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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed-zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else if (!strcmp(arg, "dimmed_zebra"))
 289                 return COLOR_MOVED_ZEBRA_DIM;
 290         else
 291                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 292 }
 293
 294 static int parse_color_moved_ws(const char *arg)
 295 {
 296         int ret = 0;
 297         struct string_list l = STRING_LIST_INIT_DUP;
 298         struct string_list_item *i;
 299
 300         string_list_split(&l, arg, ',', -1);
 301
 302         for_each_string_list_item(i, &l) {
 303                 struct strbuf sb = STRBUF_INIT;
 304                 strbuf_addstr(&sb, i->string);
 305                 strbuf_trim(&sb);
 306
 307                 if (!strcmp(sb.buf, "no"))
 308                         ret = 0;
 309                 else if (!strcmp(sb.buf, "ignore-space-change"))
 310                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 311                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 312                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 313                 else if (!strcmp(sb.buf, "ignore-all-space"))
 314                         ret |= XDF_IGNORE_WHITESPACE;
 315                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 316                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 317                 else
 318                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 319
 320                 strbuf_release(&sb);
 321         }
 322
 323         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 324             (ret & XDF_WHITESPACE_FLAGS))
 325                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 326
 327         string_list_clear(&l, 0);
 328
 329         return ret;
 330 }
 331
 332 int git_diff_ui_config(const char *var, const char *value, void *cb)
 333 {
 334         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 335                 diff_use_color_default = git_config_colorbool(var, value);
 336                 return 0;
 337         }
 338         if (!strcmp(var, "diff.colormoved")) {
 339                 int cm = parse_color_moved(value);
 340                 if (cm < 0)
 341                         return -1;
 342                 diff_color_moved_default = cm;
 343                 return 0;
 344         }
 345         if (!strcmp(var, "diff.colormovedws")) {
 346                 int cm = parse_color_moved_ws(value);
 347                 if (cm < 0)
 348                         return -1;
 349                 diff_color_moved_ws_default = cm;
 350                 return 0;
 351         }
 352         if (!strcmp(var, "diff.context")) {
 353                 diff_context_default = git_config_int(var, value);
 354                 if (diff_context_default < 0)
 355                         return -1;
 356                 return 0;
 357         }
 358         if (!strcmp(var, "diff.interhunkcontext")) {
 359                 diff_interhunk_context_default = git_config_int(var, value);
 360                 if (diff_interhunk_context_default < 0)
 361                         return -1;
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.renames")) {
 365                 diff_detect_rename_default = git_config_rename(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.autorefreshindex")) {
 369                 diff_auto_refresh_index = git_config_bool(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.mnemonicprefix")) {
 373                 diff_mnemonic_prefix = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.noprefix")) {
 377                 diff_no_prefix = git_config_bool(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.statgraphwidth")) {
 381                 diff_stat_graph_width = git_config_int(var, value);
 382                 return 0;
 383         }
 384         if (!strcmp(var, "diff.external"))
 385                 return git_config_string(&external_diff_cmd_cfg, var, value);
 386         if (!strcmp(var, "diff.wordregex"))
 387                 return git_config_string(&diff_word_regex_cfg, var, value);
 388         if (!strcmp(var, "diff.orderfile"))
 389                 return git_config_pathname(&diff_order_file_cfg, var, value);
 390
 391         if (!strcmp(var, "diff.ignoresubmodules"))
 392                 handle_ignore_submodules_arg(&default_diff_options, value);
 393
 394         if (!strcmp(var, "diff.submodule")) {
 395                 if (parse_submodule_params(&default_diff_options, value))
 396                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 397                                 value);
 398                 return 0;
 399         }
 400
 401         if (!strcmp(var, "diff.algorithm")) {
 402                 diff_algorithm = parse_algorithm_value(value);
 403                 if (diff_algorithm < 0)
 404                         return -1;
 405                 return 0;
 406         }
 407
 408         if (!strcmp(var, "diff.wserrorhighlight")) {
 409                 int val = parse_ws_error_highlight(value);
 410                 if (val < 0)
 411                         return -1;
 412                 ws_error_highlight_default = val;
 413                 return 0;
 414         }
 415
 416         if (git_color_config(var, value, cb) < 0)
 417                 return -1;
 418
 419         return git_diff_basic_config(var, value, cb);
 420 }
 421
 422 int git_diff_basic_config(const char *var, const char *value, void *cb)
 423 {
 424         const char *name;
 425
 426         if (!strcmp(var, "diff.renamelimit")) {
 427                 diff_rename_limit_default = git_config_int(var, value);
 428                 return 0;
 429         }
 430
 431         if (userdiff_config(var, value) < 0)
 432                 return -1;
 433
 434         if (skip_prefix(var, "diff.color.", &name) ||
 435             skip_prefix(var, "color.diff.", &name)) {
 436                 int slot = parse_diff_color_slot(name);
 437                 if (slot < 0)
 438                         return 0;
 439                 if (!value)
 440                         return config_error_nonbool(var);
 441                 return color_parse(value, diff_colors[slot]);
 442         }
 443
 444         /* like GNU diff's --suppress-blank-empty option  */
 445         if (!strcmp(var, "diff.suppressblankempty") ||
 446                         /* for backwards compatibility */
 447                         !strcmp(var, "diff.suppress-blank-empty")) {
 448                 diff_suppress_blank_empty = git_config_bool(var, value);
 449                 return 0;
 450         }
 451
 452         if (!strcmp(var, "diff.dirstat")) {
 453                 struct strbuf errmsg = STRBUF_INIT;
 454                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 455                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 456                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 457                                 errmsg.buf);
 458                 strbuf_release(&errmsg);
 459                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 460                 return 0;
 461         }
 462
 463         if (git_diff_heuristic_config(var, value, cb) < 0)
 464                 return -1;
 465
 466         return git_default_config(var, value, cb);
 467 }
 468
 469 static char *quote_two(const char *one, const char *two)
 470 {
 471         int need_one = quote_c_style(one, NULL, NULL, 1);
 472         int need_two = quote_c_style(two, NULL, NULL, 1);
 473         struct strbuf res = STRBUF_INIT;
 474
 475         if (need_one + need_two) {
 476                 strbuf_addch(&res, '"');
 477                 quote_c_style(one, &res, NULL, 1);
 478                 quote_c_style(two, &res, NULL, 1);
 479                 strbuf_addch(&res, '"');
 480         } else {
 481                 strbuf_addstr(&res, one);
 482                 strbuf_addstr(&res, two);
 483         }
 484         return strbuf_detach(&res, NULL);
 485 }
 486
 487 static const char *external_diff(void)
 488 {
 489         static const char *external_diff_cmd = NULL;
 490         static int done_preparing = 0;
 491
 492         if (done_preparing)
 493                 return external_diff_cmd;
 494         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 495         if (!external_diff_cmd)
 496                 external_diff_cmd = external_diff_cmd_cfg;
 497         done_preparing = 1;
 498         return external_diff_cmd;
 499 }
 500
 501 /*
 502  * Keep track of files used for diffing. Sometimes such an entry
 503  * refers to a temporary file, sometimes to an existing file, and
 504  * sometimes to "/dev/null".
 505  */
 506 static struct diff_tempfile {
 507         /*
 508          * filename external diff should read from, or NULL if this
 509          * entry is currently not in use:
 510          */
 511         const char *name;
 512
 513         char hex[GIT_MAX_HEXSZ + 1];
 514         char mode[10];
 515
 516         /*
 517          * If this diff_tempfile instance refers to a temporary file,
 518          * this tempfile object is used to manage its lifetime.
 519          */
 520         struct tempfile *tempfile;
 521 } diff_temp[2];
 522
 523 struct emit_callback {
 524         int color_diff;
 525         unsigned ws_rule;
 526         int blank_at_eof_in_preimage;
 527         int blank_at_eof_in_postimage;
 528         int lno_in_preimage;
 529         int lno_in_postimage;
 530         const char **label_path;
 531         struct diff_words_data *diff_words;
 532         struct diff_options *opt;
 533         struct strbuf *header;
 534 };
 535
 536 static int count_lines(const char *data, int size)
 537 {
 538         int count, ch, completely_empty = 1, nl_just_seen = 0;
 539         count = 0;
 540         while (0 < size--) {
 541                 ch = *data++;
 542                 if (ch == '\n') {
 543                         count++;
 544                         nl_just_seen = 1;
 545                         completely_empty = 0;
 546                 }
 547                 else {
 548                         nl_just_seen = 0;
 549                         completely_empty = 0;
 550                 }
 551         }
 552         if (completely_empty)
 553                 return 0;
 554         if (!nl_just_seen)
 555                 count++; /* no trailing newline */
 556         return count;
 557 }
 558
 559 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 560                        struct diff_filespec *one)
 561 {
 562         if (!DIFF_FILE_VALID(one)) {
 563                 mf->ptr = (char *)""; /* does not matter */
 564                 mf->size = 0;
 565                 return 0;
 566         }
 567         else if (diff_populate_filespec(r, one, 0))
 568                 return -1;
 569
 570         mf->ptr = one->data;
 571         mf->size = one->size;
 572         return 0;
 573 }
 574
 575 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 576 static unsigned long diff_filespec_size(struct repository *r,
 577                                         struct diff_filespec *one)
 578 {
 579         if (!DIFF_FILE_VALID(one))
 580                 return 0;
 581         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 582         return one->size;
 583 }
 584
 585 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 586 {
 587         char *ptr = mf->ptr;
 588         long size = mf->size;
 589         int cnt = 0;
 590
 591         if (!size)
 592                 return cnt;
 593         ptr += size - 1; /* pointing at the very end */
 594         if (*ptr != '\n')
 595                 ; /* incomplete line */
 596         else
 597                 ptr--; /* skip the last LF */
 598         while (mf->ptr < ptr) {
 599                 char *prev_eol;
 600                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 601                         if (*prev_eol == '\n')
 602                                 break;
 603                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 604                         break;
 605                 cnt++;
 606                 ptr = prev_eol - 1;
 607         }
 608         return cnt;
 609 }
 610
 611 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 612                                struct emit_callback *ecbdata)
 613 {
 614         int l1, l2, at;
 615         unsigned ws_rule = ecbdata->ws_rule;
 616         l1 = count_trailing_blank(mf1, ws_rule);
 617         l2 = count_trailing_blank(mf2, ws_rule);
 618         if (l2 <= l1) {
 619                 ecbdata->blank_at_eof_in_preimage = 0;
 620                 ecbdata->blank_at_eof_in_postimage = 0;
 621                 return;
 622         }
 623         at = count_lines(mf1->ptr, mf1->size);
 624         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 625
 626         at = count_lines(mf2->ptr, mf2->size);
 627         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 628 }
 629
 630 static void emit_line_0(struct diff_options *o,
 631                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 632                         int first, const char *line, int len)
 633 {
 634         int has_trailing_newline, has_trailing_carriage_return;
 635         int needs_reset = 0; /* at the end of the line */
 636         FILE *file = o->file;
 637
 638         fputs(diff_line_prefix(o), file);
 639
 640         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 641         if (has_trailing_newline)
 642                 len--;
 643
 644         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 645         if (has_trailing_carriage_return)
 646                 len--;
 647
 648         if (!len && !first)
 649                 goto end_of_line;
 650
 651         if (reverse && want_color(o->use_color)) {
 652                 fputs(GIT_COLOR_REVERSE, file);
 653                 needs_reset = 1;
 654         }
 655
 656         if (set_sign) {
 657                 fputs(set_sign, file);
 658                 needs_reset = 1;
 659         }
 660
 661         if (first)
 662                 fputc(first, file);
 663
 664         if (!len)
 665                 goto end_of_line;
 666
 667         if (set) {
 668                 if (set_sign && set != set_sign)
 669                         fputs(reset, file);
 670                 fputs(set, file);
 671                 needs_reset = 1;
 672         }
 673         fwrite(line, len, 1, file);
 674         needs_reset = 1; /* 'line' may contain color codes. */
 675
 676 end_of_line:
 677         if (needs_reset)
 678                 fputs(reset, file);
 679         if (has_trailing_carriage_return)
 680                 fputc('\r', file);
 681         if (has_trailing_newline)
 682                 fputc('\n', file);
 683 }
 684
 685 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 686                       const char *line, int len)
 687 {
 688         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 689 }
 690
 691 enum diff_symbol {
 692         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 693         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 694         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 695         DIFF_SYMBOL_BINARY_DIFF_BODY,
 696         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 697         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 698         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 699         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 700         DIFF_SYMBOL_STATS_LINE,
 701         DIFF_SYMBOL_WORD_DIFF,
 702         DIFF_SYMBOL_STAT_SEP,
 703         DIFF_SYMBOL_SUMMARY,
 704         DIFF_SYMBOL_SUBMODULE_ADD,
 705         DIFF_SYMBOL_SUBMODULE_DEL,
 706         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 707         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 708         DIFF_SYMBOL_SUBMODULE_HEADER,
 709         DIFF_SYMBOL_SUBMODULE_ERROR,
 710         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 711         DIFF_SYMBOL_REWRITE_DIFF,
 712         DIFF_SYMBOL_BINARY_FILES,
 713         DIFF_SYMBOL_HEADER,
 714         DIFF_SYMBOL_FILEPAIR_PLUS,
 715         DIFF_SYMBOL_FILEPAIR_MINUS,
 716         DIFF_SYMBOL_WORDS_PORCELAIN,
 717         DIFF_SYMBOL_WORDS,
 718         DIFF_SYMBOL_CONTEXT,
 719         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 720         DIFF_SYMBOL_PLUS,
 721         DIFF_SYMBOL_MINUS,
 722         DIFF_SYMBOL_NO_LF_EOF,
 723         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 724         DIFF_SYMBOL_CONTEXT_MARKER,
 725         DIFF_SYMBOL_SEPARATOR
 726 };
 727 /*
 728  * Flags for content lines:
 729  * 0..12 are whitespace rules
 730  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 731  * 16 is marking if the line is blank at EOF
 732  */
 733 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 734 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 735 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 736 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 737 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 738
 739 /*
 740  * This struct is used when we need to buffer the output of the diff output.
 741  *
 742  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 743  * into the pre/post image file. This pointer could be a union with the
 744  * line pointer. By storing an offset into the file instead of the literal line,
 745  * we can decrease the memory footprint for the buffered output. At first we
 746  * may want to only have indirection for the content lines, but we could also
 747  * enhance the state for emitting prefabricated lines, e.g. the similarity
 748  * score line or hunk/file headers would only need to store a number or path
 749  * and then the output can be constructed later on depending on state.
 750  */
 751 struct emitted_diff_symbol {
 752         const char *line;
 753         int len;
 754         int flags;
 755         int indent_off;   /* Offset to first non-whitespace character */
 756         int indent_width; /* The visual width of the indentation */
 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 struct moved_block {
 788         struct moved_entry *match;
 789         int wsd; /* The whitespace delta of this block */
 790 };
 791
 792 static void moved_block_clear(struct moved_block *b)
 793 {
 794         memset(b, 0, sizeof(*b));
 795 }
 796
 797 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 798 {
 799         unsigned int off = 0;
 800         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 801         const char *s = es->line;
 802         const int len = es->len;
 803
 804         /* skip any \v \f \r at start of indentation */
 805         while (s[off] == '\f' || s[off] == '\v' ||
 806                (s[off] == '\r' && off < len - 1))
 807                 off++;
 808
 809         /* calculate the visual width of indentation */
 810         while(1) {
 811                 if (s[off] == ' ') {
 812                         width++;
 813                         off++;
 814                 } else if (s[off] == '\t') {
 815                         width += tab_width - (width % tab_width);
 816                         while (s[++off] == '\t')
 817                                 width += tab_width;
 818                 } else {
 819                         break;
 820                 }
 821         }
 822
 823         es->indent_off = off;
 824         es->indent_width = width;
 825 }
 826
 827 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 828                             const struct emitted_diff_symbol *b,
 829                             int *out)
 830 {
 831         int a_len = a->len,
 832             b_len = b->len,
 833             a_off = a->indent_off,
 834             a_width = a->indent_width,
 835             b_off = b->indent_off,
 836             b_width = b->indent_width;
 837         int delta;
 838
 839         if (a->s == DIFF_SYMBOL_PLUS)
 840                 delta = a_width - b_width;
 841         else
 842                 delta = b_width - a_width;
 843
 844         if (a_len - a_off != b_len - b_off ||
 845             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 846                 return 0;
 847
 848         *out = delta;
 849
 850         return 1;
 851 }
 852
 853 static int cmp_in_block_with_wsd(const struct diff_options *o,
 854                                  const struct moved_entry *cur,
 855                                  const struct moved_entry *match,
 856                                  struct moved_block *pmb,
 857                                  int n)
 858 {
 859         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 860         int al = cur->es->len, bl = match->es->len, cl = l->len;
 861         const char *a = cur->es->line,
 862                    *b = match->es->line,
 863                    *c = l->line;
 864         int a_off = cur->es->indent_off,
 865             a_width = cur->es->indent_width,
 866             c_off = l->indent_off,
 867             c_width = l->indent_width;
 868         int delta;
 869
 870         /*
 871          * We need to check if 'cur' is equal to 'match'.  As those
 872          * are from the same (+/-) side, we do not need to adjust for
 873          * indent changes. However these were found using fuzzy
 874          * matching so we do have to check if they are equal. Here we
 875          * just check the lengths. We delay calling memcmp() to check
 876          * the contents until later as if the length comparison for a
 877          * and c fails we can avoid the call all together.
 878          */
 879         if (al != bl)
 880                 return 1;
 881
 882         /*
 883          * The indent changes of the block are known and stored in pmb->wsd;
 884          * however we need to check if the indent changes of the current line
 885          * match those of the current block and that the text of 'l' and 'cur'
 886          * after the indentation match.
 887          */
 888         if (cur->es->s == DIFF_SYMBOL_PLUS)
 889                 delta = a_width - c_width;
 890         else
 891                 delta = c_width - a_width;
 892
 893         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 894                  !memcmp(a, b, al) && !
 895                  memcmp(a + a_off, c + c_off, al - a_off));
 896 }
 897
 898 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 899                            const void *entry,
 900                            const void *entry_or_key,
 901                            const void *keydata)
 902 {
 903         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 904         const struct moved_entry *a = entry;
 905         const struct moved_entry *b = entry_or_key;
 906         unsigned flags = diffopt->color_moved_ws_handling
 907                          & XDF_WHITESPACE_FLAGS;
 908
 909         if (diffopt->color_moved_ws_handling &
 910             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 911                 /*
 912                  * As there is not specific white space config given,
 913                  * we'd need to check for a new block, so ignore all
 914                  * white space. The setup of the white space
 915                  * configuration for the next block is done else where
 916                  */
 917                 flags |= XDF_IGNORE_WHITESPACE;
 918
 919         return !xdiff_compare_lines(a->es->line, a->es->len,
 920                                     b->es->line, b->es->len,
 921                                     flags);
 922 }
 923
 924 static struct moved_entry *prepare_entry(struct diff_options *o,
 925                                          int line_no)
 926 {
 927         struct moved_entry *ret = xmalloc(sizeof(*ret));
 928         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 929         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 930
 931         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 932         ret->es = l;
 933         ret->next_line = NULL;
 934
 935         return ret;
 936 }
 937
 938 static void add_lines_to_move_detection(struct diff_options *o,
 939                                         struct hashmap *add_lines,
 940                                         struct hashmap *del_lines)
 941 {
 942         struct moved_entry *prev_line = NULL;
 943
 944         int n;
 945         for (n = 0; n < o->emitted_symbols->nr; n++) {
 946                 struct hashmap *hm;
 947                 struct moved_entry *key;
 948
 949                 switch (o->emitted_symbols->buf[n].s) {
 950                 case DIFF_SYMBOL_PLUS:
 951                         hm = add_lines;
 952                         break;
 953                 case DIFF_SYMBOL_MINUS:
 954                         hm = del_lines;
 955                         break;
 956                 default:
 957                         prev_line = NULL;
 958                         continue;
 959                 }
 960
 961                 if (o->color_moved_ws_handling &
 962                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 963                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
 964                 key = prepare_entry(o, n);
 965                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 966                         prev_line->next_line = key;
 967
 968                 hashmap_add(hm, key);
 969                 prev_line = key;
 970         }
 971 }
 972
 973 static void pmb_advance_or_null(struct diff_options *o,
 974                                 struct moved_entry *match,
 975                                 struct hashmap *hm,
 976                                 struct moved_block *pmb,
 977                                 int pmb_nr)
 978 {
 979         int i;
 980         for (i = 0; i < pmb_nr; i++) {
 981                 struct moved_entry *prev = pmb[i].match;
 982                 struct moved_entry *cur = (prev && prev->next_line) ?
 983                                 prev->next_line : NULL;
 984                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 985                         pmb[i].match = cur;
 986                 } else {
 987                         pmb[i].match = NULL;
 988                 }
 989         }
 990 }
 991
 992 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 993                                             struct moved_entry *match,
 994                                             struct hashmap *hm,
 995                                             struct moved_block *pmb,
 996                                             int pmb_nr, int n)
 997 {
 998         int i;
 999         char *got_match = xcalloc(1, pmb_nr);
1000
1001         for (; match; match = hashmap_get_next(hm, match)) {
1002                 for (i = 0; i < pmb_nr; i++) {
1003                         struct moved_entry *prev = pmb[i].match;
1004                         struct moved_entry *cur = (prev && prev->next_line) ?
1005                                         prev->next_line : NULL;
1006                         if (!cur)
1007                                 continue;
1008                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1009                                 got_match[i] |= 1;
1010                 }
1011         }
1012
1013         for (i = 0; i < pmb_nr; i++) {
1014                 if (got_match[i]) {
1015                         /* Advance to the next line */
1016                         pmb[i].match = pmb[i].match->next_line;
1017                 } else {
1018                         moved_block_clear(&pmb[i]);
1019                 }
1020         }
1021
1022         free(got_match);
1023 }
1024
1025 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1026                                          int pmb_nr)
1027 {
1028         int lp, rp;
1029
1030         /* Shrink the set of potential block to the remaining running */
1031         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1032                 while (lp < pmb_nr && pmb[lp].match)
1033                         lp++;
1034                 /* lp points at the first NULL now */
1035
1036                 while (rp > -1 && !pmb[rp].match)
1037                         rp--;
1038                 /* rp points at the last non-NULL */
1039
1040                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1041                         pmb[lp] = pmb[rp];
1042                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1043                         rp--;
1044                         lp++;
1045                 }
1046         }
1047
1048         /* Remember the number of running sets */
1049         return rp + 1;
1050 }
1051
1052 /*
1053  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1054  *
1055  * Otherwise, if the last block has fewer alphanumeric characters than
1056  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1057  * that block.
1058  *
1059  * The last block consists of the (n - block_length)'th line up to but not
1060  * including the nth line.
1061  *
1062  * Returns 0 if the last block is empty or is unset by this function, non zero
1063  * otherwise.
1064  *
1065  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1066  * Think of a way to unify them.
1067  */
1068 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1069 {
1070         int i, alnum_count = 0;
1071         if (o->color_moved == COLOR_MOVED_PLAIN)
1072                 return block_length;
1073         for (i = 1; i < block_length + 1; i++) {
1074                 const char *c = o->emitted_symbols->buf[n - i].line;
1075                 for (; *c; c++) {
1076                         if (!isalnum(*c))
1077                                 continue;
1078                         alnum_count++;
1079                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1080                                 return 1;
1081                 }
1082         }
1083         for (i = 1; i < block_length + 1; i++)
1084                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1085         return 0;
1086 }
1087
1088 /* Find blocks of moved code, delegate actual coloring decision to helper */
1089 static void mark_color_as_moved(struct diff_options *o,
1090                                 struct hashmap *add_lines,
1091                                 struct hashmap *del_lines)
1092 {
1093         struct moved_block *pmb = NULL; /* potentially moved blocks */
1094         int pmb_nr = 0, pmb_alloc = 0;
1095         int n, flipped_block = 0, block_length = 0;
1096
1097
1098         for (n = 0; n < o->emitted_symbols->nr; n++) {
1099                 struct hashmap *hm = NULL;
1100                 struct moved_entry *key;
1101                 struct moved_entry *match = NULL;
1102                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1103                 enum diff_symbol last_symbol = 0;
1104
1105                 switch (l->s) {
1106                 case DIFF_SYMBOL_PLUS:
1107                         hm = del_lines;
1108                         key = prepare_entry(o, n);
1109                         match = hashmap_get(hm, key, NULL);
1110                         free(key);
1111                         break;
1112                 case DIFF_SYMBOL_MINUS:
1113                         hm = add_lines;
1114                         key = prepare_entry(o, n);
1115                         match = hashmap_get(hm, key, NULL);
1116                         free(key);
1117                         break;
1118                 default:
1119                         flipped_block = 0;
1120                 }
1121
1122                 if (!match) {
1123                         int i;
1124
1125                         adjust_last_block(o, n, block_length);
1126                         for(i = 0; i < pmb_nr; i++)
1127                                 moved_block_clear(&pmb[i]);
1128                         pmb_nr = 0;
1129                         block_length = 0;
1130                         flipped_block = 0;
1131                         last_symbol = l->s;
1132                         continue;
1133                 }
1134
1135                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1136                         last_symbol = l->s;
1137                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1138                         continue;
1139                 }
1140
1141                 if (o->color_moved_ws_handling &
1142                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1143                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1144                 else
1145                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1146
1147                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1148
1149                 if (pmb_nr == 0) {
1150                         /*
1151                          * The current line is the start of a new block.
