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         while (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                         if (*begin == *end)
2065                                 (*begin)++;
2066                         else
2067                                 return *begin > *end;
2068                 } else {
2069                         return -1;
2070                 }
2071         }
2072
2073         /* find the next word */
2074         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2075                 (*begin)++;
2076         if (*begin >= buffer->size)
2077                 return -1;
2078
2079         /* find the end of the word */
2080         *end = *begin + 1;
2081         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2082                 (*end)++;
2083
2084         return 0;
2085 }
2086
2087 /*
2088  * This function splits the words in buffer->text, stores the list with
2089  * newline separator into out, and saves the offsets of the original words
2090  * in buffer->orig.
2091  */
2092 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2093                 regex_t *word_regex)
2094 {
2095         int i, j;
2096         long alloc = 0;
2097
2098         out->size = 0;
2099         out->ptr = NULL;
2100
2101         /* fake an empty "0th" word */
2102         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2103         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2104         buffer->orig_nr = 1;
2105
2106         for (i = 0; i < buffer->text.size; i++) {
2107                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2108                         return;
2109
2110                 /* store original boundaries */
2111                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2112                                 buffer->orig_alloc);
2113                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2114                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2115                 buffer->orig_nr++;
2116
2117                 /* store one word */
2118                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2119                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2120                 out->ptr[out->size + j - i] = '\n';
2121                 out->size += j - i + 1;
2122
2123                 i = j - 1;
2124         }
2125 }
2126
2127 /* this executes the word diff on the accumulated buffers */
2128 static void diff_words_show(struct diff_words_data *diff_words)
2129 {
2130         xpparam_t xpp;
2131         xdemitconf_t xecfg;
2132         mmfile_t minus, plus;
2133         struct diff_words_style *style = diff_words->style;
2134
2135         struct diff_options *opt = diff_words->opt;
2136         const char *line_prefix;
2137
2138         assert(opt);
2139         line_prefix = diff_line_prefix(opt);
2140
2141         /* special case: only removal */
2142         if (!diff_words->plus.text.size) {
2143                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2144                                  line_prefix, strlen(line_prefix), 0);
2145                 fn_out_diff_words_write_helper(diff_words->opt,
2146                         &style->old_word, style->newline,
2147                         diff_words->minus.text.size,
2148                         diff_words->minus.text.ptr);
2149                 diff_words->minus.text.size = 0;
2150                 return;
2151         }
2152
2153         diff_words->current_plus = diff_words->plus.text.ptr;
2154         diff_words->last_minus = 0;
2155
2156         memset(&xpp, 0, sizeof(xpp));
2157         memset(&xecfg, 0, sizeof(xecfg));
2158         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2159         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2160         xpp.flags = 0;
2161         /* as only the hunk header will be parsed, we need a 0-context */
2162         xecfg.ctxlen = 0;
2163         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2164                           diff_words, &xpp, &xecfg))
2165                 die("unable to generate word diff");
2166         free(minus.ptr);
2167         free(plus.ptr);
2168         if (diff_words->current_plus != diff_words->plus.text.ptr +
2169                         diff_words->plus.text.size) {
2170                 if (color_words_output_graph_prefix(diff_words))
2171                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2172                                          line_prefix, strlen(line_prefix), 0);
2173                 fn_out_diff_words_write_helper(diff_words->opt,
2174                         &style->ctx, style->newline,
2175                         diff_words->plus.text.ptr + diff_words->plus.text.size
2176                         - diff_words->current_plus, diff_words->current_plus);
2177         }
2178         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2179 }
2180
2181 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2182 static void diff_words_flush(struct emit_callback *ecbdata)
2183 {
2184         struct diff_options *wo = ecbdata->diff_words->opt;
2185
2186         if (ecbdata->diff_words->minus.text.size ||
2187             ecbdata->diff_words->plus.text.size)
2188                 diff_words_show(ecbdata->diff_words);
2189
2190         if (wo->emitted_symbols) {
2191                 struct diff_options *o = ecbdata->opt;
2192                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2193                 int i;
2194
2195                 /*
2196                  * NEEDSWORK:
2197                  * Instead of appending each, concat all words to a line?
2198                  */
2199                 for (i = 0; i < wol->nr; i++)
2200                         append_emitted_diff_symbol(o, &wol->buf[i]);
2201
2202                 for (i = 0; i < wol->nr; i++)
2203                         free((void *)wol->buf[i].line);
2204
2205                 wol->nr = 0;
2206         }
2207 }
2208
2209 static void diff_filespec_load_driver(struct diff_filespec *one,
2210                                       struct index_state *istate)
2211 {
2212         /* Use already-loaded driver */
2213         if (one->driver)
2214                 return;
2215
2216         if (S_ISREG(one->mode))
2217                 one->driver = userdiff_find_by_path(istate, one->path);
2218
2219         /* Fallback to default settings */
2220         if (!one->driver)
2221                 one->driver = userdiff_find_by_name("default");
2222 }
2223
2224 static const char *userdiff_word_regex(struct diff_filespec *one,
2225                                        struct index_state *istate)
2226 {
2227         diff_filespec_load_driver(one, istate);
2228         return one->driver->word_regex;
2229 }
2230
2231 static void init_diff_words_data(struct emit_callback *ecbdata,
2232                                  struct diff_options *orig_opts,
2233                                  struct diff_filespec *one,
2234                                  struct diff_filespec *two)
2235 {
2236         int i;
2237         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2238         memcpy(o, orig_opts, sizeof(struct diff_options));
2239
2240         CALLOC_ARRAY(ecbdata->diff_words, 1);
2241         ecbdata->diff_words->type = o->word_diff;
2242         ecbdata->diff_words->opt = o;
2243
2244         if (orig_opts->emitted_symbols)
2245                 CALLOC_ARRAY(o->emitted_symbols, 1);
2246
2247         if (!o->word_regex)
2248                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2249         if (!o->word_regex)
2250                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2251         if (!o->word_regex)
2252                 o->word_regex = diff_word_regex_cfg;
2253         if (o->word_regex) {
2254                 ecbdata->diff_words->word_regex = (regex_t *)
2255                         xmalloc(sizeof(regex_t));
2256                 if (regcomp(ecbdata->diff_words->word_regex,
2257                             o->word_regex,
2258                             REG_EXTENDED | REG_NEWLINE))
2259                         die("invalid regular expression: %s",
2260                             o->word_regex);
2261         }
2262         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2263                 if (o->word_diff == diff_words_styles[i].type) {
2264                         ecbdata->diff_words->style =
2265                                 &diff_words_styles[i];
2266                         break;
2267                 }
2268         }
2269         if (want_color(o->use_color)) {
2270                 struct diff_words_style *st = ecbdata->diff_words->style;
2271                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2272                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2273                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2274         }
2275 }
2276
2277 static void free_diff_words_data(struct emit_callback *ecbdata)
2278 {
2279         if (ecbdata->diff_words) {
2280                 diff_words_flush(ecbdata);
2281                 free (ecbdata->diff_words->opt->emitted_symbols);
2282                 free (ecbdata->diff_words->opt);
2283                 free (ecbdata->diff_words->minus.text.ptr);
2284                 free (ecbdata->diff_words->minus.orig);
2285                 free (ecbdata->diff_words->plus.text.ptr);
2286                 free (ecbdata->diff_words->plus.orig);
2287                 if (ecbdata->diff_words->word_regex) {
2288                         regfree(ecbdata->diff_words->word_regex);
2289                         free(ecbdata->diff_words->word_regex);
2290                 }
2291                 FREE_AND_NULL(ecbdata->diff_words);
2292         }
2293 }
2294
2295 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2296 {
2297         if (want_color(diff_use_color))
2298                 return diff_colors[ix];
2299         return "";
2300 }
2301
2302 const char *diff_line_prefix(struct diff_options *opt)
2303 {
2304         struct strbuf *msgbuf;
2305         if (!opt->output_prefix)
2306                 return "";
2307
2308         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2309         return msgbuf->buf;
2310 }
2311
2312 static unsigned long sane_truncate_line(char *line, unsigned long len)
2313 {
2314         const char *cp;
2315         unsigned long allot;
2316         size_t l = len;
2317
2318         cp = line;
2319         allot = l;
2320         while (0 < l) {
2321                 (void) utf8_width(&cp, &l);
2322                 if (!cp)
2323                         break; /* truncated in the middle? */
2324         }
2325         return allot - l;
2326 }
2327
2328 static void find_lno(const char *line, struct emit_callback *ecbdata)
2329 {
2330         const char *p;
2331         ecbdata->lno_in_preimage = 0;
2332         ecbdata->lno_in_postimage = 0;
2333         p = strchr(line, '-');
2334         if (!p)
2335                 return; /* cannot happen */
2336         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2337         p = strchr(p, '+');
2338         if (!p)
2339                 return; /* cannot happen */
2340         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2341 }
2342
2343 static int fn_out_consume(void *priv, char *line, unsigned long len)
2344 {
2345         struct emit_callback *ecbdata = priv;
2346         struct diff_options *o = ecbdata->opt;
2347
2348         o->found_changes = 1;
2349
2350         if (ecbdata->header) {
2351                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2352                                  ecbdata->header->buf, ecbdata->header->len, 0);
2353                 strbuf_reset(ecbdata->header);
2354                 ecbdata->header = NULL;
2355         }
2356
2357         if (ecbdata->label_path[0]) {
2358                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2359                                  ecbdata->label_path[0],
2360                                  strlen(ecbdata->label_path[0]), 0);
2361                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2362                                  ecbdata->label_path[1],
2363                                  strlen(ecbdata->label_path[1]), 0);
2364                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2365         }
2366
2367         if (diff_suppress_blank_empty
2368             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2369                 line[0] = '\n';
2370                 len = 1;
2371         }
2372
2373         if (line[0] == '@') {
2374                 if (ecbdata->diff_words)
2375                         diff_words_flush(ecbdata);
2376                 len = sane_truncate_line(line, len);
2377                 find_lno(line, ecbdata);
2378                 emit_hunk_header(ecbdata, line, len);
2379                 return 0;
2380         }
2381
2382         if (ecbdata->diff_words) {
2383                 enum diff_symbol s =
2384                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2385                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2386                 if (line[0] == '-') {
2387                         diff_words_append(line, len,
2388                                           &ecbdata->diff_words->minus);
2389                         return 0;
2390                 } else if (line[0] == '+') {
2391                         diff_words_append(line, len,
2392                                           &ecbdata->diff_words->plus);
2393                         return 0;
2394                 } else if (starts_with(line, "\\ ")) {
2395                         /*
2396                          * Eat the "no newline at eof" marker as if we
2397                          * saw a "+" or "-" line with nothing on it,
2398                          * and return without diff_words_flush() to
2399                          * defer processing. If this is the end of
2400                          * preimage, more "+" lines may come after it.
2401                          */
2402                         return 0;
2403                 }
2404                 diff_words_flush(ecbdata);
2405                 emit_diff_symbol(o, s, line, len, 0);
2406                 return 0;
2407         }
2408
2409         switch (line[0]) {
2410         case '+':
2411                 ecbdata->lno_in_postimage++;
2412                 emit_add_line(ecbdata, line + 1, len - 1);
2413                 break;
2414         case '-':
2415                 ecbdata->lno_in_preimage++;
2416                 emit_del_line(ecbdata, line + 1, len - 1);
2417                 break;
2418         case ' ':
2419                 ecbdata->lno_in_postimage++;
2420                 ecbdata->lno_in_preimage++;
2421                 emit_context_line(ecbdata, line + 1, len - 1);
2422                 break;
2423         default:
2424                 /* incomplete line at the end */
2425                 ecbdata->lno_in_preimage++;
2426                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2427                                  line, len, 0);
2428                 break;
2429         }
2430         return 0;
2431 }
2432
2433 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2434 {
2435         const char *old_name = a;
2436         const char *new_name = b;
2437         int pfx_length, sfx_length;
2438         int pfx_adjust_for_slash;
2439         int len_a = strlen(a);
2440         int len_b = strlen(b);
2441         int a_midlen, b_midlen;
2442         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2443         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2444
2445         if (qlen_a || qlen_b) {
2446                 quote_c_style(a, name, NULL, 0);
2447                 strbuf_addstr(name, " => ");
2448                 quote_c_style(b, name, NULL, 0);
2449                 return;
2450         }
2451
2452         /* Find common prefix */
2453         pfx_length = 0;
2454         while (*old_name && *new_name && *old_name == *new_name) {
2455                 if (*old_name == '/')
2456                         pfx_length = old_name - a + 1;
2457                 old_name++;
2458                 new_name++;
2459         }
2460
2461         /* Find common suffix */
2462         old_name = a + len_a;
2463         new_name = b + len_b;
2464         sfx_length = 0;
2465         /*
2466          * If there is a common prefix, it must end in a slash.  In
2467          * that case we let this loop run 1 into the prefix to see the
2468          * same slash.
2469          *
2470          * If there is no common prefix, we cannot do this as it would
2471          * underrun the input strings.
2472          */
2473         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2474         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2475                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2476                *old_name == *new_name) {
2477                 if (*old_name == '/')
2478                         sfx_length = len_a - (old_name - a);
2479                 old_name--;
2480                 new_name--;
2481         }
2482
2483         /*
2484          * pfx{mid-a => mid-b}sfx
2485          * {pfx-a => pfx-b}sfx
2486          * pfx{sfx-a => sfx-b}
2487          * name-a => name-b
2488          */
2489         a_midlen = len_a - pfx_length - sfx_length;
2490         b_midlen = len_b - pfx_length - sfx_length;
2491         if (a_midlen < 0)
2492                 a_midlen = 0;
2493         if (b_midlen < 0)
2494                 b_midlen = 0;
2495
2496         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2497         if (pfx_length + sfx_length) {
2498                 strbuf_add(name, a, pfx_length);
2499                 strbuf_addch(name, '{');
2500         }
2501         strbuf_add(name, a + pfx_length, a_midlen);
2502         strbuf_addstr(name, " => ");
2503         strbuf_add(name, b + pfx_length, b_midlen);
2504         if (pfx_length + sfx_length) {
2505                 strbuf_addch(name, '}');
2506                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2507         }
2508 }
2509
2510 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2511                                           const char *name_a,
2512                                           const char *name_b)
2513 {
2514         struct diffstat_file *x;
2515         CALLOC_ARRAY(x, 1);
2516         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2517         diffstat->files[diffstat->nr++] = x;
2518         if (name_b) {
2519                 x->from_name = xstrdup(name_a);
2520                 x->name = xstrdup(name_b);
2521                 x->is_renamed = 1;
2522         }
2523         else {
2524                 x->from_name = NULL;
2525                 x->name = xstrdup(name_a);
2526         }
2527         return x;
2528 }
2529
2530 static int diffstat_consume(void *priv, char *line, unsigned long len)
2531 {
2532         struct diffstat_t *diffstat = priv;
2533         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2534
2535         if (line[0] == '+')
2536                 x->added++;
2537         else if (line[0] == '-')
2538                 x->deleted++;
2539         return 0;
2540 }
2541
2542 const char mime_boundary_leader[] = "------------";
2543
2544 static int scale_linear(int it, int width, int max_change)
2545 {
2546         if (!it)
2547                 return 0;
2548         /*
2549          * make sure that at least one '-' or '+' is printed if
2550          * there is any change to this path. The easiest way is to
2551          * scale linearly as if the allotted width is one column shorter
2552          * than it is, and then add 1 to the result.
2553          */
2554         return 1 + (it * (width - 1) / max_change);
2555 }
2556
2557 static void show_graph(struct strbuf *out, char ch, int cnt,
2558                        const char *set, const char *reset)
2559 {
2560         if (cnt <= 0)
2561                 return;
2562         strbuf_addstr(out, set);
2563         strbuf_addchars(out, ch, cnt);
2564         strbuf_addstr(out, reset);
2565 }
2566
2567 static void fill_print_name(struct diffstat_file *file)
2568 {
2569         struct strbuf pname = STRBUF_INIT;
2570
2571         if (file->print_name)
2572                 return;
2573
2574         if (file->is_renamed)
2575                 pprint_rename(&pname, file->from_name, file->name);
2576         else
2577                 quote_c_style(file->name, &pname, NULL, 0);
2578
2579         if (file->comments)
2580                 strbuf_addf(&pname, " (%s)", file->comments);
2581
2582         file->print_name = strbuf_detach(&pname, NULL);
2583 }
2584
2585 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2586                 int files, int insertions, int deletions)
2587 {
2588         struct strbuf sb = STRBUF_INIT;
2589
2590         if (!files) {
2591                 assert(insertions == 0 && deletions == 0);
2592                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2593                                  NULL, 0, 0);
2594                 return;
2595         }
2596
2597         strbuf_addf(&sb,
2598                     (files == 1) ? " %d file changed" : " %d files changed",
2599                     files);
2600
2601         /*
2602          * For binary diff, the caller may want to print "x files
2603          * changed" with insertions == 0 && deletions == 0.
2604          *
2605          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2606          * is probably less confusing (i.e skip over "2 files changed
2607          * but nothing about added/removed lines? Is this a bug in Git?").
2608          */
2609         if (insertions || deletions == 0) {
2610                 strbuf_addf(&sb,
2611                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2612                             insertions);
2613         }
2614
2615         if (deletions || insertions == 0) {
2616                 strbuf_addf(&sb,
2617                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2618                             deletions);
2619         }
2620         strbuf_addch(&sb, '\n');
2621         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2622                          sb.buf, sb.len, 0);
2623         strbuf_release(&sb);
2624 }
2625
2626 void print_stat_summary(FILE *fp, int files,
2627                         int insertions, int deletions)
2628 {
2629         struct diff_options o;
2630         memset(&o, 0, sizeof(o));
2631         o.file = fp;
2632
2633         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2634 }
2635
2636 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2637 {
2638         int i, len, add, del, adds = 0, dels = 0;
2639         uintmax_t max_change = 0, max_len = 0;
2640         int total_files = data->nr, count;
2641         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2642         const char *reset, *add_c, *del_c;
2643         int extra_shown = 0;
2644         const char *line_prefix = diff_line_prefix(options);
2645         struct strbuf out = STRBUF_INIT;
2646
2647         if (data->nr == 0)
2648                 return;
2649
2650         count = options->stat_count ? options->stat_count : data->nr;
2651
2652         reset = diff_get_color_opt(options, DIFF_RESET);
2653         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2654         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2655
2656         /*
2657          * Find the longest filename and max number of changes
2658          */
2659         for (i = 0; (i < count) && (i < data->nr); i++) {
2660                 struct diffstat_file *file = data->files[i];
2661                 uintmax_t change = file->added + file->deleted;
2662
2663                 if (!file->is_interesting && (change == 0)) {
2664                         count++; /* not shown == room for one more */
2665                         continue;
2666                 }
2667                 fill_print_name(file);
2668                 len = strlen(file->print_name);
2669                 if (max_len < len)
2670                         max_len = len;
2671
2672                 if (file->is_unmerged) {
2673                         /* "Unmerged" is 8 characters */
2674                         bin_width = bin_width < 8 ? 8 : bin_width;
2675                         continue;
2676                 }
2677                 if (file->is_binary) {
2678                         /* "Bin XXX -> YYY bytes" */
2679                         int w = 14 + decimal_width(file->added)
2680                                 + decimal_width(file->deleted);
2681                         bin_width = bin_width < w ? w : bin_width;
2682                         /* Display change counts aligned with "Bin" */
2683                         number_width = 3;
2684                         continue;
2685                 }
2686
2687                 if (max_change < change)
2688                         max_change = change;
2689         }
2690         count = i; /* where we can stop scanning in data->files[] */
2691
2692         /*
2693          * We have width = stat_width or term_columns() columns total.
2694          * We want a maximum of min(max_len, stat_name_width) for the name part.
2695          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2696          * We also need 1 for " " and 4 + decimal_width(max_change)
2697          * for " | NNNN " and one the empty column at the end, altogether
2698          * 6 + decimal_width(max_change).
2699          *
2700          * If there's not enough space, we will use the smaller of
2701          * stat_name_width (if set) and 5/8*width for the filename,
2702          * and the rest for constant elements + graph part, but no more
2703          * than stat_graph_width for the graph part.
2704          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2705          * for the standard terminal size).
2706          *
2707          * In other words: stat_width limits the maximum width, and
2708          * stat_name_width fixes the maximum width of the filename,
2709          * and is also used to divide available columns if there
2710          * aren't enough.
