git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73 };
  74
  75 static const char *color_diff_slots[] = {
  76         [DIFF_CONTEXT]                = "context",
  77         [DIFF_METAINFO]               = "meta",
  78         [DIFF_FRAGINFO]               = "frag",
  79         [DIFF_FILE_OLD]               = "old",
  80         [DIFF_FILE_NEW]               = "new",
  81         [DIFF_COMMIT]                 = "commit",
  82         [DIFF_WHITESPACE]             = "whitespace",
  83         [DIFF_FUNCINFO]               = "func",
  84         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  85         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  86         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  87         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  88         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  89         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  90         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  91         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  92 };
  93
  94 static NORETURN void die_want_option(const char *option_name)
  95 {
  96         die(_("option '%s' requires a value"), option_name);
  97 }
  98
  99 define_list_config_array_extra(color_diff_slots, {"plain"});
 100
 101 static int parse_diff_color_slot(const char *var)
 102 {
 103         if (!strcasecmp(var, "plain"))
 104                 return DIFF_CONTEXT;
 105         return LOOKUP_CONFIG(color_diff_slots, var);
 106 }
 107
 108 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 109                                 struct strbuf *errmsg)
 110 {
 111         char *params_copy = xstrdup(params_string);
 112         struct string_list params = STRING_LIST_INIT_NODUP;
 113         int ret = 0;
 114         int i;
 115
 116         if (*params_copy)
 117                 string_list_split_in_place(&params, params_copy, ',', -1);
 118         for (i = 0; i < params.nr; i++) {
 119                 const char *p = params.items[i].string;
 120                 if (!strcmp(p, "changes")) {
 121                         options->flags.dirstat_by_line = 0;
 122                         options->flags.dirstat_by_file = 0;
 123                 } else if (!strcmp(p, "lines")) {
 124                         options->flags.dirstat_by_line = 1;
 125                         options->flags.dirstat_by_file = 0;
 126                 } else if (!strcmp(p, "files")) {
 127                         options->flags.dirstat_by_line = 0;
 128                         options->flags.dirstat_by_file = 1;
 129                 } else if (!strcmp(p, "noncumulative")) {
 130                         options->flags.dirstat_cumulative = 0;
 131                 } else if (!strcmp(p, "cumulative")) {
 132                         options->flags.dirstat_cumulative = 1;
 133                 } else if (isdigit(*p)) {
 134                         char *end;
 135                         int permille = strtoul(p, &end, 10) * 10;
 136                         if (*end == '.' && isdigit(*++end)) {
 137                                 /* only use first digit */
 138                                 permille += *end - '0';
 139                                 /* .. and ignore any further digits */
 140                                 while (isdigit(*++end))
 141                                         ; /* nothing */
 142                         }
 143                         if (!*end)
 144                                 options->dirstat_permille = permille;
 145                         else {
 146                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 147                                             p);
 148                                 ret++;
 149                         }
 150                 } else {
 151                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 152                         ret++;
 153                 }
 154
 155         }
 156         string_list_clear(&params, 0);
 157         free(params_copy);
 158         return ret;
 159 }
 160
 161 static int parse_submodule_params(struct diff_options *options, const char *value)
 162 {
 163         if (!strcmp(value, "log"))
 164                 options->submodule_format = DIFF_SUBMODULE_LOG;
 165         else if (!strcmp(value, "short"))
 166                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 167         else if (!strcmp(value, "diff"))
 168                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 169         else
 170                 return -1;
 171         return 0;
 172 }
 173
 174 int git_config_rename(const char *var, const char *value)
 175 {
 176         if (!value)
 177                 return DIFF_DETECT_RENAME;
 178         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 179                 return  DIFF_DETECT_COPY;
 180         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 181 }
 182
 183 long parse_algorithm_value(const char *value)
 184 {
 185         if (!value)
 186                 return -1;
 187         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 188                 return 0;
 189         else if (!strcasecmp(value, "minimal"))
 190                 return XDF_NEED_MINIMAL;
 191         else if (!strcasecmp(value, "patience"))
 192                 return XDF_PATIENCE_DIFF;
 193         else if (!strcasecmp(value, "histogram"))
 194                 return XDF_HISTOGRAM_DIFF;
 195         return -1;
 196 }
 197
 198 static int parse_one_token(const char **arg, const char *token)
 199 {
 200         const char *rest;
 201         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 202                 *arg = rest;
 203                 return 1;
 204         }
 205         return 0;
 206 }
 207
 208 static int parse_ws_error_highlight(const char *arg)
 209 {
 210         const char *orig_arg = arg;
 211         unsigned val = 0;
 212
 213         while (*arg) {
 214                 if (parse_one_token(&arg, "none"))
 215                         val = 0;
 216                 else if (parse_one_token(&arg, "default"))
 217                         val = WSEH_NEW;
 218                 else if (parse_one_token(&arg, "all"))
 219                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 220                 else if (parse_one_token(&arg, "new"))
 221                         val |= WSEH_NEW;
 222                 else if (parse_one_token(&arg, "old"))
 223                         val |= WSEH_OLD;
 224                 else if (parse_one_token(&arg, "context"))
 225                         val |= WSEH_CONTEXT;
 226                 else {
 227                         return -1 - (int)(arg - orig_arg);
 228                 }
 229                 if (*arg)
 230                         arg++;
 231         }
 232         return val;
 233 }
 234
 235 /*
 236  * These are to give UI layer defaults.
 237  * The core-level commands such as git-diff-files should
 238  * never be affected by the setting of diff.renames
 239  * the user happens to have in the configuration file.
 240  */
 241 void init_diff_ui_defaults(void)
 242 {
 243         diff_detect_rename_default = DIFF_DETECT_RENAME;
 244 }
 245
 246 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 247 {
 248         if (!strcmp(var, "diff.indentheuristic"))
 249                 diff_indent_heuristic = git_config_bool(var, value);
 250         return 0;
 251 }
 252
 253 static int parse_color_moved(const char *arg)
 254 {
 255         switch (git_parse_maybe_bool(arg)) {
 256         case 0:
 257                 return COLOR_MOVED_NO;
 258         case 1:
 259                 return COLOR_MOVED_DEFAULT;
 260         default:
 261                 break;
 262         }
 263
 264         if (!strcmp(arg, "no"))
 265                 return COLOR_MOVED_NO;
 266         else if (!strcmp(arg, "plain"))
 267                 return COLOR_MOVED_PLAIN;
 268         else if (!strcmp(arg, "blocks"))
 269                 return COLOR_MOVED_BLOCKS;
 270         else if (!strcmp(arg, "zebra"))
 271                 return COLOR_MOVED_ZEBRA;
 272         else if (!strcmp(arg, "default"))
 273                 return COLOR_MOVED_DEFAULT;
 274         else if (!strcmp(arg, "dimmed_zebra"))
 275                 return COLOR_MOVED_ZEBRA_DIM;
 276         else
 277                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 278 }
 279
 280 static int parse_color_moved_ws(const char *arg)
 281 {
 282         int ret = 0;
 283         struct string_list l = STRING_LIST_INIT_DUP;
 284         struct string_list_item *i;
 285
 286         string_list_split(&l, arg, ',', -1);
 287
 288         for_each_string_list_item(i, &l) {
 289                 struct strbuf sb = STRBUF_INIT;
 290                 strbuf_addstr(&sb, i->string);
 291                 strbuf_trim(&sb);
 292
 293                 if (!strcmp(sb.buf, "ignore-space-change"))
 294                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 295                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 296                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 297                 else if (!strcmp(sb.buf, "ignore-all-space"))
 298                         ret |= XDF_IGNORE_WHITESPACE;
 299                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 300                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 301                 else
 302                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 303
 304                 strbuf_release(&sb);
 305         }
 306
 307         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 308             (ret & XDF_WHITESPACE_FLAGS))
 309                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 310
 311         string_list_clear(&l, 0);
 312
 313         return ret;
 314 }
 315
 316 int git_diff_ui_config(const char *var, const char *value, void *cb)
 317 {
 318         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 319                 diff_use_color_default = git_config_colorbool(var, value);
 320                 return 0;
 321         }
 322         if (!strcmp(var, "diff.colormoved")) {
 323                 int cm = parse_color_moved(value);
 324                 if (cm < 0)
 325                         return -1;
 326                 diff_color_moved_default = cm;
 327                 return 0;
 328         }
 329         if (!strcmp(var, "diff.colormovedws")) {
 330                 int cm = parse_color_moved_ws(value);
 331                 if (cm < 0)
 332                         return -1;
 333                 diff_color_moved_ws_default = cm;
 334                 return 0;
 335         }
 336         if (!strcmp(var, "diff.context")) {
 337                 diff_context_default = git_config_int(var, value);
 338                 if (diff_context_default < 0)
 339                         return -1;
 340                 return 0;
 341         }
 342         if (!strcmp(var, "diff.interhunkcontext")) {
 343                 diff_interhunk_context_default = git_config_int(var, value);
 344                 if (diff_interhunk_context_default < 0)
 345                         return -1;
 346                 return 0;
 347         }
 348         if (!strcmp(var, "diff.renames")) {
 349                 diff_detect_rename_default = git_config_rename(var, value);
 350                 return 0;
 351         }
 352         if (!strcmp(var, "diff.autorefreshindex")) {
 353                 diff_auto_refresh_index = git_config_bool(var, value);
 354                 return 0;
 355         }
 356         if (!strcmp(var, "diff.mnemonicprefix")) {
 357                 diff_mnemonic_prefix = git_config_bool(var, value);
 358                 return 0;
 359         }
 360         if (!strcmp(var, "diff.noprefix")) {
 361                 diff_no_prefix = git_config_bool(var, value);
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.statgraphwidth")) {
 365                 diff_stat_graph_width = git_config_int(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.external"))
 369                 return git_config_string(&external_diff_cmd_cfg, var, value);
 370         if (!strcmp(var, "diff.wordregex"))
 371                 return git_config_string(&diff_word_regex_cfg, var, value);
 372         if (!strcmp(var, "diff.orderfile"))
 373                 return git_config_pathname(&diff_order_file_cfg, var, value);
 374
 375         if (!strcmp(var, "diff.ignoresubmodules"))
 376                 handle_ignore_submodules_arg(&default_diff_options, value);
 377
 378         if (!strcmp(var, "diff.submodule")) {
 379                 if (parse_submodule_params(&default_diff_options, value))
 380                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 381                                 value);
 382                 return 0;
 383         }
 384
 385         if (!strcmp(var, "diff.algorithm")) {
 386                 diff_algorithm = parse_algorithm_value(value);
 387                 if (diff_algorithm < 0)
 388                         return -1;
 389                 return 0;
 390         }
 391
 392         if (!strcmp(var, "diff.wserrorhighlight")) {
 393                 int val = parse_ws_error_highlight(value);
 394                 if (val < 0)
 395                         return -1;
 396                 ws_error_highlight_default = val;
 397                 return 0;
 398         }
 399
 400         if (git_color_config(var, value, cb) < 0)
 401                 return -1;
 402
 403         return git_diff_basic_config(var, value, cb);
 404 }
 405
 406 int git_diff_basic_config(const char *var, const char *value, void *cb)
 407 {
 408         const char *name;
 409
 410         if (!strcmp(var, "diff.renamelimit")) {
 411                 diff_rename_limit_default = git_config_int(var, value);
 412                 return 0;
 413         }
 414
 415         if (userdiff_config(var, value) < 0)
 416                 return -1;
 417
 418         if (skip_prefix(var, "diff.color.", &name) ||
 419             skip_prefix(var, "color.diff.", &name)) {
 420                 int slot = parse_diff_color_slot(name);
 421                 if (slot < 0)
 422                         return 0;
 423                 if (!value)
 424                         return config_error_nonbool(var);
 425                 return color_parse(value, diff_colors[slot]);
 426         }
 427
 428         /* like GNU diff's --suppress-blank-empty option  */
 429         if (!strcmp(var, "diff.suppressblankempty") ||
 430                         /* for backwards compatibility */
 431                         !strcmp(var, "diff.suppress-blank-empty")) {
 432                 diff_suppress_blank_empty = git_config_bool(var, value);
 433                 return 0;
 434         }
 435
 436         if (!strcmp(var, "diff.dirstat")) {
 437                 struct strbuf errmsg = STRBUF_INIT;
 438                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 439                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 440                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 441                                 errmsg.buf);
 442                 strbuf_release(&errmsg);
 443                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 444                 return 0;
 445         }
 446
 447         if (git_diff_heuristic_config(var, value, cb) < 0)
 448                 return -1;
 449
 450         return git_default_config(var, value, cb);
 451 }
 452
 453 static char *quote_two(const char *one, const char *two)
 454 {
 455         int need_one = quote_c_style(one, NULL, NULL, 1);
 456         int need_two = quote_c_style(two, NULL, NULL, 1);
 457         struct strbuf res = STRBUF_INIT;
 458
 459         if (need_one + need_two) {
 460                 strbuf_addch(&res, '"');
 461                 quote_c_style(one, &res, NULL, 1);
 462                 quote_c_style(two, &res, NULL, 1);
 463                 strbuf_addch(&res, '"');
 464         } else {
 465                 strbuf_addstr(&res, one);
 466                 strbuf_addstr(&res, two);
 467         }
 468         return strbuf_detach(&res, NULL);
 469 }
 470
 471 static const char *external_diff(void)
 472 {
 473         static const char *external_diff_cmd = NULL;
 474         static int done_preparing = 0;
 475
 476         if (done_preparing)
 477                 return external_diff_cmd;
 478         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 479         if (!external_diff_cmd)
 480                 external_diff_cmd = external_diff_cmd_cfg;
 481         done_preparing = 1;
 482         return external_diff_cmd;
 483 }
 484
 485 /*
 486  * Keep track of files used for diffing. Sometimes such an entry
 487  * refers to a temporary file, sometimes to an existing file, and
 488  * sometimes to "/dev/null".
 489  */
 490 static struct diff_tempfile {
 491         /*
 492          * filename external diff should read from, or NULL if this
 493          * entry is currently not in use:
 494          */
 495         const char *name;
 496
 497         char hex[GIT_MAX_HEXSZ + 1];
 498         char mode[10];
 499
 500         /*
 501          * If this diff_tempfile instance refers to a temporary file,
 502          * this tempfile object is used to manage its lifetime.
 503          */
 504         struct tempfile *tempfile;
 505 } diff_temp[2];
 506
 507 struct emit_callback {
 508         int color_diff;
 509         unsigned ws_rule;
 510         int blank_at_eof_in_preimage;
 511         int blank_at_eof_in_postimage;
 512         int lno_in_preimage;
 513         int lno_in_postimage;
 514         const char **label_path;
 515         struct diff_words_data *diff_words;
 516         struct diff_options *opt;
 517         struct strbuf *header;
 518 };
 519
 520 static int count_lines(const char *data, int size)
 521 {
 522         int count, ch, completely_empty = 1, nl_just_seen = 0;
 523         count = 0;
 524         while (0 < size--) {
 525                 ch = *data++;
 526                 if (ch == '\n') {
 527                         count++;
 528                         nl_just_seen = 1;
 529                         completely_empty = 0;
 530                 }
 531                 else {
 532                         nl_just_seen = 0;
 533                         completely_empty = 0;
 534                 }
 535         }
 536         if (completely_empty)
 537                 return 0;
 538         if (!nl_just_seen)
 539                 count++; /* no trailing newline */
 540         return count;
 541 }
 542
 543 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 544 {
 545         if (!DIFF_FILE_VALID(one)) {
 546                 mf->ptr = (char *)""; /* does not matter */
 547                 mf->size = 0;
 548                 return 0;
 549         }
 550         else if (diff_populate_filespec(one, 0))
 551                 return -1;
 552
 553         mf->ptr = one->data;
 554         mf->size = one->size;
 555         return 0;
 556 }
 557
 558 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 559 static unsigned long diff_filespec_size(struct diff_filespec *one)
 560 {
 561         if (!DIFF_FILE_VALID(one))
 562                 return 0;
 563         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 564         return one->size;
 565 }
 566
 567 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 568 {
 569         char *ptr = mf->ptr;
 570         long size = mf->size;
 571         int cnt = 0;
 572
 573         if (!size)
 574                 return cnt;
 575         ptr += size - 1; /* pointing at the very end */
 576         if (*ptr != '\n')
 577                 ; /* incomplete line */
 578         else
 579                 ptr--; /* skip the last LF */
 580         while (mf->ptr < ptr) {
 581                 char *prev_eol;
 582                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 583                         if (*prev_eol == '\n')
 584                                 break;
 585                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 586                         break;
 587                 cnt++;
 588                 ptr = prev_eol - 1;
 589         }
 590         return cnt;
 591 }
 592
 593 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 594                                struct emit_callback *ecbdata)
 595 {
 596         int l1, l2, at;
 597         unsigned ws_rule = ecbdata->ws_rule;
 598         l1 = count_trailing_blank(mf1, ws_rule);
 599         l2 = count_trailing_blank(mf2, ws_rule);
 600         if (l2 <= l1) {
 601                 ecbdata->blank_at_eof_in_preimage = 0;
 602                 ecbdata->blank_at_eof_in_postimage = 0;
 603                 return;
 604         }
 605         at = count_lines(mf1->ptr, mf1->size);
 606         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 607
 608         at = count_lines(mf2->ptr, mf2->size);
 609         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 610 }
 611
 612 static void emit_line_0(struct diff_options *o,
 613                         const char *set, unsigned reverse, const char *reset,
 614                         int first, const char *line, int len)
 615 {
 616         int has_trailing_newline, has_trailing_carriage_return;
 617         int nofirst;
 618         FILE *file = o->file;
 619
 620         if (first)
 621                 fputs(diff_line_prefix(o), file);
 622         else if (!len)
 623                 return;
 624
 625         if (len == 0) {
 626                 has_trailing_newline = (first == '\n');
 627                 has_trailing_carriage_return = (!has_trailing_newline &&
 628                                                 (first == '\r'));
 629                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 630         } else {
 631                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 632                 if (has_trailing_newline)
 633                         len--;
 634                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 635                 if (has_trailing_carriage_return)
 636                         len--;
 637                 nofirst = 0;
 638         }
 639
 640         if (len || !nofirst) {
 641                 if (reverse && want_color(o->use_color))
 642                         fputs(GIT_COLOR_REVERSE, file);
 643                 fputs(set, file);
 644                 if (first && !nofirst)
 645                         fputc(first, file);
 646                 fwrite(line, len, 1, file);
 647                 fputs(reset, file);
 648         }
 649         if (has_trailing_carriage_return)
 650                 fputc('\r', file);
 651         if (has_trailing_newline)
 652                 fputc('\n', file);
 653 }
 654
 655 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 656                       const char *line, int len)
 657 {
 658         emit_line_0(o, set, 0, reset, line[0], line+1, len-1);
 659 }
 660
 661 enum diff_symbol {
 662         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 663         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 664         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 665         DIFF_SYMBOL_BINARY_DIFF_BODY,
 666         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 667         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 668         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 669         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 670         DIFF_SYMBOL_STATS_LINE,
 671         DIFF_SYMBOL_WORD_DIFF,
 672         DIFF_SYMBOL_STAT_SEP,
 673         DIFF_SYMBOL_SUMMARY,
 674         DIFF_SYMBOL_SUBMODULE_ADD,
 675         DIFF_SYMBOL_SUBMODULE_DEL,
 676         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 677         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 678         DIFF_SYMBOL_SUBMODULE_HEADER,
 679         DIFF_SYMBOL_SUBMODULE_ERROR,
 680         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 681         DIFF_SYMBOL_REWRITE_DIFF,
 682         DIFF_SYMBOL_BINARY_FILES,
 683         DIFF_SYMBOL_HEADER,
 684         DIFF_SYMBOL_FILEPAIR_PLUS,
 685         DIFF_SYMBOL_FILEPAIR_MINUS,
 686         DIFF_SYMBOL_WORDS_PORCELAIN,
 687         DIFF_SYMBOL_WORDS,
 688         DIFF_SYMBOL_CONTEXT,
 689         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 690         DIFF_SYMBOL_PLUS,
 691         DIFF_SYMBOL_MINUS,
 692         DIFF_SYMBOL_NO_LF_EOF,
 693         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 694         DIFF_SYMBOL_CONTEXT_MARKER,
 695         DIFF_SYMBOL_SEPARATOR
 696 };
 697 /*
 698  * Flags for content lines:
 699  * 0..12 are whitespace rules
 700  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 701  * 16 is marking if the line is blank at EOF
 702  */
 703 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 704 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 705 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 706 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 707 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 708
 709 /*
 710  * This struct is used when we need to buffer the output of the diff output.
 711  *
 712  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 713  * into the pre/post image file. This pointer could be a union with the
 714  * line pointer. By storing an offset into the file instead of the literal line,
 715  * we can decrease the memory footprint for the buffered output. At first we
 716  * may want to only have indirection for the content lines, but we could also
 717  * enhance the state for emitting prefabricated lines, e.g. the similarity
 718  * score line or hunk/file headers would only need to store a number or path
 719  * and then the output can be constructed later on depending on state.
 720  */
 721 struct emitted_diff_symbol {
 722         const char *line;
 723         int len;
 724         int flags;
 725         enum diff_symbol s;
 726 };
 727 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 728
 729 struct emitted_diff_symbols {
 730         struct emitted_diff_symbol *buf;
 731         int nr, alloc;
 732 };
 733 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 734
 735 static void append_emitted_diff_symbol(struct diff_options *o,
 736                                        struct emitted_diff_symbol *e)
 737 {
 738         struct emitted_diff_symbol *f;
 739
 740         ALLOC_GROW(o->emitted_symbols->buf,
 741                    o->emitted_symbols->nr + 1,
 742                    o->emitted_symbols->alloc);
 743         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 744
 745         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 746         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 747 }
 748
 749 struct moved_entry {
 750         struct hashmap_entry ent;
 751         const struct emitted_diff_symbol *es;
 752         struct moved_entry *next_line;
 753         struct ws_delta *wsd;
 754 };
 755
 756 /**
 757  * The struct ws_delta holds white space differences between moved lines, i.e.
 758  * between '+' and '-' lines that have been detected to be a move.
 759  * The string contains the difference in leading white spaces, before the
 760  * rest of the line is compared using the white space config for move
 761  * coloring. The current_longer indicates if the first string in the
 762  * comparision is longer than the second.
 763  */
 764 struct ws_delta {
 765         char *string;
 766         unsigned int current_longer : 1;
 767 };
 768 #define WS_DELTA_INIT { NULL, 0 }
 769
 770 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 771                              const struct emitted_diff_symbol *b,
 772                              struct ws_delta *out)
 773 {
 774         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 775         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 776         int d = longer->len - shorter->len;
 777
 778         out->string = xmemdupz(longer->line, d);
 779         out->current_longer = (a == longer);
 780
 781         return !strncmp(longer->line + d, shorter->line, shorter->len);
 782 }
 783
 784 static int cmp_in_block_with_wsd(const struct diff_options *o,
 785                                  const struct moved_entry *cur,
 786                                  const struct moved_entry *match,
 787                                  struct moved_entry *pmb,
 788                                  int n)
 789 {
 790         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 791         int al = cur->es->len, cl = l->len;
 792         const char *a = cur->es->line,
 793                    *b = match->es->line,
 794                    *c = l->line;
 795
 796         int wslen;
 797
 798         /*
 799          * We need to check if 'cur' is equal to 'match'.
 800          * As those are from the same (+/-) side, we do not need to adjust for
 801          * indent changes. However these were found using fuzzy matching
 802          * so we do have to check if they are equal.
 803          */
 804         if (strcmp(a, b))
 805                 return 1;
 806
 807         if (!pmb->wsd)
 808                 /*
 809                  * No white space delta was carried forward? This can happen
 810                  * when we exit early in this function and do not carry
 811                  * forward ws.
 812                  */
 813                 return 1;
 814
 815         /*
 816          * The indent changes of the block are known and carried forward in
 817          * pmb->wsd; however we need to check if the indent changes of the
 818          * current line are still the same as before.
 819          *
 820          * To do so we need to compare 'l' to 'cur', adjusting the
 821          * one of them for the white spaces, depending which was longer.
