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