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 unsigned 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                         ret |= COLOR_MOVED_WS_ERROR;
 317                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 318                 }
 319
 320                 strbuf_release(&sb);
 321         }
 322
 323         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 324             (ret & XDF_WHITESPACE_FLAGS)) {
 325                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
 326                 ret |= COLOR_MOVED_WS_ERROR;
 327         }
 328
 329         string_list_clear(&l, 0);
 330
 331         return ret;
 332 }
 333
 334 int git_diff_ui_config(const char *var, const char *value, void *cb)
 335 {
 336         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 337                 diff_use_color_default = git_config_colorbool(var, value);
 338                 return 0;
 339         }
 340         if (!strcmp(var, "diff.colormoved")) {
 341                 int cm = parse_color_moved(value);
 342                 if (cm < 0)
 343                         return -1;
 344                 diff_color_moved_default = cm;
 345                 return 0;
 346         }
 347         if (!strcmp(var, "diff.colormovedws")) {
 348                 unsigned cm = parse_color_moved_ws(value);
 349                 if (cm & COLOR_MOVED_WS_ERROR)
 350                         return -1;
 351                 diff_color_moved_ws_default = cm;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.context")) {
 355                 diff_context_default = git_config_int(var, value);
 356                 if (diff_context_default < 0)
 357                         return -1;
 358                 return 0;
 359         }
 360         if (!strcmp(var, "diff.interhunkcontext")) {
 361                 diff_interhunk_context_default = git_config_int(var, value);
 362                 if (diff_interhunk_context_default < 0)
 363                         return -1;
 364                 return 0;
 365         }
 366         if (!strcmp(var, "diff.renames")) {
 367                 diff_detect_rename_default = git_config_rename(var, value);
 368                 return 0;
 369         }
 370         if (!strcmp(var, "diff.autorefreshindex")) {
 371                 diff_auto_refresh_index = git_config_bool(var, value);
 372                 return 0;
 373         }
 374         if (!strcmp(var, "diff.mnemonicprefix")) {
 375                 diff_mnemonic_prefix = git_config_bool(var, value);
 376                 return 0;
 377         }
 378         if (!strcmp(var, "diff.noprefix")) {
 379                 diff_no_prefix = git_config_bool(var, value);
 380                 return 0;
 381         }
 382         if (!strcmp(var, "diff.statgraphwidth")) {
 383                 diff_stat_graph_width = git_config_int(var, value);
 384                 return 0;
 385         }
 386         if (!strcmp(var, "diff.external"))
 387                 return git_config_string(&external_diff_cmd_cfg, var, value);
 388         if (!strcmp(var, "diff.wordregex"))
 389                 return git_config_string(&diff_word_regex_cfg, var, value);
 390         if (!strcmp(var, "diff.orderfile"))
 391                 return git_config_pathname(&diff_order_file_cfg, var, value);
 392
 393         if (!strcmp(var, "diff.ignoresubmodules"))
 394                 handle_ignore_submodules_arg(&default_diff_options, value);
 395
 396         if (!strcmp(var, "diff.submodule")) {
 397                 if (parse_submodule_params(&default_diff_options, value))
 398                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 399                                 value);
 400                 return 0;
 401         }
 402
 403         if (!strcmp(var, "diff.algorithm")) {
 404                 diff_algorithm = parse_algorithm_value(value);
 405                 if (diff_algorithm < 0)
 406                         return -1;
 407                 return 0;
 408         }
 409
 410         if (!strcmp(var, "diff.wserrorhighlight")) {
 411                 int val = parse_ws_error_highlight(value);
 412                 if (val < 0)
 413                         return -1;
 414                 ws_error_highlight_default = val;
 415                 return 0;
 416         }
 417
 418         if (git_color_config(var, value, cb) < 0)
 419                 return -1;
 420
 421         return git_diff_basic_config(var, value, cb);
 422 }
 423
 424 int git_diff_basic_config(const char *var, const char *value, void *cb)
 425 {
 426         const char *name;
 427
 428         if (!strcmp(var, "diff.renamelimit")) {
 429                 diff_rename_limit_default = git_config_int(var, value);
 430                 return 0;
 431         }
 432
 433         if (userdiff_config(var, value) < 0)
 434                 return -1;
 435
 436         if (skip_prefix(var, "diff.color.", &name) ||
 437             skip_prefix(var, "color.diff.", &name)) {
 438                 int slot = parse_diff_color_slot(name);
 439                 if (slot < 0)
 440                         return 0;
 441                 if (!value)
 442                         return config_error_nonbool(var);
 443                 return color_parse(value, diff_colors[slot]);
 444         }
 445
 446         /* like GNU diff's --suppress-blank-empty option  */
 447         if (!strcmp(var, "diff.suppressblankempty") ||
 448                         /* for backwards compatibility */
 449                         !strcmp(var, "diff.suppress-blank-empty")) {
 450                 diff_suppress_blank_empty = git_config_bool(var, value);
 451                 return 0;
 452         }
 453
 454         if (!strcmp(var, "diff.dirstat")) {
 455                 struct strbuf errmsg = STRBUF_INIT;
 456                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 457                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 458                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 459                                 errmsg.buf);
 460                 strbuf_release(&errmsg);
 461                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 462                 return 0;
 463         }
 464
 465         if (git_diff_heuristic_config(var, value, cb) < 0)
 466                 return -1;
 467
 468         return git_default_config(var, value, cb);
 469 }
 470
 471 static char *quote_two(const char *one, const char *two)
 472 {
 473         int need_one = quote_c_style(one, NULL, NULL, 1);
 474         int need_two = quote_c_style(two, NULL, NULL, 1);
 475         struct strbuf res = STRBUF_INIT;
 476
 477         if (need_one + need_two) {
 478                 strbuf_addch(&res, '"');
 479                 quote_c_style(one, &res, NULL, 1);
 480                 quote_c_style(two, &res, NULL, 1);
 481                 strbuf_addch(&res, '"');
 482         } else {
 483                 strbuf_addstr(&res, one);
 484                 strbuf_addstr(&res, two);
 485         }
 486         return strbuf_detach(&res, NULL);
 487 }
 488
 489 static const char *external_diff(void)
 490 {
 491         static const char *external_diff_cmd = NULL;
 492         static int done_preparing = 0;
 493
 494         if (done_preparing)
 495                 return external_diff_cmd;
 496         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 497         if (!external_diff_cmd)
 498                 external_diff_cmd = external_diff_cmd_cfg;
 499         done_preparing = 1;
 500         return external_diff_cmd;
 501 }
 502
 503 /*
 504  * Keep track of files used for diffing. Sometimes such an entry
 505  * refers to a temporary file, sometimes to an existing file, and
 506  * sometimes to "/dev/null".
 507  */
 508 static struct diff_tempfile {
 509         /*
 510          * filename external diff should read from, or NULL if this
 511          * entry is currently not in use:
 512          */
 513         const char *name;
 514
 515         char hex[GIT_MAX_HEXSZ + 1];
 516         char mode[10];
 517
 518         /*
 519          * If this diff_tempfile instance refers to a temporary file,
 520          * this tempfile object is used to manage its lifetime.
 521          */
 522         struct tempfile *tempfile;
 523 } diff_temp[2];
 524
 525 struct emit_callback {
 526         int color_diff;
 527         unsigned ws_rule;
 528         int blank_at_eof_in_preimage;
 529         int blank_at_eof_in_postimage;
 530         int lno_in_preimage;
 531         int lno_in_postimage;
 532         const char **label_path;
 533         struct diff_words_data *diff_words;
 534         struct diff_options *opt;
 535         struct strbuf *header;
 536 };
 537
 538 static int count_lines(const char *data, int size)
 539 {
 540         int count, ch, completely_empty = 1, nl_just_seen = 0;
 541         count = 0;
 542         while (0 < size--) {
 543                 ch = *data++;
 544                 if (ch == '\n') {
 545                         count++;
 546                         nl_just_seen = 1;
 547                         completely_empty = 0;
 548                 }
 549                 else {
 550                         nl_just_seen = 0;
 551                         completely_empty = 0;
 552                 }
 553         }
 554         if (completely_empty)
 555                 return 0;
 556         if (!nl_just_seen)
 557                 count++; /* no trailing newline */
 558         return count;
 559 }
 560
 561 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 562                        struct diff_filespec *one)
 563 {
 564         if (!DIFF_FILE_VALID(one)) {
 565                 mf->ptr = (char *)""; /* does not matter */
 566                 mf->size = 0;
 567                 return 0;
 568         }
 569         else if (diff_populate_filespec(r, one, 0))
 570                 return -1;
 571
 572         mf->ptr = one->data;
 573         mf->size = one->size;
 574         return 0;
 575 }
 576
 577 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 578 static unsigned long diff_filespec_size(struct repository *r,
 579                                         struct diff_filespec *one)
 580 {
 581         if (!DIFF_FILE_VALID(one))
 582                 return 0;
 583         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 584         return one->size;
 585 }
 586
 587 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 588 {
 589         char *ptr = mf->ptr;
 590         long size = mf->size;
 591         int cnt = 0;
 592
 593         if (!size)
 594                 return cnt;
 595         ptr += size - 1; /* pointing at the very end */
 596         if (*ptr != '\n')
 597                 ; /* incomplete line */
 598         else
 599                 ptr--; /* skip the last LF */
 600         while (mf->ptr < ptr) {
 601                 char *prev_eol;
 602                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 603                         if (*prev_eol == '\n')
 604                                 break;
 605                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 606                         break;
 607                 cnt++;
 608                 ptr = prev_eol - 1;
 609         }
 610         return cnt;
 611 }
 612
 613 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 614                                struct emit_callback *ecbdata)
 615 {
 616         int l1, l2, at;
 617         unsigned ws_rule = ecbdata->ws_rule;
 618         l1 = count_trailing_blank(mf1, ws_rule);
 619         l2 = count_trailing_blank(mf2, ws_rule);
 620         if (l2 <= l1) {
 621                 ecbdata->blank_at_eof_in_preimage = 0;
 622                 ecbdata->blank_at_eof_in_postimage = 0;
 623                 return;
 624         }
 625         at = count_lines(mf1->ptr, mf1->size);
 626         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 627
 628         at = count_lines(mf2->ptr, mf2->size);
 629         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 630 }
 631
 632 static void emit_line_0(struct diff_options *o,
 633                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 634                         int first, const char *line, int len)
 635 {
 636         int has_trailing_newline, has_trailing_carriage_return;
 637         int needs_reset = 0; /* at the end of the line */
 638         FILE *file = o->file;
 639
 640         fputs(diff_line_prefix(o), file);
 641
 642         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 643         if (has_trailing_newline)
 644                 len--;
 645
 646         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 647         if (has_trailing_carriage_return)
 648                 len--;
 649
 650         if (!len && !first)
 651                 goto end_of_line;
 652
 653         if (reverse && want_color(o->use_color)) {
 654                 fputs(GIT_COLOR_REVERSE, file);
 655                 needs_reset = 1;
 656         }
 657
 658         if (set_sign) {
 659                 fputs(set_sign, file);
 660                 needs_reset = 1;
 661         }
 662
 663         if (first)
 664                 fputc(first, file);
 665
 666         if (!len)
 667                 goto end_of_line;
 668
 669         if (set) {
 670                 if (set_sign && set != set_sign)
 671                         fputs(reset, file);
 672                 fputs(set, file);
 673                 needs_reset = 1;
 674         }
 675         fwrite(line, len, 1, file);
 676         needs_reset = 1; /* 'line' may contain color codes. */
 677
 678 end_of_line:
 679         if (needs_reset)
 680                 fputs(reset, file);
 681         if (has_trailing_carriage_return)
 682                 fputc('\r', file);
 683         if (has_trailing_newline)
 684                 fputc('\n', file);
 685 }
 686
 687 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 688                       const char *line, int len)
 689 {
 690         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 691 }
 692
 693 enum diff_symbol {
 694         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 695         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 696         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 697         DIFF_SYMBOL_BINARY_DIFF_BODY,
 698         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 699         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 700         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 701         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 702         DIFF_SYMBOL_STATS_LINE,
 703         DIFF_SYMBOL_WORD_DIFF,
 704         DIFF_SYMBOL_STAT_SEP,
 705         DIFF_SYMBOL_SUMMARY,
 706         DIFF_SYMBOL_SUBMODULE_ADD,
 707         DIFF_SYMBOL_SUBMODULE_DEL,
 708         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 709         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 710         DIFF_SYMBOL_SUBMODULE_HEADER,
 711         DIFF_SYMBOL_SUBMODULE_ERROR,
 712         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 713         DIFF_SYMBOL_REWRITE_DIFF,
 714         DIFF_SYMBOL_BINARY_FILES,
 715         DIFF_SYMBOL_HEADER,
 716         DIFF_SYMBOL_FILEPAIR_PLUS,
 717         DIFF_SYMBOL_FILEPAIR_MINUS,
 718         DIFF_SYMBOL_WORDS_PORCELAIN,
 719         DIFF_SYMBOL_WORDS,
 720         DIFF_SYMBOL_CONTEXT,
 721         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 722         DIFF_SYMBOL_PLUS,
 723         DIFF_SYMBOL_MINUS,
 724         DIFF_SYMBOL_NO_LF_EOF,
 725         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 726         DIFF_SYMBOL_CONTEXT_MARKER,
 727         DIFF_SYMBOL_SEPARATOR
 728 };
 729 /*
 730  * Flags for content lines:
 731  * 0..12 are whitespace rules
 732  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 733  * 16 is marking if the line is blank at EOF
 734  */
 735 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 736 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 737 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 738 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 739 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 740
 741 /*
 742  * This struct is used when we need to buffer the output of the diff output.
 743  *
 744  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 745  * into the pre/post image file. This pointer could be a union with the
 746  * line pointer. By storing an offset into the file instead of the literal line,
 747  * we can decrease the memory footprint for the buffered output. At first we
 748  * may want to only have indirection for the content lines, but we could also
 749  * enhance the state for emitting prefabricated lines, e.g. the similarity
 750  * score line or hunk/file headers would only need to store a number or path
 751  * and then the output can be constructed later on depending on state.
 752  */
 753 struct emitted_diff_symbol {
 754         const char *line;
 755         int len;
 756         int flags;
 757         enum diff_symbol s;
 758 };
 759 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 760
 761 struct emitted_diff_symbols {
 762         struct emitted_diff_symbol *buf;
 763         int nr, alloc;
 764 };
 765 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 766
 767 static void append_emitted_diff_symbol(struct diff_options *o,
 768                                        struct emitted_diff_symbol *e)
 769 {
 770         struct emitted_diff_symbol *f;
 771
 772         ALLOC_GROW(o->emitted_symbols->buf,
 773                    o->emitted_symbols->nr + 1,
 774                    o->emitted_symbols->alloc);
 775         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 776
 777         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 778         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 779 }
 780
 781 struct moved_entry {
 782         struct hashmap_entry ent;
 783         const struct emitted_diff_symbol *es;
 784         struct moved_entry *next_line;
 785 };
 786
 787 /**
 788  * The struct ws_delta holds white space differences between moved lines, i.e.
 789  * between '+' and '-' lines that have been detected to be a move.
 790  * The string contains the difference in leading white spaces, before the
 791  * rest of the line is compared using the white space config for move
 792  * coloring. The current_longer indicates if the first string in the
 793  * comparision is longer than the second.
 794  */
 795 struct ws_delta {
 796         char *string;
 797         unsigned int current_longer : 1;
 798 };
 799 #define WS_DELTA_INIT { NULL, 0 }
 800
 801 struct moved_block {
 802         struct moved_entry *match;
 803         struct ws_delta wsd;
 804 };
 805
 806 static void moved_block_clear(struct moved_block *b)
 807 {
 808         FREE_AND_NULL(b->wsd.string);
 809         b->match = NULL;
 810 }
 811
 812 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 813                              const struct emitted_diff_symbol *b,
 814                              struct ws_delta *out)
 815 {
 816         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 817         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 818         int d = longer->len - shorter->len;
 819
 820         if (strncmp(longer->line + d, shorter->line, shorter->len))
 821                 return 0;
 822
 823         out->string = xmemdupz(longer->line, d);
 824         out->current_longer = (a == longer);
 825
 826         return 1;
 827 }
 828
 829 static int cmp_in_block_with_wsd(const struct diff_options *o,
 830                                  const struct moved_entry *cur,
 831                                  const struct moved_entry *match,
 832                                  struct moved_block *pmb,
 833                                  int n)
 834 {
 835         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 836         int al = cur->es->len, cl = l->len;
 837         const char *a = cur->es->line,
 838                    *b = match->es->line,
 839                    *c = l->line;
 840
 841         int wslen;
 842
 843         /*
 844          * We need to check if 'cur' is equal to 'match'.
 845          * As those are from the same (+/-) side, we do not need to adjust for
 846          * indent changes. However these were found using fuzzy matching
 847          * so we do have to check if they are equal.
 848          */
 849         if (strcmp(a, b))
 850                 return 1;
 851
 852         if (!pmb->wsd.string)
 853                 /*
 854                  * The white space delta is not active? This can happen
 855                  * when we exit early in this function.
 856                  */
 857                 return 1;
 858
 859         /*
 860          * The indent changes of the block are known and stored in
 861          * pmb->wsd; however we need to check if the indent changes of the
 862          * current line are still the same as before.
 863          *
 864          * To do so we need to compare 'l' to 'cur', adjusting the
 865          * one of them for the white spaces, depending which was longer.
 866          */
 867
 868         wslen = strlen(pmb->wsd.string);
 869         if (pmb->wsd.current_longer) {
 870                 c += wslen;
 871                 cl -= wslen;
 872         } else {
 873                 a += wslen;
 874                 al -= wslen;
 875         }
 876
 877         if (al != cl || memcmp(a, c, al))
 878                 return 1;
 879
 880         return 0;
 881 }
 882
 883 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 884                            const void *entry,
 885                            const void *entry_or_key,
 886                            const void *keydata)
 887 {
 888         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 889         const struct moved_entry *a = entry;
 890         const struct moved_entry *b = entry_or_key;
 891         unsigned flags = diffopt->color_moved_ws_handling
 892                          & XDF_WHITESPACE_FLAGS;
 893
 894         if (diffopt->color_moved_ws_handling &
 895             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 896                 /*
 897                  * As there is not specific white space config given,
 898                  * we'd need to check for a new block, so ignore all
 899                  * white space. The setup of the white space
 900                  * configuration for the next block is done else where
 901                  */
 902                 flags |= XDF_IGNORE_WHITESPACE;
 903
 904         return !xdiff_compare_lines(a->es->line, a->es->len,
 905                                     b->es->line, b->es->len,
 906                                     flags);
 907 }
 908
 909 static struct moved_entry *prepare_entry(struct diff_options *o,
 910                                          int line_no)
 911 {
 912         struct moved_entry *ret = xmalloc(sizeof(*ret));
 913         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 914         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 915
 916         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 917         ret->es = l;
 918         ret->next_line = NULL;
 919
 920         return ret;
 921 }
 922
 923 static void add_lines_to_move_detection(struct diff_options *o,
 924                                         struct hashmap *add_lines,
 925                                         struct hashmap *del_lines)
 926 {
 927         struct moved_entry *prev_line = NULL;
 928
 929         int n;
 930         for (n = 0; n < o->emitted_symbols->nr; n++) {
 931                 struct hashmap *hm;
 932                 struct moved_entry *key;
 933
 934                 switch (o->emitted_symbols->buf[n].s) {
 935                 case DIFF_SYMBOL_PLUS:
 936                         hm = add_lines;
 937                         break;
 938                 case DIFF_SYMBOL_MINUS:
 939                         hm = del_lines;
 940                         break;
 941                 default:
 942                         prev_line = NULL;
 943                         continue;
 944                 }
 945
 946                 key = prepare_entry(o, n);
 947                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 948                         prev_line->next_line = key;
 949
 950                 hashmap_add(hm, key);
 951                 prev_line = key;
 952         }
 953 }
 954
 955 static void pmb_advance_or_null(struct diff_options *o,
 956                                 struct moved_entry *match,
 957                                 struct hashmap *hm,
 958                                 struct moved_block *pmb,
 959                                 int pmb_nr)
 960 {
 961         int i;
 962         for (i = 0; i < pmb_nr; i++) {
 963                 struct moved_entry *prev = pmb[i].match;
 964                 struct moved_entry *cur = (prev && prev->next_line) ?
 965                                 prev->next_line : NULL;
 966                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 967                         pmb[i].match = cur;
 968                 } else {
 969                         pmb[i].match = NULL;
 970                 }
 971         }
 972 }
 973
 974 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 975                                             struct moved_entry *match,
 976                                             struct hashmap *hm,
 977                                             struct moved_block *pmb,
 978                                             int pmb_nr, int n)
 979 {
 980         int i;
 981         char *got_match = xcalloc(1, pmb_nr);
 982
 983         for (; match; match = hashmap_get_next(hm, match)) {
 984                 for (i = 0; i < pmb_nr; i++) {
 985                         struct moved_entry *prev = pmb[i].match;
 986                         struct moved_entry *cur = (prev && prev->next_line) ?
 987                                         prev->next_line : NULL;
 988                         if (!cur)
 989                                 continue;
 990                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
 991                                 got_match[i] |= 1;
 992                 }
 993         }
 994
 995         for (i = 0; i < pmb_nr; i++) {
 996                 if (got_match[i]) {
 997                         /* Advance to the next line */
 998                         pmb[i].match = pmb[i].match->next_line;
 999                 } else {
1000                         moved_block_clear(&pmb[i]);
1001                 }
1002         }
1003
1004         free(got_match);
1005 }
1006
1007 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1008                                          int pmb_nr)
1009 {
1010         int lp, rp;
1011
1012         /* Shrink the set of potential block to the remaining running */
1013         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1014                 while (lp < pmb_nr && pmb[lp].match)
1015                         lp++;
1016                 /* lp points at the first NULL now */
1017
1018                 while (rp > -1 && !pmb[rp].match)
1019                         rp--;
1020                 /* rp points at the last non-NULL */
1021
1022                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1023                         pmb[lp] = pmb[rp];
1024                         pmb[rp].match = NULL;
1025                         pmb[rp].wsd.string = NULL;
1026                         rp--;
1027                         lp++;
1028                 }
1029         }
1030
1031         /* Remember the number of running sets */
1032         return rp + 1;
1033 }
1034
1035 /*
1036  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1037  *
1038  * Otherwise, if the last block has fewer alphanumeric characters than
1039  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1040  * that block.
1041  *
1042  * The last block consists of the (n - block_length)'th line up to but not
1043  * including the nth line.
1044  *
1045  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1046  * Think of a way to unify them.
1047  */
1048 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1049 {
1050         int i, alnum_count = 0;
1051         if (o->color_moved == COLOR_MOVED_PLAIN)
1052                 return;
1053         for (i = 1; i < block_length + 1; i++) {
1054                 const char *c = o->emitted_symbols->buf[n - i].line;
1055                 for (; *c; c++) {
1056                         if (!isalnum(*c))
1057                                 continue;
1058                         alnum_count++;
1059                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1060                                 return;
1061                 }
1062         }
1063         for (i = 1; i < block_length + 1; i++)
1064                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1065 }
1066
1067 /* Find blocks of moved code, delegate actual coloring decision to helper */
1068 static void mark_color_as_moved(struct diff_options *o,
1069                                 struct hashmap *add_lines,
1070                                 struct hashmap *del_lines)
1071 {
1072         struct moved_block *pmb = NULL; /* potentially moved blocks */
1073         int pmb_nr = 0, pmb_alloc = 0;
1074         int n, flipped_block = 1, block_length = 0;
1075
1076
1077         for (n = 0; n < o->emitted_symbols->nr; n++) {
1078                 struct hashmap *hm = NULL;
1079                 struct moved_entry *key;
1080                 struct moved_entry *match = NULL;
1081                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1082
1083                 switch (l->s) {
1084                 case DIFF_SYMBOL_PLUS:
1085                         hm = del_lines;
1086                         key = prepare_entry(o, n);
1087                         match = hashmap_get(hm, key, NULL);
1088                         free(key);
1089                         break;
1090                 case DIFF_SYMBOL_MINUS:
1091                         hm = add_lines;
1092                         key = prepare_entry(o, n);
1093                         match = hashmap_get(hm, key, NULL);
1094                         free(key);
1095                         break;
1096                 default:
1097                         flipped_block = 1;
1098                 }
1099
1100                 if (!match) {
1101                         int i;
1102
1103                         adjust_last_block(o, n, block_length);
1104                         for(i = 0; i < pmb_nr; i++)
1105                                 moved_block_clear(&pmb[i]);
1106                         pmb_nr = 0;
1107                         block_length = 0;
1108                         continue;
1109                 }
1110
1111                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1112
1113                 if (o->color_moved == COLOR_MOVED_PLAIN)
1114                         continue;
1115
1116                 if (o->color_moved_ws_handling &
1117                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1118                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1119                 else
1120                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1121
1122                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1123
1124                 if (pmb_nr == 0) {
1125                         /*
1126                          * The current line is the start of a new block.