 822          */
 823
 824         wslen = strlen(pmb->wsd->string);
 825         if (pmb->wsd->current_longer) {
 826                 c += wslen;
 827                 cl -= wslen;
 828         } else {
 829                 a += wslen;
 830                 al -= wslen;
 831         }
 832
 833         if (strcmp(a, c))
 834                 return 1;
 835
 836         return 0;
 837 }
 838
 839 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 840                            const void *entry,
 841                            const void *entry_or_key,
 842                            const void *keydata)
 843 {
 844         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 845         const struct moved_entry *a = entry;
 846         const struct moved_entry *b = entry_or_key;
 847         unsigned flags = diffopt->color_moved_ws_handling
 848                          & XDF_WHITESPACE_FLAGS;
 849
 850         if (diffopt->color_moved_ws_handling &
 851             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 852                 /*
 853                  * As there is not specific white space config given,
 854                  * we'd need to check for a new block, so ignore all
 855                  * white space. The setup of the white space
 856                  * configuration for the next block is done else where
 857                  */
 858                 flags |= XDF_IGNORE_WHITESPACE;
 859
 860         return !xdiff_compare_lines(a->es->line, a->es->len,
 861                                     b->es->line, b->es->len,
 862                                     flags);
 863 }
 864
 865 static struct moved_entry *prepare_entry(struct diff_options *o,
 866                                          int line_no)
 867 {
 868         struct moved_entry *ret = xmalloc(sizeof(*ret));
 869         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 870         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 871
 872         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 873         ret->es = l;
 874         ret->next_line = NULL;
 875         ret->wsd = NULL;
 876
 877         return ret;
 878 }
 879
 880 static void add_lines_to_move_detection(struct diff_options *o,
 881                                         struct hashmap *add_lines,
 882                                         struct hashmap *del_lines)
 883 {
 884         struct moved_entry *prev_line = NULL;
 885
 886         int n;
 887         for (n = 0; n < o->emitted_symbols->nr; n++) {
 888                 struct hashmap *hm;
 889                 struct moved_entry *key;
 890
 891                 switch (o->emitted_symbols->buf[n].s) {
 892                 case DIFF_SYMBOL_PLUS:
 893                         hm = add_lines;
 894                         break;
 895                 case DIFF_SYMBOL_MINUS:
 896                         hm = del_lines;
 897                         break;
 898                 default:
 899                         prev_line = NULL;
 900                         continue;
 901                 }
 902
 903                 key = prepare_entry(o, n);
 904                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 905                         prev_line->next_line = key;
 906
 907                 hashmap_add(hm, key);
 908                 prev_line = key;
 909         }
 910 }
 911
 912 static void pmb_advance_or_null(struct diff_options *o,
 913                                 struct moved_entry *match,
 914                                 struct hashmap *hm,
 915                                 struct moved_entry **pmb,
 916                                 int pmb_nr)
 917 {
 918         int i;
 919         for (i = 0; i < pmb_nr; i++) {
 920                 struct moved_entry *prev = pmb[i];
 921                 struct moved_entry *cur = (prev && prev->next_line) ?
 922                                 prev->next_line : NULL;
 923                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 924                         pmb[i] = cur;
 925                 } else {
 926                         pmb[i] = NULL;
 927                 }
 928         }
 929 }
 930
 931 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 932                                             struct moved_entry *match,
 933                                             struct hashmap *hm,
 934                                             struct moved_entry **pmb,
 935                                             int pmb_nr, int n)
 936 {
 937         int i;
 938         char *got_match = xcalloc(1, pmb_nr);
 939
 940         for (; match; match = hashmap_get_next(hm, match)) {
 941                 for (i = 0; i < pmb_nr; i++) {
 942                         struct moved_entry *prev = pmb[i];
 943                         struct moved_entry *cur = (prev && prev->next_line) ?
 944                                         prev->next_line : NULL;
 945                         if (!cur)
 946                                 continue;
 947                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 948                                 got_match[i] |= 1;
 949                 }
 950         }
 951
 952         for (i = 0; i < pmb_nr; i++) {
 953                 if (got_match[i]) {
 954                         /* Carry the white space delta forward */
 955                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 956                         pmb[i] = pmb[i]->next_line;
 957                 } else
 958                         pmb[i] = NULL;
 959         }
 960 }
 961
 962 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 963                                          int pmb_nr)
 964 {
 965         int lp, rp;
 966
 967         /* Shrink the set of potential block to the remaining running */
 968         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 969                 while (lp < pmb_nr && pmb[lp])
 970                         lp++;
 971                 /* lp points at the first NULL now */
 972
 973                 while (rp > -1 && !pmb[rp])
 974                         rp--;
 975                 /* rp points at the last non-NULL */
 976
 977                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 978                         pmb[lp] = pmb[rp];
 979                         if (pmb[rp]->wsd) {
 980                                 free(pmb[rp]->wsd->string);
 981                                 FREE_AND_NULL(pmb[rp]->wsd);
 982                         }
 983                         pmb[rp] = NULL;
 984                         rp--;
 985                         lp++;
 986                 }
 987         }
 988
 989         /* Remember the number of running sets */
 990         return rp + 1;
 991 }
 992
 993 /*
 994  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 995  *
 996  * Otherwise, if the last block has fewer alphanumeric characters than
 997  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 998  * that block.
 999  *
1000  * The last block consists of the (n - block_length)'th line up to but not
1001  * including the nth line.
1002  *
1003  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1004  * Think of a way to unify them.
1005  */
1006 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1007 {
1008         int i, alnum_count = 0;
1009         if (o->color_moved == COLOR_MOVED_PLAIN)
1010                 return;
1011         for (i = 1; i < block_length + 1; i++) {
1012                 const char *c = o->emitted_symbols->buf[n - i].line;
1013                 for (; *c; c++) {
1014                         if (!isalnum(*c))
1015                                 continue;
1016                         alnum_count++;
1017                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1018                                 return;
1019                 }
1020         }
1021         for (i = 1; i < block_length + 1; i++)
1022                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1023 }
1024
1025 /* Find blocks of moved code, delegate actual coloring decision to helper */
1026 static void mark_color_as_moved(struct diff_options *o,
1027                                 struct hashmap *add_lines,
1028                                 struct hashmap *del_lines)
1029 {
1030         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1031         int pmb_nr = 0, pmb_alloc = 0;
1032         int n, flipped_block = 1, block_length = 0;
1033
1034
1035         for (n = 0; n < o->emitted_symbols->nr; n++) {
1036                 struct hashmap *hm = NULL;
1037                 struct moved_entry *key;
1038                 struct moved_entry *match = NULL;
1039                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1040
1041                 switch (l->s) {
1042                 case DIFF_SYMBOL_PLUS:
1043                         hm = del_lines;
1044                         key = prepare_entry(o, n);
1045                         match = hashmap_get(hm, key, NULL);
1046                         free(key);
1047                         break;
1048                 case DIFF_SYMBOL_MINUS:
1049                         hm = add_lines;
1050                         key = prepare_entry(o, n);
1051                         match = hashmap_get(hm, key, NULL);
1052                         free(key);
1053                         break;
1054                 default:
1055                         flipped_block = 1;
1056                 }
1057
1058                 if (!match) {
1059                         adjust_last_block(o, n, block_length);
1060                         pmb_nr = 0;
1061                         block_length = 0;
1062                         continue;
1063                 }
1064
1065                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1066
1067                 if (o->color_moved == COLOR_MOVED_PLAIN)
1068                         continue;
1069
1070                 if (o->color_moved_ws_handling &
1071                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1072                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1073                 else
1074                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1075
1076                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1077
1078                 if (pmb_nr == 0) {
1079                         /*
1080                          * The current line is the start of a new block.
1081                          * Setup the set of potential blocks.
1082                          */
1083                         for (; match; match = hashmap_get_next(hm, match)) {
1084                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1085                                 if (o->color_moved_ws_handling &
1086                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1087                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1088                                         if (compute_ws_delta(l, match->es, wsd)) {
1089                                                 match->wsd = wsd;
1090                                                 pmb[pmb_nr++] = match;
1091                                         } else
1092                                                 free(wsd);
1093                                 } else {
1094                                         pmb[pmb_nr++] = match;
1095                                 }
1096                         }
1097
1098                         flipped_block = (flipped_block + 1) % 2;
1099
1100                         adjust_last_block(o, n, block_length);
1101                         block_length = 0;
1102                 }
1103
1104                 block_length++;
1105
1106                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1107                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1108         }
1109         adjust_last_block(o, n, block_length);
1110
1111         free(pmb);
1112 }
1113
1114 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1115   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1116 static void dim_moved_lines(struct diff_options *o)
1117 {
1118         int n;
1119         for (n = 0; n < o->emitted_symbols->nr; n++) {
1120                 struct emitted_diff_symbol *prev = (n != 0) ?
1121                                 &o->emitted_symbols->buf[n - 1] : NULL;
1122                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1123                 struct emitted_diff_symbol *next =
1124                                 (n < o->emitted_symbols->nr - 1) ?
1125                                 &o->emitted_symbols->buf[n + 1] : NULL;
1126
1127                 /* Not a plus or minus line? */
1128                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1129                         continue;
1130
1131                 /* Not a moved line? */
1132                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1133                         continue;
1134
1135                 /*
1136                  * If prev or next are not a plus or minus line,
1137                  * pretend they don't exist
1138                  */
1139                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1140                             prev->s != DIFF_SYMBOL_MINUS)
1141                         prev = NULL;
1142                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1143                             next->s != DIFF_SYMBOL_MINUS)
1144                         next = NULL;
1145
1146                 /* Inside a block? */
1147                 if ((prev &&
1148                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1149                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1150                     (next &&
1151                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1152                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1153                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1154                         continue;
1155                 }
1156
1157                 /* Check if we are at an interesting bound: */
1158                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1159                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1160                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1161                         continue;
1162                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1163                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1164                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1165                         continue;
1166
1167                 /*
1168                  * The boundary to prev and next are not interesting,
1169                  * so this line is not interesting as a whole
1170                  */
1171                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1172         }
1173 }
1174
1175 static void emit_line_ws_markup(struct diff_options *o,
1176                                 const char *set, const char *reset,
1177                                 const char *line, int len,
1178                                 const char *set_sign, char sign,
1179                                 unsigned ws_rule, int blank_at_eof)
1180 {
1181         const char *ws = NULL;
1182
1183         if (o->ws_error_highlight & ws_rule) {
1184                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1185                 if (!*ws)
1186                         ws = NULL;
1187         }
1188
1189         if (!ws && !set_sign)
1190                 emit_line_0(o, set, 0, reset, sign, line, len);
1191         else if (!ws) {
1192                 /* Emit just the prefix, then the rest. */
1193                 emit_line_0(o, set_sign ? set_sign : set, !!set_sign, reset,
1194                             sign, "", 0);
1195                 emit_line_0(o, set, 0, reset, 0, line, len);
1196         } else if (blank_at_eof)
1197                 /* Blank line at EOF - paint '+' as well */
1198                 emit_line_0(o, ws, 0, reset, sign, line, len);
1199         else {
1200                 /* Emit just the prefix, then the rest. */
1201                 emit_line_0(o, set_sign ? set_sign : set, !!set_sign, reset,
1202                             sign, "", 0);
1203                 ws_check_emit(line, len, ws_rule,
1204                               o->file, set, reset, ws);
1205         }
1206 }
1207
1208 static void emit_diff_symbol_from_struct(struct diff_options *o,
1209                                          struct emitted_diff_symbol *eds)
1210 {
1211         static const char *nneof = " No newline at end of file\n";
1212         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1213         struct strbuf sb = STRBUF_INIT;
1214
1215         enum diff_symbol s = eds->s;
1216         const char *line = eds->line;
1217         int len = eds->len;
1218         unsigned flags = eds->flags;
1219
1220         switch (s) {
1221         case DIFF_SYMBOL_NO_LF_EOF:
1222                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1223                 reset = diff_get_color_opt(o, DIFF_RESET);
1224                 putc('\n', o->file);
1225                 emit_line_0(o, context, 0, reset, '\\',
1226                             nneof, strlen(nneof));
1227                 break;
1228         case DIFF_SYMBOL_SUBMODULE_HEADER:
1229         case DIFF_SYMBOL_SUBMODULE_ERROR:
1230         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1231         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1232         case DIFF_SYMBOL_SUMMARY:
1233         case DIFF_SYMBOL_STATS_LINE:
1234         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1235         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1236                 emit_line(o, "", "", line, len);
1237                 break;
1238         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1239         case DIFF_SYMBOL_CONTEXT_MARKER:
1240                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1241                 reset = diff_get_color_opt(o, DIFF_RESET);
1242                 emit_line(o, context, reset, line, len);
1243                 break;
1244         case DIFF_SYMBOL_SEPARATOR:
1245                 fprintf(o->file, "%s%c",
1246                         diff_line_prefix(o),
1247                         o->line_termination);
1248                 break;
1249         case DIFF_SYMBOL_CONTEXT:
1250                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1251                 reset = diff_get_color_opt(o, DIFF_RESET);
1252                 set_sign = NULL;
1253                 if (o->flags.dual_color_diffed_diffs) {
1254                         char c = !len ? 0 : line[0];
1255
1256                         if (c == '+')
1257                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1258                         else if (c == '@')
1259                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1260                         else if (c == '-')
1261                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1262                 }
1263                 emit_line_ws_markup(o, set, reset, line, len, set_sign, ' ',
1264                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1265                 break;
1266         case DIFF_SYMBOL_PLUS:
1267                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1268                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1269                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1270                 case DIFF_SYMBOL_MOVED_LINE |
1271                      DIFF_SYMBOL_MOVED_LINE_ALT |
1272                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1273                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1274                         break;
1275                 case DIFF_SYMBOL_MOVED_LINE |
1276                      DIFF_SYMBOL_MOVED_LINE_ALT:
1277                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1278                         break;
1279                 case DIFF_SYMBOL_MOVED_LINE |
1280                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1281                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1282                         break;
1283                 case DIFF_SYMBOL_MOVED_LINE:
1284                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1285                         break;
1286                 default:
1287                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1288                 }
1289                 reset = diff_get_color_opt(o, DIFF_RESET);
1290                 if (!o->flags.dual_color_diffed_diffs)
1291                         set_sign = NULL;
1292                 else {
1293                         char c = !len ? 0 : line[0];
1294
1295                         set_sign = set;
1296                         if (c == '-')
1297                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1298                         else if (c == '@')
1299                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1300                         else if (c != '+')
1301                                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1302                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1303                 }
1304                 emit_line_ws_markup(o, set, reset, line, len, set_sign, '+',
1305                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1306                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1307                 break;
1308         case DIFF_SYMBOL_MINUS:
1309                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1310                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1311                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1312                 case DIFF_SYMBOL_MOVED_LINE |
1313                      DIFF_SYMBOL_MOVED_LINE_ALT |
1314                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1315                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1316                         break;
1317                 case DIFF_SYMBOL_MOVED_LINE |
1318                      DIFF_SYMBOL_MOVED_LINE_ALT:
1319                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1320                         break;
1321                 case DIFF_SYMBOL_MOVED_LINE |
1322                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1323                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1324                         break;
1325                 case DIFF_SYMBOL_MOVED_LINE:
1326                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1327                         break;
1328                 default:
1329                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1330                 }
1331                 reset = diff_get_color_opt(o, DIFF_RESET);
1332                 if (!o->flags.dual_color_diffed_diffs)
1333                         set_sign = NULL;
1334                 else {
1335                         char c = !len ? 0 : line[0];
1336
1337                         set_sign = set;
1338                         if (c == '+')
1339                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1340                         else if (c == '@')
1341                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1342                         else if (c != '-')
1343                                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1344                 }
1345                 emit_line_ws_markup(o, set, reset, line, len, set_sign, '-',
1346                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1347                 break;
1348         case DIFF_SYMBOL_WORDS_PORCELAIN:
1349                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1350                 reset = diff_get_color_opt(o, DIFF_RESET);
1351                 emit_line(o, context, reset, line, len);
1352                 fputs("~\n", o->file);
1353                 break;
1354         case DIFF_SYMBOL_WORDS:
1355                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1356                 reset = diff_get_color_opt(o, DIFF_RESET);
1357                 /*
1358                  * Skip the prefix character, if any.  With
1359                  * diff_suppress_blank_empty, there may be
1360                  * none.
1361                  */
1362                 if (line[0] != '\n') {
1363                         line++;
1364                         len--;
1365                 }
1366                 emit_line(o, context, reset, line, len);
1367                 break;
1368         case DIFF_SYMBOL_FILEPAIR_PLUS:
1369                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1370                 reset = diff_get_color_opt(o, DIFF_RESET);
1371                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1372                         line, reset,
1373                         strchr(line, ' ') ? "\t" : "");
1374                 break;
1375         case DIFF_SYMBOL_FILEPAIR_MINUS:
1376                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1377                 reset = diff_get_color_opt(o, DIFF_RESET);
1378                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1379                         line, reset,
1380                         strchr(line, ' ') ? "\t" : "");
1381                 break;
1382         case DIFF_SYMBOL_BINARY_FILES:
1383         case DIFF_SYMBOL_HEADER:
1384                 fprintf(o->file, "%s", line);
1385                 break;
1386         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1387                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1388                 break;
1389         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1390                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1391                 break;
1392         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1393                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1394                 break;
1395         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1396                 fputs(diff_line_prefix(o), o->file);
1397                 fputc('\n', o->file);
1398                 break;
1399         case DIFF_SYMBOL_REWRITE_DIFF:
1400                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1401                 reset = diff_get_color_opt(o, DIFF_RESET);
1402                 emit_line(o, fraginfo, reset, line, len);
1403                 break;
1404         case DIFF_SYMBOL_SUBMODULE_ADD:
1405                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1406                 reset = diff_get_color_opt(o, DIFF_RESET);
1407                 emit_line(o, set, reset, line, len);
1408                 break;
1409         case DIFF_SYMBOL_SUBMODULE_DEL:
1410                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1411                 reset = diff_get_color_opt(o, DIFF_RESET);
1412                 emit_line(o, set, reset, line, len);
1413                 break;
1414         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1415                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1416                         diff_line_prefix(o), line);
1417                 break;
1418         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1419                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1420                         diff_line_prefix(o), line);
1421                 break;
1422         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1423                 emit_line(o, "", "", " 0 files changed\n",
1424                           strlen(" 0 files changed\n"));
1425                 break;
1426         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1427                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1428                 break;
1429         case DIFF_SYMBOL_WORD_DIFF:
1430                 fprintf(o->file, "%.*s", len, line);
1431                 break;
1432         case DIFF_SYMBOL_STAT_SEP:
1433                 fputs(o->stat_sep, o->file);
1434                 break;
1435         default:
1436                 BUG("unknown diff symbol");
1437         }
1438         strbuf_release(&sb);
1439 }
1440
1441 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1442                              const char *line, int len, unsigned flags)
1443 {
1444         struct emitted_diff_symbol e = {line, len, flags, s};
1445
1446         if (o->emitted_symbols)
1447                 append_emitted_diff_symbol(o, &e);
1448         else
1449                 emit_diff_symbol_from_struct(o, &e);
1450 }
1451
1452 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1453 {
1454         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1455 }
1456
1457 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1458 {
1459         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1460 }
1461
1462 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1463 {
1464         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1465                          path, strlen(path), 0);
1466 }
1467
1468 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1469 {
1470         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1471                          path, strlen(path), 0);
1472 }
1473
1474 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1475 {
1476         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1477                          header, strlen(header), 0);
1478 }
1479
1480 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1481 {
1482         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1483 }
1484
1485 void diff_emit_submodule_pipethrough(struct diff_options *o,
1486                                      const char *line, int len)
1487 {
1488         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1489 }
1490
1491 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1492 {
1493         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1494               ecbdata->blank_at_eof_in_preimage &&
1495               ecbdata->blank_at_eof_in_postimage &&
1496               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1497               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1498                 return 0;
1499         return ws_blank_line(line, len, ecbdata->ws_rule);
1500 }
1501
1502 static void emit_add_line(const char *reset,
1503                           struct emit_callback *ecbdata,
1504                           const char *line, int len)
1505 {
1506         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1507         if (new_blank_line_at_eof(ecbdata, line, len))
1508                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1509
1510         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1511 }
1512
1513 static void emit_del_line(const char *reset,
1514                           struct emit_callback *ecbdata,
1515                           const char *line, int len)
1516 {
1517         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1518         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1519 }
1520
1521 static void emit_context_line(const char *reset,
1522                               struct emit_callback *ecbdata,
1523                               const char *line, int len)
1524 {
1525         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1526         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1527 }
1528
1529 static void emit_hunk_header(struct emit_callback *ecbdata,
1530                              const char *line, int len)
1531 {
1532         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1533         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1534         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1535         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1536         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1537         static const char atat[2] = { '@', '@' };
1538         const char *cp, *ep;
1539         struct strbuf msgbuf = STRBUF_INIT;
1540         int org_len = len;
1541         int i = 1;
1542
1543         /*
1544          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1545          * it always is at least 10 bytes long.
1546          */
1547         if (len < 10 ||
1548             memcmp(line, atat, 2) ||
1549             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1550                 emit_diff_symbol(ecbdata->opt,
1551                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1552                 return;
1553         }
1554         ep += 2; /* skip over @@ */
1555
1556         /* The hunk header in fraginfo color */
1557         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1558                 strbuf_addstr(&msgbuf, reverse);
1559         strbuf_addstr(&msgbuf, frag);
1560         strbuf_add(&msgbuf, line, ep - line);
1561         strbuf_addstr(&msgbuf, reset);
1562
1563         /*
1564          * trailing "\r\n"
1565          */
1566         for ( ; i < 3; i++)
1567                 if (line[len - i] == '\r' || line[len - i] == '\n')
1568                         len--;
1569
1570         /* blank before the func header */
1571         for (cp = ep; ep - line < len; ep++)
1572                 if (*ep != ' ' && *ep != '\t')
1573                         break;
1574         if (ep != cp) {
1575                 strbuf_addstr(&msgbuf, context);
1576                 strbuf_add(&msgbuf, cp, ep - cp);
1577                 strbuf_addstr(&msgbuf, reset);
1578         }
1579
1580         if (ep < line + len) {
1581                 strbuf_addstr(&msgbuf, func);
1582                 strbuf_add(&msgbuf, ep, line + len - ep);
1583                 strbuf_addstr(&msgbuf, reset);
1584         }
1585
1586         strbuf_add(&msgbuf, line + len, org_len - len);
1587         strbuf_complete_line(&msgbuf);
1588         emit_diff_symbol(ecbdata->opt,
1589                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1590         strbuf_release(&msgbuf);
1591 }
1592
1593 static struct diff_tempfile *claim_diff_tempfile(void) {
1594         int i;
1595         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1596                 if (!diff_temp[i].name)
1597                         return diff_temp + i;
1598         BUG("diff is failing to clean up its tempfiles");
1599 }
1600
1601 static void remove_tempfile(void)
1602 {
1603         int i;
1604         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1605                 if (is_tempfile_active(diff_temp[i].tempfile))
1606                         delete_tempfile(&diff_temp[i].tempfile);
1607                 diff_temp[i].name = NULL;
1608         }
1609 }
1610
1611 static void add_line_count(struct strbuf *out, int count)
1612 {
1613         switch (count) {
1614         case 0:
1615                 strbuf_addstr(out, "0,0");
1616                 break;
1617         case 1:
1618                 strbuf_addstr(out, "1");
1619                 break;
1620         default:
1621                 strbuf_addf(out, "1,%d", count);
1622                 break;
1623         }
1624 }
1625
1626 static void emit_rewrite_lines(struct emit_callback *ecb,
1627                                int prefix, const char *data, int size)
1628 {
1629         const char *endp = NULL;
1630         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1631
1632         while (0 < size) {
1633                 int len;
1634
1635                 endp = memchr(data, '\n', size);
1636                 len = endp ? (endp - data + 1) : size;
1637                 if (prefix != '+') {
1638                         ecb->lno_in_preimage++;
1639                         emit_del_line(reset, ecb, data, len);
1640                 } else {
1641                         ecb->lno_in_postimage++;
1642                         emit_add_line(reset, ecb, data, len);
1643                 }
1644                 size -= len;
1645                 data += len;
1646         }
1647         if (!endp)
1648                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1649 }
1650
1651 static void emit_rewrite_diff(const char *name_a,
1652                               const char *name_b,
1653                               struct diff_filespec *one,
1654                               struct diff_filespec *two,
1655                               struct userdiff_driver *textconv_one,
1656                               struct userdiff_driver *textconv_two,
1657                               struct diff_options *o)
1658 {
1659         int lc_a, lc_b;
1660         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1661         const char *a_prefix, *b_prefix;
1662         char *data_one, *data_two;
1663         size_t size_one, size_two;
1664         struct emit_callback ecbdata;
1665         struct strbuf out = STRBUF_INIT;
1666
1667         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1668                 a_prefix = o->b_prefix;
1669                 b_prefix = o->a_prefix;
1670         } else {
1671                 a_prefix = o->a_prefix;
1672                 b_prefix = o->b_prefix;
1673         }
1674
1675         name_a += (*name_a == '/');
1676         name_b += (*name_b == '/');
1677
1678         strbuf_reset(&a_name);
1679         strbuf_reset(&b_name);
1680         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1681         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1682
1683         size_one = fill_textconv(textconv_one, one, &data_one);
1684         size_two = fill_textconv(textconv_two, two, &data_two);
1685
1686         memset(&ecbdata, 0, sizeof(ecbdata));
1687         ecbdata.color_diff = want_color(o->use_color);
1688         ecbdata.ws_rule = whitespace_rule(name_b);
1689         ecbdata.opt = o;
1690         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1691                 mmfile_t mf1, mf2;
1692                 mf1.ptr = (char *)data_one;
1693                 mf2.ptr = (char *)data_two;
1694                 mf1.size = size_one;
1695                 mf2.size = size_two;
1696                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1697         }
1698         ecbdata.lno_in_preimage = 1;
1699         ecbdata.lno_in_postimage = 1;
1700
1701         lc_a = count_lines(data_one, size_one);
1702         lc_b = count_lines(data_two, size_two);
1703
1704         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1705                          a_name.buf, a_name.len, 0);
1706         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1707                          b_name.buf, b_name.len, 0);
1708
1709         strbuf_addstr(&out, "@@ -");
1710         if (!o->irreversible_delete)
1711                 add_line_count(&out, lc_a);
1712         else
1713                 strbuf_addstr(&out, "?,?");
1714         strbuf_addstr(&out, " +");
1715         add_line_count(&out, lc_b);
1716         strbuf_addstr(&out, " @@\n");
1717         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1718         strbuf_release(&out);
1719
1720         if (lc_a && !o->irreversible_delete)
1721                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1722         if (lc_b)
1723                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1724         if (textconv_one)
1725                 free((char *)data_one);
1726         if (textconv_two)
1727                 free((char *)data_two);
1728 }
1729
1730 struct diff_words_buffer {
1731         mmfile_t text;
1732         unsigned long alloc;
1733         struct diff_words_orig {
1734                 const char *begin, *end;
1735         } *orig;
1736         int orig_nr, orig_alloc;
1737 };
1738
1739 static void diff_words_append(char *line, unsigned long len,
1740                 struct diff_words_buffer *buffer)
1741 {
1742         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1743         line++;
1744         len--;
1745         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1746         buffer->text.size += len;
1747         buffer->text.ptr[buffer->text.size] = '\0';
1748 }
1749
1750 struct diff_words_style_elem {
1751         const char *prefix;
1752         const char *suffix;
1753         const char *color; /* NULL; filled in by the setup code if
1754                             * color is enabled */
1755 };
1756
1757 struct diff_words_style {
1758         enum diff_words_type type;
1759         struct diff_words_style_elem new_word, old_word, ctx;
1760         const char *newline;
1761 };
1762
1763 static struct diff_words_style diff_words_styles[] = {
1764         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1765         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1766         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1767 };
1768
1769 struct diff_words_data {
1770         struct diff_words_buffer minus, plus;
1771         const char *current_plus;
1772         int last_minus;
1773         struct diff_options *opt;
1774         regex_t *word_regex;
1775         enum diff_words_type type;
1776         struct diff_words_style *style;
1777 };
1778
1779 static int fn_out_diff_words_write_helper(struct diff_options *o,
1780                                           struct diff_words_style_elem *st_el,
1781                                           const char *newline,
1782                                           size_t count, const char *buf)
1783 {
1784         int print = 0;
1785         struct strbuf sb = STRBUF_INIT;
1786
1787         while (count) {
1788                 char *p = memchr(buf, '\n', count);
1789                 if (print)
1790                         strbuf_addstr(&sb, diff_line_prefix(o));
1791
1792                 if (p != buf) {
1793                         const char *reset = st_el->color && *st_el->color ?
