum diff_words_type type;
1875         struct diff_words_style_elem new_word, old_word, ctx;
1876         const char *newline;
1877 };
1878
1879 static struct diff_words_style diff_words_styles[] = {
1880         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1881         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1882         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1883 };
1884
1885 struct diff_words_data {
1886         struct diff_words_buffer minus, plus;
1887         const char *current_plus;
1888         int last_minus;
1889         struct diff_options *opt;
1890         regex_t *word_regex;
1891         enum diff_words_type type;
1892         struct diff_words_style *style;
1893 };
1894
1895 static int fn_out_diff_words_write_helper(struct diff_options *o,
1896                                           struct diff_words_style_elem *st_el,
1897                                           const char *newline,
1898                                           size_t count, const char *buf)
1899 {
1900         int print = 0;
1901         struct strbuf sb = STRBUF_INIT;
1902
1903         while (count) {
1904                 char *p = memchr(buf, '\n', count);
1905                 if (print)
1906                         strbuf_addstr(&sb, diff_line_prefix(o));
1907
1908                 if (p != buf) {
1909                         const char *reset = st_el->color && *st_el->color ?
1910                                             GIT_COLOR_RESET : NULL;
1911                         if (st_el->color && *st_el->color)
1912                                 strbuf_addstr(&sb, st_el->color);
1913                         strbuf_addstr(&sb, st_el->prefix);
1914                         strbuf_add(&sb, buf, p ? p - buf : count);
1915                         strbuf_addstr(&sb, st_el->suffix);
1916                         if (reset)
1917                                 strbuf_addstr(&sb, reset);
1918                 }
1919                 if (!p)
1920                         goto out;
1921
1922                 strbuf_addstr(&sb, newline);
1923                 count -= p + 1 - buf;
1924                 buf = p + 1;
1925                 print = 1;
1926                 if (count) {
1927                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1928                                          sb.buf, sb.len, 0);
1929                         strbuf_reset(&sb);
1930                 }
1931         }
1932
1933 out:
1934         if (sb.len)
1935                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1936                                  sb.buf, sb.len, 0);
1937         strbuf_release(&sb);
1938         return 0;
1939 }
1940
1941 /*
1942  * '--color-words' algorithm can be described as:
1943  *
1944  *   1. collect the minus/plus lines of a diff hunk, divided into
1945  *      minus-lines and plus-lines;
1946  *
1947  *   2. break both minus-lines and plus-lines into words and
1948  *      place them into two mmfile_t with one word for each line;
1949  *
1950  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1951  *
1952  * And for the common parts of the both file, we output the plus side text.
1953  * diff_words->current_plus is used to trace the current position of the plus file
1954  * which printed. diff_words->last_minus is used to trace the last minus word
1955  * printed.
1956  *
1957  * For '--graph' to work with '--color-words', we need to output the graph prefix
1958  * on each line of color words output. Generally, there are two conditions on
1959  * which we should output the prefix.
1960  *
1961  *   1. diff_words->last_minus == 0 &&
1962  *      diff_words->current_plus == diff_words->plus.text.ptr
1963  *
1964  *      that is: the plus text must start as a new line, and if there is no minus
1965  *      word printed, a graph prefix must be printed.
1966  *
1967  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1968  *      *(diff_words->current_plus - 1) == '\n'
1969  *
1970  *      that is: a graph prefix must be printed following a '\n'
1971  */
1972 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1973 {
1974         if ((diff_words->last_minus == 0 &&
1975                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1976                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1977                 *(diff_words->current_plus - 1) == '\n')) {
1978                 return 1;
1979         } else {
1980                 return 0;
1981         }
1982 }
1983
1984 static void fn_out_diff_words_aux(void *priv,
1985                                   long minus_first, long minus_len,
1986                                   long plus_first, long plus_len,
1987                                   const char *func, long funclen)
1988 {
1989         struct diff_words_data *diff_words = priv;
1990         struct diff_words_style *style = diff_words->style;
1991         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1992         struct diff_options *opt = diff_words->opt;
1993         const char *line_prefix;
1994
1995         assert(opt);
1996         line_prefix = diff_line_prefix(opt);
1997
1998         /* POSIX requires that first be decremented by one if len == 0... */
1999         if (minus_len) {
2000                 minus_begin = diff_words->minus.orig[minus_first].begin;
2001                 minus_end =
2002                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2003         } else
2004                 minus_begin = minus_end =
2005                         diff_words->minus.orig[minus_first].end;
2006
2007         if (plus_len) {
2008                 plus_begin = diff_words->plus.orig[plus_first].begin;
2009                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2010         } else
2011                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2012
2013         if (color_words_output_graph_prefix(diff_words)) {
2014                 fputs(line_prefix, diff_words->opt->file);
2015         }
2016         if (diff_words->current_plus != plus_begin) {
2017                 fn_out_diff_words_write_helper(diff_words->opt,
2018                                 &style->ctx, style->newline,
2019                                 plus_begin - diff_words->current_plus,
2020                                 diff_words->current_plus);
2021         }
2022         if (minus_begin != minus_end) {
2023                 fn_out_diff_words_write_helper(diff_words->opt,
2024                                 &style->old_word, style->newline,
2025                                 minus_end - minus_begin, minus_begin);
2026         }
2027         if (plus_begin != plus_end) {
2028                 fn_out_diff_words_write_helper(diff_words->opt,
2029                                 &style->new_word, style->newline,
2030                                 plus_end - plus_begin, plus_begin);
2031         }
2032
2033         diff_words->current_plus = plus_end;
2034         diff_words->last_minus = minus_first;
2035 }
2036
2037 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2038 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2039                 int *begin, int *end)
2040 {
2041         if (word_regex && *begin < buffer->size) {
2042                 regmatch_t match[1];
2043                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2044                                  buffer->size - *begin, 1, match, 0)) {
2045                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2046                                         '\n', match[0].rm_eo - match[0].rm_so);
2047                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2048                         *begin += match[0].rm_so;
2049                         return *begin >= *end;
2050                 }
2051                 return -1;
2052         }
2053
2054         /* find the next word */
2055         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2056                 (*begin)++;
2057         if (*begin >= buffer->size)
2058                 return -1;
2059
2060         /* find the end of the word */
2061         *end = *begin + 1;
2062         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2063                 (*end)++;
2064
2065         return 0;
2066 }
2067
2068 /*
2069  * This function splits the words in buffer->text, stores the list with
2070  * newline separator into out, and saves the offsets of the original words
2071  * in buffer->orig.
2072  */
2073 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2074                 regex_t *word_regex)
2075 {
2076         int i, j;
2077         long alloc = 0;
2078
2079         out->size = 0;
2080         out->ptr = NULL;
2081
2082         /* fake an empty "0th" word */
2083         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2084         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2085         buffer->orig_nr = 1;
2086
2087         for (i = 0; i < buffer->text.size; i++) {
2088                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2089                         return;
2090
2091                 /* store original boundaries */
2092                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2093                                 buffer->orig_alloc);
2094                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2095                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2096                 buffer->orig_nr++;
2097
2098                 /* store one word */
2099                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2100                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2101                 out->ptr[out->size + j - i] = '\n';
2102                 out->size += j - i + 1;
2103
2104                 i = j - 1;
2105         }
2106 }
2107
2108 /* this executes the word diff on the accumulated buffers */
2109 static void diff_words_show(struct diff_words_data *diff_words)
2110 {
2111         xpparam_t xpp;
2112         xdemitconf_t xecfg;
2113         mmfile_t minus, plus;
2114         struct diff_words_style *style = diff_words->style;
2115
2116         struct diff_options *opt = diff_words->opt;
2117         const char *line_prefix;
2118
2119         assert(opt);
2120         line_prefix = diff_line_prefix(opt);
2121
2122         /* special case: only removal */
2123         if (!diff_words->plus.text.size) {
2124                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2125                                  line_prefix, strlen(line_prefix), 0);
2126                 fn_out_diff_words_write_helper(diff_words->opt,
2127                         &style->old_word, style->newline,
2128                         diff_words->minus.text.size,
2129                         diff_words->minus.text.ptr);
2130                 diff_words->minus.text.size = 0;
2131                 return;
2132         }
2133
2134         diff_words->current_plus = diff_words->plus.text.ptr;
2135         diff_words->last_minus = 0;
2136
2137         memset(&xpp, 0, sizeof(xpp));
2138         memset(&xecfg, 0, sizeof(xecfg));
2139         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2140         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2141         xpp.flags = 0;
2142         /* as only the hunk header will be parsed, we need a 0-context */
2143         xecfg.ctxlen = 0;
2144         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2145                           diff_words, &xpp, &xecfg))
2146                 die("unable to generate word diff");
2147         free(minus.ptr);
2148         free(plus.ptr);
2149         if (diff_words->current_plus != diff_words->plus.text.ptr +
2150                         diff_words->plus.text.size) {
2151                 if (color_words_output_graph_prefix(diff_words))
2152                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2153                                          line_prefix, strlen(line_prefix), 0);
2154                 fn_out_diff_words_write_helper(diff_words->opt,
2155                         &style->ctx, style->newline,
2156                         diff_words->plus.text.ptr + diff_words->plus.text.size
2157                         - diff_words->current_plus, diff_words->current_plus);
2158         }
2159         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2160 }
2161
2162 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2163 static void diff_words_flush(struct emit_callback *ecbdata)
2164 {
2165         struct diff_options *wo = ecbdata->diff_words->opt;
2166
2167         if (ecbdata->diff_words->minus.text.size ||
2168             ecbdata->diff_words->plus.text.size)
2169                 diff_words_show(ecbdata->diff_words);
2170
2171         if (wo->emitted_symbols) {
2172                 struct diff_options *o = ecbdata->opt;
2173                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2174                 int i;
2175
2176                 /*
2177                  * NEEDSWORK:
2178                  * Instead of appending each, concat all words to a line?
2179                  */
2180                 for (i = 0; i < wol->nr; i++)
2181                         append_emitted_diff_symbol(o, &wol->buf[i]);
2182
2183                 for (i = 0; i < wol->nr; i++)
2184                         free((void *)wol->buf[i].line);
2185
2186                 wol->nr = 0;
2187         }
2188 }
2189
2190 static void diff_filespec_load_driver(struct diff_filespec *one,
2191                                       struct index_state *istate)
2192 {
2193         /* Use already-loaded driver */
2194         if (one->driver)
2195                 return;
2196
2197         if (S_ISREG(one->mode))
2198                 one->driver = userdiff_find_by_path(istate, one->path);
2199
2200         /* Fallback to default settings */
2201         if (!one->driver)
2202                 one->driver = userdiff_find_by_name("default");
2203 }
2204
2205 static const char *userdiff_word_regex(struct diff_filespec *one,
2206                                        struct index_state *istate)
2207 {
2208         diff_filespec_load_driver(one, istate);
2209         return one->driver->word_regex;
2210 }
2211
2212 static void init_diff_words_data(struct emit_callback *ecbdata,
2213                                  struct diff_options *orig_opts,
2214                                  struct diff_filespec *one,
2215                                  struct diff_filespec *two)
2216 {
2217         int i;
2218         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2219         memcpy(o, orig_opts, sizeof(struct diff_options));
2220
2221         ecbdata->diff_words =
2222                 xcalloc(1, sizeof(struct diff_words_data));
2223         ecbdata->diff_words->type = o->word_diff;
2224         ecbdata->diff_words->opt = o;
2225
2226         if (orig_opts->emitted_symbols)
2227                 o->emitted_symbols =
2228                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2229
2230         if (!o->word_regex)
2231                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2232         if (!o->word_regex)
2233                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2234         if (!o->word_regex)
2235                 o->word_regex = diff_word_regex_cfg;
2236         if (o->word_regex) {
2237                 ecbdata->diff_words->word_regex = (regex_t *)
2238                         xmalloc(sizeof(regex_t));
2239                 if (regcomp(ecbdata->diff_words->word_regex,
2240                             o->word_regex,
2241                             REG_EXTENDED | REG_NEWLINE))
2242                         die("invalid regular expression: %s",
2243                             o->word_regex);
2244         }
2245         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2246                 if (o->word_diff == diff_words_styles[i].type) {
2247                         ecbdata->diff_words->style =
2248                                 &diff_words_styles[i];
2249                         break;
2250                 }
2251         }
2252         if (want_color(o->use_color)) {
2253                 struct diff_words_style *st = ecbdata->diff_words->style;
2254                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2255                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2256                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2257         }
2258 }
2259
2260 static void free_diff_words_data(struct emit_callback *ecbdata)
2261 {
2262         if (ecbdata->diff_words) {
2263                 diff_words_flush(ecbdata);
2264                 free (ecbdata->diff_words->opt->emitted_symbols);
2265                 free (ecbdata->diff_words->opt);
2266                 free (ecbdata->diff_words->minus.text.ptr);
2267                 free (ecbdata->diff_words->minus.orig);
2268                 free (ecbdata->diff_words->plus.text.ptr);
2269                 free (ecbdata->diff_words->plus.orig);
2270                 if (ecbdata->diff_words->word_regex) {
2271                         regfree(ecbdata->diff_words->word_regex);
2272                         free(ecbdata->diff_words->word_regex);
2273                 }
2274                 FREE_AND_NULL(ecbdata->diff_words);
2275         }
2276 }
2277
2278 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2279 {
2280         if (want_color(diff_use_color))
2281                 return diff_colors[ix];
2282         return "";
2283 }
2284
2285 const char *diff_line_prefix(struct diff_options *opt)
2286 {
2287         struct strbuf *msgbuf;
2288         if (!opt->output_prefix)
2289                 return "";
2290
2291         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2292         return msgbuf->buf;
2293 }
2294
2295 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2296 {
2297         const char *cp;
2298         unsigned long allot;
2299         size_t l = len;
2300
2301         cp = line;
2302         allot = l;
2303         while (0 < l) {
2304                 (void) utf8_width(&cp, &l);
2305                 if (!cp)
2306                         break; /* truncated in the middle? */
2307         }
2308         return allot - l;
2309 }
2310
2311 static void find_lno(const char *line, struct emit_callback *ecbdata)
2312 {
2313         const char *p;
2314         ecbdata->lno_in_preimage = 0;
2315         ecbdata->lno_in_postimage = 0;
2316         p = strchr(line, '-');
2317         if (!p)
2318                 return; /* cannot happen */
2319         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2320         p = strchr(p, '+');
2321         if (!p)
2322                 return; /* cannot happen */
2323         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2324 }
2325
2326 static void fn_out_consume(void *priv, char *line, unsigned long len)
2327 {
2328         struct emit_callback *ecbdata = priv;
2329         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2330         struct diff_options *o = ecbdata->opt;
2331
2332         o->found_changes = 1;
2333
2334         if (ecbdata->header) {
2335                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2336                                  ecbdata->header->buf, ecbdata->header->len, 0);
2337                 strbuf_reset(ecbdata->header);
2338                 ecbdata->header = NULL;
2339         }
2340
2341         if (ecbdata->label_path[0]) {
2342                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2343                                  ecbdata->label_path[0],
2344                                  strlen(ecbdata->label_path[0]), 0);
2345                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2346                                  ecbdata->label_path[1],
2347                                  strlen(ecbdata->label_path[1]), 0);
2348                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2349         }
2350
2351         if (diff_suppress_blank_empty
2352             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2353                 line[0] = '\n';
2354                 len = 1;
2355         }
2356
2357         if (line[0] == '@') {
2358                 if (ecbdata->diff_words)
2359                         diff_words_flush(ecbdata);
2360                 len = sane_truncate_line(ecbdata, line, len);
2361                 find_lno(line, ecbdata);
2362                 emit_hunk_header(ecbdata, line, len);
2363                 return;
2364         }
2365
2366         if (ecbdata->diff_words) {
2367                 enum diff_symbol s =
2368                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2369                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2370                 if (line[0] == '-') {
2371                         diff_words_append(line, len,
2372                                           &ecbdata->diff_words->minus);
2373                         return;
2374                 } else if (line[0] == '+') {
2375                         diff_words_append(line, len,
2376                                           &ecbdata->diff_words->plus);
2377                         return;
2378                 } else if (starts_with(line, "\\ ")) {
2379                         /*
2380                          * Eat the "no newline at eof" marker as if we
2381                          * saw a "+" or "-" line with nothing on it,
2382                          * and return without diff_words_flush() to
2383                          * defer processing. If this is the end of
2384                          * preimage, more "+" lines may come after it.
2385                          */
2386                         return;
2387                 }
2388                 diff_words_flush(ecbdata);
2389                 emit_diff_symbol(o, s, line, len, 0);
2390                 return;
2391         }
2392
2393         switch (line[0]) {
2394         case '+':
2395                 ecbdata->lno_in_postimage++;
2396                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2397                 break;
2398         case '-':
2399                 ecbdata->lno_in_preimage++;
2400                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2401                 break;
2402         case ' ':
2403                 ecbdata->lno_in_postimage++;
2404                 ecbdata->lno_in_preimage++;
2405                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2406                 break;
2407         default:
2408                 /* incomplete line at the end */
2409                 ecbdata->lno_in_preimage++;
2410                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2411                                  line, len, 0);
2412                 break;
2413         }
2414 }
2415
2416 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2417 {
2418         const char *old_name = a;
2419         const char *new_name = b;
2420         int pfx_length, sfx_length;
2421         int pfx_adjust_for_slash;
2422         int len_a = strlen(a);
2423         int len_b = strlen(b);
2424         int a_midlen, b_midlen;
2425         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2426         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2427
2428         if (qlen_a || qlen_b) {
2429                 quote_c_style(a, name, NULL, 0);
2430                 strbuf_addstr(name, " => ");
2431                 quote_c_style(b, name, NULL, 0);
2432                 return;
2433         }
2434
2435         /* Find common prefix */
2436         pfx_length = 0;
2437         while (*old_name && *new_name && *old_name == *new_name) {
2438                 if (*old_name == '/')
2439                         pfx_length = old_name - a + 1;
2440                 old_name++;
2441                 new_name++;
2442         }
2443
2444         /* Find common suffix */
2445         old_name = a + len_a;
2446         new_name = b + len_b;
2447         sfx_length = 0;
2448         /*
2449          * If there is a common prefix, it must end in a slash.  In
2450          * that case we let this loop run 1 into the prefix to see the
2451          * same slash.
2452          *
2453          * If there is no common prefix, we cannot do this as it would
2454          * underrun the input strings.