1127                          * Setup the set of potential blocks.
1128                          */
1129                         for (; match; match = hashmap_get_next(hm, match)) {
1130                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1131                                 if (o->color_moved_ws_handling &
1132                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1133                                         if (compute_ws_delta(l, match->es,
1134                                                              &pmb[pmb_nr].wsd))
1135                                                 pmb[pmb_nr++].match = match;
1136                                 } else {
1137                                         pmb[pmb_nr].wsd.string = NULL;
1138                                         pmb[pmb_nr++].match = match;
1139                                 }
1140                         }
1141
1142                         flipped_block = (flipped_block + 1) % 2;
1143
1144                         adjust_last_block(o, n, block_length);
1145                         block_length = 0;
1146                 }
1147
1148                 block_length++;
1149
1150                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1151                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1152         }
1153         adjust_last_block(o, n, block_length);
1154
1155         for(n = 0; n < pmb_nr; n++)
1156                 moved_block_clear(&pmb[n]);
1157         free(pmb);
1158 }
1159
1160 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1161   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1162 static void dim_moved_lines(struct diff_options *o)
1163 {
1164         int n;
1165         for (n = 0; n < o->emitted_symbols->nr; n++) {
1166                 struct emitted_diff_symbol *prev = (n != 0) ?
1167                                 &o->emitted_symbols->buf[n - 1] : NULL;
1168                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1169                 struct emitted_diff_symbol *next =
1170                                 (n < o->emitted_symbols->nr - 1) ?
1171                                 &o->emitted_symbols->buf[n + 1] : NULL;
1172
1173                 /* Not a plus or minus line? */
1174                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1175                         continue;
1176
1177                 /* Not a moved line? */
1178                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1179                         continue;
1180
1181                 /*
1182                  * If prev or next are not a plus or minus line,
1183                  * pretend they don't exist
1184                  */
1185                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1186                             prev->s != DIFF_SYMBOL_MINUS)
1187                         prev = NULL;
1188                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1189                             next->s != DIFF_SYMBOL_MINUS)
1190                         next = NULL;
1191
1192                 /* Inside a block? */
1193                 if ((prev &&
1194                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1195                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1196                     (next &&
1197                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1198                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1199                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1200                         continue;
1201                 }
1202
1203                 /* Check if we are at an interesting bound: */
1204                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1205                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1206                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1207                         continue;
1208                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1209                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1210                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1211                         continue;
1212
1213                 /*
1214                  * The boundary to prev and next are not interesting,
1215                  * so this line is not interesting as a whole
1216                  */
1217                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1218         }
1219 }
1220
1221 static void emit_line_ws_markup(struct diff_options *o,
1222                                 const char *set_sign, const char *set,
1223                                 const char *reset,
1224                                 int sign_index, const char *line, int len,
1225                                 unsigned ws_rule, int blank_at_eof)
1226 {
1227         const char *ws = NULL;
1228         int sign = o->output_indicators[sign_index];
1229
1230         if (o->ws_error_highlight & ws_rule) {
1231                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1232                 if (!*ws)
1233                         ws = NULL;
1234         }
1235
1236         if (!ws && !set_sign)
1237                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1238         else if (!ws) {
1239                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1240         } else if (blank_at_eof)
1241                 /* Blank line at EOF - paint '+' as well */
1242                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1243         else {
1244                 /* Emit just the prefix, then the rest. */
1245                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1246                             sign, "", 0);
1247                 ws_check_emit(line, len, ws_rule,
1248                               o->file, set, reset, ws);
1249         }
1250 }
1251
1252 static void emit_diff_symbol_from_struct(struct diff_options *o,
1253                                          struct emitted_diff_symbol *eds)
1254 {
1255         static const char *nneof = " No newline at end of file\n";
1256         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1257         struct strbuf sb = STRBUF_INIT;
1258
1259         enum diff_symbol s = eds->s;
1260         const char *line = eds->line;
1261         int len = eds->len;
1262         unsigned flags = eds->flags;
1263
1264         switch (s) {
1265         case DIFF_SYMBOL_NO_LF_EOF:
1266                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1267                 reset = diff_get_color_opt(o, DIFF_RESET);
1268                 putc('\n', o->file);
1269                 emit_line_0(o, context, NULL, 0, reset, '\\',
1270                             nneof, strlen(nneof));
1271                 break;
1272         case DIFF_SYMBOL_SUBMODULE_HEADER:
1273         case DIFF_SYMBOL_SUBMODULE_ERROR:
1274         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1275         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1276         case DIFF_SYMBOL_SUMMARY:
1277         case DIFF_SYMBOL_STATS_LINE:
1278         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1279         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1280                 emit_line(o, "", "", line, len);
1281                 break;
1282         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1283         case DIFF_SYMBOL_CONTEXT_MARKER:
1284                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1285                 reset = diff_get_color_opt(o, DIFF_RESET);
1286                 emit_line(o, context, reset, line, len);
1287                 break;
1288         case DIFF_SYMBOL_SEPARATOR:
1289                 fprintf(o->file, "%s%c",
1290                         diff_line_prefix(o),
1291                         o->line_termination);
1292                 break;
1293         case DIFF_SYMBOL_CONTEXT:
1294                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1295                 reset = diff_get_color_opt(o, DIFF_RESET);
1296                 set_sign = NULL;
1297                 if (o->flags.dual_color_diffed_diffs) {
1298                         char c = !len ? 0 : line[0];
1299
1300                         if (c == '+')
1301                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1302                         else if (c == '@')
1303                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1304                         else if (c == '-')
1305                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1306                 }
1307                 emit_line_ws_markup(o, set_sign, set, reset,
1308                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1309                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1310                 break;
1311         case DIFF_SYMBOL_PLUS:
1312                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1313                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1314                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1315                 case DIFF_SYMBOL_MOVED_LINE |
1316                      DIFF_SYMBOL_MOVED_LINE_ALT |
1317                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1318                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1319                         break;
1320                 case DIFF_SYMBOL_MOVED_LINE |
1321                      DIFF_SYMBOL_MOVED_LINE_ALT:
1322                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1323                         break;
1324                 case DIFF_SYMBOL_MOVED_LINE |
1325                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1326                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1327                         break;
1328                 case DIFF_SYMBOL_MOVED_LINE:
1329                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1330                         break;
1331                 default:
1332                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1333                 }
1334                 reset = diff_get_color_opt(o, DIFF_RESET);
1335                 if (!o->flags.dual_color_diffed_diffs)
1336                         set_sign = NULL;
1337                 else {
1338                         char c = !len ? 0 : line[0];
1339
1340                         set_sign = set;
1341                         if (c == '-')
1342                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1343                         else if (c == '@')
1344                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1345                         else if (c == '+')
1346                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1347                         else
1348                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1349                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1350                 }
1351                 emit_line_ws_markup(o, set_sign, set, reset,
1352                                     OUTPUT_INDICATOR_NEW, line, len,
1353                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1354                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1355                 break;
1356         case DIFF_SYMBOL_MINUS:
1357                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1358                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1359                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1360                 case DIFF_SYMBOL_MOVED_LINE |
1361                      DIFF_SYMBOL_MOVED_LINE_ALT |
1362                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1363                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1364                         break;
1365                 case DIFF_SYMBOL_MOVED_LINE |
1366                      DIFF_SYMBOL_MOVED_LINE_ALT:
1367                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1368                         break;
1369                 case DIFF_SYMBOL_MOVED_LINE |
1370                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1371                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1372                         break;
1373                 case DIFF_SYMBOL_MOVED_LINE:
1374                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1375                         break;
1376                 default:
1377                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1378                 }
1379                 reset = diff_get_color_opt(o, DIFF_RESET);
1380                 if (!o->flags.dual_color_diffed_diffs)
1381                         set_sign = NULL;
1382                 else {
1383                         char c = !len ? 0 : line[0];
1384
1385                         set_sign = set;
1386                         if (c == '+')
1387                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1388                         else if (c == '@')
1389                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1390                         else if (c == '-')
1391                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1392                         else
1393                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1394                 }
1395                 emit_line_ws_markup(o, set_sign, set, reset,
1396                                     OUTPUT_INDICATOR_OLD, line, len,
1397                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1398                 break;
1399         case DIFF_SYMBOL_WORDS_PORCELAIN:
1400                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1401                 reset = diff_get_color_opt(o, DIFF_RESET);
1402                 emit_line(o, context, reset, line, len);
1403                 fputs("~\n", o->file);
1404                 break;
1405         case DIFF_SYMBOL_WORDS:
1406                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1407                 reset = diff_get_color_opt(o, DIFF_RESET);
1408                 /*
1409                  * Skip the prefix character, if any.  With
1410                  * diff_suppress_blank_empty, there may be
1411                  * none.
1412                  */
1413                 if (line[0] != '\n') {
1414                         line++;
1415                         len--;
1416                 }
1417                 emit_line(o, context, reset, line, len);
1418                 break;
1419         case DIFF_SYMBOL_FILEPAIR_PLUS:
1420                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1421                 reset = diff_get_color_opt(o, DIFF_RESET);
1422                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1423                         line, reset,
1424                         strchr(line, ' ') ? "\t" : "");
1425                 break;
1426         case DIFF_SYMBOL_FILEPAIR_MINUS:
1427                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1428                 reset = diff_get_color_opt(o, DIFF_RESET);
1429                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1430                         line, reset,
1431                         strchr(line, ' ') ? "\t" : "");
1432                 break;
1433         case DIFF_SYMBOL_BINARY_FILES:
1434         case DIFF_SYMBOL_HEADER:
1435                 fprintf(o->file, "%s", line);
1436                 break;
1437         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1438                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1439                 break;
1440         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1441                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1442                 break;
1443         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1444                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1445                 break;
1446         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1447                 fputs(diff_line_prefix(o), o->file);
1448                 fputc('\n', o->file);
1449                 break;
1450         case DIFF_SYMBOL_REWRITE_DIFF:
1451                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1452                 reset = diff_get_color_opt(o, DIFF_RESET);
1453                 emit_line(o, fraginfo, reset, line, len);
1454                 break;
1455         case DIFF_SYMBOL_SUBMODULE_ADD:
1456                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1457                 reset = diff_get_color_opt(o, DIFF_RESET);
1458                 emit_line(o, set, reset, line, len);
1459                 break;
1460         case DIFF_SYMBOL_SUBMODULE_DEL:
1461                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1462                 reset = diff_get_color_opt(o, DIFF_RESET);
1463                 emit_line(o, set, reset, line, len);
1464                 break;
1465         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1466                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1467                         diff_line_prefix(o), line);
1468                 break;
1469         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1470                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1471                         diff_line_prefix(o), line);
1472                 break;
1473         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1474                 emit_line(o, "", "", " 0 files changed\n",
1475                           strlen(" 0 files changed\n"));
1476                 break;
1477         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1478                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1479                 break;
1480         case DIFF_SYMBOL_WORD_DIFF:
1481                 fprintf(o->file, "%.*s", len, line);
1482                 break;
1483         case DIFF_SYMBOL_STAT_SEP:
1484                 fputs(o->stat_sep, o->file);
1485                 break;
1486         default:
1487                 BUG("unknown diff symbol");
1488         }
1489         strbuf_release(&sb);
1490 }
1491
1492 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1493                              const char *line, int len, unsigned flags)
1494 {
1495         struct emitted_diff_symbol e = {line, len, flags, s};
1496
1497         if (o->emitted_symbols)
1498                 append_emitted_diff_symbol(o, &e);
1499         else
1500                 emit_diff_symbol_from_struct(o, &e);
1501 }
1502
1503 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1504 {
1505         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1506 }
1507
1508 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1509 {
1510         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1511 }
1512
1513 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1514 {
1515         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1516                          path, strlen(path), 0);
1517 }
1518
1519 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1520 {
1521         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1522                          path, strlen(path), 0);
1523 }
1524
1525 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1526 {
1527         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1528                          header, strlen(header), 0);
1529 }
1530
1531 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1532 {
1533         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1534 }
1535
1536 void diff_emit_submodule_pipethrough(struct diff_options *o,
1537                                      const char *line, int len)
1538 {
1539         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1540 }
1541
1542 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1543 {
1544         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1545               ecbdata->blank_at_eof_in_preimage &&
1546               ecbdata->blank_at_eof_in_postimage &&
1547               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1548               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1549                 return 0;
1550         return ws_blank_line(line, len, ecbdata->ws_rule);
1551 }
1552
1553 static void emit_add_line(const char *reset,
1554                           struct emit_callback *ecbdata,
1555                           const char *line, int len)
1556 {
1557         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1558         if (new_blank_line_at_eof(ecbdata, line, len))
1559                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1560
1561         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1562 }
1563
1564 static void emit_del_line(const char *reset,
1565                           struct emit_callback *ecbdata,
1566                           const char *line, int len)
1567 {
1568         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1569         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1570 }
1571
1572 static void emit_context_line(const char *reset,
1573                               struct emit_callback *ecbdata,
1574                               const char *line, int len)
1575 {
1576         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1577         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1578 }
1579
1580 static void emit_hunk_header(struct emit_callback *ecbdata,
1581                              const char *line, int len)
1582 {
1583         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1584         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1585         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1586         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1587         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1588         static const char atat[2] = { '@', '@' };
1589         const char *cp, *ep;
1590         struct strbuf msgbuf = STRBUF_INIT;
1591         int org_len = len;
1592         int i = 1;
1593
1594         /*
1595          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1596          * it always is at least 10 bytes long.
1597          */
1598         if (len < 10 ||
1599             memcmp(line, atat, 2) ||
1600             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1601                 emit_diff_symbol(ecbdata->opt,
1602                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1603                 return;
1604         }
1605         ep += 2; /* skip over @@ */
1606
1607         /* The hunk header in fraginfo color */
1608         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1609                 strbuf_addstr(&msgbuf, reverse);
1610         strbuf_addstr(&msgbuf, frag);
1611         strbuf_add(&msgbuf, line, ep - line);
1612         strbuf_addstr(&msgbuf, reset);
1613
1614         /*
1615          * trailing "\r\n"
1616          */
1617         for ( ; i < 3; i++)
1618                 if (line[len - i] == '\r' || line[len - i] == '\n')
1619                         len--;
1620
1621         /* blank before the func header */
1622         for (cp = ep; ep - line < len; ep++)
1623                 if (*ep != ' ' && *ep != '\t')
1624                         break;
1625         if (ep != cp) {
1626                 strbuf_addstr(&msgbuf, context);
1627                 strbuf_add(&msgbuf, cp, ep - cp);
1628                 strbuf_addstr(&msgbuf, reset);
1629         }
1630
1631         if (ep < line + len) {
1632                 strbuf_addstr(&msgbuf, func);
1633                 strbuf_add(&msgbuf, ep, line + len - ep);
1634                 strbuf_addstr(&msgbuf, reset);
1635         }
1636
1637         strbuf_add(&msgbuf, line + len, org_len - len);
1638         strbuf_complete_line(&msgbuf);
1639         emit_diff_symbol(ecbdata->opt,
1640                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1641         strbuf_release(&msgbuf);
1642 }
1643
1644 static struct diff_tempfile *claim_diff_tempfile(void) {
1645         int i;
1646         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1647                 if (!diff_temp[i].name)
1648                         return diff_temp + i;
1649         BUG("diff is failing to clean up its tempfiles");
1650 }
1651
1652 static void remove_tempfile(void)
1653 {
1654         int i;
1655         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1656                 if (is_tempfile_active(diff_temp[i].tempfile))
1657                         delete_tempfile(&diff_temp[i].tempfile);
1658                 diff_temp[i].name = NULL;
1659         }
1660 }
1661
1662 static void add_line_count(struct strbuf *out, int count)
1663 {
1664         switch (count) {
1665         case 0:
1666                 strbuf_addstr(out, "0,0");
1667                 break;
1668         case 1:
1669                 strbuf_addstr(out, "1");
1670                 break;
1671         default:
1672                 strbuf_addf(out, "1,%d", count);
1673                 break;
1674         }
1675 }
1676
1677 static void emit_rewrite_lines(struct emit_callback *ecb,
1678                                int prefix, const char *data, int size)
1679 {
1680         const char *endp = NULL;
1681         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1682
1683         while (0 < size) {
1684                 int len;
1685
1686                 endp = memchr(data, '\n', size);
1687                 len = endp ? (endp - data + 1) : size;
1688                 if (prefix != '+') {
1689                         ecb->lno_in_preimage++;
1690                         emit_del_line(reset, ecb, data, len);
1691                 } else {
1692                         ecb->lno_in_postimage++;
1693                         emit_add_line(reset, ecb, data, len);
1694                 }
1695                 size -= len;
1696                 data += len;
1697         }
1698         if (!endp)
1699                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1700 }
1701
1702 static void emit_rewrite_diff(const char *name_a,
1703                               const char *name_b,
1704                               struct diff_filespec *one,
1705                               struct diff_filespec *two,
1706                               struct userdiff_driver *textconv_one,
1707                               struct userdiff_driver *textconv_two,
1708                               struct diff_options *o)
1709 {
1710         int lc_a, lc_b;
1711         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1712         const char *a_prefix, *b_prefix;
1713         char *data_one, *data_two;
1714         size_t size_one, size_two;
1715         struct emit_callback ecbdata;
1716         struct strbuf out = STRBUF_INIT;
1717
1718         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1719                 a_prefix = o->b_prefix;
1720                 b_prefix = o->a_prefix;
1721         } else {
1722                 a_prefix = o->a_prefix;
1723                 b_prefix = o->b_prefix;
1724         }
1725
1726         name_a += (*name_a == '/');
1727         name_b += (*name_b == '/');
1728
1729         strbuf_reset(&a_name);
1730         strbuf_reset(&b_name);
1731         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1732         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1733
1734         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1735         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1736
1737         memset(&ecbdata, 0, sizeof(ecbdata));
1738         ecbdata.color_diff = want_color(o->use_color);
1739         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1740         ecbdata.opt = o;
1741         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1742                 mmfile_t mf1, mf2;
1743                 mf1.ptr = (char *)data_one;
1744                 mf2.ptr = (char *)data_two;
1745                 mf1.size = size_one;
1746                 mf2.size = size_two;
1747                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1748         }
1749         ecbdata.lno_in_preimage = 1;
1750         ecbdata.lno_in_postimage = 1;
1751
1752         lc_a = count_lines(data_one, size_one);
1753         lc_b = count_lines(data_two, size_two);
1754
1755         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1756                          a_name.buf, a_name.len, 0);
1757         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1758                          b_name.buf, b_name.len, 0);
1759
1760         strbuf_addstr(&out, "@@ -");
1761         if (!o->irreversible_delete)
1762                 add_line_count(&out, lc_a);
1763         else
1764                 strbuf_addstr(&out, "?,?");
1765         strbuf_addstr(&out, " +");
1766         add_line_count(&out, lc_b);
1767         strbuf_addstr(&out, " @@\n");
1768         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1769         strbuf_release(&out);
1770
1771         if (lc_a && !o->irreversible_delete)
1772                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1773         if (lc_b)
1774                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1775         if (textconv_one)
1776                 free((char *)data_one);
1777         if (textconv_two)
1778                 free((char *)data_two);
1779 }
1780
1781 struct diff_words_buffer {
1782         mmfile_t text;
1783         unsigned long alloc;
1784         struct diff_words_orig {
1785                 const char *begin, *end;
1786         } *orig;
1787         int orig_nr, orig_alloc;
1788 };
1789
1790 static void diff_words_append(char *line, unsigned long len,
1791                 struct diff_words_buffer *buffer)
1792 {
1793         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1794         line++;
1795         len--;
1796         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1797         buffer->text.size += len;
1798         buffer->text.ptr[buffer->text.size] = '\0';
1799 }
1800
1801 struct diff_words_style_elem {
1802         const char *prefix;
1803         const char *suffix;
1804         const char *color; /* NULL; filled in by the setup code if
1805                             * color is enabled */
1806 };
1807
1808 struct diff_words_style {
1809         enum diff_words_type type;
1810         struct diff_words_style_elem new_word, old_word, ctx;
1811         const char *newline;
1812 };
1813
1814 static struct diff_words_style diff_words_styles[] = {
1815         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1816         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1817         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1818 };
1819
1820 struct diff_words_data {
1821         struct diff_words_buffer minus, plus;
1822         const char *current_plus;
1823         int last_minus;
1824         struct diff_options *opt;
1825         regex_t *word_regex;
1826         enum diff_words_type type;
1827         struct diff_words_style *style;
1828 };
1829
1830 static int fn_out_diff_words_write_helper(struct diff_options *o,
1831                                           struct diff_words_style_elem *st_el,
1832                                           const char *newline,
1833                                           size_t count, const char *buf)
1834 {
1835         int print = 0;
1836         struct strbuf sb = STRBUF_INIT;
1837
1838         while (count) {
1839                 char *p = memchr(buf, '\n', count);
1840                 if (print)
1841                         strbuf_addstr(&sb, diff_line_prefix(o));
1842
1843                 if (p != buf) {
1844                         const char *reset = st_el->color && *st_el->color ?
1845                                             GIT_COLOR_RESET : NULL;
1846                         if (st_el->color && *st_el->color)
1847                                 strbuf_addstr(&sb, st_el->color);
1848                         strbuf_addstr(&sb, st_el->prefix);
1849                         strbuf_add(&sb, buf, p ? p - buf : count);
1850                         strbuf_addstr(&sb, st_el->suffix);
1851                         if (reset)
1852                                 strbuf_addstr(&sb, reset);
1853                 }
1854                 if (!p)
1855                         goto out;
1856
1857                 strbuf_addstr(&sb, newline);
1858                 count -= p + 1 - buf;
1859                 buf = p + 1;
1860                 print = 1;
1861                 if (count) {
1862                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1863                                          sb.buf, sb.len, 0);
1864                         strbuf_reset(&sb);
1865                 }
1866         }
1867
1868 out:
1869         if (sb.len)
1870                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1871                                  sb.buf, sb.len, 0);
1872         strbuf_release(&sb);
1873         return 0;
1874 }
1875
1876 /*
1877  * '--color-words' algorithm can be described as:
1878  *
1879  *   1. collect the minus/plus lines of a diff hunk, divided into
1880  *      minus-lines and plus-lines;
1881  *
1882  *   2. break both minus-lines and plus-lines into words and
1883  *      place them into two mmfile_t with one word for each line;
1884  *
1885  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1886  *
1887  * And for the common parts of the both file, we output the plus side text.
1888  * diff_words->current_plus is used to trace the current position of the plus file
1889  * which printed. diff_words->last_minus is used to trace the last minus word
1890  * printed.
1891  *
1892  * For '--graph' to work with '--color-words', we need to output the graph prefix
1893  * on each line of color words output. Generally, there are two conditions on
1894  * which we should output the prefix.
1895  *
1896  *   1. diff_words->last_minus == 0 &&
1897  *      diff_words->current_plus == diff_words->plus.text.ptr
1898  *
1899  *      that is: the plus text must start as a new line, and if there is no minus
1900  *      word printed, a graph prefix must be printed.