1794                                             GIT_COLOR_RESET : NULL;
1795                         if (st_el->color && *st_el->color)
1796                                 strbuf_addstr(&sb, st_el->color);
1797                         strbuf_addstr(&sb, st_el->prefix);
1798                         strbuf_add(&sb, buf, p ? p - buf : count);
1799                         strbuf_addstr(&sb, st_el->suffix);
1800                         if (reset)
1801                                 strbuf_addstr(&sb, reset);
1802                 }
1803                 if (!p)
1804                         goto out;
1805
1806                 strbuf_addstr(&sb, newline);
1807                 count -= p + 1 - buf;
1808                 buf = p + 1;
1809                 print = 1;
1810                 if (count) {
1811                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1812                                          sb.buf, sb.len, 0);
1813                         strbuf_reset(&sb);
1814                 }
1815         }
1816
1817 out:
1818         if (sb.len)
1819                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1820                                  sb.buf, sb.len, 0);
1821         strbuf_release(&sb);
1822         return 0;
1823 }
1824
1825 /*
1826  * '--color-words' algorithm can be described as:
1827  *
1828  *   1. collect the minus/plus lines of a diff hunk, divided into
1829  *      minus-lines and plus-lines;
1830  *
1831  *   2. break both minus-lines and plus-lines into words and
1832  *      place them into two mmfile_t with one word for each line;
1833  *
1834  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1835  *
1836  * And for the common parts of the both file, we output the plus side text.
1837  * diff_words->current_plus is used to trace the current position of the plus file
1838  * which printed. diff_words->last_minus is used to trace the last minus word
1839  * printed.
1840  *
1841  * For '--graph' to work with '--color-words', we need to output the graph prefix
1842  * on each line of color words output. Generally, there are two conditions on
1843  * which we should output the prefix.
1844  *
1845  *   1. diff_words->last_minus == 0 &&
1846  *      diff_words->current_plus == diff_words->plus.text.ptr
1847  *
1848  *      that is: the plus text must start as a new line, and if there is no minus
1849  *      word printed, a graph prefix must be printed.
1850  *
1851  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1852  *      *(diff_words->current_plus - 1) == '\n'
1853  *
1854  *      that is: a graph prefix must be printed following a '\n'
1855  */
1856 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1857 {
1858         if ((diff_words->last_minus == 0 &&
1859                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1860                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1861                 *(diff_words->current_plus - 1) == '\n')) {
1862                 return 1;
1863         } else {
1864                 return 0;
1865         }
1866 }
1867
1868 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1869 {
1870         struct diff_words_data *diff_words = priv;
1871         struct diff_words_style *style = diff_words->style;
1872         int minus_first, minus_len, plus_first, plus_len;
1873         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1874         struct diff_options *opt = diff_words->opt;
1875         const char *line_prefix;
1876
1877         if (line[0] != '@' || parse_hunk_header(line, len,
1878                         &minus_first, &minus_len, &plus_first, &plus_len))
1879                 return;
1880
1881         assert(opt);
1882         line_prefix = diff_line_prefix(opt);
1883
1884         /* POSIX requires that first be decremented by one if len == 0... */
1885         if (minus_len) {
1886                 minus_begin = diff_words->minus.orig[minus_first].begin;
1887                 minus_end =
1888                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1889         } else
1890                 minus_begin = minus_end =
1891                         diff_words->minus.orig[minus_first].end;
1892
1893         if (plus_len) {
1894                 plus_begin = diff_words->plus.orig[plus_first].begin;
1895                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1896         } else
1897                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1898
1899         if (color_words_output_graph_prefix(diff_words)) {
1900                 fputs(line_prefix, diff_words->opt->file);
1901         }
1902         if (diff_words->current_plus != plus_begin) {
1903                 fn_out_diff_words_write_helper(diff_words->opt,
1904                                 &style->ctx, style->newline,
1905                                 plus_begin - diff_words->current_plus,
1906                                 diff_words->current_plus);
1907         }
1908         if (minus_begin != minus_end) {
1909                 fn_out_diff_words_write_helper(diff_words->opt,
1910                                 &style->old_word, style->newline,
1911                                 minus_end - minus_begin, minus_begin);
1912         }
1913         if (plus_begin != plus_end) {
1914                 fn_out_diff_words_write_helper(diff_words->opt,
1915                                 &style->new_word, style->newline,
1916                                 plus_end - plus_begin, plus_begin);
1917         }
1918
1919         diff_words->current_plus = plus_end;
1920         diff_words->last_minus = minus_first;
1921 }
1922
1923 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1924 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1925                 int *begin, int *end)
1926 {
1927         if (word_regex && *begin < buffer->size) {
1928                 regmatch_t match[1];
1929                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1930                                  buffer->size - *begin, 1, match, 0)) {
1931                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1932                                         '\n', match[0].rm_eo - match[0].rm_so);
1933                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1934                         *begin += match[0].rm_so;
1935                         return *begin >= *end;
1936                 }
1937                 return -1;
1938         }
1939
1940         /* find the next word */
1941         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1942                 (*begin)++;
1943         if (*begin >= buffer->size)
1944                 return -1;
1945
1946         /* find the end of the word */
1947         *end = *begin + 1;
1948         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1949                 (*end)++;
1950
1951         return 0;
1952 }
1953
1954 /*
1955  * This function splits the words in buffer->text, stores the list with
1956  * newline separator into out, and saves the offsets of the original words
1957  * in buffer->orig.
1958  */
1959 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1960                 regex_t *word_regex)
1961 {
1962         int i, j;
1963         long alloc = 0;
1964
1965         out->size = 0;
1966         out->ptr = NULL;
1967
1968         /* fake an empty "0th" word */
1969         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1970         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1971         buffer->orig_nr = 1;
1972
1973         for (i = 0; i < buffer->text.size; i++) {
1974                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1975                         return;
1976
1977                 /* store original boundaries */
1978                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1979                                 buffer->orig_alloc);
1980                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1981                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1982                 buffer->orig_nr++;
1983
1984                 /* store one word */
1985                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1986                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1987                 out->ptr[out->size + j - i] = '\n';
1988                 out->size += j - i + 1;
1989
1990                 i = j - 1;
1991         }
1992 }
1993
1994 /* this executes the word diff on the accumulated buffers */
1995 static void diff_words_show(struct diff_words_data *diff_words)
1996 {
1997         xpparam_t xpp;
1998         xdemitconf_t xecfg;
1999         mmfile_t minus, plus;
2000         struct diff_words_style *style = diff_words->style;
2001
2002         struct diff_options *opt = diff_words->opt;
2003         const char *line_prefix;
2004
2005         assert(opt);
2006         line_prefix = diff_line_prefix(opt);
2007
2008         /* special case: only removal */
2009         if (!diff_words->plus.text.size) {
2010                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2011                                  line_prefix, strlen(line_prefix), 0);
2012                 fn_out_diff_words_write_helper(diff_words->opt,
2013                         &style->old_word, style->newline,
2014                         diff_words->minus.text.size,
2015                         diff_words->minus.text.ptr);
2016                 diff_words->minus.text.size = 0;
2017                 return;
2018         }
2019
2020         diff_words->current_plus = diff_words->plus.text.ptr;
2021         diff_words->last_minus = 0;
2022
2023         memset(&xpp, 0, sizeof(xpp));
2024         memset(&xecfg, 0, sizeof(xecfg));
2025         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2026         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2027         xpp.flags = 0;
2028         /* as only the hunk header will be parsed, we need a 0-context */
2029         xecfg.ctxlen = 0;
2030         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2031                           &xpp, &xecfg))
2032                 die("unable to generate word diff");
2033         free(minus.ptr);
2034         free(plus.ptr);
2035         if (diff_words->current_plus != diff_words->plus.text.ptr +
2036                         diff_words->plus.text.size) {
2037                 if (color_words_output_graph_prefix(diff_words))
2038                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2039                                          line_prefix, strlen(line_prefix), 0);
2040                 fn_out_diff_words_write_helper(diff_words->opt,
2041                         &style->ctx, style->newline,
2042                         diff_words->plus.text.ptr + diff_words->plus.text.size
2043                         - diff_words->current_plus, diff_words->current_plus);
2044         }
2045         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2046 }
2047
2048 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2049 static void diff_words_flush(struct emit_callback *ecbdata)
2050 {
2051         struct diff_options *wo = ecbdata->diff_words->opt;
2052
2053         if (ecbdata->diff_words->minus.text.size ||
2054             ecbdata->diff_words->plus.text.size)
2055                 diff_words_show(ecbdata->diff_words);
2056
2057         if (wo->emitted_symbols) {
2058                 struct diff_options *o = ecbdata->opt;
2059                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2060                 int i;
2061
2062                 /*
2063                  * NEEDSWORK:
2064                  * Instead of appending each, concat all words to a line?
2065                  */
2066                 for (i = 0; i < wol->nr; i++)
2067                         append_emitted_diff_symbol(o, &wol->buf[i]);
2068
2069                 for (i = 0; i < wol->nr; i++)
2070                         free((void *)wol->buf[i].line);
2071
2072                 wol->nr = 0;
2073         }
2074 }
2075
2076 static void diff_filespec_load_driver(struct diff_filespec *one)
2077 {
2078         /* Use already-loaded driver */
2079         if (one->driver)
2080                 return;
2081
2082         if (S_ISREG(one->mode))
2083                 one->driver = userdiff_find_by_path(one->path);
2084
2085         /* Fallback to default settings */
2086         if (!one->driver)
2087                 one->driver = userdiff_find_by_name("default");
2088 }
2089
2090 static const char *userdiff_word_regex(struct diff_filespec *one)
2091 {
2092         diff_filespec_load_driver(one);
2093         return one->driver->word_regex;
2094 }
2095
2096 static void init_diff_words_data(struct emit_callback *ecbdata,
2097                                  struct diff_options *orig_opts,
2098                                  struct diff_filespec *one,
2099                                  struct diff_filespec *two)
2100 {
2101         int i;
2102         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2103         memcpy(o, orig_opts, sizeof(struct diff_options));
2104
2105         ecbdata->diff_words =
2106                 xcalloc(1, sizeof(struct diff_words_data));
2107         ecbdata->diff_words->type = o->word_diff;
2108         ecbdata->diff_words->opt = o;
2109
2110         if (orig_opts->emitted_symbols)
2111                 o->emitted_symbols =
2112                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2113
2114         if (!o->word_regex)
2115                 o->word_regex = userdiff_word_regex(one);
2116         if (!o->word_regex)
2117                 o->word_regex = userdiff_word_regex(two);
2118         if (!o->word_regex)
2119                 o->word_regex = diff_word_regex_cfg;
2120         if (o->word_regex) {
2121                 ecbdata->diff_words->word_regex = (regex_t *)
2122                         xmalloc(sizeof(regex_t));
2123                 if (regcomp(ecbdata->diff_words->word_regex,
2124                             o->word_regex,
2125                             REG_EXTENDED | REG_NEWLINE))
2126                         die ("Invalid regular expression: %s",
2127                              o->word_regex);
2128         }
2129         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2130                 if (o->word_diff == diff_words_styles[i].type) {
2131                         ecbdata->diff_words->style =
2132                                 &diff_words_styles[i];
2133                         break;
2134                 }
2135         }
2136         if (want_color(o->use_color)) {
2137                 struct diff_words_style *st = ecbdata->diff_words->style;
2138                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2139                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2140                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2141         }
2142 }
2143
2144 static void free_diff_words_data(struct emit_callback *ecbdata)
2145 {
2146         if (ecbdata->diff_words) {
2147                 diff_words_flush(ecbdata);
2148                 free (ecbdata->diff_words->opt->emitted_symbols);
2149                 free (ecbdata->diff_words->opt);
2150                 free (ecbdata->diff_words->minus.text.ptr);
2151                 free (ecbdata->diff_words->minus.orig);
2152                 free (ecbdata->diff_words->plus.text.ptr);
2153                 free (ecbdata->diff_words->plus.orig);
2154                 if (ecbdata->diff_words->word_regex) {
2155                         regfree(ecbdata->diff_words->word_regex);
2156                         free(ecbdata->diff_words->word_regex);
2157                 }
2158                 FREE_AND_NULL(ecbdata->diff_words);
2159         }
2160 }
2161
2162 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2163 {
2164         if (want_color(diff_use_color))
2165                 return diff_colors[ix];
2166         return "";
2167 }
2168
2169 const char *diff_line_prefix(struct diff_options *opt)
2170 {
2171         struct strbuf *msgbuf;
2172         if (!opt->output_prefix)
2173                 return "";
2174
2175         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2176         return msgbuf->buf;
2177 }
2178
2179 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2180 {
2181         const char *cp;
2182         unsigned long allot;
2183         size_t l = len;
2184
2185         cp = line;
2186         allot = l;
2187         while (0 < l) {
2188                 (void) utf8_width(&cp, &l);
2189                 if (!cp)
2190                         break; /* truncated in the middle? */
2191         }
2192         return allot - l;
2193 }
2194
2195 static void find_lno(const char *line, struct emit_callback *ecbdata)
2196 {
2197         const char *p;
2198         ecbdata->lno_in_preimage = 0;
2199         ecbdata->lno_in_postimage = 0;
2200         p = strchr(line, '-');
2201         if (!p)
2202                 return; /* cannot happen */
2203         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2204         p = strchr(p, '+');
2205         if (!p)
2206                 return; /* cannot happen */
2207         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2208 }
2209
2210 static void fn_out_consume(void *priv, char *line, unsigned long len)
2211 {
2212         struct emit_callback *ecbdata = priv;
2213         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2214         struct diff_options *o = ecbdata->opt;
2215
2216         o->found_changes = 1;
2217
2218         if (ecbdata->header) {
2219                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2220                                  ecbdata->header->buf, ecbdata->header->len, 0);
2221                 strbuf_reset(ecbdata->header);
2222                 ecbdata->header = NULL;
2223         }
2224
2225         if (ecbdata->label_path[0]) {
2226                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2227                                  ecbdata->label_path[0],
2228                                  strlen(ecbdata->label_path[0]), 0);
2229                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2230                                  ecbdata->label_path[1],
2231                                  strlen(ecbdata->label_path[1]), 0);
2232                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2233         }
2234
2235         if (diff_suppress_blank_empty
2236             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2237                 line[0] = '\n';
2238                 len = 1;
2239         }
2240
2241         if (line[0] == '@') {
2242                 if (ecbdata->diff_words)
2243                         diff_words_flush(ecbdata);
2244                 len = sane_truncate_line(ecbdata, line, len);
2245                 find_lno(line, ecbdata);
2246                 emit_hunk_header(ecbdata, line, len);
2247                 return;
2248         }
2249
2250         if (ecbdata->diff_words) {
2251                 enum diff_symbol s =
2252                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2253                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2254                 if (line[0] == '-') {
2255                         diff_words_append(line, len,
2256                                           &ecbdata->diff_words->minus);
2257                         return;
2258                 } else if (line[0] == '+') {
2259                         diff_words_append(line, len,
2260                                           &ecbdata->diff_words->plus);
2261                         return;
2262                 } else if (starts_with(line, "\\ ")) {
2263                         /*
2264                          * Eat the "no newline at eof" marker as if we
2265                          * saw a "+" or "-" line with nothing on it,
2266                          * and return without diff_words_flush() to
2267                          * defer processing. If this is the end of
2268                          * preimage, more "+" lines may come after it.
2269                          */
2270                         return;
2271                 }
2272                 diff_words_flush(ecbdata);
2273                 emit_diff_symbol(o, s, line, len, 0);
2274                 return;
2275         }
2276
2277         switch (line[0]) {
2278         case '+':
2279                 ecbdata->lno_in_postimage++;
2280                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2281                 break;
2282         case '-':
2283                 ecbdata->lno_in_preimage++;
2284                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2285                 break;
2286         case ' ':
2287                 ecbdata->lno_in_postimage++;
2288                 ecbdata->lno_in_preimage++;
2289                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2290                 break;
2291         default:
2292                 /* incomplete line at the end */
2293                 ecbdata->lno_in_preimage++;
2294                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2295                                  line, len, 0);
2296                 break;
2297         }
2298 }
2299
2300 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2301 {
2302         const char *old_name = a;
2303         const char *new_name = b;
2304         int pfx_length, sfx_length;
2305         int pfx_adjust_for_slash;
2306         int len_a = strlen(a);
2307         int len_b = strlen(b);
2308         int a_midlen, b_midlen;
2309         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2310         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2311
2312         if (qlen_a || qlen_b) {
2313                 quote_c_style(a, name, NULL, 0);
2314                 strbuf_addstr(name, " => ");
2315                 quote_c_style(b, name, NULL, 0);
2316                 return;
2317         }
2318
2319         /* Find common prefix */
2320         pfx_length = 0;
2321         while (*old_name && *new_name && *old_name == *new_name) {
2322                 if (*old_name == '/')
2323                         pfx_length = old_name - a + 1;
2324                 old_name++;
2325                 new_name++;
2326         }
2327
2328         /* Find common suffix */
2329         old_name = a + len_a;
2330         new_name = b + len_b;
2331         sfx_length = 0;
2332         /*
2333          * If there is a common prefix, it must end in a slash.  In
2334          * that case we let this loop run 1 into the prefix to see the
2335          * same slash.
2336          *
2337          * If there is no common prefix, we cannot do this as it would
2338          * underrun the input strings.
2339          */
2340         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2341         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2342                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2343                *old_name == *new_name) {
2344                 if (*old_name == '/')
2345                         sfx_length = len_a - (old_name - a);
2346                 old_name--;
2347                 new_name--;
2348         }
2349
2350         /*
2351          * pfx{mid-a => mid-b}sfx
2352          * {pfx-a => pfx-b}sfx
2353          * pfx{sfx-a => sfx-b}
2354          * name-a => name-b
2355          */
2356         a_midlen = len_a - pfx_length - sfx_length;
2357         b_midlen = len_b - pfx_length - sfx_length;
2358         if (a_midlen < 0)
2359                 a_midlen = 0;
2360         if (b_midlen < 0)
2361                 b_midlen = 0;
2362
2363         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2364         if (pfx_length + sfx_length) {
2365                 strbuf_add(name, a, pfx_length);
2366                 strbuf_addch(name, '{');
2367         }
2368         strbuf_add(name, a + pfx_length, a_midlen);
2369         strbuf_addstr(name, " => ");
2370         strbuf_add(name, b + pfx_length, b_midlen);
2371         if (pfx_length + sfx_length) {
2372                 strbuf_addch(name, '}');
2373                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2374         }
2375 }
2376
2377 struct diffstat_t {
2378         int nr;
2379         int alloc;
2380         struct diffstat_file {
2381                 char *from_name;
2382                 char *name;
2383                 char *print_name;
2384                 const char *comments;
2385                 unsigned is_unmerged:1;
2386                 unsigned is_binary:1;
2387                 unsigned is_renamed:1;
2388                 unsigned is_interesting:1;
2389                 uintmax_t added, deleted;
2390         } **files;
2391 };
2392
2393 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2394                                           const char *name_a,
2395                                           const char *name_b)
2396 {
2397         struct diffstat_file *x;
2398         x = xcalloc(1, sizeof(*x));
2399         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2400         diffstat->files[diffstat->nr++] = x;
2401         if (name_b) {
2402                 x->from_name = xstrdup(name_a);
2403                 x->name = xstrdup(name_b);
2404                 x->is_renamed = 1;
2405         }
2406         else {
2407                 x->from_name = NULL;
2408                 x->name = xstrdup(name_a);
2409         }
2410         return x;
2411 }
2412
2413 static void diffstat_consume(void *priv, char *line, unsigned long len)
2414 {
2415         struct diffstat_t *diffstat = priv;
2416         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2417
2418         if (line[0] == '+')
2419                 x->added++;
2420         else if (line[0] == '-')
2421                 x->deleted++;
2422 }
2423
2424 const char mime_boundary_leader[] = "------------";
2425
2426 static int scale_linear(int it, int width, int max_change)
2427 {
2428         if (!it)
2429                 return 0;
2430         /*
2431          * make sure that at least one '-' or '+' is printed if
2432          * there is any change to this path. The easiest way is to
2433          * scale linearly as if the alloted width is one column shorter
2434          * than it is, and then add 1 to the result.
2435          */
2436         return 1 + (it * (width - 1) / max_change);
2437 }
2438
2439 static void show_graph(struct strbuf *out, char ch, int cnt,
2440                        const char *set, const char *reset)
2441 {
2442         if (cnt <= 0)
2443                 return;
2444         strbuf_addstr(out, set);
2445         strbuf_addchars(out, ch, cnt);
2446         strbuf_addstr(out, reset);
2447 }
2448
2449 static void fill_print_name(struct diffstat_file *file)
2450 {
2451         struct strbuf pname = STRBUF_INIT;
2452
2453         if (file->print_name)
2454                 return;
2455
2456         if (file->is_renamed)
2457                 pprint_rename(&pname, file->from_name, file->name);
2458         else
2459                 quote_c_style(file->name, &pname, NULL, 0);
2460
2461         if (file->comments)
2462                 strbuf_addf(&pname, " (%s)", file->comments);
2463
2464         file->print_name = strbuf_detach(&pname, NULL);
2465 }
2466
2467 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2468                 int files, int insertions, int deletions)
2469 {
2470         struct strbuf sb = STRBUF_INIT;
2471
2472         if (!files) {
2473                 assert(insertions == 0 && deletions == 0);
2474                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2475                                  NULL, 0, 0);
2476                 return;
2477         }
2478
2479         strbuf_addf(&sb,
2480                     (files == 1) ? " %d file changed" : " %d files changed",
2481                     files);
2482
2483         /*
2484          * For binary diff, the caller may want to print "x files
2485          * changed" with insertions == 0 && deletions == 0.
2486          *
2487          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2488          * is probably less confusing (i.e skip over "2 files changed
2489          * but nothing about added/removed lines? Is this a bug in Git?").