2455          */
2456         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2457         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2458                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2459                *old_name == *new_name) {
2460                 if (*old_name == '/')
2461                         sfx_length = len_a - (old_name - a);
2462                 old_name--;
2463                 new_name--;
2464         }
2465
2466         /*
2467          * pfx{mid-a => mid-b}sfx
2468          * {pfx-a => pfx-b}sfx
2469          * pfx{sfx-a => sfx-b}
2470          * name-a => name-b
2471          */
2472         a_midlen = len_a - pfx_length - sfx_length;
2473         b_midlen = len_b - pfx_length - sfx_length;
2474         if (a_midlen < 0)
2475                 a_midlen = 0;
2476         if (b_midlen < 0)
2477                 b_midlen = 0;
2478
2479         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2480         if (pfx_length + sfx_length) {
2481                 strbuf_add(name, a, pfx_length);
2482                 strbuf_addch(name, '{');
2483         }
2484         strbuf_add(name, a + pfx_length, a_midlen);
2485         strbuf_addstr(name, " => ");
2486         strbuf_add(name, b + pfx_length, b_midlen);
2487         if (pfx_length + sfx_length) {
2488                 strbuf_addch(name, '}');
2489                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2490         }
2491 }
2492
2493 struct diffstat_t {
2494         int nr;
2495         int alloc;
2496         struct diffstat_file {
2497                 char *from_name;
2498                 char *name;
2499                 char *print_name;
2500                 const char *comments;
2501                 unsigned is_unmerged:1;
2502                 unsigned is_binary:1;
2503                 unsigned is_renamed:1;
2504                 unsigned is_interesting:1;
2505                 uintmax_t added, deleted;
2506         } **files;
2507 };
2508
2509 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2510                                           const char *name_a,
2511                                           const char *name_b)
2512 {
2513         struct diffstat_file *x;
2514         x = xcalloc(1, sizeof(*x));
2515         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2516         diffstat->files[diffstat->nr++] = x;
2517         if (name_b) {
2518                 x->from_name = xstrdup(name_a);
2519                 x->name = xstrdup(name_b);
2520                 x->is_renamed = 1;
2521         }
2522         else {
2523                 x->from_name = NULL;
2524                 x->name = xstrdup(name_a);
2525         }
2526         return x;
2527 }
2528
2529 static void diffstat_consume(void *priv, char *line, unsigned long len)
2530 {
2531         struct diffstat_t *diffstat = priv;
2532         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2533
2534         if (line[0] == '+')
2535                 x->added++;
2536         else if (line[0] == '-')
2537                 x->deleted++;
2538 }
2539
2540 const char mime_boundary_leader[] = "------------";
2541
2542 static int scale_linear(int it, int width, int max_change)
2543 {
2544         if (!it)
2545                 return 0;
2546         /*
2547          * make sure that at least one '-' or '+' is printed if
2548          * there is any change to this path. The easiest way is to
2549          * scale linearly as if the alloted width is one column shorter
2550          * than it is, and then add 1 to the result.
2551          */
2552         return 1 + (it * (width - 1) / max_change);
2553 }
2554
2555 static void show_graph(struct strbuf *out, char ch, int cnt,
2556                        const char *set, const char *reset)
2557 {
2558         if (cnt <= 0)
2559                 return;
2560         strbuf_addstr(out, set);
2561         strbuf_addchars(out, ch, cnt);
2562         strbuf_addstr(out, reset);
2563 }
2564
2565 static void fill_print_name(struct diffstat_file *file)
2566 {
2567         struct strbuf pname = STRBUF_INIT;
2568
2569         if (file->print_name)
2570                 return;
2571
2572         if (file->is_renamed)
2573                 pprint_rename(&pname, file->from_name, file->name);
2574         else
2575                 quote_c_style(file->name, &pname, NULL, 0);
2576
2577         if (file->comments)
2578                 strbuf_addf(&pname, " (%s)", file->comments);
2579
2580         file->print_name = strbuf_detach(&pname, NULL);
2581 }
2582
2583 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2584                 int files, int insertions, int deletions)
2585 {
2586         struct strbuf sb = STRBUF_INIT;
2587
2588         if (!files) {
2589                 assert(insertions == 0 && deletions == 0);
2590                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2591                                  NULL, 0, 0);
2592                 return;
2593         }
2594
2595         strbuf_addf(&sb,
2596                     (files == 1) ? " %d file changed" : " %d files changed",
2597                     files);
2598
2599         /*
2600          * For binary diff, the caller may want to print "x files
2601          * changed" with insertions == 0 && deletions == 0.
2602          *
2603          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2604          * is probably less confusing (i.e skip over "2 files changed
2605          * but nothing about added/removed lines? Is this a bug in Git?").
2606          */
2607         if (insertions || deletions == 0) {
2608                 strbuf_addf(&sb,
2609                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2610                             insertions);
2611         }
2612
2613         if (deletions || insertions == 0) {
2614                 strbuf_addf(&sb,
2615                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2616                             deletions);
2617         }
2618         strbuf_addch(&sb, '\n');
2619         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2620                          sb.buf, sb.len, 0);
2621         strbuf_release(&sb);
2622 }
2623
2624 void print_stat_summary(FILE *fp, int files,
2625                         int insertions, int deletions)
2626 {
2627         struct diff_options o;
2628         memset(&o, 0, sizeof(o));
2629         o.file = fp;
2630
2631         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2632 }
2633
2634 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2635 {
2636         int i, len, add, del, adds = 0, dels = 0;
2637         uintmax_t max_change = 0, max_len = 0;
2638         int total_files = data->nr, count;
2639         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2640         const char *reset, *add_c, *del_c;
2641         int extra_shown = 0;
2642         const char *line_prefix = diff_line_prefix(options);
2643         struct strbuf out = STRBUF_INIT;
2644
2645         if (data->nr == 0)
2646                 return;
2647
2648         count = options->stat_count ? options->stat_count : data->nr;
2649
2650         reset = diff_get_color_opt(options, DIFF_RESET);
2651         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2652         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2653
2654         /*
2655          * Find the longest filename and max number of changes
2656          */
2657         for (i = 0; (i < count) && (i < data->nr); i++) {
2658                 struct diffstat_file *file = data->files[i];
2659                 uintmax_t change = file->added + file->deleted;
2660
2661                 if (!file->is_interesting && (change == 0)) {
2662                         count++; /* not shown == room for one more */
2663                         continue;
2664                 }
2665                 fill_print_name(file);
2666                 len = strlen(file->print_name);
2667                 if (max_len < len)
2668                         max_len = len;
2669
2670                 if (file->is_unmerged) {
2671                         /* "Unmerged" is 8 characters */
2672                         bin_width = bin_width < 8 ? 8 : bin_width;
2673                         continue;
2674                 }
2675                 if (file->is_binary) {
2676                         /* "Bin XXX -> YYY bytes" */
2677                         int w = 14 + decimal_width(file->added)
2678                                 + decimal_width(file->deleted);
2679                         bin_width = bin_width < w ? w : bin_width;
2680                         /* Display change counts aligned with "Bin" */
2681                         number_width = 3;
2682                         continue;
2683                 }
2684
2685                 if (max_change < change)
2686                         max_change = change;
2687         }
2688         count = i; /* where we can stop scanning in data->files[] */
2689
2690         /*
2691          * We have width = stat_width or term_columns() columns total.
2692          * We want a maximum of min(max_len, stat_name_width) for the name part.
2693          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2694          * We also need 1 for " " and 4 + decimal_width(max_change)
2695          * for " | NNNN " and one the empty column at the end, altogether
2696          * 6 + decimal_width(max_change).
2697          *
2698          * If there's not enough space, we will use the smaller of
2699          * stat_name_width (if set) and 5/8*width for the filename,
2700          * and the rest for constant elements + graph part, but no more
2701          * than stat_graph_width for the graph part.
2702          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2703          * for the standard terminal size).
2704          *
2705          * In other words: stat_width limits the maximum width, and
2706          * stat_name_width fixes the maximum width of the filename,
2707          * and is also used to divide available columns if there
2708          * aren't enough.
2709          *
2710          * Binary files are displayed with "Bin XXX -> YYY bytes"
2711          * instead of the change count and graph. This part is treated
2712          * similarly to the graph part, except that it is not
2713          * "scaled". If total width is too small to accommodate the
2714          * guaranteed minimum width of the filename part and the
2715          * separators and this message, this message will "overflow"
2716          * making the line longer than the maximum width.
2717          */
2718
2719         if (options->stat_width == -1)
2720                 width = term_columns() - strlen(line_prefix);
2721         else
2722                 width = options->stat_width ? options->stat_width : 80;
2723         number_width = decimal_width(max_change) > number_width ?
2724                 decimal_width(max_change) : number_width;
2725
2726         if (options->stat_graph_width == -1)
2727                 options->stat_graph_width = diff_stat_graph_width;
2728
2729         /*
2730          * Guarantee 3/8*16==6 for the graph part
2731          * and 5/8*16==10 for the filename part
2732          */
2733         if (width < 16 + 6 + number_width)
2734                 width = 16 + 6 + number_width;
2735
2736         /*
2737          * First assign sizes that are wanted, ignoring available width.
2738          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2739          * starting from "XXX" should fit in graph_width.
2740          */
2741         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2742         if (options->stat_graph_width &&
2743             options->stat_graph_width < graph_width)
2744                 graph_width = options->stat_graph_width;
2745
2746         name_width = (options->stat_name_width > 0 &&
2747                       options->stat_name_width < max_len) ?
2748                 options->stat_name_width : max_len;
2749
2750         /*
2751          * Adjust adjustable widths not to exceed maximum width
2752          */
2753         if (name_width + number_width + 6 + graph_width > width) {
2754                 if (graph_width > width * 3/8 - number_width - 6) {
2755                         graph_width = width * 3/8 - number_width - 6;
2756                         if (graph_width < 6)
2757                                 graph_width = 6;
2758                 }
2759
2760                 if (options->stat_graph_width &&
2761                     graph_width > options->stat_graph_width)
2762                         graph_width = options->stat_graph_width;
2763                 if (name_width > width - number_width - 6 - graph_width)
2764                         name_width = width - number_width - 6 - graph_width;
2765                 else
2766                         graph_width = width - number_width - 6 - name_width;
2767         }
2768
2769         /*
2770          * From here name_width is the width of the name area,
2771          * and graph_width is the width of the graph area.
2772          * max_change is used to scale graph properly.
2773          */
2774         for (i = 0; i < count; i++) {
2775                 const char *prefix = "";
2776                 struct diffstat_file *file = data->files[i];
2777                 char *name = file->print_name;
2778                 uintmax_t added = file->added;
2779                 uintmax_t deleted = file->deleted;
2780                 int name_len;
2781
2782                 if (!file->is_interesting && (added + deleted == 0))
2783                         continue;
2784
2785                 /*
2786                  * "scale" the filename
2787                  */
2788                 len = name_width;
2789                 name_len = strlen(name);
2790                 if (name_width < name_len) {
2791                         char *slash;
2792                         prefix = "...";
2793                         len -= 3;
2794                         name += name_len - len;
2795                         slash = strchr(name, '/');
2796                         if (slash)
2797                                 name = slash;
2798                 }
2799
2800                 if (file->is_binary) {
2801                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2802                         strbuf_addf(&out, " %*s", number_width, "Bin");
2803                         if (!added && !deleted) {
2804                                 strbuf_addch(&out, '\n');
2805                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2806                                                  out.buf, out.len, 0);
2807                                 strbuf_reset(&out);
2808                                 continue;
2809                         }
2810                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2811                                 del_c, deleted, reset);
2812                         strbuf_addstr(&out, " -> ");
2813                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2814                                 add_c, added, reset);
2815                         strbuf_addstr(&out, " bytes\n");
2816                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2817                                          out.buf, out.len, 0);
2818                         strbuf_reset(&out);
2819                         continue;
2820                 }
2821                 else if (file->is_unmerged) {
2822                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2823                         strbuf_addstr(&out, " Unmerged\n");
2824                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2825                                          out.buf, out.len, 0);
2826                         strbuf_reset(&out);
2827                         continue;
2828                 }
2829
2830                 /*
2831                  * scale the add/delete
2832                  */
2833                 add = added;
2834                 del = deleted;
2835
2836                 if (graph_width <= max_change) {
2837                         int total = scale_linear(add + del, graph_width, max_change);
2838                         if (total < 2 && add && del)
2839                                 /* width >= 2 due to the sanity check */
2840                                 total = 2;
2841                         if (add < del) {
2842                                 add = scale_linear(add, graph_width, max_change);
2843                                 del = total - add;
2844                         } else {
2845                                 del = scale_linear(del, graph_width, max_change);
2846                                 add = total - del;
2847                         }
2848                 }
2849                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2850                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2851                         number_width, added + deleted,
2852                         added + deleted ? " " : "");
2853                 show_graph(&out, '+', add, add_c, reset);
2854                 show_graph(&out, '-', del, del_c, reset);
2855                 strbuf_addch(&out, '\n');
2856                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2857                                  out.buf, out.len, 0);
2858                 strbuf_reset(&out);
2859         }
2860
2861         for (i = 0; i < data->nr; i++) {
2862                 struct diffstat_file *file = data->files[i];
2863                 uintmax_t added = file->added;
2864                 uintmax_t deleted = file->deleted;
2865
2866                 if (file->is_unmerged ||
2867                     (!file->is_interesting && (added + deleted == 0))) {
2868                         total_files--;
2869                         continue;
2870                 }
2871
2872                 if (!file->is_binary) {
2873                         adds += added;
2874                         dels += deleted;
2875                 }
2876                 if (i < count)
2877                         continue;
2878                 if (!extra_shown)
2879                         emit_diff_symbol(options,
2880                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2881                                          NULL, 0, 0);
2882                 extra_shown = 1;
2883         }
2884
2885         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2886         strbuf_release(&out);
2887 }
2888
2889 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2890 {
2891         int i, adds = 0, dels = 0, total_files = data->nr;
2892
2893         if (data->nr == 0)
2894                 return;
2895
2896         for (i = 0; i < data->nr; i++) {
2897                 int added = data->files[i]->added;
2898                 int deleted = data->files[i]->deleted;
2899
2900                 if (data->files[i]->is_unmerged ||
2901                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2902                         total_files--;
2903                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2904                         adds += added;
2905                         dels += deleted;
2906                 }
2907         }
2908         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2909 }
2910
2911 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2912 {
2913         int i;
2914
2915         if (data->nr == 0)
2916                 return;
2917
2918         for (i = 0; i < data->nr; i++) {
2919                 struct diffstat_file *file = data->files[i];
2920
2921                 fprintf(options->file, "%s", diff_line_prefix(options));
2922
2923                 if (file->is_binary)
2924                         fprintf(options->file, "-\t-\t");
2925                 else
2926                         fprintf(options->file,
2927                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2928                                 file->added, file->deleted);
2929                 if (options->line_termination) {
2930                         fill_print_name(file);
2931                         if (!file->is_renamed)
2932                                 write_name_quoted(file->name, options->file,
2933                                                   options->line_termination);
2934                         else {
2935                                 fputs(file->print_name, options->file);
2936                                 putc(options->line_termination, options->file);
2937                         }
2938                 } else {
2939                         if (file->is_renamed) {
2940                                 putc('\0', options->file);
2941                                 write_name_quoted(file->from_name, options->file, '\0');
2942                         }
2943                         write_name_quoted(file->name, options->file, '\0');
2944                 }
2945         }
2946 }
2947
2948 struct dirstat_file {
2949         const char *name;
2950         unsigned long changed;
2951 };
2952
2953 struct dirstat_dir {
2954         struct dirstat_file *files;
2955         int alloc, nr, permille, cumulative;
2956 };
2957
2958 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2959                 unsigned long changed, const char *base, int baselen)
2960 {
2961         unsigned long sum_changes = 0;
2962         unsigned int sources = 0;
2963         const char *line_prefix = diff_line_prefix(opt);
2964
2965         while (dir->nr) {
2966                 struct dirstat_file *f = dir->files;
2967                 int namelen = strlen(f->name);
2968                 unsigned long changes;
2969                 char *slash;
2970
2971                 if (namelen < baselen)
2972                         break;
2973                 if (memcmp(f->name, base, baselen))
2974                         break;
2975                 slash = strchr(f->name + baselen, '/');
2976                 if (slash) {
2977                         int newbaselen = slash + 1 - f->name;
2978                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2979                         sources++;
2980                 } else {
2981                         changes = f->changed;
2982                         dir->files++;
2983                         dir->nr--;
2984                         sources += 2;
2985                 }
2986                 sum_changes += changes;
2987         }
2988
2989         /*
2990          * We don't report dirstat's for
2991          *  - the top level
2992          *  - or cases where everything came from a single directory
2993          *    under this directory (sources == 1).
2994          */
2995         if (baselen && sources != 1) {
2996                 if (sum_changes) {
2997                         int permille = sum_changes * 1000 / changed;
2998                         if (permille >= dir->permille) {
2999                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3000                                         permille / 10, permille % 10, baselen, base);
3001                                 if (!dir->cumulative)
3002                                         return 0;
3003                         }
3004                 }
3005         }
3006         return sum_changes;
3007 }
3008
3009 static int dirstat_compare(const void *_a, const void *_b)
3010 {
3011         const struct dirstat_file *a = _a;
3012         const struct dirstat_file *b = _b;
3013         return strcmp(a->name, b->name);
3014 }
3015
3016 static void show_dirstat(struct diff_options *options)
3017 {
3018         int i;
3019         unsigned long changed;
3020         struct dirstat_dir dir;
3021         struct diff_queue_struct *q = &diff_queued_diff;
3022
3023         dir.files = NULL;
3024         dir.alloc = 0;
3025         dir.nr = 0;
3026         dir.permille = options->dirstat_permille;
3027         dir.cumulative = options->flags.dirstat_cumulative;
3028
3029         changed = 0;
3030         for (i = 0; i < q->nr; i++) {
3031                 struct diff_filepair *p = q->queue[i];
3032                 const char *name;
3033                 unsigned long copied, added, damage;
3034
3035                 name = p->two->path ? p->two->path : p->one->path;
3036
3037                 if (p->one->oid_valid && p->two->oid_valid &&
3038                     oideq(&p->one->oid, &p->two->oid)) {
3039                         /*
3040                          * The SHA1 has not changed, so pre-/post-content is
3041                          * identical. We can therefore skip looking at the
3042                          * file contents altogether.
3043                          */
3044                         damage = 0;
3045                         goto found_damage;
3046                 }
3047
3048                 if (options->flags.dirstat_by_file) {
3049                         /*
3050                          * In --dirstat-by-file mode, we don't really need to
3051                          * look at the actual file contents at all.
3052                          * The fact that the SHA1 changed is enough for us to
3053                          * add this file to the list of results
3054                          * (with each file contributing equal damage).
3055                          */
3056                         damage = 1;
3057                         goto found_damage;
3058                 }
3059
3060                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3061                         diff_populate_filespec(options->repo, p->one, 0);
3062                         diff_populate_filespec(options->repo, p->two, 0);
3063                         diffcore_count_changes(options->repo,
3064                                                p->one, p->two, NULL, NULL,
3065                                                &copied, &added);
3066                         diff_free_filespec_data(p->one);
3067                         diff_free_filespec_data(p->two);
3068                 } else if (DIFF_FILE_VALID(p->one)) {
3069                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3070                         copied = added = 0;
3071                         diff_free_filespec_data(p->one);
3072                 } else if (DIFF_FILE_VALID(p->two)) {
3073                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3074                         copied = 0;
3075                         added = p->two->size;
3076                         diff_free_filespec_data(p->two);
3077                 } else
3078                         continue;
3079
3080                 /*
3081                  * Original minus copied is the removed material,
3082                  * added is the new material.  They are both damages
3083                  * made to the preimage.
3084                  * If the resulting damage is zero, we know that
3085                  * diffcore_count_changes() considers the two entries to
3086                  * be identical, but since the oid changed, we
3087                  * know that there must have been _some_ kind of change,
3088                  * so we force all entries to have damage > 0.
