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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed_zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else
 289                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed_zebra', 'plain'"));
 290 }
 291
 292 static int parse_color_moved_ws(const char *arg)
 293 {
 294         int ret = 0;
 295         struct string_list l = STRING_LIST_INIT_DUP;
 296         struct string_list_item *i;
 297
 298         string_list_split(&l, arg, ',', -1);
 299
 300         for_each_string_list_item(i, &l) {
 301                 struct strbuf sb = STRBUF_INIT;
 302                 strbuf_addstr(&sb, i->string);
 303                 strbuf_trim(&sb);
 304
 305                 if (!strcmp(sb.buf, "ignore-space-change"))
 306                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 307                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 308                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 309                 else if (!strcmp(sb.buf, "ignore-all-space"))
 310                         ret |= XDF_IGNORE_WHITESPACE;
 311                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 312                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 313                 else
 314                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 315
 316                 strbuf_release(&sb);
 317         }
 318
 319         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 320             (ret & XDF_WHITESPACE_FLAGS))
 321                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 322
 323         string_list_clear(&l, 0);
 324
 325         return ret;
 326 }
 327
 328 int git_diff_ui_config(const char *var, const char *value, void *cb)
 329 {
 330         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 331                 diff_use_color_default = git_config_colorbool(var, value);
 332                 return 0;
 333         }
 334         if (!strcmp(var, "diff.colormoved")) {
 335                 int cm = parse_color_moved(value);
 336                 if (cm < 0)
 337                         return -1;
 338                 diff_color_moved_default = cm;
 339                 return 0;
 340         }
 341         if (!strcmp(var, "diff.colormovedws")) {
 342                 int cm = parse_color_moved_ws(value);
 343                 if (cm < 0)
 344                         return -1;
 345                 diff_color_moved_ws_default = cm;
 346                 return 0;
 347         }
 348         if (!strcmp(var, "diff.context")) {
 349                 diff_context_default = git_config_int(var, value);
 350                 if (diff_context_default < 0)
 351                         return -1;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.interhunkcontext")) {
 355                 diff_interhunk_context_default = git_config_int(var, value);
 356                 if (diff_interhunk_context_default < 0)
 357                         return -1;
 358                 return 0;
 359         }
 360         if (!strcmp(var, "diff.renames")) {
 361                 diff_detect_rename_default = git_config_rename(var, value);
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.autorefreshindex")) {
 365                 diff_auto_refresh_index = git_config_bool(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.mnemonicprefix")) {
 369                 diff_mnemonic_prefix = git_config_bool(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.noprefix")) {
 373                 diff_no_prefix = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.statgraphwidth")) {
 377                 diff_stat_graph_width = git_config_int(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.external"))
 381                 return git_config_string(&external_diff_cmd_cfg, var, value);
 382         if (!strcmp(var, "diff.wordregex"))
 383                 return git_config_string(&diff_word_regex_cfg, var, value);
 384         if (!strcmp(var, "diff.orderfile"))
 385                 return git_config_pathname(&diff_order_file_cfg, var, value);
 386
 387         if (!strcmp(var, "diff.ignoresubmodules"))
 388                 handle_ignore_submodules_arg(&default_diff_options, value);
 389
 390         if (!strcmp(var, "diff.submodule")) {
 391                 if (parse_submodule_params(&default_diff_options, value))
 392                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 393                                 value);
 394                 return 0;
 395         }
 396
 397         if (!strcmp(var, "diff.algorithm")) {
 398                 diff_algorithm = parse_algorithm_value(value);
 399                 if (diff_algorithm < 0)
 400                         return -1;
 401                 return 0;
 402         }
 403
 404         if (!strcmp(var, "diff.wserrorhighlight")) {
 405                 int val = parse_ws_error_highlight(value);
 406                 if (val < 0)
 407                         return -1;
 408                 ws_error_highlight_default = val;
 409                 return 0;
 410         }
 411
 412         if (git_color_config(var, value, cb) < 0)
 413                 return -1;
 414
 415         return git_diff_basic_config(var, value, cb);
 416 }
 417
 418 int git_diff_basic_config(const char *var, const char *value, void *cb)
 419 {
 420         const char *name;
 421
 422         if (!strcmp(var, "diff.renamelimit")) {
 423                 diff_rename_limit_default = git_config_int(var, value);
 424                 return 0;
 425         }
 426
 427         if (userdiff_config(var, value) < 0)
 428                 return -1;
 429
 430         if (skip_prefix(var, "diff.color.", &name) ||
 431             skip_prefix(var, "color.diff.", &name)) {
 432                 int slot = parse_diff_color_slot(name);
 433                 if (slot < 0)
 434                         return 0;
 435                 if (!value)
 436                         return config_error_nonbool(var);
 437                 return color_parse(value, diff_colors[slot]);
 438         }
 439
 440         /* like GNU diff's --suppress-blank-empty option  */
 441         if (!strcmp(var, "diff.suppressblankempty") ||
 442                         /* for backwards compatibility */
 443                         !strcmp(var, "diff.suppress-blank-empty")) {
 444                 diff_suppress_blank_empty = git_config_bool(var, value);
 445                 return 0;
 446         }
 447
 448         if (!strcmp(var, "diff.dirstat")) {
 449                 struct strbuf errmsg = STRBUF_INIT;
 450                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 451                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 452                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 453                                 errmsg.buf);
 454                 strbuf_release(&errmsg);
 455                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 456                 return 0;
 457         }
 458
 459         if (git_diff_heuristic_config(var, value, cb) < 0)
 460                 return -1;
 461
 462         return git_default_config(var, value, cb);
 463 }
 464
 465 static char *quote_two(const char *one, const char *two)
 466 {
 467         int need_one = quote_c_style(one, NULL, NULL, 1);
 468         int need_two = quote_c_style(two, NULL, NULL, 1);
 469         struct strbuf res = STRBUF_INIT;
 470
 471         if (need_one + need_two) {
 472                 strbuf_addch(&res, '"');
 473                 quote_c_style(one, &res, NULL, 1);
 474                 quote_c_style(two, &res, NULL, 1);
 475                 strbuf_addch(&res, '"');
 476         } else {
 477                 strbuf_addstr(&res, one);
 478                 strbuf_addstr(&res, two);
 479         }
 480         return strbuf_detach(&res, NULL);
 481 }
 482
 483 static const char *external_diff(void)
 484 {
 485         static const char *external_diff_cmd = NULL;
 486         static int done_preparing = 0;
 487
 488         if (done_preparing)
 489                 return external_diff_cmd;
 490         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 491         if (!external_diff_cmd)
 492                 external_diff_cmd = external_diff_cmd_cfg;
 493         done_preparing = 1;
 494         return external_diff_cmd;
 495 }
 496
 497 /*
 498  * Keep track of files used for diffing. Sometimes such an entry
 499  * refers to a temporary file, sometimes to an existing file, and
 500  * sometimes to "/dev/null".
 501  */
 502 static struct diff_tempfile {
 503         /*
 504          * filename external diff should read from, or NULL if this
 505          * entry is currently not in use:
 506          */
 507         const char *name;
 508
 509         char hex[GIT_MAX_HEXSZ + 1];
 510         char mode[10];
 511
 512         /*
 513          * If this diff_tempfile instance refers to a temporary file,
 514          * this tempfile object is used to manage its lifetime.
 515          */
 516         struct tempfile *tempfile;
 517 } diff_temp[2];
 518
 519 struct emit_callback {
 520         int color_diff;
 521         unsigned ws_rule;
 522         int blank_at_eof_in_preimage;
 523         int blank_at_eof_in_postimage;
 524         int lno_in_preimage;
 525         int lno_in_postimage;
 526         const char **label_path;
 527         struct diff_words_data *diff_words;
 528         struct diff_options *opt;
 529         struct strbuf *header;
 530 };
 531
 532 static int count_lines(const char *data, int size)
 533 {
 534         int count, ch, completely_empty = 1, nl_just_seen = 0;
 535         count = 0;
 536         while (0 < size--) {
 537                 ch = *data++;
 538                 if (ch == '\n') {
 539                         count++;
 540                         nl_just_seen = 1;
 541                         completely_empty = 0;
 542                 }
 543                 else {
 544                         nl_just_seen = 0;
 545                         completely_empty = 0;
 546                 }
 547         }
 548         if (completely_empty)
 549                 return 0;
 550         if (!nl_just_seen)
 551                 count++; /* no trailing newline */
 552         return count;
 553 }
 554
 555 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 556 {
 557         if (!DIFF_FILE_VALID(one)) {
 558                 mf->ptr = (char *)""; /* does not matter */
 559                 mf->size = 0;
 560                 return 0;
 561         }
 562         else if (diff_populate_filespec(one, 0))
 563                 return -1;
 564
 565         mf->ptr = one->data;
 566         mf->size = one->size;
 567         return 0;
 568 }
 569
 570 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 571 static unsigned long diff_filespec_size(struct diff_filespec *one)
 572 {
 573         if (!DIFF_FILE_VALID(one))
 574                 return 0;
 575         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 576         return one->size;
 577 }
 578
 579 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 580 {
 581         char *ptr = mf->ptr;
 582         long size = mf->size;
 583         int cnt = 0;
 584
 585         if (!size)
 586                 return cnt;
 587         ptr += size - 1; /* pointing at the very end */
 588         if (*ptr != '\n')
 589                 ; /* incomplete line */
 590         else
 591                 ptr--; /* skip the last LF */
 592         while (mf->ptr < ptr) {
 593                 char *prev_eol;
 594                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 595                         if (*prev_eol == '\n')
 596                                 break;
 597                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 598                         break;
 599                 cnt++;
 600                 ptr = prev_eol - 1;
 601         }
 602         return cnt;
 603 }
 604
 605 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 606                                struct emit_callback *ecbdata)
 607 {
 608         int l1, l2, at;
 609         unsigned ws_rule = ecbdata->ws_rule;
 610         l1 = count_trailing_blank(mf1, ws_rule);
 611         l2 = count_trailing_blank(mf2, ws_rule);
 612         if (l2 <= l1) {
 613                 ecbdata->blank_at_eof_in_preimage = 0;
 614                 ecbdata->blank_at_eof_in_postimage = 0;
 615                 return;
 616         }
 617         at = count_lines(mf1->ptr, mf1->size);
 618         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 619
 620         at = count_lines(mf2->ptr, mf2->size);
 621         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 622 }
 623
 624 static void emit_line_0(struct diff_options *o,
 625                         const char *set, unsigned reverse, const char *reset,
 626                         int first, const char *line, int len)
 627 {
 628         int has_trailing_newline, has_trailing_carriage_return;
 629         int nofirst;
 630         FILE *file = o->file;
 631
 632         if (first)
 633                 fputs(diff_line_prefix(o), file);
 634         else if (!len)
 635                 return;
 636
 637         if (len == 0) {
 638                 has_trailing_newline = (first == '\n');
 639                 has_trailing_carriage_return = (!has_trailing_newline &&
 640                                                 (first == '\r'));
 641                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 642         } else {
 643                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 644                 if (has_trailing_newline)
 645                         len--;
 646                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 647                 if (has_trailing_carriage_return)
 648                         len--;
 649                 nofirst = 0;
 650         }
 651
 652         if (len || !nofirst) {
 653                 if (reverse && want_color(o->use_color))
 654                         fputs(GIT_COLOR_REVERSE, file);
 655                 fputs(set, file);
 656                 if (first && !nofirst)
 657                         fputc(first, file);
 658                 fwrite(line, len, 1, file);
 659                 fputs(reset, file);
 660         }
 661         if (has_trailing_carriage_return)
 662                 fputc('\r', file);
 663         if (has_trailing_newline)
 664                 fputc('\n', file);
 665 }
 666
 667 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 668                       const char *line, int len)
 669 {
 670         emit_line_0(o, set, 0, reset, line[0], line+1, len-1);
 671 }
 672
 673 enum diff_symbol {
 674         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 675         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 676         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 677         DIFF_SYMBOL_BINARY_DIFF_BODY,
 678         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 679         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 680         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 681         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 682         DIFF_SYMBOL_STATS_LINE,
 683         DIFF_SYMBOL_WORD_DIFF,
 684         DIFF_SYMBOL_STAT_SEP,
 685         DIFF_SYMBOL_SUMMARY,
 686         DIFF_SYMBOL_SUBMODULE_ADD,
 687         DIFF_SYMBOL_SUBMODULE_DEL,
 688         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 689         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 690         DIFF_SYMBOL_SUBMODULE_HEADER,
 691         DIFF_SYMBOL_SUBMODULE_ERROR,
 692         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 693         DIFF_SYMBOL_REWRITE_DIFF,
 694         DIFF_SYMBOL_BINARY_FILES,
 695         DIFF_SYMBOL_HEADER,
 696         DIFF_SYMBOL_FILEPAIR_PLUS,
 697         DIFF_SYMBOL_FILEPAIR_MINUS,
 698         DIFF_SYMBOL_WORDS_PORCELAIN,
 699         DIFF_SYMBOL_WORDS,
 700         DIFF_SYMBOL_CONTEXT,
 701         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 702         DIFF_SYMBOL_PLUS,
 703         DIFF_SYMBOL_MINUS,
 704         DIFF_SYMBOL_NO_LF_EOF,
 705         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 706         DIFF_SYMBOL_CONTEXT_MARKER,
 707         DIFF_SYMBOL_SEPARATOR
 708 };
 709 /*
 710  * Flags for content lines:
 711  * 0..12 are whitespace rules
 712  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 713  * 16 is marking if the line is blank at EOF
 714  */
 715 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 716 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 717 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 718 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 719 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 720
 721 /*
 722  * This struct is used when we need to buffer the output of the diff output.
 723  *
 724  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 725  * into the pre/post image file. This pointer could be a union with the
 726  * line pointer. By storing an offset into the file instead of the literal line,
 727  * we can decrease the memory footprint for the buffered output. At first we
 728  * may want to only have indirection for the content lines, but we could also
 729  * enhance the state for emitting prefabricated lines, e.g. the similarity
 730  * score line or hunk/file headers would only need to store a number or path
 731  * and then the output can be constructed later on depending on state.
 732  */
 733 struct emitted_diff_symbol {
 734         const char *line;
 735         int len;
 736         int flags;
 737         enum diff_symbol s;
 738 };
 739 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 740
 741 struct emitted_diff_symbols {
 742         struct emitted_diff_symbol *buf;
 743         int nr, alloc;
 744 };
 745 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 746
 747 static void append_emitted_diff_symbol(struct diff_options *o,
 748                                        struct emitted_diff_symbol *e)
 749 {
 750         struct emitted_diff_symbol *f;
 751
 752         ALLOC_GROW(o->emitted_symbols->buf,
 753                    o->emitted_symbols->nr + 1,
 754                    o->emitted_symbols->alloc);
 755         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 756
 757         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 758         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 759 }
 760
 761 struct moved_entry {
 762         struct hashmap_entry ent;
 763         const struct emitted_diff_symbol *es;
 764         struct moved_entry *next_line;
 765         struct ws_delta *wsd;
 766 };
 767
 768 /**
 769  * The struct ws_delta holds white space differences between moved lines, i.e.
 770  * between '+' and '-' lines that have been detected to be a move.
 771  * The string contains the difference in leading white spaces, before the
 772  * rest of the line is compared using the white space config for move
 773  * coloring. The current_longer indicates if the first string in the
 774  * comparision is longer than the second.
 775  */
 776 struct ws_delta {
 777         char *string;
 778         unsigned int current_longer : 1;
 779 };
 780 #define WS_DELTA_INIT { NULL, 0 }
 781
 782 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 783                              const struct emitted_diff_symbol *b,
 784                              struct ws_delta *out)
 785 {
 786         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 787         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 788         int d = longer->len - shorter->len;
 789
 790         out->string = xmemdupz(longer->line, d);
 791         out->current_longer = (a == longer);
 792
 793         return !strncmp(longer->line + d, shorter->line, shorter->len);
 794 }
 795
 796 static int cmp_in_block_with_wsd(const struct diff_options *o,
 797                                  const struct moved_entry *cur,
 798                                  const struct moved_entry *match,
 799                                  struct moved_entry *pmb,
 800                                  int n)
 801 {
 802         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 803         int al = cur->es->len, cl = l->len;
 804         const char *a = cur->es->line,
 805                    *b = match->es->line,
 806                    *c = l->line;
 807
 808         int wslen;
 809
 810         /*
 811          * We need to check if 'cur' is equal to 'match'.
 812          * As those are from the same (+/-) side, we do not need to adjust for
 813          * indent changes. However these were found using fuzzy matching
 814          * so we do have to check if they are equal.
 815          */
 816         if (strcmp(a, b))
 817                 return 1;
 818
 819         if (!pmb->wsd)
 820                 /*
 821                  * No white space delta was carried forward? This can happen
 822                  * when we exit early in this function and do not carry
 823                  * forward ws.
 824                  */
 825                 return 1;
 826
 827         /*
 828          * The indent changes of the block are known and carried forward in
 829          * pmb->wsd; however we need to check if the indent changes of the
 830          * current line are still the same as before.
 831          *
 832          * To do so we need to compare 'l' to 'cur', adjusting the
 833          * one of them for the white spaces, depending which was longer.
 834          */
 835
 836         wslen = strlen(pmb->wsd->string);
 837         if (pmb->wsd->current_longer) {
 838                 c += wslen;
 839                 cl -= wslen;
 840         } else {
 841                 a += wslen;
 842                 al -= wslen;
 843         }
 844
 845         if (strcmp(a, c))
 846                 return 1;
 847
 848         return 0;
 849 }
 850
 851 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 852                            const void *entry,
 853                            const void *entry_or_key,
 854                            const void *keydata)
 855 {
 856         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 857         const struct moved_entry *a = entry;
 858         const struct moved_entry *b = entry_or_key;
 859         unsigned flags = diffopt->color_moved_ws_handling
 860                          & XDF_WHITESPACE_FLAGS;
 861
 862         if (diffopt->color_moved_ws_handling &
 863             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 864                 /*
 865                  * As there is not specific white space config given,
 866                  * we'd need to check for a new block, so ignore all
 867                  * white space. The setup of the white space
 868                  * configuration for the next block is done else where
 869                  */
 870                 flags |= XDF_IGNORE_WHITESPACE;
 871
 872         return !xdiff_compare_lines(a->es->line, a->es->len,
 873                                     b->es->line, b->es->len,
 874                                     flags);
 875 }
 876
 877 static struct moved_entry *prepare_entry(struct diff_options *o,
 878                                          int line_no)
 879 {
 880         struct moved_entry *ret = xmalloc(sizeof(*ret));
 881         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 882         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 883
 884         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 885         ret->es = l;
 886         ret->next_line = NULL;
 887         ret->wsd = NULL;
 888
 889         return ret;
 890 }
 891
 892 static void add_lines_to_move_detection(struct diff_options *o,
 893                                         struct hashmap *add_lines,
 894                                         struct hashmap *del_lines)
 895 {
 896         struct moved_entry *prev_line = NULL;
 897
 898         int n;
 899         for (n = 0; n < o->emitted_symbols->nr; n++) {
 900                 struct hashmap *hm;
 901                 struct moved_entry *key;
 902
 903                 switch (o->emitted_symbols->buf[n].s) {
 904                 case DIFF_SYMBOL_PLUS:
 905                         hm = add_lines;
 906                         break;
 907                 case DIFF_SYMBOL_MINUS:
 908                         hm = del_lines;
 909                         break;
 910                 default:
 911                         prev_line = NULL;
 912                         continue;
 913                 }
 914
 915                 key = prepare_entry(o, n);
 916                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 917                         prev_line->next_line = key;
 918
 919                 hashmap_add(hm, key);
 920                 prev_line = key;
 921         }
 922 }
 923
 924 static void pmb_advance_or_null(struct diff_options *o,
 925                                 struct moved_entry *match,
 926                                 struct hashmap *hm,
 927                                 struct moved_entry **pmb,
 928                                 int pmb_nr)
 929 {
 930         int i;
 931         for (i = 0; i < pmb_nr; i++) {
 932                 struct moved_entry *prev = pmb[i];
 933                 struct moved_entry *cur = (prev && prev->next_line) ?
 934                                 prev->next_line : NULL;
 935                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 936                         pmb[i] = cur;
 937                 } else {
 938                         pmb[i] = NULL;
 939                 }
 940         }
 941 }
 942
 943 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 944                                             struct moved_entry *match,
 945                                             struct hashmap *hm,
 946                                             struct moved_entry **pmb,
 947                                             int pmb_nr, int n)
 948 {
 949         int i;
 950         char *got_match = xcalloc(1, pmb_nr);
 951
 952         for (; match; match = hashmap_get_next(hm, match)) {
 953                 for (i = 0; i < pmb_nr; i++) {
 954                         struct moved_entry *prev = pmb[i];
 955                         struct moved_entry *cur = (prev && prev->next_line) ?
 956                                         prev->next_line : NULL;
 957                         if (!cur)
 958                                 continue;
 959                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 960                                 got_match[i] |= 1;
 961                 }
 962         }
 963
 964         for (i = 0; i < pmb_nr; i++) {
 965                 if (got_match[i]) {
 966                         /* Carry the white space delta forward */
 967                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 968                         pmb[i] = pmb[i]->next_line;
 969                 } else
 970                         pmb[i] = NULL;
 971         }
 972 }
 973
 974 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 975                                          int pmb_nr)
 976 {
 977         int lp, rp;
 978
 979         /* Shrink the set of potential block to the remaining running */
 980         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 981                 while (lp < pmb_nr && pmb[lp])
 982                         lp++;
 983                 /* lp points at the first NULL now */
 984
 985                 while (rp > -1 && !pmb[rp])
 986                         rp--;
 987                 /* rp points at the last non-NULL */
 988
 989                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 990                         pmb[lp] = pmb[rp];
 991                         if (pmb[rp]->wsd) {
 992                                 free(pmb[rp]->wsd->string);
 993                                 FREE_AND_NULL(pmb[rp]->wsd);
 994                         }
 995                         pmb[rp] = NULL;
 996                         rp--;
 997                         lp++;
 998                 }
 999         }
1000
1001         /* Remember the number of running sets */
1002         return rp + 1;
1003 }
1004
1005 /*
1006  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1007  *
1008  * Otherwise, if the last block has fewer alphanumeric characters than
1009  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1010  * that block.
1011  *
1012  * The last block consists of the (n - block_length)'th line up to but not
1013  * including the nth line.
1014  *
1015  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1016  * Think of a way to unify them.
1017  */
1018 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1019 {
1020         int i, alnum_count = 0;
1021         if (o->color_moved == COLOR_MOVED_PLAIN)
1022                 return;
1023         for (i = 1; i < block_length + 1; i++) {
1024                 const char *c = o->emitted_symbols->buf[n - i].line;
1025                 for (; *c; c++) {
1026                         if (!isalnum(*c))
1027                                 continue;
1028                         alnum_count++;
1029                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1030                                 return;
1031                 }
1032         }
1033         for (i = 1; i < block_length + 1; i++)
1034                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1035 }
1036
1037 /* Find blocks of moved code, delegate actual coloring decision to helper */
1038 static void mark_color_as_moved(struct diff_options *o,
1039                                 struct hashmap *add_lines,
1040                                 struct hashmap *del_lines)
1041 {
1042         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1043         int pmb_nr = 0, pmb_alloc = 0;
1044         int n, flipped_block = 1, block_length = 0;
1045
1046
1047         for (n = 0; n < o->emitted_symbols->nr; n++) {
1048                 struct hashmap *hm = NULL;
1049                 struct moved_entry *key;
1050                 struct moved_entry *match = NULL;
1051                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1052
1053                 switch (l->s) {
1054                 case DIFF_SYMBOL_PLUS:
1055                         hm = del_lines;
1056                         key = prepare_entry(o, n);
1057                         match = hashmap_get(hm, key, NULL);
1058                         free(key);
1059                         break;
1060                 case DIFF_SYMBOL_MINUS:
1061                         hm = add_lines;
1062                         key = prepare_entry(o, n);
1063                         match = hashmap_get(hm, key, NULL);
1064                         free(key);
1065                         break;
1066                 default:
1067                         flipped_block = 1;
1068                 }
1069
1070                 if (!match) {
1071                         adjust_last_block(o, n, block_length);
1072                         pmb_nr = 0;
1073                         block_length = 0;
1074                         continue;
1075                 }
1076
1077                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1078
1079                 if (o->color_moved == COLOR_MOVED_PLAIN)
1080                         continue;
1081
1082                 if (o->color_moved_ws_handling &
1083                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1084                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1085                 else
1086                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1087
1088                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1089
1090                 if (pmb_nr == 0) {
1091                         /*
1092                          * The current line is the start of a new block.
