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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed-zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else if (!strcmp(arg, "dimmed_zebra"))
 289                 return COLOR_MOVED_ZEBRA_DIM;
 290         else
 291                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 292 }
 293
 294 static int parse_color_moved_ws(const char *arg)
 295 {
 296         int ret = 0;
 297         struct string_list l = STRING_LIST_INIT_DUP;
 298         struct string_list_item *i;
 299
 300         string_list_split(&l, arg, ',', -1);
 301
 302         for_each_string_list_item(i, &l) {
 303                 struct strbuf sb = STRBUF_INIT;
 304                 strbuf_addstr(&sb, i->string);
 305                 strbuf_trim(&sb);
 306
 307                 if (!strcmp(sb.buf, "ignore-space-change"))
 308                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 309                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 310                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 311                 else if (!strcmp(sb.buf, "ignore-all-space"))
 312                         ret |= XDF_IGNORE_WHITESPACE;
 313                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 314                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 315                 else
 316                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 317
 318                 strbuf_release(&sb);
 319         }
 320
 321         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 322             (ret & XDF_WHITESPACE_FLAGS))
 323                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 324
 325         string_list_clear(&l, 0);
 326
 327         return ret;
 328 }
 329
 330 int git_diff_ui_config(const char *var, const char *value, void *cb)
 331 {
 332         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 333                 diff_use_color_default = git_config_colorbool(var, value);
 334                 return 0;
 335         }
 336         if (!strcmp(var, "diff.colormoved")) {
 337                 int cm = parse_color_moved(value);
 338                 if (cm < 0)
 339                         return -1;
 340                 diff_color_moved_default = cm;
 341                 return 0;
 342         }
 343         if (!strcmp(var, "diff.colormovedws")) {
 344                 int cm = parse_color_moved_ws(value);
 345                 if (cm < 0)
 346                         return -1;
 347                 diff_color_moved_ws_default = cm;
 348                 return 0;
 349         }
 350         if (!strcmp(var, "diff.context")) {
 351                 diff_context_default = git_config_int(var, value);
 352                 if (diff_context_default < 0)
 353                         return -1;
 354                 return 0;
 355         }
 356         if (!strcmp(var, "diff.interhunkcontext")) {
 357                 diff_interhunk_context_default = git_config_int(var, value);
 358                 if (diff_interhunk_context_default < 0)
 359                         return -1;
 360                 return 0;
 361         }
 362         if (!strcmp(var, "diff.renames")) {
 363                 diff_detect_rename_default = git_config_rename(var, value);
 364                 return 0;
 365         }
 366         if (!strcmp(var, "diff.autorefreshindex")) {
 367                 diff_auto_refresh_index = git_config_bool(var, value);
 368                 return 0;
 369         }
 370         if (!strcmp(var, "diff.mnemonicprefix")) {
 371                 diff_mnemonic_prefix = git_config_bool(var, value);
 372                 return 0;
 373         }
 374         if (!strcmp(var, "diff.noprefix")) {
 375                 diff_no_prefix = git_config_bool(var, value);
 376                 return 0;
 377         }
 378         if (!strcmp(var, "diff.statgraphwidth")) {
 379                 diff_stat_graph_width = git_config_int(var, value);
 380                 return 0;
 381         }
 382         if (!strcmp(var, "diff.external"))
 383                 return git_config_string(&external_diff_cmd_cfg, var, value);
 384         if (!strcmp(var, "diff.wordregex"))
 385                 return git_config_string(&diff_word_regex_cfg, var, value);
 386         if (!strcmp(var, "diff.orderfile"))
 387                 return git_config_pathname(&diff_order_file_cfg, var, value);
 388
 389         if (!strcmp(var, "diff.ignoresubmodules"))
 390                 handle_ignore_submodules_arg(&default_diff_options, value);
 391
 392         if (!strcmp(var, "diff.submodule")) {
 393                 if (parse_submodule_params(&default_diff_options, value))
 394                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 395                                 value);
 396                 return 0;
 397         }
 398
 399         if (!strcmp(var, "diff.algorithm")) {
 400                 diff_algorithm = parse_algorithm_value(value);
 401                 if (diff_algorithm < 0)
 402                         return -1;
 403                 return 0;
 404         }
 405
 406         if (!strcmp(var, "diff.wserrorhighlight")) {
 407                 int val = parse_ws_error_highlight(value);
 408                 if (val < 0)
 409                         return -1;
 410                 ws_error_highlight_default = val;
 411                 return 0;
 412         }
 413
 414         if (git_color_config(var, value, cb) < 0)
 415                 return -1;
 416
 417         return git_diff_basic_config(var, value, cb);
 418 }
 419
 420 int git_diff_basic_config(const char *var, const char *value, void *cb)
 421 {
 422         const char *name;
 423
 424         if (!strcmp(var, "diff.renamelimit")) {
 425                 diff_rename_limit_default = git_config_int(var, value);
 426                 return 0;
 427         }
 428
 429         if (userdiff_config(var, value) < 0)
 430                 return -1;
 431
 432         if (skip_prefix(var, "diff.color.", &name) ||
 433             skip_prefix(var, "color.diff.", &name)) {
 434                 int slot = parse_diff_color_slot(name);
 435                 if (slot < 0)
 436                         return 0;
 437                 if (!value)
 438                         return config_error_nonbool(var);
 439                 return color_parse(value, diff_colors[slot]);
 440         }
 441
 442         /* like GNU diff's --suppress-blank-empty option  */
 443         if (!strcmp(var, "diff.suppressblankempty") ||
 444                         /* for backwards compatibility */
 445                         !strcmp(var, "diff.suppress-blank-empty")) {
 446                 diff_suppress_blank_empty = git_config_bool(var, value);
 447                 return 0;
 448         }
 449
 450         if (!strcmp(var, "diff.dirstat")) {
 451                 struct strbuf errmsg = STRBUF_INIT;
 452                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 453                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 454                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 455                                 errmsg.buf);
 456                 strbuf_release(&errmsg);
 457                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 458                 return 0;
 459         }
 460
 461         if (git_diff_heuristic_config(var, value, cb) < 0)
 462                 return -1;
 463
 464         return git_default_config(var, value, cb);
 465 }
 466
 467 static char *quote_two(const char *one, const char *two)
 468 {
 469         int need_one = quote_c_style(one, NULL, NULL, 1);
 470         int need_two = quote_c_style(two, NULL, NULL, 1);
 471         struct strbuf res = STRBUF_INIT;
 472
 473         if (need_one + need_two) {
 474                 strbuf_addch(&res, '"');
 475                 quote_c_style(one, &res, NULL, 1);
 476                 quote_c_style(two, &res, NULL, 1);
 477                 strbuf_addch(&res, '"');
 478         } else {
 479                 strbuf_addstr(&res, one);
 480                 strbuf_addstr(&res, two);
 481         }
 482         return strbuf_detach(&res, NULL);
 483 }
 484
 485 static const char *external_diff(void)
 486 {
 487         static const char *external_diff_cmd = NULL;
 488         static int done_preparing = 0;
 489
 490         if (done_preparing)
 491                 return external_diff_cmd;
 492         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 493         if (!external_diff_cmd)
 494                 external_diff_cmd = external_diff_cmd_cfg;
 495         done_preparing = 1;
 496         return external_diff_cmd;
 497 }
 498
 499 /*
 500  * Keep track of files used for diffing. Sometimes such an entry
 501  * refers to a temporary file, sometimes to an existing file, and
 502  * sometimes to "/dev/null".
 503  */
 504 static struct diff_tempfile {
 505         /*
 506          * filename external diff should read from, or NULL if this
 507          * entry is currently not in use:
 508          */
 509         const char *name;
 510
 511         char hex[GIT_MAX_HEXSZ + 1];
 512         char mode[10];
 513
 514         /*
 515          * If this diff_tempfile instance refers to a temporary file,
 516          * this tempfile object is used to manage its lifetime.
 517          */
 518         struct tempfile *tempfile;
 519 } diff_temp[2];
 520
 521 struct emit_callback {
 522         int color_diff;
 523         unsigned ws_rule;
 524         int blank_at_eof_in_preimage;
 525         int blank_at_eof_in_postimage;
 526         int lno_in_preimage;
 527         int lno_in_postimage;
 528         const char **label_path;
 529         struct diff_words_data *diff_words;
 530         struct diff_options *opt;
 531         struct strbuf *header;
 532 };
 533
 534 static int count_lines(const char *data, int size)
 535 {
 536         int count, ch, completely_empty = 1, nl_just_seen = 0;
 537         count = 0;
 538         while (0 < size--) {
 539                 ch = *data++;
 540                 if (ch == '\n') {
 541                         count++;
 542                         nl_just_seen = 1;
 543                         completely_empty = 0;
 544                 }
 545                 else {
 546                         nl_just_seen = 0;
 547                         completely_empty = 0;
 548                 }
 549         }
 550         if (completely_empty)
 551                 return 0;
 552         if (!nl_just_seen)
 553                 count++; /* no trailing newline */
 554         return count;
 555 }
 556
 557 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 558                        struct diff_filespec *one)
 559 {
 560         if (!DIFF_FILE_VALID(one)) {
 561                 mf->ptr = (char *)""; /* does not matter */
 562                 mf->size = 0;
 563                 return 0;
 564         }
 565         else if (diff_populate_filespec(r, one, 0))
 566                 return -1;
 567
 568         mf->ptr = one->data;
 569         mf->size = one->size;
 570         return 0;
 571 }
 572
 573 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 574 static unsigned long diff_filespec_size(struct repository *r,
 575                                         struct diff_filespec *one)
 576 {
 577         if (!DIFF_FILE_VALID(one))
 578                 return 0;
 579         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 580         return one->size;
 581 }
 582
 583 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 584 {
 585         char *ptr = mf->ptr;
 586         long size = mf->size;
 587         int cnt = 0;
 588
 589         if (!size)
 590                 return cnt;
 591         ptr += size - 1; /* pointing at the very end */
 592         if (*ptr != '\n')
 593                 ; /* incomplete line */
 594         else
 595                 ptr--; /* skip the last LF */
 596         while (mf->ptr < ptr) {
 597                 char *prev_eol;
 598                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 599                         if (*prev_eol == '\n')
 600                                 break;
 601                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 602                         break;
 603                 cnt++;
 604                 ptr = prev_eol - 1;
 605         }
 606         return cnt;
 607 }
 608
 609 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 610                                struct emit_callback *ecbdata)
 611 {
 612         int l1, l2, at;
 613         unsigned ws_rule = ecbdata->ws_rule;
 614         l1 = count_trailing_blank(mf1, ws_rule);
 615         l2 = count_trailing_blank(mf2, ws_rule);
 616         if (l2 <= l1) {
 617                 ecbdata->blank_at_eof_in_preimage = 0;
 618                 ecbdata->blank_at_eof_in_postimage = 0;
 619                 return;
 620         }
 621         at = count_lines(mf1->ptr, mf1->size);
 622         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 623
 624         at = count_lines(mf2->ptr, mf2->size);
 625         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 626 }
 627
 628 static void emit_line_0(struct diff_options *o,
 629                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 630                         int first, const char *line, int len)
 631 {
 632         int has_trailing_newline, has_trailing_carriage_return;
 633         int needs_reset = 0; /* at the end of the line */
 634         FILE *file = o->file;
 635
 636         fputs(diff_line_prefix(o), file);
 637
 638         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 639         if (has_trailing_newline)
 640                 len--;
 641
 642         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 643         if (has_trailing_carriage_return)
 644                 len--;
 645
 646         if (!len && !first)
 647                 goto end_of_line;
 648
 649         if (reverse && want_color(o->use_color)) {
 650                 fputs(GIT_COLOR_REVERSE, file);
 651                 needs_reset = 1;
 652         }
 653
 654         if (set_sign) {
 655                 fputs(set_sign, file);
 656                 needs_reset = 1;
 657         }
 658
 659         if (first)
 660                 fputc(first, file);
 661
 662         if (!len)
 663                 goto end_of_line;
 664
 665         if (set) {
 666                 if (set_sign && set != set_sign)
 667                         fputs(reset, file);
 668                 fputs(set, file);
 669                 needs_reset = 1;
 670         }
 671         fwrite(line, len, 1, file);
 672         needs_reset = 1; /* 'line' may contain color codes. */
 673
 674 end_of_line:
 675         if (needs_reset)
 676                 fputs(reset, file);
 677         if (has_trailing_carriage_return)
 678                 fputc('\r', file);
 679         if (has_trailing_newline)
 680                 fputc('\n', file);
 681 }
 682
 683 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 684                       const char *line, int len)
 685 {
 686         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 687 }
 688
 689 enum diff_symbol {
 690         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 691         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 692         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 693         DIFF_SYMBOL_BINARY_DIFF_BODY,
 694         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 695         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 696         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 697         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 698         DIFF_SYMBOL_STATS_LINE,
 699         DIFF_SYMBOL_WORD_DIFF,
 700         DIFF_SYMBOL_STAT_SEP,
 701         DIFF_SYMBOL_SUMMARY,
 702         DIFF_SYMBOL_SUBMODULE_ADD,
 703         DIFF_SYMBOL_SUBMODULE_DEL,
 704         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 705         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 706         DIFF_SYMBOL_SUBMODULE_HEADER,
 707         DIFF_SYMBOL_SUBMODULE_ERROR,
 708         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 709         DIFF_SYMBOL_REWRITE_DIFF,
 710         DIFF_SYMBOL_BINARY_FILES,
 711         DIFF_SYMBOL_HEADER,
 712         DIFF_SYMBOL_FILEPAIR_PLUS,
 713         DIFF_SYMBOL_FILEPAIR_MINUS,
 714         DIFF_SYMBOL_WORDS_PORCELAIN,
 715         DIFF_SYMBOL_WORDS,
 716         DIFF_SYMBOL_CONTEXT,
 717         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 718         DIFF_SYMBOL_PLUS,
 719         DIFF_SYMBOL_MINUS,
 720         DIFF_SYMBOL_NO_LF_EOF,
 721         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 722         DIFF_SYMBOL_CONTEXT_MARKER,
 723         DIFF_SYMBOL_SEPARATOR
 724 };
 725 /*
 726  * Flags for content lines:
 727  * 0..12 are whitespace rules
 728  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 729  * 16 is marking if the line is blank at EOF
 730  */
 731 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 732 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 733 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 734 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 735 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 736
 737 /*
 738  * This struct is used when we need to buffer the output of the diff output.
 739  *
 740  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 741  * into the pre/post image file. This pointer could be a union with the
 742  * line pointer. By storing an offset into the file instead of the literal line,
 743  * we can decrease the memory footprint for the buffered output. At first we
 744  * may want to only have indirection for the content lines, but we could also
 745  * enhance the state for emitting prefabricated lines, e.g. the similarity
 746  * score line or hunk/file headers would only need to store a number or path
 747  * and then the output can be constructed later on depending on state.
 748  */
 749 struct emitted_diff_symbol {
 750         const char *line;
 751         int len;
 752         int flags;
 753         enum diff_symbol s;
 754 };
 755 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 756
 757 struct emitted_diff_symbols {
 758         struct emitted_diff_symbol *buf;
 759         int nr, alloc;
 760 };
 761 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 762
 763 static void append_emitted_diff_symbol(struct diff_options *o,
 764                                        struct emitted_diff_symbol *e)
 765 {
 766         struct emitted_diff_symbol *f;
 767
 768         ALLOC_GROW(o->emitted_symbols->buf,
 769                    o->emitted_symbols->nr + 1,
 770                    o->emitted_symbols->alloc);
 771         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 772
 773         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 774         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 775 }
 776
 777 struct moved_entry {
 778         struct hashmap_entry ent;
 779         const struct emitted_diff_symbol *es;
 780         struct moved_entry *next_line;
 781         struct ws_delta *wsd;
 782 };
 783
 784 /**
 785  * The struct ws_delta holds white space differences between moved lines, i.e.
 786  * between '+' and '-' lines that have been detected to be a move.
 787  * The string contains the difference in leading white spaces, before the
 788  * rest of the line is compared using the white space config for move
 789  * coloring. The current_longer indicates if the first string in the
 790  * comparision is longer than the second.
 791  */
 792 struct ws_delta {
 793         char *string;
 794         unsigned int current_longer : 1;
 795 };
 796 #define WS_DELTA_INIT { NULL, 0 }
 797
 798 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 799                              const struct emitted_diff_symbol *b,
 800                              struct ws_delta *out)
 801 {
 802         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 803         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 804         int d = longer->len - shorter->len;
 805
 806         out->string = xmemdupz(longer->line, d);
 807         out->current_longer = (a == longer);
 808
 809         return !strncmp(longer->line + d, shorter->line, shorter->len);
 810 }
 811
 812 static int cmp_in_block_with_wsd(const struct diff_options *o,
 813                                  const struct moved_entry *cur,
 814                                  const struct moved_entry *match,
 815                                  struct moved_entry *pmb,
 816                                  int n)
 817 {
 818         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 819         int al = cur->es->len, cl = l->len;
 820         const char *a = cur->es->line,
 821                    *b = match->es->line,
 822                    *c = l->line;
 823
 824         int wslen;
 825
 826         /*
 827          * We need to check if 'cur' is equal to 'match'.
 828          * As those are from the same (+/-) side, we do not need to adjust for
 829          * indent changes. However these were found using fuzzy matching
 830          * so we do have to check if they are equal.
 831          */
 832         if (strcmp(a, b))
 833                 return 1;
 834
 835         if (!pmb->wsd)
 836                 /*
 837                  * No white space delta was carried forward? This can happen
 838                  * when we exit early in this function and do not carry
 839                  * forward ws.
 840                  */
 841                 return 1;
 842
 843         /*
 844          * The indent changes of the block are known and carried forward in
 845          * pmb->wsd; however we need to check if the indent changes of the
 846          * current line are still the same as before.
 847          *
 848          * To do so we need to compare 'l' to 'cur', adjusting the
 849          * one of them for the white spaces, depending which was longer.
 850          */
 851
 852         wslen = strlen(pmb->wsd->string);
 853         if (pmb->wsd->current_longer) {
 854                 c += wslen;
 855                 cl -= wslen;
 856         } else {
 857                 a += wslen;
 858                 al -= wslen;
 859         }
 860
 861         if (strcmp(a, c))
 862                 return 1;
 863
 864         return 0;
 865 }
 866
 867 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 868                            const void *entry,
 869                            const void *entry_or_key,
 870                            const void *keydata)
 871 {
 872         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 873         const struct moved_entry *a = entry;
 874         const struct moved_entry *b = entry_or_key;
 875         unsigned flags = diffopt->color_moved_ws_handling
 876                          & XDF_WHITESPACE_FLAGS;
 877
 878         if (diffopt->color_moved_ws_handling &
 879             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 880                 /*
 881                  * As there is not specific white space config given,
 882                  * we'd need to check for a new block, so ignore all
 883                  * white space. The setup of the white space
 884                  * configuration for the next block is done else where
 885                  */
 886                 flags |= XDF_IGNORE_WHITESPACE;
 887
 888         return !xdiff_compare_lines(a->es->line, a->es->len,
 889                                     b->es->line, b->es->len,
 890                                     flags);
 891 }
 892
 893 static struct moved_entry *prepare_entry(struct diff_options *o,
 894                                          int line_no)
 895 {
 896         struct moved_entry *ret = xmalloc(sizeof(*ret));
 897         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 898         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 899
 900         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 901         ret->es = l;
 902         ret->next_line = NULL;
 903         ret->wsd = NULL;
 904
 905         return ret;
 906 }
 907
 908 static void add_lines_to_move_detection(struct diff_options *o,
 909                                         struct hashmap *add_lines,
 910                                         struct hashmap *del_lines)
 911 {
 912         struct moved_entry *prev_line = NULL;
 913
 914         int n;
 915         for (n = 0; n < o->emitted_symbols->nr; n++) {
 916                 struct hashmap *hm;
 917                 struct moved_entry *key;
 918
 919                 switch (o->emitted_symbols->buf[n].s) {
 920                 case DIFF_SYMBOL_PLUS:
 921                         hm = add_lines;
 922                         break;
 923                 case DIFF_SYMBOL_MINUS:
 924                         hm = del_lines;
 925                         break;
 926                 default:
 927                         prev_line = NULL;
 928                         continue;
 929                 }
 930
 931                 key = prepare_entry(o, n);
 932                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 933                         prev_line->next_line = key;
 934
 935                 hashmap_add(hm, key);
 936                 prev_line = key;
 937         }
 938 }
 939
 940 static void pmb_advance_or_null(struct diff_options *o,
 941                                 struct moved_entry *match,
 942                                 struct hashmap *hm,
 943                                 struct moved_entry **pmb,
 944                                 int pmb_nr)
 945 {
 946         int i;
 947         for (i = 0; i < pmb_nr; i++) {
 948                 struct moved_entry *prev = pmb[i];
 949                 struct moved_entry *cur = (prev && prev->next_line) ?
 950                                 prev->next_line : NULL;
 951                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 952                         pmb[i] = cur;
 953                 } else {
 954                         pmb[i] = NULL;
 955                 }
 956         }
 957 }
 958
 959 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 960                                             struct moved_entry *match,
 961                                             struct hashmap *hm,
 962                                             struct moved_entry **pmb,
 963                                             int pmb_nr, int n)
 964 {
 965         int i;
 966         char *got_match = xcalloc(1, pmb_nr);
 967
 968         for (; match; match = hashmap_get_next(hm, match)) {
 969                 for (i = 0; i < pmb_nr; i++) {
 970                         struct moved_entry *prev = pmb[i];
 971                         struct moved_entry *cur = (prev && prev->next_line) ?
 972                                         prev->next_line : NULL;
 973                         if (!cur)
 974                                 continue;
 975                         if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
 976                                 got_match[i] |= 1;
 977                 }
 978         }
 979
 980         for (i = 0; i < pmb_nr; i++) {
 981                 if (got_match[i]) {
 982                         /* Carry the white space delta forward */
 983                         pmb[i]->next_line->wsd = pmb[i]->wsd;
 984                         pmb[i] = pmb[i]->next_line;
 985                 } else {
 986                         if (pmb[i]->wsd) {
 987                                 free(pmb[i]->wsd->string);
 988                                 FREE_AND_NULL(pmb[i]->wsd);
 989                         }
 990                         pmb[i] = NULL;
 991                 }
 992         }
 993 }
 994
 995 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 996                                          int pmb_nr)
 997 {
 998         int lp, rp;
 999
1000         /* Shrink the set of potential block to the remaining running */
1001         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1002                 while (lp < pmb_nr && pmb[lp])
1003                         lp++;
1004                 /* lp points at the first NULL now */
1005
1006                 while (rp > -1 && !pmb[rp])
1007                         rp--;
1008                 /* rp points at the last non-NULL */
1009
1010                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1011                         pmb[lp] = pmb[rp];
1012                         pmb[rp] = NULL;
1013                         rp--;
1014                         lp++;
1015                 }
1016         }
1017
1018         /* Remember the number of running sets */
1019         return rp + 1;
1020 }
1021
1022 /*
1023  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1024  *
1025  * Otherwise, if the last block has fewer alphanumeric characters than
1026  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1027  * that block.
1028  *
1029  * The last block consists of the (n - block_length)'th line up to but not
1030  * including the nth line.
1031  *
1032  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1033  * Think of a way to unify them.
1034  */
1035 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1036 {
1037         int i, alnum_count = 0;
1038         if (o->color_moved == COLOR_MOVED_PLAIN)
1039                 return;
1040         for (i = 1; i < block_length + 1; i++) {
1041                 const char *c = o->emitted_symbols->buf[n - i].line;
1042                 for (; *c; c++) {
1043                         if (!isalnum(*c))
1044                                 continue;
1045                         alnum_count++;
1046                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1047                                 return;
1048                 }
1049         }
1050         for (i = 1; i < block_length + 1; i++)
1051                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1052 }
1053
1054 /* Find blocks of moved code, delegate actual coloring decision to helper */
1055 static void mark_color_as_moved(struct diff_options *o,
1056                                 struct hashmap *add_lines,
1057                                 struct hashmap *del_lines)
1058 {
1059         struct moved_entry **pmb = NULL; /* potentially moved blocks */
1060         int pmb_nr = 0, pmb_alloc = 0;
1061         int n, flipped_block = 1, block_length = 0;
1062
1063
1064         for (n = 0; n < o->emitted_symbols->nr; n++) {
1065                 struct hashmap *hm = NULL;
1066                 struct moved_entry *key;
1067                 struct moved_entry *match = NULL;
1068                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1069
1070                 switch (l->s) {
1071                 case DIFF_SYMBOL_PLUS:
1072                         hm = del_lines;
1073                         key = prepare_entry(o, n);
1074                         match = hashmap_get(hm, key, NULL);
1075                         free(key);
1076                         break;
1077                 case DIFF_SYMBOL_MINUS:
1078                         hm = add_lines;
1079                         key = prepare_entry(o, n);
1080                         match = hashmap_get(hm, key, NULL);
1081                         free(key);
1082                         break;
1083                 default:
1084                         flipped_block = 1;
1085                 }
1086
1087                 if (!match) {
1088                         adjust_last_block(o, n, block_length);
1089                         pmb_nr = 0;
1090                         block_length = 0;
1091                         continue;
1092                 }
1093
1094                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1095
1096                 if (o->color_moved == COLOR_MOVED_PLAIN)
1097                         continue;
1098
1099                 if (o->color_moved_ws_handling &
1100                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1101                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1102                 else
1103                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1104
1105                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1106
1107                 if (pmb_nr == 0) {
1108                         /*
1109                          * The current line is the start of a new block.