3089                  */
3090                 damage = (p->one->size - copied) + added;
3091                 if (!damage)
3092                         damage = 1;
3093
3094 found_damage:
3095                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3096                 dir.files[dir.nr].name = name;
3097                 dir.files[dir.nr].changed = damage;
3098                 changed += damage;
3099                 dir.nr++;
3100         }
3101
3102         /* This can happen even with many files, if everything was renames */
3103         if (!changed)
3104                 return;
3105
3106         /* Show all directories with more than x% of the changes */
3107         QSORT(dir.files, dir.nr, dirstat_compare);
3108         gather_dirstat(options, &dir, changed, "", 0);
3109 }
3110
3111 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3112 {
3113         int i;
3114         unsigned long changed;
3115         struct dirstat_dir dir;
3116
3117         if (data->nr == 0)
3118                 return;
3119
3120         dir.files = NULL;
3121         dir.alloc = 0;
3122         dir.nr = 0;
3123         dir.permille = options->dirstat_permille;
3124         dir.cumulative = options->flags.dirstat_cumulative;
3125
3126         changed = 0;
3127         for (i = 0; i < data->nr; i++) {
3128                 struct diffstat_file *file = data->files[i];
3129                 unsigned long damage = file->added + file->deleted;
3130                 if (file->is_binary)
3131                         /*
3132                          * binary files counts bytes, not lines. Must find some
3133                          * way to normalize binary bytes vs. textual lines.
3134                          * The following heuristic assumes that there are 64
3135                          * bytes per "line".
3136                          * This is stupid and ugly, but very cheap...
3137                          */
3138                         damage = DIV_ROUND_UP(damage, 64);
3139                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3140                 dir.files[dir.nr].name = file->name;
3141                 dir.files[dir.nr].changed = damage;
3142                 changed += damage;
3143                 dir.nr++;
3144         }
3145
3146         /* This can happen even with many files, if everything was renames */
3147         if (!changed)
3148                 return;
3149
3150         /* Show all directories with more than x% of the changes */
3151         QSORT(dir.files, dir.nr, dirstat_compare);
3152         gather_dirstat(options, &dir, changed, "", 0);
3153 }
3154
3155 static void free_diffstat_info(struct diffstat_t *diffstat)
3156 {
3157         int i;
3158         for (i = 0; i < diffstat->nr; i++) {
3159                 struct diffstat_file *f = diffstat->files[i];
3160                 free(f->print_name);
3161                 free(f->name);
3162                 free(f->from_name);
3163                 free(f);
3164         }
3165         free(diffstat->files);
3166 }
3167
3168 struct checkdiff_t {
3169         const char *filename;
3170         int lineno;
3171         int conflict_marker_size;
3172         struct diff_options *o;
3173         unsigned ws_rule;
3174         unsigned status;
3175 };
3176
3177 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3178 {
3179         char firstchar;
3180         int cnt;
3181
3182         if (len < marker_size + 1)
3183                 return 0;
3184         firstchar = line[0];
3185         switch (firstchar) {
3186         case '=': case '>': case '<': case '|':
3187                 break;
3188         default:
3189                 return 0;
3190         }
3191         for (cnt = 1; cnt < marker_size; cnt++)
3192                 if (line[cnt] != firstchar)
3193                         return 0;
3194         /* line[1] thru line[marker_size-1] are same as firstchar */
3195         if (len < marker_size + 1 || !isspace(line[marker_size]))
3196                 return 0;
3197         return 1;
3198 }
3199
3200 static void checkdiff_consume_hunk(void *priv,
3201                                    long ob, long on, long nb, long nn,
3202                                    const char *func, long funclen)
3203
3204 {
3205         struct checkdiff_t *data = priv;
3206         data->lineno = nb - 1;
3207 }
3208
3209 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3210 {
3211         struct checkdiff_t *data = priv;
3212         int marker_size = data->conflict_marker_size;
3213         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3214         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3215         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3216         char *err;
3217         const char *line_prefix;
3218
3219         assert(data->o);
3220         line_prefix = diff_line_prefix(data->o);
3221
3222         if (line[0] == '+') {
3223                 unsigned bad;
3224                 data->lineno++;
3225                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3226                         data->status |= 1;
3227                         fprintf(data->o->file,
3228                                 "%s%s:%d: leftover conflict marker\n",
3229                                 line_prefix, data->filename, data->lineno);
3230                 }
3231                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3232                 if (!bad)
3233                         return;
3234                 data->status |= bad;
3235                 err = whitespace_error_string(bad);
3236                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3237                         line_prefix, data->filename, data->lineno, err);
3238                 free(err);
3239                 emit_line(data->o, set, reset, line, 1);
3240                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3241                               data->o->file, set, reset, ws);
3242         } else if (line[0] == ' ') {
3243                 data->lineno++;
3244         }
3245 }
3246
3247 static unsigned char *deflate_it(char *data,
3248                                  unsigned long size,
3249                                  unsigned long *result_size)
3250 {
3251         int bound;
3252         unsigned char *deflated;
3253         git_zstream stream;
3254
3255         git_deflate_init(&stream, zlib_compression_level);
3256         bound = git_deflate_bound(&stream, size);
3257         deflated = xmalloc(bound);
3258         stream.next_out = deflated;
3259         stream.avail_out = bound;
3260
3261         stream.next_in = (unsigned char *)data;
3262         stream.avail_in = size;
3263         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3264                 ; /* nothing */
3265         git_deflate_end(&stream);
3266         *result_size = stream.total_out;
3267         return deflated;
3268 }
3269
3270 static void emit_binary_diff_body(struct diff_options *o,
3271                                   mmfile_t *one, mmfile_t *two)
3272 {
3273         void *cp;
3274         void *delta;
3275         void *deflated;
3276         void *data;
3277         unsigned long orig_size;
3278         unsigned long delta_size;
3279         unsigned long deflate_size;
3280         unsigned long data_size;
3281
3282         /* We could do deflated delta, or we could do just deflated two,
3283          * whichever is smaller.
3284          */
3285         delta = NULL;
3286         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3287         if (one->size && two->size) {
3288                 delta = diff_delta(one->ptr, one->size,
3289                                    two->ptr, two->size,
3290                                    &delta_size, deflate_size);
3291                 if (delta) {
3292                         void *to_free = delta;
3293                         orig_size = delta_size;
3294                         delta = deflate_it(delta, delta_size, &delta_size);
3295                         free(to_free);
3296                 }
3297         }
3298
3299         if (delta && delta_size < deflate_size) {
3300                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3301                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3302                                  s, strlen(s), 0);
3303                 free(s);
3304                 free(deflated);
3305                 data = delta;
3306                 data_size = delta_size;
3307         } else {
3308                 char *s = xstrfmt("%lu", two->size);
3309                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3310                                  s, strlen(s), 0);
3311                 free(s);
3312                 free(delta);
3313                 data = deflated;
3314                 data_size = deflate_size;
3315         }
3316
3317         /* emit data encoded in base85 */
3318         cp = data;
3319         while (data_size) {
3320                 int len;
3321                 int bytes = (52 < data_size) ? 52 : data_size;
3322                 char line[71];
3323                 data_size -= bytes;
3324                 if (bytes <= 26)
3325                         line[0] = bytes + 'A' - 1;
3326                 else
3327                         line[0] = bytes - 26 + 'a' - 1;
3328                 encode_85(line + 1, cp, bytes);
3329                 cp = (char *) cp + bytes;
3330
3331                 len = strlen(line);
3332                 line[len++] = '\n';
3333                 line[len] = '\0';
3334
3335                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3336                                  line, len, 0);
3337         }
3338         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3339         free(data);
3340 }
3341
3342 static void emit_binary_diff(struct diff_options *o,
3343                              mmfile_t *one, mmfile_t *two)
3344 {
3345         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3346         emit_binary_diff_body(o, one, two);
3347         emit_binary_diff_body(o, two, one);
3348 }
3349
3350 int diff_filespec_is_binary(struct repository *r,
3351                             struct diff_filespec *one)
3352 {
3353         if (one->is_binary == -1) {
3354                 diff_filespec_load_driver(one, r->index);
3355                 if (one->driver->binary != -1)
3356                         one->is_binary = one->driver->binary;
3357                 else {
3358                         if (!one->data && DIFF_FILE_VALID(one))
3359                                 diff_populate_filespec(r, one, CHECK_BINARY);
3360                         if (one->is_binary == -1 && one->data)
3361                                 one->is_binary = buffer_is_binary(one->data,
3362                                                 one->size);
3363                         if (one->is_binary == -1)
3364                                 one->is_binary = 0;
3365                 }
3366         }
3367         return one->is_binary;
3368 }
3369
3370 static const struct userdiff_funcname *
3371 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3372 {
3373         diff_filespec_load_driver(one, o->repo->index);
3374         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3375 }
3376
3377 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3378 {
3379         if (!options->a_prefix)
3380                 options->a_prefix = a;
3381         if (!options->b_prefix)
3382                 options->b_prefix = b;
3383 }
3384
3385 struct userdiff_driver *get_textconv(struct repository *r,
3386                                      struct diff_filespec *one)
3387 {
3388         if (!DIFF_FILE_VALID(one))
3389                 return NULL;
3390
3391         diff_filespec_load_driver(one, r->index);
3392         return userdiff_get_textconv(r, one->driver);
3393 }
3394
3395 static void builtin_diff(const char *name_a,
3396                          const char *name_b,
3397                          struct diff_filespec *one,
3398                          struct diff_filespec *two,
3399                          const char *xfrm_msg,
3400                          int must_show_header,
3401                          struct diff_options *o,
3402                          int complete_rewrite)
3403 {
3404         mmfile_t mf1, mf2;
3405         const char *lbl[2];
3406         char *a_one, *b_two;
3407         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3408         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3409         const char *a_prefix, *b_prefix;
3410         struct userdiff_driver *textconv_one = NULL;
3411         struct userdiff_driver *textconv_two = NULL;
3412         struct strbuf header = STRBUF_INIT;
3413         const char *line_prefix = diff_line_prefix(o);
3414
3415         diff_set_mnemonic_prefix(o, "a/", "b/");
3416         if (o->flags.reverse_diff) {
3417                 a_prefix = o->b_prefix;
3418                 b_prefix = o->a_prefix;
3419         } else {
3420                 a_prefix = o->a_prefix;
3421                 b_prefix = o->b_prefix;
3422         }
3423
3424         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3425             (!one->mode || S_ISGITLINK(one->mode)) &&
3426             (!two->mode || S_ISGITLINK(two->mode))) {
3427                 show_submodule_summary(o, one->path ? one->path : two->path,
3428                                 &one->oid, &two->oid,
3429                                 two->dirty_submodule);
3430                 return;
3431         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3432                    (!one->mode || S_ISGITLINK(one->mode)) &&
3433                    (!two->mode || S_ISGITLINK(two->mode))) {
3434                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3435                                 &one->oid, &two->oid,
3436                                 two->dirty_submodule);
3437                 return;
3438         }
3439
3440         if (o->flags.allow_textconv) {
3441                 textconv_one = get_textconv(o->repo, one);
3442                 textconv_two = get_textconv(o->repo, two);
3443         }
3444
3445         /* Never use a non-valid filename anywhere if at all possible */
3446         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3447         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3448
3449         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3450         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3451         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3452         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3453         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3454         if (lbl[0][0] == '/') {
3455                 /* /dev/null */
3456                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3457                 if (xfrm_msg)
3458                         strbuf_addstr(&header, xfrm_msg);
3459                 must_show_header = 1;
3460         }
3461         else if (lbl[1][0] == '/') {
3462                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3463                 if (xfrm_msg)
3464                         strbuf_addstr(&header, xfrm_msg);
3465                 must_show_header = 1;
3466         }
3467         else {
3468                 if (one->mode != two->mode) {
3469                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3470                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3471                         must_show_header = 1;
3472                 }
3473                 if (xfrm_msg)
3474                         strbuf_addstr(&header, xfrm_msg);
3475
3476                 /*
3477                  * we do not run diff between different kind
3478                  * of objects.
3479                  */
3480                 if ((one->mode ^ two->mode) & S_IFMT)
3481                         goto free_ab_and_return;
3482                 if (complete_rewrite &&
3483                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3484                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3485                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3486                                          header.buf, header.len, 0);
3487                         strbuf_reset(&header);
3488                         emit_rewrite_diff(name_a, name_b, one, two,
3489                                           textconv_one, textconv_two, o);
3490                         o->found_changes = 1;
3491                         goto free_ab_and_return;
3492                 }
3493         }
3494
3495         if (o->irreversible_delete && lbl[1][0] == '/') {
3496                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3497                                  header.len, 0);
3498                 strbuf_reset(&header);
3499                 goto free_ab_and_return;
3500         } else if (!o->flags.text &&
3501                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3502                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3503                 struct strbuf sb = STRBUF_INIT;
3504                 if (!one->data && !two->data &&
3505                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3506                     !o->flags.binary) {
3507                         if (oideq(&one->oid, &two->oid)) {
3508                                 if (must_show_header)
3509                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3510                                                          header.buf, header.len,
3511                                                          0);
3512                                 goto free_ab_and_return;
3513                         }
3514                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3515                                          header.buf, header.len, 0);
3516                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3517                                     diff_line_prefix(o), lbl[0], lbl[1]);
3518                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3519                                          sb.buf, sb.len, 0);
3520                         strbuf_release(&sb);
3521                         goto free_ab_and_return;
3522                 }
3523                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3524                     fill_mmfile(o->repo, &mf2, two) < 0)
3525                         die("unable to read files to diff");
3526                 /* Quite common confusing case */
3527                 if (mf1.size == mf2.size &&
3528                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3529                         if (must_show_header)
3530                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3531                                                  header.buf, header.len, 0);
3532                         goto free_ab_and_return;
3533                 }
3534                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3535                 strbuf_reset(&header);
3536                 if (o->flags.binary)
3537                         emit_binary_diff(o, &mf1, &mf2);
3538                 else {
3539                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3540                                     diff_line_prefix(o), lbl[0], lbl[1]);
3541                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3542                                          sb.buf, sb.len, 0);
3543                         strbuf_release(&sb);
3544                 }
3545                 o->found_changes = 1;
3546         } else {
3547                 /* Crazy xdl interfaces.. */
3548                 const char *diffopts;
3549                 const char *v;
3550                 xpparam_t xpp;
3551                 xdemitconf_t xecfg;
3552                 struct emit_callback ecbdata;
3553                 const struct userdiff_funcname *pe;
3554
3555                 if (must_show_header) {
3556                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3557                                          header.buf, header.len, 0);
3558                         strbuf_reset(&header);
3559                 }
3560
3561                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3562                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3563
3564                 pe = diff_funcname_pattern(o, one);
3565                 if (!pe)
3566                         pe = diff_funcname_pattern(o, two);
3567
3568                 memset(&xpp, 0, sizeof(xpp));
3569                 memset(&xecfg, 0, sizeof(xecfg));
3570                 memset(&ecbdata, 0, sizeof(ecbdata));
3571                 if (o->flags.suppress_diff_headers)
3572                         lbl[0] = NULL;
3573                 ecbdata.label_path = lbl;
3574                 ecbdata.color_diff = want_color(o->use_color);
3575                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3576                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3577                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3578                 ecbdata.opt = o;
3579                 if (header.len && !o->flags.suppress_diff_headers)
3580                         ecbdata.header = &header;
3581                 xpp.flags = o->xdl_opts;
3582                 xpp.anchors = o->anchors;
3583                 xpp.anchors_nr = o->anchors_nr;
3584                 xecfg.ctxlen = o->context;
3585                 xecfg.interhunkctxlen = o->interhunkcontext;
3586                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3587                 if (o->flags.funccontext)
3588                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3589                 if (pe)
3590                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3591
3592                 diffopts = getenv("GIT_DIFF_OPTS");
3593                 if (!diffopts)
3594                         ;
3595                 else if (skip_prefix(diffopts, "--unified=", &v))
3596                         xecfg.ctxlen = strtoul(v, NULL, 10);
3597                 else if (skip_prefix(diffopts, "-u", &v))
3598                         xecfg.ctxlen = strtoul(v, NULL, 10);
3599
3600                 if (o->word_diff)
3601                         init_diff_words_data(&ecbdata, o, one, two);
3602                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3603                                   &ecbdata, &xpp, &xecfg))
3604                         die("unable to generate diff for %s", one->path);
3605                 if (o->word_diff)
3606                         free_diff_words_data(&ecbdata);
3607                 if (textconv_one)
3608                         free(mf1.ptr);
3609                 if (textconv_two)
3610                         free(mf2.ptr);
3611                 xdiff_clear_find_func(&xecfg);
3612         }
3613
3614  free_ab_and_return:
3615         strbuf_release(&header);
3616         diff_free_filespec_data(one);
3617         diff_free_filespec_data(two);
3618         free(a_one);
3619         free(b_two);
3620         return;
3621 }
3622
3623 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3624 {
3625         if (!is_renamed) {
3626                 if (p->status == DIFF_STATUS_ADDED) {
3627                         if (S_ISLNK(p->two->mode))
3628                                 return "new +l";
3629                         else if ((p->two->mode & 0777) == 0755)
3630                                 return "new +x";
3631                         else
3632                                 return "new";
3633                 } else if (p->status == DIFF_STATUS_DELETED)
3634                         return "gone";
3635         }
3636         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3637                 return "mode -l";
3638         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3639                 return "mode +l";
3640         else if ((p->one->mode & 0777) == 0644 &&
3641                  (p->two->mode & 0777) == 0755)
3642                 return "mode +x";
3643         else if ((p->one->mode & 0777) == 0755 &&
3644                  (p->two->mode & 0777) == 0644)
3645                 return "mode -x";
3646         return NULL;
3647 }
3648
3649 static void builtin_diffstat(const char *name_a, const char *name_b,
3650                              struct diff_filespec *one,
3651                              struct diff_filespec *two,
3652                              struct diffstat_t *diffstat,
3653                              struct diff_options *o,
3654                              struct diff_filepair *p)
3655 {
3656         mmfile_t mf1, mf2;
3657         struct diffstat_file *data;
3658         int same_contents;
3659         int complete_rewrite = 0;
3660
3661         if (!DIFF_PAIR_UNMERGED(p)) {
3662                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3663                         complete_rewrite = 1;
3664         }
3665
3666         data = diffstat_add(diffstat, name_a, name_b);
3667         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3668         if (o->flags.stat_with_summary)
3669                 data->comments = get_compact_summary(p, data->is_renamed);
3670
3671         if (!one || !two) {
3672                 data->is_unmerged = 1;
3673                 return;
3674         }
3675
3676         same_contents = oideq(&one->oid, &two->oid);
3677
3678         if (diff_filespec_is_binary(o->repo, one) ||
3679             diff_filespec_is_binary(o->repo, two)) {
3680                 data->is_binary = 1;
3681                 if (same_contents) {
3682                         data->added = 0;
3683                         data->deleted = 0;
3684                 } else {
3685                         data->added = diff_filespec_size(o->repo, two);
3686                         data->deleted = diff_filespec_size(o->repo, one);
3687                 }
3688         }
3689
3690         else if (complete_rewrite) {
3691                 diff_populate_filespec(o->repo, one, 0);
3692                 diff_populate_filespec(o->repo, two, 0);
3693                 data->deleted = count_lines(one->data, one->size);
3694                 data->added = count_lines(two->data, two->size);
3695         }
3696
3697         else if (!same_contents) {
3698                 /* Crazy xdl interfaces.. */
3699                 xpparam_t xpp;
3700                 xdemitconf_t xecfg;
3701
3702                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3703                     fill_mmfile(o->repo, &mf2, two) < 0)
3704                         die("unable to read files to diff");
3705
3706                 memset(&xpp, 0, sizeof(xpp));
3707                 memset(&xecfg, 0, sizeof(xecfg));
3708                 xpp.flags = o->xdl_opts;
3709                 xpp.anchors = o->anchors;
3710                 xpp.anchors_nr = o->anchors_nr;
3711                 xecfg.ctxlen = o->context;
3712                 xecfg.interhunkctxlen = o->interhunkcontext;
3713                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3714                                   diffstat_consume, diffstat, &xpp, &xecfg))
3715                         die("unable to generate diffstat for %s", one->path);
3716         }
3717
3718         diff_free_filespec_data(one);
3719         diff_free_filespec_data(two);
3720 }
3721
3722 static void builtin_checkdiff(const char *name_a, const char *name_b,
3723                               const char *attr_path,
3724                               struct diff_filespec *one,
3725                               struct diff_filespec *two,
3726                               struct diff_options *o)
3727 {
3728         mmfile_t mf1, mf2;
3729         struct checkdiff_t data;
3730
3731         if (!two)
3732                 return;
3733
3734         memset(&data, 0, sizeof(data));
3735         data.filename = name_b ? name_b : name_a;
3736         data.lineno = 0;
3737         data.o = o;
3738         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3739         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3740
3741         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3742             fill_mmfile(o->repo, &mf2, two) < 0)
3743                 die("unable to read files to diff");
3744
3745         /*
3746          * All the other codepaths check both sides, but not checking
3747          * the "old" side here is deliberate.  We are checking the newly
3748          * introduced changes, and as long as the "new" side is text, we
3749          * can and should check what it introduces.