1093                          * Setup the set of potential blocks.
1094                          */
1095                         for (; match; match = hashmap_get_next(hm, match)) {
1096                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1097                                 if (o->color_moved_ws_handling &
1098                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1099                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1100                                         if (compute_ws_delta(l, match->es, wsd)) {
1101                                                 match->wsd = wsd;
1102                                                 pmb[pmb_nr++] = match;
1103                                         } else
1104                                                 free(wsd);
1105                                 } else {
1106                                         pmb[pmb_nr++] = match;
1107                                 }
1108                         }
1109
1110                         flipped_block = (flipped_block + 1) % 2;
1111
1112                         adjust_last_block(o, n, block_length);
1113                         block_length = 0;
1114                 }
1115
1116                 block_length++;
1117
1118                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1119                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1120         }
1121         adjust_last_block(o, n, block_length);
1122
1123         free(pmb);
1124 }
1125
1126 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1127   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1128 static void dim_moved_lines(struct diff_options *o)
1129 {
1130         int n;
1131         for (n = 0; n < o->emitted_symbols->nr; n++) {
1132                 struct emitted_diff_symbol *prev = (n != 0) ?
1133                                 &o->emitted_symbols->buf[n - 1] : NULL;
1134                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1135                 struct emitted_diff_symbol *next =
1136                                 (n < o->emitted_symbols->nr - 1) ?
1137                                 &o->emitted_symbols->buf[n + 1] : NULL;
1138
1139                 /* Not a plus or minus line? */
1140                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1141                         continue;
1142
1143                 /* Not a moved line? */
1144                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1145                         continue;
1146
1147                 /*
1148                  * If prev or next are not a plus or minus line,
1149                  * pretend they don't exist
1150                  */
1151                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1152                             prev->s != DIFF_SYMBOL_MINUS)
1153                         prev = NULL;
1154                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1155                             next->s != DIFF_SYMBOL_MINUS)
1156                         next = NULL;
1157
1158                 /* Inside a block? */
1159                 if ((prev &&
1160                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1161                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1162                     (next &&
1163                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1164                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1165                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1166                         continue;
1167                 }
1168
1169                 /* Check if we are at an interesting bound: */
1170                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1171                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1172                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1173                         continue;
1174                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1175                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1176                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1177                         continue;
1178
1179                 /*
1180                  * The boundary to prev and next are not interesting,
1181                  * so this line is not interesting as a whole
1182                  */
1183                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1184         }
1185 }
1186
1187 static void emit_line_ws_markup(struct diff_options *o,
1188                                 const char *set, const char *reset,
1189                                 const char *line, int len,
1190                                 const char *set_sign, char sign,
1191                                 unsigned ws_rule, int blank_at_eof)
1192 {
1193         const char *ws = NULL;
1194
1195         if (o->ws_error_highlight & ws_rule) {
1196                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1197                 if (!*ws)
1198                         ws = NULL;
1199         }
1200
1201         if (!ws && !set_sign)
1202                 emit_line_0(o, set, 0, reset, sign, line, len);
1203         else if (!ws) {
1204                 /* Emit just the prefix, then the rest. */
1205                 emit_line_0(o, set_sign, !!set_sign, reset, sign, "", 0);
1206                 emit_line_0(o, set, 0, reset, 0, line, len);
1207         } else if (blank_at_eof)
1208                 /* Blank line at EOF - paint '+' as well */
1209                 emit_line_0(o, ws, 0, reset, sign, line, len);
1210         else {
1211                 /* Emit just the prefix, then the rest. */
1212                 emit_line_0(o, set_sign ? set_sign : set, !!set_sign, reset,
1213                             sign, "", 0);
1214                 ws_check_emit(line, len, ws_rule,
1215                               o->file, set, reset, ws);
1216         }
1217 }
1218
1219 static void emit_diff_symbol_from_struct(struct diff_options *o,
1220                                          struct emitted_diff_symbol *eds)
1221 {
1222         static const char *nneof = " No newline at end of file\n";
1223         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1224         struct strbuf sb = STRBUF_INIT;
1225
1226         enum diff_symbol s = eds->s;
1227         const char *line = eds->line;
1228         int len = eds->len;
1229         unsigned flags = eds->flags;
1230
1231         switch (s) {
1232         case DIFF_SYMBOL_NO_LF_EOF:
1233                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1234                 reset = diff_get_color_opt(o, DIFF_RESET);
1235                 putc('\n', o->file);
1236                 emit_line_0(o, context, 0, reset, '\\',
1237                             nneof, strlen(nneof));
1238                 break;
1239         case DIFF_SYMBOL_SUBMODULE_HEADER:
1240         case DIFF_SYMBOL_SUBMODULE_ERROR:
1241         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1242         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1243         case DIFF_SYMBOL_SUMMARY:
1244         case DIFF_SYMBOL_STATS_LINE:
1245         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1246         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1247                 emit_line(o, "", "", line, len);
1248                 break;
1249         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1250         case DIFF_SYMBOL_CONTEXT_MARKER:
1251                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1252                 reset = diff_get_color_opt(o, DIFF_RESET);
1253                 emit_line(o, context, reset, line, len);
1254                 break;
1255         case DIFF_SYMBOL_SEPARATOR:
1256                 fprintf(o->file, "%s%c",
1257                         diff_line_prefix(o),
1258                         o->line_termination);
1259                 break;
1260         case DIFF_SYMBOL_CONTEXT:
1261                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1262                 reset = diff_get_color_opt(o, DIFF_RESET);
1263                 set_sign = NULL;
1264                 if (o->flags.dual_color_diffed_diffs) {
1265                         char c = !len ? 0 : line[0];
1266
1267                         if (c == '+')
1268                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1269                         else if (c == '@')
1270                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1271                         else if (c == '-')
1272                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1273                 }
1274                 emit_line_ws_markup(o, set, reset, line, len, set_sign, ' ',
1275                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1276                 break;
1277         case DIFF_SYMBOL_PLUS:
1278                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1279                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1280                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1281                 case DIFF_SYMBOL_MOVED_LINE |
1282                      DIFF_SYMBOL_MOVED_LINE_ALT |
1283                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1284                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1285                         break;
1286                 case DIFF_SYMBOL_MOVED_LINE |
1287                      DIFF_SYMBOL_MOVED_LINE_ALT:
1288                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1289                         break;
1290                 case DIFF_SYMBOL_MOVED_LINE |
1291                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1292                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1293                         break;
1294                 case DIFF_SYMBOL_MOVED_LINE:
1295                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1296                         break;
1297                 default:
1298                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1299                 }
1300                 reset = diff_get_color_opt(o, DIFF_RESET);
1301                 if (!o->flags.dual_color_diffed_diffs)
1302                         set_sign = NULL;
1303                 else {
1304                         char c = !len ? 0 : line[0];
1305
1306                         set_sign = set;
1307                         if (c == '-')
1308                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1309                         else if (c == '@')
1310                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1311                         else if (c == '+')
1312                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1313                         else
1314                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1315                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1316                 }
1317                 emit_line_ws_markup(o, set, reset, line, len, set_sign, '+',
1318                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1319                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1320                 break;
1321         case DIFF_SYMBOL_MINUS:
1322                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1323                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1324                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1325                 case DIFF_SYMBOL_MOVED_LINE |
1326                      DIFF_SYMBOL_MOVED_LINE_ALT |
1327                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1328                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1329                         break;
1330                 case DIFF_SYMBOL_MOVED_LINE |
1331                      DIFF_SYMBOL_MOVED_LINE_ALT:
1332                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1333                         break;
1334                 case DIFF_SYMBOL_MOVED_LINE |
1335                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1336                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1337                         break;
1338                 case DIFF_SYMBOL_MOVED_LINE:
1339                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1340                         break;
1341                 default:
1342                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1343                 }
1344                 reset = diff_get_color_opt(o, DIFF_RESET);
1345                 if (!o->flags.dual_color_diffed_diffs)
1346                         set_sign = NULL;
1347                 else {
1348                         char c = !len ? 0 : line[0];
1349
1350                         set_sign = set;
1351                         if (c == '+')
1352                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1353                         else if (c == '@')
1354                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1355                         else if (c == '-')
1356                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1357                         else
1358                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1359                 }
1360                 emit_line_ws_markup(o, set, reset, line, len, set_sign, '-',
1361                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1362                 break;
1363         case DIFF_SYMBOL_WORDS_PORCELAIN:
1364                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1365                 reset = diff_get_color_opt(o, DIFF_RESET);
1366                 emit_line(o, context, reset, line, len);
1367                 fputs("~\n", o->file);
1368                 break;
1369         case DIFF_SYMBOL_WORDS:
1370                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1371                 reset = diff_get_color_opt(o, DIFF_RESET);
1372                 /*
1373                  * Skip the prefix character, if any.  With
1374                  * diff_suppress_blank_empty, there may be
1375                  * none.
1376                  */
1377                 if (line[0] != '\n') {
1378                         line++;
1379                         len--;
1380                 }
1381                 emit_line(o, context, reset, line, len);
1382                 break;
1383         case DIFF_SYMBOL_FILEPAIR_PLUS:
1384                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1385                 reset = diff_get_color_opt(o, DIFF_RESET);
1386                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1387                         line, reset,
1388                         strchr(line, ' ') ? "\t" : "");
1389                 break;
1390         case DIFF_SYMBOL_FILEPAIR_MINUS:
1391                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1392                 reset = diff_get_color_opt(o, DIFF_RESET);
1393                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1394                         line, reset,
1395                         strchr(line, ' ') ? "\t" : "");
1396                 break;
1397         case DIFF_SYMBOL_BINARY_FILES:
1398         case DIFF_SYMBOL_HEADER:
1399                 fprintf(o->file, "%s", line);
1400                 break;
1401         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1402                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1403                 break;
1404         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1405                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1406                 break;
1407         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1408                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1409                 break;
1410         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1411                 fputs(diff_line_prefix(o), o->file);
1412                 fputc('\n', o->file);
1413                 break;
1414         case DIFF_SYMBOL_REWRITE_DIFF:
1415                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1416                 reset = diff_get_color_opt(o, DIFF_RESET);
1417                 emit_line(o, fraginfo, reset, line, len);
1418                 break;
1419         case DIFF_SYMBOL_SUBMODULE_ADD:
1420                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1421                 reset = diff_get_color_opt(o, DIFF_RESET);
1422                 emit_line(o, set, reset, line, len);
1423                 break;
1424         case DIFF_SYMBOL_SUBMODULE_DEL:
1425                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1426                 reset = diff_get_color_opt(o, DIFF_RESET);
1427                 emit_line(o, set, reset, line, len);
1428                 break;
1429         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1430                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1431                         diff_line_prefix(o), line);
1432                 break;
1433         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1434                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1435                         diff_line_prefix(o), line);
1436                 break;
1437         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1438                 emit_line(o, "", "", " 0 files changed\n",
1439                           strlen(" 0 files changed\n"));
1440                 break;
1441         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1442                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1443                 break;
1444         case DIFF_SYMBOL_WORD_DIFF:
1445                 fprintf(o->file, "%.*s", len, line);
1446                 break;
1447         case DIFF_SYMBOL_STAT_SEP:
1448                 fputs(o->stat_sep, o->file);
1449                 break;
1450         default:
1451                 BUG("unknown diff symbol");
1452         }
1453         strbuf_release(&sb);
1454 }
1455
1456 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1457                              const char *line, int len, unsigned flags)
1458 {
1459         struct emitted_diff_symbol e = {line, len, flags, s};
1460
1461         if (o->emitted_symbols)
1462                 append_emitted_diff_symbol(o, &e);
1463         else
1464                 emit_diff_symbol_from_struct(o, &e);
1465 }
1466
1467 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1468 {
1469         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1470 }
1471
1472 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1473 {
1474         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1475 }
1476
1477 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1478 {
1479         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1480                          path, strlen(path), 0);
1481 }
1482
1483 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1484 {
1485         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1486                          path, strlen(path), 0);
1487 }
1488
1489 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1490 {
1491         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1492                          header, strlen(header), 0);
1493 }
1494
1495 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1496 {
1497         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1498 }
1499
1500 void diff_emit_submodule_pipethrough(struct diff_options *o,
1501                                      const char *line, int len)
1502 {
1503         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1504 }
1505
1506 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1507 {
1508         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1509               ecbdata->blank_at_eof_in_preimage &&
1510               ecbdata->blank_at_eof_in_postimage &&
1511               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1512               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1513                 return 0;
1514         return ws_blank_line(line, len, ecbdata->ws_rule);
1515 }
1516
1517 static void emit_add_line(const char *reset,
1518                           struct emit_callback *ecbdata,
1519                           const char *line, int len)
1520 {
1521         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1522         if (new_blank_line_at_eof(ecbdata, line, len))
1523                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1524
1525         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1526 }
1527
1528 static void emit_del_line(const char *reset,
1529                           struct emit_callback *ecbdata,
1530                           const char *line, int len)
1531 {
1532         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1533         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1534 }
1535
1536 static void emit_context_line(const char *reset,
1537                               struct emit_callback *ecbdata,
1538                               const char *line, int len)
1539 {
1540         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1541         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1542 }
1543
1544 static void emit_hunk_header(struct emit_callback *ecbdata,
1545                              const char *line, int len)
1546 {
1547         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1548         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1549         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1550         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1551         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1552         static const char atat[2] = { '@', '@' };
1553         const char *cp, *ep;
1554         struct strbuf msgbuf = STRBUF_INIT;
1555         int org_len = len;
1556         int i = 1;
1557
1558         /*
1559          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1560          * it always is at least 10 bytes long.
1561          */
1562         if (len < 10 ||
1563             memcmp(line, atat, 2) ||
1564             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1565                 emit_diff_symbol(ecbdata->opt,
1566                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1567                 return;
1568         }
1569         ep += 2; /* skip over @@ */
1570
1571         /* The hunk header in fraginfo color */
1572         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1573                 strbuf_addstr(&msgbuf, reverse);
1574         strbuf_addstr(&msgbuf, frag);
1575         strbuf_add(&msgbuf, line, ep - line);
1576         strbuf_addstr(&msgbuf, reset);
1577
1578         /*
1579          * trailing "\r\n"
1580          */
1581         for ( ; i < 3; i++)
1582                 if (line[len - i] == '\r' || line[len - i] == '\n')
1583                         len--;
1584
1585         /* blank before the func header */
1586         for (cp = ep; ep - line < len; ep++)
1587                 if (*ep != ' ' && *ep != '\t')
1588                         break;
1589         if (ep != cp) {
1590                 strbuf_addstr(&msgbuf, context);
1591                 strbuf_add(&msgbuf, cp, ep - cp);
1592                 strbuf_addstr(&msgbuf, reset);
1593         }
1594
1595         if (ep < line + len) {
1596                 strbuf_addstr(&msgbuf, func);
1597                 strbuf_add(&msgbuf, ep, line + len - ep);
1598                 strbuf_addstr(&msgbuf, reset);
1599         }
1600
1601         strbuf_add(&msgbuf, line + len, org_len - len);
1602         strbuf_complete_line(&msgbuf);
1603         emit_diff_symbol(ecbdata->opt,
1604                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1605         strbuf_release(&msgbuf);
1606 }
1607
1608 static struct diff_tempfile *claim_diff_tempfile(void) {
1609         int i;
1610         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1611                 if (!diff_temp[i].name)
1612                         return diff_temp + i;
1613         BUG("diff is failing to clean up its tempfiles");
1614 }
1615
1616 static void remove_tempfile(void)
1617 {
1618         int i;
1619         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1620                 if (is_tempfile_active(diff_temp[i].tempfile))
1621                         delete_tempfile(&diff_temp[i].tempfile);
1622                 diff_temp[i].name = NULL;
1623         }
1624 }
1625
1626 static void add_line_count(struct strbuf *out, int count)
1627 {
1628         switch (count) {
1629         case 0:
1630                 strbuf_addstr(out, "0,0");
1631                 break;
1632         case 1:
1633                 strbuf_addstr(out, "1");
1634                 break;
1635         default:
1636                 strbuf_addf(out, "1,%d", count);
1637                 break;
1638         }
1639 }
1640
1641 static void emit_rewrite_lines(struct emit_callback *ecb,
1642                                int prefix, const char *data, int size)
1643 {
1644         const char *endp = NULL;
1645         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1646
1647         while (0 < size) {
1648                 int len;
1649
1650                 endp = memchr(data, '\n', size);
1651                 len = endp ? (endp - data + 1) : size;
1652                 if (prefix != '+') {
1653                         ecb->lno_in_preimage++;
1654                         emit_del_line(reset, ecb, data, len);
1655                 } else {
1656                         ecb->lno_in_postimage++;
1657                         emit_add_line(reset, ecb, data, len);
1658                 }
1659                 size -= len;
1660                 data += len;
1661         }
1662         if (!endp)
1663                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1664 }
1665
1666 static void emit_rewrite_diff(const char *name_a,
1667                               const char *name_b,
1668                               struct diff_filespec *one,
1669                               struct diff_filespec *two,
1670                               struct userdiff_driver *textconv_one,
1671                               struct userdiff_driver *textconv_two,
1672                               struct diff_options *o)
1673 {
1674         int lc_a, lc_b;
1675         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1676         const char *a_prefix, *b_prefix;
1677         char *data_one, *data_two;
1678         size_t size_one, size_two;
1679         struct emit_callback ecbdata;
1680         struct strbuf out = STRBUF_INIT;
1681
1682         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1683                 a_prefix = o->b_prefix;
1684                 b_prefix = o->a_prefix;
1685         } else {
1686                 a_prefix = o->a_prefix;
1687                 b_prefix = o->b_prefix;
1688         }
1689
1690         name_a += (*name_a == '/');
1691         name_b += (*name_b == '/');
1692
1693         strbuf_reset(&a_name);
1694         strbuf_reset(&b_name);
1695         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1696         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1697
1698         size_one = fill_textconv(textconv_one, one, &data_one);
1699         size_two = fill_textconv(textconv_two, two, &data_two);
1700
1701         memset(&ecbdata, 0, sizeof(ecbdata));
1702         ecbdata.color_diff = want_color(o->use_color);
1703         ecbdata.ws_rule = whitespace_rule(name_b);
1704         ecbdata.opt = o;
1705         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1706                 mmfile_t mf1, mf2;
1707                 mf1.ptr = (char *)data_one;
1708                 mf2.ptr = (char *)data_two;
1709                 mf1.size = size_one;
1710                 mf2.size = size_two;
1711                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1712         }
1713         ecbdata.lno_in_preimage = 1;
1714         ecbdata.lno_in_postimage = 1;
1715
1716         lc_a = count_lines(data_one, size_one);
1717         lc_b = count_lines(data_two, size_two);
1718
1719         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1720                          a_name.buf, a_name.len, 0);
1721         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1722                          b_name.buf, b_name.len, 0);
1723
1724         strbuf_addstr(&out, "@@ -");
1725         if (!o->irreversible_delete)
1726                 add_line_count(&out, lc_a);
1727         else
1728                 strbuf_addstr(&out, "?,?");
1729         strbuf_addstr(&out, " +");
1730         add_line_count(&out, lc_b);
1731         strbuf_addstr(&out, " @@\n");
1732         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1733         strbuf_release(&out);
1734
1735         if (lc_a && !o->irreversible_delete)
1736                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1737         if (lc_b)
1738                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1739         if (textconv_one)
1740                 free((char *)data_one);
1741         if (textconv_two)
1742                 free((char *)data_two);
1743 }
1744
1745 struct diff_words_buffer {
1746         mmfile_t text;
1747         unsigned long alloc;
1748         struct diff_words_orig {
1749                 const char *begin, *end;
1750         } *orig;
1751         int orig_nr, orig_alloc;
1752 };
1753
1754 static void diff_words_append(char *line, unsigned long len,
1755                 struct diff_words_buffer *buffer)
1756 {
1757         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1758         line++;
1759         len--;
1760         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1761         buffer->text.size += len;
1762         buffer->text.ptr[buffer->text.size] = '\0';
1763 }
1764
1765 struct diff_words_style_elem {
1766         const char *prefix;
1767         const char *suffix;
1768         const char *color; /* NULL; filled in by the setup code if
1769                             * color is enabled */
1770 };
1771
1772 struct diff_words_style {
1773         enum diff_words_type type;
1774         struct diff_words_style_elem new_word, old_word, ctx;
1775         const char *newline;
1776 };
1777
1778 static struct diff_words_style diff_words_styles[] = {
1779         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1780         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1781         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1782 };
1783
1784 struct diff_words_data {
1785         struct diff_words_buffer minus, plus;
1786         const char *current_plus;
1787         int last_minus;
1788         struct diff_options *opt;
1789         regex_t *word_regex;
1790         enum diff_words_type type;
1791         struct diff_words_style *style;
1792 };
1793
1794 static int fn_out_diff_words_write_helper(struct diff_options *o,
1795                                           struct diff_words_style_elem *st_el,
1796                                           const char *newline,
1797                                           size_t count, const char *buf)
1798 {
1799         int print = 0;
1800         struct strbuf sb = STRBUF_INIT;
1801
1802         while (count) {
1803                 char *p = memchr(buf, '\n', count);
1804                 if (print)
1805                         strbuf_addstr(&sb, diff_line_prefix(o));
1806
1807                 if (p != buf) {
1808                         const char *reset = st_el->color && *st_el->color ?
1809                                             GIT_COLOR_RESET : NULL;
1810                         if (st_el->color && *st_el->color)
1811                                 strbuf_addstr(&sb, st_el->color);
1812                         strbuf_addstr(&sb, st_el->prefix);
1813                         strbuf_add(&sb, buf, p ? p - buf : count);
1814                         strbuf_addstr(&sb, st_el->suffix);
1815                         if (reset)
1816                                 strbuf_addstr(&sb, reset);
1817                 }
1818                 if (!p)
1819                         goto out;
1820
1821                 strbuf_addstr(&sb, newline);
1822                 count -= p + 1 - buf;
1823                 buf = p + 1;
1824                 print = 1;
1825                 if (count) {
1826                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1827                                          sb.buf, sb.len, 0);
1828                         strbuf_reset(&sb);
1829                 }
1830         }
1831
1832 out:
1833         if (sb.len)
1834                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1835                                  sb.buf, sb.len, 0);
1836         strbuf_release(&sb);
1837         return 0;
1838 }
1839
1840 /*
1841  * '--color-words' algorithm can be described as:
1842  *
1843  *   1. collect the minus/plus lines of a diff hunk, divided into
1844  *      minus-lines and plus-lines;
1845  *
1846  *   2. break both minus-lines and plus-lines into words and
1847  *      place them into two mmfile_t with one word for each line;
1848  *
1849  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1850  *
1851  * And for the common parts of the both file, we output the plus side text.
1852  * diff_words->current_plus is used to trace the current position of the plus file
1853  * which printed. diff_words->last_minus is used to trace the last minus word
1854  * printed.
1855  *
1856  * For '--graph' to work with '--color-words', we need to output the graph prefix
1857  * on each line of color words output. Generally, there are two conditions on
1858  * which we should output the prefix.
1859  *
1860  *   1. diff_words->last_minus == 0 &&
1861  *      diff_words->current_plus == diff_words->plus.text.ptr
1862  *
1863  *      that is: the plus text must start as a new line, and if there is no minus
1864  *      word printed, a graph prefix must be printed.