1110                          * Setup the set of potential blocks.
1111                          */
1112                         for (; match; match = hashmap_get_next(hm, match)) {
1113                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1114                                 if (o->color_moved_ws_handling &
1115                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1116                                         struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
1117                                         if (compute_ws_delta(l, match->es, wsd)) {
1118                                                 match->wsd = wsd;
1119                                                 pmb[pmb_nr++] = match;
1120                                         } else
1121                                                 free(wsd);
1122                                 } else {
1123                                         pmb[pmb_nr++] = match;
1124                                 }
1125                         }
1126
1127                         flipped_block = (flipped_block + 1) % 2;
1128
1129                         adjust_last_block(o, n, block_length);
1130                         block_length = 0;
1131                 }
1132
1133                 block_length++;
1134
1135                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1136                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1137         }
1138         adjust_last_block(o, n, block_length);
1139
1140         free(pmb);
1141 }
1142
1143 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1144   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1145 static void dim_moved_lines(struct diff_options *o)
1146 {
1147         int n;
1148         for (n = 0; n < o->emitted_symbols->nr; n++) {
1149                 struct emitted_diff_symbol *prev = (n != 0) ?
1150                                 &o->emitted_symbols->buf[n - 1] : NULL;
1151                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1152                 struct emitted_diff_symbol *next =
1153                                 (n < o->emitted_symbols->nr - 1) ?
1154                                 &o->emitted_symbols->buf[n + 1] : NULL;
1155
1156                 /* Not a plus or minus line? */
1157                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1158                         continue;
1159
1160                 /* Not a moved line? */
1161                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1162                         continue;
1163
1164                 /*
1165                  * If prev or next are not a plus or minus line,
1166                  * pretend they don't exist
1167                  */
1168                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1169                             prev->s != DIFF_SYMBOL_MINUS)
1170                         prev = NULL;
1171                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1172                             next->s != DIFF_SYMBOL_MINUS)
1173                         next = NULL;
1174
1175                 /* Inside a block? */
1176                 if ((prev &&
1177                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1178                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1179                     (next &&
1180                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1181                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1182                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1183                         continue;
1184                 }
1185
1186                 /* Check if we are at an interesting bound: */
1187                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1188                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1189                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1190                         continue;
1191                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1192                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1193                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1194                         continue;
1195
1196                 /*
1197                  * The boundary to prev and next are not interesting,
1198                  * so this line is not interesting as a whole
1199                  */
1200                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1201         }
1202 }
1203
1204 static void emit_line_ws_markup(struct diff_options *o,
1205                                 const char *set_sign, const char *set,
1206                                 const char *reset,
1207                                 char sign, const char *line, int len,
1208                                 unsigned ws_rule, int blank_at_eof)
1209 {
1210         const char *ws = NULL;
1211
1212         if (o->ws_error_highlight & ws_rule) {
1213                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1214                 if (!*ws)
1215                         ws = NULL;
1216         }
1217
1218         if (!ws && !set_sign)
1219                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1220         else if (!ws) {
1221                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1222         } else if (blank_at_eof)
1223                 /* Blank line at EOF - paint '+' as well */
1224                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1225         else {
1226                 /* Emit just the prefix, then the rest. */
1227                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1228                             sign, "", 0);
1229                 ws_check_emit(line, len, ws_rule,
1230                               o->file, set, reset, ws);
1231         }
1232 }
1233
1234 static void emit_diff_symbol_from_struct(struct diff_options *o,
1235                                          struct emitted_diff_symbol *eds)
1236 {
1237         static const char *nneof = " No newline at end of file\n";
1238         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1239         struct strbuf sb = STRBUF_INIT;
1240
1241         enum diff_symbol s = eds->s;
1242         const char *line = eds->line;
1243         int len = eds->len;
1244         unsigned flags = eds->flags;
1245
1246         switch (s) {
1247         case DIFF_SYMBOL_NO_LF_EOF:
1248                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1249                 reset = diff_get_color_opt(o, DIFF_RESET);
1250                 putc('\n', o->file);
1251                 emit_line_0(o, context, NULL, 0, reset, '\\',
1252                             nneof, strlen(nneof));
1253                 break;
1254         case DIFF_SYMBOL_SUBMODULE_HEADER:
1255         case DIFF_SYMBOL_SUBMODULE_ERROR:
1256         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1257         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1258         case DIFF_SYMBOL_SUMMARY:
1259         case DIFF_SYMBOL_STATS_LINE:
1260         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1261         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1262                 emit_line(o, "", "", line, len);
1263                 break;
1264         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1265         case DIFF_SYMBOL_CONTEXT_MARKER:
1266                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1267                 reset = diff_get_color_opt(o, DIFF_RESET);
1268                 emit_line(o, context, reset, line, len);
1269                 break;
1270         case DIFF_SYMBOL_SEPARATOR:
1271                 fprintf(o->file, "%s%c",
1272                         diff_line_prefix(o),
1273                         o->line_termination);
1274                 break;
1275         case DIFF_SYMBOL_CONTEXT:
1276                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1277                 reset = diff_get_color_opt(o, DIFF_RESET);
1278                 set_sign = NULL;
1279                 if (o->flags.dual_color_diffed_diffs) {
1280                         char c = !len ? 0 : line[0];
1281
1282                         if (c == '+')
1283                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1284                         else if (c == '@')
1285                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1286                         else if (c == '-')
1287                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1288                 }
1289                 emit_line_ws_markup(o, set_sign, set, reset,
1290                                     o->output_indicators[OUTPUT_INDICATOR_CONTEXT],
1291                                     line, len,
1292                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1293                 break;
1294         case DIFF_SYMBOL_PLUS:
1295                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1296                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1297                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1298                 case DIFF_SYMBOL_MOVED_LINE |
1299                      DIFF_SYMBOL_MOVED_LINE_ALT |
1300                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1301                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1302                         break;
1303                 case DIFF_SYMBOL_MOVED_LINE |
1304                      DIFF_SYMBOL_MOVED_LINE_ALT:
1305                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1306                         break;
1307                 case DIFF_SYMBOL_MOVED_LINE |
1308                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1309                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1310                         break;
1311                 case DIFF_SYMBOL_MOVED_LINE:
1312                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1313                         break;
1314                 default:
1315                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1316                 }
1317                 reset = diff_get_color_opt(o, DIFF_RESET);
1318                 if (!o->flags.dual_color_diffed_diffs)
1319                         set_sign = NULL;
1320                 else {
1321                         char c = !len ? 0 : line[0];
1322
1323                         set_sign = set;
1324                         if (c == '-')
1325                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1326                         else if (c == '@')
1327                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1328                         else if (c == '+')
1329                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1330                         else
1331                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1332                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1333                 }
1334                 emit_line_ws_markup(o, set_sign, set, reset,
1335                                     o->output_indicators[OUTPUT_INDICATOR_NEW],
1336                                     line, len,
1337                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1338                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1339                 break;
1340         case DIFF_SYMBOL_MINUS:
1341                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1342                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1343                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1344                 case DIFF_SYMBOL_MOVED_LINE |
1345                      DIFF_SYMBOL_MOVED_LINE_ALT |
1346                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1347                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1348                         break;
1349                 case DIFF_SYMBOL_MOVED_LINE |
1350                      DIFF_SYMBOL_MOVED_LINE_ALT:
1351                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1352                         break;
1353                 case DIFF_SYMBOL_MOVED_LINE |
1354                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1355                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1356                         break;
1357                 case DIFF_SYMBOL_MOVED_LINE:
1358                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1359                         break;
1360                 default:
1361                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1362                 }
1363                 reset = diff_get_color_opt(o, DIFF_RESET);
1364                 if (!o->flags.dual_color_diffed_diffs)
1365                         set_sign = NULL;
1366                 else {
1367                         char c = !len ? 0 : line[0];
1368
1369                         set_sign = set;
1370                         if (c == '+')
1371                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1372                         else if (c == '@')
1373                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1374                         else if (c == '-')
1375                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1376                         else
1377                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1378                 }
1379                 emit_line_ws_markup(o, set_sign, set, reset,
1380                                     o->output_indicators[OUTPUT_INDICATOR_OLD],
1381                                     line, len,
1382                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1383                 break;
1384         case DIFF_SYMBOL_WORDS_PORCELAIN:
1385                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1386                 reset = diff_get_color_opt(o, DIFF_RESET);
1387                 emit_line(o, context, reset, line, len);
1388                 fputs("~\n", o->file);
1389                 break;
1390         case DIFF_SYMBOL_WORDS:
1391                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1392                 reset = diff_get_color_opt(o, DIFF_RESET);
1393                 /*
1394                  * Skip the prefix character, if any.  With
1395                  * diff_suppress_blank_empty, there may be
1396                  * none.
1397                  */
1398                 if (line[0] != '\n') {
1399                         line++;
1400                         len--;
1401                 }
1402                 emit_line(o, context, reset, line, len);
1403                 break;
1404         case DIFF_SYMBOL_FILEPAIR_PLUS:
1405                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1406                 reset = diff_get_color_opt(o, DIFF_RESET);
1407                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1408                         line, reset,
1409                         strchr(line, ' ') ? "\t" : "");
1410                 break;
1411         case DIFF_SYMBOL_FILEPAIR_MINUS:
1412                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1413                 reset = diff_get_color_opt(o, DIFF_RESET);
1414                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1415                         line, reset,
1416                         strchr(line, ' ') ? "\t" : "");
1417                 break;
1418         case DIFF_SYMBOL_BINARY_FILES:
1419         case DIFF_SYMBOL_HEADER:
1420                 fprintf(o->file, "%s", line);
1421                 break;
1422         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1423                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1424                 break;
1425         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1426                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1427                 break;
1428         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1429                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1430                 break;
1431         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1432                 fputs(diff_line_prefix(o), o->file);
1433                 fputc('\n', o->file);
1434                 break;
1435         case DIFF_SYMBOL_REWRITE_DIFF:
1436                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1437                 reset = diff_get_color_opt(o, DIFF_RESET);
1438                 emit_line(o, fraginfo, reset, line, len);
1439                 break;
1440         case DIFF_SYMBOL_SUBMODULE_ADD:
1441                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1442                 reset = diff_get_color_opt(o, DIFF_RESET);
1443                 emit_line(o, set, reset, line, len);
1444                 break;
1445         case DIFF_SYMBOL_SUBMODULE_DEL:
1446                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1447                 reset = diff_get_color_opt(o, DIFF_RESET);
1448                 emit_line(o, set, reset, line, len);
1449                 break;
1450         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1451                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1452                         diff_line_prefix(o), line);
1453                 break;
1454         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1455                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1456                         diff_line_prefix(o), line);
1457                 break;
1458         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1459                 emit_line(o, "", "", " 0 files changed\n",
1460                           strlen(" 0 files changed\n"));
1461                 break;
1462         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1463                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1464                 break;
1465         case DIFF_SYMBOL_WORD_DIFF:
1466                 fprintf(o->file, "%.*s", len, line);
1467                 break;
1468         case DIFF_SYMBOL_STAT_SEP:
1469                 fputs(o->stat_sep, o->file);
1470                 break;
1471         default:
1472                 BUG("unknown diff symbol");
1473         }
1474         strbuf_release(&sb);
1475 }
1476
1477 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1478                              const char *line, int len, unsigned flags)
1479 {
1480         struct emitted_diff_symbol e = {line, len, flags, s};
1481
1482         if (o->emitted_symbols)
1483                 append_emitted_diff_symbol(o, &e);
1484         else
1485                 emit_diff_symbol_from_struct(o, &e);
1486 }
1487
1488 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1489 {
1490         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1491 }
1492
1493 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1494 {
1495         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1496 }
1497
1498 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1499 {
1500         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1501                          path, strlen(path), 0);
1502 }
1503
1504 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1505 {
1506         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1507                          path, strlen(path), 0);
1508 }
1509
1510 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1511 {
1512         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1513                          header, strlen(header), 0);
1514 }
1515
1516 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1517 {
1518         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1519 }
1520
1521 void diff_emit_submodule_pipethrough(struct diff_options *o,
1522                                      const char *line, int len)
1523 {
1524         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1525 }
1526
1527 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1528 {
1529         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1530               ecbdata->blank_at_eof_in_preimage &&
1531               ecbdata->blank_at_eof_in_postimage &&
1532               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1533               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1534                 return 0;
1535         return ws_blank_line(line, len, ecbdata->ws_rule);
1536 }
1537
1538 static void emit_add_line(const char *reset,
1539                           struct emit_callback *ecbdata,
1540                           const char *line, int len)
1541 {
1542         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1543         if (new_blank_line_at_eof(ecbdata, line, len))
1544                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1545
1546         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1547 }
1548
1549 static void emit_del_line(const char *reset,
1550                           struct emit_callback *ecbdata,
1551                           const char *line, int len)
1552 {
1553         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1554         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1555 }
1556
1557 static void emit_context_line(const char *reset,
1558                               struct emit_callback *ecbdata,
1559                               const char *line, int len)
1560 {
1561         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1562         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1563 }
1564
1565 static void emit_hunk_header(struct emit_callback *ecbdata,
1566                              const char *line, int len)
1567 {
1568         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1569         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1570         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1571         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1572         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1573         static const char atat[2] = { '@', '@' };
1574         const char *cp, *ep;
1575         struct strbuf msgbuf = STRBUF_INIT;
1576         int org_len = len;
1577         int i = 1;
1578
1579         /*
1580          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1581          * it always is at least 10 bytes long.
1582          */
1583         if (len < 10 ||
1584             memcmp(line, atat, 2) ||
1585             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1586                 emit_diff_symbol(ecbdata->opt,
1587                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1588                 return;
1589         }
1590         ep += 2; /* skip over @@ */
1591
1592         /* The hunk header in fraginfo color */
1593         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1594                 strbuf_addstr(&msgbuf, reverse);
1595         strbuf_addstr(&msgbuf, frag);
1596         strbuf_add(&msgbuf, line, ep - line);
1597         strbuf_addstr(&msgbuf, reset);
1598
1599         /*
1600          * trailing "\r\n"
1601          */
1602         for ( ; i < 3; i++)
1603                 if (line[len - i] == '\r' || line[len - i] == '\n')
1604                         len--;
1605
1606         /* blank before the func header */
1607         for (cp = ep; ep - line < len; ep++)
1608                 if (*ep != ' ' && *ep != '\t')
1609                         break;
1610         if (ep != cp) {
1611                 strbuf_addstr(&msgbuf, context);
1612                 strbuf_add(&msgbuf, cp, ep - cp);
1613                 strbuf_addstr(&msgbuf, reset);
1614         }
1615
1616         if (ep < line + len) {
1617                 strbuf_addstr(&msgbuf, func);
1618                 strbuf_add(&msgbuf, ep, line + len - ep);
1619                 strbuf_addstr(&msgbuf, reset);
1620         }
1621
1622         strbuf_add(&msgbuf, line + len, org_len - len);
1623         strbuf_complete_line(&msgbuf);
1624         emit_diff_symbol(ecbdata->opt,
1625                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1626         strbuf_release(&msgbuf);
1627 }
1628
1629 static struct diff_tempfile *claim_diff_tempfile(void) {
1630         int i;
1631         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1632                 if (!diff_temp[i].name)
1633                         return diff_temp + i;
1634         BUG("diff is failing to clean up its tempfiles");
1635 }
1636
1637 static void remove_tempfile(void)
1638 {
1639         int i;
1640         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1641                 if (is_tempfile_active(diff_temp[i].tempfile))
1642                         delete_tempfile(&diff_temp[i].tempfile);
1643                 diff_temp[i].name = NULL;
1644         }
1645 }
1646
1647 static void add_line_count(struct strbuf *out, int count)
1648 {
1649         switch (count) {
1650         case 0:
1651                 strbuf_addstr(out, "0,0");
1652                 break;
1653         case 1:
1654                 strbuf_addstr(out, "1");
1655                 break;
1656         default:
1657                 strbuf_addf(out, "1,%d", count);
1658                 break;
1659         }
1660 }
1661
1662 static void emit_rewrite_lines(struct emit_callback *ecb,
1663                                int prefix, const char *data, int size)
1664 {
1665         const char *endp = NULL;
1666         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1667
1668         while (0 < size) {
1669                 int len;
1670
1671                 endp = memchr(data, '\n', size);
1672                 len = endp ? (endp - data + 1) : size;
1673                 if (prefix != '+') {
1674                         ecb->lno_in_preimage++;
1675                         emit_del_line(reset, ecb, data, len);
1676                 } else {
1677                         ecb->lno_in_postimage++;
1678                         emit_add_line(reset, ecb, data, len);
1679                 }
1680                 size -= len;
1681                 data += len;
1682         }
1683         if (!endp)
1684                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1685 }
1686
1687 static void emit_rewrite_diff(const char *name_a,
1688                               const char *name_b,
1689                               struct diff_filespec *one,
1690                               struct diff_filespec *two,
1691                               struct userdiff_driver *textconv_one,
1692                               struct userdiff_driver *textconv_two,
1693                               struct diff_options *o)
1694 {
1695         int lc_a, lc_b;
1696         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1697         const char *a_prefix, *b_prefix;
1698         char *data_one, *data_two;
1699         size_t size_one, size_two;
1700         struct emit_callback ecbdata;
1701         struct strbuf out = STRBUF_INIT;
1702
1703         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1704                 a_prefix = o->b_prefix;
1705                 b_prefix = o->a_prefix;
1706         } else {
1707                 a_prefix = o->a_prefix;
1708                 b_prefix = o->b_prefix;
1709         }
1710
1711         name_a += (*name_a == '/');
1712         name_b += (*name_b == '/');
1713
1714         strbuf_reset(&a_name);
1715         strbuf_reset(&b_name);
1716         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1717         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1718
1719         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1720         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1721
1722         memset(&ecbdata, 0, sizeof(ecbdata));
1723         ecbdata.color_diff = want_color(o->use_color);
1724         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1725         ecbdata.opt = o;
1726         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1727                 mmfile_t mf1, mf2;
1728                 mf1.ptr = (char *)data_one;
1729                 mf2.ptr = (char *)data_two;
1730                 mf1.size = size_one;
1731                 mf2.size = size_two;
1732                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1733         }
1734         ecbdata.lno_in_preimage = 1;
1735         ecbdata.lno_in_postimage = 1;
1736
1737         lc_a = count_lines(data_one, size_one);
1738         lc_b = count_lines(data_two, size_two);
1739
1740         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1741                          a_name.buf, a_name.len, 0);
1742         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1743                          b_name.buf, b_name.len, 0);
1744
1745         strbuf_addstr(&out, "@@ -");
1746         if (!o->irreversible_delete)
1747                 add_line_count(&out, lc_a);
1748         else
1749                 strbuf_addstr(&out, "?,?");
1750         strbuf_addstr(&out, " +");
1751         add_line_count(&out, lc_b);
1752         strbuf_addstr(&out, " @@\n");
1753         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1754         strbuf_release(&out);
1755
1756         if (lc_a && !o->irreversible_delete)
1757                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1758         if (lc_b)
1759                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1760         if (textconv_one)
1761                 free((char *)data_one);
1762         if (textconv_two)
1763                 free((char *)data_two);
1764 }
1765
1766 struct diff_words_buffer {
1767         mmfile_t text;
1768         unsigned long alloc;
1769         struct diff_words_orig {
1770                 const char *begin, *end;
1771         } *orig;
1772         int orig_nr, orig_alloc;
1773 };
1774
1775 static void diff_words_append(char *line, unsigned long len,
1776                 struct diff_words_buffer *buffer)
1777 {
1778         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1779         line++;
1780         len--;
1781         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1782         buffer->text.size += len;
1783         buffer->text.ptr[buffer->text.size] = '\0';
1784 }
1785
1786 struct diff_words_style_elem {
1787         const char *prefix;
1788         const char *suffix;
1789         const char *color; /* NULL; filled in by the setup code if
1790                             * color is enabled */
1791 };
1792
1793 struct diff_words_style {
1794         enum diff_words_type type;
1795         struct diff_words_style_elem new_word, old_word, ctx;
1796         const char *newline;
1797 };
1798
1799 static struct diff_words_style diff_words_styles[] = {
1800         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1801         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1802         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1803 };
1804
1805 struct diff_words_data {
1806         struct diff_words_buffer minus, plus;
1807         const char *current_plus;
1808         int last_minus;
1809         struct diff_options *opt;
1810         regex_t *word_regex;
1811         enum diff_words_type type;
1812         struct diff_words_style *style;
1813 };
1814
1815 static int fn_out_diff_words_write_helper(struct diff_options *o,
1816                                           struct diff_words_style_elem *st_el,
1817                                           const char *newline,
1818                                           size_t count, const char *buf)
1819 {
1820         int print = 0;
1821         struct strbuf sb = STRBUF_INIT;
1822
1823         while (count) {
1824                 char *p = memchr(buf, '\n', count);
1825                 if (print)
1826                         strbuf_addstr(&sb, diff_line_prefix(o));
1827
1828                 if (p != buf) {
1829                         const char *reset = st_el->color && *st_el->color ?
1830                                             GIT_COLOR_RESET : NULL;
1831                         if (st_el->color && *st_el->color)
1832                                 strbuf_addstr(&sb, st_el->color);
1833                         strbuf_addstr(&sb, st_el->prefix);
1834                         strbuf_add(&sb, buf, p ? p - buf : count);
1835                         strbuf_addstr(&sb, st_el->suffix);
1836                         if (reset)
1837                                 strbuf_addstr(&sb, reset);
1838                 }
1839                 if (!p)
1840                         goto out;
1841
1842                 strbuf_addstr(&sb, newline);
1843                 count -= p + 1 - buf;
1844                 buf = p + 1;
1845                 print = 1;
1846                 if (count) {
1847                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1848                                          sb.buf, sb.len, 0);
1849                         strbuf_reset(&sb);
1850                 }
1851         }
1852
1853 out:
1854         if (sb.len)
1855                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1856                                  sb.buf, sb.len, 0);
1857         strbuf_release(&sb);
1858         return 0;
1859 }
1860
1861 /*
1862  * '--color-words' algorithm can be described as:
1863  *
1864  *   1. collect the minus/plus lines of a diff hunk, divided into
1865  *      minus-lines and plus-lines;
1866  *
1867  *   2. break both minus-lines and plus-lines into words and
1868  *      place them into two mmfile_t with one word for each line;
1869  *
1870  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1871  *
1872  * And for the common parts of the both file, we output the plus side text.
1873  * diff_words->current_plus is used to trace the current position of the plus file
1874  * which printed. diff_words->last_minus is used to trace the last minus word
1875  * printed.
1876  *
1877  * For '--graph' to work with '--color-words', we need to output the graph prefix
1878  * on each line of color words output. Generally, there are two conditions on
1879  * which we should output the prefix.
1880  *
1881  *   1. diff_words->last_minus == 0 &&
1882  *      diff_words->current_plus == diff_words->plus.text.ptr
1883  *
1884  *      that is: the plus text must start as a new line, and if there is no minus
1885  *      word printed, a graph prefix must be printed.