2711          *
2712          * Binary files are displayed with "Bin XXX -> YYY bytes"
2713          * instead of the change count and graph. This part is treated
2714          * similarly to the graph part, except that it is not
2715          * "scaled". If total width is too small to accommodate the
2716          * guaranteed minimum width of the filename part and the
2717          * separators and this message, this message will "overflow"
2718          * making the line longer than the maximum width.
2719          */
2720
2721         if (options->stat_width == -1)
2722                 width = term_columns() - strlen(line_prefix);
2723         else
2724                 width = options->stat_width ? options->stat_width : 80;
2725         number_width = decimal_width(max_change) > number_width ?
2726                 decimal_width(max_change) : number_width;
2727
2728         if (options->stat_graph_width == -1)
2729                 options->stat_graph_width = diff_stat_graph_width;
2730
2731         /*
2732          * Guarantee 3/8*16==6 for the graph part
2733          * and 5/8*16==10 for the filename part
2734          */
2735         if (width < 16 + 6 + number_width)
2736                 width = 16 + 6 + number_width;
2737
2738         /*
2739          * First assign sizes that are wanted, ignoring available width.
2740          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2741          * starting from "XXX" should fit in graph_width.
2742          */
2743         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2744         if (options->stat_graph_width &&
2745             options->stat_graph_width < graph_width)
2746                 graph_width = options->stat_graph_width;
2747
2748         name_width = (options->stat_name_width > 0 &&
2749                       options->stat_name_width < max_len) ?
2750                 options->stat_name_width : max_len;
2751
2752         /*
2753          * Adjust adjustable widths not to exceed maximum width
2754          */
2755         if (name_width + number_width + 6 + graph_width > width) {
2756                 if (graph_width > width * 3/8 - number_width - 6) {
2757                         graph_width = width * 3/8 - number_width - 6;
2758                         if (graph_width < 6)
2759                                 graph_width = 6;
2760                 }
2761
2762                 if (options->stat_graph_width &&
2763                     graph_width > options->stat_graph_width)
2764                         graph_width = options->stat_graph_width;
2765                 if (name_width > width - number_width - 6 - graph_width)
2766                         name_width = width - number_width - 6 - graph_width;
2767                 else
2768                         graph_width = width - number_width - 6 - name_width;
2769         }
2770
2771         /*
2772          * From here name_width is the width of the name area,
2773          * and graph_width is the width of the graph area.
2774          * max_change is used to scale graph properly.
2775          */
2776         for (i = 0; i < count; i++) {
2777                 const char *prefix = "";
2778                 struct diffstat_file *file = data->files[i];
2779                 char *name = file->print_name;
2780                 uintmax_t added = file->added;
2781                 uintmax_t deleted = file->deleted;
2782                 int name_len;
2783
2784                 if (!file->is_interesting && (added + deleted == 0))
2785                         continue;
2786
2787                 /*
2788                  * "scale" the filename
2789                  */
2790                 len = name_width;
2791                 name_len = strlen(name);
2792                 if (name_width < name_len) {
2793                         char *slash;
2794                         prefix = "...";
2795                         len -= 3;
2796                         name += name_len - len;
2797                         slash = strchr(name, '/');
2798                         if (slash)
2799                                 name = slash;
2800                 }
2801
2802                 if (file->is_binary) {
2803                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2804                         strbuf_addf(&out, " %*s", number_width, "Bin");
2805                         if (!added && !deleted) {
2806                                 strbuf_addch(&out, '\n');
2807                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2808                                                  out.buf, out.len, 0);
2809                                 strbuf_reset(&out);
2810                                 continue;
2811                         }
2812                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2813                                 del_c, deleted, reset);
2814                         strbuf_addstr(&out, " -> ");
2815                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2816                                 add_c, added, reset);
2817                         strbuf_addstr(&out, " bytes\n");
2818                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2819                                          out.buf, out.len, 0);
2820                         strbuf_reset(&out);
2821                         continue;
2822                 }
2823                 else if (file->is_unmerged) {
2824                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2825                         strbuf_addstr(&out, " Unmerged\n");
2826                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2827                                          out.buf, out.len, 0);
2828                         strbuf_reset(&out);
2829                         continue;
2830                 }
2831
2832                 /*
2833                  * scale the add/delete
2834                  */
2835                 add = added;
2836                 del = deleted;
2837
2838                 if (graph_width <= max_change) {
2839                         int total = scale_linear(add + del, graph_width, max_change);
2840                         if (total < 2 && add && del)
2841                                 /* width >= 2 due to the sanity check */
2842                                 total = 2;
2843                         if (add < del) {
2844                                 add = scale_linear(add, graph_width, max_change);
2845                                 del = total - add;
2846                         } else {
2847                                 del = scale_linear(del, graph_width, max_change);
2848                                 add = total - del;
2849                         }
2850                 }
2851                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2852                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2853                         number_width, added + deleted,
2854                         added + deleted ? " " : "");
2855                 show_graph(&out, '+', add, add_c, reset);
2856                 show_graph(&out, '-', del, del_c, reset);
2857                 strbuf_addch(&out, '\n');
2858                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2859                                  out.buf, out.len, 0);
2860                 strbuf_reset(&out);
2861         }
2862
2863         for (i = 0; i < data->nr; i++) {
2864                 struct diffstat_file *file = data->files[i];
2865                 uintmax_t added = file->added;
2866                 uintmax_t deleted = file->deleted;
2867
2868                 if (file->is_unmerged ||
2869                     (!file->is_interesting && (added + deleted == 0))) {
2870                         total_files--;
2871                         continue;
2872                 }
2873
2874                 if (!file->is_binary) {
2875                         adds += added;
2876                         dels += deleted;
2877                 }
2878                 if (i < count)
2879                         continue;
2880                 if (!extra_shown)
2881                         emit_diff_symbol(options,
2882                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2883                                          NULL, 0, 0);
2884                 extra_shown = 1;
2885         }
2886
2887         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2888         strbuf_release(&out);
2889 }
2890
2891 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2892 {
2893         int i, adds = 0, dels = 0, total_files = data->nr;
2894
2895         if (data->nr == 0)
2896                 return;
2897
2898         for (i = 0; i < data->nr; i++) {
2899                 int added = data->files[i]->added;
2900                 int deleted = data->files[i]->deleted;
2901
2902                 if (data->files[i]->is_unmerged ||
2903                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2904                         total_files--;
2905                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2906                         adds += added;
2907                         dels += deleted;
2908                 }
2909         }
2910         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2911 }
2912
2913 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2914 {
2915         int i;
2916
2917         if (data->nr == 0)
2918                 return;
2919
2920         for (i = 0; i < data->nr; i++) {
2921                 struct diffstat_file *file = data->files[i];
2922
2923                 fprintf(options->file, "%s", diff_line_prefix(options));
2924
2925                 if (file->is_binary)
2926                         fprintf(options->file, "-\t-\t");
2927                 else
2928                         fprintf(options->file,
2929                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2930                                 file->added, file->deleted);
2931                 if (options->line_termination) {
2932                         fill_print_name(file);
2933                         if (!file->is_renamed)
2934                                 write_name_quoted(file->name, options->file,
2935                                                   options->line_termination);
2936                         else {
2937                                 fputs(file->print_name, options->file);
2938                                 putc(options->line_termination, options->file);
2939                         }
2940                 } else {
2941                         if (file->is_renamed) {
2942                                 putc('\0', options->file);
2943                                 write_name_quoted(file->from_name, options->file, '\0');
2944                         }
2945                         write_name_quoted(file->name, options->file, '\0');
2946                 }
2947         }
2948 }
2949
2950 struct dirstat_file {
2951         const char *name;
2952         unsigned long changed;
2953 };
2954
2955 struct dirstat_dir {
2956         struct dirstat_file *files;
2957         int alloc, nr, permille, cumulative;
2958 };
2959
2960 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2961                 unsigned long changed, const char *base, int baselen)
2962 {
2963         unsigned long sum_changes = 0;
2964         unsigned int sources = 0;
2965         const char *line_prefix = diff_line_prefix(opt);
2966
2967         while (dir->nr) {
2968                 struct dirstat_file *f = dir->files;
2969                 int namelen = strlen(f->name);
2970                 unsigned long changes;
2971                 char *slash;
2972
2973                 if (namelen < baselen)
2974                         break;
2975                 if (memcmp(f->name, base, baselen))
2976                         break;
2977                 slash = strchr(f->name + baselen, '/');
2978                 if (slash) {
2979                         int newbaselen = slash + 1 - f->name;
2980                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2981                         sources++;
2982                 } else {
2983                         changes = f->changed;
2984                         dir->files++;
2985                         dir->nr--;
2986                         sources += 2;
2987                 }
2988                 sum_changes += changes;
2989         }
2990
2991         /*
2992          * We don't report dirstat's for
2993          *  - the top level
2994          *  - or cases where everything came from a single directory
2995          *    under this directory (sources == 1).
2996          */
2997         if (baselen && sources != 1) {
2998                 if (sum_changes) {
2999                         int permille = sum_changes * 1000 / changed;
3000                         if (permille >= dir->permille) {
3001                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3002                                         permille / 10, permille % 10, baselen, base);
3003                                 if (!dir->cumulative)
3004                                         return 0;
3005                         }
3006                 }
3007         }
3008         return sum_changes;
3009 }
3010
3011 static int dirstat_compare(const void *_a, const void *_b)
3012 {
3013         const struct dirstat_file *a = _a;
3014         const struct dirstat_file *b = _b;
3015         return strcmp(a->name, b->name);
3016 }
3017
3018 static void show_dirstat(struct diff_options *options)
3019 {
3020         int i;
3021         unsigned long changed;
3022         struct dirstat_dir dir;
3023         struct diff_queue_struct *q = &diff_queued_diff;
3024
3025         dir.files = NULL;
3026         dir.alloc = 0;
3027         dir.nr = 0;
3028         dir.permille = options->dirstat_permille;
3029         dir.cumulative = options->flags.dirstat_cumulative;
3030
3031         changed = 0;
3032         for (i = 0; i < q->nr; i++) {
3033                 struct diff_filepair *p = q->queue[i];
3034                 const char *name;
3035                 unsigned long copied, added, damage;
3036                 struct diff_populate_filespec_options dpf_options = {
3037                         .check_size_only = 1,
3038                 };
3039
3040                 name = p->two->path ? p->two->path : p->one->path;
3041
3042                 if (p->one->oid_valid && p->two->oid_valid &&
3043                     oideq(&p->one->oid, &p->two->oid)) {
3044                         /*
3045                          * The SHA1 has not changed, so pre-/post-content is
3046                          * identical. We can therefore skip looking at the
3047                          * file contents altogether.
3048                          */
3049                         damage = 0;
3050                         goto found_damage;
3051                 }
3052
3053                 if (options->flags.dirstat_by_file) {
3054                         /*
3055                          * In --dirstat-by-file mode, we don't really need to
3056                          * look at the actual file contents at all.
3057                          * The fact that the SHA1 changed is enough for us to
3058                          * add this file to the list of results
3059                          * (with each file contributing equal damage).
3060                          */
3061                         damage = 1;
3062                         goto found_damage;
3063                 }
3064
3065                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3066                         diff_populate_filespec(options->repo, p->one, NULL);
3067                         diff_populate_filespec(options->repo, p->two, NULL);
3068                         diffcore_count_changes(options->repo,
3069                                                p->one, p->two, NULL, NULL,
3070                                                &copied, &added);
3071                         diff_free_filespec_data(p->one);
3072                         diff_free_filespec_data(p->two);
3073                 } else if (DIFF_FILE_VALID(p->one)) {
3074                         diff_populate_filespec(options->repo, p->one, &dpf_options);
3075                         copied = added = 0;
3076                         diff_free_filespec_data(p->one);
3077                 } else if (DIFF_FILE_VALID(p->two)) {
3078                         diff_populate_filespec(options->repo, p->two, &dpf_options);
3079                         copied = 0;
3080                         added = p->two->size;
3081                         diff_free_filespec_data(p->two);
3082                 } else
3083                         continue;
3084
3085                 /*
3086                  * Original minus copied is the removed material,
3087                  * added is the new material.  They are both damages
3088                  * made to the preimage.
3089                  * If the resulting damage is zero, we know that
3090                  * diffcore_count_changes() considers the two entries to
3091                  * be identical, but since the oid changed, we
3092                  * know that there must have been _some_ kind of change,
3093                  * so we force all entries to have damage > 0.
3094                  */
3095                 damage = (p->one->size - copied) + added;
3096                 if (!damage)
3097                         damage = 1;
3098
3099 found_damage:
3100                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3101                 dir.files[dir.nr].name = name;
3102                 dir.files[dir.nr].changed = damage;
3103                 changed += damage;
3104                 dir.nr++;
3105         }
3106
3107         /* This can happen even with many files, if everything was renames */
3108         if (!changed)
3109                 return;
3110
3111         /* Show all directories with more than x% of the changes */
3112         QSORT(dir.files, dir.nr, dirstat_compare);
3113         gather_dirstat(options, &dir, changed, "", 0);
3114 }
3115
3116 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3117 {
3118         int i;
3119         unsigned long changed;
3120         struct dirstat_dir dir;
3121
3122         if (data->nr == 0)
3123                 return;
3124
3125         dir.files = NULL;
3126         dir.alloc = 0;
3127         dir.nr = 0;
3128         dir.permille = options->dirstat_permille;
3129         dir.cumulative = options->flags.dirstat_cumulative;
3130
3131         changed = 0;
3132         for (i = 0; i < data->nr; i++) {
3133                 struct diffstat_file *file = data->files[i];
3134                 unsigned long damage = file->added + file->deleted;
3135                 if (file->is_binary)
3136                         /*
3137                          * binary files counts bytes, not lines. Must find some
3138                          * way to normalize binary bytes vs. textual lines.
3139                          * The following heuristic assumes that there are 64
3140                          * bytes per "line".
3141                          * This is stupid and ugly, but very cheap...
3142                          */
3143                         damage = DIV_ROUND_UP(damage, 64);
3144                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3145                 dir.files[dir.nr].name = file->name;
3146                 dir.files[dir.nr].changed = damage;
3147                 changed += damage;
3148                 dir.nr++;
3149         }
3150
3151         /* This can happen even with many files, if everything was renames */
3152         if (!changed)
3153                 return;
3154
3155         /* Show all directories with more than x% of the changes */
3156         QSORT(dir.files, dir.nr, dirstat_compare);
3157         gather_dirstat(options, &dir, changed, "", 0);
3158 }
3159
3160 static void free_diffstat_file(struct diffstat_file *f)
3161 {
3162         free(f->print_name);
3163         free(f->name);
3164         free(f->from_name);
3165         free(f);
3166 }
3167
3168 void free_diffstat_info(struct diffstat_t *diffstat)
3169 {
3170         int i;
3171         for (i = 0; i < diffstat->nr; i++)
3172                 free_diffstat_file(diffstat->files[i]);
3173         free(diffstat->files);
3174 }
3175
3176 struct checkdiff_t {
3177         const char *filename;
3178         int lineno;
3179         int conflict_marker_size;
3180         struct diff_options *o;
3181         unsigned ws_rule;
3182         unsigned status;
3183 };
3184
3185 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3186 {
3187         char firstchar;
3188         int cnt;
3189
3190         if (len < marker_size + 1)
3191                 return 0;
3192         firstchar = line[0];
3193         switch (firstchar) {
3194         case '=': case '>': case '<': case '|':
3195                 break;
3196         default:
3197                 return 0;
3198         }
3199         for (cnt = 1; cnt < marker_size; cnt++)
3200                 if (line[cnt] != firstchar)
3201                         return 0;
3202         /* line[1] through line[marker_size-1] are same as firstchar */
3203         if (len < marker_size + 1 || !isspace(line[marker_size]))
3204                 return 0;
3205         return 1;
3206 }
3207
3208 static void checkdiff_consume_hunk(void *priv,
3209                                    long ob, long on, long nb, long nn,
3210                                    const char *func, long funclen)
3211
3212 {
3213         struct checkdiff_t *data = priv;
3214         data->lineno = nb - 1;
3215 }
3216
3217 static int checkdiff_consume(void *priv, char *line, unsigned long len)
3218 {
3219         struct checkdiff_t *data = priv;
3220         int marker_size = data->conflict_marker_size;
3221         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3222         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3223         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3224         char *err;
3225         const char *line_prefix;
3226
3227         assert(data->o);
3228         line_prefix = diff_line_prefix(data->o);
3229
3230         if (line[0] == '+') {
3231                 unsigned bad;
3232                 data->lineno++;
3233                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3234                         data->status |= 1;
3235                         fprintf(data->o->file,
3236                                 "%s%s:%d: leftover conflict marker\n",
3237                                 line_prefix, data->filename, data->lineno);
3238                 }
3239                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3240                 if (!bad)
3241                         return 0;
3242                 data->status |= bad;
3243                 err = whitespace_error_string(bad);
3244                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3245                         line_prefix, data->filename, data->lineno, err);
3246                 free(err);
3247                 emit_line(data->o, set, reset, line, 1);
3248                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3249                               data->o->file, set, reset, ws);
3250         } else if (line[0] == ' ') {
3251                 data->lineno++;
3252         }
3253         return 0;
3254 }
3255
3256 static unsigned char *deflate_it(char *data,
3257                                  unsigned long size,
3258                                  unsigned long *result_size)
3259 {
3260         int bound;
3261         unsigned char *deflated;
3262         git_zstream stream;
3263
3264         git_deflate_init(&stream, zlib_compression_level);
3265         bound = git_deflate_bound(&stream, size);
3266         deflated = xmalloc(bound);
3267         stream.next_out = deflated;
3268         stream.avail_out = bound;
3269
3270         stream.next_in = (unsigned char *)data;
3271         stream.avail_in = size;
3272         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3273                 ; /* nothing */
3274         git_deflate_end(&stream);
3275         *result_size = stream.total_out;
3276         return deflated;
3277 }
3278
3279 static void emit_binary_diff_body(struct diff_options *o,
3280                                   mmfile_t *one, mmfile_t *two)
3281 {
3282         void *cp;
3283         void *delta;
3284         void *deflated;
3285         void *data;
3286         unsigned long orig_size;
3287         unsigned long delta_size;
3288         unsigned long deflate_size;
3289         unsigned long data_size;
3290
3291         /* We could do deflated delta, or we could do just deflated two,
3292          * whichever is smaller.
