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