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 "strvec.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "parse-options.h"
  27 #include "help.h"
  28 #include "promisor-remote.h"
  29
  30 #ifdef NO_FAST_WORKING_DIRECTORY
  31 #define FAST_WORKING_DIRECTORY 0
  32 #else
  33 #define FAST_WORKING_DIRECTORY 1
  34 #endif
  35
  36 static int diff_detect_rename_default;
  37 static int diff_indent_heuristic = 1;
  38 static int diff_rename_limit_default = 400;
  39 static int diff_suppress_blank_empty;
  40 static int diff_use_color_default = -1;
  41 static int diff_color_moved_default;
  42 static int diff_color_moved_ws_default;
  43 static int diff_context_default = 3;
  44 static int diff_interhunk_context_default;
  45 static const char *diff_word_regex_cfg;
  46 static const char *external_diff_cmd_cfg;
  47 static const char *diff_order_file_cfg;
  48 int diff_auto_refresh_index = 1;
  49 static int diff_mnemonic_prefix;
  50 static int diff_no_prefix;
  51 static int diff_relative;
  52 static int diff_stat_graph_width;
  53 static int diff_dirstat_permille_default = 30;
  54 static struct diff_options default_diff_options;
  55 static long diff_algorithm;
  56 static unsigned ws_error_highlight_default = WSEH_NEW;
  57
  58 static char diff_colors[][COLOR_MAXLEN] = {
  59         GIT_COLOR_RESET,
  60         GIT_COLOR_NORMAL,       /* CONTEXT */
  61         GIT_COLOR_BOLD,         /* METAINFO */
  62         GIT_COLOR_CYAN,         /* FRAGINFO */
  63         GIT_COLOR_RED,          /* OLD */
  64         GIT_COLOR_GREEN,        /* NEW */
  65         GIT_COLOR_YELLOW,       /* COMMIT */
  66         GIT_COLOR_BG_RED,       /* WHITESPACE */
  67         GIT_COLOR_NORMAL,       /* FUNCINFO */
  68         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  69         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  70         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  71         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  72         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  73         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  74         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  75         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  76         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  77         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  78         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  79         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  80         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  81         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  82 };
  83
  84 static const char *color_diff_slots[] = {
  85         [DIFF_CONTEXT]                = "context",
  86         [DIFF_METAINFO]               = "meta",
  87         [DIFF_FRAGINFO]               = "frag",
  88         [DIFF_FILE_OLD]               = "old",
  89         [DIFF_FILE_NEW]               = "new",
  90         [DIFF_COMMIT]                 = "commit",
  91         [DIFF_WHITESPACE]             = "whitespace",
  92         [DIFF_FUNCINFO]               = "func",
  93         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  94         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  95         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  96         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  97         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  98         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  99         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
 100         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
 101         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 102         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 103         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 104         [DIFF_CONTEXT_BOLD]           = "contextBold",
 105         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 106         [DIFF_FILE_NEW_BOLD]          = "newBold",
 107 };
 108
 109 define_list_config_array_extra(color_diff_slots, {"plain"});
 110
 111 static int parse_diff_color_slot(const char *var)
 112 {
 113         if (!strcasecmp(var, "plain"))
 114                 return DIFF_CONTEXT;
 115         return LOOKUP_CONFIG(color_diff_slots, var);
 116 }
 117
 118 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 119                                 struct strbuf *errmsg)
 120 {
 121         char *params_copy = xstrdup(params_string);
 122         struct string_list params = STRING_LIST_INIT_NODUP;
 123         int ret = 0;
 124         int i;
 125
 126         if (*params_copy)
 127                 string_list_split_in_place(&params, params_copy, ',', -1);
 128         for (i = 0; i < params.nr; i++) {
 129                 const char *p = params.items[i].string;
 130                 if (!strcmp(p, "changes")) {
 131                         options->flags.dirstat_by_line = 0;
 132                         options->flags.dirstat_by_file = 0;
 133                 } else if (!strcmp(p, "lines")) {
 134                         options->flags.dirstat_by_line = 1;
 135                         options->flags.dirstat_by_file = 0;
 136                 } else if (!strcmp(p, "files")) {
 137                         options->flags.dirstat_by_line = 0;
 138                         options->flags.dirstat_by_file = 1;
 139                 } else if (!strcmp(p, "noncumulative")) {
 140                         options->flags.dirstat_cumulative = 0;
 141                 } else if (!strcmp(p, "cumulative")) {
 142                         options->flags.dirstat_cumulative = 1;
 143                 } else if (isdigit(*p)) {
 144                         char *end;
 145                         int permille = strtoul(p, &end, 10) * 10;
 146                         if (*end == '.' && isdigit(*++end)) {
 147                                 /* only use first digit */
 148                                 permille += *end - '0';
 149                                 /* .. and ignore any further digits */
 150                                 while (isdigit(*++end))
 151                                         ; /* nothing */
 152                         }
 153                         if (!*end)
 154                                 options->dirstat_permille = permille;
 155                         else {
 156                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 157                                             p);
 158                                 ret++;
 159                         }
 160                 } else {
 161                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 162                         ret++;
 163                 }
 164
 165         }
 166         string_list_clear(&params, 0);
 167         free(params_copy);
 168         return ret;
 169 }
 170
 171 static int parse_submodule_params(struct diff_options *options, const char *value)
 172 {
 173         if (!strcmp(value, "log"))
 174                 options->submodule_format = DIFF_SUBMODULE_LOG;
 175         else if (!strcmp(value, "short"))
 176                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 177         else if (!strcmp(value, "diff"))
 178                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 179         /*
 180          * Please update $__git_diff_submodule_formats in
 181          * git-completion.bash when you add new formats.
 182          */
 183         else
 184                 return -1;
 185         return 0;
 186 }
 187
 188 int git_config_rename(const char *var, const char *value)
 189 {
 190         if (!value)
 191                 return DIFF_DETECT_RENAME;
 192         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 193                 return  DIFF_DETECT_COPY;
 194         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 195 }
 196
 197 long parse_algorithm_value(const char *value)
 198 {
 199         if (!value)
 200                 return -1;
 201         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 202                 return 0;
 203         else if (!strcasecmp(value, "minimal"))
 204                 return XDF_NEED_MINIMAL;
 205         else if (!strcasecmp(value, "patience"))
 206                 return XDF_PATIENCE_DIFF;
 207         else if (!strcasecmp(value, "histogram"))
 208                 return XDF_HISTOGRAM_DIFF;
 209         /*
 210          * Please update $__git_diff_algorithms in git-completion.bash
 211          * when you add new algorithms.
 212          */
 213         return -1;
 214 }
 215
 216 static int parse_one_token(const char **arg, const char *token)
 217 {
 218         const char *rest;
 219         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 220                 *arg = rest;
 221                 return 1;
 222         }
 223         return 0;
 224 }
 225
 226 static int parse_ws_error_highlight(const char *arg)
 227 {
 228         const char *orig_arg = arg;
 229         unsigned val = 0;
 230
 231         while (*arg) {
 232                 if (parse_one_token(&arg, "none"))
 233                         val = 0;
 234                 else if (parse_one_token(&arg, "default"))
 235                         val = WSEH_NEW;
 236                 else if (parse_one_token(&arg, "all"))
 237                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 238                 else if (parse_one_token(&arg, "new"))
 239                         val |= WSEH_NEW;
 240                 else if (parse_one_token(&arg, "old"))
 241                         val |= WSEH_OLD;
 242                 else if (parse_one_token(&arg, "context"))
 243                         val |= WSEH_CONTEXT;
 244                 else {
 245                         return -1 - (int)(arg - orig_arg);
 246                 }
 247                 if (*arg)
 248                         arg++;
 249         }
 250         return val;
 251 }
 252
 253 /*
 254  * These are to give UI layer defaults.
 255  * The core-level commands such as git-diff-files should
 256  * never be affected by the setting of diff.renames
 257  * the user happens to have in the configuration file.
 258  */
 259 void init_diff_ui_defaults(void)
 260 {
 261         diff_detect_rename_default = DIFF_DETECT_RENAME;
 262 }
 263
 264 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 265 {
 266         if (!strcmp(var, "diff.indentheuristic"))
 267                 diff_indent_heuristic = git_config_bool(var, value);
 268         return 0;
 269 }
 270
 271 static int parse_color_moved(const char *arg)
 272 {
 273         switch (git_parse_maybe_bool(arg)) {
 274         case 0:
 275                 return COLOR_MOVED_NO;
 276         case 1:
 277                 return COLOR_MOVED_DEFAULT;
 278         default:
 279                 break;
 280         }
 281
 282         if (!strcmp(arg, "no"))
 283                 return COLOR_MOVED_NO;
 284         else if (!strcmp(arg, "plain"))
 285                 return COLOR_MOVED_PLAIN;
 286         else if (!strcmp(arg, "blocks"))
 287                 return COLOR_MOVED_BLOCKS;
 288         else if (!strcmp(arg, "zebra"))
 289                 return COLOR_MOVED_ZEBRA;
 290         else if (!strcmp(arg, "default"))
 291                 return COLOR_MOVED_DEFAULT;
 292         else if (!strcmp(arg, "dimmed-zebra"))
 293                 return COLOR_MOVED_ZEBRA_DIM;
 294         else if (!strcmp(arg, "dimmed_zebra"))
 295                 return COLOR_MOVED_ZEBRA_DIM;
 296         else
 297                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 298 }
 299
 300 static unsigned parse_color_moved_ws(const char *arg)
 301 {
 302         int ret = 0;
 303         struct string_list l = STRING_LIST_INIT_DUP;
 304         struct string_list_item *i;
 305
 306         string_list_split(&l, arg, ',', -1);
 307
 308         for_each_string_list_item(i, &l) {
 309                 struct strbuf sb = STRBUF_INIT;
 310                 strbuf_addstr(&sb, i->string);
 311                 strbuf_trim(&sb);
 312
 313                 if (!strcmp(sb.buf, "no"))
 314                         ret = 0;
 315                 else if (!strcmp(sb.buf, "ignore-space-change"))
 316                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 317                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 318                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 319                 else if (!strcmp(sb.buf, "ignore-all-space"))
 320                         ret |= XDF_IGNORE_WHITESPACE;
 321                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 322                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 323                 else {
 324                         ret |= COLOR_MOVED_WS_ERROR;
 325                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 326                 }
 327
 328                 strbuf_release(&sb);
 329         }
 330
 331         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 332             (ret & XDF_WHITESPACE_FLAGS)) {
 333                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 334                 ret |= COLOR_MOVED_WS_ERROR;
 335         }
 336
 337         string_list_clear(&l, 0);
 338
 339         return ret;
 340 }
 341
 342 int git_diff_ui_config(const char *var, const char *value, void *cb)
 343 {
 344         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 345                 diff_use_color_default = git_config_colorbool(var, value);
 346                 return 0;
 347         }
 348         if (!strcmp(var, "diff.colormoved")) {
 349                 int cm = parse_color_moved(value);
 350                 if (cm < 0)
 351                         return -1;
 352                 diff_color_moved_default = cm;
 353                 return 0;
 354         }
 355         if (!strcmp(var, "diff.colormovedws")) {
 356                 unsigned cm = parse_color_moved_ws(value);
 357                 if (cm & COLOR_MOVED_WS_ERROR)
 358                         return -1;
 359                 diff_color_moved_ws_default = cm;
 360                 return 0;
 361         }
 362         if (!strcmp(var, "diff.context")) {
 363                 diff_context_default = git_config_int(var, value);
 364                 if (diff_context_default < 0)
 365                         return -1;
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.interhunkcontext")) {
 369                 diff_interhunk_context_default = git_config_int(var, value);
 370                 if (diff_interhunk_context_default < 0)
 371                         return -1;
 372                 return 0;
 373         }
 374         if (!strcmp(var, "diff.renames")) {
 375                 diff_detect_rename_default = git_config_rename(var, value);
 376                 return 0;
 377         }
 378         if (!strcmp(var, "diff.autorefreshindex")) {
 379                 diff_auto_refresh_index = git_config_bool(var, value);
 380                 return 0;
 381         }
 382         if (!strcmp(var, "diff.mnemonicprefix")) {
 383                 diff_mnemonic_prefix = git_config_bool(var, value);
 384                 return 0;
 385         }
 386         if (!strcmp(var, "diff.noprefix")) {
 387                 diff_no_prefix = git_config_bool(var, value);
 388                 return 0;
 389         }
 390         if (!strcmp(var, "diff.relative")) {
 391                 diff_relative = git_config_bool(var, value);
 392                 return 0;
 393         }
 394         if (!strcmp(var, "diff.statgraphwidth")) {
 395                 diff_stat_graph_width = git_config_int(var, value);
 396                 return 0;
 397         }
 398         if (!strcmp(var, "diff.external"))
 399                 return git_config_string(&external_diff_cmd_cfg, var, value);
 400         if (!strcmp(var, "diff.wordregex"))
 401                 return git_config_string(&diff_word_regex_cfg, var, value);
 402         if (!strcmp(var, "diff.orderfile"))
 403                 return git_config_pathname(&diff_order_file_cfg, var, value);
 404
 405         if (!strcmp(var, "diff.ignoresubmodules"))
 406                 handle_ignore_submodules_arg(&default_diff_options, value);
 407
 408         if (!strcmp(var, "diff.submodule")) {
 409                 if (parse_submodule_params(&default_diff_options, value))
 410                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 411                                 value);
 412                 return 0;
 413         }
 414
 415         if (!strcmp(var, "diff.algorithm")) {
 416                 diff_algorithm = parse_algorithm_value(value);
 417                 if (diff_algorithm < 0)
 418                         return -1;
 419                 return 0;
 420         }
 421
 422         if (git_color_config(var, value, cb) < 0)
 423                 return -1;
 424
 425         return git_diff_basic_config(var, value, cb);
 426 }
 427
 428 int git_diff_basic_config(const char *var, const char *value, void *cb)
 429 {
 430         const char *name;
 431
 432         if (!strcmp(var, "diff.renamelimit")) {
 433                 diff_rename_limit_default = git_config_int(var, value);
 434                 return 0;
 435         }
 436
 437         if (userdiff_config(var, value) < 0)
 438                 return -1;
 439
 440         if (skip_prefix(var, "diff.color.", &name) ||
 441             skip_prefix(var, "color.diff.", &name)) {
 442                 int slot = parse_diff_color_slot(name);
 443                 if (slot < 0)
 444                         return 0;
 445                 if (!value)
 446                         return config_error_nonbool(var);
 447                 return color_parse(value, diff_colors[slot]);
 448         }
 449
 450         if (!strcmp(var, "diff.wserrorhighlight")) {
 451                 int val = parse_ws_error_highlight(value);
 452                 if (val < 0)
 453                         return -1;
 454                 ws_error_highlight_default = val;
 455                 return 0;
 456         }
 457
 458         /* like GNU diff's --suppress-blank-empty option  */
 459         if (!strcmp(var, "diff.suppressblankempty") ||
 460                         /* for backwards compatibility */
 461                         !strcmp(var, "diff.suppress-blank-empty")) {
 462                 diff_suppress_blank_empty = git_config_bool(var, value);
 463                 return 0;
 464         }
 465
 466         if (!strcmp(var, "diff.dirstat")) {
 467                 struct strbuf errmsg = STRBUF_INIT;
 468                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 469                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 470                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 471                                 errmsg.buf);
 472                 strbuf_release(&errmsg);
 473                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 474                 return 0;
 475         }
 476
 477         if (git_diff_heuristic_config(var, value, cb) < 0)
 478                 return -1;
 479
 480         return git_default_config(var, value, cb);
 481 }
 482
 483 static char *quote_two(const char *one, const char *two)
 484 {
 485         int need_one = quote_c_style(one, NULL, NULL, CQUOTE_NODQ);
 486         int need_two = quote_c_style(two, NULL, NULL, CQUOTE_NODQ);
 487         struct strbuf res = STRBUF_INIT;
 488
 489         if (need_one + need_two) {
 490                 strbuf_addch(&res, '"');
 491                 quote_c_style(one, &res, NULL, CQUOTE_NODQ);
 492                 quote_c_style(two, &res, NULL, CQUOTE_NODQ);
 493                 strbuf_addch(&res, '"');
 494         } else {
 495                 strbuf_addstr(&res, one);
 496                 strbuf_addstr(&res, two);
 497         }
 498         return strbuf_detach(&res, NULL);
 499 }
 500
 501 static const char *external_diff(void)
 502 {
 503         static const char *external_diff_cmd = NULL;
 504         static int done_preparing = 0;
 505
 506         if (done_preparing)
 507                 return external_diff_cmd;
 508         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 509         if (!external_diff_cmd)
 510                 external_diff_cmd = external_diff_cmd_cfg;
 511         done_preparing = 1;
 512         return external_diff_cmd;
 513 }
 514
 515 /*
 516  * Keep track of files used for diffing. Sometimes such an entry
 517  * refers to a temporary file, sometimes to an existing file, and
 518  * sometimes to "/dev/null".
 519  */
 520 static struct diff_tempfile {
 521         /*
 522          * filename external diff should read from, or NULL if this
 523          * entry is currently not in use:
 524          */
 525         const char *name;
 526
 527         char hex[GIT_MAX_HEXSZ + 1];
 528         char mode[10];
 529
 530         /*
 531          * If this diff_tempfile instance refers to a temporary file,
 532          * this tempfile object is used to manage its lifetime.
 533          */
 534         struct tempfile *tempfile;
 535 } diff_temp[2];
 536
 537 struct emit_callback {
 538         int color_diff;
 539         unsigned ws_rule;
 540         int blank_at_eof_in_preimage;
 541         int blank_at_eof_in_postimage;
 542         int lno_in_preimage;
 543         int lno_in_postimage;
 544         const char **label_path;
 545         struct diff_words_data *diff_words;
 546         struct diff_options *opt;
 547         struct strbuf *header;
 548 };
 549
 550 static int count_lines(const char *data, int size)
 551 {
 552         int count, ch, completely_empty = 1, nl_just_seen = 0;
 553         count = 0;
 554         while (0 < size--) {
 555                 ch = *data++;
 556                 if (ch == '\n') {
 557                         count++;
 558                         nl_just_seen = 1;
 559                         completely_empty = 0;
 560                 }
 561                 else {
 562                         nl_just_seen = 0;
 563                         completely_empty = 0;
 564                 }
 565         }
 566         if (completely_empty)
 567                 return 0;
 568         if (!nl_just_seen)
 569                 count++; /* no trailing newline */
 570         return count;
 571 }
 572
 573 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 574                        struct diff_filespec *one)
 575 {
 576         if (!DIFF_FILE_VALID(one)) {
 577                 mf->ptr = (char *)""; /* does not matter */
 578                 mf->size = 0;
 579                 return 0;
 580         }
 581         else if (diff_populate_filespec(r, one, NULL))
 582                 return -1;
 583
 584         mf->ptr = one->data;
 585         mf->size = one->size;
 586         return 0;
 587 }
 588
 589 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 590 static unsigned long diff_filespec_size(struct repository *r,
 591                                         struct diff_filespec *one)
 592 {
 593         struct diff_populate_filespec_options dpf_options = {
 594                 .check_size_only = 1,
 595         };
 596
 597         if (!DIFF_FILE_VALID(one))
 598                 return 0;
 599         diff_populate_filespec(r, one, &dpf_options);
 600         return one->size;
 601 }
 602
 603 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 604 {
 605         char *ptr = mf->ptr;
 606         long size = mf->size;
 607         int cnt = 0;
 608
 609         if (!size)
 610                 return cnt;
 611         ptr += size - 1; /* pointing at the very end */
 612         if (*ptr != '\n')
 613                 ; /* incomplete line */
 614         else
 615                 ptr--; /* skip the last LF */
 616         while (mf->ptr < ptr) {
 617                 char *prev_eol;
 618                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 619                         if (*prev_eol == '\n')
 620                                 break;
 621                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 622                         break;
 623                 cnt++;
 624                 ptr = prev_eol - 1;
 625         }
 626         return cnt;
 627 }
 628
 629 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 630                                struct emit_callback *ecbdata)
 631 {
 632         int l1, l2, at;
 633         unsigned ws_rule = ecbdata->ws_rule;
 634         l1 = count_trailing_blank(mf1, ws_rule);
 635         l2 = count_trailing_blank(mf2, ws_rule);
 636         if (l2 <= l1) {
 637                 ecbdata->blank_at_eof_in_preimage = 0;
 638                 ecbdata->blank_at_eof_in_postimage = 0;
 639                 return;
 640         }
 641         at = count_lines(mf1->ptr, mf1->size);
 642         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 643
 644         at = count_lines(mf2->ptr, mf2->size);
 645         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 646 }
 647
 648 static void emit_line_0(struct diff_options *o,
 649                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 650                         int first, const char *line, int len)
 651 {
 652         int has_trailing_newline, has_trailing_carriage_return;
 653         int needs_reset = 0; /* at the end of the line */
 654         FILE *file = o->file;
 655
 656         fputs(diff_line_prefix(o), file);
 657
 658         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 659         if (has_trailing_newline)
 660                 len--;
 661
 662         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 663         if (has_trailing_carriage_return)
 664                 len--;
 665
 666         if (!len && !first)
 667                 goto end_of_line;
 668
 669         if (reverse && want_color(o->use_color)) {
 670                 fputs(GIT_COLOR_REVERSE, file);
 671                 needs_reset = 1;
 672         }
 673
 674         if (set_sign) {
 675                 fputs(set_sign, file);
 676                 needs_reset = 1;
 677         }
 678
 679         if (first)
 680                 fputc(first, file);
 681
 682         if (!len)
 683                 goto end_of_line;
 684
 685         if (set) {
 686                 if (set_sign && set != set_sign)
 687                         fputs(reset, file);
 688                 fputs(set, file);
 689                 needs_reset = 1;
 690         }
 691         fwrite(line, len, 1, file);
 692         needs_reset = 1; /* 'line' may contain color codes. */
 693
 694 end_of_line:
 695         if (needs_reset)
 696                 fputs(reset, file);
 697         if (has_trailing_carriage_return)
 698                 fputc('\r', file);
 699         if (has_trailing_newline)
 700                 fputc('\n', file);
 701 }
 702
 703 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 704                       const char *line, int len)
 705 {
 706         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 707 }
 708
 709 enum diff_symbol {
 710         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 711         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 712         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 713         DIFF_SYMBOL_BINARY_DIFF_BODY,
 714         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 715         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 716         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 717         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 718         DIFF_SYMBOL_STATS_LINE,
 719         DIFF_SYMBOL_WORD_DIFF,
 720         DIFF_SYMBOL_STAT_SEP,
 721         DIFF_SYMBOL_SUMMARY,
 722         DIFF_SYMBOL_SUBMODULE_ADD,
 723         DIFF_SYMBOL_SUBMODULE_DEL,
 724         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 725         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 726         DIFF_SYMBOL_SUBMODULE_HEADER,
 727         DIFF_SYMBOL_SUBMODULE_ERROR,
 728         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 729         DIFF_SYMBOL_REWRITE_DIFF,
 730         DIFF_SYMBOL_BINARY_FILES,
 731         DIFF_SYMBOL_HEADER,
 732         DIFF_SYMBOL_FILEPAIR_PLUS,
 733         DIFF_SYMBOL_FILEPAIR_MINUS,
 734         DIFF_SYMBOL_WORDS_PORCELAIN,
 735         DIFF_SYMBOL_WORDS,
 736         DIFF_SYMBOL_CONTEXT,
 737         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 738         DIFF_SYMBOL_PLUS,
 739         DIFF_SYMBOL_MINUS,
 740         DIFF_SYMBOL_NO_LF_EOF,
 741         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 742         DIFF_SYMBOL_CONTEXT_MARKER,
 743         DIFF_SYMBOL_SEPARATOR
 744 };
 745 /*
 746  * Flags for content lines:
 747  * 0..12 are whitespace rules
 748  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 749  * 16 is marking if the line is blank at EOF
 750  */
 751 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 752 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 753 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 754 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 755 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 756
 757 /*
 758  * This struct is used when we need to buffer the output of the diff output.
 759  *
 760  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 761  * into the pre/post image file. This pointer could be a union with the
 762  * line pointer. By storing an offset into the file instead of the literal line,
 763  * we can decrease the memory footprint for the buffered output. At first we
 764  * may want to only have indirection for the content lines, but we could also
 765  * enhance the state for emitting prefabricated lines, e.g. the similarity
 766  * score line or hunk/file headers would only need to store a number or path
 767  * and then the output can be constructed later on depending on state.
 768  */
 769 struct emitted_diff_symbol {
 770         const char *line;
 771         int len;
 772         int flags;
 773         int indent_off;   /* Offset to first non-whitespace character */
 774         int indent_width; /* The visual width of the indentation */
 775         enum diff_symbol s;
 776 };
 777 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 778
 779 struct emitted_diff_symbols {
 780         struct emitted_diff_symbol *buf;
 781         int nr, alloc;
 782 };
 783 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 784
 785 static void append_emitted_diff_symbol(struct diff_options *o,
 786                                        struct emitted_diff_symbol *e)
 787 {
 788         struct emitted_diff_symbol *f;
 789
 790         ALLOC_GROW(o->emitted_symbols->buf,
 791                    o->emitted_symbols->nr + 1,
 792                    o->emitted_symbols->alloc);
 793         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 794
 795         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 796         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 797 }
 798
 799 struct moved_entry {
 800         struct hashmap_entry ent;
 801         const struct emitted_diff_symbol *es;
 802         struct moved_entry *next_line;
 803 };
 804
 805 struct moved_block {
 806         struct moved_entry *match;
 807         int wsd; /* The whitespace delta of this block */
 808 };
 809
 810 static void moved_block_clear(struct moved_block *b)
 811 {
 812         memset(b, 0, sizeof(*b));
 813 }
 814
 815 #define INDENT_BLANKLINE INT_MIN
 816
 817 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 818 {
 819         unsigned int off = 0, i;
 820         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 821         const char *s = es->line;
 822         const int len = es->len;
 823
 824         /* skip any \v \f \r at start of indentation */
 825         while (s[off] == '\f' || s[off] == '\v' ||
 826                (s[off] == '\r' && off < len - 1))
 827                 off++;
 828
 829         /* calculate the visual width of indentation */
 830         while(1) {
 831                 if (s[off] == ' ') {
 832                         width++;
 833                         off++;
 834                 } else if (s[off] == '\t') {
 835                         width += tab_width - (width % tab_width);
 836                         while (s[++off] == '\t')
 837                                 width += tab_width;
 838                 } else {
 839                         break;
 840                 }
 841         }
 842
 843         /* check if this line is blank */
 844         for (i = off; i < len; i++)
 845                 if (!isspace(s[i]))
 846                     break;
 847
 848         if (i == len) {
 849                 es->indent_width = INDENT_BLANKLINE;
 850                 es->indent_off = len;
 851         } else {
 852                 es->indent_off = off;
 853                 es->indent_width = width;
 854         }
 855 }
 856
 857 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 858                             const struct emitted_diff_symbol *b,
 859                             int *out)
 860 {
 861         int a_len = a->len,
 862             b_len = b->len,
 863             a_off = a->indent_off,
 864             a_width = a->indent_width,
 865             b_off = b->indent_off,
 866             b_width = b->indent_width;
 867         int delta;
 868
 869         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 870                 *out = INDENT_BLANKLINE;
 871                 return 1;
 872         }
 873
 874         if (a->s == DIFF_SYMBOL_PLUS)
 875                 delta = a_width - b_width;
 876         else
 877                 delta = b_width - a_width;
 878
 879         if (a_len - a_off != b_len - b_off ||
 880             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 881                 return 0;
 882
 883         *out = delta;
 884
 885         return 1;
 886 }
 887
 888 static int cmp_in_block_with_wsd(const struct diff_options *o,
 889                                  const struct moved_entry *cur,
 890                                  const struct moved_entry *match,
 891                                  struct moved_block *pmb,
 892                                  int n)
 893 {
 894         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 895         int al = cur->es->len, bl = match->es->len, cl = l->len;
 896         const char *a = cur->es->line,
 897                    *b = match->es->line,
 898                    *c = l->line;
 899         int a_off = cur->es->indent_off,
 900             a_width = cur->es->indent_width,
 901             c_off = l->indent_off,
 902             c_width = l->indent_width;
 903         int delta;
 904
 905         /*
 906          * We need to check if 'cur' is equal to 'match'.  As those
 907          * are from the same (+/-) side, we do not need to adjust for
 908          * indent changes. However these were found using fuzzy
 909          * matching so we do have to check if they are equal. Here we
 910          * just check the lengths. We delay calling memcmp() to check
 911          * the contents until later as if the length comparison for a
 912          * and c fails we can avoid the call all together.