1152                          * Setup the set of potential blocks.
1153                          */
1154                         for (; match; match = hashmap_get_next(hm, match)) {
1155                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1156                                 if (o->color_moved_ws_handling &
1157                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1158                                         if (compute_ws_delta(l, match->es,
1159                                                              &pmb[pmb_nr].wsd))
1160                                                 pmb[pmb_nr++].match = match;
1161                                 } else {
1162                                         pmb[pmb_nr].wsd = 0;
1163                                         pmb[pmb_nr++].match = match;
1164                                 }
1165                         }
1166
1167                         if (adjust_last_block(o, n, block_length) &&
1168                             pmb_nr && last_symbol != l->s)
1169                                 flipped_block = (flipped_block + 1) % 2;
1170                         else
1171                                 flipped_block = 0;
1172
1173                         block_length = 0;
1174                 }
1175
1176                 if (pmb_nr) {
1177                         block_length++;
1178                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1179                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1180                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1181                 }
1182                 last_symbol = l->s;
1183         }
1184         adjust_last_block(o, n, block_length);
1185
1186         for(n = 0; n < pmb_nr; n++)
1187                 moved_block_clear(&pmb[n]);
1188         free(pmb);
1189 }
1190
1191 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1192   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1193 static void dim_moved_lines(struct diff_options *o)
1194 {
1195         int n;
1196         for (n = 0; n < o->emitted_symbols->nr; n++) {
1197                 struct emitted_diff_symbol *prev = (n != 0) ?
1198                                 &o->emitted_symbols->buf[n - 1] : NULL;
1199                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1200                 struct emitted_diff_symbol *next =
1201                                 (n < o->emitted_symbols->nr - 1) ?
1202                                 &o->emitted_symbols->buf[n + 1] : NULL;
1203
1204                 /* Not a plus or minus line? */
1205                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1206                         continue;
1207
1208                 /* Not a moved line? */
1209                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1210                         continue;
1211
1212                 /*
1213                  * If prev or next are not a plus or minus line,
1214                  * pretend they don't exist
1215                  */
1216                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1217                             prev->s != DIFF_SYMBOL_MINUS)
1218                         prev = NULL;
1219                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1220                             next->s != DIFF_SYMBOL_MINUS)
1221                         next = NULL;
1222
1223                 /* Inside a block? */
1224                 if ((prev &&
1225                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1226                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1227                     (next &&
1228                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1229                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1230                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1231                         continue;
1232                 }
1233
1234                 /* Check if we are at an interesting bound: */
1235                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1236                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1237                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1238                         continue;
1239                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1240                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1241                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1242                         continue;
1243
1244                 /*
1245                  * The boundary to prev and next are not interesting,
1246                  * so this line is not interesting as a whole
1247                  */
1248                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1249         }
1250 }
1251
1252 static void emit_line_ws_markup(struct diff_options *o,
1253                                 const char *set_sign, const char *set,
1254                                 const char *reset,
1255                                 int sign_index, const char *line, int len,
1256                                 unsigned ws_rule, int blank_at_eof)
1257 {
1258         const char *ws = NULL;
1259         int sign = o->output_indicators[sign_index];
1260
1261         if (o->ws_error_highlight & ws_rule) {
1262                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1263                 if (!*ws)
1264                         ws = NULL;
1265         }
1266
1267         if (!ws && !set_sign)
1268                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1269         else if (!ws) {
1270                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1271         } else if (blank_at_eof)
1272                 /* Blank line at EOF - paint '+' as well */
1273                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1274         else {
1275                 /* Emit just the prefix, then the rest. */
1276                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1277                             sign, "", 0);
1278                 ws_check_emit(line, len, ws_rule,
1279                               o->file, set, reset, ws);
1280         }
1281 }
1282
1283 static void emit_diff_symbol_from_struct(struct diff_options *o,
1284                                          struct emitted_diff_symbol *eds)
1285 {
1286         static const char *nneof = " No newline at end of file\n";
1287         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1288         struct strbuf sb = STRBUF_INIT;
1289
1290         enum diff_symbol s = eds->s;
1291         const char *line = eds->line;
1292         int len = eds->len;
1293         unsigned flags = eds->flags;
1294
1295         switch (s) {
1296         case DIFF_SYMBOL_NO_LF_EOF:
1297                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1298                 reset = diff_get_color_opt(o, DIFF_RESET);
1299                 putc('\n', o->file);
1300                 emit_line_0(o, context, NULL, 0, reset, '\\',
1301                             nneof, strlen(nneof));
1302                 break;
1303         case DIFF_SYMBOL_SUBMODULE_HEADER:
1304         case DIFF_SYMBOL_SUBMODULE_ERROR:
1305         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1306         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1307         case DIFF_SYMBOL_SUMMARY:
1308         case DIFF_SYMBOL_STATS_LINE:
1309         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1310         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1311                 emit_line(o, "", "", line, len);
1312                 break;
1313         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1314         case DIFF_SYMBOL_CONTEXT_MARKER:
1315                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1316                 reset = diff_get_color_opt(o, DIFF_RESET);
1317                 emit_line(o, context, reset, line, len);
1318                 break;
1319         case DIFF_SYMBOL_SEPARATOR:
1320                 fprintf(o->file, "%s%c",
1321                         diff_line_prefix(o),
1322                         o->line_termination);
1323                 break;
1324         case DIFF_SYMBOL_CONTEXT:
1325                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1326                 reset = diff_get_color_opt(o, DIFF_RESET);
1327                 set_sign = NULL;
1328                 if (o->flags.dual_color_diffed_diffs) {
1329                         char c = !len ? 0 : line[0];
1330
1331                         if (c == '+')
1332                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1333                         else if (c == '@')
1334                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1335                         else if (c == '-')
1336                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1337                 }
1338                 emit_line_ws_markup(o, set_sign, set, reset,
1339                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1340                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1341                 break;
1342         case DIFF_SYMBOL_PLUS:
1343                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1344                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1345                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1346                 case DIFF_SYMBOL_MOVED_LINE |
1347                      DIFF_SYMBOL_MOVED_LINE_ALT |
1348                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1349                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1350                         break;
1351                 case DIFF_SYMBOL_MOVED_LINE |
1352                      DIFF_SYMBOL_MOVED_LINE_ALT:
1353                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1354                         break;
1355                 case DIFF_SYMBOL_MOVED_LINE |
1356                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1357                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1358                         break;
1359                 case DIFF_SYMBOL_MOVED_LINE:
1360                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1361                         break;
1362                 default:
1363                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1364                 }
1365                 reset = diff_get_color_opt(o, DIFF_RESET);
1366                 if (!o->flags.dual_color_diffed_diffs)
1367                         set_sign = NULL;
1368                 else {
1369                         char c = !len ? 0 : line[0];
1370
1371                         set_sign = set;
1372                         if (c == '-')
1373                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1374                         else if (c == '@')
1375                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1376                         else if (c == '+')
1377                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1378                         else
1379                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1380                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1381                 }
1382                 emit_line_ws_markup(o, set_sign, set, reset,
1383                                     OUTPUT_INDICATOR_NEW, line, len,
1384                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1385                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1386                 break;
1387         case DIFF_SYMBOL_MINUS:
1388                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1389                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1390                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1391                 case DIFF_SYMBOL_MOVED_LINE |
1392                      DIFF_SYMBOL_MOVED_LINE_ALT |
1393                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1394                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1395                         break;
1396                 case DIFF_SYMBOL_MOVED_LINE |
1397                      DIFF_SYMBOL_MOVED_LINE_ALT:
1398                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1399                         break;
1400                 case DIFF_SYMBOL_MOVED_LINE |
1401                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1402                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1403                         break;
1404                 case DIFF_SYMBOL_MOVED_LINE:
1405                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1406                         break;
1407                 default:
1408                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1409                 }
1410                 reset = diff_get_color_opt(o, DIFF_RESET);
1411                 if (!o->flags.dual_color_diffed_diffs)
1412                         set_sign = NULL;
1413                 else {
1414                         char c = !len ? 0 : line[0];
1415
1416                         set_sign = set;
1417                         if (c == '+')
1418                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1419                         else if (c == '@')
1420                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1421                         else if (c == '-')
1422                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1423                         else
1424                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1425                 }
1426                 emit_line_ws_markup(o, set_sign, set, reset,
1427                                     OUTPUT_INDICATOR_OLD, line, len,
1428                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1429                 break;
1430         case DIFF_SYMBOL_WORDS_PORCELAIN:
1431                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1432                 reset = diff_get_color_opt(o, DIFF_RESET);
1433                 emit_line(o, context, reset, line, len);
1434                 fputs("~\n", o->file);
1435                 break;
1436         case DIFF_SYMBOL_WORDS:
1437                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1438                 reset = diff_get_color_opt(o, DIFF_RESET);
1439                 /*
1440                  * Skip the prefix character, if any.  With
1441                  * diff_suppress_blank_empty, there may be
1442                  * none.
1443                  */
1444                 if (line[0] != '\n') {
1445                         line++;
1446                         len--;
1447                 }
1448                 emit_line(o, context, reset, line, len);
1449                 break;
1450         case DIFF_SYMBOL_FILEPAIR_PLUS:
1451                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1452                 reset = diff_get_color_opt(o, DIFF_RESET);
1453                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1454                         line, reset,
1455                         strchr(line, ' ') ? "\t" : "");
1456                 break;
1457         case DIFF_SYMBOL_FILEPAIR_MINUS:
1458                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1459                 reset = diff_get_color_opt(o, DIFF_RESET);
1460                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1461                         line, reset,
1462                         strchr(line, ' ') ? "\t" : "");
1463                 break;
1464         case DIFF_SYMBOL_BINARY_FILES:
1465         case DIFF_SYMBOL_HEADER:
1466                 fprintf(o->file, "%s", line);
1467                 break;
1468         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1469                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1470                 break;
1471         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1472                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1473                 break;
1474         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1475                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1476                 break;
1477         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1478                 fputs(diff_line_prefix(o), o->file);
1479                 fputc('\n', o->file);
1480                 break;
1481         case DIFF_SYMBOL_REWRITE_DIFF:
1482                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1483                 reset = diff_get_color_opt(o, DIFF_RESET);
1484                 emit_line(o, fraginfo, reset, line, len);
1485                 break;
1486         case DIFF_SYMBOL_SUBMODULE_ADD:
1487                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1488                 reset = diff_get_color_opt(o, DIFF_RESET);
1489                 emit_line(o, set, reset, line, len);
1490                 break;
1491         case DIFF_SYMBOL_SUBMODULE_DEL:
1492                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1493                 reset = diff_get_color_opt(o, DIFF_RESET);
1494                 emit_line(o, set, reset, line, len);
1495                 break;
1496         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1497                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1498                         diff_line_prefix(o), line);
1499                 break;
1500         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1501                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1502                         diff_line_prefix(o), line);
1503                 break;
1504         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1505                 emit_line(o, "", "", " 0 files changed\n",
1506                           strlen(" 0 files changed\n"));
1507                 break;
1508         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1509                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1510                 break;
1511         case DIFF_SYMBOL_WORD_DIFF:
1512                 fprintf(o->file, "%.*s", len, line);
1513                 break;
1514         case DIFF_SYMBOL_STAT_SEP:
1515                 fputs(o->stat_sep, o->file);
1516                 break;
1517         default:
1518                 BUG("unknown diff symbol");
1519         }
1520         strbuf_release(&sb);
1521 }
1522
1523 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1524                              const char *line, int len, unsigned flags)
1525 {
1526         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1527
1528         if (o->emitted_symbols)
1529                 append_emitted_diff_symbol(o, &e);
1530         else
1531                 emit_diff_symbol_from_struct(o, &e);
1532 }
1533
1534 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1535 {
1536         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1537 }
1538
1539 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1540 {
1541         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1542 }
1543
1544 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1545 {
1546         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1547                          path, strlen(path), 0);
1548 }
1549
1550 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1551 {
1552         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1553                          path, strlen(path), 0);
1554 }
1555
1556 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1557 {
1558         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1559                          header, strlen(header), 0);
1560 }
1561
1562 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1563 {
1564         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1565 }
1566
1567 void diff_emit_submodule_pipethrough(struct diff_options *o,
1568                                      const char *line, int len)
1569 {
1570         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1571 }
1572
1573 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1574 {
1575         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1576               ecbdata->blank_at_eof_in_preimage &&
1577               ecbdata->blank_at_eof_in_postimage &&
1578               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1579               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1580                 return 0;
1581         return ws_blank_line(line, len, ecbdata->ws_rule);
1582 }
1583
1584 static void emit_add_line(const char *reset,
1585                           struct emit_callback *ecbdata,
1586                           const char *line, int len)
1587 {
1588         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1589         if (new_blank_line_at_eof(ecbdata, line, len))
1590                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1591
1592         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1593 }
1594
1595 static void emit_del_line(const char *reset,
1596                           struct emit_callback *ecbdata,
1597                           const char *line, int len)
1598 {
1599         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1600         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1601 }
1602
1603 static void emit_context_line(const char *reset,
1604                               struct emit_callback *ecbdata,
1605                               const char *line, int len)
1606 {
1607         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1608         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1609 }
1610
1611 static void emit_hunk_header(struct emit_callback *ecbdata,
1612                              const char *line, int len)
1613 {
1614         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1615         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1616         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1617         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1618         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1619         static const char atat[2] = { '@', '@' };
1620         const char *cp, *ep;
1621         struct strbuf msgbuf = STRBUF_INIT;
1622         int org_len = len;
1623         int i = 1;
1624
1625         /*
1626          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1627          * it always is at least 10 bytes long.
1628          */
1629         if (len < 10 ||
1630             memcmp(line, atat, 2) ||
1631             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1632                 emit_diff_symbol(ecbdata->opt,
1633                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1634                 return;
1635         }
1636         ep += 2; /* skip over @@ */
1637
1638         /* The hunk header in fraginfo color */
1639         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1640                 strbuf_addstr(&msgbuf, reverse);
1641         strbuf_addstr(&msgbuf, frag);
1642         strbuf_add(&msgbuf, line, ep - line);
1643         strbuf_addstr(&msgbuf, reset);
1644
1645         /*
1646          * trailing "\r\n"
1647          */
1648         for ( ; i < 3; i++)
1649                 if (line[len - i] == '\r' || line[len - i] == '\n')
1650                         len--;
1651
1652         /* blank before the func header */
1653         for (cp = ep; ep - line < len; ep++)
1654                 if (*ep != ' ' && *ep != '\t')
1655                         break;
1656         if (ep != cp) {
1657                 strbuf_addstr(&msgbuf, context);
1658                 strbuf_add(&msgbuf, cp, ep - cp);
1659                 strbuf_addstr(&msgbuf, reset);
1660         }
1661
1662         if (ep < line + len) {
1663                 strbuf_addstr(&msgbuf, func);
1664                 strbuf_add(&msgbuf, ep, line + len - ep);
1665                 strbuf_addstr(&msgbuf, reset);
1666         }
1667
1668         strbuf_add(&msgbuf, line + len, org_len - len);
1669         strbuf_complete_line(&msgbuf);
1670         emit_diff_symbol(ecbdata->opt,
1671                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1672         strbuf_release(&msgbuf);
1673 }
1674
1675 static struct diff_tempfile *claim_diff_tempfile(void) {
1676         int i;
1677         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1678                 if (!diff_temp[i].name)
1679                         return diff_temp + i;
1680         BUG("diff is failing to clean up its tempfiles");
1681 }
1682
1683 static void remove_tempfile(void)
1684 {
1685         int i;
1686         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1687                 if (is_tempfile_active(diff_temp[i].tempfile))
1688                         delete_tempfile(&diff_temp[i].tempfile);
1689                 diff_temp[i].name = NULL;
1690         }
1691 }
1692
1693 static void add_line_count(struct strbuf *out, int count)
1694 {
1695         switch (count) {
1696         case 0:
1697                 strbuf_addstr(out, "0,0");
1698                 break;
1699         case 1:
1700                 strbuf_addstr(out, "1");
1701                 break;
1702         default:
1703                 strbuf_addf(out, "1,%d", count);
1704                 break;
1705         }
1706 }
1707
1708 static void emit_rewrite_lines(struct emit_callback *ecb,
1709                                int prefix, const char *data, int size)
1710 {
1711         const char *endp = NULL;
1712         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1713
1714         while (0 < size) {
1715                 int len;
1716
1717                 endp = memchr(data, '\n', size);
1718                 len = endp ? (endp - data + 1) : size;
1719                 if (prefix != '+') {
1720                         ecb->lno_in_preimage++;
1721                         emit_del_line(reset, ecb, data, len);
1722                 } else {
1723                         ecb->lno_in_postimage++;
1724                         emit_add_line(reset, ecb, data, len);
1725                 }
1726                 size -= len;
1727                 data += len;
1728         }
1729         if (!endp)
1730                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1731 }
1732
1733 static void emit_rewrite_diff(const char *name_a,
1734                               const char *name_b,
1735                               struct diff_filespec *one,
1736                               struct diff_filespec *two,
1737                               struct userdiff_driver *textconv_one,
1738                               struct userdiff_driver *textconv_two,
1739                               struct diff_options *o)
1740 {
1741         int lc_a, lc_b;
1742         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1743         const char *a_prefix, *b_prefix;
1744         char *data_one, *data_two;
1745         size_t size_one, size_two;
1746         struct emit_callback ecbdata;
1747         struct strbuf out = STRBUF_INIT;
1748
1749         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1750                 a_prefix = o->b_prefix;
1751                 b_prefix = o->a_prefix;
1752         } else {
1753                 a_prefix = o->a_prefix;
1754                 b_prefix = o->b_prefix;
1755         }
1756
1757         name_a += (*name_a == '/');
1758         name_b += (*name_b == '/');
1759
1760         strbuf_reset(&a_name);
1761         strbuf_reset(&b_name);
1762         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1763         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1764
1765         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1766         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1767
1768         memset(&ecbdata, 0, sizeof(ecbdata));
1769         ecbdata.color_diff = want_color(o->use_color);
1770         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1771         ecbdata.opt = o;
1772         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1773                 mmfile_t mf1, mf2;
1774                 mf1.ptr = (char *)data_one;
1775                 mf2.ptr = (char *)data_two;
1776                 mf1.size = size_one;
1777                 mf2.size = size_two;
1778                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1779         }
1780         ecbdata.lno_in_preimage = 1;
1781         ecbdata.lno_in_postimage = 1;
1782
1783         lc_a = count_lines(data_one, size_one);
1784         lc_b = count_lines(data_two, size_two);
1785
1786         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1787                          a_name.buf, a_name.len, 0);
1788         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1789                          b_name.buf, b_name.len, 0);
1790
1791         strbuf_addstr(&out, "@@ -");
1792         if (!o->irreversible_delete)
1793                 add_line_count(&out, lc_a);
1794         else
1795                 strbuf_addstr(&out, "?,?");
1796         strbuf_addstr(&out, " +");
1797         add_line_count(&out, lc_b);
1798         strbuf_addstr(&out, " @@\n");
1799         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1800         strbuf_release(&out);
1801
1802         if (lc_a && !o->irreversible_delete)
1803                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1804         if (lc_b)
1805                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1806         if (textconv_one)
1807                 free((char *)data_one);
1808         if (textconv_two)
1809                 free((char *)data_two);
1810 }
1811
1812 struct diff_words_buffer {
1813         mmfile_t text;
1814         unsigned long alloc;
1815         struct diff_words_orig {
1816                 const char *begin, *end;
1817         } *orig;
1818         int orig_nr, orig_alloc;
1819 };
1820
1821 static void diff_words_append(char *line, unsigned long len,
1822                 struct diff_words_buffer *buffer)
1823 {
1824         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1825         line++;
1826         len--;
1827         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1828         buffer->text.size += len;
1829         buffer->text.ptr[buffer->text.size] = '\0';
1830 }
1831
1832 struct diff_words_style_elem {
1833         const char *prefix;
1834         const char *suffix;
1835         const char *color; /* NULL; filled in by the setup code if
1836                             * color is enabled */
1837 };
1838
1839 struct diff_words_style {
1840         enum diff_words_type type;
1841         struct diff_words_style_elem new_word, old_word, ctx;
1842         const char *newline;
1843 };
1844
1845 static struct diff_words_style diff_words_styles[] = {
1846         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1847         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1848         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1849 };
1850
1851 struct diff_words_data {
1852         struct diff_words_buffer minus, plus;
1853         const char *current_plus;
1854         int last_minus;
1855         struct diff_options *opt;
1856         regex_t *word_regex;
1857         enum diff_words_type type;
1858         struct diff_words_style *style;
1859 };
1860
1861 static int fn_out_diff_words_write_helper(struct diff_options *o,
1862                                           struct diff_words_style_elem *st_el,
1863                                           const char *newline,
1864                                           size_t count, const char *buf)
1865 {
1866         int print = 0;
1867         struct strbuf sb = STRBUF_INIT;
1868
1869         while (count) {
1870                 char *p = memchr(buf, '\n', count);
1871                 if (print)
1872                         strbuf_addstr(&sb, diff_line_prefix(o));
1873
1874                 if (p != buf) {
1875                         const char *reset = st_el->color && *st_el->color ?
1876                                             GIT_COLOR_RESET : NULL;
1877                         if (st_el->color && *st_el->color)
1878                                 strbuf_addstr(&sb, st_el->color);
1879                         strbuf_addstr(&sb, st_el->prefix);
1880                         strbuf_add(&sb, buf, p ? p - buf : count);
1881                         strbuf_addstr(&sb, st_el->suffix);
1882                         if (reset)
1883                                 strbuf_addstr(&sb, reset);
1884                 }
1885                 if (!p)
1886                         goto out;
1887
1888                 strbuf_addstr(&sb, newline);
1889                 count -= p + 1 - buf;
1890                 buf = p + 1;
1891                 print = 1;
1892                 if (count) {
1893                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1894                                          sb.buf, sb.len, 0);
1895                         strbuf_reset(&sb);
1896                 }
1897         }
1898
1899 out:
1900         if (sb.len)
1901                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1902                                  sb.buf, sb.len, 0);
1903         strbuf_release(&sb);
1904         return 0;
1905 }
1906
1907 /*
1908  * '--color-words' algorithm can be described as:
1909  *
1910  *   1. collect the minus/plus lines of a diff hunk, divided into
1911  *      minus-lines and plus-lines;
1912  *
1913  *   2. break both minus-lines and plus-lines into words and
1914  *      place them into two mmfile_t with one word for each line;
1915  *
1916  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1917  *
1918  * And for the common parts of the both file, we output the plus side text.
1919  * diff_words->current_plus is used to trace the current position of the plus file
1920  * which printed. diff_words->last_minus is used to trace the last minus word
1921  * printed.
1922  *
1923  * For '--graph' to work with '--color-words', we need to output the graph prefix
1924  * on each line of color words output. Generally, there are two conditions on
1925  * which we should output the prefix.
1926  *
1927  *   1. diff_words->last_minus == 0 &&
1928  *      diff_words->current_plus == diff_words->plus.text.ptr
1929  *
1930  *      that is: the plus text must start as a new line, and if there is no minus
1931  *      word printed, a graph prefix must be printed.