3293          */
3294         delta = NULL;
3295         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3296         if (one->size && two->size) {
3297                 delta = diff_delta(one->ptr, one->size,
3298                                    two->ptr, two->size,
3299                                    &delta_size, deflate_size);
3300                 if (delta) {
3301                         void *to_free = delta;
3302                         orig_size = delta_size;
3303                         delta = deflate_it(delta, delta_size, &delta_size);
3304                         free(to_free);
3305                 }
3306         }
3307
3308         if (delta && delta_size < deflate_size) {
3309                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3310                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3311                                  s, strlen(s), 0);
3312                 free(s);
3313                 free(deflated);
3314                 data = delta;
3315                 data_size = delta_size;
3316         } else {
3317                 char *s = xstrfmt("%lu", two->size);
3318                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3319                                  s, strlen(s), 0);
3320                 free(s);
3321                 free(delta);
3322                 data = deflated;
3323                 data_size = deflate_size;
3324         }
3325
3326         /* emit data encoded in base85 */
3327         cp = data;
3328         while (data_size) {
3329                 int len;
3330                 int bytes = (52 < data_size) ? 52 : data_size;
3331                 char line[71];
3332                 data_size -= bytes;
3333                 if (bytes <= 26)
3334                         line[0] = bytes + 'A' - 1;
3335                 else
3336                         line[0] = bytes - 26 + 'a' - 1;
3337                 encode_85(line + 1, cp, bytes);
3338                 cp = (char *) cp + bytes;
3339
3340                 len = strlen(line);
3341                 line[len++] = '\n';
3342                 line[len] = '\0';
3343
3344                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3345                                  line, len, 0);
3346         }
3347         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3348         free(data);
3349 }
3350
3351 static void emit_binary_diff(struct diff_options *o,
3352                              mmfile_t *one, mmfile_t *two)
3353 {
3354         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3355         emit_binary_diff_body(o, one, two);
3356         emit_binary_diff_body(o, two, one);
3357 }
3358
3359 int diff_filespec_is_binary(struct repository *r,
3360                             struct diff_filespec *one)
3361 {
3362         struct diff_populate_filespec_options dpf_options = {
3363                 .check_binary = 1,
3364         };
3365
3366         if (one->is_binary == -1) {
3367                 diff_filespec_load_driver(one, r->index);
3368                 if (one->driver->binary != -1)
3369                         one->is_binary = one->driver->binary;
3370                 else {
3371                         if (!one->data && DIFF_FILE_VALID(one))
3372                                 diff_populate_filespec(r, one, &dpf_options);
3373                         if (one->is_binary == -1 && one->data)
3374                                 one->is_binary = buffer_is_binary(one->data,
3375                                                 one->size);
3376                         if (one->is_binary == -1)
3377                                 one->is_binary = 0;
3378                 }
3379         }
3380         return one->is_binary;
3381 }
3382
3383 static const struct userdiff_funcname *
3384 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3385 {
3386         diff_filespec_load_driver(one, o->repo->index);
3387         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3388 }
3389
3390 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3391 {
3392         if (!options->a_prefix)
3393                 options->a_prefix = a;
3394         if (!options->b_prefix)
3395                 options->b_prefix = b;
3396 }
3397
3398 struct userdiff_driver *get_textconv(struct repository *r,
3399                                      struct diff_filespec *one)
3400 {
3401         if (!DIFF_FILE_VALID(one))
3402                 return NULL;
3403
3404         diff_filespec_load_driver(one, r->index);
3405         return userdiff_get_textconv(r, one->driver);
3406 }
3407
3408 static void builtin_diff(const char *name_a,
3409                          const char *name_b,
3410                          struct diff_filespec *one,
3411                          struct diff_filespec *two,
3412                          const char *xfrm_msg,
3413                          int must_show_header,
3414                          struct diff_options *o,
3415                          int complete_rewrite)
3416 {
3417         mmfile_t mf1, mf2;
3418         const char *lbl[2];
3419         char *a_one, *b_two;
3420         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3421         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3422         const char *a_prefix, *b_prefix;
3423         struct userdiff_driver *textconv_one = NULL;
3424         struct userdiff_driver *textconv_two = NULL;
3425         struct strbuf header = STRBUF_INIT;
3426         const char *line_prefix = diff_line_prefix(o);
3427
3428         diff_set_mnemonic_prefix(o, "a/", "b/");
3429         if (o->flags.reverse_diff) {
3430                 a_prefix = o->b_prefix;
3431                 b_prefix = o->a_prefix;
3432         } else {
3433                 a_prefix = o->a_prefix;
3434                 b_prefix = o->b_prefix;
3435         }
3436
3437         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3438             (!one->mode || S_ISGITLINK(one->mode)) &&
3439             (!two->mode || S_ISGITLINK(two->mode))) {
3440                 show_submodule_diff_summary(o, one->path ? one->path : two->path,
3441                                 &one->oid, &two->oid,
3442                                 two->dirty_submodule);
3443                 return;
3444         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3445                    (!one->mode || S_ISGITLINK(one->mode)) &&
3446                    (!two->mode || S_ISGITLINK(two->mode))) {
3447                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3448                                 &one->oid, &two->oid,
3449                                 two->dirty_submodule);
3450                 return;
3451         }
3452
3453         if (o->flags.allow_textconv) {
3454                 textconv_one = get_textconv(o->repo, one);
3455                 textconv_two = get_textconv(o->repo, two);
3456         }
3457
3458         /* Never use a non-valid filename anywhere if at all possible */
3459         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3460         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3461
3462         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3463         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3464         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3465         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3466         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3467         if (lbl[0][0] == '/') {
3468                 /* /dev/null */
3469                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3470                 if (xfrm_msg)
3471                         strbuf_addstr(&header, xfrm_msg);
3472                 must_show_header = 1;
3473         }
3474         else if (lbl[1][0] == '/') {
3475                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3476                 if (xfrm_msg)
3477                         strbuf_addstr(&header, xfrm_msg);
3478                 must_show_header = 1;
3479         }
3480         else {
3481                 if (one->mode != two->mode) {
3482                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3483                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3484                         must_show_header = 1;
3485                 }
3486                 if (xfrm_msg)
3487                         strbuf_addstr(&header, xfrm_msg);
3488
3489                 /*
3490                  * we do not run diff between different kind
3491                  * of objects.
3492                  */
3493                 if ((one->mode ^ two->mode) & S_IFMT)
3494                         goto free_ab_and_return;
3495                 if (complete_rewrite &&
3496                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3497                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3498                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3499                                          header.buf, header.len, 0);
3500                         strbuf_reset(&header);
3501                         emit_rewrite_diff(name_a, name_b, one, two,
3502                                           textconv_one, textconv_two, o);
3503                         o->found_changes = 1;
3504                         goto free_ab_and_return;
3505                 }
3506         }
3507
3508         if (o->irreversible_delete && lbl[1][0] == '/') {
3509                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3510                                  header.len, 0);
3511                 strbuf_reset(&header);
3512                 goto free_ab_and_return;
3513         } else if (!o->flags.text &&
3514                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3515                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3516                 struct strbuf sb = STRBUF_INIT;
3517                 if (!one->data && !two->data &&
3518                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3519                     !o->flags.binary) {
3520                         if (oideq(&one->oid, &two->oid)) {
3521                                 if (must_show_header)
3522                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3523                                                          header.buf, header.len,
3524                                                          0);
3525                                 goto free_ab_and_return;
3526                         }
3527                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3528                                          header.buf, header.len, 0);
3529                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3530                                     diff_line_prefix(o), lbl[0], lbl[1]);
3531                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3532                                          sb.buf, sb.len, 0);
3533                         strbuf_release(&sb);
3534                         goto free_ab_and_return;
3535                 }
3536                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3537                     fill_mmfile(o->repo, &mf2, two) < 0)
3538                         die("unable to read files to diff");
3539                 /* Quite common confusing case */
3540                 if (mf1.size == mf2.size &&
3541                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3542                         if (must_show_header)
3543                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3544                                                  header.buf, header.len, 0);
3545                         goto free_ab_and_return;
3546                 }
3547                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3548                 strbuf_reset(&header);
3549                 if (o->flags.binary)
3550                         emit_binary_diff(o, &mf1, &mf2);
3551                 else {
3552                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3553                                     diff_line_prefix(o), lbl[0], lbl[1]);
3554                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3555                                          sb.buf, sb.len, 0);
3556                         strbuf_release(&sb);
3557                 }
3558                 o->found_changes = 1;
3559         } else {
3560                 /* Crazy xdl interfaces.. */
3561                 const char *diffopts;
3562                 const char *v;
3563                 xpparam_t xpp;
3564                 xdemitconf_t xecfg;
3565                 struct emit_callback ecbdata;
3566                 const struct userdiff_funcname *pe;
3567
3568                 if (must_show_header) {
3569                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3570                                          header.buf, header.len, 0);
3571                         strbuf_reset(&header);
3572                 }
3573
3574                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3575                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3576
3577                 pe = diff_funcname_pattern(o, one);
3578                 if (!pe)
3579                         pe = diff_funcname_pattern(o, two);
3580
3581                 memset(&xpp, 0, sizeof(xpp));
3582                 memset(&xecfg, 0, sizeof(xecfg));
3583                 memset(&ecbdata, 0, sizeof(ecbdata));
3584                 if (o->flags.suppress_diff_headers)
3585                         lbl[0] = NULL;
3586                 ecbdata.label_path = lbl;
3587                 ecbdata.color_diff = want_color(o->use_color);
3588                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3589                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3590                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3591                 ecbdata.opt = o;
3592                 if (header.len && !o->flags.suppress_diff_headers)
3593                         ecbdata.header = &header;
3594                 xpp.flags = o->xdl_opts;
3595                 xpp.ignore_regex = o->ignore_regex;
3596                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3597                 xpp.anchors = o->anchors;
3598                 xpp.anchors_nr = o->anchors_nr;
3599                 xecfg.ctxlen = o->context;
3600                 xecfg.interhunkctxlen = o->interhunkcontext;
3601                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3602                 if (o->flags.funccontext)
3603                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3604                 if (pe)
3605                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3606
3607                 diffopts = getenv("GIT_DIFF_OPTS");
3608                 if (!diffopts)
3609                         ;
3610                 else if (skip_prefix(diffopts, "--unified=", &v))
3611                         xecfg.ctxlen = strtoul(v, NULL, 10);
3612                 else if (skip_prefix(diffopts, "-u", &v))
3613                         xecfg.ctxlen = strtoul(v, NULL, 10);
3614
3615                 if (o->word_diff)
3616                         init_diff_words_data(&ecbdata, o, one, two);
3617                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3618                                   &ecbdata, &xpp, &xecfg))
3619                         die("unable to generate diff for %s", one->path);
3620                 if (o->word_diff)
3621                         free_diff_words_data(&ecbdata);
3622                 if (textconv_one)
3623                         free(mf1.ptr);
3624                 if (textconv_two)
3625                         free(mf2.ptr);
3626                 xdiff_clear_find_func(&xecfg);
3627         }
3628
3629  free_ab_and_return:
3630         strbuf_release(&header);
3631         diff_free_filespec_data(one);
3632         diff_free_filespec_data(two);
3633         free(a_one);
3634         free(b_two);
3635         return;
3636 }
3637
3638 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3639 {
3640         if (!is_renamed) {
3641                 if (p->status == DIFF_STATUS_ADDED) {
3642                         if (S_ISLNK(p->two->mode))
3643                                 return "new +l";
3644                         else if ((p->two->mode & 0777) == 0755)
3645                                 return "new +x";
3646                         else
3647                                 return "new";
3648                 } else if (p->status == DIFF_STATUS_DELETED)
3649                         return "gone";
3650         }
3651         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3652                 return "mode -l";
3653         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3654                 return "mode +l";
3655         else if ((p->one->mode & 0777) == 0644 &&
3656                  (p->two->mode & 0777) == 0755)
3657                 return "mode +x";
3658         else if ((p->one->mode & 0777) == 0755 &&
3659                  (p->two->mode & 0777) == 0644)
3660                 return "mode -x";
3661         return NULL;
3662 }
3663
3664 static void builtin_diffstat(const char *name_a, const char *name_b,
3665                              struct diff_filespec *one,
3666                              struct diff_filespec *two,
3667                              struct diffstat_t *diffstat,
3668                              struct diff_options *o,
3669                              struct diff_filepair *p)
3670 {
3671         mmfile_t mf1, mf2;
3672         struct diffstat_file *data;
3673         int may_differ;
3674         int complete_rewrite = 0;
3675
3676         if (!DIFF_PAIR_UNMERGED(p)) {
3677                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3678                         complete_rewrite = 1;
3679         }
3680
3681         data = diffstat_add(diffstat, name_a, name_b);
3682         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3683         if (o->flags.stat_with_summary)
3684                 data->comments = get_compact_summary(p, data->is_renamed);
3685
3686         if (!one || !two) {
3687                 data->is_unmerged = 1;
3688                 return;
3689         }
3690
3691         /* saves some reads if true, not a guarantee of diff outcome */
3692         may_differ = !(one->oid_valid && two->oid_valid &&
3693                         oideq(&one->oid, &two->oid));
3694
3695         if (diff_filespec_is_binary(o->repo, one) ||
3696             diff_filespec_is_binary(o->repo, two)) {
3697                 data->is_binary = 1;
3698                 if (!may_differ) {
3699                         data->added = 0;
3700                         data->deleted = 0;
3701                 } else {
3702                         data->added = diff_filespec_size(o->repo, two);
3703                         data->deleted = diff_filespec_size(o->repo, one);
3704                 }
3705         }
3706
3707         else if (complete_rewrite) {
3708                 diff_populate_filespec(o->repo, one, NULL);
3709                 diff_populate_filespec(o->repo, two, NULL);
3710                 data->deleted = count_lines(one->data, one->size);
3711                 data->added = count_lines(two->data, two->size);
3712         }
3713
3714         else if (may_differ) {
3715                 /* Crazy xdl interfaces.. */
3716                 xpparam_t xpp;
3717                 xdemitconf_t xecfg;
3718
3719                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3720                     fill_mmfile(o->repo, &mf2, two) < 0)
3721                         die("unable to read files to diff");
3722
3723                 memset(&xpp, 0, sizeof(xpp));
3724                 memset(&xecfg, 0, sizeof(xecfg));
3725                 xpp.flags = o->xdl_opts;
3726                 xpp.ignore_regex = o->ignore_regex;
3727                 xpp.ignore_regex_nr = o->ignore_regex_nr;
3728                 xpp.anchors = o->anchors;
3729                 xpp.anchors_nr = o->anchors_nr;
3730                 xecfg.ctxlen = o->context;
3731                 xecfg.interhunkctxlen = o->interhunkcontext;
3732                 xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
3733                 if (xdi_diff_outf(&mf1, &mf2, NULL,
3734                                   diffstat_consume, diffstat, &xpp, &xecfg))
3735                         die("unable to generate diffstat for %s", one->path);
3736
3737                 if (DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two)) {
3738                         struct diffstat_file *file =
3739                                 diffstat->files[diffstat->nr - 1];
3740                         /*
3741                          * Omit diffstats of modified files where nothing changed.
3742                          * Even if may_differ, this might be the case due to
3743                          * ignoring whitespace changes, etc.
3744                          *
3745                          * But note that we special-case additions, deletions,
3746                          * renames, and mode changes as adding an empty file,
3747                          * for example is still of interest.
3748                          */
3749                         if ((p->status == DIFF_STATUS_MODIFIED)
3750                                 && !file->added
3751                                 && !file->deleted
3752                                 && one->mode == two->mode) {
3753                                 free_diffstat_file(file);
3754                                 diffstat->nr--;
3755                         }
3756                 }
3757         }
3758
3759         diff_free_filespec_data(one);
3760         diff_free_filespec_data(two);
3761 }
3762
3763 static void builtin_checkdiff(const char *name_a, const char *name_b,
3764                               const char *attr_path,
3765                               struct diff_filespec *one,
3766                               struct diff_filespec *two,
3767                               struct diff_options *o)
3768 {
3769         mmfile_t mf1, mf2;
3770         struct checkdiff_t data;
3771
3772         if (!two)
3773                 return;
3774
3775         memset(&data, 0, sizeof(data));
3776         data.filename = name_b ? name_b : name_a;
3777         data.lineno = 0;
3778         data.o = o;
3779         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3780         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3781
3782         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3783             fill_mmfile(o->repo, &mf2, two) < 0)
3784                 die("unable to read files to diff");
3785
3786         /*
3787          * All the other codepaths check both sides, but not checking
3788          * the "old" side here is deliberate.  We are checking the newly
3789          * introduced changes, and as long as the "new" side is text, we
3790          * can and should check what it introduces.
3791          */
3792         if (diff_filespec_is_binary(o->repo, two))
3793                 goto free_and_return;
3794         else {
3795                 /* Crazy xdl interfaces.. */
3796                 xpparam_t xpp;
3797                 xdemitconf_t xecfg;
3798
3799                 memset(&xpp, 0, sizeof(xpp));
3800                 memset(&xecfg, 0, sizeof(xecfg));
3801                 xecfg.ctxlen = 1; /* at least one context line */
3802                 xpp.flags = 0;
3803                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3804                                   checkdiff_consume, &data,
3805                                   &xpp, &xecfg))
3806                         die("unable to generate checkdiff for %s", one->path);
3807
3808                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3809                         struct emit_callback ecbdata;
3810                         int blank_at_eof;
3811
3812                         ecbdata.ws_rule = data.ws_rule;
3813                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3814                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3815
3816                         if (blank_at_eof) {
3817                                 static char *err;
3818                                 if (!err)
3819                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3820                                 fprintf(o->file, "%s:%d: %s.\n",
3821                                         data.filename, blank_at_eof, err);
3822                                 data.status = 1; /* report errors */
3823                         }
3824                 }
3825         }
3826  free_and_return:
3827         diff_free_filespec_data(one);
3828         diff_free_filespec_data(two);
3829         if (data.status)
3830                 o->flags.check_failed = 1;
3831 }
3832
3833 struct diff_filespec *alloc_filespec(const char *path)
3834 {
3835         struct diff_filespec *spec;
3836
3837         FLEXPTR_ALLOC_STR(spec, path, path);
3838         spec->count = 1;
3839         spec->is_binary = -1;
3840         return spec;
3841 }
3842
3843 void free_filespec(struct diff_filespec *spec)
3844 {
3845         if (!--spec->count) {
3846                 diff_free_filespec_data(spec);
3847                 free(spec);
3848         }
3849 }
3850
3851 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3852                    int oid_valid, unsigned short mode)
3853 {
3854         if (mode) {
3855                 spec->mode = canon_mode(mode);
3856                 oidcpy(&spec->oid, oid);
3857                 spec->oid_valid = oid_valid;
3858         }
3859 }
3860
3861 /*
3862  * Given a name and sha1 pair, if the index tells us the file in
3863  * the work tree has that object contents, return true, so that
3864  * prepare_temp_file() does not have to inflate and extract.
3865  */
3866 static int reuse_worktree_file(struct index_state *istate,
3867                                const char *name,
3868                                const struct object_id *oid,
3869                                int want_file)
3870 {
3871         const struct cache_entry *ce;
3872         struct stat st;
3873         int pos, len;
3874
3875         /*
3876          * We do not read the cache ourselves here, because the
3877          * benchmark with my previous version that always reads cache
3878          * shows that it makes things worse for diff-tree comparing
3879          * two linux-2.6 kernel trees in an already checked out work
3880          * tree.  This is because most diff-tree comparisons deal with
3881          * only a small number of files, while reading the cache is
3882          * expensive for a large project, and its cost outweighs the
3883          * savings we get by not inflating the object to a temporary
3884          * file.  Practically, this code only helps when we are used
3885          * by diff-cache --cached, which does read the cache before
3886          * calling us.
3887          */
3888         if (!istate->cache)
3889                 return 0;
3890
3891         /* We want to avoid the working directory if our caller
3892          * doesn't need the data in a normal file, this system
3893          * is rather slow with its stat/open/mmap/close syscalls,
3894          * and the object is contained in a pack file.  The pack
3895          * is probably already open and will be faster to obtain
3896          * the data through than the working directory.  Loose
3897          * objects however would tend to be slower as they need
3898          * to be individually opened and inflated.
3899          */
3900         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3901                 return 0;
3902
3903         /*
3904          * Similarly, if we'd have to convert the file contents anyway, that
3905          * makes the optimization not worthwhile.
3906          */
3907         if (!want_file && would_convert_to_git(istate, name))
3908                 return 0;
3909
3910         len = strlen(name);
3911         pos = index_name_pos(istate, name, len);
3912         if (pos < 0)
3913                 return 0;
3914         ce = istate->cache[pos];
3915
3916         /*
3917          * This is not the sha1 we are looking for, or
3918          * unreusable because it is not a regular file.
3919          */
3920         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3921                 return 0;
3922
3923         /*
3924          * If ce is marked as "assume unchanged", there is no
3925          * guarantee that work tree matches what we are looking for.
3926          */
3927         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3928                 return 0;
3929
3930         /*
3931          * If ce matches the file in the work tree, we can reuse it.
3932          */
3933         if (ce_uptodate(ce) ||
3934             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3935                 return 1;
3936
3937         return 0;
3938 }
3939
3940 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3941 {
3942         struct strbuf buf = STRBUF_INIT;
3943         char *dirty = "";
3944
3945         /* Are we looking at the work tree? */
3946         if (s->dirty_submodule)
3947                 dirty = "-dirty";
3948
3949         strbuf_addf(&buf, "Subproject commit %s%s\n",
3950                     oid_to_hex(&s->oid), dirty);
3951         s->size = buf.len;
3952         if (size_only) {
3953                 s->data = NULL;
3954                 strbuf_release(&buf);
3955         } else {
3956                 s->data = strbuf_detach(&buf, NULL);
3957                 s->should_free = 1;
3958         }
3959         return 0;
3960 }
3961
3962 /*
3963  * While doing rename detection and pickaxe operation, we may need to
3964  * grab the data for the blob (or file) for our own in-core comparison.
3965  * diff_filespec has data and size fields for this purpose.
3966  */
3967 int diff_populate_filespec(struct repository *r,
3968                            struct diff_filespec *s,
3969                            const struct diff_populate_filespec_options *options)
3970 {
3971         int size_only = options ? options->check_size_only : 0;
3972         int check_binary = options ? options->check_binary : 0;
3973         int err = 0;
3974         int conv_flags = global_conv_flags_eol;
3975         /*
3976          * demote FAIL to WARN to allow inspecting the situation
3977          * instead of refusing.