 913          */
 914         if (al != bl)
 915                 return 1;
 916
 917         /* If 'l' and 'cur' are both blank then they match. */
 918         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 919                 return 0;
 920
 921         /*
 922          * The indent changes of the block are known and stored in pmb->wsd;
 923          * however we need to check if the indent changes of the current line
 924          * match those of the current block and that the text of 'l' and 'cur'
 925          * after the indentation match.
 926          */
 927         if (cur->es->s == DIFF_SYMBOL_PLUS)
 928                 delta = a_width - c_width;
 929         else
 930                 delta = c_width - a_width;
 931
 932         /*
 933          * If the previous lines of this block were all blank then set its
 934          * whitespace delta.
 935          */
 936         if (pmb->wsd == INDENT_BLANKLINE)
 937                 pmb->wsd = delta;
 938
 939         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 940                  !memcmp(a, b, al) && !
 941                  memcmp(a + a_off, c + c_off, al - a_off));
 942 }
 943
 944 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 945                            const struct hashmap_entry *eptr,
 946                            const struct hashmap_entry *entry_or_key,
 947                            const void *keydata)
 948 {
 949         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 950         const struct moved_entry *a, *b;
 951         unsigned flags = diffopt->color_moved_ws_handling
 952                          & XDF_WHITESPACE_FLAGS;
 953
 954         a = container_of(eptr, const struct moved_entry, ent);
 955         b = container_of(entry_or_key, const struct moved_entry, ent);
 956
 957         if (diffopt->color_moved_ws_handling &
 958             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 959                 /*
 960                  * As there is not specific white space config given,
 961                  * we'd need to check for a new block, so ignore all
 962                  * white space. The setup of the white space
 963                  * configuration for the next block is done else where
 964                  */
 965                 flags |= XDF_IGNORE_WHITESPACE;
 966
 967         return !xdiff_compare_lines(a->es->line, a->es->len,
 968                                     b->es->line, b->es->len,
 969                                     flags);
 970 }
 971
 972 static struct moved_entry *prepare_entry(struct diff_options *o,
 973                                          int line_no)
 974 {
 975         struct moved_entry *ret = xmalloc(sizeof(*ret));
 976         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 977         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 978         unsigned int hash = xdiff_hash_string(l->line, l->len, flags);
 979
 980         hashmap_entry_init(&ret->ent, hash);
 981         ret->es = l;
 982         ret->next_line = NULL;
 983
 984         return ret;
 985 }
 986
 987 static void add_lines_to_move_detection(struct diff_options *o,
 988                                         struct hashmap *add_lines,
 989                                         struct hashmap *del_lines)
 990 {
 991         struct moved_entry *prev_line = NULL;
 992
 993         int n;
 994         for (n = 0; n < o->emitted_symbols->nr; n++) {
 995                 struct hashmap *hm;
 996                 struct moved_entry *key;
 997
 998                 switch (o->emitted_symbols->buf[n].s) {
 999                 case DIFF_SYMBOL_PLUS:
1000                         hm = add_lines;
1001                         break;
1002                 case DIFF_SYMBOL_MINUS:
1003                         hm = del_lines;
1004                         break;
1005                 default:
1006                         prev_line = NULL;
1007                         continue;
1008                 }
1009
1010                 if (o->color_moved_ws_handling &
1011                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1012                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1013                 key = prepare_entry(o, n);
1014                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1015                         prev_line->next_line = key;
1016
1017                 hashmap_add(hm, &key->ent);
1018                 prev_line = key;
1019         }
1020 }
1021
1022 static void pmb_advance_or_null(struct diff_options *o,
1023                                 struct moved_entry *match,
1024                                 struct hashmap *hm,
1025                                 struct moved_block *pmb,
1026                                 int pmb_nr)
1027 {
1028         int i;
1029         for (i = 0; i < pmb_nr; i++) {
1030                 struct moved_entry *prev = pmb[i].match;
1031                 struct moved_entry *cur = (prev && prev->next_line) ?
1032                                 prev->next_line : NULL;
1033                 if (cur && !hm->cmpfn(o, &cur->ent, &match->ent, NULL)) {
1034                         pmb[i].match = cur;
1035                 } else {
1036                         pmb[i].match = NULL;
1037                 }
1038         }
1039 }
1040
1041 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1042                                             struct moved_entry *match,
1043                                             struct hashmap *hm,
1044                                             struct moved_block *pmb,
1045                                             int pmb_nr, int n)
1046 {
1047         int i;
1048         char *got_match = xcalloc(1, pmb_nr);
1049
1050         hashmap_for_each_entry_from(hm, match, ent) {
1051                 for (i = 0; i < pmb_nr; i++) {
1052                         struct moved_entry *prev = pmb[i].match;
1053                         struct moved_entry *cur = (prev && prev->next_line) ?
1054                                         prev->next_line : NULL;
1055                         if (!cur)
1056                                 continue;
1057                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1058                                 got_match[i] |= 1;
1059                 }
1060         }
1061
1062         for (i = 0; i < pmb_nr; i++) {
1063                 if (got_match[i]) {
1064                         /* Advance to the next line */
1065                         pmb[i].match = pmb[i].match->next_line;
1066                 } else {
1067                         moved_block_clear(&pmb[i]);
1068                 }
1069         }
1070
1071         free(got_match);
1072 }
1073
1074 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1075                                          int pmb_nr)
1076 {
1077         int lp, rp;
1078
1079         /* Shrink the set of potential block to the remaining running */
1080         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1081                 while (lp < pmb_nr && pmb[lp].match)
1082                         lp++;
1083                 /* lp points at the first NULL now */
1084
1085                 while (rp > -1 && !pmb[rp].match)
1086                         rp--;
1087                 /* rp points at the last non-NULL */
1088
1089                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1090                         pmb[lp] = pmb[rp];
1091                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1092                         rp--;
1093                         lp++;
1094                 }
1095         }
1096
1097         /* Remember the number of running sets */
1098         return rp + 1;
1099 }
1100
1101 /*
1102  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1103  *
1104  * Otherwise, if the last block has fewer alphanumeric characters than
1105  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1106  * that block.
1107  *
1108  * The last block consists of the (n - block_length)'th line up to but not
1109  * including the nth line.
1110  *
1111  * Returns 0 if the last block is empty or is unset by this function, non zero
1112  * otherwise.
1113  *
1114  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1115  * Think of a way to unify them.
1116  */
1117 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1118 {
1119         int i, alnum_count = 0;
1120         if (o->color_moved == COLOR_MOVED_PLAIN)
1121                 return block_length;
1122         for (i = 1; i < block_length + 1; i++) {
1123                 const char *c = o->emitted_symbols->buf[n - i].line;
1124                 for (; *c; c++) {
1125                         if (!isalnum(*c))
1126                                 continue;
1127                         alnum_count++;
1128                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1129                                 return 1;
1130                 }
1131         }
1132         for (i = 1; i < block_length + 1; i++)
1133                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1134         return 0;
1135 }
1136
1137 /* Find blocks of moved code, delegate actual coloring decision to helper */
1138 static void mark_color_as_moved(struct diff_options *o,
1139                                 struct hashmap *add_lines,
1140                                 struct hashmap *del_lines)
1141 {
1142         struct moved_block *pmb = NULL; /* potentially moved blocks */
1143         int pmb_nr = 0, pmb_alloc = 0;
1144         int n, flipped_block = 0, block_length = 0;
1145
1146
1147         for (n = 0; n < o->emitted_symbols->nr; n++) {
1148                 struct hashmap *hm = NULL;
1149                 struct moved_entry *key;
1150                 struct moved_entry *match = NULL;
1151                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1152                 enum diff_symbol last_symbol = 0;
1153
1154                 switch (l->s) {
1155                 case DIFF_SYMBOL_PLUS:
1156                         hm = del_lines;
1157                         key = prepare_entry(o, n);
1158                         match = hashmap_get_entry(hm, key, ent, NULL);
1159                         free(key);
1160                         break;
1161                 case DIFF_SYMBOL_MINUS:
1162                         hm = add_lines;
1163                         key = prepare_entry(o, n);
1164                         match = hashmap_get_entry(hm, key, ent, NULL);
1165                         free(key);
1166                         break;
1167                 default:
1168                         flipped_block = 0;
1169                 }
1170
1171                 if (!match) {
1172                         int i;
1173
1174                         adjust_last_block(o, n, block_length);
1175                         for(i = 0; i < pmb_nr; i++)
1176                                 moved_block_clear(&pmb[i]);
1177                         pmb_nr = 0;
1178                         block_length = 0;
1179                         flipped_block = 0;
1180                         last_symbol = l->s;
1181                         continue;
1182                 }
1183
1184                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1185                         last_symbol = l->s;
1186                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1187                         continue;
1188                 }
1189
1190                 if (o->color_moved_ws_handling &
1191                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1192                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1193                 else
1194                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1195
1196                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1197
1198                 if (pmb_nr == 0) {
1199                         /*
1200                          * The current line is the start of a new block.
1201                          * Setup the set of potential blocks.
1202                          */
1203                         hashmap_for_each_entry_from(hm, match, ent) {
1204                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1205                                 if (o->color_moved_ws_handling &
1206                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1207                                         if (compute_ws_delta(l, match->es,
1208                                                              &pmb[pmb_nr].wsd))
1209                                                 pmb[pmb_nr++].match = match;
1210                                 } else {
1211                                         pmb[pmb_nr].wsd = 0;
1212                                         pmb[pmb_nr++].match = match;
1213                                 }
1214                         }
1215
1216                         if (adjust_last_block(o, n, block_length) &&
1217                             pmb_nr && last_symbol != l->s)
1218                                 flipped_block = (flipped_block + 1) % 2;
1219                         else
1220                                 flipped_block = 0;
1221
1222                         block_length = 0;
1223                 }
1224
1225                 if (pmb_nr) {
1226                         block_length++;
1227                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1228                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1229                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1230                 }
1231                 last_symbol = l->s;
1232         }
1233         adjust_last_block(o, n, block_length);
1234
1235         for(n = 0; n < pmb_nr; n++)
1236                 moved_block_clear(&pmb[n]);
1237         free(pmb);
1238 }
1239
1240 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1241   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1242 static void dim_moved_lines(struct diff_options *o)
1243 {
1244         int n;
1245         for (n = 0; n < o->emitted_symbols->nr; n++) {
1246                 struct emitted_diff_symbol *prev = (n != 0) ?
1247                                 &o->emitted_symbols->buf[n - 1] : NULL;
1248                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1249                 struct emitted_diff_symbol *next =
1250                                 (n < o->emitted_symbols->nr - 1) ?
1251                                 &o->emitted_symbols->buf[n + 1] : NULL;
1252
1253                 /* Not a plus or minus line? */
1254                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1255                         continue;
1256
1257                 /* Not a moved line? */
1258                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1259                         continue;
1260
1261                 /*
1262                  * If prev or next are not a plus or minus line,
1263                  * pretend they don't exist
1264                  */
1265                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1266                             prev->s != DIFF_SYMBOL_MINUS)
1267                         prev = NULL;
1268                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1269                             next->s != DIFF_SYMBOL_MINUS)
1270                         next = NULL;
1271
1272                 /* Inside a block? */
1273                 if ((prev &&
1274                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1275                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1276                     (next &&
1277                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1278                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1279                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1280                         continue;
1281                 }
1282
1283                 /* Check if we are at an interesting bound: */
1284                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1285                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1286                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1287                         continue;
1288                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1289                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1290                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1291                         continue;
1292
1293                 /*
1294                  * The boundary to prev and next are not interesting,
1295                  * so this line is not interesting as a whole
1296                  */
1297                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1298         }
1299 }
1300
1301 static void emit_line_ws_markup(struct diff_options *o,
1302                                 const char *set_sign, const char *set,
1303                                 const char *reset,
1304                                 int sign_index, const char *line, int len,
1305                                 unsigned ws_rule, int blank_at_eof)
1306 {
1307         const char *ws = NULL;
1308         int sign = o->output_indicators[sign_index];
1309
1310         if (o->ws_error_highlight & ws_rule) {
1311                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1312                 if (!*ws)
1313                         ws = NULL;
1314         }
1315
1316         if (!ws && !set_sign)
1317                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1318         else if (!ws) {
1319                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1320         } else if (blank_at_eof)
1321                 /* Blank line at EOF - paint '+' as well */
1322                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1323         else {
1324                 /* Emit just the prefix, then the rest. */
1325                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1326                             sign, "", 0);
1327                 ws_check_emit(line, len, ws_rule,
1328                               o->file, set, reset, ws);
1329         }
1330 }
1331
1332 static void emit_diff_symbol_from_struct(struct diff_options *o,
1333                                          struct emitted_diff_symbol *eds)
1334 {
1335         static const char *nneof = " No newline at end of file\n";
1336         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1337         struct strbuf sb = STRBUF_INIT;
1338
1339         enum diff_symbol s = eds->s;
1340         const char *line = eds->line;
1341         int len = eds->len;
1342         unsigned flags = eds->flags;
1343
1344         switch (s) {
1345         case DIFF_SYMBOL_NO_LF_EOF:
1346                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1347                 reset = diff_get_color_opt(o, DIFF_RESET);
1348                 putc('\n', o->file);
1349                 emit_line_0(o, context, NULL, 0, reset, '\\',
1350                             nneof, strlen(nneof));
1351                 break;
1352         case DIFF_SYMBOL_SUBMODULE_HEADER:
1353         case DIFF_SYMBOL_SUBMODULE_ERROR:
1354         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1355         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1356         case DIFF_SYMBOL_SUMMARY:
1357         case DIFF_SYMBOL_STATS_LINE:
1358         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1359         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1360                 emit_line(o, "", "", line, len);
1361                 break;
1362         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1363         case DIFF_SYMBOL_CONTEXT_MARKER:
1364                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1365                 reset = diff_get_color_opt(o, DIFF_RESET);
1366                 emit_line(o, context, reset, line, len);
1367                 break;
1368         case DIFF_SYMBOL_SEPARATOR:
1369                 fprintf(o->file, "%s%c",
1370                         diff_line_prefix(o),
1371                         o->line_termination);
1372                 break;
1373         case DIFF_SYMBOL_CONTEXT:
1374                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1375                 reset = diff_get_color_opt(o, DIFF_RESET);
1376                 set_sign = NULL;
1377                 if (o->flags.dual_color_diffed_diffs) {
1378                         char c = !len ? 0 : line[0];
1379
1380                         if (c == '+')
1381                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1382                         else if (c == '@')
1383                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1384                         else if (c == '-')
1385                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1386                 }
1387                 emit_line_ws_markup(o, set_sign, set, reset,
1388                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1389                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1390                 break;
1391         case DIFF_SYMBOL_PLUS:
1392                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1393                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1394                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1395                 case DIFF_SYMBOL_MOVED_LINE |
1396                      DIFF_SYMBOL_MOVED_LINE_ALT |
1397                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1398                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1399                         break;
1400                 case DIFF_SYMBOL_MOVED_LINE |
1401                      DIFF_SYMBOL_MOVED_LINE_ALT:
1402                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1403                         break;
1404                 case DIFF_SYMBOL_MOVED_LINE |
1405                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1406                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1407                         break;
1408                 case DIFF_SYMBOL_MOVED_LINE:
1409                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1410                         break;
1411                 default:
1412                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1413                 }
1414                 reset = diff_get_color_opt(o, DIFF_RESET);
1415                 if (!o->flags.dual_color_diffed_diffs)
1416                         set_sign = NULL;
1417                 else {
1418                         char c = !len ? 0 : line[0];
1419
1420                         set_sign = set;
1421                         if (c == '-')
1422                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1423                         else if (c == '@')
1424                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1425                         else if (c == '+')
1426                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1427                         else
1428                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1429                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1430                 }
1431                 emit_line_ws_markup(o, set_sign, set, reset,
1432                                     OUTPUT_INDICATOR_NEW, line, len,
1433                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1434                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1435                 break;
1436         case DIFF_SYMBOL_MINUS:
1437                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1438                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1439                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1440                 case DIFF_SYMBOL_MOVED_LINE |
1441                      DIFF_SYMBOL_MOVED_LINE_ALT |
1442                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1443                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1444                         break;
1445                 case DIFF_SYMBOL_MOVED_LINE |
1446                      DIFF_SYMBOL_MOVED_LINE_ALT:
1447                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1448                         break;
1449                 case DIFF_SYMBOL_MOVED_LINE |
1450                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1451                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1452                         break;
1453                 case DIFF_SYMBOL_MOVED_LINE:
1454                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1455                         break;
1456                 default:
1457                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1458                 }
1459                 reset = diff_get_color_opt(o, DIFF_RESET);
1460                 if (!o->flags.dual_color_diffed_diffs)
1461                         set_sign = NULL;
1462                 else {
1463                         char c = !len ? 0 : line[0];
1464
1465                         set_sign = set;
1466                         if (c == '+')
1467                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1468                         else if (c == '@')
1469                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1470                         else if (c == '-')
1471                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1472                         else
1473                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1474                 }
1475                 emit_line_ws_markup(o, set_sign, set, reset,
1476                                     OUTPUT_INDICATOR_OLD, line, len,
1477                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1478                 break;
1479         case DIFF_SYMBOL_WORDS_PORCELAIN:
1480                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1481                 reset = diff_get_color_opt(o, DIFF_RESET);
1482                 emit_line(o, context, reset, line, len);
1483                 fputs("~\n", o->file);
1484                 break;
1485         case DIFF_SYMBOL_WORDS:
1486                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1487                 reset = diff_get_color_opt(o, DIFF_RESET);
1488                 /*
1489                  * Skip the prefix character, if any.  With
1490                  * diff_suppress_blank_empty, there may be
1491                  * none.
1492                  */
1493                 if (line[0] != '\n') {
1494                         line++;
1495                         len--;
1496                 }
1497                 emit_line(o, context, reset, line, len);
1498                 break;
1499         case DIFF_SYMBOL_FILEPAIR_PLUS:
1500                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1501                 reset = diff_get_color_opt(o, DIFF_RESET);
1502                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1503                         line, reset,
1504                         strchr(line, ' ') ? "\t" : "");
1505                 break;
1506         case DIFF_SYMBOL_FILEPAIR_MINUS:
1507                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1508                 reset = diff_get_color_opt(o, DIFF_RESET);
1509                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1510                         line, reset,
1511                         strchr(line, ' ') ? "\t" : "");
1512                 break;
1513         case DIFF_SYMBOL_BINARY_FILES:
1514         case DIFF_SYMBOL_HEADER:
1515                 fprintf(o->file, "%s", line);
1516                 break;
1517         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1518                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1519                 break;
1520         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1521                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1522                 break;
1523         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1524                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1525                 break;
1526         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1527                 fputs(diff_line_prefix(o), o->file);
1528                 fputc('\n', o->file);
1529                 break;
1530         case DIFF_SYMBOL_REWRITE_DIFF:
1531                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1532                 reset = diff_get_color_opt(o, DIFF_RESET);
1533                 emit_line(o, fraginfo, reset, line, len);
1534                 break;
1535         case DIFF_SYMBOL_SUBMODULE_ADD:
1536                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1537                 reset = diff_get_color_opt(o, DIFF_RESET);
1538                 emit_line(o, set, reset, line, len);
1539                 break;
1540         case DIFF_SYMBOL_SUBMODULE_DEL:
1541                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1542                 reset = diff_get_color_opt(o, DIFF_RESET);
1543                 emit_line(o, set, reset, line, len);
1544                 break;
1545         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1546                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1547                         diff_line_prefix(o), line);
1548                 break;
1549         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1550                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1551                         diff_line_prefix(o), line);
1552                 break;
1553         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1554                 emit_line(o, "", "", " 0 files changed\n",
1555                           strlen(" 0 files changed\n"));
1556                 break;
1557         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1558                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1559                 break;
1560         case DIFF_SYMBOL_WORD_DIFF:
1561                 fprintf(o->file, "%.*s", len, line);
1562                 break;
1563         case DIFF_SYMBOL_STAT_SEP:
1564                 fputs(o->stat_sep, o->file);
1565                 break;
1566         default:
1567                 BUG("unknown diff symbol");
1568         }
1569         strbuf_release(&sb);
1570 }
1571
1572 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1573                              const char *line, int len, unsigned flags)
1574 {
1575         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1576
1577         if (o->emitted_symbols)
1578                 append_emitted_diff_symbol(o, &e);
1579         else
1580                 emit_diff_symbol_from_struct(o, &e);
1581 }
1582
1583 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1584 {
1585         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1586 }
1587
1588 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1589 {
1590         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1591 }
1592
1593 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1594 {
1595         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1596                          path, strlen(path), 0);
1597 }
1598
1599 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1600 {
1601         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1602                          path, strlen(path), 0);
1603 }
1604
1605 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1606 {
1607         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1608                          header, strlen(header), 0);
1609 }
1610
1611 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1612 {
1613         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1614 }
1615
1616 void diff_emit_submodule_pipethrough(struct diff_options *o,
1617                                      const char *line, int len)
1618 {
1619         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1620 }
1621
1622 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1623 {
1624         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1625               ecbdata->blank_at_eof_in_preimage &&
1626               ecbdata->blank_at_eof_in_postimage &&
1627               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1628               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1629                 return 0;
1630         return ws_blank_line(line, len, ecbdata->ws_rule);
1631 }
1632
1633 static void emit_add_line(struct emit_callback *ecbdata,
1634                           const char *line, int len)
1635 {
1636         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1637         if (new_blank_line_at_eof(ecbdata, line, len))
1638                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1639
1640         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1641 }
1642
1643 static void emit_del_line(struct emit_callback *ecbdata,
1644                           const char *line, int len)
1645 {
1646         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1647         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1648 }
1649
1650 static void emit_context_line(struct emit_callback *ecbdata,
1651                               const char *line, int len)
1652 {
1653         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1654         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1655 }
1656
1657 static void emit_hunk_header(struct emit_callback *ecbdata,
1658                              const char *line, int len)
1659 {
1660         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1661         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1662         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1663         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1664         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1665         static const char atat[2] = { '@', '@' };
1666         const char *cp, *ep;
1667         struct strbuf msgbuf = STRBUF_INIT;
1668         int org_len = len;
1669         int i = 1;
1670
1671         /*
1672          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1673          * it always is at least 10 bytes long.
1674          */
1675         if (len < 10 ||
1676             memcmp(line, atat, 2) ||
1677             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1678                 emit_diff_symbol(ecbdata->opt,
1679                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1680                 return;
1681         }
1682         ep += 2; /* skip over @@ */
1683
1684         /* The hunk header in fraginfo color */
1685         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1686                 strbuf_addstr(&msgbuf, reverse);
1687         strbuf_addstr(&msgbuf, frag);
1688         if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1689                 strbuf_add(&msgbuf, atat, sizeof(atat));
1690         else
1691                 strbuf_add(&msgbuf, line, ep - line);
1692         strbuf_addstr(&msgbuf, reset);
1693
1694         /*
1695          * trailing "\r\n"
1696          */
1697         for ( ; i < 3; i++)
1698                 if (line[len - i] == '\r' || line[len - i] == '\n')
1699                         len--;
1700
1701         /* blank before the func header */
1702         for (cp = ep; ep - line < len; ep++)
1703                 if (*ep != ' ' && *ep != '\t')
1704                         break;
1705         if (ep != cp) {
1706                 strbuf_addstr(&msgbuf, context);
1707                 strbuf_add(&msgbuf, cp, ep - cp);
1708                 strbuf_addstr(&msgbuf, reset);
1709         }
1710
1711         if (ep < line + len) {
1712                 strbuf_addstr(&msgbuf, func);
1713                 strbuf_add(&msgbuf, ep, line + len - ep);
1714                 strbuf_addstr(&msgbuf, reset);
1715         }
1716
1717         strbuf_add(&msgbuf, line + len, org_len - len);
1718         strbuf_complete_line(&msgbuf);
1719         emit_diff_symbol(ecbdata->opt,
1720                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1721         strbuf_release(&msgbuf);
1722 }
1723
1724 static struct diff_tempfile *claim_diff_tempfile(void)
1725 {
1726         int i;
1727         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1728                 if (!diff_temp[i].name)
1729                         return diff_temp + i;
1730         BUG("diff is failing to clean up its tempfiles");
1731 }
1732
1733 static void remove_tempfile(void)
1734 {
1735         int i;
1736         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1737                 if (is_tempfile_active(diff_temp[i].tempfile))
1738                         delete_tempfile(&diff_temp[i].tempfile);
1739                 diff_temp[i].name = NULL;
1740         }
1741 }
1742
1743 static void add_line_count(struct strbuf *out, int count)
1744 {
1745         switch (count) {
1746         case 0:
1747                 strbuf_addstr(out, "0,0");
1748                 break;
1749         case 1:
1750                 strbuf_addstr(out, "1");
1751                 break;
1752         default:
1753                 strbuf_addf(out, "1,%d", count);
1754                 break;
1755         }
1756 }
1757
1758 static void emit_rewrite_lines(struct emit_callback *ecb,
1759                                int prefix, const char *data, int size)
1760 {
1761         const char *endp = NULL;
1762
1763         while (0 < size) {
1764                 int len;
1765
1766                 endp = memchr(data, '\n', size);
1767                 len = endp ? (endp - data + 1) : size;
1768                 if (prefix != '+') {
1769                         ecb->lno_in_preimage++;
1770                         emit_del_line(ecb, data, len);
1771                 } else {
1772                         ecb->lno_in_postimage++;
1773                         emit_add_line(ecb, data, len);
1774                 }
1775                 size -= len;
1776                 data += len;
1777         }
1778         if (!endp)
1779                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1780 }
1781
1782 static void emit_rewrite_diff(const char *name_a,
1783                               const char *name_b,
1784                               struct diff_filespec *one,
1785                               struct diff_filespec *two,
1786                               struct userdiff_driver *textconv_one,
1787                               struct userdiff_driver *textconv_two,
1788                               struct diff_options *o)
1789 {
1790         int lc_a, lc_b;
1791         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1792         const char *a_prefix, *b_prefix;
1793         char *data_one, *data_two;
1794         size_t size_one, size_two;
1795         struct emit_callback ecbdata;
1796         struct strbuf out = STRBUF_INIT;
1797
1798         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1799                 a_prefix = o->b_prefix;
1800                 b_prefix = o->a_prefix;
1801         } else {
1802                 a_prefix = o->a_prefix;
1803                 b_prefix = o->b_prefix;
1804         }
1805
1806         name_a += (*name_a == '/');
1807         name_b += (*name_b == '/');
1808
1809         strbuf_reset(&a_name);
1810         strbuf_reset(&b_name);
1811         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1812         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1813
1814         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1815         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1816
1817         memset(&ecbdata, 0, sizeof(ecbdata));
1818         ecbdata.color_diff = want_color(o->use_color);
1819         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1820         ecbdata.opt = o;
1821         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1822                 mmfile_t mf1, mf2;
1823                 mf1.ptr = (char *)data_one;
1824                 mf2.ptr = (char *)data_two;
1825                 mf1.size = size_one;
1826                 mf2.size = size_two;
1827                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1828         }
1829         ecbdata.lno_in_preimage = 1;
1830         ecbdata.lno_in_postimage = 1;
1831
1832         lc_a = count_lines(data_one, size_one);
1833         lc_b = count_lines(data_two, size_two);
1834
1835         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1836                          a_name.buf, a_name.len, 0);
1837         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1838                          b_name.buf, b_name.len, 0);
1839
1840         strbuf_addstr(&out, "@@ -");
1841         if (!o->irreversible_delete)
1842                 add_line_count(&out, lc_a);
1843         else
1844                 strbuf_addstr(&out, "?,?");
1845         strbuf_addstr(&out, " +");
1846         add_line_count(&out, lc_b);
1847         strbuf_addstr(&out, " @@\n");
1848         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1849         strbuf_release(&out);
1850
1851         if (lc_a && !o->irreversible_delete)
1852                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1853         if (lc_b)
1854                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1855         if (textconv_one)
1856                 free((char *)data_one);
1857         if (textconv_two)
1858                 free((char *)data_two);
1859 }
1860
1861 struct diff_words_buffer {
1862         mmfile_t text;
1863         unsigned long alloc;
1864         struct diff_words_orig {
1865                 const char *begin, *end;
1866         } *orig;
1867         int orig_nr, orig_alloc;
1868 };
1869
1870 static void diff_words_append(char *line, unsigned long len,
1871                 struct diff_words_buffer *buffer)
1872 {
1873         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1874         line++;
1875         len--;
1876         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1877         buffer->text.size += len;
1878         buffer->text.ptr[buffer->text.size] = '\0';
1879 }
1880
1881 struct diff_words_style_elem {
1882         const char *prefix;
1883         const char *suffix;
1884         const char *color; /* NULL; filled in by the setup code if
1885                             * color is enabled */
1886 };
1887
1888 struct diff_words_style {
1889         enum diff_words_type type;
1890         struct diff_words_style_elem new_word, old_word, ctx;
1891         const char *newline;
1892 };
1893
1894 static struct diff_words_style diff_words_styles[] = {
1895         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1896         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1897         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1898 };
1899
1900 struct diff_words_data {
1901         struct diff_words_buffer minus, plus;
1902         const char *current_plus;
1903         int last_minus;
1904         struct diff_options *opt;
1905         regex_t *word_regex;
1906         enum diff_words_type type;
1907         struct diff_words_style *style;
1908 };
1909
1910 static int fn_out_diff_words_write_helper(struct diff_options *o,
1911                                           struct diff_words_style_elem *st_el,
1912                                           const char *newline,
1913                                           size_t count, const char *buf)
1914 {
1915         int print = 0;
1916         struct strbuf sb = STRBUF_INIT;
1917
1918         while (count) {
1919                 char *p = memchr(buf, '\n', count);
1920                 if (print)
1921                         strbuf_addstr(&sb, diff_line_prefix(o));
1922
1923                 if (p != buf) {
1924                         const char *reset = st_el->color && *st_el->color ?