1932  *
1933  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1934  *      *(diff_words->current_plus - 1) == '\n'
1935  *
1936  *      that is: a graph prefix must be printed following a '\n'
1937  */
1938 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1939 {
1940         if ((diff_words->last_minus == 0 &&
1941                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1942                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1943                 *(diff_words->current_plus - 1) == '\n')) {
1944                 return 1;
1945         } else {
1946                 return 0;
1947         }
1948 }
1949
1950 static void fn_out_diff_words_aux(void *priv,
1951                                   long minus_first, long minus_len,
1952                                   long plus_first, long plus_len,
1953                                   const char *func, long funclen)
1954 {
1955         struct diff_words_data *diff_words = priv;
1956         struct diff_words_style *style = diff_words->style;
1957         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1958         struct diff_options *opt = diff_words->opt;
1959         const char *line_prefix;
1960
1961         assert(opt);
1962         line_prefix = diff_line_prefix(opt);
1963
1964         /* POSIX requires that first be decremented by one if len == 0... */
1965         if (minus_len) {
1966                 minus_begin = diff_words->minus.orig[minus_first].begin;
1967                 minus_end =
1968                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1969         } else
1970                 minus_begin = minus_end =
1971                         diff_words->minus.orig[minus_first].end;
1972
1973         if (plus_len) {
1974                 plus_begin = diff_words->plus.orig[plus_first].begin;
1975                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1976         } else
1977                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1978
1979         if (color_words_output_graph_prefix(diff_words)) {
1980                 fputs(line_prefix, diff_words->opt->file);
1981         }
1982         if (diff_words->current_plus != plus_begin) {
1983                 fn_out_diff_words_write_helper(diff_words->opt,
1984                                 &style->ctx, style->newline,
1985                                 plus_begin - diff_words->current_plus,
1986                                 diff_words->current_plus);
1987         }
1988         if (minus_begin != minus_end) {
1989                 fn_out_diff_words_write_helper(diff_words->opt,
1990                                 &style->old_word, style->newline,
1991                                 minus_end - minus_begin, minus_begin);
1992         }
1993         if (plus_begin != plus_end) {
1994                 fn_out_diff_words_write_helper(diff_words->opt,
1995                                 &style->new_word, style->newline,
1996                                 plus_end - plus_begin, plus_begin);
1997         }
1998
1999         diff_words->current_plus = plus_end;
2000         diff_words->last_minus = minus_first;
2001 }
2002
2003 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2004 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2005                 int *begin, int *end)
2006 {
2007         if (word_regex && *begin < buffer->size) {
2008                 regmatch_t match[1];
2009                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2010                                  buffer->size - *begin, 1, match, 0)) {
2011                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2012                                         '\n', match[0].rm_eo - match[0].rm_so);
2013                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2014                         *begin += match[0].rm_so;
2015                         return *begin >= *end;
2016                 }
2017                 return -1;
2018         }
2019
2020         /* find the next word */
2021         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2022                 (*begin)++;
2023         if (*begin >= buffer->size)
2024                 return -1;
2025
2026         /* find the end of the word */
2027         *end = *begin + 1;
2028         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2029                 (*end)++;
2030
2031         return 0;
2032 }
2033
2034 /*
2035  * This function splits the words in buffer->text, stores the list with
2036  * newline separator into out, and saves the offsets of the original words
2037  * in buffer->orig.
2038  */
2039 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2040                 regex_t *word_regex)
2041 {
2042         int i, j;
2043         long alloc = 0;
2044
2045         out->size = 0;
2046         out->ptr = NULL;
2047
2048         /* fake an empty "0th" word */
2049         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2050         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2051         buffer->orig_nr = 1;
2052
2053         for (i = 0; i < buffer->text.size; i++) {
2054                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2055                         return;
2056
2057                 /* store original boundaries */
2058                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2059                                 buffer->orig_alloc);
2060                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2061                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2062                 buffer->orig_nr++;
2063
2064                 /* store one word */
2065                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2066                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2067                 out->ptr[out->size + j - i] = '\n';
2068                 out->size += j - i + 1;
2069
2070                 i = j - 1;
2071         }
2072 }
2073
2074 /* this executes the word diff on the accumulated buffers */
2075 static void diff_words_show(struct diff_words_data *diff_words)
2076 {
2077         xpparam_t xpp;
2078         xdemitconf_t xecfg;
2079         mmfile_t minus, plus;
2080         struct diff_words_style *style = diff_words->style;
2081
2082         struct diff_options *opt = diff_words->opt;
2083         const char *line_prefix;
2084
2085         assert(opt);
2086         line_prefix = diff_line_prefix(opt);
2087
2088         /* special case: only removal */
2089         if (!diff_words->plus.text.size) {
2090                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2091                                  line_prefix, strlen(line_prefix), 0);
2092                 fn_out_diff_words_write_helper(diff_words->opt,
2093                         &style->old_word, style->newline,
2094                         diff_words->minus.text.size,
2095                         diff_words->minus.text.ptr);
2096                 diff_words->minus.text.size = 0;
2097                 return;
2098         }
2099
2100         diff_words->current_plus = diff_words->plus.text.ptr;
2101         diff_words->last_minus = 0;
2102
2103         memset(&xpp, 0, sizeof(xpp));
2104         memset(&xecfg, 0, sizeof(xecfg));
2105         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2106         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2107         xpp.flags = 0;
2108         /* as only the hunk header will be parsed, we need a 0-context */
2109         xecfg.ctxlen = 0;
2110         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2111                           diff_words, &xpp, &xecfg))
2112                 die("unable to generate word diff");
2113         free(minus.ptr);
2114         free(plus.ptr);
2115         if (diff_words->current_plus != diff_words->plus.text.ptr +
2116                         diff_words->plus.text.size) {
2117                 if (color_words_output_graph_prefix(diff_words))
2118                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2119                                          line_prefix, strlen(line_prefix), 0);
2120                 fn_out_diff_words_write_helper(diff_words->opt,
2121                         &style->ctx, style->newline,
2122                         diff_words->plus.text.ptr + diff_words->plus.text.size
2123                         - diff_words->current_plus, diff_words->current_plus);
2124         }
2125         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2126 }
2127
2128 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2129 static void diff_words_flush(struct emit_callback *ecbdata)
2130 {
2131         struct diff_options *wo = ecbdata->diff_words->opt;
2132
2133         if (ecbdata->diff_words->minus.text.size ||
2134             ecbdata->diff_words->plus.text.size)
2135                 diff_words_show(ecbdata->diff_words);
2136
2137         if (wo->emitted_symbols) {
2138                 struct diff_options *o = ecbdata->opt;
2139                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2140                 int i;
2141
2142                 /*
2143                  * NEEDSWORK:
2144                  * Instead of appending each, concat all words to a line?
2145                  */
2146                 for (i = 0; i < wol->nr; i++)
2147                         append_emitted_diff_symbol(o, &wol->buf[i]);
2148
2149                 for (i = 0; i < wol->nr; i++)
2150                         free((void *)wol->buf[i].line);
2151
2152                 wol->nr = 0;
2153         }
2154 }
2155
2156 static void diff_filespec_load_driver(struct diff_filespec *one,
2157                                       struct index_state *istate)
2158 {
2159         /* Use already-loaded driver */
2160         if (one->driver)
2161                 return;
2162
2163         if (S_ISREG(one->mode))
2164                 one->driver = userdiff_find_by_path(istate, one->path);
2165
2166         /* Fallback to default settings */
2167         if (!one->driver)
2168                 one->driver = userdiff_find_by_name("default");
2169 }
2170
2171 static const char *userdiff_word_regex(struct diff_filespec *one,
2172                                        struct index_state *istate)
2173 {
2174         diff_filespec_load_driver(one, istate);
2175         return one->driver->word_regex;
2176 }
2177
2178 static void init_diff_words_data(struct emit_callback *ecbdata,
2179                                  struct diff_options *orig_opts,
2180                                  struct diff_filespec *one,
2181                                  struct diff_filespec *two)
2182 {
2183         int i;
2184         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2185         memcpy(o, orig_opts, sizeof(struct diff_options));
2186
2187         ecbdata->diff_words =
2188                 xcalloc(1, sizeof(struct diff_words_data));
2189         ecbdata->diff_words->type = o->word_diff;
2190         ecbdata->diff_words->opt = o;
2191
2192         if (orig_opts->emitted_symbols)
2193                 o->emitted_symbols =
2194                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2195
2196         if (!o->word_regex)
2197                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2198         if (!o->word_regex)
2199                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2200         if (!o->word_regex)
2201                 o->word_regex = diff_word_regex_cfg;
2202         if (o->word_regex) {
2203                 ecbdata->diff_words->word_regex = (regex_t *)
2204                         xmalloc(sizeof(regex_t));
2205                 if (regcomp(ecbdata->diff_words->word_regex,
2206                             o->word_regex,
2207                             REG_EXTENDED | REG_NEWLINE))
2208                         die("invalid regular expression: %s",
2209                             o->word_regex);
2210         }
2211         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2212                 if (o->word_diff == diff_words_styles[i].type) {
2213                         ecbdata->diff_words->style =
2214                                 &diff_words_styles[i];
2215                         break;
2216                 }
2217         }
2218         if (want_color(o->use_color)) {
2219                 struct diff_words_style *st = ecbdata->diff_words->style;
2220                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2221                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2222                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2223         }
2224 }
2225
2226 static void free_diff_words_data(struct emit_callback *ecbdata)
2227 {
2228         if (ecbdata->diff_words) {
2229                 diff_words_flush(ecbdata);
2230                 free (ecbdata->diff_words->opt->emitted_symbols);
2231                 free (ecbdata->diff_words->opt);
2232                 free (ecbdata->diff_words->minus.text.ptr);
2233                 free (ecbdata->diff_words->minus.orig);
2234                 free (ecbdata->diff_words->plus.text.ptr);
2235                 free (ecbdata->diff_words->plus.orig);
2236                 if (ecbdata->diff_words->word_regex) {
2237                         regfree(ecbdata->diff_words->word_regex);
2238                         free(ecbdata->diff_words->word_regex);
2239                 }
2240                 FREE_AND_NULL(ecbdata->diff_words);
2241         }
2242 }
2243
2244 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2245 {
2246         if (want_color(diff_use_color))
2247                 return diff_colors[ix];
2248         return "";
2249 }
2250
2251 const char *diff_line_prefix(struct diff_options *opt)
2252 {
2253         struct strbuf *msgbuf;
2254         if (!opt->output_prefix)
2255                 return "";
2256
2257         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2258         return msgbuf->buf;
2259 }
2260
2261 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2262 {
2263         const char *cp;
2264         unsigned long allot;
2265         size_t l = len;
2266
2267         cp = line;
2268         allot = l;
2269         while (0 < l) {
2270                 (void) utf8_width(&cp, &l);
2271                 if (!cp)
2272                         break; /* truncated in the middle? */
2273         }
2274         return allot - l;
2275 }
2276
2277 static void find_lno(const char *line, struct emit_callback *ecbdata)
2278 {
2279         const char *p;
2280         ecbdata->lno_in_preimage = 0;
2281         ecbdata->lno_in_postimage = 0;
2282         p = strchr(line, '-');
2283         if (!p)
2284                 return; /* cannot happen */
2285         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2286         p = strchr(p, '+');
2287         if (!p)
2288                 return; /* cannot happen */
2289         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2290 }
2291
2292 static void fn_out_consume(void *priv, char *line, unsigned long len)
2293 {
2294         struct emit_callback *ecbdata = priv;
2295         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2296         struct diff_options *o = ecbdata->opt;
2297
2298         o->found_changes = 1;
2299
2300         if (ecbdata->header) {
2301                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2302                                  ecbdata->header->buf, ecbdata->header->len, 0);
2303                 strbuf_reset(ecbdata->header);
2304                 ecbdata->header = NULL;
2305         }
2306
2307         if (ecbdata->label_path[0]) {
2308                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2309                                  ecbdata->label_path[0],
2310                                  strlen(ecbdata->label_path[0]), 0);
2311                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2312                                  ecbdata->label_path[1],
2313                                  strlen(ecbdata->label_path[1]), 0);
2314                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2315         }
2316
2317         if (diff_suppress_blank_empty
2318             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2319                 line[0] = '\n';
2320                 len = 1;
2321         }
2322
2323         if (line[0] == '@') {
2324                 if (ecbdata->diff_words)
2325                         diff_words_flush(ecbdata);
2326                 len = sane_truncate_line(ecbdata, line, len);
2327                 find_lno(line, ecbdata);
2328                 emit_hunk_header(ecbdata, line, len);
2329                 return;
2330         }
2331
2332         if (ecbdata->diff_words) {
2333                 enum diff_symbol s =
2334                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2335                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2336                 if (line[0] == '-') {
2337                         diff_words_append(line, len,
2338                                           &ecbdata->diff_words->minus);
2339                         return;
2340                 } else if (line[0] == '+') {
2341                         diff_words_append(line, len,
2342                                           &ecbdata->diff_words->plus);
2343                         return;
2344                 } else if (starts_with(line, "\\ ")) {
2345                         /*
2346                          * Eat the "no newline at eof" marker as if we
2347                          * saw a "+" or "-" line with nothing on it,
2348                          * and return without diff_words_flush() to
2349                          * defer processing. If this is the end of
2350                          * preimage, more "+" lines may come after it.
2351                          */
2352                         return;
2353                 }
2354                 diff_words_flush(ecbdata);
2355                 emit_diff_symbol(o, s, line, len, 0);
2356                 return;
2357         }
2358
2359         switch (line[0]) {
2360         case '+':
2361                 ecbdata->lno_in_postimage++;
2362                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2363                 break;
2364         case '-':
2365                 ecbdata->lno_in_preimage++;
2366                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2367                 break;
2368         case ' ':
2369                 ecbdata->lno_in_postimage++;
2370                 ecbdata->lno_in_preimage++;
2371                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2372                 break;
2373         default:
2374                 /* incomplete line at the end */
2375                 ecbdata->lno_in_preimage++;
2376                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2377                                  line, len, 0);
2378                 break;
2379         }
2380 }
2381
2382 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2383 {
2384         const char *old_name = a;
2385         const char *new_name = b;
2386         int pfx_length, sfx_length;
2387         int pfx_adjust_for_slash;
2388         int len_a = strlen(a);
2389         int len_b = strlen(b);
2390         int a_midlen, b_midlen;
2391         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2392         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2393
2394         if (qlen_a || qlen_b) {
2395                 quote_c_style(a, name, NULL, 0);
2396                 strbuf_addstr(name, " => ");
2397                 quote_c_style(b, name, NULL, 0);
2398                 return;
2399         }
2400
2401         /* Find common prefix */
2402         pfx_length = 0;
2403         while (*old_name && *new_name && *old_name == *new_name) {
2404                 if (*old_name == '/')
2405                         pfx_length = old_name - a + 1;
2406                 old_name++;
2407                 new_name++;
2408         }
2409
2410         /* Find common suffix */
2411         old_name = a + len_a;
2412         new_name = b + len_b;
2413         sfx_length = 0;
2414         /*
2415          * If there is a common prefix, it must end in a slash.  In
2416          * that case we let this loop run 1 into the prefix to see the
2417          * same slash.
2418          *
2419          * If there is no common prefix, we cannot do this as it would
2420          * underrun the input strings.
2421          */
2422         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2423         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2424                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2425                *old_name == *new_name) {
2426                 if (*old_name == '/')
2427                         sfx_length = len_a - (old_name - a);
2428                 old_name--;
2429                 new_name--;
2430         }
2431
2432         /*
2433          * pfx{mid-a => mid-b}sfx
2434          * {pfx-a => pfx-b}sfx
2435          * pfx{sfx-a => sfx-b}
2436          * name-a => name-b
2437          */
2438         a_midlen = len_a - pfx_length - sfx_length;
2439         b_midlen = len_b - pfx_length - sfx_length;
2440         if (a_midlen < 0)
2441                 a_midlen = 0;
2442         if (b_midlen < 0)
2443                 b_midlen = 0;
2444
2445         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2446         if (pfx_length + sfx_length) {
2447                 strbuf_add(name, a, pfx_length);
2448                 strbuf_addch(name, '{');
2449         }
2450         strbuf_add(name, a + pfx_length, a_midlen);
2451         strbuf_addstr(name, " => ");
2452         strbuf_add(name, b + pfx_length, b_midlen);
2453         if (pfx_length + sfx_length) {
2454                 strbuf_addch(name, '}');
2455                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2456         }
2457 }
2458
2459 struct diffstat_t {
2460         int nr;
2461         int alloc;
2462         struct diffstat_file {
2463                 char *from_name;
2464                 char *name;
2465                 char *print_name;
2466                 const char *comments;
2467                 unsigned is_unmerged:1;
2468                 unsigned is_binary:1;
2469                 unsigned is_renamed:1;
2470                 unsigned is_interesting:1;
2471                 uintmax_t added, deleted;
2472         } **files;
2473 };
2474
2475 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2476                                           const char *name_a,
2477                                           const char *name_b)
2478 {
2479         struct diffstat_file *x;
2480         x = xcalloc(1, sizeof(*x));
2481         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2482         diffstat->files[diffstat->nr++] = x;
2483         if (name_b) {
2484                 x->from_name = xstrdup(name_a);
2485                 x->name = xstrdup(name_b);
2486                 x->is_renamed = 1;
2487         }
2488         else {
2489                 x->from_name = NULL;
2490                 x->name = xstrdup(name_a);
2491         }
2492         return x;
2493 }
2494
2495 static void diffstat_consume(void *priv, char *line, unsigned long len)
2496 {
2497         struct diffstat_t *diffstat = priv;
2498         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2499
2500         if (line[0] == '+')
2501                 x->added++;
2502         else if (line[0] == '-')
2503                 x->deleted++;
2504 }
2505
2506 const char mime_boundary_leader[] = "------------";
2507
2508 static int scale_linear(int it, int width, int max_change)
2509 {
2510         if (!it)
2511                 return 0;
2512         /*
2513          * make sure that at least one '-' or '+' is printed if
2514          * there is any change to this path. The easiest way is to
2515          * scale linearly as if the alloted width is one column shorter
2516          * than it is, and then add 1 to the result.
2517          */
2518         return 1 + (it * (width - 1) / max_change);
2519 }
2520
2521 static void show_graph(struct strbuf *out, char ch, int cnt,
2522                        const char *set, const char *reset)
2523 {
2524         if (cnt <= 0)
2525                 return;
2526         strbuf_addstr(out, set);
2527         strbuf_addchars(out, ch, cnt);
2528         strbuf_addstr(out, reset);
2529 }
2530
2531 static void fill_print_name(struct diffstat_file *file)
2532 {
2533         struct strbuf pname = STRBUF_INIT;
2534
2535         if (file->print_name)
2536                 return;
2537
2538         if (file->is_renamed)
2539                 pprint_rename(&pname, file->from_name, file->name);
2540         else
2541                 quote_c_style(file->name, &pname, NULL, 0);
2542
2543         if (file->comments)
2544                 strbuf_addf(&pname, " (%s)", file->comments);
2545
2546         file->print_name = strbuf_detach(&pname, NULL);
2547 }
2548
2549 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2550                 int files, int insertions, int deletions)
2551 {
2552         struct strbuf sb = STRBUF_INIT;
2553
2554         if (!files) {
2555                 assert(insertions == 0 && deletions == 0);
2556                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2557                                  NULL, 0, 0);
2558                 return;
2559         }
2560
2561         strbuf_addf(&sb,
2562                     (files == 1) ? " %d file changed" : " %d files changed",
2563                     files);
2564
2565         /*
2566          * For binary diff, the caller may want to print "x files
2567          * changed" with insertions == 0 && deletions == 0.
2568          *
2569          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2570          * is probably less confusing (i.e skip over "2 files changed
2571          * but nothing about added/removed lines? Is this a bug in Git?").
2572          */
2573         if (insertions || deletions == 0) {
2574                 strbuf_addf(&sb,
2575                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2576                             insertions);
2577         }
2578
2579         if (deletions || insertions == 0) {
2580                 strbuf_addf(&sb,
2581                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2582                             deletions);
2583         }
2584         strbuf_addch(&sb, '\n');
2585         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2586                          sb.buf, sb.len, 0);
2587         strbuf_release(&sb);
2588 }
2589
2590 void print_stat_summary(FILE *fp, int files,
2591                         int insertions, int deletions)
2592 {
2593         struct diff_options o;
2594         memset(&o, 0, sizeof(o));
2595         o.file = fp;
2596
2597         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2598 }
2599
2600 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2601 {
2602         int i, len, add, del, adds = 0, dels = 0;
2603         uintmax_t max_change = 0, max_len = 0;
2604         int total_files = data->nr, count;
2605         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2606         const char *reset, *add_c, *del_c;
2607         int extra_shown = 0;
2608         const char *line_prefix = diff_line_prefix(options);
2609         struct strbuf out = STRBUF_INIT;
2610
2611         if (data->nr == 0)
2612                 return;
2613
2614         count = options->stat_count ? options->stat_count : data->nr;
2615
2616         reset = diff_get_color_opt(options, DIFF_RESET);
2617         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2618         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2619
2620         /*
2621          * Find the longest filename and max number of changes
2622          */
2623         for (i = 0; (i < count) && (i < data->nr); i++) {
2624                 struct diffstat_file *file = data->files[i];
2625                 uintmax_t change = file->added + file->deleted;
2626
2627                 if (!file->is_interesting && (change == 0)) {
2628                         count++; /* not shown == room for one more */
2629                         continue;
2630                 }
2631                 fill_print_name(file);
2632                 len = strlen(file->print_name);
2633                 if (max_len < len)
2634                         max_len = len;
2635
2636                 if (file->is_unmerged) {
2637                         /* "Unmerged" is 8 characters */
2638                         bin_width = bin_width < 8 ? 8 : bin_width;
2639                         continue;
2640                 }
2641                 if (file->is_binary) {
2642                         /* "Bin XXX -> YYY bytes" */
2643                         int w = 14 + decimal_width(file->added)
2644                                 + decimal_width(file->deleted);
2645                         bin_width = bin_width < w ? w : bin_width;
2646                         /* Display change counts aligned with "Bin" */
2647                         number_width = 3;
2648                         continue;
2649                 }
2650
2651                 if (max_change < change)
2652                         max_change = change;
2653         }
2654         count = i; /* where we can stop scanning in data->files[] */
2655
2656         /*
2657          * We have width = stat_width or term_columns() columns total.
2658          * We want a maximum of min(max_len, stat_name_width) for the name part.
2659          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2660          * We also need 1 for " " and 4 + decimal_width(max_change)
2661          * for " | NNNN " and one the empty column at the end, altogether
2662          * 6 + decimal_width(max_change).
2663          *
2664          * If there's not enough space, we will use the smaller of
2665          * stat_name_width (if set) and 5/8*width for the filename,
2666          * and the rest for constant elements + graph part, but no more
2667          * than stat_graph_width for the graph part.
2668          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2669          * for the standard terminal size).
2670          *
2671          * In other words: stat_width limits the maximum width, and
2672          * stat_name_width fixes the maximum width of the filename,
2673          * and is also used to divide available columns if there
2674          * aren't enough.
2675          *
2676          * Binary files are displayed with "Bin XXX -> YYY bytes"
2677          * instead of the change count and graph. This part is treated
2678          * similarly to the graph part, except that it is not
2679          * "scaled". If total width is too small to accommodate the
2680          * guaranteed minimum width of the filename part and the
2681          * separators and this message, this message will "overflow"
2682          * making the line longer than the maximum width.
2683          */
2684
2685         if (options->stat_width == -1)
2686                 width = term_columns() - strlen(line_prefix);
2687         else
2688                 width = options->stat_width ? options->stat_width : 80;
2689         number_width = decimal_width(max_change) > number_width ?
2690                 decimal_width(max_change) : number_width;
2691
2692         if (options->stat_graph_width == -1)
2693                 options->stat_graph_width = diff_stat_graph_width;
2694
2695         /*
2696          * Guarantee 3/8*16==6 for the graph part
2697          * and 5/8*16==10 for the filename part
2698          */
2699         if (width < 16 + 6 + number_width)
2700                 width = 16 + 6 + number_width;
2701
2702         /*
2703          * First assign sizes that are wanted, ignoring available width.
2704          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2705          * starting from "XXX" should fit in graph_width.
2706          */
2707         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2708         if (options->stat_graph_width &&
2709             options->stat_graph_width < graph_width)
2710                 graph_width = options->stat_graph_width;
2711
2712         name_width = (options->stat_name_width > 0 &&
2713                       options->stat_name_width < max_len) ?
2714                 options->stat_name_width : max_len;
2715
2716         /*
2717          * Adjust adjustable widths not to exceed maximum width
2718          */
2719         if (name_width + number_width + 6 + graph_width > width) {
2720                 if (graph_width > width * 3/8 - number_width - 6) {
2721                         graph_width = width * 3/8 - number_width - 6;
2722                         if (graph_width < 6)
2723                                 graph_width = 6;
2724                 }
2725
2726                 if (options->stat_graph_width &&
2727                     graph_width > options->stat_graph_width)
2728                         graph_width = options->stat_graph_width;
2729                 if (name_width > width - number_width - 6 - graph_width)
2730                         name_width = width - number_width - 6 - graph_width;
2731                 else
2732                         graph_width = width - number_width - 6 - name_width;
2733         }
2734
2735         /*
2736          * From here name_width is the width of the name area,
2737          * and graph_width is the width of the graph area.