3978          */
3979         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3980                 conv_flags = CONV_EOL_RNDTRP_WARN;
3981
3982         if (!DIFF_FILE_VALID(s))
3983                 die("internal error: asking to populate invalid file.");
3984         if (S_ISDIR(s->mode))
3985                 return -1;
3986
3987         if (s->data)
3988                 return 0;
3989
3990         if (size_only && 0 < s->size)
3991                 return 0;
3992
3993         if (S_ISGITLINK(s->mode))
3994                 return diff_populate_gitlink(s, size_only);
3995
3996         if (!s->oid_valid ||
3997             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3998                 struct strbuf buf = STRBUF_INIT;
3999                 struct stat st;
4000                 int fd;
4001
4002                 if (lstat(s->path, &st) < 0) {
4003                 err_empty:
4004                         err = -1;
4005                 empty:
4006                         s->data = (char *)"";
4007                         s->size = 0;
4008                         return err;
4009                 }
4010                 s->size = xsize_t(st.st_size);
4011                 if (!s->size)
4012                         goto empty;
4013                 if (S_ISLNK(st.st_mode)) {
4014                         struct strbuf sb = STRBUF_INIT;
4015
4016                         if (strbuf_readlink(&sb, s->path, s->size))
4017                                 goto err_empty;
4018                         s->size = sb.len;
4019                         s->data = strbuf_detach(&sb, NULL);
4020                         s->should_free = 1;
4021                         return 0;
4022                 }
4023
4024                 /*
4025                  * Even if the caller would be happy with getting
4026                  * only the size, we cannot return early at this
4027                  * point if the path requires us to run the content
4028                  * conversion.
4029                  */
4030                 if (size_only && !would_convert_to_git(r->index, s->path))
4031                         return 0;
4032
4033                 /*
4034                  * Note: this check uses xsize_t(st.st_size) that may
4035                  * not be the true size of the blob after it goes
4036                  * through convert_to_git().  This may not strictly be
4037                  * correct, but the whole point of big_file_threshold
4038                  * and is_binary check being that we want to avoid
4039                  * opening the file and inspecting the contents, this
4040                  * is probably fine.
4041                  */
4042                 if (check_binary &&
4043                     s->size > big_file_threshold && s->is_binary == -1) {
4044                         s->is_binary = 1;
4045                         return 0;
4046                 }
4047                 fd = open(s->path, O_RDONLY);
4048                 if (fd < 0)
4049                         goto err_empty;
4050                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4051                 close(fd);
4052                 s->should_munmap = 1;
4053
4054                 /*
4055                  * Convert from working tree format to canonical git format
4056                  */
4057                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4058                         size_t size = 0;
4059                         munmap(s->data, s->size);
4060                         s->should_munmap = 0;
4061                         s->data = strbuf_detach(&buf, &size);
4062                         s->size = size;
4063                         s->should_free = 1;
4064                 }
4065         }
4066         else {
4067                 struct object_info info = {
4068                         .sizep = &s->size
4069                 };
4070
4071                 if (!(size_only || check_binary))
4072                         /*
4073                          * Set contentp, since there is no chance that merely
4074                          * the size is sufficient.
4075                          */
4076                         info.contentp = &s->data;
4077
4078                 if (options && options->missing_object_cb) {
4079                         if (!oid_object_info_extended(r, &s->oid, &info,
4080                                                       OBJECT_INFO_LOOKUP_REPLACE |
4081                                                       OBJECT_INFO_SKIP_FETCH_OBJECT))
4082                                 goto object_read;
4083                         options->missing_object_cb(options->missing_object_data);
4084                 }
4085                 if (oid_object_info_extended(r, &s->oid, &info,
4086                                              OBJECT_INFO_LOOKUP_REPLACE))
4087                         die("unable to read %s", oid_to_hex(&s->oid));
4088
4089 object_read:
4090                 if (size_only || check_binary) {
4091                         if (size_only)
4092                                 return 0;
4093                         if (s->size > big_file_threshold && s->is_binary == -1) {
4094                                 s->is_binary = 1;
4095                                 return 0;
4096                         }
4097                 }
4098                 if (!info.contentp) {
4099                         info.contentp = &s->data;
4100                         if (oid_object_info_extended(r, &s->oid, &info,
4101                                                      OBJECT_INFO_LOOKUP_REPLACE))
4102                                 die("unable to read %s", oid_to_hex(&s->oid));
4103                 }
4104                 s->should_free = 1;
4105         }
4106         return 0;
4107 }
4108
4109 void diff_free_filespec_blob(struct diff_filespec *s)
4110 {
4111         if (s->should_free)
4112                 free(s->data);
4113         else if (s->should_munmap)
4114                 munmap(s->data, s->size);
4115
4116         if (s->should_free || s->should_munmap) {
4117                 s->should_free = s->should_munmap = 0;
4118                 s->data = NULL;
4119         }
4120 }
4121
4122 void diff_free_filespec_data(struct diff_filespec *s)
4123 {
4124         if (!s)
4125                 return;
4126
4127         diff_free_filespec_blob(s);
4128         FREE_AND_NULL(s->cnt_data);
4129 }
4130
4131 static void prep_temp_blob(struct index_state *istate,
4132                            const char *path, struct diff_tempfile *temp,
4133                            void *blob,
4134                            unsigned long size,
4135                            const struct object_id *oid,
4136                            int mode)
4137 {
4138         struct strbuf buf = STRBUF_INIT;
4139         struct strbuf tempfile = STRBUF_INIT;
4140         char *path_dup = xstrdup(path);
4141         const char *base = basename(path_dup);
4142         struct checkout_metadata meta;
4143
4144         init_checkout_metadata(&meta, NULL, NULL, oid);
4145
4146         /* Generate "XXXXXX_basename.ext" */
4147         strbuf_addstr(&tempfile, "XXXXXX_");
4148         strbuf_addstr(&tempfile, base);
4149
4150         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4151         if (!temp->tempfile)
4152                 die_errno("unable to create temp-file");
4153         if (convert_to_working_tree(istate, path,
4154                         (const char *)blob, (size_t)size, &buf, &meta)) {
4155                 blob = buf.buf;
4156                 size = buf.len;
4157         }
4158         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4159             close_tempfile_gently(temp->tempfile))
4160                 die_errno("unable to write temp-file");
4161         temp->name = get_tempfile_path(temp->tempfile);
4162         oid_to_hex_r(temp->hex, oid);
4163         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4164         strbuf_release(&buf);
4165         strbuf_release(&tempfile);
4166         free(path_dup);
4167 }
4168
4169 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4170                                                const char *name,
4171                                                struct diff_filespec *one)
4172 {
4173         struct diff_tempfile *temp = claim_diff_tempfile();
4174
4175         if (!DIFF_FILE_VALID(one)) {
4176         not_a_valid_file:
4177                 /* A '-' entry produces this for file-2, and
4178                  * a '+' entry produces this for file-1.
4179                  */
4180                 temp->name = "/dev/null";
4181                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4182                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4183                 return temp;
4184         }
4185
4186         if (!S_ISGITLINK(one->mode) &&
4187             (!one->oid_valid ||
4188              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4189                 struct stat st;
4190                 if (lstat(name, &st) < 0) {
4191                         if (errno == ENOENT)
4192                                 goto not_a_valid_file;
4193                         die_errno("stat(%s)", name);
4194                 }
4195                 if (S_ISLNK(st.st_mode)) {
4196                         struct strbuf sb = STRBUF_INIT;
4197                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4198                                 die_errno("readlink(%s)", name);
4199                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4200                                        (one->oid_valid ?
4201                                         &one->oid : null_oid()),
4202                                        (one->oid_valid ?
4203                                         one->mode : S_IFLNK));
4204                         strbuf_release(&sb);
4205                 }
4206                 else {
4207                         /* we can borrow from the file in the work tree */
4208                         temp->name = name;
4209                         if (!one->oid_valid)
4210                                 oid_to_hex_r(temp->hex, null_oid());
4211                         else
4212                                 oid_to_hex_r(temp->hex, &one->oid);
4213                         /* Even though we may sometimes borrow the
4214                          * contents from the work tree, we always want
4215                          * one->mode.  mode is trustworthy even when
4216                          * !(one->oid_valid), as long as
4217                          * DIFF_FILE_VALID(one).
4218                          */
4219                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4220                 }
4221                 return temp;
4222         }
4223         else {
4224                 if (diff_populate_filespec(r, one, NULL))
4225                         die("cannot read data blob for %s", one->path);
4226                 prep_temp_blob(r->index, name, temp,
4227                                one->data, one->size,
4228                                &one->oid, one->mode);
4229         }
4230         return temp;
4231 }
4232
4233 static void add_external_diff_name(struct repository *r,
4234                                    struct strvec *argv,
4235                                    const char *name,
4236                                    struct diff_filespec *df)
4237 {
4238         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4239         strvec_push(argv, temp->name);
4240         strvec_push(argv, temp->hex);
4241         strvec_push(argv, temp->mode);
4242 }
4243
4244 /* An external diff command takes:
4245  *
4246  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4247  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4248  *
4249  */
4250 static void run_external_diff(const char *pgm,
4251                               const char *name,
4252                               const char *other,
4253                               struct diff_filespec *one,
4254                               struct diff_filespec *two,
4255                               const char *xfrm_msg,
4256                               struct diff_options *o)
4257 {
4258         struct strvec argv = STRVEC_INIT;
4259         struct strvec env = STRVEC_INIT;
4260         struct diff_queue_struct *q = &diff_queued_diff;
4261
4262         strvec_push(&argv, pgm);
4263         strvec_push(&argv, name);
4264
4265         if (one && two) {
4266                 add_external_diff_name(o->repo, &argv, name, one);
4267                 if (!other)
4268                         add_external_diff_name(o->repo, &argv, name, two);
4269                 else {
4270                         add_external_diff_name(o->repo, &argv, other, two);
4271                         strvec_push(&argv, other);
4272                         strvec_push(&argv, xfrm_msg);
4273                 }
4274         }
4275
4276         strvec_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4277         strvec_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4278
4279         diff_free_filespec_data(one);
4280         diff_free_filespec_data(two);
4281         if (run_command_v_opt_cd_env(argv.v, RUN_USING_SHELL, NULL, env.v))
4282                 die(_("external diff died, stopping at %s"), name);
4283
4284         remove_tempfile();
4285         strvec_clear(&argv);
4286         strvec_clear(&env);
4287 }
4288
4289 static int similarity_index(struct diff_filepair *p)
4290 {
4291         return p->score * 100 / MAX_SCORE;
4292 }
4293
4294 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4295 {
4296         if (startup_info->have_repository)
4297                 return find_unique_abbrev(oid, abbrev);
4298         else {
4299                 char *hex = oid_to_hex(oid);
4300                 if (abbrev < 0)
4301                         abbrev = FALLBACK_DEFAULT_ABBREV;
4302                 if (abbrev > the_hash_algo->hexsz)
4303                         BUG("oid abbreviation out of range: %d", abbrev);
4304                 if (abbrev)
4305                         hex[abbrev] = '\0';
4306                 return hex;
4307         }
4308 }
4309
4310 static void fill_metainfo(struct strbuf *msg,
4311                           const char *name,
4312                           const char *other,
4313                           struct diff_filespec *one,
4314                           struct diff_filespec *two,
4315                           struct diff_options *o,
4316                           struct diff_filepair *p,
4317                           int *must_show_header,
4318                           int use_color)
4319 {
4320         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4321         const char *reset = diff_get_color(use_color, DIFF_RESET);
4322         const char *line_prefix = diff_line_prefix(o);
4323
4324         *must_show_header = 1;
4325         strbuf_init(msg, PATH_MAX * 2 + 300);
4326         switch (p->status) {
4327         case DIFF_STATUS_COPIED:
4328                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4329                             line_prefix, set, similarity_index(p));
4330                 strbuf_addf(msg, "%s\n%s%scopy from ",
4331                             reset,  line_prefix, set);
4332                 quote_c_style(name, msg, NULL, 0);
4333                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4334                 quote_c_style(other, msg, NULL, 0);
4335                 strbuf_addf(msg, "%s\n", reset);
4336                 break;
4337         case DIFF_STATUS_RENAMED:
4338                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4339                             line_prefix, set, similarity_index(p));
4340                 strbuf_addf(msg, "%s\n%s%srename from ",
4341                             reset, line_prefix, set);
4342                 quote_c_style(name, msg, NULL, 0);
4343                 strbuf_addf(msg, "%s\n%s%srename to ",
4344                             reset, line_prefix, set);
4345                 quote_c_style(other, msg, NULL, 0);
4346                 strbuf_addf(msg, "%s\n", reset);
4347                 break;
4348         case DIFF_STATUS_MODIFIED:
4349                 if (p->score) {
4350                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4351                                     line_prefix,
4352                                     set, similarity_index(p), reset);
4353                         break;
4354                 }
4355                 /* fallthru */
4356         default:
4357                 *must_show_header = 0;
4358         }
4359         if (one && two && !oideq(&one->oid, &two->oid)) {
4360                 const unsigned hexsz = the_hash_algo->hexsz;
4361                 int abbrev = o->abbrev ? o->abbrev : DEFAULT_ABBREV;
4362
4363                 if (o->flags.full_index)
4364                         abbrev = hexsz;
4365
4366                 if (o->flags.binary) {
4367                         mmfile_t mf;
4368                         if ((!fill_mmfile(o->repo, &mf, one) &&
4369                              diff_filespec_is_binary(o->repo, one)) ||
4370                             (!fill_mmfile(o->repo, &mf, two) &&
4371                              diff_filespec_is_binary(o->repo, two)))
4372                                 abbrev = hexsz;
4373                 }
4374                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4375                             diff_abbrev_oid(&one->oid, abbrev),
4376                             diff_abbrev_oid(&two->oid, abbrev));
4377                 if (one->mode == two->mode)
4378                         strbuf_addf(msg, " %06o", one->mode);
4379                 strbuf_addf(msg, "%s\n", reset);
4380         }
4381 }
4382
4383 static void run_diff_cmd(const char *pgm,
4384                          const char *name,
4385                          const char *other,
4386                          const char *attr_path,
4387                          struct diff_filespec *one,
4388                          struct diff_filespec *two,
4389                          struct strbuf *msg,
4390                          struct diff_options *o,
4391                          struct diff_filepair *p)
4392 {
4393         const char *xfrm_msg = NULL;
4394         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4395         int must_show_header = 0;
4396
4397
4398         if (o->flags.allow_external) {
4399                 struct userdiff_driver *drv;
4400
4401                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4402                 if (drv && drv->external)
4403                         pgm = drv->external;
4404         }
4405
4406         if (msg) {
4407                 /*
4408                  * don't use colors when the header is intended for an
4409                  * external diff driver
4410                  */
4411                 fill_metainfo(msg, name, other, one, two, o, p,
4412                               &must_show_header,
4413                               want_color(o->use_color) && !pgm);
4414                 xfrm_msg = msg->len ? msg->buf : NULL;
4415         }
4416
4417         if (pgm) {
4418                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4419                 return;
4420         }
4421         if (one && two)
4422                 builtin_diff(name, other ? other : name,
4423                              one, two, xfrm_msg, must_show_header,
4424                              o, complete_rewrite);
4425         else
4426                 fprintf(o->file, "* Unmerged path %s\n", name);
4427 }
4428
4429 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4430 {
4431         if (DIFF_FILE_VALID(one)) {
4432                 if (!one->oid_valid) {
4433                         struct stat st;
4434                         if (one->is_stdin) {
4435                                 oidclr(&one->oid);
4436                                 return;
4437                         }
4438                         if (lstat(one->path, &st) < 0)
4439                                 die_errno("stat '%s'", one->path);
4440                         if (index_path(istate, &one->oid, one->path, &st, 0))
4441                                 die("cannot hash %s", one->path);
4442                 }
4443         }
4444         else
4445                 oidclr(&one->oid);
4446 }
4447
4448 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4449 {
4450         /* Strip the prefix but do not molest /dev/null and absolute paths */
4451         if (*namep && !is_absolute_path(*namep)) {
4452                 *namep += prefix_length;
4453                 if (**namep == '/')
4454                         ++*namep;
4455         }
4456         if (*otherp && !is_absolute_path(*otherp)) {
4457                 *otherp += prefix_length;
4458                 if (**otherp == '/')
4459                         ++*otherp;
4460         }
4461 }
4462
4463 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4464 {
4465         const char *pgm = external_diff();
4466         struct strbuf msg;
4467         struct diff_filespec *one = p->one;
4468         struct diff_filespec *two = p->two;
4469         const char *name;
4470         const char *other;
4471         const char *attr_path;
4472
4473         name  = one->path;
4474         other = (strcmp(name, two->path) ? two->path : NULL);
4475         attr_path = name;
4476         if (o->prefix_length)
4477                 strip_prefix(o->prefix_length, &name, &other);
4478
4479         if (!o->flags.allow_external)
4480                 pgm = NULL;
4481
4482         if (DIFF_PAIR_UNMERGED(p)) {
4483                 run_diff_cmd(pgm, name, NULL, attr_path,
4484                              NULL, NULL, NULL, o, p);
4485                 return;
4486         }
4487
4488         diff_fill_oid_info(one, o->repo->index);
4489         diff_fill_oid_info(two, o->repo->index);
4490
4491         if (!pgm &&
4492             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4493             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4494                 /*
4495                  * a filepair that changes between file and symlink
4496                  * needs to be split into deletion and creation.
4497                  */
4498                 struct diff_filespec *null = alloc_filespec(two->path);
4499                 run_diff_cmd(NULL, name, other, attr_path,
4500                              one, null, &msg,
4501                              o, p);
4502                 free(null);
4503                 strbuf_release(&msg);
4504
4505                 null = alloc_filespec(one->path);
4506                 run_diff_cmd(NULL, name, other, attr_path,
4507                              null, two, &msg, o, p);
4508                 free(null);
4509         }
4510         else
4511                 run_diff_cmd(pgm, name, other, attr_path,
4512                              one, two, &msg, o, p);
4513
4514         strbuf_release(&msg);
4515 }
4516
4517 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4518                          struct diffstat_t *diffstat)
4519 {
4520         const char *name;
4521         const char *other;
4522
4523         if (DIFF_PAIR_UNMERGED(p)) {
4524                 /* unmerged */
4525                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4526                                  diffstat, o, p);
4527                 return;
4528         }
4529
4530         name = p->one->path;
4531         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4532
4533         if (o->prefix_length)
4534                 strip_prefix(o->prefix_length, &name, &other);
4535
4536         diff_fill_oid_info(p->one, o->repo->index);
4537         diff_fill_oid_info(p->two, o->repo->index);
4538
4539         builtin_diffstat(name, other, p->one, p->two,
4540                          diffstat, o, p);
4541 }
4542
4543 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4544 {
4545         const char *name;
4546         const char *other;
4547         const char *attr_path;
4548
4549         if (DIFF_PAIR_UNMERGED(p)) {
4550                 /* unmerged */
4551                 return;
4552         }
4553
4554         name = p->one->path;
4555         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4556         attr_path = other ? other : name;
4557
4558         if (o->prefix_length)
4559                 strip_prefix(o->prefix_length, &name, &other);
4560
4561         diff_fill_oid_info(p->one, o->repo->index);
4562         diff_fill_oid_info(p->two, o->repo->index);
4563
4564         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4565 }
4566
4567 static void prep_parse_options(struct diff_options *options);
4568
4569 void repo_diff_setup(struct repository *r, struct diff_options *options)
4570 {
4571         memcpy(options, &default_diff_options, sizeof(*options));
4572
4573         options->file = stdout;
4574         options->repo = r;
4575
4576         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4577         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4578         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4579         options->abbrev = DEFAULT_ABBREV;
4580         options->line_termination = '\n';
4581         options->break_opt = -1;
4582         options->rename_limit = -1;
4583         options->dirstat_permille = diff_dirstat_permille_default;
4584         options->context = diff_context_default;
4585         options->interhunkcontext = diff_interhunk_context_default;
4586         options->ws_error_highlight = ws_error_highlight_default;
4587         options->flags.rename_empty = 1;
4588         options->flags.relative_name = diff_relative;
4589         options->objfind = NULL;
4590
4591         /* pathchange left =NULL by default */
4592         options->change = diff_change;
4593         options->add_remove = diff_addremove;
4594         options->use_color = diff_use_color_default;
4595         options->detect_rename = diff_detect_rename_default;
4596         options->xdl_opts |= diff_algorithm;
4597         if (diff_indent_heuristic)
4598                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4599
4600         options->orderfile = diff_order_file_cfg;
4601
4602         if (!options->flags.ignore_submodule_set)
4603                 options->flags.ignore_untracked_in_submodules = 1;
4604
4605         if (diff_no_prefix) {
4606                 options->a_prefix = options->b_prefix = "";
4607         } else if (!diff_mnemonic_prefix) {
4608                 options->a_prefix = "a/";
4609                 options->b_prefix = "b/";
4610         }
4611
4612         options->color_moved = diff_color_moved_default;
4613         options->color_moved_ws_handling = diff_color_moved_ws_default;
4614
4615         prep_parse_options(options);
4616 }
4617
4618 void diff_setup_done(struct diff_options *options)
4619 {
4620         unsigned check_mask = DIFF_FORMAT_NAME |
4621                               DIFF_FORMAT_NAME_STATUS |
4622                               DIFF_FORMAT_CHECKDIFF |
4623                               DIFF_FORMAT_NO_OUTPUT;
4624         /*
4625          * This must be signed because we're comparing against a potentially
4626          * negative value.