1925                                             GIT_COLOR_RESET : NULL;
1926                         if (st_el->color && *st_el->color)
1927                                 strbuf_addstr(&sb, st_el->color);
1928                         strbuf_addstr(&sb, st_el->prefix);
1929                         strbuf_add(&sb, buf, p ? p - buf : count);
1930                         strbuf_addstr(&sb, st_el->suffix);
1931                         if (reset)
1932                                 strbuf_addstr(&sb, reset);
1933                 }
1934                 if (!p)
1935                         goto out;
1936
1937                 strbuf_addstr(&sb, newline);
1938                 count -= p + 1 - buf;
1939                 buf = p + 1;
1940                 print = 1;
1941                 if (count) {
1942                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1943                                          sb.buf, sb.len, 0);
1944                         strbuf_reset(&sb);
1945                 }
1946         }
1947
1948 out:
1949         if (sb.len)
1950                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1951                                  sb.buf, sb.len, 0);
1952         strbuf_release(&sb);
1953         return 0;
1954 }
1955
1956 /*
1957  * '--color-words' algorithm can be described as:
1958  *
1959  *   1. collect the minus/plus lines of a diff hunk, divided into
1960  *      minus-lines and plus-lines;
1961  *
1962  *   2. break both minus-lines and plus-lines into words and
1963  *      place them into two mmfile_t with one word for each line;
1964  *
1965  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1966  *
1967  * And for the common parts of the both file, we output the plus side text.
1968  * diff_words->current_plus is used to trace the current position of the plus file
1969  * which printed. diff_words->last_minus is used to trace the last minus word
1970  * printed.
1971  *
1972  * For '--graph' to work with '--color-words', we need to output the graph prefix
1973  * on each line of color words output. Generally, there are two conditions on
1974  * which we should output the prefix.
1975  *
1976  *   1. diff_words->last_minus == 0 &&
1977  *      diff_words->current_plus == diff_words->plus.text.ptr
1978  *
1979  *      that is: the plus text must start as a new line, and if there is no minus
1980  *      word printed, a graph prefix must be printed.
1981  *
1982  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1983  *      *(diff_words->current_plus - 1) == '\n'
1984  *
1985  *      that is: a graph prefix must be printed following a '\n'
1986  */
1987 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1988 {
1989         if ((diff_words->last_minus == 0 &&
1990                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1991                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1992                 *(diff_words->current_plus - 1) == '\n')) {
1993                 return 1;
1994         } else {
1995                 return 0;
1996         }
1997 }
1998
1999 static void fn_out_diff_words_aux(void *priv,
2000                                   long minus_first, long minus_len,
2001                                   long plus_first, long plus_len,
2002                                   const char *func, long funclen)
2003 {
2004         struct diff_words_data *diff_words = priv;
2005         struct diff_words_style *style = diff_words->style;
2006         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
2007         struct diff_options *opt = diff_words->opt;
2008         const char *line_prefix;
2009
2010         assert(opt);
2011         line_prefix = diff_line_prefix(opt);
2012
2013         /* POSIX requires that first be decremented by one if len == 0... */
2014         if (minus_len) {
2015                 minus_begin = diff_words->minus.orig[minus_first].begin;
2016                 minus_end =
2017                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2018         } else
2019                 minus_begin = minus_end =
2020                         diff_words->minus.orig[minus_first].end;
2021
2022         if (plus_len) {
2023                 plus_begin = diff_words->plus.orig[plus_first].begin;
2024                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2025         } else
2026                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2027
2028         if (color_words_output_graph_prefix(diff_words)) {
2029                 fputs(line_prefix, diff_words->opt->file);
2030         }
2031         if (diff_words->current_plus != plus_begin) {
2032                 fn_out_diff_words_write_helper(diff_words->opt,
2033                                 &style->ctx, style->newline,
2034                                 plus_begin - diff_words->current_plus,
2035                                 diff_words->current_plus);
2036         }
2037         if (minus_begin != minus_end) {
2038                 fn_out_diff_words_write_helper(diff_words->opt,
2039                                 &style->old_word, style->newline,
2040                                 minus_end - minus_begin, minus_begin);
2041         }
2042         if (plus_begin != plus_end) {
2043                 fn_out_diff_words_write_helper(diff_words->opt,
2044                                 &style->new_word, style->newline,
2045                                 plus_end - plus_begin, plus_begin);
2046         }
2047
2048         diff_words->current_plus = plus_end;
2049         diff_words->last_minus = minus_first;
2050 }
2051
2052 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2053 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2054                 int *begin, int *end)
2055 {
2056         if (word_regex && *begin < buffer->size) {
2057                 regmatch_t match[1];
2058                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2059                                  buffer->size - *begin, 1, match, 0)) {
2060                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2061                                         '\n', match[0].rm_eo - match[0].rm_so);
2062                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2063                         *begin += match[0].rm_so;
2064                         return *begin >= *end;
2065                 }
2066                 return -1;
2067         }
2068
2069         /* find the next word */
2070         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2071                 (*begin)++;
2072         if (*begin >= buffer->size)
2073                 return -1;
2074
2075         /* find the end of the word */
2076         *end = *begin + 1;
2077         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2078                 (*end)++;
2079
2080         return 0;
2081 }
2082
2083 /*
2084  * This function splits the words in buffer->text, stores the list with
2085  * newline separator into out, and saves the offsets of the original words
2086  * in buffer->orig.
2087  */
2088 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2089                 regex_t *word_regex)
2090 {
2091         int i, j;
2092         long alloc = 0;
2093
2094         out->size = 0;
2095         out->ptr = NULL;
2096
2097         /* fake an empty "0th" word */
2098         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2099         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2100         buffer->orig_nr = 1;
2101
2102         for (i = 0; i < buffer->text.size; i++) {
2103                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2104                         return;
2105
2106                 /* store original boundaries */
2107                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2108                                 buffer->orig_alloc);
2109                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2110                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2111                 buffer->orig_nr++;
2112
2113                 /* store one word */
2114                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2115                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2116                 out->ptr[out->size + j - i] = '\n';
2117                 out->size += j - i + 1;
2118
2119                 i = j - 1;
2120         }
2121 }
2122
2123 /* this executes the word diff on the accumulated buffers */
2124 static void diff_words_show(struct diff_words_data *diff_words)
2125 {
2126         xpparam_t xpp;
2127         xdemitconf_t xecfg;
2128         mmfile_t minus, plus;
2129         struct diff_words_style *style = diff_words->style;
2130
2131         struct diff_options *opt = diff_words->opt;
2132         const char *line_prefix;
2133
2134         assert(opt);
2135         line_prefix = diff_line_prefix(opt);
2136
2137         /* special case: only removal */
2138         if (!diff_words->plus.text.size) {
2139                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2140                                  line_prefix, strlen(line_prefix), 0);
2141                 fn_out_diff_words_write_helper(diff_words->opt,
2142                         &style->old_word, style->newline,
2143                         diff_words->minus.text.size,
2144                         diff_words->minus.text.ptr);
2145                 diff_words->minus.text.size = 0;
2146                 return;
2147         }
2148
2149         diff_words->current_plus = diff_words->plus.text.ptr;
2150         diff_words->last_minus = 0;
2151
2152         memset(&xpp, 0, sizeof(xpp));
2153         memset(&xecfg, 0, sizeof(xecfg));
2154         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2155         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2156         xpp.flags = 0;
2157         /* as only the hunk header will be parsed, we need a 0-context */
2158         xecfg.ctxlen = 0;
2159         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2160                           diff_words, &xpp, &xecfg))
2161                 die("unable to generate word diff");
2162         free(minus.ptr);
2163         free(plus.ptr);
2164         if (diff_words->current_plus != diff_words->plus.text.ptr +
2165                         diff_words->plus.text.size) {
2166                 if (color_words_output_graph_prefix(diff_words))
2167                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2168                                          line_prefix, strlen(line_prefix), 0);
2169                 fn_out_diff_words_write_helper(diff_words->opt,
2170                         &style->ctx, style->newline,
2171                         diff_words->plus.text.ptr + diff_words->plus.text.size
2172                         - diff_words->current_plus, diff_words->current_plus);
2173         }
2174         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2175 }
2176
2177 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2178 static void diff_words_flush(struct emit_callback *ecbdata)
2179 {
2180         struct diff_options *wo = ecbdata->diff_words->opt;
2181
2182         if (ecbdata->diff_words->minus.text.size ||
2183             ecbdata->diff_words->plus.text.size)
2184                 diff_words_show(ecbdata->diff_words);
2185
2186         if (wo->emitted_symbols) {
2187                 struct diff_options *o = ecbdata->opt;
2188                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2189                 int i;
2190
2191                 /*
2192                  * NEEDSWORK:
2193                  * Instead of appending each, concat all words to a line?
2194                  */
2195                 for (i = 0; i < wol->nr; i++)
2196                         append_emitted_diff_symbol(o, &wol->buf[i]);
2197
2198                 for (i = 0; i < wol->nr; i++)
2199                         free((void *)wol->buf[i].line);
2200
2201                 wol->nr = 0;
2202         }
2203 }
2204
2205 static void diff_filespec_load_driver(struct diff_filespec *one,
2206                                       struct index_state *istate)
2207 {
2208         /* Use already-loaded driver */
2209         if (one->driver)
2210                 return;
2211
2212         if (S_ISREG(one->mode))
2213                 one->driver = userdiff_find_by_path(istate, one->path);
2214
2215         /* Fallback to default settings */
2216         if (!one->driver)
2217                 one->driver = userdiff_find_by_name("default");
2218 }
2219
2220 static const char *userdiff_word_regex(struct diff_filespec *one,
2221                                        struct index_state *istate)
2222 {
2223         diff_filespec_load_driver(one, istate);
2224         return one->driver->word_regex;
2225 }
2226
2227 static void init_diff_words_data(struct emit_callback *ecbdata,
2228                                  struct diff_options *orig_opts,
2229                                  struct diff_filespec *one,
2230                                  struct diff_filespec *two)
2231 {
2232         int i;
2233         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2234         memcpy(o, orig_opts, sizeof(struct diff_options));
2235
2236         ecbdata->diff_words =
2237                 xcalloc(1, sizeof(struct diff_words_data));
2238         ecbdata->diff_words->type = o->word_diff;
2239         ecbdata->diff_words->opt = o;
2240
2241         if (orig_opts->emitted_symbols)
2242                 o->emitted_symbols =
2243                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2244
2245         if (!o->word_regex)
2246                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2247         if (!o->word_regex)
2248                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2249         if (!o->word_regex)
2250                 o->word_regex = diff_word_regex_cfg;
2251         if (o->word_regex) {
2252                 ecbdata->diff_words->word_regex = (regex_t *)
2253                         xmalloc(sizeof(regex_t));
2254                 if (regcomp(ecbdata->diff_words->word_regex,
2255                             o->word_regex,
2256                             REG_EXTENDED | REG_NEWLINE))
2257                         die("invalid regular expression: %s",
2258                             o->word_regex);
2259         }
2260         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2261                 if (o->word_diff == diff_words_styles[i].type) {
2262                         ecbdata->diff_words->style =
2263                                 &diff_words_styles[i];
2264                         break;
2265                 }
2266         }
2267         if (want_color(o->use_color)) {
2268                 struct diff_words_style *st = ecbdata->diff_words->style;
2269                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2270                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2271                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2272         }
2273 }
2274
2275 static void free_diff_words_data(struct emit_callback *ecbdata)
2276 {
2277         if (ecbdata->diff_words) {
2278                 diff_words_flush(ecbdata);
2279                 free (ecbdata->diff_words->opt->emitted_symbols);
2280                 free (ecbdata->diff_words->opt);
2281                 free (ecbdata->diff_words->minus.text.ptr);
2282                 free (ecbdata->diff_words->minus.orig);
2283                 free (ecbdata->diff_words->plus.text.ptr);
2284                 free (ecbdata->diff_words->plus.orig);
2285                 if (ecbdata->diff_words->word_regex) {
2286                         regfree(ecbdata->diff_words->word_regex);
2287                         free(ecbdata->diff_words->word_regex);
2288                 }
2289                 FREE_AND_NULL(ecbdata->diff_words);
2290         }
2291 }
2292
2293 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2294 {
2295         if (want_color(diff_use_color))
2296                 return diff_colors[ix];
2297         return "";
2298 }
2299
2300 const char *diff_line_prefix(struct diff_options *opt)
2301 {
2302         struct strbuf *msgbuf;
2303         if (!opt->output_prefix)
2304                 return "";
2305
2306         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2307         return msgbuf->buf;
2308 }
2309
2310 static unsigned long sane_truncate_line(char *line, unsigned long len)
2311 {
2312         const char *cp;
2313         unsigned long allot;
2314         size_t l = len;
2315
2316         cp = line;
2317         allot = l;
2318         while (0 < l) {
2319                 (void) utf8_width(&cp, &l);
2320                 if (!cp)
2321                         break; /* truncated in the middle? */
2322         }
2323         return allot - l;
2324 }
2325
2326 static void find_lno(const char *line, struct emit_callback *ecbdata)
2327 {
2328         const char *p;
2329         ecbdata->lno_in_preimage = 0;
2330         ecbdata->lno_in_postimage = 0;
2331         p = strchr(line, '-');
2332         if (!p)
2333                 return; /* cannot happen */
2334         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2335         p = strchr(p, '+');
2336         if (!p)
2337                 return; /* cannot happen */
2338         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2339 }
2340
2341 static void fn_out_consume(void *priv, char *line, unsigned long len)
2342 {
2343         struct emit_callback *ecbdata = priv;
2344         struct diff_options *o = ecbdata->opt;
2345
2346         o->found_changes = 1;
2347
2348         if (ecbdata->header) {
2349                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2350                                  ecbdata->header->buf, ecbdata->header->len, 0);
2351                 strbuf_reset(ecbdata->header);
2352                 ecbdata->header = NULL;
2353         }
2354
2355         if (ecbdata->label_path[0]) {
2356                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2357                                  ecbdata->label_path[0],
2358                                  strlen(ecbdata->label_path[0]), 0);
2359                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2360                                  ecbdata->label_path[1],
2361                                  strlen(ecbdata->label_path[1]), 0);
2362                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2363         }
2364
2365         if (diff_suppress_blank_empty
2366             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2367                 line[0] = '\n';
2368                 len = 1;
2369         }
2370
2371         if (line[0] == '@') {
2372                 if (ecbdata->diff_words)
2373                         diff_words_flush(ecbdata);
2374                 len = sane_truncate_line(line, len);
2375                 find_lno(line, ecbdata);
2376                 emit_hunk_header(ecbdata, line, len);
2377                 return;
2378         }
2379
2380         if (ecbdata->diff_words) {
2381                 enum diff_symbol s =
2382                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2383                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2384                 if (line[0] == '-') {
2385                         diff_words_append(line, len,
2386                                           &ecbdata->diff_words->minus);
2387                         return;
2388                 } else if (line[0] == '+') {
2389                         diff_words_append(line, len,
2390                                           &ecbdata->diff_words->plus);
2391                         return;
2392                 } else if (starts_with(line, "\\ ")) {
2393                         /*
2394                          * Eat the "no newline at eof" marker as if we
2395                          * saw a "+" or "-" line with nothing on it,
2396                          * and return without diff_words_flush() to
2397                          * defer processing. If this is the end of
2398                          * preimage, more "+" lines may come after it.
2399                          */
2400                         return;
2401                 }
2402                 diff_words_flush(ecbdata);
2403                 emit_diff_symbol(o, s, line, len, 0);
2404                 return;
2405         }
2406
2407         switch (line[0]) {
2408         case '+':
2409                 ecbdata->lno_in_postimage++;
2410                 emit_add_line(ecbdata, line + 1, len - 1);
2411                 break;
2412         case '-':
2413                 ecbdata->lno_in_preimage++;
2414                 emit_del_line(ecbdata, line + 1, len - 1);
2415                 break;
2416         case ' ':
2417                 ecbdata->lno_in_postimage++;
2418                 ecbdata->lno_in_preimage++;
2419                 emit_context_line(ecbdata, line + 1, len - 1);
2420                 break;
2421         default:
2422                 /* incomplete line at the end */
2423                 ecbdata->lno_in_preimage++;
2424                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2425                                  line, len, 0);
2426                 break;
2427         }
2428 }
2429
2430 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2431 {
2432         const char *old_name = a;
2433         const char *new_name = b;
2434         int pfx_length, sfx_length;
2435         int pfx_adjust_for_slash;
2436         int len_a = strlen(a);
2437         int len_b = strlen(b);
2438         int a_midlen, b_midlen;
2439         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2440         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2441
2442         if (qlen_a || qlen_b) {
2443                 quote_c_style(a, name, NULL, 0);
2444                 strbuf_addstr(name, " => ");
2445                 quote_c_style(b, name, NULL, 0);
2446                 return;
2447         }
2448
2449         /* Find common prefix */
2450         pfx_length = 0;
2451         while (*old_name && *new_name && *old_name == *new_name) {
2452                 if (*old_name == '/')
2453                         pfx_length = old_name - a + 1;
2454                 old_name++;
2455                 new_name++;
2456         }
2457
2458         /* Find common suffix */
2459         old_name = a + len_a;
2460         new_name = b + len_b;
2461         sfx_length = 0;
2462         /*
2463          * If there is a common prefix, it must end in a slash.  In
2464          * that case we let this loop run 1 into the prefix to see the
2465          * same slash.
2466          *
2467          * If there is no common prefix, we cannot do this as it would
2468          * underrun the input strings.
2469          */
2470         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2471         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2472                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2473                *old_name == *new_name) {
2474                 if (*old_name == '/')
2475                         sfx_length = len_a - (old_name - a);
2476                 old_name--;
2477                 new_name--;
2478         }
2479
2480         /*
2481          * pfx{mid-a => mid-b}sfx
2482          * {pfx-a => pfx-b}sfx
2483          * pfx{sfx-a => sfx-b}
2484          * name-a => name-b
2485          */
2486         a_midlen = len_a - pfx_length - sfx_length;
2487         b_midlen = len_b - pfx_length - sfx_length;
2488         if (a_midlen < 0)
2489                 a_midlen = 0;
2490         if (b_midlen < 0)
2491                 b_midlen = 0;
2492
2493         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2494         if (pfx_length + sfx_length) {
2495                 strbuf_add(name, a, pfx_length);
2496                 strbuf_addch(name, '{');
2497         }
2498         strbuf_add(name, a + pfx_length, a_midlen);
2499         strbuf_addstr(name, " => ");
2500         strbuf_add(name, b + pfx_length, b_midlen);
2501         if (pfx_length + sfx_length) {
2502                 strbuf_addch(name, '}');
2503                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2504         }
2505 }
2506
2507 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2508                                           const char *name_a,
2509                                           const char *name_b)
2510 {
2511         struct diffstat_file *x;
2512         x = xcalloc(1, sizeof(*x));
2513         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2514         diffstat->files[diffstat->nr++] = x;
2515         if (name_b) {
2516                 x->from_name = xstrdup(name_a);
2517                 x->name = xstrdup(name_b);
2518                 x->is_renamed = 1;
2519         }
2520         else {
2521                 x->from_name = NULL;
2522                 x->name = xstrdup(name_a);
2523         }
2524         return x;
2525 }
2526
2527 static void diffstat_consume(void *priv, char *line, unsigned long len)
2528 {
2529         struct diffstat_t *diffstat = priv;
2530         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2531
2532         if (line[0] == '+')
2533                 x->added++;
2534         else if (line[0] == '-')
2535                 x->deleted++;
2536 }
2537
2538 const char mime_boundary_leader[] = "------------";
2539
2540 static int scale_linear(int it, int width, int max_change)
2541 {
2542         if (!it)
2543                 return 0;
2544         /*
2545          * make sure that at least one '-' or '+' is printed if
2546          * there is any change to this path. The easiest way is to
2547          * scale linearly as if the allotted width is one column shorter
2548          * than it is, and then add 1 to the result.
2549          */
2550         return 1 + (it * (width - 1) / max_change);
2551 }
2552
2553 static void show_graph(struct strbuf *out, char ch, int cnt,
2554                        const char *set, const char *reset)
2555 {
2556         if (cnt <= 0)
2557                 return;
2558         strbuf_addstr(out, set);
2559         strbuf_addchars(out, ch, cnt);
2560         strbuf_addstr(out, reset);
2561 }
2562
2563 static void fill_print_name(struct diffstat_file *file)
2564 {
2565         struct strbuf pname = STRBUF_INIT;
2566
2567         if (file->print_name)
2568                 return;
2569
2570         if (file->is_renamed)
2571                 pprint_rename(&pname, file->from_name, file->name);
2572         else
2573                 quote_c_style(file->name, &pname, NULL, 0);
2574
2575         if (file->comments)
2576                 strbuf_addf(&pname, " (%s)", file->comments);
2577
2578         file->print_name = strbuf_detach(&pname, NULL);
2579 }
2580
2581 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2582                 int files, int insertions, int deletions)
2583 {
2584         struct strbuf sb = STRBUF_INIT;
2585
2586         if (!files) {
2587                 assert(insertions == 0 && deletions == 0);
2588                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2589                                  NULL, 0, 0);
2590                 return;
2591         }
2592
2593         strbuf_addf(&sb,
2594                     (files == 1) ? " %d file changed" : " %d files changed",
2595                     files);
2596
2597         /*
2598          * For binary diff, the caller may want to print "x files
2599          * changed" with insertions == 0 && deletions == 0.
2600          *
2601          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2602          * is probably less confusing (i.e skip over "2 files changed
2603          * but nothing about added/removed lines? Is this a bug in Git?").
2604          */
2605         if (insertions || deletions == 0) {
2606                 strbuf_addf(&sb,
2607                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2608                             insertions);
2609         }
2610
2611         if (deletions || insertions == 0) {
2612                 strbuf_addf(&sb,
2613                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2614                             deletions);
2615         }
2616         strbuf_addch(&sb, '\n');
2617         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2618                          sb.buf, sb.len, 0);
2619         strbuf_release(&sb);
2620 }
2621
2622 void print_stat_summary(FILE *fp, int files,
2623                         int insertions, int deletions)
2624 {
2625         struct diff_options o;
2626         memset(&o, 0, sizeof(o));
2627         o.file = fp;
2628
2629         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2630 }
2631
2632 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2633 {
2634         int i, len, add, del, adds = 0, dels = 0;
2635         uintmax_t max_change = 0, max_len = 0;
2636         int total_files = data->nr, count;
2637         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2638         const char *reset, *add_c, *del_c;
2639         int extra_shown = 0;
2640         const char *line_prefix = diff_line_prefix(options);
2641         struct strbuf out = STRBUF_INIT;
2642
2643         if (data->nr == 0)
2644                 return;
2645
2646         count = options->stat_count ? options->stat_count : data->nr;
2647
2648         reset = diff_get_color_opt(options, DIFF_RESET);
2649         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2650         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2651
2652         /*
2653          * Find the longest filename and max number of changes
2654          */
2655         for (i = 0; (i < count) && (i < data->nr); i++) {
2656                 struct diffstat_file *file = data->files[i];
2657                 uintmax_t change = file->added + file->deleted;
2658
2659                 if (!file->is_interesting && (change == 0)) {
2660                         count++; /* not shown == room for one more */
2661                         continue;
2662                 }
2663                 fill_print_name(file);
2664                 len = strlen(file->print_name);
2665                 if (max_len < len)
2666                         max_len = len;
2667
2668                 if (file->is_unmerged) {
2669                         /* "Unmerged" is 8 characters */
2670                         bin_width = bin_width < 8 ? 8 : bin_width;
2671                         continue;
2672                 }
2673                 if (file->is_binary) {
2674                         /* "Bin XXX -> YYY bytes" */
2675                         int w = 14 + decimal_width(file->added)
2676                                 + decimal_width(file->deleted);
2677                         bin_width = bin_width < w ? w : bin_width;
2678                         /* Display change counts aligned with "Bin" */
2679                         number_width = 3;
2680                         continue;
2681                 }
2682
2683                 if (max_change < change)
2684                         max_change = change;
2685         }
2686         count = i; /* where we can stop scanning in data->files[] */
2687
2688         /*
2689          * We have width = stat_width or term_columns() columns total.
2690          * We want a maximum of min(max_len, stat_name_width) for the name part.
2691          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2692          * We also need 1 for " " and 4 + decimal_width(max_change)
2693          * for " | NNNN " and one the empty column at the end, altogether
2694          * 6 + decimal_width(max_change).
2695          *
2696          * If there's not enough space, we will use the smaller of
2697          * stat_name_width (if set) and 5/8*width for the filename,
2698          * and the rest for constant elements + graph part, but no more
2699          * than stat_graph_width for the graph part.
2700          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2701          * for the standard terminal size).
2702          *
2703          * In other words: stat_width limits the maximum width, and
2704          * stat_name_width fixes the maximum width of the filename,
2705          * and is also used to divide available columns if there
2706          * aren't enough.
2707          *
2708          * Binary files are displayed with "Bin XXX -> YYY bytes"
2709          * instead of the change count and graph. This part is treated
2710          * similarly to the graph part, except that it is not
2711          * "scaled". If total width is too small to accommodate the
2712          * guaranteed minimum width of the filename part and the
2713          * separators and this message, this message will "overflow"
2714          * making the line longer than the maximum width.