1886  *
1887  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1888  *      *(diff_words->current_plus - 1) == '\n'
1889  *
1890  *      that is: a graph prefix must be printed following a '\n'
1891  */
1892 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1893 {
1894         if ((diff_words->last_minus == 0 &&
1895                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1896                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1897                 *(diff_words->current_plus - 1) == '\n')) {
1898                 return 1;
1899         } else {
1900                 return 0;
1901         }
1902 }
1903
1904 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1905 {
1906         struct diff_words_data *diff_words = priv;
1907         struct diff_words_style *style = diff_words->style;
1908         int minus_first, minus_len, plus_first, plus_len;
1909         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1910         struct diff_options *opt = diff_words->opt;
1911         const char *line_prefix;
1912
1913         if (line[0] != '@' || parse_hunk_header(line, len,
1914                         &minus_first, &minus_len, &plus_first, &plus_len))
1915                 return;
1916
1917         assert(opt);
1918         line_prefix = diff_line_prefix(opt);
1919
1920         /* POSIX requires that first be decremented by one if len == 0... */
1921         if (minus_len) {
1922                 minus_begin = diff_words->minus.orig[minus_first].begin;
1923                 minus_end =
1924                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1925         } else
1926                 minus_begin = minus_end =
1927                         diff_words->minus.orig[minus_first].end;
1928
1929         if (plus_len) {
1930                 plus_begin = diff_words->plus.orig[plus_first].begin;
1931                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1932         } else
1933                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1934
1935         if (color_words_output_graph_prefix(diff_words)) {
1936                 fputs(line_prefix, diff_words->opt->file);
1937         }
1938         if (diff_words->current_plus != plus_begin) {
1939                 fn_out_diff_words_write_helper(diff_words->opt,
1940                                 &style->ctx, style->newline,
1941                                 plus_begin - diff_words->current_plus,
1942                                 diff_words->current_plus);
1943         }
1944         if (minus_begin != minus_end) {
1945                 fn_out_diff_words_write_helper(diff_words->opt,
1946                                 &style->old_word, style->newline,
1947                                 minus_end - minus_begin, minus_begin);
1948         }
1949         if (plus_begin != plus_end) {
1950                 fn_out_diff_words_write_helper(diff_words->opt,
1951                                 &style->new_word, style->newline,
1952                                 plus_end - plus_begin, plus_begin);
1953         }
1954
1955         diff_words->current_plus = plus_end;
1956         diff_words->last_minus = minus_first;
1957 }
1958
1959 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1960 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1961                 int *begin, int *end)
1962 {
1963         if (word_regex && *begin < buffer->size) {
1964                 regmatch_t match[1];
1965                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1966                                  buffer->size - *begin, 1, match, 0)) {
1967                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1968                                         '\n', match[0].rm_eo - match[0].rm_so);
1969                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1970                         *begin += match[0].rm_so;
1971                         return *begin >= *end;
1972                 }
1973                 return -1;
1974         }
1975
1976         /* find the next word */
1977         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1978                 (*begin)++;
1979         if (*begin >= buffer->size)
1980                 return -1;
1981
1982         /* find the end of the word */
1983         *end = *begin + 1;
1984         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1985                 (*end)++;
1986
1987         return 0;
1988 }
1989
1990 /*
1991  * This function splits the words in buffer->text, stores the list with
1992  * newline separator into out, and saves the offsets of the original words
1993  * in buffer->orig.
1994  */
1995 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1996                 regex_t *word_regex)
1997 {
1998         int i, j;
1999         long alloc = 0;
2000
2001         out->size = 0;
2002         out->ptr = NULL;
2003
2004         /* fake an empty "0th" word */
2005         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2006         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2007         buffer->orig_nr = 1;
2008
2009         for (i = 0; i < buffer->text.size; i++) {
2010                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2011                         return;
2012
2013                 /* store original boundaries */
2014                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2015                                 buffer->orig_alloc);
2016                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2017                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2018                 buffer->orig_nr++;
2019
2020                 /* store one word */
2021                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2022                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2023                 out->ptr[out->size + j - i] = '\n';
2024                 out->size += j - i + 1;
2025
2026                 i = j - 1;
2027         }
2028 }
2029
2030 /* this executes the word diff on the accumulated buffers */
2031 static void diff_words_show(struct diff_words_data *diff_words)
2032 {
2033         xpparam_t xpp;
2034         xdemitconf_t xecfg;
2035         mmfile_t minus, plus;
2036         struct diff_words_style *style = diff_words->style;
2037
2038         struct diff_options *opt = diff_words->opt;
2039         const char *line_prefix;
2040
2041         assert(opt);
2042         line_prefix = diff_line_prefix(opt);
2043
2044         /* special case: only removal */
2045         if (!diff_words->plus.text.size) {
2046                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2047                                  line_prefix, strlen(line_prefix), 0);
2048                 fn_out_diff_words_write_helper(diff_words->opt,
2049                         &style->old_word, style->newline,
2050                         diff_words->minus.text.size,
2051                         diff_words->minus.text.ptr);
2052                 diff_words->minus.text.size = 0;
2053                 return;
2054         }
2055
2056         diff_words->current_plus = diff_words->plus.text.ptr;
2057         diff_words->last_minus = 0;
2058
2059         memset(&xpp, 0, sizeof(xpp));
2060         memset(&xecfg, 0, sizeof(xecfg));
2061         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2062         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2063         xpp.flags = 0;
2064         /* as only the hunk header will be parsed, we need a 0-context */
2065         xecfg.ctxlen = 0;
2066         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2067                           &xpp, &xecfg))
2068                 die("unable to generate word diff");
2069         free(minus.ptr);
2070         free(plus.ptr);
2071         if (diff_words->current_plus != diff_words->plus.text.ptr +
2072                         diff_words->plus.text.size) {
2073                 if (color_words_output_graph_prefix(diff_words))
2074                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2075                                          line_prefix, strlen(line_prefix), 0);
2076                 fn_out_diff_words_write_helper(diff_words->opt,
2077                         &style->ctx, style->newline,
2078                         diff_words->plus.text.ptr + diff_words->plus.text.size
2079                         - diff_words->current_plus, diff_words->current_plus);
2080         }
2081         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2082 }
2083
2084 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2085 static void diff_words_flush(struct emit_callback *ecbdata)
2086 {
2087         struct diff_options *wo = ecbdata->diff_words->opt;
2088
2089         if (ecbdata->diff_words->minus.text.size ||
2090             ecbdata->diff_words->plus.text.size)
2091                 diff_words_show(ecbdata->diff_words);
2092
2093         if (wo->emitted_symbols) {
2094                 struct diff_options *o = ecbdata->opt;
2095                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2096                 int i;
2097
2098                 /*
2099                  * NEEDSWORK:
2100                  * Instead of appending each, concat all words to a line?
2101                  */
2102                 for (i = 0; i < wol->nr; i++)
2103                         append_emitted_diff_symbol(o, &wol->buf[i]);
2104
2105                 for (i = 0; i < wol->nr; i++)
2106                         free((void *)wol->buf[i].line);
2107
2108                 wol->nr = 0;
2109         }
2110 }
2111
2112 static void diff_filespec_load_driver(struct diff_filespec *one,
2113                                       struct index_state *istate)
2114 {
2115         /* Use already-loaded driver */
2116         if (one->driver)
2117                 return;
2118
2119         if (S_ISREG(one->mode))
2120                 one->driver = userdiff_find_by_path(istate, one->path);
2121
2122         /* Fallback to default settings */
2123         if (!one->driver)
2124                 one->driver = userdiff_find_by_name("default");
2125 }
2126
2127 static const char *userdiff_word_regex(struct diff_filespec *one,
2128                                        struct index_state *istate)
2129 {
2130         diff_filespec_load_driver(one, istate);
2131         return one->driver->word_regex;
2132 }
2133
2134 static void init_diff_words_data(struct emit_callback *ecbdata,
2135                                  struct diff_options *orig_opts,
2136                                  struct diff_filespec *one,
2137                                  struct diff_filespec *two)
2138 {
2139         int i;
2140         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2141         memcpy(o, orig_opts, sizeof(struct diff_options));
2142
2143         ecbdata->diff_words =
2144                 xcalloc(1, sizeof(struct diff_words_data));
2145         ecbdata->diff_words->type = o->word_diff;
2146         ecbdata->diff_words->opt = o;
2147
2148         if (orig_opts->emitted_symbols)
2149                 o->emitted_symbols =
2150                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2151
2152         if (!o->word_regex)
2153                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2154         if (!o->word_regex)
2155                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2156         if (!o->word_regex)
2157                 o->word_regex = diff_word_regex_cfg;
2158         if (o->word_regex) {
2159                 ecbdata->diff_words->word_regex = (regex_t *)
2160                         xmalloc(sizeof(regex_t));
2161                 if (regcomp(ecbdata->diff_words->word_regex,
2162                             o->word_regex,
2163                             REG_EXTENDED | REG_NEWLINE))
2164                         die("invalid regular expression: %s",
2165                             o->word_regex);
2166         }
2167         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2168                 if (o->word_diff == diff_words_styles[i].type) {
2169                         ecbdata->diff_words->style =
2170                                 &diff_words_styles[i];
2171                         break;
2172                 }
2173         }
2174         if (want_color(o->use_color)) {
2175                 struct diff_words_style *st = ecbdata->diff_words->style;
2176                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2177                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2178                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2179         }
2180 }
2181
2182 static void free_diff_words_data(struct emit_callback *ecbdata)
2183 {
2184         if (ecbdata->diff_words) {
2185                 diff_words_flush(ecbdata);
2186                 free (ecbdata->diff_words->opt->emitted_symbols);
2187                 free (ecbdata->diff_words->opt);
2188                 free (ecbdata->diff_words->minus.text.ptr);
2189                 free (ecbdata->diff_words->minus.orig);
2190                 free (ecbdata->diff_words->plus.text.ptr);
2191                 free (ecbdata->diff_words->plus.orig);
2192                 if (ecbdata->diff_words->word_regex) {
2193                         regfree(ecbdata->diff_words->word_regex);
2194                         free(ecbdata->diff_words->word_regex);
2195                 }
2196                 FREE_AND_NULL(ecbdata->diff_words);
2197         }
2198 }
2199
2200 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2201 {
2202         if (want_color(diff_use_color))
2203                 return diff_colors[ix];
2204         return "";
2205 }
2206
2207 const char *diff_line_prefix(struct diff_options *opt)
2208 {
2209         struct strbuf *msgbuf;
2210         if (!opt->output_prefix)
2211                 return "";
2212
2213         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2214         return msgbuf->buf;
2215 }
2216
2217 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2218 {
2219         const char *cp;
2220         unsigned long allot;
2221         size_t l = len;
2222
2223         cp = line;
2224         allot = l;
2225         while (0 < l) {
2226                 (void) utf8_width(&cp, &l);
2227                 if (!cp)
2228                         break; /* truncated in the middle? */
2229         }
2230         return allot - l;
2231 }
2232
2233 static void find_lno(const char *line, struct emit_callback *ecbdata)
2234 {
2235         const char *p;
2236         ecbdata->lno_in_preimage = 0;
2237         ecbdata->lno_in_postimage = 0;
2238         p = strchr(line, '-');
2239         if (!p)
2240                 return; /* cannot happen */
2241         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2242         p = strchr(p, '+');
2243         if (!p)
2244                 return; /* cannot happen */
2245         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2246 }
2247
2248 static void fn_out_consume(void *priv, char *line, unsigned long len)
2249 {
2250         struct emit_callback *ecbdata = priv;
2251         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2252         struct diff_options *o = ecbdata->opt;
2253
2254         o->found_changes = 1;
2255
2256         if (ecbdata->header) {
2257                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2258                                  ecbdata->header->buf, ecbdata->header->len, 0);
2259                 strbuf_reset(ecbdata->header);
2260                 ecbdata->header = NULL;
2261         }
2262
2263         if (ecbdata->label_path[0]) {
2264                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2265                                  ecbdata->label_path[0],
2266                                  strlen(ecbdata->label_path[0]), 0);
2267                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2268                                  ecbdata->label_path[1],
2269                                  strlen(ecbdata->label_path[1]), 0);
2270                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2271         }
2272
2273         if (diff_suppress_blank_empty
2274             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2275                 line[0] = '\n';
2276                 len = 1;
2277         }
2278
2279         if (line[0] == '@') {
2280                 if (ecbdata->diff_words)
2281                         diff_words_flush(ecbdata);
2282                 len = sane_truncate_line(ecbdata, line, len);
2283                 find_lno(line, ecbdata);
2284                 emit_hunk_header(ecbdata, line, len);
2285                 return;
2286         }
2287
2288         if (ecbdata->diff_words) {
2289                 enum diff_symbol s =
2290                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2291                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2292                 if (line[0] == '-') {
2293                         diff_words_append(line, len,
2294                                           &ecbdata->diff_words->minus);
2295                         return;
2296                 } else if (line[0] == '+') {
2297                         diff_words_append(line, len,
2298                                           &ecbdata->diff_words->plus);
2299                         return;
2300                 } else if (starts_with(line, "\\ ")) {
2301                         /*
2302                          * Eat the "no newline at eof" marker as if we
2303                          * saw a "+" or "-" line with nothing on it,
2304                          * and return without diff_words_flush() to
2305                          * defer processing. If this is the end of
2306                          * preimage, more "+" lines may come after it.
2307                          */
2308                         return;
2309                 }
2310                 diff_words_flush(ecbdata);
2311                 emit_diff_symbol(o, s, line, len, 0);
2312                 return;
2313         }
2314
2315         switch (line[0]) {
2316         case '+':
2317                 ecbdata->lno_in_postimage++;
2318                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2319                 break;
2320         case '-':
2321                 ecbdata->lno_in_preimage++;
2322                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2323                 break;
2324         case ' ':
2325                 ecbdata->lno_in_postimage++;
2326                 ecbdata->lno_in_preimage++;
2327                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2328                 break;
2329         default:
2330                 /* incomplete line at the end */
2331                 ecbdata->lno_in_preimage++;
2332                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2333                                  line, len, 0);
2334                 break;
2335         }
2336 }
2337
2338 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2339 {
2340         const char *old_name = a;
2341         const char *new_name = b;
2342         int pfx_length, sfx_length;
2343         int pfx_adjust_for_slash;
2344         int len_a = strlen(a);
2345         int len_b = strlen(b);
2346         int a_midlen, b_midlen;
2347         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2348         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2349
2350         if (qlen_a || qlen_b) {
2351                 quote_c_style(a, name, NULL, 0);
2352                 strbuf_addstr(name, " => ");
2353                 quote_c_style(b, name, NULL, 0);
2354                 return;
2355         }
2356
2357         /* Find common prefix */
2358         pfx_length = 0;
2359         while (*old_name && *new_name && *old_name == *new_name) {
2360                 if (*old_name == '/')
2361                         pfx_length = old_name - a + 1;
2362                 old_name++;
2363                 new_name++;
2364         }
2365
2366         /* Find common suffix */
2367         old_name = a + len_a;
2368         new_name = b + len_b;
2369         sfx_length = 0;
2370         /*
2371          * If there is a common prefix, it must end in a slash.  In
2372          * that case we let this loop run 1 into the prefix to see the
2373          * same slash.
2374          *
2375          * If there is no common prefix, we cannot do this as it would
2376          * underrun the input strings.
2377          */
2378         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2379         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2380                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2381                *old_name == *new_name) {
2382                 if (*old_name == '/')
2383                         sfx_length = len_a - (old_name - a);
2384                 old_name--;
2385                 new_name--;
2386         }
2387
2388         /*
2389          * pfx{mid-a => mid-b}sfx
2390          * {pfx-a => pfx-b}sfx
2391          * pfx{sfx-a => sfx-b}
2392          * name-a => name-b
2393          */
2394         a_midlen = len_a - pfx_length - sfx_length;
2395         b_midlen = len_b - pfx_length - sfx_length;
2396         if (a_midlen < 0)
2397                 a_midlen = 0;
2398         if (b_midlen < 0)
2399                 b_midlen = 0;
2400
2401         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2402         if (pfx_length + sfx_length) {
2403                 strbuf_add(name, a, pfx_length);
2404                 strbuf_addch(name, '{');
2405         }
2406         strbuf_add(name, a + pfx_length, a_midlen);
2407         strbuf_addstr(name, " => ");
2408         strbuf_add(name, b + pfx_length, b_midlen);
2409         if (pfx_length + sfx_length) {
2410                 strbuf_addch(name, '}');
2411                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2412         }
2413 }
2414
2415 struct diffstat_t {
2416         int nr;
2417         int alloc;
2418         struct diffstat_file {
2419                 char *from_name;
2420                 char *name;
2421                 char *print_name;
2422                 const char *comments;
2423                 unsigned is_unmerged:1;
2424                 unsigned is_binary:1;
2425                 unsigned is_renamed:1;
2426                 unsigned is_interesting:1;
2427                 uintmax_t added, deleted;
2428         } **files;
2429 };
2430
2431 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2432                                           const char *name_a,
2433                                           const char *name_b)
2434 {
2435         struct diffstat_file *x;
2436         x = xcalloc(1, sizeof(*x));
2437         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2438         diffstat->files[diffstat->nr++] = x;
2439         if (name_b) {
2440                 x->from_name = xstrdup(name_a);
2441                 x->name = xstrdup(name_b);
2442                 x->is_renamed = 1;
2443         }
2444         else {
2445                 x->from_name = NULL;
2446                 x->name = xstrdup(name_a);
2447         }
2448         return x;
2449 }
2450
2451 static void diffstat_consume(void *priv, char *line, unsigned long len)
2452 {
2453         struct diffstat_t *diffstat = priv;
2454         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2455
2456         if (line[0] == '+')
2457                 x->added++;
2458         else if (line[0] == '-')
2459                 x->deleted++;
2460 }
2461
2462 const char mime_boundary_leader[] = "------------";
2463
2464 static int scale_linear(int it, int width, int max_change)
2465 {
2466         if (!it)
2467                 return 0;
2468         /*
2469          * make sure that at least one '-' or '+' is printed if
2470          * there is any change to this path. The easiest way is to
2471          * scale linearly as if the alloted width is one column shorter
2472          * than it is, and then add 1 to the result.
2473          */
2474         return 1 + (it * (width - 1) / max_change);
2475 }
2476
2477 static void show_graph(struct strbuf *out, char ch, int cnt,
2478                        const char *set, const char *reset)
2479 {
2480         if (cnt <= 0)
2481                 return;
2482         strbuf_addstr(out, set);
2483         strbuf_addchars(out, ch, cnt);
2484         strbuf_addstr(out, reset);
2485 }
2486
2487 static void fill_print_name(struct diffstat_file *file)
2488 {
2489         struct strbuf pname = STRBUF_INIT;
2490
2491         if (file->print_name)
2492                 return;
2493
2494         if (file->is_renamed)
2495                 pprint_rename(&pname, file->from_name, file->name);
2496         else
2497                 quote_c_style(file->name, &pname, NULL, 0);
2498
2499         if (file->comments)
2500                 strbuf_addf(&pname, " (%s)", file->comments);
2501
2502         file->print_name = strbuf_detach(&pname, NULL);
2503 }
2504
2505 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2506                 int files, int insertions, int deletions)
2507 {
2508         struct strbuf sb = STRBUF_INIT;
2509
2510         if (!files) {
2511                 assert(insertions == 0 && deletions == 0);
2512                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2513                                  NULL, 0, 0);
2514                 return;
2515         }
2516
2517         strbuf_addf(&sb,
2518                     (files == 1) ? " %d file changed" : " %d files changed",
2519                     files);
2520
2521         /*
2522          * For binary diff, the caller may want to print "x files
2523          * changed" with insertions == 0 && deletions == 0.
2524          *
2525          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2526          * is probably less confusing (i.e skip over "2 files changed
2527          * but nothing about added/removed lines? Is this a bug in Git?").
2528          */
2529         if (insertions || deletions == 0) {
2530                 strbuf_addf(&sb,
2531                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2532                             insertions);
2533         }
2534
2535         if (deletions || insertions == 0) {
2536                 strbuf_addf(&sb,
2537                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2538                             deletions);
2539         }
2540         strbuf_addch(&sb, '\n');
2541         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2542                          sb.buf, sb.len, 0);
2543         strbuf_release(&sb);
2544 }
2545
2546 void print_stat_summary(FILE *fp, int files,
2547                         int insertions, int deletions)
2548 {
2549         struct diff_options o;
2550         memset(&o, 0, sizeof(o));
2551         o.file = fp;
2552
2553         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2554 }
2555
2556 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2557 {
2558         int i, len, add, del, adds = 0, dels = 0;
2559         uintmax_t max_change = 0, max_len = 0;
2560         int total_files = data->nr, count;
2561         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2562         const char *reset, *add_c, *del_c;
2563         int extra_shown = 0;
2564         const char *line_prefix = diff_line_prefix(options);
2565         struct strbuf out = STRBUF_INIT;
2566
2567         if (data->nr == 0)
2568                 return;
2569
2570         count = options->stat_count ? options->stat_count : data->nr;
2571
2572         reset = diff_get_color_opt(options, DIFF_RESET);
2573         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2574         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2575
2576         /*
2577          * Find the longest filename and max number of changes
2578          */
2579         for (i = 0; (i < count) && (i < data->nr); i++) {
2580                 struct diffstat_file *file = data->files[i];
2581                 uintmax_t change = file->added + file->deleted;
2582
2583                 if (!file->is_interesting && (change == 0)) {
2584                         count++; /* not shown == room for one more */
2585                         continue;
2586                 }
2587                 fill_print_name(file);
2588                 len = strlen(file->print_name);
2589                 if (max_len < len)
2590                         max_len = len;
2591
2592                 if (file->is_unmerged) {
2593                         /* "Unmerged" is 8 characters */
2594                         bin_width = bin_width < 8 ? 8 : bin_width;
2595                         continue;
2596                 }
2597                 if (file->is_binary) {
2598                         /* "Bin XXX -> YYY bytes" */
2599                         int w = 14 + decimal_width(file->added)
2600                                 + decimal_width(file->deleted);
2601                         bin_width = bin_width < w ? w : bin_width;
2602                         /* Display change counts aligned with "Bin" */
2603                         number_width = 3;
2604                         continue;
2605                 }
2606
2607                 if (max_change < change)
2608                         max_change = change;
2609         }
2610         count = i; /* where we can stop scanning in data->files[] */
2611
2612         /*
2613          * We have width = stat_width or term_columns() columns total.
2614          * We want a maximum of min(max_len, stat_name_width) for the name part.
2615          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2616          * We also need 1 for " " and 4 + decimal_width(max_change)
2617          * for " | NNNN " and one the empty column at the end, altogether
2618          * 6 + decimal_width(max_change).
2619          *
2620          * If there's not enough space, we will use the smaller of
2621          * stat_name_width (if set) and 5/8*width for the filename,
2622          * and the rest for constant elements + graph part, but no more
2623          * than stat_graph_width for the graph part.
2624          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2625          * for the standard terminal size).
2626          *
2627          * In other words: stat_width limits the maximum width, and
2628          * stat_name_width fixes the maximum width of the filename,
2629          * and is also used to divide available columns if there
2630          * aren't enough.
2631          *
2632          * Binary files are displayed with "Bin XXX -> YYY bytes"
2633          * instead of the change count and graph. This part is treated
2634          * similarly to the graph part, except that it is not
2635          * "scaled". If total width is too small to accommodate the
2636          * guaranteed minimum width of the filename part and the
2637          * separators and this message, this message will "overflow"
2638          * making the line longer than the maximum width.
2639          */
2640
2641         if (options->stat_width == -1)
2642                 width = term_columns() - strlen(line_prefix);
2643         else
2644                 width = options->stat_width ? options->stat_width : 80;
2645         number_width = decimal_width(max_change) > number_width ?
2646                 decimal_width(max_change) : number_width;
2647
2648         if (options->stat_graph_width == -1)
2649                 options->stat_graph_width = diff_stat_graph_width;
2650
2651         /*
2652          * Guarantee 3/8*16==6 for the graph part
2653          * and 5/8*16==10 for the filename part
2654          */
2655         if (width < 16 + 6 + number_width)
2656                 width = 16 + 6 + number_width;
2657
2658         /*
2659          * First assign sizes that are wanted, ignoring available width.
2660          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2661          * starting from "XXX" should fit in graph_width.
2662          */
2663         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2664         if (options->stat_graph_width &&
2665             options->stat_graph_width < graph_width)
2666                 graph_width = options->stat_graph_width;
2667
2668         name_width = (options->stat_name_width > 0 &&
2669                       options->stat_name_width < max_len) ?
2670                 options->stat_name_width : max_len;
2671
2672         /*
2673          * Adjust adjustable widths not to exceed maximum width
2674          */
2675         if (name_width + number_width + 6 + graph_width > width) {
2676                 if (graph_width > width * 3/8 - number_width - 6) {
2677                         graph_width = width * 3/8 - number_width - 6;
2678                         if (graph_width < 6)
2679                                 graph_width = 6;
2680                 }
2681
2682                 if (options->stat_graph_width &&
2683                     graph_width > options->stat_graph_width)
2684                         graph_width = options->stat_graph_width;
2685                 if (name_width > width - number_width - 6 - graph_width)
2686                         name_width = width - number_width - 6 - graph_width;
2687                 else
2688                         graph_width = width - number_width - 6 - name_width;
2689         }
2690
2691         /*
2692          * From here name_width is the width of the name area,
2693          * and graph_width is the width of the graph area.
