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