2715          */
2716
2717         if (options->stat_width == -1)
2718                 width = term_columns() - strlen(line_prefix);
2719         else
2720                 width = options->stat_width ? options->stat_width : 80;
2721         number_width = decimal_width(max_change) > number_width ?
2722                 decimal_width(max_change) : number_width;
2723
2724         if (options->stat_graph_width == -1)
2725                 options->stat_graph_width = diff_stat_graph_width;
2726
2727         /*
2728          * Guarantee 3/8*16==6 for the graph part
2729          * and 5/8*16==10 for the filename part
2730          */
2731         if (width < 16 + 6 + number_width)
2732                 width = 16 + 6 + number_width;
2733
2734         /*
2735          * First assign sizes that are wanted, ignoring available width.
2736          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2737          * starting from "XXX" should fit in graph_width.
2738          */
2739         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2740         if (options->stat_graph_width &&
2741             options->stat_graph_width < graph_width)
2742                 graph_width = options->stat_graph_width;
2743
2744         name_width = (options->stat_name_width > 0 &&
2745                       options->stat_name_width < max_len) ?
2746                 options->stat_name_width : max_len;
2747
2748         /*
2749          * Adjust adjustable widths not to exceed maximum width
2750          */
2751         if (name_width + number_width + 6 + graph_width > width) {
2752                 if (graph_width > width * 3/8 - number_width - 6) {
2753                         graph_width = width * 3/8 - number_width - 6;
2754                         if (graph_width < 6)
2755                                 graph_width = 6;
2756                 }
2757
2758                 if (options->stat_graph_width &&
2759                     graph_width > options->stat_graph_width)
2760                         graph_width = options->stat_graph_width;
2761                 if (name_width > width - number_width - 6 - graph_width)
2762                         name_width = width - number_width - 6 - graph_width;
2763                 else
2764                         graph_width = width - number_width - 6 - name_width;
2765         }
2766
2767         /*
2768          * From here name_width is the width of the name area,
2769          * and graph_width is the width of the graph area.
2770          * max_change is used to scale graph properly.
2771          */
2772         for (i = 0; i < count; i++) {
2773                 const char *prefix = "";
2774                 struct diffstat_file *file = data->files[i];
2775                 char *name = file->print_name;
2776                 uintmax_t added = file->added;
2777                 uintmax_t deleted = file->deleted;
2778                 int name_len;
2779
2780                 if (!file->is_interesting && (added + deleted == 0))
2781                         continue;
2782
2783                 /*
2784                  * "scale" the filename
2785                  */
2786                 len = name_width;
2787                 name_len = strlen(name);
2788                 if (name_width < name_len) {
2789                         char *slash;
2790                         prefix = "...";
2791                         len -= 3;
2792                         name += name_len - len;
2793                         slash = strchr(name, '/');
2794                         if (slash)
2795                                 name = slash;
2796                 }
2797
2798                 if (file->is_binary) {
2799                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2800                         strbuf_addf(&out, " %*s", number_width, "Bin");
2801                         if (!added && !deleted) {
2802                                 strbuf_addch(&out, '\n');
2803                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2804                                                  out.buf, out.len, 0);
2805                                 strbuf_reset(&out);
2806                                 continue;
2807                         }
2808                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2809                                 del_c, deleted, reset);
2810                         strbuf_addstr(&out, " -> ");
2811                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2812                                 add_c, added, reset);
2813                         strbuf_addstr(&out, " bytes\n");
2814                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2815                                          out.buf, out.len, 0);
2816                         strbuf_reset(&out);
2817                         continue;
2818                 }
2819                 else if (file->is_unmerged) {
2820                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2821                         strbuf_addstr(&out, " Unmerged\n");
2822                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2823                                          out.buf, out.len, 0);
2824                         strbuf_reset(&out);
2825                         continue;
2826                 }
2827
2828                 /*
2829                  * scale the add/delete
2830                  */
2831                 add = added;
2832                 del = deleted;
2833
2834                 if (graph_width <= max_change) {
2835                         int total = scale_linear(add + del, graph_width, max_change);
2836                         if (total < 2 && add && del)
2837                                 /* width >= 2 due to the sanity check */
2838                                 total = 2;
2839                         if (add < del) {
2840                                 add = scale_linear(add, graph_width, max_change);
2841                                 del = total - add;
2842                         } else {
2843                                 del = scale_linear(del, graph_width, max_change);
2844                                 add = total - del;
2845                         }
2846                 }
2847                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2848                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2849                         number_width, added + deleted,
2850                         added + deleted ? " " : "");
2851                 show_graph(&out, '+', add, add_c, reset);
2852                 show_graph(&out, '-', del, del_c, reset);
2853                 strbuf_addch(&out, '\n');
2854                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2855                                  out.buf, out.len, 0);
2856                 strbuf_reset(&out);
2857         }
2858
2859         for (i = 0; i < data->nr; i++) {
2860                 struct diffstat_file *file = data->files[i];
2861                 uintmax_t added = file->added;
2862                 uintmax_t deleted = file->deleted;
2863
2864                 if (file->is_unmerged ||
2865                     (!file->is_interesting && (added + deleted == 0))) {
2866                         total_files--;
2867                         continue;
2868                 }
2869
2870                 if (!file->is_binary) {
2871                         adds += added;
2872                         dels += deleted;
2873                 }
2874                 if (i < count)
2875                         continue;
2876                 if (!extra_shown)
2877                         emit_diff_symbol(options,
2878                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2879                                          NULL, 0, 0);
2880                 extra_shown = 1;
2881         }
2882
2883         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2884         strbuf_release(&out);
2885 }
2886
2887 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2888 {
2889         int i, adds = 0, dels = 0, total_files = data->nr;
2890
2891         if (data->nr == 0)
2892                 return;
2893
2894         for (i = 0; i < data->nr; i++) {
2895                 int added = data->files[i]->added;
2896                 int deleted = data->files[i]->deleted;
2897
2898                 if (data->files[i]->is_unmerged ||
2899                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2900                         total_files--;
2901                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2902                         adds += added;
2903                         dels += deleted;
2904                 }
2905         }
2906         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2907 }
2908
2909 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2910 {
2911         int i;
2912
2913         if (data->nr == 0)
2914                 return;
2915
2916         for (i = 0; i < data->nr; i++) {
2917                 struct diffstat_file *file = data->files[i];
2918
2919                 fprintf(options->file, "%s", diff_line_prefix(options));
2920
2921                 if (file->is_binary)
2922                         fprintf(options->file, "-\t-\t");
2923                 else
2924                         fprintf(options->file,
2925                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2926                                 file->added, file->deleted);
2927                 if (options->line_termination) {
2928                         fill_print_name(file);
2929                         if (!file->is_renamed)
2930                                 write_name_quoted(file->name, options->file,
2931                                                   options->line_termination);
2932                         else {
2933                                 fputs(file->print_name, options->file);
2934                                 putc(options->line_termination, options->file);
2935                         }
2936                 } else {
2937                         if (file->is_renamed) {
2938                                 putc('\0', options->file);
2939                                 write_name_quoted(file->from_name, options->file, '\0');
2940                         }
2941                         write_name_quoted(file->name, options->file, '\0');
2942                 }
2943         }
2944 }
2945
2946 struct dirstat_file {
2947         const char *name;
2948         unsigned long changed;
2949 };
2950
2951 struct dirstat_dir {
2952         struct dirstat_file *files;
2953         int alloc, nr, permille, cumulative;
2954 };
2955
2956 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2957                 unsigned long changed, const char *base, int baselen)
2958 {
2959         unsigned long sum_changes = 0;
2960         unsigned int sources = 0;
2961         const char *line_prefix = diff_line_prefix(opt);
2962
2963         while (dir->nr) {
2964                 struct dirstat_file *f = dir->files;
2965                 int namelen = strlen(f->name);
2966                 unsigned long changes;
2967                 char *slash;
2968
2969                 if (namelen < baselen)
2970                         break;
2971                 if (memcmp(f->name, base, baselen))
2972                         break;
2973                 slash = strchr(f->name + baselen, '/');
2974                 if (slash) {
2975                         int newbaselen = slash + 1 - f->name;
2976                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2977                         sources++;
2978                 } else {
2979                         changes = f->changed;
2980                         dir->files++;
2981                         dir->nr--;
2982                         sources += 2;
2983                 }
2984                 sum_changes += changes;
2985         }
2986
2987         /*
2988          * We don't report dirstat's for
2989          *  - the top level
2990          *  - or cases where everything came from a single directory
2991          *    under this directory (sources == 1).
2992          */
2993         if (baselen && sources != 1) {
2994                 if (sum_changes) {
2995                         int permille = sum_changes * 1000 / changed;
2996                         if (permille >= dir->permille) {
2997                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2998                                         permille / 10, permille % 10, baselen, base);
2999                                 if (!dir->cumulative)
3000                                         return 0;
3001                         }
3002                 }
3003         }
3004         return sum_changes;
3005 }
3006
3007 static int dirstat_compare(const void *_a, const void *_b)
3008 {
3009         const struct dirstat_file *a = _a;
3010         const struct dirstat_file *b = _b;
3011         return strcmp(a->name, b->name);
3012 }
3013
3014 static void show_dirstat(struct diff_options *options)
3015 {
3016         int i;
3017         unsigned long changed;
3018         struct dirstat_dir dir;
3019         struct diff_queue_struct *q = &diff_queued_diff;
3020
3021         dir.files = NULL;
3022         dir.alloc = 0;
3023         dir.nr = 0;
3024         dir.permille = options->dirstat_permille;
3025         dir.cumulative = options->flags.dirstat_cumulative;
3026
3027         changed = 0;
3028         for (i = 0; i < q->nr; i++) {
3029                 struct diff_filepair *p = q->queue[i];
3030                 const char *name;
3031                 unsigned long copied, added, damage;
3032                 struct diff_populate_filespec_options dpf_options = {
3033                         .check_size_only = 1,
3034                 };
3035
3036                 name = p->two->path ? p->two->path : p->one->path;
3037
3038                 if (p->one->oid_valid && p->two->oid_valid &&
3039                     oideq(&p->one->oid, &p->two->oid)) {
3040                         /*
3041                          * The SHA1 has not changed, so pre-/post-content is
3042                          * identical. We can therefore skip looking at the
3043                          * file contents altogether.
3044                          */
3045                         damage = 0;
3046                         goto found_damage;
3047                 }
3048
3049                 if (options->flags.dirstat_by_file) {
3050                         /*
3051                          * In --dirstat-by-file mode, we don't really need to
3052                          * look at the actual file contents at all.
3053                          * The fact that the SHA1 changed is enough for us to
3054                          * add this file to the list of results
3055                          * (with each file contributing equal damage).
3056                          */
3057                         damage = 1;
3058                         goto found_damage;
3059                 }
3060
3061                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3062                         diff_populate_filespec(options->repo, p->one, NULL);
3063                         diff_populate_filespec(options->repo, p->two, NULL);
3064                         diffcore_count_changes(options->repo,
3065                                                p->one, p->two, NULL, NULL,
3066                                                &copied, &added);
3067                         diff_free_filespec_data(p->one);
3068                         diff_free_filespec_data(p->two);
3069                 } else if (DIFF_FILE_VALID(p->one)) {
3070                         diff_populate_filespec(options->repo, p->one, &dpf_options);
3071                         copied = added = 0;
3072                         diff_free_filespec_data(p->one);
3073                 } else if (DIFF_FILE_VALID(p->two)) {
3074                         diff_populate_filespec(options->repo, p->two, &dpf_options);
3075                         copied = 0;
3076                         added = p->two->size;
3077                         diff_free_filespec_data(p->two);
3078                 } else
3079                         continue;
3080
3081                 /*
3082                  * Original minus copied is the removed material,
3083                  * added is the new material.  They are both damages
3084                  * made to the preimage.
3085                  * If the resulting damage is zero, we know that
3086                  * diffcore_count_changes() considers the two entries to
3087                  * be identical, but since the oid changed, we
3088                  * know that there must have been _some_ kind of change,
3089                  * so we force all entries to have damage > 0.
3090                  */
3091                 damage = (p->one->size - copied) + added;
3092                 if (!damage)
3093                         damage = 1;
3094
3095 found_damage:
3096                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3097                 dir.files[dir.nr].name = name;
3098                 dir.files[dir.nr].changed = damage;
3099                 changed += damage;
3100                 dir.nr++;
3101         }
3102
3103         /* This can happen even with many files, if everything was renames */
3104         if (!changed)
3105                 return;
3106
3107         /* Show all directories with more than x% of the changes */
3108         QSORT(dir.files, dir.nr, dirstat_compare);
3109         gather_dirstat(options, &dir, changed, "", 0);
3110 }
3111
3112 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3113 {
3114         int i;
3115         unsigned long changed;
3116         struct dirstat_dir dir;
3117
3118         if (data->nr == 0)
3119                 return;
3120
3121         dir.files = NULL;
3122         dir.alloc = 0;
3123         dir.nr = 0;
3124         dir.permille = options->dirstat_permille;
3125         dir.cumulative = options->flags.dirstat_cumulative;
3126
3127         changed = 0;
3128         for (i = 0; i < data->nr; i++) {
3129                 struct diffstat_file *file = data->files[i];
3130                 unsigned long damage = file->added + file->deleted;
3131                 if (file->is_binary)
3132                         /*
3133                          * binary files counts bytes, not lines. Must find some
3134                          * way to normalize binary bytes vs. textual lines.
3135                          * The following heuristic assumes that there are 64
3136                          * bytes per "line".
3137                          * This is stupid and ugly, but very cheap...
3138                          */
3139                         damage = DIV_ROUND_UP(damage, 64);
3140                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3141                 dir.files[dir.nr].name = file->name;
3142                 dir.files[dir.nr].changed = damage;
3143                 changed += damage;
3144                 dir.nr++;
3145         }
3146
3147         /* This can happen even with many files, if everything was renames */
3148         if (!changed)
3149                 return;
3150
3151         /* Show all directories with more than x% of the changes */
3152         QSORT(dir.files, dir.nr, dirstat_compare);
3153         gather_dirstat(options, &dir, changed, "", 0);
3154 }
3155
3156 static void free_diffstat_file(struct diffstat_file *f)
3157 {
3158         free(f->print_name);
3159         free(f->name);
3160         free(f->from_name);
3161         free(f);
3162 }
3163
3164 void free_diffstat_info(struct diffstat_t *diffstat)
3165 {
3166         int i;
3167         for (i = 0; i < diffstat->nr; i++)
3168                 free_diffstat_file(diffstat->files[i]);
3169         free(diffstat->files);
3170 }
3171
3172 struct checkdiff_t {
3173         const char *filename;
3174         int lineno;
3175         int conflict_marker_size;
3176         struct diff_options *o;
3177         unsigned ws_rule;
3178         unsigned status;
3179 };
3180
3181 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3182 {
3183         char firstchar;
3184         int cnt;
3185
3186         if (len < marker_size + 1)
3187                 return 0;
3188         firstchar = line[0];
3189         switch (firstchar) {
3190         case '=': case '>': case '<': case '|':
3191                 break;
3192         default:
3193                 return 0;
3194         }
3195         for (cnt = 1; cnt < marker_size; cnt++)
3196                 if (line[cnt] != firstchar)
3197                         return 0;
3198         /* line[1] through line[marker_size-1] are same as firstchar */
3199         if (len < marker_size + 1 || !isspace(line[marker_size]))
3200                 return 0;
3201         return 1;
3202 }
3203
3204 static void checkdiff_consume_hunk(void *priv,
3205                                    long ob, long on, long nb, long nn,
3206                                    const char *func, long funclen)
3207
3208 {
3209         struct checkdiff_t *data = priv;
3210         data->lineno = nb - 1;
3211 }
3212
3213 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3214 {
3215         struct checkdiff_t *data = priv;
3216         int marker_size = data->conflict_marker_size;
3217         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3218         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3219         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3220         char *err;
3221         const char *line_prefix;
3222
3223         assert(data->o);
3224         line_prefix = diff_line_prefix(data->o);
3225
3226         if (line[0] == '+') {
3227                 unsigned bad;
3228                 data->lineno++;
3229                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3230                         data->status |= 1;
3231                         fprintf(data->o->file,
3232                                 "%s%s:%d: leftover conflict marker\n",
3233                                 line_prefix, data->filename, data->lineno);
3234                 }
3235                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3236                 if (!bad)
3237                         return;
3238                 data->status |= bad;
3239                 err = whitespace_error_string(bad);
3240                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3241                         line_prefix, data->filename, data->lineno, err);
3242                 free(err);
3243                 emit_line(data->o, set, reset, line, 1);
3244                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3245                               data->o->file, set, reset, ws);
3246         } else if (line[0] == ' ') {
3247                 data->lineno++;
3248         }
3249 }
3250
3251 static unsigned char *deflate_it(char *data,
3252                                  unsigned long size,
3253                                  unsigned long *result_size)
3254 {
3255         int bound;
3256         unsigned char *deflated;
3257         git_zstream stream;
3258
3259         git_deflate_init(&stream, zlib_compression_level);
3260         bound = git_deflate_bound(&stream, size);
3261         deflated = xmalloc(bound);
3262         stream.next_out = deflated;
3263         stream.avail_out = bound;
3264
3265         stream.next_in = (unsigned char *)data;
3266         stream.avail_in = size;
3267         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3268                 ; /* nothing */
3269         git_deflate_end(&stream);
3270         *result_size = stream.total_out;
3271         return deflated;
3272 }
3273
3274 static void emit_binary_diff_body(struct diff_options *o,
3275                                   mmfile_t *one, mmfile_t *two)
3276 {
3277         void *cp;
3278         void *delta;
3279         void *deflated;
3280         void *data;
3281         unsigned long orig_size;
3282         unsigned long delta_size;
3283         unsigned long deflate_size;
3284         unsigned long data_size;
3285
3286         /* We could do deflated delta, or we could do just deflated two,
3287          * whichever is smaller.
3288          */
3289         delta = NULL;
3290         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3291         if (one->size && two->size) {
3292                 delta = diff_delta(one->ptr, one->size,
3293                                    two->ptr, two->size,
3294                                    &delta_size, deflate_size);
3295                 if (delta) {
3296                         void *to_free = delta;
3297                         orig_size = delta_size;
3298                         delta = deflate_it(delta, delta_size, &delta_size);
3299                         free(to_free);
3300                 }
3301         }
3302
3303         if (delta && delta_size < deflate_size) {
3304                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3305                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3306                                  s, strlen(s), 0);
3307                 free(s);
3308                 free(deflated);
3309                 data = delta;
3310                 data_size = delta_size;
3311         } else {
3312                 char *s = xstrfmt("%lu", two->size);
3313                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3314                                  s, strlen(s), 0);
3315                 free(s);
3316                 free(delta);
3317                 data = deflated;
3318                 data_size = deflate_size;
3319         }
3320
3321         /* emit data encoded in base85 */
3322         cp = data;
3323         while (data_size) {
3324                 int len;
3325                 int bytes = (52 < data_size) ? 52 : data_size;
3326                 char line[71];
3327                 data_size -= bytes;
3328                 if (bytes <= 26)
3329                         line[0] = bytes + 'A' - 1;
3330                 else
3331                         line[0] = bytes - 26 + 'a' - 1;
3332                 encode_85(line + 1, cp, bytes);
3333                 cp = (char *) cp + bytes;
3334
3335                 len = strlen(line);
3336                 line[len++] = '\n';
3337                 line[len] = '\0';
3338
3339                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3340                                  line, len, 0);
3341         }
3342         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3343         free(data);
3344 }
3345
3346 static void emit_binary_diff(struct diff_options *o,
3347                              mmfile_t *one, mmfile_t *two)
3348 {
3349         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3350         emit_binary_diff_body(o, one, two);
3351         emit_binary_diff_body(o, two, one);
3352 }
3353
3354 int diff_filespec_is_binary(struct repository *r,
3355                             struct diff_filespec *one)
3356 {
3357         struct diff_populate_filespec_options dpf_options = {
3358                 .check_binary = 1,
3359         };
3360
3361         if (one->is_binary == -1) {
3362                 diff_filespec_load_driver(one, r->index);
3363                 if (one->driver->binary != -1)
3364                         one->is_binary = one->driver->binary;
3365                 else {
3366                         if (!one->data && DIFF_FILE_VALID(one))
3367                                 diff_populate_filespec(r, one, &dpf_options);
3368                         if (one->is_binary == -1 && one->data)
3369                                 one->is_binary = buffer_is_binary(one->data,
3370                                                 one->size);
3371                         if (one->is_binary == -1)
3372                                 one->is_binary = 0;
3373                 }
3374         }
3375         return one->is_binary;
3376 }
3377
3378 static const struct userdiff_funcname *
3379 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3380 {
3381         diff_filespec_load_driver(one, o->repo->index);
3382         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3383 }
3384
3385 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3386 {
3387         if (!options->a_prefix)
3388                 options->a_prefix = a;
3389         if (!options->b_prefix)
3390                 options->b_prefix = b;
3391 }
3392
3393 struct userdiff_driver *get_textconv(struct repository *r,
3394                                      struct diff_filespec *one)
3395 {
3396         if (!DIFF_FILE_VALID(one))
3397                 return NULL;
3398
3399         diff_filespec_load_driver(one, r->index);
3400         return userdiff_get_textconv(r, one->driver);
3401 }
3402
3403 static void builtin_diff(const char *name_a,
3404                          const char *name_b,
3405                          struct diff_filespec *one,
3406                          struct diff_filespec *two,
3407                          const char *xfrm_msg,
3408                          int must_show_header,
3409                          struct diff_options *o,
3410                          int complete_rewrite)
3411 {
3412         mmfile_t mf1, mf2;
3413         const char *lbl[2];
3414         char *a_one, *b_two;
3415         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3416         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3417         const char *a_prefix, *b_prefix;
3418         struct userdiff_driver *textconv_one = NULL;
3419         struct userdiff_driver *textconv_two = NULL;
3420         struct strbuf header = STRBUF_INIT;
3421         const char *line_prefix = diff_line_prefix(o);
3422
3423         diff_set_mnemonic_prefix(o, "a/", "b/");
3424         if (o->flags.reverse_diff) {
3425                 a_prefix = o->b_prefix;
3426                 b_prefix = o->a_prefix;
3427         } else {
3428                 a_prefix = o->a_prefix;
3429                 b_prefix = o->b_prefix;
3430         }
3431
3432         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3433             (!one->mode || S_ISGITLINK(one->mode)) &&
3434             (!two->mode || S_ISGITLINK(two->mode))) {
3435                 show_submodule_diff_summary(o, one->path ? one->path : two->path,
3436                                 &one->oid, &two->oid,
3437                                 two->dirty_submodule);
3438                 return;
3439         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3440                    (!one->mode || S_ISGITLINK(one->mode)) &&
3441                    (!two->mode || S_ISGITLINK(two->mode))) {
3442                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3443                                 &one->oid, &two->oid,
3444                                 two->dirty_submodule);
3445                 return;
3446         }
3447
3448         if (o->flags.allow_textconv) {
3449                 textconv_one = get_textconv(o->repo, one);
3450                 textconv_two = get_textconv(o->repo, two);
3451         }
3452
3453         /* Never use a non-valid filename anywhere if at all possible */
3454         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3455         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3456
3457         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3458         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3459         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3460         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3461         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3462         if (lbl[0][0] == '/') {
3463                 /* /dev/null */
3464                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3465                 if (xfrm_msg)
3466                         strbuf_addstr(&header, xfrm_msg);
3467                 must_show_header = 1;
3468         }
3469         else if (lbl[1][0] == '/') {
3470                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3471                 if (xfrm_msg)
3472                         strbuf_addstr(&header, xfrm_msg);
3473                 must_show_header = 1;
3474         }
3475         else {
3476                 if (one->mode != two->mode) {
3477                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3478                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3479                         must_show_header = 1;
3480                 }
3481                 if (xfrm_msg)
3482                         strbuf_addstr(&header, xfrm_msg);
3483
3484                 /*
3485                  * we do not run diff between different kind
3486                  * of objects.