2694          * max_change is used to scale graph properly.
2695          */
2696         for (i = 0; i < count; i++) {
2697                 const char *prefix = "";
2698                 struct diffstat_file *file = data->files[i];
2699                 char *name = file->print_name;
2700                 uintmax_t added = file->added;
2701                 uintmax_t deleted = file->deleted;
2702                 int name_len;
2703
2704                 if (!file->is_interesting && (added + deleted == 0))
2705                         continue;
2706
2707                 /*
2708                  * "scale" the filename
2709                  */
2710                 len = name_width;
2711                 name_len = strlen(name);
2712                 if (name_width < name_len) {
2713                         char *slash;
2714                         prefix = "...";
2715                         len -= 3;
2716                         name += name_len - len;
2717                         slash = strchr(name, '/');
2718                         if (slash)
2719                                 name = slash;
2720                 }
2721
2722                 if (file->is_binary) {
2723                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2724                         strbuf_addf(&out, " %*s", number_width, "Bin");
2725                         if (!added && !deleted) {
2726                                 strbuf_addch(&out, '\n');
2727                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2728                                                  out.buf, out.len, 0);
2729                                 strbuf_reset(&out);
2730                                 continue;
2731                         }
2732                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2733                                 del_c, deleted, reset);
2734                         strbuf_addstr(&out, " -> ");
2735                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2736                                 add_c, added, reset);
2737                         strbuf_addstr(&out, " bytes\n");
2738                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2739                                          out.buf, out.len, 0);
2740                         strbuf_reset(&out);
2741                         continue;
2742                 }
2743                 else if (file->is_unmerged) {
2744                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2745                         strbuf_addstr(&out, " Unmerged\n");
2746                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2747                                          out.buf, out.len, 0);
2748                         strbuf_reset(&out);
2749                         continue;
2750                 }
2751
2752                 /*
2753                  * scale the add/delete
2754                  */
2755                 add = added;
2756                 del = deleted;
2757
2758                 if (graph_width <= max_change) {
2759                         int total = scale_linear(add + del, graph_width, max_change);
2760                         if (total < 2 && add && del)
2761                                 /* width >= 2 due to the sanity check */
2762                                 total = 2;
2763                         if (add < del) {
2764                                 add = scale_linear(add, graph_width, max_change);
2765                                 del = total - add;
2766                         } else {
2767                                 del = scale_linear(del, graph_width, max_change);
2768                                 add = total - del;
2769                         }
2770                 }
2771                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2772                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2773                         number_width, added + deleted,
2774                         added + deleted ? " " : "");
2775                 show_graph(&out, '+', add, add_c, reset);
2776                 show_graph(&out, '-', del, del_c, reset);
2777                 strbuf_addch(&out, '\n');
2778                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2779                                  out.buf, out.len, 0);
2780                 strbuf_reset(&out);
2781         }
2782
2783         for (i = 0; i < data->nr; i++) {
2784                 struct diffstat_file *file = data->files[i];
2785                 uintmax_t added = file->added;
2786                 uintmax_t deleted = file->deleted;
2787
2788                 if (file->is_unmerged ||
2789                     (!file->is_interesting && (added + deleted == 0))) {
2790                         total_files--;
2791                         continue;
2792                 }
2793
2794                 if (!file->is_binary) {
2795                         adds += added;
2796                         dels += deleted;
2797                 }
2798                 if (i < count)
2799                         continue;
2800                 if (!extra_shown)
2801                         emit_diff_symbol(options,
2802                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2803                                          NULL, 0, 0);
2804                 extra_shown = 1;
2805         }
2806
2807         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2808         strbuf_release(&out);
2809 }
2810
2811 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2812 {
2813         int i, adds = 0, dels = 0, total_files = data->nr;
2814
2815         if (data->nr == 0)
2816                 return;
2817
2818         for (i = 0; i < data->nr; i++) {
2819                 int added = data->files[i]->added;
2820                 int deleted = data->files[i]->deleted;
2821
2822                 if (data->files[i]->is_unmerged ||
2823                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2824                         total_files--;
2825                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2826                         adds += added;
2827                         dels += deleted;
2828                 }
2829         }
2830         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2831 }
2832
2833 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2834 {
2835         int i;
2836
2837         if (data->nr == 0)
2838                 return;
2839
2840         for (i = 0; i < data->nr; i++) {
2841                 struct diffstat_file *file = data->files[i];
2842
2843                 fprintf(options->file, "%s", diff_line_prefix(options));
2844
2845                 if (file->is_binary)
2846                         fprintf(options->file, "-\t-\t");
2847                 else
2848                         fprintf(options->file,
2849                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2850                                 file->added, file->deleted);
2851                 if (options->line_termination) {
2852                         fill_print_name(file);
2853                         if (!file->is_renamed)
2854                                 write_name_quoted(file->name, options->file,
2855                                                   options->line_termination);
2856                         else {
2857                                 fputs(file->print_name, options->file);
2858                                 putc(options->line_termination, options->file);
2859                         }
2860                 } else {
2861                         if (file->is_renamed) {
2862                                 putc('\0', options->file);
2863                                 write_name_quoted(file->from_name, options->file, '\0');
2864                         }
2865                         write_name_quoted(file->name, options->file, '\0');
2866                 }
2867         }
2868 }
2869
2870 struct dirstat_file {
2871         const char *name;
2872         unsigned long changed;
2873 };
2874
2875 struct dirstat_dir {
2876         struct dirstat_file *files;
2877         int alloc, nr, permille, cumulative;
2878 };
2879
2880 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2881                 unsigned long changed, const char *base, int baselen)
2882 {
2883         unsigned long sum_changes = 0;
2884         unsigned int sources = 0;
2885         const char *line_prefix = diff_line_prefix(opt);
2886
2887         while (dir->nr) {
2888                 struct dirstat_file *f = dir->files;
2889                 int namelen = strlen(f->name);
2890                 unsigned long changes;
2891                 char *slash;
2892
2893                 if (namelen < baselen)
2894                         break;
2895                 if (memcmp(f->name, base, baselen))
2896                         break;
2897                 slash = strchr(f->name + baselen, '/');
2898                 if (slash) {
2899                         int newbaselen = slash + 1 - f->name;
2900                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2901                         sources++;
2902                 } else {
2903                         changes = f->changed;
2904                         dir->files++;
2905                         dir->nr--;
2906                         sources += 2;
2907                 }
2908                 sum_changes += changes;
2909         }
2910
2911         /*
2912          * We don't report dirstat's for
2913          *  - the top level
2914          *  - or cases where everything came from a single directory
2915          *    under this directory (sources == 1).
2916          */
2917         if (baselen && sources != 1) {
2918                 if (sum_changes) {
2919                         int permille = sum_changes * 1000 / changed;
2920                         if (permille >= dir->permille) {
2921                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2922                                         permille / 10, permille % 10, baselen, base);
2923                                 if (!dir->cumulative)
2924                                         return 0;
2925                         }
2926                 }
2927         }
2928         return sum_changes;
2929 }
2930
2931 static int dirstat_compare(const void *_a, const void *_b)
2932 {
2933         const struct dirstat_file *a = _a;
2934         const struct dirstat_file *b = _b;
2935         return strcmp(a->name, b->name);
2936 }
2937
2938 static void show_dirstat(struct diff_options *options)
2939 {
2940         int i;
2941         unsigned long changed;
2942         struct dirstat_dir dir;
2943         struct diff_queue_struct *q = &diff_queued_diff;
2944
2945         dir.files = NULL;
2946         dir.alloc = 0;
2947         dir.nr = 0;
2948         dir.permille = options->dirstat_permille;
2949         dir.cumulative = options->flags.dirstat_cumulative;
2950
2951         changed = 0;
2952         for (i = 0; i < q->nr; i++) {
2953                 struct diff_filepair *p = q->queue[i];
2954                 const char *name;
2955                 unsigned long copied, added, damage;
2956
2957                 name = p->two->path ? p->two->path : p->one->path;
2958
2959                 if (p->one->oid_valid && p->two->oid_valid &&
2960                     oideq(&p->one->oid, &p->two->oid)) {
2961                         /*
2962                          * The SHA1 has not changed, so pre-/post-content is
2963                          * identical. We can therefore skip looking at the
2964                          * file contents altogether.
2965                          */
2966                         damage = 0;
2967                         goto found_damage;
2968                 }
2969
2970                 if (options->flags.dirstat_by_file) {
2971                         /*
2972                          * In --dirstat-by-file mode, we don't really need to
2973                          * look at the actual file contents at all.
2974                          * The fact that the SHA1 changed is enough for us to
2975                          * add this file to the list of results
2976                          * (with each file contributing equal damage).
2977                          */
2978                         damage = 1;
2979                         goto found_damage;
2980                 }
2981
2982                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2983                         diff_populate_filespec(options->repo, p->one, 0);
2984                         diff_populate_filespec(options->repo, p->two, 0);
2985                         diffcore_count_changes(options->repo,
2986                                                p->one, p->two, NULL, NULL,
2987                                                &copied, &added);
2988                         diff_free_filespec_data(p->one);
2989                         diff_free_filespec_data(p->two);
2990                 } else if (DIFF_FILE_VALID(p->one)) {
2991                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
2992                         copied = added = 0;
2993                         diff_free_filespec_data(p->one);
2994                 } else if (DIFF_FILE_VALID(p->two)) {
2995                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
2996                         copied = 0;
2997                         added = p->two->size;
2998                         diff_free_filespec_data(p->two);
2999                 } else
3000                         continue;
3001
3002                 /*
3003                  * Original minus copied is the removed material,
3004                  * added is the new material.  They are both damages
3005                  * made to the preimage.
3006                  * If the resulting damage is zero, we know that
3007                  * diffcore_count_changes() considers the two entries to
3008                  * be identical, but since the oid changed, we
3009                  * know that there must have been _some_ kind of change,
3010                  * so we force all entries to have damage > 0.
3011                  */
3012                 damage = (p->one->size - copied) + added;
3013                 if (!damage)
3014                         damage = 1;
3015
3016 found_damage:
3017                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3018                 dir.files[dir.nr].name = name;
3019                 dir.files[dir.nr].changed = damage;
3020                 changed += damage;
3021                 dir.nr++;
3022         }
3023
3024         /* This can happen even with many files, if everything was renames */
3025         if (!changed)
3026                 return;
3027
3028         /* Show all directories with more than x% of the changes */
3029         QSORT(dir.files, dir.nr, dirstat_compare);
3030         gather_dirstat(options, &dir, changed, "", 0);
3031 }
3032
3033 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3034 {
3035         int i;
3036         unsigned long changed;
3037         struct dirstat_dir dir;
3038
3039         if (data->nr == 0)
3040                 return;
3041
3042         dir.files = NULL;
3043         dir.alloc = 0;
3044         dir.nr = 0;
3045         dir.permille = options->dirstat_permille;
3046         dir.cumulative = options->flags.dirstat_cumulative;
3047
3048         changed = 0;
3049         for (i = 0; i < data->nr; i++) {
3050                 struct diffstat_file *file = data->files[i];
3051                 unsigned long damage = file->added + file->deleted;
3052                 if (file->is_binary)
3053                         /*
3054                          * binary files counts bytes, not lines. Must find some
3055                          * way to normalize binary bytes vs. textual lines.
3056                          * The following heuristic assumes that there are 64
3057                          * bytes per "line".
3058                          * This is stupid and ugly, but very cheap...
3059                          */
3060                         damage = DIV_ROUND_UP(damage, 64);
3061                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3062                 dir.files[dir.nr].name = file->name;
3063                 dir.files[dir.nr].changed = damage;
3064                 changed += damage;
3065                 dir.nr++;
3066         }
3067
3068         /* This can happen even with many files, if everything was renames */
3069         if (!changed)
3070                 return;
3071
3072         /* Show all directories with more than x% of the changes */
3073         QSORT(dir.files, dir.nr, dirstat_compare);
3074         gather_dirstat(options, &dir, changed, "", 0);
3075 }
3076
3077 static void free_diffstat_info(struct diffstat_t *diffstat)
3078 {
3079         int i;
3080         for (i = 0; i < diffstat->nr; i++) {
3081                 struct diffstat_file *f = diffstat->files[i];
3082                 free(f->print_name);
3083                 free(f->name);
3084                 free(f->from_name);
3085                 free(f);
3086         }
3087         free(diffstat->files);
3088 }
3089
3090 struct checkdiff_t {
3091         const char *filename;
3092         int lineno;
3093         int conflict_marker_size;
3094         struct diff_options *o;
3095         unsigned ws_rule;
3096         unsigned status;
3097 };
3098
3099 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3100 {
3101         char firstchar;
3102         int cnt;
3103
3104         if (len < marker_size + 1)
3105                 return 0;
3106         firstchar = line[0];
3107         switch (firstchar) {
3108         case '=': case '>': case '<': case '|':
3109                 break;
3110         default:
3111                 return 0;
3112         }
3113         for (cnt = 1; cnt < marker_size; cnt++)
3114                 if (line[cnt] != firstchar)
3115                         return 0;
3116         /* line[1] thru line[marker_size-1] are same as firstchar */
3117         if (len < marker_size + 1 || !isspace(line[marker_size]))
3118                 return 0;
3119         return 1;
3120 }
3121
3122 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3123 {
3124         struct checkdiff_t *data = priv;
3125         int marker_size = data->conflict_marker_size;
3126         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3127         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3128         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3129         char *err;
3130         const char *line_prefix;
3131
3132         assert(data->o);
3133         line_prefix = diff_line_prefix(data->o);
3134
3135         if (line[0] == '+') {
3136                 unsigned bad;
3137                 data->lineno++;
3138                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3139                         data->status |= 1;
3140                         fprintf(data->o->file,
3141                                 "%s%s:%d: leftover conflict marker\n",
3142                                 line_prefix, data->filename, data->lineno);
3143                 }
3144                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3145                 if (!bad)
3146                         return;
3147                 data->status |= bad;
3148                 err = whitespace_error_string(bad);
3149                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3150                         line_prefix, data->filename, data->lineno, err);
3151                 free(err);
3152                 emit_line(data->o, set, reset, line, 1);
3153                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3154                               data->o->file, set, reset, ws);
3155         } else if (line[0] == ' ') {
3156                 data->lineno++;
3157         } else if (line[0] == '@') {
3158                 char *plus = strchr(line, '+');
3159                 if (plus)
3160                         data->lineno = strtol(plus, NULL, 10) - 1;
3161                 else
3162                         die("invalid diff");
3163         }
3164 }
3165
3166 static unsigned char *deflate_it(char *data,
3167                                  unsigned long size,
3168                                  unsigned long *result_size)
3169 {
3170         int bound;
3171         unsigned char *deflated;
3172         git_zstream stream;
3173
3174         git_deflate_init(&stream, zlib_compression_level);
3175         bound = git_deflate_bound(&stream, size);
3176         deflated = xmalloc(bound);
3177         stream.next_out = deflated;
3178         stream.avail_out = bound;
3179
3180         stream.next_in = (unsigned char *)data;
3181         stream.avail_in = size;
3182         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3183                 ; /* nothing */
3184         git_deflate_end(&stream);
3185         *result_size = stream.total_out;
3186         return deflated;
3187 }
3188
3189 static void emit_binary_diff_body(struct diff_options *o,
3190                                   mmfile_t *one, mmfile_t *two)
3191 {
3192         void *cp;
3193         void *delta;
3194         void *deflated;
3195         void *data;
3196         unsigned long orig_size;
3197         unsigned long delta_size;
3198         unsigned long deflate_size;
3199         unsigned long data_size;
3200
3201         /* We could do deflated delta, or we could do just deflated two,
3202          * whichever is smaller.
3203          */
3204         delta = NULL;
3205         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3206         if (one->size && two->size) {
3207                 delta = diff_delta(one->ptr, one->size,
3208                                    two->ptr, two->size,
3209                                    &delta_size, deflate_size);
3210                 if (delta) {
3211                         void *to_free = delta;
3212                         orig_size = delta_size;
3213                         delta = deflate_it(delta, delta_size, &delta_size);
3214                         free(to_free);
3215                 }
3216         }
3217
3218         if (delta && delta_size < deflate_size) {
3219                 char *s = xstrfmt("%lu", orig_size);
3220                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3221                                  s, strlen(s), 0);
3222                 free(s);
3223                 free(deflated);
3224                 data = delta;
3225                 data_size = delta_size;
3226         } else {
3227                 char *s = xstrfmt("%lu", two->size);
3228                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3229                                  s, strlen(s), 0);
3230                 free(s);
3231                 free(delta);
3232                 data = deflated;
3233                 data_size = deflate_size;
3234         }
3235
3236         /* emit data encoded in base85 */
3237         cp = data;
3238         while (data_size) {
3239                 int len;
3240                 int bytes = (52 < data_size) ? 52 : data_size;
3241                 char line[71];
3242                 data_size -= bytes;
3243                 if (bytes <= 26)
3244                         line[0] = bytes + 'A' - 1;
3245                 else
3246                         line[0] = bytes - 26 + 'a' - 1;
3247                 encode_85(line + 1, cp, bytes);
3248                 cp = (char *) cp + bytes;
3249
3250                 len = strlen(line);
3251                 line[len++] = '\n';
3252                 line[len] = '\0';
3253
3254                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3255                                  line, len, 0);
3256         }
3257         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3258         free(data);
3259 }
3260
3261 static void emit_binary_diff(struct diff_options *o,
3262                              mmfile_t *one, mmfile_t *two)
3263 {
3264         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3265         emit_binary_diff_body(o, one, two);
3266         emit_binary_diff_body(o, two, one);
3267 }
3268
3269 int diff_filespec_is_binary(struct repository *r,
3270                             struct diff_filespec *one)
3271 {
3272         if (one->is_binary == -1) {
3273                 diff_filespec_load_driver(one, r->index);
3274                 if (one->driver->binary != -1)
3275                         one->is_binary = one->driver->binary;
3276                 else {
3277                         if (!one->data && DIFF_FILE_VALID(one))
3278                                 diff_populate_filespec(r, one, CHECK_BINARY);
3279                         if (one->is_binary == -1 && one->data)
3280                                 one->is_binary = buffer_is_binary(one->data,
3281                                                 one->size);
3282                         if (one->is_binary == -1)
3283                                 one->is_binary = 0;
3284                 }
3285         }
3286         return one->is_binary;
3287 }
3288
3289 static const struct userdiff_funcname *
3290 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3291 {
3292         diff_filespec_load_driver(one, o->repo->index);
3293         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3294 }
3295
3296 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3297 {
3298         if (!options->a_prefix)
3299                 options->a_prefix = a;
3300         if (!options->b_prefix)
3301                 options->b_prefix = b;
3302 }
3303
3304 struct userdiff_driver *get_textconv(struct index_state *istate,
3305                                      struct diff_filespec *one)
3306 {
3307         if (!DIFF_FILE_VALID(one))
3308                 return NULL;
3309
3310         diff_filespec_load_driver(one, istate);
3311         return userdiff_get_textconv(one->driver);
3312 }
3313
3314 static void builtin_diff(const char *name_a,
3315                          const char *name_b,
3316                          struct diff_filespec *one,
3317                          struct diff_filespec *two,
3318                          const char *xfrm_msg,
3319                          int must_show_header,
3320                          struct diff_options *o,
3321                          int complete_rewrite)
3322 {
3323         mmfile_t mf1, mf2;
3324         const char *lbl[2];
3325         char *a_one, *b_two;
3326         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3327         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3328         const char *a_prefix, *b_prefix;
3329         struct userdiff_driver *textconv_one = NULL;
3330         struct userdiff_driver *textconv_two = NULL;
3331         struct strbuf header = STRBUF_INIT;
3332         const char *line_prefix = diff_line_prefix(o);
3333
3334         diff_set_mnemonic_prefix(o, "a/", "b/");
3335         if (o->flags.reverse_diff) {
3336                 a_prefix = o->b_prefix;
3337                 b_prefix = o->a_prefix;
3338         } else {
3339                 a_prefix = o->a_prefix;
3340                 b_prefix = o->b_prefix;
3341         }
3342
3343         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3344             (!one->mode || S_ISGITLINK(one->mode)) &&
3345             (!two->mode || S_ISGITLINK(two->mode))) {
3346                 show_submodule_summary(o, one->path ? one->path : two->path,
3347                                 &one->oid, &two->oid,
3348                                 two->dirty_submodule);
3349                 return;
3350         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3351                    (!one->mode || S_ISGITLINK(one->mode)) &&
3352                    (!two->mode || S_ISGITLINK(two->mode))) {
3353                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3354                                 &one->oid, &two->oid,
3355                                 two->dirty_submodule);
3356                 return;
3357         }
3358
3359         if (o->flags.allow_textconv) {
3360                 textconv_one = get_textconv(o->repo->index, one);
3361                 textconv_two = get_textconv(o->repo->index, two);
3362         }
3363
3364         /* Never use a non-valid filename anywhere if at all possible */
3365         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3366         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3367
3368         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3369         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3370         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3371         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3372         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3373         if (lbl[0][0] == '/') {
3374                 /* /dev/null */
3375                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3376                 if (xfrm_msg)
3377                         strbuf_addstr(&header, xfrm_msg);
3378                 must_show_header = 1;
3379         }
3380         else if (lbl[1][0] == '/') {
3381                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3382                 if (xfrm_msg)
3383                         strbuf_addstr(&header, xfrm_msg);
3384                 must_show_header = 1;
3385         }
3386         else {
3387                 if (one->mode != two->mode) {
3388                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3389                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3390                         must_show_header = 1;
3391                 }
3392                 if (xfrm_msg)
3393                         strbuf_addstr(&header, xfrm_msg);
3394
3395                 /*
3396                  * we do not run diff between different kind
3397                  * of objects.