2738          * max_change is used to scale graph properly.
2739          */
2740         for (i = 0; i < count; i++) {
2741                 const char *prefix = "";
2742                 struct diffstat_file *file = data->files[i];
2743                 char *name = file->print_name;
2744                 uintmax_t added = file->added;
2745                 uintmax_t deleted = file->deleted;
2746                 int name_len;
2747
2748                 if (!file->is_interesting && (added + deleted == 0))
2749                         continue;
2750
2751                 /*
2752                  * "scale" the filename
2753                  */
2754                 len = name_width;
2755                 name_len = strlen(name);
2756                 if (name_width < name_len) {
2757                         char *slash;
2758                         prefix = "...";
2759                         len -= 3;
2760                         name += name_len - len;
2761                         slash = strchr(name, '/');
2762                         if (slash)
2763                                 name = slash;
2764                 }
2765
2766                 if (file->is_binary) {
2767                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2768                         strbuf_addf(&out, " %*s", number_width, "Bin");
2769                         if (!added && !deleted) {
2770                                 strbuf_addch(&out, '\n');
2771                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2772                                                  out.buf, out.len, 0);
2773                                 strbuf_reset(&out);
2774                                 continue;
2775                         }
2776                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2777                                 del_c, deleted, reset);
2778                         strbuf_addstr(&out, " -> ");
2779                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2780                                 add_c, added, reset);
2781                         strbuf_addstr(&out, " bytes\n");
2782                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2783                                          out.buf, out.len, 0);
2784                         strbuf_reset(&out);
2785                         continue;
2786                 }
2787                 else if (file->is_unmerged) {
2788                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2789                         strbuf_addstr(&out, " Unmerged\n");
2790                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2791                                          out.buf, out.len, 0);
2792                         strbuf_reset(&out);
2793                         continue;
2794                 }
2795
2796                 /*
2797                  * scale the add/delete
2798                  */
2799                 add = added;
2800                 del = deleted;
2801
2802                 if (graph_width <= max_change) {
2803                         int total = scale_linear(add + del, graph_width, max_change);
2804                         if (total < 2 && add && del)
2805                                 /* width >= 2 due to the sanity check */
2806                                 total = 2;
2807                         if (add < del) {
2808                                 add = scale_linear(add, graph_width, max_change);
2809                                 del = total - add;
2810                         } else {
2811                                 del = scale_linear(del, graph_width, max_change);
2812                                 add = total - del;
2813                         }
2814                 }
2815                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2816                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2817                         number_width, added + deleted,
2818                         added + deleted ? " " : "");
2819                 show_graph(&out, '+', add, add_c, reset);
2820                 show_graph(&out, '-', del, del_c, reset);
2821                 strbuf_addch(&out, '\n');
2822                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2823                                  out.buf, out.len, 0);
2824                 strbuf_reset(&out);
2825         }
2826
2827         for (i = 0; i < data->nr; i++) {
2828                 struct diffstat_file *file = data->files[i];
2829                 uintmax_t added = file->added;
2830                 uintmax_t deleted = file->deleted;
2831
2832                 if (file->is_unmerged ||
2833                     (!file->is_interesting && (added + deleted == 0))) {
2834                         total_files--;
2835                         continue;
2836                 }
2837
2838                 if (!file->is_binary) {
2839                         adds += added;
2840                         dels += deleted;
2841                 }
2842                 if (i < count)
2843                         continue;
2844                 if (!extra_shown)
2845                         emit_diff_symbol(options,
2846                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2847                                          NULL, 0, 0);
2848                 extra_shown = 1;
2849         }
2850
2851         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2852         strbuf_release(&out);
2853 }
2854
2855 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2856 {
2857         int i, adds = 0, dels = 0, total_files = data->nr;
2858
2859         if (data->nr == 0)
2860                 return;
2861
2862         for (i = 0; i < data->nr; i++) {
2863                 int added = data->files[i]->added;
2864                 int deleted = data->files[i]->deleted;
2865
2866                 if (data->files[i]->is_unmerged ||
2867                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2868                         total_files--;
2869                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2870                         adds += added;
2871                         dels += deleted;
2872                 }
2873         }
2874         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2875 }
2876
2877 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2878 {
2879         int i;
2880
2881         if (data->nr == 0)
2882                 return;
2883
2884         for (i = 0; i < data->nr; i++) {
2885                 struct diffstat_file *file = data->files[i];
2886
2887                 fprintf(options->file, "%s", diff_line_prefix(options));
2888
2889                 if (file->is_binary)
2890                         fprintf(options->file, "-\t-\t");
2891                 else
2892                         fprintf(options->file,
2893                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2894                                 file->added, file->deleted);
2895                 if (options->line_termination) {
2896                         fill_print_name(file);
2897                         if (!file->is_renamed)
2898                                 write_name_quoted(file->name, options->file,
2899                                                   options->line_termination);
2900                         else {
2901                                 fputs(file->print_name, options->file);
2902                                 putc(options->line_termination, options->file);
2903                         }
2904                 } else {
2905                         if (file->is_renamed) {
2906                                 putc('\0', options->file);
2907                                 write_name_quoted(file->from_name, options->file, '\0');
2908                         }
2909                         write_name_quoted(file->name, options->file, '\0');
2910                 }
2911         }
2912 }
2913
2914 struct dirstat_file {
2915         const char *name;
2916         unsigned long changed;
2917 };
2918
2919 struct dirstat_dir {
2920         struct dirstat_file *files;
2921         int alloc, nr, permille, cumulative;
2922 };
2923
2924 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2925                 unsigned long changed, const char *base, int baselen)
2926 {
2927         unsigned long sum_changes = 0;
2928         unsigned int sources = 0;
2929         const char *line_prefix = diff_line_prefix(opt);
2930
2931         while (dir->nr) {
2932                 struct dirstat_file *f = dir->files;
2933                 int namelen = strlen(f->name);
2934                 unsigned long changes;
2935                 char *slash;
2936
2937                 if (namelen < baselen)
2938                         break;
2939                 if (memcmp(f->name, base, baselen))
2940                         break;
2941                 slash = strchr(f->name + baselen, '/');
2942                 if (slash) {
2943                         int newbaselen = slash + 1 - f->name;
2944                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2945                         sources++;
2946                 } else {
2947                         changes = f->changed;
2948                         dir->files++;
2949                         dir->nr--;
2950                         sources += 2;
2951                 }
2952                 sum_changes += changes;
2953         }
2954
2955         /*
2956          * We don't report dirstat's for
2957          *  - the top level
2958          *  - or cases where everything came from a single directory
2959          *    under this directory (sources == 1).
2960          */
2961         if (baselen && sources != 1) {
2962                 if (sum_changes) {
2963                         int permille = sum_changes * 1000 / changed;
2964                         if (permille >= dir->permille) {
2965                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2966                                         permille / 10, permille % 10, baselen, base);
2967                                 if (!dir->cumulative)
2968                                         return 0;
2969                         }
2970                 }
2971         }
2972         return sum_changes;
2973 }
2974
2975 static int dirstat_compare(const void *_a, const void *_b)
2976 {
2977         const struct dirstat_file *a = _a;
2978         const struct dirstat_file *b = _b;
2979         return strcmp(a->name, b->name);
2980 }
2981
2982 static void show_dirstat(struct diff_options *options)
2983 {
2984         int i;
2985         unsigned long changed;
2986         struct dirstat_dir dir;
2987         struct diff_queue_struct *q = &diff_queued_diff;
2988
2989         dir.files = NULL;
2990         dir.alloc = 0;
2991         dir.nr = 0;
2992         dir.permille = options->dirstat_permille;
2993         dir.cumulative = options->flags.dirstat_cumulative;
2994
2995         changed = 0;
2996         for (i = 0; i < q->nr; i++) {
2997                 struct diff_filepair *p = q->queue[i];
2998                 const char *name;
2999                 unsigned long copied, added, damage;
3000
3001                 name = p->two->path ? p->two->path : p->one->path;
3002
3003                 if (p->one->oid_valid && p->two->oid_valid &&
3004                     oideq(&p->one->oid, &p->two->oid)) {
3005                         /*
3006                          * The SHA1 has not changed, so pre-/post-content is
3007                          * identical. We can therefore skip looking at the
3008                          * file contents altogether.
3009                          */
3010                         damage = 0;
3011                         goto found_damage;
3012                 }
3013
3014                 if (options->flags.dirstat_by_file) {
3015                         /*
3016                          * In --dirstat-by-file mode, we don't really need to
3017                          * look at the actual file contents at all.
3018                          * The fact that the SHA1 changed is enough for us to
3019                          * add this file to the list of results
3020                          * (with each file contributing equal damage).
3021                          */
3022                         damage = 1;
3023                         goto found_damage;
3024                 }
3025
3026                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3027                         diff_populate_filespec(options->repo, p->one, 0);
3028                         diff_populate_filespec(options->repo, p->two, 0);
3029                         diffcore_count_changes(options->repo,
3030                                                p->one, p->two, NULL, NULL,
3031                                                &copied, &added);
3032                         diff_free_filespec_data(p->one);
3033                         diff_free_filespec_data(p->two);
3034                 } else if (DIFF_FILE_VALID(p->one)) {
3035                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3036                         copied = added = 0;
3037                         diff_free_filespec_data(p->one);
3038                 } else if (DIFF_FILE_VALID(p->two)) {
3039                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3040                         copied = 0;
3041                         added = p->two->size;
3042                         diff_free_filespec_data(p->two);
3043                 } else
3044                         continue;
3045
3046                 /*
3047                  * Original minus copied is the removed material,
3048                  * added is the new material.  They are both damages
3049                  * made to the preimage.
3050                  * If the resulting damage is zero, we know that
3051                  * diffcore_count_changes() considers the two entries to
3052                  * be identical, but since the oid changed, we
3053                  * know that there must have been _some_ kind of change,
3054                  * so we force all entries to have damage > 0.
3055                  */
3056                 damage = (p->one->size - copied) + added;
3057                 if (!damage)
3058                         damage = 1;
3059
3060 found_damage:
3061                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3062                 dir.files[dir.nr].name = name;
3063                 dir.files[dir.nr].changed = damage;
3064                 changed += damage;
3065                 dir.nr++;
3066         }
3067
3068         /* This can happen even with many files, if everything was renames */
3069         if (!changed)
3070                 return;
3071
3072         /* Show all directories with more than x% of the changes */
3073         QSORT(dir.files, dir.nr, dirstat_compare);
3074         gather_dirstat(options, &dir, changed, "", 0);
3075 }
3076
3077 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3078 {
3079         int i;
3080         unsigned long changed;
3081         struct dirstat_dir dir;
3082
3083         if (data->nr == 0)
3084                 return;
3085
3086         dir.files = NULL;
3087         dir.alloc = 0;
3088         dir.nr = 0;
3089         dir.permille = options->dirstat_permille;
3090         dir.cumulative = options->flags.dirstat_cumulative;
3091
3092         changed = 0;
3093         for (i = 0; i < data->nr; i++) {
3094                 struct diffstat_file *file = data->files[i];
3095                 unsigned long damage = file->added + file->deleted;
3096                 if (file->is_binary)
3097                         /*
3098                          * binary files counts bytes, not lines. Must find some
3099                          * way to normalize binary bytes vs. textual lines.
3100                          * The following heuristic assumes that there are 64
3101                          * bytes per "line".
3102                          * This is stupid and ugly, but very cheap...
3103                          */
3104                         damage = DIV_ROUND_UP(damage, 64);
3105                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3106                 dir.files[dir.nr].name = file->name;
3107                 dir.files[dir.nr].changed = damage;
3108                 changed += damage;
3109                 dir.nr++;
3110         }
3111
3112         /* This can happen even with many files, if everything was renames */
3113         if (!changed)
3114                 return;
3115
3116         /* Show all directories with more than x% of the changes */
3117         QSORT(dir.files, dir.nr, dirstat_compare);
3118         gather_dirstat(options, &dir, changed, "", 0);
3119 }
3120
3121 static void free_diffstat_info(struct diffstat_t *diffstat)
3122 {
3123         int i;
3124         for (i = 0; i < diffstat->nr; i++) {
3125                 struct diffstat_file *f = diffstat->files[i];
3126                 free(f->print_name);
3127                 free(f->name);
3128                 free(f->from_name);
3129                 free(f);
3130         }
3131         free(diffstat->files);
3132 }
3133
3134 struct checkdiff_t {
3135         const char *filename;
3136         int lineno;
3137         int conflict_marker_size;
3138         struct diff_options *o;
3139         unsigned ws_rule;
3140         unsigned status;
3141 };
3142
3143 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3144 {
3145         char firstchar;
3146         int cnt;
3147
3148         if (len < marker_size + 1)
3149                 return 0;
3150         firstchar = line[0];
3151         switch (firstchar) {
3152         case '=': case '>': case '<': case '|':
3153                 break;
3154         default:
3155                 return 0;
3156         }
3157         for (cnt = 1; cnt < marker_size; cnt++)
3158                 if (line[cnt] != firstchar)
3159                         return 0;
3160         /* line[1] thru line[marker_size-1] are same as firstchar */
3161         if (len < marker_size + 1 || !isspace(line[marker_size]))
3162                 return 0;
3163         return 1;
3164 }
3165
3166 static void checkdiff_consume_hunk(void *priv,
3167                                    long ob, long on, long nb, long nn,
3168                                    const char *func, long funclen)
3169
3170 {
3171         struct checkdiff_t *data = priv;
3172         data->lineno = nb - 1;
3173 }
3174
3175 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3176 {
3177         struct checkdiff_t *data = priv;
3178         int marker_size = data->conflict_marker_size;
3179         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3180         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3181         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3182         char *err;
3183         const char *line_prefix;
3184
3185         assert(data->o);
3186         line_prefix = diff_line_prefix(data->o);
3187
3188         if (line[0] == '+') {
3189                 unsigned bad;
3190                 data->lineno++;
3191                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3192                         data->status |= 1;
3193                         fprintf(data->o->file,
3194                                 "%s%s:%d: leftover conflict marker\n",
3195                                 line_prefix, data->filename, data->lineno);
3196                 }
3197                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3198                 if (!bad)
3199                         return;
3200                 data->status |= bad;
3201                 err = whitespace_error_string(bad);
3202                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3203                         line_prefix, data->filename, data->lineno, err);
3204                 free(err);
3205                 emit_line(data->o, set, reset, line, 1);
3206                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3207                               data->o->file, set, reset, ws);
3208         } else if (line[0] == ' ') {
3209                 data->lineno++;
3210         }
3211 }
3212
3213 static unsigned char *deflate_it(char *data,
3214                                  unsigned long size,
3215                                  unsigned long *result_size)
3216 {
3217         int bound;
3218         unsigned char *deflated;
3219         git_zstream stream;
3220
3221         git_deflate_init(&stream, zlib_compression_level);
3222         bound = git_deflate_bound(&stream, size);
3223         deflated = xmalloc(bound);
3224         stream.next_out = deflated;
3225         stream.avail_out = bound;
3226
3227         stream.next_in = (unsigned char *)data;
3228         stream.avail_in = size;
3229         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3230                 ; /* nothing */
3231         git_deflate_end(&stream);
3232         *result_size = stream.total_out;
3233         return deflated;
3234 }
3235
3236 static void emit_binary_diff_body(struct diff_options *o,
3237                                   mmfile_t *one, mmfile_t *two)
3238 {
3239         void *cp;
3240         void *delta;
3241         void *deflated;
3242         void *data;
3243         unsigned long orig_size;
3244         unsigned long delta_size;
3245         unsigned long deflate_size;
3246         unsigned long data_size;
3247
3248         /* We could do deflated delta, or we could do just deflated two,
3249          * whichever is smaller.
3250          */
3251         delta = NULL;
3252         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3253         if (one->size && two->size) {
3254                 delta = diff_delta(one->ptr, one->size,
3255                                    two->ptr, two->size,
3256                                    &delta_size, deflate_size);
3257                 if (delta) {
3258                         void *to_free = delta;
3259                         orig_size = delta_size;
3260                         delta = deflate_it(delta, delta_size, &delta_size);
3261                         free(to_free);
3262                 }
3263         }
3264
3265         if (delta && delta_size < deflate_size) {
3266                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3267                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3268                                  s, strlen(s), 0);
3269                 free(s);
3270                 free(deflated);
3271                 data = delta;
3272                 data_size = delta_size;
3273         } else {
3274                 char *s = xstrfmt("%lu", two->size);
3275                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3276                                  s, strlen(s), 0);
3277                 free(s);
3278                 free(delta);
3279                 data = deflated;
3280                 data_size = deflate_size;
3281         }
3282
3283         /* emit data encoded in base85 */
3284         cp = data;
3285         while (data_size) {
3286                 int len;
3287                 int bytes = (52 < data_size) ? 52 : data_size;
3288                 char line[71];
3289                 data_size -= bytes;
3290                 if (bytes <= 26)
3291                         line[0] = bytes + 'A' - 1;
3292                 else
3293                         line[0] = bytes - 26 + 'a' - 1;
3294                 encode_85(line + 1, cp, bytes);
3295                 cp = (char *) cp + bytes;
3296
3297                 len = strlen(line);
3298                 line[len++] = '\n';
3299                 line[len] = '\0';
3300
3301                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3302                                  line, len, 0);
3303         }
3304         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3305         free(data);
3306 }
3307
3308 static void emit_binary_diff(struct diff_options *o,
3309                              mmfile_t *one, mmfile_t *two)
3310 {
3311         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3312         emit_binary_diff_body(o, one, two);
3313         emit_binary_diff_body(o, two, one);
3314 }
3315
3316 int diff_filespec_is_binary(struct repository *r,
3317                             struct diff_filespec *one)
3318 {
3319         if (one->is_binary == -1) {
3320                 diff_filespec_load_driver(one, r->index);
3321                 if (one->driver->binary != -1)
3322                         one->is_binary = one->driver->binary;
3323                 else {
3324                         if (!one->data && DIFF_FILE_VALID(one))
3325                                 diff_populate_filespec(r, one, CHECK_BINARY);
3326                         if (one->is_binary == -1 && one->data)
3327                                 one->is_binary = buffer_is_binary(one->data,
3328                                                 one->size);
3329                         if (one->is_binary == -1)
3330                                 one->is_binary = 0;
3331                 }
3332         }
3333         return one->is_binary;
3334 }
3335
3336 static const struct userdiff_funcname *
3337 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3338 {
3339         diff_filespec_load_driver(one, o->repo->index);
3340         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3341 }
3342
3343 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3344 {
3345         if (!options->a_prefix)
3346                 options->a_prefix = a;
3347         if (!options->b_prefix)
3348                 options->b_prefix = b;
3349 }
3350
3351 struct userdiff_driver *get_textconv(struct index_state *istate,
3352                                      struct diff_filespec *one)
3353 {
3354         if (!DIFF_FILE_VALID(one))
3355                 return NULL;
3356
3357         diff_filespec_load_driver(one, istate);
3358         return userdiff_get_textconv(one->driver);
3359 }
3360
3361 static void builtin_diff(const char *name_a,
3362                          const char *name_b,
3363                          struct diff_filespec *one,
3364                          struct diff_filespec *two,
3365                          const char *xfrm_msg,
3366                          int must_show_header,
3367                          struct diff_options *o,
3368                          int complete_rewrite)
3369 {
3370         mmfile_t mf1, mf2;
3371         const char *lbl[2];
3372         char *a_one, *b_two;
3373         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3374         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3375         const char *a_prefix, *b_prefix;
3376         struct userdiff_driver *textconv_one = NULL;
3377         struct userdiff_driver *textconv_two = NULL;
3378         struct strbuf header = STRBUF_INIT;
3379         const char *line_prefix = diff_line_prefix(o);
3380
3381         diff_set_mnemonic_prefix(o, "a/", "b/");
3382         if (o->flags.reverse_diff) {
3383                 a_prefix = o->b_prefix;
3384                 b_prefix = o->a_prefix;
3385         } else {
3386                 a_prefix = o->a_prefix;
3387                 b_prefix = o->b_prefix;
3388         }
3389
3390         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3391             (!one->mode || S_ISGITLINK(one->mode)) &&
3392             (!two->mode || S_ISGITLINK(two->mode))) {
3393                 show_submodule_summary(o, one->path ? one->path : two->path,
3394                                 &one->oid, &two->oid,
3395                                 two->dirty_submodule);
3396                 return;
3397         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3398                    (!one->mode || S_ISGITLINK(one->mode)) &&
3399                    (!two->mode || S_ISGITLINK(two->mode))) {
3400                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3401                                 &one->oid, &two->oid,
3402                                 two->dirty_submodule);
3403                 return;
3404         }
3405
3406         if (o->flags.allow_textconv) {
3407                 textconv_one = get_textconv(o->repo->index, one);
3408                 textconv_two = get_textconv(o->repo->index, two);
3409         }
3410
3411         /* Never use a non-valid filename anywhere if at all possible */
3412         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3413         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3414
3415         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3416         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3417         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3418         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3419         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3420         if (lbl[0][0] == '/') {
3421                 /* /dev/null */
3422                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3423                 if (xfrm_msg)
3424                         strbuf_addstr(&header, xfrm_msg);
3425                 must_show_header = 1;
3426         }
3427         else if (lbl[1][0] == '/') {
3428                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3429                 if (xfrm_msg)
3430                         strbuf_addstr(&header, xfrm_msg);
3431                 must_show_header = 1;
3432         }
3433         else {
3434                 if (one->mode != two->mode) {
3435                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3436                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3437                         must_show_header = 1;
3438                 }
3439                 if (xfrm_msg)
3440                         strbuf_addstr(&header, xfrm_msg);
3441
3442                 /*
3443                  * we do not run diff between different kind
3444                  * of objects.