1901  *
1902  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1903  *      *(diff_words->current_plus - 1) == '\n'
1904  *
1905  *      that is: a graph prefix must be printed following a '\n'
1906  */
1907 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1908 {
1909         if ((diff_words->last_minus == 0 &&
1910                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1911                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1912                 *(diff_words->current_plus - 1) == '\n')) {
1913                 return 1;
1914         } else {
1915                 return 0;
1916         }
1917 }
1918
1919 static void fn_out_diff_words_aux(void *priv,
1920                                   long minus_first, long minus_len,
1921                                   long plus_first, long plus_len,
1922                                   const char *func, long funclen)
1923 {
1924         struct diff_words_data *diff_words = priv;
1925         struct diff_words_style *style = diff_words->style;
1926         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1927         struct diff_options *opt = diff_words->opt;
1928         const char *line_prefix;
1929
1930         assert(opt);
1931         line_prefix = diff_line_prefix(opt);
1932
1933         /* POSIX requires that first be decremented by one if len == 0... */
1934         if (minus_len) {
1935                 minus_begin = diff_words->minus.orig[minus_first].begin;
1936                 minus_end =
1937                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1938         } else
1939                 minus_begin = minus_end =
1940                         diff_words->minus.orig[minus_first].end;
1941
1942         if (plus_len) {
1943                 plus_begin = diff_words->plus.orig[plus_first].begin;
1944                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1945         } else
1946                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1947
1948         if (color_words_output_graph_prefix(diff_words)) {
1949                 fputs(line_prefix, diff_words->opt->file);
1950         }
1951         if (diff_words->current_plus != plus_begin) {
1952                 fn_out_diff_words_write_helper(diff_words->opt,
1953                                 &style->ctx, style->newline,
1954                                 plus_begin - diff_words->current_plus,
1955                                 diff_words->current_plus);
1956         }
1957         if (minus_begin != minus_end) {
1958                 fn_out_diff_words_write_helper(diff_words->opt,
1959                                 &style->old_word, style->newline,
1960                                 minus_end - minus_begin, minus_begin);
1961         }
1962         if (plus_begin != plus_end) {
1963                 fn_out_diff_words_write_helper(diff_words->opt,
1964                                 &style->new_word, style->newline,
1965                                 plus_end - plus_begin, plus_begin);
1966         }
1967
1968         diff_words->current_plus = plus_end;
1969         diff_words->last_minus = minus_first;
1970 }
1971
1972 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1973 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1974                 int *begin, int *end)
1975 {
1976         if (word_regex && *begin < buffer->size) {
1977                 regmatch_t match[1];
1978                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1979                                  buffer->size - *begin, 1, match, 0)) {
1980                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1981                                         '\n', match[0].rm_eo - match[0].rm_so);
1982                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1983                         *begin += match[0].rm_so;
1984                         return *begin >= *end;
1985                 }
1986                 return -1;
1987         }
1988
1989         /* find the next word */
1990         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1991                 (*begin)++;
1992         if (*begin >= buffer->size)
1993                 return -1;
1994
1995         /* find the end of the word */
1996         *end = *begin + 1;
1997         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1998                 (*end)++;
1999
2000         return 0;
2001 }
2002
2003 /*
2004  * This function splits the words in buffer->text, stores the list with
2005  * newline separator into out, and saves the offsets of the original words
2006  * in buffer->orig.
2007  */
2008 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2009                 regex_t *word_regex)
2010 {
2011         int i, j;
2012         long alloc = 0;
2013
2014         out->size = 0;
2015         out->ptr = NULL;
2016
2017         /* fake an empty "0th" word */
2018         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2019         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2020         buffer->orig_nr = 1;
2021
2022         for (i = 0; i < buffer->text.size; i++) {
2023                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2024                         return;
2025
2026                 /* store original boundaries */
2027                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2028                                 buffer->orig_alloc);
2029                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2030                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2031                 buffer->orig_nr++;
2032
2033                 /* store one word */
2034                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2035                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2036                 out->ptr[out->size + j - i] = '\n';
2037                 out->size += j - i + 1;
2038
2039                 i = j - 1;
2040         }
2041 }
2042
2043 /* this executes the word diff on the accumulated buffers */
2044 static void diff_words_show(struct diff_words_data *diff_words)
2045 {
2046         xpparam_t xpp;
2047         xdemitconf_t xecfg;
2048         mmfile_t minus, plus;
2049         struct diff_words_style *style = diff_words->style;
2050
2051         struct diff_options *opt = diff_words->opt;
2052         const char *line_prefix;
2053
2054         assert(opt);
2055         line_prefix = diff_line_prefix(opt);
2056
2057         /* special case: only removal */
2058         if (!diff_words->plus.text.size) {
2059                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2060                                  line_prefix, strlen(line_prefix), 0);
2061                 fn_out_diff_words_write_helper(diff_words->opt,
2062                         &style->old_word, style->newline,
2063                         diff_words->minus.text.size,
2064                         diff_words->minus.text.ptr);
2065                 diff_words->minus.text.size = 0;
2066                 return;
2067         }
2068
2069         diff_words->current_plus = diff_words->plus.text.ptr;
2070         diff_words->last_minus = 0;
2071
2072         memset(&xpp, 0, sizeof(xpp));
2073         memset(&xecfg, 0, sizeof(xecfg));
2074         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2075         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2076         xpp.flags = 0;
2077         /* as only the hunk header will be parsed, we need a 0-context */
2078         xecfg.ctxlen = 0;
2079         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2080                           diff_words, &xpp, &xecfg))
2081                 die("unable to generate word diff");
2082         free(minus.ptr);
2083         free(plus.ptr);
2084         if (diff_words->current_plus != diff_words->plus.text.ptr +
2085                         diff_words->plus.text.size) {
2086                 if (color_words_output_graph_prefix(diff_words))
2087                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2088                                          line_prefix, strlen(line_prefix), 0);
2089                 fn_out_diff_words_write_helper(diff_words->opt,
2090                         &style->ctx, style->newline,
2091                         diff_words->plus.text.ptr + diff_words->plus.text.size
2092                         - diff_words->current_plus, diff_words->current_plus);
2093         }
2094         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2095 }
2096
2097 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2098 static void diff_words_flush(struct emit_callback *ecbdata)
2099 {
2100         struct diff_options *wo = ecbdata->diff_words->opt;
2101
2102         if (ecbdata->diff_words->minus.text.size ||
2103             ecbdata->diff_words->plus.text.size)
2104                 diff_words_show(ecbdata->diff_words);
2105
2106         if (wo->emitted_symbols) {
2107                 struct diff_options *o = ecbdata->opt;
2108                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2109                 int i;
2110
2111                 /*
2112                  * NEEDSWORK:
2113                  * Instead of appending each, concat all words to a line?
2114                  */
2115                 for (i = 0; i < wol->nr; i++)
2116                         append_emitted_diff_symbol(o, &wol->buf[i]);
2117
2118                 for (i = 0; i < wol->nr; i++)
2119                         free((void *)wol->buf[i].line);
2120
2121                 wol->nr = 0;
2122         }
2123 }
2124
2125 static void diff_filespec_load_driver(struct diff_filespec *one,
2126                                       struct index_state *istate)
2127 {
2128         /* Use already-loaded driver */
2129         if (one->driver)
2130                 return;
2131
2132         if (S_ISREG(one->mode))
2133                 one->driver = userdiff_find_by_path(istate, one->path);
2134
2135         /* Fallback to default settings */
2136         if (!one->driver)
2137                 one->driver = userdiff_find_by_name("default");
2138 }
2139
2140 static const char *userdiff_word_regex(struct diff_filespec *one,
2141                                        struct index_state *istate)
2142 {
2143         diff_filespec_load_driver(one, istate);
2144         return one->driver->word_regex;
2145 }
2146
2147 static void init_diff_words_data(struct emit_callback *ecbdata,
2148                                  struct diff_options *orig_opts,
2149                                  struct diff_filespec *one,
2150                                  struct diff_filespec *two)
2151 {
2152         int i;
2153         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2154         memcpy(o, orig_opts, sizeof(struct diff_options));
2155
2156         ecbdata->diff_words =
2157                 xcalloc(1, sizeof(struct diff_words_data));
2158         ecbdata->diff_words->type = o->word_diff;
2159         ecbdata->diff_words->opt = o;
2160
2161         if (orig_opts->emitted_symbols)
2162                 o->emitted_symbols =
2163                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2164
2165         if (!o->word_regex)
2166                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2167         if (!o->word_regex)
2168                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2169         if (!o->word_regex)
2170                 o->word_regex = diff_word_regex_cfg;
2171         if (o->word_regex) {
2172                 ecbdata->diff_words->word_regex = (regex_t *)
2173                         xmalloc(sizeof(regex_t));
2174                 if (regcomp(ecbdata->diff_words->word_regex,
2175                             o->word_regex,
2176                             REG_EXTENDED | REG_NEWLINE))
2177                         die("invalid regular expression: %s",
2178                             o->word_regex);
2179         }
2180         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2181                 if (o->word_diff == diff_words_styles[i].type) {
2182                         ecbdata->diff_words->style =
2183                                 &diff_words_styles[i];
2184                         break;
2185                 }
2186         }
2187         if (want_color(o->use_color)) {
2188                 struct diff_words_style *st = ecbdata->diff_words->style;
2189                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2190                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2191                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2192         }
2193 }
2194
2195 static void free_diff_words_data(struct emit_callback *ecbdata)
2196 {
2197         if (ecbdata->diff_words) {
2198                 diff_words_flush(ecbdata);
2199                 free (ecbdata->diff_words->opt->emitted_symbols);
2200                 free (ecbdata->diff_words->opt);
2201                 free (ecbdata->diff_words->minus.text.ptr);
2202                 free (ecbdata->diff_words->minus.orig);
2203                 free (ecbdata->diff_words->plus.text.ptr);
2204                 free (ecbdata->diff_words->plus.orig);
2205                 if (ecbdata->diff_words->word_regex) {
2206                         regfree(ecbdata->diff_words->word_regex);
2207                         free(ecbdata->diff_words->word_regex);
2208                 }
2209                 FREE_AND_NULL(ecbdata->diff_words);
2210         }
2211 }
2212
2213 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2214 {
2215         if (want_color(diff_use_color))
2216                 return diff_colors[ix];
2217         return "";
2218 }
2219
2220 const char *diff_line_prefix(struct diff_options *opt)
2221 {
2222         struct strbuf *msgbuf;
2223         if (!opt->output_prefix)
2224                 return "";
2225
2226         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2227         return msgbuf->buf;
2228 }
2229
2230 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2231 {
2232         const char *cp;
2233         unsigned long allot;
2234         size_t l = len;
2235
2236         cp = line;
2237         allot = l;
2238         while (0 < l) {
2239                 (void) utf8_width(&cp, &l);
2240                 if (!cp)
2241                         break; /* truncated in the middle? */
2242         }
2243         return allot - l;
2244 }
2245
2246 static void find_lno(const char *line, struct emit_callback *ecbdata)
2247 {
2248         const char *p;
2249         ecbdata->lno_in_preimage = 0;
2250         ecbdata->lno_in_postimage = 0;
2251         p = strchr(line, '-');
2252         if (!p)
2253                 return; /* cannot happen */
2254         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2255         p = strchr(p, '+');
2256         if (!p)
2257                 return; /* cannot happen */
2258         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2259 }
2260
2261 static void fn_out_consume(void *priv, char *line, unsigned long len)
2262 {
2263         struct emit_callback *ecbdata = priv;
2264         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2265         struct diff_options *o = ecbdata->opt;
2266
2267         o->found_changes = 1;
2268
2269         if (ecbdata->header) {
2270                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2271                                  ecbdata->header->buf, ecbdata->header->len, 0);
2272                 strbuf_reset(ecbdata->header);
2273                 ecbdata->header = NULL;
2274         }
2275
2276         if (ecbdata->label_path[0]) {
2277                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2278                                  ecbdata->label_path[0],
2279                                  strlen(ecbdata->label_path[0]), 0);
2280                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2281                                  ecbdata->label_path[1],
2282                                  strlen(ecbdata->label_path[1]), 0);
2283                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2284         }
2285
2286         if (diff_suppress_blank_empty
2287             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2288                 line[0] = '\n';
2289                 len = 1;
2290         }
2291
2292         if (line[0] == '@') {
2293                 if (ecbdata->diff_words)
2294                         diff_words_flush(ecbdata);
2295                 len = sane_truncate_line(ecbdata, line, len);
2296                 find_lno(line, ecbdata);
2297                 emit_hunk_header(ecbdata, line, len);
2298                 return;
2299         }
2300
2301         if (ecbdata->diff_words) {
2302                 enum diff_symbol s =
2303                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2304                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2305                 if (line[0] == '-') {
2306                         diff_words_append(line, len,
2307                                           &ecbdata->diff_words->minus);
2308                         return;
2309                 } else if (line[0] == '+') {
2310                         diff_words_append(line, len,
2311                                           &ecbdata->diff_words->plus);
2312                         return;
2313                 } else if (starts_with(line, "\\ ")) {
2314                         /*
2315                          * Eat the "no newline at eof" marker as if we
2316                          * saw a "+" or "-" line with nothing on it,
2317                          * and return without diff_words_flush() to
2318                          * defer processing. If this is the end of
2319                          * preimage, more "+" lines may come after it.
2320                          */
2321                         return;
2322                 }
2323                 diff_words_flush(ecbdata);
2324                 emit_diff_symbol(o, s, line, len, 0);
2325                 return;
2326         }
2327
2328         switch (line[0]) {
2329         case '+':
2330                 ecbdata->lno_in_postimage++;
2331                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2332                 break;
2333         case '-':
2334                 ecbdata->lno_in_preimage++;
2335                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2336                 break;
2337         case ' ':
2338                 ecbdata->lno_in_postimage++;
2339                 ecbdata->lno_in_preimage++;
2340                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2341                 break;
2342         default:
2343                 /* incomplete line at the end */
2344                 ecbdata->lno_in_preimage++;
2345                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2346                                  line, len, 0);
2347                 break;
2348         }
2349 }
2350
2351 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2352 {
2353         const char *old_name = a;
2354         const char *new_name = b;
2355         int pfx_length, sfx_length;
2356         int pfx_adjust_for_slash;
2357         int len_a = strlen(a);
2358         int len_b = strlen(b);
2359         int a_midlen, b_midlen;
2360         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2361         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2362
2363         if (qlen_a || qlen_b) {
2364                 quote_c_style(a, name, NULL, 0);
2365                 strbuf_addstr(name, " => ");
2366                 quote_c_style(b, name, NULL, 0);
2367                 return;
2368         }
2369
2370         /* Find common prefix */
2371         pfx_length = 0;
2372         while (*old_name && *new_name && *old_name == *new_name) {
2373                 if (*old_name == '/')
2374                         pfx_length = old_name - a + 1;
2375                 old_name++;
2376                 new_name++;
2377         }
2378
2379         /* Find common suffix */
2380         old_name = a + len_a;
2381         new_name = b + len_b;
2382         sfx_length = 0;
2383         /*
2384          * If there is a common prefix, it must end in a slash.  In
2385          * that case we let this loop run 1 into the prefix to see the
2386          * same slash.
2387          *
2388          * If there is no common prefix, we cannot do this as it would
2389          * underrun the input strings.
2390          */
2391         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2392         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2393                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2394                *old_name == *new_name) {
2395                 if (*old_name == '/')
2396                         sfx_length = len_a - (old_name - a);
2397                 old_name--;
2398                 new_name--;
2399         }
2400
2401         /*
2402          * pfx{mid-a => mid-b}sfx
2403          * {pfx-a => pfx-b}sfx
2404          * pfx{sfx-a => sfx-b}
2405          * name-a => name-b
2406          */
2407         a_midlen = len_a - pfx_length - sfx_length;
2408         b_midlen = len_b - pfx_length - sfx_length;
2409         if (a_midlen < 0)
2410                 a_midlen = 0;
2411         if (b_midlen < 0)
2412                 b_midlen = 0;
2413
2414         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2415         if (pfx_length + sfx_length) {
2416                 strbuf_add(name, a, pfx_length);
2417                 strbuf_addch(name, '{');
2418         }
2419         strbuf_add(name, a + pfx_length, a_midlen);
2420         strbuf_addstr(name, " => ");
2421         strbuf_add(name, b + pfx_length, b_midlen);
2422         if (pfx_length + sfx_length) {
2423                 strbuf_addch(name, '}');
2424                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2425         }
2426 }
2427
2428 struct diffstat_t {
2429         int nr;
2430         int alloc;
2431         struct diffstat_file {
2432                 char *from_name;
2433                 char *name;
2434                 char *print_name;
2435                 const char *comments;
2436                 unsigned is_unmerged:1;
2437                 unsigned is_binary:1;
2438                 unsigned is_renamed:1;
2439                 unsigned is_interesting:1;
2440                 uintmax_t added, deleted;
2441         } **files;
2442 };
2443
2444 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2445                                           const char *name_a,
2446                                           const char *name_b)
2447 {
2448         struct diffstat_file *x;
2449         x = xcalloc(1, sizeof(*x));
2450         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2451         diffstat->files[diffstat->nr++] = x;
2452         if (name_b) {
2453                 x->from_name = xstrdup(name_a);
2454                 x->name = xstrdup(name_b);
2455                 x->is_renamed = 1;
2456         }
2457         else {
2458                 x->from_name = NULL;
2459                 x->name = xstrdup(name_a);
2460         }
2461         return x;
2462 }
2463
2464 static void diffstat_consume(void *priv, char *line, unsigned long len)
2465 {
2466         struct diffstat_t *diffstat = priv;
2467         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2468
2469         if (line[0] == '+')
2470                 x->added++;
2471         else if (line[0] == '-')
2472                 x->deleted++;
2473 }
2474
2475 const char mime_boundary_leader[] = "------------";
2476
2477 static int scale_linear(int it, int width, int max_change)
2478 {
2479         if (!it)
2480                 return 0;
2481         /*
2482          * make sure that at least one '-' or '+' is printed if
2483          * there is any change to this path. The easiest way is to
2484          * scale linearly as if the alloted width is one column shorter
2485          * than it is, and then add 1 to the result.
2486          */
2487         return 1 + (it * (width - 1) / max_change);
2488 }
2489
2490 static void show_graph(struct strbuf *out, char ch, int cnt,
2491                        const char *set, const char *reset)
2492 {
2493         if (cnt <= 0)
2494                 return;
2495         strbuf_addstr(out, set);
2496         strbuf_addchars(out, ch, cnt);
2497         strbuf_addstr(out, reset);
2498 }
2499
2500 static void fill_print_name(struct diffstat_file *file)
2501 {
2502         struct strbuf pname = STRBUF_INIT;
2503
2504         if (file->print_name)
2505                 return;
2506
2507         if (file->is_renamed)
2508                 pprint_rename(&pname, file->from_name, file->name);
2509         else
2510                 quote_c_style(file->name, &pname, NULL, 0);
2511
2512         if (file->comments)
2513                 strbuf_addf(&pname, " (%s)", file->comments);
2514
2515         file->print_name = strbuf_detach(&pname, NULL);
2516 }
2517
2518 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2519                 int files, int insertions, int deletions)
2520 {
2521         struct strbuf sb = STRBUF_INIT;
2522
2523         if (!files) {
2524                 assert(insertions == 0 && deletions == 0);
2525                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2526                                  NULL, 0, 0);
2527                 return;
2528         }
2529
2530         strbuf_addf(&sb,
2531                     (files == 1) ? " %d file changed" : " %d files changed",
2532                     files);
2533
2534         /*
2535          * For binary diff, the caller may want to print "x files
2536          * changed" with insertions == 0 && deletions == 0.
2537          *
2538          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2539          * is probably less confusing (i.e skip over "2 files changed
2540          * but nothing about added/removed lines? Is this a bug in Git?").
2541          */
2542         if (insertions || deletions == 0) {
2543                 strbuf_addf(&sb,
2544                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2545                             insertions);
2546         }
2547
2548         if (deletions || insertions == 0) {
2549                 strbuf_addf(&sb,
2550                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2551                             deletions);
2552         }
2553         strbuf_addch(&sb, '\n');
2554         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2555                          sb.buf, sb.len, 0);
2556         strbuf_release(&sb);
2557 }
2558
2559 void print_stat_summary(FILE *fp, int files,
2560                         int insertions, int deletions)
2561 {
2562         struct diff_options o;
2563         memset(&o, 0, sizeof(o));
2564         o.file = fp;
2565
2566         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2567 }
2568
2569 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2570 {
2571         int i, len, add, del, adds = 0, dels = 0;
2572         uintmax_t max_change = 0, max_len = 0;
2573         int total_files = data->nr, count;
2574         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2575         const char *reset, *add_c, *del_c;
2576         int extra_shown = 0;
2577         const char *line_prefix = diff_line_prefix(options);
2578         struct strbuf out = STRBUF_INIT;
2579
2580         if (data->nr == 0)
2581                 return;
2582
2583         count = options->stat_count ? options->stat_count : data->nr;
2584
2585         reset = diff_get_color_opt(options, DIFF_RESET);
2586         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2587         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2588
2589         /*
2590          * Find the longest filename and max number of changes
2591          */
2592         for (i = 0; (i < count) && (i < data->nr); i++) {
2593                 struct diffstat_file *file = data->files[i];
2594                 uintmax_t change = file->added + file->deleted;
2595
2596                 if (!file->is_interesting && (change == 0)) {
2597                         count++; /* not shown == room for one more */
2598                         continue;
2599                 }
2600                 fill_print_name(file);
2601                 len = strlen(file->print_name);
2602                 if (max_len < len)
2603                         max_len = len;
2604
2605                 if (file->is_unmerged) {
2606                         /* "Unmerged" is 8 characters */
2607                         bin_width = bin_width < 8 ? 8 : bin_width;
2608                         continue;
2609                 }
2610                 if (file->is_binary) {
2611                         /* "Bin XXX -> YYY bytes" */
2612                         int w = 14 + decimal_width(file->added)
2613                                 + decimal_width(file->deleted);
2614                         bin_width = bin_width < w ? w : bin_width;
2615                         /* Display change counts aligned with "Bin" */
2616                         number_width = 3;
2617                         continue;
2618                 }
2619
2620                 if (max_change < change)
2621                         max_change = change;
2622         }
2623         count = i; /* where we can stop scanning in data->files[] */
2624
2625         /*
2626          * We have width = stat_width or term_columns() columns total.
2627          * We want a maximum of min(max_len, stat_name_width) for the name part.
2628          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2629          * We also need 1 for " " and 4 + decimal_width(max_change)
2630          * for " | NNNN " and one the empty column at the end, altogether
2631          * 6 + decimal_width(max_change).
2632          *
2633          * If there's not enough space, we will use the smaller of
2634          * stat_name_width (if set) and 5/8*width for the filename,
2635          * and the rest for constant elements + graph part, but no more
2636          * than stat_graph_width for the graph part.
2637          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2638          * for the standard terminal size).
2639          *
2640          * In other words: stat_width limits the maximum width, and
2641          * stat_name_width fixes the maximum width of the filename,
2642          * and is also used to divide available columns if there
2643          * aren't enough.
2644          *
2645          * Binary files are displayed with "Bin XXX -> YYY bytes"
2646          * instead of the change count and graph. This part is treated
2647          * similarly to the graph part, except that it is not
2648          * "scaled". If total width is too small to accommodate the
2649          * guaranteed minimum width of the filename part and the
2650          * separators and this message, this message will "overflow"
2651          * making the line longer than the maximum width.
2652          */
2653
2654         if (options->stat_width == -1)
2655                 width = term_columns() - strlen(line_prefix);
2656         else
2657                 width = options->stat_width ? options->stat_width : 80;
2658         number_width = decimal_width(max_change) > number_width ?
2659                 decimal_width(max_change) : number_width;
2660
2661         if (options->stat_graph_width == -1)
2662                 options->stat_graph_width = diff_stat_graph_width;
2663
2664         /*
2665          * Guarantee 3/8*16==6 for the graph part
2666          * and 5/8*16==10 for the filename part
2667          */
2668         if (width < 16 + 6 + number_width)
2669                 width = 16 + 6 + number_width;
2670
2671         /*
2672          * First assign sizes that are wanted, ignoring available width.
2673          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2674          * starting from "XXX" should fit in graph_width.
2675          */
2676         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2677         if (options->stat_graph_width &&
2678             options->stat_graph_width < graph_width)
2679                 graph_width = options->stat_graph_width;
2680
2681         name_width = (options->stat_name_width > 0 &&
2682                       options->stat_name_width < max_len) ?
2683                 options->stat_name_width : max_len;
2684
2685         /*
2686          * Adjust adjustable widths not to exceed maximum width
2687          */
2688         if (name_width + number_width + 6 + graph_width > width) {
2689                 if (graph_width > width * 3/8 - number_width - 6) {
2690                         graph_width = width * 3/8 - number_width - 6;
2691                         if (graph_width < 6)
2692                                 graph_width = 6;
2693                 }
2694
2695                 if (options->stat_graph_width &&
2696                     graph_width > options->stat_graph_width)
2697                         graph_width = options->stat_graph_width;
2698                 if (name_width > width - number_width - 6 - graph_width)
2699                         name_width = width - number_width - 6 - graph_width;
2700                 else
2701                         graph_width = width - number_width - 6 - name_width;
2702         }
2703
2704         /*
2705          * From here name_width is the width of the name area,
2706          * and graph_width is the width of the graph area.
2707          * max_change is used to scale graph properly.