3398                  */
3399                 if ((one->mode ^ two->mode) & S_IFMT)
3400                         goto free_ab_and_return;
3401                 if (complete_rewrite &&
3402                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3403                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3404                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3405                                          header.buf, header.len, 0);
3406                         strbuf_reset(&header);
3407                         emit_rewrite_diff(name_a, name_b, one, two,
3408                                           textconv_one, textconv_two, o);
3409                         o->found_changes = 1;
3410                         goto free_ab_and_return;
3411                 }
3412         }
3413
3414         if (o->irreversible_delete && lbl[1][0] == '/') {
3415                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3416                                  header.len, 0);
3417                 strbuf_reset(&header);
3418                 goto free_ab_and_return;
3419         } else if (!o->flags.text &&
3420                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3421                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3422                 struct strbuf sb = STRBUF_INIT;
3423                 if (!one->data && !two->data &&
3424                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3425                     !o->flags.binary) {
3426                         if (oideq(&one->oid, &two->oid)) {
3427                                 if (must_show_header)
3428                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3429                                                          header.buf, header.len,
3430                                                          0);
3431                                 goto free_ab_and_return;
3432                         }
3433                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3434                                          header.buf, header.len, 0);
3435                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3436                                     diff_line_prefix(o), lbl[0], lbl[1]);
3437                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3438                                          sb.buf, sb.len, 0);
3439                         strbuf_release(&sb);
3440                         goto free_ab_and_return;
3441                 }
3442                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3443                     fill_mmfile(o->repo, &mf2, two) < 0)
3444                         die("unable to read files to diff");
3445                 /* Quite common confusing case */
3446                 if (mf1.size == mf2.size &&
3447                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3448                         if (must_show_header)
3449                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3450                                                  header.buf, header.len, 0);
3451                         goto free_ab_and_return;
3452                 }
3453                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3454                 strbuf_reset(&header);
3455                 if (o->flags.binary)
3456                         emit_binary_diff(o, &mf1, &mf2);
3457                 else {
3458                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3459                                     diff_line_prefix(o), lbl[0], lbl[1]);
3460                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3461                                          sb.buf, sb.len, 0);
3462                         strbuf_release(&sb);
3463                 }
3464                 o->found_changes = 1;
3465         } else {
3466                 /* Crazy xdl interfaces.. */
3467                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3468                 const char *v;
3469                 xpparam_t xpp;
3470                 xdemitconf_t xecfg;
3471                 struct emit_callback ecbdata;
3472                 const struct userdiff_funcname *pe;
3473
3474                 if (must_show_header) {
3475                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3476                                          header.buf, header.len, 0);
3477                         strbuf_reset(&header);
3478                 }
3479
3480                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3481                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3482
3483                 pe = diff_funcname_pattern(o, one);
3484                 if (!pe)
3485                         pe = diff_funcname_pattern(o, two);
3486
3487                 memset(&xpp, 0, sizeof(xpp));
3488                 memset(&xecfg, 0, sizeof(xecfg));
3489                 memset(&ecbdata, 0, sizeof(ecbdata));
3490                 if (o->flags.suppress_diff_headers)
3491                         lbl[0] = NULL;
3492                 ecbdata.label_path = lbl;
3493                 ecbdata.color_diff = want_color(o->use_color);
3494                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3495                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3496                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3497                 ecbdata.opt = o;
3498                 if (header.len && !o->flags.suppress_diff_headers)
3499                         ecbdata.header = &header;
3500                 xpp.flags = o->xdl_opts;
3501                 xpp.anchors = o->anchors;
3502                 xpp.anchors_nr = o->anchors_nr;
3503                 xecfg.ctxlen = o->context;
3504                 xecfg.interhunkctxlen = o->interhunkcontext;
3505                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3506                 if (o->flags.funccontext)
3507                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3508                 if (pe)
3509                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3510                 if (!diffopts)
3511                         ;
3512                 else if (skip_prefix(diffopts, "--unified=", &v))
3513                         xecfg.ctxlen = strtoul(v, NULL, 10);
3514                 else if (skip_prefix(diffopts, "-u", &v))
3515                         xecfg.ctxlen = strtoul(v, NULL, 10);
3516                 if (o->word_diff)
3517                         init_diff_words_data(&ecbdata, o, one, two);
3518                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3519                                   &xpp, &xecfg))
3520                         die("unable to generate diff for %s", one->path);
3521                 if (o->word_diff)
3522                         free_diff_words_data(&ecbdata);
3523                 if (textconv_one)
3524                         free(mf1.ptr);
3525                 if (textconv_two)
3526                         free(mf2.ptr);
3527                 xdiff_clear_find_func(&xecfg);
3528         }
3529
3530  free_ab_and_return:
3531         strbuf_release(&header);
3532         diff_free_filespec_data(one);
3533         diff_free_filespec_data(two);
3534         free(a_one);
3535         free(b_two);
3536         return;
3537 }
3538
3539 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3540 {
3541         if (!is_renamed) {
3542                 if (p->status == DIFF_STATUS_ADDED) {
3543                         if (S_ISLNK(p->two->mode))
3544                                 return "new +l";
3545                         else if ((p->two->mode & 0777) == 0755)
3546                                 return "new +x";
3547                         else
3548                                 return "new";
3549                 } else if (p->status == DIFF_STATUS_DELETED)
3550                         return "gone";
3551         }
3552         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3553                 return "mode -l";
3554         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3555                 return "mode +l";
3556         else if ((p->one->mode & 0777) == 0644 &&
3557                  (p->two->mode & 0777) == 0755)
3558                 return "mode +x";
3559         else if ((p->one->mode & 0777) == 0755 &&
3560                  (p->two->mode & 0777) == 0644)
3561                 return "mode -x";
3562         return NULL;
3563 }
3564
3565 static void builtin_diffstat(const char *name_a, const char *name_b,
3566                              struct diff_filespec *one,
3567                              struct diff_filespec *two,
3568                              struct diffstat_t *diffstat,
3569                              struct diff_options *o,
3570                              struct diff_filepair *p)
3571 {
3572         mmfile_t mf1, mf2;
3573         struct diffstat_file *data;
3574         int same_contents;
3575         int complete_rewrite = 0;
3576
3577         if (!DIFF_PAIR_UNMERGED(p)) {
3578                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3579                         complete_rewrite = 1;
3580         }
3581
3582         data = diffstat_add(diffstat, name_a, name_b);
3583         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3584         if (o->flags.stat_with_summary)
3585                 data->comments = get_compact_summary(p, data->is_renamed);
3586
3587         if (!one || !two) {
3588                 data->is_unmerged = 1;
3589                 return;
3590         }
3591
3592         same_contents = oideq(&one->oid, &two->oid);
3593
3594         if (diff_filespec_is_binary(o->repo, one) ||
3595             diff_filespec_is_binary(o->repo, two)) {
3596                 data->is_binary = 1;
3597                 if (same_contents) {
3598                         data->added = 0;
3599                         data->deleted = 0;
3600                 } else {
3601                         data->added = diff_filespec_size(o->repo, two);
3602                         data->deleted = diff_filespec_size(o->repo, one);
3603                 }
3604         }
3605
3606         else if (complete_rewrite) {
3607                 diff_populate_filespec(o->repo, one, 0);
3608                 diff_populate_filespec(o->repo, two, 0);
3609                 data->deleted = count_lines(one->data, one->size);
3610                 data->added = count_lines(two->data, two->size);
3611         }
3612
3613         else if (!same_contents) {
3614                 /* Crazy xdl interfaces.. */
3615                 xpparam_t xpp;
3616                 xdemitconf_t xecfg;
3617
3618                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3619                     fill_mmfile(o->repo, &mf2, two) < 0)
3620                         die("unable to read files to diff");
3621
3622                 memset(&xpp, 0, sizeof(xpp));
3623                 memset(&xecfg, 0, sizeof(xecfg));
3624                 xpp.flags = o->xdl_opts;
3625                 xpp.anchors = o->anchors;
3626                 xpp.anchors_nr = o->anchors_nr;
3627                 xecfg.ctxlen = o->context;
3628                 xecfg.interhunkctxlen = o->interhunkcontext;
3629                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3630                                   &xpp, &xecfg))
3631                         die("unable to generate diffstat for %s", one->path);
3632         }
3633
3634         diff_free_filespec_data(one);
3635         diff_free_filespec_data(two);
3636 }
3637
3638 static void builtin_checkdiff(const char *name_a, const char *name_b,
3639                               const char *attr_path,
3640                               struct diff_filespec *one,
3641                               struct diff_filespec *two,
3642                               struct diff_options *o)
3643 {
3644         mmfile_t mf1, mf2;
3645         struct checkdiff_t data;
3646
3647         if (!two)
3648                 return;
3649
3650         memset(&data, 0, sizeof(data));
3651         data.filename = name_b ? name_b : name_a;
3652         data.lineno = 0;
3653         data.o = o;
3654         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3655         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3656
3657         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3658             fill_mmfile(o->repo, &mf2, two) < 0)
3659                 die("unable to read files to diff");
3660
3661         /*
3662          * All the other codepaths check both sides, but not checking
3663          * the "old" side here is deliberate.  We are checking the newly
3664          * introduced changes, and as long as the "new" side is text, we
3665          * can and should check what it introduces.
3666          */
3667         if (diff_filespec_is_binary(o->repo, two))
3668                 goto free_and_return;
3669         else {
3670                 /* Crazy xdl interfaces.. */
3671                 xpparam_t xpp;
3672                 xdemitconf_t xecfg;
3673
3674                 memset(&xpp, 0, sizeof(xpp));
3675                 memset(&xecfg, 0, sizeof(xecfg));
3676                 xecfg.ctxlen = 1; /* at least one context line */
3677                 xpp.flags = 0;
3678                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3679                                   &xpp, &xecfg))
3680                         die("unable to generate checkdiff for %s", one->path);
3681
3682                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3683                         struct emit_callback ecbdata;
3684                         int blank_at_eof;
3685
3686                         ecbdata.ws_rule = data.ws_rule;
3687                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3688                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3689
3690                         if (blank_at_eof) {
3691                                 static char *err;
3692                                 if (!err)
3693                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3694                                 fprintf(o->file, "%s:%d: %s.\n",
3695                                         data.filename, blank_at_eof, err);
3696                                 data.status = 1; /* report errors */
3697                         }
3698                 }
3699         }
3700  free_and_return:
3701         diff_free_filespec_data(one);
3702         diff_free_filespec_data(two);
3703         if (data.status)
3704                 o->flags.check_failed = 1;
3705 }
3706
3707 struct diff_filespec *alloc_filespec(const char *path)
3708 {
3709         struct diff_filespec *spec;
3710
3711         FLEXPTR_ALLOC_STR(spec, path, path);
3712         spec->count = 1;
3713         spec->is_binary = -1;
3714         return spec;
3715 }
3716
3717 void free_filespec(struct diff_filespec *spec)
3718 {
3719         if (!--spec->count) {
3720                 diff_free_filespec_data(spec);
3721                 free(spec);
3722         }
3723 }
3724
3725 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3726                    int oid_valid, unsigned short mode)
3727 {
3728         if (mode) {
3729                 spec->mode = canon_mode(mode);
3730                 oidcpy(&spec->oid, oid);
3731                 spec->oid_valid = oid_valid;
3732         }
3733 }
3734
3735 /*
3736  * Given a name and sha1 pair, if the index tells us the file in
3737  * the work tree has that object contents, return true, so that
3738  * prepare_temp_file() does not have to inflate and extract.
3739  */
3740 static int reuse_worktree_file(struct index_state *istate,
3741                                const char *name,
3742                                const struct object_id *oid,
3743                                int want_file)
3744 {
3745         const struct cache_entry *ce;
3746         struct stat st;
3747         int pos, len;
3748
3749         /*
3750          * We do not read the cache ourselves here, because the
3751          * benchmark with my previous version that always reads cache
3752          * shows that it makes things worse for diff-tree comparing
3753          * two linux-2.6 kernel trees in an already checked out work
3754          * tree.  This is because most diff-tree comparisons deal with
3755          * only a small number of files, while reading the cache is
3756          * expensive for a large project, and its cost outweighs the
3757          * savings we get by not inflating the object to a temporary
3758          * file.  Practically, this code only helps when we are used
3759          * by diff-cache --cached, which does read the cache before
3760          * calling us.
3761          */
3762         if (!istate->cache)
3763                 return 0;
3764
3765         /* We want to avoid the working directory if our caller
3766          * doesn't need the data in a normal file, this system
3767          * is rather slow with its stat/open/mmap/close syscalls,
3768          * and the object is contained in a pack file.  The pack
3769          * is probably already open and will be faster to obtain
3770          * the data through than the working directory.  Loose
3771          * objects however would tend to be slower as they need
3772          * to be individually opened and inflated.
3773          */
3774         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3775                 return 0;
3776
3777         /*
3778          * Similarly, if we'd have to convert the file contents anyway, that
3779          * makes the optimization not worthwhile.
3780          */
3781         if (!want_file && would_convert_to_git(istate, name))
3782                 return 0;
3783
3784         len = strlen(name);
3785         pos = index_name_pos(istate, name, len);
3786         if (pos < 0)
3787                 return 0;
3788         ce = istate->cache[pos];
3789
3790         /*
3791          * This is not the sha1 we are looking for, or
3792          * unreusable because it is not a regular file.
3793          */
3794         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3795                 return 0;
3796
3797         /*
3798          * If ce is marked as "assume unchanged", there is no
3799          * guarantee that work tree matches what we are looking for.
3800          */
3801         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3802                 return 0;
3803
3804         /*
3805          * If ce matches the file in the work tree, we can reuse it.
3806          */
3807         if (ce_uptodate(ce) ||
3808             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3809                 return 1;
3810
3811         return 0;
3812 }
3813
3814 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3815 {
3816         struct strbuf buf = STRBUF_INIT;
3817         char *dirty = "";
3818
3819         /* Are we looking at the work tree? */
3820         if (s->dirty_submodule)
3821                 dirty = "-dirty";
3822
3823         strbuf_addf(&buf, "Subproject commit %s%s\n",
3824                     oid_to_hex(&s->oid), dirty);
3825         s->size = buf.len;
3826         if (size_only) {
3827                 s->data = NULL;
3828                 strbuf_release(&buf);
3829         } else {
3830                 s->data = strbuf_detach(&buf, NULL);
3831                 s->should_free = 1;
3832         }
3833         return 0;
3834 }
3835
3836 /*
3837  * While doing rename detection and pickaxe operation, we may need to
3838  * grab the data for the blob (or file) for our own in-core comparison.
3839  * diff_filespec has data and size fields for this purpose.
3840  */
3841 int diff_populate_filespec(struct repository *r,
3842                            struct diff_filespec *s,
3843                            unsigned int flags)
3844 {
3845         int size_only = flags & CHECK_SIZE_ONLY;
3846         int err = 0;
3847         int conv_flags = global_conv_flags_eol;
3848         /*
3849          * demote FAIL to WARN to allow inspecting the situation
3850          * instead of refusing.
3851          */
3852         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3853                 conv_flags = CONV_EOL_RNDTRP_WARN;
3854
3855         if (!DIFF_FILE_VALID(s))
3856                 die("internal error: asking to populate invalid file.");
3857         if (S_ISDIR(s->mode))
3858                 return -1;
3859
3860         if (s->data)
3861                 return 0;
3862
3863         if (size_only && 0 < s->size)
3864                 return 0;
3865
3866         if (S_ISGITLINK(s->mode))
3867                 return diff_populate_gitlink(s, size_only);
3868
3869         if (!s->oid_valid ||
3870             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3871                 struct strbuf buf = STRBUF_INIT;
3872                 struct stat st;
3873                 int fd;
3874
3875                 if (lstat(s->path, &st) < 0) {
3876                 err_empty:
3877                         err = -1;
3878                 empty:
3879                         s->data = (char *)"";
3880                         s->size = 0;
3881                         return err;
3882                 }
3883                 s->size = xsize_t(st.st_size);
3884                 if (!s->size)
3885                         goto empty;
3886                 if (S_ISLNK(st.st_mode)) {
3887                         struct strbuf sb = STRBUF_INIT;
3888
3889                         if (strbuf_readlink(&sb, s->path, s->size))
3890                                 goto err_empty;
3891                         s->size = sb.len;
3892                         s->data = strbuf_detach(&sb, NULL);
3893                         s->should_free = 1;
3894                         return 0;
3895                 }
3896
3897                 /*
3898                  * Even if the caller would be happy with getting
3899                  * only the size, we cannot return early at this
3900                  * point if the path requires us to run the content
3901                  * conversion.
3902                  */
3903                 if (size_only && !would_convert_to_git(r->index, s->path))
3904                         return 0;
3905
3906                 /*
3907                  * Note: this check uses xsize_t(st.st_size) that may
3908                  * not be the true size of the blob after it goes
3909                  * through convert_to_git().  This may not strictly be
3910                  * correct, but the whole point of big_file_threshold
3911                  * and is_binary check being that we want to avoid
3912                  * opening the file and inspecting the contents, this
3913                  * is probably fine.
3914                  */
3915                 if ((flags & CHECK_BINARY) &&
3916                     s->size > big_file_threshold && s->is_binary == -1) {
3917                         s->is_binary = 1;
3918                         return 0;
3919                 }
3920                 fd = open(s->path, O_RDONLY);
3921                 if (fd < 0)
3922                         goto err_empty;
3923                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3924                 close(fd);
3925                 s->should_munmap = 1;
3926
3927                 /*
3928                  * Convert from working tree format to canonical git format
3929                  */
3930                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3931                         size_t size = 0;
3932                         munmap(s->data, s->size);
3933                         s->should_munmap = 0;
3934                         s->data = strbuf_detach(&buf, &size);
3935                         s->size = size;
3936                         s->should_free = 1;
3937                 }
3938         }
3939         else {
3940                 enum object_type type;
3941                 if (size_only || (flags & CHECK_BINARY)) {
3942                         type = oid_object_info(r, &s->oid, &s->size);
3943                         if (type < 0)
3944                                 die("unable to read %s",
3945                                     oid_to_hex(&s->oid));
3946                         if (size_only)
3947                                 return 0;
3948                         if (s->size > big_file_threshold && s->is_binary == -1) {
3949                                 s->is_binary = 1;
3950                                 return 0;
3951                         }
3952                 }
3953                 s->data = read_object_file(&s->oid, &type, &s->size);
3954                 if (!s->data)
3955                         die("unable to read %s", oid_to_hex(&s->oid));
3956                 s->should_free = 1;
3957         }
3958         return 0;
3959 }
3960
3961 void diff_free_filespec_blob(struct diff_filespec *s)
3962 {
3963         if (s->should_free)
3964                 free(s->data);
3965         else if (s->should_munmap)
3966                 munmap(s->data, s->size);
3967
3968         if (s->should_free || s->should_munmap) {
3969                 s->should_free = s->should_munmap = 0;
3970                 s->data = NULL;
3971         }
3972 }
3973
3974 void diff_free_filespec_data(struct diff_filespec *s)
3975 {
3976         diff_free_filespec_blob(s);
3977         FREE_AND_NULL(s->cnt_data);
3978 }
3979
3980 static void prep_temp_blob(struct index_state *istate,
3981                            const char *path, struct diff_tempfile *temp,
3982                            void *blob,
3983                            unsigned long size,
3984                            const struct object_id *oid,
3985                            int mode)
3986 {
3987         struct strbuf buf = STRBUF_INIT;
3988         struct strbuf tempfile = STRBUF_INIT;
3989         char *path_dup = xstrdup(path);
3990         const char *base = basename(path_dup);
3991
3992         /* Generate "XXXXXX_basename.ext" */
3993         strbuf_addstr(&tempfile, "XXXXXX_");
3994         strbuf_addstr(&tempfile, base);
3995
3996         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
3997         if (!temp->tempfile)
3998                 die_errno("unable to create temp-file");
3999         if (convert_to_working_tree(istate, path,
4000                         (const char *)blob, (size_t)size, &buf)) {
4001                 blob = buf.buf;
4002                 size = buf.len;
4003         }
4004         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4005             close_tempfile_gently(temp->tempfile))
4006                 die_errno("unable to write temp-file");
4007         temp->name = get_tempfile_path(temp->tempfile);
4008         oid_to_hex_r(temp->hex, oid);
4009         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4010         strbuf_release(&buf);
4011         strbuf_release(&tempfile);
4012         free(path_dup);
4013 }
4014
4015 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4016                                                const char *name,
4017                                                struct diff_filespec *one)
4018 {
4019         struct diff_tempfile *temp = claim_diff_tempfile();
4020
4021         if (!DIFF_FILE_VALID(one)) {
4022         not_a_valid_file:
4023                 /* A '-' entry produces this for file-2, and
4024                  * a '+' entry produces this for file-1.
4025                  */
4026                 temp->name = "/dev/null";
4027                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4028                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4029                 return temp;
4030         }
4031
4032         if (!S_ISGITLINK(one->mode) &&
4033             (!one->oid_valid ||
4034              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4035                 struct stat st;
4036                 if (lstat(name, &st) < 0) {
4037                         if (errno == ENOENT)
4038                                 goto not_a_valid_file;
4039                         die_errno("stat(%s)", name);
4040                 }
4041                 if (S_ISLNK(st.st_mode)) {
4042                         struct strbuf sb = STRBUF_INIT;
4043                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4044                                 die_errno("readlink(%s)", name);
4045                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4046                                        (one->oid_valid ?
4047                                         &one->oid : &null_oid),
4048                                        (one->oid_valid ?
4049                                         one->mode : S_IFLNK));
4050                         strbuf_release(&sb);
4051                 }
4052                 else {
4053                         /* we can borrow from the file in the work tree */
4054                         temp->name = name;
4055                         if (!one->oid_valid)
4056                                 oid_to_hex_r(temp->hex, &null_oid);
4057                         else
4058                                 oid_to_hex_r(temp->hex, &one->oid);
4059                         /* Even though we may sometimes borrow the
4060                          * contents from the work tree, we always want
4061                          * one->mode.  mode is trustworthy even when
4062                          * !(one->oid_valid), as long as
4063                          * DIFF_FILE_VALID(one).
4064                          */
4065                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4066                 }
4067                 return temp;
4068         }
4069         else {
4070                 if (diff_populate_filespec(r, one, 0))
4071                         die("cannot read data blob for %s", one->path);
4072                 prep_temp_blob(r->index, name, temp,
4073                                one->data, one->size,
4074                                &one->oid, one->mode);
4075         }
4076         return temp;
4077 }
4078
4079 static void add_external_diff_name(struct repository *r,
4080                                    struct argv_array *argv,
4081                                    const char *name,
4082                                    struct diff_filespec *df)
4083 {
4084         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4085         argv_array_push(argv, temp->name);
4086         argv_array_push(argv, temp->hex);
4087         argv_array_push(argv, temp->mode);
4088 }
4089
4090 /* An external diff command takes:
4091  *
4092  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4093  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4094  *
4095  */
4096 static void run_external_diff(const char *pgm,
4097                               const char *name,
4098                               const char *other,
4099                               struct diff_filespec *one,
4100                               struct diff_filespec *two,
4101                               const char *xfrm_msg,
4102                               int complete_rewrite,
4103                               struct diff_options *o)
4104 {
4105         struct argv_array argv = ARGV_ARRAY_INIT;
4106         struct argv_array env = ARGV_ARRAY_INIT;
4107         struct diff_queue_struct *q = &diff_queued_diff;
4108
4109         argv_array_push(&argv, pgm);
4110         argv_array_push(&argv, name);
4111
4112         if (one && two) {
4113                 add_external_diff_name(o->repo, &argv, name, one);
4114                 if (!other)
4115                         add_external_diff_name(o->repo, &argv, name, two);
4116                 else {
4117                         add_external_diff_name(o->repo, &argv, other, two);
4118                         argv_array_push(&argv, other);
4119                         argv_array_push(&argv, xfrm_msg);
4120                 }
4121         }
4122
4123         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4124         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4125
4126         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4127                 die(_("external diff died, stopping at %s"), name);
4128
4129         remove_tempfile();
4130         argv_array_clear(&argv);
4131         argv_array_clear(&env);
4132 }
4133
4134 static int similarity_index(struct diff_filepair *p)
4135 {
4136         return p->score * 100 / MAX_SCORE;
4137 }
4138
4139 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4140 {
4141         if (startup_info->have_repository)
4142                 return find_unique_abbrev(oid, abbrev);
4143         else {
4144                 char *hex = oid_to_hex(oid);
4145                 if (abbrev < 0)
4146                         abbrev = FALLBACK_DEFAULT_ABBREV;
4147                 if (abbrev > the_hash_algo->hexsz)
4148                         BUG("oid abbreviation out of range: %d", abbrev);
4149                 if (abbrev)
4150                         hex[abbrev] = '\0';
4151                 return hex;
4152         }
4153 }
4154
4155 static void fill_metainfo(struct strbuf *msg,
4156                           const char *name,
4157                           const char *other,
4158                           struct diff_filespec *one,
4159                           struct diff_filespec *two,
4160                           struct diff_options *o,
4161                           struct diff_filepair *p,
4162                           int *must_show_header,
4163                           int use_color)
4164 {
4165         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4166         const char *reset = diff_get_color(use_color, DIFF_RESET);
4167         const char *line_prefix = diff_line_prefix(o);
4168
4169         *must_show_header = 1;
4170         strbuf_init(msg, PATH_MAX * 2 + 300);
4171         switch (p->status) {
4172         case DIFF_STATUS_COPIED:
4173                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4174                             line_prefix, set, similarity_index(p));
4175                 strbuf_addf(msg, "%s\n%s%scopy from ",
4176                             reset,  line_prefix, set);
4177                 quote_c_style(name, msg, NULL, 0);
4178                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4179                 quote_c_style(other, msg, NULL, 0);
4180                 strbuf_addf(msg, "%s\n", reset);
4181                 break;
4182         case DIFF_STATUS_RENAMED:
4183                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4184                             line_prefix, set, similarity_index(p));
4185                 strbuf_addf(msg, "%s\n%s%srename from ",
4186                             reset, line_prefix, set);
4187                 quote_c_style(name, msg, NULL, 0);
4188                 strbuf_addf(msg, "%s\n%s%srename to ",
4189                             reset, line_prefix, set);
4190                 quote_c_style(other, msg, NULL, 0);
4191                 strbuf_addf(msg, "%s\n", reset);
4192                 break;
4193         case DIFF_STATUS_MODIFIED:
4194                 if (p->score) {
4195                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4196                                     line_prefix,
4197                                     set, similarity_index(p), reset);
4198                         break;
4199                 }
4200                 /* fallthru */
4201         default:
4202                 *must_show_header = 0;
4203         }
4204         if (one && two && !oideq(&one->oid, &two->oid)) {
4205                 const unsigned hexsz = the_hash_algo->hexsz;
4206                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4207
4208                 if (o->flags.binary) {
4209                         mmfile_t mf;
4210                         if ((!fill_mmfile(o->repo, &mf, one) &&
4211                              diff_filespec_is_binary(o->repo, one)) ||
4212                             (!fill_mmfile(o->repo, &mf, two) &&
4213                              diff_filespec_is_binary(o->repo, two)))
4214                                 abbrev = hexsz;
4215                 }
4216                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4217                             diff_abbrev_oid(&one->oid, abbrev),
4218                             diff_abbrev_oid(&two->oid, abbrev));
4219                 if (one->mode == two->mode)
4220                         strbuf_addf(msg, " %06o", one->mode);
4221                 strbuf_addf(msg, "%s\n", reset);
4222         }
4223 }
4224
4225 static void run_diff_cmd(const char *pgm,
4226                          const char *name,
4227                          const char *other,
4228                          const char *attr_path,
4229                          struct diff_filespec *one,
4230                          struct diff_filespec *two,
4231                          struct strbuf *msg,
4232                          struct diff_options *o,
4233                          struct diff_filepair *p)
4234 {
4235         const char *xfrm_msg = NULL;
4236         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4237         int must_show_header = 0;
4238
4239
4240         if (o->flags.allow_external) {
4241                 struct userdiff_driver *drv;
4242
4243                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4244                 if (drv && drv->external)
4245                         pgm = drv->external;
4246         }
4247
4248         if (msg) {
4249                 /*
4250                  * don't use colors when the header is intended for an
4251                  * external diff driver
4252                  */
4253                 fill_metainfo(msg, name, other, one, two, o, p,
4254                               &must_show_header,
4255                               want_color(o->use_color) && !pgm);
4256                 xfrm_msg = msg->len ? msg->buf : NULL;
4257         }
4258
4259         if (pgm) {
4260                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4261                                   complete_rewrite, o);
4262                 return;
4263         }
4264         if (one && two)
4265                 builtin_diff(name, other ? other : name,
4266                              one, two, xfrm_msg, must_show_header,
4267                              o, complete_rewrite);
4268         else
4269                 fprintf(o->file, "* Unmerged path %s\n", name);
4270 }
4271
4272 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4273 {
4274         if (DIFF_FILE_VALID(one)) {
4275                 if (!one->oid_valid) {
4276                         struct stat st;
4277                         if (one->is_stdin) {
4278                                 oidclr(&one->oid);
4279                                 return;
4280                         }
4281                         if (lstat(one->path, &st) < 0)
4282                                 die_errno("stat '%s'", one->path);
4283                         if (index_path(istate, &one->oid, one->path, &st, 0))
4284                                 die("cannot hash %s", one->path);
4285                 }
4286         }
4287         else
4288                 oidclr(&one->oid);
4289 }
4290
4291 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4292 {
4293         /* Strip the prefix but do not molest /dev/null and absolute paths */
4294         if (*namep && **namep != '/') {
4295                 *namep += prefix_length;
4296                 if (**namep == '/')
4297                         ++*namep;
4298         }
4299         if (*otherp && **otherp != '/') {
4300                 *otherp += prefix_length;
4301                 if (**otherp == '/')
4302                         ++*otherp;
4303         }
4304 }
4305
4306 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4307 {
4308         const char *pgm = external_diff();
4309         struct strbuf msg;
4310         struct diff_filespec *one = p->one;
4311         struct diff_filespec *two = p->two;
4312         const char *name;
4313         const char *other;
4314         const char *attr_path;
4315
4316         name  = one->path;
4317         other = (strcmp(name, two->path) ? two->path : NULL);
4318         attr_path = name;
4319         if (o->prefix_length)
4320                 strip_prefix(o->prefix_length, &name, &other);
4321
4322         if (!o->flags.allow_external)
4323                 pgm = NULL;
4324
4325         if (DIFF_PAIR_UNMERGED(p)) {
4326                 run_diff_cmd(pgm, name, NULL, attr_path,
4327                              NULL, NULL, NULL, o, p);
4328                 return;
4329         }
4330
4331         diff_fill_oid_info(one, o->repo->index);
4332         diff_fill_oid_info(two, o->repo->index);
4333
4334         if (!pgm &&
4335             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4336             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4337                 /*
4338                  * a filepair that changes between file and symlink
4339                  * needs to be split into deletion and creation.