3445                  */
3446                 if ((one->mode ^ two->mode) & S_IFMT)
3447                         goto free_ab_and_return;
3448                 if (complete_rewrite &&
3449                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3450                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3451                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3452                                          header.buf, header.len, 0);
3453                         strbuf_reset(&header);
3454                         emit_rewrite_diff(name_a, name_b, one, two,
3455                                           textconv_one, textconv_two, o);
3456                         o->found_changes = 1;
3457                         goto free_ab_and_return;
3458                 }
3459         }
3460
3461         if (o->irreversible_delete && lbl[1][0] == '/') {
3462                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3463                                  header.len, 0);
3464                 strbuf_reset(&header);
3465                 goto free_ab_and_return;
3466         } else if (!o->flags.text &&
3467                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3468                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3469                 struct strbuf sb = STRBUF_INIT;
3470                 if (!one->data && !two->data &&
3471                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3472                     !o->flags.binary) {
3473                         if (oideq(&one->oid, &two->oid)) {
3474                                 if (must_show_header)
3475                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3476                                                          header.buf, header.len,
3477                                                          0);
3478                                 goto free_ab_and_return;
3479                         }
3480                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3481                                          header.buf, header.len, 0);
3482                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3483                                     diff_line_prefix(o), lbl[0], lbl[1]);
3484                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3485                                          sb.buf, sb.len, 0);
3486                         strbuf_release(&sb);
3487                         goto free_ab_and_return;
3488                 }
3489                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3490                     fill_mmfile(o->repo, &mf2, two) < 0)
3491                         die("unable to read files to diff");
3492                 /* Quite common confusing case */
3493                 if (mf1.size == mf2.size &&
3494                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3495                         if (must_show_header)
3496                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3497                                                  header.buf, header.len, 0);
3498                         goto free_ab_and_return;
3499                 }
3500                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3501                 strbuf_reset(&header);
3502                 if (o->flags.binary)
3503                         emit_binary_diff(o, &mf1, &mf2);
3504                 else {
3505                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3506                                     diff_line_prefix(o), lbl[0], lbl[1]);
3507                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3508                                          sb.buf, sb.len, 0);
3509                         strbuf_release(&sb);
3510                 }
3511                 o->found_changes = 1;
3512         } else {
3513                 /* Crazy xdl interfaces.. */
3514                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3515                 const char *v;
3516                 xpparam_t xpp;
3517                 xdemitconf_t xecfg;
3518                 struct emit_callback ecbdata;
3519                 const struct userdiff_funcname *pe;
3520
3521                 if (must_show_header) {
3522                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3523                                          header.buf, header.len, 0);
3524                         strbuf_reset(&header);
3525                 }
3526
3527                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3528                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3529
3530                 pe = diff_funcname_pattern(o, one);
3531                 if (!pe)
3532                         pe = diff_funcname_pattern(o, two);
3533
3534                 memset(&xpp, 0, sizeof(xpp));
3535                 memset(&xecfg, 0, sizeof(xecfg));
3536                 memset(&ecbdata, 0, sizeof(ecbdata));
3537                 if (o->flags.suppress_diff_headers)
3538                         lbl[0] = NULL;
3539                 ecbdata.label_path = lbl;
3540                 ecbdata.color_diff = want_color(o->use_color);
3541                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3542                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3543                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3544                 ecbdata.opt = o;
3545                 if (header.len && !o->flags.suppress_diff_headers)
3546                         ecbdata.header = &header;
3547                 xpp.flags = o->xdl_opts;
3548                 xpp.anchors = o->anchors;
3549                 xpp.anchors_nr = o->anchors_nr;
3550                 xecfg.ctxlen = o->context;
3551                 xecfg.interhunkctxlen = o->interhunkcontext;
3552                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3553                 if (o->flags.funccontext)
3554                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3555                 if (pe)
3556                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3557                 if (!diffopts)
3558                         ;
3559                 else if (skip_prefix(diffopts, "--unified=", &v))
3560                         xecfg.ctxlen = strtoul(v, NULL, 10);
3561                 else if (skip_prefix(diffopts, "-u", &v))
3562                         xecfg.ctxlen = strtoul(v, NULL, 10);
3563                 if (o->word_diff)
3564                         init_diff_words_data(&ecbdata, o, one, two);
3565                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3566                                   &ecbdata, &xpp, &xecfg))
3567                         die("unable to generate diff for %s", one->path);
3568                 if (o->word_diff)
3569                         free_diff_words_data(&ecbdata);
3570                 if (textconv_one)
3571                         free(mf1.ptr);
3572                 if (textconv_two)
3573                         free(mf2.ptr);
3574                 xdiff_clear_find_func(&xecfg);
3575         }
3576
3577  free_ab_and_return:
3578         strbuf_release(&header);
3579         diff_free_filespec_data(one);
3580         diff_free_filespec_data(two);
3581         free(a_one);
3582         free(b_two);
3583         return;
3584 }
3585
3586 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3587 {
3588         if (!is_renamed) {
3589                 if (p->status == DIFF_STATUS_ADDED) {
3590                         if (S_ISLNK(p->two->mode))
3591                                 return "new +l";
3592                         else if ((p->two->mode & 0777) == 0755)
3593                                 return "new +x";
3594                         else
3595                                 return "new";
3596                 } else if (p->status == DIFF_STATUS_DELETED)
3597                         return "gone";
3598         }
3599         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3600                 return "mode -l";
3601         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3602                 return "mode +l";
3603         else if ((p->one->mode & 0777) == 0644 &&
3604                  (p->two->mode & 0777) == 0755)
3605                 return "mode +x";
3606         else if ((p->one->mode & 0777) == 0755 &&
3607                  (p->two->mode & 0777) == 0644)
3608                 return "mode -x";
3609         return NULL;
3610 }
3611
3612 static void builtin_diffstat(const char *name_a, const char *name_b,
3613                              struct diff_filespec *one,
3614                              struct diff_filespec *two,
3615                              struct diffstat_t *diffstat,
3616                              struct diff_options *o,
3617                              struct diff_filepair *p)
3618 {
3619         mmfile_t mf1, mf2;
3620         struct diffstat_file *data;
3621         int same_contents;
3622         int complete_rewrite = 0;
3623
3624         if (!DIFF_PAIR_UNMERGED(p)) {
3625                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3626                         complete_rewrite = 1;
3627         }
3628
3629         data = diffstat_add(diffstat, name_a, name_b);
3630         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3631         if (o->flags.stat_with_summary)
3632                 data->comments = get_compact_summary(p, data->is_renamed);
3633
3634         if (!one || !two) {
3635                 data->is_unmerged = 1;
3636                 return;
3637         }
3638
3639         same_contents = oideq(&one->oid, &two->oid);
3640
3641         if (diff_filespec_is_binary(o->repo, one) ||
3642             diff_filespec_is_binary(o->repo, two)) {
3643                 data->is_binary = 1;
3644                 if (same_contents) {
3645                         data->added = 0;
3646                         data->deleted = 0;
3647                 } else {
3648                         data->added = diff_filespec_size(o->repo, two);
3649                         data->deleted = diff_filespec_size(o->repo, one);
3650                 }
3651         }
3652
3653         else if (complete_rewrite) {
3654                 diff_populate_filespec(o->repo, one, 0);
3655                 diff_populate_filespec(o->repo, two, 0);
3656                 data->deleted = count_lines(one->data, one->size);
3657                 data->added = count_lines(two->data, two->size);
3658         }
3659
3660         else if (!same_contents) {
3661                 /* Crazy xdl interfaces.. */
3662                 xpparam_t xpp;
3663                 xdemitconf_t xecfg;
3664
3665                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3666                     fill_mmfile(o->repo, &mf2, two) < 0)
3667                         die("unable to read files to diff");
3668
3669                 memset(&xpp, 0, sizeof(xpp));
3670                 memset(&xecfg, 0, sizeof(xecfg));
3671                 xpp.flags = o->xdl_opts;
3672                 xpp.anchors = o->anchors;
3673                 xpp.anchors_nr = o->anchors_nr;
3674                 xecfg.ctxlen = o->context;
3675                 xecfg.interhunkctxlen = o->interhunkcontext;
3676                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3677                                   diffstat_consume, diffstat, &xpp, &xecfg))
3678                         die("unable to generate diffstat for %s", one->path);
3679         }
3680
3681         diff_free_filespec_data(one);
3682         diff_free_filespec_data(two);
3683 }
3684
3685 static void builtin_checkdiff(const char *name_a, const char *name_b,
3686                               const char *attr_path,
3687                               struct diff_filespec *one,
3688                               struct diff_filespec *two,
3689                               struct diff_options *o)
3690 {
3691         mmfile_t mf1, mf2;
3692         struct checkdiff_t data;
3693
3694         if (!two)
3695                 return;
3696
3697         memset(&data, 0, sizeof(data));
3698         data.filename = name_b ? name_b : name_a;
3699         data.lineno = 0;
3700         data.o = o;
3701         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3702         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3703
3704         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3705             fill_mmfile(o->repo, &mf2, two) < 0)
3706                 die("unable to read files to diff");
3707
3708         /*
3709          * All the other codepaths check both sides, but not checking
3710          * the "old" side here is deliberate.  We are checking the newly
3711          * introduced changes, and as long as the "new" side is text, we
3712          * can and should check what it introduces.
3713          */
3714         if (diff_filespec_is_binary(o->repo, two))
3715                 goto free_and_return;
3716         else {
3717                 /* Crazy xdl interfaces.. */
3718                 xpparam_t xpp;
3719                 xdemitconf_t xecfg;
3720
3721                 memset(&xpp, 0, sizeof(xpp));
3722                 memset(&xecfg, 0, sizeof(xecfg));
3723                 xecfg.ctxlen = 1; /* at least one context line */
3724                 xpp.flags = 0;
3725                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3726                                   checkdiff_consume, &data,
3727                                   &xpp, &xecfg))
3728                         die("unable to generate checkdiff for %s", one->path);
3729
3730                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3731                         struct emit_callback ecbdata;
3732                         int blank_at_eof;
3733
3734                         ecbdata.ws_rule = data.ws_rule;
3735                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3736                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3737
3738                         if (blank_at_eof) {
3739                                 static char *err;
3740                                 if (!err)
3741                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3742                                 fprintf(o->file, "%s:%d: %s.\n",
3743                                         data.filename, blank_at_eof, err);
3744                                 data.status = 1; /* report errors */
3745                         }
3746                 }
3747         }
3748  free_and_return:
3749         diff_free_filespec_data(one);
3750         diff_free_filespec_data(two);
3751         if (data.status)
3752                 o->flags.check_failed = 1;
3753 }
3754
3755 struct diff_filespec *alloc_filespec(const char *path)
3756 {
3757         struct diff_filespec *spec;
3758
3759         FLEXPTR_ALLOC_STR(spec, path, path);
3760         spec->count = 1;
3761         spec->is_binary = -1;
3762         return spec;
3763 }
3764
3765 void free_filespec(struct diff_filespec *spec)
3766 {
3767         if (!--spec->count) {
3768                 diff_free_filespec_data(spec);
3769                 free(spec);
3770         }
3771 }
3772
3773 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3774                    int oid_valid, unsigned short mode)
3775 {
3776         if (mode) {
3777                 spec->mode = canon_mode(mode);
3778                 oidcpy(&spec->oid, oid);
3779                 spec->oid_valid = oid_valid;
3780         }
3781 }
3782
3783 /*
3784  * Given a name and sha1 pair, if the index tells us the file in
3785  * the work tree has that object contents, return true, so that
3786  * prepare_temp_file() does not have to inflate and extract.
3787  */
3788 static int reuse_worktree_file(struct index_state *istate,
3789                                const char *name,
3790                                const struct object_id *oid,
3791                                int want_file)
3792 {
3793         const struct cache_entry *ce;
3794         struct stat st;
3795         int pos, len;
3796
3797         /*
3798          * We do not read the cache ourselves here, because the
3799          * benchmark with my previous version that always reads cache
3800          * shows that it makes things worse for diff-tree comparing
3801          * two linux-2.6 kernel trees in an already checked out work
3802          * tree.  This is because most diff-tree comparisons deal with
3803          * only a small number of files, while reading the cache is
3804          * expensive for a large project, and its cost outweighs the
3805          * savings we get by not inflating the object to a temporary
3806          * file.  Practically, this code only helps when we are used
3807          * by diff-cache --cached, which does read the cache before
3808          * calling us.
3809          */
3810         if (!istate->cache)
3811                 return 0;
3812
3813         /* We want to avoid the working directory if our caller
3814          * doesn't need the data in a normal file, this system
3815          * is rather slow with its stat/open/mmap/close syscalls,
3816          * and the object is contained in a pack file.  The pack
3817          * is probably already open and will be faster to obtain
3818          * the data through than the working directory.  Loose
3819          * objects however would tend to be slower as they need
3820          * to be individually opened and inflated.
3821          */
3822         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3823                 return 0;
3824
3825         /*
3826          * Similarly, if we'd have to convert the file contents anyway, that
3827          * makes the optimization not worthwhile.
3828          */
3829         if (!want_file && would_convert_to_git(istate, name))
3830                 return 0;
3831
3832         len = strlen(name);
3833         pos = index_name_pos(istate, name, len);
3834         if (pos < 0)
3835                 return 0;
3836         ce = istate->cache[pos];
3837
3838         /*
3839          * This is not the sha1 we are looking for, or
3840          * unreusable because it is not a regular file.
3841          */
3842         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3843                 return 0;
3844
3845         /*
3846          * If ce is marked as "assume unchanged", there is no
3847          * guarantee that work tree matches what we are looking for.
3848          */
3849         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3850                 return 0;
3851
3852         /*
3853          * If ce matches the file in the work tree, we can reuse it.
3854          */
3855         if (ce_uptodate(ce) ||
3856             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3857                 return 1;
3858
3859         return 0;
3860 }
3861
3862 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3863 {
3864         struct strbuf buf = STRBUF_INIT;
3865         char *dirty = "";
3866
3867         /* Are we looking at the work tree? */
3868         if (s->dirty_submodule)
3869                 dirty = "-dirty";
3870
3871         strbuf_addf(&buf, "Subproject commit %s%s\n",
3872                     oid_to_hex(&s->oid), dirty);
3873         s->size = buf.len;
3874         if (size_only) {
3875                 s->data = NULL;
3876                 strbuf_release(&buf);
3877         } else {
3878                 s->data = strbuf_detach(&buf, NULL);
3879                 s->should_free = 1;
3880         }
3881         return 0;
3882 }
3883
3884 /*
3885  * While doing rename detection and pickaxe operation, we may need to
3886  * grab the data for the blob (or file) for our own in-core comparison.
3887  * diff_filespec has data and size fields for this purpose.
3888  */
3889 int diff_populate_filespec(struct repository *r,
3890                            struct diff_filespec *s,
3891                            unsigned int flags)
3892 {
3893         int size_only = flags & CHECK_SIZE_ONLY;
3894         int err = 0;
3895         int conv_flags = global_conv_flags_eol;
3896         /*
3897          * demote FAIL to WARN to allow inspecting the situation
3898          * instead of refusing.
3899          */
3900         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3901                 conv_flags = CONV_EOL_RNDTRP_WARN;
3902
3903         if (!DIFF_FILE_VALID(s))
3904                 die("internal error: asking to populate invalid file.");
3905         if (S_ISDIR(s->mode))
3906                 return -1;
3907
3908         if (s->data)
3909                 return 0;
3910
3911         if (size_only && 0 < s->size)
3912                 return 0;
3913
3914         if (S_ISGITLINK(s->mode))
3915                 return diff_populate_gitlink(s, size_only);
3916
3917         if (!s->oid_valid ||
3918             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3919                 struct strbuf buf = STRBUF_INIT;
3920                 struct stat st;
3921                 int fd;
3922
3923                 if (lstat(s->path, &st) < 0) {
3924                 err_empty:
3925                         err = -1;
3926                 empty:
3927                         s->data = (char *)"";
3928                         s->size = 0;
3929                         return err;
3930                 }
3931                 s->size = xsize_t(st.st_size);
3932                 if (!s->size)
3933                         goto empty;
3934                 if (S_ISLNK(st.st_mode)) {
3935                         struct strbuf sb = STRBUF_INIT;
3936
3937                         if (strbuf_readlink(&sb, s->path, s->size))
3938                                 goto err_empty;
3939                         s->size = sb.len;
3940                         s->data = strbuf_detach(&sb, NULL);
3941                         s->should_free = 1;
3942                         return 0;
3943                 }
3944
3945                 /*
3946                  * Even if the caller would be happy with getting
3947                  * only the size, we cannot return early at this
3948                  * point if the path requires us to run the content
3949                  * conversion.
3950                  */
3951                 if (size_only && !would_convert_to_git(r->index, s->path))
3952                         return 0;
3953
3954                 /*
3955                  * Note: this check uses xsize_t(st.st_size) that may
3956                  * not be the true size of the blob after it goes
3957                  * through convert_to_git().  This may not strictly be
3958                  * correct, but the whole point of big_file_threshold
3959                  * and is_binary check being that we want to avoid
3960                  * opening the file and inspecting the contents, this
3961                  * is probably fine.
3962                  */
3963                 if ((flags & CHECK_BINARY) &&
3964                     s->size > big_file_threshold && s->is_binary == -1) {
3965                         s->is_binary = 1;
3966                         return 0;
3967                 }
3968                 fd = open(s->path, O_RDONLY);
3969                 if (fd < 0)
3970                         goto err_empty;
3971                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3972                 close(fd);
3973                 s->should_munmap = 1;
3974
3975                 /*
3976                  * Convert from working tree format to canonical git format
3977                  */
3978                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3979                         size_t size = 0;
3980                         munmap(s->data, s->size);
3981                         s->should_munmap = 0;
3982                         s->data = strbuf_detach(&buf, &size);
3983                         s->size = size;
3984                         s->should_free = 1;
3985                 }
3986         }
3987         else {
3988                 enum object_type type;
3989                 if (size_only || (flags & CHECK_BINARY)) {
3990                         type = oid_object_info(r, &s->oid, &s->size);
3991                         if (type < 0)
3992                                 die("unable to read %s",
3993                                     oid_to_hex(&s->oid));
3994                         if (size_only)
3995                                 return 0;
3996                         if (s->size > big_file_threshold && s->is_binary == -1) {
3997                                 s->is_binary = 1;
3998                                 return 0;
3999                         }
4000                 }
4001                 s->data = read_object_file(&s->oid, &type, &s->size);
4002                 if (!s->data)
4003                         die("unable to read %s", oid_to_hex(&s->oid));
4004                 s->should_free = 1;
4005         }
4006         return 0;
4007 }
4008
4009 void diff_free_filespec_blob(struct diff_filespec *s)
4010 {
4011         if (s->should_free)
4012                 free(s->data);
4013         else if (s->should_munmap)
4014                 munmap(s->data, s->size);
4015
4016         if (s->should_free || s->should_munmap) {
4017                 s->should_free = s->should_munmap = 0;
4018                 s->data = NULL;
4019         }
4020 }
4021
4022 void diff_free_filespec_data(struct diff_filespec *s)
4023 {
4024         diff_free_filespec_blob(s);
4025         FREE_AND_NULL(s->cnt_data);
4026 }
4027
4028 static void prep_temp_blob(struct index_state *istate,
4029                            const char *path, struct diff_tempfile *temp,
4030                            void *blob,
4031                            unsigned long size,
4032                            const struct object_id *oid,
4033                            int mode)
4034 {
4035         struct strbuf buf = STRBUF_INIT;
4036         struct strbuf tempfile = STRBUF_INIT;
4037         char *path_dup = xstrdup(path);
4038         const char *base = basename(path_dup);
4039
4040         /* Generate "XXXXXX_basename.ext" */
4041         strbuf_addstr(&tempfile, "XXXXXX_");
4042         strbuf_addstr(&tempfile, base);
4043
4044         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4045         if (!temp->tempfile)
4046                 die_errno("unable to create temp-file");
4047         if (convert_to_working_tree(istate, path,
4048                         (const char *)blob, (size_t)size, &buf)) {
4049                 blob = buf.buf;
4050                 size = buf.len;
4051         }
4052         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4053             close_tempfile_gently(temp->tempfile))
4054                 die_errno("unable to write temp-file");
4055         temp->name = get_tempfile_path(temp->tempfile);
4056         oid_to_hex_r(temp->hex, oid);
4057         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4058         strbuf_release(&buf);
4059         strbuf_release(&tempfile);
4060         free(path_dup);
4061 }
4062
4063 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4064                                                const char *name,
4065                                                struct diff_filespec *one)
4066 {
4067         struct diff_tempfile *temp = claim_diff_tempfile();
4068
4069         if (!DIFF_FILE_VALID(one)) {
4070         not_a_valid_file:
4071                 /* A '-' entry produces this for file-2, and
4072                  * a '+' entry produces this for file-1.
4073                  */
4074                 temp->name = "/dev/null";
4075                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4076                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4077                 return temp;
4078         }
4079
4080         if (!S_ISGITLINK(one->mode) &&
4081             (!one->oid_valid ||
4082              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4083                 struct stat st;
4084                 if (lstat(name, &st) < 0) {
4085                         if (errno == ENOENT)
4086                                 goto not_a_valid_file;
4087                         die_errno("stat(%s)", name);
4088                 }
4089                 if (S_ISLNK(st.st_mode)) {
4090                         struct strbuf sb = STRBUF_INIT;
4091                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4092                                 die_errno("readlink(%s)", name);
4093                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4094                                        (one->oid_valid ?
4095                                         &one->oid : &null_oid),
4096                                        (one->oid_valid ?
4097                                         one->mode : S_IFLNK));
4098                         strbuf_release(&sb);
4099                 }
4100                 else {
4101                         /* we can borrow from the file in the work tree */
4102                         temp->name = name;
4103                         if (!one->oid_valid)
4104                                 oid_to_hex_r(temp->hex, &null_oid);
4105                         else
4106                                 oid_to_hex_r(temp->hex, &one->oid);
4107                         /* Even though we may sometimes borrow the
4108                          * contents from the work tree, we always want
4109                          * one->mode.  mode is trustworthy even when
4110                          * !(one->oid_valid), as long as
4111                          * DIFF_FILE_VALID(one).
4112                          */
4113                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4114                 }
4115                 return temp;
4116         }
4117         else {
4118                 if (diff_populate_filespec(r, one, 0))
4119                         die("cannot read data blob for %s", one->path);
4120                 prep_temp_blob(r->index, name, temp,
4121                                one->data, one->size,
4122                                &one->oid, one->mode);
4123         }
4124         return temp;
4125 }
4126
4127 static void add_external_diff_name(struct repository *r,
4128                                    struct argv_array *argv,
4129                                    const char *name,
4130                                    struct diff_filespec *df)
4131 {
4132         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4133         argv_array_push(argv, temp->name);
4134         argv_array_push(argv, temp->hex);
4135         argv_array_push(argv, temp->mode);
4136 }
4137
4138 /* An external diff command takes:
4139  *
4140  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4141  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4142  *
4143  */
4144 static void run_external_diff(const char *pgm,
4145                               const char *name,
4146                               const char *other,
4147                               struct diff_filespec *one,
4148                               struct diff_filespec *two,
4149                               const char *xfrm_msg,
4150                               int complete_rewrite,
4151                               struct diff_options *o)
4152 {
4153         struct argv_array argv = ARGV_ARRAY_INIT;
4154         struct argv_array env = ARGV_ARRAY_INIT;
4155         struct diff_queue_struct *q = &diff_queued_diff;
4156
4157         argv_array_push(&argv, pgm);
4158         argv_array_push(&argv, name);
4159
4160         if (one && two) {
4161                 add_external_diff_name(o->repo, &argv, name, one);
4162                 if (!other)
4163                         add_external_diff_name(o->repo, &argv, name, two);
4164                 else {
4165                         add_external_diff_name(o->repo, &argv, other, two);
4166                         argv_array_push(&argv, other);
4167                         argv_array_push(&argv, xfrm_msg);
4168                 }
4169         }
4170
4171         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4172         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4173
4174         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4175                 die(_("external diff died, stopping at %s"), name);
4176
4177         remove_tempfile();
4178         argv_array_clear(&argv);
4179         argv_array_clear(&env);
4180 }
4181
4182 static int similarity_index(struct diff_filepair *p)
4183 {
4184         return p->score * 100 / MAX_SCORE;
4185 }
4186
4187 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4188 {
4189         if (startup_info->have_repository)
4190                 return find_unique_abbrev(oid, abbrev);
4191         else {
4192                 char *hex = oid_to_hex(oid);
4193                 if (abbrev < 0)
4194                         abbrev = FALLBACK_DEFAULT_ABBREV;
4195                 if (abbrev > the_hash_algo->hexsz)
4196                         BUG("oid abbreviation out of range: %d", abbrev);
4197                 if (abbrev)
4198                         hex[abbrev] = '\0';
4199                 return hex;
4200         }
4201 }
4202
4203 static void fill_metainfo(struct strbuf *msg,
4204                           const char *name,
4205                           const char *other,
4206                           struct diff_filespec *one,
4207                           struct diff_filespec *two,
4208                           struct diff_options *o,
4209                           struct diff_filepair *p,
4210                           int *must_show_header,
4211                           int use_color)
4212 {
4213         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4214         const char *reset = diff_get_color(use_color, DIFF_RESET);
4215         const char *line_prefix = diff_line_prefix(o);
4216
4217         *must_show_header = 1;
4218         strbuf_init(msg, PATH_MAX * 2 + 300);
4219         switch (p->status) {
4220         case DIFF_STATUS_COPIED:
4221                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4222                             line_prefix, set, similarity_index(p));
4223                 strbuf_addf(msg, "%s\n%s%scopy from ",
4224                             reset,  line_prefix, set);
4225                 quote_c_style(name, msg, NULL, 0);
4226                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4227                 quote_c_style(other, msg, NULL, 0);
4228                 strbuf_addf(msg, "%s\n", reset);
4229                 break;
4230         case DIFF_STATUS_RENAMED:
4231                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4232                             line_prefix, set, similarity_index(p));
4233                 strbuf_addf(msg, "%s\n%s%srename from ",
4234                             reset, line_prefix, set);
4235                 quote_c_style(name, msg, NULL, 0);
4236                 strbuf_addf(msg, "%s\n%s%srename to ",
4237                             reset, line_prefix, set);
4238                 quote_c_style(other, msg, NULL, 0);
4239                 strbuf_addf(msg, "%s\n", reset);
4240                 break;
4241         case DIFF_STATUS_MODIFIED:
4242                 if (p->score) {
4243                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4244                                     line_prefix,
4245                                     set, similarity_index(p), reset);
4246                         break;
4247                 }
4248                 /* fallthru */
4249         default:
4250                 *must_show_header = 0;
4251         }
4252         if (one && two && !oideq(&one->oid, &two->oid)) {
4253                 const unsigned hexsz = the_hash_algo->hexsz;
4254                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4255
4256                 if (o->flags.binary) {
4257                         mmfile_t mf;
4258                         if ((!fill_mmfile(o->repo, &mf, one) &&
4259                              diff_filespec_is_binary(o->repo, one)) ||
4260                             (!fill_mmfile(o->repo, &mf, two) &&
4261                              diff_filespec_is_binary(o->repo, two)))
4262                                 abbrev = hexsz;
4263                 }
4264                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4265                             diff_abbrev_oid(&one->oid, abbrev),
4266                             diff_abbrev_oid(&two->oid, abbrev));
4267                 if (one->mode == two->mode)
4268                         strbuf_addf(msg, " %06o", one->mode);
4269                 strbuf_addf(msg, "%s\n", reset);
4270         }
4271 }
4272
4273 static void run_diff_cmd(const char *pgm,
4274                          const char *name,
4275                          const char *other,
4276                          const char *attr_path,
4277                          struct diff_filespec *one,
4278                          struct diff_filespec *two,
4279                          struct strbuf *msg,
4280                          struct diff_options *o,
4281                          struct diff_filepair *p)
4282 {
4283         const char *xfrm_msg = NULL;
4284         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4285         int must_show_header = 0;
4286
4287
4288         if (o->flags.allow_external) {
4289                 struct userdiff_driver *drv;
4290
4291                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4292                 if (drv && drv->external)
4293                         pgm = drv->external;
4294         }
4295
4296         if (msg) {
4297                 /*
4298                  * don't use colors when the header is intended for an
4299                  * external diff driver
4300                  */
4301                 fill_metainfo(msg, name, other, one, two, o, p,
4302                               &must_show_header,
4303                               want_color(o->use_color) && !pgm);
4304                 xfrm_msg = msg->len ? msg->buf : NULL;
4305         }
4306
4307         if (pgm) {
4308                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4309                                   complete_rewrite, o);
4310                 return;
4311         }
4312         if (one && two)
4313                 builtin_diff(name, other ? other : name,
4314                              one, two, xfrm_msg, must_show_header,
4315                              o, complete_rewrite);
4316         else
4317                 fprintf(o->file, "* Unmerged path %s\n", name);
4318 }
4319
4320 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4321 {
4322         if (DIFF_FILE_VALID(one)) {
4323                 if (!one->oid_valid) {
4324                         struct stat st;
4325                         if (one->is_stdin) {
4326                                 oidclr(&one->oid);
4327                                 return;
4328                         }
4329                         if (lstat(one->path, &st) < 0)
4330                                 die_errno("stat '%s'", one->path);
4331                         if (index_path(istate, &one->oid, one->path, &st, 0))
4332                                 die("cannot hash %s", one->path);
4333                 }
4334         }
4335         else
4336                 oidclr(&one->oid);
4337 }
4338
4339 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4340 {
4341         /* Strip the prefix but do not molest /dev/null and absolute paths */
4342         if (*namep && !is_absolute_path(*namep)) {
4343                 *namep += prefix_length;
4344                 if (**namep == '/')
4345                         ++*namep;
4346         }
4347         if (*otherp && !is_absolute_path(*otherp)) {
4348                 *otherp += prefix_length;
4349                 if (**otherp == '/')
4350                         ++*otherp;
4351         }
4352 }
4353
4354 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4355 {
4356         const char *pgm = external_diff();
4357         struct strbuf msg;
4358         struct diff_filespec *one = p->one;
4359         struct diff_filespec *two = p->two;
4360         const char *name;
4361         const char *other;
4362         const char *attr_path;
4363
4364         name  = one->path;
4365         other = (strcmp(name, two->path) ? two->path : NULL);
4366         attr_path = name;
4367         if (o->prefix_length)
4368                 strip_prefix(o->prefix_length, &name, &other);
4369
4370         if (!o->flags.allow_external)
4371                 pgm = NULL;
4372
4373         if (DIFF_PAIR_UNMERGED(p)) {
4374                 run_diff_cmd(pgm, name, NULL, attr_path,
4375                              NULL, NULL, NULL, o, p);
4376                 return;
4377         }
4378
4379         diff_fill_oid_info(one, o->repo->index);
4380         diff_fill_oid_info(two, o->repo->index);
4381
4382         if (!pgm &&
4383             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4384             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4385                 /*
4386                  * a filepair that changes between file and symlink
4387                  * needs to be split into deletion and creation.