1865  *
1866  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1867  *      *(diff_words->current_plus - 1) == '\n'
1868  *
1869  *      that is: a graph prefix must be printed following a '\n'
1870  */
1871 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1872 {
1873         if ((diff_words->last_minus == 0 &&
1874                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1875                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1876                 *(diff_words->current_plus - 1) == '\n')) {
1877                 return 1;
1878         } else {
1879                 return 0;
1880         }
1881 }
1882
1883 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1884 {
1885         struct diff_words_data *diff_words = priv;
1886         struct diff_words_style *style = diff_words->style;
1887         int minus_first, minus_len, plus_first, plus_len;
1888         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1889         struct diff_options *opt = diff_words->opt;
1890         const char *line_prefix;
1891
1892         if (line[0] != '@' || parse_hunk_header(line, len,
1893                         &minus_first, &minus_len, &plus_first, &plus_len))
1894                 return;
1895
1896         assert(opt);
1897         line_prefix = diff_line_prefix(opt);
1898
1899         /* POSIX requires that first be decremented by one if len == 0... */
1900         if (minus_len) {
1901                 minus_begin = diff_words->minus.orig[minus_first].begin;
1902                 minus_end =
1903                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1904         } else
1905                 minus_begin = minus_end =
1906                         diff_words->minus.orig[minus_first].end;
1907
1908         if (plus_len) {
1909                 plus_begin = diff_words->plus.orig[plus_first].begin;
1910                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1911         } else
1912                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1913
1914         if (color_words_output_graph_prefix(diff_words)) {
1915                 fputs(line_prefix, diff_words->opt->file);
1916         }
1917         if (diff_words->current_plus != plus_begin) {
1918                 fn_out_diff_words_write_helper(diff_words->opt,
1919                                 &style->ctx, style->newline,
1920                                 plus_begin - diff_words->current_plus,
1921                                 diff_words->current_plus);
1922         }
1923         if (minus_begin != minus_end) {
1924                 fn_out_diff_words_write_helper(diff_words->opt,
1925                                 &style->old_word, style->newline,
1926                                 minus_end - minus_begin, minus_begin);
1927         }
1928         if (plus_begin != plus_end) {
1929                 fn_out_diff_words_write_helper(diff_words->opt,
1930                                 &style->new_word, style->newline,
1931                                 plus_end - plus_begin, plus_begin);
1932         }
1933
1934         diff_words->current_plus = plus_end;
1935         diff_words->last_minus = minus_first;
1936 }
1937
1938 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1939 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1940                 int *begin, int *end)
1941 {
1942         if (word_regex && *begin < buffer->size) {
1943                 regmatch_t match[1];
1944                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1945                                  buffer->size - *begin, 1, match, 0)) {
1946                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1947                                         '\n', match[0].rm_eo - match[0].rm_so);
1948                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1949                         *begin += match[0].rm_so;
1950                         return *begin >= *end;
1951                 }
1952                 return -1;
1953         }
1954
1955         /* find the next word */
1956         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1957                 (*begin)++;
1958         if (*begin >= buffer->size)
1959                 return -1;
1960
1961         /* find the end of the word */
1962         *end = *begin + 1;
1963         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1964                 (*end)++;
1965
1966         return 0;
1967 }
1968
1969 /*
1970  * This function splits the words in buffer->text, stores the list with
1971  * newline separator into out, and saves the offsets of the original words
1972  * in buffer->orig.
1973  */
1974 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1975                 regex_t *word_regex)
1976 {
1977         int i, j;
1978         long alloc = 0;
1979
1980         out->size = 0;
1981         out->ptr = NULL;
1982
1983         /* fake an empty "0th" word */
1984         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1985         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1986         buffer->orig_nr = 1;
1987
1988         for (i = 0; i < buffer->text.size; i++) {
1989                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1990                         return;
1991
1992                 /* store original boundaries */
1993                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1994                                 buffer->orig_alloc);
1995                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1996                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1997                 buffer->orig_nr++;
1998
1999                 /* store one word */
2000                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2001                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2002                 out->ptr[out->size + j - i] = '\n';
2003                 out->size += j - i + 1;
2004
2005                 i = j - 1;
2006         }
2007 }
2008
2009 /* this executes the word diff on the accumulated buffers */
2010 static void diff_words_show(struct diff_words_data *diff_words)
2011 {
2012         xpparam_t xpp;
2013         xdemitconf_t xecfg;
2014         mmfile_t minus, plus;
2015         struct diff_words_style *style = diff_words->style;
2016
2017         struct diff_options *opt = diff_words->opt;
2018         const char *line_prefix;
2019
2020         assert(opt);
2021         line_prefix = diff_line_prefix(opt);
2022
2023         /* special case: only removal */
2024         if (!diff_words->plus.text.size) {
2025                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2026                                  line_prefix, strlen(line_prefix), 0);
2027                 fn_out_diff_words_write_helper(diff_words->opt,
2028                         &style->old_word, style->newline,
2029                         diff_words->minus.text.size,
2030                         diff_words->minus.text.ptr);
2031                 diff_words->minus.text.size = 0;
2032                 return;
2033         }
2034
2035         diff_words->current_plus = diff_words->plus.text.ptr;
2036         diff_words->last_minus = 0;
2037
2038         memset(&xpp, 0, sizeof(xpp));
2039         memset(&xecfg, 0, sizeof(xecfg));
2040         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2041         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2042         xpp.flags = 0;
2043         /* as only the hunk header will be parsed, we need a 0-context */
2044         xecfg.ctxlen = 0;
2045         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2046                           &xpp, &xecfg))
2047                 die("unable to generate word diff");
2048         free(minus.ptr);
2049         free(plus.ptr);
2050         if (diff_words->current_plus != diff_words->plus.text.ptr +
2051                         diff_words->plus.text.size) {
2052                 if (color_words_output_graph_prefix(diff_words))
2053                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2054                                          line_prefix, strlen(line_prefix), 0);
2055                 fn_out_diff_words_write_helper(diff_words->opt,
2056                         &style->ctx, style->newline,
2057                         diff_words->plus.text.ptr + diff_words->plus.text.size
2058                         - diff_words->current_plus, diff_words->current_plus);
2059         }
2060         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2061 }
2062
2063 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2064 static void diff_words_flush(struct emit_callback *ecbdata)
2065 {
2066         struct diff_options *wo = ecbdata->diff_words->opt;
2067
2068         if (ecbdata->diff_words->minus.text.size ||
2069             ecbdata->diff_words->plus.text.size)
2070                 diff_words_show(ecbdata->diff_words);
2071
2072         if (wo->emitted_symbols) {
2073                 struct diff_options *o = ecbdata->opt;
2074                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2075                 int i;
2076
2077                 /*
2078                  * NEEDSWORK:
2079                  * Instead of appending each, concat all words to a line?
2080                  */
2081                 for (i = 0; i < wol->nr; i++)
2082                         append_emitted_diff_symbol(o, &wol->buf[i]);
2083
2084                 for (i = 0; i < wol->nr; i++)
2085                         free((void *)wol->buf[i].line);
2086
2087                 wol->nr = 0;
2088         }
2089 }
2090
2091 static void diff_filespec_load_driver(struct diff_filespec *one)
2092 {
2093         /* Use already-loaded driver */
2094         if (one->driver)
2095                 return;
2096
2097         if (S_ISREG(one->mode))
2098                 one->driver = userdiff_find_by_path(one->path);
2099
2100         /* Fallback to default settings */
2101         if (!one->driver)
2102                 one->driver = userdiff_find_by_name("default");
2103 }
2104
2105 static const char *userdiff_word_regex(struct diff_filespec *one)
2106 {
2107         diff_filespec_load_driver(one);
2108         return one->driver->word_regex;
2109 }
2110
2111 static void init_diff_words_data(struct emit_callback *ecbdata,
2112                                  struct diff_options *orig_opts,
2113                                  struct diff_filespec *one,
2114                                  struct diff_filespec *two)
2115 {
2116         int i;
2117         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2118         memcpy(o, orig_opts, sizeof(struct diff_options));
2119
2120         ecbdata->diff_words =
2121                 xcalloc(1, sizeof(struct diff_words_data));
2122         ecbdata->diff_words->type = o->word_diff;
2123         ecbdata->diff_words->opt = o;
2124
2125         if (orig_opts->emitted_symbols)
2126                 o->emitted_symbols =
2127                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2128
2129         if (!o->word_regex)
2130                 o->word_regex = userdiff_word_regex(one);
2131         if (!o->word_regex)
2132                 o->word_regex = userdiff_word_regex(two);
2133         if (!o->word_regex)
2134                 o->word_regex = diff_word_regex_cfg;
2135         if (o->word_regex) {
2136                 ecbdata->diff_words->word_regex = (regex_t *)
2137                         xmalloc(sizeof(regex_t));
2138                 if (regcomp(ecbdata->diff_words->word_regex,
2139                             o->word_regex,
2140                             REG_EXTENDED | REG_NEWLINE))
2141                         die ("Invalid regular expression: %s",
2142                              o->word_regex);
2143         }
2144         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2145                 if (o->word_diff == diff_words_styles[i].type) {
2146                         ecbdata->diff_words->style =
2147                                 &diff_words_styles[i];
2148                         break;
2149                 }
2150         }
2151         if (want_color(o->use_color)) {
2152                 struct diff_words_style *st = ecbdata->diff_words->style;
2153                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2154                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2155                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2156         }
2157 }
2158
2159 static void free_diff_words_data(struct emit_callback *ecbdata)
2160 {
2161         if (ecbdata->diff_words) {
2162                 diff_words_flush(ecbdata);
2163                 free (ecbdata->diff_words->opt->emitted_symbols);
2164                 free (ecbdata->diff_words->opt);
2165                 free (ecbdata->diff_words->minus.text.ptr);
2166                 free (ecbdata->diff_words->minus.orig);
2167                 free (ecbdata->diff_words->plus.text.ptr);
2168                 free (ecbdata->diff_words->plus.orig);
2169                 if (ecbdata->diff_words->word_regex) {
2170                         regfree(ecbdata->diff_words->word_regex);
2171                         free(ecbdata->diff_words->word_regex);
2172                 }
2173                 FREE_AND_NULL(ecbdata->diff_words);
2174         }
2175 }
2176
2177 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2178 {
2179         if (want_color(diff_use_color))
2180                 return diff_colors[ix];
2181         return "";
2182 }
2183
2184 const char *diff_line_prefix(struct diff_options *opt)
2185 {
2186         struct strbuf *msgbuf;
2187         if (!opt->output_prefix)
2188                 return "";
2189
2190         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2191         return msgbuf->buf;
2192 }
2193
2194 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2195 {
2196         const char *cp;
2197         unsigned long allot;
2198         size_t l = len;
2199
2200         cp = line;
2201         allot = l;
2202         while (0 < l) {
2203                 (void) utf8_width(&cp, &l);
2204                 if (!cp)
2205                         break; /* truncated in the middle? */
2206         }
2207         return allot - l;
2208 }
2209
2210 static void find_lno(const char *line, struct emit_callback *ecbdata)
2211 {
2212         const char *p;
2213         ecbdata->lno_in_preimage = 0;
2214         ecbdata->lno_in_postimage = 0;
2215         p = strchr(line, '-');
2216         if (!p)
2217                 return; /* cannot happen */
2218         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2219         p = strchr(p, '+');
2220         if (!p)
2221                 return; /* cannot happen */
2222         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2223 }
2224
2225 static void fn_out_consume(void *priv, char *line, unsigned long len)
2226 {
2227         struct emit_callback *ecbdata = priv;
2228         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2229         struct diff_options *o = ecbdata->opt;
2230
2231         o->found_changes = 1;
2232
2233         if (ecbdata->header) {
2234                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2235                                  ecbdata->header->buf, ecbdata->header->len, 0);
2236                 strbuf_reset(ecbdata->header);
2237                 ecbdata->header = NULL;
2238         }
2239
2240         if (ecbdata->label_path[0]) {
2241                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2242                                  ecbdata->label_path[0],
2243                                  strlen(ecbdata->label_path[0]), 0);
2244                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2245                                  ecbdata->label_path[1],
2246                                  strlen(ecbdata->label_path[1]), 0);
2247                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2248         }
2249
2250         if (diff_suppress_blank_empty
2251             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2252                 line[0] = '\n';
2253                 len = 1;
2254         }
2255
2256         if (line[0] == '@') {
2257                 if (ecbdata->diff_words)
2258                         diff_words_flush(ecbdata);
2259                 len = sane_truncate_line(ecbdata, line, len);
2260                 find_lno(line, ecbdata);
2261                 emit_hunk_header(ecbdata, line, len);
2262                 return;
2263         }
2264
2265         if (ecbdata->diff_words) {
2266                 enum diff_symbol s =
2267                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2268                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2269                 if (line[0] == '-') {
2270                         diff_words_append(line, len,
2271                                           &ecbdata->diff_words->minus);
2272                         return;
2273                 } else if (line[0] == '+') {
2274                         diff_words_append(line, len,
2275                                           &ecbdata->diff_words->plus);
2276                         return;
2277                 } else if (starts_with(line, "\\ ")) {
2278                         /*
2279                          * Eat the "no newline at eof" marker as if we
2280                          * saw a "+" or "-" line with nothing on it,
2281                          * and return without diff_words_flush() to
2282                          * defer processing. If this is the end of
2283                          * preimage, more "+" lines may come after it.
2284                          */
2285                         return;
2286                 }
2287                 diff_words_flush(ecbdata);
2288                 emit_diff_symbol(o, s, line, len, 0);
2289                 return;
2290         }
2291
2292         switch (line[0]) {
2293         case '+':
2294                 ecbdata->lno_in_postimage++;
2295                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2296                 break;
2297         case '-':
2298                 ecbdata->lno_in_preimage++;
2299                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2300                 break;
2301         case ' ':
2302                 ecbdata->lno_in_postimage++;
2303                 ecbdata->lno_in_preimage++;
2304                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2305                 break;
2306         default:
2307                 /* incomplete line at the end */
2308                 ecbdata->lno_in_preimage++;
2309                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2310                                  line, len, 0);
2311                 break;
2312         }
2313 }
2314
2315 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2316 {
2317         const char *old_name = a;
2318         const char *new_name = b;
2319         int pfx_length, sfx_length;
2320         int pfx_adjust_for_slash;
2321         int len_a = strlen(a);
2322         int len_b = strlen(b);
2323         int a_midlen, b_midlen;
2324         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2325         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2326
2327         if (qlen_a || qlen_b) {
2328                 quote_c_style(a, name, NULL, 0);
2329                 strbuf_addstr(name, " => ");
2330                 quote_c_style(b, name, NULL, 0);
2331                 return;
2332         }
2333
2334         /* Find common prefix */
2335         pfx_length = 0;
2336         while (*old_name && *new_name && *old_name == *new_name) {
2337                 if (*old_name == '/')
2338                         pfx_length = old_name - a + 1;
2339                 old_name++;
2340                 new_name++;
2341         }
2342
2343         /* Find common suffix */
2344         old_name = a + len_a;
2345         new_name = b + len_b;
2346         sfx_length = 0;
2347         /*
2348          * If there is a common prefix, it must end in a slash.  In
2349          * that case we let this loop run 1 into the prefix to see the
2350          * same slash.
2351          *
2352          * If there is no common prefix, we cannot do this as it would
2353          * underrun the input strings.
2354          */
2355         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2356         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2357                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2358                *old_name == *new_name) {
2359                 if (*old_name == '/')
2360                         sfx_length = len_a - (old_name - a);
2361                 old_name--;
2362                 new_name--;
2363         }
2364
2365         /*
2366          * pfx{mid-a => mid-b}sfx
2367          * {pfx-a => pfx-b}sfx
2368          * pfx{sfx-a => sfx-b}
2369          * name-a => name-b
2370          */
2371         a_midlen = len_a - pfx_length - sfx_length;
2372         b_midlen = len_b - pfx_length - sfx_length;
2373         if (a_midlen < 0)
2374                 a_midlen = 0;
2375         if (b_midlen < 0)
2376                 b_midlen = 0;
2377
2378         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2379         if (pfx_length + sfx_length) {
2380                 strbuf_add(name, a, pfx_length);
2381                 strbuf_addch(name, '{');
2382         }
2383         strbuf_add(name, a + pfx_length, a_midlen);
2384         strbuf_addstr(name, " => ");
2385         strbuf_add(name, b + pfx_length, b_midlen);
2386         if (pfx_length + sfx_length) {
2387                 strbuf_addch(name, '}');
2388                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2389         }
2390 }
2391
2392 struct diffstat_t {
2393         int nr;
2394         int alloc;
2395         struct diffstat_file {
2396                 char *from_name;
2397                 char *name;
2398                 char *print_name;
2399                 const char *comments;
2400                 unsigned is_unmerged:1;
2401                 unsigned is_binary:1;
2402                 unsigned is_renamed:1;
2403                 unsigned is_interesting:1;
2404                 uintmax_t added, deleted;
2405         } **files;
2406 };
2407
2408 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2409                                           const char *name_a,
2410                                           const char *name_b)
2411 {
2412         struct diffstat_file *x;
2413         x = xcalloc(1, sizeof(*x));
2414         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2415         diffstat->files[diffstat->nr++] = x;
2416         if (name_b) {
2417                 x->from_name = xstrdup(name_a);
2418                 x->name = xstrdup(name_b);
2419                 x->is_renamed = 1;
2420         }
2421         else {
2422                 x->from_name = NULL;
2423                 x->name = xstrdup(name_a);
2424         }
2425         return x;
2426 }
2427
2428 static void diffstat_consume(void *priv, char *line, unsigned long len)
2429 {
2430         struct diffstat_t *diffstat = priv;
2431         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2432
2433         if (line[0] == '+')
2434                 x->added++;
2435         else if (line[0] == '-')
2436                 x->deleted++;
2437 }
2438
2439 const char mime_boundary_leader[] = "------------";
2440
2441 static int scale_linear(int it, int width, int max_change)
2442 {
2443         if (!it)
2444                 return 0;
2445         /*
2446          * make sure that at least one '-' or '+' is printed if
2447          * there is any change to this path. The easiest way is to
2448          * scale linearly as if the alloted width is one column shorter
2449          * than it is, and then add 1 to the result.
2450          */
2451         return 1 + (it * (width - 1) / max_change);
2452 }
2453
2454 static void show_graph(struct strbuf *out, char ch, int cnt,
2455                        const char *set, const char *reset)
2456 {
2457         if (cnt <= 0)
2458                 return;
2459         strbuf_addstr(out, set);
2460         strbuf_addchars(out, ch, cnt);
2461         strbuf_addstr(out, reset);
2462 }
2463
2464 static void fill_print_name(struct diffstat_file *file)
2465 {
2466         struct strbuf pname = STRBUF_INIT;
2467
2468         if (file->print_name)
2469                 return;
2470
2471         if (file->is_renamed)
2472                 pprint_rename(&pname, file->from_name, file->name);
2473         else
2474                 quote_c_style(file->name, &pname, NULL, 0);
2475
2476         if (file->comments)
2477                 strbuf_addf(&pname, " (%s)", file->comments);
2478
2479         file->print_name = strbuf_detach(&pname, NULL);
2480 }
2481
2482 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2483                 int files, int insertions, int deletions)
2484 {
2485         struct strbuf sb = STRBUF_INIT;
2486
2487         if (!files) {
2488                 assert(insertions == 0 && deletions == 0);
2489                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2490                                  NULL, 0, 0);
2491                 return;
2492         }
2493
2494         strbuf_addf(&sb,
2495                     (files == 1) ? " %d file changed" : " %d files changed",
2496                     files);
2497
2498         /*
2499          * For binary diff, the caller may want to print "x files
2500          * changed" with insertions == 0 && deletions == 0.
2501          *
2502          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2503          * is probably less confusing (i.e skip over "2 files changed
2504          * but nothing about added/removed lines? Is this a bug in Git?").
2505          */
2506         if (insertions || deletions == 0) {
2507                 strbuf_addf(&sb,
2508                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2509                             insertions);
2510         }
2511
2512         if (deletions || insertions == 0) {
2513                 strbuf_addf(&sb,
2514                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2515                             deletions);
2516         }
2517         strbuf_addch(&sb, '\n');
2518         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2519                          sb.buf, sb.len, 0);
2520         strbuf_release(&sb);
2521 }
2522
2523 void print_stat_summary(FILE *fp, int files,
2524                         int insertions, int deletions)
2525 {
2526         struct diff_options o;
2527         memset(&o, 0, sizeof(o));
2528         o.file = fp;
2529
2530         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2531 }
2532
2533 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2534 {
2535         int i, len, add, del, adds = 0, dels = 0;
2536         uintmax_t max_change = 0, max_len = 0;
2537         int total_files = data->nr, count;
2538         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2539         const char *reset, *add_c, *del_c;
2540         int extra_shown = 0;
2541         const char *line_prefix = diff_line_prefix(options);
2542         struct strbuf out = STRBUF_INIT;
2543
2544         if (data->nr == 0)
2545                 return;
2546
2547         count = options->stat_count ? options->stat_count : data->nr;
2548
2549         reset = diff_get_color_opt(options, DIFF_RESET);
2550         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2551         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2552
2553         /*
2554          * Find the longest filename and max number of changes
2555          */
2556         for (i = 0; (i < count) && (i < data->nr); i++) {
2557                 struct diffstat_file *file = data->files[i];
2558                 uintmax_t change = file->added + file->deleted;
2559
2560                 if (!file->is_interesting && (change == 0)) {
2561                         count++; /* not shown == room for one more */
2562                         continue;
2563                 }
2564                 fill_print_name(file);
2565                 len = strlen(file->print_name);
2566                 if (max_len < len)
2567                         max_len = len;
2568
2569                 if (file->is_unmerged) {
2570                         /* "Unmerged" is 8 characters */
2571                         bin_width = bin_width < 8 ? 8 : bin_width;
2572                         continue;
2573                 }
2574                 if (file->is_binary) {
2575                         /* "Bin XXX -> YYY bytes" */
2576                         int w = 14 + decimal_width(file->added)
2577                                 + decimal_width(file->deleted);
2578                         bin_width = bin_width < w ? w : bin_width;
2579                         /* Display change counts aligned with "Bin" */
2580                         number_width = 3;
2581                         continue;
2582                 }
2583
2584                 if (max_change < change)
2585                         max_change = change;
2586         }
2587         count = i; /* where we can stop scanning in data->files[] */
2588
2589         /*
2590          * We have width = stat_width or term_columns() columns total.
2591          * We want a maximum of min(max_len, stat_name_width) for the name part.
2592          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2593          * We also need 1 for " " and 4 + decimal_width(max_change)
2594          * for " | NNNN " and one the empty column at the end, altogether
2595          * 6 + decimal_width(max_change).
2596          *
2597          * If there's not enough space, we will use the smaller of
2598          * stat_name_width (if set) and 5/8*width for the filename,
2599          * and the rest for constant elements + graph part, but no more
2600          * than stat_graph_width for the graph part.
2601          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2602          * for the standard terminal size).
2603          *
2604          * In other words: stat_width limits the maximum width, and
2605          * stat_name_width fixes the maximum width of the filename,
2606          * and is also used to divide available columns if there
2607          * aren't enough.
2608          *
2609          * Binary files are displayed with "Bin XXX -> YYY bytes"
2610          * instead of the change count and graph. This part is treated
2611          * similarly to the graph part, except that it is not
2612          * "scaled". If total width is too small to accommodate the
2613          * guaranteed minimum width of the filename part and the
2614          * separators and this message, this message will "overflow"
2615          * making the line longer than the maximum width.
2616          */
2617
2618         if (options->stat_width == -1)
2619                 width = term_columns() - strlen(line_prefix);
2620         else
2621                 width = options->stat_width ? options->stat_width : 80;
2622         number_width = decimal_width(max_change) > number_width ?
2623                 decimal_width(max_change) : number_width;
2624
2625         if (options->stat_graph_width == -1)
2626                 options->stat_graph_width = diff_stat_graph_width;
2627
2628         /*
2629          * Guarantee 3/8*16==6 for the graph part
2630          * and 5/8*16==10 for the filename part
2631          */
2632         if (width < 16 + 6 + number_width)
2633                 width = 16 + 6 + number_width;
2634
2635         /*
2636          * First assign sizes that are wanted, ignoring available width.
2637          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2638          * starting from "XXX" should fit in graph_width.
2639          */
2640         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2641         if (options->stat_graph_width &&
2642             options->stat_graph_width < graph_width)
2643                 graph_width = options->stat_graph_width;
2644
2645         name_width = (options->stat_name_width > 0 &&
2646                       options->stat_name_width < max_len) ?
2647                 options->stat_name_width : max_len;
2648
2649         /*
2650          * Adjust adjustable widths not to exceed maximum width
2651          */
2652         if (name_width + number_width + 6 + graph_width > width) {
2653                 if (graph_width > width * 3/8 - number_width - 6) {
2654                         graph_width = width * 3/8 - number_width - 6;
2655                         if (graph_width < 6)
2656                                 graph_width = 6;
2657                 }
2658
2659                 if (options->stat_graph_width &&
2660                     graph_width > options->stat_graph_width)
2661                         graph_width = options->stat_graph_width;
2662                 if (name_width > width - number_width - 6 - graph_width)
2663                         name_width = width - number_width - 6 - graph_width;
2664                 else
2665                         graph_width = width - number_width - 6 - name_width;
2666         }
2667
2668         /*
2669          * From here name_width is the width of the name area,
2670          * and graph_width is the width of the graph area.