4340                  */
4341                 struct diff_filespec *null = alloc_filespec(two->path);
4342                 run_diff_cmd(NULL, name, other, attr_path,
4343                              one, null, &msg,
4344                              o, p);
4345                 free(null);
4346                 strbuf_release(&msg);
4347
4348                 null = alloc_filespec(one->path);
4349                 run_diff_cmd(NULL, name, other, attr_path,
4350                              null, two, &msg, o, p);
4351                 free(null);
4352         }
4353         else
4354                 run_diff_cmd(pgm, name, other, attr_path,
4355                              one, two, &msg, o, p);
4356
4357         strbuf_release(&msg);
4358 }
4359
4360 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4361                          struct diffstat_t *diffstat)
4362 {
4363         const char *name;
4364         const char *other;
4365
4366         if (DIFF_PAIR_UNMERGED(p)) {
4367                 /* unmerged */
4368                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4369                                  diffstat, o, p);
4370                 return;
4371         }
4372
4373         name = p->one->path;
4374         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4375
4376         if (o->prefix_length)
4377                 strip_prefix(o->prefix_length, &name, &other);
4378
4379         diff_fill_oid_info(p->one, o->repo->index);
4380         diff_fill_oid_info(p->two, o->repo->index);
4381
4382         builtin_diffstat(name, other, p->one, p->two,
4383                          diffstat, o, p);
4384 }
4385
4386 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4387 {
4388         const char *name;
4389         const char *other;
4390         const char *attr_path;
4391
4392         if (DIFF_PAIR_UNMERGED(p)) {
4393                 /* unmerged */
4394                 return;
4395         }
4396
4397         name = p->one->path;
4398         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4399         attr_path = other ? other : name;
4400
4401         if (o->prefix_length)
4402                 strip_prefix(o->prefix_length, &name, &other);
4403
4404         diff_fill_oid_info(p->one, o->repo->index);
4405         diff_fill_oid_info(p->two, o->repo->index);
4406
4407         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4408 }
4409
4410 void repo_diff_setup(struct repository *r, struct diff_options *options)
4411 {
4412         memcpy(options, &default_diff_options, sizeof(*options));
4413
4414         options->file = stdout;
4415         options->repo = r;
4416
4417         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4418         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4419         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4420         options->abbrev = DEFAULT_ABBREV;
4421         options->line_termination = '\n';
4422         options->break_opt = -1;
4423         options->rename_limit = -1;
4424         options->dirstat_permille = diff_dirstat_permille_default;
4425         options->context = diff_context_default;
4426         options->interhunkcontext = diff_interhunk_context_default;
4427         options->ws_error_highlight = ws_error_highlight_default;
4428         options->flags.rename_empty = 1;
4429         options->objfind = NULL;
4430
4431         /* pathchange left =NULL by default */
4432         options->change = diff_change;
4433         options->add_remove = diff_addremove;
4434         options->use_color = diff_use_color_default;
4435         options->detect_rename = diff_detect_rename_default;
4436         options->xdl_opts |= diff_algorithm;
4437         if (diff_indent_heuristic)
4438                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4439
4440         options->orderfile = diff_order_file_cfg;
4441
4442         if (diff_no_prefix) {
4443                 options->a_prefix = options->b_prefix = "";
4444         } else if (!diff_mnemonic_prefix) {
4445                 options->a_prefix = "a/";
4446                 options->b_prefix = "b/";
4447         }
4448
4449         options->color_moved = diff_color_moved_default;
4450         options->color_moved_ws_handling = diff_color_moved_ws_default;
4451 }
4452
4453 void diff_setup_done(struct diff_options *options)
4454 {
4455         unsigned check_mask = DIFF_FORMAT_NAME |
4456                               DIFF_FORMAT_NAME_STATUS |
4457                               DIFF_FORMAT_CHECKDIFF |
4458                               DIFF_FORMAT_NO_OUTPUT;
4459         /*
4460          * This must be signed because we're comparing against a potentially
4461          * negative value.
4462          */
4463         const int hexsz = the_hash_algo->hexsz;
4464
4465         if (options->set_default)
4466                 options->set_default(options);
4467
4468         if (HAS_MULTI_BITS(options->output_format & check_mask))
4469                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4470
4471         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4472                 die(_("-G, -S and --find-object are mutually exclusive"));
4473
4474         /*
4475          * Most of the time we can say "there are changes"
4476          * only by checking if there are changed paths, but
4477          * --ignore-whitespace* options force us to look
4478          * inside contents.
4479          */
4480
4481         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4482                 options->flags.diff_from_contents = 1;
4483         else
4484                 options->flags.diff_from_contents = 0;
4485
4486         if (options->flags.find_copies_harder)
4487                 options->detect_rename = DIFF_DETECT_COPY;
4488
4489         if (!options->flags.relative_name)
4490                 options->prefix = NULL;
4491         if (options->prefix)
4492                 options->prefix_length = strlen(options->prefix);
4493         else
4494                 options->prefix_length = 0;
4495
4496         if (options->output_format & (DIFF_FORMAT_NAME |
4497                                       DIFF_FORMAT_NAME_STATUS |
4498                                       DIFF_FORMAT_CHECKDIFF |
4499                                       DIFF_FORMAT_NO_OUTPUT))
4500                 options->output_format &= ~(DIFF_FORMAT_RAW |
4501                                             DIFF_FORMAT_NUMSTAT |
4502                                             DIFF_FORMAT_DIFFSTAT |
4503                                             DIFF_FORMAT_SHORTSTAT |
4504                                             DIFF_FORMAT_DIRSTAT |
4505                                             DIFF_FORMAT_SUMMARY |
4506                                             DIFF_FORMAT_PATCH);
4507
4508         /*
4509          * These cases always need recursive; we do not drop caller-supplied
4510          * recursive bits for other formats here.
4511          */
4512         if (options->output_format & (DIFF_FORMAT_PATCH |
4513                                       DIFF_FORMAT_NUMSTAT |
4514                                       DIFF_FORMAT_DIFFSTAT |
4515                                       DIFF_FORMAT_SHORTSTAT |
4516                                       DIFF_FORMAT_DIRSTAT |
4517                                       DIFF_FORMAT_SUMMARY |
4518                                       DIFF_FORMAT_CHECKDIFF))
4519                 options->flags.recursive = 1;
4520         /*
4521          * Also pickaxe would not work very well if you do not say recursive
4522          */
4523         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4524                 options->flags.recursive = 1;
4525         /*
4526          * When patches are generated, submodules diffed against the work tree
4527          * must be checked for dirtiness too so it can be shown in the output
4528          */
4529         if (options->output_format & DIFF_FORMAT_PATCH)
4530                 options->flags.dirty_submodules = 1;
4531
4532         if (options->detect_rename && options->rename_limit < 0)
4533                 options->rename_limit = diff_rename_limit_default;
4534         if (hexsz < options->abbrev)
4535                 options->abbrev = hexsz; /* full */
4536
4537         /*
4538          * It does not make sense to show the first hit we happened
4539          * to have found.  It does not make sense not to return with
4540          * exit code in such a case either.
4541          */
4542         if (options->flags.quick) {
4543                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4544                 options->flags.exit_with_status = 1;
4545         }
4546
4547         options->diff_path_counter = 0;
4548
4549         if (options->flags.follow_renames && options->pathspec.nr != 1)
4550                 die(_("--follow requires exactly one pathspec"));
4551
4552         if (!options->use_color || external_diff())
4553                 options->color_moved = 0;
4554 }
4555
4556 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4557 {
4558         char c, *eq;
4559         int len;
4560
4561         if (*arg != '-')
4562                 return 0;
4563         c = *++arg;
4564         if (!c)
4565                 return 0;
4566         if (c == arg_short) {
4567                 c = *++arg;
4568                 if (!c)
4569                         return 1;
4570                 if (val && isdigit(c)) {
4571                         char *end;
4572                         int n = strtoul(arg, &end, 10);
4573                         if (*end)
4574                                 return 0;
4575                         *val = n;
4576                         return 1;
4577                 }
4578                 return 0;
4579         }
4580         if (c != '-')
4581                 return 0;
4582         arg++;
4583         eq = strchrnul(arg, '=');
4584         len = eq - arg;
4585         if (!len || strncmp(arg, arg_long, len))
4586                 return 0;
4587         if (*eq) {
4588                 int n;
4589                 char *end;
4590                 if (!isdigit(*++eq))
4591                         return 0;
4592                 n = strtoul(eq, &end, 10);
4593                 if (*end)
4594                         return 0;
4595                 *val = n;
4596         }
4597         return 1;
4598 }
4599
4600 static int diff_scoreopt_parse(const char *opt);
4601
4602 static inline int short_opt(char opt, const char **argv,
4603                             const char **optarg)
4604 {
4605         const char *arg = argv[0];
4606         if (arg[0] != '-' || arg[1] != opt)
4607                 return 0;
4608         if (arg[2] != '\0') {
4609                 *optarg = arg + 2;
4610                 return 1;
4611         }
4612         if (!argv[1])
4613                 die("Option '%c' requires a value", opt);
4614         *optarg = argv[1];
4615         return 2;
4616 }
4617
4618 int parse_long_opt(const char *opt, const char **argv,
4619                    const char **optarg)
4620 {
4621         const char *arg = argv[0];
4622         if (!skip_prefix(arg, "--", &arg))
4623                 return 0;
4624         if (!skip_prefix(arg, opt, &arg))
4625                 return 0;
4626         if (*arg == '=') { /* stuck form: --option=value */
4627                 *optarg = arg + 1;
4628                 return 1;
4629         }
4630         if (*arg != '\0')
4631                 return 0;
4632         /* separate form: --option value */
4633         if (!argv[1])
4634                 die("Option '--%s' requires a value", opt);
4635         *optarg = argv[1];
4636         return 2;
4637 }
4638
4639 static int stat_opt(struct diff_options *options, const char **av)
4640 {
4641         const char *arg = av[0];
4642         char *end;
4643         int width = options->stat_width;
4644         int name_width = options->stat_name_width;
4645         int graph_width = options->stat_graph_width;
4646         int count = options->stat_count;
4647         int argcount = 1;
4648
4649         if (!skip_prefix(arg, "--stat", &arg))
4650                 BUG("stat option does not begin with --stat: %s", arg);
4651         end = (char *)arg;
4652
4653         switch (*arg) {
4654         case '-':
4655                 if (skip_prefix(arg, "-width", &arg)) {
4656                         if (*arg == '=')
4657                                 width = strtoul(arg + 1, &end, 10);
4658                         else if (!*arg && !av[1])
4659                                 die_want_option("--stat-width");
4660                         else if (!*arg) {
4661                                 width = strtoul(av[1], &end, 10);
4662                                 argcount = 2;
4663                         }
4664                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4665                         if (*arg == '=')
4666                                 name_width = strtoul(arg + 1, &end, 10);
4667                         else if (!*arg && !av[1])
4668                                 die_want_option("--stat-name-width");
4669                         else if (!*arg) {
4670                                 name_width = strtoul(av[1], &end, 10);
4671                                 argcount = 2;
4672                         }
4673                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4674                         if (*arg == '=')
4675                                 graph_width = strtoul(arg + 1, &end, 10);
4676                         else if (!*arg && !av[1])
4677                                 die_want_option("--stat-graph-width");
4678                         else if (!*arg) {
4679                                 graph_width = strtoul(av[1], &end, 10);
4680                                 argcount = 2;
4681                         }
4682                 } else if (skip_prefix(arg, "-count", &arg)) {
4683                         if (*arg == '=')
4684                                 count = strtoul(arg + 1, &end, 10);
4685                         else if (!*arg && !av[1])
4686                                 die_want_option("--stat-count");
4687                         else if (!*arg) {
4688                                 count = strtoul(av[1], &end, 10);
4689                                 argcount = 2;
4690                         }
4691                 }
4692                 break;
4693         case '=':
4694                 width = strtoul(arg+1, &end, 10);
4695                 if (*end == ',')
4696                         name_width = strtoul(end+1, &end, 10);
4697                 if (*end == ',')
4698                         count = strtoul(end+1, &end, 10);
4699         }
4700
4701         /* Important! This checks all the error cases! */
4702         if (*end)
4703                 return 0;
4704         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4705         options->stat_name_width = name_width;
4706         options->stat_graph_width = graph_width;
4707         options->stat_width = width;
4708         options->stat_count = count;
4709         return argcount;
4710 }
4711
4712 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4713 {
4714         struct strbuf errmsg = STRBUF_INIT;
4715         if (parse_dirstat_params(options, params, &errmsg))
4716                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4717                     errmsg.buf);
4718         strbuf_release(&errmsg);
4719         /*
4720          * The caller knows a dirstat-related option is given from the command
4721          * line; allow it to say "return this_function();"
4722          */
4723         options->output_format |= DIFF_FORMAT_DIRSTAT;
4724         return 1;
4725 }
4726
4727 static int parse_submodule_opt(struct diff_options *options, const char *value)
4728 {
4729         if (parse_submodule_params(options, value))
4730                 die(_("Failed to parse --submodule option parameter: '%s'"),
4731                         value);
4732         return 1;
4733 }
4734
4735 static const char diff_status_letters[] = {
4736         DIFF_STATUS_ADDED,
4737         DIFF_STATUS_COPIED,
4738         DIFF_STATUS_DELETED,
4739         DIFF_STATUS_MODIFIED,
4740         DIFF_STATUS_RENAMED,
4741         DIFF_STATUS_TYPE_CHANGED,
4742         DIFF_STATUS_UNKNOWN,
4743         DIFF_STATUS_UNMERGED,
4744         DIFF_STATUS_FILTER_AON,
4745         DIFF_STATUS_FILTER_BROKEN,
4746         '\0',
4747 };
4748
4749 static unsigned int filter_bit['Z' + 1];
4750
4751 static void prepare_filter_bits(void)
4752 {
4753         int i;
4754
4755         if (!filter_bit[DIFF_STATUS_ADDED]) {
4756                 for (i = 0; diff_status_letters[i]; i++)
4757                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4758         }
4759 }
4760
4761 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4762 {
4763         return opt->filter & filter_bit[(int) status];
4764 }
4765
4766 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4767 {
4768         int i, optch;
4769
4770         prepare_filter_bits();
4771
4772         /*
4773          * If there is a negation e.g. 'd' in the input, and we haven't
4774          * initialized the filter field with another --diff-filter, start
4775          * from full set of bits, except for AON.
4776          */
4777         if (!opt->filter) {
4778                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4779                         if (optch < 'a' || 'z' < optch)
4780                                 continue;
4781                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4782                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4783                         break;
4784                 }
4785         }
4786
4787         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4788                 unsigned int bit;
4789                 int negate;
4790
4791                 if ('a' <= optch && optch <= 'z') {
4792                         negate = 1;
4793                         optch = toupper(optch);
4794                 } else {
4795                         negate = 0;
4796                 }
4797
4798                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4799                 if (!bit)
4800                         return optarg[i];
4801                 if (negate)
4802                         opt->filter &= ~bit;
4803                 else
4804                         opt->filter |= bit;
4805         }
4806         return 0;
4807 }
4808
4809 static void enable_patch_output(int *fmt) {
4810         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4811         *fmt |= DIFF_FORMAT_PATCH;
4812 }
4813
4814 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4815 {
4816         int val = parse_ws_error_highlight(arg);
4817
4818         if (val < 0) {
4819                 error("unknown value after ws-error-highlight=%.*s",
4820                       -1 - val, arg);
4821                 return 0;
4822         }
4823         opt->ws_error_highlight = val;
4824         return 1;
4825 }
4826
4827 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4828 {
4829         struct object_id oid;
4830
4831         if (get_oid(arg, &oid))
4832                 return error("unable to resolve '%s'", arg);
4833
4834         if (!opt->objfind)
4835                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4836
4837         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4838         opt->flags.recursive = 1;
4839         opt->flags.tree_in_recursive = 1;
4840         oidset_insert(opt->objfind, &oid);
4841         return 1;
4842 }
4843
4844 int diff_opt_parse(struct diff_options *options,
4845                    const char **av, int ac, const char *prefix)
4846 {
4847         const char *arg = av[0];
4848         const char *optarg;
4849         int argcount;
4850
4851         if (!prefix)
4852                 prefix = "";
4853
4854         /* Output format options */
4855         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4856             || opt_arg(arg, 'U', "unified", &options->context))
4857                 enable_patch_output(&options->output_format);
4858         else if (!strcmp(arg, "--raw"))
4859                 options->output_format |= DIFF_FORMAT_RAW;
4860         else if (!strcmp(arg, "--patch-with-raw")) {
4861                 enable_patch_output(&options->output_format);
4862                 options->output_format |= DIFF_FORMAT_RAW;
4863         } else if (!strcmp(arg, "--numstat"))
4864                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4865         else if (!strcmp(arg, "--shortstat"))
4866                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4867         else if (skip_prefix(arg, "-X", &arg) ||
4868                  skip_to_optional_arg(arg, "--dirstat", &arg))
4869                 return parse_dirstat_opt(options, arg);
4870         else if (!strcmp(arg, "--cumulative"))
4871                 return parse_dirstat_opt(options, "cumulative");
4872         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4873                 parse_dirstat_opt(options, "files");
4874                 return parse_dirstat_opt(options, arg);
4875         }
4876         else if (!strcmp(arg, "--check"))
4877                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4878         else if (!strcmp(arg, "--summary"))
4879                 options->output_format |= DIFF_FORMAT_SUMMARY;
4880         else if (!strcmp(arg, "--patch-with-stat")) {
4881                 enable_patch_output(&options->output_format);
4882                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4883         } else if (!strcmp(arg, "--name-only"))
4884                 options->output_format |= DIFF_FORMAT_NAME;
4885         else if (!strcmp(arg, "--name-status"))
4886                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4887         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4888                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4889         else if (starts_with(arg, "--stat"))
4890                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4891                 return stat_opt(options, av);
4892         else if (!strcmp(arg, "--compact-summary")) {
4893                  options->flags.stat_with_summary = 1;
4894                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4895         } else if (!strcmp(arg, "--no-compact-summary"))
4896                  options->flags.stat_with_summary = 0;
4897         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4898                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4899         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4900                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4901         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4902                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4903
4904         /* renames options */
4905         else if (starts_with(arg, "-B") ||
4906                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4907                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4908                         return error("invalid argument to -B: %s", arg+2);
4909         }
4910         else if (starts_with(arg, "-M") ||
4911                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4912                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4913                         return error("invalid argument to -M: %s", arg+2);
4914                 options->detect_rename = DIFF_DETECT_RENAME;
4915         }
4916         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4917                 options->irreversible_delete = 1;
4918         }
4919         else if (starts_with(arg, "-C") ||
4920                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4921                 if (options->detect_rename == DIFF_DETECT_COPY)
4922                         options->flags.find_copies_harder = 1;
4923                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4924                         return error("invalid argument to -C: %s", arg+2);
4925                 options->detect_rename = DIFF_DETECT_COPY;
4926         }
4927         else if (!strcmp(arg, "--no-renames"))
4928                 options->detect_rename = 0;
4929         else if (!strcmp(arg, "--rename-empty"))
4930                 options->flags.rename_empty = 1;
4931         else if (!strcmp(arg, "--no-rename-empty"))
4932                 options->flags.rename_empty = 0;
4933         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4934                 options->flags.relative_name = 1;
4935                 if (arg)
4936                         options->prefix = arg;
4937         }
4938
4939         /* xdiff options */
4940         else if (!strcmp(arg, "--minimal"))
4941                 DIFF_XDL_SET(options, NEED_MINIMAL);
4942         else if (!strcmp(arg, "--no-minimal"))
4943                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4944         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4945                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4946         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4947                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4948         else if (!strcmp(arg, "--ignore-space-at-eol"))
4949                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4950         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4951                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4952         else if (!strcmp(arg, "--ignore-blank-lines"))
4953                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4954         else if (!strcmp(arg, "--indent-heuristic"))
4955                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4956         else if (!strcmp(arg, "--no-indent-heuristic"))
4957                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4958         else if (!strcmp(arg, "--patience")) {
4959                 int i;
4960                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4961                 /*
4962                  * Both --patience and --anchored use PATIENCE_DIFF
4963                  * internally, so remove any anchors previously
4964                  * specified.