4388                  */
4389                 struct diff_filespec *null = alloc_filespec(two->path);
4390                 run_diff_cmd(NULL, name, other, attr_path,
4391                              one, null, &msg,
4392                              o, p);
4393                 free(null);
4394                 strbuf_release(&msg);
4395
4396                 null = alloc_filespec(one->path);
4397                 run_diff_cmd(NULL, name, other, attr_path,
4398                              null, two, &msg, o, p);
4399                 free(null);
4400         }
4401         else
4402                 run_diff_cmd(pgm, name, other, attr_path,
4403                              one, two, &msg, o, p);
4404
4405         strbuf_release(&msg);
4406 }
4407
4408 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4409                          struct diffstat_t *diffstat)
4410 {
4411         const char *name;
4412         const char *other;
4413
4414         if (DIFF_PAIR_UNMERGED(p)) {
4415                 /* unmerged */
4416                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4417                                  diffstat, o, p);
4418                 return;
4419         }
4420
4421         name = p->one->path;
4422         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4423
4424         if (o->prefix_length)
4425                 strip_prefix(o->prefix_length, &name, &other);
4426
4427         diff_fill_oid_info(p->one, o->repo->index);
4428         diff_fill_oid_info(p->two, o->repo->index);
4429
4430         builtin_diffstat(name, other, p->one, p->two,
4431                          diffstat, o, p);
4432 }
4433
4434 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4435 {
4436         const char *name;
4437         const char *other;
4438         const char *attr_path;
4439
4440         if (DIFF_PAIR_UNMERGED(p)) {
4441                 /* unmerged */
4442                 return;
4443         }
4444
4445         name = p->one->path;
4446         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4447         attr_path = other ? other : name;
4448
4449         if (o->prefix_length)
4450                 strip_prefix(o->prefix_length, &name, &other);
4451
4452         diff_fill_oid_info(p->one, o->repo->index);
4453         diff_fill_oid_info(p->two, o->repo->index);
4454
4455         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4456 }
4457
4458 void repo_diff_setup(struct repository *r, struct diff_options *options)
4459 {
4460         memcpy(options, &default_diff_options, sizeof(*options));
4461
4462         options->file = stdout;
4463         options->repo = r;
4464
4465         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4466         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4467         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4468         options->abbrev = DEFAULT_ABBREV;
4469         options->line_termination = '\n';
4470         options->break_opt = -1;
4471         options->rename_limit = -1;
4472         options->dirstat_permille = diff_dirstat_permille_default;
4473         options->context = diff_context_default;
4474         options->interhunkcontext = diff_interhunk_context_default;
4475         options->ws_error_highlight = ws_error_highlight_default;
4476         options->flags.rename_empty = 1;
4477         options->objfind = NULL;
4478
4479         /* pathchange left =NULL by default */
4480         options->change = diff_change;
4481         options->add_remove = diff_addremove;
4482         options->use_color = diff_use_color_default;
4483         options->detect_rename = diff_detect_rename_default;
4484         options->xdl_opts |= diff_algorithm;
4485         if (diff_indent_heuristic)
4486                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4487
4488         options->orderfile = diff_order_file_cfg;
4489
4490         if (diff_no_prefix) {
4491                 options->a_prefix = options->b_prefix = "";
4492         } else if (!diff_mnemonic_prefix) {
4493                 options->a_prefix = "a/";
4494                 options->b_prefix = "b/";
4495         }
4496
4497         options->color_moved = diff_color_moved_default;
4498         options->color_moved_ws_handling = diff_color_moved_ws_default;
4499 }
4500
4501 void diff_setup_done(struct diff_options *options)
4502 {
4503         unsigned check_mask = DIFF_FORMAT_NAME |
4504                               DIFF_FORMAT_NAME_STATUS |
4505                               DIFF_FORMAT_CHECKDIFF |
4506                               DIFF_FORMAT_NO_OUTPUT;
4507         /*
4508          * This must be signed because we're comparing against a potentially
4509          * negative value.
4510          */
4511         const int hexsz = the_hash_algo->hexsz;
4512
4513         if (options->set_default)
4514                 options->set_default(options);
4515
4516         if (HAS_MULTI_BITS(options->output_format & check_mask))
4517                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4518
4519         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4520                 die(_("-G, -S and --find-object are mutually exclusive"));
4521
4522         /*
4523          * Most of the time we can say "there are changes"
4524          * only by checking if there are changed paths, but
4525          * --ignore-whitespace* options force us to look
4526          * inside contents.
4527          */
4528
4529         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4530                 options->flags.diff_from_contents = 1;
4531         else
4532                 options->flags.diff_from_contents = 0;
4533
4534         if (options->flags.find_copies_harder)
4535                 options->detect_rename = DIFF_DETECT_COPY;
4536
4537         if (!options->flags.relative_name)
4538                 options->prefix = NULL;
4539         if (options->prefix)
4540                 options->prefix_length = strlen(options->prefix);
4541         else
4542                 options->prefix_length = 0;
4543
4544         if (options->output_format & (DIFF_FORMAT_NAME |
4545                                       DIFF_FORMAT_NAME_STATUS |
4546                                       DIFF_FORMAT_CHECKDIFF |
4547                                       DIFF_FORMAT_NO_OUTPUT))
4548                 options->output_format &= ~(DIFF_FORMAT_RAW |
4549                                             DIFF_FORMAT_NUMSTAT |
4550                                             DIFF_FORMAT_DIFFSTAT |
4551                                             DIFF_FORMAT_SHORTSTAT |
4552                                             DIFF_FORMAT_DIRSTAT |
4553                                             DIFF_FORMAT_SUMMARY |
4554                                             DIFF_FORMAT_PATCH);
4555
4556         /*
4557          * These cases always need recursive; we do not drop caller-supplied
4558          * recursive bits for other formats here.
4559          */
4560         if (options->output_format & (DIFF_FORMAT_PATCH |
4561                                       DIFF_FORMAT_NUMSTAT |
4562                                       DIFF_FORMAT_DIFFSTAT |
4563                                       DIFF_FORMAT_SHORTSTAT |
4564                                       DIFF_FORMAT_DIRSTAT |
4565                                       DIFF_FORMAT_SUMMARY |
4566                                       DIFF_FORMAT_CHECKDIFF))
4567                 options->flags.recursive = 1;
4568         /*
4569          * Also pickaxe would not work very well if you do not say recursive
4570          */
4571         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4572                 options->flags.recursive = 1;
4573         /*
4574          * When patches are generated, submodules diffed against the work tree
4575          * must be checked for dirtiness too so it can be shown in the output
4576          */
4577         if (options->output_format & DIFF_FORMAT_PATCH)
4578                 options->flags.dirty_submodules = 1;
4579
4580         if (options->detect_rename && options->rename_limit < 0)
4581                 options->rename_limit = diff_rename_limit_default;
4582         if (hexsz < options->abbrev)
4583                 options->abbrev = hexsz; /* full */
4584
4585         /*
4586          * It does not make sense to show the first hit we happened
4587          * to have found.  It does not make sense not to return with
4588          * exit code in such a case either.
4589          */
4590         if (options->flags.quick) {
4591                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4592                 options->flags.exit_with_status = 1;
4593         }
4594
4595         options->diff_path_counter = 0;
4596
4597         if (options->flags.follow_renames && options->pathspec.nr != 1)
4598                 die(_("--follow requires exactly one pathspec"));
4599
4600         if (!options->use_color || external_diff())
4601                 options->color_moved = 0;
4602 }
4603
4604 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4605 {
4606         char c, *eq;
4607         int len;
4608
4609         if (*arg != '-')
4610                 return 0;
4611         c = *++arg;
4612         if (!c)
4613                 return 0;
4614         if (c == arg_short) {
4615                 c = *++arg;
4616                 if (!c)
4617                         return 1;
4618                 if (val && isdigit(c)) {
4619                         char *end;
4620                         int n = strtoul(arg, &end, 10);
4621                         if (*end)
4622                                 return 0;
4623                         *val = n;
4624                         return 1;
4625                 }
4626                 return 0;
4627         }
4628         if (c != '-')
4629                 return 0;
4630         arg++;
4631         eq = strchrnul(arg, '=');
4632         len = eq - arg;
4633         if (!len || strncmp(arg, arg_long, len))
4634                 return 0;
4635         if (*eq) {
4636                 int n;
4637                 char *end;
4638                 if (!isdigit(*++eq))
4639                         return 0;
4640                 n = strtoul(eq, &end, 10);
4641                 if (*end)
4642                         return 0;
4643                 *val = n;
4644         }
4645         return 1;
4646 }
4647
4648 static int diff_scoreopt_parse(const char *opt);
4649
4650 static inline int short_opt(char opt, const char **argv,
4651                             const char **optarg)
4652 {
4653         const char *arg = argv[0];
4654         if (arg[0] != '-' || arg[1] != opt)
4655                 return 0;
4656         if (arg[2] != '\0') {
4657                 *optarg = arg + 2;
4658                 return 1;
4659         }
4660         if (!argv[1])
4661                 die("Option '%c' requires a value", opt);
4662         *optarg = argv[1];
4663         return 2;
4664 }
4665
4666 int parse_long_opt(const char *opt, const char **argv,
4667                    const char **optarg)
4668 {
4669         const char *arg = argv[0];
4670         if (!skip_prefix(arg, "--", &arg))
4671                 return 0;
4672         if (!skip_prefix(arg, opt, &arg))
4673                 return 0;
4674         if (*arg == '=') { /* stuck form: --option=value */
4675                 *optarg = arg + 1;
4676                 return 1;
4677         }
4678         if (*arg != '\0')
4679                 return 0;
4680         /* separate form: --option value */
4681         if (!argv[1])
4682                 die("Option '--%s' requires a value", opt);
4683         *optarg = argv[1];
4684         return 2;
4685 }
4686
4687 static int stat_opt(struct diff_options *options, const char **av)
4688 {
4689         const char *arg = av[0];
4690         char *end;
4691         int width = options->stat_width;
4692         int name_width = options->stat_name_width;
4693         int graph_width = options->stat_graph_width;
4694         int count = options->stat_count;
4695         int argcount = 1;
4696
4697         if (!skip_prefix(arg, "--stat", &arg))
4698                 BUG("stat option does not begin with --stat: %s", arg);
4699         end = (char *)arg;
4700
4701         switch (*arg) {
4702         case '-':
4703                 if (skip_prefix(arg, "-width", &arg)) {
4704                         if (*arg == '=')
4705                                 width = strtoul(arg + 1, &end, 10);
4706                         else if (!*arg && !av[1])
4707                                 die_want_option("--stat-width");
4708                         else if (!*arg) {
4709                                 width = strtoul(av[1], &end, 10);
4710                                 argcount = 2;
4711                         }
4712                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4713                         if (*arg == '=')
4714                                 name_width = strtoul(arg + 1, &end, 10);
4715                         else if (!*arg && !av[1])
4716                                 die_want_option("--stat-name-width");
4717                         else if (!*arg) {
4718                                 name_width = strtoul(av[1], &end, 10);
4719                                 argcount = 2;
4720                         }
4721                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4722                         if (*arg == '=')
4723                                 graph_width = strtoul(arg + 1, &end, 10);
4724                         else if (!*arg && !av[1])
4725                                 die_want_option("--stat-graph-width");
4726                         else if (!*arg) {
4727                                 graph_width = strtoul(av[1], &end, 10);
4728                                 argcount = 2;
4729                         }
4730                 } else if (skip_prefix(arg, "-count", &arg)) {
4731                         if (*arg == '=')
4732                                 count = strtoul(arg + 1, &end, 10);
4733                         else if (!*arg && !av[1])
4734                                 die_want_option("--stat-count");
4735                         else if (!*arg) {
4736                                 count = strtoul(av[1], &end, 10);
4737                                 argcount = 2;
4738                         }
4739                 }
4740                 break;
4741         case '=':
4742                 width = strtoul(arg+1, &end, 10);
4743                 if (*end == ',')
4744                         name_width = strtoul(end+1, &end, 10);
4745                 if (*end == ',')
4746                         count = strtoul(end+1, &end, 10);
4747         }
4748
4749         /* Important! This checks all the error cases! */
4750         if (*end)
4751                 return 0;
4752         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4753         options->stat_name_width = name_width;
4754         options->stat_graph_width = graph_width;
4755         options->stat_width = width;
4756         options->stat_count = count;
4757         return argcount;
4758 }
4759
4760 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4761 {
4762         struct strbuf errmsg = STRBUF_INIT;
4763         if (parse_dirstat_params(options, params, &errmsg))
4764                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4765                     errmsg.buf);
4766         strbuf_release(&errmsg);
4767         /*
4768          * The caller knows a dirstat-related option is given from the command
4769          * line; allow it to say "return this_function();"
4770          */
4771         options->output_format |= DIFF_FORMAT_DIRSTAT;
4772         return 1;
4773 }
4774
4775 static int parse_submodule_opt(struct diff_options *options, const char *value)
4776 {
4777         if (parse_submodule_params(options, value))
4778                 die(_("Failed to parse --submodule option parameter: '%s'"),
4779                         value);
4780         return 1;
4781 }
4782
4783 static const char diff_status_letters[] = {
4784         DIFF_STATUS_ADDED,
4785         DIFF_STATUS_COPIED,
4786         DIFF_STATUS_DELETED,
4787         DIFF_STATUS_MODIFIED,
4788         DIFF_STATUS_RENAMED,
4789         DIFF_STATUS_TYPE_CHANGED,
4790         DIFF_STATUS_UNKNOWN,
4791         DIFF_STATUS_UNMERGED,
4792         DIFF_STATUS_FILTER_AON,
4793         DIFF_STATUS_FILTER_BROKEN,
4794         '\0',
4795 };
4796
4797 static unsigned int filter_bit['Z' + 1];
4798
4799 static void prepare_filter_bits(void)
4800 {
4801         int i;
4802
4803         if (!filter_bit[DIFF_STATUS_ADDED]) {
4804                 for (i = 0; diff_status_letters[i]; i++)
4805                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4806         }
4807 }
4808
4809 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4810 {
4811         return opt->filter & filter_bit[(int) status];
4812 }
4813
4814 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4815 {
4816         int i, optch;
4817
4818         prepare_filter_bits();
4819
4820         /*
4821          * If there is a negation e.g. 'd' in the input, and we haven't
4822          * initialized the filter field with another --diff-filter, start
4823          * from full set of bits, except for AON.
4824          */
4825         if (!opt->filter) {
4826                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4827                         if (optch < 'a' || 'z' < optch)
4828                                 continue;
4829                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4830                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4831                         break;
4832                 }
4833         }
4834
4835         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4836                 unsigned int bit;
4837                 int negate;
4838
4839                 if ('a' <= optch && optch <= 'z') {
4840                         negate = 1;
4841                         optch = toupper(optch);
4842                 } else {
4843                         negate = 0;
4844                 }
4845
4846                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4847                 if (!bit)
4848                         return optarg[i];
4849                 if (negate)
4850                         opt->filter &= ~bit;
4851                 else
4852                         opt->filter |= bit;
4853         }
4854         return 0;
4855 }
4856
4857 static void enable_patch_output(int *fmt) {
4858         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4859         *fmt |= DIFF_FORMAT_PATCH;
4860 }
4861
4862 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4863 {
4864         int val = parse_ws_error_highlight(arg);
4865
4866         if (val < 0) {
4867                 error("unknown value after ws-error-highlight=%.*s",
4868                       -1 - val, arg);
4869                 return 0;
4870         }
4871         opt->ws_error_highlight = val;
4872         return 1;
4873 }
4874
4875 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4876 {
4877         struct object_id oid;
4878
4879         if (get_oid(arg, &oid))
4880                 return error("unable to resolve '%s'", arg);
4881
4882         if (!opt->objfind)
4883                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4884
4885         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4886         opt->flags.recursive = 1;
4887         opt->flags.tree_in_recursive = 1;
4888         oidset_insert(opt->objfind, &oid);
4889         return 1;
4890 }
4891
4892 int diff_opt_parse(struct diff_options *options,
4893                    const char **av, int ac, const char *prefix)
4894 {
4895         const char *arg = av[0];
4896         const char *optarg;
4897         int argcount;
4898
4899         if (!prefix)
4900                 prefix = "";
4901
4902         /* Output format options */
4903         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4904             || opt_arg(arg, 'U', "unified", &options->context))
4905                 enable_patch_output(&options->output_format);
4906         else if (!strcmp(arg, "--raw"))
4907                 options->output_format |= DIFF_FORMAT_RAW;
4908         else if (!strcmp(arg, "--patch-with-raw")) {
4909                 enable_patch_output(&options->output_format);
4910                 options->output_format |= DIFF_FORMAT_RAW;
4911         } else if (!strcmp(arg, "--numstat"))
4912                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4913         else if (!strcmp(arg, "--shortstat"))
4914                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4915         else if (skip_prefix(arg, "-X", &arg) ||
4916                  skip_to_optional_arg(arg, "--dirstat", &arg))
4917                 return parse_dirstat_opt(options, arg);
4918         else if (!strcmp(arg, "--cumulative"))
4919                 return parse_dirstat_opt(options, "cumulative");
4920         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4921                 parse_dirstat_opt(options, "files");
4922                 return parse_dirstat_opt(options, arg);
4923         }
4924         else if (!strcmp(arg, "--check"))
4925                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4926         else if (!strcmp(arg, "--summary"))
4927                 options->output_format |= DIFF_FORMAT_SUMMARY;
4928         else if (!strcmp(arg, "--patch-with-stat")) {
4929                 enable_patch_output(&options->output_format);
4930                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4931         } else if (!strcmp(arg, "--name-only"))
4932                 options->output_format |= DIFF_FORMAT_NAME;
4933         else if (!strcmp(arg, "--name-status"))
4934                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4935         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4936                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4937         else if (starts_with(arg, "--stat"))
4938                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4939                 return stat_opt(options, av);
4940         else if (!strcmp(arg, "--compact-summary")) {
4941                  options->flags.stat_with_summary = 1;
4942                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4943         } else if (!strcmp(arg, "--no-compact-summary"))
4944                  options->flags.stat_with_summary = 0;
4945         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4946                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4947         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4948                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4949         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4950                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4951
4952         /* renames options */
4953         else if (starts_with(arg, "-B") ||
4954                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4955                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4956                         return error("invalid argument to -B: %s", arg+2);
4957         }
4958         else if (starts_with(arg, "-M") ||
4959                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4960                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4961                         return error("invalid argument to -M: %s", arg+2);
4962                 options->detect_rename = DIFF_DETECT_RENAME;
4963         }
4964         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4965                 options->irreversible_delete = 1;
4966         }
4967         else if (starts_with(arg, "-C") ||
4968                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4969                 if (options->detect_rename == DIFF_DETECT_COPY)
4970                         options->flags.find_copies_harder = 1;
4971                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4972                         return error("invalid argument to -C: %s", arg+2);
4973                 options->detect_rename = DIFF_DETECT_COPY;
4974         }
4975         else if (!strcmp(arg, "--no-renames"))
4976                 options->detect_rename = 0;
4977         else if (!strcmp(arg, "--rename-empty"))
4978                 options->flags.rename_empty = 1;
4979         else if (!strcmp(arg, "--no-rename-empty"))
4980                 options->flags.rename_empty = 0;
4981         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4982                 options->flags.relative_name = 1;
4983                 if (arg)
4984                         options->prefix = arg;
4985         }
4986
4987         /* xdiff options */
4988         else if (!strcmp(arg, "--minimal"))
4989                 DIFF_XDL_SET(options, NEED_MINIMAL);
4990         else if (!strcmp(arg, "--no-minimal"))
4991                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4992         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4993                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4994         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4995                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4996         else if (!strcmp(arg, "--ignore-space-at-eol"))
4997                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4998         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4999                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
5000         else if (!strcmp(arg, "--ignore-blank-lines"))
5001                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
5002         else if (!strcmp(arg, "--indent-heuristic"))
5003                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5004         else if (!strcmp(arg, "--no-indent-heuristic"))
5005                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5006         else if (!strcmp(arg, "--patience")) {
5007                 int i;
5008                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5009                 /*
5010                  * Both --patience and --anchored use PATIENCE_DIFF
5011                  * internally, so remove any anchors previously
5012                  * specified.