3487                  */
3488                 if ((one->mode ^ two->mode) & S_IFMT)
3489                         goto free_ab_and_return;
3490                 if (complete_rewrite &&
3491                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3492                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3493                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3494                                          header.buf, header.len, 0);
3495                         strbuf_reset(&header);
3496                         emit_rewrite_diff(name_a, name_b, one, two,
3497                                           textconv_one, textconv_two, o);
3498                         o->found_changes = 1;
3499                         goto free_ab_and_return;
3500                 }
3501         }
3502
3503         if (o->irreversible_delete && lbl[1][0] == '/') {
3504                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3505                                  header.len, 0);
3506                 strbuf_reset(&header);
3507                 goto free_ab_and_return;
3508         } else if (!o->flags.text &&
3509                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3510                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3511                 struct strbuf sb = STRBUF_INIT;
3512                 if (!one->data && !two->data &&
3513                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3514                     !o->flags.binary) {
3515                         if (oideq(&one->oid, &two->oid)) {
3516                                 if (must_show_header)
3517                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3518                                                          header.buf, header.len,
3519                                                          0);
3520                                 goto free_ab_and_return;
3521                         }
3522                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3523                                          header.buf, header.len, 0);
3524                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3525                                     diff_line_prefix(o), lbl[0], lbl[1]);
3526                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3527                                          sb.buf, sb.len, 0);
3528                         strbuf_release(&sb);
3529                         goto free_ab_and_return;
3530                 }
3531                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3532                     fill_mmfile(o->repo, &mf2, two) < 0)
3533                         die("unable to read files to diff");
3534                 /* Quite common confusing case */
3535                 if (mf1.size == mf2.size &&
3536                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3537                         if (must_show_header)
3538                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3539                                                  header.buf, header.len, 0);
3540                         goto free_ab_and_return;
3541                 }
3542                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3543                 strbuf_reset(&header);
3544                 if (o->flags.binary)
3545                         emit_binary_diff(o, &mf1, &mf2);
3546                 else {
3547                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3548                                     diff_line_prefix(o), lbl[0], lbl[1]);
3549                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3550                                          sb.buf, sb.len, 0);
3551                         strbuf_release(&sb);
3552                 }
3553                 o->found_changes = 1;
3554         } else {
3555                 /* Crazy xdl interfaces.. */
3556                 const char *diffopts;
3557                 const char *v;
3558                 xpparam_t xpp;
3559                 xdemitconf_t xecfg;
3560                 struct emit_callback ecbdata;
3561                 const struct userdiff_funcname *pe;
3562
3563                 if (must_show_header) {
3564                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3565                                          header.buf, header.len, 0);
3566                         strbuf_reset(&header);
3567                 }
3568
3569                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3570                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3571
3572                 pe = diff_funcname_pattern(o, one);
3573                 if (!pe)
3574                         pe = diff_funcname_pattern(o, two);
3575
3576                 memset(&xpp, 0, sizeof(xpp));
3577                 memset(&xecfg, 0, sizeof(xecfg));
3578                 memset(&ecbdata, 0, sizeof(ecbdata));
3579                 if (o->flags.suppress_diff_headers)
3580                         lbl[0] = NULL;
3581                 ecbdata.label_path = lbl;
3582                 ecbdata.color_diff = want_color(o->use_color);
3583                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3584                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3585                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3586                 ecbdata.opt = o;
3587                 if (header.len && !o->flags.suppress_diff_headers)
3588                         ecbdata.header = &header;
3589                 xpp.flags = o->xdl_opts;
3590                 xpp.ignore_regex = o->ignore_regex;
3591                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3592                 xpp.anchors = o->anchors;
3593                 xpp.anchors_nr = o->anchors_nr;
3594                 xecfg.ctxlen = o->context;
3595                 xecfg.interhunkctxlen = o->interhunkcontext;
3596                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3597                 if (o->flags.funccontext)
3598                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3599                 if (pe)
3600                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3601
3602                 diffopts = getenv("GIT_DIFF_OPTS");
3603                 if (!diffopts)
3604                         ;
3605                 else if (skip_prefix(diffopts, "--unified=", &v))
3606                         xecfg.ctxlen = strtoul(v, NULL, 10);
3607                 else if (skip_prefix(diffopts, "-u", &v))
3608                         xecfg.ctxlen = strtoul(v, NULL, 10);
3609
3610                 if (o->word_diff)
3611                         init_diff_words_data(&ecbdata, o, one, two);
3612                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3613                                   &ecbdata, &xpp, &xecfg))
3614                         die("unable to generate diff for %s", one->path);
3615                 if (o->word_diff)
3616                         free_diff_words_data(&ecbdata);
3617                 if (textconv_one)
3618                         free(mf1.ptr);
3619                 if (textconv_two)
3620                         free(mf2.ptr);
3621                 xdiff_clear_find_func(&xecfg);
3622         }
3623
3624  free_ab_and_return:
3625         strbuf_release(&header);
3626         diff_free_filespec_data(one);
3627         diff_free_filespec_data(two);
3628         free(a_one);
3629         free(b_two);
3630         return;
3631 }
3632
3633 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3634 {
3635         if (!is_renamed) {
3636                 if (p->status == DIFF_STATUS_ADDED) {
3637                         if (S_ISLNK(p->two->mode))
3638                                 return "new +l";
3639                         else if ((p->two->mode & 0777) == 0755)
3640                                 return "new +x";
3641                         else
3642                                 return "new";
3643                 } else if (p->status == DIFF_STATUS_DELETED)
3644                         return "gone";
3645         }
3646         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3647                 return "mode -l";
3648         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3649                 return "mode +l";
3650         else if ((p->one->mode & 0777) == 0644 &&
3651                  (p->two->mode & 0777) == 0755)
3652                 return "mode +x";
3653         else if ((p->one->mode & 0777) == 0755 &&
3654                  (p->two->mode & 0777) == 0644)
3655                 return "mode -x";
3656         return NULL;
3657 }
3658
3659 static void builtin_diffstat(const char *name_a, const char *name_b,
3660                              struct diff_filespec *one,
3661                              struct diff_filespec *two,
3662                              struct diffstat_t *diffstat,
3663                              struct diff_options *o,
3664                              struct diff_filepair *p)
3665 {
3666         mmfile_t mf1, mf2;
3667         struct diffstat_file *data;
3668         int may_differ;
3669         int complete_rewrite = 0;
3670
3671         if (!DIFF_PAIR_UNMERGED(p)) {
3672                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3673                         complete_rewrite = 1;
3674         }
3675
3676         data = diffstat_add(diffstat, name_a, name_b);
3677         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3678         if (o->flags.stat_with_summary)
3679                 data->comments = get_compact_summary(p, data->is_renamed);
3680
3681         if (!one || !two) {
3682                 data->is_unmerged = 1;
3683                 return;
3684         }
3685
3686         /* saves some reads if true, not a guarantee of diff outcome */
3687         may_differ = !(one->oid_valid && two->oid_valid &&
3688                         oideq(&one->oid, &two->oid));
3689
3690         if (diff_filespec_is_binary(o->repo, one) ||
3691             diff_filespec_is_binary(o->repo, two)) {
3692                 data->is_binary = 1;
3693                 if (!may_differ) {
3694                         data->added = 0;
3695                         data->deleted = 0;
3696                 } else {
3697                         data->added = diff_filespec_size(o->repo, two);
3698                         data->deleted = diff_filespec_size(o->repo, one);
3699                 }
3700         }
3701
3702         else if (complete_rewrite) {
3703                 diff_populate_filespec(o->repo, one, NULL);
3704                 diff_populate_filespec(o->repo, two, NULL);
3705                 data->deleted = count_lines(one->data, one->size);
3706                 data->added = count_lines(two->data, two->size);
3707         }
3708
3709         else if (may_differ) {
3710                 /* Crazy xdl interfaces.. */
3711                 xpparam_t xpp;
3712                 xdemitconf_t xecfg;
3713
3714                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3715                     fill_mmfile(o->repo, &mf2, two) < 0)
3716                         die("unable to read files to diff");
3717
3718                 memset(&xpp, 0, sizeof(xpp));
3719                 memset(&xecfg, 0, sizeof(xecfg));
3720                 xpp.flags = o->xdl_opts;
3721                 xpp.ignore_regex = o->ignore_regex;
3722                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3723                 xpp.anchors = o->anchors;
3724                 xpp.anchors_nr = o->anchors_nr;
3725                 xecfg.ctxlen = o->context;
3726                 xecfg.interhunkctxlen = o->interhunkcontext;
3727                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3728                                   diffstat_consume, diffstat, &xpp, &xecfg))
3729                         die("unable to generate diffstat for %s", one->path);
3730
3731                 if (DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two)) {
3732                         struct diffstat_file *file =
3733                                 diffstat->files[diffstat->nr - 1];
3734                         /*
3735                          * Omit diffstats of modified files where nothing changed.
3736                          * Even if may_differ, this might be the case due to
3737                          * ignoring whitespace changes, etc.
3738                          *
3739                          * But note that we special-case additions, deletions,
3740                          * renames, and mode changes as adding an empty file,
3741                          * for example is still of interest.
3742                          */
3743                         if ((p->status == DIFF_STATUS_MODIFIED)
3744                                 && !file->added
3745                                 && !file->deleted
3746                                 && one->mode == two->mode) {
3747                                 free_diffstat_file(file);
3748                                 diffstat->nr--;
3749                         }
3750                 }
3751         }
3752
3753         diff_free_filespec_data(one);
3754         diff_free_filespec_data(two);
3755 }
3756
3757 static void builtin_checkdiff(const char *name_a, const char *name_b,
3758                               const char *attr_path,
3759                               struct diff_filespec *one,
3760                               struct diff_filespec *two,
3761                               struct diff_options *o)
3762 {
3763         mmfile_t mf1, mf2;
3764         struct checkdiff_t data;
3765
3766         if (!two)
3767                 return;
3768
3769         memset(&data, 0, sizeof(data));
3770         data.filename = name_b ? name_b : name_a;
3771         data.lineno = 0;
3772         data.o = o;
3773         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3774         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3775
3776         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3777             fill_mmfile(o->repo, &mf2, two) < 0)
3778                 die("unable to read files to diff");
3779
3780         /*
3781          * All the other codepaths check both sides, but not checking
3782          * the "old" side here is deliberate.  We are checking the newly
3783          * introduced changes, and as long as the "new" side is text, we
3784          * can and should check what it introduces.
3785          */
3786         if (diff_filespec_is_binary(o->repo, two))
3787                 goto free_and_return;
3788         else {
3789                 /* Crazy xdl interfaces.. */
3790                 xpparam_t xpp;
3791                 xdemitconf_t xecfg;
3792
3793                 memset(&xpp, 0, sizeof(xpp));
3794                 memset(&xecfg, 0, sizeof(xecfg));
3795                 xecfg.ctxlen = 1; /* at least one context line */
3796                 xpp.flags = 0;
3797                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3798                                   checkdiff_consume, &data,
3799                                   &xpp, &xecfg))
3800                         die("unable to generate checkdiff for %s", one->path);
3801
3802                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3803                         struct emit_callback ecbdata;
3804                         int blank_at_eof;
3805
3806                         ecbdata.ws_rule = data.ws_rule;
3807                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3808                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3809
3810                         if (blank_at_eof) {
3811                                 static char *err;
3812                                 if (!err)
3813                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3814                                 fprintf(o->file, "%s:%d: %s.\n",
3815                                         data.filename, blank_at_eof, err);
3816                                 data.status = 1; /* report errors */
3817                         }
3818                 }
3819         }
3820  free_and_return:
3821         diff_free_filespec_data(one);
3822         diff_free_filespec_data(two);
3823         if (data.status)
3824                 o->flags.check_failed = 1;
3825 }
3826
3827 struct diff_filespec *alloc_filespec(const char *path)
3828 {
3829         struct diff_filespec *spec;
3830
3831         FLEXPTR_ALLOC_STR(spec, path, path);
3832         spec->count = 1;
3833         spec->is_binary = -1;
3834         return spec;
3835 }
3836
3837 void free_filespec(struct diff_filespec *spec)
3838 {
3839         if (!--spec->count) {
3840                 diff_free_filespec_data(spec);
3841                 free(spec);
3842         }
3843 }
3844
3845 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3846                    int oid_valid, unsigned short mode)
3847 {
3848         if (mode) {
3849                 spec->mode = canon_mode(mode);
3850                 oidcpy(&spec->oid, oid);
3851                 spec->oid_valid = oid_valid;
3852         }
3853 }
3854
3855 /*
3856  * Given a name and sha1 pair, if the index tells us the file in
3857  * the work tree has that object contents, return true, so that
3858  * prepare_temp_file() does not have to inflate and extract.
3859  */
3860 static int reuse_worktree_file(struct index_state *istate,
3861                                const char *name,
3862                                const struct object_id *oid,
3863                                int want_file)
3864 {
3865         const struct cache_entry *ce;
3866         struct stat st;
3867         int pos, len;
3868
3869         /*
3870          * We do not read the cache ourselves here, because the
3871          * benchmark with my previous version that always reads cache
3872          * shows that it makes things worse for diff-tree comparing
3873          * two linux-2.6 kernel trees in an already checked out work
3874          * tree.  This is because most diff-tree comparisons deal with
3875          * only a small number of files, while reading the cache is
3876          * expensive for a large project, and its cost outweighs the
3877          * savings we get by not inflating the object to a temporary
3878          * file.  Practically, this code only helps when we are used
3879          * by diff-cache --cached, which does read the cache before
3880          * calling us.
3881          */
3882         if (!istate->cache)
3883                 return 0;
3884
3885         /* We want to avoid the working directory if our caller
3886          * doesn't need the data in a normal file, this system
3887          * is rather slow with its stat/open/mmap/close syscalls,
3888          * and the object is contained in a pack file.  The pack
3889          * is probably already open and will be faster to obtain
3890          * the data through than the working directory.  Loose
3891          * objects however would tend to be slower as they need
3892          * to be individually opened and inflated.
3893          */
3894         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3895                 return 0;
3896
3897         /*
3898          * Similarly, if we'd have to convert the file contents anyway, that
3899          * makes the optimization not worthwhile.
3900          */
3901         if (!want_file && would_convert_to_git(istate, name))
3902                 return 0;
3903
3904         len = strlen(name);
3905         pos = index_name_pos(istate, name, len);
3906         if (pos < 0)
3907                 return 0;
3908         ce = istate->cache[pos];
3909
3910         /*
3911          * This is not the sha1 we are looking for, or
3912          * unreusable because it is not a regular file.
3913          */
3914         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3915                 return 0;
3916
3917         /*
3918          * If ce is marked as "assume unchanged", there is no
3919          * guarantee that work tree matches what we are looking for.
3920          */
3921         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3922                 return 0;
3923
3924         /*
3925          * If ce matches the file in the work tree, we can reuse it.
3926          */
3927         if (ce_uptodate(ce) ||
3928             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3929                 return 1;
3930
3931         return 0;
3932 }
3933
3934 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3935 {
3936         struct strbuf buf = STRBUF_INIT;
3937         char *dirty = "";
3938
3939         /* Are we looking at the work tree? */
3940         if (s->dirty_submodule)
3941                 dirty = "-dirty";
3942
3943         strbuf_addf(&buf, "Subproject commit %s%s\n",
3944                     oid_to_hex(&s->oid), dirty);
3945         s->size = buf.len;
3946         if (size_only) {
3947                 s->data = NULL;
3948                 strbuf_release(&buf);
3949         } else {
3950                 s->data = strbuf_detach(&buf, NULL);
3951                 s->should_free = 1;
3952         }
3953         return 0;
3954 }
3955
3956 /*
3957  * While doing rename detection and pickaxe operation, we may need to
3958  * grab the data for the blob (or file) for our own in-core comparison.
3959  * diff_filespec has data and size fields for this purpose.
3960  */
3961 int diff_populate_filespec(struct repository *r,
3962                            struct diff_filespec *s,
3963                            const struct diff_populate_filespec_options *options)
3964 {
3965         int size_only = options ? options->check_size_only : 0;
3966         int check_binary = options ? options->check_binary : 0;
3967         int err = 0;
3968         int conv_flags = global_conv_flags_eol;
3969         /*
3970          * demote FAIL to WARN to allow inspecting the situation
3971          * instead of refusing.
3972          */
3973         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3974                 conv_flags = CONV_EOL_RNDTRP_WARN;
3975
3976         if (!DIFF_FILE_VALID(s))
3977                 die("internal error: asking to populate invalid file.");
3978         if (S_ISDIR(s->mode))
3979                 return -1;
3980
3981         if (s->data)
3982                 return 0;
3983
3984         if (size_only && 0 < s->size)
3985                 return 0;
3986
3987         if (S_ISGITLINK(s->mode))
3988                 return diff_populate_gitlink(s, size_only);
3989
3990         if (!s->oid_valid ||
3991             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3992                 struct strbuf buf = STRBUF_INIT;
3993                 struct stat st;
3994                 int fd;
3995
3996                 if (lstat(s->path, &st) < 0) {
3997                 err_empty:
3998                         err = -1;
3999                 empty:
4000                         s->data = (char *)"";
4001                         s->size = 0;
4002                         return err;
4003                 }
4004                 s->size = xsize_t(st.st_size);
4005                 if (!s->size)
4006                         goto empty;
4007                 if (S_ISLNK(st.st_mode)) {
4008                         struct strbuf sb = STRBUF_INIT;
4009
4010                         if (strbuf_readlink(&sb, s->path, s->size))
4011                                 goto err_empty;
4012                         s->size = sb.len;
4013                         s->data = strbuf_detach(&sb, NULL);
4014                         s->should_free = 1;
4015                         return 0;
4016                 }
4017
4018                 /*
4019                  * Even if the caller would be happy with getting
4020                  * only the size, we cannot return early at this
4021                  * point if the path requires us to run the content
4022                  * conversion.
4023                  */
4024                 if (size_only && !would_convert_to_git(r->index, s->path))
4025                         return 0;
4026
4027                 /*
4028                  * Note: this check uses xsize_t(st.st_size) that may
4029                  * not be the true size of the blob after it goes
4030                  * through convert_to_git().  This may not strictly be
4031                  * correct, but the whole point of big_file_threshold
4032                  * and is_binary check being that we want to avoid
4033                  * opening the file and inspecting the contents, this
4034                  * is probably fine.
4035                  */
4036                 if (check_binary &&
4037                     s->size > big_file_threshold && s->is_binary == -1) {
4038                         s->is_binary = 1;
4039                         return 0;
4040                 }
4041                 fd = open(s->path, O_RDONLY);
4042                 if (fd < 0)
4043                         goto err_empty;
4044                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4045                 close(fd);
4046                 s->should_munmap = 1;
4047
4048                 /*
4049                  * Convert from working tree format to canonical git format
4050                  */
4051                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4052                         size_t size = 0;
4053                         munmap(s->data, s->size);
4054                         s->should_munmap = 0;
4055                         s->data = strbuf_detach(&buf, &size);
4056                         s->size = size;
4057                         s->should_free = 1;
4058                 }
4059         }
4060         else {
4061                 struct object_info info = {
4062                         .sizep = &s->size
4063                 };
4064
4065                 if (!(size_only || check_binary))
4066                         /*
4067                          * Set contentp, since there is no chance that merely
4068                          * the size is sufficient.
4069                          */
4070                         info.contentp = &s->data;
4071
4072                 if (options && options->missing_object_cb) {
4073                         if (!oid_object_info_extended(r, &s->oid, &info,
4074                                                       OBJECT_INFO_LOOKUP_REPLACE |
4075                                                       OBJECT_INFO_SKIP_FETCH_OBJECT))
4076                                 goto object_read;
4077                         options->missing_object_cb(options->missing_object_data);
4078                 }
4079                 if (oid_object_info_extended(r, &s->oid, &info,
4080                                              OBJECT_INFO_LOOKUP_REPLACE))
4081                         die("unable to read %s", oid_to_hex(&s->oid));
4082
4083 object_read:
4084                 if (size_only || check_binary) {
4085                         if (size_only)
4086                                 return 0;
4087                         if (s->size > big_file_threshold && s->is_binary == -1) {
4088                                 s->is_binary = 1;
4089                                 return 0;
4090                         }
4091                 }
4092                 if (!info.contentp) {
4093                         info.contentp = &s->data;
4094                         if (oid_object_info_extended(r, &s->oid, &info,
4095                                                      OBJECT_INFO_LOOKUP_REPLACE))
4096                                 die("unable to read %s", oid_to_hex(&s->oid));
4097                 }
4098                 s->should_free = 1;
4099         }
4100         return 0;
4101 }
4102
4103 void diff_free_filespec_blob(struct diff_filespec *s)
4104 {
4105         if (s->should_free)
4106                 free(s->data);
4107         else if (s->should_munmap)
4108                 munmap(s->data, s->size);
4109
4110         if (s->should_free || s->should_munmap) {
4111                 s->should_free = s->should_munmap = 0;
4112                 s->data = NULL;
4113         }
4114 }
4115
4116 void diff_free_filespec_data(struct diff_filespec *s)
4117 {
4118         if (!s)
4119                 return;
4120
4121         diff_free_filespec_blob(s);
4122         FREE_AND_NULL(s->cnt_data);
4123 }
4124
4125 static void prep_temp_blob(struct index_state *istate,
4126                            const char *path, struct diff_tempfile *temp,
4127                            void *blob,
4128                            unsigned long size,
4129                            const struct object_id *oid,
4130                            int mode)
4131 {
4132         struct strbuf buf = STRBUF_INIT;
4133         struct strbuf tempfile = STRBUF_INIT;
4134         char *path_dup = xstrdup(path);
4135         const char *base = basename(path_dup);
4136         struct checkout_metadata meta;
4137
4138         init_checkout_metadata(&meta, NULL, NULL, oid);
4139
4140         /* Generate "XXXXXX_basename.ext" */
4141         strbuf_addstr(&tempfile, "XXXXXX_");
4142         strbuf_addstr(&tempfile, base);
4143
4144         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4145         if (!temp->tempfile)
4146                 die_errno("unable to create temp-file");
4147         if (convert_to_working_tree(istate, path,
4148                         (const char *)blob, (size_t)size, &buf, &meta)) {
4149                 blob = buf.buf;
4150                 size = buf.len;
4151         }
4152         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4153             close_tempfile_gently(temp->tempfile))
4154                 die_errno("unable to write temp-file");
4155         temp->name = get_tempfile_path(temp->tempfile);
4156         oid_to_hex_r(temp->hex, oid);
4157         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4158         strbuf_release(&buf);
4159         strbuf_release(&tempfile);
4160         free(path_dup);
4161 }
4162
4163 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4164                                                const char *name,
4165                                                struct diff_filespec *one)
4166 {
4167         struct diff_tempfile *temp = claim_diff_tempfile();
4168
4169         if (!DIFF_FILE_VALID(one)) {
4170         not_a_valid_file:
4171                 /* A '-' entry produces this for file-2, and
4172                  * a '+' entry produces this for file-1.
4173                  */
4174                 temp->name = "/dev/null";
4175                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4176                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4177                 return temp;
4178         }
4179
4180         if (!S_ISGITLINK(one->mode) &&
4181             (!one->oid_valid ||
4182              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4183                 struct stat st;
4184                 if (lstat(name, &st) < 0) {
4185                         if (errno == ENOENT)
4186                                 goto not_a_valid_file;
4187                         die_errno("stat(%s)", name);
4188                 }
4189                 if (S_ISLNK(st.st_mode)) {
4190                         struct strbuf sb = STRBUF_INIT;
4191                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4192                                 die_errno("readlink(%s)", name);
4193                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4194                                        (one->oid_valid ?
4195                                         &one->oid : &null_oid),
4196                                        (one->oid_valid ?
4197                                         one->mode : S_IFLNK));
4198                         strbuf_release(&sb);
4199                 }
4200                 else {
4201                         /* we can borrow from the file in the work tree */
4202                         temp->name = name;
4203                         if (!one->oid_valid)
4204                                 oid_to_hex_r(temp->hex, &null_oid);
4205                         else
4206                                 oid_to_hex_r(temp->hex, &one->oid);
4207                         /* Even though we may sometimes borrow the
4208                          * contents from the work tree, we always want
4209                          * one->mode.  mode is trustworthy even when
4210                          * !(one->oid_valid), as long as
4211                          * DIFF_FILE_VALID(one).
4212                          */
4213                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4214                 }
4215                 return temp;
4216         }
4217         else {
4218                 if (diff_populate_filespec(r, one, NULL))
4219                         die("cannot read data blob for %s", one->path);
4220                 prep_temp_blob(r->index, name, temp,
4221                                one->data, one->size,
4222                                &one->oid, one->mode);
4223         }
4224         return temp;
4225 }
4226
4227 static void add_external_diff_name(struct repository *r,
4228                                    struct strvec *argv,
4229                                    const char *name,
4230                                    struct diff_filespec *df)
4231 {
4232         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4233         strvec_push(argv, temp->name);
4234         strvec_push(argv, temp->hex);
4235         strvec_push(argv, temp->mode);
4236 }
4237
4238 /* An external diff command takes:
4239  *
4240  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4241  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4242  *
4243  */
4244 static void run_external_diff(const char *pgm,
4245                               const char *name,
4246                               const char *other,
4247                               struct diff_filespec *one,
4248                               struct diff_filespec *two,
4249                               const char *xfrm_msg,
4250                               struct diff_options *o)
4251 {
4252         struct strvec argv = STRVEC_INIT;
4253         struct strvec env = STRVEC_INIT;
4254         struct diff_queue_struct *q = &diff_queued_diff;
4255
4256         strvec_push(&argv, pgm);
4257         strvec_push(&argv, name);
4258
4259         if (one && two) {
4260                 add_external_diff_name(o->repo, &argv, name, one);
4261                 if (!other)
4262                         add_external_diff_name(o->repo, &argv, name, two);
4263                 else {
4264                         add_external_diff_name(o->repo, &argv, other, two);
4265                         strvec_push(&argv, other);
4266                         strvec_push(&argv, xfrm_msg);
4267                 }
4268         }
4269
4270         strvec_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4271         strvec_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4272
4273         diff_free_filespec_data(one);
4274         diff_free_filespec_data(two);
4275         if (run_command_v_opt_cd_env(argv.v, RUN_USING_SHELL, NULL, env.v))
4276                 die(_("external diff died, stopping at %s"), name);
4277
4278         remove_tempfile();
4279         strvec_clear(&argv);
4280         strvec_clear(&env);
4281 }
4282
4283 static int similarity_index(struct diff_filepair *p)
4284 {
4285         return p->score * 100 / MAX_SCORE;
4286 }
4287
4288 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4289 {
4290         if (startup_info->have_repository)
4291                 return find_unique_abbrev(oid, abbrev);
4292         else {
4293                 char *hex = oid_to_hex(oid);
4294                 if (abbrev < 0)
4295                         abbrev = FALLBACK_DEFAULT_ABBREV;
4296                 if (abbrev > the_hash_algo->hexsz)
4297                         BUG("oid abbreviation out of range: %d", abbrev);
4298                 if (abbrev)
4299                         hex[abbrev] = '\0';
4300                 return hex;
4301         }
4302 }
4303
4304 static void fill_metainfo(struct strbuf *msg,
4305                           const char *name,
4306                           const char *other,
4307                           struct diff_filespec *one,
4308                           struct diff_filespec *two,
4309                           struct diff_options *o,
4310                           struct diff_filepair *p,
4311                           int *must_show_header,
4312                           int use_color)
4313 {
4314         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4315         const char *reset = diff_get_color(use_color, DIFF_RESET);
4316         const char *line_prefix = diff_line_prefix(o);
4317
4318         *must_show_header = 1;
4319         strbuf_init(msg, PATH_MAX * 2 + 300);
4320         switch (p->status) {
4321         case DIFF_STATUS_COPIED:
4322                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4323                             line_prefix, set, similarity_index(p));
4324                 strbuf_addf(msg, "%s\n%s%scopy from ",
4325                             reset,  line_prefix, set);
4326                 quote_c_style(name, msg, NULL, 0);
4327                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4328                 quote_c_style(other, msg, NULL, 0);
4329                 strbuf_addf(msg, "%s\n", reset);
4330                 break;
4331         case DIFF_STATUS_RENAMED:
4332                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4333                             line_prefix, set, similarity_index(p));
4334                 strbuf_addf(msg, "%s\n%s%srename from ",
4335                             reset, line_prefix, set);
4336                 quote_c_style(name, msg, NULL, 0);
4337                 strbuf_addf(msg, "%s\n%s%srename to ",
4338                             reset, line_prefix, set);
4339                 quote_c_style(other, msg, NULL, 0);
4340                 strbuf_addf(msg, "%s\n", reset);
4341                 break;
4342         case DIFF_STATUS_MODIFIED:
4343                 if (p->score) {
4344                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4345                                     line_prefix,
4346                                     set, similarity_index(p), reset);
4347                         break;
4348                 }
4349                 /* fallthru */
4350         default:
4351                 *must_show_header = 0;
4352         }
4353         if (one && two && !oideq(&one->oid, &two->oid)) {
4354                 const unsigned hexsz = the_hash_algo->hexsz;
4355                 int abbrev = o->abbrev ? o->abbrev : DEFAULT_ABBREV;
4356
4357                 if (o->flags.full_index)
4358                         abbrev = hexsz;
4359
4360                 if (o->flags.binary) {
4361                         mmfile_t mf;
4362                         if ((!fill_mmfile(o->repo, &mf, one) &&
4363                              diff_filespec_is_binary(o->repo, one)) ||
4364                             (!fill_mmfile(o->repo, &mf, two) &&
4365                              diff_filespec_is_binary(o->repo, two)))
4366                                 abbrev = hexsz;
4367                 }
4368                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4369                             diff_abbrev_oid(&one->oid, abbrev),
4370                             diff_abbrev_oid(&two->oid, abbrev));
4371                 if (one->mode == two->mode)
4372                         strbuf_addf(msg, " %06o", one->mode);
4373                 strbuf_addf(msg, "%s\n", reset);
4374         }
4375 }
4376
4377 static void run_diff_cmd(const char *pgm,
4378                          const char *name,
4379                          const char *other,
4380                          const char *attr_path,
4381                          struct diff_filespec *one,
4382                          struct diff_filespec *two,
4383                          struct strbuf *msg,
4384                          struct diff_options *o,
4385                          struct diff_filepair *p)
4386 {
4387         const char *xfrm_msg = NULL;
4388         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4389         int must_show_header = 0;
4390
4391
4392         if (o->flags.allow_external) {
4393                 struct userdiff_driver *drv;
4394
4395                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4396                 if (drv && drv->external)
4397                         pgm = drv->external;
4398         }
4399
4400         if (msg) {
4401                 /*
4402                  * don't use colors when the header is intended for an
4403                  * external diff driver
4404                  */
4405                 fill_metainfo(msg, name, other, one, two, o, p,
4406                               &must_show_header,
4407                               want_color(o->use_color) && !pgm);
4408                 xfrm_msg = msg->len ? msg->buf : NULL;
4409         }
4410
4411         if (pgm) {
4412                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4413                 return;
4414         }
4415         if (one && two)
4416                 builtin_diff(name, other ? other : name,
4417                              one, two, xfrm_msg, must_show_header,
4418                              o, complete_rewrite);
4419         else
4420                 fprintf(o->file, "* Unmerged path %s\n", name);
4421 }
4422
4423 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4424 {
4425         if (DIFF_FILE_VALID(one)) {
4426                 if (!one->oid_valid) {
4427                         struct stat st;
4428                         if (one->is_stdin) {
4429                                 oidclr(&one->oid);
4430                                 return;
4431                         }
4432                         if (lstat(one->path, &st) < 0)
4433                                 die_errno("stat '%s'", one->path);
4434                         if (index_path(istate, &one->oid, one->path, &st, 0))
4435                                 die("cannot hash %s", one->path);
4436                 }
4437         }
4438         else
4439                 oidclr(&one->oid);
4440 }
4441
4442 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4443 {
4444         /* Strip the prefix but do not molest /dev/null and absolute paths */
4445         if (*namep && !is_absolute_path(*namep)) {
4446                 *namep += prefix_length;
4447                 if (**namep == '/')
4448                         ++*namep;
4449         }
4450         if (*otherp && !is_absolute_path(*otherp)) {
4451                 *otherp += prefix_length;
4452                 if (**otherp == '/')
4453                         ++*otherp;
4454         }
4455 }
4456
4457 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4458 {
4459         const char *pgm = external_diff();
4460         struct strbuf msg;
4461         struct diff_filespec *one = p->one;
4462         struct diff_filespec *two = p->two;
4463         const char *name;
4464         const char *other;
4465         const char *attr_path;
4466
4467         name  = one->path;
4468         other = (strcmp(name, two->path) ? two->path : NULL);
4469         attr_path = name;
4470         if (o->prefix_length)
4471                 strip_prefix(o->prefix_length, &name, &other);
4472
4473         if (!o->flags.allow_external)
4474                 pgm = NULL;
4475
4476         if (DIFF_PAIR_UNMERGED(p)) {
4477                 run_diff_cmd(pgm, name, NULL, attr_path,
4478                              NULL, NULL, NULL, o, p);
4479                 return;
4480         }
4481
4482         diff_fill_oid_info(one, o->repo->index);
4483         diff_fill_oid_info(two, o->repo->index);
4484
4485         if (!pgm &&
4486             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4487             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4488                 /*
4489                  * a filepair that changes between file and symlink
4490                  * needs to be split into deletion and creation.