4627          */
4628         const int hexsz = the_hash_algo->hexsz;
4629
4630         if (options->set_default)
4631                 options->set_default(options);
4632
4633         if (HAS_MULTI_BITS(options->output_format & check_mask))
4634                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4635
4636         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4637                 die(_("-G, -S and --find-object are mutually exclusive"));
4638
4639         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_G_REGEX_MASK))
4640                 die(_("-G and --pickaxe-regex are mutually exclusive, use --pickaxe-regex with -S"));
4641
4642         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_ALL_OBJFIND_MASK))
4643                 die(_("---pickaxe-all and --find-object are mutually exclusive, use --pickaxe-all with -G and -S"));
4644
4645         /*
4646          * Most of the time we can say "there are changes"
4647          * only by checking if there are changed paths, but
4648          * --ignore-whitespace* options force us to look
4649          * inside contents.
4650          */
4651
4652         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS) ||
4653             options->ignore_regex_nr)
4654                 options->flags.diff_from_contents = 1;
4655         else
4656                 options->flags.diff_from_contents = 0;
4657
4658         if (options->flags.find_copies_harder)
4659                 options->detect_rename = DIFF_DETECT_COPY;
4660
4661         if (!options->flags.relative_name)
4662                 options->prefix = NULL;
4663         if (options->prefix)
4664                 options->prefix_length = strlen(options->prefix);
4665         else
4666                 options->prefix_length = 0;
4667
4668         if (options->output_format & (DIFF_FORMAT_NAME |
4669                                       DIFF_FORMAT_NAME_STATUS |
4670                                       DIFF_FORMAT_CHECKDIFF |
4671                                       DIFF_FORMAT_NO_OUTPUT))
4672                 options->output_format &= ~(DIFF_FORMAT_RAW |
4673                                             DIFF_FORMAT_NUMSTAT |
4674                                             DIFF_FORMAT_DIFFSTAT |
4675                                             DIFF_FORMAT_SHORTSTAT |
4676                                             DIFF_FORMAT_DIRSTAT |
4677                                             DIFF_FORMAT_SUMMARY |
4678                                             DIFF_FORMAT_PATCH);
4679
4680         /*
4681          * These cases always need recursive; we do not drop caller-supplied
4682          * recursive bits for other formats here.
4683          */
4684         if (options->output_format & (DIFF_FORMAT_PATCH |
4685                                       DIFF_FORMAT_NUMSTAT |
4686                                       DIFF_FORMAT_DIFFSTAT |
4687                                       DIFF_FORMAT_SHORTSTAT |
4688                                       DIFF_FORMAT_DIRSTAT |
4689                                       DIFF_FORMAT_SUMMARY |
4690                                       DIFF_FORMAT_CHECKDIFF))
4691                 options->flags.recursive = 1;
4692         /*
4693          * Also pickaxe would not work very well if you do not say recursive
4694          */
4695         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4696                 options->flags.recursive = 1;
4697         /*
4698          * When patches are generated, submodules diffed against the work tree
4699          * must be checked for dirtiness too so it can be shown in the output
4700          */
4701         if (options->output_format & DIFF_FORMAT_PATCH)
4702                 options->flags.dirty_submodules = 1;
4703
4704         if (options->detect_rename && options->rename_limit < 0)
4705                 options->rename_limit = diff_rename_limit_default;
4706         if (hexsz < options->abbrev)
4707                 options->abbrev = hexsz; /* full */
4708
4709         /*
4710          * It does not make sense to show the first hit we happened
4711          * to have found.  It does not make sense not to return with
4712          * exit code in such a case either.
4713          */
4714         if (options->flags.quick) {
4715                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4716                 options->flags.exit_with_status = 1;
4717         }
4718
4719         options->diff_path_counter = 0;
4720
4721         if (options->flags.follow_renames && options->pathspec.nr != 1)
4722                 die(_("--follow requires exactly one pathspec"));
4723
4724         if (!options->use_color || external_diff())
4725                 options->color_moved = 0;
4726
4727         FREE_AND_NULL(options->parseopts);
4728 }
4729
4730 int parse_long_opt(const char *opt, const char **argv,
4731                    const char **optarg)
4732 {
4733         const char *arg = argv[0];
4734         if (!skip_prefix(arg, "--", &arg))
4735                 return 0;
4736         if (!skip_prefix(arg, opt, &arg))
4737                 return 0;
4738         if (*arg == '=') { /* stuck form: --option=value */
4739                 *optarg = arg + 1;
4740                 return 1;
4741         }
4742         if (*arg != '\0')
4743                 return 0;
4744         /* separate form: --option value */
4745         if (!argv[1])
4746                 die("Option '--%s' requires a value", opt);
4747         *optarg = argv[1];
4748         return 2;
4749 }
4750
4751 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4752 {
4753         struct diff_options *options = opt->value;
4754         int width = options->stat_width;
4755         int name_width = options->stat_name_width;
4756         int graph_width = options->stat_graph_width;
4757         int count = options->stat_count;
4758         char *end;
4759
4760         BUG_ON_OPT_NEG(unset);
4761
4762         if (!strcmp(opt->long_name, "stat")) {
4763                 if (value) {
4764                         width = strtoul(value, &end, 10);
4765                         if (*end == ',')
4766                                 name_width = strtoul(end+1, &end, 10);
4767                         if (*end == ',')
4768                                 count = strtoul(end+1, &end, 10);
4769                         if (*end)
4770                                 return error(_("invalid --stat value: %s"), value);
4771                 }
4772         } else if (!strcmp(opt->long_name, "stat-width")) {
4773                 width = strtoul(value, &end, 10);
4774                 if (*end)
4775                         return error(_("%s expects a numerical value"),
4776                                      opt->long_name);
4777         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4778                 name_width = strtoul(value, &end, 10);
4779                 if (*end)
4780                         return error(_("%s expects a numerical value"),
4781                                      opt->long_name);
4782         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4783                 graph_width = strtoul(value, &end, 10);
4784                 if (*end)
4785                         return error(_("%s expects a numerical value"),
4786                                      opt->long_name);
4787         } else if (!strcmp(opt->long_name, "stat-count")) {
4788                 count = strtoul(value, &end, 10);
4789                 if (*end)
4790                         return error(_("%s expects a numerical value"),
4791                                      opt->long_name);
4792         } else
4793                 BUG("%s should not get here", opt->long_name);
4794
4795         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4796         options->stat_name_width = name_width;
4797         options->stat_graph_width = graph_width;
4798         options->stat_width = width;
4799         options->stat_count = count;
4800         return 0;
4801 }
4802
4803 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4804 {
4805         struct strbuf errmsg = STRBUF_INIT;
4806         if (parse_dirstat_params(options, params, &errmsg))
4807                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4808                     errmsg.buf);
4809         strbuf_release(&errmsg);
4810         /*
4811          * The caller knows a dirstat-related option is given from the command
4812          * line; allow it to say "return this_function();"
4813          */
4814         options->output_format |= DIFF_FORMAT_DIRSTAT;
4815         return 1;
4816 }
4817
4818 static const char diff_status_letters[] = {
4819         DIFF_STATUS_ADDED,
4820         DIFF_STATUS_COPIED,
4821         DIFF_STATUS_DELETED,
4822         DIFF_STATUS_MODIFIED,
4823         DIFF_STATUS_RENAMED,
4824         DIFF_STATUS_TYPE_CHANGED,
4825         DIFF_STATUS_UNKNOWN,
4826         DIFF_STATUS_UNMERGED,
4827         DIFF_STATUS_FILTER_AON,
4828         DIFF_STATUS_FILTER_BROKEN,
4829         '\0',
4830 };
4831
4832 static unsigned int filter_bit['Z' + 1];
4833
4834 static void prepare_filter_bits(void)
4835 {
4836         int i;
4837
4838         if (!filter_bit[DIFF_STATUS_ADDED]) {
4839                 for (i = 0; diff_status_letters[i]; i++)
4840                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4841         }
4842 }
4843
4844 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4845 {
4846         return opt->filter & filter_bit[(int) status];
4847 }
4848
4849 unsigned diff_filter_bit(char status)
4850 {
4851         prepare_filter_bits();
4852         return filter_bit[(int) status];
4853 }
4854
4855 static int diff_opt_diff_filter(const struct option *option,
4856                                 const char *optarg, int unset)
4857 {
4858         struct diff_options *opt = option->value;
4859         int i, optch;
4860
4861         BUG_ON_OPT_NEG(unset);
4862         prepare_filter_bits();
4863
4864         /*
4865          * If there is a negation e.g. 'd' in the input, and we haven't
4866          * initialized the filter field with another --diff-filter, start
4867          * from full set of bits, except for AON.
4868          */
4869         if (!opt->filter) {
4870                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4871                         if (optch < 'a' || 'z' < optch)
4872                                 continue;
4873                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4874                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4875                         break;
4876                 }
4877         }
4878
4879         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4880                 unsigned int bit;
4881                 int negate;
4882
4883                 if ('a' <= optch && optch <= 'z') {
4884                         negate = 1;
4885                         optch = toupper(optch);
4886                 } else {
4887                         negate = 0;
4888                 }
4889
4890                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4891                 if (!bit)
4892                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4893                                      optarg[i], optarg);
4894                 if (negate)
4895                         opt->filter &= ~bit;
4896                 else
4897                         opt->filter |= bit;
4898         }
4899         return 0;
4900 }
4901
4902 static void enable_patch_output(int *fmt)
4903 {
4904         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4905         *fmt |= DIFF_FORMAT_PATCH;
4906 }
4907
4908 static int diff_opt_ws_error_highlight(const struct option *option,
4909                                        const char *arg, int unset)
4910 {
4911         struct diff_options *opt = option->value;
4912         int val = parse_ws_error_highlight(arg);
4913
4914         BUG_ON_OPT_NEG(unset);
4915         if (val < 0)
4916                 return error(_("unknown value after ws-error-highlight=%.*s"),
4917                              -1 - val, arg);
4918         opt->ws_error_highlight = val;
4919         return 0;
4920 }
4921
4922 static int diff_opt_find_object(const struct option *option,
4923                                 const char *arg, int unset)
4924 {
4925         struct diff_options *opt = option->value;
4926         struct object_id oid;
4927
4928         BUG_ON_OPT_NEG(unset);
4929         if (get_oid(arg, &oid))
4930                 return error(_("unable to resolve '%s'"), arg);
4931
4932         if (!opt->objfind)
4933                 CALLOC_ARRAY(opt->objfind, 1);
4934
4935         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4936         opt->flags.recursive = 1;
4937         opt->flags.tree_in_recursive = 1;
4938         oidset_insert(opt->objfind, &oid);
4939         return 0;
4940 }
4941
4942 static int diff_opt_anchored(const struct option *opt,
4943                              const char *arg, int unset)
4944 {
4945         struct diff_options *options = opt->value;
4946
4947         BUG_ON_OPT_NEG(unset);
4948         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4949         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4950                    options->anchors_alloc);
4951         options->anchors[options->anchors_nr++] = xstrdup(arg);
4952         return 0;
4953 }
4954
4955 static int diff_opt_binary(const struct option *opt,
4956                            const char *arg, int unset)
4957 {
4958         struct diff_options *options = opt->value;
4959
4960         BUG_ON_OPT_NEG(unset);
4961         BUG_ON_OPT_ARG(arg);
4962         enable_patch_output(&options->output_format);
4963         options->flags.binary = 1;
4964         return 0;
4965 }
4966
4967 static int diff_opt_break_rewrites(const struct option *opt,
4968                                    const char *arg, int unset)
4969 {
4970         int *break_opt = opt->value;
4971         int opt1, opt2;
4972
4973         BUG_ON_OPT_NEG(unset);
4974         if (!arg)
4975                 arg = "";
4976         opt1 = parse_rename_score(&arg);
4977         if (*arg == 0)
4978                 opt2 = 0;
4979         else if (*arg != '/')
4980                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4981         else {
4982                 arg++;
4983                 opt2 = parse_rename_score(&arg);
4984         }
4985         if (*arg != 0)
4986                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4987         *break_opt = opt1 | (opt2 << 16);
4988         return 0;
4989 }
4990
4991 static int diff_opt_char(const struct option *opt,
4992                          const char *arg, int unset)
4993 {
4994         char *value = opt->value;
4995
4996         BUG_ON_OPT_NEG(unset);
4997         if (arg[1])
4998                 return error(_("%s expects a character, got '%s'"),
4999                              opt->long_name, arg);
5000         *value = arg[0];
5001         return 0;
5002 }
5003
5004 static int diff_opt_color_moved(const struct option *opt,
5005                                 const char *arg, int unset)
5006 {
5007         struct diff_options *options = opt->value;
5008
5009         if (unset) {
5010                 options->color_moved = COLOR_MOVED_NO;
5011         } else if (!arg) {
5012                 if (diff_color_moved_default)
5013                         options->color_moved = diff_color_moved_default;
5014                 if (options->color_moved == COLOR_MOVED_NO)
5015                         options->color_moved = COLOR_MOVED_DEFAULT;
5016         } else {
5017                 int cm = parse_color_moved(arg);
5018                 if (cm < 0)
5019                         return error(_("bad --color-moved argument: %s"), arg);
5020                 options->color_moved = cm;
5021         }
5022         return 0;
5023 }
5024
5025 static int diff_opt_color_moved_ws(const struct option *opt,
5026                                    const char *arg, int unset)
5027 {
5028         struct diff_options *options = opt->value;
5029         unsigned cm;
5030
5031         if (unset) {
5032                 options->color_moved_ws_handling = 0;
5033                 return 0;
5034         }
5035
5036         cm = parse_color_moved_ws(arg);
5037         if (cm & COLOR_MOVED_WS_ERROR)
5038                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
5039         options->color_moved_ws_handling = cm;
5040         return 0;
5041 }
5042
5043 static int diff_opt_color_words(const struct option *opt,
5044                                 const char *arg, int unset)
5045 {
5046         struct diff_options *options = opt->value;
5047
5048         BUG_ON_OPT_NEG(unset);
5049         options->use_color = 1;
5050         options->word_diff = DIFF_WORDS_COLOR;
5051         options->word_regex = arg;
5052         return 0;
5053 }
5054
5055 static int diff_opt_compact_summary(const struct option *opt,
5056                                     const char *arg, int unset)
5057 {
5058         struct diff_options *options = opt->value;
5059
5060         BUG_ON_OPT_ARG(arg);
5061         if (unset) {
5062                 options->flags.stat_with_summary = 0;
5063         } else {
5064                 options->flags.stat_with_summary = 1;
5065                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5066         }
5067         return 0;
5068 }
5069
5070 static int diff_opt_diff_algorithm(const struct option *opt,
5071                                    const char *arg, int unset)
5072 {
5073         struct diff_options *options = opt->value;
5074         long value = parse_algorithm_value(arg);
5075
5076         BUG_ON_OPT_NEG(unset);
5077         if (value < 0)
5078                 return error(_("option diff-algorithm accepts \"myers\", "
5079                                "\"minimal\", \"patience\" and \"histogram\""));
5080
5081         /* clear out previous settings */
5082         DIFF_XDL_CLR(options, NEED_MINIMAL);
5083         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5084         options->xdl_opts |= value;
5085         return 0;
5086 }
5087
5088 static int diff_opt_dirstat(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 (!strcmp(opt->long_name, "cumulative")) {
5095                 if (arg)
5096                         BUG("how come --cumulative take a value?");
5097                 arg = "cumulative";
5098         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5099                 parse_dirstat_opt(options, "files");
5100         parse_dirstat_opt(options, arg ? arg : "");
5101         return 0;
5102 }
5103
5104 static int diff_opt_find_copies(const struct option *opt,
5105                                 const char *arg, int unset)
5106 {
5107         struct diff_options *options = opt->value;
5108
5109         BUG_ON_OPT_NEG(unset);
5110         if (!arg)
5111                 arg = "";
5112         options->rename_score = parse_rename_score(&arg);
5113         if (*arg != 0)
5114                 return error(_("invalid argument to %s"), opt->long_name);
5115
5116         if (options->detect_rename == DIFF_DETECT_COPY)
5117                 options->flags.find_copies_harder = 1;
5118         else
5119                 options->detect_rename = DIFF_DETECT_COPY;
5120
5121         return 0;
5122 }
5123
5124 static int diff_opt_find_renames(const struct option *opt,
5125                                  const char *arg, int unset)
5126 {
5127         struct diff_options *options = opt->value;
5128
5129         BUG_ON_OPT_NEG(unset);
5130         if (!arg)
5131                 arg = "";
5132         options->rename_score = parse_rename_score(&arg);
5133         if (*arg != 0)
5134                 return error(_("invalid argument to %s"), opt->long_name);
5135
5136         options->detect_rename = DIFF_DETECT_RENAME;
5137         return 0;
5138 }
5139
5140 static int diff_opt_follow(const struct option *opt,
5141                            const char *arg, int unset)
5142 {
5143         struct diff_options *options = opt->value;
5144
5145         BUG_ON_OPT_ARG(arg);
5146         if (unset) {
5147                 options->flags.follow_renames = 0;
5148                 options->flags.default_follow_renames = 0;
5149         } else {
5150                 options->flags.follow_renames = 1;
5151         }
5152         return 0;
5153 }
5154
5155 static int diff_opt_ignore_submodules(const struct option *opt,
5156                                       const char *arg, int unset)
5157 {
5158         struct diff_options *options = opt->value;
5159
5160         BUG_ON_OPT_NEG(unset);
5161         if (!arg)
5162                 arg = "all";
5163         options->flags.override_submodule_config = 1;
5164         handle_ignore_submodules_arg(options, arg);
5165         return 0;
5166 }
5167
5168 static int diff_opt_line_prefix(const struct option *opt,
5169                                 const char *optarg, int unset)
5170 {
5171         struct diff_options *options = opt->value;
5172
5173         BUG_ON_OPT_NEG(unset);
5174         options->line_prefix = optarg;
5175         options->line_prefix_length = strlen(options->line_prefix);
5176         graph_setup_line_prefix(options);
5177         return 0;
5178 }
5179
5180 static int diff_opt_no_prefix(const struct option *opt,
5181                               const char *optarg, int unset)
5182 {
5183         struct diff_options *options = opt->value;
5184
5185         BUG_ON_OPT_NEG(unset);
5186         BUG_ON_OPT_ARG(optarg);
5187         options->a_prefix = "";
5188         options->b_prefix = "";
5189         return 0;
5190 }
5191
5192 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5193                                              const struct option *opt,
5194                                              const char *arg, int unset)
5195 {
5196         struct diff_options *options = opt->value;
5197         char *path;
5198
5199         BUG_ON_OPT_NEG(unset);
5200         path = prefix_filename(ctx->prefix, arg);
5201         options->file = xfopen(path, "w");
5202         options->close_file = 1;
5203         if (options->use_color != GIT_COLOR_ALWAYS)
5204                 options->use_color = GIT_COLOR_NEVER;
5205         free(path);
5206         return 0;
5207 }
5208
5209 static int diff_opt_patience(const struct option *opt,
5210                              const char *arg, int unset)
5211 {
5212         struct diff_options *options = opt->value;
5213         int i;
5214
5215         BUG_ON_OPT_NEG(unset);
5216         BUG_ON_OPT_ARG(arg);
5217         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5218         /*
5219          * Both --patience and --anchored use PATIENCE_DIFF
5220          * internally, so remove any anchors previously
5221          * specified.
