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