5013                  */
5014                 for (i = 0; i < options->anchors_nr; i++)
5015                         free(options->anchors[i]);
5016                 options->anchors_nr = 0;
5017         } else if (!strcmp(arg, "--histogram"))
5018                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5019         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5020                 long value = parse_algorithm_value(optarg);
5021                 if (value < 0)
5022                         return error("option diff-algorithm accepts \"myers\", "
5023                                      "\"minimal\", \"patience\" and \"histogram\"");
5024                 /* clear out previous settings */
5025                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5026                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5027                 options->xdl_opts |= value;
5028                 return argcount;
5029         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5030                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5031                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5032                            options->anchors_alloc);
5033                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5034         }
5035
5036         /* flags options */
5037         else if (!strcmp(arg, "--binary")) {
5038                 enable_patch_output(&options->output_format);
5039                 options->flags.binary = 1;
5040         }
5041         else if (!strcmp(arg, "--full-index"))
5042                 options->flags.full_index = 1;
5043         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5044                 options->flags.text = 1;
5045         else if (!strcmp(arg, "-R"))
5046                 options->flags.reverse_diff = 1;
5047         else if (!strcmp(arg, "--find-copies-harder"))
5048                 options->flags.find_copies_harder = 1;
5049         else if (!strcmp(arg, "--follow"))
5050                 options->flags.follow_renames = 1;
5051         else if (!strcmp(arg, "--no-follow")) {
5052                 options->flags.follow_renames = 0;
5053                 options->flags.default_follow_renames = 0;
5054         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5055                 int value = git_config_colorbool(NULL, arg);
5056                 if (value < 0)
5057                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5058                 options->use_color = value;
5059         }
5060         else if (!strcmp(arg, "--no-color"))
5061                 options->use_color = 0;
5062         else if (!strcmp(arg, "--color-moved")) {
5063                 if (diff_color_moved_default)
5064                         options->color_moved = diff_color_moved_default;
5065                 if (options->color_moved == COLOR_MOVED_NO)
5066                         options->color_moved = COLOR_MOVED_DEFAULT;
5067         } else if (!strcmp(arg, "--no-color-moved"))
5068                 options->color_moved = COLOR_MOVED_NO;
5069         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5070                 int cm = parse_color_moved(arg);
5071                 if (cm < 0)
5072                         die("bad --color-moved argument: %s", arg);
5073                 options->color_moved = cm;
5074         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5075                 options->color_moved_ws_handling = 0;
5076         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5077                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5078         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5079                 options->use_color = 1;
5080                 options->word_diff = DIFF_WORDS_COLOR;
5081         }
5082         else if (!strcmp(arg, "--word-diff")) {
5083                 if (options->word_diff == DIFF_WORDS_NONE)
5084                         options->word_diff = DIFF_WORDS_PLAIN;
5085         }
5086         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5087                 if (!strcmp(arg, "plain"))
5088                         options->word_diff = DIFF_WORDS_PLAIN;
5089                 else if (!strcmp(arg, "color")) {
5090                         options->use_color = 1;
5091                         options->word_diff = DIFF_WORDS_COLOR;
5092                 }
5093                 else if (!strcmp(arg, "porcelain"))
5094                         options->word_diff = DIFF_WORDS_PORCELAIN;
5095                 else if (!strcmp(arg, "none"))
5096                         options->word_diff = DIFF_WORDS_NONE;
5097                 else
5098                         die("bad --word-diff argument: %s", arg);
5099         }
5100         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5101                 if (options->word_diff == DIFF_WORDS_NONE)
5102                         options->word_diff = DIFF_WORDS_PLAIN;
5103                 options->word_regex = optarg;
5104                 return argcount;
5105         }
5106         else if (!strcmp(arg, "--exit-code"))
5107                 options->flags.exit_with_status = 1;
5108         else if (!strcmp(arg, "--quiet"))
5109                 options->flags.quick = 1;
5110         else if (!strcmp(arg, "--ext-diff"))
5111                 options->flags.allow_external = 1;
5112         else if (!strcmp(arg, "--no-ext-diff"))
5113                 options->flags.allow_external = 0;
5114         else if (!strcmp(arg, "--textconv")) {
5115                 options->flags.allow_textconv = 1;
5116                 options->flags.textconv_set_via_cmdline = 1;
5117         } else if (!strcmp(arg, "--no-textconv"))
5118                 options->flags.allow_textconv = 0;
5119         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5120                 options->flags.override_submodule_config = 1;
5121                 handle_ignore_submodules_arg(options, arg);
5122         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5123                 return parse_submodule_opt(options, arg);
5124         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5125                 return parse_ws_error_highlight_opt(options, arg);
5126         else if (!strcmp(arg, "--ita-invisible-in-index"))
5127                 options->ita_invisible_in_index = 1;
5128         else if (!strcmp(arg, "--ita-visible-in-index"))
5129                 options->ita_invisible_in_index = 0;
5130
5131         /* misc options */
5132         else if (!strcmp(arg, "-z"))
5133                 options->line_termination = 0;
5134         else if ((argcount = short_opt('l', av, &optarg))) {
5135                 options->rename_limit = strtoul(optarg, NULL, 10);
5136                 return argcount;
5137         }
5138         else if ((argcount = short_opt('S', av, &optarg))) {
5139                 options->pickaxe = optarg;
5140                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5141                 return argcount;
5142         } else if ((argcount = short_opt('G', av, &optarg))) {
5143                 options->pickaxe = optarg;
5144                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5145                 return argcount;
5146         }
5147         else if (!strcmp(arg, "--pickaxe-all"))
5148                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5149         else if (!strcmp(arg, "--pickaxe-regex"))
5150                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5151         else if ((argcount = short_opt('O', av, &optarg))) {
5152                 options->orderfile = prefix_filename(prefix, optarg);
5153                 return argcount;
5154         } else if (skip_prefix(arg, "--find-object=", &arg))
5155                 return parse_objfind_opt(options, arg);
5156         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5157                 int offending = parse_diff_filter_opt(optarg, options);
5158                 if (offending)
5159                         die("unknown change class '%c' in --diff-filter=%s",
5160                             offending, optarg);
5161                 return argcount;
5162         }
5163         else if (!strcmp(arg, "--no-abbrev"))
5164                 options->abbrev = 0;
5165         else if (!strcmp(arg, "--abbrev"))
5166                 options->abbrev = DEFAULT_ABBREV;
5167         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5168                 options->abbrev = strtoul(arg, NULL, 10);
5169                 if (options->abbrev < MINIMUM_ABBREV)
5170                         options->abbrev = MINIMUM_ABBREV;
5171                 else if (the_hash_algo->hexsz < options->abbrev)
5172                         options->abbrev = the_hash_algo->hexsz;
5173         }
5174         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5175                 options->a_prefix = optarg;
5176                 return argcount;
5177         }
5178         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5179                 options->line_prefix = optarg;
5180                 options->line_prefix_length = strlen(options->line_prefix);
5181                 graph_setup_line_prefix(options);
5182                 return argcount;
5183         }
5184         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5185                 options->b_prefix = optarg;
5186                 return argcount;
5187         }
5188         else if (!strcmp(arg, "--no-prefix"))
5189                 options->a_prefix = options->b_prefix = "";
5190         else if (opt_arg(arg, '\0', "inter-hunk-context",
5191                          &options->interhunkcontext))
5192                 ;
5193         else if (!strcmp(arg, "-W"))
5194                 options->flags.funccontext = 1;
5195         else if (!strcmp(arg, "--function-context"))
5196                 options->flags.funccontext = 1;
5197         else if (!strcmp(arg, "--no-function-context"))
5198                 options->flags.funccontext = 0;
5199         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5200                 char *path = prefix_filename(prefix, optarg);
5201                 options->file = xfopen(path, "w");
5202                 options->close_file = 1;
5203                 if (options->use_color != GIT_COLOR_ALWAYS)
5204                         options->use_color = GIT_COLOR_NEVER;
5205                 free(path);
5206                 return argcount;
5207         } else
5208                 return 0;
5209         return 1;
5210 }
5211
5212 int parse_rename_score(const char **cp_p)
5213 {
5214         unsigned long num, scale;
5215         int ch, dot;
5216         const char *cp = *cp_p;
5217
5218         num = 0;
5219         scale = 1;
5220         dot = 0;
5221         for (;;) {
5222                 ch = *cp;
5223                 if ( !dot && ch == '.' ) {
5224                         scale = 1;
5225                         dot = 1;
5226                 } else if ( ch == '%' ) {
5227                         scale = dot ? scale*100 : 100;
5228                         cp++;   /* % is always at the end */
5229                         break;
5230                 } else if ( ch >= '0' && ch <= '9' ) {
5231                         if ( scale < 100000 ) {
5232                                 scale *= 10;
5233                                 num = (num*10) + (ch-'0');
5234                         }
5235                 } else {
5236                         break;
5237                 }
5238                 cp++;
5239         }
5240         *cp_p = cp;
5241
5242         /* user says num divided by scale and we say internally that
5243          * is MAX_SCORE * num / scale.
5244          */
5245         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5246 }
5247
5248 static int diff_scoreopt_parse(const char *opt)
5249 {
5250         int opt1, opt2, cmd;
5251
5252         if (*opt++ != '-')
5253                 return -1;
5254         cmd = *opt++;
5255         if (cmd == '-') {
5256                 /* convert the long-form arguments into short-form versions */
5257                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5258                         if (*opt == 0 || *opt++ == '=')
5259                                 cmd = 'B';
5260                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5261                         if (*opt == 0 || *opt++ == '=')
5262                                 cmd = 'C';
5263                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5264                         if (*opt == 0 || *opt++ == '=')
5265                                 cmd = 'M';
5266                 }
5267         }
5268         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5269                 return -1; /* that is not a -M, -C, or -B option */
5270
5271         opt1 = parse_rename_score(&opt);
5272         if (cmd != 'B')
5273                 opt2 = 0;
5274         else {
5275                 if (*opt == 0)
5276                         opt2 = 0;
5277                 else if (*opt != '/')
5278                         return -1; /* we expect -B80/99 or -B80 */
5279                 else {
5280                         opt++;
5281                         opt2 = parse_rename_score(&opt);
5282                 }
5283         }
5284         if (*opt != 0)
5285                 return -1;
5286         return opt1 | (opt2 << 16);
5287 }
5288
5289 struct diff_queue_struct diff_queued_diff;
5290
5291 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5292 {
5293         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5294         queue->queue[queue->nr++] = dp;
5295 }
5296
5297 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5298                                  struct diff_filespec *one,
5299                                  struct diff_filespec *two)
5300 {
5301         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5302         dp->one = one;
5303         dp->two = two;
5304         if (queue)
5305                 diff_q(queue, dp);
5306         return dp;
5307 }
5308
5309 void diff_free_filepair(struct diff_filepair *p)
5310 {
5311         free_filespec(p->one);
5312         free_filespec(p->two);
5313         free(p);
5314 }
5315
5316 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5317 {
5318         int abblen;
5319         const char *abbrev;
5320
5321         /* Do we want all 40 hex characters? */
5322         if (len == the_hash_algo->hexsz)
5323                 return oid_to_hex(oid);
5324
5325         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5326         abbrev = diff_abbrev_oid(oid, len);
5327
5328         if (!print_sha1_ellipsis())
5329                 return abbrev;
5330
5331         abblen = strlen(abbrev);
5332
5333         /*
5334          * In well-behaved cases, where the abbreviated result is the
5335          * same as the requested length, append three dots after the
5336          * abbreviation (hence the whole logic is limited to the case
5337          * where abblen < 37); when the actual abbreviated result is a
5338          * bit longer than the requested length, we reduce the number
5339          * of dots so that they match the well-behaved ones.  However,
5340          * if the actual abbreviation is longer than the requested
5341          * length by more than three, we give up on aligning, and add
5342          * three dots anyway, to indicate that the output is not the
5343          * full object name.  Yes, this may be suboptimal, but this
5344          * appears only in "diff --raw --abbrev" output and it is not
5345          * worth the effort to change it now.  Note that this would
5346          * likely to work fine when the automatic sizing of default
5347          * abbreviation length is used--we would be fed -1 in "len" in
5348          * that case, and will end up always appending three-dots, but
5349          * the automatic sizing is supposed to give abblen that ensures
5350          * uniqueness across all objects (statistically speaking).
5351          */
5352         if (abblen < the_hash_algo->hexsz - 3) {
5353                 static char hex[GIT_MAX_HEXSZ + 1];
5354                 if (len < abblen && abblen <= len + 2)
5355                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5356                 else
5357                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5358                 return hex;
5359         }
5360
5361         return oid_to_hex(oid);
5362 }
5363
5364 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5365 {
5366         int line_termination = opt->line_termination;
5367         int inter_name_termination = line_termination ? '\t' : '\0';
5368
5369         fprintf(opt->file, "%s", diff_line_prefix(opt));
5370         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5371                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5372                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5373                 fprintf(opt->file, "%s ",
5374                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5375         }
5376         if (p->score) {
5377                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5378                         inter_name_termination);
5379         } else {
5380                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5381         }
5382
5383         if (p->status == DIFF_STATUS_COPIED ||
5384             p->status == DIFF_STATUS_RENAMED) {
5385                 const char *name_a, *name_b;
5386                 name_a = p->one->path;
5387                 name_b = p->two->path;
5388                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5389                 write_name_quoted(name_a, opt->file, inter_name_termination);
5390                 write_name_quoted(name_b, opt->file, line_termination);
5391         } else {
5392                 const char *name_a, *name_b;
5393                 name_a = p->one->mode ? p->one->path : p->two->path;
5394                 name_b = NULL;
5395                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5396                 write_name_quoted(name_a, opt->file, line_termination);
5397         }
5398 }
5399
5400 int diff_unmodified_pair(struct diff_filepair *p)
5401 {
5402         /* This function is written stricter than necessary to support
5403          * the currently implemented transformers, but the idea is to
5404          * let transformers to produce diff_filepairs any way they want,
5405          * and filter and clean them up here before producing the output.
5406          */
5407         struct diff_filespec *one = p->one, *two = p->two;
5408
5409         if (DIFF_PAIR_UNMERGED(p))
5410                 return 0; /* unmerged is interesting */
5411
5412         /* deletion, addition, mode or type change
5413          * and rename are all interesting.
5414          */
5415         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5416             DIFF_PAIR_MODE_CHANGED(p) ||
5417             strcmp(one->path, two->path))
5418                 return 0;
5419
5420         /* both are valid and point at the same path.  that is, we are
5421          * dealing with a change.
5422          */
5423         if (one->oid_valid && two->oid_valid &&
5424             oideq(&one->oid, &two->oid) &&
5425             !one->dirty_submodule && !two->dirty_submodule)
5426                 return 1; /* no change */
5427         if (!one->oid_valid && !two->oid_valid)
5428                 return 1; /* both look at the same file on the filesystem. */
5429         return 0;
5430 }
5431
5432 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5433 {
5434         if (diff_unmodified_pair(p))
5435                 return;
5436
5437         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5438             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5439                 return; /* no tree diffs in patch format */
5440
5441         run_diff(p, o);
5442 }
5443
5444 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5445                             struct diffstat_t *diffstat)
5446 {
5447         if (diff_unmodified_pair(p))
5448                 return;
5449
5450         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5451             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5452                 return; /* no useful stat for tree diffs */
5453
5454         run_diffstat(p, o, diffstat);
5455 }
5456
5457 static void diff_flush_checkdiff(struct diff_filepair *p,
5458                 struct diff_options *o)
5459 {
5460         if (diff_unmodified_pair(p))
5461                 return;
5462
5463         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5464             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5465                 return; /* nothing to check in tree diffs */
5466
5467         run_checkdiff(p, o);
5468 }
5469
5470 int diff_queue_is_empty(void)
5471 {
5472         struct diff_queue_struct *q = &diff_queued_diff;
5473         int i;
5474         for (i = 0; i < q->nr; i++)
5475                 if (!diff_unmodified_pair(q->queue[i]))
5476                         return 0;
5477         return 1;
5478 }
5479
5480 #if DIFF_DEBUG
5481 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5482 {
5483         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5484                 x, one ? one : "",
5485                 s->path,
5486                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5487                 s->mode,
5488                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5489         fprintf(stderr, "queue[%d] %s size %lu\n",
5490                 x, one ? one : "",
5491                 s->size);
5492 }
5493
5494 void diff_debug_filepair(const struct diff_filepair *p, int i)
5495 {
5496         diff_debug_filespec(p->one, i, "one");
5497         diff_debug_filespec(p->two, i, "two");
5498         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5499                 p->score, p->status ? p->status : '?',
5500                 p->one->rename_used, p->broken_pair);
5501 }
5502
5503 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5504 {
5505         int i;
5506         if (msg)
5507                 fprintf(stderr, "%s\n", msg);
5508         fprintf(stderr, "q->nr = %d\n", q->nr);
5509         for (i = 0; i < q->nr; i++) {
5510                 struct diff_filepair *p = q->queue[i];
5511                 diff_debug_filepair(p, i);
5512         }
5513 }
5514 #endif
5515
5516 static void diff_resolve_rename_copy(void)
5517 {
5518         int i;
5519         struct diff_filepair *p;
5520         struct diff_queue_struct *q = &diff_queued_diff;
5521
5522         diff_debug_queue("resolve-rename-copy", q);
5523
5524         for (i = 0; i < q->nr; i++) {
5525                 p = q->queue[i];
5526                 p->status = 0; /* undecided */
5527                 if (DIFF_PAIR_UNMERGED(p))
5528                         p->status = DIFF_STATUS_UNMERGED;
5529                 else if (!DIFF_FILE_VALID(p->one))
5530                         p->status = DIFF_STATUS_ADDED;
5531                 else if (!DIFF_FILE_VALID(p->two))
5532                         p->status = DIFF_STATUS_DELETED;
5533                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5534                         p->status = DIFF_STATUS_TYPE_CHANGED;
5535
5536                 /* from this point on, we are dealing with a pair
5537                  * whose both sides are valid and of the same type, i.e.
5538                  * either in-place edit or rename/copy edit.
5539                  */
5540                 else if (DIFF_PAIR_RENAME(p)) {
5541                         /*
5542                          * A rename might have re-connected a broken
5543                          * pair up, causing the pathnames to be the
5544                          * same again. If so, that's not a rename at
5545                          * all, just a modification..
5546                          *
5547                          * Otherwise, see if this source was used for
5548                          * multiple renames, in which case we decrement
5549                          * the count, and call it a copy.
5550                          */
5551                         if (!strcmp(p->one->path, p->two->path))
5552                                 p->status = DIFF_STATUS_MODIFIED;
5553                         else if (--p->one->rename_used > 0)
5554                                 p->status = DIFF_STATUS_COPIED;
5555                         else
5556                                 p->status = DIFF_STATUS_RENAMED;
5557                 }
5558                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5559                          p->one->mode != p->two->mode ||
5560                          p->one->dirty_submodule ||
5561                          p->two->dirty_submodule ||
5562                          is_null_oid(&p->one->oid))
5563                         p->status = DIFF_STATUS_MODIFIED;
5564                 else {
5565                         /* This is a "no-change" entry and should not
5566                          * happen anymore, but prepare for broken callers.