2708          */
2709         for (i = 0; i < count; i++) {
2710                 const char *prefix = "";
2711                 struct diffstat_file *file = data->files[i];
2712                 char *name = file->print_name;
2713                 uintmax_t added = file->added;
2714                 uintmax_t deleted = file->deleted;
2715                 int name_len;
2716
2717                 if (!file->is_interesting && (added + deleted == 0))
2718                         continue;
2719
2720                 /*
2721                  * "scale" the filename
2722                  */
2723                 len = name_width;
2724                 name_len = strlen(name);
2725                 if (name_width < name_len) {
2726                         char *slash;
2727                         prefix = "...";
2728                         len -= 3;
2729                         name += name_len - len;
2730                         slash = strchr(name, '/');
2731                         if (slash)
2732                                 name = slash;
2733                 }
2734
2735                 if (file->is_binary) {
2736                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2737                         strbuf_addf(&out, " %*s", number_width, "Bin");
2738                         if (!added && !deleted) {
2739                                 strbuf_addch(&out, '\n');
2740                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2741                                                  out.buf, out.len, 0);
2742                                 strbuf_reset(&out);
2743                                 continue;
2744                         }
2745                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2746                                 del_c, deleted, reset);
2747                         strbuf_addstr(&out, " -> ");
2748                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2749                                 add_c, added, reset);
2750                         strbuf_addstr(&out, " bytes\n");
2751                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2752                                          out.buf, out.len, 0);
2753                         strbuf_reset(&out);
2754                         continue;
2755                 }
2756                 else if (file->is_unmerged) {
2757                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2758                         strbuf_addstr(&out, " Unmerged\n");
2759                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2760                                          out.buf, out.len, 0);
2761                         strbuf_reset(&out);
2762                         continue;
2763                 }
2764
2765                 /*
2766                  * scale the add/delete
2767                  */
2768                 add = added;
2769                 del = deleted;
2770
2771                 if (graph_width <= max_change) {
2772                         int total = scale_linear(add + del, graph_width, max_change);
2773                         if (total < 2 && add && del)
2774                                 /* width >= 2 due to the sanity check */
2775                                 total = 2;
2776                         if (add < del) {
2777                                 add = scale_linear(add, graph_width, max_change);
2778                                 del = total - add;
2779                         } else {
2780                                 del = scale_linear(del, graph_width, max_change);
2781                                 add = total - del;
2782                         }
2783                 }
2784                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2785                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2786                         number_width, added + deleted,
2787                         added + deleted ? " " : "");
2788                 show_graph(&out, '+', add, add_c, reset);
2789                 show_graph(&out, '-', del, del_c, reset);
2790                 strbuf_addch(&out, '\n');
2791                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2792                                  out.buf, out.len, 0);
2793                 strbuf_reset(&out);
2794         }
2795
2796         for (i = 0; i < data->nr; i++) {
2797                 struct diffstat_file *file = data->files[i];
2798                 uintmax_t added = file->added;
2799                 uintmax_t deleted = file->deleted;
2800
2801                 if (file->is_unmerged ||
2802                     (!file->is_interesting && (added + deleted == 0))) {
2803                         total_files--;
2804                         continue;
2805                 }
2806
2807                 if (!file->is_binary) {
2808                         adds += added;
2809                         dels += deleted;
2810                 }
2811                 if (i < count)
2812                         continue;
2813                 if (!extra_shown)
2814                         emit_diff_symbol(options,
2815                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2816                                          NULL, 0, 0);
2817                 extra_shown = 1;
2818         }
2819
2820         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2821         strbuf_release(&out);
2822 }
2823
2824 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2825 {
2826         int i, adds = 0, dels = 0, total_files = data->nr;
2827
2828         if (data->nr == 0)
2829                 return;
2830
2831         for (i = 0; i < data->nr; i++) {
2832                 int added = data->files[i]->added;
2833                 int deleted = data->files[i]->deleted;
2834
2835                 if (data->files[i]->is_unmerged ||
2836                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2837                         total_files--;
2838                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2839                         adds += added;
2840                         dels += deleted;
2841                 }
2842         }
2843         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2844 }
2845
2846 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2847 {
2848         int i;
2849
2850         if (data->nr == 0)
2851                 return;
2852
2853         for (i = 0; i < data->nr; i++) {
2854                 struct diffstat_file *file = data->files[i];
2855
2856                 fprintf(options->file, "%s", diff_line_prefix(options));
2857
2858                 if (file->is_binary)
2859                         fprintf(options->file, "-\t-\t");
2860                 else
2861                         fprintf(options->file,
2862                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2863                                 file->added, file->deleted);
2864                 if (options->line_termination) {
2865                         fill_print_name(file);
2866                         if (!file->is_renamed)
2867                                 write_name_quoted(file->name, options->file,
2868                                                   options->line_termination);
2869                         else {
2870                                 fputs(file->print_name, options->file);
2871                                 putc(options->line_termination, options->file);
2872                         }
2873                 } else {
2874                         if (file->is_renamed) {
2875                                 putc('\0', options->file);
2876                                 write_name_quoted(file->from_name, options->file, '\0');
2877                         }
2878                         write_name_quoted(file->name, options->file, '\0');
2879                 }
2880         }
2881 }
2882
2883 struct dirstat_file {
2884         const char *name;
2885         unsigned long changed;
2886 };
2887
2888 struct dirstat_dir {
2889         struct dirstat_file *files;
2890         int alloc, nr, permille, cumulative;
2891 };
2892
2893 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2894                 unsigned long changed, const char *base, int baselen)
2895 {
2896         unsigned long sum_changes = 0;
2897         unsigned int sources = 0;
2898         const char *line_prefix = diff_line_prefix(opt);
2899
2900         while (dir->nr) {
2901                 struct dirstat_file *f = dir->files;
2902                 int namelen = strlen(f->name);
2903                 unsigned long changes;
2904                 char *slash;
2905
2906                 if (namelen < baselen)
2907                         break;
2908                 if (memcmp(f->name, base, baselen))
2909                         break;
2910                 slash = strchr(f->name + baselen, '/');
2911                 if (slash) {
2912                         int newbaselen = slash + 1 - f->name;
2913                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2914                         sources++;
2915                 } else {
2916                         changes = f->changed;
2917                         dir->files++;
2918                         dir->nr--;
2919                         sources += 2;
2920                 }
2921                 sum_changes += changes;
2922         }
2923
2924         /*
2925          * We don't report dirstat's for
2926          *  - the top level
2927          *  - or cases where everything came from a single directory
2928          *    under this directory (sources == 1).
2929          */
2930         if (baselen && sources != 1) {
2931                 if (sum_changes) {
2932                         int permille = sum_changes * 1000 / changed;
2933                         if (permille >= dir->permille) {
2934                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2935                                         permille / 10, permille % 10, baselen, base);
2936                                 if (!dir->cumulative)
2937                                         return 0;
2938                         }
2939                 }
2940         }
2941         return sum_changes;
2942 }
2943
2944 static int dirstat_compare(const void *_a, const void *_b)
2945 {
2946         const struct dirstat_file *a = _a;
2947         const struct dirstat_file *b = _b;
2948         return strcmp(a->name, b->name);
2949 }
2950
2951 static void show_dirstat(struct diff_options *options)
2952 {
2953         int i;
2954         unsigned long changed;
2955         struct dirstat_dir dir;
2956         struct diff_queue_struct *q = &diff_queued_diff;
2957
2958         dir.files = NULL;
2959         dir.alloc = 0;
2960         dir.nr = 0;
2961         dir.permille = options->dirstat_permille;
2962         dir.cumulative = options->flags.dirstat_cumulative;
2963
2964         changed = 0;
2965         for (i = 0; i < q->nr; i++) {
2966                 struct diff_filepair *p = q->queue[i];
2967                 const char *name;
2968                 unsigned long copied, added, damage;
2969
2970                 name = p->two->path ? p->two->path : p->one->path;
2971
2972                 if (p->one->oid_valid && p->two->oid_valid &&
2973                     oideq(&p->one->oid, &p->two->oid)) {
2974                         /*
2975                          * The SHA1 has not changed, so pre-/post-content is
2976                          * identical. We can therefore skip looking at the
2977                          * file contents altogether.
2978                          */
2979                         damage = 0;
2980                         goto found_damage;
2981                 }
2982
2983                 if (options->flags.dirstat_by_file) {
2984                         /*
2985                          * In --dirstat-by-file mode, we don't really need to
2986                          * look at the actual file contents at all.
2987                          * The fact that the SHA1 changed is enough for us to
2988                          * add this file to the list of results
2989                          * (with each file contributing equal damage).
2990                          */
2991                         damage = 1;
2992                         goto found_damage;
2993                 }
2994
2995                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2996                         diff_populate_filespec(options->repo, p->one, 0);
2997                         diff_populate_filespec(options->repo, p->two, 0);
2998                         diffcore_count_changes(options->repo,
2999                                                p->one, p->two, NULL, NULL,
3000                                                &copied, &added);
3001                         diff_free_filespec_data(p->one);
3002                         diff_free_filespec_data(p->two);
3003                 } else if (DIFF_FILE_VALID(p->one)) {
3004                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3005                         copied = added = 0;
3006                         diff_free_filespec_data(p->one);
3007                 } else if (DIFF_FILE_VALID(p->two)) {
3008                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3009                         copied = 0;
3010                         added = p->two->size;
3011                         diff_free_filespec_data(p->two);
3012                 } else
3013                         continue;
3014
3015                 /*
3016                  * Original minus copied is the removed material,
3017                  * added is the new material.  They are both damages
3018                  * made to the preimage.
3019                  * If the resulting damage is zero, we know that
3020                  * diffcore_count_changes() considers the two entries to
3021                  * be identical, but since the oid changed, we
3022                  * know that there must have been _some_ kind of change,
3023                  * so we force all entries to have damage > 0.
3024                  */
3025                 damage = (p->one->size - copied) + added;
3026                 if (!damage)
3027                         damage = 1;
3028
3029 found_damage:
3030                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3031                 dir.files[dir.nr].name = name;
3032                 dir.files[dir.nr].changed = damage;
3033                 changed += damage;
3034                 dir.nr++;
3035         }
3036
3037         /* This can happen even with many files, if everything was renames */
3038         if (!changed)
3039                 return;
3040
3041         /* Show all directories with more than x% of the changes */
3042         QSORT(dir.files, dir.nr, dirstat_compare);
3043         gather_dirstat(options, &dir, changed, "", 0);
3044 }
3045
3046 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3047 {
3048         int i;
3049         unsigned long changed;
3050         struct dirstat_dir dir;
3051
3052         if (data->nr == 0)
3053                 return;
3054
3055         dir.files = NULL;
3056         dir.alloc = 0;
3057         dir.nr = 0;
3058         dir.permille = options->dirstat_permille;
3059         dir.cumulative = options->flags.dirstat_cumulative;
3060
3061         changed = 0;
3062         for (i = 0; i < data->nr; i++) {
3063                 struct diffstat_file *file = data->files[i];
3064                 unsigned long damage = file->added + file->deleted;
3065                 if (file->is_binary)
3066                         /*
3067                          * binary files counts bytes, not lines. Must find some
3068                          * way to normalize binary bytes vs. textual lines.
3069                          * The following heuristic assumes that there are 64
3070                          * bytes per "line".
3071                          * This is stupid and ugly, but very cheap...
3072                          */
3073                         damage = DIV_ROUND_UP(damage, 64);
3074                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3075                 dir.files[dir.nr].name = file->name;
3076                 dir.files[dir.nr].changed = damage;
3077                 changed += damage;
3078                 dir.nr++;
3079         }
3080
3081         /* This can happen even with many files, if everything was renames */
3082         if (!changed)
3083                 return;
3084
3085         /* Show all directories with more than x% of the changes */
3086         QSORT(dir.files, dir.nr, dirstat_compare);
3087         gather_dirstat(options, &dir, changed, "", 0);
3088 }
3089
3090 static void free_diffstat_info(struct diffstat_t *diffstat)
3091 {
3092         int i;
3093         for (i = 0; i < diffstat->nr; i++) {
3094                 struct diffstat_file *f = diffstat->files[i];
3095                 free(f->print_name);
3096                 free(f->name);
3097                 free(f->from_name);
3098                 free(f);
3099         }
3100         free(diffstat->files);
3101 }
3102
3103 struct checkdiff_t {
3104         const char *filename;
3105         int lineno;
3106         int conflict_marker_size;
3107         struct diff_options *o;
3108         unsigned ws_rule;
3109         unsigned status;
3110 };
3111
3112 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3113 {
3114         char firstchar;
3115         int cnt;
3116
3117         if (len < marker_size + 1)
3118                 return 0;
3119         firstchar = line[0];
3120         switch (firstchar) {
3121         case '=': case '>': case '<': case '|':
3122                 break;
3123         default:
3124                 return 0;
3125         }
3126         for (cnt = 1; cnt < marker_size; cnt++)
3127                 if (line[cnt] != firstchar)
3128                         return 0;
3129         /* line[1] thru line[marker_size-1] are same as firstchar */
3130         if (len < marker_size + 1 || !isspace(line[marker_size]))
3131                 return 0;
3132         return 1;
3133 }
3134
3135 static void checkdiff_consume_hunk(void *priv,
3136                                    long ob, long on, long nb, long nn,
3137                                    const char *func, long funclen)
3138
3139 {
3140         struct checkdiff_t *data = priv;
3141         data->lineno = nb - 1;
3142 }
3143
3144 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3145 {
3146         struct checkdiff_t *data = priv;
3147         int marker_size = data->conflict_marker_size;
3148         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3149         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3150         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3151         char *err;
3152         const char *line_prefix;
3153
3154         assert(data->o);
3155         line_prefix = diff_line_prefix(data->o);
3156
3157         if (line[0] == '+') {
3158                 unsigned bad;
3159                 data->lineno++;
3160                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3161                         data->status |= 1;
3162                         fprintf(data->o->file,
3163                                 "%s%s:%d: leftover conflict marker\n",
3164                                 line_prefix, data->filename, data->lineno);
3165                 }
3166                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3167                 if (!bad)
3168                         return;
3169                 data->status |= bad;
3170                 err = whitespace_error_string(bad);
3171                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3172                         line_prefix, data->filename, data->lineno, err);
3173                 free(err);
3174                 emit_line(data->o, set, reset, line, 1);
3175                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3176                               data->o->file, set, reset, ws);
3177         } else if (line[0] == ' ') {
3178                 data->lineno++;
3179         }
3180 }
3181
3182 static unsigned char *deflate_it(char *data,
3183                                  unsigned long size,
3184                                  unsigned long *result_size)
3185 {
3186         int bound;
3187         unsigned char *deflated;
3188         git_zstream stream;
3189
3190         git_deflate_init(&stream, zlib_compression_level);
3191         bound = git_deflate_bound(&stream, size);
3192         deflated = xmalloc(bound);
3193         stream.next_out = deflated;
3194         stream.avail_out = bound;
3195
3196         stream.next_in = (unsigned char *)data;
3197         stream.avail_in = size;
3198         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3199                 ; /* nothing */
3200         git_deflate_end(&stream);
3201         *result_size = stream.total_out;
3202         return deflated;
3203 }
3204
3205 static void emit_binary_diff_body(struct diff_options *o,
3206                                   mmfile_t *one, mmfile_t *two)
3207 {
3208         void *cp;
3209         void *delta;
3210         void *deflated;
3211         void *data;
3212         unsigned long orig_size;
3213         unsigned long delta_size;
3214         unsigned long deflate_size;
3215         unsigned long data_size;
3216
3217         /* We could do deflated delta, or we could do just deflated two,
3218          * whichever is smaller.
3219          */
3220         delta = NULL;
3221         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3222         if (one->size && two->size) {
3223                 delta = diff_delta(one->ptr, one->size,
3224                                    two->ptr, two->size,
3225                                    &delta_size, deflate_size);
3226                 if (delta) {
3227                         void *to_free = delta;
3228                         orig_size = delta_size;
3229                         delta = deflate_it(delta, delta_size, &delta_size);
3230                         free(to_free);
3231                 }
3232         }
3233
3234         if (delta && delta_size < deflate_size) {
3235                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3236                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3237                                  s, strlen(s), 0);
3238                 free(s);
3239                 free(deflated);
3240                 data = delta;
3241                 data_size = delta_size;
3242         } else {
3243                 char *s = xstrfmt("%lu", two->size);
3244                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3245                                  s, strlen(s), 0);
3246                 free(s);
3247                 free(delta);
3248                 data = deflated;
3249                 data_size = deflate_size;
3250         }
3251
3252         /* emit data encoded in base85 */
3253         cp = data;
3254         while (data_size) {
3255                 int len;
3256                 int bytes = (52 < data_size) ? 52 : data_size;
3257                 char line[71];
3258                 data_size -= bytes;
3259                 if (bytes <= 26)
3260                         line[0] = bytes + 'A' - 1;
3261                 else
3262                         line[0] = bytes - 26 + 'a' - 1;
3263                 encode_85(line + 1, cp, bytes);
3264                 cp = (char *) cp + bytes;
3265
3266                 len = strlen(line);
3267                 line[len++] = '\n';
3268                 line[len] = '\0';
3269
3270                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3271                                  line, len, 0);
3272         }
3273         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3274         free(data);
3275 }
3276
3277 static void emit_binary_diff(struct diff_options *o,
3278                              mmfile_t *one, mmfile_t *two)
3279 {
3280         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3281         emit_binary_diff_body(o, one, two);
3282         emit_binary_diff_body(o, two, one);
3283 }
3284
3285 int diff_filespec_is_binary(struct repository *r,
3286                             struct diff_filespec *one)
3287 {
3288         if (one->is_binary == -1) {
3289                 diff_filespec_load_driver(one, r->index);
3290                 if (one->driver->binary != -1)
3291                         one->is_binary = one->driver->binary;
3292                 else {
3293                         if (!one->data && DIFF_FILE_VALID(one))
3294                                 diff_populate_filespec(r, one, CHECK_BINARY);
3295                         if (one->is_binary == -1 && one->data)
3296                                 one->is_binary = buffer_is_binary(one->data,
3297                                                 one->size);
3298                         if (one->is_binary == -1)
3299                                 one->is_binary = 0;
3300                 }
3301         }
3302         return one->is_binary;
3303 }
3304
3305 static const struct userdiff_funcname *
3306 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3307 {
3308         diff_filespec_load_driver(one, o->repo->index);
3309         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3310 }
3311
3312 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3313 {
3314         if (!options->a_prefix)
3315                 options->a_prefix = a;
3316         if (!options->b_prefix)
3317                 options->b_prefix = b;
3318 }
3319
3320 struct userdiff_driver *get_textconv(struct repository *r,
3321                                      struct diff_filespec *one)
3322 {
3323         if (!DIFF_FILE_VALID(one))
3324                 return NULL;
3325
3326         diff_filespec_load_driver(one, r->index);
3327         return userdiff_get_textconv(r, one->driver);
3328 }
3329
3330 static void builtin_diff(const char *name_a,
3331                          const char *name_b,
3332                          struct diff_filespec *one,
3333                          struct diff_filespec *two,
3334                          const char *xfrm_msg,
3335                          int must_show_header,
3336                          struct diff_options *o,
3337                          int complete_rewrite)
3338 {
3339         mmfile_t mf1, mf2;
3340         const char *lbl[2];
3341         char *a_one, *b_two;
3342         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3343         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3344         const char *a_prefix, *b_prefix;
3345         struct userdiff_driver *textconv_one = NULL;
3346         struct userdiff_driver *textconv_two = NULL;
3347         struct strbuf header = STRBUF_INIT;
3348         const char *line_prefix = diff_line_prefix(o);
3349
3350         diff_set_mnemonic_prefix(o, "a/", "b/");
3351         if (o->flags.reverse_diff) {
3352                 a_prefix = o->b_prefix;
3353                 b_prefix = o->a_prefix;
3354         } else {
3355                 a_prefix = o->a_prefix;
3356                 b_prefix = o->b_prefix;
3357         }
3358
3359         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3360             (!one->mode || S_ISGITLINK(one->mode)) &&
3361             (!two->mode || S_ISGITLINK(two->mode))) {
3362                 show_submodule_summary(o, one->path ? one->path : two->path,
3363                                 &one->oid, &two->oid,
3364                                 two->dirty_submodule);
3365                 return;
3366         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3367                    (!one->mode || S_ISGITLINK(one->mode)) &&
3368                    (!two->mode || S_ISGITLINK(two->mode))) {
3369                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3370                                 &one->oid, &two->oid,
3371                                 two->dirty_submodule);
3372                 return;
3373         }
3374
3375         if (o->flags.allow_textconv) {
3376                 textconv_one = get_textconv(o->repo, one);
3377                 textconv_two = get_textconv(o->repo, two);
3378         }
3379
3380         /* Never use a non-valid filename anywhere if at all possible */
3381         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3382         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3383
3384         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3385         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3386         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3387         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3388         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3389         if (lbl[0][0] == '/') {
3390                 /* /dev/null */
3391                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3392                 if (xfrm_msg)
3393                         strbuf_addstr(&header, xfrm_msg);
3394                 must_show_header = 1;
3395         }
3396         else if (lbl[1][0] == '/') {
3397                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3398                 if (xfrm_msg)
3399                         strbuf_addstr(&header, xfrm_msg);
3400                 must_show_header = 1;
3401         }
3402         else {
3403                 if (one->mode != two->mode) {
3404                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3405                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3406                         must_show_header = 1;
3407                 }
3408                 if (xfrm_msg)
3409                         strbuf_addstr(&header, xfrm_msg);
3410
3411                 /*
3412                  * we do not run diff between different kind
3413                  * of objects.