3750          */
3751         if (diff_filespec_is_binary(o->repo, two))
3752                 goto free_and_return;
3753         else {
3754                 /* Crazy xdl interfaces.. */
3755                 xpparam_t xpp;
3756                 xdemitconf_t xecfg;
3757
3758                 memset(&xpp, 0, sizeof(xpp));
3759                 memset(&xecfg, 0, sizeof(xecfg));
3760                 xecfg.ctxlen = 1; /* at least one context line */
3761                 xpp.flags = 0;
3762                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3763                                   checkdiff_consume, &data,
3764                                   &xpp, &xecfg))
3765                         die("unable to generate checkdiff for %s", one->path);
3766
3767                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3768                         struct emit_callback ecbdata;
3769                         int blank_at_eof;
3770
3771                         ecbdata.ws_rule = data.ws_rule;
3772                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3773                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3774
3775                         if (blank_at_eof) {
3776                                 static char *err;
3777                                 if (!err)
3778                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3779                                 fprintf(o->file, "%s:%d: %s.\n",
3780                                         data.filename, blank_at_eof, err);
3781                                 data.status = 1; /* report errors */
3782                         }
3783                 }
3784         }
3785  free_and_return:
3786         diff_free_filespec_data(one);
3787         diff_free_filespec_data(two);
3788         if (data.status)
3789                 o->flags.check_failed = 1;
3790 }
3791
3792 struct diff_filespec *alloc_filespec(const char *path)
3793 {
3794         struct diff_filespec *spec;
3795
3796         FLEXPTR_ALLOC_STR(spec, path, path);
3797         spec->count = 1;
3798         spec->is_binary = -1;
3799         return spec;
3800 }
3801
3802 void free_filespec(struct diff_filespec *spec)
3803 {
3804         if (!--spec->count) {
3805                 diff_free_filespec_data(spec);
3806                 free(spec);
3807         }
3808 }
3809
3810 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3811                    int oid_valid, unsigned short mode)
3812 {
3813         if (mode) {
3814                 spec->mode = canon_mode(mode);
3815                 oidcpy(&spec->oid, oid);
3816                 spec->oid_valid = oid_valid;
3817         }
3818 }
3819
3820 /*
3821  * Given a name and sha1 pair, if the index tells us the file in
3822  * the work tree has that object contents, return true, so that
3823  * prepare_temp_file() does not have to inflate and extract.
3824  */
3825 static int reuse_worktree_file(struct index_state *istate,
3826                                const char *name,
3827                                const struct object_id *oid,
3828                                int want_file)
3829 {
3830         const struct cache_entry *ce;
3831         struct stat st;
3832         int pos, len;
3833
3834         /*
3835          * We do not read the cache ourselves here, because the
3836          * benchmark with my previous version that always reads cache
3837          * shows that it makes things worse for diff-tree comparing
3838          * two linux-2.6 kernel trees in an already checked out work
3839          * tree.  This is because most diff-tree comparisons deal with
3840          * only a small number of files, while reading the cache is
3841          * expensive for a large project, and its cost outweighs the
3842          * savings we get by not inflating the object to a temporary
3843          * file.  Practically, this code only helps when we are used
3844          * by diff-cache --cached, which does read the cache before
3845          * calling us.
3846          */
3847         if (!istate->cache)
3848                 return 0;
3849
3850         /* We want to avoid the working directory if our caller
3851          * doesn't need the data in a normal file, this system
3852          * is rather slow with its stat/open/mmap/close syscalls,
3853          * and the object is contained in a pack file.  The pack
3854          * is probably already open and will be faster to obtain
3855          * the data through than the working directory.  Loose
3856          * objects however would tend to be slower as they need
3857          * to be individually opened and inflated.
3858          */
3859         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3860                 return 0;
3861
3862         /*
3863          * Similarly, if we'd have to convert the file contents anyway, that
3864          * makes the optimization not worthwhile.
3865          */
3866         if (!want_file && would_convert_to_git(istate, name))
3867                 return 0;
3868
3869         len = strlen(name);
3870         pos = index_name_pos(istate, name, len);
3871         if (pos < 0)
3872                 return 0;
3873         ce = istate->cache[pos];
3874
3875         /*
3876          * This is not the sha1 we are looking for, or
3877          * unreusable because it is not a regular file.
3878          */
3879         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3880                 return 0;
3881
3882         /*
3883          * If ce is marked as "assume unchanged", there is no
3884          * guarantee that work tree matches what we are looking for.
3885          */
3886         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3887                 return 0;
3888
3889         /*
3890          * If ce matches the file in the work tree, we can reuse it.
3891          */
3892         if (ce_uptodate(ce) ||
3893             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3894                 return 1;
3895
3896         return 0;
3897 }
3898
3899 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3900 {
3901         struct strbuf buf = STRBUF_INIT;
3902         char *dirty = "";
3903
3904         /* Are we looking at the work tree? */
3905         if (s->dirty_submodule)
3906                 dirty = "-dirty";
3907
3908         strbuf_addf(&buf, "Subproject commit %s%s\n",
3909                     oid_to_hex(&s->oid), dirty);
3910         s->size = buf.len;
3911         if (size_only) {
3912                 s->data = NULL;
3913                 strbuf_release(&buf);
3914         } else {
3915                 s->data = strbuf_detach(&buf, NULL);
3916                 s->should_free = 1;
3917         }
3918         return 0;
3919 }
3920
3921 /*
3922  * While doing rename detection and pickaxe operation, we may need to
3923  * grab the data for the blob (or file) for our own in-core comparison.
3924  * diff_filespec has data and size fields for this purpose.
3925  */
3926 int diff_populate_filespec(struct repository *r,
3927                            struct diff_filespec *s,
3928                            unsigned int flags)
3929 {
3930         int size_only = flags & CHECK_SIZE_ONLY;
3931         int err = 0;
3932         int conv_flags = global_conv_flags_eol;
3933         /*
3934          * demote FAIL to WARN to allow inspecting the situation
3935          * instead of refusing.
3936          */
3937         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3938                 conv_flags = CONV_EOL_RNDTRP_WARN;
3939
3940         if (!DIFF_FILE_VALID(s))
3941                 die("internal error: asking to populate invalid file.");
3942         if (S_ISDIR(s->mode))
3943                 return -1;
3944
3945         if (s->data)
3946                 return 0;
3947
3948         if (size_only && 0 < s->size)
3949                 return 0;
3950
3951         if (S_ISGITLINK(s->mode))
3952                 return diff_populate_gitlink(s, size_only);
3953
3954         if (!s->oid_valid ||
3955             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3956                 struct strbuf buf = STRBUF_INIT;
3957                 struct stat st;
3958                 int fd;
3959
3960                 if (lstat(s->path, &st) < 0) {
3961                 err_empty:
3962                         err = -1;
3963                 empty:
3964                         s->data = (char *)"";
3965                         s->size = 0;
3966                         return err;
3967                 }
3968                 s->size = xsize_t(st.st_size);
3969                 if (!s->size)
3970                         goto empty;
3971                 if (S_ISLNK(st.st_mode)) {
3972                         struct strbuf sb = STRBUF_INIT;
3973
3974                         if (strbuf_readlink(&sb, s->path, s->size))
3975                                 goto err_empty;
3976                         s->size = sb.len;
3977                         s->data = strbuf_detach(&sb, NULL);
3978                         s->should_free = 1;
3979                         return 0;
3980                 }
3981
3982                 /*
3983                  * Even if the caller would be happy with getting
3984                  * only the size, we cannot return early at this
3985                  * point if the path requires us to run the content
3986                  * conversion.
3987                  */
3988                 if (size_only && !would_convert_to_git(r->index, s->path))
3989                         return 0;
3990
3991                 /*
3992                  * Note: this check uses xsize_t(st.st_size) that may
3993                  * not be the true size of the blob after it goes
3994                  * through convert_to_git().  This may not strictly be
3995                  * correct, but the whole point of big_file_threshold
3996                  * and is_binary check being that we want to avoid
3997                  * opening the file and inspecting the contents, this
3998                  * is probably fine.
3999                  */
4000                 if ((flags & CHECK_BINARY) &&
4001                     s->size > big_file_threshold && s->is_binary == -1) {
4002                         s->is_binary = 1;
4003                         return 0;
4004                 }
4005                 fd = open(s->path, O_RDONLY);
4006                 if (fd < 0)
4007                         goto err_empty;
4008                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4009                 close(fd);
4010                 s->should_munmap = 1;
4011
4012                 /*
4013                  * Convert from working tree format to canonical git format
4014                  */
4015                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4016                         size_t size = 0;
4017                         munmap(s->data, s->size);
4018                         s->should_munmap = 0;
4019                         s->data = strbuf_detach(&buf, &size);
4020                         s->size = size;
4021                         s->should_free = 1;
4022                 }
4023         }
4024         else {
4025                 enum object_type type;
4026                 if (size_only || (flags & CHECK_BINARY)) {
4027                         type = oid_object_info(r, &s->oid, &s->size);
4028                         if (type < 0)
4029                                 die("unable to read %s",
4030                                     oid_to_hex(&s->oid));
4031                         if (size_only)
4032                                 return 0;
4033                         if (s->size > big_file_threshold && s->is_binary == -1) {
4034                                 s->is_binary = 1;
4035                                 return 0;
4036                         }
4037                 }
4038                 s->data = read_object_file(&s->oid, &type, &s->size);
4039                 if (!s->data)
4040                         die("unable to read %s", oid_to_hex(&s->oid));
4041                 s->should_free = 1;
4042         }
4043         return 0;
4044 }
4045
4046 void diff_free_filespec_blob(struct diff_filespec *s)
4047 {
4048         if (s->should_free)
4049                 free(s->data);
4050         else if (s->should_munmap)
4051                 munmap(s->data, s->size);
4052
4053         if (s->should_free || s->should_munmap) {
4054                 s->should_free = s->should_munmap = 0;
4055                 s->data = NULL;
4056         }
4057 }
4058
4059 void diff_free_filespec_data(struct diff_filespec *s)
4060 {
4061         diff_free_filespec_blob(s);
4062         FREE_AND_NULL(s->cnt_data);
4063 }
4064
4065 static void prep_temp_blob(struct index_state *istate,
4066                            const char *path, struct diff_tempfile *temp,
4067                            void *blob,
4068                            unsigned long size,
4069                            const struct object_id *oid,
4070                            int mode)
4071 {
4072         struct strbuf buf = STRBUF_INIT;
4073         struct strbuf tempfile = STRBUF_INIT;
4074         char *path_dup = xstrdup(path);
4075         const char *base = basename(path_dup);
4076
4077         /* Generate "XXXXXX_basename.ext" */
4078         strbuf_addstr(&tempfile, "XXXXXX_");
4079         strbuf_addstr(&tempfile, base);
4080
4081         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4082         if (!temp->tempfile)
4083                 die_errno("unable to create temp-file");
4084         if (convert_to_working_tree(istate, path,
4085                         (const char *)blob, (size_t)size, &buf)) {
4086                 blob = buf.buf;
4087                 size = buf.len;
4088         }
4089         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4090             close_tempfile_gently(temp->tempfile))
4091                 die_errno("unable to write temp-file");
4092         temp->name = get_tempfile_path(temp->tempfile);
4093         oid_to_hex_r(temp->hex, oid);
4094         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4095         strbuf_release(&buf);
4096         strbuf_release(&tempfile);
4097         free(path_dup);
4098 }
4099
4100 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4101                                                const char *name,
4102                                                struct diff_filespec *one)
4103 {
4104         struct diff_tempfile *temp = claim_diff_tempfile();
4105
4106         if (!DIFF_FILE_VALID(one)) {
4107         not_a_valid_file:
4108                 /* A '-' entry produces this for file-2, and
4109                  * a '+' entry produces this for file-1.
4110                  */
4111                 temp->name = "/dev/null";
4112                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4113                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4114                 return temp;
4115         }
4116
4117         if (!S_ISGITLINK(one->mode) &&
4118             (!one->oid_valid ||
4119              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4120                 struct stat st;
4121                 if (lstat(name, &st) < 0) {
4122                         if (errno == ENOENT)
4123                                 goto not_a_valid_file;
4124                         die_errno("stat(%s)", name);
4125                 }
4126                 if (S_ISLNK(st.st_mode)) {
4127                         struct strbuf sb = STRBUF_INIT;
4128                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4129                                 die_errno("readlink(%s)", name);
4130                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4131                                        (one->oid_valid ?
4132                                         &one->oid : &null_oid),
4133                                        (one->oid_valid ?
4134                                         one->mode : S_IFLNK));
4135                         strbuf_release(&sb);
4136                 }
4137                 else {
4138                         /* we can borrow from the file in the work tree */
4139                         temp->name = name;
4140                         if (!one->oid_valid)
4141                                 oid_to_hex_r(temp->hex, &null_oid);
4142                         else
4143                                 oid_to_hex_r(temp->hex, &one->oid);
4144                         /* Even though we may sometimes borrow the
4145                          * contents from the work tree, we always want
4146                          * one->mode.  mode is trustworthy even when
4147                          * !(one->oid_valid), as long as
4148                          * DIFF_FILE_VALID(one).