4491                  */
4492                 struct diff_filespec *null = alloc_filespec(two->path);
4493                 run_diff_cmd(NULL, name, other, attr_path,
4494                              one, null, &msg,
4495                              o, p);
4496                 free(null);
4497                 strbuf_release(&msg);
4498
4499                 null = alloc_filespec(one->path);
4500                 run_diff_cmd(NULL, name, other, attr_path,
4501                              null, two, &msg, o, p);
4502                 free(null);
4503         }
4504         else
4505                 run_diff_cmd(pgm, name, other, attr_path,
4506                              one, two, &msg, o, p);
4507
4508         strbuf_release(&msg);
4509 }
4510
4511 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4512                          struct diffstat_t *diffstat)
4513 {
4514         const char *name;
4515         const char *other;
4516
4517         if (DIFF_PAIR_UNMERGED(p)) {
4518                 /* unmerged */
4519                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4520                                  diffstat, o, p);
4521                 return;
4522         }
4523
4524         name = p->one->path;
4525         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4526
4527         if (o->prefix_length)
4528                 strip_prefix(o->prefix_length, &name, &other);
4529
4530         diff_fill_oid_info(p->one, o->repo->index);
4531         diff_fill_oid_info(p->two, o->repo->index);
4532
4533         builtin_diffstat(name, other, p->one, p->two,
4534                          diffstat, o, p);
4535 }
4536
4537 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4538 {
4539         const char *name;
4540         const char *other;
4541         const char *attr_path;
4542
4543         if (DIFF_PAIR_UNMERGED(p)) {
4544                 /* unmerged */
4545                 return;
4546         }
4547
4548         name = p->one->path;
4549         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4550         attr_path = other ? other : name;
4551
4552         if (o->prefix_length)
4553                 strip_prefix(o->prefix_length, &name, &other);
4554
4555         diff_fill_oid_info(p->one, o->repo->index);
4556         diff_fill_oid_info(p->two, o->repo->index);
4557
4558         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4559 }
4560
4561 static void prep_parse_options(struct diff_options *options);
4562
4563 void repo_diff_setup(struct repository *r, struct diff_options *options)
4564 {
4565         memcpy(options, &default_diff_options, sizeof(*options));
4566
4567         options->file = stdout;
4568         options->repo = r;
4569
4570         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4571         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4572         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4573         options->abbrev = DEFAULT_ABBREV;
4574         options->line_termination = '\n';
4575         options->break_opt = -1;
4576         options->rename_limit = -1;
4577         options->dirstat_permille = diff_dirstat_permille_default;
4578         options->context = diff_context_default;
4579         options->interhunkcontext = diff_interhunk_context_default;
4580         options->ws_error_highlight = ws_error_highlight_default;
4581         options->flags.rename_empty = 1;
4582         options->flags.relative_name = diff_relative;
4583         options->objfind = NULL;
4584
4585         /* pathchange left =NULL by default */
4586         options->change = diff_change;
4587         options->add_remove = diff_addremove;
4588         options->use_color = diff_use_color_default;
4589         options->detect_rename = diff_detect_rename_default;
4590         options->xdl_opts |= diff_algorithm;
4591         if (diff_indent_heuristic)
4592                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4593
4594         options->orderfile = diff_order_file_cfg;
4595
4596         if (diff_no_prefix) {
4597                 options->a_prefix = options->b_prefix = "";
4598         } else if (!diff_mnemonic_prefix) {
4599                 options->a_prefix = "a/";
4600                 options->b_prefix = "b/";
4601         }
4602
4603         options->color_moved = diff_color_moved_default;
4604         options->color_moved_ws_handling = diff_color_moved_ws_default;
4605
4606         prep_parse_options(options);
4607 }
4608
4609 void diff_setup_done(struct diff_options *options)
4610 {
4611         unsigned check_mask = DIFF_FORMAT_NAME |
4612                               DIFF_FORMAT_NAME_STATUS |
4613                               DIFF_FORMAT_CHECKDIFF |
4614                               DIFF_FORMAT_NO_OUTPUT;
4615         /*
4616          * This must be signed because we're comparing against a potentially
4617          * negative value.
4618          */
4619         const int hexsz = the_hash_algo->hexsz;
4620
4621         if (options->set_default)
4622                 options->set_default(options);
4623
4624         if (HAS_MULTI_BITS(options->output_format & check_mask))
4625                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4626
4627         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4628                 die(_("-G, -S and --find-object are mutually exclusive"));
4629
4630         /*
4631          * Most of the time we can say "there are changes"
4632          * only by checking if there are changed paths, but
4633          * --ignore-whitespace* options force us to look
4634          * inside contents.
4635          */
4636
4637         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4638                 options->flags.diff_from_contents = 1;
4639         else
4640                 options->flags.diff_from_contents = 0;
4641
4642         if (options->flags.find_copies_harder)
4643                 options->detect_rename = DIFF_DETECT_COPY;
4644
4645         if (!options->flags.relative_name)
4646                 options->prefix = NULL;
4647         if (options->prefix)
4648                 options->prefix_length = strlen(options->prefix);
4649         else
4650                 options->prefix_length = 0;
4651
4652         if (options->output_format & (DIFF_FORMAT_NAME |
4653                                       DIFF_FORMAT_NAME_STATUS |
4654                                       DIFF_FORMAT_CHECKDIFF |
4655                                       DIFF_FORMAT_NO_OUTPUT))
4656                 options->output_format &= ~(DIFF_FORMAT_RAW |
4657                                             DIFF_FORMAT_NUMSTAT |
4658                                             DIFF_FORMAT_DIFFSTAT |
4659                                             DIFF_FORMAT_SHORTSTAT |
4660                                             DIFF_FORMAT_DIRSTAT |
4661                                             DIFF_FORMAT_SUMMARY |
4662                                             DIFF_FORMAT_PATCH);
4663
4664         /*
4665          * These cases always need recursive; we do not drop caller-supplied
4666          * recursive bits for other formats here.
4667          */
4668         if (options->output_format & (DIFF_FORMAT_PATCH |
4669                                       DIFF_FORMAT_NUMSTAT |
4670                                       DIFF_FORMAT_DIFFSTAT |
4671                                       DIFF_FORMAT_SHORTSTAT |
4672                                       DIFF_FORMAT_DIRSTAT |
4673                                       DIFF_FORMAT_SUMMARY |
4674                                       DIFF_FORMAT_CHECKDIFF))
4675                 options->flags.recursive = 1;
4676         /*
4677          * Also pickaxe would not work very well if you do not say recursive
4678          */
4679         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4680                 options->flags.recursive = 1;
4681         /*
4682          * When patches are generated, submodules diffed against the work tree
4683          * must be checked for dirtiness too so it can be shown in the output
4684          */
4685         if (options->output_format & DIFF_FORMAT_PATCH)
4686                 options->flags.dirty_submodules = 1;
4687
4688         if (options->detect_rename && options->rename_limit < 0)
4689                 options->rename_limit = diff_rename_limit_default;
4690         if (hexsz < options->abbrev)
4691                 options->abbrev = hexsz; /* full */
4692
4693         /*
4694          * It does not make sense to show the first hit we happened
4695          * to have found.  It does not make sense not to return with
4696          * exit code in such a case either.
4697          */
4698         if (options->flags.quick) {
4699                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4700                 options->flags.exit_with_status = 1;
4701         }
4702
4703         options->diff_path_counter = 0;
4704
4705         if (options->flags.follow_renames && options->pathspec.nr != 1)
4706                 die(_("--follow requires exactly one pathspec"));
4707
4708         if (!options->use_color || external_diff())
4709                 options->color_moved = 0;
4710
4711         FREE_AND_NULL(options->parseopts);
4712 }
4713
4714 int parse_long_opt(const char *opt, const char **argv,
4715                    const char **optarg)
4716 {
4717         const char *arg = argv[0];
4718         if (!skip_prefix(arg, "--", &arg))
4719                 return 0;
4720         if (!skip_prefix(arg, opt, &arg))
4721                 return 0;
4722         if (*arg == '=') { /* stuck form: --option=value */
4723                 *optarg = arg + 1;
4724                 return 1;
4725         }
4726         if (*arg != '\0')
4727                 return 0;
4728         /* separate form: --option value */
4729         if (!argv[1])
4730                 die("Option '--%s' requires a value", opt);
4731         *optarg = argv[1];
4732         return 2;
4733 }
4734
4735 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4736 {
4737         struct diff_options *options = opt->value;
4738         int width = options->stat_width;
4739         int name_width = options->stat_name_width;
4740         int graph_width = options->stat_graph_width;
4741         int count = options->stat_count;
4742         char *end;
4743
4744         BUG_ON_OPT_NEG(unset);
4745
4746         if (!strcmp(opt->long_name, "stat")) {
4747                 if (value) {
4748                         width = strtoul(value, &end, 10);
4749                         if (*end == ',')
4750                                 name_width = strtoul(end+1, &end, 10);
4751                         if (*end == ',')
4752                                 count = strtoul(end+1, &end, 10);
4753                         if (*end)
4754                                 return error(_("invalid --stat value: %s"), value);
4755                 }
4756         } else if (!strcmp(opt->long_name, "stat-width")) {
4757                 width = strtoul(value, &end, 10);
4758                 if (*end)
4759                         return error(_("%s expects a numerical value"),
4760                                      opt->long_name);
4761         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4762                 name_width = strtoul(value, &end, 10);
4763                 if (*end)
4764                         return error(_("%s expects a numerical value"),
4765                                      opt->long_name);
4766         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4767                 graph_width = strtoul(value, &end, 10);
4768                 if (*end)
4769                         return error(_("%s expects a numerical value"),
4770                                      opt->long_name);
4771         } else if (!strcmp(opt->long_name, "stat-count")) {
4772                 count = strtoul(value, &end, 10);
4773                 if (*end)
4774                         return error(_("%s expects a numerical value"),
4775                                      opt->long_name);
4776         } else
4777                 BUG("%s should not get here", opt->long_name);
4778
4779         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4780         options->stat_name_width = name_width;
4781         options->stat_graph_width = graph_width;
4782         options->stat_width = width;
4783         options->stat_count = count;
4784         return 0;
4785 }
4786
4787 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4788 {
4789         struct strbuf errmsg = STRBUF_INIT;
4790         if (parse_dirstat_params(options, params, &errmsg))
4791                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4792                     errmsg.buf);
4793         strbuf_release(&errmsg);
4794         /*
4795          * The caller knows a dirstat-related option is given from the command
4796          * line; allow it to say "return this_function();"
4797          */
4798         options->output_format |= DIFF_FORMAT_DIRSTAT;
4799         return 1;
4800 }
4801
4802 static const char diff_status_letters[] = {
4803         DIFF_STATUS_ADDED,
4804         DIFF_STATUS_COPIED,
4805         DIFF_STATUS_DELETED,
4806         DIFF_STATUS_MODIFIED,
4807         DIFF_STATUS_RENAMED,
4808         DIFF_STATUS_TYPE_CHANGED,
4809         DIFF_STATUS_UNKNOWN,
4810         DIFF_STATUS_UNMERGED,
4811         DIFF_STATUS_FILTER_AON,
4812         DIFF_STATUS_FILTER_BROKEN,
4813         '\0',
4814 };
4815
4816 static unsigned int filter_bit['Z' + 1];
4817
4818 static void prepare_filter_bits(void)
4819 {
4820         int i;
4821
4822         if (!filter_bit[DIFF_STATUS_ADDED]) {
4823                 for (i = 0; diff_status_letters[i]; i++)
4824                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4825         }
4826 }
4827
4828 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4829 {
4830         return opt->filter & filter_bit[(int) status];
4831 }
4832
4833 unsigned diff_filter_bit(char status)
4834 {
4835         prepare_filter_bits();
4836         return filter_bit[(int) status];
4837 }
4838
4839 static int diff_opt_diff_filter(const struct option *option,
4840                                 const char *optarg, int unset)
4841 {
4842         struct diff_options *opt = option->value;
4843         int i, optch;
4844
4845         BUG_ON_OPT_NEG(unset);
4846         prepare_filter_bits();
4847
4848         /*
4849          * If there is a negation e.g. 'd' in the input, and we haven't
4850          * initialized the filter field with another --diff-filter, start
4851          * from full set of bits, except for AON.
4852          */
4853         if (!opt->filter) {
4854                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4855                         if (optch < 'a' || 'z' < optch)
4856                                 continue;
4857                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4858                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4859                         break;
4860                 }
4861         }
4862
4863         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4864                 unsigned int bit;
4865                 int negate;
4866
4867                 if ('a' <= optch && optch <= 'z') {
4868                         negate = 1;
4869                         optch = toupper(optch);
4870                 } else {
4871                         negate = 0;
4872                 }
4873
4874                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4875                 if (!bit)
4876                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4877                                      optarg[i], optarg);
4878                 if (negate)
4879                         opt->filter &= ~bit;
4880                 else
4881                         opt->filter |= bit;
4882         }
4883         return 0;
4884 }
4885
4886 static void enable_patch_output(int *fmt)
4887 {
4888         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4889         *fmt |= DIFF_FORMAT_PATCH;
4890 }
4891
4892 static int diff_opt_ws_error_highlight(const struct option *option,
4893                                        const char *arg, int unset)
4894 {
4895         struct diff_options *opt = option->value;
4896         int val = parse_ws_error_highlight(arg);
4897
4898         BUG_ON_OPT_NEG(unset);
4899         if (val < 0)
4900                 return error(_("unknown value after ws-error-highlight=%.*s"),
4901                              -1 - val, arg);
4902         opt->ws_error_highlight = val;
4903         return 0;
4904 }
4905
4906 static int diff_opt_find_object(const struct option *option,
4907                                 const char *arg, int unset)
4908 {
4909         struct diff_options *opt = option->value;
4910         struct object_id oid;
4911
4912         BUG_ON_OPT_NEG(unset);
4913         if (get_oid(arg, &oid))
4914                 return error(_("unable to resolve '%s'"), arg);
4915
4916         if (!opt->objfind)
4917                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4918
4919         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4920         opt->flags.recursive = 1;
4921         opt->flags.tree_in_recursive = 1;
4922         oidset_insert(opt->objfind, &oid);
4923         return 0;
4924 }
4925
4926 static int diff_opt_anchored(const struct option *opt,
4927                              const char *arg, int unset)
4928 {
4929         struct diff_options *options = opt->value;
4930
4931         BUG_ON_OPT_NEG(unset);
4932         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4933         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4934                    options->anchors_alloc);
4935         options->anchors[options->anchors_nr++] = xstrdup(arg);
4936         return 0;
4937 }
4938
4939 static int diff_opt_binary(const struct option *opt,
4940                            const char *arg, int unset)
4941 {
4942         struct diff_options *options = opt->value;
4943
4944         BUG_ON_OPT_NEG(unset);
4945         BUG_ON_OPT_ARG(arg);
4946         enable_patch_output(&options->output_format);
4947         options->flags.binary = 1;
4948         return 0;
4949 }
4950
4951 static int diff_opt_break_rewrites(const struct option *opt,
4952                                    const char *arg, int unset)
4953 {
4954         int *break_opt = opt->value;
4955         int opt1, opt2;
4956
4957         BUG_ON_OPT_NEG(unset);
4958         if (!arg)
4959                 arg = "";
4960         opt1 = parse_rename_score(&arg);
4961         if (*arg == 0)
4962                 opt2 = 0;
4963         else if (*arg != '/')
4964                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4965         else {
4966                 arg++;
4967                 opt2 = parse_rename_score(&arg);
4968         }
4969         if (*arg != 0)
4970                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4971         *break_opt = opt1 | (opt2 << 16);
4972         return 0;
4973 }
4974
4975 static int diff_opt_char(const struct option *opt,
4976                          const char *arg, int unset)
4977 {
4978         char *value = opt->value;
4979
4980         BUG_ON_OPT_NEG(unset);
4981         if (arg[1])
4982                 return error(_("%s expects a character, got '%s'"),
4983                              opt->long_name, arg);
4984         *value = arg[0];
4985         return 0;
4986 }
4987
4988 static int diff_opt_color_moved(const struct option *opt,
4989                                 const char *arg, int unset)
4990 {
4991         struct diff_options *options = opt->value;
4992
4993         if (unset) {
4994                 options->color_moved = COLOR_MOVED_NO;
4995         } else if (!arg) {
4996                 if (diff_color_moved_default)
4997                         options->color_moved = diff_color_moved_default;
4998                 if (options->color_moved == COLOR_MOVED_NO)
4999                         options->color_moved = COLOR_MOVED_DEFAULT;
5000         } else {
5001                 int cm = parse_color_moved(arg);
5002                 if (cm < 0)
5003                         return error(_("bad --color-moved argument: %s"), arg);
5004                 options->color_moved = cm;
5005         }
5006         return 0;
5007 }
5008
5009 static int diff_opt_color_moved_ws(const struct option *opt,
5010                                    const char *arg, int unset)
5011 {
5012         struct diff_options *options = opt->value;
5013         unsigned cm;
5014
5015         if (unset) {
5016                 options->color_moved_ws_handling = 0;
5017                 return 0;
5018         }
5019
5020         cm = parse_color_moved_ws(arg);
5021         if (cm & COLOR_MOVED_WS_ERROR)
5022                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
5023         options->color_moved_ws_handling = cm;
5024         return 0;
5025 }
5026
5027 static int diff_opt_color_words(const struct option *opt,
5028                                 const char *arg, int unset)
5029 {
5030         struct diff_options *options = opt->value;
5031
5032         BUG_ON_OPT_NEG(unset);
5033         options->use_color = 1;
5034         options->word_diff = DIFF_WORDS_COLOR;
5035         options->word_regex = arg;
5036         return 0;
5037 }
5038
5039 static int diff_opt_compact_summary(const struct option *opt,
5040                                     const char *arg, int unset)
5041 {
5042         struct diff_options *options = opt->value;
5043
5044         BUG_ON_OPT_ARG(arg);
5045         if (unset) {
5046                 options->flags.stat_with_summary = 0;
5047         } else {
5048                 options->flags.stat_with_summary = 1;
5049                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5050         }
5051         return 0;
5052 }
5053
5054 static int diff_opt_diff_algorithm(const struct option *opt,
5055                                    const char *arg, int unset)
5056 {
5057         struct diff_options *options = opt->value;
5058         long value = parse_algorithm_value(arg);
5059
5060         BUG_ON_OPT_NEG(unset);
5061         if (value < 0)
5062                 return error(_("option diff-algorithm accepts \"myers\", "
5063                                "\"minimal\", \"patience\" and \"histogram\""));
5064
5065         /* clear out previous settings */
5066         DIFF_XDL_CLR(options, NEED_MINIMAL);
5067         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5068         options->xdl_opts |= value;
5069         return 0;
5070 }
5071
5072 static int diff_opt_dirstat(const struct option *opt,
5073                             const char *arg, int unset)
5074 {
5075         struct diff_options *options = opt->value;
5076
5077         BUG_ON_OPT_NEG(unset);
5078         if (!strcmp(opt->long_name, "cumulative")) {
5079                 if (arg)
5080                         BUG("how come --cumulative take a value?");
5081                 arg = "cumulative";
5082         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5083                 parse_dirstat_opt(options, "files");
5084         parse_dirstat_opt(options, arg ? arg : "");
5085         return 0;
5086 }
5087
5088 static int diff_opt_find_copies(const struct option *opt,
5089                                 const char *arg, int unset)
5090 {
5091         struct diff_options *options = opt->value;
5092
5093         BUG_ON_OPT_NEG(unset);
5094         if (!arg)
5095                 arg = "";
5096         options->rename_score = parse_rename_score(&arg);
5097         if (*arg != 0)
5098                 return error(_("invalid argument to %s"), opt->long_name);
5099
5100         if (options->detect_rename == DIFF_DETECT_COPY)
5101                 options->flags.find_copies_harder = 1;
5102         else
5103                 options->detect_rename = DIFF_DETECT_COPY;
5104
5105         return 0;
5106 }
5107
5108 static int diff_opt_find_renames(const struct option *opt,
5109                                  const char *arg, int unset)
5110 {
5111         struct diff_options *options = opt->value;
5112
5113         BUG_ON_OPT_NEG(unset);
5114         if (!arg)
5115                 arg = "";
5116         options->rename_score = parse_rename_score(&arg);
5117         if (*arg != 0)
5118                 return error(_("invalid argument to %s"), opt->long_name);
5119
5120         options->detect_rename = DIFF_DETECT_RENAME;
5121         return 0;
5122 }
5123
5124 static int diff_opt_follow(const struct option *opt,
5125                            const char *arg, int unset)
5126 {
5127         struct diff_options *options = opt->value;
5128
5129         BUG_ON_OPT_ARG(arg);
5130         if (unset) {
5131                 options->flags.follow_renames = 0;
5132                 options->flags.default_follow_renames = 0;
5133         } else {
5134                 options->flags.follow_renames = 1;
5135         }
5136         return 0;
5137 }
5138
5139 static int diff_opt_ignore_submodules(const struct option *opt,
5140                                       const char *arg, int unset)
5141 {
5142         struct diff_options *options = opt->value;
5143
5144         BUG_ON_OPT_NEG(unset);
5145         if (!arg)
5146                 arg = "all";
5147         options->flags.override_submodule_config = 1;
5148         handle_ignore_submodules_arg(options, arg);
5149         return 0;
5150 }
5151
5152 static int diff_opt_line_prefix(const struct option *opt,
5153                                 const char *optarg, int unset)
5154 {
5155         struct diff_options *options = opt->value;
5156
5157         BUG_ON_OPT_NEG(unset);
5158         options->line_prefix = optarg;
5159         options->line_prefix_length = strlen(options->line_prefix);
5160         graph_setup_line_prefix(options);
5161         return 0;
5162 }
5163
5164 static int diff_opt_no_prefix(const struct option *opt,
5165                               const char *optarg, int unset)
5166 {
5167         struct diff_options *options = opt->value;
5168
5169         BUG_ON_OPT_NEG(unset);
5170         BUG_ON_OPT_ARG(optarg);
5171         options->a_prefix = "";
5172         options->b_prefix = "";
5173         return 0;
5174 }
5175
5176 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5177                                              const struct option *opt,
5178                                              const char *arg, int unset)
5179 {
5180         struct diff_options *options = opt->value;
5181         char *path;
5182
5183         BUG_ON_OPT_NEG(unset);
5184         path = prefix_filename(ctx->prefix, arg);
5185         options->file = xfopen(path, "w");
5186         options->close_file = 1;
5187         if (options->use_color != GIT_COLOR_ALWAYS)
5188                 options->use_color = GIT_COLOR_NEVER;
5189         free(path);
5190         return 0;
5191 }
5192
5193 static int diff_opt_patience(const struct option *opt,
5194                              const char *arg, int unset)
5195 {
5196         struct diff_options *options = opt->value;
5197         int i;
5198
5199         BUG_ON_OPT_NEG(unset);
5200         BUG_ON_OPT_ARG(arg);
5201         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5202         /*
5203          * Both --patience and --anchored use PATIENCE_DIFF
5204          * internally, so remove any anchors previously
5205          * specified.