5222          */
5223         for (i = 0; i < options->anchors_nr; i++)
5224                 free(options->anchors[i]);
5225         options->anchors_nr = 0;
5226         return 0;
5227 }
5228
5229 static int diff_opt_ignore_regex(const struct option *opt,
5230                                  const char *arg, int unset)
5231 {
5232         struct diff_options *options = opt->value;
5233         regex_t *regex;
5234
5235         BUG_ON_OPT_NEG(unset);
5236         regex = xmalloc(sizeof(*regex));
5237         if (regcomp(regex, arg, REG_EXTENDED | REG_NEWLINE))
5238                 return error(_("invalid regex given to -I: '%s'"), arg);
5239         ALLOC_GROW(options->ignore_regex, options->ignore_regex_nr + 1,
5240                    options->ignore_regex_alloc);
5241         options->ignore_regex[options->ignore_regex_nr++] = regex;
5242         return 0;
5243 }
5244
5245 static int diff_opt_pickaxe_regex(const struct option *opt,
5246                                   const char *arg, int unset)
5247 {
5248         struct diff_options *options = opt->value;
5249
5250         BUG_ON_OPT_NEG(unset);
5251         options->pickaxe = arg;
5252         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5253         return 0;
5254 }
5255
5256 static int diff_opt_pickaxe_string(const struct option *opt,
5257                                    const char *arg, int unset)
5258 {
5259         struct diff_options *options = opt->value;
5260
5261         BUG_ON_OPT_NEG(unset);
5262         options->pickaxe = arg;
5263         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5264         return 0;
5265 }
5266
5267 static int diff_opt_relative(const struct option *opt,
5268                              const char *arg, int unset)
5269 {
5270         struct diff_options *options = opt->value;
5271
5272         options->flags.relative_name = !unset;
5273         if (arg)
5274                 options->prefix = arg;
5275         return 0;
5276 }
5277
5278 static int diff_opt_submodule(const struct option *opt,
5279                               const char *arg, int unset)
5280 {
5281         struct diff_options *options = opt->value;
5282
5283         BUG_ON_OPT_NEG(unset);
5284         if (!arg)
5285                 arg = "log";
5286         if (parse_submodule_params(options, arg))
5287                 return error(_("failed to parse --submodule option parameter: '%s'"),
5288                              arg);
5289         return 0;
5290 }
5291
5292 static int diff_opt_textconv(const struct option *opt,
5293                              const char *arg, int unset)
5294 {
5295         struct diff_options *options = opt->value;
5296
5297         BUG_ON_OPT_ARG(arg);
5298         if (unset) {
5299                 options->flags.allow_textconv = 0;
5300         } else {
5301                 options->flags.allow_textconv = 1;
5302                 options->flags.textconv_set_via_cmdline = 1;
5303         }
5304         return 0;
5305 }
5306
5307 static int diff_opt_unified(const struct option *opt,
5308                             const char *arg, int unset)
5309 {
5310         struct diff_options *options = opt->value;
5311         char *s;
5312
5313         BUG_ON_OPT_NEG(unset);
5314
5315         if (arg) {
5316                 options->context = strtol(arg, &s, 10);
5317                 if (*s)
5318                         return error(_("%s expects a numerical value"), "--unified");
5319         }
5320         enable_patch_output(&options->output_format);
5321
5322         return 0;
5323 }
5324
5325 static int diff_opt_word_diff(const struct option *opt,
5326                               const char *arg, int unset)
5327 {
5328         struct diff_options *options = opt->value;
5329
5330         BUG_ON_OPT_NEG(unset);
5331         if (arg) {
5332                 if (!strcmp(arg, "plain"))
5333                         options->word_diff = DIFF_WORDS_PLAIN;
5334                 else if (!strcmp(arg, "color")) {
5335                         options->use_color = 1;
5336                         options->word_diff = DIFF_WORDS_COLOR;
5337                 }
5338                 else if (!strcmp(arg, "porcelain"))
5339                         options->word_diff = DIFF_WORDS_PORCELAIN;
5340                 else if (!strcmp(arg, "none"))
5341                         options->word_diff = DIFF_WORDS_NONE;
5342                 else
5343                         return error(_("bad --word-diff argument: %s"), arg);
5344         } else {
5345                 if (options->word_diff == DIFF_WORDS_NONE)
5346                         options->word_diff = DIFF_WORDS_PLAIN;
5347         }
5348         return 0;
5349 }
5350
5351 static int diff_opt_word_diff_regex(const struct option *opt,
5352                                     const char *arg, int unset)
5353 {
5354         struct diff_options *options = opt->value;
5355
5356         BUG_ON_OPT_NEG(unset);
5357         if (options->word_diff == DIFF_WORDS_NONE)
5358                 options->word_diff = DIFF_WORDS_PLAIN;
5359         options->word_regex = arg;
5360         return 0;
5361 }
5362
5363 static int diff_opt_rotate_to(const struct option *opt, const char *arg, int unset)
5364 {
5365         struct diff_options *options = opt->value;
5366
5367         BUG_ON_OPT_NEG(unset);
5368         if (!strcmp(opt->long_name, "skip-to"))
5369                 options->skip_instead_of_rotate = 1;
5370         else
5371                 options->skip_instead_of_rotate = 0;
5372         options->rotate_to = arg;
5373         return 0;
5374 }
5375
5376 static void prep_parse_options(struct diff_options *options)
5377 {
5378         struct option parseopts[] = {
5379                 OPT_GROUP(N_("Diff output format options")),
5380                 OPT_BITOP('p', "patch", &options->output_format,
5381                           N_("generate patch"),
5382                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5383                 OPT_BIT_F('s', "no-patch", &options->output_format,
5384                           N_("suppress diff output"),
5385                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5386                 OPT_BITOP('u', NULL, &options->output_format,
5387                           N_("generate patch"),
5388                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5389                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5390                                N_("generate diffs with <n> lines context"),
5391                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5392                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5393                          N_("generate diffs with <n> lines context")),
5394                 OPT_BIT_F(0, "raw", &options->output_format,
5395                           N_("generate the diff in raw format"),
5396                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5397                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5398                           N_("synonym for '-p --raw'"),
5399                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5400                           DIFF_FORMAT_NO_OUTPUT),
5401                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5402                           N_("synonym for '-p --stat'"),
5403                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5404                           DIFF_FORMAT_NO_OUTPUT),
5405                 OPT_BIT_F(0, "numstat", &options->output_format,
5406                           N_("machine friendly --stat"),
5407                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5408                 OPT_BIT_F(0, "shortstat", &options->output_format,
5409                           N_("output only the last line of --stat"),
5410                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5411                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5412                                N_("output the distribution of relative amount of changes for each sub-directory"),
5413                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5414                                diff_opt_dirstat),
5415                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5416                                N_("synonym for --dirstat=cumulative"),
5417                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5418                                diff_opt_dirstat),
5419                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5420                                N_("synonym for --dirstat=files,param1,param2..."),
5421                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5422                                diff_opt_dirstat),
5423                 OPT_BIT_F(0, "check", &options->output_format,
5424                           N_("warn if changes introduce conflict markers or whitespace errors"),
5425                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5426                 OPT_BIT_F(0, "summary", &options->output_format,
5427                           N_("condensed summary such as creations, renames and mode changes"),
5428                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5429                 OPT_BIT_F(0, "name-only", &options->output_format,
5430                           N_("show only names of changed files"),
5431                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5432                 OPT_BIT_F(0, "name-status", &options->output_format,
5433                           N_("show only names and status of changed files"),
5434                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5435                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5436                                N_("generate diffstat"),
5437                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5438                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5439                                N_("generate diffstat with a given width"),
5440                                PARSE_OPT_NONEG, diff_opt_stat),
5441                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5442                                N_("generate diffstat with a given name width"),
5443                                PARSE_OPT_NONEG, diff_opt_stat),
5444                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5445                                N_("generate diffstat with a given graph width"),
5446                                PARSE_OPT_NONEG, diff_opt_stat),
5447                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5448                                N_("generate diffstat with limited lines"),
5449                                PARSE_OPT_NONEG, diff_opt_stat),
5450                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5451                                N_("generate compact summary in diffstat"),
5452                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5453                 OPT_CALLBACK_F(0, "binary", options, NULL,
5454                                N_("output a binary diff that can be applied"),
5455                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5456                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5457                          N_("show full pre- and post-image object names on the \"index\" lines")),
5458                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5459                                N_("show colored diff")),
5460                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5461                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5462                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5463                 OPT_SET_INT('z', NULL, &options->line_termination,
5464                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5465                             0),
5466                 OPT__ABBREV(&options->abbrev),
5467                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5468                              N_("show the given source prefix instead of \"a/\""),
5469                              PARSE_OPT_NONEG),
5470                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5471                              N_("show the given destination prefix instead of \"b/\""),
5472                              PARSE_OPT_NONEG),
5473                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5474                                N_("prepend an additional prefix to every line of output"),
5475                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5476                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5477                                N_("do not show any source or destination prefix"),
5478                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5479                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5480                               N_("show context between diff hunks up to the specified number of lines"),
5481                               PARSE_OPT_NONEG),
5482                 OPT_CALLBACK_F(0, "output-indicator-new",
5483                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5484                                N_("<char>"),
5485                                N_("specify the character to indicate a new line instead of '+'"),
5486                                PARSE_OPT_NONEG, diff_opt_char),
5487                 OPT_CALLBACK_F(0, "output-indicator-old",
5488                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5489                                N_("<char>"),
5490                                N_("specify the character to indicate an old line instead of '-'"),
5491                                PARSE_OPT_NONEG, diff_opt_char),
5492                 OPT_CALLBACK_F(0, "output-indicator-context",
5493                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5494                                N_("<char>"),
5495                                N_("specify the character to indicate a context instead of ' '"),
5496                                PARSE_OPT_NONEG, diff_opt_char),
5497
5498                 OPT_GROUP(N_("Diff rename options")),
5499                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5500                                N_("break complete rewrite changes into pairs of delete and create"),
5501                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5502                                diff_opt_break_rewrites),
5503                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5504                                N_("detect renames"),
5505                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5506                                diff_opt_find_renames),
5507                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5508                               N_("omit the preimage for deletes"),
5509                               1, PARSE_OPT_NONEG),
5510                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5511                                N_("detect copies"),
5512                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5513                                diff_opt_find_copies),
5514                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5515                          N_("use unmodified files as source to find copies")),
5516                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5517                               N_("disable rename detection"),
5518                               0, PARSE_OPT_NONEG),
5519                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5520                          N_("use empty blobs as rename source")),
5521                 OPT_CALLBACK_F(0, "follow", options, NULL,
5522                                N_("continue listing the history of a file beyond renames"),
5523                                PARSE_OPT_NOARG, diff_opt_follow),
5524                 OPT_INTEGER('l', NULL, &options->rename_limit,
5525                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5526
5527                 OPT_GROUP(N_("Diff algorithm options")),
5528                 OPT_BIT(0, "minimal", &options->xdl_opts,
5529                         N_("produce the smallest possible diff"),
5530                         XDF_NEED_MINIMAL),
5531                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5532                           N_("ignore whitespace when comparing lines"),
5533                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5534                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5535                           N_("ignore changes in amount of whitespace"),
5536                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5537                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5538                           N_("ignore changes in whitespace at EOL"),
5539                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5540                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5541                           N_("ignore carrier-return at the end of line"),
5542                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5543                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5544                           N_("ignore changes whose lines are all blank"),
5545                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5546                 OPT_CALLBACK_F('I', "ignore-matching-lines", options, N_("<regex>"),
5547                                N_("ignore changes whose all lines match <regex>"),
5548                                0, diff_opt_ignore_regex),
5549                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5550                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5551                         XDF_INDENT_HEURISTIC),
5552                 OPT_CALLBACK_F(0, "patience", options, NULL,
5553                                N_("generate diff using the \"patience diff\" algorithm"),
5554                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5555                                diff_opt_patience),
5556                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5557                           N_("generate diff using the \"histogram diff\" algorithm"),
5558                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5559                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5560                                N_("choose a diff algorithm"),
5561                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5562                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5563                                N_("generate diff using the \"anchored diff\" algorithm"),
5564                                PARSE_OPT_NONEG, diff_opt_anchored),
5565                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5566                                N_("show word diff, using <mode> to delimit changed words"),
5567                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5568                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5569                                N_("use <regex> to decide what a word is"),
5570                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5571                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5572                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5573                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5574                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5575                                N_("moved lines of code are colored differently"),
5576                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5577                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5578                                N_("how white spaces are ignored in --color-moved"),
5579                                0, diff_opt_color_moved_ws),
5580
5581                 OPT_GROUP(N_("Other diff options")),
5582                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5583                                N_("when run from subdir, exclude changes outside and show relative paths"),
5584                                PARSE_OPT_OPTARG,
5585                                diff_opt_relative),
5586                 OPT_BOOL('a', "text", &options->flags.text,
5587                          N_("treat all files as text")),
5588                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5589                          N_("swap two inputs, reverse the diff")),
5590                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5591                          N_("exit with 1 if there were differences, 0 otherwise")),
5592                 OPT_BOOL(0, "quiet", &options->flags.quick,
5593                          N_("disable all output of the program")),
5594                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5595                          N_("allow an external diff helper to be executed")),
5596                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5597                                N_("run external text conversion filters when comparing binary files"),
5598                                PARSE_OPT_NOARG, diff_opt_textconv),
5599                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5600                                N_("ignore changes to submodules in the diff generation"),
5601                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5602                                diff_opt_ignore_submodules),
5603                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5604                                N_("specify how differences in submodules are shown"),
5605                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5606                                diff_opt_submodule),
5607                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5608                               N_("hide 'git add -N' entries from the index"),
5609                               1, PARSE_OPT_NONEG),
5610                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5611                               N_("treat 'git add -N' entries as real in the index"),
5612                               0, PARSE_OPT_NONEG),
5613                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5614                                N_("look for differences that change the number of occurrences of the specified string"),
5615                                0, diff_opt_pickaxe_string),
5616                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5617                                N_("look for differences that change the number of occurrences of the specified regex"),
5618                                0, diff_opt_pickaxe_regex),
5619                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5620                           N_("show all changes in the changeset with -S or -G"),
5621                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5622                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5623                           N_("treat <string> in -S as extended POSIX regular expression"),
5624                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5625                 OPT_FILENAME('O', NULL, &options->orderfile,
5626                              N_("control the order in which files appear in the output")),
5627                 OPT_CALLBACK_F(0, "rotate-to", options, N_("<path>"),
5628                                N_("show the change in the specified path first"),
5629                                PARSE_OPT_NONEG, diff_opt_rotate_to),
5630                 OPT_CALLBACK_F(0, "skip-to", options, N_("<path>"),
5631                                N_("skip the output to the specified path"),
5632                                PARSE_OPT_NONEG, diff_opt_rotate_to),
5633                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5634                                N_("look for differences that change the number of occurrences of the specified object"),
5635                                PARSE_OPT_NONEG, diff_opt_find_object),
5636                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5637                                N_("select files by diff type"),
5638                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5639                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5640                   N_("Output to a specific file"),
5641                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5642
5643                 OPT_END()
5644         };
5645
5646         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5647         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5648 }
5649
5650 int diff_opt_parse(struct diff_options *options,
5651                    const char **av, int ac, const char *prefix)
5652 {
5653         if (!prefix)
5654                 prefix = "";
5655
5656         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5657                            PARSE_OPT_KEEP_DASHDASH |
5658                            PARSE_OPT_KEEP_UNKNOWN |
5659                            PARSE_OPT_NO_INTERNAL_HELP |
5660                            PARSE_OPT_ONE_SHOT |
5661                            PARSE_OPT_STOP_AT_NON_OPTION);
5662
5663         return ac;
5664 }
5665
5666 int parse_rename_score(const char **cp_p)
5667 {
5668         unsigned long num, scale;
5669         int ch, dot;
5670         const char *cp = *cp_p;
5671
5672         num = 0;
5673         scale = 1;
5674         dot = 0;
5675         for (;;) {
5676                 ch = *cp;
5677                 if ( !dot && ch == '.' ) {
5678                         scale = 1;
5679                         dot = 1;
5680                 } else if ( ch == '%' ) {
5681                         scale = dot ? scale*100 : 100;
5682                         cp++;   /* % is always at the end */
5683                         break;
5684                 } else if ( ch >= '0' && ch <= '9' ) {
5685                         if ( scale < 100000 ) {
5686                                 scale *= 10;
5687                                 num = (num*10) + (ch-'0');
5688                         }
5689                 } else {
5690                         break;
5691                 }
5692                 cp++;
5693         }
5694         *cp_p = cp;
5695
5696         /* user says num divided by scale and we say internally that
5697          * is MAX_SCORE * num / scale.
5698          */
5699         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5700 }
5701
5702 struct diff_queue_struct diff_queued_diff;
5703
5704 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5705 {
5706         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5707         queue->queue[queue->nr++] = dp;
5708 }
5709
5710 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5711                                  struct diff_filespec *one,
5712                                  struct diff_filespec *two)
5713 {
5714         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5715         dp->one = one;
5716         dp->two = two;
5717         if (queue)
5718                 diff_q(queue, dp);
5719         return dp;
5720 }
5721
5722 void diff_free_filepair(struct diff_filepair *p)
5723 {
5724         free_filespec(p->one);
5725         free_filespec(p->two);
5726         free(p);
5727 }
5728
5729 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5730 {
5731         int abblen;
5732         const char *abbrev;
5733
5734         /* Do we want all 40 hex characters? */
5735         if (len == the_hash_algo->hexsz)
5736                 return oid_to_hex(oid);
5737
5738         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5739         abbrev = diff_abbrev_oid(oid, len);
5740
5741         if (!print_sha1_ellipsis())
5742                 return abbrev;
5743
5744         abblen = strlen(abbrev);
5745
5746         /*
5747          * In well-behaved cases, where the abbreviated result is the
5748          * same as the requested length, append three dots after the
5749          * abbreviation (hence the whole logic is limited to the case
5750          * where abblen < 37); when the actual abbreviated result is a
5751          * bit longer than the requested length, we reduce the number
5752          * of dots so that they match the well-behaved ones.  However,
5753          * if the actual abbreviation is longer than the requested
5754          * length by more than three, we give up on aligning, and add
5755          * three dots anyway, to indicate that the output is not the
5756          * full object name.  Yes, this may be suboptimal, but this
5757          * appears only in "diff --raw --abbrev" output and it is not
5758          * worth the effort to change it now.  Note that this would
5759          * likely to work fine when the automatic sizing of default
5760          * abbreviation length is used--we would be fed -1 in "len" in
5761          * that case, and will end up always appending three-dots, but
5762          * the automatic sizing is supposed to give abblen that ensures
5763          * uniqueness across all objects (statistically speaking).
5764          */
5765         if (abblen < the_hash_algo->hexsz - 3) {
5766                 static char hex[GIT_MAX_HEXSZ + 1];
5767                 if (len < abblen && abblen <= len + 2)
5768                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5769                 else
5770                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5771                 return hex;
5772         }
5773
5774         return oid_to_hex(oid);
5775 }
5776
5777 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5778 {
5779         int line_termination = opt->line_termination;
5780         int inter_name_termination = line_termination ? '\t' : '\0';
5781
5782         fprintf(opt->file, "%s", diff_line_prefix(opt));
5783         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5784                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5785                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5786                 fprintf(opt->file, "%s ",
5787                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5788         }
5789         if (p->score) {
5790                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5791                         inter_name_termination);
5792         } else {
5793                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5794         }
5795
5796         if (p->status == DIFF_STATUS_COPIED ||
5797             p->status == DIFF_STATUS_RENAMED) {
5798                 const char *name_a, *name_b;
5799                 name_a = p->one->path;
5800                 name_b = p->two->path;
5801                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5802                 write_name_quoted(name_a, opt->file, inter_name_termination);
5803                 write_name_quoted(name_b, opt->file, line_termination);
5804         } else {
5805                 const char *name_a, *name_b;
5806                 name_a = p->one->mode ? p->one->path : p->two->path;
5807                 name_b = NULL;
5808                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5809                 write_name_quoted(name_a, opt->file, line_termination);
5810         }
5811 }
5812
5813 int diff_unmodified_pair(struct diff_filepair *p)
5814 {
5815         /* This function is written stricter than necessary to support
5816          * the currently implemented transformers, but the idea is to
5817          * let transformers to produce diff_filepairs any way they want,
5818          * and filter and clean them up here before producing the output.