2671          * max_change is used to scale graph properly.
2672          */
2673         for (i = 0; i < count; i++) {
2674                 const char *prefix = "";
2675                 struct diffstat_file *file = data->files[i];
2676                 char *name = file->print_name;
2677                 uintmax_t added = file->added;
2678                 uintmax_t deleted = file->deleted;
2679                 int name_len;
2680
2681                 if (!file->is_interesting && (added + deleted == 0))
2682                         continue;
2683
2684                 /*
2685                  * "scale" the filename
2686                  */
2687                 len = name_width;
2688                 name_len = strlen(name);
2689                 if (name_width < name_len) {
2690                         char *slash;
2691                         prefix = "...";
2692                         len -= 3;
2693                         name += name_len - len;
2694                         slash = strchr(name, '/');
2695                         if (slash)
2696                                 name = slash;
2697                 }
2698
2699                 if (file->is_binary) {
2700                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2701                         strbuf_addf(&out, " %*s", number_width, "Bin");
2702                         if (!added && !deleted) {
2703                                 strbuf_addch(&out, '\n');
2704                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2705                                                  out.buf, out.len, 0);
2706                                 strbuf_reset(&out);
2707                                 continue;
2708                         }
2709                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2710                                 del_c, deleted, reset);
2711                         strbuf_addstr(&out, " -> ");
2712                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2713                                 add_c, added, reset);
2714                         strbuf_addstr(&out, " bytes\n");
2715                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2716                                          out.buf, out.len, 0);
2717                         strbuf_reset(&out);
2718                         continue;
2719                 }
2720                 else if (file->is_unmerged) {
2721                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2722                         strbuf_addstr(&out, " Unmerged\n");
2723                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2724                                          out.buf, out.len, 0);
2725                         strbuf_reset(&out);
2726                         continue;
2727                 }
2728
2729                 /*
2730                  * scale the add/delete
2731                  */
2732                 add = added;
2733                 del = deleted;
2734
2735                 if (graph_width <= max_change) {
2736                         int total = scale_linear(add + del, graph_width, max_change);
2737                         if (total < 2 && add && del)
2738                                 /* width >= 2 due to the sanity check */
2739                                 total = 2;
2740                         if (add < del) {
2741                                 add = scale_linear(add, graph_width, max_change);
2742                                 del = total - add;
2743                         } else {
2744                                 del = scale_linear(del, graph_width, max_change);
2745                                 add = total - del;
2746                         }
2747                 }
2748                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2749                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2750                         number_width, added + deleted,
2751                         added + deleted ? " " : "");
2752                 show_graph(&out, '+', add, add_c, reset);
2753                 show_graph(&out, '-', del, del_c, reset);
2754                 strbuf_addch(&out, '\n');
2755                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2756                                  out.buf, out.len, 0);
2757                 strbuf_reset(&out);
2758         }
2759
2760         for (i = 0; i < data->nr; i++) {
2761                 struct diffstat_file *file = data->files[i];
2762                 uintmax_t added = file->added;
2763                 uintmax_t deleted = file->deleted;
2764
2765                 if (file->is_unmerged ||
2766                     (!file->is_interesting && (added + deleted == 0))) {
2767                         total_files--;
2768                         continue;
2769                 }
2770
2771                 if (!file->is_binary) {
2772                         adds += added;
2773                         dels += deleted;
2774                 }
2775                 if (i < count)
2776                         continue;
2777                 if (!extra_shown)
2778                         emit_diff_symbol(options,
2779                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2780                                          NULL, 0, 0);
2781                 extra_shown = 1;
2782         }
2783
2784         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2785         strbuf_release(&out);
2786 }
2787
2788 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2789 {
2790         int i, adds = 0, dels = 0, total_files = data->nr;
2791
2792         if (data->nr == 0)
2793                 return;
2794
2795         for (i = 0; i < data->nr; i++) {
2796                 int added = data->files[i]->added;
2797                 int deleted = data->files[i]->deleted;
2798
2799                 if (data->files[i]->is_unmerged ||
2800                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2801                         total_files--;
2802                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2803                         adds += added;
2804                         dels += deleted;
2805                 }
2806         }
2807         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2808 }
2809
2810 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2811 {
2812         int i;
2813
2814         if (data->nr == 0)
2815                 return;
2816
2817         for (i = 0; i < data->nr; i++) {
2818                 struct diffstat_file *file = data->files[i];
2819
2820                 fprintf(options->file, "%s", diff_line_prefix(options));
2821
2822                 if (file->is_binary)
2823                         fprintf(options->file, "-\t-\t");
2824                 else
2825                         fprintf(options->file,
2826                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2827                                 file->added, file->deleted);
2828                 if (options->line_termination) {
2829                         fill_print_name(file);
2830                         if (!file->is_renamed)
2831                                 write_name_quoted(file->name, options->file,
2832                                                   options->line_termination);
2833                         else {
2834                                 fputs(file->print_name, options->file);
2835                                 putc(options->line_termination, options->file);
2836                         }
2837                 } else {
2838                         if (file->is_renamed) {
2839                                 putc('\0', options->file);
2840                                 write_name_quoted(file->from_name, options->file, '\0');
2841                         }
2842                         write_name_quoted(file->name, options->file, '\0');
2843                 }
2844         }
2845 }
2846
2847 struct dirstat_file {
2848         const char *name;
2849         unsigned long changed;
2850 };
2851
2852 struct dirstat_dir {
2853         struct dirstat_file *files;
2854         int alloc, nr, permille, cumulative;
2855 };
2856
2857 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2858                 unsigned long changed, const char *base, int baselen)
2859 {
2860         unsigned long sum_changes = 0;
2861         unsigned int sources = 0;
2862         const char *line_prefix = diff_line_prefix(opt);
2863
2864         while (dir->nr) {
2865                 struct dirstat_file *f = dir->files;
2866                 int namelen = strlen(f->name);
2867                 unsigned long changes;
2868                 char *slash;
2869
2870                 if (namelen < baselen)
2871                         break;
2872                 if (memcmp(f->name, base, baselen))
2873                         break;
2874                 slash = strchr(f->name + baselen, '/');
2875                 if (slash) {
2876                         int newbaselen = slash + 1 - f->name;
2877                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2878                         sources++;
2879                 } else {
2880                         changes = f->changed;
2881                         dir->files++;
2882                         dir->nr--;
2883                         sources += 2;
2884                 }
2885                 sum_changes += changes;
2886         }
2887
2888         /*
2889          * We don't report dirstat's for
2890          *  - the top level
2891          *  - or cases where everything came from a single directory
2892          *    under this directory (sources == 1).
2893          */
2894         if (baselen && sources != 1) {
2895                 if (sum_changes) {
2896                         int permille = sum_changes * 1000 / changed;
2897                         if (permille >= dir->permille) {
2898                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2899                                         permille / 10, permille % 10, baselen, base);
2900                                 if (!dir->cumulative)
2901                                         return 0;
2902                         }
2903                 }
2904         }
2905         return sum_changes;
2906 }
2907
2908 static int dirstat_compare(const void *_a, const void *_b)
2909 {
2910         const struct dirstat_file *a = _a;
2911         const struct dirstat_file *b = _b;
2912         return strcmp(a->name, b->name);
2913 }
2914
2915 static void show_dirstat(struct diff_options *options)
2916 {
2917         int i;
2918         unsigned long changed;
2919         struct dirstat_dir dir;
2920         struct diff_queue_struct *q = &diff_queued_diff;
2921
2922         dir.files = NULL;
2923         dir.alloc = 0;
2924         dir.nr = 0;
2925         dir.permille = options->dirstat_permille;
2926         dir.cumulative = options->flags.dirstat_cumulative;
2927
2928         changed = 0;
2929         for (i = 0; i < q->nr; i++) {
2930                 struct diff_filepair *p = q->queue[i];
2931                 const char *name;
2932                 unsigned long copied, added, damage;
2933                 int content_changed;
2934
2935                 name = p->two->path ? p->two->path : p->one->path;
2936
2937                 if (p->one->oid_valid && p->two->oid_valid)
2938                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2939                 else
2940                         content_changed = 1;
2941
2942                 if (!content_changed) {
2943                         /*
2944                          * The SHA1 has not changed, so pre-/post-content is
2945                          * identical. We can therefore skip looking at the
2946                          * file contents altogether.
2947                          */
2948                         damage = 0;
2949                         goto found_damage;
2950                 }
2951
2952                 if (options->flags.dirstat_by_file) {
2953                         /*
2954                          * In --dirstat-by-file mode, we don't really need to
2955                          * look at the actual file contents at all.
2956                          * The fact that the SHA1 changed is enough for us to
2957                          * add this file to the list of results
2958                          * (with each file contributing equal damage).
2959                          */
2960                         damage = 1;
2961                         goto found_damage;
2962                 }
2963
2964                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2965                         diff_populate_filespec(p->one, 0);
2966                         diff_populate_filespec(p->two, 0);
2967                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2968                                                &copied, &added);
2969                         diff_free_filespec_data(p->one);
2970                         diff_free_filespec_data(p->two);
2971                 } else if (DIFF_FILE_VALID(p->one)) {
2972                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2973                         copied = added = 0;
2974                         diff_free_filespec_data(p->one);
2975                 } else if (DIFF_FILE_VALID(p->two)) {
2976                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2977                         copied = 0;
2978                         added = p->two->size;
2979                         diff_free_filespec_data(p->two);
2980                 } else
2981                         continue;
2982
2983                 /*
2984                  * Original minus copied is the removed material,
2985                  * added is the new material.  They are both damages
2986                  * made to the preimage.
2987                  * If the resulting damage is zero, we know that
2988                  * diffcore_count_changes() considers the two entries to
2989                  * be identical, but since content_changed is true, we
2990                  * know that there must have been _some_ kind of change,
2991                  * so we force all entries to have damage > 0.
2992                  */
2993                 damage = (p->one->size - copied) + added;
2994                 if (!damage)
2995                         damage = 1;
2996
2997 found_damage:
2998                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2999                 dir.files[dir.nr].name = name;
3000                 dir.files[dir.nr].changed = damage;
3001                 changed += damage;
3002                 dir.nr++;
3003         }
3004
3005         /* This can happen even with many files, if everything was renames */
3006         if (!changed)
3007                 return;
3008
3009         /* Show all directories with more than x% of the changes */
3010         QSORT(dir.files, dir.nr, dirstat_compare);
3011         gather_dirstat(options, &dir, changed, "", 0);
3012 }
3013
3014 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3015 {
3016         int i;
3017         unsigned long changed;
3018         struct dirstat_dir dir;
3019
3020         if (data->nr == 0)
3021                 return;
3022
3023         dir.files = NULL;
3024         dir.alloc = 0;
3025         dir.nr = 0;
3026         dir.permille = options->dirstat_permille;
3027         dir.cumulative = options->flags.dirstat_cumulative;
3028
3029         changed = 0;
3030         for (i = 0; i < data->nr; i++) {
3031                 struct diffstat_file *file = data->files[i];
3032                 unsigned long damage = file->added + file->deleted;
3033                 if (file->is_binary)
3034                         /*
3035                          * binary files counts bytes, not lines. Must find some
3036                          * way to normalize binary bytes vs. textual lines.
3037                          * The following heuristic assumes that there are 64
3038                          * bytes per "line".
3039                          * This is stupid and ugly, but very cheap...
3040                          */
3041                         damage = DIV_ROUND_UP(damage, 64);
3042                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3043                 dir.files[dir.nr].name = file->name;
3044                 dir.files[dir.nr].changed = damage;
3045                 changed += damage;
3046                 dir.nr++;
3047         }
3048
3049         /* This can happen even with many files, if everything was renames */
3050         if (!changed)
3051                 return;
3052
3053         /* Show all directories with more than x% of the changes */
3054         QSORT(dir.files, dir.nr, dirstat_compare);
3055         gather_dirstat(options, &dir, changed, "", 0);
3056 }
3057
3058 static void free_diffstat_info(struct diffstat_t *diffstat)
3059 {
3060         int i;
3061         for (i = 0; i < diffstat->nr; i++) {
3062                 struct diffstat_file *f = diffstat->files[i];
3063                 free(f->print_name);
3064                 free(f->name);
3065                 free(f->from_name);
3066                 free(f);
3067         }
3068         free(diffstat->files);
3069 }
3070
3071 struct checkdiff_t {
3072         const char *filename;
3073         int lineno;
3074         int conflict_marker_size;
3075         struct diff_options *o;
3076         unsigned ws_rule;
3077         unsigned status;
3078 };
3079
3080 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3081 {
3082         char firstchar;
3083         int cnt;
3084
3085         if (len < marker_size + 1)
3086                 return 0;
3087         firstchar = line[0];
3088         switch (firstchar) {
3089         case '=': case '>': case '<': case '|':
3090                 break;
3091         default:
3092                 return 0;
3093         }
3094         for (cnt = 1; cnt < marker_size; cnt++)
3095                 if (line[cnt] != firstchar)
3096                         return 0;
3097         /* line[1] thru line[marker_size-1] are same as firstchar */
3098         if (len < marker_size + 1 || !isspace(line[marker_size]))
3099                 return 0;
3100         return 1;
3101 }
3102
3103 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3104 {
3105         struct checkdiff_t *data = priv;
3106         int marker_size = data->conflict_marker_size;
3107         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3108         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3109         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3110         char *err;
3111         const char *line_prefix;
3112
3113         assert(data->o);
3114         line_prefix = diff_line_prefix(data->o);
3115
3116         if (line[0] == '+') {
3117                 unsigned bad;
3118                 data->lineno++;
3119                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3120                         data->status |= 1;
3121                         fprintf(data->o->file,
3122                                 "%s%s:%d: leftover conflict marker\n",
3123                                 line_prefix, data->filename, data->lineno);
3124                 }
3125                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3126                 if (!bad)
3127                         return;
3128                 data->status |= bad;
3129                 err = whitespace_error_string(bad);
3130                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3131                         line_prefix, data->filename, data->lineno, err);
3132                 free(err);
3133                 emit_line(data->o, set, reset, line, 1);
3134                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3135                               data->o->file, set, reset, ws);
3136         } else if (line[0] == ' ') {
3137                 data->lineno++;
3138         } else if (line[0] == '@') {
3139                 char *plus = strchr(line, '+');
3140                 if (plus)
3141                         data->lineno = strtol(plus, NULL, 10) - 1;
3142                 else
3143                         die("invalid diff");
3144         }
3145 }
3146
3147 static unsigned char *deflate_it(char *data,
3148                                  unsigned long size,
3149                                  unsigned long *result_size)
3150 {
3151         int bound;
3152         unsigned char *deflated;
3153         git_zstream stream;
3154
3155         git_deflate_init(&stream, zlib_compression_level);
3156         bound = git_deflate_bound(&stream, size);
3157         deflated = xmalloc(bound);
3158         stream.next_out = deflated;
3159         stream.avail_out = bound;
3160
3161         stream.next_in = (unsigned char *)data;
3162         stream.avail_in = size;
3163         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3164                 ; /* nothing */
3165         git_deflate_end(&stream);
3166         *result_size = stream.total_out;
3167         return deflated;
3168 }
3169
3170 static void emit_binary_diff_body(struct diff_options *o,
3171                                   mmfile_t *one, mmfile_t *two)
3172 {
3173         void *cp;
3174         void *delta;
3175         void *deflated;
3176         void *data;
3177         unsigned long orig_size;
3178         unsigned long delta_size;
3179         unsigned long deflate_size;
3180         unsigned long data_size;
3181
3182         /* We could do deflated delta, or we could do just deflated two,
3183          * whichever is smaller.
3184          */
3185         delta = NULL;
3186         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3187         if (one->size && two->size) {
3188                 delta = diff_delta(one->ptr, one->size,
3189                                    two->ptr, two->size,
3190                                    &delta_size, deflate_size);
3191                 if (delta) {
3192                         void *to_free = delta;
3193                         orig_size = delta_size;
3194                         delta = deflate_it(delta, delta_size, &delta_size);
3195                         free(to_free);
3196                 }
3197         }
3198
3199         if (delta && delta_size < deflate_size) {
3200                 char *s = xstrfmt("%lu", orig_size);
3201                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3202                                  s, strlen(s), 0);
3203                 free(s);
3204                 free(deflated);
3205                 data = delta;
3206                 data_size = delta_size;
3207         } else {
3208                 char *s = xstrfmt("%lu", two->size);
3209                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3210                                  s, strlen(s), 0);
3211                 free(s);
3212                 free(delta);
3213                 data = deflated;
3214                 data_size = deflate_size;
3215         }
3216
3217         /* emit data encoded in base85 */
3218         cp = data;
3219         while (data_size) {
3220                 int len;
3221                 int bytes = (52 < data_size) ? 52 : data_size;
3222                 char line[71];
3223                 data_size -= bytes;
3224                 if (bytes <= 26)
3225                         line[0] = bytes + 'A' - 1;
3226                 else
3227                         line[0] = bytes - 26 + 'a' - 1;
3228                 encode_85(line + 1, cp, bytes);
3229                 cp = (char *) cp + bytes;
3230
3231                 len = strlen(line);
3232                 line[len++] = '\n';
3233                 line[len] = '\0';
3234
3235                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3236                                  line, len, 0);
3237         }
3238         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3239         free(data);
3240 }
3241
3242 static void emit_binary_diff(struct diff_options *o,
3243                              mmfile_t *one, mmfile_t *two)
3244 {
3245         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3246         emit_binary_diff_body(o, one, two);
3247         emit_binary_diff_body(o, two, one);
3248 }
3249
3250 int diff_filespec_is_binary(struct diff_filespec *one)
3251 {
3252         if (one->is_binary == -1) {
3253                 diff_filespec_load_driver(one);
3254                 if (one->driver->binary != -1)
3255                         one->is_binary = one->driver->binary;
3256                 else {
3257                         if (!one->data && DIFF_FILE_VALID(one))
3258                                 diff_populate_filespec(one, CHECK_BINARY);
3259                         if (one->is_binary == -1 && one->data)
3260                                 one->is_binary = buffer_is_binary(one->data,
3261                                                 one->size);
3262                         if (one->is_binary == -1)
3263                                 one->is_binary = 0;
3264                 }
3265         }
3266         return one->is_binary;
3267 }
3268
3269 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3270 {
3271         diff_filespec_load_driver(one);
3272         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3273 }
3274
3275 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3276 {
3277         if (!options->a_prefix)
3278                 options->a_prefix = a;
3279         if (!options->b_prefix)
3280                 options->b_prefix = b;
3281 }
3282
3283 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3284 {
3285         if (!DIFF_FILE_VALID(one))
3286                 return NULL;
3287
3288         diff_filespec_load_driver(one);
3289         return userdiff_get_textconv(one->driver);
3290 }
3291
3292 static void builtin_diff(const char *name_a,
3293                          const char *name_b,
3294                          struct diff_filespec *one,
3295                          struct diff_filespec *two,
3296                          const char *xfrm_msg,
3297                          int must_show_header,
3298                          struct diff_options *o,
3299                          int complete_rewrite)
3300 {
3301         mmfile_t mf1, mf2;
3302         const char *lbl[2];
3303         char *a_one, *b_two;
3304         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3305         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3306         const char *a_prefix, *b_prefix;
3307         struct userdiff_driver *textconv_one = NULL;
3308         struct userdiff_driver *textconv_two = NULL;
3309         struct strbuf header = STRBUF_INIT;
3310         const char *line_prefix = diff_line_prefix(o);
3311
3312         diff_set_mnemonic_prefix(o, "a/", "b/");
3313         if (o->flags.reverse_diff) {
3314                 a_prefix = o->b_prefix;
3315                 b_prefix = o->a_prefix;
3316         } else {
3317                 a_prefix = o->a_prefix;
3318                 b_prefix = o->b_prefix;
3319         }
3320
3321         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3322             (!one->mode || S_ISGITLINK(one->mode)) &&
3323             (!two->mode || S_ISGITLINK(two->mode))) {
3324                 show_submodule_summary(o, one->path ? one->path : two->path,
3325                                 &one->oid, &two->oid,
3326                                 two->dirty_submodule);
3327                 return;
3328         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3329                    (!one->mode || S_ISGITLINK(one->mode)) &&
3330                    (!two->mode || S_ISGITLINK(two->mode))) {
3331                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3332                                 &one->oid, &two->oid,
3333                                 two->dirty_submodule);
3334                 return;
3335         }
3336
3337         if (o->flags.allow_textconv) {
3338                 textconv_one = get_textconv(one);
3339                 textconv_two = get_textconv(two);
3340         }
3341
3342         /* Never use a non-valid filename anywhere if at all possible */
3343         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3344         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3345
3346         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3347         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3348         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3349         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3350         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3351         if (lbl[0][0] == '/') {
3352                 /* /dev/null */
3353                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3354                 if (xfrm_msg)
3355                         strbuf_addstr(&header, xfrm_msg);
3356                 must_show_header = 1;
3357         }
3358         else if (lbl[1][0] == '/') {
3359                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3360                 if (xfrm_msg)
3361                         strbuf_addstr(&header, xfrm_msg);
3362                 must_show_header = 1;
3363         }
3364         else {
3365                 if (one->mode != two->mode) {
3366                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3367                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3368                         must_show_header = 1;
3369                 }
3370                 if (xfrm_msg)
3371                         strbuf_addstr(&header, xfrm_msg);
3372
3373                 /*
3374                  * we do not run diff between different kind
3375                  * of objects.