5206          */
5207         for (i = 0; i < options->anchors_nr; i++)
5208                 free(options->anchors[i]);
5209         options->anchors_nr = 0;
5210         return 0;
5211 }
5212
5213 static int diff_opt_ignore_regex(const struct option *opt,
5214                                  const char *arg, int unset)
5215 {
5216         struct diff_options *options = opt->value;
5217         regex_t *regex;
5218
5219         BUG_ON_OPT_NEG(unset);
5220         regex = xmalloc(sizeof(*regex));
5221         if (regcomp(regex, arg, REG_EXTENDED | REG_NEWLINE))
5222                 return error(_("invalid regex given to -I: '%s'"), arg);
5223         ALLOC_GROW(options->ignore_regex, options->ignore_regex_nr + 1,
5224                    options->ignore_regex_alloc);
5225         options->ignore_regex[options->ignore_regex_nr++] = regex;
5226         return 0;
5227 }
5228
5229 static int diff_opt_pickaxe_regex(const struct option *opt,
5230                                   const char *arg, int unset)
5231 {
5232         struct diff_options *options = opt->value;
5233
5234         BUG_ON_OPT_NEG(unset);
5235         options->pickaxe = arg;
5236         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5237         return 0;
5238 }
5239
5240 static int diff_opt_pickaxe_string(const struct option *opt,
5241                                    const char *arg, int unset)
5242 {
5243         struct diff_options *options = opt->value;
5244
5245         BUG_ON_OPT_NEG(unset);
5246         options->pickaxe = arg;
5247         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5248         return 0;
5249 }
5250
5251 static int diff_opt_relative(const struct option *opt,
5252                              const char *arg, int unset)
5253 {
5254         struct diff_options *options = opt->value;
5255
5256         options->flags.relative_name = !unset;
5257         if (arg)
5258                 options->prefix = arg;
5259         return 0;
5260 }
5261
5262 static int diff_opt_submodule(const struct option *opt,
5263                               const char *arg, int unset)
5264 {
5265         struct diff_options *options = opt->value;
5266
5267         BUG_ON_OPT_NEG(unset);
5268         if (!arg)
5269                 arg = "log";
5270         if (parse_submodule_params(options, arg))
5271                 return error(_("failed to parse --submodule option parameter: '%s'"),
5272                              arg);
5273         return 0;
5274 }
5275
5276 static int diff_opt_textconv(const struct option *opt,
5277                              const char *arg, int unset)
5278 {
5279         struct diff_options *options = opt->value;
5280
5281         BUG_ON_OPT_ARG(arg);
5282         if (unset) {
5283                 options->flags.allow_textconv = 0;
5284         } else {
5285                 options->flags.allow_textconv = 1;
5286                 options->flags.textconv_set_via_cmdline = 1;
5287         }
5288         return 0;
5289 }
5290
5291 static int diff_opt_unified(const struct option *opt,
5292                             const char *arg, int unset)
5293 {
5294         struct diff_options *options = opt->value;
5295         char *s;
5296
5297         BUG_ON_OPT_NEG(unset);
5298
5299         if (arg) {
5300                 options->context = strtol(arg, &s, 10);
5301                 if (*s)
5302                         return error(_("%s expects a numerical value"), "--unified");
5303         }
5304         enable_patch_output(&options->output_format);
5305
5306         return 0;
5307 }
5308
5309 static int diff_opt_word_diff(const struct option *opt,
5310                               const char *arg, int unset)
5311 {
5312         struct diff_options *options = opt->value;
5313
5314         BUG_ON_OPT_NEG(unset);
5315         if (arg) {
5316                 if (!strcmp(arg, "plain"))
5317                         options->word_diff = DIFF_WORDS_PLAIN;
5318                 else if (!strcmp(arg, "color")) {
5319                         options->use_color = 1;
5320                         options->word_diff = DIFF_WORDS_COLOR;
5321                 }
5322                 else if (!strcmp(arg, "porcelain"))
5323                         options->word_diff = DIFF_WORDS_PORCELAIN;
5324                 else if (!strcmp(arg, "none"))
5325                         options->word_diff = DIFF_WORDS_NONE;
5326                 else
5327                         return error(_("bad --word-diff argument: %s"), arg);
5328         } else {
5329                 if (options->word_diff == DIFF_WORDS_NONE)
5330                         options->word_diff = DIFF_WORDS_PLAIN;
5331         }
5332         return 0;
5333 }
5334
5335 static int diff_opt_word_diff_regex(const struct option *opt,
5336                                     const char *arg, int unset)
5337 {
5338         struct diff_options *options = opt->value;
5339
5340         BUG_ON_OPT_NEG(unset);
5341         if (options->word_diff == DIFF_WORDS_NONE)
5342                 options->word_diff = DIFF_WORDS_PLAIN;
5343         options->word_regex = arg;
5344         return 0;
5345 }
5346
5347 static void prep_parse_options(struct diff_options *options)
5348 {
5349         struct option parseopts[] = {
5350                 OPT_GROUP(N_("Diff output format options")),
5351                 OPT_BITOP('p', "patch", &options->output_format,
5352                           N_("generate patch"),
5353                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5354                 OPT_BIT_F('s', "no-patch", &options->output_format,
5355                           N_("suppress diff output"),
5356                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5357                 OPT_BITOP('u', NULL, &options->output_format,
5358                           N_("generate patch"),
5359                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5360                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5361                                N_("generate diffs with <n> lines context"),
5362                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5363                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5364                          N_("generate diffs with <n> lines context")),
5365                 OPT_BIT_F(0, "raw", &options->output_format,
5366                           N_("generate the diff in raw format"),
5367                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5368                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5369                           N_("synonym for '-p --raw'"),
5370                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5371                           DIFF_FORMAT_NO_OUTPUT),
5372                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5373                           N_("synonym for '-p --stat'"),
5374                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5375                           DIFF_FORMAT_NO_OUTPUT),
5376                 OPT_BIT_F(0, "numstat", &options->output_format,
5377                           N_("machine friendly --stat"),
5378                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5379                 OPT_BIT_F(0, "shortstat", &options->output_format,
5380                           N_("output only the last line of --stat"),
5381                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5382                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5383                                N_("output the distribution of relative amount of changes for each sub-directory"),
5384                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5385                                diff_opt_dirstat),
5386                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5387                                N_("synonym for --dirstat=cumulative"),
5388                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5389                                diff_opt_dirstat),
5390                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5391                                N_("synonym for --dirstat=files,param1,param2..."),
5392                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5393                                diff_opt_dirstat),
5394                 OPT_BIT_F(0, "check", &options->output_format,
5395                           N_("warn if changes introduce conflict markers or whitespace errors"),
5396                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5397                 OPT_BIT_F(0, "summary", &options->output_format,
5398                           N_("condensed summary such as creations, renames and mode changes"),
5399                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5400                 OPT_BIT_F(0, "name-only", &options->output_format,
5401                           N_("show only names of changed files"),
5402                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5403                 OPT_BIT_F(0, "name-status", &options->output_format,
5404                           N_("show only names and status of changed files"),
5405                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5406                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5407                                N_("generate diffstat"),
5408                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5409                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5410                                N_("generate diffstat with a given width"),
5411                                PARSE_OPT_NONEG, diff_opt_stat),
5412                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5413                                N_("generate diffstat with a given name width"),
5414                                PARSE_OPT_NONEG, diff_opt_stat),
5415                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5416                                N_("generate diffstat with a given graph width"),
5417                                PARSE_OPT_NONEG, diff_opt_stat),
5418                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5419                                N_("generate diffstat with limited lines"),
5420                                PARSE_OPT_NONEG, diff_opt_stat),
5421                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5422                                N_("generate compact summary in diffstat"),
5423                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5424                 OPT_CALLBACK_F(0, "binary", options, NULL,
5425                                N_("output a binary diff that can be applied"),
5426                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5427                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5428                          N_("show full pre- and post-image object names on the \"index\" lines")),
5429                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5430                                N_("show colored diff")),
5431                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5432                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5433                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5434                 OPT_SET_INT('z', NULL, &options->line_termination,
5435                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5436                             0),
5437                 OPT__ABBREV(&options->abbrev),
5438                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5439                              N_("show the given source prefix instead of \"a/\""),
5440                              PARSE_OPT_NONEG),
5441                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5442                              N_("show the given destination prefix instead of \"b/\""),
5443                              PARSE_OPT_NONEG),
5444                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5445                                N_("prepend an additional prefix to every line of output"),
5446                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5447                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5448                                N_("do not show any source or destination prefix"),
5449                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5450                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5451                               N_("show context between diff hunks up to the specified number of lines"),
5452                               PARSE_OPT_NONEG),
5453                 OPT_CALLBACK_F(0, "output-indicator-new",
5454                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5455                                N_("<char>"),
5456                                N_("specify the character to indicate a new line instead of '+'"),
5457                                PARSE_OPT_NONEG, diff_opt_char),
5458                 OPT_CALLBACK_F(0, "output-indicator-old",
5459                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5460                                N_("<char>"),
5461                                N_("specify the character to indicate an old line instead of '-'"),
5462                                PARSE_OPT_NONEG, diff_opt_char),
5463                 OPT_CALLBACK_F(0, "output-indicator-context",
5464                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5465                                N_("<char>"),
5466                                N_("specify the character to indicate a context instead of ' '"),
5467                                PARSE_OPT_NONEG, diff_opt_char),
5468
5469                 OPT_GROUP(N_("Diff rename options")),
5470                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5471                                N_("break complete rewrite changes into pairs of delete and create"),
5472                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5473                                diff_opt_break_rewrites),
5474                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5475                                N_("detect renames"),
5476                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5477                                diff_opt_find_renames),
5478                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5479                               N_("omit the preimage for deletes"),
5480                               1, PARSE_OPT_NONEG),
5481                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5482                                N_("detect copies"),
5483                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5484                                diff_opt_find_copies),
5485                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5486                          N_("use unmodified files as source to find copies")),
5487                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5488                               N_("disable rename detection"),
5489                               0, PARSE_OPT_NONEG),
5490                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5491                          N_("use empty blobs as rename source")),
5492                 OPT_CALLBACK_F(0, "follow", options, NULL,
5493                                N_("continue listing the history of a file beyond renames"),
5494                                PARSE_OPT_NOARG, diff_opt_follow),
5495                 OPT_INTEGER('l', NULL, &options->rename_limit,
5496                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5497
5498                 OPT_GROUP(N_("Diff algorithm options")),
5499                 OPT_BIT(0, "minimal", &options->xdl_opts,
5500                         N_("produce the smallest possible diff"),
5501                         XDF_NEED_MINIMAL),
5502                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5503                           N_("ignore whitespace when comparing lines"),
5504                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5505                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5506                           N_("ignore changes in amount of whitespace"),
5507                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5508                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5509                           N_("ignore changes in whitespace at EOL"),
5510                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5511                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5512                           N_("ignore carrier-return at the end of line"),
5513                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5514                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5515                           N_("ignore changes whose lines are all blank"),
5516                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5517                 OPT_CALLBACK_F('I', "ignore-matching-lines", options, N_("<regex>"),
5518                                N_("ignore changes whose all lines match <regex>"),
5519                                0, diff_opt_ignore_regex),
5520                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5521                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5522                         XDF_INDENT_HEURISTIC),
5523                 OPT_CALLBACK_F(0, "patience", options, NULL,
5524                                N_("generate diff using the \"patience diff\" algorithm"),
5525                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5526                                diff_opt_patience),
5527                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5528                           N_("generate diff using the \"histogram diff\" algorithm"),
5529                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5530                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5531                                N_("choose a diff algorithm"),
5532                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5533                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5534                                N_("generate diff using the \"anchored diff\" algorithm"),
5535                                PARSE_OPT_NONEG, diff_opt_anchored),
5536                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5537                                N_("show word diff, using <mode> to delimit changed words"),
5538                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5539                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5540                                N_("use <regex> to decide what a word is"),
5541                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5542                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5543                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5544                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5545                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5546                                N_("moved lines of code are colored differently"),
5547                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5548                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5549                                N_("how white spaces are ignored in --color-moved"),
5550                                0, diff_opt_color_moved_ws),
5551
5552                 OPT_GROUP(N_("Other diff options")),
5553                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5554                                N_("when run from subdir, exclude changes outside and show relative paths"),
5555                                PARSE_OPT_OPTARG,
5556                                diff_opt_relative),
5557                 OPT_BOOL('a', "text", &options->flags.text,
5558                          N_("treat all files as text")),
5559                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5560                          N_("swap two inputs, reverse the diff")),
5561                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5562                          N_("exit with 1 if there were differences, 0 otherwise")),
5563                 OPT_BOOL(0, "quiet", &options->flags.quick,
5564                          N_("disable all output of the program")),
5565                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5566                          N_("allow an external diff helper to be executed")),
5567                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5568                                N_("run external text conversion filters when comparing binary files"),
5569                                PARSE_OPT_NOARG, diff_opt_textconv),
5570                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5571                                N_("ignore changes to submodules in the diff generation"),
5572                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5573                                diff_opt_ignore_submodules),
5574                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5575                                N_("specify how differences in submodules are shown"),
5576                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5577                                diff_opt_submodule),
5578                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5579                               N_("hide 'git add -N' entries from the index"),
5580                               1, PARSE_OPT_NONEG),
5581                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5582                               N_("treat 'git add -N' entries as real in the index"),
5583                               0, PARSE_OPT_NONEG),
5584                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5585                                N_("look for differences that change the number of occurrences of the specified string"),
5586                                0, diff_opt_pickaxe_string),
5587                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5588                                N_("look for differences that change the number of occurrences of the specified regex"),
5589                                0, diff_opt_pickaxe_regex),
5590                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5591                           N_("show all changes in the changeset with -S or -G"),
5592                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5593                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5594                           N_("treat <string> in -S as extended POSIX regular expression"),
5595                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5596                 OPT_FILENAME('O', NULL, &options->orderfile,
5597                              N_("control the order in which files appear in the output")),
5598                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5599                                N_("look for differences that change the number of occurrences of the specified object"),
5600                                PARSE_OPT_NONEG, diff_opt_find_object),
5601                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5602                                N_("select files by diff type"),
5603                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5604                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5605                   N_("Output to a specific file"),
5606                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5607
5608                 OPT_END()
5609         };
5610
5611         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5612         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5613 }
5614
5615 int diff_opt_parse(struct diff_options *options,
5616                    const char **av, int ac, const char *prefix)
5617 {
5618         if (!prefix)
5619                 prefix = "";
5620
5621         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5622                            PARSE_OPT_KEEP_DASHDASH |
5623                            PARSE_OPT_KEEP_UNKNOWN |
5624                            PARSE_OPT_NO_INTERNAL_HELP |
5625                            PARSE_OPT_ONE_SHOT |
5626                            PARSE_OPT_STOP_AT_NON_OPTION);
5627
5628         return ac;
5629 }
5630
5631 int parse_rename_score(const char **cp_p)
5632 {
5633         unsigned long num, scale;
5634         int ch, dot;
5635         const char *cp = *cp_p;
5636
5637         num = 0;
5638         scale = 1;
5639         dot = 0;
5640         for (;;) {
5641                 ch = *cp;
5642                 if ( !dot && ch == '.' ) {
5643                         scale = 1;
5644                         dot = 1;
5645                 } else if ( ch == '%' ) {
5646                         scale = dot ? scale*100 : 100;
5647                         cp++;   /* % is always at the end */
5648                         break;
5649                 } else if ( ch >= '0' && ch <= '9' ) {
5650                         if ( scale < 100000 ) {
5651                                 scale *= 10;
5652                                 num = (num*10) + (ch-'0');
5653                         }
5654                 } else {
5655                         break;
5656                 }
5657                 cp++;
5658         }
5659         *cp_p = cp;
5660
5661         /* user says num divided by scale and we say internally that
5662          * is MAX_SCORE * num / scale.
5663          */
5664         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5665 }
5666
5667 struct diff_queue_struct diff_queued_diff;
5668
5669 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5670 {
5671         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5672         queue->queue[queue->nr++] = dp;
5673 }
5674
5675 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5676                                  struct diff_filespec *one,
5677                                  struct diff_filespec *two)
5678 {
5679         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5680         dp->one = one;
5681         dp->two = two;
5682         if (queue)
5683                 diff_q(queue, dp);
5684         return dp;
5685 }
5686
5687 void diff_free_filepair(struct diff_filepair *p)
5688 {
5689         free_filespec(p->one);
5690         free_filespec(p->two);
5691         free(p);
5692 }
5693
5694 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5695 {
5696         int abblen;
5697         const char *abbrev;
5698
5699         /* Do we want all 40 hex characters? */
5700         if (len == the_hash_algo->hexsz)
5701                 return oid_to_hex(oid);
5702
5703         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5704         abbrev = diff_abbrev_oid(oid, len);
5705
5706         if (!print_sha1_ellipsis())
5707                 return abbrev;
5708
5709         abblen = strlen(abbrev);
5710
5711         /*
5712          * In well-behaved cases, where the abbreviated result is the
5713          * same as the requested length, append three dots after the
5714          * abbreviation (hence the whole logic is limited to the case
5715          * where abblen < 37); when the actual abbreviated result is a
5716          * bit longer than the requested length, we reduce the number
5717          * of dots so that they match the well-behaved ones.  However,
5718          * if the actual abbreviation is longer than the requested
5719          * length by more than three, we give up on aligning, and add
5720          * three dots anyway, to indicate that the output is not the
5721          * full object name.  Yes, this may be suboptimal, but this
5722          * appears only in "diff --raw --abbrev" output and it is not
5723          * worth the effort to change it now.  Note that this would
5724          * likely to work fine when the automatic sizing of default
5725          * abbreviation length is used--we would be fed -1 in "len" in
5726          * that case, and will end up always appending three-dots, but
5727          * the automatic sizing is supposed to give abblen that ensures
5728          * uniqueness across all objects (statistically speaking).
5729          */
5730         if (abblen < the_hash_algo->hexsz - 3) {
5731                 static char hex[GIT_MAX_HEXSZ + 1];
5732                 if (len < abblen && abblen <= len + 2)
5733                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5734                 else
5735                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5736                 return hex;
5737         }
5738
5739         return oid_to_hex(oid);
5740 }
5741
5742 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5743 {
5744         int line_termination = opt->line_termination;
5745         int inter_name_termination = line_termination ? '\t' : '\0';
5746
5747         fprintf(opt->file, "%s", diff_line_prefix(opt));
5748         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5749                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5750                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5751                 fprintf(opt->file, "%s ",
5752                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5753         }
5754         if (p->score) {
5755                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5756                         inter_name_termination);
5757         } else {
5758                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5759         }
5760
5761         if (p->status == DIFF_STATUS_COPIED ||
5762             p->status == DIFF_STATUS_RENAMED) {
5763                 const char *name_a, *name_b;
5764                 name_a = p->one->path;
5765                 name_b = p->two->path;
5766                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5767                 write_name_quoted(name_a, opt->file, inter_name_termination);
5768                 write_name_quoted(name_b, opt->file, line_termination);
5769         } else {
5770                 const char *name_a, *name_b;
5771                 name_a = p->one->mode ? p->one->path : p->two->path;
5772                 name_b = NULL;
5773                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5774                 write_name_quoted(name_a, opt->file, line_termination);
5775         }
5776 }
5777
5778 int diff_unmodified_pair(struct diff_filepair *p)
5779 {
5780         /* This function is written stricter than necessary to support
5781          * the currently implemented transformers, but the idea is to
5782          * let transformers to produce diff_filepairs any way they want,
5783          * and filter and clean them up here before producing the output.
5784          */
5785         struct diff_filespec *one = p->one, *two = p->two;
5786
5787         if (DIFF_PAIR_UNMERGED(p))
5788                 return 0; /* unmerged is interesting */
5789
5790         /* deletion, addition, mode or type change
5791          * and rename are all interesting.
5792          */
5793         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5794             DIFF_PAIR_MODE_CHANGED(p) ||
5795             strcmp(one->path, two->path))
5796                 return 0;
5797
5798         /* both are valid and point at the same path.  that is, we are
5799          * dealing with a change.
5800          */
5801         if (one->oid_valid && two->oid_valid &&
5802             oideq(&one->oid, &two->oid) &&
5803             !one->dirty_submodule && !two->dirty_submodule)
5804                 return 1; /* no change */
5805         if (!one->oid_valid && !two->oid_valid)
5806                 return 1; /* both look at the same file on the filesystem. */
5807         return 0;
5808 }
5809
5810 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5811 {
5812         if (diff_unmodified_pair(p))
5813                 return;
5814
5815         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5816             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5817                 return; /* no tree diffs in patch format */
5818
5819         run_diff(p, o);
5820 }
5821
5822 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5823                             struct diffstat_t *diffstat)
5824 {
5825         if (diff_unmodified_pair(p))
5826                 return;
5827
5828         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5829             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5830                 return; /* no useful stat for tree diffs */
5831
5832         run_diffstat(p, o, diffstat);
5833 }
5834
5835 static void diff_flush_checkdiff(struct diff_filepair *p,
5836                 struct diff_options *o)
5837 {
5838         if (diff_unmodified_pair(p))
5839                 return;
5840
5841         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5842             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5843                 return; /* nothing to check in tree diffs */
5844
5845         run_checkdiff(p, o);
5846 }
5847
5848 int diff_queue_is_empty(void)
5849 {
5850         struct diff_queue_struct *q = &diff_queued_diff;
5851         int i;
5852         for (i = 0; i < q->nr; i++)
5853                 if (!diff_unmodified_pair(q->queue[i]))
5854                         return 0;
5855         return 1;
5856 }
5857
5858 #if DIFF_DEBUG
5859 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5860 {
5861         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5862                 x, one ? one : "",
5863                 s->path,
5864                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5865                 s->mode,
5866                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5867         fprintf(stderr, "queue[%d] %s size %lu\n",
5868                 x, one ? one : "",
5869                 s->size);
5870 }
5871
5872 void diff_debug_filepair(const struct diff_filepair *p, int i)
5873 {
5874         diff_debug_filespec(p->one, i, "one");
5875         diff_debug_filespec(p->two, i, "two");
5876         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5877                 p->score, p->status ? p->status : '?',
5878                 p->one->rename_used, p->broken_pair);
5879 }
5880
5881 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5882 {
5883         int i;
5884         if (msg)
5885                 fprintf(stderr, "%s\n", msg);
5886         fprintf(stderr, "q->nr = %d\n", q->nr);
5887         for (i = 0; i < q->nr; i++) {
5888                 struct diff_filepair *p = q->queue[i];
5889                 diff_debug_filepair(p, i);
5890         }
5891 }
5892 #endif
5893
5894 static void diff_resolve_rename_copy(void)
5895 {
5896         int i;
5897         struct diff_filepair *p;
5898         struct diff_queue_struct *q = &diff_queued_diff;
5899
5900         diff_debug_queue("resolve-rename-copy", q);
5901
5902         for (i = 0; i < q->nr; i++) {
5903                 p = q->queue[i];
5904                 p->status = 0; /* undecided */
5905                 if (DIFF_PAIR_UNMERGED(p))
5906                         p->status = DIFF_STATUS_UNMERGED;
5907                 else if (!DIFF_FILE_VALID(p->one))
5908                         p->status = DIFF_STATUS_ADDED;
5909                 else if (!DIFF_FILE_VALID(p->two))
5910                         p->status = DIFF_STATUS_DELETED;
5911                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5912                         p->status = DIFF_STATUS_TYPE_CHANGED;
5913
5914                 /* from this point on, we are dealing with a pair
5915                  * whose both sides are valid and of the same type, i.e.
5916                  * either in-place edit or rename/copy edit.
5917                  */
5918                 else if (DIFF_PAIR_RENAME(p)) {
5919                         /*
5920                          * A rename might have re-connected a broken
5921                          * pair up, causing the pathnames to be the
5922                          * same again. If so, that's not a rename at
5923                          * all, just a modification..
5924                          *
5925                          * Otherwise, see if this source was used for
5926                          * multiple renames, in which case we decrement
5927                          * the count, and call it a copy.
5928                          */
5929                         if (!strcmp(p->one->path, p->two->path))
5930                                 p->status = DIFF_STATUS_MODIFIED;
5931                         else if (--p->one->rename_used > 0)
5932                                 p->status = DIFF_STATUS_COPIED;
5933                         else
5934                                 p->status = DIFF_STATUS_RENAMED;
5935                 }
5936                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5937                          p->one->mode != p->two->mode ||
5938                          p->one->dirty_submodule ||
5939                          p->two->dirty_submodule ||
5940                          is_null_oid(&p->one->oid))
5941                         p->status = DIFF_STATUS_MODIFIED;
5942                 else {
5943                         /* This is a "no-change" entry and should not
5944                          * happen anymore, but prepare for broken callers.