5819          */
5820         struct diff_filespec *one = p->one, *two = p->two;
5821
5822         if (DIFF_PAIR_UNMERGED(p))
5823                 return 0; /* unmerged is interesting */
5824
5825         /* deletion, addition, mode or type change
5826          * and rename are all interesting.
5827          */
5828         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5829             DIFF_PAIR_MODE_CHANGED(p) ||
5830             strcmp(one->path, two->path))
5831                 return 0;
5832
5833         /* both are valid and point at the same path.  that is, we are
5834          * dealing with a change.
5835          */
5836         if (one->oid_valid && two->oid_valid &&
5837             oideq(&one->oid, &two->oid) &&
5838             !one->dirty_submodule && !two->dirty_submodule)
5839                 return 1; /* no change */
5840         if (!one->oid_valid && !two->oid_valid)
5841                 return 1; /* both look at the same file on the filesystem. */
5842         return 0;
5843 }
5844
5845 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5846 {
5847         if (diff_unmodified_pair(p))
5848                 return;
5849
5850         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5851             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5852                 return; /* no tree diffs in patch format */
5853
5854         run_diff(p, o);
5855 }
5856
5857 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5858                             struct diffstat_t *diffstat)
5859 {
5860         if (diff_unmodified_pair(p))
5861                 return;
5862
5863         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5864             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5865                 return; /* no useful stat for tree diffs */
5866
5867         run_diffstat(p, o, diffstat);
5868 }
5869
5870 static void diff_flush_checkdiff(struct diff_filepair *p,
5871                 struct diff_options *o)
5872 {
5873         if (diff_unmodified_pair(p))
5874                 return;
5875
5876         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5877             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5878                 return; /* nothing to check in tree diffs */
5879
5880         run_checkdiff(p, o);
5881 }
5882
5883 int diff_queue_is_empty(void)
5884 {
5885         struct diff_queue_struct *q = &diff_queued_diff;
5886         int i;
5887         for (i = 0; i < q->nr; i++)
5888                 if (!diff_unmodified_pair(q->queue[i]))
5889                         return 0;
5890         return 1;
5891 }
5892
5893 #if DIFF_DEBUG
5894 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5895 {
5896         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5897                 x, one ? one : "",
5898                 s->path,
5899                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5900                 s->mode,
5901                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5902         fprintf(stderr, "queue[%d] %s size %lu\n",
5903                 x, one ? one : "",
5904                 s->size);
5905 }
5906
5907 void diff_debug_filepair(const struct diff_filepair *p, int i)
5908 {
5909         diff_debug_filespec(p->one, i, "one");
5910         diff_debug_filespec(p->two, i, "two");
5911         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5912                 p->score, p->status ? p->status : '?',
5913                 p->one->rename_used, p->broken_pair);
5914 }
5915
5916 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5917 {
5918         int i;
5919         if (msg)
5920                 fprintf(stderr, "%s\n", msg);
5921         fprintf(stderr, "q->nr = %d\n", q->nr);
5922         for (i = 0; i < q->nr; i++) {
5923                 struct diff_filepair *p = q->queue[i];
5924                 diff_debug_filepair(p, i);
5925         }
5926 }
5927 #endif
5928
5929 static void diff_resolve_rename_copy(void)
5930 {
5931         int i;
5932         struct diff_filepair *p;
5933         struct diff_queue_struct *q = &diff_queued_diff;
5934
5935         diff_debug_queue("resolve-rename-copy", q);
5936
5937         for (i = 0; i < q->nr; i++) {
5938                 p = q->queue[i];
5939                 p->status = 0; /* undecided */
5940                 if (DIFF_PAIR_UNMERGED(p))
5941                         p->status = DIFF_STATUS_UNMERGED;
5942                 else if (!DIFF_FILE_VALID(p->one))
5943                         p->status = DIFF_STATUS_ADDED;
5944                 else if (!DIFF_FILE_VALID(p->two))
5945                         p->status = DIFF_STATUS_DELETED;
5946                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5947                         p->status = DIFF_STATUS_TYPE_CHANGED;
5948
5949                 /* from this point on, we are dealing with a pair
5950                  * whose both sides are valid and of the same type, i.e.
5951                  * either in-place edit or rename/copy edit.
5952                  */
5953                 else if (DIFF_PAIR_RENAME(p)) {
5954                         /*
5955                          * A rename might have re-connected a broken
5956                          * pair up, causing the pathnames to be the
5957                          * same again. If so, that's not a rename at
5958                          * all, just a modification..
5959                          *
5960                          * Otherwise, see if this source was used for
5961                          * multiple renames, in which case we decrement
5962                          * the count, and call it a copy.
5963                          */
5964                         if (!strcmp(p->one->path, p->two->path))
5965                                 p->status = DIFF_STATUS_MODIFIED;
5966                         else if (--p->one->rename_used > 0)
5967                                 p->status = DIFF_STATUS_COPIED;
5968                         else
5969                                 p->status = DIFF_STATUS_RENAMED;
5970                 }
5971                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5972                          p->one->mode != p->two->mode ||
5973                          p->one->dirty_submodule ||
5974                          p->two->dirty_submodule ||
5975                          is_null_oid(&p->one->oid))
5976                         p->status = DIFF_STATUS_MODIFIED;
5977                 else {
5978                         /* This is a "no-change" entry and should not
5979                          * happen anymore, but prepare for broken callers.
5980                          */
5981                         error("feeding unmodified %s to diffcore",
5982                               p->one->path);
5983                         p->status = DIFF_STATUS_UNKNOWN;
5984                 }
5985         }
5986         diff_debug_queue("resolve-rename-copy done", q);
5987 }
5988
5989 static int check_pair_status(struct diff_filepair *p)
5990 {
5991         switch (p->status) {
5992         case DIFF_STATUS_UNKNOWN:
5993                 return 0;
5994         case 0:
5995                 die("internal error in diff-resolve-rename-copy");
5996         default:
5997                 return 1;
5998         }
5999 }
6000
6001 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
6002 {
6003         int fmt = opt->output_format;
6004
6005         if (fmt & DIFF_FORMAT_CHECKDIFF)
6006                 diff_flush_checkdiff(p, opt);
6007         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
6008                 diff_flush_raw(p, opt);
6009         else if (fmt & DIFF_FORMAT_NAME) {
6010                 const char *name_a, *name_b;
6011                 name_a = p->two->path;
6012                 name_b = NULL;
6013                 strip_prefix(opt->prefix_length, &name_a, &name_b);
6014                 fprintf(opt->file, "%s", diff_line_prefix(opt));
6015                 write_name_quoted(name_a, opt->file, opt->line_termination);
6016         }
6017 }
6018
6019 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
6020 {
6021         struct strbuf sb = STRBUF_INIT;
6022         if (fs->mode)
6023                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
6024         else
6025                 strbuf_addf(&sb, " %s ", newdelete);
6026
6027         quote_c_style(fs->path, &sb, NULL, 0);
6028         strbuf_addch(&sb, '\n');
6029         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6030                          sb.buf, sb.len, 0);
6031         strbuf_release(&sb);
6032 }
6033
6034 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
6035                 int show_name)
6036 {
6037         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
6038                 struct strbuf sb = STRBUF_INIT;
6039                 strbuf_addf(&sb, " mode change %06o => %06o",
6040                             p->one->mode, p->two->mode);
6041                 if (show_name) {
6042                         strbuf_addch(&sb, ' ');
6043                         quote_c_style(p->two->path, &sb, NULL, 0);
6044                 }
6045                 strbuf_addch(&sb, '\n');
6046                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6047                                  sb.buf, sb.len, 0);
6048                 strbuf_release(&sb);
6049         }
6050 }
6051
6052 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
6053                 struct diff_filepair *p)
6054 {
6055         struct strbuf sb = STRBUF_INIT;
6056         struct strbuf names = STRBUF_INIT;
6057
6058         pprint_rename(&names, p->one->path, p->two->path);
6059         strbuf_addf(&sb, " %s %s (%d%%)\n",
6060                     renamecopy, names.buf, similarity_index(p));
6061         strbuf_release(&names);
6062         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6063                                  sb.buf, sb.len, 0);
6064         show_mode_change(opt, p, 0);
6065         strbuf_release(&sb);
6066 }
6067
6068 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
6069 {
6070         switch(p->status) {
6071         case DIFF_STATUS_DELETED:
6072                 show_file_mode_name(opt, "delete", p->one);
6073                 break;
6074         case DIFF_STATUS_ADDED:
6075                 show_file_mode_name(opt, "create", p->two);
6076                 break;
6077         case DIFF_STATUS_COPIED:
6078                 show_rename_copy(opt, "copy", p);
6079                 break;
6080         case DIFF_STATUS_RENAMED:
6081                 show_rename_copy(opt, "rename", p);
6082                 break;
6083         default:
6084                 if (p->score) {
6085                         struct strbuf sb = STRBUF_INIT;
6086                         strbuf_addstr(&sb, " rewrite ");
6087                         quote_c_style(p->two->path, &sb, NULL, 0);
6088                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
6089                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6090                                          sb.buf, sb.len, 0);
6091                         strbuf_release(&sb);
6092                 }
6093                 show_mode_change(opt, p, !p->score);
6094                 break;
6095         }
6096 }
6097
6098 struct patch_id_t {
6099         git_hash_ctx *ctx;
6100         int patchlen;
6101 };
6102
6103 static int remove_space(char *line, int len)
6104 {
6105         int i;
6106         char *dst = line;
6107         unsigned char c;
6108
6109         for (i = 0; i < len; i++)
6110                 if (!isspace((c = line[i])))
6111                         *dst++ = c;
6112
6113         return dst - line;
6114 }
6115
6116 void flush_one_hunk(struct object_id *result, git_hash_ctx *ctx)
6117 {
6118         unsigned char hash[GIT_MAX_RAWSZ];
6119         unsigned short carry = 0;
6120         int i;
6121
6122         the_hash_algo->final_fn(hash, ctx);
6123         the_hash_algo->init_fn(ctx);
6124         /* 20-byte sum, with carry */
6125         for (i = 0; i < the_hash_algo->rawsz; ++i) {
6126                 carry += result->hash[i] + hash[i];
6127                 result->hash[i] = carry;
6128                 carry >>= 8;
6129         }
6130 }
6131
6132 static int patch_id_consume(void *priv, char *line, unsigned long len)
6133 {
6134         struct patch_id_t *data = priv;
6135         int new_len;
6136
6137         if (len > 12 && starts_with(line, "\\ "))
6138                 return 0;
6139         new_len = remove_space(line, len);
6140
6141         the_hash_algo->update_fn(data->ctx, line, new_len);
6142         data->patchlen += new_len;
6143         return 0;
6144 }
6145
6146 static void patch_id_add_string(git_hash_ctx *ctx, const char *str)
6147 {
6148         the_hash_algo->update_fn(ctx, str, strlen(str));
6149 }
6150
6151 static void patch_id_add_mode(git_hash_ctx *ctx, unsigned mode)
6152 {
6153         /* large enough for 2^32 in octal */
6154         char buf[12];
6155         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6156         the_hash_algo->update_fn(ctx, buf, len);
6157 }
6158
6159 /* returns 0 upon success, and writes result into oid */
6160 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6161 {
6162         struct diff_queue_struct *q = &diff_queued_diff;
6163         int i;
6164         git_hash_ctx ctx;
6165         struct patch_id_t data;
6166
6167         the_hash_algo->init_fn(&ctx);
6168         memset(&data, 0, sizeof(struct patch_id_t));
6169         data.ctx = &ctx;
6170         oidclr(oid);
6171
6172         for (i = 0; i < q->nr; i++) {
6173                 xpparam_t xpp;
6174                 xdemitconf_t xecfg;
6175                 mmfile_t mf1, mf2;
6176                 struct diff_filepair *p = q->queue[i];
6177                 int len1, len2;
6178
6179                 memset(&xpp, 0, sizeof(xpp));
6180                 memset(&xecfg, 0, sizeof(xecfg));
6181                 if (p->status == 0)
6182                         return error("internal diff status error");
6183                 if (p->status == DIFF_STATUS_UNKNOWN)
6184                         continue;
6185                 if (diff_unmodified_pair(p))
6186                         continue;
6187                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6188                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6189                         continue;
6190                 if (DIFF_PAIR_UNMERGED(p))
6191                         continue;
6192
6193                 diff_fill_oid_info(p->one, options->repo->index);
6194                 diff_fill_oid_info(p->two, options->repo->index);
6195
6196                 len1 = remove_space(p->one->path, strlen(p->one->path));
6197                 len2 = remove_space(p->two->path, strlen(p->two->path));
6198                 patch_id_add_string(&ctx, "diff--git");
6199                 patch_id_add_string(&ctx, "a/");
6200                 the_hash_algo->update_fn(&ctx, p->one->path, len1);
6201                 patch_id_add_string(&ctx, "b/");
6202                 the_hash_algo->update_fn(&ctx, p->two->path, len2);
6203
6204                 if (p->one->mode == 0) {
6205                         patch_id_add_string(&ctx, "newfilemode");
6206                         patch_id_add_mode(&ctx, p->two->mode);
6207                         patch_id_add_string(&ctx, "---/dev/null");
6208                         patch_id_add_string(&ctx, "+++b/");
6209                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6210                 } else if (p->two->mode == 0) {
6211                         patch_id_add_string(&ctx, "deletedfilemode");
6212                         patch_id_add_mode(&ctx, p->one->mode);
6213                         patch_id_add_string(&ctx, "---a/");
6214                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6215                         patch_id_add_string(&ctx, "+++/dev/null");
6216                 } else {
6217                         patch_id_add_string(&ctx, "---a/");
6218                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6219                         patch_id_add_string(&ctx, "+++b/");
6220                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6221                 }
6222
6223                 if (diff_header_only)
6224                         continue;
6225
6226                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6227                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6228                         return error("unable to read files to diff");
6229
6230                 if (diff_filespec_is_binary(options->repo, p->one) ||
6231                     diff_filespec_is_binary(options->repo, p->two)) {
6232                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->one->oid),
6233                                         the_hash_algo->hexsz);
6234                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
6235                                         the_hash_algo->hexsz);
6236                         continue;
6237                 }
6238
6239                 xpp.flags = 0;
6240                 xecfg.ctxlen = 3;
6241                 xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
6242                 if (xdi_diff_outf(&mf1, &mf2, NULL,
6243                                   patch_id_consume, &data, &xpp, &xecfg))
6244                         return error("unable to generate patch-id diff for %s",
6245                                      p->one->path);
6246
6247                 if (stable)
6248                         flush_one_hunk(oid, &ctx);
6249         }
6250
6251         if (!stable)
6252                 the_hash_algo->final_oid_fn(oid, &ctx);
6253
6254         return 0;
6255 }
6256
6257 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6258 {
6259         struct diff_queue_struct *q = &diff_queued_diff;
6260         int i;
6261         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6262
6263         for (i = 0; i < q->nr; i++)
6264                 diff_free_filepair(q->queue[i]);
6265
6266         free(q->queue);
6267         DIFF_QUEUE_CLEAR(q);
6268
6269         return result;
6270 }
6271
6272 static int is_summary_empty(const struct diff_queue_struct *q)
6273 {
6274         int i;
6275
6276         for (i = 0; i < q->nr; i++) {
6277                 const struct diff_filepair *p = q->queue[i];
6278
6279                 switch (p->status) {
6280                 case DIFF_STATUS_DELETED:
6281                 case DIFF_STATUS_ADDED:
6282                 case DIFF_STATUS_COPIED:
6283                 case DIFF_STATUS_RENAMED:
6284                         return 0;
6285                 default:
6286                         if (p->score)
6287                                 return 0;
6288                         if (p->one->mode && p->two->mode &&
6289                             p->one->mode != p->two->mode)
6290                                 return 0;
6291                         break;
6292                 }
6293         }
6294         return 1;
6295 }
6296
6297 static const char rename_limit_warning[] =
6298 N_("inexact rename detection was skipped due to too many files.");
6299
6300 static const char degrade_cc_to_c_warning[] =
6301 N_("only found copies from modified paths due to too many files.");
6302
6303 static const char rename_limit_advice[] =
6304 N_("you may want to set your %s variable to at least "
6305    "%d and retry the command.");
6306
6307 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6308 {
6309         fflush(stdout);
6310         if (degraded_cc)
6311                 warning(_(degrade_cc_to_c_warning));
6312         else if (needed)
6313                 warning(_(rename_limit_warning));
6314         else
6315                 return;
6316         if (0 < needed)
6317                 warning(_(rename_limit_advice), varname, needed);
6318 }
6319
6320 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6321 {
6322         int i;
6323         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6324         struct diff_queue_struct *q = &diff_queued_diff;
6325
6326         if (WSEH_NEW & WS_RULE_MASK)
6327                 BUG("WS rules bit mask overlaps with diff symbol flags");
6328
6329         if (o->color_moved)
6330                 o->emitted_symbols = &esm;
6331
6332         for (i = 0; i < q->nr; i++) {
6333                 struct diff_filepair *p = q->queue[i];
6334                 if (check_pair_status(p))
6335                         diff_flush_patch(p, o);
6336         }
6337
6338         if (o->emitted_symbols) {
6339                 if (o->color_moved) {
6340                         struct hashmap add_lines, del_lines;
6341
6342                         if (o->color_moved_ws_handling &
6343                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6344                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6345
6346                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6347                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6348
6349                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6350                         mark_color_as_moved(o, &add_lines, &del_lines);
6351                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6352                                 dim_moved_lines(o);
6353
6354                         hashmap_clear_and_free(&add_lines, struct moved_entry,
6355                                                 ent);
6356                         hashmap_clear_and_free(&del_lines, struct moved_entry,
6357                                                 ent);
6358                 }
6359
6360                 for (i = 0; i < esm.nr; i++)
6361                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6362
6363                 for (i = 0; i < esm.nr; i++)
6364                         free((void *)esm.buf[i].line);
6365                 esm.nr = 0;
6366
6367                 o->emitted_symbols = NULL;
6368         }
6369 }
6370
6371 static void diff_free_file(struct diff_options *options)
6372 {
6373         if (options->close_file)
6374                 fclose(options->file);
6375 }
6376
6377 static void diff_free_ignore_regex(struct diff_options *options)
6378 {
6379         int i;
6380
6381         for (i = 0; i < options->ignore_regex_nr; i++) {
6382                 regfree(options->ignore_regex[i]);
6383                 free(options->ignore_regex[i]);
6384         }
6385         free(options->ignore_regex);
6386 }
6387
6388 void diff_free(struct diff_options *options)
6389 {
6390         if (options->no_free)
6391                 return;
6392
6393         diff_free_file(options);
6394         diff_free_ignore_regex(options);
6395 }
6396
6397 void diff_flush(struct diff_options *options)
6398 {
6399         struct diff_queue_struct *q = &diff_queued_diff;
6400         int i, output_format = options->output_format;
6401         int separator = 0;
6402         int dirstat_by_line = 0;
6403
6404         /*
6405          * Order: raw, stat, summary, patch
6406          * or:    name/name-status/checkdiff (other bits clear)
6407          */
6408         if (!q->nr)
6409                 goto free_queue;
6410
6411         if (output_format & (DIFF_FORMAT_RAW |
6412                              DIFF_FORMAT_NAME |
6413                              DIFF_FORMAT_NAME_STATUS |
6414                              DIFF_FORMAT_CHECKDIFF)) {
6415                 for (i = 0; i < q->nr; i++) {
6416                         struct diff_filepair *p = q->queue[i];
6417                         if (check_pair_status(p))
6418                                 flush_one_pair(p, options);
6419                 }
6420                 separator++;
6421         }
6422
6423         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6424                 dirstat_by_line = 1;
6425
6426         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6427             dirstat_by_line) {
6428                 struct diffstat_t diffstat;
6429
6430                 compute_diffstat(options, &diffstat, q);
6431                 if (output_format & DIFF_FORMAT_NUMSTAT)
6432                         show_numstat(&diffstat, options);
6433                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6434                         show_stats(&diffstat, options);
6435                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6436                         show_shortstats(&diffstat, options);
6437                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6438                         show_dirstat_by_line(&diffstat, options);
6439                 free_diffstat_info(&diffstat);
6440                 separator++;
6441         }
6442         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6443                 show_dirstat(options);
6444
6445         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6446                 for (i = 0; i < q->nr; i++) {
6447                         diff_summary(options, q->queue[i]);
6448                 }
6449                 separator++;
6450         }
6451
6452         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6453             options->flags.exit_with_status &&
6454             options->flags.diff_from_contents) {
6455                 /*
6456                  * run diff_flush_patch for the exit status. setting
6457                  * options->file to /dev/null should be safe, because we
6458                  * aren't supposed to produce any output anyway.