2490          */
2491         if (insertions || deletions == 0) {
2492                 strbuf_addf(&sb,
2493                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2494                             insertions);
2495         }
2496
2497         if (deletions || insertions == 0) {
2498                 strbuf_addf(&sb,
2499                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2500                             deletions);
2501         }
2502         strbuf_addch(&sb, '\n');
2503         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2504                          sb.buf, sb.len, 0);
2505         strbuf_release(&sb);
2506 }
2507
2508 void print_stat_summary(FILE *fp, int files,
2509                         int insertions, int deletions)
2510 {
2511         struct diff_options o;
2512         memset(&o, 0, sizeof(o));
2513         o.file = fp;
2514
2515         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2516 }
2517
2518 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2519 {
2520         int i, len, add, del, adds = 0, dels = 0;
2521         uintmax_t max_change = 0, max_len = 0;
2522         int total_files = data->nr, count;
2523         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2524         const char *reset, *add_c, *del_c;
2525         int extra_shown = 0;
2526         const char *line_prefix = diff_line_prefix(options);
2527         struct strbuf out = STRBUF_INIT;
2528
2529         if (data->nr == 0)
2530                 return;
2531
2532         count = options->stat_count ? options->stat_count : data->nr;
2533
2534         reset = diff_get_color_opt(options, DIFF_RESET);
2535         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2536         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2537
2538         /*
2539          * Find the longest filename and max number of changes
2540          */
2541         for (i = 0; (i < count) && (i < data->nr); i++) {
2542                 struct diffstat_file *file = data->files[i];
2543                 uintmax_t change = file->added + file->deleted;
2544
2545                 if (!file->is_interesting && (change == 0)) {
2546                         count++; /* not shown == room for one more */
2547                         continue;
2548                 }
2549                 fill_print_name(file);
2550                 len = strlen(file->print_name);
2551                 if (max_len < len)
2552                         max_len = len;
2553
2554                 if (file->is_unmerged) {
2555                         /* "Unmerged" is 8 characters */
2556                         bin_width = bin_width < 8 ? 8 : bin_width;
2557                         continue;
2558                 }
2559                 if (file->is_binary) {
2560                         /* "Bin XXX -> YYY bytes" */
2561                         int w = 14 + decimal_width(file->added)
2562                                 + decimal_width(file->deleted);
2563                         bin_width = bin_width < w ? w : bin_width;
2564                         /* Display change counts aligned with "Bin" */
2565                         number_width = 3;
2566                         continue;
2567                 }
2568
2569                 if (max_change < change)
2570                         max_change = change;
2571         }
2572         count = i; /* where we can stop scanning in data->files[] */
2573
2574         /*
2575          * We have width = stat_width or term_columns() columns total.
2576          * We want a maximum of min(max_len, stat_name_width) for the name part.
2577          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2578          * We also need 1 for " " and 4 + decimal_width(max_change)
2579          * for " | NNNN " and one the empty column at the end, altogether
2580          * 6 + decimal_width(max_change).
2581          *
2582          * If there's not enough space, we will use the smaller of
2583          * stat_name_width (if set) and 5/8*width for the filename,
2584          * and the rest for constant elements + graph part, but no more
2585          * than stat_graph_width for the graph part.
2586          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2587          * for the standard terminal size).
2588          *
2589          * In other words: stat_width limits the maximum width, and
2590          * stat_name_width fixes the maximum width of the filename,
2591          * and is also used to divide available columns if there
2592          * aren't enough.
2593          *
2594          * Binary files are displayed with "Bin XXX -> YYY bytes"
2595          * instead of the change count and graph. This part is treated
2596          * similarly to the graph part, except that it is not
2597          * "scaled". If total width is too small to accommodate the
2598          * guaranteed minimum width of the filename part and the
2599          * separators and this message, this message will "overflow"
2600          * making the line longer than the maximum width.
2601          */
2602
2603         if (options->stat_width == -1)
2604                 width = term_columns() - strlen(line_prefix);
2605         else
2606                 width = options->stat_width ? options->stat_width : 80;
2607         number_width = decimal_width(max_change) > number_width ?
2608                 decimal_width(max_change) : number_width;
2609
2610         if (options->stat_graph_width == -1)
2611                 options->stat_graph_width = diff_stat_graph_width;
2612
2613         /*
2614          * Guarantee 3/8*16==6 for the graph part
2615          * and 5/8*16==10 for the filename part
2616          */
2617         if (width < 16 + 6 + number_width)
2618                 width = 16 + 6 + number_width;
2619
2620         /*
2621          * First assign sizes that are wanted, ignoring available width.
2622          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2623          * starting from "XXX" should fit in graph_width.
2624          */
2625         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2626         if (options->stat_graph_width &&
2627             options->stat_graph_width < graph_width)
2628                 graph_width = options->stat_graph_width;
2629
2630         name_width = (options->stat_name_width > 0 &&
2631                       options->stat_name_width < max_len) ?
2632                 options->stat_name_width : max_len;
2633
2634         /*
2635          * Adjust adjustable widths not to exceed maximum width
2636          */
2637         if (name_width + number_width + 6 + graph_width > width) {
2638                 if (graph_width > width * 3/8 - number_width - 6) {
2639                         graph_width = width * 3/8 - number_width - 6;
2640                         if (graph_width < 6)
2641                                 graph_width = 6;
2642                 }
2643
2644                 if (options->stat_graph_width &&
2645                     graph_width > options->stat_graph_width)
2646                         graph_width = options->stat_graph_width;
2647                 if (name_width > width - number_width - 6 - graph_width)
2648                         name_width = width - number_width - 6 - graph_width;
2649                 else
2650                         graph_width = width - number_width - 6 - name_width;
2651         }
2652
2653         /*
2654          * From here name_width is the width of the name area,
2655          * and graph_width is the width of the graph area.
2656          * max_change is used to scale graph properly.
2657          */
2658         for (i = 0; i < count; i++) {
2659                 const char *prefix = "";
2660                 struct diffstat_file *file = data->files[i];
2661                 char *name = file->print_name;
2662                 uintmax_t added = file->added;
2663                 uintmax_t deleted = file->deleted;
2664                 int name_len;
2665
2666                 if (!file->is_interesting && (added + deleted == 0))
2667                         continue;
2668
2669                 /*
2670                  * "scale" the filename
2671                  */
2672                 len = name_width;
2673                 name_len = strlen(name);
2674                 if (name_width < name_len) {
2675                         char *slash;
2676                         prefix = "...";
2677                         len -= 3;
2678                         name += name_len - len;
2679                         slash = strchr(name, '/');
2680                         if (slash)
2681                                 name = slash;
2682                 }
2683
2684                 if (file->is_binary) {
2685                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2686                         strbuf_addf(&out, " %*s", number_width, "Bin");
2687                         if (!added && !deleted) {
2688                                 strbuf_addch(&out, '\n');
2689                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2690                                                  out.buf, out.len, 0);
2691                                 strbuf_reset(&out);
2692                                 continue;
2693                         }
2694                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2695                                 del_c, deleted, reset);
2696                         strbuf_addstr(&out, " -> ");
2697                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2698                                 add_c, added, reset);
2699                         strbuf_addstr(&out, " bytes\n");
2700                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2701                                          out.buf, out.len, 0);
2702                         strbuf_reset(&out);
2703                         continue;
2704                 }
2705                 else if (file->is_unmerged) {
2706                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2707                         strbuf_addstr(&out, " Unmerged\n");
2708                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2709                                          out.buf, out.len, 0);
2710                         strbuf_reset(&out);
2711                         continue;
2712                 }
2713
2714                 /*
2715                  * scale the add/delete
2716                  */
2717                 add = added;
2718                 del = deleted;
2719
2720                 if (graph_width <= max_change) {
2721                         int total = scale_linear(add + del, graph_width, max_change);
2722                         if (total < 2 && add && del)
2723                                 /* width >= 2 due to the sanity check */
2724                                 total = 2;
2725                         if (add < del) {
2726                                 add = scale_linear(add, graph_width, max_change);
2727                                 del = total - add;
2728                         } else {
2729                                 del = scale_linear(del, graph_width, max_change);
2730                                 add = total - del;
2731                         }
2732                 }
2733                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2734                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2735                         number_width, added + deleted,
2736                         added + deleted ? " " : "");
2737                 show_graph(&out, '+', add, add_c, reset);
2738                 show_graph(&out, '-', del, del_c, reset);
2739                 strbuf_addch(&out, '\n');
2740                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2741                                  out.buf, out.len, 0);
2742                 strbuf_reset(&out);
2743         }
2744
2745         for (i = 0; i < data->nr; i++) {
2746                 struct diffstat_file *file = data->files[i];
2747                 uintmax_t added = file->added;
2748                 uintmax_t deleted = file->deleted;
2749
2750                 if (file->is_unmerged ||
2751                     (!file->is_interesting && (added + deleted == 0))) {
2752                         total_files--;
2753                         continue;
2754                 }
2755
2756                 if (!file->is_binary) {
2757                         adds += added;
2758                         dels += deleted;
2759                 }
2760                 if (i < count)
2761                         continue;
2762                 if (!extra_shown)
2763                         emit_diff_symbol(options,
2764                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2765                                          NULL, 0, 0);
2766                 extra_shown = 1;
2767         }
2768
2769         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2770         strbuf_release(&out);
2771 }
2772
2773 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2774 {
2775         int i, adds = 0, dels = 0, total_files = data->nr;
2776
2777         if (data->nr == 0)
2778                 return;
2779
2780         for (i = 0; i < data->nr; i++) {
2781                 int added = data->files[i]->added;
2782                 int deleted = data->files[i]->deleted;
2783
2784                 if (data->files[i]->is_unmerged ||
2785                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2786                         total_files--;
2787                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2788                         adds += added;
2789                         dels += deleted;
2790                 }
2791         }
2792         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2793 }
2794
2795 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2796 {
2797         int i;
2798
2799         if (data->nr == 0)
2800                 return;
2801
2802         for (i = 0; i < data->nr; i++) {
2803                 struct diffstat_file *file = data->files[i];
2804
2805                 fprintf(options->file, "%s", diff_line_prefix(options));
2806
2807                 if (file->is_binary)
2808                         fprintf(options->file, "-\t-\t");
2809                 else
2810                         fprintf(options->file,
2811                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2812                                 file->added, file->deleted);
2813                 if (options->line_termination) {
2814                         fill_print_name(file);
2815                         if (!file->is_renamed)
2816                                 write_name_quoted(file->name, options->file,
2817                                                   options->line_termination);
2818                         else {
2819                                 fputs(file->print_name, options->file);
2820                                 putc(options->line_termination, options->file);
2821                         }
2822                 } else {
2823                         if (file->is_renamed) {
2824                                 putc('\0', options->file);
2825                                 write_name_quoted(file->from_name, options->file, '\0');
2826                         }
2827                         write_name_quoted(file->name, options->file, '\0');
2828                 }
2829         }
2830 }
2831
2832 struct dirstat_file {
2833         const char *name;
2834         unsigned long changed;
2835 };
2836
2837 struct dirstat_dir {
2838         struct dirstat_file *files;
2839         int alloc, nr, permille, cumulative;
2840 };
2841
2842 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2843                 unsigned long changed, const char *base, int baselen)
2844 {
2845         unsigned long sum_changes = 0;
2846         unsigned int sources = 0;
2847         const char *line_prefix = diff_line_prefix(opt);
2848
2849         while (dir->nr) {
2850                 struct dirstat_file *f = dir->files;
2851                 int namelen = strlen(f->name);
2852                 unsigned long changes;
2853                 char *slash;
2854
2855                 if (namelen < baselen)
2856                         break;
2857                 if (memcmp(f->name, base, baselen))
2858                         break;
2859                 slash = strchr(f->name + baselen, '/');
2860                 if (slash) {
2861                         int newbaselen = slash + 1 - f->name;
2862                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2863                         sources++;
2864                 } else {
2865                         changes = f->changed;
2866                         dir->files++;
2867                         dir->nr--;
2868                         sources += 2;
2869                 }
2870                 sum_changes += changes;
2871         }
2872
2873         /*
2874          * We don't report dirstat's for
2875          *  - the top level
2876          *  - or cases where everything came from a single directory
2877          *    under this directory (sources == 1).
2878          */
2879         if (baselen && sources != 1) {
2880                 if (sum_changes) {
2881                         int permille = sum_changes * 1000 / changed;
2882                         if (permille >= dir->permille) {
2883                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2884                                         permille / 10, permille % 10, baselen, base);
2885                                 if (!dir->cumulative)
2886                                         return 0;
2887                         }
2888                 }
2889         }
2890         return sum_changes;
2891 }
2892
2893 static int dirstat_compare(const void *_a, const void *_b)
2894 {
2895         const struct dirstat_file *a = _a;
2896         const struct dirstat_file *b = _b;
2897         return strcmp(a->name, b->name);
2898 }
2899
2900 static void show_dirstat(struct diff_options *options)
2901 {
2902         int i;
2903         unsigned long changed;
2904         struct dirstat_dir dir;
2905         struct diff_queue_struct *q = &diff_queued_diff;
2906
2907         dir.files = NULL;
2908         dir.alloc = 0;
2909         dir.nr = 0;
2910         dir.permille = options->dirstat_permille;
2911         dir.cumulative = options->flags.dirstat_cumulative;
2912
2913         changed = 0;
2914         for (i = 0; i < q->nr; i++) {
2915                 struct diff_filepair *p = q->queue[i];
2916                 const char *name;
2917                 unsigned long copied, added, damage;
2918                 int content_changed;
2919
2920                 name = p->two->path ? p->two->path : p->one->path;
2921
2922                 if (p->one->oid_valid && p->two->oid_valid)
2923                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2924                 else
2925                         content_changed = 1;
2926
2927                 if (!content_changed) {
2928                         /*
2929                          * The SHA1 has not changed, so pre-/post-content is
2930                          * identical. We can therefore skip looking at the
2931                          * file contents altogether.
2932                          */
2933                         damage = 0;
2934                         goto found_damage;
2935                 }
2936
2937                 if (options->flags.dirstat_by_file) {
2938                         /*
2939                          * In --dirstat-by-file mode, we don't really need to
2940                          * look at the actual file contents at all.
2941                          * The fact that the SHA1 changed is enough for us to
2942                          * add this file to the list of results
2943                          * (with each file contributing equal damage).
2944                          */
2945                         damage = 1;
2946                         goto found_damage;
2947                 }
2948
2949                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2950                         diff_populate_filespec(p->one, 0);
2951                         diff_populate_filespec(p->two, 0);
2952                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2953                                                &copied, &added);
2954                         diff_free_filespec_data(p->one);
2955                         diff_free_filespec_data(p->two);
2956                 } else if (DIFF_FILE_VALID(p->one)) {
2957                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2958                         copied = added = 0;
2959                         diff_free_filespec_data(p->one);
2960                 } else if (DIFF_FILE_VALID(p->two)) {
2961                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2962                         copied = 0;
2963                         added = p->two->size;
2964                         diff_free_filespec_data(p->two);
2965                 } else
2966                         continue;
2967
2968                 /*
2969                  * Original minus copied is the removed material,
2970                  * added is the new material.  They are both damages
2971                  * made to the preimage.
2972                  * If the resulting damage is zero, we know that
2973                  * diffcore_count_changes() considers the two entries to
2974                  * be identical, but since content_changed is true, we
2975                  * know that there must have been _some_ kind of change,
2976                  * so we force all entries to have damage > 0.
2977                  */
2978                 damage = (p->one->size - copied) + added;
2979                 if (!damage)
2980                         damage = 1;
2981
2982 found_damage:
2983                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2984                 dir.files[dir.nr].name = name;
2985                 dir.files[dir.nr].changed = damage;
2986                 changed += damage;
2987                 dir.nr++;
2988         }
2989
2990         /* This can happen even with many files, if everything was renames */
2991         if (!changed)
2992                 return;
2993
2994         /* Show all directories with more than x% of the changes */
2995         QSORT(dir.files, dir.nr, dirstat_compare);
2996         gather_dirstat(options, &dir, changed, "", 0);
2997 }
2998
2999 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3000 {
3001         int i;
3002         unsigned long changed;
3003         struct dirstat_dir dir;
3004
3005         if (data->nr == 0)
3006                 return;
3007
3008         dir.files = NULL;
3009         dir.alloc = 0;
3010         dir.nr = 0;
3011         dir.permille = options->dirstat_permille;
3012         dir.cumulative = options->flags.dirstat_cumulative;
3013
3014         changed = 0;
3015         for (i = 0; i < data->nr; i++) {
3016                 struct diffstat_file *file = data->files[i];
3017                 unsigned long damage = file->added + file->deleted;
3018                 if (file->is_binary)
3019                         /*
3020                          * binary files counts bytes, not lines. Must find some
3021                          * way to normalize binary bytes vs. textual lines.
3022                          * The following heuristic assumes that there are 64
3023                          * bytes per "line".
3024                          * This is stupid and ugly, but very cheap...
3025                          */
3026                         damage = DIV_ROUND_UP(damage, 64);
3027                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3028                 dir.files[dir.nr].name = file->name;
3029                 dir.files[dir.nr].changed = damage;
3030                 changed += damage;
3031                 dir.nr++;
3032         }
3033
3034         /* This can happen even with many files, if everything was renames */
3035         if (!changed)
3036                 return;
3037
3038         /* Show all directories with more than x% of the changes */
3039         QSORT(dir.files, dir.nr, dirstat_compare);
3040         gather_dirstat(options, &dir, changed, "", 0);
3041 }
3042
3043 static void free_diffstat_info(struct diffstat_t *diffstat)
3044 {
3045         int i;
3046         for (i = 0; i < diffstat->nr; i++) {
3047                 struct diffstat_file *f = diffstat->files[i];
3048                 free(f->print_name);
3049                 free(f->name);
3050                 free(f->from_name);
3051                 free(f);
3052         }
3053         free(diffstat->files);
3054 }
3055
3056 struct checkdiff_t {
3057         const char *filename;
3058         int lineno;
3059         int conflict_marker_size;
3060         struct diff_options *o;
3061         unsigned ws_rule;
3062         unsigned status;
3063 };
3064
3065 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3066 {
3067         char firstchar;
3068         int cnt;
3069
3070         if (len < marker_size + 1)
3071                 return 0;
3072         firstchar = line[0];
3073         switch (firstchar) {
3074         case '=': case '>': case '<': case '|':
3075                 break;
3076         default:
3077                 return 0;
3078         }
3079         for (cnt = 1; cnt < marker_size; cnt++)
3080                 if (line[cnt] != firstchar)
3081                         return 0;
3082         /* line[1] thru line[marker_size-1] are same as firstchar */
3083         if (len < marker_size + 1 || !isspace(line[marker_size]))
3084                 return 0;
3085         return 1;
3086 }
3087
3088 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3089 {
3090         struct checkdiff_t *data = priv;
3091         int marker_size = data->conflict_marker_size;
3092         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3093         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3094         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3095         char *err;
3096         const char *line_prefix;
3097
3098         assert(data->o);
3099         line_prefix = diff_line_prefix(data->o);
3100
3101         if (line[0] == '+') {
3102                 unsigned bad;
3103                 data->lineno++;
3104                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3105                         data->status |= 1;
3106                         fprintf(data->o->file,
3107                                 "%s%s:%d: leftover conflict marker\n",
3108                                 line_prefix, data->filename, data->lineno);
3109                 }
3110                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3111                 if (!bad)
3112                         return;
3113                 data->status |= bad;
3114                 err = whitespace_error_string(bad);
3115                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3116                         line_prefix, data->filename, data->lineno, err);
3117                 free(err);
3118                 emit_line(data->o, set, reset, line, 1);
3119                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3120                               data->o->file, set, reset, ws);
3121         } else if (line[0] == ' ') {
3122                 data->lineno++;
3123         } else if (line[0] == '@') {
3124                 char *plus = strchr(line, '+');
3125                 if (plus)
3126                         data->lineno = strtol(plus, NULL, 10) - 1;
3127                 else
3128                         die("invalid diff");
3129         }
3130 }
3131
3132 static unsigned char *deflate_it(char *data,
3133                                  unsigned long size,
3134                                  unsigned long *result_size)
3135 {
3136         int bound;
3137         unsigned char *deflated;
3138         git_zstream stream;
3139
3140         git_deflate_init(&stream, zlib_compression_level);
3141         bound = git_deflate_bound(&stream, size);
3142         deflated = xmalloc(bound);
3143         stream.next_out = deflated;
3144         stream.avail_out = bound;
3145
3146         stream.next_in = (unsigned char *)data;
3147         stream.avail_in = size;
3148         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3149                 ; /* nothing */
3150         git_deflate_end(&stream);
3151         *result_size = stream.total_out;
3152         return deflated;
3153 }
3154
3155 static void emit_binary_diff_body(struct diff_options *o,
3156                                   mmfile_t *one, mmfile_t *two)
3157 {
3158         void *cp;
3159         void *delta;
3160         void *deflated;
3161         void *data;
3162         unsigned long orig_size;
3163         unsigned long delta_size;
3164         unsigned long deflate_size;
3165         unsigned long data_size;
3166
3167         /* We could do deflated delta, or we could do just deflated two,
3168          * whichever is smaller.
3169          */
3170         delta = NULL;
3171         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3172         if (one->size && two->size) {
3173                 delta = diff_delta(one->ptr, one->size,
3174                                    two->ptr, two->size,
3175                                    &delta_size, deflate_size);
3176                 if (delta) {
3177                         void *to_free = delta;
3178                         orig_size = delta_size;
3179                         delta = deflate_it(delta, delta_size, &delta_size);
3180                         free(to_free);
3181                 }
3182         }
3183
3184         if (delta && delta_size < deflate_size) {
3185                 char *s = xstrfmt("%lu", orig_size);
3186                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3187                                  s, strlen(s), 0);
3188                 free(s);
3189                 free(deflated);
3190                 data = delta;
3191                 data_size = delta_size;
3192         } else {
3193                 char *s = xstrfmt("%lu", two->size);
3194                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3195                                  s, strlen(s), 0);
3196                 free(s);
3197                 free(delta);
3198                 data = deflated;
3199                 data_size = deflate_size;
3200         }
3201
3202         /* emit data encoded in base85 */
3203         cp = data;
3204         while (data_size) {
3205                 int len;
3206                 int bytes = (52 < data_size) ? 52 : data_size;
3207                 char line[71];
3208                 data_size -= bytes;
3209                 if (bytes <= 26)
3210                         line[0] = bytes + 'A' - 1;
3211                 else
3212                         line[0] = bytes - 26 + 'a' - 1;
3213                 encode_85(line + 1, cp, bytes);
3214                 cp = (char *) cp + bytes;
3215
3216                 len = strlen(line);
3217                 line[len++] = '\n';
3218                 line[len] = '\0';
3219
3220                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3221                                  line, len, 0);
3222         }
3223         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3224         free(data);
3225 }
3226
3227 static void emit_binary_diff(struct diff_options *o,
3228                              mmfile_t *one, mmfile_t *two)
3229 {
3230         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3231         emit_binary_diff_body(o, one, two);
3232         emit_binary_diff_body(o, two, one);
3233 }
3234
3235 int diff_filespec_is_binary(struct diff_filespec *one)
3236 {
3237         if (one->is_binary == -1) {
3238                 diff_filespec_load_driver(one);
3239                 if (one->driver->binary != -1)
3240                         one->is_binary = one->driver->binary;
3241                 else {
3242                         if (!one->data && DIFF_FILE_VALID(one))
3243                                 diff_populate_filespec(one, CHECK_BINARY);
3244                         if (one->is_binary == -1 && one->data)
3245                                 one->is_binary = buffer_is_binary(one->data,
3246                                                 one->size);
3247                         if (one->is_binary == -1)
3248                                 one->is_binary = 0;
3249                 }
3250         }
3251         return one->is_binary;
3252 }
3253
3254 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3255 {
3256         diff_filespec_load_driver(one);
3257         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3258 }
3259
3260 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3261 {
3262         if (!options->a_prefix)
3263                 options->a_prefix = a;
3264         if (!options->b_prefix)
3265                 options->b_prefix = b;
3266 }
3267
3268 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3269 {
3270         if (!DIFF_FILE_VALID(one))
3271                 return NULL;
3272
3273         diff_filespec_load_driver(one);
3274         return userdiff_get_textconv(one->driver);
3275 }
3276
3277 static void builtin_diff(const char *name_a,
3278                          const char *name_b,
3279                          struct diff_filespec *one,
3280                          struct diff_filespec *two,
3281                          const char *xfrm_msg,
3282                          int must_show_header,
3283                          struct diff_options *o,
3284                          int complete_rewrite)
3285 {
3286         mmfile_t mf1, mf2;
3287         const char *lbl[2];
3288         char *a_one, *b_two;
3289         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3290         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3291         const char *a_prefix, *b_prefix;
3292         struct userdiff_driver *textconv_one = NULL;
3293         struct userdiff_driver *textconv_two = NULL;
3294         struct strbuf header = STRBUF_INIT;
3295         const char *line_prefix = diff_line_prefix(o);
3296
3297         diff_set_mnemonic_prefix(o, "a/", "b/");
3298         if (o->flags.reverse_diff) {
3299                 a_prefix = o->b_prefix;
3300                 b_prefix = o->a_prefix;
3301         } else {
3302                 a_prefix = o->a_prefix;
3303                 b_prefix = o->b_prefix;
3304         }
3305
3306         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3307             (!one->mode || S_ISGITLINK(one->mode)) &&
3308             (!two->mode || S_ISGITLINK(two->mode))) {
3309                 show_submodule_summary(o, one->path ? one->path : two->path,
3310                                 &one->oid, &two->oid,
3311                                 two->dirty_submodule);
3312                 return;
3313         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3314                    (!one->mode || S_ISGITLINK(one->mode)) &&
3315                    (!two->mode || S_ISGITLINK(two->mode))) {
3316                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3317                                 &one->oid, &two->oid,
3318                                 two->dirty_submodule);
3319                 return;
3320         }
3321
3322         if (o->flags.allow_textconv) {
3323                 textconv_one = get_textconv(one);
3324                 textconv_two = get_textconv(two);
3325         }
3326
3327         /* Never use a non-valid filename anywhere if at all possible */
3328         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3329         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3330
3331         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3332         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3333         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3334         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3335         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3336         if (lbl[0][0] == '/') {
3337                 /* /dev/null */
3338                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3339                 if (xfrm_msg)
3340                         strbuf_addstr(&header, xfrm_msg);
3341                 must_show_header = 1;
3342         }
3343         else if (lbl[1][0] == '/') {
3344                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3345                 if (xfrm_msg)
3346                         strbuf_addstr(&header, xfrm_msg);
3347                 must_show_header = 1;
3348         }
3349         else {
3350                 if (one->mode != two->mode) {
3351                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3352                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3353                         must_show_header = 1;
3354                 }
3355                 if (xfrm_msg)
3356                         strbuf_addstr(&header, xfrm_msg);
3357
3358                 /*
3359                  * we do not run diff between different kind
3360                  * of objects.