5945                          */
5946                         error("feeding unmodified %s to diffcore",
5947                               p->one->path);
5948                         p->status = DIFF_STATUS_UNKNOWN;
5949                 }
5950         }
5951         diff_debug_queue("resolve-rename-copy done", q);
5952 }
5953
5954 static int check_pair_status(struct diff_filepair *p)
5955 {
5956         switch (p->status) {
5957         case DIFF_STATUS_UNKNOWN:
5958                 return 0;
5959         case 0:
5960                 die("internal error in diff-resolve-rename-copy");
5961         default:
5962                 return 1;
5963         }
5964 }
5965
5966 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5967 {
5968         int fmt = opt->output_format;
5969
5970         if (fmt & DIFF_FORMAT_CHECKDIFF)
5971                 diff_flush_checkdiff(p, opt);
5972         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5973                 diff_flush_raw(p, opt);
5974         else if (fmt & DIFF_FORMAT_NAME) {
5975                 const char *name_a, *name_b;
5976                 name_a = p->two->path;
5977                 name_b = NULL;
5978                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5979                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5980                 write_name_quoted(name_a, opt->file, opt->line_termination);
5981         }
5982 }
5983
5984 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5985 {
5986         struct strbuf sb = STRBUF_INIT;
5987         if (fs->mode)
5988                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5989         else
5990                 strbuf_addf(&sb, " %s ", newdelete);
5991
5992         quote_c_style(fs->path, &sb, NULL, 0);
5993         strbuf_addch(&sb, '\n');
5994         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5995                          sb.buf, sb.len, 0);
5996         strbuf_release(&sb);
5997 }
5998
5999 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
6000                 int show_name)
6001 {
6002         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
6003                 struct strbuf sb = STRBUF_INIT;
6004                 strbuf_addf(&sb, " mode change %06o => %06o",
6005                             p->one->mode, p->two->mode);
6006                 if (show_name) {
6007                         strbuf_addch(&sb, ' ');
6008                         quote_c_style(p->two->path, &sb, NULL, 0);
6009                 }
6010                 strbuf_addch(&sb, '\n');
6011                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6012                                  sb.buf, sb.len, 0);
6013                 strbuf_release(&sb);
6014         }
6015 }
6016
6017 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
6018                 struct diff_filepair *p)
6019 {
6020         struct strbuf sb = STRBUF_INIT;
6021         struct strbuf names = STRBUF_INIT;
6022
6023         pprint_rename(&names, p->one->path, p->two->path);
6024         strbuf_addf(&sb, " %s %s (%d%%)\n",
6025                     renamecopy, names.buf, similarity_index(p));
6026         strbuf_release(&names);
6027         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6028                                  sb.buf, sb.len, 0);
6029         show_mode_change(opt, p, 0);
6030         strbuf_release(&sb);
6031 }
6032
6033 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
6034 {
6035         switch(p->status) {
6036         case DIFF_STATUS_DELETED:
6037                 show_file_mode_name(opt, "delete", p->one);
6038                 break;
6039         case DIFF_STATUS_ADDED:
6040                 show_file_mode_name(opt, "create", p->two);
6041                 break;
6042         case DIFF_STATUS_COPIED:
6043                 show_rename_copy(opt, "copy", p);
6044                 break;
6045         case DIFF_STATUS_RENAMED:
6046                 show_rename_copy(opt, "rename", p);
6047                 break;
6048         default:
6049                 if (p->score) {
6050                         struct strbuf sb = STRBUF_INIT;
6051                         strbuf_addstr(&sb, " rewrite ");
6052                         quote_c_style(p->two->path, &sb, NULL, 0);
6053                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
6054                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6055                                          sb.buf, sb.len, 0);
6056                         strbuf_release(&sb);
6057                 }
6058                 show_mode_change(opt, p, !p->score);
6059                 break;
6060         }
6061 }
6062
6063 struct patch_id_t {
6064         git_hash_ctx *ctx;
6065         int patchlen;
6066 };
6067
6068 static int remove_space(char *line, int len)
6069 {
6070         int i;
6071         char *dst = line;
6072         unsigned char c;
6073
6074         for (i = 0; i < len; i++)
6075                 if (!isspace((c = line[i])))
6076                         *dst++ = c;
6077
6078         return dst - line;
6079 }
6080
6081 void flush_one_hunk(struct object_id *result, git_hash_ctx *ctx)
6082 {
6083         unsigned char hash[GIT_MAX_RAWSZ];
6084         unsigned short carry = 0;
6085         int i;
6086
6087         the_hash_algo->final_fn(hash, ctx);
6088         the_hash_algo->init_fn(ctx);
6089         /* 20-byte sum, with carry */
6090         for (i = 0; i < the_hash_algo->rawsz; ++i) {
6091                 carry += result->hash[i] + hash[i];
6092                 result->hash[i] = carry;
6093                 carry >>= 8;
6094         }
6095 }
6096
6097 static void patch_id_consume(void *priv, char *line, unsigned long len)
6098 {
6099         struct patch_id_t *data = priv;
6100         int new_len;
6101
6102         if (len > 12 && starts_with(line, "\\ "))
6103                 return;
6104         new_len = remove_space(line, len);
6105
6106         the_hash_algo->update_fn(data->ctx, line, new_len);
6107         data->patchlen += new_len;
6108 }
6109
6110 static void patch_id_add_string(git_hash_ctx *ctx, const char *str)
6111 {
6112         the_hash_algo->update_fn(ctx, str, strlen(str));
6113 }
6114
6115 static void patch_id_add_mode(git_hash_ctx *ctx, unsigned mode)
6116 {
6117         /* large enough for 2^32 in octal */
6118         char buf[12];
6119         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6120         the_hash_algo->update_fn(ctx, buf, len);
6121 }
6122
6123 /* returns 0 upon success, and writes result into oid */
6124 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6125 {
6126         struct diff_queue_struct *q = &diff_queued_diff;
6127         int i;
6128         git_hash_ctx ctx;
6129         struct patch_id_t data;
6130
6131         the_hash_algo->init_fn(&ctx);
6132         memset(&data, 0, sizeof(struct patch_id_t));
6133         data.ctx = &ctx;
6134         oidclr(oid);
6135
6136         for (i = 0; i < q->nr; i++) {
6137                 xpparam_t xpp;
6138                 xdemitconf_t xecfg;
6139                 mmfile_t mf1, mf2;
6140                 struct diff_filepair *p = q->queue[i];
6141                 int len1, len2;
6142
6143                 memset(&xpp, 0, sizeof(xpp));
6144                 memset(&xecfg, 0, sizeof(xecfg));
6145                 if (p->status == 0)
6146                         return error("internal diff status error");
6147                 if (p->status == DIFF_STATUS_UNKNOWN)
6148                         continue;
6149                 if (diff_unmodified_pair(p))
6150                         continue;
6151                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6152                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6153                         continue;
6154                 if (DIFF_PAIR_UNMERGED(p))
6155                         continue;
6156
6157                 diff_fill_oid_info(p->one, options->repo->index);
6158                 diff_fill_oid_info(p->two, options->repo->index);
6159
6160                 len1 = remove_space(p->one->path, strlen(p->one->path));
6161                 len2 = remove_space(p->two->path, strlen(p->two->path));
6162                 patch_id_add_string(&ctx, "diff--git");
6163                 patch_id_add_string(&ctx, "a/");
6164                 the_hash_algo->update_fn(&ctx, p->one->path, len1);
6165                 patch_id_add_string(&ctx, "b/");
6166                 the_hash_algo->update_fn(&ctx, p->two->path, len2);
6167
6168                 if (p->one->mode == 0) {
6169                         patch_id_add_string(&ctx, "newfilemode");
6170                         patch_id_add_mode(&ctx, p->two->mode);
6171                         patch_id_add_string(&ctx, "---/dev/null");
6172                         patch_id_add_string(&ctx, "+++b/");
6173                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6174                 } else if (p->two->mode == 0) {
6175                         patch_id_add_string(&ctx, "deletedfilemode");
6176                         patch_id_add_mode(&ctx, p->one->mode);
6177                         patch_id_add_string(&ctx, "---a/");
6178                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6179                         patch_id_add_string(&ctx, "+++/dev/null");
6180                 } else {
6181                         patch_id_add_string(&ctx, "---a/");
6182                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6183                         patch_id_add_string(&ctx, "+++b/");
6184                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6185                 }
6186
6187                 if (diff_header_only)
6188                         continue;
6189
6190                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6191                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6192                         return error("unable to read files to diff");
6193
6194                 if (diff_filespec_is_binary(options->repo, p->one) ||
6195                     diff_filespec_is_binary(options->repo, p->two)) {
6196                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->one->oid),
6197                                         the_hash_algo->hexsz);
6198                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
6199                                         the_hash_algo->hexsz);
6200                         continue;
6201                 }
6202
6203                 xpp.flags = 0;
6204                 xecfg.ctxlen = 3;
6205                 xecfg.flags = 0;
6206                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6207                                   patch_id_consume, &data, &xpp, &xecfg))
6208                         return error("unable to generate patch-id diff for %s",
6209                                      p->one->path);
6210
6211                 if (stable)
6212                         flush_one_hunk(oid, &ctx);
6213         }
6214
6215         if (!stable)
6216                 the_hash_algo->final_fn(oid->hash, &ctx);
6217
6218         return 0;
6219 }
6220
6221 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6222 {
6223         struct diff_queue_struct *q = &diff_queued_diff;
6224         int i;
6225         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6226
6227         for (i = 0; i < q->nr; i++)
6228                 diff_free_filepair(q->queue[i]);
6229
6230         free(q->queue);
6231         DIFF_QUEUE_CLEAR(q);
6232
6233         return result;
6234 }
6235
6236 static int is_summary_empty(const struct diff_queue_struct *q)
6237 {
6238         int i;
6239
6240         for (i = 0; i < q->nr; i++) {
6241                 const struct diff_filepair *p = q->queue[i];
6242
6243                 switch (p->status) {
6244                 case DIFF_STATUS_DELETED:
6245                 case DIFF_STATUS_ADDED:
6246                 case DIFF_STATUS_COPIED:
6247                 case DIFF_STATUS_RENAMED:
6248                         return 0;
6249                 default:
6250                         if (p->score)
6251                                 return 0;
6252                         if (p->one->mode && p->two->mode &&
6253                             p->one->mode != p->two->mode)
6254                                 return 0;
6255                         break;
6256                 }
6257         }
6258         return 1;
6259 }
6260
6261 static const char rename_limit_warning[] =
6262 N_("inexact rename detection was skipped due to too many files.");
6263
6264 static const char degrade_cc_to_c_warning[] =
6265 N_("only found copies from modified paths due to too many files.");
6266
6267 static const char rename_limit_advice[] =
6268 N_("you may want to set your %s variable to at least "
6269    "%d and retry the command.");
6270
6271 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6272 {
6273         fflush(stdout);
6274         if (degraded_cc)
6275                 warning(_(degrade_cc_to_c_warning));
6276         else if (needed)
6277                 warning(_(rename_limit_warning));
6278         else
6279                 return;
6280         if (0 < needed)
6281                 warning(_(rename_limit_advice), varname, needed);
6282 }
6283
6284 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6285 {
6286         int i;
6287         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6288         struct diff_queue_struct *q = &diff_queued_diff;
6289
6290         if (WSEH_NEW & WS_RULE_MASK)
6291                 BUG("WS rules bit mask overlaps with diff symbol flags");
6292
6293         if (o->color_moved)
6294                 o->emitted_symbols = &esm;
6295
6296         for (i = 0; i < q->nr; i++) {
6297                 struct diff_filepair *p = q->queue[i];
6298                 if (check_pair_status(p))
6299                         diff_flush_patch(p, o);
6300         }
6301
6302         if (o->emitted_symbols) {
6303                 if (o->color_moved) {
6304                         struct hashmap add_lines, del_lines;
6305
6306                         if (o->color_moved_ws_handling &
6307                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6308                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6309
6310                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6311                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6312
6313                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6314                         mark_color_as_moved(o, &add_lines, &del_lines);
6315                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6316                                 dim_moved_lines(o);
6317
6318                         hashmap_free_entries(&add_lines, struct moved_entry,
6319                                                 ent);
6320                         hashmap_free_entries(&del_lines, struct moved_entry,
6321                                                 ent);
6322                 }
6323
6324                 for (i = 0; i < esm.nr; i++)
6325                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6326
6327                 for (i = 0; i < esm.nr; i++)
6328                         free((void *)esm.buf[i].line);
6329                 esm.nr = 0;
6330
6331                 o->emitted_symbols = NULL;
6332         }
6333 }
6334
6335 void diff_flush(struct diff_options *options)
6336 {
6337         struct diff_queue_struct *q = &diff_queued_diff;
6338         int i, output_format = options->output_format;
6339         int separator = 0;
6340         int dirstat_by_line = 0;
6341
6342         /*
6343          * Order: raw, stat, summary, patch
6344          * or:    name/name-status/checkdiff (other bits clear)
6345          */
6346         if (!q->nr)
6347                 goto free_queue;
6348
6349         if (output_format & (DIFF_FORMAT_RAW |
6350                              DIFF_FORMAT_NAME |
6351                              DIFF_FORMAT_NAME_STATUS |
6352                              DIFF_FORMAT_CHECKDIFF)) {
6353                 for (i = 0; i < q->nr; i++) {
6354                         struct diff_filepair *p = q->queue[i];
6355                         if (check_pair_status(p))
6356                                 flush_one_pair(p, options);
6357                 }
6358                 separator++;
6359         }
6360
6361         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6362                 dirstat_by_line = 1;
6363
6364         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6365             dirstat_by_line) {
6366                 struct diffstat_t diffstat;
6367
6368                 compute_diffstat(options, &diffstat, q);
6369                 if (output_format & DIFF_FORMAT_NUMSTAT)
6370                         show_numstat(&diffstat, options);
6371                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6372                         show_stats(&diffstat, options);
6373                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6374                         show_shortstats(&diffstat, options);
6375                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6376                         show_dirstat_by_line(&diffstat, options);
6377                 free_diffstat_info(&diffstat);
6378                 separator++;
6379         }
6380         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6381                 show_dirstat(options);
6382
6383         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6384                 for (i = 0; i < q->nr; i++) {
6385                         diff_summary(options, q->queue[i]);
6386                 }
6387                 separator++;
6388         }
6389
6390         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6391             options->flags.exit_with_status &&
6392             options->flags.diff_from_contents) {
6393                 /*
6394                  * run diff_flush_patch for the exit status. setting
6395                  * options->file to /dev/null should be safe, because we
6396                  * aren't supposed to produce any output anyway.
6397                  */
6398                 if (options->close_file)
6399                         fclose(options->file);
6400                 options->file = xfopen("/dev/null", "w");
6401                 options->close_file = 1;
6402                 options->color_moved = 0;
6403                 for (i = 0; i < q->nr; i++) {
6404                         struct diff_filepair *p = q->queue[i];
6405                         if (check_pair_status(p))
6406                                 diff_flush_patch(p, options);
6407                         if (options->found_changes)
6408                                 break;
6409                 }
6410         }
6411
6412         if (output_format & DIFF_FORMAT_PATCH) {
6413                 if (separator) {
6414                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6415                         if (options->stat_sep)
6416                                 /* attach patch instead of inline */
6417                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6418                                                  NULL, 0, 0);
6419                 }
6420
6421                 diff_flush_patch_all_file_pairs(options);
6422         }
6423
6424         if (output_format & DIFF_FORMAT_CALLBACK)
6425                 options->format_callback(q, options, options->format_callback_data);
6426
6427         for (i = 0; i < q->nr; i++)
6428                 diff_free_filepair(q->queue[i]);
6429 free_queue:
6430         free(q->queue);
6431         DIFF_QUEUE_CLEAR(q);
6432         if (options->close_file)
6433                 fclose(options->file);
6434
6435         /*
6436          * Report the content-level differences with HAS_CHANGES;
6437          * diff_addremove/diff_change does not set the bit when
6438          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6439          */
6440         if (options->flags.diff_from_contents) {
6441                 if (options->found_changes)
6442                         options->flags.has_changes = 1;
6443                 else
6444                         options->flags.has_changes = 0;
6445         }
6446 }
6447
6448 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6449 {
6450         return (((p->status == DIFF_STATUS_MODIFIED) &&
6451                  ((p->score &&
6452                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6453                   (!p->score &&
6454                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6455                 ((p->status != DIFF_STATUS_MODIFIED) &&
6456                  filter_bit_tst(p->status, options)));
6457 }
6458
6459 static void diffcore_apply_filter(struct diff_options *options)
6460 {
6461         int i;
6462         struct diff_queue_struct *q = &diff_queued_diff;
6463         struct diff_queue_struct outq;
6464
6465         DIFF_QUEUE_CLEAR(&outq);
6466
6467         if (!options->filter)
6468                 return;
6469
6470         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6471                 int found;
6472                 for (i = found = 0; !found && i < q->nr; i++) {
6473                         if (match_filter(options, q->queue[i]))
6474                                 found++;
6475                 }
6476                 if (found)
6477                         return;
6478
6479                 /* otherwise we will clear the whole queue
6480                  * by copying the empty outq at the end of this
6481                  * function, but first clear the current entries
6482                  * in the queue.
6483                  */
6484                 for (i = 0; i < q->nr; i++)
6485                         diff_free_filepair(q->queue[i]);
6486         }
6487         else {
6488                 /* Only the matching ones */
6489                 for (i = 0; i < q->nr; i++) {
6490                         struct diff_filepair *p = q->queue[i];
6491                         if (match_filter(options, p))
6492                                 diff_q(&outq, p);
6493                         else
6494                                 diff_free_filepair(p);
6495                 }
6496         }
6497         free(q->queue);
6498         *q = outq;
6499 }
6500
6501 /* Check whether two filespecs with the same mode and size are identical */
6502 static int diff_filespec_is_identical(struct repository *r,
6503                                       struct diff_filespec *one,
6504                                       struct diff_filespec *two)
6505 {
6506         if (S_ISGITLINK(one->mode))
6507                 return 0;
6508         if (diff_populate_filespec(r, one, NULL))
6509                 return 0;
6510         if (diff_populate_filespec(r, two, NULL))
6511                 return 0;
6512         return !memcmp(one->data, two->data, one->size);
6513 }
6514
6515 static int diff_filespec_check_stat_unmatch(struct repository *r,
6516                                             struct diff_filepair *p)
6517 {
6518         struct diff_populate_filespec_options dpf_options = {
6519                 .check_size_only = 1,
6520                 .missing_object_cb = diff_queued_diff_prefetch,
6521                 .missing_object_data = r,
6522         };
6523
6524         if (p->done_skip_stat_unmatch)
6525                 return p->skip_stat_unmatch_result;
6526
6527         p->done_skip_stat_unmatch = 1;
6528         p->skip_stat_unmatch_result = 0;
6529         /*
6530          * 1. Entries that come from stat info dirtiness
6531          *    always have both sides (iow, not create/delete),
6532          *    one side of the object name is unknown, with
6533          *    the same mode and size.  Keep the ones that
6534          *    do not match these criteria.  They have real
6535          *    differences.
6536          *
6537          * 2. At this point, the file is known to be modified,
6538          *    with the same mode and size, and the object
6539          *    name of one side is unknown.  Need to inspect
6540          *    the identical contents.
6541          */
6542         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6543             !DIFF_FILE_VALID(p->two) ||
6544             (p->one->oid_valid && p->two->oid_valid) ||
6545             (p->one->mode != p->two->mode) ||
6546             diff_populate_filespec(r, p->one, &dpf_options) ||
6547             diff_populate_filespec(r, p->two, &dpf_options) ||
6548             (p->one->size != p->two->size) ||
6549             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6550                 p->skip_stat_unmatch_result = 1;
6551         return p->skip_stat_unmatch_result;
6552 }
6553
6554 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6555 {
6556         int i;
6557         struct diff_queue_struct *q = &diff_queued_diff;
6558         struct diff_queue_struct outq;
6559         DIFF_QUEUE_CLEAR(&outq);
6560
6561         for (i = 0; i < q->nr; i++) {
6562                 struct diff_filepair *p = q->queue[i];
6563
6564                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6565                         diff_q(&outq, p);
6566                 else {
6567                         /*
6568                          * The caller can subtract 1 from skip_stat_unmatch
6569                          * to determine how many paths were dirty only
6570                          * due to stat info mismatch.
6571                          */
6572                         if (!diffopt->flags.no_index)
6573                                 diffopt->skip_stat_unmatch++;
6574                         diff_free_filepair(p);
6575                 }
6576         }
6577         free(q->queue);
6578         *q = outq;
6579 }
6580
6581 static int diffnamecmp(const void *a_, const void *b_)
6582 {
6583         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6584         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6585         const char *name_a, *name_b;
6586
6587         name_a = a->one ? a->one->path : a->two->path;
6588         name_b = b->one ? b->one->path : b->two->path;
6589         return strcmp(name_a, name_b);
6590 }
6591
6592 void diffcore_fix_diff_index(void)
6593 {
6594         struct diff_queue_struct *q = &diff_queued_diff;
6595         QSORT(q->queue, q->nr, diffnamecmp);
6596 }
6597
6598 void diff_add_if_missing(struct repository *r,
6599                          struct oid_array *to_fetch,
6600                          const struct diff_filespec *filespec)
6601 {
6602         if (filespec && filespec->oid_valid &&
6603             !S_ISGITLINK(filespec->mode) &&
6604             oid_object_info_extended(r, &filespec->oid, NULL,
6605                                      OBJECT_INFO_FOR_PREFETCH))
6606                 oid_array_append(to_fetch, &filespec->oid);
6607 }
6608
6609 void diff_queued_diff_prefetch(void *repository)
6610 {
6611         struct repository *repo = repository;
6612         int i;
6613         struct diff_queue_struct *q = &diff_queued_diff;
6614         struct oid_array to_fetch = OID_ARRAY_INIT;
6615
6616         for (i = 0; i < q->nr; i++) {
6617                 struct diff_filepair *p = q->queue[i];
6618                 diff_add_if_missing(repo, &to_fetch, p->one);
6619                 diff_add_if_missing(repo, &to_fetch, p->two);
6620         }
6621
6622         /*
6623          * NEEDSWORK: Consider deduplicating the OIDs sent.
6624          */
6625         promisor_remote_get_direct(repo, to_fetch.oid, to_fetch.nr);
6626
6627         oid_array_clear(&to_fetch);
6628 }
6629
6630 void diffcore_std(struct diff_options *options)
6631 {
6632         int output_formats_to_prefetch = DIFF_FORMAT_DIFFSTAT |
6633                 DIFF_FORMAT_NUMSTAT |
6634                 DIFF_FORMAT_PATCH |
6635                 DIFF_FORMAT_SHORTSTAT |
6636                 DIFF_FORMAT_DIRSTAT;
6637
6638         /*
6639          * Check if the user requested a blob-data-requiring diff output and/or
6640          * break-rewrite detection (which requires blob data). If yes, prefetch
6641          * the diff pairs.
6642          *
6643          * If no prefetching occurs, diffcore_rename() will prefetch if it
6644          * decides that it needs inexact rename detection.
6645          */
6646         if (options->repo == the_repository && has_promisor_remote() &&
6647             (options->output_format & output_formats_to_prefetch ||
6648              options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
6649                 diff_queued_diff_prefetch(options->repo);
6650
6651         /* NOTE please keep the following in sync with diff_tree_combined() */
6652         if (options->skip_stat_unmatch)
6653                 diffcore_skip_stat_unmatch(options);
6654         if (!options->found_follow) {
6655                 /* See try_to_follow_renames() in tree-diff.c */
6656                 if (options->break_opt != -1)
6657                         diffcore_break(options->repo,
6658                                        options->break_opt);
6659                 if (options->detect_rename)
6660                         diffcore_rename(options);
6661                 if (options->break_opt != -1)
6662                         diffcore_merge_broken();
6663         }
6664         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6665                 diffcore_pickaxe(options);
6666         if (options->orderfile)
6667                 diffcore_order(options->orderfile);
6668         if (!options->found_follow)
6669                 /* See try_to_follow_renames() in tree-diff.c */
6670                 diff_resolve_rename_copy();
6671         diffcore_apply_filter(options);
6672
6673         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6674                 options->flags.has_changes = 1;
6675         else
6676                 options->flags.has_changes = 0;
6677
6678         options->found_follow = 0;
6679 }
6680
6681 int diff_result_code(struct diff_options *opt, int status)
6682 {
6683         int result = 0;
6684
6685         diff_warn_rename_limit("diff.renameLimit",
6686                                opt->needed_rename_limit,
6687                                opt->degraded_cc_to_c);
6688         if (!opt->flags.exit_with_status &&
6689             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6690                 return status;
6691         if (opt->flags.exit_with_status &&
6692             opt->flags.has_changes)
6693                 result |= 01;
6694         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6695             opt->flags.check_failed)
6696                 result |= 02;
6697         return result;
6698 }
6699
6700 int diff_can_quit_early(struct diff_options *opt)
6701 {
6702         return (opt->flags.quick &&
6703                 !opt->filter &&
6704                 opt->flags.has_changes);
6705 }
6706
6707 /*
6708  * Shall changes to this submodule be ignored?
6709  *
6710  * Submodule changes can be configured to be ignored separately for each path,
6711  * but that configuration can be overridden from the command line.
6712  */
6713 static int is_submodule_ignored(const char *path, struct diff_options *options)
6714 {
6715         int ignored = 0;
6716         struct diff_flags orig_flags = options->flags;
6717         if (!options->flags.override_submodule_config)
6718                 set_diffopt_flags_from_submodule_config(options, path);
6719         if (options->flags.ignore_submodules)
6720                 ignored = 1;
6721         options->flags = orig_flags;
6722         return ignored;
6723 }
6724
6725 void compute_diffstat(struct diff_options *options,
6726                       struct diffstat_t *diffstat,
6727                       struct diff_queue_struct *q)
6728 {
6729         int i;
6730
6731         memset(diffstat, 0, sizeof(struct diffstat_t));
6732         for (i = 0; i < q->nr; i++) {
6733                 struct diff_filepair *p = q->queue[i];
6734                 if (check_pair_status(p))
6735                         diff_flush_stat(p, options, diffstat);
6736         }
6737 }
6738
6739 void diff_addremove(struct diff_options *options,
6740                     int addremove, unsigned mode,
6741                     const struct object_id *oid,
6742                     int oid_valid,
6743                     const char *concatpath, unsigned dirty_submodule)
6744 {
6745         struct diff_filespec *one, *two;
6746
6747         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6748                 return;
6749
6750         /* This may look odd, but it is a preparation for
6751          * feeding "there are unchanged files which should
6752          * not produce diffs, but when you are doing copy
6753          * detection you would need them, so here they are"
6754          * entries to the diff-core.  They will be prefixed
6755          * with something like '=' or '*' (I haven't decided
6756          * which but should not make any difference).
6757          * Feeding the same new and old to diff_change()
6758          * also has the same effect.
6759          * Before the final output happens, they are pruned after
6760          * merged into rename/copy pairs as appropriate.
6761          */
6762         if (options->flags.reverse_diff)
6763                 addremove = (addremove == '+' ? '-' :
6764                              addremove == '-' ? '+' : addremove);
6765
6766         if (options->prefix &&
6767             strncmp(concatpath, options->prefix, options->prefix_length))
6768                 return;
6769
6770         one = alloc_filespec(concatpath);
6771         two = alloc_filespec(concatpath);
6772
6773         if (addremove != '+')
6774                 fill_filespec(one, oid, oid_valid, mode);
6775         if (addremove != '-') {
6776                 fill_filespec(two, oid, oid_valid, mode);
6777                 two->dirty_submodule = dirty_submodule;
6778         }
6779
6780         diff_queue(&diff_queued_diff, one, two);
6781         if (!options->flags.diff_from_contents)
6782                 options->flags.has_changes = 1;
6783 }
6784
6785 void diff_change(struct diff_options *options,
6786                  unsigned old_mode, unsigned new_mode,
6787                  const struct object_id *old_oid,
6788                  const struct object_id *new_oid,
6789                  int old_oid_valid, int new_oid_valid,
6790                  const char *concatpath,
6791                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6792 {
6793         struct diff_filespec *one, *two;
6794         struct diff_filepair *p;
6795
6796         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6797             is_submodule_ignored(concatpath, options))
6798                 return;
6799
6800         if (options->flags.reverse_diff) {
6801                 SWAP(old_mode, new_mode);
6802                 SWAP(old_oid, new_oid);
6803                 SWAP(old_oid_valid, new_oid_valid);
6804                 SWAP(old_dirty_submodule, new_dirty_submodule);
6805         }
6806
6807         if (options->prefix &&
6808             strncmp(concatpath, options->prefix, options->prefix_length))
6809                 return;
6810
6811         one = alloc_filespec(concatpath);
6812         two = alloc_filespec(concatpath);
6813         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6814         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6815         one->dirty_submodule = old_dirty_submodule;
6816         two->dirty_submodule = new_dirty_submodule;
6817         p = diff_queue(&diff_queued_diff, one, two);
6818
6819         if (options->flags.diff_from_contents)
6820                 return;
6821
6822         if (options->flags.quick && options->skip_stat_unmatch &&
6823             !diff_filespec_check_stat_unmatch(options->repo, p)) {
6824                 diff_free_filespec_data(p->one);
6825                 diff_free_filespec_data(p->two);
6826                 return;
6827         }
6828
6829         options->flags.has_changes = 1;
6830 }
6831
6832 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6833 {
6834         struct diff_filepair *pair;
6835         struct diff_filespec *one, *two;
6836
6837         if (options->prefix &&
6838             strncmp(path, options->prefix, options->prefix_length))
6839                 return NULL;
6840
6841         one = alloc_filespec(path);
6842         two = alloc_filespec(path);
6843         pair = diff_queue(&diff_queued_diff, one, two);
6844         pair->is_unmerged = 1;
6845         return pair;
6846 }
6847
6848 static char *run_textconv(struct repository *r,
6849                           const char *pgm,
6850                           struct diff_filespec *spec,
6851                           size_t *outsize)
6852 {
6853         struct diff_tempfile *temp;
6854         const char *argv[3];
6855         const char **arg = argv;
6856         struct child_process child = CHILD_PROCESS_INIT;
6857         struct strbuf buf = STRBUF_INIT;
6858         int err = 0;
6859
6860         temp = prepare_temp_file(r, spec->path, spec);
6861         *arg++ = pgm;
6862         *arg++ = temp->name;
6863         *arg = NULL;
6864
6865         child.use_shell = 1;
6866         child.argv = argv;
6867         child.out = -1;
6868         if (start_command(&child)) {
6869                 remove_tempfile();
6870                 return NULL;
6871         }
6872
6873         if (strbuf_read(&buf, child.out, 0) < 0)
6874                 err = error("error reading from textconv command '%s'", pgm);
6875         close(child.out);
6876
6877         if (finish_command(&child) || err) {
6878                 strbuf_release(&buf);
6879                 remove_tempfile();
6880                 return NULL;
6881         }
6882         remove_tempfile();
6883
6884         return strbuf_detach(&buf, outsize);
6885 }
6886
6887 size_t fill_textconv(struct repository *r,
6888                      struct userdiff_driver *driver,
6889                      struct diff_filespec *df,
6890                      char **outbuf)
6891 {
6892         size_t size;
6893
6894         if (!driver) {
6895                 if (!DIFF_FILE_VALID(df)) {
6896                         *outbuf = "";
6897                         return 0;
6898                 }
6899                 if (diff_populate_filespec(r, df, NULL))
6900                         die("unable to read files to diff");
6901                 *outbuf = df->data;
6902                 return df->size;
6903         }
6904
6905         if (!driver->textconv)
6906                 BUG("fill_textconv called with non-textconv driver");
6907
6908         if (driver->textconv_cache && df->oid_valid) {
6909                 *outbuf = notes_cache_get(driver->textconv_cache,
6910                                           &df->oid,
6911                                           &size);
6912                 if (*outbuf)
6913                         return size;
6914         }
6915
6916         *outbuf = run_textconv(r, driver->textconv, df, &size);
6917         if (!*outbuf)
6918                 die("unable to read files to diff");
6919
6920         if (driver->textconv_cache && df->oid_valid) {
6921                 /* ignore errors, as we might be in a readonly repository */
6922                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6923                                 size);
6924                 /*
6925                  * we could save up changes and flush them all at the end,
6926                  * but we would need an extra call after all diffing is done.
6927                  * Since generating a cache entry is the slow path anyway,
6928                  * this extra overhead probably isn't a big deal.
6929                  */
6930                 notes_cache_write(driver->textconv_cache);
6931         }
6932
6933         return size;
6934 }
6935
6936 int textconv_object(struct repository *r,
6937                     const char *path,
6938                     unsigned mode,
6939                     const struct object_id *oid,
6940                     int oid_valid,
6941                     char **buf,
6942                     unsigned long *buf_size)
6943 {
6944         struct diff_filespec *df;
6945         struct userdiff_driver *textconv;
6946
6947         df = alloc_filespec(path);
6948         fill_filespec(df, oid, oid_valid, mode);
6949         textconv = get_textconv(r, df);
6950         if (!textconv) {
6951                 free_filespec(df);
6952                 return 0;
6953         }
6954
6955         *buf_size = fill_textconv(r, textconv, df, buf);
6956         free_filespec(df);
6957         return 1;
6958 }
6959
6960 void setup_diff_pager(struct diff_options *opt)
6961 {
6962         /*
6963          * If the user asked for our exit code, then either they want --quiet
6964          * or --exit-code. We should definitely not bother with a pager in the
6965          * former case, as we will generate no output. Since we still properly
6966          * report our exit code even when a pager is run, we _could_ run a
6967          * pager with --exit-code. But since we have not done so historically,
6968          * and because it is easy to find people oneline advising "git diff
6969          * --exit-code" in hooks and other scripts, we do not do so.
6970          */
6971         if (!opt->flags.exit_with_status &&
6972             check_pager_config("diff") != 0)
6973                 setup_pager();
6974 }