3376                  */
3377                 if ((one->mode ^ two->mode) & S_IFMT)
3378                         goto free_ab_and_return;
3379                 if (complete_rewrite &&
3380                     (textconv_one || !diff_filespec_is_binary(one)) &&
3381                     (textconv_two || !diff_filespec_is_binary(two))) {
3382                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3383                                          header.buf, header.len, 0);
3384                         strbuf_reset(&header);
3385                         emit_rewrite_diff(name_a, name_b, one, two,
3386                                                 textconv_one, textconv_two, o);
3387                         o->found_changes = 1;
3388                         goto free_ab_and_return;
3389                 }
3390         }
3391
3392         if (o->irreversible_delete && lbl[1][0] == '/') {
3393                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3394                                  header.len, 0);
3395                 strbuf_reset(&header);
3396                 goto free_ab_and_return;
3397         } else if (!o->flags.text &&
3398             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3399               (!textconv_two && diff_filespec_is_binary(two)) )) {
3400                 struct strbuf sb = STRBUF_INIT;
3401                 if (!one->data && !two->data &&
3402                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3403                     !o->flags.binary) {
3404                         if (!oidcmp(&one->oid, &two->oid)) {
3405                                 if (must_show_header)
3406                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3407                                                          header.buf, header.len,
3408                                                          0);
3409                                 goto free_ab_and_return;
3410                         }
3411                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3412                                          header.buf, header.len, 0);
3413                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3414                                     diff_line_prefix(o), lbl[0], lbl[1]);
3415                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3416                                          sb.buf, sb.len, 0);
3417                         strbuf_release(&sb);
3418                         goto free_ab_and_return;
3419                 }
3420                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3421                         die("unable to read files to diff");
3422                 /* Quite common confusing case */
3423                 if (mf1.size == mf2.size &&
3424                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3425                         if (must_show_header)
3426                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3427                                                  header.buf, header.len, 0);
3428                         goto free_ab_and_return;
3429                 }
3430                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3431                 strbuf_reset(&header);
3432                 if (o->flags.binary)
3433                         emit_binary_diff(o, &mf1, &mf2);
3434                 else {
3435                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3436                                     diff_line_prefix(o), lbl[0], lbl[1]);
3437                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3438                                          sb.buf, sb.len, 0);
3439                         strbuf_release(&sb);
3440                 }
3441                 o->found_changes = 1;
3442         } else {
3443                 /* Crazy xdl interfaces.. */
3444                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3445                 const char *v;
3446                 xpparam_t xpp;
3447                 xdemitconf_t xecfg;
3448                 struct emit_callback ecbdata;
3449                 const struct userdiff_funcname *pe;
3450
3451                 if (must_show_header) {
3452                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3453                                          header.buf, header.len, 0);
3454                         strbuf_reset(&header);
3455                 }
3456
3457                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3458                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3459
3460                 pe = diff_funcname_pattern(one);
3461                 if (!pe)
3462                         pe = diff_funcname_pattern(two);
3463
3464                 memset(&xpp, 0, sizeof(xpp));
3465                 memset(&xecfg, 0, sizeof(xecfg));
3466                 memset(&ecbdata, 0, sizeof(ecbdata));
3467                 if (o->flags.suppress_diff_headers)
3468                         lbl[0] = NULL;
3469                 ecbdata.label_path = lbl;
3470                 ecbdata.color_diff = want_color(o->use_color);
3471                 ecbdata.ws_rule = whitespace_rule(name_b);
3472                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3473                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3474                 ecbdata.opt = o;
3475                 if (header.len && !o->flags.suppress_diff_headers)
3476                         ecbdata.header = &header;
3477                 xpp.flags = o->xdl_opts;
3478                 xpp.anchors = o->anchors;
3479                 xpp.anchors_nr = o->anchors_nr;
3480                 xecfg.ctxlen = o->context;
3481                 xecfg.interhunkctxlen = o->interhunkcontext;
3482                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3483                 if (o->flags.funccontext)
3484                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3485                 if (pe)
3486                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3487                 if (!diffopts)
3488                         ;
3489                 else if (skip_prefix(diffopts, "--unified=", &v))
3490                         xecfg.ctxlen = strtoul(v, NULL, 10);
3491                 else if (skip_prefix(diffopts, "-u", &v))
3492                         xecfg.ctxlen = strtoul(v, NULL, 10);
3493                 if (o->word_diff)
3494                         init_diff_words_data(&ecbdata, o, one, two);
3495                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3496                                   &xpp, &xecfg))
3497                         die("unable to generate diff for %s", one->path);
3498                 if (o->word_diff)
3499                         free_diff_words_data(&ecbdata);
3500                 if (textconv_one)
3501                         free(mf1.ptr);
3502                 if (textconv_two)
3503                         free(mf2.ptr);
3504                 xdiff_clear_find_func(&xecfg);
3505         }
3506
3507  free_ab_and_return:
3508         strbuf_release(&header);
3509         diff_free_filespec_data(one);
3510         diff_free_filespec_data(two);
3511         free(a_one);
3512         free(b_two);
3513         return;
3514 }
3515
3516 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3517 {
3518         if (!is_renamed) {
3519                 if (p->status == DIFF_STATUS_ADDED) {
3520                         if (S_ISLNK(p->two->mode))
3521                                 return "new +l";
3522                         else if ((p->two->mode & 0777) == 0755)
3523                                 return "new +x";
3524                         else
3525                                 return "new";
3526                 } else if (p->status == DIFF_STATUS_DELETED)
3527                         return "gone";
3528         }
3529         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3530                 return "mode -l";
3531         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3532                 return "mode +l";
3533         else if ((p->one->mode & 0777) == 0644 &&
3534                  (p->two->mode & 0777) == 0755)
3535                 return "mode +x";
3536         else if ((p->one->mode & 0777) == 0755 &&
3537                  (p->two->mode & 0777) == 0644)
3538                 return "mode -x";
3539         return NULL;
3540 }
3541
3542 static void builtin_diffstat(const char *name_a, const char *name_b,
3543                              struct diff_filespec *one,
3544                              struct diff_filespec *two,
3545                              struct diffstat_t *diffstat,
3546                              struct diff_options *o,
3547                              struct diff_filepair *p)
3548 {
3549         mmfile_t mf1, mf2;
3550         struct diffstat_file *data;
3551         int same_contents;
3552         int complete_rewrite = 0;
3553
3554         if (!DIFF_PAIR_UNMERGED(p)) {
3555                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3556                         complete_rewrite = 1;
3557         }
3558
3559         data = diffstat_add(diffstat, name_a, name_b);
3560         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3561         if (o->flags.stat_with_summary)
3562                 data->comments = get_compact_summary(p, data->is_renamed);
3563
3564         if (!one || !two) {
3565                 data->is_unmerged = 1;
3566                 return;
3567         }
3568
3569         same_contents = !oidcmp(&one->oid, &two->oid);
3570
3571         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3572                 data->is_binary = 1;
3573                 if (same_contents) {
3574                         data->added = 0;
3575                         data->deleted = 0;
3576                 } else {
3577                         data->added = diff_filespec_size(two);
3578                         data->deleted = diff_filespec_size(one);
3579                 }
3580         }
3581
3582         else if (complete_rewrite) {
3583                 diff_populate_filespec(one, 0);
3584                 diff_populate_filespec(two, 0);
3585                 data->deleted = count_lines(one->data, one->size);
3586                 data->added = count_lines(two->data, two->size);
3587         }
3588
3589         else if (!same_contents) {
3590                 /* Crazy xdl interfaces.. */
3591                 xpparam_t xpp;
3592                 xdemitconf_t xecfg;
3593
3594                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3595                         die("unable to read files to diff");
3596
3597                 memset(&xpp, 0, sizeof(xpp));
3598                 memset(&xecfg, 0, sizeof(xecfg));
3599                 xpp.flags = o->xdl_opts;
3600                 xpp.anchors = o->anchors;
3601                 xpp.anchors_nr = o->anchors_nr;
3602                 xecfg.ctxlen = o->context;
3603                 xecfg.interhunkctxlen = o->interhunkcontext;
3604                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3605                                   &xpp, &xecfg))
3606                         die("unable to generate diffstat for %s", one->path);
3607         }
3608
3609         diff_free_filespec_data(one);
3610         diff_free_filespec_data(two);
3611 }
3612
3613 static void builtin_checkdiff(const char *name_a, const char *name_b,
3614                               const char *attr_path,
3615                               struct diff_filespec *one,
3616                               struct diff_filespec *two,
3617                               struct diff_options *o)
3618 {
3619         mmfile_t mf1, mf2;
3620         struct checkdiff_t data;
3621
3622         if (!two)
3623                 return;
3624
3625         memset(&data, 0, sizeof(data));
3626         data.filename = name_b ? name_b : name_a;
3627         data.lineno = 0;
3628         data.o = o;
3629         data.ws_rule = whitespace_rule(attr_path);
3630         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3631
3632         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3633                 die("unable to read files to diff");
3634
3635         /*
3636          * All the other codepaths check both sides, but not checking
3637          * the "old" side here is deliberate.  We are checking the newly
3638          * introduced changes, and as long as the "new" side is text, we
3639          * can and should check what it introduces.
3640          */
3641         if (diff_filespec_is_binary(two))
3642                 goto free_and_return;
3643         else {
3644                 /* Crazy xdl interfaces.. */
3645                 xpparam_t xpp;
3646                 xdemitconf_t xecfg;
3647
3648                 memset(&xpp, 0, sizeof(xpp));
3649                 memset(&xecfg, 0, sizeof(xecfg));
3650                 xecfg.ctxlen = 1; /* at least one context line */
3651                 xpp.flags = 0;
3652                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3653                                   &xpp, &xecfg))
3654                         die("unable to generate checkdiff for %s", one->path);
3655
3656                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3657                         struct emit_callback ecbdata;
3658                         int blank_at_eof;
3659
3660                         ecbdata.ws_rule = data.ws_rule;
3661                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3662                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3663
3664                         if (blank_at_eof) {
3665                                 static char *err;
3666                                 if (!err)
3667                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3668                                 fprintf(o->file, "%s:%d: %s.\n",
3669                                         data.filename, blank_at_eof, err);
3670                                 data.status = 1; /* report errors */
3671                         }
3672                 }
3673         }
3674  free_and_return:
3675         diff_free_filespec_data(one);
3676         diff_free_filespec_data(two);
3677         if (data.status)
3678                 o->flags.check_failed = 1;
3679 }
3680
3681 struct diff_filespec *alloc_filespec(const char *path)
3682 {
3683         struct diff_filespec *spec;
3684
3685         FLEXPTR_ALLOC_STR(spec, path, path);
3686         spec->count = 1;
3687         spec->is_binary = -1;
3688         return spec;
3689 }
3690
3691 void free_filespec(struct diff_filespec *spec)
3692 {
3693         if (!--spec->count) {
3694                 diff_free_filespec_data(spec);
3695                 free(spec);
3696         }
3697 }
3698
3699 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3700                    int oid_valid, unsigned short mode)
3701 {
3702         if (mode) {
3703                 spec->mode = canon_mode(mode);
3704                 oidcpy(&spec->oid, oid);
3705                 spec->oid_valid = oid_valid;
3706         }
3707 }
3708
3709 /*
3710  * Given a name and sha1 pair, if the index tells us the file in
3711  * the work tree has that object contents, return true, so that
3712  * prepare_temp_file() does not have to inflate and extract.
3713  */
3714 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3715 {
3716         const struct cache_entry *ce;
3717         struct stat st;
3718         int pos, len;
3719
3720         /*
3721          * We do not read the cache ourselves here, because the
3722          * benchmark with my previous version that always reads cache
3723          * shows that it makes things worse for diff-tree comparing
3724          * two linux-2.6 kernel trees in an already checked out work
3725          * tree.  This is because most diff-tree comparisons deal with
3726          * only a small number of files, while reading the cache is
3727          * expensive for a large project, and its cost outweighs the
3728          * savings we get by not inflating the object to a temporary
3729          * file.  Practically, this code only helps when we are used
3730          * by diff-cache --cached, which does read the cache before
3731          * calling us.
3732          */
3733         if (!active_cache)
3734                 return 0;
3735
3736         /* We want to avoid the working directory if our caller
3737          * doesn't need the data in a normal file, this system
3738          * is rather slow with its stat/open/mmap/close syscalls,
3739          * and the object is contained in a pack file.  The pack
3740          * is probably already open and will be faster to obtain
3741          * the data through than the working directory.  Loose
3742          * objects however would tend to be slower as they need
3743          * to be individually opened and inflated.
3744          */
3745         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3746                 return 0;
3747
3748         /*
3749          * Similarly, if we'd have to convert the file contents anyway, that
3750          * makes the optimization not worthwhile.
3751          */
3752         if (!want_file && would_convert_to_git(&the_index, name))
3753                 return 0;
3754
3755         len = strlen(name);
3756         pos = cache_name_pos(name, len);
3757         if (pos < 0)
3758                 return 0;
3759         ce = active_cache[pos];
3760
3761         /*
3762          * This is not the sha1 we are looking for, or
3763          * unreusable because it is not a regular file.
3764          */
3765         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3766                 return 0;
3767
3768         /*
3769          * If ce is marked as "assume unchanged", there is no
3770          * guarantee that work tree matches what we are looking for.
3771          */
3772         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3773                 return 0;
3774
3775         /*
3776          * If ce matches the file in the work tree, we can reuse it.
3777          */
3778         if (ce_uptodate(ce) ||
3779             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3780                 return 1;
3781
3782         return 0;
3783 }
3784
3785 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3786 {
3787         struct strbuf buf = STRBUF_INIT;
3788         char *dirty = "";
3789
3790         /* Are we looking at the work tree? */
3791         if (s->dirty_submodule)
3792                 dirty = "-dirty";
3793
3794         strbuf_addf(&buf, "Subproject commit %s%s\n",
3795                     oid_to_hex(&s->oid), dirty);
3796         s->size = buf.len;
3797         if (size_only) {
3798                 s->data = NULL;
3799                 strbuf_release(&buf);
3800         } else {
3801                 s->data = strbuf_detach(&buf, NULL);
3802                 s->should_free = 1;
3803         }
3804         return 0;
3805 }
3806
3807 /*
3808  * While doing rename detection and pickaxe operation, we may need to
3809  * grab the data for the blob (or file) for our own in-core comparison.
3810  * diff_filespec has data and size fields for this purpose.
3811  */
3812 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3813 {
3814         int size_only = flags & CHECK_SIZE_ONLY;
3815         int err = 0;
3816         int conv_flags = global_conv_flags_eol;
3817         /*
3818          * demote FAIL to WARN to allow inspecting the situation
3819          * instead of refusing.
3820          */
3821         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3822                 conv_flags = CONV_EOL_RNDTRP_WARN;
3823
3824         if (!DIFF_FILE_VALID(s))
3825                 die("internal error: asking to populate invalid file.");
3826         if (S_ISDIR(s->mode))
3827                 return -1;
3828
3829         if (s->data)
3830                 return 0;
3831
3832         if (size_only && 0 < s->size)
3833                 return 0;
3834
3835         if (S_ISGITLINK(s->mode))
3836                 return diff_populate_gitlink(s, size_only);
3837
3838         if (!s->oid_valid ||
3839             reuse_worktree_file(s->path, &s->oid, 0)) {
3840                 struct strbuf buf = STRBUF_INIT;
3841                 struct stat st;
3842                 int fd;
3843
3844                 if (lstat(s->path, &st) < 0) {
3845                 err_empty:
3846                         err = -1;
3847                 empty:
3848                         s->data = (char *)"";
3849                         s->size = 0;
3850                         return err;
3851                 }
3852                 s->size = xsize_t(st.st_size);
3853                 if (!s->size)
3854                         goto empty;
3855                 if (S_ISLNK(st.st_mode)) {
3856                         struct strbuf sb = STRBUF_INIT;
3857
3858                         if (strbuf_readlink(&sb, s->path, s->size))
3859                                 goto err_empty;
3860                         s->size = sb.len;
3861                         s->data = strbuf_detach(&sb, NULL);
3862                         s->should_free = 1;
3863                         return 0;
3864                 }
3865
3866                 /*
3867                  * Even if the caller would be happy with getting
3868                  * only the size, we cannot return early at this
3869                  * point if the path requires us to run the content
3870                  * conversion.
3871                  */
3872                 if (size_only && !would_convert_to_git(&the_index, s->path))
3873                         return 0;
3874
3875                 /*
3876                  * Note: this check uses xsize_t(st.st_size) that may
3877                  * not be the true size of the blob after it goes
3878                  * through convert_to_git().  This may not strictly be
3879                  * correct, but the whole point of big_file_threshold
3880                  * and is_binary check being that we want to avoid
3881                  * opening the file and inspecting the contents, this
3882                  * is probably fine.
3883                  */
3884                 if ((flags & CHECK_BINARY) &&
3885                     s->size > big_file_threshold && s->is_binary == -1) {
3886                         s->is_binary = 1;
3887                         return 0;
3888                 }
3889                 fd = open(s->path, O_RDONLY);
3890                 if (fd < 0)
3891                         goto err_empty;
3892                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3893                 close(fd);
3894                 s->should_munmap = 1;
3895
3896                 /*
3897                  * Convert from working tree format to canonical git format
3898                  */
3899                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, conv_flags)) {
3900                         size_t size = 0;
3901                         munmap(s->data, s->size);
3902                         s->should_munmap = 0;
3903                         s->data = strbuf_detach(&buf, &size);
3904                         s->size = size;
3905                         s->should_free = 1;
3906                 }
3907         }
3908         else {
3909                 enum object_type type;
3910                 if (size_only || (flags & CHECK_BINARY)) {
3911                         type = oid_object_info(the_repository, &s->oid,
3912                                                &s->size);
3913                         if (type < 0)
3914                                 die("unable to read %s",
3915                                     oid_to_hex(&s->oid));
3916                         if (size_only)
3917                                 return 0;
3918                         if (s->size > big_file_threshold && s->is_binary == -1) {
3919                                 s->is_binary = 1;
3920                                 return 0;
3921                         }
3922                 }
3923                 s->data = read_object_file(&s->oid, &type, &s->size);
3924                 if (!s->data)
3925                         die("unable to read %s", oid_to_hex(&s->oid));
3926                 s->should_free = 1;
3927         }
3928         return 0;
3929 }
3930
3931 void diff_free_filespec_blob(struct diff_filespec *s)
3932 {
3933         if (s->should_free)
3934                 free(s->data);
3935         else if (s->should_munmap)
3936                 munmap(s->data, s->size);
3937
3938         if (s->should_free || s->should_munmap) {
3939                 s->should_free = s->should_munmap = 0;
3940                 s->data = NULL;
3941         }
3942 }
3943
3944 void diff_free_filespec_data(struct diff_filespec *s)
3945 {
3946         diff_free_filespec_blob(s);
3947         FREE_AND_NULL(s->cnt_data);
3948 }
3949
3950 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3951                            void *blob,
3952                            unsigned long size,
3953                            const struct object_id *oid,
3954                            int mode)
3955 {
3956         struct strbuf buf = STRBUF_INIT;
3957         struct strbuf tempfile = STRBUF_INIT;
3958         char *path_dup = xstrdup(path);
3959         const char *base = basename(path_dup);
3960
3961         /* Generate "XXXXXX_basename.ext" */
3962         strbuf_addstr(&tempfile, "XXXXXX_");
3963         strbuf_addstr(&tempfile, base);
3964
3965         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3966         if (!temp->tempfile)
3967                 die_errno("unable to create temp-file");
3968         if (convert_to_working_tree(path,
3969                         (const char *)blob, (size_t)size, &buf)) {
3970                 blob = buf.buf;
3971                 size = buf.len;
3972         }
3973         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3974             close_tempfile_gently(temp->tempfile))
3975                 die_errno("unable to write temp-file");
3976         temp->name = get_tempfile_path(temp->tempfile);
3977         oid_to_hex_r(temp->hex, oid);
3978         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3979         strbuf_release(&buf);
3980         strbuf_release(&tempfile);
3981         free(path_dup);
3982 }
3983
3984 static struct diff_tempfile *prepare_temp_file(const char *name,
3985                 struct diff_filespec *one)
3986 {
3987         struct diff_tempfile *temp = claim_diff_tempfile();
3988
3989         if (!DIFF_FILE_VALID(one)) {
3990         not_a_valid_file:
3991                 /* A '-' entry produces this for file-2, and
3992                  * a '+' entry produces this for file-1.
3993                  */
3994                 temp->name = "/dev/null";
3995                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3996                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3997                 return temp;
3998         }
3999
4000         if (!S_ISGITLINK(one->mode) &&
4001             (!one->oid_valid ||
4002              reuse_worktree_file(name, &one->oid, 1))) {
4003                 struct stat st;
4004                 if (lstat(name, &st) < 0) {
4005                         if (errno == ENOENT)
4006                                 goto not_a_valid_file;
4007                         die_errno("stat(%s)", name);
4008                 }
4009                 if (S_ISLNK(st.st_mode)) {
4010                         struct strbuf sb = STRBUF_INIT;
4011                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4012                                 die_errno("readlink(%s)", name);
4013                         prep_temp_blob(name, temp, sb.buf, sb.len,
4014                                        (one->oid_valid ?
4015                                         &one->oid : &null_oid),
4016                                        (one->oid_valid ?
4017                                         one->mode : S_IFLNK));
4018                         strbuf_release(&sb);
4019                 }
4020                 else {
4021                         /* we can borrow from the file in the work tree */
4022                         temp->name = name;
4023                         if (!one->oid_valid)
4024                                 oid_to_hex_r(temp->hex, &null_oid);
4025                         else
4026                                 oid_to_hex_r(temp->hex, &one->oid);
4027                         /* Even though we may sometimes borrow the
4028                          * contents from the work tree, we always want
4029                          * one->mode.  mode is trustworthy even when
4030                          * !(one->oid_valid), as long as
4031                          * DIFF_FILE_VALID(one).
4032                          */
4033                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4034                 }
4035                 return temp;
4036         }
4037         else {
4038                 if (diff_populate_filespec(one, 0))
4039                         die("cannot read data blob for %s", one->path);
4040                 prep_temp_blob(name, temp, one->data, one->size,
4041                                &one->oid, one->mode);
4042         }
4043         return temp;
4044 }
4045
4046 static void add_external_diff_name(struct argv_array *argv,
4047                                    const char *name,
4048                                    struct diff_filespec *df)
4049 {
4050         struct diff_tempfile *temp = prepare_temp_file(name, df);
4051         argv_array_push(argv, temp->name);
4052         argv_array_push(argv, temp->hex);
4053         argv_array_push(argv, temp->mode);
4054 }
4055
4056 /* An external diff command takes:
4057  *
4058  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4059  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4060  *
4061  */
4062 static void run_external_diff(const char *pgm,
4063                               const char *name,
4064                               const char *other,
4065                               struct diff_filespec *one,
4066                               struct diff_filespec *two,
4067                               const char *xfrm_msg,
4068                               int complete_rewrite,
4069                               struct diff_options *o)
4070 {
4071         struct argv_array argv = ARGV_ARRAY_INIT;
4072         struct argv_array env = ARGV_ARRAY_INIT;
4073         struct diff_queue_struct *q = &diff_queued_diff;
4074
4075         argv_array_push(&argv, pgm);
4076         argv_array_push(&argv, name);
4077
4078         if (one && two) {
4079                 add_external_diff_name(&argv, name, one);
4080                 if (!other)
4081                         add_external_diff_name(&argv, name, two);
4082                 else {
4083                         add_external_diff_name(&argv, other, two);
4084                         argv_array_push(&argv, other);
4085                         argv_array_push(&argv, xfrm_msg);
4086                 }
4087         }
4088
4089         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4090         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4091
4092         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4093                 die(_("external diff died, stopping at %s"), name);
4094
4095         remove_tempfile();
4096         argv_array_clear(&argv);
4097         argv_array_clear(&env);
4098 }
4099
4100 static int similarity_index(struct diff_filepair *p)
4101 {
4102         return p->score * 100 / MAX_SCORE;
4103 }
4104
4105 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4106 {
4107         if (startup_info->have_repository)
4108                 return find_unique_abbrev(oid, abbrev);
4109         else {
4110                 char *hex = oid_to_hex(oid);
4111                 if (abbrev < 0)
4112                         abbrev = FALLBACK_DEFAULT_ABBREV;
4113                 if (abbrev > the_hash_algo->hexsz)
4114                         BUG("oid abbreviation out of range: %d", abbrev);
4115                 if (abbrev)
4116                         hex[abbrev] = '\0';
4117                 return hex;
4118         }
4119 }
4120
4121 static void fill_metainfo(struct strbuf *msg,
4122                           const char *name,
4123                           const char *other,
4124                           struct diff_filespec *one,
4125                           struct diff_filespec *two,
4126                           struct diff_options *o,
4127                           struct diff_filepair *p,
4128                           int *must_show_header,
4129                           int use_color)
4130 {
4131         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4132         const char *reset = diff_get_color(use_color, DIFF_RESET);
4133         const char *line_prefix = diff_line_prefix(o);
4134
4135         *must_show_header = 1;
4136         strbuf_init(msg, PATH_MAX * 2 + 300);
4137         switch (p->status) {
4138         case DIFF_STATUS_COPIED:
4139                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4140                             line_prefix, set, similarity_index(p));
4141                 strbuf_addf(msg, "%s\n%s%scopy from ",
4142                             reset,  line_prefix, set);
4143                 quote_c_style(name, msg, NULL, 0);
4144                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4145                 quote_c_style(other, msg, NULL, 0);
4146                 strbuf_addf(msg, "%s\n", reset);
4147                 break;
4148         case DIFF_STATUS_RENAMED:
4149                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4150                             line_prefix, set, similarity_index(p));
4151                 strbuf_addf(msg, "%s\n%s%srename from ",
4152                             reset, line_prefix, set);
4153                 quote_c_style(name, msg, NULL, 0);
4154                 strbuf_addf(msg, "%s\n%s%srename to ",
4155                             reset, line_prefix, set);
4156                 quote_c_style(other, msg, NULL, 0);
4157                 strbuf_addf(msg, "%s\n", reset);
4158                 break;
4159         case DIFF_STATUS_MODIFIED:
4160                 if (p->score) {
4161                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4162                                     line_prefix,
4163                                     set, similarity_index(p), reset);
4164                         break;
4165                 }
4166                 /* fallthru */
4167         default:
4168                 *must_show_header = 0;
4169         }
4170         if (one && two && oidcmp(&one->oid, &two->oid)) {
4171                 const unsigned hexsz = the_hash_algo->hexsz;
4172                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4173
4174                 if (o->flags.binary) {
4175                         mmfile_t mf;
4176                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
4177                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
4178                                 abbrev = hexsz;
4179                 }
4180                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4181                             diff_abbrev_oid(&one->oid, abbrev),
4182                             diff_abbrev_oid(&two->oid, abbrev));
4183                 if (one->mode == two->mode)
4184                         strbuf_addf(msg, " %06o", one->mode);
4185                 strbuf_addf(msg, "%s\n", reset);
4186         }
4187 }
4188
4189 static void run_diff_cmd(const char *pgm,
4190                          const char *name,
4191                          const char *other,
4192                          const char *attr_path,
4193                          struct diff_filespec *one,
4194                          struct diff_filespec *two,
4195                          struct strbuf *msg,
4196                          struct diff_options *o,
4197                          struct diff_filepair *p)
4198 {
4199         const char *xfrm_msg = NULL;
4200         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4201         int must_show_header = 0;
4202
4203
4204         if (o->flags.allow_external) {
4205                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
4206                 if (drv && drv->external)
4207                         pgm = drv->external;
4208         }
4209
4210         if (msg) {
4211                 /*
4212                  * don't use colors when the header is intended for an
4213                  * external diff driver
4214                  */
4215                 fill_metainfo(msg, name, other, one, two, o, p,
4216                               &must_show_header,
4217                               want_color(o->use_color) && !pgm);
4218                 xfrm_msg = msg->len ? msg->buf : NULL;
4219         }
4220
4221         if (pgm) {
4222                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4223                                   complete_rewrite, o);
4224                 return;
4225         }
4226         if (one && two)
4227                 builtin_diff(name, other ? other : name,
4228                              one, two, xfrm_msg, must_show_header,
4229                              o, complete_rewrite);
4230         else
4231                 fprintf(o->file, "* Unmerged path %s\n", name);
4232 }
4233
4234 static void diff_fill_oid_info(struct diff_filespec *one)
4235 {
4236         if (DIFF_FILE_VALID(one)) {
4237                 if (!one->oid_valid) {
4238                         struct stat st;
4239                         if (one->is_stdin) {
4240                                 oidclr(&one->oid);
4241                                 return;
4242                         }
4243                         if (lstat(one->path, &st) < 0)
4244                                 die_errno("stat '%s'", one->path);
4245                         if (index_path(&one->oid, one->path, &st, 0))
4246                                 die("cannot hash %s", one->path);
4247                 }
4248         }
4249         else
4250                 oidclr(&one->oid);
4251 }
4252
4253 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4254 {
4255         /* Strip the prefix but do not molest /dev/null and absolute paths */
4256         if (*namep && **namep != '/') {
4257                 *namep += prefix_length;
4258                 if (**namep == '/')
4259                         ++*namep;
4260         }
4261         if (*otherp && **otherp != '/') {
4262                 *otherp += prefix_length;
4263                 if (**otherp == '/')
4264                         ++*otherp;
4265         }
4266 }
4267
4268 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4269 {
4270         const char *pgm = external_diff();
4271         struct strbuf msg;
4272         struct diff_filespec *one = p->one;
4273         struct diff_filespec *two = p->two;
4274         const char *name;
4275         const char *other;
4276         const char *attr_path;
4277
4278         name  = one->path;
4279         other = (strcmp(name, two->path) ? two->path : NULL);
4280         attr_path = name;
4281         if (o->prefix_length)
4282                 strip_prefix(o->prefix_length, &name, &other);
4283
4284         if (!o->flags.allow_external)
4285                 pgm = NULL;
4286
4287         if (DIFF_PAIR_UNMERGED(p)) {
4288                 run_diff_cmd(pgm, name, NULL, attr_path,
4289                              NULL, NULL, NULL, o, p);
4290                 return;
4291         }
4292
4293         diff_fill_oid_info(one);
4294         diff_fill_oid_info(two);
4295
4296         if (!pgm &&
4297             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4298             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4299                 /*
4300                  * a filepair that changes between file and symlink
4301                  * needs to be split into deletion and creation.