4149                          */
4150                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4151                 }
4152                 return temp;
4153         }
4154         else {
4155                 if (diff_populate_filespec(r, one, 0))
4156                         die("cannot read data blob for %s", one->path);
4157                 prep_temp_blob(r->index, name, temp,
4158                                one->data, one->size,
4159                                &one->oid, one->mode);
4160         }
4161         return temp;
4162 }
4163
4164 static void add_external_diff_name(struct repository *r,
4165                                    struct argv_array *argv,
4166                                    const char *name,
4167                                    struct diff_filespec *df)
4168 {
4169         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4170         argv_array_push(argv, temp->name);
4171         argv_array_push(argv, temp->hex);
4172         argv_array_push(argv, temp->mode);
4173 }
4174
4175 /* An external diff command takes:
4176  *
4177  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4178  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4179  *
4180  */
4181 static void run_external_diff(const char *pgm,
4182                               const char *name,
4183                               const char *other,
4184                               struct diff_filespec *one,
4185                               struct diff_filespec *two,
4186                               const char *xfrm_msg,
4187                               int complete_rewrite,
4188                               struct diff_options *o)
4189 {
4190         struct argv_array argv = ARGV_ARRAY_INIT;
4191         struct argv_array env = ARGV_ARRAY_INIT;
4192         struct diff_queue_struct *q = &diff_queued_diff;
4193
4194         argv_array_push(&argv, pgm);
4195         argv_array_push(&argv, name);
4196
4197         if (one && two) {
4198                 add_external_diff_name(o->repo, &argv, name, one);
4199                 if (!other)
4200                         add_external_diff_name(o->repo, &argv, name, two);
4201                 else {
4202                         add_external_diff_name(o->repo, &argv, other, two);
4203                         argv_array_push(&argv, other);
4204                         argv_array_push(&argv, xfrm_msg);
4205                 }
4206         }
4207
4208         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4209         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4210
4211         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4212                 die(_("external diff died, stopping at %s"), name);
4213
4214         remove_tempfile();
4215         argv_array_clear(&argv);
4216         argv_array_clear(&env);
4217 }
4218
4219 static int similarity_index(struct diff_filepair *p)
4220 {
4221         return p->score * 100 / MAX_SCORE;
4222 }
4223
4224 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4225 {
4226         if (startup_info->have_repository)
4227                 return find_unique_abbrev(oid, abbrev);
4228         else {
4229                 char *hex = oid_to_hex(oid);
4230                 if (abbrev < 0)
4231                         abbrev = FALLBACK_DEFAULT_ABBREV;
4232                 if (abbrev > the_hash_algo->hexsz)
4233                         BUG("oid abbreviation out of range: %d", abbrev);
4234                 if (abbrev)
4235                         hex[abbrev] = '\0';
4236                 return hex;
4237         }
4238 }
4239
4240 static void fill_metainfo(struct strbuf *msg,
4241                           const char *name,
4242                           const char *other,
4243                           struct diff_filespec *one,
4244                           struct diff_filespec *two,
4245                           struct diff_options *o,
4246                           struct diff_filepair *p,
4247                           int *must_show_header,
4248                           int use_color)
4249 {
4250         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4251         const char *reset = diff_get_color(use_color, DIFF_RESET);
4252         const char *line_prefix = diff_line_prefix(o);
4253
4254         *must_show_header = 1;
4255         strbuf_init(msg, PATH_MAX * 2 + 300);
4256         switch (p->status) {
4257         case DIFF_STATUS_COPIED:
4258                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4259                             line_prefix, set, similarity_index(p));
4260                 strbuf_addf(msg, "%s\n%s%scopy from ",
4261                             reset,  line_prefix, set);
4262                 quote_c_style(name, msg, NULL, 0);
4263                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4264                 quote_c_style(other, msg, NULL, 0);
4265                 strbuf_addf(msg, "%s\n", reset);
4266                 break;
4267         case DIFF_STATUS_RENAMED:
4268                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4269                             line_prefix, set, similarity_index(p));
4270                 strbuf_addf(msg, "%s\n%s%srename from ",
4271                             reset, line_prefix, set);
4272                 quote_c_style(name, msg, NULL, 0);
4273                 strbuf_addf(msg, "%s\n%s%srename to ",
4274                             reset, line_prefix, set);
4275                 quote_c_style(other, msg, NULL, 0);
4276                 strbuf_addf(msg, "%s\n", reset);
4277                 break;
4278         case DIFF_STATUS_MODIFIED:
4279                 if (p->score) {
4280                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4281                                     line_prefix,
4282                                     set, similarity_index(p), reset);
4283                         break;
4284                 }
4285                 /* fallthru */
4286         default:
4287                 *must_show_header = 0;
4288         }
4289         if (one && two && !oideq(&one->oid, &two->oid)) {
4290                 const unsigned hexsz = the_hash_algo->hexsz;
4291                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4292
4293                 if (o->flags.binary) {
4294                         mmfile_t mf;
4295                         if ((!fill_mmfile(o->repo, &mf, one) &&
4296                              diff_filespec_is_binary(o->repo, one)) ||
4297                             (!fill_mmfile(o->repo, &mf, two) &&
4298                              diff_filespec_is_binary(o->repo, two)))
4299                                 abbrev = hexsz;
4300                 }
4301                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4302                             diff_abbrev_oid(&one->oid, abbrev),
4303                             diff_abbrev_oid(&two->oid, abbrev));
4304                 if (one->mode == two->mode)
4305                         strbuf_addf(msg, " %06o", one->mode);
4306                 strbuf_addf(msg, "%s\n", reset);
4307         }
4308 }
4309
4310 static void run_diff_cmd(const char *pgm,
4311                          const char *name,
4312                          const char *other,
4313                          const char *attr_path,
4314                          struct diff_filespec *one,
4315                          struct diff_filespec *two,
4316                          struct strbuf *msg,
4317                          struct diff_options *o,
4318                          struct diff_filepair *p)
4319 {
4320         const char *xfrm_msg = NULL;
4321         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4322         int must_show_header = 0;
4323
4324
4325         if (o->flags.allow_external) {
4326                 struct userdiff_driver *drv;
4327
4328                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4329                 if (drv && drv->external)
4330                         pgm = drv->external;
4331         }
4332
4333         if (msg) {
4334                 /*
4335                  * don't use colors when the header is intended for an
4336                  * external diff driver
4337                  */
4338                 fill_metainfo(msg, name, other, one, two, o, p,
4339                               &must_show_header,
4340                               want_color(o->use_color) && !pgm);
4341                 xfrm_msg = msg->len ? msg->buf : NULL;
4342         }
4343
4344         if (pgm) {
4345                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4346                                   complete_rewrite, o);
4347                 return;
4348         }
4349         if (one && two)
4350                 builtin_diff(name, other ? other : name,
4351                              one, two, xfrm_msg, must_show_header,
4352                              o, complete_rewrite);
4353         else
4354                 fprintf(o->file, "* Unmerged path %s\n", name);
4355 }
4356
4357 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4358 {
4359         if (DIFF_FILE_VALID(one)) {
4360                 if (!one->oid_valid) {
4361                         struct stat st;
4362                         if (one->is_stdin) {
4363                                 oidclr(&one->oid);
4364                                 return;
4365                         }
4366                         if (lstat(one->path, &st) < 0)
4367                                 die_errno("stat '%s'", one->path);
4368                         if (index_path(istate, &one->oid, one->path, &st, 0))
4369                                 die("cannot hash %s", one->path);
4370                 }
4371         }
4372         else
4373                 oidclr(&one->oid);
4374 }
4375
4376 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4377 {
4378         /* Strip the prefix but do not molest /dev/null and absolute paths */
4379         if (*namep && !is_absolute_path(*namep)) {
4380                 *namep += prefix_length;
4381                 if (**namep == '/')
4382                         ++*namep;
4383         }
4384         if (*otherp && !is_absolute_path(*otherp)) {
4385                 *otherp += prefix_length;
4386                 if (**otherp == '/')
4387                         ++*otherp;
4388         }
4389 }
4390
4391 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4392 {
4393         const char *pgm = external_diff();
4394         struct strbuf msg;
4395         struct diff_filespec *one = p->one;
4396         struct diff_filespec *two = p->two;
4397         const char *name;
4398         const char *other;
4399         const char *attr_path;
4400
4401         name  = one->path;
4402         other = (strcmp(name, two->path) ? two->path : NULL);
4403         attr_path = name;
4404         if (o->prefix_length)
4405                 strip_prefix(o->prefix_length, &name, &other);
4406
4407         if (!o->flags.allow_external)
4408                 pgm = NULL;
4409
4410         if (DIFF_PAIR_UNMERGED(p)) {
4411                 run_diff_cmd(pgm, name, NULL, attr_path,
4412                              NULL, NULL, NULL, o, p);
4413                 return;
4414         }
4415
4416         diff_fill_oid_info(one, o->repo->index);
4417         diff_fill_oid_info(two, o->repo->index);
4418
4419         if (!pgm &&
4420             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4421             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4422                 /*
4423                  * a filepair that changes between file and symlink
4424                  * needs to be split into deletion and creation.
4425                  */
4426                 struct diff_filespec *null = alloc_filespec(two->path);
4427                 run_diff_cmd(NULL, name, other, attr_path,
4428                              one, null, &msg,
4429                              o, p);
4430                 free(null);
4431                 strbuf_release(&msg);
4432
4433                 null = alloc_filespec(one->path);
4434                 run_diff_cmd(NULL, name, other, attr_path,
4435                              null, two, &msg, o, p);
4436                 free(null);
4437         }
4438         else
4439                 run_diff_cmd(pgm, name, other, attr_path,
4440                              one, two, &msg, o, p);
4441
4442         strbuf_release(&msg);
4443 }
4444
4445 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4446                          struct diffstat_t *diffstat)
4447 {
4448         const char *name;
4449         const char *other;
4450
4451         if (DIFF_PAIR_UNMERGED(p)) {
4452                 /* unmerged */
4453                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4454                                  diffstat, o, p);
4455                 return;
4456         }
4457
4458         name = p->one->path;
4459         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4460
4461         if (o->prefix_length)
4462                 strip_prefix(o->prefix_length, &name, &other);
4463
4464         diff_fill_oid_info(p->one, o->repo->index);
4465         diff_fill_oid_info(p->two, o->repo->index);
4466
4467         builtin_diffstat(name, other, p->one, p->two,
4468                          diffstat, o, p);
4469 }
4470
4471 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4472 {
4473         const char *name;
4474         const char *other;
4475         const char *attr_path;
4476
4477         if (DIFF_PAIR_UNMERGED(p)) {
4478                 /* unmerged */
4479                 return;
4480         }
4481
4482         name = p->one->path;
4483         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4484         attr_path = other ? other : name;
4485
4486         if (o->prefix_length)
4487                 strip_prefix(o->prefix_length, &name, &other);
4488
4489         diff_fill_oid_info(p->one, o->repo->index);
4490         diff_fill_oid_info(p->two, o->repo->index);
4491
4492         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4493 }
4494
4495 static void prep_parse_options(struct diff_options *options);
4496
4497 void repo_diff_setup(struct repository *r, struct diff_options *options)
4498 {
4499         memcpy(options, &default_diff_options, sizeof(*options));
4500
4501         options->file = stdout;
4502         options->repo = r;
4503
4504         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4505         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4506         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4507         options->abbrev = DEFAULT_ABBREV;
4508         options->line_termination = '\n';
4509         options->break_opt = -1;
4510         options->rename_limit = -1;
4511         options->dirstat_permille = diff_dirstat_permille_default;
4512         options->context = diff_context_default;
4513         options->interhunkcontext = diff_interhunk_context_default;
4514         options->ws_error_highlight = ws_error_highlight_default;
4515         options->flags.rename_empty = 1;
4516         options->objfind = NULL;
4517
4518         /* pathchange left =NULL by default */
4519         options->change = diff_change;
4520         options->add_remove = diff_addremove;
4521         options->use_color = diff_use_color_default;
4522         options->detect_rename = diff_detect_rename_default;
4523         options->xdl_opts |= diff_algorithm;
4524         if (diff_indent_heuristic)
4525                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4526
4527         options->orderfile = diff_order_file_cfg;
4528
4529         if (diff_no_prefix) {
4530                 options->a_prefix = options->b_prefix = "";
4531         } else if (!diff_mnemonic_prefix) {
4532                 options->a_prefix = "a/";
4533                 options->b_prefix = "b/";
4534         }
4535
4536         options->color_moved = diff_color_moved_default;
4537         options->color_moved_ws_handling = diff_color_moved_ws_default;
4538
4539         prep_parse_options(options);
4540 }
4541
4542 void diff_setup_done(struct diff_options *options)
4543 {
4544         unsigned check_mask = DIFF_FORMAT_NAME |
4545                               DIFF_FORMAT_NAME_STATUS |
4546                               DIFF_FORMAT_CHECKDIFF |
4547                               DIFF_FORMAT_NO_OUTPUT;
4548         /*
4549          * This must be signed because we're comparing against a potentially
4550          * negative value.
4551          */
4552         const int hexsz = the_hash_algo->hexsz;
4553
4554         if (options->set_default)
4555                 options->set_default(options);
4556
4557         if (HAS_MULTI_BITS(options->output_format & check_mask))
4558                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4559
4560         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4561                 die(_("-G, -S and --find-object are mutually exclusive"));
4562
4563         /*
4564          * Most of the time we can say "there are changes"
4565          * only by checking if there are changed paths, but
4566          * --ignore-whitespace* options force us to look
4567          * inside contents.
4568          */
4569
4570         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4571                 options->flags.diff_from_contents = 1;
4572         else
4573                 options->flags.diff_from_contents = 0;
4574
4575         if (options->flags.find_copies_harder)
4576                 options->detect_rename = DIFF_DETECT_COPY;
4577
4578         if (!options->flags.relative_name)
4579                 options->prefix = NULL;
4580         if (options->prefix)
4581                 options->prefix_length = strlen(options->prefix);
4582         else
4583                 options->prefix_length = 0;
4584
4585         if (options->output_format & (DIFF_FORMAT_NAME |
4586                                       DIFF_FORMAT_NAME_STATUS |
4587                                       DIFF_FORMAT_CHECKDIFF |
4588                                       DIFF_FORMAT_NO_OUTPUT))
4589                 options->output_format &= ~(DIFF_FORMAT_RAW |
4590                                             DIFF_FORMAT_NUMSTAT |
4591                                             DIFF_FORMAT_DIFFSTAT |
4592                                             DIFF_FORMAT_SHORTSTAT |
4593                                             DIFF_FORMAT_DIRSTAT |
4594                                             DIFF_FORMAT_SUMMARY |
4595                                             DIFF_FORMAT_PATCH);
4596
4597         /*
4598          * These cases always need recursive; we do not drop caller-supplied
4599          * recursive bits for other formats here.
4600          */
4601         if (options->output_format & (DIFF_FORMAT_PATCH |
4602                                       DIFF_FORMAT_NUMSTAT |
4603                                       DIFF_FORMAT_DIFFSTAT |
4604                                       DIFF_FORMAT_SHORTSTAT |
4605                                       DIFF_FORMAT_DIRSTAT |
4606                                       DIFF_FORMAT_SUMMARY |
4607                                       DIFF_FORMAT_CHECKDIFF))
4608                 options->flags.recursive = 1;
4609         /*
4610          * Also pickaxe would not work very well if you do not say recursive
4611          */
4612         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4613                 options->flags.recursive = 1;
4614         /*
4615          * When patches are generated, submodules diffed against the work tree
4616          * must be checked for dirtiness too so it can be shown in the output
4617          */
4618         if (options->output_format & DIFF_FORMAT_PATCH)
4619                 options->flags.dirty_submodules = 1;
4620
4621         if (options->detect_rename && options->rename_limit < 0)
4622                 options->rename_limit = diff_rename_limit_default;
4623         if (hexsz < options->abbrev)
4624                 options->abbrev = hexsz; /* full */
4625
4626         /*
4627          * It does not make sense to show the first hit we happened
4628          * to have found.  It does not make sense not to return with
4629          * exit code in such a case either.
4630          */
4631         if (options->flags.quick) {
4632                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4633                 options->flags.exit_with_status = 1;
4634         }
4635
4636         options->diff_path_counter = 0;
4637
4638         if (options->flags.follow_renames && options->pathspec.nr != 1)
4639                 die(_("--follow requires exactly one pathspec"));
4640
4641         if (!options->use_color || external_diff())
4642                 options->color_moved = 0;
4643
4644         FREE_AND_NULL(options->parseopts);
4645 }
4646
4647 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4648 {
4649         char c, *eq;
4650         int len;
4651
4652         if (*arg != '-')
4653                 return 0;
4654         c = *++arg;
4655         if (!c)
4656                 return 0;
4657         if (c == arg_short) {
4658                 c = *++arg;
4659                 if (!c)
4660                         return 1;
4661                 if (val && isdigit(c)) {
4662                         char *end;
4663                         int n = strtoul(arg, &end, 10);
4664                         if (*end)
4665                                 return 0;
4666                         *val = n;
4667                         return 1;
4668                 }
4669                 return 0;
4670         }
4671         if (c != '-')
4672                 return 0;
4673         arg++;
4674         eq = strchrnul(arg, '=');
4675         len = eq - arg;
4676         if (!len || strncmp(arg, arg_long, len))
4677                 return 0;
4678         if (*eq) {
4679                 int n;
4680                 char *end;
4681                 if (!isdigit(*++eq))
4682                         return 0;
4683                 n = strtoul(eq, &end, 10);
4684                 if (*end)
4685                         return 0;
4686                 *val = n;
4687         }
4688         return 1;
4689 }
4690
4691 static int diff_scoreopt_parse(const char *opt);
4692
4693 static inline int short_opt(char opt, const char **argv,
4694                             const char **optarg)
4695 {
4696         const char *arg = argv[0];
4697         if (arg[0] != '-' || arg[1] != opt)
4698                 return 0;
4699         if (arg[2] != '\0') {
4700                 *optarg = arg + 2;
4701                 return 1;
4702         }
4703         if (!argv[1])
4704                 die("Option '%c' requires a value", opt);
4705         *optarg = argv[1];
4706         return 2;
4707 }
4708
4709 int parse_long_opt(const char *opt, const char **argv,
4710                    const char **optarg)
4711 {
4712         const char *arg = argv[0];
4713         if (!skip_prefix(arg, "--", &arg))
4714                 return 0;
4715         if (!skip_prefix(arg, opt, &arg))
4716                 return 0;
4717         if (*arg == '=') { /* stuck form: --option=value */
4718                 *optarg = arg + 1;
4719                 return 1;
4720         }
4721         if (*arg != '\0')
4722                 return 0;
4723         /* separate form: --option value */
4724         if (!argv[1])
4725                 die("Option '--%s' requires a value", opt);
4726         *optarg = argv[1];
4727         return 2;
4728 }
4729
4730 static int stat_opt(struct diff_options *options, const char **av)
4731 {
4732         const char *arg = av[0];
4733         char *end;
4734         int width = options->stat_width;
4735         int name_width = options->stat_name_width;
4736         int graph_width = options->stat_graph_width;
4737         int count = options->stat_count;
4738         int argcount = 1;
4739
4740         if (!skip_prefix(arg, "--stat", &arg))
4741                 BUG("stat option does not begin with --stat: %s", arg);
4742         end = (char *)arg;
4743
4744         switch (*arg) {
4745         case '-':
4746                 if (skip_prefix(arg, "-width", &arg)) {
4747                         if (*arg == '=')
4748                                 width = strtoul(arg + 1, &end, 10);
4749                         else if (!*arg && !av[1])
4750                                 die_want_option("--stat-width");
4751                         else if (!*arg) {
4752                                 width = strtoul(av[1], &end, 10);
4753                                 argcount = 2;
4754                         }
4755                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4756                         if (*arg == '=')
4757                                 name_width = strtoul(arg + 1, &end, 10);
4758                         else if (!*arg && !av[1])
4759                                 die_want_option("--stat-name-width");
4760                         else if (!*arg) {
4761                                 name_width = strtoul(av[1], &end, 10);
4762                                 argcount = 2;
4763                         }
4764                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4765                         if (*arg == '=')
4766                                 graph_width = strtoul(arg + 1, &end, 10);
4767                         else if (!*arg && !av[1])
4768                                 die_want_option("--stat-graph-width");
4769                         else if (!*arg) {
4770                                 graph_width = strtoul(av[1], &end, 10);
4771                                 argcount = 2;
4772                         }
4773                 } else if (skip_prefix(arg, "-count", &arg)) {
4774                         if (*arg == '=')
4775                                 count = strtoul(arg + 1, &end, 10);
4776                         else if (!*arg && !av[1])
4777                                 die_want_option("--stat-count");
4778                         else if (!*arg) {
4779                                 count = strtoul(av[1], &end, 10);
4780                                 argcount = 2;
4781                         }
4782                 }
4783                 break;
4784         case '=':
4785                 width = strtoul(arg+1, &end, 10);
4786                 if (*end == ',')
4787                         name_width = strtoul(end+1, &end, 10);
4788                 if (*end == ',')
4789                         count = strtoul(end+1, &end, 10);
4790         }
4791
4792         /* Important! This checks all the error cases! */
4793         if (*end)
4794                 return 0;
4795         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4796         options->stat_name_width = name_width;
4797         options->stat_graph_width = graph_width;
4798         options->stat_width = width;
4799         options->stat_count = count;
4800         return argcount;
4801 }
4802
4803 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4804 {
4805         struct strbuf errmsg = STRBUF_INIT;
4806         if (parse_dirstat_params(options, params, &errmsg))
4807                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4808                     errmsg.buf);
4809         strbuf_release(&errmsg);
4810         /*
4811          * The caller knows a dirstat-related option is given from the command
4812          * line; allow it to say "return this_function();"
4813          */
4814         options->output_format |= DIFF_FORMAT_DIRSTAT;
4815         return 1;
4816 }
4817
4818 static int parse_submodule_opt(struct diff_options *options, const char *value)
4819 {
4820         if (parse_submodule_params(options, value))
4821                 die(_("Failed to parse --submodule option parameter: '%s'"),
4822                         value);
4823         return 1;
4824 }
4825
4826 static const char diff_status_letters[] = {
4827         DIFF_STATUS_ADDED,
4828         DIFF_STATUS_COPIED,
4829         DIFF_STATUS_DELETED,
4830         DIFF_STATUS_MODIFIED,
4831         DIFF_STATUS_RENAMED,
4832         DIFF_STATUS_TYPE_CHANGED,
4833         DIFF_STATUS_UNKNOWN,
4834         DIFF_STATUS_UNMERGED,
4835         DIFF_STATUS_FILTER_AON,
4836         DIFF_STATUS_FILTER_BROKEN,
4837         '\0',
4838 };
4839
4840 static unsigned int filter_bit['Z' + 1];
4841
4842 static void prepare_filter_bits(void)
4843 {
4844         int i;
4845
4846         if (!filter_bit[DIFF_STATUS_ADDED]) {
4847                 for (i = 0; diff_status_letters[i]; i++)
4848                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4849         }
4850 }
4851
4852 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4853 {
4854         return opt->filter & filter_bit[(int) status];
4855 }
4856
4857 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4858 {
4859         int i, optch;
4860
4861         prepare_filter_bits();
4862
4863         /*
4864          * If there is a negation e.g. 'd' in the input, and we haven't
4865          * initialized the filter field with another --diff-filter, start
4866          * from full set of bits, except for AON.