6459                  */
6460                 diff_free_file(options);
6461                 options->file = xfopen("/dev/null", "w");
6462                 options->close_file = 1;
6463                 options->color_moved = 0;
6464                 for (i = 0; i < q->nr; i++) {
6465                         struct diff_filepair *p = q->queue[i];
6466                         if (check_pair_status(p))
6467                                 diff_flush_patch(p, options);
6468                         if (options->found_changes)
6469                                 break;
6470                 }
6471         }
6472
6473         if (output_format & DIFF_FORMAT_PATCH) {
6474                 if (separator) {
6475                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6476                         if (options->stat_sep)
6477                                 /* attach patch instead of inline */
6478                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6479                                                  NULL, 0, 0);
6480                 }
6481
6482                 diff_flush_patch_all_file_pairs(options);
6483         }
6484
6485         if (output_format & DIFF_FORMAT_CALLBACK)
6486                 options->format_callback(q, options, options->format_callback_data);
6487
6488         for (i = 0; i < q->nr; i++)
6489                 diff_free_filepair(q->queue[i]);
6490 free_queue:
6491         free(q->queue);
6492         DIFF_QUEUE_CLEAR(q);
6493         diff_free(options);
6494
6495         /*
6496          * Report the content-level differences with HAS_CHANGES;
6497          * diff_addremove/diff_change does not set the bit when
6498          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6499          */
6500         if (options->flags.diff_from_contents) {
6501                 if (options->found_changes)
6502                         options->flags.has_changes = 1;
6503                 else
6504                         options->flags.has_changes = 0;
6505         }
6506 }
6507
6508 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6509 {
6510         return (((p->status == DIFF_STATUS_MODIFIED) &&
6511                  ((p->score &&
6512                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6513                   (!p->score &&
6514                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6515                 ((p->status != DIFF_STATUS_MODIFIED) &&
6516                  filter_bit_tst(p->status, options)));
6517 }
6518
6519 static void diffcore_apply_filter(struct diff_options *options)
6520 {
6521         int i;
6522         struct diff_queue_struct *q = &diff_queued_diff;
6523         struct diff_queue_struct outq;
6524
6525         DIFF_QUEUE_CLEAR(&outq);
6526
6527         if (!options->filter)
6528                 return;
6529
6530         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6531                 int found;
6532                 for (i = found = 0; !found && i < q->nr; i++) {
6533                         if (match_filter(options, q->queue[i]))
6534                                 found++;
6535                 }
6536                 if (found)
6537                         return;
6538
6539                 /* otherwise we will clear the whole queue
6540                  * by copying the empty outq at the end of this
6541                  * function, but first clear the current entries
6542                  * in the queue.
6543                  */
6544                 for (i = 0; i < q->nr; i++)
6545                         diff_free_filepair(q->queue[i]);
6546         }
6547         else {
6548                 /* Only the matching ones */
6549                 for (i = 0; i < q->nr; i++) {
6550                         struct diff_filepair *p = q->queue[i];
6551                         if (match_filter(options, p))
6552                                 diff_q(&outq, p);
6553                         else
6554                                 diff_free_filepair(p);
6555                 }
6556         }
6557         free(q->queue);
6558         *q = outq;
6559 }
6560
6561 /* Check whether two filespecs with the same mode and size are identical */
6562 static int diff_filespec_is_identical(struct repository *r,
6563                                       struct diff_filespec *one,
6564                                       struct diff_filespec *two)
6565 {
6566         if (S_ISGITLINK(one->mode))
6567                 return 0;
6568         if (diff_populate_filespec(r, one, NULL))
6569                 return 0;
6570         if (diff_populate_filespec(r, two, NULL))
6571                 return 0;
6572         return !memcmp(one->data, two->data, one->size);
6573 }
6574
6575 static int diff_filespec_check_stat_unmatch(struct repository *r,
6576                                             struct diff_filepair *p)
6577 {
6578         struct diff_populate_filespec_options dpf_options = {
6579                 .check_size_only = 1,
6580                 .missing_object_cb = diff_queued_diff_prefetch,
6581                 .missing_object_data = r,
6582         };
6583
6584         if (p->done_skip_stat_unmatch)
6585                 return p->skip_stat_unmatch_result;
6586
6587         p->done_skip_stat_unmatch = 1;
6588         p->skip_stat_unmatch_result = 0;
6589         /*
6590          * 1. Entries that come from stat info dirtiness
6591          *    always have both sides (iow, not create/delete),
6592          *    one side of the object name is unknown, with
6593          *    the same mode and size.  Keep the ones that
6594          *    do not match these criteria.  They have real
6595          *    differences.
6596          *
6597          * 2. At this point, the file is known to be modified,
6598          *    with the same mode and size, and the object
6599          *    name of one side is unknown.  Need to inspect
6600          *    the identical contents.
6601          */
6602         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6603             !DIFF_FILE_VALID(p->two) ||
6604             (p->one->oid_valid && p->two->oid_valid) ||
6605             (p->one->mode != p->two->mode) ||
6606             diff_populate_filespec(r, p->one, &dpf_options) ||
6607             diff_populate_filespec(r, p->two, &dpf_options) ||
6608             (p->one->size != p->two->size) ||
6609             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6610                 p->skip_stat_unmatch_result = 1;
6611         return p->skip_stat_unmatch_result;
6612 }
6613
6614 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6615 {
6616         int i;
6617         struct diff_queue_struct *q = &diff_queued_diff;
6618         struct diff_queue_struct outq;
6619         DIFF_QUEUE_CLEAR(&outq);
6620
6621         for (i = 0; i < q->nr; i++) {
6622                 struct diff_filepair *p = q->queue[i];
6623
6624                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6625                         diff_q(&outq, p);
6626                 else {
6627                         /*
6628                          * The caller can subtract 1 from skip_stat_unmatch
6629                          * to determine how many paths were dirty only
6630                          * due to stat info mismatch.
6631                          */
6632                         if (!diffopt->flags.no_index)
6633                                 diffopt->skip_stat_unmatch++;
6634                         diff_free_filepair(p);
6635                 }
6636         }
6637         free(q->queue);
6638         *q = outq;
6639 }
6640
6641 static int diffnamecmp(const void *a_, const void *b_)
6642 {
6643         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6644         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6645         const char *name_a, *name_b;
6646
6647         name_a = a->one ? a->one->path : a->two->path;
6648         name_b = b->one ? b->one->path : b->two->path;
6649         return strcmp(name_a, name_b);
6650 }
6651
6652 void diffcore_fix_diff_index(void)
6653 {
6654         struct diff_queue_struct *q = &diff_queued_diff;
6655         QSORT(q->queue, q->nr, diffnamecmp);
6656 }
6657
6658 void diff_add_if_missing(struct repository *r,
6659                          struct oid_array *to_fetch,
6660                          const struct diff_filespec *filespec)
6661 {
6662         if (filespec && filespec->oid_valid &&
6663             !S_ISGITLINK(filespec->mode) &&
6664             oid_object_info_extended(r, &filespec->oid, NULL,
6665                                      OBJECT_INFO_FOR_PREFETCH))
6666                 oid_array_append(to_fetch, &filespec->oid);
6667 }
6668
6669 void diff_queued_diff_prefetch(void *repository)
6670 {
6671         struct repository *repo = repository;
6672         int i;
6673         struct diff_queue_struct *q = &diff_queued_diff;
6674         struct oid_array to_fetch = OID_ARRAY_INIT;
6675
6676         for (i = 0; i < q->nr; i++) {
6677                 struct diff_filepair *p = q->queue[i];
6678                 diff_add_if_missing(repo, &to_fetch, p->one);
6679                 diff_add_if_missing(repo, &to_fetch, p->two);
6680         }
6681
6682         /*
6683          * NEEDSWORK: Consider deduplicating the OIDs sent.
6684          */
6685         promisor_remote_get_direct(repo, to_fetch.oid, to_fetch.nr);
6686
6687         oid_array_clear(&to_fetch);
6688 }
6689
6690 void diffcore_std(struct diff_options *options)
6691 {
6692         int output_formats_to_prefetch = DIFF_FORMAT_DIFFSTAT |
6693                 DIFF_FORMAT_NUMSTAT |
6694                 DIFF_FORMAT_PATCH |
6695                 DIFF_FORMAT_SHORTSTAT |
6696                 DIFF_FORMAT_DIRSTAT;
6697
6698         /*
6699          * Check if the user requested a blob-data-requiring diff output and/or
6700          * break-rewrite detection (which requires blob data). If yes, prefetch
6701          * the diff pairs.
6702          *
6703          * If no prefetching occurs, diffcore_rename() will prefetch if it
6704          * decides that it needs inexact rename detection.
6705          */
6706         if (options->repo == the_repository && has_promisor_remote() &&
6707             (options->output_format & output_formats_to_prefetch ||
6708              options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
6709                 diff_queued_diff_prefetch(options->repo);
6710
6711         /* NOTE please keep the following in sync with diff_tree_combined() */
6712         if (options->skip_stat_unmatch)
6713                 diffcore_skip_stat_unmatch(options);
6714         if (!options->found_follow) {
6715                 /* See try_to_follow_renames() in tree-diff.c */
6716                 if (options->break_opt != -1)
6717                         diffcore_break(options->repo,
6718                                        options->break_opt);
6719                 if (options->detect_rename)
6720                         diffcore_rename(options);
6721                 if (options->break_opt != -1)
6722                         diffcore_merge_broken();
6723         }
6724         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6725                 diffcore_pickaxe(options);
6726         if (options->orderfile)
6727                 diffcore_order(options->orderfile);
6728         if (options->rotate_to)
6729                 diffcore_rotate(options);
6730         if (!options->found_follow)
6731                 /* See try_to_follow_renames() in tree-diff.c */
6732                 diff_resolve_rename_copy();
6733         diffcore_apply_filter(options);
6734
6735         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6736                 options->flags.has_changes = 1;
6737         else
6738                 options->flags.has_changes = 0;
6739
6740         options->found_follow = 0;
6741 }
6742
6743 int diff_result_code(struct diff_options *opt, int status)
6744 {
6745         int result = 0;
6746
6747         diff_warn_rename_limit("diff.renameLimit",
6748                                opt->needed_rename_limit,
6749                                opt->degraded_cc_to_c);
6750         if (!opt->flags.exit_with_status &&
6751             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6752                 return status;
6753         if (opt->flags.exit_with_status &&
6754             opt->flags.has_changes)
6755                 result |= 01;
6756         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6757             opt->flags.check_failed)
6758                 result |= 02;
6759         return result;
6760 }
6761
6762 int diff_can_quit_early(struct diff_options *opt)
6763 {
6764         return (opt->flags.quick &&
6765                 !opt->filter &&
6766                 opt->flags.has_changes);
6767 }
6768
6769 /*
6770  * Shall changes to this submodule be ignored?
6771  *
6772  * Submodule changes can be configured to be ignored separately for each path,
6773  * but that configuration can be overridden from the command line.
6774  */
6775 static int is_submodule_ignored(const char *path, struct diff_options *options)
6776 {
6777         int ignored = 0;
6778         struct diff_flags orig_flags = options->flags;
6779         if (!options->flags.override_submodule_config)
6780                 set_diffopt_flags_from_submodule_config(options, path);
6781         if (options->flags.ignore_submodules)
6782                 ignored = 1;
6783         options->flags = orig_flags;
6784         return ignored;
6785 }
6786
6787 void compute_diffstat(struct diff_options *options,
6788                       struct diffstat_t *diffstat,
6789                       struct diff_queue_struct *q)
6790 {
6791         int i;
6792
6793         memset(diffstat, 0, sizeof(struct diffstat_t));
6794         for (i = 0; i < q->nr; i++) {
6795                 struct diff_filepair *p = q->queue[i];
6796                 if (check_pair_status(p))
6797                         diff_flush_stat(p, options, diffstat);
6798         }
6799 }
6800
6801 void diff_addremove(struct diff_options *options,
6802                     int addremove, unsigned mode,
6803                     const struct object_id *oid,
6804                     int oid_valid,
6805                     const char *concatpath, unsigned dirty_submodule)
6806 {
6807         struct diff_filespec *one, *two;
6808
6809         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6810                 return;
6811
6812         /* This may look odd, but it is a preparation for
6813          * feeding "there are unchanged files which should
6814          * not produce diffs, but when you are doing copy
6815          * detection you would need them, so here they are"
6816          * entries to the diff-core.  They will be prefixed
6817          * with something like '=' or '*' (I haven't decided
6818          * which but should not make any difference).
6819          * Feeding the same new and old to diff_change()
6820          * also has the same effect.
6821          * Before the final output happens, they are pruned after
6822          * merged into rename/copy pairs as appropriate.
6823          */
6824         if (options->flags.reverse_diff)
6825                 addremove = (addremove == '+' ? '-' :
6826                              addremove == '-' ? '+' : addremove);
6827
6828         if (options->prefix &&
6829             strncmp(concatpath, options->prefix, options->prefix_length))
6830                 return;
6831
6832         one = alloc_filespec(concatpath);
6833         two = alloc_filespec(concatpath);
6834
6835         if (addremove != '+')
6836                 fill_filespec(one, oid, oid_valid, mode);
6837         if (addremove != '-') {
6838                 fill_filespec(two, oid, oid_valid, mode);
6839                 two->dirty_submodule = dirty_submodule;
6840         }
6841
6842         diff_queue(&diff_queued_diff, one, two);
6843         if (!options->flags.diff_from_contents)
6844                 options->flags.has_changes = 1;
6845 }
6846
6847 void diff_change(struct diff_options *options,
6848                  unsigned old_mode, unsigned new_mode,
6849                  const struct object_id *old_oid,
6850                  const struct object_id *new_oid,
6851                  int old_oid_valid, int new_oid_valid,
6852                  const char *concatpath,
6853                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6854 {
6855         struct diff_filespec *one, *two;
6856         struct diff_filepair *p;
6857
6858         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6859             is_submodule_ignored(concatpath, options))
6860                 return;
6861
6862         if (options->flags.reverse_diff) {
6863                 SWAP(old_mode, new_mode);
6864                 SWAP(old_oid, new_oid);
6865                 SWAP(old_oid_valid, new_oid_valid);
6866                 SWAP(old_dirty_submodule, new_dirty_submodule);
6867         }
6868
6869         if (options->prefix &&
6870             strncmp(concatpath, options->prefix, options->prefix_length))
6871                 return;
6872
6873         one = alloc_filespec(concatpath);
6874         two = alloc_filespec(concatpath);
6875         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6876         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6877         one->dirty_submodule = old_dirty_submodule;
6878         two->dirty_submodule = new_dirty_submodule;
6879         p = diff_queue(&diff_queued_diff, one, two);
6880
6881         if (options->flags.diff_from_contents)
6882                 return;
6883
6884         if (options->flags.quick && options->skip_stat_unmatch &&
6885             !diff_filespec_check_stat_unmatch(options->repo, p)) {
6886                 diff_free_filespec_data(p->one);
6887                 diff_free_filespec_data(p->two);
6888                 return;
6889         }
6890
6891         options->flags.has_changes = 1;
6892 }
6893
6894 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6895 {
6896         struct diff_filepair *pair;
6897         struct diff_filespec *one, *two;
6898
6899         if (options->prefix &&
6900             strncmp(path, options->prefix, options->prefix_length))
6901                 return NULL;
6902
6903         one = alloc_filespec(path);
6904         two = alloc_filespec(path);
6905         pair = diff_queue(&diff_queued_diff, one, two);
6906         pair->is_unmerged = 1;
6907         return pair;
6908 }
6909
6910 static char *run_textconv(struct repository *r,
6911                           const char *pgm,
6912                           struct diff_filespec *spec,
6913                           size_t *outsize)
6914 {
6915         struct diff_tempfile *temp;
6916         const char *argv[3];
6917         const char **arg = argv;
6918         struct child_process child = CHILD_PROCESS_INIT;
6919         struct strbuf buf = STRBUF_INIT;
6920         int err = 0;
6921
6922         temp = prepare_temp_file(r, spec->path, spec);
6923         *arg++ = pgm;
6924         *arg++ = temp->name;
6925         *arg = NULL;
6926
6927         child.use_shell = 1;
6928         child.argv = argv;
6929         child.out = -1;
6930         if (start_command(&child)) {
6931                 remove_tempfile();
6932                 return NULL;
6933         }
6934
6935         if (strbuf_read(&buf, child.out, 0) < 0)
6936                 err = error("error reading from textconv command '%s'", pgm);
6937         close(child.out);
6938
6939         if (finish_command(&child) || err) {
6940                 strbuf_release(&buf);
6941                 remove_tempfile();
6942                 return NULL;
6943         }
6944         remove_tempfile();
6945
6946         return strbuf_detach(&buf, outsize);
6947 }
6948
6949 size_t fill_textconv(struct repository *r,
6950                      struct userdiff_driver *driver,
6951                      struct diff_filespec *df,
6952                      char **outbuf)
6953 {
6954         size_t size;
6955
6956         if (!driver) {
6957                 if (!DIFF_FILE_VALID(df)) {
6958                         *outbuf = "";
6959                         return 0;
6960                 }
6961                 if (diff_populate_filespec(r, df, NULL))
6962                         die("unable to read files to diff");
6963                 *outbuf = df->data;
6964                 return df->size;
6965         }
6966
6967         if (!driver->textconv)
6968                 BUG("fill_textconv called with non-textconv driver");
6969
6970         if (driver->textconv_cache && df->oid_valid) {
6971                 *outbuf = notes_cache_get(driver->textconv_cache,
6972                                           &df->oid,
6973                                           &size);
6974                 if (*outbuf)
6975                         return size;
6976         }
6977
6978         *outbuf = run_textconv(r, driver->textconv, df, &size);
6979         if (!*outbuf)
6980                 die("unable to read files to diff");
6981
6982         if (driver->textconv_cache && df->oid_valid) {
6983                 /* ignore errors, as we might be in a readonly repository */
6984                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6985                                 size);
6986                 /*
6987                  * we could save up changes and flush them all at the end,
6988                  * but we would need an extra call after all diffing is done.
6989                  * Since generating a cache entry is the slow path anyway,
6990                  * this extra overhead probably isn't a big deal.
6991                  */
6992                 notes_cache_write(driver->textconv_cache);
6993         }
6994
6995         return size;
6996 }
6997
6998 int textconv_object(struct repository *r,
6999                     const char *path,
7000                     unsigned mode,
7001                     const struct object_id *oid,
7002                     int oid_valid,
7003                     char **buf,
7004                     unsigned long *buf_size)
7005 {
7006         struct diff_filespec *df;
7007         struct userdiff_driver *textconv;
7008
7009         df = alloc_filespec(path);
7010         fill_filespec(df, oid, oid_valid, mode);
7011         textconv = get_textconv(r, df);
7012         if (!textconv) {
7013                 free_filespec(df);
7014                 return 0;
7015         }
7016
7017         *buf_size = fill_textconv(r, textconv, df, buf);
7018         free_filespec(df);
7019         return 1;
7020 }
7021
7022 void setup_diff_pager(struct diff_options *opt)
7023 {
7024         /*
7025          * If the user asked for our exit code, then either they want --quiet
7026          * or --exit-code. We should definitely not bother with a pager in the
7027          * former case, as we will generate no output. Since we still properly
7028          * report our exit code even when a pager is run, we _could_ run a
7029          * pager with --exit-code. But since we have not done so historically,
7030          * and because it is easy to find people oneline advising "git diff
7031          * --exit-code" in hooks and other scripts, we do not do so.
7032          */
7033         if (!opt->flags.exit_with_status &&
7034             check_pager_config("diff") != 0)
7035                 setup_pager();
7036 }