4965                  */
4966                 for (i = 0; i < options->anchors_nr; i++)
4967                         free(options->anchors[i]);
4968                 options->anchors_nr = 0;
4969         } else if (!strcmp(arg, "--histogram"))
4970                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4971         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4972                 long value = parse_algorithm_value(optarg);
4973                 if (value < 0)
4974                         return error("option diff-algorithm accepts \"myers\", "
4975                                      "\"minimal\", \"patience\" and \"histogram\"");
4976                 /* clear out previous settings */
4977                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4978                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4979                 options->xdl_opts |= value;
4980                 return argcount;
4981         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4982                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4983                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4984                            options->anchors_alloc);
4985                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4986         }
4987
4988         /* flags options */
4989         else if (!strcmp(arg, "--binary")) {
4990                 enable_patch_output(&options->output_format);
4991                 options->flags.binary = 1;
4992         }
4993         else if (!strcmp(arg, "--full-index"))
4994                 options->flags.full_index = 1;
4995         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4996                 options->flags.text = 1;
4997         else if (!strcmp(arg, "-R"))
4998                 options->flags.reverse_diff = 1;
4999         else if (!strcmp(arg, "--find-copies-harder"))
5000                 options->flags.find_copies_harder = 1;
5001         else if (!strcmp(arg, "--follow"))
5002                 options->flags.follow_renames = 1;
5003         else if (!strcmp(arg, "--no-follow")) {
5004                 options->flags.follow_renames = 0;
5005                 options->flags.default_follow_renames = 0;
5006         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5007                 int value = git_config_colorbool(NULL, arg);
5008                 if (value < 0)
5009                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5010                 options->use_color = value;
5011         }
5012         else if (!strcmp(arg, "--no-color"))
5013                 options->use_color = 0;
5014         else if (!strcmp(arg, "--color-moved")) {
5015                 if (diff_color_moved_default)
5016                         options->color_moved = diff_color_moved_default;
5017                 if (options->color_moved == COLOR_MOVED_NO)
5018                         options->color_moved = COLOR_MOVED_DEFAULT;
5019         } else if (!strcmp(arg, "--no-color-moved"))
5020                 options->color_moved = COLOR_MOVED_NO;
5021         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5022                 int cm = parse_color_moved(arg);
5023                 if (cm < 0)
5024                         die("bad --color-moved argument: %s", arg);
5025                 options->color_moved = cm;
5026         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5027                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5028         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5029                 options->use_color = 1;
5030                 options->word_diff = DIFF_WORDS_COLOR;
5031         }
5032         else if (!strcmp(arg, "--word-diff")) {
5033                 if (options->word_diff == DIFF_WORDS_NONE)
5034                         options->word_diff = DIFF_WORDS_PLAIN;
5035         }
5036         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5037                 if (!strcmp(arg, "plain"))
5038                         options->word_diff = DIFF_WORDS_PLAIN;
5039                 else if (!strcmp(arg, "color")) {
5040                         options->use_color = 1;
5041                         options->word_diff = DIFF_WORDS_COLOR;
5042                 }
5043                 else if (!strcmp(arg, "porcelain"))
5044                         options->word_diff = DIFF_WORDS_PORCELAIN;
5045                 else if (!strcmp(arg, "none"))
5046                         options->word_diff = DIFF_WORDS_NONE;
5047                 else
5048                         die("bad --word-diff argument: %s", arg);
5049         }
5050         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5051                 if (options->word_diff == DIFF_WORDS_NONE)
5052                         options->word_diff = DIFF_WORDS_PLAIN;
5053                 options->word_regex = optarg;
5054                 return argcount;
5055         }
5056         else if (!strcmp(arg, "--exit-code"))
5057                 options->flags.exit_with_status = 1;
5058         else if (!strcmp(arg, "--quiet"))
5059                 options->flags.quick = 1;
5060         else if (!strcmp(arg, "--ext-diff"))
5061                 options->flags.allow_external = 1;
5062         else if (!strcmp(arg, "--no-ext-diff"))
5063                 options->flags.allow_external = 0;
5064         else if (!strcmp(arg, "--textconv")) {
5065                 options->flags.allow_textconv = 1;
5066                 options->flags.textconv_set_via_cmdline = 1;
5067         } else if (!strcmp(arg, "--no-textconv"))
5068                 options->flags.allow_textconv = 0;
5069         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5070                 options->flags.override_submodule_config = 1;
5071                 handle_ignore_submodules_arg(options, arg);
5072         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5073                 return parse_submodule_opt(options, arg);
5074         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5075                 return parse_ws_error_highlight_opt(options, arg);
5076         else if (!strcmp(arg, "--ita-invisible-in-index"))
5077                 options->ita_invisible_in_index = 1;
5078         else if (!strcmp(arg, "--ita-visible-in-index"))
5079                 options->ita_invisible_in_index = 0;
5080
5081         /* misc options */
5082         else if (!strcmp(arg, "-z"))
5083                 options->line_termination = 0;
5084         else if ((argcount = short_opt('l', av, &optarg))) {
5085                 options->rename_limit = strtoul(optarg, NULL, 10);
5086                 return argcount;
5087         }
5088         else if ((argcount = short_opt('S', av, &optarg))) {
5089                 options->pickaxe = optarg;
5090                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5091                 return argcount;
5092         } else if ((argcount = short_opt('G', av, &optarg))) {
5093                 options->pickaxe = optarg;
5094                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5095                 return argcount;
5096         }
5097         else if (!strcmp(arg, "--pickaxe-all"))
5098                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5099         else if (!strcmp(arg, "--pickaxe-regex"))
5100                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5101         else if ((argcount = short_opt('O', av, &optarg))) {
5102                 options->orderfile = prefix_filename(prefix, optarg);
5103                 return argcount;
5104         } else if (skip_prefix(arg, "--find-object=", &arg))
5105                 return parse_objfind_opt(options, arg);
5106         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5107                 int offending = parse_diff_filter_opt(optarg, options);
5108                 if (offending)
5109                         die("unknown change class '%c' in --diff-filter=%s",
5110                             offending, optarg);
5111                 return argcount;
5112         }
5113         else if (!strcmp(arg, "--no-abbrev"))
5114                 options->abbrev = 0;
5115         else if (!strcmp(arg, "--abbrev"))
5116                 options->abbrev = DEFAULT_ABBREV;
5117         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5118                 options->abbrev = strtoul(arg, NULL, 10);
5119                 if (options->abbrev < MINIMUM_ABBREV)
5120                         options->abbrev = MINIMUM_ABBREV;
5121                 else if (the_hash_algo->hexsz < options->abbrev)
5122                         options->abbrev = the_hash_algo->hexsz;
5123         }
5124         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5125                 options->a_prefix = optarg;
5126                 return argcount;
5127         }
5128         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5129                 options->line_prefix = optarg;
5130                 options->line_prefix_length = strlen(options->line_prefix);
5131                 graph_setup_line_prefix(options);
5132                 return argcount;
5133         }
5134         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5135                 options->b_prefix = optarg;
5136                 return argcount;
5137         }
5138         else if (!strcmp(arg, "--no-prefix"))
5139                 options->a_prefix = options->b_prefix = "";
5140         else if (opt_arg(arg, '\0', "inter-hunk-context",
5141                          &options->interhunkcontext))
5142                 ;
5143         else if (!strcmp(arg, "-W"))
5144                 options->flags.funccontext = 1;
5145         else if (!strcmp(arg, "--function-context"))
5146                 options->flags.funccontext = 1;
5147         else if (!strcmp(arg, "--no-function-context"))
5148                 options->flags.funccontext = 0;
5149         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5150                 char *path = prefix_filename(prefix, optarg);
5151                 options->file = xfopen(path, "w");
5152                 options->close_file = 1;
5153                 if (options->use_color != GIT_COLOR_ALWAYS)
5154                         options->use_color = GIT_COLOR_NEVER;
5155                 free(path);
5156                 return argcount;
5157         } else
5158                 return 0;
5159         return 1;
5160 }
5161
5162 int parse_rename_score(const char **cp_p)
5163 {
5164         unsigned long num, scale;
5165         int ch, dot;
5166         const char *cp = *cp_p;
5167
5168         num = 0;
5169         scale = 1;
5170         dot = 0;
5171         for (;;) {
5172                 ch = *cp;
5173                 if ( !dot && ch == '.' ) {
5174                         scale = 1;
5175                         dot = 1;
5176                 } else if ( ch == '%' ) {
5177                         scale = dot ? scale*100 : 100;
5178                         cp++;   /* % is always at the end */
5179                         break;
5180                 } else if ( ch >= '0' && ch <= '9' ) {
5181                         if ( scale < 100000 ) {
5182                                 scale *= 10;
5183                                 num = (num*10) + (ch-'0');
5184                         }
5185                 } else {
5186                         break;
5187                 }
5188                 cp++;
5189         }
5190         *cp_p = cp;
5191
5192         /* user says num divided by scale and we say internally that
5193          * is MAX_SCORE * num / scale.
5194          */
5195         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5196 }
5197
5198 static int diff_scoreopt_parse(const char *opt)
5199 {
5200         int opt1, opt2, cmd;
5201
5202         if (*opt++ != '-')
5203                 return -1;
5204         cmd = *opt++;
5205         if (cmd == '-') {
5206                 /* convert the long-form arguments into short-form versions */
5207                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5208                         if (*opt == 0 || *opt++ == '=')
5209                                 cmd = 'B';
5210                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5211                         if (*opt == 0 || *opt++ == '=')
5212                                 cmd = 'C';
5213                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5214                         if (*opt == 0 || *opt++ == '=')
5215                                 cmd = 'M';
5216                 }
5217         }
5218         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5219                 return -1; /* that is not a -M, -C, or -B option */
5220
5221         opt1 = parse_rename_score(&opt);
5222         if (cmd != 'B')
5223                 opt2 = 0;
5224         else {
5225                 if (*opt == 0)
5226                         opt2 = 0;
5227                 else if (*opt != '/')
5228                         return -1; /* we expect -B80/99 or -B80 */
5229                 else {
5230                         opt++;
5231                         opt2 = parse_rename_score(&opt);
5232                 }
5233         }
5234         if (*opt != 0)
5235                 return -1;
5236         return opt1 | (opt2 << 16);
5237 }
5238
5239 struct diff_queue_struct diff_queued_diff;
5240
5241 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5242 {
5243         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5244         queue->queue[queue->nr++] = dp;
5245 }
5246
5247 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5248                                  struct diff_filespec *one,
5249                                  struct diff_filespec *two)
5250 {
5251         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5252         dp->one = one;
5253         dp->two = two;
5254         if (queue)
5255                 diff_q(queue, dp);
5256         return dp;
5257 }
5258
5259 void diff_free_filepair(struct diff_filepair *p)
5260 {
5261         free_filespec(p->one);
5262         free_filespec(p->two);
5263         free(p);
5264 }
5265
5266 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5267 {
5268         int abblen;
5269         const char *abbrev;
5270
5271         /* Do we want all 40 hex characters? */
5272         if (len == the_hash_algo->hexsz)
5273                 return oid_to_hex(oid);
5274
5275         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5276         abbrev = diff_abbrev_oid(oid, len);
5277
5278         if (!print_sha1_ellipsis())
5279                 return abbrev;
5280
5281         abblen = strlen(abbrev);
5282
5283         /*
5284          * In well-behaved cases, where the abbreviated result is the
5285          * same as the requested length, append three dots after the
5286          * abbreviation (hence the whole logic is limited to the case
5287          * where abblen < 37); when the actual abbreviated result is a
5288          * bit longer than the requested length, we reduce the number
5289          * of dots so that they match the well-behaved ones.  However,
5290          * if the actual abbreviation is longer than the requested
5291          * length by more than three, we give up on aligning, and add
5292          * three dots anyway, to indicate that the output is not the
5293          * full object name.  Yes, this may be suboptimal, but this
5294          * appears only in "diff --raw --abbrev" output and it is not
5295          * worth the effort to change it now.  Note that this would
5296          * likely to work fine when the automatic sizing of default
5297          * abbreviation length is used--we would be fed -1 in "len" in
5298          * that case, and will end up always appending three-dots, but
5299          * the automatic sizing is supposed to give abblen that ensures
5300          * uniqueness across all objects (statistically speaking).
5301          */
5302         if (abblen < the_hash_algo->hexsz - 3) {
5303                 static char hex[GIT_MAX_HEXSZ + 1];
5304                 if (len < abblen && abblen <= len + 2)
5305                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5306                 else
5307                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5308                 return hex;
5309         }
5310
5311         return oid_to_hex(oid);
5312 }
5313
5314 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5315 {
5316         int line_termination = opt->line_termination;
5317         int inter_name_termination = line_termination ? '\t' : '\0';
5318
5319         fprintf(opt->file, "%s", diff_line_prefix(opt));
5320         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5321                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5322                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5323                 fprintf(opt->file, "%s ",
5324                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5325         }
5326         if (p->score) {
5327                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5328                         inter_name_termination);
5329         } else {
5330                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5331         }
5332
5333         if (p->status == DIFF_STATUS_COPIED ||
5334             p->status == DIFF_STATUS_RENAMED) {
5335                 const char *name_a, *name_b;
5336                 name_a = p->one->path;
5337                 name_b = p->two->path;
5338                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5339                 write_name_quoted(name_a, opt->file, inter_name_termination);
5340                 write_name_quoted(name_b, opt->file, line_termination);
5341         } else {
5342                 const char *name_a, *name_b;
5343                 name_a = p->one->mode ? p->one->path : p->two->path;
5344                 name_b = NULL;
5345                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5346                 write_name_quoted(name_a, opt->file, line_termination);
5347         }
5348 }
5349
5350 int diff_unmodified_pair(struct diff_filepair *p)
5351 {
5352         /* This function is written stricter than necessary to support
5353          * the currently implemented transformers, but the idea is to
5354          * let transformers to produce diff_filepairs any way they want,
5355          * and filter and clean them up here before producing the output.
5356          */
5357         struct diff_filespec *one = p->one, *two = p->two;
5358
5359         if (DIFF_PAIR_UNMERGED(p))
5360                 return 0; /* unmerged is interesting */
5361
5362         /* deletion, addition, mode or type change
5363          * and rename are all interesting.
5364          */
5365         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5366             DIFF_PAIR_MODE_CHANGED(p) ||
5367             strcmp(one->path, two->path))
5368                 return 0;
5369
5370         /* both are valid and point at the same path.  that is, we are
5371          * dealing with a change.
5372          */
5373         if (one->oid_valid && two->oid_valid &&
5374             oideq(&one->oid, &two->oid) &&
5375             !one->dirty_submodule && !two->dirty_submodule)
5376                 return 1; /* no change */
5377         if (!one->oid_valid && !two->oid_valid)
5378                 return 1; /* both look at the same file on the filesystem. */
5379         return 0;
5380 }
5381
5382 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5383 {
5384         if (diff_unmodified_pair(p))
5385                 return;
5386
5387         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5388             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5389                 return; /* no tree diffs in patch format */
5390
5391         run_diff(p, o);
5392 }
5393
5394 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5395                             struct diffstat_t *diffstat)
5396 {
5397         if (diff_unmodified_pair(p))
5398                 return;
5399
5400         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5401             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5402                 return; /* no useful stat for tree diffs */
5403
5404         run_diffstat(p, o, diffstat);
5405 }
5406
5407 static void diff_flush_checkdiff(struct diff_filepair *p,
5408                 struct diff_options *o)
5409 {
5410         if (diff_unmodified_pair(p))
5411                 return;
5412
5413         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5414             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5415                 return; /* nothing to check in tree diffs */
5416
5417         run_checkdiff(p, o);
5418 }
5419
5420 int diff_queue_is_empty(void)
5421 {
5422         struct diff_queue_struct *q = &diff_queued_diff;
5423         int i;
5424         for (i = 0; i < q->nr; i++)
5425                 if (!diff_unmodified_pair(q->queue[i]))
5426                         return 0;
5427         return 1;
5428 }
5429
5430 #if DIFF_DEBUG
5431 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5432 {
5433         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5434                 x, one ? one : "",
5435                 s->path,
5436                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5437                 s->mode,
5438                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5439         fprintf(stderr, "queue[%d] %s size %lu\n",
5440                 x, one ? one : "",
5441                 s->size);
5442 }
5443
5444 void diff_debug_filepair(const struct diff_filepair *p, int i)
5445 {
5446         diff_debug_filespec(p->one, i, "one");
5447         diff_debug_filespec(p->two, i, "two");
5448         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5449                 p->score, p->status ? p->status : '?',
5450                 p->one->rename_used, p->broken_pair);
5451 }
5452
5453 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5454 {
5455         int i;
5456         if (msg)
5457                 fprintf(stderr, "%s\n", msg);
5458         fprintf(stderr, "q->nr = %d\n", q->nr);
5459         for (i = 0; i < q->nr; i++) {
5460                 struct diff_filepair *p = q->queue[i];
5461                 diff_debug_filepair(p, i);
5462         }
5463 }
5464 #endif
5465
5466 static void diff_resolve_rename_copy(void)
5467 {
5468         int i;
5469         struct diff_filepair *p;
5470         struct diff_queue_struct *q = &diff_queued_diff;
5471
5472         diff_debug_queue("resolve-rename-copy", q);
5473
5474         for (i = 0; i < q->nr; i++) {
5475                 p = q->queue[i];
5476                 p->status = 0; /* undecided */
5477                 if (DIFF_PAIR_UNMERGED(p))
5478                         p->status = DIFF_STATUS_UNMERGED;
5479                 else if (!DIFF_FILE_VALID(p->one))
5480                         p->status = DIFF_STATUS_ADDED;
5481                 else if (!DIFF_FILE_VALID(p->two))
5482                         p->status = DIFF_STATUS_DELETED;
5483                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5484                         p->status = DIFF_STATUS_TYPE_CHANGED;
5485
5486                 /* from this point on, we are dealing with a pair
5487                  * whose both sides are valid and of the same type, i.e.
5488                  * either in-place edit or rename/copy edit.
5489                  */
5490                 else if (DIFF_PAIR_RENAME(p)) {
5491                         /*
5492                          * A rename might have re-connected a broken
5493                          * pair up, causing the pathnames to be the
5494                          * same again. If so, that's not a rename at
5495                          * all, just a modification..
5496                          *
5497                          * Otherwise, see if this source was used for
5498                          * multiple renames, in which case we decrement
5499                          * the count, and call it a copy.
5500                          */
5501                         if (!strcmp(p->one->path, p->two->path))
5502                                 p->status = DIFF_STATUS_MODIFIED;
5503                         else if (--p->one->rename_used > 0)
5504                                 p->status = DIFF_STATUS_COPIED;
5505                         else
5506                                 p->status = DIFF_STATUS_RENAMED;
5507                 }
5508                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5509                          p->one->mode != p->two->mode ||
5510                          p->one->dirty_submodule ||
5511                          p->two->dirty_submodule ||
5512                          is_null_oid(&p->one->oid))
5513                         p->status = DIFF_STATUS_MODIFIED;
5514                 else {
5515                         /* This is a "no-change" entry and should not
5516                          * happen anymore, but prepare for broken callers.