5567                          */
5568                         error("feeding unmodified %s to diffcore",
5569                               p->one->path);
5570                         p->status = DIFF_STATUS_UNKNOWN;
5571                 }
5572         }
5573         diff_debug_queue("resolve-rename-copy done", q);
5574 }
5575
5576 static int check_pair_status(struct diff_filepair *p)
5577 {
5578         switch (p->status) {
5579         case DIFF_STATUS_UNKNOWN:
5580                 return 0;
5581         case 0:
5582                 die("internal error in diff-resolve-rename-copy");
5583         default:
5584                 return 1;
5585         }
5586 }
5587
5588 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5589 {
5590         int fmt = opt->output_format;
5591
5592         if (fmt & DIFF_FORMAT_CHECKDIFF)
5593                 diff_flush_checkdiff(p, opt);
5594         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5595                 diff_flush_raw(p, opt);
5596         else if (fmt & DIFF_FORMAT_NAME) {
5597                 const char *name_a, *name_b;
5598                 name_a = p->two->path;
5599                 name_b = NULL;
5600                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5601                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5602                 write_name_quoted(name_a, opt->file, opt->line_termination);
5603         }
5604 }
5605
5606 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5607 {
5608         struct strbuf sb = STRBUF_INIT;
5609         if (fs->mode)
5610                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5611         else
5612                 strbuf_addf(&sb, " %s ", newdelete);
5613
5614         quote_c_style(fs->path, &sb, NULL, 0);
5615         strbuf_addch(&sb, '\n');
5616         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5617                          sb.buf, sb.len, 0);
5618         strbuf_release(&sb);
5619 }
5620
5621 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5622                 int show_name)
5623 {
5624         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5625                 struct strbuf sb = STRBUF_INIT;
5626                 strbuf_addf(&sb, " mode change %06o => %06o",
5627                             p->one->mode, p->two->mode);
5628                 if (show_name) {
5629                         strbuf_addch(&sb, ' ');
5630                         quote_c_style(p->two->path, &sb, NULL, 0);
5631                 }
5632                 strbuf_addch(&sb, '\n');
5633                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5634                                  sb.buf, sb.len, 0);
5635                 strbuf_release(&sb);
5636         }
5637 }
5638
5639 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5640                 struct diff_filepair *p)
5641 {
5642         struct strbuf sb = STRBUF_INIT;
5643         struct strbuf names = STRBUF_INIT;
5644
5645         pprint_rename(&names, p->one->path, p->two->path);
5646         strbuf_addf(&sb, " %s %s (%d%%)\n",
5647                     renamecopy, names.buf, similarity_index(p));
5648         strbuf_release(&names);
5649         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5650                                  sb.buf, sb.len, 0);
5651         show_mode_change(opt, p, 0);
5652         strbuf_release(&sb);
5653 }
5654
5655 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5656 {
5657         switch(p->status) {
5658         case DIFF_STATUS_DELETED:
5659                 show_file_mode_name(opt, "delete", p->one);
5660                 break;
5661         case DIFF_STATUS_ADDED:
5662                 show_file_mode_name(opt, "create", p->two);
5663                 break;
5664         case DIFF_STATUS_COPIED:
5665                 show_rename_copy(opt, "copy", p);
5666                 break;
5667         case DIFF_STATUS_RENAMED:
5668                 show_rename_copy(opt, "rename", p);
5669                 break;
5670         default:
5671                 if (p->score) {
5672                         struct strbuf sb = STRBUF_INIT;
5673                         strbuf_addstr(&sb, " rewrite ");
5674                         quote_c_style(p->two->path, &sb, NULL, 0);
5675                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5676                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5677                                          sb.buf, sb.len, 0);
5678                         strbuf_release(&sb);
5679                 }
5680                 show_mode_change(opt, p, !p->score);
5681                 break;
5682         }
5683 }
5684
5685 struct patch_id_t {
5686         git_SHA_CTX *ctx;
5687         int patchlen;
5688 };
5689
5690 static int remove_space(char *line, int len)
5691 {
5692         int i;
5693         char *dst = line;
5694         unsigned char c;
5695
5696         for (i = 0; i < len; i++)
5697                 if (!isspace((c = line[i])))
5698                         *dst++ = c;
5699
5700         return dst - line;
5701 }
5702
5703 static void patch_id_consume(void *priv, char *line, unsigned long len)
5704 {
5705         struct patch_id_t *data = priv;
5706         int new_len;
5707
5708         new_len = remove_space(line, len);
5709
5710         git_SHA1_Update(data->ctx, line, new_len);
5711         data->patchlen += new_len;
5712 }
5713
5714 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5715 {
5716         git_SHA1_Update(ctx, str, strlen(str));
5717 }
5718
5719 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5720 {
5721         /* large enough for 2^32 in octal */
5722         char buf[12];
5723         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5724         git_SHA1_Update(ctx, buf, len);
5725 }
5726
5727 /* returns 0 upon success, and writes result into sha1 */
5728 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5729 {
5730         struct diff_queue_struct *q = &diff_queued_diff;
5731         int i;
5732         git_SHA_CTX ctx;
5733         struct patch_id_t data;
5734
5735         git_SHA1_Init(&ctx);
5736         memset(&data, 0, sizeof(struct patch_id_t));
5737         data.ctx = &ctx;
5738
5739         for (i = 0; i < q->nr; i++) {
5740                 xpparam_t xpp;
5741                 xdemitconf_t xecfg;
5742                 mmfile_t mf1, mf2;
5743                 struct diff_filepair *p = q->queue[i];
5744                 int len1, len2;
5745
5746                 memset(&xpp, 0, sizeof(xpp));
5747                 memset(&xecfg, 0, sizeof(xecfg));
5748                 if (p->status == 0)
5749                         return error("internal diff status error");
5750                 if (p->status == DIFF_STATUS_UNKNOWN)
5751                         continue;
5752                 if (diff_unmodified_pair(p))
5753                         continue;
5754                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5755                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5756                         continue;
5757                 if (DIFF_PAIR_UNMERGED(p))
5758                         continue;
5759
5760                 diff_fill_oid_info(p->one, options->repo->index);
5761                 diff_fill_oid_info(p->two, options->repo->index);
5762
5763                 len1 = remove_space(p->one->path, strlen(p->one->path));
5764                 len2 = remove_space(p->two->path, strlen(p->two->path));
5765                 patch_id_add_string(&ctx, "diff--git");
5766                 patch_id_add_string(&ctx, "a/");
5767                 git_SHA1_Update(&ctx, p->one->path, len1);
5768                 patch_id_add_string(&ctx, "b/");
5769                 git_SHA1_Update(&ctx, p->two->path, len2);
5770
5771                 if (p->one->mode == 0) {
5772                         patch_id_add_string(&ctx, "newfilemode");
5773                         patch_id_add_mode(&ctx, p->two->mode);
5774                         patch_id_add_string(&ctx, "---/dev/null");
5775                         patch_id_add_string(&ctx, "+++b/");
5776                         git_SHA1_Update(&ctx, p->two->path, len2);
5777                 } else if (p->two->mode == 0) {
5778                         patch_id_add_string(&ctx, "deletedfilemode");
5779                         patch_id_add_mode(&ctx, p->one->mode);
5780                         patch_id_add_string(&ctx, "---a/");
5781                         git_SHA1_Update(&ctx, p->one->path, len1);
5782                         patch_id_add_string(&ctx, "+++/dev/null");
5783                 } else {
5784                         patch_id_add_string(&ctx, "---a/");
5785                         git_SHA1_Update(&ctx, p->one->path, len1);
5786                         patch_id_add_string(&ctx, "+++b/");
5787                         git_SHA1_Update(&ctx, p->two->path, len2);
5788                 }
5789
5790                 if (diff_header_only)
5791                         continue;
5792
5793                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5794                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5795                         return error("unable to read files to diff");
5796
5797                 if (diff_filespec_is_binary(options->repo, p->one) ||
5798                     diff_filespec_is_binary(options->repo, p->two)) {
5799                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5800                                         GIT_SHA1_HEXSZ);
5801                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5802                                         GIT_SHA1_HEXSZ);
5803                         continue;
5804                 }
5805
5806                 xpp.flags = 0;
5807                 xecfg.ctxlen = 3;
5808                 xecfg.flags = 0;
5809                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5810                                   patch_id_consume, &data, &xpp, &xecfg))
5811                         return error("unable to generate patch-id diff for %s",
5812                                      p->one->path);
5813         }
5814
5815         git_SHA1_Final(oid->hash, &ctx);
5816         return 0;
5817 }
5818
5819 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5820 {
5821         struct diff_queue_struct *q = &diff_queued_diff;
5822         int i;
5823         int result = diff_get_patch_id(options, oid, diff_header_only);
5824
5825         for (i = 0; i < q->nr; i++)
5826                 diff_free_filepair(q->queue[i]);
5827
5828         free(q->queue);
5829         DIFF_QUEUE_CLEAR(q);
5830
5831         return result;
5832 }
5833
5834 static int is_summary_empty(const struct diff_queue_struct *q)
5835 {
5836         int i;
5837
5838         for (i = 0; i < q->nr; i++) {
5839                 const struct diff_filepair *p = q->queue[i];
5840
5841                 switch (p->status) {
5842                 case DIFF_STATUS_DELETED:
5843                 case DIFF_STATUS_ADDED:
5844                 case DIFF_STATUS_COPIED:
5845                 case DIFF_STATUS_RENAMED:
5846                         return 0;
5847                 default:
5848                         if (p->score)
5849                                 return 0;
5850                         if (p->one->mode && p->two->mode &&
5851                             p->one->mode != p->two->mode)
5852                                 return 0;
5853                         break;
5854                 }
5855         }
5856         return 1;
5857 }
5858
5859 static const char rename_limit_warning[] =
5860 N_("inexact rename detection was skipped due to too many files.");
5861
5862 static const char degrade_cc_to_c_warning[] =
5863 N_("only found copies from modified paths due to too many files.");
5864
5865 static const char rename_limit_advice[] =
5866 N_("you may want to set your %s variable to at least "
5867    "%d and retry the command.");
5868
5869 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5870 {
5871         fflush(stdout);
5872         if (degraded_cc)
5873                 warning(_(degrade_cc_to_c_warning));
5874         else if (needed)
5875                 warning(_(rename_limit_warning));
5876         else
5877                 return;
5878         if (0 < needed)
5879                 warning(_(rename_limit_advice), varname, needed);
5880 }
5881
5882 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5883 {
5884         int i;
5885         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5886         struct diff_queue_struct *q = &diff_queued_diff;
5887
5888         if (WSEH_NEW & WS_RULE_MASK)
5889                 BUG("WS rules bit mask overlaps with diff symbol flags");
5890
5891         if (o->color_moved)
5892                 o->emitted_symbols = &esm;
5893
5894         for (i = 0; i < q->nr; i++) {
5895                 struct diff_filepair *p = q->queue[i];
5896                 if (check_pair_status(p))
5897                         diff_flush_patch(p, o);
5898         }
5899
5900         if (o->emitted_symbols) {
5901                 if (o->color_moved) {
5902                         struct hashmap add_lines, del_lines;
5903
5904                         if (o->color_moved_ws_handling &
5905                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5906                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5907
5908                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5909                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5910
5911                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5912                         mark_color_as_moved(o, &add_lines, &del_lines);
5913                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5914                                 dim_moved_lines(o);
5915
5916                         hashmap_free(&add_lines, 1);
5917                         hashmap_free(&del_lines, 1);
5918                 }
5919
5920                 for (i = 0; i < esm.nr; i++)
5921                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5922
5923                 for (i = 0; i < esm.nr; i++)
5924                         free((void *)esm.buf[i].line);
5925         }
5926         esm.nr = 0;
5927 }
5928
5929 void diff_flush(struct diff_options *options)
5930 {
5931         struct diff_queue_struct *q = &diff_queued_diff;
5932         int i, output_format = options->output_format;
5933         int separator = 0;
5934         int dirstat_by_line = 0;
5935
5936         /*
5937          * Order: raw, stat, summary, patch
5938          * or:    name/name-status/checkdiff (other bits clear)
5939          */
5940         if (!q->nr)
5941                 goto free_queue;
5942
5943         if (output_format & (DIFF_FORMAT_RAW |
5944                              DIFF_FORMAT_NAME |
5945                              DIFF_FORMAT_NAME_STATUS |
5946                              DIFF_FORMAT_CHECKDIFF)) {
5947                 for (i = 0; i < q->nr; i++) {
5948                         struct diff_filepair *p = q->queue[i];
5949                         if (check_pair_status(p))
5950                                 flush_one_pair(p, options);
5951                 }
5952                 separator++;
5953         }
5954
5955         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5956                 dirstat_by_line = 1;
5957
5958         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5959             dirstat_by_line) {
5960                 struct diffstat_t diffstat;
5961
5962                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5963                 for (i = 0; i < q->nr; i++) {
5964                         struct diff_filepair *p = q->queue[i];
5965                         if (check_pair_status(p))
5966                                 diff_flush_stat(p, options, &diffstat);
5967                 }
5968                 if (output_format & DIFF_FORMAT_NUMSTAT)
5969                         show_numstat(&diffstat, options);
5970                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5971                         show_stats(&diffstat, options);
5972                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5973                         show_shortstats(&diffstat, options);
5974                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5975                         show_dirstat_by_line(&diffstat, options);
5976                 free_diffstat_info(&diffstat);
5977                 separator++;
5978         }
5979         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5980                 show_dirstat(options);
5981
5982         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5983                 for (i = 0; i < q->nr; i++) {
5984                         diff_summary(options, q->queue[i]);
5985                 }
5986                 separator++;
5987         }
5988
5989         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5990             options->flags.exit_with_status &&
5991             options->flags.diff_from_contents) {
5992                 /*
5993                  * run diff_flush_patch for the exit status. setting
5994                  * options->file to /dev/null should be safe, because we
5995                  * aren't supposed to produce any output anyway.
5996                  */
5997                 if (options->close_file)
5998                         fclose(options->file);
5999                 options->file = xfopen("/dev/null", "w");
6000                 options->close_file = 1;
6001                 options->color_moved = 0;
6002                 for (i = 0; i < q->nr; i++) {
6003                         struct diff_filepair *p = q->queue[i];
6004                         if (check_pair_status(p))
6005                                 diff_flush_patch(p, options);
6006                         if (options->found_changes)
6007                                 break;
6008                 }
6009         }
6010
6011         if (output_format & DIFF_FORMAT_PATCH) {
6012                 if (separator) {
6013                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6014                         if (options->stat_sep)
6015                                 /* attach patch instead of inline */
6016                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6017                                                  NULL, 0, 0);
6018                 }
6019
6020                 diff_flush_patch_all_file_pairs(options);
6021         }
6022
6023         if (output_format & DIFF_FORMAT_CALLBACK)
6024                 options->format_callback(q, options, options->format_callback_data);
6025
6026         for (i = 0; i < q->nr; i++)
6027                 diff_free_filepair(q->queue[i]);
6028 free_queue:
6029         free(q->queue);
6030         DIFF_QUEUE_CLEAR(q);
6031         if (options->close_file)
6032                 fclose(options->file);
6033
6034         /*
6035          * Report the content-level differences with HAS_CHANGES;
6036          * diff_addremove/diff_change does not set the bit when
6037          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6038          */
6039         if (options->flags.diff_from_contents) {
6040                 if (options->found_changes)
6041                         options->flags.has_changes = 1;
6042                 else
6043                         options->flags.has_changes = 0;
6044         }
6045 }
6046
6047 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6048 {
6049         return (((p->status == DIFF_STATUS_MODIFIED) &&
6050                  ((p->score &&
6051                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6052                   (!p->score &&
6053                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6054                 ((p->status != DIFF_STATUS_MODIFIED) &&
6055                  filter_bit_tst(p->status, options)));
6056 }
6057
6058 static void diffcore_apply_filter(struct diff_options *options)
6059 {
6060         int i;
6061         struct diff_queue_struct *q = &diff_queued_diff;
6062         struct diff_queue_struct outq;
6063
6064         DIFF_QUEUE_CLEAR(&outq);
6065
6066         if (!options->filter)
6067                 return;
6068
6069         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6070                 int found;
6071                 for (i = found = 0; !found && i < q->nr; i++) {
6072                         if (match_filter(options, q->queue[i]))
6073                                 found++;
6074                 }
6075                 if (found)
6076                         return;
6077
6078                 /* otherwise we will clear the whole queue
6079                  * by copying the empty outq at the end of this
6080                  * function, but first clear the current entries
6081                  * in the queue.
6082                  */
6083                 for (i = 0; i < q->nr; i++)
6084                         diff_free_filepair(q->queue[i]);
6085         }
6086         else {
6087                 /* Only the matching ones */
6088                 for (i = 0; i < q->nr; i++) {
6089                         struct diff_filepair *p = q->queue[i];
6090                         if (match_filter(options, p))
6091                                 diff_q(&outq, p);
6092                         else
6093                                 diff_free_filepair(p);
6094                 }
6095         }
6096         free(q->queue);
6097         *q = outq;
6098 }
6099
6100 /* Check whether two filespecs with the same mode and size are identical */
6101 static int diff_filespec_is_identical(struct repository *r,
6102                                       struct diff_filespec *one,
6103                                       struct diff_filespec *two)
6104 {
6105         if (S_ISGITLINK(one->mode))
6106                 return 0;
6107         if (diff_populate_filespec(r, one, 0))
6108                 return 0;
6109         if (diff_populate_filespec(r, two, 0))
6110                 return 0;
6111         return !memcmp(one->data, two->data, one->size);
6112 }
6113
6114 static int diff_filespec_check_stat_unmatch(struct repository *r,
6115                                             struct diff_filepair *p)
6116 {
6117         if (p->done_skip_stat_unmatch)
6118                 return p->skip_stat_unmatch_result;
6119
6120         p->done_skip_stat_unmatch = 1;
6121         p->skip_stat_unmatch_result = 0;
6122         /*
6123          * 1. Entries that come from stat info dirtiness
6124          *    always have both sides (iow, not create/delete),
6125          *    one side of the object name is unknown, with
6126          *    the same mode and size.  Keep the ones that
6127          *    do not match these criteria.  They have real
6128          *    differences.
6129          *
6130          * 2. At this point, the file is known to be modified,
6131          *    with the same mode and size, and the object
6132          *    name of one side is unknown.  Need to inspect
6133          *    the identical contents.
6134          */
6135         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6136             !DIFF_FILE_VALID(p->two) ||
6137             (p->one->oid_valid && p->two->oid_valid) ||
6138             (p->one->mode != p->two->mode) ||
6139             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6140             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6141             (p->one->size != p->two->size) ||
6142             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6143                 p->skip_stat_unmatch_result = 1;
6144         return p->skip_stat_unmatch_result;
6145 }
6146
6147 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6148 {
6149         int i;
6150         struct diff_queue_struct *q = &diff_queued_diff;
6151         struct diff_queue_struct outq;
6152         DIFF_QUEUE_CLEAR(&outq);
6153
6154         for (i = 0; i < q->nr; i++) {
6155                 struct diff_filepair *p = q->queue[i];
6156
6157                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6158                         diff_q(&outq, p);
6159                 else {
6160                         /*
6161                          * The caller can subtract 1 from skip_stat_unmatch
6162                          * to determine how many paths were dirty only
6163                          * due to stat info mismatch.
6164                          */
6165                         if (!diffopt->flags.no_index)
6166                                 diffopt->skip_stat_unmatch++;
6167                         diff_free_filepair(p);
6168                 }
6169         }
6170         free(q->queue);
6171         *q = outq;
6172 }
6173
6174 static int diffnamecmp(const void *a_, const void *b_)
6175 {
6176         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6177         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6178         const char *name_a, *name_b;
6179
6180         name_a = a->one ? a->one->path : a->two->path;
6181         name_b = b->one ? b->one->path : b->two->path;
6182         return strcmp(name_a, name_b);
6183 }
6184
6185 void diffcore_fix_diff_index(struct diff_options *options)
6186 {
6187         struct diff_queue_struct *q = &diff_queued_diff;
6188         QSORT(q->queue, q->nr, diffnamecmp);
6189 }
6190
6191 void diffcore_std(struct diff_options *options)
6192 {
6193         /* NOTE please keep the following in sync with diff_tree_combined() */
6194         if (options->skip_stat_unmatch)
6195                 diffcore_skip_stat_unmatch(options);
6196         if (!options->found_follow) {
6197                 /* See try_to_follow_renames() in tree-diff.c */
6198                 if (options->break_opt != -1)
6199                         diffcore_break(options->repo,
6200                                        options->break_opt);
6201                 if (options->detect_rename)
6202                         diffcore_rename(options);
6203                 if (options->break_opt != -1)
6204                         diffcore_merge_broken();
6205         }
6206         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6207                 diffcore_pickaxe(options);
6208         if (options->orderfile)
6209                 diffcore_order(options->orderfile);
6210         if (!options->found_follow)
6211                 /* See try_to_follow_renames() in tree-diff.c */
6212                 diff_resolve_rename_copy();
6213         diffcore_apply_filter(options);
6214
6215         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6216                 options->flags.has_changes = 1;
6217         else
6218                 options->flags.has_changes = 0;
6219
6220         options->found_follow = 0;
6221 }
6222
6223 int diff_result_code(struct diff_options *opt, int status)
6224 {
6225         int result = 0;
6226
6227         diff_warn_rename_limit("diff.renameLimit",
6228                                opt->needed_rename_limit,
6229                                opt->degraded_cc_to_c);
6230         if (!opt->flags.exit_with_status &&
6231             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6232                 return status;
6233         if (opt->flags.exit_with_status &&
6234             opt->flags.has_changes)
6235                 result |= 01;
6236         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6237             opt->flags.check_failed)
6238                 result |= 02;
6239         return result;
6240 }
6241
6242 int diff_can_quit_early(struct diff_options *opt)
6243 {
6244         return (opt->flags.quick &&
6245                 !opt->filter &&
6246                 opt->flags.has_changes);
6247 }
6248
6249 /*
6250  * Shall changes to this submodule be ignored?
6251  *
6252  * Submodule changes can be configured to be ignored separately for each path,
6253  * but that configuration can be overridden from the command line.
6254  */
6255 static int is_submodule_ignored(const char *path, struct diff_options *options)
6256 {
6257         int ignored = 0;
6258         struct diff_flags orig_flags = options->flags;
6259         if (!options->flags.override_submodule_config)
6260                 set_diffopt_flags_from_submodule_config(options, path);
6261         if (options->flags.ignore_submodules)
6262                 ignored = 1;
6263         options->flags = orig_flags;
6264         return ignored;
6265 }
6266
6267 void diff_addremove(struct diff_options *options,
6268                     int addremove, unsigned mode,
6269                     const struct object_id *oid,
6270                     int oid_valid,
6271                     const char *concatpath, unsigned dirty_submodule)
6272 {
6273         struct diff_filespec *one, *two;
6274
6275         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6276                 return;
6277
6278         /* This may look odd, but it is a preparation for
6279          * feeding "there are unchanged files which should
6280          * not produce diffs, but when you are doing copy
6281          * detection you would need them, so here they are"
6282          * entries to the diff-core.  They will be prefixed
6283          * with something like '=' or '*' (I haven't decided
6284          * which but should not make any difference).
6285          * Feeding the same new and old to diff_change()
6286          * also has the same effect.
6287          * Before the final output happens, they are pruned after
6288          * merged into rename/copy pairs as appropriate.
6289          */
6290         if (options->flags.reverse_diff)
6291                 addremove = (addremove == '+' ? '-' :
6292                              addremove == '-' ? '+' : addremove);
6293
6294         if (options->prefix &&
6295             strncmp(concatpath, options->prefix, options->prefix_length))
6296                 return;
6297
6298         one = alloc_filespec(concatpath);
6299         two = alloc_filespec(concatpath);
6300
6301         if (addremove != '+')
6302                 fill_filespec(one, oid, oid_valid, mode);
6303         if (addremove != '-') {
6304                 fill_filespec(two, oid, oid_valid, mode);
6305                 two->dirty_submodule = dirty_submodule;
6306         }
6307
6308         diff_queue(&diff_queued_diff, one, two);
6309         if (!options->flags.diff_from_contents)
6310                 options->flags.has_changes = 1;
6311 }
6312
6313 void diff_change(struct diff_options *options,
6314                  unsigned old_mode, unsigned new_mode,
6315                  const struct object_id *old_oid,
6316                  const struct object_id *new_oid,
6317                  int old_oid_valid, int new_oid_valid,
6318                  const char *concatpath,
6319                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6320 {
6321         struct diff_filespec *one, *two;
6322         struct diff_filepair *p;
6323
6324         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6325             is_submodule_ignored(concatpath, options))
6326                 return;
6327
6328         if (options->flags.reverse_diff) {
6329                 SWAP(old_mode, new_mode);
6330                 SWAP(old_oid, new_oid);
6331                 SWAP(old_oid_valid, new_oid_valid);
6332                 SWAP(old_dirty_submodule, new_dirty_submodule);
6333         }
6334
6335         if (options->prefix &&
6336             strncmp(concatpath, options->prefix, options->prefix_length))
6337                 return;
6338
6339         one = alloc_filespec(concatpath);
6340         two = alloc_filespec(concatpath);
6341         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6342         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6343         one->dirty_submodule = old_dirty_submodule;
6344         two->dirty_submodule = new_dirty_submodule;
6345         p = diff_queue(&diff_queued_diff, one, two);
6346
6347         if (options->flags.diff_from_contents)
6348                 return;
6349
6350         if (options->flags.quick && options->skip_stat_unmatch &&
6351             !diff_filespec_check_stat_unmatch(options->repo, p))
6352                 return;
6353
6354         options->flags.has_changes = 1;
6355 }
6356
6357 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6358 {
6359         struct diff_filepair *pair;
6360         struct diff_filespec *one, *two;
6361
6362         if (options->prefix &&
6363             strncmp(path, options->prefix, options->prefix_length))
6364                 return NULL;
6365
6366         one = alloc_filespec(path);
6367         two = alloc_filespec(path);
6368         pair = diff_queue(&diff_queued_diff, one, two);
6369         pair->is_unmerged = 1;
6370         return pair;
6371 }
6372
6373 static char *run_textconv(struct repository *r,
6374                           const char *pgm,
6375                           struct diff_filespec *spec,
6376                           size_t *outsize)
6377 {
6378         struct diff_tempfile *temp;
6379         const char *argv[3];
6380         const char **arg = argv;
6381         struct child_process child = CHILD_PROCESS_INIT;
6382         struct strbuf buf = STRBUF_INIT;
6383         int err = 0;
6384
6385         temp = prepare_temp_file(r, spec->path, spec);
6386         *arg++ = pgm;
6387         *arg++ = temp->name;
6388         *arg = NULL;
6389
6390         child.use_shell = 1;
6391         child.argv = argv;
6392         child.out = -1;
6393         if (start_command(&child)) {
6394                 remove_tempfile();
6395                 return NULL;
6396         }
6397
6398         if (strbuf_read(&buf, child.out, 0) < 0)
6399                 err = error("error reading from textconv command '%s'", pgm);
6400         close(child.out);
6401
6402         if (finish_command(&child) || err) {
6403                 strbuf_release(&buf);
6404                 remove_tempfile();
6405                 return NULL;
6406         }
6407         remove_tempfile();
6408
6409         return strbuf_detach(&buf, outsize);
6410 }
6411
6412 size_t fill_textconv(struct repository *r,
6413                      struct userdiff_driver *driver,
6414                      struct diff_filespec *df,
6415                      char **outbuf)
6416 {
6417         size_t size;
6418
6419         if (!driver) {
6420                 if (!DIFF_FILE_VALID(df)) {
6421                         *outbuf = "";
6422                         return 0;
6423                 }
6424                 if (diff_populate_filespec(r, df, 0))
6425                         die("unable to read files to diff");
6426                 *outbuf = df->data;
6427                 return df->size;
6428         }
6429
6430         if (!driver->textconv)
6431                 BUG("fill_textconv called with non-textconv driver");
6432
6433         if (driver->textconv_cache && df->oid_valid) {
6434                 *outbuf = notes_cache_get(driver->textconv_cache,
6435                                           &df->oid,
6436                                           &size);
6437                 if (*outbuf)
6438                         return size;
6439         }
6440
6441         *outbuf = run_textconv(r, driver->textconv, df, &size);
6442         if (!*outbuf)
6443                 die("unable to read files to diff");
6444
6445         if (driver->textconv_cache && df->oid_valid) {
6446                 /* ignore errors, as we might be in a readonly repository */
6447                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6448                                 size);
6449                 /*
6450                  * we could save up changes and flush them all at the end,
6451                  * but we would need an extra call after all diffing is done.
6452                  * Since generating a cache entry is the slow path anyway,
6453                  * this extra overhead probably isn't a big deal.
6454                  */
6455                 notes_cache_write(driver->textconv_cache);
6456         }
6457
6458         return size;
6459 }
6460
6461 int textconv_object(struct repository *r,
6462                     const char *path,
6463                     unsigned mode,
6464                     const struct object_id *oid,
6465                     int oid_valid,
6466                     char **buf,
6467                     unsigned long *buf_size)
6468 {
6469         struct diff_filespec *df;
6470         struct userdiff_driver *textconv;
6471
6472         df = alloc_filespec(path);
6473         fill_filespec(df, oid, oid_valid, mode);
6474         textconv = get_textconv(r->index, df);
6475         if (!textconv) {
6476                 free_filespec(df);
6477                 return 0;
6478         }
6479
6480         *buf_size = fill_textconv(r, textconv, df, buf);
6481         free_filespec(df);
6482         return 1;
6483 }
6484
6485 void setup_diff_pager(struct diff_options *opt)
6486 {
6487         /*
6488          * If the user asked for our exit code, then either they want --quiet
6489          * or --exit-code. We should definitely not bother with a pager in the
6490          * former case, as we will generate no output. Since we still properly
6491          * report our exit code even when a pager is run, we _could_ run a
6492          * pager with --exit-code. But since we have not done so historically,
6493          * and because it is easy to find people oneline advising "git diff
6494          * --exit-code" in hooks and other scripts, we do not do so.
6495          */
6496         if (!opt->flags.exit_with_status &&
6497             check_pager_config("diff") != 0)
6498                 setup_pager();
6499 }