4302                  */
4303                 struct diff_filespec *null = alloc_filespec(two->path);
4304                 run_diff_cmd(NULL, name, other, attr_path,
4305                              one, null, &msg, o, p);
4306                 free(null);
4307                 strbuf_release(&msg);
4308
4309                 null = alloc_filespec(one->path);
4310                 run_diff_cmd(NULL, name, other, attr_path,
4311                              null, two, &msg, o, p);
4312                 free(null);
4313         }
4314         else
4315                 run_diff_cmd(pgm, name, other, attr_path,
4316                              one, two, &msg, o, p);
4317
4318         strbuf_release(&msg);
4319 }
4320
4321 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4322                          struct diffstat_t *diffstat)
4323 {
4324         const char *name;
4325         const char *other;
4326
4327         if (DIFF_PAIR_UNMERGED(p)) {
4328                 /* unmerged */
4329                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4330                 return;
4331         }
4332
4333         name = p->one->path;
4334         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4335
4336         if (o->prefix_length)
4337                 strip_prefix(o->prefix_length, &name, &other);
4338
4339         diff_fill_oid_info(p->one);
4340         diff_fill_oid_info(p->two);
4341
4342         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4343 }
4344
4345 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4346 {
4347         const char *name;
4348         const char *other;
4349         const char *attr_path;
4350
4351         if (DIFF_PAIR_UNMERGED(p)) {
4352                 /* unmerged */
4353                 return;
4354         }
4355
4356         name = p->one->path;
4357         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4358         attr_path = other ? other : name;
4359
4360         if (o->prefix_length)
4361                 strip_prefix(o->prefix_length, &name, &other);
4362
4363         diff_fill_oid_info(p->one);
4364         diff_fill_oid_info(p->two);
4365
4366         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4367 }
4368
4369 void diff_setup(struct diff_options *options)
4370 {
4371         memcpy(options, &default_diff_options, sizeof(*options));
4372
4373         options->file = stdout;
4374
4375         options->abbrev = DEFAULT_ABBREV;
4376         options->line_termination = '\n';
4377         options->break_opt = -1;
4378         options->rename_limit = -1;
4379         options->dirstat_permille = diff_dirstat_permille_default;
4380         options->context = diff_context_default;
4381         options->interhunkcontext = diff_interhunk_context_default;
4382         options->ws_error_highlight = ws_error_highlight_default;
4383         options->flags.rename_empty = 1;
4384         options->objfind = NULL;
4385
4386         /* pathchange left =NULL by default */
4387         options->change = diff_change;
4388         options->add_remove = diff_addremove;
4389         options->use_color = diff_use_color_default;
4390         options->detect_rename = diff_detect_rename_default;
4391         options->xdl_opts |= diff_algorithm;
4392         if (diff_indent_heuristic)
4393                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4394
4395         options->orderfile = diff_order_file_cfg;
4396
4397         if (diff_no_prefix) {
4398                 options->a_prefix = options->b_prefix = "";
4399         } else if (!diff_mnemonic_prefix) {
4400                 options->a_prefix = "a/";
4401                 options->b_prefix = "b/";
4402         }
4403
4404         options->color_moved = diff_color_moved_default;
4405         options->color_moved_ws_handling = diff_color_moved_ws_default;
4406 }
4407
4408 void diff_setup_done(struct diff_options *options)
4409 {
4410         unsigned check_mask = DIFF_FORMAT_NAME |
4411                               DIFF_FORMAT_NAME_STATUS |
4412                               DIFF_FORMAT_CHECKDIFF |
4413                               DIFF_FORMAT_NO_OUTPUT;
4414         /*
4415          * This must be signed because we're comparing against a potentially
4416          * negative value.
4417          */
4418         const int hexsz = the_hash_algo->hexsz;
4419
4420         if (options->set_default)
4421                 options->set_default(options);
4422
4423         if (HAS_MULTI_BITS(options->output_format & check_mask))
4424                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4425
4426         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4427                 die(_("-G, -S and --find-object are mutually exclusive"));
4428
4429         /*
4430          * Most of the time we can say "there are changes"
4431          * only by checking if there are changed paths, but
4432          * --ignore-whitespace* options force us to look
4433          * inside contents.
4434          */
4435
4436         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4437                 options->flags.diff_from_contents = 1;
4438         else
4439                 options->flags.diff_from_contents = 0;
4440
4441         if (options->flags.find_copies_harder)
4442                 options->detect_rename = DIFF_DETECT_COPY;
4443
4444         if (!options->flags.relative_name)
4445                 options->prefix = NULL;
4446         if (options->prefix)
4447                 options->prefix_length = strlen(options->prefix);
4448         else
4449                 options->prefix_length = 0;
4450
4451         if (options->output_format & (DIFF_FORMAT_NAME |
4452                                       DIFF_FORMAT_NAME_STATUS |
4453                                       DIFF_FORMAT_CHECKDIFF |
4454                                       DIFF_FORMAT_NO_OUTPUT))
4455                 options->output_format &= ~(DIFF_FORMAT_RAW |
4456                                             DIFF_FORMAT_NUMSTAT |
4457                                             DIFF_FORMAT_DIFFSTAT |
4458                                             DIFF_FORMAT_SHORTSTAT |
4459                                             DIFF_FORMAT_DIRSTAT |
4460                                             DIFF_FORMAT_SUMMARY |
4461                                             DIFF_FORMAT_PATCH);
4462
4463         /*
4464          * These cases always need recursive; we do not drop caller-supplied
4465          * recursive bits for other formats here.
4466          */
4467         if (options->output_format & (DIFF_FORMAT_PATCH |
4468                                       DIFF_FORMAT_NUMSTAT |
4469                                       DIFF_FORMAT_DIFFSTAT |
4470                                       DIFF_FORMAT_SHORTSTAT |
4471                                       DIFF_FORMAT_DIRSTAT |
4472                                       DIFF_FORMAT_SUMMARY |
4473                                       DIFF_FORMAT_CHECKDIFF))
4474                 options->flags.recursive = 1;
4475         /*
4476          * Also pickaxe would not work very well if you do not say recursive
4477          */
4478         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4479                 options->flags.recursive = 1;
4480         /*
4481          * When patches are generated, submodules diffed against the work tree
4482          * must be checked for dirtiness too so it can be shown in the output
4483          */
4484         if (options->output_format & DIFF_FORMAT_PATCH)
4485                 options->flags.dirty_submodules = 1;
4486
4487         if (options->detect_rename && options->rename_limit < 0)
4488                 options->rename_limit = diff_rename_limit_default;
4489         if (options->setup & DIFF_SETUP_USE_CACHE) {
4490                 if (!active_cache)
4491                         /* read-cache does not die even when it fails
4492                          * so it is safe for us to do this here.  Also
4493                          * it does not smudge active_cache or active_nr
4494                          * when it fails, so we do not have to worry about
4495                          * cleaning it up ourselves either.
4496                          */
4497                         read_cache();
4498         }
4499         if (hexsz < options->abbrev)
4500                 options->abbrev = hexsz; /* full */
4501
4502         /*
4503          * It does not make sense to show the first hit we happened
4504          * to have found.  It does not make sense not to return with
4505          * exit code in such a case either.
4506          */
4507         if (options->flags.quick) {
4508                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4509                 options->flags.exit_with_status = 1;
4510         }
4511
4512         options->diff_path_counter = 0;
4513
4514         if (options->flags.follow_renames && options->pathspec.nr != 1)
4515                 die(_("--follow requires exactly one pathspec"));
4516
4517         if (!options->use_color || external_diff())
4518                 options->color_moved = 0;
4519 }
4520
4521 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4522 {
4523         char c, *eq;
4524         int len;
4525
4526         if (*arg != '-')
4527                 return 0;
4528         c = *++arg;
4529         if (!c)
4530                 return 0;
4531         if (c == arg_short) {
4532                 c = *++arg;
4533                 if (!c)
4534                         return 1;
4535                 if (val && isdigit(c)) {
4536                         char *end;
4537                         int n = strtoul(arg, &end, 10);
4538                         if (*end)
4539                                 return 0;
4540                         *val = n;
4541                         return 1;
4542                 }
4543                 return 0;
4544         }
4545         if (c != '-')
4546                 return 0;
4547         arg++;
4548         eq = strchrnul(arg, '=');
4549         len = eq - arg;
4550         if (!len || strncmp(arg, arg_long, len))
4551                 return 0;
4552         if (*eq) {
4553                 int n;
4554                 char *end;
4555                 if (!isdigit(*++eq))
4556                         return 0;
4557                 n = strtoul(eq, &end, 10);
4558                 if (*end)
4559                         return 0;
4560                 *val = n;
4561         }
4562         return 1;
4563 }
4564
4565 static int diff_scoreopt_parse(const char *opt);
4566
4567 static inline int short_opt(char opt, const char **argv,
4568                             const char **optarg)
4569 {
4570         const char *arg = argv[0];
4571         if (arg[0] != '-' || arg[1] != opt)
4572                 return 0;
4573         if (arg[2] != '\0') {
4574                 *optarg = arg + 2;
4575                 return 1;
4576         }
4577         if (!argv[1])
4578                 die("Option '%c' requires a value", opt);
4579         *optarg = argv[1];
4580         return 2;
4581 }
4582
4583 int parse_long_opt(const char *opt, const char **argv,
4584                    const char **optarg)
4585 {
4586         const char *arg = argv[0];
4587         if (!skip_prefix(arg, "--", &arg))
4588                 return 0;
4589         if (!skip_prefix(arg, opt, &arg))
4590                 return 0;
4591         if (*arg == '=') { /* stuck form: --option=value */
4592                 *optarg = arg + 1;
4593                 return 1;
4594         }
4595         if (*arg != '\0')
4596                 return 0;
4597         /* separate form: --option value */
4598         if (!argv[1])
4599                 die("Option '--%s' requires a value", opt);
4600         *optarg = argv[1];
4601         return 2;
4602 }
4603
4604 static int stat_opt(struct diff_options *options, const char **av)
4605 {
4606         const char *arg = av[0];
4607         char *end;
4608         int width = options->stat_width;
4609         int name_width = options->stat_name_width;
4610         int graph_width = options->stat_graph_width;
4611         int count = options->stat_count;
4612         int argcount = 1;
4613
4614         if (!skip_prefix(arg, "--stat", &arg))
4615                 BUG("stat option does not begin with --stat: %s", arg);
4616         end = (char *)arg;
4617
4618         switch (*arg) {
4619         case '-':
4620                 if (skip_prefix(arg, "-width", &arg)) {
4621                         if (*arg == '=')
4622                                 width = strtoul(arg + 1, &end, 10);
4623                         else if (!*arg && !av[1])
4624                                 die_want_option("--stat-width");
4625                         else if (!*arg) {
4626                                 width = strtoul(av[1], &end, 10);
4627                                 argcount = 2;
4628                         }
4629                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4630                         if (*arg == '=')
4631                                 name_width = strtoul(arg + 1, &end, 10);
4632                         else if (!*arg && !av[1])
4633                                 die_want_option("--stat-name-width");
4634                         else if (!*arg) {
4635                                 name_width = strtoul(av[1], &end, 10);
4636                                 argcount = 2;
4637                         }
4638                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4639                         if (*arg == '=')
4640                                 graph_width = strtoul(arg + 1, &end, 10);
4641                         else if (!*arg && !av[1])
4642                                 die_want_option("--stat-graph-width");
4643                         else if (!*arg) {
4644                                 graph_width = strtoul(av[1], &end, 10);
4645                                 argcount = 2;
4646                         }
4647                 } else if (skip_prefix(arg, "-count", &arg)) {
4648                         if (*arg == '=')
4649                                 count = strtoul(arg + 1, &end, 10);
4650                         else if (!*arg && !av[1])
4651                                 die_want_option("--stat-count");
4652                         else if (!*arg) {
4653                                 count = strtoul(av[1], &end, 10);
4654                                 argcount = 2;
4655                         }
4656                 }
4657                 break;
4658         case '=':
4659                 width = strtoul(arg+1, &end, 10);
4660                 if (*end == ',')
4661                         name_width = strtoul(end+1, &end, 10);
4662                 if (*end == ',')
4663                         count = strtoul(end+1, &end, 10);
4664         }
4665
4666         /* Important! This checks all the error cases! */
4667         if (*end)
4668                 return 0;
4669         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4670         options->stat_name_width = name_width;
4671         options->stat_graph_width = graph_width;
4672         options->stat_width = width;
4673         options->stat_count = count;
4674         return argcount;
4675 }
4676
4677 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4678 {
4679         struct strbuf errmsg = STRBUF_INIT;
4680         if (parse_dirstat_params(options, params, &errmsg))
4681                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4682                     errmsg.buf);
4683         strbuf_release(&errmsg);
4684         /*
4685          * The caller knows a dirstat-related option is given from the command
4686          * line; allow it to say "return this_function();"
4687          */
4688         options->output_format |= DIFF_FORMAT_DIRSTAT;
4689         return 1;
4690 }
4691
4692 static int parse_submodule_opt(struct diff_options *options, const char *value)
4693 {
4694         if (parse_submodule_params(options, value))
4695                 die(_("Failed to parse --submodule option parameter: '%s'"),
4696                         value);
4697         return 1;
4698 }
4699
4700 static const char diff_status_letters[] = {
4701         DIFF_STATUS_ADDED,
4702         DIFF_STATUS_COPIED,
4703         DIFF_STATUS_DELETED,
4704         DIFF_STATUS_MODIFIED,
4705         DIFF_STATUS_RENAMED,
4706         DIFF_STATUS_TYPE_CHANGED,
4707         DIFF_STATUS_UNKNOWN,
4708         DIFF_STATUS_UNMERGED,
4709         DIFF_STATUS_FILTER_AON,
4710         DIFF_STATUS_FILTER_BROKEN,
4711         '\0',
4712 };
4713
4714 static unsigned int filter_bit['Z' + 1];
4715
4716 static void prepare_filter_bits(void)
4717 {
4718         int i;
4719
4720         if (!filter_bit[DIFF_STATUS_ADDED]) {
4721                 for (i = 0; diff_status_letters[i]; i++)
4722                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4723         }
4724 }
4725
4726 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4727 {
4728         return opt->filter & filter_bit[(int) status];
4729 }
4730
4731 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4732 {
4733         int i, optch;
4734
4735         prepare_filter_bits();
4736
4737         /*
4738          * If there is a negation e.g. 'd' in the input, and we haven't
4739          * initialized the filter field with another --diff-filter, start
4740          * from full set of bits, except for AON.
4741          */
4742         if (!opt->filter) {
4743                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4744                         if (optch < 'a' || 'z' < optch)
4745                                 continue;
4746                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4747                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4748                         break;
4749                 }
4750         }
4751
4752         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4753                 unsigned int bit;
4754                 int negate;
4755
4756                 if ('a' <= optch && optch <= 'z') {
4757                         negate = 1;
4758                         optch = toupper(optch);
4759                 } else {
4760                         negate = 0;
4761                 }
4762
4763                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4764                 if (!bit)
4765                         return optarg[i];
4766                 if (negate)
4767                         opt->filter &= ~bit;
4768                 else
4769                         opt->filter |= bit;
4770         }
4771         return 0;
4772 }
4773
4774 static void enable_patch_output(int *fmt) {
4775         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4776         *fmt |= DIFF_FORMAT_PATCH;
4777 }
4778
4779 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4780 {
4781         int val = parse_ws_error_highlight(arg);
4782
4783         if (val < 0) {
4784                 error("unknown value after ws-error-highlight=%.*s",
4785                       -1 - val, arg);
4786                 return 0;
4787         }
4788         opt->ws_error_highlight = val;
4789         return 1;
4790 }
4791
4792 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4793 {
4794         struct object_id oid;
4795
4796         if (get_oid(arg, &oid))
4797                 return error("unable to resolve '%s'", arg);
4798
4799         if (!opt->objfind)
4800                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4801
4802         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4803         opt->flags.recursive = 1;
4804         opt->flags.tree_in_recursive = 1;
4805         oidset_insert(opt->objfind, &oid);
4806         return 1;
4807 }
4808
4809 int diff_opt_parse(struct diff_options *options,
4810                    const char **av, int ac, const char *prefix)
4811 {
4812         const char *arg = av[0];
4813         const char *optarg;
4814         int argcount;
4815
4816         if (!prefix)
4817                 prefix = "";
4818
4819         /* Output format options */
4820         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4821             || opt_arg(arg, 'U', "unified", &options->context))
4822                 enable_patch_output(&options->output_format);
4823         else if (!strcmp(arg, "--raw"))
4824                 options->output_format |= DIFF_FORMAT_RAW;
4825         else if (!strcmp(arg, "--patch-with-raw")) {
4826                 enable_patch_output(&options->output_format);
4827                 options->output_format |= DIFF_FORMAT_RAW;
4828         } else if (!strcmp(arg, "--numstat"))
4829                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4830         else if (!strcmp(arg, "--shortstat"))
4831                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4832         else if (skip_prefix(arg, "-X", &arg) ||
4833                  skip_to_optional_arg(arg, "--dirstat", &arg))
4834                 return parse_dirstat_opt(options, arg);
4835         else if (!strcmp(arg, "--cumulative"))
4836                 return parse_dirstat_opt(options, "cumulative");
4837         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4838                 parse_dirstat_opt(options, "files");
4839                 return parse_dirstat_opt(options, arg);
4840         }
4841         else if (!strcmp(arg, "--check"))
4842                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4843         else if (!strcmp(arg, "--summary"))
4844                 options->output_format |= DIFF_FORMAT_SUMMARY;
4845         else if (!strcmp(arg, "--patch-with-stat")) {
4846                 enable_patch_output(&options->output_format);
4847                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4848         } else if (!strcmp(arg, "--name-only"))
4849                 options->output_format |= DIFF_FORMAT_NAME;
4850         else if (!strcmp(arg, "--name-status"))
4851                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4852         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4853                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4854         else if (starts_with(arg, "--stat"))
4855                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4856                 return stat_opt(options, av);
4857         else if (!strcmp(arg, "--compact-summary")) {
4858                  options->flags.stat_with_summary = 1;
4859                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4860         } else if (!strcmp(arg, "--no-compact-summary"))
4861                  options->flags.stat_with_summary = 0;
4862
4863         /* renames options */
4864         else if (starts_with(arg, "-B") ||
4865                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4866                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4867                         return error("invalid argument to -B: %s", arg+2);
4868         }
4869         else if (starts_with(arg, "-M") ||
4870                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4871                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4872                         return error("invalid argument to -M: %s", arg+2);
4873                 options->detect_rename = DIFF_DETECT_RENAME;
4874         }
4875         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4876                 options->irreversible_delete = 1;
4877         }
4878         else if (starts_with(arg, "-C") ||
4879                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4880                 if (options->detect_rename == DIFF_DETECT_COPY)
4881                         options->flags.find_copies_harder = 1;
4882                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4883                         return error("invalid argument to -C: %s", arg+2);
4884                 options->detect_rename = DIFF_DETECT_COPY;
4885         }
4886         else if (!strcmp(arg, "--no-renames"))
4887                 options->detect_rename = 0;
4888         else if (!strcmp(arg, "--rename-empty"))
4889                 options->flags.rename_empty = 1;
4890         else if (!strcmp(arg, "--no-rename-empty"))
4891                 options->flags.rename_empty = 0;
4892         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4893                 options->flags.relative_name = 1;
4894                 if (arg)
4895                         options->prefix = arg;
4896         }
4897
4898         /* xdiff options */
4899         else if (!strcmp(arg, "--minimal"))
4900                 DIFF_XDL_SET(options, NEED_MINIMAL);
4901         else if (!strcmp(arg, "--no-minimal"))
4902                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4903         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4904                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4905         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4906                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4907         else if (!strcmp(arg, "--ignore-space-at-eol"))
4908                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4909         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4910                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4911         else if (!strcmp(arg, "--ignore-blank-lines"))
4912                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4913         else if (!strcmp(arg, "--indent-heuristic"))
4914                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4915         else if (!strcmp(arg, "--no-indent-heuristic"))
4916                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4917         else if (!strcmp(arg, "--patience")) {
4918                 int i;
4919                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4920                 /*
4921                  * Both --patience and --anchored use PATIENCE_DIFF
4922                  * internally, so remove any anchors previously
4923                  * specified.