3414                  */
3415                 if ((one->mode ^ two->mode) & S_IFMT)
3416                         goto free_ab_and_return;
3417                 if (complete_rewrite &&
3418                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3419                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3420                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3421                                          header.buf, header.len, 0);
3422                         strbuf_reset(&header);
3423                         emit_rewrite_diff(name_a, name_b, one, two,
3424                                           textconv_one, textconv_two, o);
3425                         o->found_changes = 1;
3426                         goto free_ab_and_return;
3427                 }
3428         }
3429
3430         if (o->irreversible_delete && lbl[1][0] == '/') {
3431                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3432                                  header.len, 0);
3433                 strbuf_reset(&header);
3434                 goto free_ab_and_return;
3435         } else if (!o->flags.text &&
3436                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3437                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3438                 struct strbuf sb = STRBUF_INIT;
3439                 if (!one->data && !two->data &&
3440                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3441                     !o->flags.binary) {
3442                         if (oideq(&one->oid, &two->oid)) {
3443                                 if (must_show_header)
3444                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3445                                                          header.buf, header.len,
3446                                                          0);
3447                                 goto free_ab_and_return;
3448                         }
3449                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3450                                          header.buf, header.len, 0);
3451                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3452                                     diff_line_prefix(o), lbl[0], lbl[1]);
3453                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3454                                          sb.buf, sb.len, 0);
3455                         strbuf_release(&sb);
3456                         goto free_ab_and_return;
3457                 }
3458                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3459                     fill_mmfile(o->repo, &mf2, two) < 0)
3460                         die("unable to read files to diff");
3461                 /* Quite common confusing case */
3462                 if (mf1.size == mf2.size &&
3463                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3464                         if (must_show_header)
3465                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3466                                                  header.buf, header.len, 0);
3467                         goto free_ab_and_return;
3468                 }
3469                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3470                 strbuf_reset(&header);
3471                 if (o->flags.binary)
3472                         emit_binary_diff(o, &mf1, &mf2);
3473                 else {
3474                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3475                                     diff_line_prefix(o), lbl[0], lbl[1]);
3476                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3477                                          sb.buf, sb.len, 0);
3478                         strbuf_release(&sb);
3479                 }
3480                 o->found_changes = 1;
3481         } else {
3482                 /* Crazy xdl interfaces.. */
3483                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3484                 const char *v;
3485                 xpparam_t xpp;
3486                 xdemitconf_t xecfg;
3487                 struct emit_callback ecbdata;
3488                 const struct userdiff_funcname *pe;
3489
3490                 if (must_show_header) {
3491                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3492                                          header.buf, header.len, 0);
3493                         strbuf_reset(&header);
3494                 }
3495
3496                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3497                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3498
3499                 pe = diff_funcname_pattern(o, one);
3500                 if (!pe)
3501                         pe = diff_funcname_pattern(o, two);
3502
3503                 memset(&xpp, 0, sizeof(xpp));
3504                 memset(&xecfg, 0, sizeof(xecfg));
3505                 memset(&ecbdata, 0, sizeof(ecbdata));
3506                 if (o->flags.suppress_diff_headers)
3507                         lbl[0] = NULL;
3508                 ecbdata.label_path = lbl;
3509                 ecbdata.color_diff = want_color(o->use_color);
3510                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3511                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3512                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3513                 ecbdata.opt = o;
3514                 if (header.len && !o->flags.suppress_diff_headers)
3515                         ecbdata.header = &header;
3516                 xpp.flags = o->xdl_opts;
3517                 xpp.anchors = o->anchors;
3518                 xpp.anchors_nr = o->anchors_nr;
3519                 xecfg.ctxlen = o->context;
3520                 xecfg.interhunkctxlen = o->interhunkcontext;
3521                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3522                 if (o->flags.funccontext)
3523                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3524                 if (pe)
3525                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3526                 if (!diffopts)
3527                         ;
3528                 else if (skip_prefix(diffopts, "--unified=", &v))
3529                         xecfg.ctxlen = strtoul(v, NULL, 10);
3530                 else if (skip_prefix(diffopts, "-u", &v))
3531                         xecfg.ctxlen = strtoul(v, NULL, 10);
3532                 if (o->word_diff)
3533                         init_diff_words_data(&ecbdata, o, one, two);
3534                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3535                                   &ecbdata, &xpp, &xecfg))
3536                         die("unable to generate diff for %s", one->path);
3537                 if (o->word_diff)
3538                         free_diff_words_data(&ecbdata);
3539                 if (textconv_one)
3540                         free(mf1.ptr);
3541                 if (textconv_two)
3542                         free(mf2.ptr);
3543                 xdiff_clear_find_func(&xecfg);
3544         }
3545
3546  free_ab_and_return:
3547         strbuf_release(&header);
3548         diff_free_filespec_data(one);
3549         diff_free_filespec_data(two);
3550         free(a_one);
3551         free(b_two);
3552         return;
3553 }
3554
3555 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3556 {
3557         if (!is_renamed) {
3558                 if (p->status == DIFF_STATUS_ADDED) {
3559                         if (S_ISLNK(p->two->mode))
3560                                 return "new +l";
3561                         else if ((p->two->mode & 0777) == 0755)
3562                                 return "new +x";
3563                         else
3564                                 return "new";
3565                 } else if (p->status == DIFF_STATUS_DELETED)
3566                         return "gone";
3567         }
3568         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3569                 return "mode -l";
3570         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3571                 return "mode +l";
3572         else if ((p->one->mode & 0777) == 0644 &&
3573                  (p->two->mode & 0777) == 0755)
3574                 return "mode +x";
3575         else if ((p->one->mode & 0777) == 0755 &&
3576                  (p->two->mode & 0777) == 0644)
3577                 return "mode -x";
3578         return NULL;
3579 }
3580
3581 static void builtin_diffstat(const char *name_a, const char *name_b,
3582                              struct diff_filespec *one,
3583                              struct diff_filespec *two,
3584                              struct diffstat_t *diffstat,
3585                              struct diff_options *o,
3586                              struct diff_filepair *p)
3587 {
3588         mmfile_t mf1, mf2;
3589         struct diffstat_file *data;
3590         int same_contents;
3591         int complete_rewrite = 0;
3592
3593         if (!DIFF_PAIR_UNMERGED(p)) {
3594                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3595                         complete_rewrite = 1;
3596         }
3597
3598         data = diffstat_add(diffstat, name_a, name_b);
3599         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3600         if (o->flags.stat_with_summary)
3601                 data->comments = get_compact_summary(p, data->is_renamed);
3602
3603         if (!one || !two) {
3604                 data->is_unmerged = 1;
3605                 return;
3606         }
3607
3608         same_contents = oideq(&one->oid, &two->oid);
3609
3610         if (diff_filespec_is_binary(o->repo, one) ||
3611             diff_filespec_is_binary(o->repo, two)) {
3612                 data->is_binary = 1;
3613                 if (same_contents) {
3614                         data->added = 0;
3615                         data->deleted = 0;
3616                 } else {
3617                         data->added = diff_filespec_size(o->repo, two);
3618                         data->deleted = diff_filespec_size(o->repo, one);
3619                 }
3620         }
3621
3622         else if (complete_rewrite) {
3623                 diff_populate_filespec(o->repo, one, 0);
3624                 diff_populate_filespec(o->repo, two, 0);
3625                 data->deleted = count_lines(one->data, one->size);
3626                 data->added = count_lines(two->data, two->size);
3627         }
3628
3629         else if (!same_contents) {
3630                 /* Crazy xdl interfaces.. */
3631                 xpparam_t xpp;
3632                 xdemitconf_t xecfg;
3633
3634                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3635                     fill_mmfile(o->repo, &mf2, two) < 0)
3636                         die("unable to read files to diff");
3637
3638                 memset(&xpp, 0, sizeof(xpp));
3639                 memset(&xecfg, 0, sizeof(xecfg));
3640                 xpp.flags = o->xdl_opts;
3641                 xpp.anchors = o->anchors;
3642                 xpp.anchors_nr = o->anchors_nr;
3643                 xecfg.ctxlen = o->context;
3644                 xecfg.interhunkctxlen = o->interhunkcontext;
3645                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3646                                   diffstat_consume, diffstat, &xpp, &xecfg))
3647                         die("unable to generate diffstat for %s", one->path);
3648         }
3649
3650         diff_free_filespec_data(one);
3651         diff_free_filespec_data(two);
3652 }
3653
3654 static void builtin_checkdiff(const char *name_a, const char *name_b,
3655                               const char *attr_path,
3656                               struct diff_filespec *one,
3657                               struct diff_filespec *two,
3658                               struct diff_options *o)
3659 {
3660         mmfile_t mf1, mf2;
3661         struct checkdiff_t data;
3662
3663         if (!two)
3664                 return;
3665
3666         memset(&data, 0, sizeof(data));
3667         data.filename = name_b ? name_b : name_a;
3668         data.lineno = 0;
3669         data.o = o;
3670         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3671         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3672
3673         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3674             fill_mmfile(o->repo, &mf2, two) < 0)
3675                 die("unable to read files to diff");
3676
3677         /*
3678          * All the other codepaths check both sides, but not checking
3679          * the "old" side here is deliberate.  We are checking the newly
3680          * introduced changes, and as long as the "new" side is text, we
3681          * can and should check what it introduces.
3682          */
3683         if (diff_filespec_is_binary(o->repo, two))
3684                 goto free_and_return;
3685         else {
3686                 /* Crazy xdl interfaces.. */
3687                 xpparam_t xpp;
3688                 xdemitconf_t xecfg;
3689
3690                 memset(&xpp, 0, sizeof(xpp));
3691                 memset(&xecfg, 0, sizeof(xecfg));
3692                 xecfg.ctxlen = 1; /* at least one context line */
3693                 xpp.flags = 0;
3694                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3695                                   checkdiff_consume, &data,
3696                                   &xpp, &xecfg))
3697                         die("unable to generate checkdiff for %s", one->path);
3698
3699                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3700                         struct emit_callback ecbdata;
3701                         int blank_at_eof;
3702
3703                         ecbdata.ws_rule = data.ws_rule;
3704                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3705                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3706
3707                         if (blank_at_eof) {
3708                                 static char *err;
3709                                 if (!err)
3710                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3711                                 fprintf(o->file, "%s:%d: %s.\n",
3712                                         data.filename, blank_at_eof, err);
3713                                 data.status = 1; /* report errors */
3714                         }
3715                 }
3716         }
3717  free_and_return:
3718         diff_free_filespec_data(one);
3719         diff_free_filespec_data(two);
3720         if (data.status)
3721                 o->flags.check_failed = 1;
3722 }
3723
3724 struct diff_filespec *alloc_filespec(const char *path)
3725 {
3726         struct diff_filespec *spec;
3727
3728         FLEXPTR_ALLOC_STR(spec, path, path);
3729         spec->count = 1;
3730         spec->is_binary = -1;
3731         return spec;
3732 }
3733
3734 void free_filespec(struct diff_filespec *spec)
3735 {
3736         if (!--spec->count) {
3737                 diff_free_filespec_data(spec);
3738                 free(spec);
3739         }
3740 }
3741
3742 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3743                    int oid_valid, unsigned short mode)
3744 {
3745         if (mode) {
3746                 spec->mode = canon_mode(mode);
3747                 oidcpy(&spec->oid, oid);
3748                 spec->oid_valid = oid_valid;
3749         }
3750 }
3751
3752 /*
3753  * Given a name and sha1 pair, if the index tells us the file in
3754  * the work tree has that object contents, return true, so that
3755  * prepare_temp_file() does not have to inflate and extract.
3756  */
3757 static int reuse_worktree_file(struct index_state *istate,
3758                                const char *name,
3759                                const struct object_id *oid,
3760                                int want_file)
3761 {
3762         const struct cache_entry *ce;
3763         struct stat st;
3764         int pos, len;
3765
3766         /*
3767          * We do not read the cache ourselves here, because the
3768          * benchmark with my previous version that always reads cache
3769          * shows that it makes things worse for diff-tree comparing
3770          * two linux-2.6 kernel trees in an already checked out work
3771          * tree.  This is because most diff-tree comparisons deal with
3772          * only a small number of files, while reading the cache is
3773          * expensive for a large project, and its cost outweighs the
3774          * savings we get by not inflating the object to a temporary
3775          * file.  Practically, this code only helps when we are used
3776          * by diff-cache --cached, which does read the cache before
3777          * calling us.
3778          */
3779         if (!istate->cache)
3780                 return 0;
3781
3782         /* We want to avoid the working directory if our caller
3783          * doesn't need the data in a normal file, this system
3784          * is rather slow with its stat/open/mmap/close syscalls,
3785          * and the object is contained in a pack file.  The pack
3786          * is probably already open and will be faster to obtain
3787          * the data through than the working directory.  Loose
3788          * objects however would tend to be slower as they need
3789          * to be individually opened and inflated.
3790          */
3791         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3792                 return 0;
3793
3794         /*
3795          * Similarly, if we'd have to convert the file contents anyway, that
3796          * makes the optimization not worthwhile.
3797          */
3798         if (!want_file && would_convert_to_git(istate, name))
3799                 return 0;
3800
3801         len = strlen(name);
3802         pos = index_name_pos(istate, name, len);
3803         if (pos < 0)
3804                 return 0;
3805         ce = istate->cache[pos];
3806
3807         /*
3808          * This is not the sha1 we are looking for, or
3809          * unreusable because it is not a regular file.
3810          */
3811         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3812                 return 0;
3813
3814         /*
3815          * If ce is marked as "assume unchanged", there is no
3816          * guarantee that work tree matches what we are looking for.
3817          */
3818         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3819                 return 0;
3820
3821         /*
3822          * If ce matches the file in the work tree, we can reuse it.
3823          */
3824         if (ce_uptodate(ce) ||
3825             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3826                 return 1;
3827
3828         return 0;
3829 }
3830
3831 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3832 {
3833         struct strbuf buf = STRBUF_INIT;
3834         char *dirty = "";
3835
3836         /* Are we looking at the work tree? */
3837         if (s->dirty_submodule)
3838                 dirty = "-dirty";
3839
3840         strbuf_addf(&buf, "Subproject commit %s%s\n",
3841                     oid_to_hex(&s->oid), dirty);
3842         s->size = buf.len;
3843         if (size_only) {
3844                 s->data = NULL;
3845                 strbuf_release(&buf);
3846         } else {
3847                 s->data = strbuf_detach(&buf, NULL);
3848                 s->should_free = 1;
3849         }
3850         return 0;
3851 }
3852
3853 /*
3854  * While doing rename detection and pickaxe operation, we may need to
3855  * grab the data for the blob (or file) for our own in-core comparison.
3856  * diff_filespec has data and size fields for this purpose.
3857  */
3858 int diff_populate_filespec(struct repository *r,
3859                            struct diff_filespec *s,
3860                            unsigned int flags)
3861 {
3862         int size_only = flags & CHECK_SIZE_ONLY;
3863         int err = 0;
3864         int conv_flags = global_conv_flags_eol;
3865         /*
3866          * demote FAIL to WARN to allow inspecting the situation
3867          * instead of refusing.
3868          */
3869         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3870                 conv_flags = CONV_EOL_RNDTRP_WARN;
3871
3872         if (!DIFF_FILE_VALID(s))
3873                 die("internal error: asking to populate invalid file.");
3874         if (S_ISDIR(s->mode))
3875                 return -1;
3876
3877         if (s->data)
3878                 return 0;
3879
3880         if (size_only && 0 < s->size)
3881                 return 0;
3882
3883         if (S_ISGITLINK(s->mode))
3884                 return diff_populate_gitlink(s, size_only);
3885
3886         if (!s->oid_valid ||
3887             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3888                 struct strbuf buf = STRBUF_INIT;
3889                 struct stat st;
3890                 int fd;
3891
3892                 if (lstat(s->path, &st) < 0) {
3893                 err_empty:
3894                         err = -1;
3895                 empty:
3896                         s->data = (char *)"";
3897                         s->size = 0;
3898                         return err;
3899                 }
3900                 s->size = xsize_t(st.st_size);
3901                 if (!s->size)
3902                         goto empty;
3903                 if (S_ISLNK(st.st_mode)) {
3904                         struct strbuf sb = STRBUF_INIT;
3905
3906                         if (strbuf_readlink(&sb, s->path, s->size))
3907                                 goto err_empty;
3908                         s->size = sb.len;
3909                         s->data = strbuf_detach(&sb, NULL);
3910                         s->should_free = 1;
3911                         return 0;
3912                 }
3913
3914                 /*
3915                  * Even if the caller would be happy with getting
3916                  * only the size, we cannot return early at this
3917                  * point if the path requires us to run the content
3918                  * conversion.
3919                  */
3920                 if (size_only && !would_convert_to_git(r->index, s->path))
3921                         return 0;
3922
3923                 /*
3924                  * Note: this check uses xsize_t(st.st_size) that may
3925                  * not be the true size of the blob after it goes
3926                  * through convert_to_git().  This may not strictly be
3927                  * correct, but the whole point of big_file_threshold
3928                  * and is_binary check being that we want to avoid
3929                  * opening the file and inspecting the contents, this
3930                  * is probably fine.
3931                  */
3932                 if ((flags & CHECK_BINARY) &&
3933                     s->size > big_file_threshold && s->is_binary == -1) {
3934                         s->is_binary = 1;
3935                         return 0;
3936                 }
3937                 fd = open(s->path, O_RDONLY);
3938                 if (fd < 0)
3939                         goto err_empty;
3940                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3941                 close(fd);
3942                 s->should_munmap = 1;
3943
3944                 /*
3945                  * Convert from working tree format to canonical git format
3946                  */
3947                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3948                         size_t size = 0;
3949                         munmap(s->data, s->size);
3950                         s->should_munmap = 0;
3951                         s->data = strbuf_detach(&buf, &size);
3952                         s->size = size;
3953                         s->should_free = 1;
3954                 }
3955         }
3956         else {
3957                 enum object_type type;
3958                 if (size_only || (flags & CHECK_BINARY)) {
3959                         type = oid_object_info(r, &s->oid, &s->size);
3960                         if (type < 0)
3961                                 die("unable to read %s",
3962                                     oid_to_hex(&s->oid));
3963                         if (size_only)
3964                                 return 0;
3965                         if (s->size > big_file_threshold && s->is_binary == -1) {
3966                                 s->is_binary = 1;
3967                                 return 0;
3968                         }
3969                 }
3970                 s->data = read_object_file(&s->oid, &type, &s->size);
3971                 if (!s->data)
3972                         die("unable to read %s", oid_to_hex(&s->oid));
3973                 s->should_free = 1;
3974         }
3975         return 0;
3976 }
3977
3978 void diff_free_filespec_blob(struct diff_filespec *s)
3979 {
3980         if (s->should_free)
3981                 free(s->data);
3982         else if (s->should_munmap)
3983                 munmap(s->data, s->size);
3984
3985         if (s->should_free || s->should_munmap) {
3986                 s->should_free = s->should_munmap = 0;
3987                 s->data = NULL;
3988         }
3989 }
3990
3991 void diff_free_filespec_data(struct diff_filespec *s)
3992 {
3993         diff_free_filespec_blob(s);
3994         FREE_AND_NULL(s->cnt_data);
3995 }
3996
3997 static void prep_temp_blob(struct index_state *istate,
3998                            const char *path, struct diff_tempfile *temp,
3999                            void *blob,
4000                            unsigned long size,
4001                            const struct object_id *oid,
4002                            int mode)
4003 {
4004         struct strbuf buf = STRBUF_INIT;
4005         struct strbuf tempfile = STRBUF_INIT;
4006         char *path_dup = xstrdup(path);
4007         const char *base = basename(path_dup);
4008
4009         /* Generate "XXXXXX_basename.ext" */
4010         strbuf_addstr(&tempfile, "XXXXXX_");
4011         strbuf_addstr(&tempfile, base);
4012
4013         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4014         if (!temp->tempfile)
4015                 die_errno("unable to create temp-file");
4016         if (convert_to_working_tree(istate, path,
4017                         (const char *)blob, (size_t)size, &buf)) {
4018                 blob = buf.buf;
4019                 size = buf.len;
4020         }
4021         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4022             close_tempfile_gently(temp->tempfile))
4023                 die_errno("unable to write temp-file");
4024         temp->name = get_tempfile_path(temp->tempfile);
4025         oid_to_hex_r(temp->hex, oid);
4026         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4027         strbuf_release(&buf);
4028         strbuf_release(&tempfile);
4029         free(path_dup);
4030 }
4031
4032 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4033                                                const char *name,
4034                                                struct diff_filespec *one)
4035 {
4036         struct diff_tempfile *temp = claim_diff_tempfile();
4037
4038         if (!DIFF_FILE_VALID(one)) {
4039         not_a_valid_file:
4040                 /* A '-' entry produces this for file-2, and
4041                  * a '+' entry produces this for file-1.
4042                  */
4043                 temp->name = "/dev/null";
4044                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4045                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4046                 return temp;
4047         }
4048
4049         if (!S_ISGITLINK(one->mode) &&
4050             (!one->oid_valid ||
4051              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4052                 struct stat st;
4053                 if (lstat(name, &st) < 0) {
4054                         if (errno == ENOENT)
4055                                 goto not_a_valid_file;
4056                         die_errno("stat(%s)", name);
4057                 }
4058                 if (S_ISLNK(st.st_mode)) {
4059                         struct strbuf sb = STRBUF_INIT;
4060                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4061                                 die_errno("readlink(%s)", name);
4062                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4063                                        (one->oid_valid ?
4064                                         &one->oid : &null_oid),
4065                                        (one->oid_valid ?
4066                                         one->mode : S_IFLNK));
4067                         strbuf_release(&sb);
4068                 }
4069                 else {
4070                         /* we can borrow from the file in the work tree */
4071                         temp->name = name;
4072                         if (!one->oid_valid)
4073                                 oid_to_hex_r(temp->hex, &null_oid);
4074                         else
4075                                 oid_to_hex_r(temp->hex, &one->oid);
4076                         /* Even though we may sometimes borrow the
4077                          * contents from the work tree, we always want
4078                          * one->mode.  mode is trustworthy even when
4079                          * !(one->oid_valid), as long as
4080                          * DIFF_FILE_VALID(one).
4081                          */
4082                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4083                 }
4084                 return temp;
4085         }
4086         else {
4087                 if (diff_populate_filespec(r, one, 0))
4088                         die("cannot read data blob for %s", one->path);
4089                 prep_temp_blob(r->index, name, temp,
4090                                one->data, one->size,
4091                                &one->oid, one->mode);
4092         }
4093         return temp;
4094 }
4095
4096 static void add_external_diff_name(struct repository *r,
4097                                    struct argv_array *argv,
4098                                    const char *name,
4099                                    struct diff_filespec *df)
4100 {
4101         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4102         argv_array_push(argv, temp->name);
4103         argv_array_push(argv, temp->hex);
4104         argv_array_push(argv, temp->mode);
4105 }
4106
4107 /* An external diff command takes:
4108  *
4109  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4110  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4111  *
4112  */
4113 static void run_external_diff(const char *pgm,
4114                               const char *name,
4115                               const char *other,
4116                               struct diff_filespec *one,
4117                               struct diff_filespec *two,
4118                               const char *xfrm_msg,
4119                               int complete_rewrite,
4120                               struct diff_options *o)
4121 {
4122         struct argv_array argv = ARGV_ARRAY_INIT;
4123         struct argv_array env = ARGV_ARRAY_INIT;
4124         struct diff_queue_struct *q = &diff_queued_diff;
4125
4126         argv_array_push(&argv, pgm);
4127         argv_array_push(&argv, name);
4128
4129         if (one && two) {
4130                 add_external_diff_name(o->repo, &argv, name, one);
4131                 if (!other)
4132                         add_external_diff_name(o->repo, &argv, name, two);
4133                 else {
4134                         add_external_diff_name(o->repo, &argv, other, two);
4135                         argv_array_push(&argv, other);
4136                         argv_array_push(&argv, xfrm_msg);
4137                 }
4138         }
4139
4140         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4141         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4142
4143         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4144                 die(_("external diff died, stopping at %s"), name);
4145
4146         remove_tempfile();
4147         argv_array_clear(&argv);
4148         argv_array_clear(&env);
4149 }
4150
4151 static int similarity_index(struct diff_filepair *p)
4152 {
4153         return p->score * 100 / MAX_SCORE;
4154 }
4155
4156 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4157 {
4158         if (startup_info->have_repository)
4159                 return find_unique_abbrev(oid, abbrev);
4160         else {
4161                 char *hex = oid_to_hex(oid);
4162                 if (abbrev < 0)
4163                         abbrev = FALLBACK_DEFAULT_ABBREV;
4164                 if (abbrev > the_hash_algo->hexsz)
4165                         BUG("oid abbreviation out of range: %d", abbrev);
4166                 if (abbrev)
4167                         hex[abbrev] = '\0';
4168                 return hex;
4169         }
4170 }
4171
4172 static void fill_metainfo(struct strbuf *msg,
4173                           const char *name,
4174                           const char *other,
4175                           struct diff_filespec *one,
4176                           struct diff_filespec *two,
4177                           struct diff_options *o,
4178                           struct diff_filepair *p,
4179                           int *must_show_header,
4180                           int use_color)
4181 {
4182         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4183         const char *reset = diff_get_color(use_color, DIFF_RESET);
4184         const char *line_prefix = diff_line_prefix(o);
4185
4186         *must_show_header = 1;
4187         strbuf_init(msg, PATH_MAX * 2 + 300);
4188         switch (p->status) {
4189         case DIFF_STATUS_COPIED:
4190                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4191                             line_prefix, set, similarity_index(p));
4192                 strbuf_addf(msg, "%s\n%s%scopy from ",
4193                             reset,  line_prefix, set);
4194                 quote_c_style(name, msg, NULL, 0);
4195                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4196                 quote_c_style(other, msg, NULL, 0);
4197                 strbuf_addf(msg, "%s\n", reset);
4198                 break;
4199         case DIFF_STATUS_RENAMED:
4200                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4201                             line_prefix, set, similarity_index(p));
4202                 strbuf_addf(msg, "%s\n%s%srename from ",
4203                             reset, line_prefix, set);
4204                 quote_c_style(name, msg, NULL, 0);
4205                 strbuf_addf(msg, "%s\n%s%srename to ",
4206                             reset, line_prefix, set);
4207                 quote_c_style(other, msg, NULL, 0);
4208                 strbuf_addf(msg, "%s\n", reset);
4209                 break;
4210         case DIFF_STATUS_MODIFIED:
4211                 if (p->score) {
4212                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4213                                     line_prefix,
4214                                     set, similarity_index(p), reset);
4215                         break;
4216                 }
4217                 /* fallthru */
4218         default:
4219                 *must_show_header = 0;
4220         }
4221         if (one && two && !oideq(&one->oid, &two->oid)) {
4222                 const unsigned hexsz = the_hash_algo->hexsz;
4223                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4224
4225                 if (o->flags.binary) {
4226                         mmfile_t mf;
4227                         if ((!fill_mmfile(o->repo, &mf, one) &&
4228                              diff_filespec_is_binary(o->repo, one)) ||
4229                             (!fill_mmfile(o->repo, &mf, two) &&
4230                              diff_filespec_is_binary(o->repo, two)))
4231                                 abbrev = hexsz;
4232                 }
4233                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4234                             diff_abbrev_oid(&one->oid, abbrev),
4235                             diff_abbrev_oid(&two->oid, abbrev));
4236                 if (one->mode == two->mode)
4237                         strbuf_addf(msg, " %06o", one->mode);
4238                 strbuf_addf(msg, "%s\n", reset);
4239         }
4240 }
4241
4242 static void run_diff_cmd(const char *pgm,
4243                          const char *name,
4244                          const char *other,
4245                          const char *attr_path,
4246                          struct diff_filespec *one,
4247                          struct diff_filespec *two,
4248                          struct strbuf *msg,
4249                          struct diff_options *o,
4250                          struct diff_filepair *p)
4251 {
4252         const char *xfrm_msg = NULL;
4253         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4254         int must_show_header = 0;
4255
4256
4257         if (o->flags.allow_external) {
4258                 struct userdiff_driver *drv;
4259
4260                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4261                 if (drv && drv->external)
4262                         pgm = drv->external;
4263         }
4264
4265         if (msg) {
4266                 /*
4267                  * don't use colors when the header is intended for an
4268                  * external diff driver
4269                  */
4270                 fill_metainfo(msg, name, other, one, two, o, p,
4271                               &must_show_header,
4272                               want_color(o->use_color) && !pgm);
4273                 xfrm_msg = msg->len ? msg->buf : NULL;
4274         }
4275
4276         if (pgm) {
4277                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4278                                   complete_rewrite, o);
4279                 return;
4280         }
4281         if (one && two)
4282                 builtin_diff(name, other ? other : name,
4283                              one, two, xfrm_msg, must_show_header,
4284                              o, complete_rewrite);
4285         else
4286                 fprintf(o->file, "* Unmerged path %s\n", name);
4287 }
4288
4289 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4290 {
4291         if (DIFF_FILE_VALID(one)) {
4292                 if (!one->oid_valid) {
4293                         struct stat st;
4294                         if (one->is_stdin) {
4295                                 oidclr(&one->oid);
4296                                 return;
4297                         }
4298                         if (lstat(one->path, &st) < 0)
4299                                 die_errno("stat '%s'", one->path);
4300                         if (index_path(istate, &one->oid, one->path, &st, 0))
4301                                 die("cannot hash %s", one->path);
4302                 }
4303         }
4304         else
4305                 oidclr(&one->oid);
4306 }
4307
4308 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4309 {
4310         /* Strip the prefix but do not molest /dev/null and absolute paths */
4311         if (*namep && !is_absolute_path(*namep)) {
4312                 *namep += prefix_length;
4313                 if (**namep == '/')
4314                         ++*namep;
4315         }
4316         if (*otherp && !is_absolute_path(*otherp)) {
4317                 *otherp += prefix_length;
4318                 if (**otherp == '/')
4319                         ++*otherp;
4320         }
4321 }
4322
4323 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4324 {
4325         const char *pgm = external_diff();
4326         struct strbuf msg;
4327         struct diff_filespec *one = p->one;
4328         struct diff_filespec *two = p->two;
4329         const char *name;
4330         const char *other;
4331         const char *attr_path;
4332
4333         name  = one->path;
4334         other = (strcmp(name, two->path) ? two->path : NULL);
4335         attr_path = name;
4336         if (o->prefix_length)
4337                 strip_prefix(o->prefix_length, &name, &other);
4338
4339         if (!o->flags.allow_external)
4340                 pgm = NULL;
4341
4342         if (DIFF_PAIR_UNMERGED(p)) {
4343                 run_diff_cmd(pgm, name, NULL, attr_path,
4344                              NULL, NULL, NULL, o, p);
4345                 return;
4346         }
4347
4348         diff_fill_oid_info(one, o->repo->index);
4349         diff_fill_oid_info(two, o->repo->index);
4350
4351         if (!pgm &&
4352             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4353             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4354                 /*
4355                  * a filepair that changes between file and symlink
4356                  * needs to be split into deletion and creation.