4867          */
4868         if (!opt->filter) {
4869                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4870                         if (optch < 'a' || 'z' < optch)
4871                                 continue;
4872                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4873                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4874                         break;
4875                 }
4876         }
4877
4878         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4879                 unsigned int bit;
4880                 int negate;
4881
4882                 if ('a' <= optch && optch <= 'z') {
4883                         negate = 1;
4884                         optch = toupper(optch);
4885                 } else {
4886                         negate = 0;
4887                 }
4888
4889                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4890                 if (!bit)
4891                         return optarg[i];
4892                 if (negate)
4893                         opt->filter &= ~bit;
4894                 else
4895                         opt->filter |= bit;
4896         }
4897         return 0;
4898 }
4899
4900 static void enable_patch_output(int *fmt)
4901 {
4902         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4903         *fmt |= DIFF_FORMAT_PATCH;
4904 }
4905
4906 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4907 {
4908         int val = parse_ws_error_highlight(arg);
4909
4910         if (val < 0) {
4911                 error("unknown value after ws-error-highlight=%.*s",
4912                       -1 - val, arg);
4913                 return 0;
4914         }
4915         opt->ws_error_highlight = val;
4916         return 1;
4917 }
4918
4919 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4920 {
4921         struct object_id oid;
4922
4923         if (get_oid(arg, &oid))
4924                 return error("unable to resolve '%s'", arg);
4925
4926         if (!opt->objfind)
4927                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4928
4929         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4930         opt->flags.recursive = 1;
4931         opt->flags.tree_in_recursive = 1;
4932         oidset_insert(opt->objfind, &oid);
4933         return 1;
4934 }
4935
4936 static int diff_opt_unified(const struct option *opt,
4937                             const char *arg, int unset)
4938 {
4939         struct diff_options *options = opt->value;
4940         char *s;
4941
4942         BUG_ON_OPT_NEG(unset);
4943
4944         options->context = strtol(arg, &s, 10);
4945         if (*s)
4946                 return error(_("%s expects a numerical value"), "--unified");
4947         enable_patch_output(&options->output_format);
4948
4949         return 0;
4950 }
4951
4952 static void prep_parse_options(struct diff_options *options)
4953 {
4954         struct option parseopts[] = {
4955                 OPT_GROUP(N_("Diff output format options")),
4956                 OPT_BITOP('p', "patch", &options->output_format,
4957                           N_("generate patch"),
4958                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4959                 OPT_BITOP('u', NULL, &options->output_format,
4960                           N_("generate patch"),
4961                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
4962                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
4963                                N_("generate diffs with <n> lines context"),
4964                                PARSE_OPT_NONEG, diff_opt_unified),
4965                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
4966                          N_("generate diffs with <n> lines context")),
4967                 OPT_BIT_F(0, "raw", &options->output_format,
4968                           N_("generate the diff in raw format"),
4969                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
4970                 OPT_END()
4971         };
4972
4973         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
4974         memcpy(options->parseopts, parseopts, sizeof(parseopts));
4975 }
4976
4977 int diff_opt_parse(struct diff_options *options,
4978                    const char **av, int ac, const char *prefix)
4979 {
4980         const char *arg = av[0];
4981         const char *optarg;
4982         int argcount;
4983
4984         if (!prefix)
4985                 prefix = "";
4986
4987         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
4988                            PARSE_OPT_KEEP_DASHDASH |
4989                            PARSE_OPT_KEEP_UNKNOWN |
4990                            PARSE_OPT_NO_INTERNAL_HELP |
4991                            PARSE_OPT_ONE_SHOT |
4992                            PARSE_OPT_STOP_AT_NON_OPTION);
4993
4994         if (ac)
4995                 return ac;
4996
4997         /* Output format options */
4998         if (!strcmp(arg, "--patch-with-raw")) {
4999                 enable_patch_output(&options->output_format);
5000                 options->output_format |= DIFF_FORMAT_RAW;
5001         } else if (!strcmp(arg, "--numstat"))
5002                 options->output_format |= DIFF_FORMAT_NUMSTAT;
5003         else if (!strcmp(arg, "--shortstat"))
5004                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
5005         else if (skip_prefix(arg, "-X", &arg) ||
5006                  skip_to_optional_arg(arg, "--dirstat", &arg))
5007                 return parse_dirstat_opt(options, arg);
5008         else if (!strcmp(arg, "--cumulative"))
5009                 return parse_dirstat_opt(options, "cumulative");
5010         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
5011                 parse_dirstat_opt(options, "files");
5012                 return parse_dirstat_opt(options, arg);
5013         }
5014         else if (!strcmp(arg, "--check"))
5015                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
5016         else if (!strcmp(arg, "--summary"))
5017                 options->output_format |= DIFF_FORMAT_SUMMARY;
5018         else if (!strcmp(arg, "--patch-with-stat")) {
5019                 enable_patch_output(&options->output_format);
5020                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5021         } else if (!strcmp(arg, "--name-only"))
5022                 options->output_format |= DIFF_FORMAT_NAME;
5023         else if (!strcmp(arg, "--name-status"))
5024                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
5025         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
5026                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
5027         else if (starts_with(arg, "--stat"))
5028                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
5029                 return stat_opt(options, av);
5030         else if (!strcmp(arg, "--compact-summary")) {
5031                  options->flags.stat_with_summary = 1;
5032                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
5033         } else if (!strcmp(arg, "--no-compact-summary"))
5034                  options->flags.stat_with_summary = 0;
5035         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
5036                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
5037         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
5038                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
5039         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
5040                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
5041
5042         /* renames options */
5043         else if (starts_with(arg, "-B") ||
5044                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
5045                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
5046                         return error("invalid argument to -B: %s", arg+2);
5047         }
5048         else if (starts_with(arg, "-M") ||
5049                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
5050                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5051                         return error("invalid argument to -M: %s", arg+2);
5052                 options->detect_rename = DIFF_DETECT_RENAME;
5053         }
5054         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
5055                 options->irreversible_delete = 1;
5056         }
5057         else if (starts_with(arg, "-C") ||
5058                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
5059                 if (options->detect_rename == DIFF_DETECT_COPY)
5060                         options->flags.find_copies_harder = 1;
5061                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
5062                         return error("invalid argument to -C: %s", arg+2);
5063                 options->detect_rename = DIFF_DETECT_COPY;
5064         }
5065         else if (!strcmp(arg, "--no-renames"))
5066                 options->detect_rename = 0;
5067         else if (!strcmp(arg, "--rename-empty"))
5068                 options->flags.rename_empty = 1;
5069         else if (!strcmp(arg, "--no-rename-empty"))
5070                 options->flags.rename_empty = 0;
5071         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
5072                 options->flags.relative_name = 1;
5073                 if (arg)
5074                         options->prefix = arg;
5075         }
5076
5077         /* xdiff options */
5078         else if (!strcmp(arg, "--minimal"))
5079                 DIFF_XDL_SET(options, NEED_MINIMAL);
5080         else if (!strcmp(arg, "--no-minimal"))
5081                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5082         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
5083                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
5084         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
5085                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
5086         else if (!strcmp(arg, "--ignore-space-at-eol"))
5087                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
5088         else if (!strcmp(arg, "--ignore-cr-at-eol"))
5089                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5090         else if (!strcmp(arg, "--ignore-blank-lines"))
5091                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5092         else if (!strcmp(arg, "--indent-heuristic"))
5093                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5094         else if (!strcmp(arg, "--no-indent-heuristic"))
5095                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5096         else if (!strcmp(arg, "--patience")) {
5097                 int i;
5098                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5099                 /*
5100                  * Both --patience and --anchored use PATIENCE_DIFF
5101                  * internally, so remove any anchors previously
5102                  * specified.
5103                  */
5104                 for (i = 0; i < options->anchors_nr; i++)
5105                         free(options->anchors[i]);
5106                 options->anchors_nr = 0;
5107         } else if (!strcmp(arg, "--histogram"))
5108                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5109         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5110                 long value = parse_algorithm_value(optarg);
5111                 if (value < 0)
5112                         return error("option diff-algorithm accepts \"myers\", "
5113                                      "\"minimal\", \"patience\" and \"histogram\"");
5114                 /* clear out previous settings */
5115                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5116                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5117                 options->xdl_opts |= value;
5118                 return argcount;
5119         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5120                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5121                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5122                            options->anchors_alloc);
5123                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5124         }
5125
5126         /* flags options */
5127         else if (!strcmp(arg, "--binary")) {
5128                 enable_patch_output(&options->output_format);
5129                 options->flags.binary = 1;
5130         }
5131         else if (!strcmp(arg, "--full-index"))
5132                 options->flags.full_index = 1;
5133         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5134                 options->flags.text = 1;
5135         else if (!strcmp(arg, "-R"))
5136                 options->flags.reverse_diff = 1;
5137         else if (!strcmp(arg, "--find-copies-harder"))
5138                 options->flags.find_copies_harder = 1;
5139         else if (!strcmp(arg, "--follow"))
5140                 options->flags.follow_renames = 1;
5141         else if (!strcmp(arg, "--no-follow")) {
5142                 options->flags.follow_renames = 0;
5143                 options->flags.default_follow_renames = 0;
5144         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5145                 int value = git_config_colorbool(NULL, arg);
5146                 if (value < 0)
5147                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5148                 options->use_color = value;
5149         }
5150         else if (!strcmp(arg, "--no-color"))
5151                 options->use_color = 0;
5152         else if (!strcmp(arg, "--color-moved")) {
5153                 if (diff_color_moved_default)
5154                         options->color_moved = diff_color_moved_default;
5155                 if (options->color_moved == COLOR_MOVED_NO)
5156                         options->color_moved = COLOR_MOVED_DEFAULT;
5157         } else if (!strcmp(arg, "--no-color-moved"))
5158                 options->color_moved = COLOR_MOVED_NO;
5159         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5160                 int cm = parse_color_moved(arg);
5161                 if (cm < 0)
5162                         return error("bad --color-moved argument: %s", arg);
5163                 options->color_moved = cm;
5164         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5165                 options->color_moved_ws_handling = 0;
5166         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5167                 unsigned cm = parse_color_moved_ws(arg);
5168                 if (cm & COLOR_MOVED_WS_ERROR)
5169                         return -1;
5170                 options->color_moved_ws_handling = cm;
5171         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5172                 options->use_color = 1;
5173                 options->word_diff = DIFF_WORDS_COLOR;
5174         }
5175         else if (!strcmp(arg, "--word-diff")) {
5176                 if (options->word_diff == DIFF_WORDS_NONE)
5177                         options->word_diff = DIFF_WORDS_PLAIN;
5178         }
5179         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5180                 if (!strcmp(arg, "plain"))
5181                         options->word_diff = DIFF_WORDS_PLAIN;
5182                 else if (!strcmp(arg, "color")) {
5183                         options->use_color = 1;
5184                         options->word_diff = DIFF_WORDS_COLOR;
5185                 }
5186                 else if (!strcmp(arg, "porcelain"))
5187                         options->word_diff = DIFF_WORDS_PORCELAIN;
5188                 else if (!strcmp(arg, "none"))
5189                         options->word_diff = DIFF_WORDS_NONE;
5190                 else
5191                         die("bad --word-diff argument: %s", arg);
5192         }
5193         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5194                 if (options->word_diff == DIFF_WORDS_NONE)
5195                         options->word_diff = DIFF_WORDS_PLAIN;
5196                 options->word_regex = optarg;
5197                 return argcount;
5198         }
5199         else if (!strcmp(arg, "--exit-code"))
5200                 options->flags.exit_with_status = 1;
5201         else if (!strcmp(arg, "--quiet"))
5202                 options->flags.quick = 1;
5203         else if (!strcmp(arg, "--ext-diff"))
5204                 options->flags.allow_external = 1;
5205         else if (!strcmp(arg, "--no-ext-diff"))
5206                 options->flags.allow_external = 0;
5207         else if (!strcmp(arg, "--textconv")) {
5208                 options->flags.allow_textconv = 1;
5209                 options->flags.textconv_set_via_cmdline = 1;
5210         } else if (!strcmp(arg, "--no-textconv"))
5211                 options->flags.allow_textconv = 0;
5212         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5213                 options->flags.override_submodule_config = 1;
5214                 handle_ignore_submodules_arg(options, arg);
5215         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5216                 return parse_submodule_opt(options, arg);
5217         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5218                 return parse_ws_error_highlight_opt(options, arg);
5219         else if (!strcmp(arg, "--ita-invisible-in-index"))
5220                 options->ita_invisible_in_index = 1;
5221         else if (!strcmp(arg, "--ita-visible-in-index"))
5222                 options->ita_invisible_in_index = 0;
5223
5224         /* misc options */
5225         else if (!strcmp(arg, "-z"))
5226                 options->line_termination = 0;
5227         else if ((argcount = short_opt('l', av, &optarg))) {
5228                 options->rename_limit = strtoul(optarg, NULL, 10);
5229                 return argcount;
5230         }
5231         else if ((argcount = short_opt('S', av, &optarg))) {
5232                 options->pickaxe = optarg;
5233                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5234                 return argcount;
5235         } else if ((argcount = short_opt('G', av, &optarg))) {
5236                 options->pickaxe = optarg;
5237                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5238                 return argcount;
5239         }
5240         else if (!strcmp(arg, "--pickaxe-all"))
5241                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5242         else if (!strcmp(arg, "--pickaxe-regex"))
5243                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5244         else if ((argcount = short_opt('O', av, &optarg))) {
5245                 options->orderfile = prefix_filename(prefix, optarg);
5246                 return argcount;
5247         } else if (skip_prefix(arg, "--find-object=", &arg))
5248                 return parse_objfind_opt(options, arg);
5249         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5250                 int offending = parse_diff_filter_opt(optarg, options);
5251                 if (offending)
5252                         die("unknown change class '%c' in --diff-filter=%s",
5253                             offending, optarg);
5254                 return argcount;
5255         }
5256         else if (!strcmp(arg, "--no-abbrev"))
5257                 options->abbrev = 0;
5258         else if (!strcmp(arg, "--abbrev"))
5259                 options->abbrev = DEFAULT_ABBREV;
5260         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5261                 options->abbrev = strtoul(arg, NULL, 10);
5262                 if (options->abbrev < MINIMUM_ABBREV)
5263                         options->abbrev = MINIMUM_ABBREV;
5264                 else if (the_hash_algo->hexsz < options->abbrev)
5265                         options->abbrev = the_hash_algo->hexsz;
5266         }
5267         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5268                 options->a_prefix = optarg;
5269                 return argcount;
5270         }
5271         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5272                 options->line_prefix = optarg;
5273                 options->line_prefix_length = strlen(options->line_prefix);
5274                 graph_setup_line_prefix(options);
5275                 return argcount;
5276         }
5277         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5278                 options->b_prefix = optarg;
5279                 return argcount;
5280         }
5281         else if (!strcmp(arg, "--no-prefix"))
5282                 options->a_prefix = options->b_prefix = "";
5283         else if (opt_arg(arg, '\0', "inter-hunk-context",
5284                          &options->interhunkcontext))
5285                 ;
5286         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5287                 char *path = prefix_filename(prefix, optarg);
5288                 options->file = xfopen(path, "w");
5289                 options->close_file = 1;
5290                 if (options->use_color != GIT_COLOR_ALWAYS)
5291                         options->use_color = GIT_COLOR_NEVER;
5292                 free(path);
5293                 return argcount;
5294         } else
5295                 return 0;
5296         return 1;
5297 }
5298
5299 int parse_rename_score(const char **cp_p)
5300 {
5301         unsigned long num, scale;
5302         int ch, dot;
5303         const char *cp = *cp_p;
5304
5305         num = 0;
5306         scale = 1;
5307         dot = 0;
5308         for (;;) {
5309                 ch = *cp;
5310                 if ( !dot && ch == '.' ) {
5311                         scale = 1;
5312                         dot = 1;
5313                 } else if ( ch == '%' ) {
5314                         scale = dot ? scale*100 : 100;
5315                         cp++;   /* % is always at the end */
5316                         break;
5317                 } else if ( ch >= '0' && ch <= '9' ) {
5318                         if ( scale < 100000 ) {
5319                                 scale *= 10;
5320                                 num = (num*10) + (ch-'0');
5321                         }
5322                 } else {
5323                         break;
5324                 }
5325                 cp++;
5326         }
5327         *cp_p = cp;
5328
5329         /* user says num divided by scale and we say internally that
5330          * is MAX_SCORE * num / scale.
5331          */
5332         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5333 }
5334
5335 static int diff_scoreopt_parse(const char *opt)
5336 {
5337         int opt1, opt2, cmd;
5338
5339         if (*opt++ != '-')
5340                 return -1;
5341         cmd = *opt++;
5342         if (cmd == '-') {
5343                 /* convert the long-form arguments into short-form versions */
5344                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5345                         if (*opt == 0 || *opt++ == '=')
5346                                 cmd = 'B';
5347                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5348                         if (*opt == 0 || *opt++ == '=')
5349                                 cmd = 'C';
5350                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5351                         if (*opt == 0 || *opt++ == '=')
5352                                 cmd = 'M';
5353                 }
5354         }
5355         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5356                 return -1; /* that is not a -M, -C, or -B option */
5357
5358         opt1 = parse_rename_score(&opt);
5359         if (cmd != 'B')
5360                 opt2 = 0;
5361         else {
5362                 if (*opt == 0)
5363                         opt2 = 0;
5364                 else if (*opt != '/')
5365                         return -1; /* we expect -B80/99 or -B80 */
5366                 else {
5367                         opt++;
5368                         opt2 = parse_rename_score(&opt);
5369                 }
5370         }
5371         if (*opt != 0)
5372                 return -1;
5373         return opt1 | (opt2 << 16);
5374 }
5375
5376 struct diff_queue_struct diff_queued_diff;
5377
5378 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5379 {
5380         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5381         queue->queue[queue->nr++] = dp;
5382 }
5383
5384 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5385                                  struct diff_filespec *one,
5386                                  struct diff_filespec *two)
5387 {
5388         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5389         dp->one = one;
5390         dp->two = two;
5391         if (queue)
5392                 diff_q(queue, dp);
5393         return dp;
5394 }
5395
5396 void diff_free_filepair(struct diff_filepair *p)
5397 {
5398         free_filespec(p->one);
5399         free_filespec(p->two);
5400         free(p);
5401 }
5402
5403 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5404 {
5405         int abblen;
5406         const char *abbrev;
5407
5408         /* Do we want all 40 hex characters? */
5409         if (len == the_hash_algo->hexsz)
5410                 return oid_to_hex(oid);
5411
5412         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5413         abbrev = diff_abbrev_oid(oid, len);
5414
5415         if (!print_sha1_ellipsis())
5416                 return abbrev;
5417
5418         abblen = strlen(abbrev);
5419
5420         /*
5421          * In well-behaved cases, where the abbreviated result is the
5422          * same as the requested length, append three dots after the
5423          * abbreviation (hence the whole logic is limited to the case
5424          * where abblen < 37); when the actual abbreviated result is a
5425          * bit longer than the requested length, we reduce the number
5426          * of dots so that they match the well-behaved ones.  However,
5427          * if the actual abbreviation is longer than the requested
5428          * length by more than three, we give up on aligning, and add
5429          * three dots anyway, to indicate that the output is not the
5430          * full object name.  Yes, this may be suboptimal, but this
5431          * appears only in "diff --raw --abbrev" output and it is not
5432          * worth the effort to change it now.  Note that this would
5433          * likely to work fine when the automatic sizing of default
5434          * abbreviation length is used--we would be fed -1 in "len" in
5435          * that case, and will end up always appending three-dots, but
5436          * the automatic sizing is supposed to give abblen that ensures
5437          * uniqueness across all objects (statistically speaking).