3361                  */
3362                 if ((one->mode ^ two->mode) & S_IFMT)
3363                         goto free_ab_and_return;
3364                 if (complete_rewrite &&
3365                     (textconv_one || !diff_filespec_is_binary(one)) &&
3366                     (textconv_two || !diff_filespec_is_binary(two))) {
3367                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3368                                          header.buf, header.len, 0);
3369                         strbuf_reset(&header);
3370                         emit_rewrite_diff(name_a, name_b, one, two,
3371                                                 textconv_one, textconv_two, o);
3372                         o->found_changes = 1;
3373                         goto free_ab_and_return;
3374                 }
3375         }
3376
3377         if (o->irreversible_delete && lbl[1][0] == '/') {
3378                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3379                                  header.len, 0);
3380                 strbuf_reset(&header);
3381                 goto free_ab_and_return;
3382         } else if (!o->flags.text &&
3383             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3384               (!textconv_two && diff_filespec_is_binary(two)) )) {
3385                 struct strbuf sb = STRBUF_INIT;
3386                 if (!one->data && !two->data &&
3387                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3388                     !o->flags.binary) {
3389                         if (!oidcmp(&one->oid, &two->oid)) {
3390                                 if (must_show_header)
3391                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3392                                                          header.buf, header.len,
3393                                                          0);
3394                                 goto free_ab_and_return;
3395                         }
3396                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3397                                          header.buf, header.len, 0);
3398                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3399                                     diff_line_prefix(o), lbl[0], lbl[1]);
3400                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3401                                          sb.buf, sb.len, 0);
3402                         strbuf_release(&sb);
3403                         goto free_ab_and_return;
3404                 }
3405                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3406                         die("unable to read files to diff");
3407                 /* Quite common confusing case */
3408                 if (mf1.size == mf2.size &&
3409                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3410                         if (must_show_header)
3411                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3412                                                  header.buf, header.len, 0);
3413                         goto free_ab_and_return;
3414                 }
3415                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3416                 strbuf_reset(&header);
3417                 if (o->flags.binary)
3418                         emit_binary_diff(o, &mf1, &mf2);
3419                 else {
3420                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3421                                     diff_line_prefix(o), lbl[0], lbl[1]);
3422                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3423                                          sb.buf, sb.len, 0);
3424                         strbuf_release(&sb);
3425                 }
3426                 o->found_changes = 1;
3427         } else {
3428                 /* Crazy xdl interfaces.. */
3429                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3430                 const char *v;
3431                 xpparam_t xpp;
3432                 xdemitconf_t xecfg;
3433                 struct emit_callback ecbdata;
3434                 const struct userdiff_funcname *pe;
3435
3436                 if (must_show_header) {
3437                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3438                                          header.buf, header.len, 0);
3439                         strbuf_reset(&header);
3440                 }
3441
3442                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3443                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3444
3445                 pe = diff_funcname_pattern(one);
3446                 if (!pe)
3447                         pe = diff_funcname_pattern(two);
3448
3449                 memset(&xpp, 0, sizeof(xpp));
3450                 memset(&xecfg, 0, sizeof(xecfg));
3451                 memset(&ecbdata, 0, sizeof(ecbdata));
3452                 if (o->flags.suppress_diff_headers)
3453                         lbl[0] = NULL;
3454                 ecbdata.label_path = lbl;
3455                 ecbdata.color_diff = want_color(o->use_color);
3456                 ecbdata.ws_rule = whitespace_rule(name_b);
3457                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3458                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3459                 ecbdata.opt = o;
3460                 if (header.len && !o->flags.suppress_diff_headers)
3461                         ecbdata.header = &header;
3462                 xpp.flags = o->xdl_opts;
3463                 xpp.anchors = o->anchors;
3464                 xpp.anchors_nr = o->anchors_nr;
3465                 xecfg.ctxlen = o->context;
3466                 xecfg.interhunkctxlen = o->interhunkcontext;
3467                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3468                 if (o->flags.funccontext)
3469                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3470                 if (pe)
3471                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3472                 if (!diffopts)
3473                         ;
3474                 else if (skip_prefix(diffopts, "--unified=", &v))
3475                         xecfg.ctxlen = strtoul(v, NULL, 10);
3476                 else if (skip_prefix(diffopts, "-u", &v))
3477                         xecfg.ctxlen = strtoul(v, NULL, 10);
3478                 if (o->word_diff)
3479                         init_diff_words_data(&ecbdata, o, one, two);
3480                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3481                                   &xpp, &xecfg))
3482                         die("unable to generate diff for %s", one->path);
3483                 if (o->word_diff)
3484                         free_diff_words_data(&ecbdata);
3485                 if (textconv_one)
3486                         free(mf1.ptr);
3487                 if (textconv_two)
3488                         free(mf2.ptr);
3489                 xdiff_clear_find_func(&xecfg);
3490         }
3491
3492  free_ab_and_return:
3493         strbuf_release(&header);
3494         diff_free_filespec_data(one);
3495         diff_free_filespec_data(two);
3496         free(a_one);
3497         free(b_two);
3498         return;
3499 }
3500
3501 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3502 {
3503         if (!is_renamed) {
3504                 if (p->status == DIFF_STATUS_ADDED) {
3505                         if (S_ISLNK(p->two->mode))
3506                                 return "new +l";
3507                         else if ((p->two->mode & 0777) == 0755)
3508                                 return "new +x";
3509                         else
3510                                 return "new";
3511                 } else if (p->status == DIFF_STATUS_DELETED)
3512                         return "gone";
3513         }
3514         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3515                 return "mode -l";
3516         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3517                 return "mode +l";
3518         else if ((p->one->mode & 0777) == 0644 &&
3519                  (p->two->mode & 0777) == 0755)
3520                 return "mode +x";
3521         else if ((p->one->mode & 0777) == 0755 &&
3522                  (p->two->mode & 0777) == 0644)
3523                 return "mode -x";
3524         return NULL;
3525 }
3526
3527 static void builtin_diffstat(const char *name_a, const char *name_b,
3528                              struct diff_filespec *one,
3529                              struct diff_filespec *two,
3530                              struct diffstat_t *diffstat,
3531                              struct diff_options *o,
3532                              struct diff_filepair *p)
3533 {
3534         mmfile_t mf1, mf2;
3535         struct diffstat_file *data;
3536         int same_contents;
3537         int complete_rewrite = 0;
3538
3539         if (!DIFF_PAIR_UNMERGED(p)) {
3540                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3541                         complete_rewrite = 1;
3542         }
3543
3544         data = diffstat_add(diffstat, name_a, name_b);
3545         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3546         if (o->flags.stat_with_summary)
3547                 data->comments = get_compact_summary(p, data->is_renamed);
3548
3549         if (!one || !two) {
3550                 data->is_unmerged = 1;
3551                 return;
3552         }
3553
3554         same_contents = !oidcmp(&one->oid, &two->oid);
3555
3556         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3557                 data->is_binary = 1;
3558                 if (same_contents) {
3559                         data->added = 0;
3560                         data->deleted = 0;
3561                 } else {
3562                         data->added = diff_filespec_size(two);
3563                         data->deleted = diff_filespec_size(one);
3564                 }
3565         }
3566
3567         else if (complete_rewrite) {
3568                 diff_populate_filespec(one, 0);
3569                 diff_populate_filespec(two, 0);
3570                 data->deleted = count_lines(one->data, one->size);
3571                 data->added = count_lines(two->data, two->size);
3572         }
3573
3574         else if (!same_contents) {
3575                 /* Crazy xdl interfaces.. */
3576                 xpparam_t xpp;
3577                 xdemitconf_t xecfg;
3578
3579                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3580                         die("unable to read files to diff");
3581
3582                 memset(&xpp, 0, sizeof(xpp));
3583                 memset(&xecfg, 0, sizeof(xecfg));
3584                 xpp.flags = o->xdl_opts;
3585                 xpp.anchors = o->anchors;
3586                 xpp.anchors_nr = o->anchors_nr;
3587                 xecfg.ctxlen = o->context;
3588                 xecfg.interhunkctxlen = o->interhunkcontext;
3589                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3590                                   &xpp, &xecfg))
3591                         die("unable to generate diffstat for %s", one->path);
3592         }
3593
3594         diff_free_filespec_data(one);
3595         diff_free_filespec_data(two);
3596 }
3597
3598 static void builtin_checkdiff(const char *name_a, const char *name_b,
3599                               const char *attr_path,
3600                               struct diff_filespec *one,
3601                               struct diff_filespec *two,
3602                               struct diff_options *o)
3603 {
3604         mmfile_t mf1, mf2;
3605         struct checkdiff_t data;
3606
3607         if (!two)
3608                 return;
3609
3610         memset(&data, 0, sizeof(data));
3611         data.filename = name_b ? name_b : name_a;
3612         data.lineno = 0;
3613         data.o = o;
3614         data.ws_rule = whitespace_rule(attr_path);
3615         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3616
3617         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3618                 die("unable to read files to diff");
3619
3620         /*
3621          * All the other codepaths check both sides, but not checking
3622          * the "old" side here is deliberate.  We are checking the newly
3623          * introduced changes, and as long as the "new" side is text, we
3624          * can and should check what it introduces.
3625          */
3626         if (diff_filespec_is_binary(two))
3627                 goto free_and_return;
3628         else {
3629                 /* Crazy xdl interfaces.. */
3630                 xpparam_t xpp;
3631                 xdemitconf_t xecfg;
3632
3633                 memset(&xpp, 0, sizeof(xpp));
3634                 memset(&xecfg, 0, sizeof(xecfg));
3635                 xecfg.ctxlen = 1; /* at least one context line */
3636                 xpp.flags = 0;
3637                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3638                                   &xpp, &xecfg))
3639                         die("unable to generate checkdiff for %s", one->path);
3640
3641                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3642                         struct emit_callback ecbdata;
3643                         int blank_at_eof;
3644
3645                         ecbdata.ws_rule = data.ws_rule;
3646                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3647                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3648
3649                         if (blank_at_eof) {
3650                                 static char *err;
3651                                 if (!err)
3652                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3653                                 fprintf(o->file, "%s:%d: %s.\n",
3654                                         data.filename, blank_at_eof, err);
3655                                 data.status = 1; /* report errors */
3656                         }
3657                 }
3658         }
3659  free_and_return:
3660         diff_free_filespec_data(one);
3661         diff_free_filespec_data(two);
3662         if (data.status)
3663                 o->flags.check_failed = 1;
3664 }
3665
3666 struct diff_filespec *alloc_filespec(const char *path)
3667 {
3668         struct diff_filespec *spec;
3669
3670         FLEXPTR_ALLOC_STR(spec, path, path);
3671         spec->count = 1;
3672         spec->is_binary = -1;
3673         return spec;
3674 }
3675
3676 void free_filespec(struct diff_filespec *spec)
3677 {
3678         if (!--spec->count) {
3679                 diff_free_filespec_data(spec);
3680                 free(spec);
3681         }
3682 }
3683
3684 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3685                    int oid_valid, unsigned short mode)
3686 {
3687         if (mode) {
3688                 spec->mode = canon_mode(mode);
3689                 oidcpy(&spec->oid, oid);
3690                 spec->oid_valid = oid_valid;
3691         }
3692 }
3693
3694 /*
3695  * Given a name and sha1 pair, if the index tells us the file in
3696  * the work tree has that object contents, return true, so that
3697  * prepare_temp_file() does not have to inflate and extract.
3698  */
3699 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3700 {
3701         const struct cache_entry *ce;
3702         struct stat st;
3703         int pos, len;
3704
3705         /*
3706          * We do not read the cache ourselves here, because the
3707          * benchmark with my previous version that always reads cache
3708          * shows that it makes things worse for diff-tree comparing
3709          * two linux-2.6 kernel trees in an already checked out work
3710          * tree.  This is because most diff-tree comparisons deal with
3711          * only a small number of files, while reading the cache is
3712          * expensive for a large project, and its cost outweighs the
3713          * savings we get by not inflating the object to a temporary
3714          * file.  Practically, this code only helps when we are used
3715          * by diff-cache --cached, which does read the cache before
3716          * calling us.
3717          */
3718         if (!active_cache)
3719                 return 0;
3720
3721         /* We want to avoid the working directory if our caller
3722          * doesn't need the data in a normal file, this system
3723          * is rather slow with its stat/open/mmap/close syscalls,
3724          * and the object is contained in a pack file.  The pack
3725          * is probably already open and will be faster to obtain
3726          * the data through than the working directory.  Loose
3727          * objects however would tend to be slower as they need
3728          * to be individually opened and inflated.
3729          */
3730         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3731                 return 0;
3732
3733         /*
3734          * Similarly, if we'd have to convert the file contents anyway, that
3735          * makes the optimization not worthwhile.
3736          */
3737         if (!want_file && would_convert_to_git(&the_index, name))
3738                 return 0;
3739
3740         len = strlen(name);
3741         pos = cache_name_pos(name, len);
3742         if (pos < 0)
3743                 return 0;
3744         ce = active_cache[pos];
3745
3746         /*
3747          * This is not the sha1 we are looking for, or
3748          * unreusable because it is not a regular file.
3749          */
3750         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3751                 return 0;
3752
3753         /*
3754          * If ce is marked as "assume unchanged", there is no
3755          * guarantee that work tree matches what we are looking for.
3756          */
3757         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3758                 return 0;
3759
3760         /*
3761          * If ce matches the file in the work tree, we can reuse it.
3762          */
3763         if (ce_uptodate(ce) ||
3764             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3765                 return 1;
3766
3767         return 0;
3768 }
3769
3770 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3771 {
3772         struct strbuf buf = STRBUF_INIT;
3773         char *dirty = "";
3774
3775         /* Are we looking at the work tree? */
3776         if (s->dirty_submodule)
3777                 dirty = "-dirty";
3778
3779         strbuf_addf(&buf, "Subproject commit %s%s\n",
3780                     oid_to_hex(&s->oid), dirty);
3781         s->size = buf.len;
3782         if (size_only) {
3783                 s->data = NULL;
3784                 strbuf_release(&buf);
3785         } else {
3786                 s->data = strbuf_detach(&buf, NULL);
3787                 s->should_free = 1;
3788         }
3789         return 0;
3790 }
3791
3792 /*
3793  * While doing rename detection and pickaxe operation, we may need to
3794  * grab the data for the blob (or file) for our own in-core comparison.
3795  * diff_filespec has data and size fields for this purpose.
3796  */
3797 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3798 {
3799         int size_only = flags & CHECK_SIZE_ONLY;
3800         int err = 0;
3801         int conv_flags = global_conv_flags_eol;
3802         /*
3803          * demote FAIL to WARN to allow inspecting the situation
3804          * instead of refusing.
3805          */
3806         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3807                 conv_flags = CONV_EOL_RNDTRP_WARN;
3808
3809         if (!DIFF_FILE_VALID(s))
3810                 die("internal error: asking to populate invalid file.");
3811         if (S_ISDIR(s->mode))
3812                 return -1;
3813
3814         if (s->data)
3815                 return 0;
3816
3817         if (size_only && 0 < s->size)
3818                 return 0;
3819
3820         if (S_ISGITLINK(s->mode))
3821                 return diff_populate_gitlink(s, size_only);
3822
3823         if (!s->oid_valid ||
3824             reuse_worktree_file(s->path, &s->oid, 0)) {
3825                 struct strbuf buf = STRBUF_INIT;
3826                 struct stat st;
3827                 int fd;
3828
3829                 if (lstat(s->path, &st) < 0) {
3830                 err_empty:
3831                         err = -1;
3832                 empty:
3833                         s->data = (char *)"";
3834                         s->size = 0;
3835                         return err;
3836                 }
3837                 s->size = xsize_t(st.st_size);
3838                 if (!s->size)
3839                         goto empty;
3840                 if (S_ISLNK(st.st_mode)) {
3841                         struct strbuf sb = STRBUF_INIT;
3842
3843                         if (strbuf_readlink(&sb, s->path, s->size))
3844                                 goto err_empty;
3845                         s->size = sb.len;
3846                         s->data = strbuf_detach(&sb, NULL);
3847                         s->should_free = 1;
3848                         return 0;
3849                 }
3850
3851                 /*
3852                  * Even if the caller would be happy with getting
3853                  * only the size, we cannot return early at this
3854                  * point if the path requires us to run the content
3855                  * conversion.
3856                  */
3857                 if (size_only && !would_convert_to_git(&the_index, s->path))
3858                         return 0;
3859
3860                 /*
3861                  * Note: this check uses xsize_t(st.st_size) that may
3862                  * not be the true size of the blob after it goes
3863                  * through convert_to_git().  This may not strictly be
3864                  * correct, but the whole point of big_file_threshold
3865                  * and is_binary check being that we want to avoid
3866                  * opening the file and inspecting the contents, this
3867                  * is probably fine.
3868                  */
3869                 if ((flags & CHECK_BINARY) &&
3870                     s->size > big_file_threshold && s->is_binary == -1) {
3871                         s->is_binary = 1;
3872                         return 0;
3873                 }
3874                 fd = open(s->path, O_RDONLY);
3875                 if (fd < 0)
3876                         goto err_empty;
3877                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3878                 close(fd);
3879                 s->should_munmap = 1;
3880
3881                 /*
3882                  * Convert from working tree format to canonical git format
3883                  */
3884                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3885                         size_t size = 0;
3886                         munmap(s->data, s->size);
3887                         s->should_munmap = 0;
3888                         s->data = strbuf_detach(&buf, &size);
3889                         s->size = size;
3890                         s->should_free = 1;
3891                 }
3892         }
3893         else {
3894                 enum object_type type;
3895                 if (size_only || (flags & CHECK_BINARY)) {
3896                         type = oid_object_info(the_repository, &s->oid,
3897                                                &s->size);
3898                         if (type < 0)
3899                                 die("unable to read %s",
3900                                     oid_to_hex(&s->oid));
3901                         if (size_only)
3902                                 return 0;
3903                         if (s->size > big_file_threshold && s->is_binary == -1) {
3904                                 s->is_binary = 1;
3905                                 return 0;
3906                         }
3907                 }
3908                 s->data = read_object_file(&s->oid, &type, &s->size);
3909                 if (!s->data)
3910                         die("unable to read %s", oid_to_hex(&s->oid));
3911                 s->should_free = 1;
3912         }
3913         return 0;
3914 }
3915
3916 void diff_free_filespec_blob(struct diff_filespec *s)
3917 {
3918         if (s->should_free)
3919                 free(s->data);
3920         else if (s->should_munmap)
3921                 munmap(s->data, s->size);
3922
3923         if (s->should_free || s->should_munmap) {
3924                 s->should_free = s->should_munmap = 0;
3925                 s->data = NULL;
3926         }
3927 }
3928
3929 void diff_free_filespec_data(struct diff_filespec *s)
3930 {
3931         diff_free_filespec_blob(s);
3932         FREE_AND_NULL(s->cnt_data);
3933 }
3934
3935 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3936                            void *blob,
3937                            unsigned long size,
3938                            const struct object_id *oid,
3939                            int mode)
3940 {
3941         struct strbuf buf = STRBUF_INIT;
3942         struct strbuf tempfile = STRBUF_INIT;
3943         char *path_dup = xstrdup(path);
3944         const char *base = basename(path_dup);
3945
3946         /* Generate "XXXXXX_basename.ext" */
3947         strbuf_addstr(&tempfile, "XXXXXX_");
3948         strbuf_addstr(&tempfile, base);
3949
3950         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3951         if (!temp->tempfile)
3952                 die_errno("unable to create temp-file");
3953         if (convert_to_working_tree(path,
3954                         (const char *)blob, (size_t)size, &buf)) {
3955                 blob = buf.buf;
3956                 size = buf.len;
3957         }
3958         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3959             close_tempfile_gently(temp->tempfile))
3960                 die_errno("unable to write temp-file");
3961         temp->name = get_tempfile_path(temp->tempfile);
3962         oid_to_hex_r(temp->hex, oid);
3963         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3964         strbuf_release(&buf);
3965         strbuf_release(&tempfile);
3966         free(path_dup);
3967 }
3968
3969 static struct diff_tempfile *prepare_temp_file(const char *name,
3970                 struct diff_filespec *one)
3971 {
3972         struct diff_tempfile *temp = claim_diff_tempfile();
3973
3974         if (!DIFF_FILE_VALID(one)) {
3975         not_a_valid_file:
3976                 /* A '-' entry produces this for file-2, and
3977                  * a '+' entry produces this for file-1.
3978                  */
3979                 temp->name = "/dev/null";
3980                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3981                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3982                 return temp;
3983         }
3984
3985         if (!S_ISGITLINK(one->mode) &&
3986             (!one->oid_valid ||
3987              reuse_worktree_file(name, &one->oid, 1))) {
3988                 struct stat st;
3989                 if (lstat(name, &st) < 0) {
3990                         if (errno == ENOENT)
3991                                 goto not_a_valid_file;
3992                         die_errno("stat(%s)", name);
3993                 }
3994                 if (S_ISLNK(st.st_mode)) {
3995                         struct strbuf sb = STRBUF_INIT;
3996                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3997                                 die_errno("readlink(%s)", name);
3998                         prep_temp_blob(name, temp, sb.buf, sb.len,
3999                                        (one->oid_valid ?
4000                                         &one->oid : &null_oid),
4001                                        (one->oid_valid ?
4002                                         one->mode : S_IFLNK));
4003                         strbuf_release(&sb);
4004                 }
4005                 else {
4006                         /* we can borrow from the file in the work tree */
4007                         temp->name = name;
4008                         if (!one->oid_valid)
4009                                 oid_to_hex_r(temp->hex, &null_oid);
4010                         else
4011                                 oid_to_hex_r(temp->hex, &one->oid);
4012                         /* Even though we may sometimes borrow the
4013                          * contents from the work tree, we always want
4014                          * one->mode.  mode is trustworthy even when
4015                          * !(one->oid_valid), as long as
4016                          * DIFF_FILE_VALID(one).