4357                  */
4358                 struct diff_filespec *null = alloc_filespec(two->path);
4359                 run_diff_cmd(NULL, name, other, attr_path,
4360                              one, null, &msg,
4361                              o, p);
4362                 free(null);
4363                 strbuf_release(&msg);
4364
4365                 null = alloc_filespec(one->path);
4366                 run_diff_cmd(NULL, name, other, attr_path,
4367                              null, two, &msg, o, p);
4368                 free(null);
4369         }
4370         else
4371                 run_diff_cmd(pgm, name, other, attr_path,
4372                              one, two, &msg, o, p);
4373
4374         strbuf_release(&msg);
4375 }
4376
4377 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4378                          struct diffstat_t *diffstat)
4379 {
4380         const char *name;
4381         const char *other;
4382
4383         if (DIFF_PAIR_UNMERGED(p)) {
4384                 /* unmerged */
4385                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4386                                  diffstat, o, p);
4387                 return;
4388         }
4389
4390         name = p->one->path;
4391         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4392
4393         if (o->prefix_length)
4394                 strip_prefix(o->prefix_length, &name, &other);
4395
4396         diff_fill_oid_info(p->one, o->repo->index);
4397         diff_fill_oid_info(p->two, o->repo->index);
4398
4399         builtin_diffstat(name, other, p->one, p->two,
4400                          diffstat, o, p);
4401 }
4402
4403 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4404 {
4405         const char *name;
4406         const char *other;
4407         const char *attr_path;
4408
4409         if (DIFF_PAIR_UNMERGED(p)) {
4410                 /* unmerged */
4411                 return;
4412         }
4413
4414         name = p->one->path;
4415         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4416         attr_path = other ? other : name;
4417
4418         if (o->prefix_length)
4419                 strip_prefix(o->prefix_length, &name, &other);
4420
4421         diff_fill_oid_info(p->one, o->repo->index);
4422         diff_fill_oid_info(p->two, o->repo->index);
4423
4424         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4425 }
4426
4427 void repo_diff_setup(struct repository *r, struct diff_options *options)
4428 {
4429         memcpy(options, &default_diff_options, sizeof(*options));
4430
4431         options->file = stdout;
4432         options->repo = r;
4433
4434         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4435         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4436         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4437         options->abbrev = DEFAULT_ABBREV;
4438         options->line_termination = '\n';
4439         options->break_opt = -1;
4440         options->rename_limit = -1;
4441         options->dirstat_permille = diff_dirstat_permille_default;
4442         options->context = diff_context_default;
4443         options->interhunkcontext = diff_interhunk_context_default;
4444         options->ws_error_highlight = ws_error_highlight_default;
4445         options->flags.rename_empty = 1;
4446         options->objfind = NULL;
4447
4448         /* pathchange left =NULL by default */
4449         options->change = diff_change;
4450         options->add_remove = diff_addremove;
4451         options->use_color = diff_use_color_default;
4452         options->detect_rename = diff_detect_rename_default;
4453         options->xdl_opts |= diff_algorithm;
4454         if (diff_indent_heuristic)
4455                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4456
4457         options->orderfile = diff_order_file_cfg;
4458
4459         if (diff_no_prefix) {
4460                 options->a_prefix = options->b_prefix = "";
4461         } else if (!diff_mnemonic_prefix) {
4462                 options->a_prefix = "a/";
4463                 options->b_prefix = "b/";
4464         }
4465
4466         options->color_moved = diff_color_moved_default;
4467         options->color_moved_ws_handling = diff_color_moved_ws_default;
4468 }
4469
4470 void diff_setup_done(struct diff_options *options)
4471 {
4472         unsigned check_mask = DIFF_FORMAT_NAME |
4473                               DIFF_FORMAT_NAME_STATUS |
4474                               DIFF_FORMAT_CHECKDIFF |
4475                               DIFF_FORMAT_NO_OUTPUT;
4476         /*
4477          * This must be signed because we're comparing against a potentially
4478          * negative value.
4479          */
4480         const int hexsz = the_hash_algo->hexsz;
4481
4482         if (options->set_default)
4483                 options->set_default(options);
4484
4485         if (HAS_MULTI_BITS(options->output_format & check_mask))
4486                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4487
4488         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4489                 die(_("-G, -S and --find-object are mutually exclusive"));
4490
4491         /*
4492          * Most of the time we can say "there are changes"
4493          * only by checking if there are changed paths, but
4494          * --ignore-whitespace* options force us to look
4495          * inside contents.
4496          */
4497
4498         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4499                 options->flags.diff_from_contents = 1;
4500         else
4501                 options->flags.diff_from_contents = 0;
4502
4503         if (options->flags.find_copies_harder)
4504                 options->detect_rename = DIFF_DETECT_COPY;
4505
4506         if (!options->flags.relative_name)
4507                 options->prefix = NULL;
4508         if (options->prefix)
4509                 options->prefix_length = strlen(options->prefix);
4510         else
4511                 options->prefix_length = 0;
4512
4513         if (options->output_format & (DIFF_FORMAT_NAME |
4514                                       DIFF_FORMAT_NAME_STATUS |
4515                                       DIFF_FORMAT_CHECKDIFF |
4516                                       DIFF_FORMAT_NO_OUTPUT))
4517                 options->output_format &= ~(DIFF_FORMAT_RAW |
4518                                             DIFF_FORMAT_NUMSTAT |
4519                                             DIFF_FORMAT_DIFFSTAT |
4520                                             DIFF_FORMAT_SHORTSTAT |
4521                                             DIFF_FORMAT_DIRSTAT |
4522                                             DIFF_FORMAT_SUMMARY |
4523                                             DIFF_FORMAT_PATCH);
4524
4525         /*
4526          * These cases always need recursive; we do not drop caller-supplied
4527          * recursive bits for other formats here.
4528          */
4529         if (options->output_format & (DIFF_FORMAT_PATCH |
4530                                       DIFF_FORMAT_NUMSTAT |
4531                                       DIFF_FORMAT_DIFFSTAT |
4532                                       DIFF_FORMAT_SHORTSTAT |
4533                                       DIFF_FORMAT_DIRSTAT |
4534                                       DIFF_FORMAT_SUMMARY |
4535                                       DIFF_FORMAT_CHECKDIFF))
4536                 options->flags.recursive = 1;
4537         /*
4538          * Also pickaxe would not work very well if you do not say recursive
4539          */
4540         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4541                 options->flags.recursive = 1;
4542         /*
4543          * When patches are generated, submodules diffed against the work tree
4544          * must be checked for dirtiness too so it can be shown in the output
4545          */
4546         if (options->output_format & DIFF_FORMAT_PATCH)
4547                 options->flags.dirty_submodules = 1;
4548
4549         if (options->detect_rename && options->rename_limit < 0)
4550                 options->rename_limit = diff_rename_limit_default;
4551         if (hexsz < options->abbrev)
4552                 options->abbrev = hexsz; /* full */
4553
4554         /*
4555          * It does not make sense to show the first hit we happened
4556          * to have found.  It does not make sense not to return with
4557          * exit code in such a case either.
4558          */
4559         if (options->flags.quick) {
4560                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4561                 options->flags.exit_with_status = 1;
4562         }
4563
4564         options->diff_path_counter = 0;
4565
4566         if (options->flags.follow_renames && options->pathspec.nr != 1)
4567                 die(_("--follow requires exactly one pathspec"));
4568
4569         if (!options->use_color || external_diff())
4570                 options->color_moved = 0;
4571 }
4572
4573 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4574 {
4575         char c, *eq;
4576         int len;
4577
4578         if (*arg != '-')
4579                 return 0;
4580         c = *++arg;
4581         if (!c)
4582                 return 0;
4583         if (c == arg_short) {
4584                 c = *++arg;
4585                 if (!c)
4586                         return 1;
4587                 if (val && isdigit(c)) {
4588                         char *end;
4589                         int n = strtoul(arg, &end, 10);
4590                         if (*end)
4591                                 return 0;
4592                         *val = n;
4593                         return 1;
4594                 }
4595                 return 0;
4596         }
4597         if (c != '-')
4598                 return 0;
4599         arg++;
4600         eq = strchrnul(arg, '=');
4601         len = eq - arg;
4602         if (!len || strncmp(arg, arg_long, len))
4603                 return 0;
4604         if (*eq) {
4605                 int n;
4606                 char *end;
4607                 if (!isdigit(*++eq))
4608                         return 0;
4609                 n = strtoul(eq, &end, 10);
4610                 if (*end)
4611                         return 0;
4612                 *val = n;
4613         }
4614         return 1;
4615 }
4616
4617 static int diff_scoreopt_parse(const char *opt);
4618
4619 static inline int short_opt(char opt, const char **argv,
4620                             const char **optarg)
4621 {
4622         const char *arg = argv[0];
4623         if (arg[0] != '-' || arg[1] != opt)
4624                 return 0;
4625         if (arg[2] != '\0') {
4626                 *optarg = arg + 2;
4627                 return 1;
4628         }
4629         if (!argv[1])
4630                 die("Option '%c' requires a value", opt);
4631         *optarg = argv[1];
4632         return 2;
4633 }
4634
4635 int parse_long_opt(const char *opt, const char **argv,
4636                    const char **optarg)
4637 {
4638         const char *arg = argv[0];
4639         if (!skip_prefix(arg, "--", &arg))
4640                 return 0;
4641         if (!skip_prefix(arg, opt, &arg))
4642                 return 0;
4643         if (*arg == '=') { /* stuck form: --option=value */
4644                 *optarg = arg + 1;
4645                 return 1;
4646         }
4647         if (*arg != '\0')
4648                 return 0;
4649         /* separate form: --option value */
4650         if (!argv[1])
4651                 die("Option '--%s' requires a value", opt);
4652         *optarg = argv[1];
4653         return 2;
4654 }
4655
4656 static int stat_opt(struct diff_options *options, const char **av)
4657 {
4658         const char *arg = av[0];
4659         char *end;
4660         int width = options->stat_width;
4661         int name_width = options->stat_name_width;
4662         int graph_width = options->stat_graph_width;
4663         int count = options->stat_count;
4664         int argcount = 1;
4665
4666         if (!skip_prefix(arg, "--stat", &arg))
4667                 BUG("stat option does not begin with --stat: %s", arg);
4668         end = (char *)arg;
4669
4670         switch (*arg) {
4671         case '-':
4672                 if (skip_prefix(arg, "-width", &arg)) {
4673                         if (*arg == '=')
4674                                 width = strtoul(arg + 1, &end, 10);
4675                         else if (!*arg && !av[1])
4676                                 die_want_option("--stat-width");
4677                         else if (!*arg) {
4678                                 width = strtoul(av[1], &end, 10);
4679                                 argcount = 2;
4680                         }
4681                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4682                         if (*arg == '=')
4683                                 name_width = strtoul(arg + 1, &end, 10);
4684                         else if (!*arg && !av[1])
4685                                 die_want_option("--stat-name-width");
4686                         else if (!*arg) {
4687                                 name_width = strtoul(av[1], &end, 10);
4688                                 argcount = 2;
4689                         }
4690                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4691                         if (*arg == '=')
4692                                 graph_width = strtoul(arg + 1, &end, 10);
4693                         else if (!*arg && !av[1])
4694                                 die_want_option("--stat-graph-width");
4695                         else if (!*arg) {
4696                                 graph_width = strtoul(av[1], &end, 10);
4697                                 argcount = 2;
4698                         }
4699                 } else if (skip_prefix(arg, "-count", &arg)) {
4700                         if (*arg == '=')
4701                                 count = strtoul(arg + 1, &end, 10);
4702                         else if (!*arg && !av[1])
4703                                 die_want_option("--stat-count");
4704                         else if (!*arg) {
4705                                 count = strtoul(av[1], &end, 10);
4706                                 argcount = 2;
4707                         }
4708                 }
4709                 break;
4710         case '=':
4711                 width = strtoul(arg+1, &end, 10);
4712                 if (*end == ',')
4713                         name_width = strtoul(end+1, &end, 10);
4714                 if (*end == ',')
4715                         count = strtoul(end+1, &end, 10);
4716         }
4717
4718         /* Important! This checks all the error cases! */
4719         if (*end)
4720                 return 0;
4721         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4722         options->stat_name_width = name_width;
4723         options->stat_graph_width = graph_width;
4724         options->stat_width = width;
4725         options->stat_count = count;
4726         return argcount;
4727 }
4728
4729 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4730 {
4731         struct strbuf errmsg = STRBUF_INIT;
4732         if (parse_dirstat_params(options, params, &errmsg))
4733                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4734                     errmsg.buf);
4735         strbuf_release(&errmsg);
4736         /*
4737          * The caller knows a dirstat-related option is given from the command
4738          * line; allow it to say "return this_function();"
4739          */
4740         options->output_format |= DIFF_FORMAT_DIRSTAT;
4741         return 1;
4742 }
4743
4744 static int parse_submodule_opt(struct diff_options *options, const char *value)
4745 {
4746         if (parse_submodule_params(options, value))
4747                 die(_("Failed to parse --submodule option parameter: '%s'"),
4748                         value);
4749         return 1;
4750 }
4751
4752 static const char diff_status_letters[] = {
4753         DIFF_STATUS_ADDED,
4754         DIFF_STATUS_COPIED,
4755         DIFF_STATUS_DELETED,
4756         DIFF_STATUS_MODIFIED,
4757         DIFF_STATUS_RENAMED,
4758         DIFF_STATUS_TYPE_CHANGED,
4759         DIFF_STATUS_UNKNOWN,
4760         DIFF_STATUS_UNMERGED,
4761         DIFF_STATUS_FILTER_AON,
4762         DIFF_STATUS_FILTER_BROKEN,
4763         '\0',
4764 };
4765
4766 static unsigned int filter_bit['Z' + 1];
4767
4768 static void prepare_filter_bits(void)
4769 {
4770         int i;
4771
4772         if (!filter_bit[DIFF_STATUS_ADDED]) {
4773                 for (i = 0; diff_status_letters[i]; i++)
4774                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4775         }
4776 }
4777
4778 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4779 {
4780         return opt->filter & filter_bit[(int) status];
4781 }
4782
4783 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4784 {
4785         int i, optch;
4786
4787         prepare_filter_bits();
4788
4789         /*
4790          * If there is a negation e.g. 'd' in the input, and we haven't
4791          * initialized the filter field with another --diff-filter, start
4792          * from full set of bits, except for AON.
4793          */
4794         if (!opt->filter) {
4795                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4796                         if (optch < 'a' || 'z' < optch)
4797                                 continue;
4798                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4799                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4800                         break;
4801                 }
4802         }
4803
4804         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4805                 unsigned int bit;
4806                 int negate;
4807
4808                 if ('a' <= optch && optch <= 'z') {
4809                         negate = 1;
4810                         optch = toupper(optch);
4811                 } else {
4812                         negate = 0;
4813                 }
4814
4815                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4816                 if (!bit)
4817                         return optarg[i];
4818                 if (negate)
4819                         opt->filter &= ~bit;
4820                 else
4821                         opt->filter |= bit;
4822         }
4823         return 0;
4824 }
4825
4826 static void enable_patch_output(int *fmt) {
4827         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4828         *fmt |= DIFF_FORMAT_PATCH;
4829 }
4830
4831 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4832 {
4833         int val = parse_ws_error_highlight(arg);
4834
4835         if (val < 0) {
4836                 error("unknown value after ws-error-highlight=%.*s",
4837                       -1 - val, arg);
4838                 return 0;
4839         }
4840         opt->ws_error_highlight = val;
4841         return 1;
4842 }
4843
4844 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4845 {
4846         struct object_id oid;
4847
4848         if (get_oid(arg, &oid))
4849                 return error("unable to resolve '%s'", arg);
4850
4851         if (!opt->objfind)
4852                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4853
4854         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4855         opt->flags.recursive = 1;
4856         opt->flags.tree_in_recursive = 1;
4857         oidset_insert(opt->objfind, &oid);
4858         return 1;
4859 }
4860
4861 int diff_opt_parse(struct diff_options *options,
4862                    const char **av, int ac, const char *prefix)
4863 {
4864         const char *arg = av[0];
4865         const char *optarg;
4866         int argcount;
4867
4868         if (!prefix)
4869                 prefix = "";
4870
4871         /* Output format options */
4872         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4873             || opt_arg(arg, 'U', "unified", &options->context))
4874                 enable_patch_output(&options->output_format);
4875         else if (!strcmp(arg, "--raw"))
4876                 options->output_format |= DIFF_FORMAT_RAW;
4877         else if (!strcmp(arg, "--patch-with-raw")) {
4878                 enable_patch_output(&options->output_format);
4879                 options->output_format |= DIFF_FORMAT_RAW;
4880         } else if (!strcmp(arg, "--numstat"))
4881                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4882         else if (!strcmp(arg, "--shortstat"))
4883                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4884         else if (skip_prefix(arg, "-X", &arg) ||
4885                  skip_to_optional_arg(arg, "--dirstat", &arg))
4886                 return parse_dirstat_opt(options, arg);
4887         else if (!strcmp(arg, "--cumulative"))
4888                 return parse_dirstat_opt(options, "cumulative");
4889         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4890                 parse_dirstat_opt(options, "files");
4891                 return parse_dirstat_opt(options, arg);
4892         }
4893         else if (!strcmp(arg, "--check"))
4894                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4895         else if (!strcmp(arg, "--summary"))
4896                 options->output_format |= DIFF_FORMAT_SUMMARY;
4897         else if (!strcmp(arg, "--patch-with-stat")) {
4898                 enable_patch_output(&options->output_format);
4899                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4900         } else if (!strcmp(arg, "--name-only"))
4901                 options->output_format |= DIFF_FORMAT_NAME;
4902         else if (!strcmp(arg, "--name-status"))
4903                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4904         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4905                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4906         else if (starts_with(arg, "--stat"))
4907                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4908                 return stat_opt(options, av);
4909         else if (!strcmp(arg, "--compact-summary")) {
4910                  options->flags.stat_with_summary = 1;
4911                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4912         } else if (!strcmp(arg, "--no-compact-summary"))
4913                  options->flags.stat_with_summary = 0;
4914         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4915                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4916         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4917                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4918         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4919                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4920
4921         /* renames options */
4922         else if (starts_with(arg, "-B") ||
4923                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4924                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4925                         return error("invalid argument to -B: %s", arg+2);
4926         }
4927         else if (starts_with(arg, "-M") ||
4928                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4929                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4930                         return error("invalid argument to -M: %s", arg+2);
4931                 options->detect_rename = DIFF_DETECT_RENAME;
4932         }
4933         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4934                 options->irreversible_delete = 1;
4935         }
4936         else if (starts_with(arg, "-C") ||
4937                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4938                 if (options->detect_rename == DIFF_DETECT_COPY)
4939                         options->flags.find_copies_harder = 1;
4940                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4941                         return error("invalid argument to -C: %s", arg+2);
4942                 options->detect_rename = DIFF_DETECT_COPY;
4943         }
4944         else if (!strcmp(arg, "--no-renames"))
4945                 options->detect_rename = 0;
4946         else if (!strcmp(arg, "--rename-empty"))
4947                 options->flags.rename_empty = 1;
4948         else if (!strcmp(arg, "--no-rename-empty"))
4949                 options->flags.rename_empty = 0;
4950         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4951                 options->flags.relative_name = 1;
4952                 if (arg)
4953                         options->prefix = arg;
4954         }
4955
4956         /* xdiff options */
4957         else if (!strcmp(arg, "--minimal"))
4958                 DIFF_XDL_SET(options, NEED_MINIMAL);
4959         else if (!strcmp(arg, "--no-minimal"))
4960                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4961         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4962                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4963         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4964                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4965         else if (!strcmp(arg, "--ignore-space-at-eol"))
4966                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4967         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4968                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4969         else if (!strcmp(arg, "--ignore-blank-lines"))
4970                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4971         else if (!strcmp(arg, "--indent-heuristic"))
4972                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4973         else if (!strcmp(arg, "--no-indent-heuristic"))
4974                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4975         else if (!strcmp(arg, "--patience")) {
4976                 int i;
4977                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4978                 /*
4979                  * Both --patience and --anchored use PATIENCE_DIFF
4980                  * internally, so remove any anchors previously
4981                  * specified.