5438          */
5439         if (abblen < the_hash_algo->hexsz - 3) {
5440                 static char hex[GIT_MAX_HEXSZ + 1];
5441                 if (len < abblen && abblen <= len + 2)
5442                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5443                 else
5444                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5445                 return hex;
5446         }
5447
5448         return oid_to_hex(oid);
5449 }
5450
5451 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5452 {
5453         int line_termination = opt->line_termination;
5454         int inter_name_termination = line_termination ? '\t' : '\0';
5455
5456         fprintf(opt->file, "%s", diff_line_prefix(opt));
5457         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5458                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5459                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5460                 fprintf(opt->file, "%s ",
5461                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5462         }
5463         if (p->score) {
5464                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5465                         inter_name_termination);
5466         } else {
5467                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5468         }
5469
5470         if (p->status == DIFF_STATUS_COPIED ||
5471             p->status == DIFF_STATUS_RENAMED) {
5472                 const char *name_a, *name_b;
5473                 name_a = p->one->path;
5474                 name_b = p->two->path;
5475                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5476                 write_name_quoted(name_a, opt->file, inter_name_termination);
5477                 write_name_quoted(name_b, opt->file, line_termination);
5478         } else {
5479                 const char *name_a, *name_b;
5480                 name_a = p->one->mode ? p->one->path : p->two->path;
5481                 name_b = NULL;
5482                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5483                 write_name_quoted(name_a, opt->file, line_termination);
5484         }
5485 }
5486
5487 int diff_unmodified_pair(struct diff_filepair *p)
5488 {
5489         /* This function is written stricter than necessary to support
5490          * the currently implemented transformers, but the idea is to
5491          * let transformers to produce diff_filepairs any way they want,
5492          * and filter and clean them up here before producing the output.
5493          */
5494         struct diff_filespec *one = p->one, *two = p->two;
5495
5496         if (DIFF_PAIR_UNMERGED(p))
5497                 return 0; /* unmerged is interesting */
5498
5499         /* deletion, addition, mode or type change
5500          * and rename are all interesting.
5501          */
5502         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5503             DIFF_PAIR_MODE_CHANGED(p) ||
5504             strcmp(one->path, two->path))
5505                 return 0;
5506
5507         /* both are valid and point at the same path.  that is, we are
5508          * dealing with a change.
5509          */
5510         if (one->oid_valid && two->oid_valid &&
5511             oideq(&one->oid, &two->oid) &&
5512             !one->dirty_submodule && !two->dirty_submodule)
5513                 return 1; /* no change */
5514         if (!one->oid_valid && !two->oid_valid)
5515                 return 1; /* both look at the same file on the filesystem. */
5516         return 0;
5517 }
5518
5519 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5520 {
5521         if (diff_unmodified_pair(p))
5522                 return;
5523
5524         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5525             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5526                 return; /* no tree diffs in patch format */
5527
5528         run_diff(p, o);
5529 }
5530
5531 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5532                             struct diffstat_t *diffstat)
5533 {
5534         if (diff_unmodified_pair(p))
5535                 return;
5536
5537         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5538             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5539                 return; /* no useful stat for tree diffs */
5540
5541         run_diffstat(p, o, diffstat);
5542 }
5543
5544 static void diff_flush_checkdiff(struct diff_filepair *p,
5545                 struct diff_options *o)
5546 {
5547         if (diff_unmodified_pair(p))
5548                 return;
5549
5550         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5551             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5552                 return; /* nothing to check in tree diffs */
5553
5554         run_checkdiff(p, o);
5555 }
5556
5557 int diff_queue_is_empty(void)
5558 {
5559         struct diff_queue_struct *q = &diff_queued_diff;
5560         int i;
5561         for (i = 0; i < q->nr; i++)
5562                 if (!diff_unmodified_pair(q->queue[i]))
5563                         return 0;
5564         return 1;
5565 }
5566
5567 #if DIFF_DEBUG
5568 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5569 {
5570         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5571                 x, one ? one : "",
5572                 s->path,
5573                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5574                 s->mode,
5575                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5576         fprintf(stderr, "queue[%d] %s size %lu\n",
5577                 x, one ? one : "",
5578                 s->size);
5579 }
5580
5581 void diff_debug_filepair(const struct diff_filepair *p, int i)
5582 {
5583         diff_debug_filespec(p->one, i, "one");
5584         diff_debug_filespec(p->two, i, "two");
5585         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5586                 p->score, p->status ? p->status : '?',
5587                 p->one->rename_used, p->broken_pair);
5588 }
5589
5590 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5591 {
5592         int i;
5593         if (msg)
5594                 fprintf(stderr, "%s\n", msg);
5595         fprintf(stderr, "q->nr = %d\n", q->nr);
5596         for (i = 0; i < q->nr; i++) {
5597                 struct diff_filepair *p = q->queue[i];
5598                 diff_debug_filepair(p, i);
5599         }
5600 }
5601 #endif
5602
5603 static void diff_resolve_rename_copy(void)
5604 {
5605         int i;
5606         struct diff_filepair *p;
5607         struct diff_queue_struct *q = &diff_queued_diff;
5608
5609         diff_debug_queue("resolve-rename-copy", q);
5610
5611         for (i = 0; i < q->nr; i++) {
5612                 p = q->queue[i];
5613                 p->status = 0; /* undecided */
5614                 if (DIFF_PAIR_UNMERGED(p))
5615                         p->status = DIFF_STATUS_UNMERGED;
5616                 else if (!DIFF_FILE_VALID(p->one))
5617                         p->status = DIFF_STATUS_ADDED;
5618                 else if (!DIFF_FILE_VALID(p->two))
5619                         p->status = DIFF_STATUS_DELETED;
5620                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5621                         p->status = DIFF_STATUS_TYPE_CHANGED;
5622
5623                 /* from this point on, we are dealing with a pair
5624                  * whose both sides are valid and of the same type, i.e.
5625                  * either in-place edit or rename/copy edit.
5626                  */
5627                 else if (DIFF_PAIR_RENAME(p)) {
5628                         /*
5629                          * A rename might have re-connected a broken
5630                          * pair up, causing the pathnames to be the
5631                          * same again. If so, that's not a rename at
5632                          * all, just a modification..
5633                          *
5634                          * Otherwise, see if this source was used for
5635                          * multiple renames, in which case we decrement
5636                          * the count, and call it a copy.
5637                          */
5638                         if (!strcmp(p->one->path, p->two->path))
5639                                 p->status = DIFF_STATUS_MODIFIED;
5640                         else if (--p->one->rename_used > 0)
5641                                 p->status = DIFF_STATUS_COPIED;
5642                         else
5643                                 p->status = DIFF_STATUS_RENAMED;
5644                 }
5645                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5646                          p->one->mode != p->two->mode ||
5647                          p->one->dirty_submodule ||
5648                          p->two->dirty_submodule ||
5649                          is_null_oid(&p->one->oid))
5650                         p->status = DIFF_STATUS_MODIFIED;
5651                 else {
5652                         /* This is a "no-change" entry and should not
5653                          * happen anymore, but prepare for broken callers.
5654                          */
5655                         error("feeding unmodified %s to diffcore",
5656                               p->one->path);
5657                         p->status = DIFF_STATUS_UNKNOWN;
5658                 }
5659         }
5660         diff_debug_queue("resolve-rename-copy done", q);
5661 }
5662
5663 static int check_pair_status(struct diff_filepair *p)
5664 {
5665         switch (p->status) {
5666         case DIFF_STATUS_UNKNOWN:
5667                 return 0;
5668         case 0:
5669                 die("internal error in diff-resolve-rename-copy");
5670         default:
5671                 return 1;
5672         }
5673 }
5674
5675 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5676 {
5677         int fmt = opt->output_format;
5678
5679         if (fmt & DIFF_FORMAT_CHECKDIFF)
5680                 diff_flush_checkdiff(p, opt);
5681         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5682                 diff_flush_raw(p, opt);
5683         else if (fmt & DIFF_FORMAT_NAME) {
5684                 const char *name_a, *name_b;
5685                 name_a = p->two->path;
5686                 name_b = NULL;
5687                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5688                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5689                 write_name_quoted(name_a, opt->file, opt->line_termination);
5690         }
5691 }
5692
5693 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5694 {
5695         struct strbuf sb = STRBUF_INIT;
5696         if (fs->mode)
5697                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5698         else
5699                 strbuf_addf(&sb, " %s ", newdelete);
5700
5701         quote_c_style(fs->path, &sb, NULL, 0);
5702         strbuf_addch(&sb, '\n');
5703         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5704                          sb.buf, sb.len, 0);
5705         strbuf_release(&sb);
5706 }
5707
5708 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5709                 int show_name)
5710 {
5711         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5712                 struct strbuf sb = STRBUF_INIT;
5713                 strbuf_addf(&sb, " mode change %06o => %06o",
5714                             p->one->mode, p->two->mode);
5715                 if (show_name) {
5716                         strbuf_addch(&sb, ' ');
5717                         quote_c_style(p->two->path, &sb, NULL, 0);
5718                 }
5719                 strbuf_addch(&sb, '\n');
5720                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5721                                  sb.buf, sb.len, 0);
5722                 strbuf_release(&sb);
5723         }
5724 }
5725
5726 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5727                 struct diff_filepair *p)
5728 {
5729         struct strbuf sb = STRBUF_INIT;
5730         struct strbuf names = STRBUF_INIT;
5731
5732         pprint_rename(&names, p->one->path, p->two->path);
5733         strbuf_addf(&sb, " %s %s (%d%%)\n",
5734                     renamecopy, names.buf, similarity_index(p));
5735         strbuf_release(&names);
5736         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5737                                  sb.buf, sb.len, 0);
5738         show_mode_change(opt, p, 0);
5739         strbuf_release(&sb);
5740 }
5741
5742 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5743 {
5744         switch(p->status) {
5745         case DIFF_STATUS_DELETED:
5746                 show_file_mode_name(opt, "delete", p->one);
5747                 break;
5748         case DIFF_STATUS_ADDED:
5749                 show_file_mode_name(opt, "create", p->two);
5750                 break;
5751         case DIFF_STATUS_COPIED:
5752                 show_rename_copy(opt, "copy", p);
5753                 break;
5754         case DIFF_STATUS_RENAMED:
5755                 show_rename_copy(opt, "rename", p);
5756                 break;
5757         default:
5758                 if (p->score) {
5759                         struct strbuf sb = STRBUF_INIT;
5760                         strbuf_addstr(&sb, " rewrite ");
5761                         quote_c_style(p->two->path, &sb, NULL, 0);
5762                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5763                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5764                                          sb.buf, sb.len, 0);
5765                         strbuf_release(&sb);
5766                 }
5767                 show_mode_change(opt, p, !p->score);
5768                 break;
5769         }
5770 }
5771
5772 struct patch_id_t {
5773         git_SHA_CTX *ctx;
5774         int patchlen;
5775 };
5776
5777 static int remove_space(char *line, int len)
5778 {
5779         int i;
5780         char *dst = line;
5781         unsigned char c;
5782
5783         for (i = 0; i < len; i++)
5784                 if (!isspace((c = line[i])))
5785                         *dst++ = c;
5786
5787         return dst - line;
5788 }
5789
5790 static void patch_id_consume(void *priv, char *line, unsigned long len)
5791 {
5792         struct patch_id_t *data = priv;
5793         int new_len;
5794
5795         new_len = remove_space(line, len);
5796
5797         git_SHA1_Update(data->ctx, line, new_len);
5798         data->patchlen += new_len;
5799 }
5800
5801 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5802 {
5803         git_SHA1_Update(ctx, str, strlen(str));
5804 }
5805
5806 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5807 {
5808         /* large enough for 2^32 in octal */
5809         char buf[12];
5810         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5811         git_SHA1_Update(ctx, buf, len);
5812 }
5813
5814 /* returns 0 upon success, and writes result into sha1 */
5815 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5816 {
5817         struct diff_queue_struct *q = &diff_queued_diff;
5818         int i;
5819         git_SHA_CTX ctx;
5820         struct patch_id_t data;
5821
5822         git_SHA1_Init(&ctx);
5823         memset(&data, 0, sizeof(struct patch_id_t));
5824         data.ctx = &ctx;
5825
5826         for (i = 0; i < q->nr; i++) {
5827                 xpparam_t xpp;
5828                 xdemitconf_t xecfg;
5829                 mmfile_t mf1, mf2;
5830                 struct diff_filepair *p = q->queue[i];
5831                 int len1, len2;
5832
5833                 memset(&xpp, 0, sizeof(xpp));
5834                 memset(&xecfg, 0, sizeof(xecfg));
5835                 if (p->status == 0)
5836                         return error("internal diff status error");
5837                 if (p->status == DIFF_STATUS_UNKNOWN)
5838                         continue;
5839                 if (diff_unmodified_pair(p))
5840                         continue;
5841                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5842                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5843                         continue;
5844                 if (DIFF_PAIR_UNMERGED(p))
5845                         continue;
5846
5847                 diff_fill_oid_info(p->one, options->repo->index);
5848                 diff_fill_oid_info(p->two, options->repo->index);
5849
5850                 len1 = remove_space(p->one->path, strlen(p->one->path));
5851                 len2 = remove_space(p->two->path, strlen(p->two->path));
5852                 patch_id_add_string(&ctx, "diff--git");
5853                 patch_id_add_string(&ctx, "a/");
5854                 git_SHA1_Update(&ctx, p->one->path, len1);
5855                 patch_id_add_string(&ctx, "b/");
5856                 git_SHA1_Update(&ctx, p->two->path, len2);
5857
5858                 if (p->one->mode == 0) {
5859                         patch_id_add_string(&ctx, "newfilemode");
5860                         patch_id_add_mode(&ctx, p->two->mode);
5861                         patch_id_add_string(&ctx, "---/dev/null");
5862                         patch_id_add_string(&ctx, "+++b/");
5863                         git_SHA1_Update(&ctx, p->two->path, len2);
5864                 } else if (p->two->mode == 0) {
5865                         patch_id_add_string(&ctx, "deletedfilemode");
5866                         patch_id_add_mode(&ctx, p->one->mode);
5867                         patch_id_add_string(&ctx, "---a/");
5868                         git_SHA1_Update(&ctx, p->one->path, len1);
5869                         patch_id_add_string(&ctx, "+++/dev/null");
5870                 } else {
5871                         patch_id_add_string(&ctx, "---a/");
5872                         git_SHA1_Update(&ctx, p->one->path, len1);
5873                         patch_id_add_string(&ctx, "+++b/");
5874                         git_SHA1_Update(&ctx, p->two->path, len2);
5875                 }
5876
5877                 if (diff_header_only)
5878                         continue;
5879
5880                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5881                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5882                         return error("unable to read files to diff");
5883
5884                 if (diff_filespec_is_binary(options->repo, p->one) ||
5885                     diff_filespec_is_binary(options->repo, p->two)) {
5886                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5887                                         GIT_SHA1_HEXSZ);
5888                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5889                                         GIT_SHA1_HEXSZ);
5890                         continue;
5891                 }
5892
5893                 xpp.flags = 0;
5894                 xecfg.ctxlen = 3;
5895                 xecfg.flags = 0;
5896                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5897                                   patch_id_consume, &data, &xpp, &xecfg))
5898                         return error("unable to generate patch-id diff for %s",
5899                                      p->one->path);
5900         }
5901
5902         git_SHA1_Final(oid->hash, &ctx);
5903         return 0;
5904 }
5905
5906 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5907 {
5908         struct diff_queue_struct *q = &diff_queued_diff;
5909         int i;
5910         int result = diff_get_patch_id(options, oid, diff_header_only);
5911
5912         for (i = 0; i < q->nr; i++)
5913                 diff_free_filepair(q->queue[i]);
5914
5915         free(q->queue);
5916         DIFF_QUEUE_CLEAR(q);
5917
5918         return result;
5919 }
5920
5921 static int is_summary_empty(const struct diff_queue_struct *q)
5922 {
5923         int i;
5924
5925         for (i = 0; i < q->nr; i++) {
5926                 const struct diff_filepair *p = q->queue[i];
5927
5928                 switch (p->status) {
5929                 case DIFF_STATUS_DELETED:
5930                 case DIFF_STATUS_ADDED:
5931                 case DIFF_STATUS_COPIED:
5932                 case DIFF_STATUS_RENAMED:
5933                         return 0;
5934                 default:
5935                         if (p->score)
5936                                 return 0;
5937                         if (p->one->mode && p->two->mode &&
5938                             p->one->mode != p->two->mode)
5939                                 return 0;
5940                         break;
5941                 }
5942         }
5943         return 1;
5944 }
5945
5946 static const char rename_limit_warning[] =
5947 N_("inexact rename detection was skipped due to too many files.");
5948
5949 static const char degrade_cc_to_c_warning[] =
5950 N_("only found copies from modified paths due to too many files.");
5951
5952 static const char rename_limit_advice[] =
5953 N_("you may want to set your %s variable to at least "
5954    "%d and retry the command.");
5955
5956 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5957 {
5958         fflush(stdout);
5959         if (degraded_cc)
5960                 warning(_(degrade_cc_to_c_warning));
5961         else if (needed)
5962                 warning(_(rename_limit_warning));
5963         else
5964                 return;
5965         if (0 < needed)
5966                 warning(_(rename_limit_advice), varname, needed);
5967 }
5968
5969 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5970 {
5971         int i;
5972         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5973         struct diff_queue_struct *q = &diff_queued_diff;
5974
5975         if (WSEH_NEW & WS_RULE_MASK)
5976                 BUG("WS rules bit mask overlaps with diff symbol flags");
5977
5978         if (o->color_moved)
5979                 o->emitted_symbols = &esm;
5980
5981         for (i = 0; i < q->nr; i++) {
5982                 struct diff_filepair *p = q->queue[i];
5983                 if (check_pair_status(p))
5984                         diff_flush_patch(p, o);
5985         }
5986
5987         if (o->emitted_symbols) {
5988                 if (o->color_moved) {
5989                         struct hashmap add_lines, del_lines;
5990
5991                         if (o->color_moved_ws_handling &
5992                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5993                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5994
5995                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5996                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5997
5998                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5999                         mark_color_as_moved(o, &add_lines, &del_lines);
6000                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6001                                 dim_moved_lines(o);
6002
6003                         hashmap_free(&add_lines, 1);
6004                         hashmap_free(&del_lines, 1);
6005                 }
6006
6007                 for (i = 0; i < esm.nr; i++)
6008                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6009
6010                 for (i = 0; i < esm.nr; i++)
6011                         free((void *)esm.buf[i].line);
6012                 esm.nr = 0;
6013
6014                 o->emitted_symbols = NULL;
6015         }
6016 }
6017
6018 void diff_flush(struct diff_options *options)
6019 {
6020         struct diff_queue_struct *q = &diff_queued_diff;
6021         int i, output_format = options->output_format;
6022         int separator = 0;
6023         int dirstat_by_line = 0;
6024
6025         /*
6026          * Order: raw, stat, summary, patch
6027          * or:    name/name-status/checkdiff (other bits clear)
6028          */
6029         if (!q->nr)
6030                 goto free_queue;
6031
6032         if (output_format & (DIFF_FORMAT_RAW |
6033                              DIFF_FORMAT_NAME |
6034                              DIFF_FORMAT_NAME_STATUS |
6035                              DIFF_FORMAT_CHECKDIFF)) {
6036                 for (i = 0; i < q->nr; i++) {
6037                         struct diff_filepair *p = q->queue[i];
6038                         if (check_pair_status(p))
6039                                 flush_one_pair(p, options);
6040                 }
6041                 separator++;
6042         }
6043
6044         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6045                 dirstat_by_line = 1;
6046
6047         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6048             dirstat_by_line) {
6049                 struct diffstat_t diffstat;
6050
6051                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6052                 for (i = 0; i < q->nr; i++) {
6053                         struct diff_filepair *p = q->queue[i];
6054                         if (check_pair_status(p))
6055                                 diff_flush_stat(p, options, &diffstat);
6056                 }
6057                 if (output_format & DIFF_FORMAT_NUMSTAT)
6058                         show_numstat(&diffstat, options);
6059                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6060                         show_stats(&diffstat, options);
6061                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6062                         show_shortstats(&diffstat, options);
6063                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6064                         show_dirstat_by_line(&diffstat, options);
6065                 free_diffstat_info(&diffstat);
6066                 separator++;
6067         }
6068         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6069                 show_dirstat(options);
6070
6071         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6072                 for (i = 0; i < q->nr; i++) {
6073                         diff_summary(options, q->queue[i]);
6074                 }
6075                 separator++;
6076         }
6077
6078         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6079             options->flags.exit_with_status &&
6080             options->flags.diff_from_contents) {
6081                 /*
6082                  * run diff_flush_patch for the exit status. setting
6083                  * options->file to /dev/null should be safe, because we
6084                  * aren't supposed to produce any output anyway.