4017                          */
4018                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4019                 }
4020                 return temp;
4021         }
4022         else {
4023                 if (diff_populate_filespec(one, 0))
4024                         die("cannot read data blob for %s", one->path);
4025                 prep_temp_blob(name, temp, one->data, one->size,
4026                                &one->oid, one->mode);
4027         }
4028         return temp;
4029 }
4030
4031 static void add_external_diff_name(struct argv_array *argv,
4032                                    const char *name,
4033                                    struct diff_filespec *df)
4034 {
4035         struct diff_tempfile *temp = prepare_temp_file(name, df);
4036         argv_array_push(argv, temp->name);
4037         argv_array_push(argv, temp->hex);
4038         argv_array_push(argv, temp->mode);
4039 }
4040
4041 /* An external diff command takes:
4042  *
4043  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4044  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4045  *
4046  */
4047 static void run_external_diff(const char *pgm,
4048                               const char *name,
4049                               const char *other,
4050                               struct diff_filespec *one,
4051                               struct diff_filespec *two,
4052                               const char *xfrm_msg,
4053                               int complete_rewrite,
4054                               struct diff_options *o)
4055 {
4056         struct argv_array argv = ARGV_ARRAY_INIT;
4057         struct argv_array env = ARGV_ARRAY_INIT;
4058         struct diff_queue_struct *q = &diff_queued_diff;
4059
4060         argv_array_push(&argv, pgm);
4061         argv_array_push(&argv, name);
4062
4063         if (one && two) {
4064                 add_external_diff_name(&argv, name, one);
4065                 if (!other)
4066                         add_external_diff_name(&argv, name, two);
4067                 else {
4068                         add_external_diff_name(&argv, other, two);
4069                         argv_array_push(&argv, other);
4070                         argv_array_push(&argv, xfrm_msg);
4071                 }
4072         }
4073
4074         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4075         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4076
4077         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4078                 die(_("external diff died, stopping at %s"), name);
4079
4080         remove_tempfile();
4081         argv_array_clear(&argv);
4082         argv_array_clear(&env);
4083 }
4084
4085 static int similarity_index(struct diff_filepair *p)
4086 {
4087         return p->score * 100 / MAX_SCORE;
4088 }
4089
4090 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4091 {
4092         if (startup_info->have_repository)
4093                 return find_unique_abbrev(oid, abbrev);
4094         else {
4095                 char *hex = oid_to_hex(oid);
4096                 if (abbrev < 0)
4097                         abbrev = FALLBACK_DEFAULT_ABBREV;
4098                 if (abbrev > the_hash_algo->hexsz)
4099                         BUG("oid abbreviation out of range: %d", abbrev);
4100                 if (abbrev)
4101                         hex[abbrev] = '\0';
4102                 return hex;
4103         }
4104 }
4105
4106 static void fill_metainfo(struct strbuf *msg,
4107                           const char *name,
4108                           const char *other,
4109                           struct diff_filespec *one,
4110                           struct diff_filespec *two,
4111                           struct diff_options *o,
4112                           struct diff_filepair *p,
4113                           int *must_show_header,
4114                           int use_color)
4115 {
4116         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4117         const char *reset = diff_get_color(use_color, DIFF_RESET);
4118         const char *line_prefix = diff_line_prefix(o);
4119
4120         *must_show_header = 1;
4121         strbuf_init(msg, PATH_MAX * 2 + 300);
4122         switch (p->status) {
4123         case DIFF_STATUS_COPIED:
4124                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4125                             line_prefix, set, similarity_index(p));
4126                 strbuf_addf(msg, "%s\n%s%scopy from ",
4127                             reset,  line_prefix, set);
4128                 quote_c_style(name, msg, NULL, 0);
4129                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4130                 quote_c_style(other, msg, NULL, 0);
4131                 strbuf_addf(msg, "%s\n", reset);
4132                 break;
4133         case DIFF_STATUS_RENAMED:
4134                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4135                             line_prefix, set, similarity_index(p));
4136                 strbuf_addf(msg, "%s\n%s%srename from ",
4137                             reset, line_prefix, set);
4138                 quote_c_style(name, msg, NULL, 0);
4139                 strbuf_addf(msg, "%s\n%s%srename to ",
4140                             reset, line_prefix, set);
4141                 quote_c_style(other, msg, NULL, 0);
4142                 strbuf_addf(msg, "%s\n", reset);
4143                 break;
4144         case DIFF_STATUS_MODIFIED:
4145                 if (p->score) {
4146                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4147                                     line_prefix,
4148                                     set, similarity_index(p), reset);
4149                         break;
4150                 }
4151                 /* fallthru */
4152         default:
4153                 *must_show_header = 0;
4154         }
4155         if (one && two && oidcmp(&one->oid, &two->oid)) {
4156                 const unsigned hexsz = the_hash_algo->hexsz;
4157                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4158
4159                 if (o->flags.binary) {
4160                         mmfile_t mf;
4161                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
4162                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
4163                                 abbrev = hexsz;
4164                 }
4165                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4166                             diff_abbrev_oid(&one->oid, abbrev),
4167                             diff_abbrev_oid(&two->oid, abbrev));
4168                 if (one->mode == two->mode)
4169                         strbuf_addf(msg, " %06o", one->mode);
4170                 strbuf_addf(msg, "%s\n", reset);
4171         }
4172 }
4173
4174 static void run_diff_cmd(const char *pgm,
4175                          const char *name,
4176                          const char *other,
4177                          const char *attr_path,
4178                          struct diff_filespec *one,
4179                          struct diff_filespec *two,
4180                          struct strbuf *msg,
4181                          struct diff_options *o,
4182                          struct diff_filepair *p)
4183 {
4184         const char *xfrm_msg = NULL;
4185         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4186         int must_show_header = 0;
4187
4188
4189         if (o->flags.allow_external) {
4190                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
4191                 if (drv && drv->external)
4192                         pgm = drv->external;
4193         }
4194
4195         if (msg) {
4196                 /*
4197                  * don't use colors when the header is intended for an
4198                  * external diff driver
4199                  */
4200                 fill_metainfo(msg, name, other, one, two, o, p,
4201                               &must_show_header,
4202                               want_color(o->use_color) && !pgm);
4203                 xfrm_msg = msg->len ? msg->buf : NULL;
4204         }
4205
4206         if (pgm) {
4207                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4208                                   complete_rewrite, o);
4209                 return;
4210         }
4211         if (one && two)
4212                 builtin_diff(name, other ? other : name,
4213                              one, two, xfrm_msg, must_show_header,
4214                              o, complete_rewrite);
4215         else
4216                 fprintf(o->file, "* Unmerged path %s\n", name);
4217 }
4218
4219 static void diff_fill_oid_info(struct diff_filespec *one)
4220 {
4221         if (DIFF_FILE_VALID(one)) {
4222                 if (!one->oid_valid) {
4223                         struct stat st;
4224                         if (one->is_stdin) {
4225                                 oidclr(&one->oid);
4226                                 return;
4227                         }
4228                         if (lstat(one->path, &st) < 0)
4229                                 die_errno("stat '%s'", one->path);
4230                         if (index_path(&one->oid, one->path, &st, 0))
4231                                 die("cannot hash %s", one->path);
4232                 }
4233         }
4234         else
4235                 oidclr(&one->oid);
4236 }
4237
4238 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4239 {
4240         /* Strip the prefix but do not molest /dev/null and absolute paths */
4241         if (*namep && **namep != '/') {
4242                 *namep += prefix_length;
4243                 if (**namep == '/')
4244                         ++*namep;
4245         }
4246         if (*otherp && **otherp != '/') {
4247                 *otherp += prefix_length;
4248                 if (**otherp == '/')
4249                         ++*otherp;
4250         }
4251 }
4252
4253 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4254 {
4255         const char *pgm = external_diff();
4256         struct strbuf msg;
4257         struct diff_filespec *one = p->one;
4258         struct diff_filespec *two = p->two;
4259         const char *name;
4260         const char *other;
4261         const char *attr_path;
4262
4263         name  = one->path;
4264         other = (strcmp(name, two->path) ? two->path : NULL);
4265         attr_path = name;
4266         if (o->prefix_length)
4267                 strip_prefix(o->prefix_length, &name, &other);
4268
4269         if (!o->flags.allow_external)
4270                 pgm = NULL;
4271
4272         if (DIFF_PAIR_UNMERGED(p)) {
4273                 run_diff_cmd(pgm, name, NULL, attr_path,
4274                              NULL, NULL, NULL, o, p);
4275                 return;
4276         }
4277
4278         diff_fill_oid_info(one);
4279         diff_fill_oid_info(two);
4280
4281         if (!pgm &&
4282             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4283             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4284                 /*
4285                  * a filepair that changes between file and symlink
4286                  * needs to be split into deletion and creation.
4287                  */
4288                 struct diff_filespec *null = alloc_filespec(two->path);
4289                 run_diff_cmd(NULL, name, other, attr_path,
4290                              one, null, &msg, o, p);
4291                 free(null);
4292                 strbuf_release(&msg);
4293
4294                 null = alloc_filespec(one->path);
4295                 run_diff_cmd(NULL, name, other, attr_path,
4296                              null, two, &msg, o, p);
4297                 free(null);
4298         }
4299         else
4300                 run_diff_cmd(pgm, name, other, attr_path,
4301                              one, two, &msg, o, p);
4302
4303         strbuf_release(&msg);
4304 }
4305
4306 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4307                          struct diffstat_t *diffstat)
4308 {
4309         const char *name;
4310         const char *other;
4311
4312         if (DIFF_PAIR_UNMERGED(p)) {
4313                 /* unmerged */
4314                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4315                 return;
4316         }
4317
4318         name = p->one->path;
4319         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4320
4321         if (o->prefix_length)
4322                 strip_prefix(o->prefix_length, &name, &other);
4323
4324         diff_fill_oid_info(p->one);
4325         diff_fill_oid_info(p->two);
4326
4327         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4328 }
4329
4330 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4331 {
4332         const char *name;
4333         const char *other;
4334         const char *attr_path;
4335
4336         if (DIFF_PAIR_UNMERGED(p)) {
4337                 /* unmerged */
4338                 return;
4339         }
4340
4341         name = p->one->path;
4342         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4343         attr_path = other ? other : name;
4344
4345         if (o->prefix_length)
4346                 strip_prefix(o->prefix_length, &name, &other);
4347
4348         diff_fill_oid_info(p->one);
4349         diff_fill_oid_info(p->two);
4350
4351         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4352 }
4353
4354 void diff_setup(struct diff_options *options)
4355 {
4356         memcpy(options, &default_diff_options, sizeof(*options));
4357
4358         options->file = stdout;
4359
4360         options->abbrev = DEFAULT_ABBREV;
4361         options->line_termination = '\n';
4362         options->break_opt = -1;
4363         options->rename_limit = -1;
4364         options->dirstat_permille = diff_dirstat_permille_default;
4365         options->context = diff_context_default;
4366         options->interhunkcontext = diff_interhunk_context_default;
4367         options->ws_error_highlight = ws_error_highlight_default;
4368         options->flags.rename_empty = 1;
4369         options->objfind = NULL;
4370
4371         /* pathchange left =NULL by default */
4372         options->change = diff_change;
4373         options->add_remove = diff_addremove;
4374         options->use_color = diff_use_color_default;
4375         options->detect_rename = diff_detect_rename_default;
4376         options->xdl_opts |= diff_algorithm;
4377         if (diff_indent_heuristic)
4378                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4379
4380         options->orderfile = diff_order_file_cfg;
4381
4382         if (diff_no_prefix) {
4383                 options->a_prefix = options->b_prefix = "";
4384         } else if (!diff_mnemonic_prefix) {
4385                 options->a_prefix = "a/";
4386                 options->b_prefix = "b/";
4387         }
4388
4389         options->color_moved = diff_color_moved_default;
4390         options->color_moved_ws_handling = diff_color_moved_ws_default;
4391 }
4392
4393 void diff_setup_done(struct diff_options *options)
4394 {
4395         unsigned check_mask = DIFF_FORMAT_NAME |
4396                               DIFF_FORMAT_NAME_STATUS |
4397                               DIFF_FORMAT_CHECKDIFF |
4398                               DIFF_FORMAT_NO_OUTPUT;
4399         /*
4400          * This must be signed because we're comparing against a potentially
4401          * negative value.
4402          */
4403         const int hexsz = the_hash_algo->hexsz;
4404
4405         if (options->set_default)
4406                 options->set_default(options);
4407
4408         if (HAS_MULTI_BITS(options->output_format & check_mask))
4409                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4410
4411         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4412                 die(_("-G, -S and --find-object are mutually exclusive"));
4413
4414         /*
4415          * Most of the time we can say "there are changes"
4416          * only by checking if there are changed paths, but
4417          * --ignore-whitespace* options force us to look
4418          * inside contents.
4419          */
4420
4421         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4422                 options->flags.diff_from_contents = 1;
4423         else
4424                 options->flags.diff_from_contents = 0;
4425
4426         if (options->flags.find_copies_harder)
4427                 options->detect_rename = DIFF_DETECT_COPY;
4428
4429         if (!options->flags.relative_name)
4430                 options->prefix = NULL;
4431         if (options->prefix)
4432                 options->prefix_length = strlen(options->prefix);
4433         else
4434                 options->prefix_length = 0;
4435
4436         if (options->output_format & (DIFF_FORMAT_NAME |
4437                                       DIFF_FORMAT_NAME_STATUS |
4438                                       DIFF_FORMAT_CHECKDIFF |
4439                                       DIFF_FORMAT_NO_OUTPUT))
4440                 options->output_format &= ~(DIFF_FORMAT_RAW |
4441                                             DIFF_FORMAT_NUMSTAT |
4442                                             DIFF_FORMAT_DIFFSTAT |
4443                                             DIFF_FORMAT_SHORTSTAT |
4444                                             DIFF_FORMAT_DIRSTAT |
4445                                             DIFF_FORMAT_SUMMARY |
4446                                             DIFF_FORMAT_PATCH);
4447
4448         /*
4449          * These cases always need recursive; we do not drop caller-supplied
4450          * recursive bits for other formats here.
4451          */
4452         if (options->output_format & (DIFF_FORMAT_PATCH |
4453                                       DIFF_FORMAT_NUMSTAT |
4454                                       DIFF_FORMAT_DIFFSTAT |
4455                                       DIFF_FORMAT_SHORTSTAT |
4456                                       DIFF_FORMAT_DIRSTAT |
4457                                       DIFF_FORMAT_SUMMARY |
4458                                       DIFF_FORMAT_CHECKDIFF))
4459                 options->flags.recursive = 1;
4460         /*
4461          * Also pickaxe would not work very well if you do not say recursive
4462          */
4463         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4464                 options->flags.recursive = 1;
4465         /*
4466          * When patches are generated, submodules diffed against the work tree
4467          * must be checked for dirtiness too so it can be shown in the output
4468          */
4469         if (options->output_format & DIFF_FORMAT_PATCH)
4470                 options->flags.dirty_submodules = 1;
4471
4472         if (options->detect_rename && options->rename_limit < 0)
4473                 options->rename_limit = diff_rename_limit_default;
4474         if (options->setup & DIFF_SETUP_USE_CACHE) {
4475                 if (!active_cache)
4476                         /* read-cache does not die even when it fails
4477                          * so it is safe for us to do this here.  Also
4478                          * it does not smudge active_cache or active_nr
4479                          * when it fails, so we do not have to worry about
4480                          * cleaning it up ourselves either.
4481                          */
4482                         read_cache();
4483         }
4484         if (hexsz < options->abbrev)
4485                 options->abbrev = hexsz; /* full */
4486
4487         /*
4488          * It does not make sense to show the first hit we happened
4489          * to have found.  It does not make sense not to return with
4490          * exit code in such a case either.
4491          */
4492         if (options->flags.quick) {
4493                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4494                 options->flags.exit_with_status = 1;
4495         }
4496
4497         options->diff_path_counter = 0;
4498
4499         if (options->flags.follow_renames && options->pathspec.nr != 1)
4500                 die(_("--follow requires exactly one pathspec"));
4501
4502         if (!options->use_color || external_diff())
4503                 options->color_moved = 0;
4504 }
4505
4506 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4507 {
4508         char c, *eq;
4509         int len;
4510
4511         if (*arg != '-')
4512                 return 0;
4513         c = *++arg;
4514         if (!c)
4515                 return 0;
4516         if (c == arg_short) {
4517                 c = *++arg;
4518                 if (!c)
4519                         return 1;
4520                 if (val && isdigit(c)) {
4521                         char *end;
4522                         int n = strtoul(arg, &end, 10);
4523                         if (*end)
4524                                 return 0;
4525                         *val = n;
4526                         return 1;
4527                 }
4528                 return 0;
4529         }
4530         if (c != '-')
4531                 return 0;
4532         arg++;
4533         eq = strchrnul(arg, '=');
4534         len = eq - arg;
4535         if (!len || strncmp(arg, arg_long, len))
4536                 return 0;
4537         if (*eq) {
4538                 int n;
4539                 char *end;
4540                 if (!isdigit(*++eq))
4541                         return 0;
4542                 n = strtoul(eq, &end, 10);
4543                 if (*end)
4544                         return 0;
4545                 *val = n;
4546         }
4547         return 1;
4548 }
4549
4550 static int diff_scoreopt_parse(const char *opt);
4551
4552 static inline int short_opt(char opt, const char **argv,
4553                             const char **optarg)
4554 {
4555         const char *arg = argv[0];
4556         if (arg[0] != '-' || arg[1] != opt)
4557                 return 0;
4558         if (arg[2] != '\0') {
4559                 *optarg = arg + 2;
4560                 return 1;
4561         }
4562         if (!argv[1])
4563                 die("Option '%c' requires a value", opt);
4564         *optarg = argv[1];
4565         return 2;
4566 }
4567
4568 int parse_long_opt(const char *opt, const char **argv,
4569                    const char **optarg)
4570 {
4571         const char *arg = argv[0];
4572         if (!skip_prefix(arg, "--", &arg))
4573                 return 0;
4574         if (!skip_prefix(arg, opt, &arg))
4575                 return 0;
4576         if (*arg == '=') { /* stuck form: --option=value */
4577                 *optarg = arg + 1;
4578                 return 1;
4579         }
4580         if (*arg != '\0')
4581                 return 0;
4582         /* separate form: --option value */
4583         if (!argv[1])
4584                 die("Option '--%s' requires a value", opt);
4585         *optarg = argv[1];
4586         return 2;
4587 }
4588
4589 static int stat_opt(struct diff_options *options, const char **av)
4590 {
4591         const char *arg = av[0];
4592         char *end;
4593         int width = options->stat_width;
4594         int name_width = options->stat_name_width;
4595         int graph_width = options->stat_graph_width;
4596         int count = options->stat_count;
4597         int argcount = 1;
4598
4599         if (!skip_prefix(arg, "--stat", &arg))
4600                 BUG("stat option does not begin with --stat: %s", arg);
4601         end = (char *)arg;
4602
4603         switch (*arg) {
4604         case '-':
4605                 if (skip_prefix(arg, "-width", &arg)) {
4606                         if (*arg == '=')
4607                                 width = strtoul(arg + 1, &end, 10);
4608                         else if (!*arg && !av[1])
4609                                 die_want_option("--stat-width");
4610                         else if (!*arg) {
4611                                 width = strtoul(av[1], &end, 10);
4612                                 argcount = 2;
4613                         }
4614                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4615                         if (*arg == '=')
4616                                 name_width = strtoul(arg + 1, &end, 10);
4617                         else if (!*arg && !av[1])
4618                                 die_want_option("--stat-name-width");
4619                         else if (!*arg) {
4620                                 name_width = strtoul(av[1], &end, 10);
4621                                 argcount = 2;
4622                         }
4623                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4624                         if (*arg == '=')
4625                                 graph_width = strtoul(arg + 1, &end, 10);
4626                         else if (!*arg && !av[1])
4627                                 die_want_option("--stat-graph-width");
4628                         else if (!*arg) {
4629                                 graph_width = strtoul(av[1], &end, 10);
4630                                 argcount = 2;
4631                         }
4632                 } else if (skip_prefix(arg, "-count", &arg)) {
4633                         if (*arg == '=')
4634                                 count = strtoul(arg + 1, &end, 10);
4635                         else if (!*arg && !av[1])
4636                                 die_want_option("--stat-count");
4637                         else if (!*arg) {
4638                                 count = strtoul(av[1], &end, 10);
4639                                 argcount = 2;
4640                         }
4641                 }
4642                 break;
4643         case '=':
4644                 width = strtoul(arg+1, &end, 10);
4645                 if (*end == ',')
4646                         name_width = strtoul(end+1, &end, 10);
4647                 if (*end == ',')
4648                         count = strtoul(end+1, &end, 10);
4649         }
4650
4651         /* Important! This checks all the error cases! */
4652         if (*end)
4653                 return 0;
4654         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4655         options->stat_name_width = name_width;
4656         options->stat_graph_width = graph_width;
4657         options->stat_width = width;
4658         options->stat_count = count;
4659         return argcount;
4660 }
4661
4662 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4663 {
4664         struct strbuf errmsg = STRBUF_INIT;
4665         if (parse_dirstat_params(options, params, &errmsg))
4666                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4667                     errmsg.buf);
4668         strbuf_release(&errmsg);
4669         /*
4670          * The caller knows a dirstat-related option is given from the command
4671          * line; allow it to say "return this_function();"
4672          */
4673         options->output_format |= DIFF_FORMAT_DIRSTAT;
4674         return 1;
4675 }
4676
4677 static int parse_submodule_opt(struct diff_options *options, const char *value)
4678 {
4679         if (parse_submodule_params(options, value))
4680                 die(_("Failed to parse --submodule option parameter: '%s'"),
4681                         value);
4682         return 1;
4683 }
4684
4685 static const char diff_status_letters[] = {
4686         DIFF_STATUS_ADDED,
4687         DIFF_STATUS_COPIED,
4688         DIFF_STATUS_DELETED,
4689         DIFF_STATUS_MODIFIED,
4690         DIFF_STATUS_RENAMED,
4691         DIFF_STATUS_TYPE_CHANGED,
4692         DIFF_STATUS_UNKNOWN,
4693         DIFF_STATUS_UNMERGED,
4694         DIFF_STATUS_FILTER_AON,
4695         DIFF_STATUS_FILTER_BROKEN,
4696         '\0',
4697 };
4698
4699 static unsigned int filter_bit['Z' + 1];
4700
4701 static void prepare_filter_bits(void)
4702 {
4703         int i;
4704
4705         if (!filter_bit[DIFF_STATUS_ADDED]) {
4706                 for (i = 0; diff_status_letters[i]; i++)
4707                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4708         }
4709 }
4710
4711 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4712 {
4713         return opt->filter & filter_bit[(int) status];
4714 }
4715
4716 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4717 {
4718         int i, optch;
4719
4720         prepare_filter_bits();
4721
4722         /*
4723          * If there is a negation e.g. 'd' in the input, and we haven't
4724          * initialized the filter field with another --diff-filter, start
4725          * from full set of bits, except for AON.
4726          */
4727         if (!opt->filter) {
4728                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4729                         if (optch < 'a' || 'z' < optch)
4730                                 continue;
4731                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4732                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4733                         break;
4734                 }
4735         }
4736
4737         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4738                 unsigned int bit;
4739                 int negate;
4740
4741                 if ('a' <= optch && optch <= 'z') {
4742                         negate = 1;
4743                         optch = toupper(optch);
4744                 } else {
4745                         negate = 0;
4746                 }
4747
4748                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4749                 if (!bit)
4750                         return optarg[i];
4751                 if (negate)
4752                         opt->filter &= ~bit;
4753                 else
4754                         opt->filter |= bit;
4755         }
4756         return 0;
4757 }
4758
4759 static void enable_patch_output(int *fmt) {
4760         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4761         *fmt |= DIFF_FORMAT_PATCH;
4762 }
4763
4764 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4765 {
4766         int val = parse_ws_error_highlight(arg);
4767
4768         if (val < 0) {
4769                 error("unknown value after ws-error-highlight=%.*s",
4770                       -1 - val, arg);
4771                 return 0;
4772         }
4773         opt->ws_error_highlight = val;
4774         return 1;
4775 }
4776
4777 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4778 {
4779         struct object_id oid;
4780
4781         if (get_oid(arg, &oid))
4782                 return error("unable to resolve '%s'", arg);
4783
4784         if (!opt->objfind)
4785                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4786
4787         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4788         opt->flags.recursive = 1;
4789         opt->flags.tree_in_recursive = 1;
4790         oidset_insert(opt->objfind, &oid);
4791         return 1;
4792 }
4793
4794 int diff_opt_parse(struct diff_options *options,
4795                    const char **av, int ac, const char *prefix)
4796 {
4797         const char *arg = av[0];
4798         const char *optarg;
4799         int argcount;
4800
4801         if (!prefix)
4802                 prefix = "";
4803
4804         /* Output format options */
4805         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4806             || opt_arg(arg, 'U', "unified", &options->context))
4807                 enable_patch_output(&options->output_format);
4808         else if (!strcmp(arg, "--raw"))
4809                 options->output_format |= DIFF_FORMAT_RAW;
4810         else if (!strcmp(arg, "--patch-with-raw")) {
4811                 enable_patch_output(&options->output_format);
4812                 options->output_format |= DIFF_FORMAT_RAW;
4813         } else if (!strcmp(arg, "--numstat"))
4814                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4815         else if (!strcmp(arg, "--shortstat"))
4816                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4817         else if (skip_prefix(arg, "-X", &arg) ||
4818                  skip_to_optional_arg(arg, "--dirstat", &arg))
4819                 return parse_dirstat_opt(options, arg);
4820         else if (!strcmp(arg, "--cumulative"))
4821                 return parse_dirstat_opt(options, "cumulative");
4822         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4823                 parse_dirstat_opt(options, "files");
4824                 return parse_dirstat_opt(options, arg);
4825         }
4826         else if (!strcmp(arg, "--check"))
4827                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4828         else if (!strcmp(arg, "--summary"))
4829                 options->output_format |= DIFF_FORMAT_SUMMARY;
4830         else if (!strcmp(arg, "--patch-with-stat")) {
4831                 enable_patch_output(&options->output_format);
4832                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4833         } else if (!strcmp(arg, "--name-only"))
4834                 options->output_format |= DIFF_FORMAT_NAME;
4835         else if (!strcmp(arg, "--name-status"))
4836                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4837         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4838                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4839         else if (starts_with(arg, "--stat"))
4840                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4841                 return stat_opt(options, av);
4842         else if (!strcmp(arg, "--compact-summary")) {
4843                  options->flags.stat_with_summary = 1;
4844                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4845         } else if (!strcmp(arg, "--no-compact-summary"))
4846                  options->flags.stat_with_summary = 0;
4847
4848         /* renames options */
4849         else if (starts_with(arg, "-B") ||
4850                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4851                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4852                         return error("invalid argument to -B: %s", arg+2);
4853         }
4854         else if (starts_with(arg, "-M") ||
4855                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4856                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4857                         return error("invalid argument to -M: %s", arg+2);
4858                 options->detect_rename = DIFF_DETECT_RENAME;
4859         }
4860         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4861                 options->irreversible_delete = 1;
4862         }
4863         else if (starts_with(arg, "-C") ||
4864                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4865                 if (options->detect_rename == DIFF_DETECT_COPY)
4866                         options->flags.find_copies_harder = 1;
4867                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4868                         return error("invalid argument to -C: %s", arg+2);
4869                 options->detect_rename = DIFF_DETECT_COPY;
4870         }
4871         else if (!strcmp(arg, "--no-renames"))
4872                 options->detect_rename = 0;
4873         else if (!strcmp(arg, "--rename-empty"))
4874                 options->flags.rename_empty = 1;
4875         else if (!strcmp(arg, "--no-rename-empty"))
4876                 options->flags.rename_empty = 0;
4877         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4878                 options->flags.relative_name = 1;
4879                 if (arg)
4880                         options->prefix = arg;
4881         }
4882
4883         /* xdiff options */
4884         else if (!strcmp(arg, "--minimal"))
4885                 DIFF_XDL_SET(options, NEED_MINIMAL);
4886         else if (!strcmp(arg, "--no-minimal"))
4887                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4888         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4889                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4890         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4891                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4892         else if (!strcmp(arg, "--ignore-space-at-eol"))
4893                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4894         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4895                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4896         else if (!strcmp(arg, "--ignore-blank-lines"))
4897                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4898         else if (!strcmp(arg, "--indent-heuristic"))
4899                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4900         else if (!strcmp(arg, "--no-indent-heuristic"))
4901                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4902         else if (!strcmp(arg, "--patience")) {
4903                 int i;
4904                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4905                 /*
4906                  * Both --patience and --anchored use PATIENCE_DIFF
4907                  * internally, so remove any anchors previously
4908                  * specified.