4924                  */
4925                 for (i = 0; i < options->anchors_nr; i++)
4926                         free(options->anchors[i]);
4927                 options->anchors_nr = 0;
4928         } else if (!strcmp(arg, "--histogram"))
4929                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4930         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4931                 long value = parse_algorithm_value(optarg);
4932                 if (value < 0)
4933                         return error("option diff-algorithm accepts \"myers\", "
4934                                      "\"minimal\", \"patience\" and \"histogram\"");
4935                 /* clear out previous settings */
4936                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4937                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4938                 options->xdl_opts |= value;
4939                 return argcount;
4940         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4941                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4942                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4943                            options->anchors_alloc);
4944                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4945         }
4946
4947         /* flags options */
4948         else if (!strcmp(arg, "--binary")) {
4949                 enable_patch_output(&options->output_format);
4950                 options->flags.binary = 1;
4951         }
4952         else if (!strcmp(arg, "--full-index"))
4953                 options->flags.full_index = 1;
4954         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4955                 options->flags.text = 1;
4956         else if (!strcmp(arg, "-R"))
4957                 options->flags.reverse_diff = 1;
4958         else if (!strcmp(arg, "--find-copies-harder"))
4959                 options->flags.find_copies_harder = 1;
4960         else if (!strcmp(arg, "--follow"))
4961                 options->flags.follow_renames = 1;
4962         else if (!strcmp(arg, "--no-follow")) {
4963                 options->flags.follow_renames = 0;
4964                 options->flags.default_follow_renames = 0;
4965         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4966                 int value = git_config_colorbool(NULL, arg);
4967                 if (value < 0)
4968                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4969                 options->use_color = value;
4970         }
4971         else if (!strcmp(arg, "--no-color"))
4972                 options->use_color = 0;
4973         else if (!strcmp(arg, "--color-moved")) {
4974                 if (diff_color_moved_default)
4975                         options->color_moved = diff_color_moved_default;
4976                 if (options->color_moved == COLOR_MOVED_NO)
4977                         options->color_moved = COLOR_MOVED_DEFAULT;
4978         } else if (!strcmp(arg, "--no-color-moved"))
4979                 options->color_moved = COLOR_MOVED_NO;
4980         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4981                 int cm = parse_color_moved(arg);
4982                 if (cm < 0)
4983                         die("bad --color-moved argument: %s", arg);
4984                 options->color_moved = cm;
4985         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
4986                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
4987         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4988                 options->use_color = 1;
4989                 options->word_diff = DIFF_WORDS_COLOR;
4990         }
4991         else if (!strcmp(arg, "--word-diff")) {
4992                 if (options->word_diff == DIFF_WORDS_NONE)
4993                         options->word_diff = DIFF_WORDS_PLAIN;
4994         }
4995         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4996                 if (!strcmp(arg, "plain"))
4997                         options->word_diff = DIFF_WORDS_PLAIN;
4998                 else if (!strcmp(arg, "color")) {
4999                         options->use_color = 1;
5000                         options->word_diff = DIFF_WORDS_COLOR;
5001                 }
5002                 else if (!strcmp(arg, "porcelain"))
5003                         options->word_diff = DIFF_WORDS_PORCELAIN;
5004                 else if (!strcmp(arg, "none"))
5005                         options->word_diff = DIFF_WORDS_NONE;
5006                 else
5007                         die("bad --word-diff argument: %s", arg);
5008         }
5009         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5010                 if (options->word_diff == DIFF_WORDS_NONE)
5011                         options->word_diff = DIFF_WORDS_PLAIN;
5012                 options->word_regex = optarg;
5013                 return argcount;
5014         }
5015         else if (!strcmp(arg, "--exit-code"))
5016                 options->flags.exit_with_status = 1;
5017         else if (!strcmp(arg, "--quiet"))
5018                 options->flags.quick = 1;
5019         else if (!strcmp(arg, "--ext-diff"))
5020                 options->flags.allow_external = 1;
5021         else if (!strcmp(arg, "--no-ext-diff"))
5022                 options->flags.allow_external = 0;
5023         else if (!strcmp(arg, "--textconv")) {
5024                 options->flags.allow_textconv = 1;
5025                 options->flags.textconv_set_via_cmdline = 1;
5026         } else if (!strcmp(arg, "--no-textconv"))
5027                 options->flags.allow_textconv = 0;
5028         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5029                 options->flags.override_submodule_config = 1;
5030                 handle_ignore_submodules_arg(options, arg);
5031         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5032                 return parse_submodule_opt(options, arg);
5033         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5034                 return parse_ws_error_highlight_opt(options, arg);
5035         else if (!strcmp(arg, "--ita-invisible-in-index"))
5036                 options->ita_invisible_in_index = 1;
5037         else if (!strcmp(arg, "--ita-visible-in-index"))
5038                 options->ita_invisible_in_index = 0;
5039
5040         /* misc options */
5041         else if (!strcmp(arg, "-z"))
5042                 options->line_termination = 0;
5043         else if ((argcount = short_opt('l', av, &optarg))) {
5044                 options->rename_limit = strtoul(optarg, NULL, 10);
5045                 return argcount;
5046         }
5047         else if ((argcount = short_opt('S', av, &optarg))) {
5048                 options->pickaxe = optarg;
5049                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5050                 return argcount;
5051         } else if ((argcount = short_opt('G', av, &optarg))) {
5052                 options->pickaxe = optarg;
5053                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5054                 return argcount;
5055         }
5056         else if (!strcmp(arg, "--pickaxe-all"))
5057                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5058         else if (!strcmp(arg, "--pickaxe-regex"))
5059                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5060         else if ((argcount = short_opt('O', av, &optarg))) {
5061                 options->orderfile = prefix_filename(prefix, optarg);
5062                 return argcount;
5063         } else if (skip_prefix(arg, "--find-object=", &arg))
5064                 return parse_objfind_opt(options, arg);
5065         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5066                 int offending = parse_diff_filter_opt(optarg, options);
5067                 if (offending)
5068                         die("unknown change class '%c' in --diff-filter=%s",
5069                             offending, optarg);
5070                 return argcount;
5071         }
5072         else if (!strcmp(arg, "--no-abbrev"))
5073                 options->abbrev = 0;
5074         else if (!strcmp(arg, "--abbrev"))
5075                 options->abbrev = DEFAULT_ABBREV;
5076         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5077                 options->abbrev = strtoul(arg, NULL, 10);
5078                 if (options->abbrev < MINIMUM_ABBREV)
5079                         options->abbrev = MINIMUM_ABBREV;
5080                 else if (the_hash_algo->hexsz < options->abbrev)
5081                         options->abbrev = the_hash_algo->hexsz;
5082         }
5083         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5084                 options->a_prefix = optarg;
5085                 return argcount;
5086         }
5087         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5088                 options->line_prefix = optarg;
5089                 options->line_prefix_length = strlen(options->line_prefix);
5090                 graph_setup_line_prefix(options);
5091                 return argcount;
5092         }
5093         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5094                 options->b_prefix = optarg;
5095                 return argcount;
5096         }
5097         else if (!strcmp(arg, "--no-prefix"))
5098                 options->a_prefix = options->b_prefix = "";
5099         else if (opt_arg(arg, '\0', "inter-hunk-context",
5100                          &options->interhunkcontext))
5101                 ;
5102         else if (!strcmp(arg, "-W"))
5103                 options->flags.funccontext = 1;
5104         else if (!strcmp(arg, "--function-context"))
5105                 options->flags.funccontext = 1;
5106         else if (!strcmp(arg, "--no-function-context"))
5107                 options->flags.funccontext = 0;
5108         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5109                 char *path = prefix_filename(prefix, optarg);
5110                 options->file = xfopen(path, "w");
5111                 options->close_file = 1;
5112                 if (options->use_color != GIT_COLOR_ALWAYS)
5113                         options->use_color = GIT_COLOR_NEVER;
5114                 free(path);
5115                 return argcount;
5116         } else
5117                 return 0;
5118         return 1;
5119 }
5120
5121 int parse_rename_score(const char **cp_p)
5122 {
5123         unsigned long num, scale;
5124         int ch, dot;
5125         const char *cp = *cp_p;
5126
5127         num = 0;
5128         scale = 1;
5129         dot = 0;
5130         for (;;) {
5131                 ch = *cp;
5132                 if ( !dot && ch == '.' ) {
5133                         scale = 1;
5134                         dot = 1;
5135                 } else if ( ch == '%' ) {
5136                         scale = dot ? scale*100 : 100;
5137                         cp++;   /* % is always at the end */
5138                         break;
5139                 } else if ( ch >= '0' && ch <= '9' ) {
5140                         if ( scale < 100000 ) {
5141                                 scale *= 10;
5142                                 num = (num*10) + (ch-'0');
5143                         }
5144                 } else {
5145                         break;
5146                 }
5147                 cp++;
5148         }
5149         *cp_p = cp;
5150
5151         /* user says num divided by scale and we say internally that
5152          * is MAX_SCORE * num / scale.
5153          */
5154         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5155 }
5156
5157 static int diff_scoreopt_parse(const char *opt)
5158 {
5159         int opt1, opt2, cmd;
5160
5161         if (*opt++ != '-')
5162                 return -1;
5163         cmd = *opt++;
5164         if (cmd == '-') {
5165                 /* convert the long-form arguments into short-form versions */
5166                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5167                         if (*opt == 0 || *opt++ == '=')
5168                                 cmd = 'B';
5169                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5170                         if (*opt == 0 || *opt++ == '=')
5171                                 cmd = 'C';
5172                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5173                         if (*opt == 0 || *opt++ == '=')
5174                                 cmd = 'M';
5175                 }
5176         }
5177         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5178                 return -1; /* that is not a -M, -C, or -B option */
5179
5180         opt1 = parse_rename_score(&opt);
5181         if (cmd != 'B')
5182                 opt2 = 0;
5183         else {
5184                 if (*opt == 0)
5185                         opt2 = 0;
5186                 else if (*opt != '/')
5187                         return -1; /* we expect -B80/99 or -B80 */
5188                 else {
5189                         opt++;
5190                         opt2 = parse_rename_score(&opt);
5191                 }
5192         }
5193         if (*opt != 0)
5194                 return -1;
5195         return opt1 | (opt2 << 16);
5196 }
5197
5198 struct diff_queue_struct diff_queued_diff;
5199
5200 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5201 {
5202         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5203         queue->queue[queue->nr++] = dp;
5204 }
5205
5206 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5207                                  struct diff_filespec *one,
5208                                  struct diff_filespec *two)
5209 {
5210         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5211         dp->one = one;
5212         dp->two = two;
5213         if (queue)
5214                 diff_q(queue, dp);
5215         return dp;
5216 }
5217
5218 void diff_free_filepair(struct diff_filepair *p)
5219 {
5220         free_filespec(p->one);
5221         free_filespec(p->two);
5222         free(p);
5223 }
5224
5225 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5226 {
5227         int abblen;
5228         const char *abbrev;
5229
5230         /* Do we want all 40 hex characters? */
5231         if (len == the_hash_algo->hexsz)
5232                 return oid_to_hex(oid);
5233
5234         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5235         abbrev = diff_abbrev_oid(oid, len);
5236
5237         if (!print_sha1_ellipsis())
5238                 return abbrev;
5239
5240         abblen = strlen(abbrev);
5241
5242         /*
5243          * In well-behaved cases, where the abbreviated result is the
5244          * same as the requested length, append three dots after the
5245          * abbreviation (hence the whole logic is limited to the case
5246          * where abblen < 37); when the actual abbreviated result is a
5247          * bit longer than the requested length, we reduce the number
5248          * of dots so that they match the well-behaved ones.  However,
5249          * if the actual abbreviation is longer than the requested
5250          * length by more than three, we give up on aligning, and add
5251          * three dots anyway, to indicate that the output is not the
5252          * full object name.  Yes, this may be suboptimal, but this
5253          * appears only in "diff --raw --abbrev" output and it is not
5254          * worth the effort to change it now.  Note that this would
5255          * likely to work fine when the automatic sizing of default
5256          * abbreviation length is used--we would be fed -1 in "len" in
5257          * that case, and will end up always appending three-dots, but
5258          * the automatic sizing is supposed to give abblen that ensures
5259          * uniqueness across all objects (statistically speaking).
5260          */
5261         if (abblen < the_hash_algo->hexsz - 3) {
5262                 static char hex[GIT_MAX_HEXSZ + 1];
5263                 if (len < abblen && abblen <= len + 2)
5264                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5265                 else
5266                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5267                 return hex;
5268         }
5269
5270         return oid_to_hex(oid);
5271 }
5272
5273 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5274 {
5275         int line_termination = opt->line_termination;
5276         int inter_name_termination = line_termination ? '\t' : '\0';
5277
5278         fprintf(opt->file, "%s", diff_line_prefix(opt));
5279         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5280                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5281                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5282                 fprintf(opt->file, "%s ",
5283                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5284         }
5285         if (p->score) {
5286                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5287                         inter_name_termination);
5288         } else {
5289                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5290         }
5291
5292         if (p->status == DIFF_STATUS_COPIED ||
5293             p->status == DIFF_STATUS_RENAMED) {
5294                 const char *name_a, *name_b;
5295                 name_a = p->one->path;
5296                 name_b = p->two->path;
5297                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5298                 write_name_quoted(name_a, opt->file, inter_name_termination);
5299                 write_name_quoted(name_b, opt->file, line_termination);
5300         } else {
5301                 const char *name_a, *name_b;
5302                 name_a = p->one->mode ? p->one->path : p->two->path;
5303                 name_b = NULL;
5304                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5305                 write_name_quoted(name_a, opt->file, line_termination);
5306         }
5307 }
5308
5309 int diff_unmodified_pair(struct diff_filepair *p)
5310 {
5311         /* This function is written stricter than necessary to support
5312          * the currently implemented transformers, but the idea is to
5313          * let transformers to produce diff_filepairs any way they want,
5314          * and filter and clean them up here before producing the output.
5315          */
5316         struct diff_filespec *one = p->one, *two = p->two;
5317
5318         if (DIFF_PAIR_UNMERGED(p))
5319                 return 0; /* unmerged is interesting */
5320
5321         /* deletion, addition, mode or type change
5322          * and rename are all interesting.
5323          */
5324         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5325             DIFF_PAIR_MODE_CHANGED(p) ||
5326             strcmp(one->path, two->path))
5327                 return 0;
5328
5329         /* both are valid and point at the same path.  that is, we are
5330          * dealing with a change.
5331          */
5332         if (one->oid_valid && two->oid_valid &&
5333             !oidcmp(&one->oid, &two->oid) &&
5334             !one->dirty_submodule && !two->dirty_submodule)
5335                 return 1; /* no change */
5336         if (!one->oid_valid && !two->oid_valid)
5337                 return 1; /* both look at the same file on the filesystem. */
5338         return 0;
5339 }
5340
5341 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5342 {
5343         if (diff_unmodified_pair(p))
5344                 return;
5345
5346         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5347             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5348                 return; /* no tree diffs in patch format */
5349
5350         run_diff(p, o);
5351 }
5352
5353 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5354                             struct diffstat_t *diffstat)
5355 {
5356         if (diff_unmodified_pair(p))
5357                 return;
5358
5359         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5360             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5361                 return; /* no useful stat for tree diffs */
5362
5363         run_diffstat(p, o, diffstat);
5364 }
5365
5366 static void diff_flush_checkdiff(struct diff_filepair *p,
5367                 struct diff_options *o)
5368 {
5369         if (diff_unmodified_pair(p))
5370                 return;
5371
5372         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5373             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5374                 return; /* nothing to check in tree diffs */
5375
5376         run_checkdiff(p, o);
5377 }
5378
5379 int diff_queue_is_empty(void)
5380 {
5381         struct diff_queue_struct *q = &diff_queued_diff;
5382         int i;
5383         for (i = 0; i < q->nr; i++)
5384                 if (!diff_unmodified_pair(q->queue[i]))
5385                         return 0;
5386         return 1;
5387 }
5388
5389 #if DIFF_DEBUG
5390 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5391 {
5392         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5393                 x, one ? one : "",
5394                 s->path,
5395                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5396                 s->mode,
5397                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5398         fprintf(stderr, "queue[%d] %s size %lu\n",
5399                 x, one ? one : "",
5400                 s->size);
5401 }
5402
5403 void diff_debug_filepair(const struct diff_filepair *p, int i)
5404 {
5405         diff_debug_filespec(p->one, i, "one");
5406         diff_debug_filespec(p->two, i, "two");
5407         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5408                 p->score, p->status ? p->status : '?',
5409                 p->one->rename_used, p->broken_pair);
5410 }
5411
5412 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5413 {
5414         int i;
5415         if (msg)
5416                 fprintf(stderr, "%s\n", msg);
5417         fprintf(stderr, "q->nr = %d\n", q->nr);
5418         for (i = 0; i < q->nr; i++) {
5419                 struct diff_filepair *p = q->queue[i];
5420                 diff_debug_filepair(p, i);
5421         }
5422 }
5423 #endif
5424
5425 static void diff_resolve_rename_copy(void)
5426 {
5427         int i;
5428         struct diff_filepair *p;
5429         struct diff_queue_struct *q = &diff_queued_diff;
5430
5431         diff_debug_queue("resolve-rename-copy", q);
5432
5433         for (i = 0; i < q->nr; i++) {
5434                 p = q->queue[i];
5435                 p->status = 0; /* undecided */
5436                 if (DIFF_PAIR_UNMERGED(p))
5437                         p->status = DIFF_STATUS_UNMERGED;
5438                 else if (!DIFF_FILE_VALID(p->one))
5439                         p->status = DIFF_STATUS_ADDED;
5440                 else if (!DIFF_FILE_VALID(p->two))
5441                         p->status = DIFF_STATUS_DELETED;
5442                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5443                         p->status = DIFF_STATUS_TYPE_CHANGED;
5444
5445                 /* from this point on, we are dealing with a pair
5446                  * whose both sides are valid and of the same type, i.e.
5447                  * either in-place edit or rename/copy edit.
5448                  */
5449                 else if (DIFF_PAIR_RENAME(p)) {
5450                         /*
5451                          * A rename might have re-connected a broken
5452                          * pair up, causing the pathnames to be the
5453                          * same again. If so, that's not a rename at
5454                          * all, just a modification..
5455                          *
5456                          * Otherwise, see if this source was used for
5457                          * multiple renames, in which case we decrement
5458                          * the count, and call it a copy.
5459                          */
5460                         if (!strcmp(p->one->path, p->two->path))
5461                                 p->status = DIFF_STATUS_MODIFIED;
5462                         else if (--p->one->rename_used > 0)
5463                                 p->status = DIFF_STATUS_COPIED;
5464                         else
5465                                 p->status = DIFF_STATUS_RENAMED;
5466                 }
5467                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5468                          p->one->mode != p->two->mode ||
5469                          p->one->dirty_submodule ||
5470                          p->two->dirty_submodule ||
5471                          is_null_oid(&p->one->oid))
5472                         p->status = DIFF_STATUS_MODIFIED;
5473                 else {
5474                         /* This is a "no-change" entry and should not
5475                          * happen anymore, but prepare for broken callers.