4982                  */
4983                 for (i = 0; i < options->anchors_nr; i++)
4984                         free(options->anchors[i]);
4985                 options->anchors_nr = 0;
4986         } else if (!strcmp(arg, "--histogram"))
4987                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4988         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4989                 long value = parse_algorithm_value(optarg);
4990                 if (value < 0)
4991                         return error("option diff-algorithm accepts \"myers\", "
4992                                      "\"minimal\", \"patience\" and \"histogram\"");
4993                 /* clear out previous settings */
4994                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4995                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4996                 options->xdl_opts |= value;
4997                 return argcount;
4998         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4999                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5000                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5001                            options->anchors_alloc);
5002                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5003         }
5004
5005         /* flags options */
5006         else if (!strcmp(arg, "--binary")) {
5007                 enable_patch_output(&options->output_format);
5008                 options->flags.binary = 1;
5009         }
5010         else if (!strcmp(arg, "--full-index"))
5011                 options->flags.full_index = 1;
5012         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5013                 options->flags.text = 1;
5014         else if (!strcmp(arg, "-R"))
5015                 options->flags.reverse_diff = 1;
5016         else if (!strcmp(arg, "--find-copies-harder"))
5017                 options->flags.find_copies_harder = 1;
5018         else if (!strcmp(arg, "--follow"))
5019                 options->flags.follow_renames = 1;
5020         else if (!strcmp(arg, "--no-follow")) {
5021                 options->flags.follow_renames = 0;
5022                 options->flags.default_follow_renames = 0;
5023         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5024                 int value = git_config_colorbool(NULL, arg);
5025                 if (value < 0)
5026                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5027                 options->use_color = value;
5028         }
5029         else if (!strcmp(arg, "--no-color"))
5030                 options->use_color = 0;
5031         else if (!strcmp(arg, "--color-moved")) {
5032                 if (diff_color_moved_default)
5033                         options->color_moved = diff_color_moved_default;
5034                 if (options->color_moved == COLOR_MOVED_NO)
5035                         options->color_moved = COLOR_MOVED_DEFAULT;
5036         } else if (!strcmp(arg, "--no-color-moved"))
5037                 options->color_moved = COLOR_MOVED_NO;
5038         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5039                 int cm = parse_color_moved(arg);
5040                 if (cm < 0)
5041                         return error("bad --color-moved argument: %s", arg);
5042                 options->color_moved = cm;
5043         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5044                 unsigned cm = parse_color_moved_ws(arg);
5045                 if (cm & COLOR_MOVED_WS_ERROR)
5046                         return -1;
5047                 options->color_moved_ws_handling = cm;
5048         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5049                 options->use_color = 1;
5050                 options->word_diff = DIFF_WORDS_COLOR;
5051         }
5052         else if (!strcmp(arg, "--word-diff")) {
5053                 if (options->word_diff == DIFF_WORDS_NONE)
5054                         options->word_diff = DIFF_WORDS_PLAIN;
5055         }
5056         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5057                 if (!strcmp(arg, "plain"))
5058                         options->word_diff = DIFF_WORDS_PLAIN;
5059                 else if (!strcmp(arg, "color")) {
5060                         options->use_color = 1;
5061                         options->word_diff = DIFF_WORDS_COLOR;
5062                 }
5063                 else if (!strcmp(arg, "porcelain"))
5064                         options->word_diff = DIFF_WORDS_PORCELAIN;
5065                 else if (!strcmp(arg, "none"))
5066                         options->word_diff = DIFF_WORDS_NONE;
5067                 else
5068                         die("bad --word-diff argument: %s", arg);
5069         }
5070         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5071                 if (options->word_diff == DIFF_WORDS_NONE)
5072                         options->word_diff = DIFF_WORDS_PLAIN;
5073                 options->word_regex = optarg;
5074                 return argcount;
5075         }
5076         else if (!strcmp(arg, "--exit-code"))
5077                 options->flags.exit_with_status = 1;
5078         else if (!strcmp(arg, "--quiet"))
5079                 options->flags.quick = 1;
5080         else if (!strcmp(arg, "--ext-diff"))
5081                 options->flags.allow_external = 1;
5082         else if (!strcmp(arg, "--no-ext-diff"))
5083                 options->flags.allow_external = 0;
5084         else if (!strcmp(arg, "--textconv")) {
5085                 options->flags.allow_textconv = 1;
5086                 options->flags.textconv_set_via_cmdline = 1;
5087         } else if (!strcmp(arg, "--no-textconv"))
5088                 options->flags.allow_textconv = 0;
5089         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5090                 options->flags.override_submodule_config = 1;
5091                 handle_ignore_submodules_arg(options, arg);
5092         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5093                 return parse_submodule_opt(options, arg);
5094         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5095                 return parse_ws_error_highlight_opt(options, arg);
5096         else if (!strcmp(arg, "--ita-invisible-in-index"))
5097                 options->ita_invisible_in_index = 1;
5098         else if (!strcmp(arg, "--ita-visible-in-index"))
5099                 options->ita_invisible_in_index = 0;
5100
5101         /* misc options */
5102         else if (!strcmp(arg, "-z"))
5103                 options->line_termination = 0;
5104         else if ((argcount = short_opt('l', av, &optarg))) {
5105                 options->rename_limit = strtoul(optarg, NULL, 10);
5106                 return argcount;
5107         }
5108         else if ((argcount = short_opt('S', av, &optarg))) {
5109                 options->pickaxe = optarg;
5110                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5111                 return argcount;
5112         } else if ((argcount = short_opt('G', av, &optarg))) {
5113                 options->pickaxe = optarg;
5114                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5115                 return argcount;
5116         }
5117         else if (!strcmp(arg, "--pickaxe-all"))
5118                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5119         else if (!strcmp(arg, "--pickaxe-regex"))
5120                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5121         else if ((argcount = short_opt('O', av, &optarg))) {
5122                 options->orderfile = prefix_filename(prefix, optarg);
5123                 return argcount;
5124         } else if (skip_prefix(arg, "--find-object=", &arg))
5125                 return parse_objfind_opt(options, arg);
5126         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5127                 int offending = parse_diff_filter_opt(optarg, options);
5128                 if (offending)
5129                         die("unknown change class '%c' in --diff-filter=%s",
5130                             offending, optarg);
5131                 return argcount;
5132         }
5133         else if (!strcmp(arg, "--no-abbrev"))
5134                 options->abbrev = 0;
5135         else if (!strcmp(arg, "--abbrev"))
5136                 options->abbrev = DEFAULT_ABBREV;
5137         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5138                 options->abbrev = strtoul(arg, NULL, 10);
5139                 if (options->abbrev < MINIMUM_ABBREV)
5140                         options->abbrev = MINIMUM_ABBREV;
5141                 else if (the_hash_algo->hexsz < options->abbrev)
5142                         options->abbrev = the_hash_algo->hexsz;
5143         }
5144         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5145                 options->a_prefix = optarg;
5146                 return argcount;
5147         }
5148         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5149                 options->line_prefix = optarg;
5150                 options->line_prefix_length = strlen(options->line_prefix);
5151                 graph_setup_line_prefix(options);
5152                 return argcount;
5153         }
5154         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5155                 options->b_prefix = optarg;
5156                 return argcount;
5157         }
5158         else if (!strcmp(arg, "--no-prefix"))
5159                 options->a_prefix = options->b_prefix = "";
5160         else if (opt_arg(arg, '\0', "inter-hunk-context",
5161                          &options->interhunkcontext))
5162                 ;
5163         else if (!strcmp(arg, "-W"))
5164                 options->flags.funccontext = 1;
5165         else if (!strcmp(arg, "--function-context"))
5166                 options->flags.funccontext = 1;
5167         else if (!strcmp(arg, "--no-function-context"))
5168                 options->flags.funccontext = 0;
5169         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5170                 char *path = prefix_filename(prefix, optarg);
5171                 options->file = xfopen(path, "w");
5172                 options->close_file = 1;
5173                 if (options->use_color != GIT_COLOR_ALWAYS)
5174                         options->use_color = GIT_COLOR_NEVER;
5175                 free(path);
5176                 return argcount;
5177         } else
5178                 return 0;
5179         return 1;
5180 }
5181
5182 int parse_rename_score(const char **cp_p)
5183 {
5184         unsigned long num, scale;
5185         int ch, dot;
5186         const char *cp = *cp_p;
5187
5188         num = 0;
5189         scale = 1;
5190         dot = 0;
5191         for (;;) {
5192                 ch = *cp;
5193                 if ( !dot && ch == '.' ) {
5194                         scale = 1;
5195                         dot = 1;
5196                 } else if ( ch == '%' ) {
5197                         scale = dot ? scale*100 : 100;
5198                         cp++;   /* % is always at the end */
5199                         break;
5200                 } else if ( ch >= '0' && ch <= '9' ) {
5201                         if ( scale < 100000 ) {
5202                                 scale *= 10;
5203                                 num = (num*10) + (ch-'0');
5204                         }
5205                 } else {
5206                         break;
5207                 }
5208                 cp++;
5209         }
5210         *cp_p = cp;
5211
5212         /* user says num divided by scale and we say internally that
5213          * is MAX_SCORE * num / scale.
5214          */
5215         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5216 }
5217
5218 static int diff_scoreopt_parse(const char *opt)
5219 {
5220         int opt1, opt2, cmd;
5221
5222         if (*opt++ != '-')
5223                 return -1;
5224         cmd = *opt++;
5225         if (cmd == '-') {
5226                 /* convert the long-form arguments into short-form versions */
5227                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5228                         if (*opt == 0 || *opt++ == '=')
5229                                 cmd = 'B';
5230                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5231                         if (*opt == 0 || *opt++ == '=')
5232                                 cmd = 'C';
5233                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5234                         if (*opt == 0 || *opt++ == '=')
5235                                 cmd = 'M';
5236                 }
5237         }
5238         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5239                 return -1; /* that is not a -M, -C, or -B option */
5240
5241         opt1 = parse_rename_score(&opt);
5242         if (cmd != 'B')
5243                 opt2 = 0;
5244         else {
5245                 if (*opt == 0)
5246                         opt2 = 0;
5247                 else if (*opt != '/')
5248                         return -1; /* we expect -B80/99 or -B80 */
5249                 else {
5250                         opt++;
5251                         opt2 = parse_rename_score(&opt);
5252                 }
5253         }
5254         if (*opt != 0)
5255                 return -1;
5256         return opt1 | (opt2 << 16);
5257 }
5258
5259 struct diff_queue_struct diff_queued_diff;
5260
5261 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5262 {
5263         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5264         queue->queue[queue->nr++] = dp;
5265 }
5266
5267 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5268                                  struct diff_filespec *one,
5269                                  struct diff_filespec *two)
5270 {
5271         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5272         dp->one = one;
5273         dp->two = two;
5274         if (queue)
5275                 diff_q(queue, dp);
5276         return dp;
5277 }
5278
5279 void diff_free_filepair(struct diff_filepair *p)
5280 {
5281         free_filespec(p->one);
5282         free_filespec(p->two);
5283         free(p);
5284 }
5285
5286 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5287 {
5288         int abblen;
5289         const char *abbrev;
5290
5291         /* Do we want all 40 hex characters? */
5292         if (len == the_hash_algo->hexsz)
5293                 return oid_to_hex(oid);
5294
5295         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5296         abbrev = diff_abbrev_oid(oid, len);
5297
5298         if (!print_sha1_ellipsis())
5299                 return abbrev;
5300
5301         abblen = strlen(abbrev);
5302
5303         /*
5304          * In well-behaved cases, where the abbreviated result is the
5305          * same as the requested length, append three dots after the
5306          * abbreviation (hence the whole logic is limited to the case
5307          * where abblen < 37); when the actual abbreviated result is a
5308          * bit longer than the requested length, we reduce the number
5309          * of dots so that they match the well-behaved ones.  However,
5310          * if the actual abbreviation is longer than the requested
5311          * length by more than three, we give up on aligning, and add
5312          * three dots anyway, to indicate that the output is not the
5313          * full object name.  Yes, this may be suboptimal, but this
5314          * appears only in "diff --raw --abbrev" output and it is not
5315          * worth the effort to change it now.  Note that this would
5316          * likely to work fine when the automatic sizing of default
5317          * abbreviation length is used--we would be fed -1 in "len" in
5318          * that case, and will end up always appending three-dots, but
5319          * the automatic sizing is supposed to give abblen that ensures
5320          * uniqueness across all objects (statistically speaking).
5321          */
5322         if (abblen < the_hash_algo->hexsz - 3) {
5323                 static char hex[GIT_MAX_HEXSZ + 1];
5324                 if (len < abblen && abblen <= len + 2)
5325                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5326                 else
5327                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5328                 return hex;
5329         }
5330
5331         return oid_to_hex(oid);
5332 }
5333
5334 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5335 {
5336         int line_termination = opt->line_termination;
5337         int inter_name_termination = line_termination ? '\t' : '\0';
5338
5339         fprintf(opt->file, "%s", diff_line_prefix(opt));
5340         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5341                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5342                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5343                 fprintf(opt->file, "%s ",
5344                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5345         }
5346         if (p->score) {
5347                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5348                         inter_name_termination);
5349         } else {
5350                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5351         }
5352
5353         if (p->status == DIFF_STATUS_COPIED ||
5354             p->status == DIFF_STATUS_RENAMED) {
5355                 const char *name_a, *name_b;
5356                 name_a = p->one->path;
5357                 name_b = p->two->path;
5358                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5359                 write_name_quoted(name_a, opt->file, inter_name_termination);
5360                 write_name_quoted(name_b, opt->file, line_termination);
5361         } else {
5362                 const char *name_a, *name_b;
5363                 name_a = p->one->mode ? p->one->path : p->two->path;
5364                 name_b = NULL;
5365                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5366                 write_name_quoted(name_a, opt->file, line_termination);
5367         }
5368 }
5369
5370 int diff_unmodified_pair(struct diff_filepair *p)
5371 {
5372         /* This function is written stricter than necessary to support
5373          * the currently implemented transformers, but the idea is to
5374          * let transformers to produce diff_filepairs any way they want,
5375          * and filter and clean them up here before producing the output.
5376          */
5377         struct diff_filespec *one = p->one, *two = p->two;
5378
5379         if (DIFF_PAIR_UNMERGED(p))
5380                 return 0; /* unmerged is interesting */
5381
5382         /* deletion, addition, mode or type change
5383          * and rename are all interesting.
5384          */
5385         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5386             DIFF_PAIR_MODE_CHANGED(p) ||
5387             strcmp(one->path, two->path))
5388                 return 0;
5389
5390         /* both are valid and point at the same path.  that is, we are
5391          * dealing with a change.
5392          */
5393         if (one->oid_valid && two->oid_valid &&
5394             oideq(&one->oid, &two->oid) &&
5395             !one->dirty_submodule && !two->dirty_submodule)
5396                 return 1; /* no change */
5397         if (!one->oid_valid && !two->oid_valid)
5398                 return 1; /* both look at the same file on the filesystem. */
5399         return 0;
5400 }
5401
5402 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5403 {
5404         if (diff_unmodified_pair(p))
5405                 return;
5406
5407         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5408             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5409                 return; /* no tree diffs in patch format */
5410
5411         run_diff(p, o);
5412 }
5413
5414 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5415                             struct diffstat_t *diffstat)
5416 {
5417         if (diff_unmodified_pair(p))
5418                 return;
5419
5420         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5421             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5422                 return; /* no useful stat for tree diffs */
5423
5424         run_diffstat(p, o, diffstat);
5425 }
5426
5427 static void diff_flush_checkdiff(struct diff_filepair *p,
5428                 struct diff_options *o)
5429 {
5430         if (diff_unmodified_pair(p))
5431                 return;
5432
5433         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5434             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5435                 return; /* nothing to check in tree diffs */
5436
5437         run_checkdiff(p, o);
5438 }
5439
5440 int diff_queue_is_empty(void)
5441 {
5442         struct diff_queue_struct *q = &diff_queued_diff;
5443         int i;
5444         for (i = 0; i < q->nr; i++)
5445                 if (!diff_unmodified_pair(q->queue[i]))
5446                         return 0;
5447         return 1;
5448 }
5449
5450 #if DIFF_DEBUG
5451 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5452 {
5453         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5454                 x, one ? one : "",
5455                 s->path,
5456                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5457                 s->mode,
5458                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5459         fprintf(stderr, "queue[%d] %s size %lu\n",
5460                 x, one ? one : "",
5461                 s->size);
5462 }
5463
5464 void diff_debug_filepair(const struct diff_filepair *p, int i)
5465 {
5466         diff_debug_filespec(p->one, i, "one");
5467         diff_debug_filespec(p->two, i, "two");
5468         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5469                 p->score, p->status ? p->status : '?',
5470                 p->one->rename_used, p->broken_pair);
5471 }
5472
5473 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5474 {
5475         int i;
5476         if (msg)
5477                 fprintf(stderr, "%s\n", msg);
5478         fprintf(stderr, "q->nr = %d\n", q->nr);
5479         for (i = 0; i < q->nr; i++) {
5480                 struct diff_filepair *p = q->queue[i];
5481                 diff_debug_filepair(p, i);
5482         }
5483 }
5484 #endif
5485
5486 static void diff_resolve_rename_copy(void)
5487 {
5488         int i;
5489         struct diff_filepair *p;
5490         struct diff_queue_struct *q = &diff_queued_diff;
5491
5492         diff_debug_queue("resolve-rename-copy", q);
5493
5494         for (i = 0; i < q->nr; i++) {
5495                 p = q->queue[i];
5496                 p->status = 0; /* undecided */
5497                 if (DIFF_PAIR_UNMERGED(p))
5498                         p->status = DIFF_STATUS_UNMERGED;
5499                 else if (!DIFF_FILE_VALID(p->one))
5500                         p->status = DIFF_STATUS_ADDED;
5501                 else if (!DIFF_FILE_VALID(p->two))
5502                         p->status = DIFF_STATUS_DELETED;
5503                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5504                         p->status = DIFF_STATUS_TYPE_CHANGED;
5505
5506                 /* from this point on, we are dealing with a pair
5507                  * whose both sides are valid and of the same type, i.e.
5508                  * either in-place edit or rename/copy edit.
5509                  */
5510                 else if (DIFF_PAIR_RENAME(p)) {
5511                         /*
5512                          * A rename might have re-connected a broken
5513                          * pair up, causing the pathnames to be the
5514                          * same again. If so, that's not a rename at
5515                          * all, just a modification..
5516                          *
5517                          * Otherwise, see if this source was used for
5518                          * multiple renames, in which case we decrement
5519                          * the count, and call it a copy.
5520                          */
5521                         if (!strcmp(p->one->path, p->two->path))
5522                                 p->status = DIFF_STATUS_MODIFIED;
5523                         else if (--p->one->rename_used > 0)
5524                                 p->status = DIFF_STATUS_COPIED;
5525                         else
5526                                 p->status = DIFF_STATUS_RENAMED;
5527                 }
5528                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5529                          p->one->mode != p->two->mode ||
5530                          p->one->dirty_submodule ||
5531                          p->two->dirty_submodule ||
5532                          is_null_oid(&p->one->oid))
5533                         p->status = DIFF_STATUS_MODIFIED;
5534                 else {
5535                         /* This is a "no-change" entry and should not
5536                          * happen anymore, but prepare for broken callers.