4909                  */
4910                 for (i = 0; i < options->anchors_nr; i++)
4911                         free(options->anchors[i]);
4912                 options->anchors_nr = 0;
4913         } else if (!strcmp(arg, "--histogram"))
4914                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4915         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4916                 long value = parse_algorithm_value(optarg);
4917                 if (value < 0)
4918                         return error("option diff-algorithm accepts \"myers\", "
4919                                      "\"minimal\", \"patience\" and \"histogram\"");
4920                 /* clear out previous settings */
4921                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4922                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4923                 options->xdl_opts |= value;
4924                 return argcount;
4925         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4926                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4927                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4928                            options->anchors_alloc);
4929                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4930         }
4931
4932         /* flags options */
4933         else if (!strcmp(arg, "--binary")) {
4934                 enable_patch_output(&options->output_format);
4935                 options->flags.binary = 1;
4936         }
4937         else if (!strcmp(arg, "--full-index"))
4938                 options->flags.full_index = 1;
4939         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4940                 options->flags.text = 1;
4941         else if (!strcmp(arg, "-R"))
4942                 options->flags.reverse_diff = 1;
4943         else if (!strcmp(arg, "--find-copies-harder"))
4944                 options->flags.find_copies_harder = 1;
4945         else if (!strcmp(arg, "--follow"))
4946                 options->flags.follow_renames = 1;
4947         else if (!strcmp(arg, "--no-follow")) {
4948                 options->flags.follow_renames = 0;
4949                 options->flags.default_follow_renames = 0;
4950         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4951                 int value = git_config_colorbool(NULL, arg);
4952                 if (value < 0)
4953                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4954                 options->use_color = value;
4955         }
4956         else if (!strcmp(arg, "--no-color"))
4957                 options->use_color = 0;
4958         else if (!strcmp(arg, "--color-moved")) {
4959                 if (diff_color_moved_default)
4960                         options->color_moved = diff_color_moved_default;
4961                 if (options->color_moved == COLOR_MOVED_NO)
4962                         options->color_moved = COLOR_MOVED_DEFAULT;
4963         } else if (!strcmp(arg, "--no-color-moved"))
4964                 options->color_moved = COLOR_MOVED_NO;
4965         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4966                 int cm = parse_color_moved(arg);
4967                 if (cm < 0)
4968                         die("bad --color-moved argument: %s", arg);
4969                 options->color_moved = cm;
4970         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
4971                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
4972         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4973                 options->use_color = 1;
4974                 options->word_diff = DIFF_WORDS_COLOR;
4975         }
4976         else if (!strcmp(arg, "--word-diff")) {
4977                 if (options->word_diff == DIFF_WORDS_NONE)
4978                         options->word_diff = DIFF_WORDS_PLAIN;
4979         }
4980         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4981                 if (!strcmp(arg, "plain"))
4982                         options->word_diff = DIFF_WORDS_PLAIN;
4983                 else if (!strcmp(arg, "color")) {
4984                         options->use_color = 1;
4985                         options->word_diff = DIFF_WORDS_COLOR;
4986                 }
4987                 else if (!strcmp(arg, "porcelain"))
4988                         options->word_diff = DIFF_WORDS_PORCELAIN;
4989                 else if (!strcmp(arg, "none"))
4990                         options->word_diff = DIFF_WORDS_NONE;
4991                 else
4992                         die("bad --word-diff argument: %s", arg);
4993         }
4994         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4995                 if (options->word_diff == DIFF_WORDS_NONE)
4996                         options->word_diff = DIFF_WORDS_PLAIN;
4997                 options->word_regex = optarg;
4998                 return argcount;
4999         }
5000         else if (!strcmp(arg, "--exit-code"))
5001                 options->flags.exit_with_status = 1;
5002         else if (!strcmp(arg, "--quiet"))
5003                 options->flags.quick = 1;
5004         else if (!strcmp(arg, "--ext-diff"))
5005                 options->flags.allow_external = 1;
5006         else if (!strcmp(arg, "--no-ext-diff"))
5007                 options->flags.allow_external = 0;
5008         else if (!strcmp(arg, "--textconv")) {
5009                 options->flags.allow_textconv = 1;
5010                 options->flags.textconv_set_via_cmdline = 1;
5011         } else if (!strcmp(arg, "--no-textconv"))
5012                 options->flags.allow_textconv = 0;
5013         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5014                 options->flags.override_submodule_config = 1;
5015                 handle_ignore_submodules_arg(options, arg);
5016         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5017                 return parse_submodule_opt(options, arg);
5018         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5019                 return parse_ws_error_highlight_opt(options, arg);
5020         else if (!strcmp(arg, "--ita-invisible-in-index"))
5021                 options->ita_invisible_in_index = 1;
5022         else if (!strcmp(arg, "--ita-visible-in-index"))
5023                 options->ita_invisible_in_index = 0;
5024
5025         /* misc options */
5026         else if (!strcmp(arg, "-z"))
5027                 options->line_termination = 0;
5028         else if ((argcount = short_opt('l', av, &optarg))) {
5029                 options->rename_limit = strtoul(optarg, NULL, 10);
5030                 return argcount;
5031         }
5032         else if ((argcount = short_opt('S', av, &optarg))) {
5033                 options->pickaxe = optarg;
5034                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5035                 return argcount;
5036         } else if ((argcount = short_opt('G', av, &optarg))) {
5037                 options->pickaxe = optarg;
5038                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5039                 return argcount;
5040         }
5041         else if (!strcmp(arg, "--pickaxe-all"))
5042                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5043         else if (!strcmp(arg, "--pickaxe-regex"))
5044                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5045         else if ((argcount = short_opt('O', av, &optarg))) {
5046                 options->orderfile = prefix_filename(prefix, optarg);
5047                 return argcount;
5048         } else if (skip_prefix(arg, "--find-object=", &arg))
5049                 return parse_objfind_opt(options, arg);
5050         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5051                 int offending = parse_diff_filter_opt(optarg, options);
5052                 if (offending)
5053                         die("unknown change class '%c' in --diff-filter=%s",
5054                             offending, optarg);
5055                 return argcount;
5056         }
5057         else if (!strcmp(arg, "--no-abbrev"))
5058                 options->abbrev = 0;
5059         else if (!strcmp(arg, "--abbrev"))
5060                 options->abbrev = DEFAULT_ABBREV;
5061         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5062                 options->abbrev = strtoul(arg, NULL, 10);
5063                 if (options->abbrev < MINIMUM_ABBREV)
5064                         options->abbrev = MINIMUM_ABBREV;
5065                 else if (the_hash_algo->hexsz < options->abbrev)
5066                         options->abbrev = the_hash_algo->hexsz;
5067         }
5068         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5069                 options->a_prefix = optarg;
5070                 return argcount;
5071         }
5072         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5073                 options->line_prefix = optarg;
5074                 options->line_prefix_length = strlen(options->line_prefix);
5075                 graph_setup_line_prefix(options);
5076                 return argcount;
5077         }
5078         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5079                 options->b_prefix = optarg;
5080                 return argcount;
5081         }
5082         else if (!strcmp(arg, "--no-prefix"))
5083                 options->a_prefix = options->b_prefix = "";
5084         else if (opt_arg(arg, '\0', "inter-hunk-context",
5085                          &options->interhunkcontext))
5086                 ;
5087         else if (!strcmp(arg, "-W"))
5088                 options->flags.funccontext = 1;
5089         else if (!strcmp(arg, "--function-context"))
5090                 options->flags.funccontext = 1;
5091         else if (!strcmp(arg, "--no-function-context"))
5092                 options->flags.funccontext = 0;
5093         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5094                 char *path = prefix_filename(prefix, optarg);
5095                 options->file = xfopen(path, "w");
5096                 options->close_file = 1;
5097                 if (options->use_color != GIT_COLOR_ALWAYS)
5098                         options->use_color = GIT_COLOR_NEVER;
5099                 free(path);
5100                 return argcount;
5101         } else
5102                 return 0;
5103         return 1;
5104 }
5105
5106 int parse_rename_score(const char **cp_p)
5107 {
5108         unsigned long num, scale;
5109         int ch, dot;
5110         const char *cp = *cp_p;
5111
5112         num = 0;
5113         scale = 1;
5114         dot = 0;
5115         for (;;) {
5116                 ch = *cp;
5117                 if ( !dot && ch == '.' ) {
5118                         scale = 1;
5119                         dot = 1;
5120                 } else if ( ch == '%' ) {
5121                         scale = dot ? scale*100 : 100;
5122                         cp++;   /* % is always at the end */
5123                         break;
5124                 } else if ( ch >= '0' && ch <= '9' ) {
5125                         if ( scale < 100000 ) {
5126                                 scale *= 10;
5127                                 num = (num*10) + (ch-'0');
5128                         }
5129                 } else {
5130                         break;
5131                 }
5132                 cp++;
5133         }
5134         *cp_p = cp;
5135
5136         /* user says num divided by scale and we say internally that
5137          * is MAX_SCORE * num / scale.
5138          */
5139         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5140 }
5141
5142 static int diff_scoreopt_parse(const char *opt)
5143 {
5144         int opt1, opt2, cmd;
5145
5146         if (*opt++ != '-')
5147                 return -1;
5148         cmd = *opt++;
5149         if (cmd == '-') {
5150                 /* convert the long-form arguments into short-form versions */
5151                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5152                         if (*opt == 0 || *opt++ == '=')
5153                                 cmd = 'B';
5154                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5155                         if (*opt == 0 || *opt++ == '=')
5156                                 cmd = 'C';
5157                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5158                         if (*opt == 0 || *opt++ == '=')
5159                                 cmd = 'M';
5160                 }
5161         }
5162         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5163                 return -1; /* that is not a -M, -C, or -B option */
5164
5165         opt1 = parse_rename_score(&opt);
5166         if (cmd != 'B')
5167                 opt2 = 0;
5168         else {
5169                 if (*opt == 0)
5170                         opt2 = 0;
5171                 else if (*opt != '/')
5172                         return -1; /* we expect -B80/99 or -B80 */
5173                 else {
5174                         opt++;
5175                         opt2 = parse_rename_score(&opt);
5176                 }
5177         }
5178         if (*opt != 0)
5179                 return -1;
5180         return opt1 | (opt2 << 16);
5181 }
5182
5183 struct diff_queue_struct diff_queued_diff;
5184
5185 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5186 {
5187         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5188         queue->queue[queue->nr++] = dp;
5189 }
5190
5191 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5192                                  struct diff_filespec *one,
5193                                  struct diff_filespec *two)
5194 {
5195         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5196         dp->one = one;
5197         dp->two = two;
5198         if (queue)
5199                 diff_q(queue, dp);
5200         return dp;
5201 }
5202
5203 void diff_free_filepair(struct diff_filepair *p)
5204 {
5205         free_filespec(p->one);
5206         free_filespec(p->two);
5207         free(p);
5208 }
5209
5210 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5211 {
5212         int abblen;
5213         const char *abbrev;
5214
5215         /* Do we want all 40 hex characters? */
5216         if (len == the_hash_algo->hexsz)
5217                 return oid_to_hex(oid);
5218
5219         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5220         abbrev = diff_abbrev_oid(oid, len);
5221
5222         if (!print_sha1_ellipsis())
5223                 return abbrev;
5224
5225         abblen = strlen(abbrev);
5226
5227         /*
5228          * In well-behaved cases, where the abbreviated result is the
5229          * same as the requested length, append three dots after the
5230          * abbreviation (hence the whole logic is limited to the case
5231          * where abblen < 37); when the actual abbreviated result is a
5232          * bit longer than the requested length, we reduce the number
5233          * of dots so that they match the well-behaved ones.  However,
5234          * if the actual abbreviation is longer than the requested
5235          * length by more than three, we give up on aligning, and add
5236          * three dots anyway, to indicate that the output is not the
5237          * full object name.  Yes, this may be suboptimal, but this
5238          * appears only in "diff --raw --abbrev" output and it is not
5239          * worth the effort to change it now.  Note that this would
5240          * likely to work fine when the automatic sizing of default
5241          * abbreviation length is used--we would be fed -1 in "len" in
5242          * that case, and will end up always appending three-dots, but
5243          * the automatic sizing is supposed to give abblen that ensures
5244          * uniqueness across all objects (statistically speaking).
5245          */
5246         if (abblen < the_hash_algo->hexsz - 3) {
5247                 static char hex[GIT_MAX_HEXSZ + 1];
5248                 if (len < abblen && abblen <= len + 2)
5249                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5250                 else
5251                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5252                 return hex;
5253         }
5254
5255         return oid_to_hex(oid);
5256 }
5257
5258 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5259 {
5260         int line_termination = opt->line_termination;
5261         int inter_name_termination = line_termination ? '\t' : '\0';
5262
5263         fprintf(opt->file, "%s", diff_line_prefix(opt));
5264         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5265                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5266                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5267                 fprintf(opt->file, "%s ",
5268                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5269         }
5270         if (p->score) {
5271                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5272                         inter_name_termination);
5273         } else {
5274                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5275         }
5276
5277         if (p->status == DIFF_STATUS_COPIED ||
5278             p->status == DIFF_STATUS_RENAMED) {
5279                 const char *name_a, *name_b;
5280                 name_a = p->one->path;
5281                 name_b = p->two->path;
5282                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5283                 write_name_quoted(name_a, opt->file, inter_name_termination);
5284                 write_name_quoted(name_b, opt->file, line_termination);
5285         } else {
5286                 const char *name_a, *name_b;
5287                 name_a = p->one->mode ? p->one->path : p->two->path;
5288                 name_b = NULL;
5289                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5290                 write_name_quoted(name_a, opt->file, line_termination);
5291         }
5292 }
5293
5294 int diff_unmodified_pair(struct diff_filepair *p)
5295 {
5296         /* This function is written stricter than necessary to support
5297          * the currently implemented transformers, but the idea is to
5298          * let transformers to produce diff_filepairs any way they want,
5299          * and filter and clean them up here before producing the output.
5300          */
5301         struct diff_filespec *one = p->one, *two = p->two;
5302
5303         if (DIFF_PAIR_UNMERGED(p))
5304                 return 0; /* unmerged is interesting */
5305
5306         /* deletion, addition, mode or type change
5307          * and rename are all interesting.
5308          */
5309         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5310             DIFF_PAIR_MODE_CHANGED(p) ||
5311             strcmp(one->path, two->path))
5312                 return 0;
5313
5314         /* both are valid and point at the same path.  that is, we are
5315          * dealing with a change.
5316          */
5317         if (one->oid_valid && two->oid_valid &&
5318             !oidcmp(&one->oid, &two->oid) &&
5319             !one->dirty_submodule && !two->dirty_submodule)
5320                 return 1; /* no change */
5321         if (!one->oid_valid && !two->oid_valid)
5322                 return 1; /* both look at the same file on the filesystem. */
5323         return 0;
5324 }
5325
5326 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5327 {
5328         if (diff_unmodified_pair(p))
5329                 return;
5330
5331         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5332             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5333                 return; /* no tree diffs in patch format */
5334
5335         run_diff(p, o);
5336 }
5337
5338 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5339                             struct diffstat_t *diffstat)
5340 {
5341         if (diff_unmodified_pair(p))
5342                 return;
5343
5344         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5345             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5346                 return; /* no useful stat for tree diffs */
5347
5348         run_diffstat(p, o, diffstat);
5349 }
5350
5351 static void diff_flush_checkdiff(struct diff_filepair *p,
5352                 struct diff_options *o)
5353 {
5354         if (diff_unmodified_pair(p))
5355                 return;
5356
5357         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5358             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5359                 return; /* nothing to check in tree diffs */
5360
5361         run_checkdiff(p, o);
5362 }
5363
5364 int diff_queue_is_empty(void)
5365 {
5366         struct diff_queue_struct *q = &diff_queued_diff;
5367         int i;
5368         for (i = 0; i < q->nr; i++)
5369                 if (!diff_unmodified_pair(q->queue[i]))
5370                         return 0;
5371         return 1;
5372 }
5373
5374 #if DIFF_DEBUG
5375 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5376 {
5377         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5378                 x, one ? one : "",
5379                 s->path,
5380                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5381                 s->mode,
5382                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5383         fprintf(stderr, "queue[%d] %s size %lu\n",
5384                 x, one ? one : "",
5385                 s->size);
5386 }
5387
5388 void diff_debug_filepair(const struct diff_filepair *p, int i)
5389 {
5390         diff_debug_filespec(p->one, i, "one");
5391         diff_debug_filespec(p->two, i, "two");
5392         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5393                 p->score, p->status ? p->status : '?',
5394                 p->one->rename_used, p->broken_pair);
5395 }
5396
5397 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5398 {
5399         int i;
5400         if (msg)
5401                 fprintf(stderr, "%s\n", msg);
5402         fprintf(stderr, "q->nr = %d\n", q->nr);
5403         for (i = 0; i < q->nr; i++) {
5404                 struct diff_filepair *p = q->queue[i];
5405                 diff_debug_filepair(p, i);
5406         }
5407 }
5408 #endif
5409
5410 static void diff_resolve_rename_copy(void)
5411 {
5412         int i;
5413         struct diff_filepair *p;
5414         struct diff_queue_struct *q = &diff_queued_diff;
5415
5416         diff_debug_queue("resolve-rename-copy", q);
5417
5418         for (i = 0; i < q->nr; i++) {
5419                 p = q->queue[i];
5420                 p->status = 0; /* undecided */
5421                 if (DIFF_PAIR_UNMERGED(p))
5422                         p->status = DIFF_STATUS_UNMERGED;
5423                 else if (!DIFF_FILE_VALID(p->one))
5424                         p->status = DIFF_STATUS_ADDED;
5425                 else if (!DIFF_FILE_VALID(p->two))
5426                         p->status = DIFF_STATUS_DELETED;
5427                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5428                         p->status = DIFF_STATUS_TYPE_CHANGED;
5429
5430                 /* from this point on, we are dealing with a pair
5431                  * whose both sides are valid and of the same type, i.e.
5432                  * either in-place edit or rename/copy edit.
5433                  */
5434                 else if (DIFF_PAIR_RENAME(p)) {
5435                         /*
5436                          * A rename might have re-connected a broken
5437                          * pair up, causing the pathnames to be the
5438                          * same again. If so, that's not a rename at
5439                          * all, just a modification..
5440                          *
5441                          * Otherwise, see if this source was used for
5442                          * multiple renames, in which case we decrement
5443                          * the count, and call it a copy.
5444                          */
5445                         if (!strcmp(p->one->path, p->two->path))
5446                                 p->status = DIFF_STATUS_MODIFIED;
5447                         else if (--p->one->rename_used > 0)
5448                                 p->status = DIFF_STATUS_COPIED;
5449                         else
5450                                 p->status = DIFF_STATUS_RENAMED;
5451                 }
5452                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5453                          p->one->mode != p->two->mode ||
5454                          p->one->dirty_submodule ||
5455                          p->two->dirty_submodule ||
5456                          is_null_oid(&p->one->oid))
5457                         p->status = DIFF_STATUS_MODIFIED;
5458                 else {
5459                         /* This is a "no-change" entry and should not
5460                          * happen anymore, but prepare for broken callers.