5517                          */
5518                         error("feeding unmodified %s to diffcore",
5519                               p->one->path);
5520                         p->status = DIFF_STATUS_UNKNOWN;
5521                 }
5522         }
5523         diff_debug_queue("resolve-rename-copy done", q);
5524 }
5525
5526 static int check_pair_status(struct diff_filepair *p)
5527 {
5528         switch (p->status) {
5529         case DIFF_STATUS_UNKNOWN:
5530                 return 0;
5531         case 0:
5532                 die("internal error in diff-resolve-rename-copy");
5533         default:
5534                 return 1;
5535         }
5536 }
5537
5538 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5539 {
5540         int fmt = opt->output_format;
5541
5542         if (fmt & DIFF_FORMAT_CHECKDIFF)
5543                 diff_flush_checkdiff(p, opt);
5544         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5545                 diff_flush_raw(p, opt);
5546         else if (fmt & DIFF_FORMAT_NAME) {
5547                 const char *name_a, *name_b;
5548                 name_a = p->two->path;
5549                 name_b = NULL;
5550                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5551                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5552                 write_name_quoted(name_a, opt->file, opt->line_termination);
5553         }
5554 }
5555
5556 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5557 {
5558         struct strbuf sb = STRBUF_INIT;
5559         if (fs->mode)
5560                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5561         else
5562                 strbuf_addf(&sb, " %s ", newdelete);
5563
5564         quote_c_style(fs->path, &sb, NULL, 0);
5565         strbuf_addch(&sb, '\n');
5566         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5567                          sb.buf, sb.len, 0);
5568         strbuf_release(&sb);
5569 }
5570
5571 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5572                 int show_name)
5573 {
5574         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5575                 struct strbuf sb = STRBUF_INIT;
5576                 strbuf_addf(&sb, " mode change %06o => %06o",
5577                             p->one->mode, p->two->mode);
5578                 if (show_name) {
5579                         strbuf_addch(&sb, ' ');
5580                         quote_c_style(p->two->path, &sb, NULL, 0);
5581                 }
5582                 strbuf_addch(&sb, '\n');
5583                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5584                                  sb.buf, sb.len, 0);
5585                 strbuf_release(&sb);
5586         }
5587 }
5588
5589 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5590                 struct diff_filepair *p)
5591 {
5592         struct strbuf sb = STRBUF_INIT;
5593         struct strbuf names = STRBUF_INIT;
5594
5595         pprint_rename(&names, p->one->path, p->two->path);
5596         strbuf_addf(&sb, " %s %s (%d%%)\n",
5597                     renamecopy, names.buf, similarity_index(p));
5598         strbuf_release(&names);
5599         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5600                                  sb.buf, sb.len, 0);
5601         show_mode_change(opt, p, 0);
5602         strbuf_release(&sb);
5603 }
5604
5605 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5606 {
5607         switch(p->status) {
5608         case DIFF_STATUS_DELETED:
5609                 show_file_mode_name(opt, "delete", p->one);
5610                 break;
5611         case DIFF_STATUS_ADDED:
5612                 show_file_mode_name(opt, "create", p->two);
5613                 break;
5614         case DIFF_STATUS_COPIED:
5615                 show_rename_copy(opt, "copy", p);
5616                 break;
5617         case DIFF_STATUS_RENAMED:
5618                 show_rename_copy(opt, "rename", p);
5619                 break;
5620         default:
5621                 if (p->score) {
5622                         struct strbuf sb = STRBUF_INIT;
5623                         strbuf_addstr(&sb, " rewrite ");
5624                         quote_c_style(p->two->path, &sb, NULL, 0);
5625                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5626                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5627                                          sb.buf, sb.len, 0);
5628                         strbuf_release(&sb);
5629                 }
5630                 show_mode_change(opt, p, !p->score);
5631                 break;
5632         }
5633 }
5634
5635 struct patch_id_t {
5636         git_SHA_CTX *ctx;
5637         int patchlen;
5638 };
5639
5640 static int remove_space(char *line, int len)
5641 {
5642         int i;
5643         char *dst = line;
5644         unsigned char c;
5645
5646         for (i = 0; i < len; i++)
5647                 if (!isspace((c = line[i])))
5648                         *dst++ = c;
5649
5650         return dst - line;
5651 }
5652
5653 static void patch_id_consume(void *priv, char *line, unsigned long len)
5654 {
5655         struct patch_id_t *data = priv;
5656         int new_len;
5657
5658         /* Ignore line numbers when computing the SHA1 of the patch */
5659         if (starts_with(line, "@@ -"))
5660                 return;
5661
5662         new_len = remove_space(line, len);
5663
5664         git_SHA1_Update(data->ctx, line, new_len);
5665         data->patchlen += new_len;
5666 }
5667
5668 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5669 {
5670         git_SHA1_Update(ctx, str, strlen(str));
5671 }
5672
5673 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5674 {
5675         /* large enough for 2^32 in octal */
5676         char buf[12];
5677         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5678         git_SHA1_Update(ctx, buf, len);
5679 }
5680
5681 /* returns 0 upon success, and writes result into sha1 */
5682 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5683 {
5684         struct diff_queue_struct *q = &diff_queued_diff;
5685         int i;
5686         git_SHA_CTX ctx;
5687         struct patch_id_t data;
5688
5689         git_SHA1_Init(&ctx);
5690         memset(&data, 0, sizeof(struct patch_id_t));
5691         data.ctx = &ctx;
5692
5693         for (i = 0; i < q->nr; i++) {
5694                 xpparam_t xpp;
5695                 xdemitconf_t xecfg;
5696                 mmfile_t mf1, mf2;
5697                 struct diff_filepair *p = q->queue[i];
5698                 int len1, len2;
5699
5700                 memset(&xpp, 0, sizeof(xpp));
5701                 memset(&xecfg, 0, sizeof(xecfg));
5702                 if (p->status == 0)
5703                         return error("internal diff status error");
5704                 if (p->status == DIFF_STATUS_UNKNOWN)
5705                         continue;
5706                 if (diff_unmodified_pair(p))
5707                         continue;
5708                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5709                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5710                         continue;
5711                 if (DIFF_PAIR_UNMERGED(p))
5712                         continue;
5713
5714                 diff_fill_oid_info(p->one, options->repo->index);
5715                 diff_fill_oid_info(p->two, options->repo->index);
5716
5717                 len1 = remove_space(p->one->path, strlen(p->one->path));
5718                 len2 = remove_space(p->two->path, strlen(p->two->path));
5719                 patch_id_add_string(&ctx, "diff--git");
5720                 patch_id_add_string(&ctx, "a/");
5721                 git_SHA1_Update(&ctx, p->one->path, len1);
5722                 patch_id_add_string(&ctx, "b/");
5723                 git_SHA1_Update(&ctx, p->two->path, len2);
5724
5725                 if (p->one->mode == 0) {
5726                         patch_id_add_string(&ctx, "newfilemode");
5727                         patch_id_add_mode(&ctx, p->two->mode);
5728                         patch_id_add_string(&ctx, "---/dev/null");
5729                         patch_id_add_string(&ctx, "+++b/");
5730                         git_SHA1_Update(&ctx, p->two->path, len2);
5731                 } else if (p->two->mode == 0) {
5732                         patch_id_add_string(&ctx, "deletedfilemode");
5733                         patch_id_add_mode(&ctx, p->one->mode);
5734                         patch_id_add_string(&ctx, "---a/");
5735                         git_SHA1_Update(&ctx, p->one->path, len1);
5736                         patch_id_add_string(&ctx, "+++/dev/null");
5737                 } else {
5738                         patch_id_add_string(&ctx, "---a/");
5739                         git_SHA1_Update(&ctx, p->one->path, len1);
5740                         patch_id_add_string(&ctx, "+++b/");
5741                         git_SHA1_Update(&ctx, p->two->path, len2);
5742                 }
5743
5744                 if (diff_header_only)
5745                         continue;
5746
5747                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5748                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5749                         return error("unable to read files to diff");
5750
5751                 if (diff_filespec_is_binary(options->repo, p->one) ||
5752                     diff_filespec_is_binary(options->repo, p->two)) {
5753                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5754                                         GIT_SHA1_HEXSZ);
5755                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5756                                         GIT_SHA1_HEXSZ);
5757                         continue;
5758                 }
5759
5760                 xpp.flags = 0;
5761                 xecfg.ctxlen = 3;
5762                 xecfg.flags = 0;
5763                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5764                                   &xpp, &xecfg))
5765                         return error("unable to generate patch-id diff for %s",
5766                                      p->one->path);
5767         }
5768
5769         git_SHA1_Final(oid->hash, &ctx);
5770         return 0;
5771 }
5772
5773 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5774 {
5775         struct diff_queue_struct *q = &diff_queued_diff;
5776         int i;
5777         int result = diff_get_patch_id(options, oid, diff_header_only);
5778
5779         for (i = 0; i < q->nr; i++)
5780                 diff_free_filepair(q->queue[i]);
5781
5782         free(q->queue);
5783         DIFF_QUEUE_CLEAR(q);
5784
5785         return result;
5786 }
5787
5788 static int is_summary_empty(const struct diff_queue_struct *q)
5789 {
5790         int i;
5791
5792         for (i = 0; i < q->nr; i++) {
5793                 const struct diff_filepair *p = q->queue[i];
5794
5795                 switch (p->status) {
5796                 case DIFF_STATUS_DELETED:
5797                 case DIFF_STATUS_ADDED:
5798                 case DIFF_STATUS_COPIED:
5799                 case DIFF_STATUS_RENAMED:
5800                         return 0;
5801                 default:
5802                         if (p->score)
5803                                 return 0;
5804                         if (p->one->mode && p->two->mode &&
5805                             p->one->mode != p->two->mode)
5806                                 return 0;
5807                         break;
5808                 }
5809         }
5810         return 1;
5811 }
5812
5813 static const char rename_limit_warning[] =
5814 N_("inexact rename detection was skipped due to too many files.");
5815
5816 static const char degrade_cc_to_c_warning[] =
5817 N_("only found copies from modified paths due to too many files.");
5818
5819 static const char rename_limit_advice[] =
5820 N_("you may want to set your %s variable to at least "
5821    "%d and retry the command.");
5822
5823 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5824 {
5825         fflush(stdout);
5826         if (degraded_cc)
5827                 warning(_(degrade_cc_to_c_warning));
5828         else if (needed)
5829                 warning(_(rename_limit_warning));
5830         else
5831                 return;
5832         if (0 < needed)
5833                 warning(_(rename_limit_advice), varname, needed);
5834 }
5835
5836 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5837 {
5838         int i;
5839         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5840         struct diff_queue_struct *q = &diff_queued_diff;
5841
5842         if (WSEH_NEW & WS_RULE_MASK)
5843                 BUG("WS rules bit mask overlaps with diff symbol flags");
5844
5845         if (o->color_moved)
5846                 o->emitted_symbols = &esm;
5847
5848         for (i = 0; i < q->nr; i++) {
5849                 struct diff_filepair *p = q->queue[i];
5850                 if (check_pair_status(p))
5851                         diff_flush_patch(p, o);
5852         }
5853
5854         if (o->emitted_symbols) {
5855                 if (o->color_moved) {
5856                         struct hashmap add_lines, del_lines;
5857
5858                         if (o->color_moved_ws_handling &
5859                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5860                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5861
5862                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5863                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5864
5865                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5866                         mark_color_as_moved(o, &add_lines, &del_lines);
5867                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5868                                 dim_moved_lines(o);
5869
5870                         hashmap_free(&add_lines, 0);
5871                         hashmap_free(&del_lines, 0);
5872                 }
5873
5874                 for (i = 0; i < esm.nr; i++)
5875                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5876
5877                 for (i = 0; i < esm.nr; i++)
5878                         free((void *)esm.buf[i].line);
5879         }
5880         esm.nr = 0;
5881 }
5882
5883 void diff_flush(struct diff_options *options)
5884 {
5885         struct diff_queue_struct *q = &diff_queued_diff;
5886         int i, output_format = options->output_format;
5887         int separator = 0;
5888         int dirstat_by_line = 0;
5889
5890         /*
5891          * Order: raw, stat, summary, patch
5892          * or:    name/name-status/checkdiff (other bits clear)
5893          */
5894         if (!q->nr)
5895                 goto free_queue;
5896
5897         if (output_format & (DIFF_FORMAT_RAW |
5898                              DIFF_FORMAT_NAME |
5899                              DIFF_FORMAT_NAME_STATUS |
5900                              DIFF_FORMAT_CHECKDIFF)) {
5901                 for (i = 0; i < q->nr; i++) {
5902                         struct diff_filepair *p = q->queue[i];
5903                         if (check_pair_status(p))
5904                                 flush_one_pair(p, options);
5905                 }
5906                 separator++;
5907         }
5908
5909         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5910                 dirstat_by_line = 1;
5911
5912         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5913             dirstat_by_line) {
5914                 struct diffstat_t diffstat;
5915
5916                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5917                 for (i = 0; i < q->nr; i++) {
5918                         struct diff_filepair *p = q->queue[i];
5919                         if (check_pair_status(p))
5920                                 diff_flush_stat(p, options, &diffstat);
5921                 }
5922                 if (output_format & DIFF_FORMAT_NUMSTAT)
5923                         show_numstat(&diffstat, options);
5924                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5925                         show_stats(&diffstat, options);
5926                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5927                         show_shortstats(&diffstat, options);
5928                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5929                         show_dirstat_by_line(&diffstat, options);
5930                 free_diffstat_info(&diffstat);
5931                 separator++;
5932         }
5933         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5934                 show_dirstat(options);
5935
5936         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5937                 for (i = 0; i < q->nr; i++) {
5938                         diff_summary(options, q->queue[i]);
5939                 }
5940                 separator++;
5941         }
5942
5943         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5944             options->flags.exit_with_status &&
5945             options->flags.diff_from_contents) {
5946                 /*
5947                  * run diff_flush_patch for the exit status. setting
5948                  * options->file to /dev/null should be safe, because we
5949                  * aren't supposed to produce any output anyway.
5950                  */
5951                 if (options->close_file)
5952                         fclose(options->file);
5953                 options->file = xfopen("/dev/null", "w");
5954                 options->close_file = 1;
5955                 options->color_moved = 0;
5956                 for (i = 0; i < q->nr; i++) {
5957                         struct diff_filepair *p = q->queue[i];
5958                         if (check_pair_status(p))
5959                                 diff_flush_patch(p, options);
5960                         if (options->found_changes)
5961                                 break;
5962                 }
5963         }
5964
5965         if (output_format & DIFF_FORMAT_PATCH) {
5966                 if (separator) {
5967                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5968                         if (options->stat_sep)
5969                                 /* attach patch instead of inline */
5970                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5971                                                  NULL, 0, 0);
5972                 }
5973
5974                 diff_flush_patch_all_file_pairs(options);
5975         }
5976
5977         if (output_format & DIFF_FORMAT_CALLBACK)
5978                 options->format_callback(q, options, options->format_callback_data);
5979
5980         for (i = 0; i < q->nr; i++)
5981                 diff_free_filepair(q->queue[i]);
5982 free_queue:
5983         free(q->queue);
5984         DIFF_QUEUE_CLEAR(q);
5985         if (options->close_file)
5986                 fclose(options->file);
5987
5988         /*
5989          * Report the content-level differences with HAS_CHANGES;
5990          * diff_addremove/diff_change does not set the bit when
5991          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5992          */
5993         if (options->flags.diff_from_contents) {
5994                 if (options->found_changes)
5995                         options->flags.has_changes = 1;
5996                 else
5997                         options->flags.has_changes = 0;
5998         }
5999 }
6000
6001 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6002 {
6003         return (((p->status == DIFF_STATUS_MODIFIED) &&
6004                  ((p->score &&
6005                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6006                   (!p->score &&
6007                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6008                 ((p->status != DIFF_STATUS_MODIFIED) &&
6009                  filter_bit_tst(p->status, options)));
6010 }
6011
6012 static void diffcore_apply_filter(struct diff_options *options)
6013 {
6014         int i;
6015         struct diff_queue_struct *q = &diff_queued_diff;
6016         struct diff_queue_struct outq;
6017
6018         DIFF_QUEUE_CLEAR(&outq);
6019
6020         if (!options->filter)
6021                 return;
6022
6023         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6024                 int found;
6025                 for (i = found = 0; !found && i < q->nr; i++) {
6026                         if (match_filter(options, q->queue[i]))
6027                                 found++;
6028                 }
6029                 if (found)
6030                         return;
6031
6032                 /* otherwise we will clear the whole queue
6033                  * by copying the empty outq at the end of this
6034                  * function, but first clear the current entries
6035                  * in the queue.
6036                  */
6037                 for (i = 0; i < q->nr; i++)
6038                         diff_free_filepair(q->queue[i]);
6039         }
6040         else {
6041                 /* Only the matching ones */
6042                 for (i = 0; i < q->nr; i++) {
6043                         struct diff_filepair *p = q->queue[i];
6044                         if (match_filter(options, p))
6045                                 diff_q(&outq, p);
6046                         else
6047                                 diff_free_filepair(p);
6048                 }
6049         }
6050         free(q->queue);
6051         *q = outq;
6052 }
6053
6054 /* Check whether two filespecs with the same mode and size are identical */
6055 static int diff_filespec_is_identical(struct repository *r,
6056                                       struct diff_filespec *one,
6057                                       struct diff_filespec *two)
6058 {
6059         if (S_ISGITLINK(one->mode))
6060                 return 0;
6061         if (diff_populate_filespec(r, one, 0))
6062                 return 0;
6063         if (diff_populate_filespec(r, two, 0))
6064                 return 0;
6065         return !memcmp(one->data, two->data, one->size);
6066 }
6067
6068 static int diff_filespec_check_stat_unmatch(struct repository *r,
6069                                             struct diff_filepair *p)
6070 {
6071         if (p->done_skip_stat_unmatch)
6072                 return p->skip_stat_unmatch_result;
6073
6074         p->done_skip_stat_unmatch = 1;
6075         p->skip_stat_unmatch_result = 0;
6076         /*
6077          * 1. Entries that come from stat info dirtiness
6078          *    always have both sides (iow, not create/delete),
6079          *    one side of the object name is unknown, with
6080          *    the same mode and size.  Keep the ones that
6081          *    do not match these criteria.  They have real
6082          *    differences.
6083          *
6084          * 2. At this point, the file is known to be modified,
6085          *    with the same mode and size, and the object
6086          *    name of one side is unknown.  Need to inspect
6087          *    the identical contents.
6088          */
6089         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6090             !DIFF_FILE_VALID(p->two) ||
6091             (p->one->oid_valid && p->two->oid_valid) ||
6092             (p->one->mode != p->two->mode) ||
6093             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6094             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6095             (p->one->size != p->two->size) ||
6096             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6097                 p->skip_stat_unmatch_result = 1;
6098         return p->skip_stat_unmatch_result;
6099 }
6100
6101 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6102 {
6103         int i;
6104         struct diff_queue_struct *q = &diff_queued_diff;
6105         struct diff_queue_struct outq;
6106         DIFF_QUEUE_CLEAR(&outq);
6107
6108         for (i = 0; i < q->nr; i++) {
6109                 struct diff_filepair *p = q->queue[i];
6110
6111                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6112                         diff_q(&outq, p);
6113                 else {
6114                         /*
6115                          * The caller can subtract 1 from skip_stat_unmatch
6116                          * to determine how many paths were dirty only
6117                          * due to stat info mismatch.
6118                          */
6119                         if (!diffopt->flags.no_index)
6120                                 diffopt->skip_stat_unmatch++;
6121                         diff_free_filepair(p);
6122                 }
6123         }
6124         free(q->queue);
6125         *q = outq;
6126 }
6127
6128 static int diffnamecmp(const void *a_, const void *b_)
6129 {
6130         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6131         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6132         const char *name_a, *name_b;
6133
6134         name_a = a->one ? a->one->path : a->two->path;
6135         name_b = b->one ? b->one->path : b->two->path;
6136         return strcmp(name_a, name_b);
6137 }
6138
6139 void diffcore_fix_diff_index(struct diff_options *options)
6140 {
6141         struct diff_queue_struct *q = &diff_queued_diff;
6142         QSORT(q->queue, q->nr, diffnamecmp);
6143 }
6144
6145 void diffcore_std(struct diff_options *options)
6146 {
6147         /* NOTE please keep the following in sync with diff_tree_combined() */
6148         if (options->skip_stat_unmatch)
6149                 diffcore_skip_stat_unmatch(options);
6150         if (!options->found_follow) {
6151                 /* See try_to_follow_renames() in tree-diff.c */
6152                 if (options->break_opt != -1)
6153                         diffcore_break(options->repo,
6154                                        options->break_opt);
6155                 if (options->detect_rename)
6156                         diffcore_rename(options);
6157                 if (options->break_opt != -1)
6158                         diffcore_merge_broken();
6159         }
6160         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6161                 diffcore_pickaxe(options);
6162         if (options->orderfile)
6163                 diffcore_order(options->orderfile);
6164         if (!options->found_follow)
6165                 /* See try_to_follow_renames() in tree-diff.c */
6166                 diff_resolve_rename_copy();
6167         diffcore_apply_filter(options);
6168
6169         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6170                 options->flags.has_changes = 1;
6171         else
6172                 options->flags.has_changes = 0;
6173
6174         options->found_follow = 0;
6175 }
6176
6177 int diff_result_code(struct diff_options *opt, int status)
6178 {
6179         int result = 0;
6180
6181         diff_warn_rename_limit("diff.renameLimit",
6182                                opt->needed_rename_limit,
6183                                opt->degraded_cc_to_c);
6184         if (!opt->flags.exit_with_status &&
6185             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6186                 return status;
6187         if (opt->flags.exit_with_status &&
6188             opt->flags.has_changes)
6189                 result |= 01;
6190         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6191             opt->flags.check_failed)
6192                 result |= 02;
6193         return result;
6194 }
6195
6196 int diff_can_quit_early(struct diff_options *opt)
6197 {
6198         return (opt->flags.quick &&
6199                 !opt->filter &&
6200                 opt->flags.has_changes);
6201 }
6202
6203 /*
6204  * Shall changes to this submodule be ignored?
6205  *
6206  * Submodule changes can be configured to be ignored separately for each path,
6207  * but that configuration can be overridden from the command line.
6208  */
6209 static int is_submodule_ignored(const char *path, struct diff_options *options)
6210 {
6211         int ignored = 0;
6212         struct diff_flags orig_flags = options->flags;
6213         if (!options->flags.override_submodule_config)
6214                 set_diffopt_flags_from_submodule_config(options, path);
6215         if (options->flags.ignore_submodules)
6216                 ignored = 1;
6217         options->flags = orig_flags;
6218         return ignored;
6219 }
6220
6221 void diff_addremove(struct diff_options *options,
6222                     int addremove, unsigned mode,
6223                     const struct object_id *oid,
6224                     int oid_valid,
6225                     const char *concatpath, unsigned dirty_submodule)
6226 {
6227         struct diff_filespec *one, *two;
6228
6229         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6230                 return;
6231
6232         /* This may look odd, but it is a preparation for
6233          * feeding "there are unchanged files which should
6234          * not produce diffs, but when you are doing copy
6235          * detection you would need them, so here they are"
6236          * entries to the diff-core.  They will be prefixed
6237          * with something like '=' or '*' (I haven't decided
6238          * which but should not make any difference).
6239          * Feeding the same new and old to diff_change()
6240          * also has the same effect.
6241          * Before the final output happens, they are pruned after
6242          * merged into rename/copy pairs as appropriate.
6243          */
6244         if (options->flags.reverse_diff)
6245                 addremove = (addremove == '+' ? '-' :
6246                              addremove == '-' ? '+' : addremove);
6247
6248         if (options->prefix &&
6249             strncmp(concatpath, options->prefix, options->prefix_length))
6250                 return;
6251
6252         one = alloc_filespec(concatpath);
6253         two = alloc_filespec(concatpath);
6254
6255         if (addremove != '+')
6256                 fill_filespec(one, oid, oid_valid, mode);
6257         if (addremove != '-') {
6258                 fill_filespec(two, oid, oid_valid, mode);
6259                 two->dirty_submodule = dirty_submodule;
6260         }
6261
6262         diff_queue(&diff_queued_diff, one, two);
6263         if (!options->flags.diff_from_contents)
6264                 options->flags.has_changes = 1;
6265 }
6266
6267 void diff_change(struct diff_options *options,
6268                  unsigned old_mode, unsigned new_mode,
6269                  const struct object_id *old_oid,
6270                  const struct object_id *new_oid,
6271                  int old_oid_valid, int new_oid_valid,
6272                  const char *concatpath,
6273                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6274 {
6275         struct diff_filespec *one, *two;
6276         struct diff_filepair *p;
6277
6278         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6279             is_submodule_ignored(concatpath, options))
6280                 return;
6281
6282         if (options->flags.reverse_diff) {
6283                 SWAP(old_mode, new_mode);
6284                 SWAP(old_oid, new_oid);
6285                 SWAP(old_oid_valid, new_oid_valid);
6286                 SWAP(old_dirty_submodule, new_dirty_submodule);
6287         }
6288
6289         if (options->prefix &&
6290             strncmp(concatpath, options->prefix, options->prefix_length))
6291                 return;
6292
6293         one = alloc_filespec(concatpath);
6294         two = alloc_filespec(concatpath);
6295         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6296         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6297         one->dirty_submodule = old_dirty_submodule;
6298         two->dirty_submodule = new_dirty_submodule;
6299         p = diff_queue(&diff_queued_diff, one, two);
6300
6301         if (options->flags.diff_from_contents)
6302                 return;
6303
6304         if (options->flags.quick && options->skip_stat_unmatch &&
6305             !diff_filespec_check_stat_unmatch(options->repo, p))
6306                 return;
6307
6308         options->flags.has_changes = 1;
6309 }
6310
6311 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6312 {
6313         struct diff_filepair *pair;
6314         struct diff_filespec *one, *two;
6315
6316         if (options->prefix &&
6317             strncmp(path, options->prefix, options->prefix_length))
6318                 return NULL;
6319
6320         one = alloc_filespec(path);
6321         two = alloc_filespec(path);
6322         pair = diff_queue(&diff_queued_diff, one, two);
6323         pair->is_unmerged = 1;
6324         return pair;
6325 }
6326
6327 static char *run_textconv(struct repository *r,
6328                           const char *pgm,
6329                           struct diff_filespec *spec,
6330                           size_t *outsize)
6331 {
6332         struct diff_tempfile *temp;
6333         const char *argv[3];
6334         const char **arg = argv;
6335         struct child_process child = CHILD_PROCESS_INIT;
6336         struct strbuf buf = STRBUF_INIT;
6337         int err = 0;
6338
6339         temp = prepare_temp_file(r, spec->path, spec);
6340         *arg++ = pgm;
6341         *arg++ = temp->name;
6342         *arg = NULL;
6343
6344         child.use_shell = 1;
6345         child.argv = argv;
6346         child.out = -1;
6347         if (start_command(&child)) {
6348                 remove_tempfile();
6349                 return NULL;
6350         }
6351
6352         if (strbuf_read(&buf, child.out, 0) < 0)
6353                 err = error("error reading from textconv command '%s'", pgm);
6354         close(child.out);
6355
6356         if (finish_command(&child) || err) {
6357                 strbuf_release(&buf);
6358                 remove_tempfile();
6359                 return NULL;
6360         }
6361         remove_tempfile();
6362
6363         return strbuf_detach(&buf, outsize);
6364 }
6365
6366 size_t fill_textconv(struct repository *r,
6367                      struct userdiff_driver *driver,
6368                      struct diff_filespec *df,
6369                      char **outbuf)
6370 {
6371         size_t size;
6372
6373         if (!driver) {
6374                 if (!DIFF_FILE_VALID(df)) {
6375                         *outbuf = "";
6376                         return 0;
6377                 }
6378                 if (diff_populate_filespec(r, df, 0))
6379                         die("unable to read files to diff");
6380                 *outbuf = df->data;
6381                 return df->size;
6382         }
6383
6384         if (!driver->textconv)
6385                 BUG("fill_textconv called with non-textconv driver");
6386
6387         if (driver->textconv_cache && df->oid_valid) {
6388                 *outbuf = notes_cache_get(driver->textconv_cache,
6389                                           &df->oid,
6390                                           &size);
6391                 if (*outbuf)
6392                         return size;
6393         }
6394
6395         *outbuf = run_textconv(r, driver->textconv, df, &size);
6396         if (!*outbuf)
6397                 die("unable to read files to diff");
6398
6399         if (driver->textconv_cache && df->oid_valid) {
6400                 /* ignore errors, as we might be in a readonly repository */
6401                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6402                                 size);
6403                 /*
6404                  * we could save up changes and flush them all at the end,
6405                  * but we would need an extra call after all diffing is done.
6406                  * Since generating a cache entry is the slow path anyway,
6407                  * this extra overhead probably isn't a big deal.
6408                  */
6409                 notes_cache_write(driver->textconv_cache);
6410         }
6411
6412         return size;
6413 }
6414
6415 int textconv_object(struct repository *r,
6416                     const char *path,
6417                     unsigned mode,
6418                     const struct object_id *oid,
6419                     int oid_valid,
6420                     char **buf,
6421                     unsigned long *buf_size)
6422 {
6423         struct diff_filespec *df;
6424         struct userdiff_driver *textconv;
6425
6426         df = alloc_filespec(path);
6427         fill_filespec(df, oid, oid_valid, mode);
6428         textconv = get_textconv(r->index, df);
6429         if (!textconv) {
6430                 free_filespec(df);
6431                 return 0;
6432         }
6433
6434         *buf_size = fill_textconv(r, textconv, df, buf);
6435         free_filespec(df);
6436         return 1;
6437 }
6438
6439 void setup_diff_pager(struct diff_options *opt)
6440 {
6441         /*
6442          * If the user asked for our exit code, then either they want --quiet
6443          * or --exit-code. We should definitely not bother with a pager in the
6444          * former case, as we will generate no output. Since we still properly
6445          * report our exit code even when a pager is run, we _could_ run a
6446          * pager with --exit-code. But since we have not done so historically,
6447          * and because it is easy to find people oneline advising "git diff
6448          * --exit-code" in hooks and other scripts, we do not do so.
6449          */
6450         if (!opt->flags.exit_with_status &&
6451             check_pager_config("diff") != 0)
6452                 setup_pager();
6453 }