5537                          */
5538                         error("feeding unmodified %s to diffcore",
5539                               p->one->path);
5540                         p->status = DIFF_STATUS_UNKNOWN;
5541                 }
5542         }
5543         diff_debug_queue("resolve-rename-copy done", q);
5544 }
5545
5546 static int check_pair_status(struct diff_filepair *p)
5547 {
5548         switch (p->status) {
5549         case DIFF_STATUS_UNKNOWN:
5550                 return 0;
5551         case 0:
5552                 die("internal error in diff-resolve-rename-copy");
5553         default:
5554                 return 1;
5555         }
5556 }
5557
5558 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5559 {
5560         int fmt = opt->output_format;
5561
5562         if (fmt & DIFF_FORMAT_CHECKDIFF)
5563                 diff_flush_checkdiff(p, opt);
5564         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5565                 diff_flush_raw(p, opt);
5566         else if (fmt & DIFF_FORMAT_NAME) {
5567                 const char *name_a, *name_b;
5568                 name_a = p->two->path;
5569                 name_b = NULL;
5570                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5571                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5572                 write_name_quoted(name_a, opt->file, opt->line_termination);
5573         }
5574 }
5575
5576 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5577 {
5578         struct strbuf sb = STRBUF_INIT;
5579         if (fs->mode)
5580                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5581         else
5582                 strbuf_addf(&sb, " %s ", newdelete);
5583
5584         quote_c_style(fs->path, &sb, NULL, 0);
5585         strbuf_addch(&sb, '\n');
5586         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5587                          sb.buf, sb.len, 0);
5588         strbuf_release(&sb);
5589 }
5590
5591 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5592                 int show_name)
5593 {
5594         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5595                 struct strbuf sb = STRBUF_INIT;
5596                 strbuf_addf(&sb, " mode change %06o => %06o",
5597                             p->one->mode, p->two->mode);
5598                 if (show_name) {
5599                         strbuf_addch(&sb, ' ');
5600                         quote_c_style(p->two->path, &sb, NULL, 0);
5601                 }
5602                 strbuf_addch(&sb, '\n');
5603                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5604                                  sb.buf, sb.len, 0);
5605                 strbuf_release(&sb);
5606         }
5607 }
5608
5609 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5610                 struct diff_filepair *p)
5611 {
5612         struct strbuf sb = STRBUF_INIT;
5613         struct strbuf names = STRBUF_INIT;
5614
5615         pprint_rename(&names, p->one->path, p->two->path);
5616         strbuf_addf(&sb, " %s %s (%d%%)\n",
5617                     renamecopy, names.buf, similarity_index(p));
5618         strbuf_release(&names);
5619         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5620                                  sb.buf, sb.len, 0);
5621         show_mode_change(opt, p, 0);
5622         strbuf_release(&sb);
5623 }
5624
5625 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5626 {
5627         switch(p->status) {
5628         case DIFF_STATUS_DELETED:
5629                 show_file_mode_name(opt, "delete", p->one);
5630                 break;
5631         case DIFF_STATUS_ADDED:
5632                 show_file_mode_name(opt, "create", p->two);
5633                 break;
5634         case DIFF_STATUS_COPIED:
5635                 show_rename_copy(opt, "copy", p);
5636                 break;
5637         case DIFF_STATUS_RENAMED:
5638                 show_rename_copy(opt, "rename", p);
5639                 break;
5640         default:
5641                 if (p->score) {
5642                         struct strbuf sb = STRBUF_INIT;
5643                         strbuf_addstr(&sb, " rewrite ");
5644                         quote_c_style(p->two->path, &sb, NULL, 0);
5645                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5646                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5647                                          sb.buf, sb.len, 0);
5648                         strbuf_release(&sb);
5649                 }
5650                 show_mode_change(opt, p, !p->score);
5651                 break;
5652         }
5653 }
5654
5655 struct patch_id_t {
5656         git_SHA_CTX *ctx;
5657         int patchlen;
5658 };
5659
5660 static int remove_space(char *line, int len)
5661 {
5662         int i;
5663         char *dst = line;
5664         unsigned char c;
5665
5666         for (i = 0; i < len; i++)
5667                 if (!isspace((c = line[i])))
5668                         *dst++ = c;
5669
5670         return dst - line;
5671 }
5672
5673 static void patch_id_consume(void *priv, char *line, unsigned long len)
5674 {
5675         struct patch_id_t *data = priv;
5676         int new_len;
5677
5678         new_len = remove_space(line, len);
5679
5680         git_SHA1_Update(data->ctx, line, new_len);
5681         data->patchlen += new_len;
5682 }
5683
5684 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5685 {
5686         git_SHA1_Update(ctx, str, strlen(str));
5687 }
5688
5689 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5690 {
5691         /* large enough for 2^32 in octal */
5692         char buf[12];
5693         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5694         git_SHA1_Update(ctx, buf, len);
5695 }
5696
5697 /* returns 0 upon success, and writes result into sha1 */
5698 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5699 {
5700         struct diff_queue_struct *q = &diff_queued_diff;
5701         int i;
5702         git_SHA_CTX ctx;
5703         struct patch_id_t data;
5704
5705         git_SHA1_Init(&ctx);
5706         memset(&data, 0, sizeof(struct patch_id_t));
5707         data.ctx = &ctx;
5708
5709         for (i = 0; i < q->nr; i++) {
5710                 xpparam_t xpp;
5711                 xdemitconf_t xecfg;
5712                 mmfile_t mf1, mf2;
5713                 struct diff_filepair *p = q->queue[i];
5714                 int len1, len2;
5715
5716                 memset(&xpp, 0, sizeof(xpp));
5717                 memset(&xecfg, 0, sizeof(xecfg));
5718                 if (p->status == 0)
5719                         return error("internal diff status error");
5720                 if (p->status == DIFF_STATUS_UNKNOWN)
5721                         continue;
5722                 if (diff_unmodified_pair(p))
5723                         continue;
5724                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5725                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5726                         continue;
5727                 if (DIFF_PAIR_UNMERGED(p))
5728                         continue;
5729
5730                 diff_fill_oid_info(p->one, options->repo->index);
5731                 diff_fill_oid_info(p->two, options->repo->index);
5732
5733                 len1 = remove_space(p->one->path, strlen(p->one->path));
5734                 len2 = remove_space(p->two->path, strlen(p->two->path));
5735                 patch_id_add_string(&ctx, "diff--git");
5736                 patch_id_add_string(&ctx, "a/");
5737                 git_SHA1_Update(&ctx, p->one->path, len1);
5738                 patch_id_add_string(&ctx, "b/");
5739                 git_SHA1_Update(&ctx, p->two->path, len2);
5740
5741                 if (p->one->mode == 0) {
5742                         patch_id_add_string(&ctx, "newfilemode");
5743                         patch_id_add_mode(&ctx, p->two->mode);
5744                         patch_id_add_string(&ctx, "---/dev/null");
5745                         patch_id_add_string(&ctx, "+++b/");
5746                         git_SHA1_Update(&ctx, p->two->path, len2);
5747                 } else if (p->two->mode == 0) {
5748                         patch_id_add_string(&ctx, "deletedfilemode");
5749                         patch_id_add_mode(&ctx, p->one->mode);
5750                         patch_id_add_string(&ctx, "---a/");
5751                         git_SHA1_Update(&ctx, p->one->path, len1);
5752                         patch_id_add_string(&ctx, "+++/dev/null");
5753                 } else {
5754                         patch_id_add_string(&ctx, "---a/");
5755                         git_SHA1_Update(&ctx, p->one->path, len1);
5756                         patch_id_add_string(&ctx, "+++b/");
5757                         git_SHA1_Update(&ctx, p->two->path, len2);
5758                 }
5759
5760                 if (diff_header_only)
5761                         continue;
5762
5763                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5764                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5765                         return error("unable to read files to diff");
5766
5767                 if (diff_filespec_is_binary(options->repo, p->one) ||
5768                     diff_filespec_is_binary(options->repo, p->two)) {
5769                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5770                                         GIT_SHA1_HEXSZ);
5771                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5772                                         GIT_SHA1_HEXSZ);
5773                         continue;
5774                 }
5775
5776                 xpp.flags = 0;
5777                 xecfg.ctxlen = 3;
5778                 xecfg.flags = 0;
5779                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5780                                   patch_id_consume, &data, &xpp, &xecfg))
5781                         return error("unable to generate patch-id diff for %s",
5782                                      p->one->path);
5783         }
5784
5785         git_SHA1_Final(oid->hash, &ctx);
5786         return 0;
5787 }
5788
5789 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5790 {
5791         struct diff_queue_struct *q = &diff_queued_diff;
5792         int i;
5793         int result = diff_get_patch_id(options, oid, diff_header_only);
5794
5795         for (i = 0; i < q->nr; i++)
5796                 diff_free_filepair(q->queue[i]);
5797
5798         free(q->queue);
5799         DIFF_QUEUE_CLEAR(q);
5800
5801         return result;
5802 }
5803
5804 static int is_summary_empty(const struct diff_queue_struct *q)
5805 {
5806         int i;
5807
5808         for (i = 0; i < q->nr; i++) {
5809                 const struct diff_filepair *p = q->queue[i];
5810
5811                 switch (p->status) {
5812                 case DIFF_STATUS_DELETED:
5813                 case DIFF_STATUS_ADDED:
5814                 case DIFF_STATUS_COPIED:
5815                 case DIFF_STATUS_RENAMED:
5816                         return 0;
5817                 default:
5818                         if (p->score)
5819                                 return 0;
5820                         if (p->one->mode && p->two->mode &&
5821                             p->one->mode != p->two->mode)
5822                                 return 0;
5823                         break;
5824                 }
5825         }
5826         return 1;
5827 }
5828
5829 static const char rename_limit_warning[] =
5830 N_("inexact rename detection was skipped due to too many files.");
5831
5832 static const char degrade_cc_to_c_warning[] =
5833 N_("only found copies from modified paths due to too many files.");
5834
5835 static const char rename_limit_advice[] =
5836 N_("you may want to set your %s variable to at least "
5837    "%d and retry the command.");
5838
5839 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5840 {
5841         fflush(stdout);
5842         if (degraded_cc)
5843                 warning(_(degrade_cc_to_c_warning));
5844         else if (needed)
5845                 warning(_(rename_limit_warning));
5846         else
5847                 return;
5848         if (0 < needed)
5849                 warning(_(rename_limit_advice), varname, needed);
5850 }
5851
5852 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5853 {
5854         int i;
5855         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5856         struct diff_queue_struct *q = &diff_queued_diff;
5857
5858         if (WSEH_NEW & WS_RULE_MASK)
5859                 BUG("WS rules bit mask overlaps with diff symbol flags");
5860
5861         if (o->color_moved)
5862                 o->emitted_symbols = &esm;
5863
5864         for (i = 0; i < q->nr; i++) {
5865                 struct diff_filepair *p = q->queue[i];
5866                 if (check_pair_status(p))
5867                         diff_flush_patch(p, o);
5868         }
5869
5870         if (o->emitted_symbols) {
5871                 if (o->color_moved) {
5872                         struct hashmap add_lines, del_lines;
5873
5874                         if (o->color_moved_ws_handling &
5875                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5876                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5877
5878                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5879                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5880
5881                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5882                         mark_color_as_moved(o, &add_lines, &del_lines);
5883                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5884                                 dim_moved_lines(o);
5885
5886                         hashmap_free(&add_lines, 1);
5887                         hashmap_free(&del_lines, 1);
5888                 }
5889
5890                 for (i = 0; i < esm.nr; i++)
5891                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5892
5893                 for (i = 0; i < esm.nr; i++)
5894                         free((void *)esm.buf[i].line);
5895         }
5896         esm.nr = 0;
5897 }
5898
5899 void diff_flush(struct diff_options *options)
5900 {
5901         struct diff_queue_struct *q = &diff_queued_diff;
5902         int i, output_format = options->output_format;
5903         int separator = 0;
5904         int dirstat_by_line = 0;
5905
5906         /*
5907          * Order: raw, stat, summary, patch
5908          * or:    name/name-status/checkdiff (other bits clear)
5909          */
5910         if (!q->nr)
5911                 goto free_queue;
5912
5913         if (output_format & (DIFF_FORMAT_RAW |
5914                              DIFF_FORMAT_NAME |
5915                              DIFF_FORMAT_NAME_STATUS |
5916                              DIFF_FORMAT_CHECKDIFF)) {
5917                 for (i = 0; i < q->nr; i++) {
5918                         struct diff_filepair *p = q->queue[i];
5919                         if (check_pair_status(p))
5920                                 flush_one_pair(p, options);
5921                 }
5922                 separator++;
5923         }
5924
5925         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5926                 dirstat_by_line = 1;
5927
5928         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5929             dirstat_by_line) {
5930                 struct diffstat_t diffstat;
5931
5932                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5933                 for (i = 0; i < q->nr; i++) {
5934                         struct diff_filepair *p = q->queue[i];
5935                         if (check_pair_status(p))
5936                                 diff_flush_stat(p, options, &diffstat);
5937                 }
5938                 if (output_format & DIFF_FORMAT_NUMSTAT)
5939                         show_numstat(&diffstat, options);
5940                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5941                         show_stats(&diffstat, options);
5942                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5943                         show_shortstats(&diffstat, options);
5944                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5945                         show_dirstat_by_line(&diffstat, options);
5946                 free_diffstat_info(&diffstat);
5947                 separator++;
5948         }
5949         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5950                 show_dirstat(options);
5951
5952         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5953                 for (i = 0; i < q->nr; i++) {
5954                         diff_summary(options, q->queue[i]);
5955                 }
5956                 separator++;
5957         }
5958
5959         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5960             options->flags.exit_with_status &&
5961             options->flags.diff_from_contents) {
5962                 /*
5963                  * run diff_flush_patch for the exit status. setting
5964                  * options->file to /dev/null should be safe, because we
5965                  * aren't supposed to produce any output anyway.
5966                  */
5967                 if (options->close_file)
5968                         fclose(options->file);
5969                 options->file = xfopen("/dev/null", "w");
5970                 options->close_file = 1;
5971                 options->color_moved = 0;
5972                 for (i = 0; i < q->nr; i++) {
5973                         struct diff_filepair *p = q->queue[i];
5974                         if (check_pair_status(p))
5975                                 diff_flush_patch(p, options);
5976                         if (options->found_changes)
5977                                 break;
5978                 }
5979         }
5980
5981         if (output_format & DIFF_FORMAT_PATCH) {
5982                 if (separator) {
5983                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5984                         if (options->stat_sep)
5985                                 /* attach patch instead of inline */
5986                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5987                                                  NULL, 0, 0);
5988                 }
5989
5990                 diff_flush_patch_all_file_pairs(options);
5991         }
5992
5993         if (output_format & DIFF_FORMAT_CALLBACK)
5994                 options->format_callback(q, options, options->format_callback_data);
5995
5996         for (i = 0; i < q->nr; i++)
5997                 diff_free_filepair(q->queue[i]);
5998 free_queue:
5999         free(q->queue);
6000         DIFF_QUEUE_CLEAR(q);
6001         if (options->close_file)
6002                 fclose(options->file);
6003
6004         /*
6005          * Report the content-level differences with HAS_CHANGES;
6006          * diff_addremove/diff_change does not set the bit when
6007          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6008          */
6009         if (options->flags.diff_from_contents) {
6010                 if (options->found_changes)
6011                         options->flags.has_changes = 1;
6012                 else
6013                         options->flags.has_changes = 0;
6014         }
6015 }
6016
6017 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6018 {
6019         return (((p->status == DIFF_STATUS_MODIFIED) &&
6020                  ((p->score &&
6021                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6022                   (!p->score &&
6023                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6024                 ((p->status != DIFF_STATUS_MODIFIED) &&
6025                  filter_bit_tst(p->status, options)));
6026 }
6027
6028 static void diffcore_apply_filter(struct diff_options *options)
6029 {
6030         int i;
6031         struct diff_queue_struct *q = &diff_queued_diff;
6032         struct diff_queue_struct outq;
6033
6034         DIFF_QUEUE_CLEAR(&outq);
6035
6036         if (!options->filter)
6037                 return;
6038
6039         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6040                 int found;
6041                 for (i = found = 0; !found && i < q->nr; i++) {
6042                         if (match_filter(options, q->queue[i]))
6043                                 found++;
6044                 }
6045                 if (found)
6046                         return;
6047
6048                 /* otherwise we will clear the whole queue
6049                  * by copying the empty outq at the end of this
6050                  * function, but first clear the current entries
6051                  * in the queue.
6052                  */
6053                 for (i = 0; i < q->nr; i++)
6054                         diff_free_filepair(q->queue[i]);
6055         }
6056         else {
6057                 /* Only the matching ones */
6058                 for (i = 0; i < q->nr; i++) {
6059                         struct diff_filepair *p = q->queue[i];
6060                         if (match_filter(options, p))
6061                                 diff_q(&outq, p);
6062                         else
6063                                 diff_free_filepair(p);
6064                 }
6065         }
6066         free(q->queue);
6067         *q = outq;
6068 }
6069
6070 /* Check whether two filespecs with the same mode and size are identical */
6071 static int diff_filespec_is_identical(struct repository *r,
6072                                       struct diff_filespec *one,
6073                                       struct diff_filespec *two)
6074 {
6075         if (S_ISGITLINK(one->mode))
6076                 return 0;
6077         if (diff_populate_filespec(r, one, 0))
6078                 return 0;
6079         if (diff_populate_filespec(r, two, 0))
6080                 return 0;
6081         return !memcmp(one->data, two->data, one->size);
6082 }
6083
6084 static int diff_filespec_check_stat_unmatch(struct repository *r,
6085                                             struct diff_filepair *p)
6086 {
6087         if (p->done_skip_stat_unmatch)
6088                 return p->skip_stat_unmatch_result;
6089
6090         p->done_skip_stat_unmatch = 1;
6091         p->skip_stat_unmatch_result = 0;
6092         /*
6093          * 1. Entries that come from stat info dirtiness
6094          *    always have both sides (iow, not create/delete),
6095          *    one side of the object name is unknown, with
6096          *    the same mode and size.  Keep the ones that
6097          *    do not match these criteria.  They have real
6098          *    differences.
6099          *
6100          * 2. At this point, the file is known to be modified,
6101          *    with the same mode and size, and the object
6102          *    name of one side is unknown.  Need to inspect
6103          *    the identical contents.
6104          */
6105         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6106             !DIFF_FILE_VALID(p->two) ||
6107             (p->one->oid_valid && p->two->oid_valid) ||
6108             (p->one->mode != p->two->mode) ||
6109             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6110             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6111             (p->one->size != p->two->size) ||
6112             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6113                 p->skip_stat_unmatch_result = 1;
6114         return p->skip_stat_unmatch_result;
6115 }
6116
6117 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6118 {
6119         int i;
6120         struct diff_queue_struct *q = &diff_queued_diff;
6121         struct diff_queue_struct outq;
6122         DIFF_QUEUE_CLEAR(&outq);
6123
6124         for (i = 0; i < q->nr; i++) {
6125                 struct diff_filepair *p = q->queue[i];
6126
6127                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6128                         diff_q(&outq, p);
6129                 else {
6130                         /*
6131                          * The caller can subtract 1 from skip_stat_unmatch
6132                          * to determine how many paths were dirty only
6133                          * due to stat info mismatch.
6134                          */
6135                         if (!diffopt->flags.no_index)
6136                                 diffopt->skip_stat_unmatch++;
6137                         diff_free_filepair(p);
6138                 }
6139         }
6140         free(q->queue);
6141         *q = outq;
6142 }
6143
6144 static int diffnamecmp(const void *a_, const void *b_)
6145 {
6146         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6147         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6148         const char *name_a, *name_b;
6149
6150         name_a = a->one ? a->one->path : a->two->path;
6151         name_b = b->one ? b->one->path : b->two->path;
6152         return strcmp(name_a, name_b);
6153 }
6154
6155 void diffcore_fix_diff_index(struct diff_options *options)
6156 {
6157         struct diff_queue_struct *q = &diff_queued_diff;
6158         QSORT(q->queue, q->nr, diffnamecmp);
6159 }
6160
6161 void diffcore_std(struct diff_options *options)
6162 {
6163         /* NOTE please keep the following in sync with diff_tree_combined() */
6164         if (options->skip_stat_unmatch)
6165                 diffcore_skip_stat_unmatch(options);
6166         if (!options->found_follow) {
6167                 /* See try_to_follow_renames() in tree-diff.c */
6168                 if (options->break_opt != -1)
6169                         diffcore_break(options->repo,
6170                                        options->break_opt);
6171                 if (options->detect_rename)
6172                         diffcore_rename(options);
6173                 if (options->break_opt != -1)
6174                         diffcore_merge_broken();
6175         }
6176         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6177                 diffcore_pickaxe(options);
6178         if (options->orderfile)
6179                 diffcore_order(options->orderfile);
6180         if (!options->found_follow)
6181                 /* See try_to_follow_renames() in tree-diff.c */
6182                 diff_resolve_rename_copy();
6183         diffcore_apply_filter(options);
6184
6185         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6186                 options->flags.has_changes = 1;
6187         else
6188                 options->flags.has_changes = 0;
6189
6190         options->found_follow = 0;
6191 }
6192
6193 int diff_result_code(struct diff_options *opt, int status)
6194 {
6195         int result = 0;
6196
6197         diff_warn_rename_limit("diff.renameLimit",
6198                                opt->needed_rename_limit,
6199                                opt->degraded_cc_to_c);
6200         if (!opt->flags.exit_with_status &&
6201             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6202                 return status;
6203         if (opt->flags.exit_with_status &&
6204             opt->flags.has_changes)
6205                 result |= 01;
6206         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6207             opt->flags.check_failed)
6208                 result |= 02;
6209         return result;
6210 }
6211
6212 int diff_can_quit_early(struct diff_options *opt)
6213 {
6214         return (opt->flags.quick &&
6215                 !opt->filter &&
6216                 opt->flags.has_changes);
6217 }
6218
6219 /*
6220  * Shall changes to this submodule be ignored?
6221  *
6222  * Submodule changes can be configured to be ignored separately for each path,
6223  * but that configuration can be overridden from the command line.
6224  */
6225 static int is_submodule_ignored(const char *path, struct diff_options *options)
6226 {
6227         int ignored = 0;
6228         struct diff_flags orig_flags = options->flags;
6229         if (!options->flags.override_submodule_config)
6230                 set_diffopt_flags_from_submodule_config(options, path);
6231         if (options->flags.ignore_submodules)
6232                 ignored = 1;
6233         options->flags = orig_flags;
6234         return ignored;
6235 }
6236
6237 void diff_addremove(struct diff_options *options,
6238                     int addremove, unsigned mode,
6239                     const struct object_id *oid,
6240                     int oid_valid,
6241                     const char *concatpath, unsigned dirty_submodule)
6242 {
6243         struct diff_filespec *one, *two;
6244
6245         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6246                 return;
6247
6248         /* This may look odd, but it is a preparation for
6249          * feeding "there are unchanged files which should
6250          * not produce diffs, but when you are doing copy
6251          * detection you would need them, so here they are"
6252          * entries to the diff-core.  They will be prefixed
6253          * with something like '=' or '*' (I haven't decided
6254          * which but should not make any difference).
6255          * Feeding the same new and old to diff_change()
6256          * also has the same effect.
6257          * Before the final output happens, they are pruned after
6258          * merged into rename/copy pairs as appropriate.
6259          */
6260         if (options->flags.reverse_diff)
6261                 addremove = (addremove == '+' ? '-' :
6262                              addremove == '-' ? '+' : addremove);
6263
6264         if (options->prefix &&
6265             strncmp(concatpath, options->prefix, options->prefix_length))
6266                 return;
6267
6268         one = alloc_filespec(concatpath);
6269         two = alloc_filespec(concatpath);
6270
6271         if (addremove != '+')
6272                 fill_filespec(one, oid, oid_valid, mode);
6273         if (addremove != '-') {
6274                 fill_filespec(two, oid, oid_valid, mode);
6275                 two->dirty_submodule = dirty_submodule;
6276         }
6277
6278         diff_queue(&diff_queued_diff, one, two);
6279         if (!options->flags.diff_from_contents)
6280                 options->flags.has_changes = 1;
6281 }
6282
6283 void diff_change(struct diff_options *options,
6284                  unsigned old_mode, unsigned new_mode,
6285                  const struct object_id *old_oid,
6286                  const struct object_id *new_oid,
6287                  int old_oid_valid, int new_oid_valid,
6288                  const char *concatpath,
6289                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6290 {
6291         struct diff_filespec *one, *two;
6292         struct diff_filepair *p;
6293
6294         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6295             is_submodule_ignored(concatpath, options))
6296                 return;
6297
6298         if (options->flags.reverse_diff) {
6299                 SWAP(old_mode, new_mode);
6300                 SWAP(old_oid, new_oid);
6301                 SWAP(old_oid_valid, new_oid_valid);
6302                 SWAP(old_dirty_submodule, new_dirty_submodule);
6303         }
6304
6305         if (options->prefix &&
6306             strncmp(concatpath, options->prefix, options->prefix_length))
6307                 return;
6308
6309         one = alloc_filespec(concatpath);
6310         two = alloc_filespec(concatpath);
6311         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6312         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6313         one->dirty_submodule = old_dirty_submodule;
6314         two->dirty_submodule = new_dirty_submodule;
6315         p = diff_queue(&diff_queued_diff, one, two);
6316
6317         if (options->flags.diff_from_contents)
6318                 return;
6319
6320         if (options->flags.quick && options->skip_stat_unmatch &&
6321             !diff_filespec_check_stat_unmatch(options->repo, p))
6322                 return;
6323
6324         options->flags.has_changes = 1;
6325 }
6326
6327 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6328 {
6329         struct diff_filepair *pair;
6330         struct diff_filespec *one, *two;
6331
6332         if (options->prefix &&
6333             strncmp(path, options->prefix, options->prefix_length))
6334                 return NULL;
6335
6336         one = alloc_filespec(path);
6337         two = alloc_filespec(path);
6338         pair = diff_queue(&diff_queued_diff, one, two);
6339         pair->is_unmerged = 1;
6340         return pair;
6341 }
6342
6343 static char *run_textconv(struct repository *r,
6344                           const char *pgm,
6345                           struct diff_filespec *spec,
6346                           size_t *outsize)
6347 {
6348         struct diff_tempfile *temp;
6349         const char *argv[3];
6350         const char **arg = argv;
6351         struct child_process child = CHILD_PROCESS_INIT;
6352         struct strbuf buf = STRBUF_INIT;
6353         int err = 0;
6354
6355         temp = prepare_temp_file(r, spec->path, spec);
6356         *arg++ = pgm;
6357         *arg++ = temp->name;
6358         *arg = NULL;
6359
6360         child.use_shell = 1;
6361         child.argv = argv;
6362         child.out = -1;
6363         if (start_command(&child)) {
6364                 remove_tempfile();
6365                 return NULL;
6366         }
6367
6368         if (strbuf_read(&buf, child.out, 0) < 0)
6369                 err = error("error reading from textconv command '%s'", pgm);
6370         close(child.out);
6371
6372         if (finish_command(&child) || err) {
6373                 strbuf_release(&buf);
6374                 remove_tempfile();
6375                 return NULL;
6376         }
6377         remove_tempfile();
6378
6379         return strbuf_detach(&buf, outsize);
6380 }
6381
6382 size_t fill_textconv(struct repository *r,
6383                      struct userdiff_driver *driver,
6384                      struct diff_filespec *df,
6385                      char **outbuf)
6386 {
6387         size_t size;
6388
6389         if (!driver) {
6390                 if (!DIFF_FILE_VALID(df)) {
6391                         *outbuf = "";
6392                         return 0;
6393                 }
6394                 if (diff_populate_filespec(r, df, 0))
6395                         die("unable to read files to diff");
6396                 *outbuf = df->data;
6397                 return df->size;
6398         }
6399
6400         if (!driver->textconv)
6401                 BUG("fill_textconv called with non-textconv driver");
6402
6403         if (driver->textconv_cache && df->oid_valid) {
6404                 *outbuf = notes_cache_get(driver->textconv_cache,
6405                                           &df->oid,
6406                                           &size);
6407                 if (*outbuf)
6408                         return size;
6409         }
6410
6411         *outbuf = run_textconv(r, driver->textconv, df, &size);
6412         if (!*outbuf)
6413                 die("unable to read files to diff");
6414
6415         if (driver->textconv_cache && df->oid_valid) {
6416                 /* ignore errors, as we might be in a readonly repository */
6417                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6418                                 size);
6419                 /*
6420                  * we could save up changes and flush them all at the end,
6421                  * but we would need an extra call after all diffing is done.
6422                  * Since generating a cache entry is the slow path anyway,
6423                  * this extra overhead probably isn't a big deal.
6424                  */
6425                 notes_cache_write(driver->textconv_cache);
6426         }
6427
6428         return size;
6429 }
6430
6431 int textconv_object(struct repository *r,
6432                     const char *path,
6433                     unsigned mode,
6434                     const struct object_id *oid,
6435                     int oid_valid,
6436                     char **buf,
6437                     unsigned long *buf_size)
6438 {
6439         struct diff_filespec *df;
6440         struct userdiff_driver *textconv;
6441
6442         df = alloc_filespec(path);
6443         fill_filespec(df, oid, oid_valid, mode);
6444         textconv = get_textconv(r, df);
6445         if (!textconv) {
6446                 free_filespec(df);
6447                 return 0;
6448         }
6449
6450         *buf_size = fill_textconv(r, textconv, df, buf);
6451         free_filespec(df);
6452         return 1;
6453 }
6454
6455 void setup_diff_pager(struct diff_options *opt)
6456 {
6457         /*
6458          * If the user asked for our exit code, then either they want --quiet
6459          * or --exit-code. We should definitely not bother with a pager in the
6460          * former case, as we will generate no output. Since we still properly
6461          * report our exit code even when a pager is run, we _could_ run a
6462          * pager with --exit-code. But since we have not done so historically,
6463          * and because it is easy to find people oneline advising "git diff
6464          * --exit-code" in hooks and other scripts, we do not do so.
6465          */
6466         if (!opt->flags.exit_with_status &&
6467             check_pager_config("diff") != 0)
6468                 setup_pager();
6469 }