5461                          */
5462                         error("feeding unmodified %s to diffcore",
5463                               p->one->path);
5464                         p->status = DIFF_STATUS_UNKNOWN;
5465                 }
5466         }
5467         diff_debug_queue("resolve-rename-copy done", q);
5468 }
5469
5470 static int check_pair_status(struct diff_filepair *p)
5471 {
5472         switch (p->status) {
5473         case DIFF_STATUS_UNKNOWN:
5474                 return 0;
5475         case 0:
5476                 die("internal error in diff-resolve-rename-copy");
5477         default:
5478                 return 1;
5479         }
5480 }
5481
5482 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5483 {
5484         int fmt = opt->output_format;
5485
5486         if (fmt & DIFF_FORMAT_CHECKDIFF)
5487                 diff_flush_checkdiff(p, opt);
5488         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5489                 diff_flush_raw(p, opt);
5490         else if (fmt & DIFF_FORMAT_NAME) {
5491                 const char *name_a, *name_b;
5492                 name_a = p->two->path;
5493                 name_b = NULL;
5494                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5495                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5496                 write_name_quoted(name_a, opt->file, opt->line_termination);
5497         }
5498 }
5499
5500 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5501 {
5502         struct strbuf sb = STRBUF_INIT;
5503         if (fs->mode)
5504                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5505         else
5506                 strbuf_addf(&sb, " %s ", newdelete);
5507
5508         quote_c_style(fs->path, &sb, NULL, 0);
5509         strbuf_addch(&sb, '\n');
5510         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5511                          sb.buf, sb.len, 0);
5512         strbuf_release(&sb);
5513 }
5514
5515 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5516                 int show_name)
5517 {
5518         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5519                 struct strbuf sb = STRBUF_INIT;
5520                 strbuf_addf(&sb, " mode change %06o => %06o",
5521                             p->one->mode, p->two->mode);
5522                 if (show_name) {
5523                         strbuf_addch(&sb, ' ');
5524                         quote_c_style(p->two->path, &sb, NULL, 0);
5525                 }
5526                 strbuf_addch(&sb, '\n');
5527                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5528                                  sb.buf, sb.len, 0);
5529                 strbuf_release(&sb);
5530         }
5531 }
5532
5533 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5534                 struct diff_filepair *p)
5535 {
5536         struct strbuf sb = STRBUF_INIT;
5537         struct strbuf names = STRBUF_INIT;
5538
5539         pprint_rename(&names, p->one->path, p->two->path);
5540         strbuf_addf(&sb, " %s %s (%d%%)\n",
5541                     renamecopy, names.buf, similarity_index(p));
5542         strbuf_release(&names);
5543         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5544                                  sb.buf, sb.len, 0);
5545         show_mode_change(opt, p, 0);
5546         strbuf_release(&sb);
5547 }
5548
5549 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5550 {
5551         switch(p->status) {
5552         case DIFF_STATUS_DELETED:
5553                 show_file_mode_name(opt, "delete", p->one);
5554                 break;
5555         case DIFF_STATUS_ADDED:
5556                 show_file_mode_name(opt, "create", p->two);
5557                 break;
5558         case DIFF_STATUS_COPIED:
5559                 show_rename_copy(opt, "copy", p);
5560                 break;
5561         case DIFF_STATUS_RENAMED:
5562                 show_rename_copy(opt, "rename", p);
5563                 break;
5564         default:
5565                 if (p->score) {
5566                         struct strbuf sb = STRBUF_INIT;
5567                         strbuf_addstr(&sb, " rewrite ");
5568                         quote_c_style(p->two->path, &sb, NULL, 0);
5569                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5570                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5571                                          sb.buf, sb.len, 0);
5572                         strbuf_release(&sb);
5573                 }
5574                 show_mode_change(opt, p, !p->score);
5575                 break;
5576         }
5577 }
5578
5579 struct patch_id_t {
5580         git_SHA_CTX *ctx;
5581         int patchlen;
5582 };
5583
5584 static int remove_space(char *line, int len)
5585 {
5586         int i;
5587         char *dst = line;
5588         unsigned char c;
5589
5590         for (i = 0; i < len; i++)
5591                 if (!isspace((c = line[i])))
5592                         *dst++ = c;
5593
5594         return dst - line;
5595 }
5596
5597 static void patch_id_consume(void *priv, char *line, unsigned long len)
5598 {
5599         struct patch_id_t *data = priv;
5600         int new_len;
5601
5602         /* Ignore line numbers when computing the SHA1 of the patch */
5603         if (starts_with(line, "@@ -"))
5604                 return;
5605
5606         new_len = remove_space(line, len);
5607
5608         git_SHA1_Update(data->ctx, line, new_len);
5609         data->patchlen += new_len;
5610 }
5611
5612 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5613 {
5614         git_SHA1_Update(ctx, str, strlen(str));
5615 }
5616
5617 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5618 {
5619         /* large enough for 2^32 in octal */
5620         char buf[12];
5621         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5622         git_SHA1_Update(ctx, buf, len);
5623 }
5624
5625 /* returns 0 upon success, and writes result into sha1 */
5626 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5627 {
5628         struct diff_queue_struct *q = &diff_queued_diff;
5629         int i;
5630         git_SHA_CTX ctx;
5631         struct patch_id_t data;
5632
5633         git_SHA1_Init(&ctx);
5634         memset(&data, 0, sizeof(struct patch_id_t));
5635         data.ctx = &ctx;
5636
5637         for (i = 0; i < q->nr; i++) {
5638                 xpparam_t xpp;
5639                 xdemitconf_t xecfg;
5640                 mmfile_t mf1, mf2;
5641                 struct diff_filepair *p = q->queue[i];
5642                 int len1, len2;
5643
5644                 memset(&xpp, 0, sizeof(xpp));
5645                 memset(&xecfg, 0, sizeof(xecfg));
5646                 if (p->status == 0)
5647                         return error("internal diff status error");
5648                 if (p->status == DIFF_STATUS_UNKNOWN)
5649                         continue;
5650                 if (diff_unmodified_pair(p))
5651                         continue;
5652                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5653                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5654                         continue;
5655                 if (DIFF_PAIR_UNMERGED(p))
5656                         continue;
5657
5658                 diff_fill_oid_info(p->one);
5659                 diff_fill_oid_info(p->two);
5660
5661                 len1 = remove_space(p->one->path, strlen(p->one->path));
5662                 len2 = remove_space(p->two->path, strlen(p->two->path));
5663                 patch_id_add_string(&ctx, "diff--git");
5664                 patch_id_add_string(&ctx, "a/");
5665                 git_SHA1_Update(&ctx, p->one->path, len1);
5666                 patch_id_add_string(&ctx, "b/");
5667                 git_SHA1_Update(&ctx, p->two->path, len2);
5668
5669                 if (p->one->mode == 0) {
5670                         patch_id_add_string(&ctx, "newfilemode");
5671                         patch_id_add_mode(&ctx, p->two->mode);
5672                         patch_id_add_string(&ctx, "---/dev/null");
5673                         patch_id_add_string(&ctx, "+++b/");
5674                         git_SHA1_Update(&ctx, p->two->path, len2);
5675                 } else if (p->two->mode == 0) {
5676                         patch_id_add_string(&ctx, "deletedfilemode");
5677                         patch_id_add_mode(&ctx, p->one->mode);
5678                         patch_id_add_string(&ctx, "---a/");
5679                         git_SHA1_Update(&ctx, p->one->path, len1);
5680                         patch_id_add_string(&ctx, "+++/dev/null");
5681                 } else {
5682                         patch_id_add_string(&ctx, "---a/");
5683                         git_SHA1_Update(&ctx, p->one->path, len1);
5684                         patch_id_add_string(&ctx, "+++b/");
5685                         git_SHA1_Update(&ctx, p->two->path, len2);
5686                 }
5687
5688                 if (diff_header_only)
5689                         continue;
5690
5691                 if (fill_mmfile(&mf1, p->one) < 0 ||
5692                     fill_mmfile(&mf2, p->two) < 0)
5693                         return error("unable to read files to diff");
5694
5695                 if (diff_filespec_is_binary(p->one) ||
5696                     diff_filespec_is_binary(p->two)) {
5697                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5698                                         GIT_SHA1_HEXSZ);
5699                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5700                                         GIT_SHA1_HEXSZ);
5701                         continue;
5702                 }
5703
5704                 xpp.flags = 0;
5705                 xecfg.ctxlen = 3;
5706                 xecfg.flags = 0;
5707                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5708                                   &xpp, &xecfg))
5709                         return error("unable to generate patch-id diff for %s",
5710                                      p->one->path);
5711         }
5712
5713         git_SHA1_Final(oid->hash, &ctx);
5714         return 0;
5715 }
5716
5717 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5718 {
5719         struct diff_queue_struct *q = &diff_queued_diff;
5720         int i;
5721         int result = diff_get_patch_id(options, oid, diff_header_only);
5722
5723         for (i = 0; i < q->nr; i++)
5724                 diff_free_filepair(q->queue[i]);
5725
5726         free(q->queue);
5727         DIFF_QUEUE_CLEAR(q);
5728
5729         return result;
5730 }
5731
5732 static int is_summary_empty(const struct diff_queue_struct *q)
5733 {
5734         int i;
5735
5736         for (i = 0; i < q->nr; i++) {
5737                 const struct diff_filepair *p = q->queue[i];
5738
5739                 switch (p->status) {
5740                 case DIFF_STATUS_DELETED:
5741                 case DIFF_STATUS_ADDED:
5742                 case DIFF_STATUS_COPIED:
5743                 case DIFF_STATUS_RENAMED:
5744                         return 0;
5745                 default:
5746                         if (p->score)
5747                                 return 0;
5748                         if (p->one->mode && p->two->mode &&
5749                             p->one->mode != p->two->mode)
5750                                 return 0;
5751                         break;
5752                 }
5753         }
5754         return 1;
5755 }
5756
5757 static const char rename_limit_warning[] =
5758 N_("inexact rename detection was skipped due to too many files.");
5759
5760 static const char degrade_cc_to_c_warning[] =
5761 N_("only found copies from modified paths due to too many files.");
5762
5763 static const char rename_limit_advice[] =
5764 N_("you may want to set your %s variable to at least "
5765    "%d and retry the command.");
5766
5767 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5768 {
5769         fflush(stdout);
5770         if (degraded_cc)
5771                 warning(_(degrade_cc_to_c_warning));
5772         else if (needed)
5773                 warning(_(rename_limit_warning));
5774         else
5775                 return;
5776         if (0 < needed)
5777                 warning(_(rename_limit_advice), varname, needed);
5778 }
5779
5780 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5781 {
5782         int i;
5783         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5784         struct diff_queue_struct *q = &diff_queued_diff;
5785
5786         if (WSEH_NEW & WS_RULE_MASK)
5787                 BUG("WS rules bit mask overlaps with diff symbol flags");
5788
5789         if (o->color_moved)
5790                 o->emitted_symbols = &esm;
5791
5792         for (i = 0; i < q->nr; i++) {
5793                 struct diff_filepair *p = q->queue[i];
5794                 if (check_pair_status(p))
5795                         diff_flush_patch(p, o);
5796         }
5797
5798         if (o->emitted_symbols) {
5799                 if (o->color_moved) {
5800                         struct hashmap add_lines, del_lines;
5801
5802                         if (o->color_moved_ws_handling &
5803                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5804                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5805
5806                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5807                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5808
5809                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5810                         mark_color_as_moved(o, &add_lines, &del_lines);
5811                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5812                                 dim_moved_lines(o);
5813
5814                         hashmap_free(&add_lines, 0);
5815                         hashmap_free(&del_lines, 0);
5816                 }
5817
5818                 for (i = 0; i < esm.nr; i++)
5819                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5820
5821                 for (i = 0; i < esm.nr; i++)
5822                         free((void *)esm.buf[i].line);
5823         }
5824         esm.nr = 0;
5825 }
5826
5827 void diff_flush(struct diff_options *options)
5828 {
5829         struct diff_queue_struct *q = &diff_queued_diff;
5830         int i, output_format = options->output_format;
5831         int separator = 0;
5832         int dirstat_by_line = 0;
5833
5834         /*
5835          * Order: raw, stat, summary, patch
5836          * or:    name/name-status/checkdiff (other bits clear)
5837          */
5838         if (!q->nr)
5839                 goto free_queue;
5840
5841         if (output_format & (DIFF_FORMAT_RAW |
5842                              DIFF_FORMAT_NAME |
5843                              DIFF_FORMAT_NAME_STATUS |
5844                              DIFF_FORMAT_CHECKDIFF)) {
5845                 for (i = 0; i < q->nr; i++) {
5846                         struct diff_filepair *p = q->queue[i];
5847                         if (check_pair_status(p))
5848                                 flush_one_pair(p, options);
5849                 }
5850                 separator++;
5851         }
5852
5853         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5854                 dirstat_by_line = 1;
5855
5856         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5857             dirstat_by_line) {
5858                 struct diffstat_t diffstat;
5859
5860                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5861                 for (i = 0; i < q->nr; i++) {
5862                         struct diff_filepair *p = q->queue[i];
5863                         if (check_pair_status(p))
5864                                 diff_flush_stat(p, options, &diffstat);
5865                 }
5866                 if (output_format & DIFF_FORMAT_NUMSTAT)
5867                         show_numstat(&diffstat, options);
5868                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5869                         show_stats(&diffstat, options);
5870                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5871                         show_shortstats(&diffstat, options);
5872                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5873                         show_dirstat_by_line(&diffstat, options);
5874                 free_diffstat_info(&diffstat);
5875                 separator++;
5876         }
5877         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5878                 show_dirstat(options);
5879
5880         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5881                 for (i = 0; i < q->nr; i++) {
5882                         diff_summary(options, q->queue[i]);
5883                 }
5884                 separator++;
5885         }
5886
5887         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5888             options->flags.exit_with_status &&
5889             options->flags.diff_from_contents) {
5890                 /*
5891                  * run diff_flush_patch for the exit status. setting
5892                  * options->file to /dev/null should be safe, because we
5893                  * aren't supposed to produce any output anyway.
5894                  */
5895                 if (options->close_file)
5896                         fclose(options->file);
5897                 options->file = xfopen("/dev/null", "w");
5898                 options->close_file = 1;
5899                 options->color_moved = 0;
5900                 for (i = 0; i < q->nr; i++) {
5901                         struct diff_filepair *p = q->queue[i];
5902                         if (check_pair_status(p))
5903                                 diff_flush_patch(p, options);
5904                         if (options->found_changes)
5905                                 break;
5906                 }
5907         }
5908
5909         if (output_format & DIFF_FORMAT_PATCH) {
5910                 if (separator) {
5911                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5912                         if (options->stat_sep)
5913                                 /* attach patch instead of inline */
5914                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5915                                                  NULL, 0, 0);
5916                 }
5917
5918                 diff_flush_patch_all_file_pairs(options);
5919         }
5920
5921         if (output_format & DIFF_FORMAT_CALLBACK)
5922                 options->format_callback(q, options, options->format_callback_data);
5923
5924         for (i = 0; i < q->nr; i++)
5925                 diff_free_filepair(q->queue[i]);
5926 free_queue:
5927         free(q->queue);
5928         DIFF_QUEUE_CLEAR(q);
5929         if (options->close_file)
5930                 fclose(options->file);
5931
5932         /*
5933          * Report the content-level differences with HAS_CHANGES;
5934          * diff_addremove/diff_change does not set the bit when
5935          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5936          */
5937         if (options->flags.diff_from_contents) {
5938                 if (options->found_changes)
5939                         options->flags.has_changes = 1;
5940                 else
5941                         options->flags.has_changes = 0;
5942         }
5943 }
5944
5945 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5946 {
5947         return (((p->status == DIFF_STATUS_MODIFIED) &&
5948                  ((p->score &&
5949                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5950                   (!p->score &&
5951                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5952                 ((p->status != DIFF_STATUS_MODIFIED) &&
5953                  filter_bit_tst(p->status, options)));
5954 }
5955
5956 static void diffcore_apply_filter(struct diff_options *options)
5957 {
5958         int i;
5959         struct diff_queue_struct *q = &diff_queued_diff;
5960         struct diff_queue_struct outq;
5961
5962         DIFF_QUEUE_CLEAR(&outq);
5963
5964         if (!options->filter)
5965                 return;
5966
5967         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5968                 int found;
5969                 for (i = found = 0; !found && i < q->nr; i++) {
5970                         if (match_filter(options, q->queue[i]))
5971                                 found++;
5972                 }
5973                 if (found)
5974                         return;
5975
5976                 /* otherwise we will clear the whole queue
5977                  * by copying the empty outq at the end of this
5978                  * function, but first clear the current entries
5979                  * in the queue.
5980                  */
5981                 for (i = 0; i < q->nr; i++)
5982                         diff_free_filepair(q->queue[i]);
5983         }
5984         else {
5985                 /* Only the matching ones */
5986                 for (i = 0; i < q->nr; i++) {
5987                         struct diff_filepair *p = q->queue[i];
5988                         if (match_filter(options, p))
5989                                 diff_q(&outq, p);
5990                         else
5991                                 diff_free_filepair(p);
5992                 }
5993         }
5994         free(q->queue);
5995         *q = outq;
5996 }
5997
5998 /* Check whether two filespecs with the same mode and size are identical */
5999 static int diff_filespec_is_identical(struct diff_filespec *one,
6000                                       struct diff_filespec *two)
6001 {
6002         if (S_ISGITLINK(one->mode))
6003                 return 0;
6004         if (diff_populate_filespec(one, 0))
6005                 return 0;
6006         if (diff_populate_filespec(two, 0))
6007                 return 0;
6008         return !memcmp(one->data, two->data, one->size);
6009 }
6010
6011 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
6012 {
6013         if (p->done_skip_stat_unmatch)
6014                 return p->skip_stat_unmatch_result;
6015
6016         p->done_skip_stat_unmatch = 1;
6017         p->skip_stat_unmatch_result = 0;
6018         /*
6019          * 1. Entries that come from stat info dirtiness
6020          *    always have both sides (iow, not create/delete),
6021          *    one side of the object name is unknown, with
6022          *    the same mode and size.  Keep the ones that
6023          *    do not match these criteria.  They have real
6024          *    differences.
6025          *
6026          * 2. At this point, the file is known to be modified,
6027          *    with the same mode and size, and the object
6028          *    name of one side is unknown.  Need to inspect
6029          *    the identical contents.
6030          */
6031         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6032             !DIFF_FILE_VALID(p->two) ||
6033             (p->one->oid_valid && p->two->oid_valid) ||
6034             (p->one->mode != p->two->mode) ||
6035             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
6036             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
6037             (p->one->size != p->two->size) ||
6038             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
6039                 p->skip_stat_unmatch_result = 1;
6040         return p->skip_stat_unmatch_result;
6041 }
6042
6043 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6044 {
6045         int i;
6046         struct diff_queue_struct *q = &diff_queued_diff;
6047         struct diff_queue_struct outq;
6048         DIFF_QUEUE_CLEAR(&outq);
6049
6050         for (i = 0; i < q->nr; i++) {
6051                 struct diff_filepair *p = q->queue[i];
6052
6053                 if (diff_filespec_check_stat_unmatch(p))
6054                         diff_q(&outq, p);
6055                 else {
6056                         /*
6057                          * The caller can subtract 1 from skip_stat_unmatch
6058                          * to determine how many paths were dirty only
6059                          * due to stat info mismatch.
6060                          */
6061                         if (!diffopt->flags.no_index)
6062                                 diffopt->skip_stat_unmatch++;
6063                         diff_free_filepair(p);
6064                 }
6065         }
6066         free(q->queue);
6067         *q = outq;
6068 }
6069
6070 static int diffnamecmp(const void *a_, const void *b_)
6071 {
6072         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6073         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6074         const char *name_a, *name_b;
6075
6076         name_a = a->one ? a->one->path : a->two->path;
6077         name_b = b->one ? b->one->path : b->two->path;
6078         return strcmp(name_a, name_b);
6079 }
6080
6081 void diffcore_fix_diff_index(struct diff_options *options)
6082 {
6083         struct diff_queue_struct *q = &diff_queued_diff;
6084         QSORT(q->queue, q->nr, diffnamecmp);
6085 }
6086
6087 void diffcore_std(struct diff_options *options)
6088 {
6089         /* NOTE please keep the following in sync with diff_tree_combined() */
6090         if (options->skip_stat_unmatch)
6091                 diffcore_skip_stat_unmatch(options);
6092         if (!options->found_follow) {
6093                 /* See try_to_follow_renames() in tree-diff.c */
6094                 if (options->break_opt != -1)
6095                         diffcore_break(options->break_opt);
6096                 if (options->detect_rename)
6097                         diffcore_rename(options);
6098                 if (options->break_opt != -1)
6099                         diffcore_merge_broken();
6100         }
6101         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6102                 diffcore_pickaxe(options);
6103         if (options->orderfile)
6104                 diffcore_order(options->orderfile);
6105         if (!options->found_follow)
6106                 /* See try_to_follow_renames() in tree-diff.c */
6107                 diff_resolve_rename_copy();
6108         diffcore_apply_filter(options);
6109
6110         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6111                 options->flags.has_changes = 1;
6112         else
6113                 options->flags.has_changes = 0;
6114
6115         options->found_follow = 0;
6116 }
6117
6118 int diff_result_code(struct diff_options *opt, int status)
6119 {
6120         int result = 0;
6121
6122         diff_warn_rename_limit("diff.renameLimit",
6123                                opt->needed_rename_limit,
6124                                opt->degraded_cc_to_c);
6125         if (!opt->flags.exit_with_status &&
6126             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6127                 return status;
6128         if (opt->flags.exit_with_status &&
6129             opt->flags.has_changes)
6130                 result |= 01;
6131         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6132             opt->flags.check_failed)
6133                 result |= 02;
6134         return result;
6135 }
6136
6137 int diff_can_quit_early(struct diff_options *opt)
6138 {
6139         return (opt->flags.quick &&
6140                 !opt->filter &&
6141                 opt->flags.has_changes);
6142 }
6143
6144 /*
6145  * Shall changes to this submodule be ignored?
6146  *
6147  * Submodule changes can be configured to be ignored separately for each path,
6148  * but that configuration can be overridden from the command line.
6149  */
6150 static int is_submodule_ignored(const char *path, struct diff_options *options)
6151 {
6152         int ignored = 0;
6153         struct diff_flags orig_flags = options->flags;
6154         if (!options->flags.override_submodule_config)
6155                 set_diffopt_flags_from_submodule_config(options, path);
6156         if (options->flags.ignore_submodules)
6157                 ignored = 1;
6158         options->flags = orig_flags;
6159         return ignored;
6160 }
6161
6162 void diff_addremove(struct diff_options *options,
6163                     int addremove, unsigned mode,
6164                     const struct object_id *oid,
6165                     int oid_valid,
6166                     const char *concatpath, unsigned dirty_submodule)
6167 {
6168         struct diff_filespec *one, *two;
6169
6170         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6171                 return;
6172
6173         /* This may look odd, but it is a preparation for
6174          * feeding "there are unchanged files which should
6175          * not produce diffs, but when you are doing copy
6176          * detection you would need them, so here they are"
6177          * entries to the diff-core.  They will be prefixed
6178          * with something like '=' or '*' (I haven't decided
6179          * which but should not make any difference).
6180          * Feeding the same new and old to diff_change()
6181          * also has the same effect.
6182          * Before the final output happens, they are pruned after
6183          * merged into rename/copy pairs as appropriate.
6184          */
6185         if (options->flags.reverse_diff)
6186                 addremove = (addremove == '+' ? '-' :
6187                              addremove == '-' ? '+' : addremove);
6188
6189         if (options->prefix &&
6190             strncmp(concatpath, options->prefix, options->prefix_length))
6191                 return;
6192
6193         one = alloc_filespec(concatpath);
6194         two = alloc_filespec(concatpath);
6195
6196         if (addremove != '+')
6197                 fill_filespec(one, oid, oid_valid, mode);
6198         if (addremove != '-') {
6199                 fill_filespec(two, oid, oid_valid, mode);
6200                 two->dirty_submodule = dirty_submodule;
6201         }
6202
6203         diff_queue(&diff_queued_diff, one, two);
6204         if (!options->flags.diff_from_contents)
6205                 options->flags.has_changes = 1;
6206 }
6207
6208 void diff_change(struct diff_options *options,
6209                  unsigned old_mode, unsigned new_mode,
6210                  const struct object_id *old_oid,
6211                  const struct object_id *new_oid,
6212                  int old_oid_valid, int new_oid_valid,
6213                  const char *concatpath,
6214                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6215 {
6216         struct diff_filespec *one, *two;
6217         struct diff_filepair *p;
6218
6219         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6220             is_submodule_ignored(concatpath, options))
6221                 return;
6222
6223         if (options->flags.reverse_diff) {
6224                 SWAP(old_mode, new_mode);
6225                 SWAP(old_oid, new_oid);
6226                 SWAP(old_oid_valid, new_oid_valid);
6227                 SWAP(old_dirty_submodule, new_dirty_submodule);
6228         }
6229
6230         if (options->prefix &&
6231             strncmp(concatpath, options->prefix, options->prefix_length))
6232                 return;
6233
6234         one = alloc_filespec(concatpath);
6235         two = alloc_filespec(concatpath);
6236         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6237         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6238         one->dirty_submodule = old_dirty_submodule;
6239         two->dirty_submodule = new_dirty_submodule;
6240         p = diff_queue(&diff_queued_diff, one, two);
6241
6242         if (options->flags.diff_from_contents)
6243                 return;
6244
6245         if (options->flags.quick && options->skip_stat_unmatch &&
6246             !diff_filespec_check_stat_unmatch(p))
6247                 return;
6248
6249         options->flags.has_changes = 1;
6250 }
6251
6252 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6253 {
6254         struct diff_filepair *pair;
6255         struct diff_filespec *one, *two;
6256
6257         if (options->prefix &&
6258             strncmp(path, options->prefix, options->prefix_length))
6259                 return NULL;
6260
6261         one = alloc_filespec(path);
6262         two = alloc_filespec(path);
6263         pair = diff_queue(&diff_queued_diff, one, two);
6264         pair->is_unmerged = 1;
6265         return pair;
6266 }
6267
6268 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6269                 size_t *outsize)
6270 {
6271         struct diff_tempfile *temp;
6272         const char *argv[3];
6273         const char **arg = argv;
6274         struct child_process child = CHILD_PROCESS_INIT;
6275         struct strbuf buf = STRBUF_INIT;
6276         int err = 0;
6277
6278         temp = prepare_temp_file(spec->path, spec);
6279         *arg++ = pgm;
6280         *arg++ = temp->name;
6281         *arg = NULL;
6282
6283         child.use_shell = 1;
6284         child.argv = argv;
6285         child.out = -1;
6286         if (start_command(&child)) {
6287                 remove_tempfile();
6288                 return NULL;
6289         }
6290
6291         if (strbuf_read(&buf, child.out, 0) < 0)
6292                 err = error("error reading from textconv command '%s'", pgm);
6293         close(child.out);
6294
6295         if (finish_command(&child) || err) {
6296                 strbuf_release(&buf);
6297                 remove_tempfile();
6298                 return NULL;
6299         }
6300         remove_tempfile();
6301
6302         return strbuf_detach(&buf, outsize);
6303 }
6304
6305 size_t fill_textconv(struct userdiff_driver *driver,
6306                      struct diff_filespec *df,
6307                      char **outbuf)
6308 {
6309         size_t size;
6310
6311         if (!driver) {
6312                 if (!DIFF_FILE_VALID(df)) {
6313                         *outbuf = "";
6314                         return 0;
6315                 }
6316                 if (diff_populate_filespec(df, 0))
6317                         die("unable to read files to diff");
6318                 *outbuf = df->data;
6319                 return df->size;
6320         }
6321
6322         if (!driver->textconv)
6323                 BUG("fill_textconv called with non-textconv driver");
6324
6325         if (driver->textconv_cache && df->oid_valid) {
6326                 *outbuf = notes_cache_get(driver->textconv_cache,
6327                                           &df->oid,
6328                                           &size);
6329                 if (*outbuf)
6330                         return size;
6331         }
6332
6333         *outbuf = run_textconv(driver->textconv, df, &size);
6334         if (!*outbuf)
6335                 die("unable to read files to diff");
6336
6337         if (driver->textconv_cache && df->oid_valid) {
6338                 /* ignore errors, as we might be in a readonly repository */
6339                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6340                                 size);
6341                 /*
6342                  * we could save up changes and flush them all at the end,
6343                  * but we would need an extra call after all diffing is done.
6344                  * Since generating a cache entry is the slow path anyway,
6345                  * this extra overhead probably isn't a big deal.
6346                  */
6347                 notes_cache_write(driver->textconv_cache);
6348         }
6349
6350         return size;
6351 }
6352
6353 int textconv_object(const char *path,
6354                     unsigned mode,
6355                     const struct object_id *oid,
6356                     int oid_valid,
6357                     char **buf,
6358                     unsigned long *buf_size)
6359 {
6360         struct diff_filespec *df;
6361         struct userdiff_driver *textconv;
6362
6363         df = alloc_filespec(path);
6364         fill_filespec(df, oid, oid_valid, mode);
6365         textconv = get_textconv(df);
6366         if (!textconv) {
6367                 free_filespec(df);
6368                 return 0;
6369         }
6370
6371         *buf_size = fill_textconv(textconv, df, buf);
6372         free_filespec(df);
6373         return 1;
6374 }
6375
6376 void setup_diff_pager(struct diff_options *opt)
6377 {
6378         /*
6379          * If the user asked for our exit code, then either they want --quiet
6380          * or --exit-code. We should definitely not bother with a pager in the
6381          * former case, as we will generate no output. Since we still properly
6382          * report our exit code even when a pager is run, we _could_ run a
6383          * pager with --exit-code. But since we have not done so historically,
6384          * and because it is easy to find people oneline advising "git diff
6385          * --exit-code" in hooks and other scripts, we do not do so.
6386          */
6387         if (!opt->flags.exit_with_status &&
6388             check_pager_config("diff") != 0)
6389                 setup_pager();
6390 }