6085                  */
6086                 if (options->close_file)
6087                         fclose(options->file);
6088                 options->file = xfopen("/dev/null", "w");
6089                 options->close_file = 1;
6090                 options->color_moved = 0;
6091                 for (i = 0; i < q->nr; i++) {
6092                         struct diff_filepair *p = q->queue[i];
6093                         if (check_pair_status(p))
6094                                 diff_flush_patch(p, options);
6095                         if (options->found_changes)
6096                                 break;
6097                 }
6098         }
6099
6100         if (output_format & DIFF_FORMAT_PATCH) {
6101                 if (separator) {
6102                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6103                         if (options->stat_sep)
6104                                 /* attach patch instead of inline */
6105                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6106                                                  NULL, 0, 0);
6107                 }
6108
6109                 diff_flush_patch_all_file_pairs(options);
6110         }
6111
6112         if (output_format & DIFF_FORMAT_CALLBACK)
6113                 options->format_callback(q, options, options->format_callback_data);
6114
6115         for (i = 0; i < q->nr; i++)
6116                 diff_free_filepair(q->queue[i]);
6117 free_queue:
6118         free(q->queue);
6119         DIFF_QUEUE_CLEAR(q);
6120         if (options->close_file)
6121                 fclose(options->file);
6122
6123         /*
6124          * Report the content-level differences with HAS_CHANGES;
6125          * diff_addremove/diff_change does not set the bit when
6126          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6127          */
6128         if (options->flags.diff_from_contents) {
6129                 if (options->found_changes)
6130                         options->flags.has_changes = 1;
6131                 else
6132                         options->flags.has_changes = 0;
6133         }
6134 }
6135
6136 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6137 {
6138         return (((p->status == DIFF_STATUS_MODIFIED) &&
6139                  ((p->score &&
6140                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6141                   (!p->score &&
6142                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6143                 ((p->status != DIFF_STATUS_MODIFIED) &&
6144                  filter_bit_tst(p->status, options)));
6145 }
6146
6147 static void diffcore_apply_filter(struct diff_options *options)
6148 {
6149         int i;
6150         struct diff_queue_struct *q = &diff_queued_diff;
6151         struct diff_queue_struct outq;
6152
6153         DIFF_QUEUE_CLEAR(&outq);
6154
6155         if (!options->filter)
6156                 return;
6157
6158         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6159                 int found;
6160                 for (i = found = 0; !found && i < q->nr; i++) {
6161                         if (match_filter(options, q->queue[i]))
6162                                 found++;
6163                 }
6164                 if (found)
6165                         return;
6166
6167                 /* otherwise we will clear the whole queue
6168                  * by copying the empty outq at the end of this
6169                  * function, but first clear the current entries
6170                  * in the queue.
6171                  */
6172                 for (i = 0; i < q->nr; i++)
6173                         diff_free_filepair(q->queue[i]);
6174         }
6175         else {
6176                 /* Only the matching ones */
6177                 for (i = 0; i < q->nr; i++) {
6178                         struct diff_filepair *p = q->queue[i];
6179                         if (match_filter(options, p))
6180                                 diff_q(&outq, p);
6181                         else
6182                                 diff_free_filepair(p);
6183                 }
6184         }
6185         free(q->queue);
6186         *q = outq;
6187 }
6188
6189 /* Check whether two filespecs with the same mode and size are identical */
6190 static int diff_filespec_is_identical(struct repository *r,
6191                                       struct diff_filespec *one,
6192                                       struct diff_filespec *two)
6193 {
6194         if (S_ISGITLINK(one->mode))
6195                 return 0;
6196         if (diff_populate_filespec(r, one, 0))
6197                 return 0;
6198         if (diff_populate_filespec(r, two, 0))
6199                 return 0;
6200         return !memcmp(one->data, two->data, one->size);
6201 }
6202
6203 static int diff_filespec_check_stat_unmatch(struct repository *r,
6204                                             struct diff_filepair *p)
6205 {
6206         if (p->done_skip_stat_unmatch)
6207                 return p->skip_stat_unmatch_result;
6208
6209         p->done_skip_stat_unmatch = 1;
6210         p->skip_stat_unmatch_result = 0;
6211         /*
6212          * 1. Entries that come from stat info dirtiness
6213          *    always have both sides (iow, not create/delete),
6214          *    one side of the object name is unknown, with
6215          *    the same mode and size.  Keep the ones that
6216          *    do not match these criteria.  They have real
6217          *    differences.
6218          *
6219          * 2. At this point, the file is known to be modified,
6220          *    with the same mode and size, and the object
6221          *    name of one side is unknown.  Need to inspect
6222          *    the identical contents.
6223          */
6224         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6225             !DIFF_FILE_VALID(p->two) ||
6226             (p->one->oid_valid && p->two->oid_valid) ||
6227             (p->one->mode != p->two->mode) ||
6228             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6229             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6230             (p->one->size != p->two->size) ||
6231             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6232                 p->skip_stat_unmatch_result = 1;
6233         return p->skip_stat_unmatch_result;
6234 }
6235
6236 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6237 {
6238         int i;
6239         struct diff_queue_struct *q = &diff_queued_diff;
6240         struct diff_queue_struct outq;
6241         DIFF_QUEUE_CLEAR(&outq);
6242
6243         for (i = 0; i < q->nr; i++) {
6244                 struct diff_filepair *p = q->queue[i];
6245
6246                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6247                         diff_q(&outq, p);
6248                 else {
6249                         /*
6250                          * The caller can subtract 1 from skip_stat_unmatch
6251                          * to determine how many paths were dirty only
6252                          * due to stat info mismatch.
6253                          */
6254                         if (!diffopt->flags.no_index)
6255                                 diffopt->skip_stat_unmatch++;
6256                         diff_free_filepair(p);
6257                 }
6258         }
6259         free(q->queue);
6260         *q = outq;
6261 }
6262
6263 static int diffnamecmp(const void *a_, const void *b_)
6264 {
6265         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6266         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6267         const char *name_a, *name_b;
6268
6269         name_a = a->one ? a->one->path : a->two->path;
6270         name_b = b->one ? b->one->path : b->two->path;
6271         return strcmp(name_a, name_b);
6272 }
6273
6274 void diffcore_fix_diff_index(struct diff_options *options)
6275 {
6276         struct diff_queue_struct *q = &diff_queued_diff;
6277         QSORT(q->queue, q->nr, diffnamecmp);
6278 }
6279
6280 void diffcore_std(struct diff_options *options)
6281 {
6282         /* NOTE please keep the following in sync with diff_tree_combined() */
6283         if (options->skip_stat_unmatch)
6284                 diffcore_skip_stat_unmatch(options);
6285         if (!options->found_follow) {
6286                 /* See try_to_follow_renames() in tree-diff.c */
6287                 if (options->break_opt != -1)
6288                         diffcore_break(options->repo,
6289                                        options->break_opt);
6290                 if (options->detect_rename)
6291                         diffcore_rename(options);
6292                 if (options->break_opt != -1)
6293                         diffcore_merge_broken();
6294         }
6295         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6296                 diffcore_pickaxe(options);
6297         if (options->orderfile)
6298                 diffcore_order(options->orderfile);
6299         if (!options->found_follow)
6300                 /* See try_to_follow_renames() in tree-diff.c */
6301                 diff_resolve_rename_copy();
6302         diffcore_apply_filter(options);
6303
6304         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6305                 options->flags.has_changes = 1;
6306         else
6307                 options->flags.has_changes = 0;
6308
6309         options->found_follow = 0;
6310 }
6311
6312 int diff_result_code(struct diff_options *opt, int status)
6313 {
6314         int result = 0;
6315
6316         diff_warn_rename_limit("diff.renameLimit",
6317                                opt->needed_rename_limit,
6318                                opt->degraded_cc_to_c);
6319         if (!opt->flags.exit_with_status &&
6320             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6321                 return status;
6322         if (opt->flags.exit_with_status &&
6323             opt->flags.has_changes)
6324                 result |= 01;
6325         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6326             opt->flags.check_failed)
6327                 result |= 02;
6328         return result;
6329 }
6330
6331 int diff_can_quit_early(struct diff_options *opt)
6332 {
6333         return (opt->flags.quick &&
6334                 !opt->filter &&
6335                 opt->flags.has_changes);
6336 }
6337
6338 /*
6339  * Shall changes to this submodule be ignored?
6340  *
6341  * Submodule changes can be configured to be ignored separately for each path,
6342  * but that configuration can be overridden from the command line.
6343  */
6344 static int is_submodule_ignored(const char *path, struct diff_options *options)
6345 {
6346         int ignored = 0;
6347         struct diff_flags orig_flags = options->flags;
6348         if (!options->flags.override_submodule_config)
6349                 set_diffopt_flags_from_submodule_config(options, path);
6350         if (options->flags.ignore_submodules)
6351                 ignored = 1;
6352         options->flags = orig_flags;
6353         return ignored;
6354 }
6355
6356 void diff_addremove(struct diff_options *options,
6357                     int addremove, unsigned mode,
6358                     const struct object_id *oid,
6359                     int oid_valid,
6360                     const char *concatpath, unsigned dirty_submodule)
6361 {
6362         struct diff_filespec *one, *two;
6363
6364         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6365                 return;
6366
6367         /* This may look odd, but it is a preparation for
6368          * feeding "there are unchanged files which should
6369          * not produce diffs, but when you are doing copy
6370          * detection you would need them, so here they are"
6371          * entries to the diff-core.  They will be prefixed
6372          * with something like '=' or '*' (I haven't decided
6373          * which but should not make any difference).
6374          * Feeding the same new and old to diff_change()
6375          * also has the same effect.
6376          * Before the final output happens, they are pruned after
6377          * merged into rename/copy pairs as appropriate.
6378          */
6379         if (options->flags.reverse_diff)
6380                 addremove = (addremove == '+' ? '-' :
6381                              addremove == '-' ? '+' : addremove);
6382
6383         if (options->prefix &&
6384             strncmp(concatpath, options->prefix, options->prefix_length))
6385                 return;
6386
6387         one = alloc_filespec(concatpath);
6388         two = alloc_filespec(concatpath);
6389
6390         if (addremove != '+')
6391                 fill_filespec(one, oid, oid_valid, mode);
6392         if (addremove != '-') {
6393                 fill_filespec(two, oid, oid_valid, mode);
6394                 two->dirty_submodule = dirty_submodule;
6395         }
6396
6397         diff_queue(&diff_queued_diff, one, two);
6398         if (!options->flags.diff_from_contents)
6399                 options->flags.has_changes = 1;
6400 }
6401
6402 void diff_change(struct diff_options *options,
6403                  unsigned old_mode, unsigned new_mode,
6404                  const struct object_id *old_oid,
6405                  const struct object_id *new_oid,
6406                  int old_oid_valid, int new_oid_valid,
6407                  const char *concatpath,
6408                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6409 {
6410         struct diff_filespec *one, *two;
6411         struct diff_filepair *p;
6412
6413         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6414             is_submodule_ignored(concatpath, options))
6415                 return;
6416
6417         if (options->flags.reverse_diff) {
6418                 SWAP(old_mode, new_mode);
6419                 SWAP(old_oid, new_oid);
6420                 SWAP(old_oid_valid, new_oid_valid);
6421                 SWAP(old_dirty_submodule, new_dirty_submodule);
6422         }
6423
6424         if (options->prefix &&
6425             strncmp(concatpath, options->prefix, options->prefix_length))
6426                 return;
6427
6428         one = alloc_filespec(concatpath);
6429         two = alloc_filespec(concatpath);
6430         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6431         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6432         one->dirty_submodule = old_dirty_submodule;
6433         two->dirty_submodule = new_dirty_submodule;
6434         p = diff_queue(&diff_queued_diff, one, two);
6435
6436         if (options->flags.diff_from_contents)
6437                 return;
6438
6439         if (options->flags.quick && options->skip_stat_unmatch &&
6440             !diff_filespec_check_stat_unmatch(options->repo, p))
6441                 return;
6442
6443         options->flags.has_changes = 1;
6444 }
6445
6446 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6447 {
6448         struct diff_filepair *pair;
6449         struct diff_filespec *one, *two;
6450
6451         if (options->prefix &&
6452             strncmp(path, options->prefix, options->prefix_length))
6453                 return NULL;
6454
6455         one = alloc_filespec(path);
6456         two = alloc_filespec(path);
6457         pair = diff_queue(&diff_queued_diff, one, two);
6458         pair->is_unmerged = 1;
6459         return pair;
6460 }
6461
6462 static char *run_textconv(struct repository *r,
6463                           const char *pgm,
6464                           struct diff_filespec *spec,
6465                           size_t *outsize)
6466 {
6467         struct diff_tempfile *temp;
6468         const char *argv[3];
6469         const char **arg = argv;
6470         struct child_process child = CHILD_PROCESS_INIT;
6471         struct strbuf buf = STRBUF_INIT;
6472         int err = 0;
6473
6474         temp = prepare_temp_file(r, spec->path, spec);
6475         *arg++ = pgm;
6476         *arg++ = temp->name;
6477         *arg = NULL;
6478
6479         child.use_shell = 1;
6480         child.argv = argv;
6481         child.out = -1;
6482         if (start_command(&child)) {
6483                 remove_tempfile();
6484                 return NULL;
6485         }
6486
6487         if (strbuf_read(&buf, child.out, 0) < 0)
6488                 err = error("error reading from textconv command '%s'", pgm);
6489         close(child.out);
6490
6491         if (finish_command(&child) || err) {
6492                 strbuf_release(&buf);
6493                 remove_tempfile();
6494                 return NULL;
6495         }
6496         remove_tempfile();
6497
6498         return strbuf_detach(&buf, outsize);
6499 }
6500
6501 size_t fill_textconv(struct repository *r,
6502                      struct userdiff_driver *driver,
6503                      struct diff_filespec *df,
6504                      char **outbuf)
6505 {
6506         size_t size;
6507
6508         if (!driver) {
6509                 if (!DIFF_FILE_VALID(df)) {
6510                         *outbuf = "";
6511                         return 0;
6512                 }
6513                 if (diff_populate_filespec(r, df, 0))
6514                         die("unable to read files to diff");
6515                 *outbuf = df->data;
6516                 return df->size;
6517         }
6518
6519         if (!driver->textconv)
6520                 BUG("fill_textconv called with non-textconv driver");
6521
6522         if (driver->textconv_cache && df->oid_valid) {
6523                 *outbuf = notes_cache_get(driver->textconv_cache,
6524                                           &df->oid,
6525                                           &size);
6526                 if (*outbuf)
6527                         return size;
6528         }
6529
6530         *outbuf = run_textconv(r, driver->textconv, df, &size);
6531         if (!*outbuf)
6532                 die("unable to read files to diff");
6533
6534         if (driver->textconv_cache && df->oid_valid) {
6535                 /* ignore errors, as we might be in a readonly repository */
6536                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6537                                 size);
6538                 /*
6539                  * we could save up changes and flush them all at the end,
6540                  * but we would need an extra call after all diffing is done.
6541                  * Since generating a cache entry is the slow path anyway,
6542                  * this extra overhead probably isn't a big deal.
6543                  */
6544                 notes_cache_write(driver->textconv_cache);
6545         }
6546
6547         return size;
6548 }
6549
6550 int textconv_object(struct repository *r,
6551                     const char *path,
6552                     unsigned mode,
6553                     const struct object_id *oid,
6554                     int oid_valid,
6555                     char **buf,
6556                     unsigned long *buf_size)
6557 {
6558         struct diff_filespec *df;
6559         struct userdiff_driver *textconv;
6560
6561         df = alloc_filespec(path);
6562         fill_filespec(df, oid, oid_valid, mode);
6563         textconv = get_textconv(r, df);
6564         if (!textconv) {
6565                 free_filespec(df);
6566                 return 0;
6567         }
6568
6569         *buf_size = fill_textconv(r, textconv, df, buf);
6570         free_filespec(df);
6571         return 1;
6572 }
6573
6574 void setup_diff_pager(struct diff_options *opt)
6575 {
6576         /*
6577          * If the user asked for our exit code, then either they want --quiet
6578          * or --exit-code. We should definitely not bother with a pager in the
6579          * former case, as we will generate no output. Since we still properly
6580          * report our exit code even when a pager is run, we _could_ run a
6581          * pager with --exit-code. But since we have not done so historically,
6582          * and because it is easy to find people oneline advising "git diff
6583          * --exit-code" in hooks and other scripts, we do not do so.
6584          */
6585         if (!opt->flags.exit_with_status &&
6586             check_pager_config("diff") != 0)
6587                 setup_pager();
6588 }