5476                          */
5477                         error("feeding unmodified %s to diffcore",
5478                               p->one->path);
5479                         p->status = DIFF_STATUS_UNKNOWN;
5480                 }
5481         }
5482         diff_debug_queue("resolve-rename-copy done", q);
5483 }
5484
5485 static int check_pair_status(struct diff_filepair *p)
5486 {
5487         switch (p->status) {
5488         case DIFF_STATUS_UNKNOWN:
5489                 return 0;
5490         case 0:
5491                 die("internal error in diff-resolve-rename-copy");
5492         default:
5493                 return 1;
5494         }
5495 }
5496
5497 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5498 {
5499         int fmt = opt->output_format;
5500
5501         if (fmt & DIFF_FORMAT_CHECKDIFF)
5502                 diff_flush_checkdiff(p, opt);
5503         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5504                 diff_flush_raw(p, opt);
5505         else if (fmt & DIFF_FORMAT_NAME) {
5506                 const char *name_a, *name_b;
5507                 name_a = p->two->path;
5508                 name_b = NULL;
5509                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5510                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5511                 write_name_quoted(name_a, opt->file, opt->line_termination);
5512         }
5513 }
5514
5515 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5516 {
5517         struct strbuf sb = STRBUF_INIT;
5518         if (fs->mode)
5519                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5520         else
5521                 strbuf_addf(&sb, " %s ", newdelete);
5522
5523         quote_c_style(fs->path, &sb, NULL, 0);
5524         strbuf_addch(&sb, '\n');
5525         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5526                          sb.buf, sb.len, 0);
5527         strbuf_release(&sb);
5528 }
5529
5530 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5531                 int show_name)
5532 {
5533         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5534                 struct strbuf sb = STRBUF_INIT;
5535                 strbuf_addf(&sb, " mode change %06o => %06o",
5536                             p->one->mode, p->two->mode);
5537                 if (show_name) {
5538                         strbuf_addch(&sb, ' ');
5539                         quote_c_style(p->two->path, &sb, NULL, 0);
5540                 }
5541                 strbuf_addch(&sb, '\n');
5542                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5543                                  sb.buf, sb.len, 0);
5544                 strbuf_release(&sb);
5545         }
5546 }
5547
5548 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5549                 struct diff_filepair *p)
5550 {
5551         struct strbuf sb = STRBUF_INIT;
5552         struct strbuf names = STRBUF_INIT;
5553
5554         pprint_rename(&names, p->one->path, p->two->path);
5555         strbuf_addf(&sb, " %s %s (%d%%)\n",
5556                     renamecopy, names.buf, similarity_index(p));
5557         strbuf_release(&names);
5558         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5559                                  sb.buf, sb.len, 0);
5560         show_mode_change(opt, p, 0);
5561         strbuf_release(&sb);
5562 }
5563
5564 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5565 {
5566         switch(p->status) {
5567         case DIFF_STATUS_DELETED:
5568                 show_file_mode_name(opt, "delete", p->one);
5569                 break;
5570         case DIFF_STATUS_ADDED:
5571                 show_file_mode_name(opt, "create", p->two);
5572                 break;
5573         case DIFF_STATUS_COPIED:
5574                 show_rename_copy(opt, "copy", p);
5575                 break;
5576         case DIFF_STATUS_RENAMED:
5577                 show_rename_copy(opt, "rename", p);
5578                 break;
5579         default:
5580                 if (p->score) {
5581                         struct strbuf sb = STRBUF_INIT;
5582                         strbuf_addstr(&sb, " rewrite ");
5583                         quote_c_style(p->two->path, &sb, NULL, 0);
5584                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5585                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5586                                          sb.buf, sb.len, 0);
5587                         strbuf_release(&sb);
5588                 }
5589                 show_mode_change(opt, p, !p->score);
5590                 break;
5591         }
5592 }
5593
5594 struct patch_id_t {
5595         git_SHA_CTX *ctx;
5596         int patchlen;
5597 };
5598
5599 static int remove_space(char *line, int len)
5600 {
5601         int i;
5602         char *dst = line;
5603         unsigned char c;
5604
5605         for (i = 0; i < len; i++)
5606                 if (!isspace((c = line[i])))
5607                         *dst++ = c;
5608
5609         return dst - line;
5610 }
5611
5612 static void patch_id_consume(void *priv, char *line, unsigned long len)
5613 {
5614         struct patch_id_t *data = priv;
5615         int new_len;
5616
5617         /* Ignore line numbers when computing the SHA1 of the patch */
5618         if (starts_with(line, "@@ -"))
5619                 return;
5620
5621         new_len = remove_space(line, len);
5622
5623         git_SHA1_Update(data->ctx, line, new_len);
5624         data->patchlen += new_len;
5625 }
5626
5627 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5628 {
5629         git_SHA1_Update(ctx, str, strlen(str));
5630 }
5631
5632 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5633 {
5634         /* large enough for 2^32 in octal */
5635         char buf[12];
5636         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5637         git_SHA1_Update(ctx, buf, len);
5638 }
5639
5640 /* returns 0 upon success, and writes result into sha1 */
5641 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5642 {
5643         struct diff_queue_struct *q = &diff_queued_diff;
5644         int i;
5645         git_SHA_CTX ctx;
5646         struct patch_id_t data;
5647
5648         git_SHA1_Init(&ctx);
5649         memset(&data, 0, sizeof(struct patch_id_t));
5650         data.ctx = &ctx;
5651
5652         for (i = 0; i < q->nr; i++) {
5653                 xpparam_t xpp;
5654                 xdemitconf_t xecfg;
5655                 mmfile_t mf1, mf2;
5656                 struct diff_filepair *p = q->queue[i];
5657                 int len1, len2;
5658
5659                 memset(&xpp, 0, sizeof(xpp));
5660                 memset(&xecfg, 0, sizeof(xecfg));
5661                 if (p->status == 0)
5662                         return error("internal diff status error");
5663                 if (p->status == DIFF_STATUS_UNKNOWN)
5664                         continue;
5665                 if (diff_unmodified_pair(p))
5666                         continue;
5667                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5668                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5669                         continue;
5670                 if (DIFF_PAIR_UNMERGED(p))
5671                         continue;
5672
5673                 diff_fill_oid_info(p->one);
5674                 diff_fill_oid_info(p->two);
5675
5676                 len1 = remove_space(p->one->path, strlen(p->one->path));
5677                 len2 = remove_space(p->two->path, strlen(p->two->path));
5678                 patch_id_add_string(&ctx, "diff--git");
5679                 patch_id_add_string(&ctx, "a/");
5680                 git_SHA1_Update(&ctx, p->one->path, len1);
5681                 patch_id_add_string(&ctx, "b/");
5682                 git_SHA1_Update(&ctx, p->two->path, len2);
5683
5684                 if (p->one->mode == 0) {
5685                         patch_id_add_string(&ctx, "newfilemode");
5686                         patch_id_add_mode(&ctx, p->two->mode);
5687                         patch_id_add_string(&ctx, "---/dev/null");
5688                         patch_id_add_string(&ctx, "+++b/");
5689                         git_SHA1_Update(&ctx, p->two->path, len2);
5690                 } else if (p->two->mode == 0) {
5691                         patch_id_add_string(&ctx, "deletedfilemode");
5692                         patch_id_add_mode(&ctx, p->one->mode);
5693                         patch_id_add_string(&ctx, "---a/");
5694                         git_SHA1_Update(&ctx, p->one->path, len1);
5695                         patch_id_add_string(&ctx, "+++/dev/null");
5696                 } else {
5697                         patch_id_add_string(&ctx, "---a/");
5698                         git_SHA1_Update(&ctx, p->one->path, len1);
5699                         patch_id_add_string(&ctx, "+++b/");
5700                         git_SHA1_Update(&ctx, p->two->path, len2);
5701                 }
5702
5703                 if (diff_header_only)
5704                         continue;
5705
5706                 if (fill_mmfile(&mf1, p->one) < 0 ||
5707                     fill_mmfile(&mf2, p->two) < 0)
5708                         return error("unable to read files to diff");
5709
5710                 if (diff_filespec_is_binary(p->one) ||
5711                     diff_filespec_is_binary(p->two)) {
5712                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5713                                         GIT_SHA1_HEXSZ);
5714                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5715                                         GIT_SHA1_HEXSZ);
5716                         continue;
5717                 }
5718
5719                 xpp.flags = 0;
5720                 xecfg.ctxlen = 3;
5721                 xecfg.flags = 0;
5722                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5723                                   &xpp, &xecfg))
5724                         return error("unable to generate patch-id diff for %s",
5725                                      p->one->path);
5726         }
5727
5728         git_SHA1_Final(oid->hash, &ctx);
5729         return 0;
5730 }
5731
5732 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5733 {
5734         struct diff_queue_struct *q = &diff_queued_diff;
5735         int i;
5736         int result = diff_get_patch_id(options, oid, diff_header_only);
5737
5738         for (i = 0; i < q->nr; i++)
5739                 diff_free_filepair(q->queue[i]);
5740
5741         free(q->queue);
5742         DIFF_QUEUE_CLEAR(q);
5743
5744         return result;
5745 }
5746
5747 static int is_summary_empty(const struct diff_queue_struct *q)
5748 {
5749         int i;
5750
5751         for (i = 0; i < q->nr; i++) {
5752                 const struct diff_filepair *p = q->queue[i];
5753
5754                 switch (p->status) {
5755                 case DIFF_STATUS_DELETED:
5756                 case DIFF_STATUS_ADDED:
5757                 case DIFF_STATUS_COPIED:
5758                 case DIFF_STATUS_RENAMED:
5759                         return 0;
5760                 default:
5761                         if (p->score)
5762                                 return 0;
5763                         if (p->one->mode && p->two->mode &&
5764                             p->one->mode != p->two->mode)
5765                                 return 0;
5766                         break;
5767                 }
5768         }
5769         return 1;
5770 }
5771
5772 static const char rename_limit_warning[] =
5773 N_("inexact rename detection was skipped due to too many files.");
5774
5775 static const char degrade_cc_to_c_warning[] =
5776 N_("only found copies from modified paths due to too many files.");
5777
5778 static const char rename_limit_advice[] =
5779 N_("you may want to set your %s variable to at least "
5780    "%d and retry the command.");
5781
5782 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5783 {
5784         fflush(stdout);
5785         if (degraded_cc)
5786                 warning(_(degrade_cc_to_c_warning));
5787         else if (needed)
5788                 warning(_(rename_limit_warning));
5789         else
5790                 return;
5791         if (0 < needed)
5792                 warning(_(rename_limit_advice), varname, needed);
5793 }
5794
5795 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5796 {
5797         int i;
5798         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5799         struct diff_queue_struct *q = &diff_queued_diff;
5800
5801         if (WSEH_NEW & WS_RULE_MASK)
5802                 BUG("WS rules bit mask overlaps with diff symbol flags");
5803
5804         if (o->color_moved)
5805                 o->emitted_symbols = &esm;
5806
5807         for (i = 0; i < q->nr; i++) {
5808                 struct diff_filepair *p = q->queue[i];
5809                 if (check_pair_status(p))
5810                         diff_flush_patch(p, o);
5811         }
5812
5813         if (o->emitted_symbols) {
5814                 if (o->color_moved) {
5815                         struct hashmap add_lines, del_lines;
5816
5817                         if (o->color_moved_ws_handling &
5818                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5819                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5820
5821                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5822                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5823
5824                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5825                         mark_color_as_moved(o, &add_lines, &del_lines);
5826                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5827                                 dim_moved_lines(o);
5828
5829                         hashmap_free(&add_lines, 0);
5830                         hashmap_free(&del_lines, 0);
5831                 }
5832
5833                 for (i = 0; i < esm.nr; i++)
5834                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5835
5836                 for (i = 0; i < esm.nr; i++)
5837                         free((void *)esm.buf[i].line);
5838         }
5839         esm.nr = 0;
5840 }
5841
5842 void diff_flush(struct diff_options *options)
5843 {
5844         struct diff_queue_struct *q = &diff_queued_diff;
5845         int i, output_format = options->output_format;
5846         int separator = 0;
5847         int dirstat_by_line = 0;
5848
5849         /*
5850          * Order: raw, stat, summary, patch
5851          * or:    name/name-status/checkdiff (other bits clear)
5852          */
5853         if (!q->nr)
5854                 goto free_queue;
5855
5856         if (output_format & (DIFF_FORMAT_RAW |
5857                              DIFF_FORMAT_NAME |
5858                              DIFF_FORMAT_NAME_STATUS |
5859                              DIFF_FORMAT_CHECKDIFF)) {
5860                 for (i = 0; i < q->nr; i++) {
5861                         struct diff_filepair *p = q->queue[i];
5862                         if (check_pair_status(p))
5863                                 flush_one_pair(p, options);
5864                 }
5865                 separator++;
5866         }
5867
5868         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5869                 dirstat_by_line = 1;
5870
5871         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5872             dirstat_by_line) {
5873                 struct diffstat_t diffstat;
5874
5875                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5876                 for (i = 0; i < q->nr; i++) {
5877                         struct diff_filepair *p = q->queue[i];
5878                         if (check_pair_status(p))
5879                                 diff_flush_stat(p, options, &diffstat);
5880                 }
5881                 if (output_format & DIFF_FORMAT_NUMSTAT)
5882                         show_numstat(&diffstat, options);
5883                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5884                         show_stats(&diffstat, options);
5885                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5886                         show_shortstats(&diffstat, options);
5887                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5888                         show_dirstat_by_line(&diffstat, options);
5889                 free_diffstat_info(&diffstat);
5890                 separator++;
5891         }
5892         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5893                 show_dirstat(options);
5894
5895         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5896                 for (i = 0; i < q->nr; i++) {
5897                         diff_summary(options, q->queue[i]);
5898                 }
5899                 separator++;
5900         }
5901
5902         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5903             options->flags.exit_with_status &&
5904             options->flags.diff_from_contents) {
5905                 /*
5906                  * run diff_flush_patch for the exit status. setting
5907                  * options->file to /dev/null should be safe, because we
5908                  * aren't supposed to produce any output anyway.
5909                  */
5910                 if (options->close_file)
5911                         fclose(options->file);
5912                 options->file = xfopen("/dev/null", "w");
5913                 options->close_file = 1;
5914                 options->color_moved = 0;
5915                 for (i = 0; i < q->nr; i++) {
5916                         struct diff_filepair *p = q->queue[i];
5917                         if (check_pair_status(p))
5918                                 diff_flush_patch(p, options);
5919                         if (options->found_changes)
5920                                 break;
5921                 }
5922         }
5923
5924         if (output_format & DIFF_FORMAT_PATCH) {
5925                 if (separator) {
5926                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5927                         if (options->stat_sep)
5928                                 /* attach patch instead of inline */
5929                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5930                                                  NULL, 0, 0);
5931                 }
5932
5933                 diff_flush_patch_all_file_pairs(options);
5934         }
5935
5936         if (output_format & DIFF_FORMAT_CALLBACK)
5937                 options->format_callback(q, options, options->format_callback_data);
5938
5939         for (i = 0; i < q->nr; i++)
5940                 diff_free_filepair(q->queue[i]);
5941 free_queue:
5942         free(q->queue);
5943         DIFF_QUEUE_CLEAR(q);
5944         if (options->close_file)
5945                 fclose(options->file);
5946
5947         /*
5948          * Report the content-level differences with HAS_CHANGES;
5949          * diff_addremove/diff_change does not set the bit when
5950          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5951          */
5952         if (options->flags.diff_from_contents) {
5953                 if (options->found_changes)
5954                         options->flags.has_changes = 1;
5955                 else
5956                         options->flags.has_changes = 0;
5957         }
5958 }
5959
5960 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5961 {
5962         return (((p->status == DIFF_STATUS_MODIFIED) &&
5963                  ((p->score &&
5964                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5965                   (!p->score &&
5966                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5967                 ((p->status != DIFF_STATUS_MODIFIED) &&
5968                  filter_bit_tst(p->status, options)));
5969 }
5970
5971 static void diffcore_apply_filter(struct diff_options *options)
5972 {
5973         int i;
5974         struct diff_queue_struct *q = &diff_queued_diff;
5975         struct diff_queue_struct outq;
5976
5977         DIFF_QUEUE_CLEAR(&outq);
5978
5979         if (!options->filter)
5980                 return;
5981
5982         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5983                 int found;
5984                 for (i = found = 0; !found && i < q->nr; i++) {
5985                         if (match_filter(options, q->queue[i]))
5986                                 found++;
5987                 }
5988                 if (found)
5989                         return;
5990
5991                 /* otherwise we will clear the whole queue
5992                  * by copying the empty outq at the end of this
5993                  * function, but first clear the current entries
5994                  * in the queue.
5995                  */
5996                 for (i = 0; i < q->nr; i++)
5997                         diff_free_filepair(q->queue[i]);
5998         }
5999         else {
6000                 /* Only the matching ones */
6001                 for (i = 0; i < q->nr; i++) {
6002                         struct diff_filepair *p = q->queue[i];
6003                         if (match_filter(options, p))
6004                                 diff_q(&outq, p);
6005                         else
6006                                 diff_free_filepair(p);
6007                 }
6008         }
6009         free(q->queue);
6010         *q = outq;
6011 }
6012
6013 /* Check whether two filespecs with the same mode and size are identical */
6014 static int diff_filespec_is_identical(struct diff_filespec *one,
6015                                       struct diff_filespec *two)
6016 {
6017         if (S_ISGITLINK(one->mode))
6018                 return 0;
6019         if (diff_populate_filespec(one, 0))
6020                 return 0;
6021         if (diff_populate_filespec(two, 0))
6022                 return 0;
6023         return !memcmp(one->data, two->data, one->size);
6024 }
6025
6026 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
6027 {
6028         if (p->done_skip_stat_unmatch)
6029                 return p->skip_stat_unmatch_result;
6030
6031         p->done_skip_stat_unmatch = 1;
6032         p->skip_stat_unmatch_result = 0;
6033         /*
6034          * 1. Entries that come from stat info dirtiness
6035          *    always have both sides (iow, not create/delete),
6036          *    one side of the object name is unknown, with
6037          *    the same mode and size.  Keep the ones that
6038          *    do not match these criteria.  They have real
6039          *    differences.
6040          *
6041          * 2. At this point, the file is known to be modified,
6042          *    with the same mode and size, and the object
6043          *    name of one side is unknown.  Need to inspect
6044          *    the identical contents.
6045          */
6046         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6047             !DIFF_FILE_VALID(p->two) ||
6048             (p->one->oid_valid && p->two->oid_valid) ||
6049             (p->one->mode != p->two->mode) ||
6050             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
6051             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
6052             (p->one->size != p->two->size) ||
6053             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
6054                 p->skip_stat_unmatch_result = 1;
6055         return p->skip_stat_unmatch_result;
6056 }
6057
6058 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6059 {
6060         int i;
6061         struct diff_queue_struct *q = &diff_queued_diff;
6062         struct diff_queue_struct outq;
6063         DIFF_QUEUE_CLEAR(&outq);
6064
6065         for (i = 0; i < q->nr; i++) {
6066                 struct diff_filepair *p = q->queue[i];
6067
6068                 if (diff_filespec_check_stat_unmatch(p))
6069                         diff_q(&outq, p);
6070                 else {
6071                         /*
6072                          * The caller can subtract 1 from skip_stat_unmatch
6073                          * to determine how many paths were dirty only
6074                          * due to stat info mismatch.
6075                          */
6076                         if (!diffopt->flags.no_index)
6077                                 diffopt->skip_stat_unmatch++;
6078                         diff_free_filepair(p);
6079                 }
6080         }
6081         free(q->queue);
6082         *q = outq;
6083 }
6084
6085 static int diffnamecmp(const void *a_, const void *b_)
6086 {
6087         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6088         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6089         const char *name_a, *name_b;
6090
6091         name_a = a->one ? a->one->path : a->two->path;
6092         name_b = b->one ? b->one->path : b->two->path;
6093         return strcmp(name_a, name_b);
6094 }
6095
6096 void diffcore_fix_diff_index(struct diff_options *options)
6097 {
6098         struct diff_queue_struct *q = &diff_queued_diff;
6099         QSORT(q->queue, q->nr, diffnamecmp);
6100 }
6101
6102 void diffcore_std(struct diff_options *options)
6103 {
6104         /* NOTE please keep the following in sync with diff_tree_combined() */
6105         if (options->skip_stat_unmatch)
6106                 diffcore_skip_stat_unmatch(options);
6107         if (!options->found_follow) {
6108                 /* See try_to_follow_renames() in tree-diff.c */
6109                 if (options->break_opt != -1)
6110                         diffcore_break(options->break_opt);
6111                 if (options->detect_rename)
6112                         diffcore_rename(options);
6113                 if (options->break_opt != -1)
6114                         diffcore_merge_broken();
6115         }
6116         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6117                 diffcore_pickaxe(options);
6118         if (options->orderfile)
6119                 diffcore_order(options->orderfile);
6120         if (!options->found_follow)
6121                 /* See try_to_follow_renames() in tree-diff.c */
6122                 diff_resolve_rename_copy();
6123         diffcore_apply_filter(options);
6124
6125         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6126                 options->flags.has_changes = 1;
6127         else
6128                 options->flags.has_changes = 0;
6129
6130         options->found_follow = 0;
6131 }
6132
6133 int diff_result_code(struct diff_options *opt, int status)
6134 {
6135         int result = 0;
6136
6137         diff_warn_rename_limit("diff.renameLimit",
6138                                opt->needed_rename_limit,
6139                                opt->degraded_cc_to_c);
6140         if (!opt->flags.exit_with_status &&
6141             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6142                 return status;
6143         if (opt->flags.exit_with_status &&
6144             opt->flags.has_changes)
6145                 result |= 01;
6146         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6147             opt->flags.check_failed)
6148                 result |= 02;
6149         return result;
6150 }
6151
6152 int diff_can_quit_early(struct diff_options *opt)
6153 {
6154         return (opt->flags.quick &&
6155                 !opt->filter &&
6156                 opt->flags.has_changes);
6157 }
6158
6159 /*
6160  * Shall changes to this submodule be ignored?
6161  *
6162  * Submodule changes can be configured to be ignored separately for each path,
6163  * but that configuration can be overridden from the command line.
6164  */
6165 static int is_submodule_ignored(const char *path, struct diff_options *options)
6166 {
6167         int ignored = 0;
6168         struct diff_flags orig_flags = options->flags;
6169         if (!options->flags.override_submodule_config)
6170                 set_diffopt_flags_from_submodule_config(options, path);
6171         if (options->flags.ignore_submodules)
6172                 ignored = 1;
6173         options->flags = orig_flags;
6174         return ignored;
6175 }
6176
6177 void diff_addremove(struct diff_options *options,
6178                     int addremove, unsigned mode,
6179                     const struct object_id *oid,
6180                     int oid_valid,
6181                     const char *concatpath, unsigned dirty_submodule)
6182 {
6183         struct diff_filespec *one, *two;
6184
6185         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6186                 return;
6187
6188         /* This may look odd, but it is a preparation for
6189          * feeding "there are unchanged files which should
6190          * not produce diffs, but when you are doing copy
6191          * detection you would need them, so here they are"
6192          * entries to the diff-core.  They will be prefixed
6193          * with something like '=' or '*' (I haven't decided
6194          * which but should not make any difference).
6195          * Feeding the same new and old to diff_change()
6196          * also has the same effect.
6197          * Before the final output happens, they are pruned after
6198          * merged into rename/copy pairs as appropriate.
6199          */
6200         if (options->flags.reverse_diff)
6201                 addremove = (addremove == '+' ? '-' :
6202                              addremove == '-' ? '+' : addremove);
6203
6204         if (options->prefix &&
6205             strncmp(concatpath, options->prefix, options->prefix_length))
6206                 return;
6207
6208         one = alloc_filespec(concatpath);
6209         two = alloc_filespec(concatpath);
6210
6211         if (addremove != '+')
6212                 fill_filespec(one, oid, oid_valid, mode);
6213         if (addremove != '-') {
6214                 fill_filespec(two, oid, oid_valid, mode);
6215                 two->dirty_submodule = dirty_submodule;
6216         }
6217
6218         diff_queue(&diff_queued_diff, one, two);
6219         if (!options->flags.diff_from_contents)
6220                 options->flags.has_changes = 1;
6221 }
6222
6223 void diff_change(struct diff_options *options,
6224                  unsigned old_mode, unsigned new_mode,
6225                  const struct object_id *old_oid,
6226                  const struct object_id *new_oid,
6227                  int old_oid_valid, int new_oid_valid,
6228                  const char *concatpath,
6229                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6230 {
6231         struct diff_filespec *one, *two;
6232         struct diff_filepair *p;
6233
6234         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6235             is_submodule_ignored(concatpath, options))
6236                 return;
6237
6238         if (options->flags.reverse_diff) {
6239                 SWAP(old_mode, new_mode);
6240                 SWAP(old_oid, new_oid);
6241                 SWAP(old_oid_valid, new_oid_valid);
6242                 SWAP(old_dirty_submodule, new_dirty_submodule);
6243         }
6244
6245         if (options->prefix &&
6246             strncmp(concatpath, options->prefix, options->prefix_length))
6247                 return;
6248
6249         one = alloc_filespec(concatpath);
6250         two = alloc_filespec(concatpath);
6251         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6252         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6253         one->dirty_submodule = old_dirty_submodule;
6254         two->dirty_submodule = new_dirty_submodule;
6255         p = diff_queue(&diff_queued_diff, one, two);
6256
6257         if (options->flags.diff_from_contents)
6258                 return;
6259
6260         if (options->flags.quick && options->skip_stat_unmatch &&
6261             !diff_filespec_check_stat_unmatch(p))
6262                 return;
6263
6264         options->flags.has_changes = 1;
6265 }
6266
6267 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6268 {
6269         struct diff_filepair *pair;
6270         struct diff_filespec *one, *two;
6271
6272         if (options->prefix &&
6273             strncmp(path, options->prefix, options->prefix_length))
6274                 return NULL;
6275
6276         one = alloc_filespec(path);
6277         two = alloc_filespec(path);
6278         pair = diff_queue(&diff_queued_diff, one, two);
6279         pair->is_unmerged = 1;
6280         return pair;
6281 }
6282
6283 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6284                 size_t *outsize)
6285 {
6286         struct diff_tempfile *temp;
6287         const char *argv[3];
6288         const char **arg = argv;
6289         struct child_process child = CHILD_PROCESS_INIT;
6290         struct strbuf buf = STRBUF_INIT;
6291         int err = 0;
6292
6293         temp = prepare_temp_file(spec->path, spec);
6294         *arg++ = pgm;
6295         *arg++ = temp->name;
6296         *arg = NULL;
6297
6298         child.use_shell = 1;
6299         child.argv = argv;
6300         child.out = -1;
6301         if (start_command(&child)) {
6302                 remove_tempfile();
6303                 return NULL;
6304         }
6305
6306         if (strbuf_read(&buf, child.out, 0) < 0)
6307                 err = error("error reading from textconv command '%s'", pgm);
6308         close(child.out);
6309
6310         if (finish_command(&child) || err) {
6311                 strbuf_release(&buf);
6312                 remove_tempfile();
6313                 return NULL;
6314         }
6315         remove_tempfile();
6316
6317         return strbuf_detach(&buf, outsize);
6318 }
6319
6320 size_t fill_textconv(struct userdiff_driver *driver,
6321                      struct diff_filespec *df,
6322                      char **outbuf)
6323 {
6324         size_t size;
6325
6326         if (!driver) {
6327                 if (!DIFF_FILE_VALID(df)) {
6328                         *outbuf = "";
6329                         return 0;
6330                 }
6331                 if (diff_populate_filespec(df, 0))
6332                         die("unable to read files to diff");
6333                 *outbuf = df->data;
6334                 return df->size;
6335         }
6336
6337         if (!driver->textconv)
6338                 BUG("fill_textconv called with non-textconv driver");
6339
6340         if (driver->textconv_cache && df->oid_valid) {
6341                 *outbuf = notes_cache_get(driver->textconv_cache,
6342                                           &df->oid,
6343                                           &size);
6344                 if (*outbuf)
6345                         return size;
6346         }
6347
6348         *outbuf = run_textconv(driver->textconv, df, &size);
6349         if (!*outbuf)
6350                 die("unable to read files to diff");
6351
6352         if (driver->textconv_cache && df->oid_valid) {
6353                 /* ignore errors, as we might be in a readonly repository */
6354                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6355                                 size);
6356                 /*
6357                  * we could save up changes and flush them all at the end,
6358                  * but we would need an extra call after all diffing is done.
6359                  * Since generating a cache entry is the slow path anyway,
6360                  * this extra overhead probably isn't a big deal.
6361                  */
6362                 notes_cache_write(driver->textconv_cache);
6363         }
6364
6365         return size;
6366 }
6367
6368 int textconv_object(const char *path,
6369                     unsigned mode,
6370                     const struct object_id *oid,
6371                     int oid_valid,
6372                     char **buf,
6373                     unsigned long *buf_size)
6374 {
6375         struct diff_filespec *df;
6376         struct userdiff_driver *textconv;
6377
6378         df = alloc_filespec(path);
6379         fill_filespec(df, oid, oid_valid, mode);
6380         textconv = get_textconv(df);
6381         if (!textconv) {
6382                 free_filespec(df);
6383                 return 0;
6384         }
6385
6386         *buf_size = fill_textconv(textconv, df, buf);
6387         free_filespec(df);
6388         return 1;
6389 }
6390
6391 void setup_diff_pager(struct diff_options *opt)
6392 {
6393         /*
6394          * If the user asked for our exit code, then either they want --quiet
6395          * or --exit-code. We should definitely not bother with a pager in the
6396          * former case, as we will generate no output. Since we still properly
6397          * report our exit code even when a pager is run, we _could_ run a
6398          * pager with --exit-code. But since we have not done so historically,
6399          * and because it is easy to find people oneline advising "git diff
6400          * --exit-code" in hooks and other scripts, we do not do so.
6401          */
6402         if (!opt->flags.exit_with_status &&
6403             check_pager_config("diff") != 0)
6404                 setup_pager();
6405 }